The type III secretion system (TTSS) encoded by Salmonella Pathogenicity Island 2 (SPI-2) is required for systemic infection and intracellular replication of Salmonella enterica serovar Typhimurium. The SPI-2 TTSS is activated after internalization of bacteria by host cells, and translocates effector proteins into and across the vacuolar membrane, where they interfere with several host cell functions. Here, we investigated the function of SsaM, a small protein encoded within SPI-2. An ssaM deletion mutant had virulence and intracellular replication defects comparable to those of a SPI-2 TTSS null mutant. Although the ssaM mutant was able to secrete the effector protein SseJ in vitro, it failed to translocate SseJ into host cells, and to secrete the translocon proteins SseB, SseC and SseD in vitro. This phenotype is similar to that of a strain carrying a mutation in the SPI-2 gene spiC, whose product is reported to be an effector involved in trafficking of the Salmonella vacuole in macrophages. Both ssaM and spiC mutants were found to oversecrete the SPI-2 effector proteins SseJ and PipB in vitro. Fractionation assays and immunofluorescence microscopy were used to investigate the localization of SsaM and SpiC in macrophages. No evidence for translocation of these proteins was obtained. The similar phenotypes of the ssaM and spiC mutants suggested that they might be involved in the same function. Pull-down and co-immune precipitation experiments showed that SpiC and SsaM interact within the bacterial cell. We propose that a complex involving SsaM and SpiC distinguishes between translocators and effector proteins, and controls their ordered secretion through the SPI-2 TTSS.
Type III secretion systems (TTSSs) are multiprotein structures found in a variety of Gram-negative bacterial pathogens, which transport proteins across the two membranes of the bacterial cell and a third membrane of the target host cell (Cornelis and Van Gijsegem, 2000). Salmonella enterica encodes two TTSSs which perform a range of virulence functions during infection. The Salmonella Pathogenicity Island 1 (SPI-1) TTSS is activated in extracellular bacteria and translocates effectors through the host cell plasma membrane. The action of some of these effectors leads to localized actin polymerization, membrane ruffling and entry of bacteria into the host cell (Galán, 2001), where they remain in a membrane-bound compartment known as the Salmonella-containing vacuole (SCV). The second TTSS, which is also encoded by a pathogenicity island (SPI-2), is activated in intracellular bacteria, probably in response to low [Ca2+] and [Fe2+], acidic pH and low osmolarity (Lee et al., 2000; Zaharik et al., 2002; Garmendia et al., 2003) in the SCV. SPI-2 gene expression commences ≈1 h after uptake of bacteria by host cells, and increases over the next few hours (Cirillo et al., 1998; Uchiya et al., 1999; Beuzón et al., 2000; Knodler et al., 2002; Kuhle and Hensel, 2002; Garmendia et al., 2003). The expression of SPI-2 genes and associated effectors is controlled by the SsrA-SsrB two component regulatory system, encoded within SPI-2 (Cirillo et al., 1998).
The maturation of the SCV in host cells is characterized by its ability to avoid extensive interactions with the cation-independent mannose-6-phosphate receptor (MPR) and enzymes such as cathepsin L and D which are normally delivered to phagosomes by the MPR (García-del Portillo and Finlay, 1995; Rathman et al., 1997). However, the SCV does interact with some late endocytic compartments, and recruits lysosomal membrane glycoproteins (lgps) including LAMP1, which are incorporated in the SCV membrane (García-del Portillo and Finlay, 1995; Rathman et al., 1997; Beuzón et al., 2002). In epithelial cells, Salmonella-induced reorganization of lgps is very dramatic, resulting in the formation of characteristic lgp-enriched tubular structures called Salmonella-induced filaments (Sifs), which extend from SCVs and appear ≈6 h after invasion (García-del Portillo et al., 1993).
SpiC was reported to be translocated into the cytosol of macrophages and to have a role in preventing interactions between SCVs and late endocytic compartments (Uchiya et al., 1999). Subsequent studies identified two host cell proteins, TassC (Lee et al., 2002) and Hook3 (Shotland et al., 2003), as host cell interacting partners of SpiC. The function of TassC is unknown, but Hook3 is a microtubule-binding linker protein associated with the Golgi network (Walenta et al., 2001). The relationship between the normal functions of these proteins and trafficking of SCVs is not clear.
Genes of SPI-2 encoding the secretion system are located in a region of 26 kb of the pathogenicity island, and are organized as four operons: regulatory, structural I, structural II and effector/chaperone (Shea et al., 1996; Cirillo et al., 1998; Hensel et al., 1998). Many of these genes or their products have significant sequence similarity to genes or proteins of other TTSS, allowing the functions and/or cellular locations of some of the proteins to be inferred. These include SsaN, SsaR, SsaS, SsaT, SsaU and SsaV (putative proto-channel components), SsaD/SpiB, SsaJ, SsaK and SsaQ (putative basal components), and SsaC/SpiA (a putative outer ring protein) (Aizawa, 2001). The functions of several other SPI-2 genes are unknown. These include ssaM, located in the structural II operon (Hensel et al., 1997). A screen of a MudJ mutant library identified a strain carrying an insertion in ssaM that was incapable of forming Sifs (Guy et al., 2000), suggesting that SsaM might have an important role in the TTSS. However, ssaM is located upstream of ssaV, which is known to be essential for secretion of SPI-2 proteins (Beuzón et al., 1999). Therefore, a polar effect of the MudJ insertion might have been responsible for the lack of Sifs in cells infected with this mutant.
In this study, we investigated the function of SsaM in more detail. We found that a strain carrying a non-polar deletion of ssaM has marked defects in virulence and intracellular replication. In vitro, SsaM is essential for secretion of SPI-2 translocon components but its absence leads to oversecretion of at least two effector proteins. A strain carrying a mutation in spiC has a phenotype very similar to that of the ssaM mutant. No evidence for secretion or translocation of either protein was obtained, but SsaM and SpiC were found to interact within the bacterial cell. We propose that a complex involving SsaM and SpiC distinguishes between translocators and effector proteins, and controls their ordered secretion through the SPI-2 TTSS.
Virulence defect of an ssaM mutant
The ssaM gene is predicted to encode a cytoplasmic protein of 122 amino acids, with a pI of 8.5. blastp database searches with the predicted amino acid sequence of SsaM identified a protein with 29% identity to SsaM, predicted to be part of a TTSS in Chromobacterium violaceum (Brazilian National Genome Project Consortium, 2003). psi-blast searches against the sequences of these two proteins identified a predicted protein with 17% identity and 41% similarity over first 93 amino acids, encoded by ORF12 in the locus of enterocyte effacement (LEE) of enteropathogenic Escherichia coli: this pathogenicity island also encodes a TTSS (Elliott et al., 1998).
To investigate the function of SsaM, the ssaM gene was mutated by replacing the sequence encoding amino acids 10–103 with the kanamycin resistance gene aphA-3 (Ménard et al., 1993). The ssaM mutant was assessed for virulence attenuation in mice. The competitive index (CI), which provides a sensitive measure of the relative degree of attenuation (Beuzón and Holden, 2001), was determined 48 h after intraperitoneal (i.p.) inoculation of a mixture of mutant and wild-type strains. The ssaM mutant had a CI of 0.0027 ± 0.00001, similar to that of a SPI-2 null mutant (Beuzón et al., 2000). The virulence of the ssaM mutant containing the wild-type allele of ssaM on a plasmid was indistinguishable from the wild-type strain (the CI of ssaM pssaM strain versus the wild-type strain was 1.012 ± 0.074). This shows that the attenuation results from loss of ssaM and not from a polar effect on another gene in the operon, or a secondary mutation elsewhere in the genome. Other SPI-2 genes that are required for systemic virulence in mice are also required for bacterial replication within macrophages (Hensel et al., 1995; 1998; Ochman et al., 1996; Cirillo et al., 1998; Beuzón et al., 2000). In keeping with this, the ssaM mutant had a noticeable replication defect in RAW macrophages, which was rescued by complementation with the plasmid containing the wild-type ssaM allele (Fig. 1).
The degree of attenuation of the ssaM mutant suggested that loss of SsaM could result in a completely inactive secretion system. If this is the case, the ssaM mutant should fail to produce SPI-2-associated phenotypes, such as Sifs. To determine whether ssaM is required for Sif formation, HeLa cells were infected with the ssaM mutant for 8 h, then fixed and labelled with an antibody against LAMP1. The ssaM mutant was completely unable to form Sifs, and this defect was not observed in the ssaM mutant carrying the complementing plasmid (results not shown).
These results suggested that the ssaM mutant might be unable to secrete or translocate other SPI-2 effectors. To test this possibility, we investigated the translocation of the SPI-2 effector SseJ (Miao and Miller, 2000; Ruiz-Albert et al., 2002) by the ssaM mutant strain. To detect SseJ, chromosomal sseJ was modified by the one-step polymerase chain reaction (PCR) method (Uzzau et al., 2001) to incorporate a sequence encoding a C-terminal double haemagglutinin (HA) epitope tag (SseJ-2HA) at the 3′ end of the gene. Consistent with previous reports (Freeman et al., 2002; 2003; Kuhle and Hensel, 2002), we found that SseJ-2HA was translocated across the vacuolar membrane and associated with Sifs in HeLa cells infected with wild-type bacteria (Fig. 2). SseJ-2HA was also translocated by the complemented ssaM mutant, but no translocation of the tagged protein was detected in cells infected with the ssaM mutant expressing SseJ-2HA (Fig. 2).
The ssaM mutant oversecretes effector proteins in vitro
The location of ssaM within the structural II operon of SPI-2, the marked virulence attenuation of the ssaM mutant and its failure to translocate SseJ suggested that SsaM might be an essential component of secretion apparatus. If so, then the ssaM mutant would be defective in the secretion of SPI-2 proteins. To determine whether this is the case, bacterial strains expressing SseJ-2HA from the chromosome were grown in a medium [magnesium minimal medium MES (MgM-MES)] that induces the expression and secretion of the SPI-2 TTSS and its effectors (Beuzón et al., 1999; Hansen-Wester et al., 2002). Under these conditions, SPI-2-secreted proteins are barely detectable in the culture supernatant, even following precipitation or other methods of concentration (data not shown), but accumulate on the bacterial cell surface and plastic surface of the tube in which the bacteria are grown (Beuzón et al., 1999). Therefore, extracellular proteins were recovered from these locations, separated by SDS-PAGE and analysed by immunoblotting. Wild-type bacteria consistently appeared to secrete slightly more flagellin protein FliC than ssaM mutant bacteria, but much greater amounts of SseJ-2HA were recovered from the cell surface of the ssaM mutant using hexadecane, and the inner surface of the tube containing this strain (Fig. 3A). Introduction of the plasmid containing the wild-type ssaM allele to the mutant strain restored the amount of extracellular SseJ to that of the wild-type strain (Fig. 3A).
We next examined the extracellular levels of another SPI-2 effector, PipB, which represents one of several SPI-2 effectors which lack a conserved sequence found in the first 150 amino acids of some other SPI-2 effectors, including SseJ and SifA (Knodler et al., 2002). In this case, PipB-2HA was expressed in S. typhimurium strains from a plasmid under the control of the pipB promoter. As for SseJ-2HA, greater quantities of PipB-2HA were found on the bacterial cell surface and on the plastic surface of the tube containing the ssaM mutant, compared with the wild-type strain (Fig. 3B). Therefore, mutation of ssaM appears to result in oversecretion of two different SPI-2 effectors in vitro.
One explanation for the apparently increased secretion of these two effectors is that the ssaM mutant strain undergoes a greater degree of lysis than the wild-type strain, thereby releasing cytoplasmic proteins non-specifically. However, similar low levels of the cytoplasmic protein RecA were found in the extracellular fractions from ssaM mutant and wild-type bacteria (Fig. 4A), suggesting that the increased extracellular levels of SseJ and PipB associated with the ssaM mutant might be caused by oversecretion.
Secretion of proteins by the SPI-2 TTSS is controlled in part by the external pH of the growth medium. After growth in MgM-MES at pH 7.5, effector and translocator proteins are synthesized in the bacterial cell, but secretion is only activated when the external pH drops to ≈5.0 (Beuzón et al., 1999). When the ssaM mutant was grown in MgM-MES at pH 7.5 or pH 5.0, SseJ-2HA was present in the cell cytosol, but it was only detected in the extracellular fraction after growth at pH 5.0 (Fig. 4B). Similar results were obtained for PipB-2HA (data not shown). Furthermore, analysis of double mutant strains showed that secretion of SseJ-2HA was abolished when a null mutation in a gene (ssaC) encoding an essential component of the SPI-2 secretion apparatus (Freeman et al., 2002) was created in the ssaM mutant strain (Fig. 4C). However, secretion was unaffected after introduction of a mutation in a gene (prgH) encoding an essential component of the SPI-1 TTSS (Kubori et al., 1998) (Fig. 4C). We conclude that the increased quantities of SseJ-2HA and PipB-2HA found outside ssaM mutant cells arise from oversecretion of these proteins through the SPI-2 TTSS.
SsaM is required for secretion of translocon proteins in vitro
Although the ssaM mutant strain oversecreted SseJ-2HA in vitro, it failed to translocate this protein in infected cells (Fig. 2). These observations could be explained if the ssaM mutant is defective in the secretion of translocon proteins. Therefore, secretion of the translocon proteins SseB, SseC and SseD was examined by immunoblotting after growth of bacterial strains under SPI-2-inducing conditions. Similar amounts of SseB, SseC and SseD were found within wild-type and ssaM mutant bacteria, but there was no detectable secretion of these proteins by the ssaM mutant, unless it carried the complementing plasmid (Fig. 5).
A spiC mutant has a similar in vitro phenotype to the ssaM mutant
The ability of the ssaM mutant to secrete SPI-2 effector(s) but not translocon proteins is shared by a strain carrying a mutation in spiC. Under SPI-2-inducing conditions, a spiC mutant secreted SseJ-HA at a level similar to that of the wild-type strain, but failed to secrete SseB or SseC (Freeman et al., 2002). We have shown previously that a spiC mutant fails to secrete SseB, SseC and SseD (Yu et al., 2002). These similarities between spiC and ssaM mutant phenotypes suggested that SpiC and SsaM might be involved in the same function within the bacterial cell. We therefore examined the levels of SseJ-2HA and PipB-2HA secreted by the spiC mutant, under the same conditions that we used to examine secretion of these proteins by the ssaM mutant. The spiC mutant oversecreted both SseJ-2HA and PipB-2HA (Fig. 6A and B). Similar to the ssaM mutant, this oversecretion required an otherwise intact SPI-2 TTSS (Fig. 6B). Therefore, the phenotypes of spiC and ssaM mutant strains in terms of secretion of SPI-2 proteins are very similar if not identical. The levels of SseJ-2HA were examined in the spiC and ssaM mutants and wild-type bacteria after removal of surface proteins by hexadecane. All three strains contained similar levels of the effector (data not shown). Therefore, the total amount of SseJ-2HA produced by the mutants is significantly higher than that produced by the wild-type strain.
To determine whether the oversecretion of SseJ-2HA and PipB-2HA by ssaM and spiC mutant bacteria was caused by increased transcriptional activity of sseJ and pipB promoters, transcriptional fusions between these promoters and a gene encoding the green fluorescent protein (GFP) were constructed and expressed from plasmids in the wild-type strain, and ssaM, spiC and ssrA mutant strains. Expression of GFP from both promoters (measured by flow cytometry) was dependent on the ssrA gene, as reported previously (Miao and Miller, 2000; Knodler et al., 2002), but there was no detectable difference in expression between the wild-type and spiC or ssaM mutant strains (Fig. 6C).
Are SsaM and SpiC translocated into host cells?
In view of previous reports describing the translocation of SpiC and a SpiC fusion protein into infected macrophages (Uchiya et al., 1999; Lee et al., 2002; Shotland et al., 2003), experiments were carried out to determine whether SsaM is also a secreted and translocated protein. A plasmid encoding SsaM tagged with a double HA epitope at its C-terminus (SsaM-2HA) was constructed and transferred into the ssaM mutant strain. This plasmid restored the secretion of translocon proteins in vitro and the ability to form Sifs in HeLa cells, confirming that the double HA tag did not interfere with the function of the protein (data not shown). However, there was no evidence that SsaM-2HA is secreted, by analysis of proteins recovered from the bacterial cell surface and plastic surface of the tube in which the bacteria were grown, after growth of bacteria in SPI-2-inducing conditions (data not shown). To investigate whether SsaM is translocated into infected cells, J774A.1 or RAW264.7 macrophages were infected for 8 or 10 h, respectively, with the ssaM mutant carrying the SsaM-2HA-expressing plasmid. Infected cells were fixed, treated with Triton X-100 at 25°C to permeabilize both host cell and bacterial membranes, and labelled with antibodies against the HA tag and Salmonella, then examined by confocal immunoflorescence microscopy. In both experiments, SsaM-2HA colocalized only with labelled bacterial cells (data not shown). As an independent test for translocation of SsaM-2HA, macrophages were infected with the ssaM mutant simultaneously expressing SsaM-2HA from the plasmid, and SseJ-2HA from the chromosome (to serve as a positive control). Macrophages were fractionated by sequential exposure to saponin and Triton X-100 at 0°C (Kuhle and Hensel, 2002). Under these experimental conditions, saponin treatment permeabilizes the plasma membrane and solubilizes proteins in the macrophage cytosol, Triton X-100 solubilizes other host cell membranes and the insoluble pellet contains bacterial cells. Proteins in these fractions were analysed by immunoblotting. SsaM-2HA (distinguished from SseJ-2HA by its size) was only detected in the fraction containing bacteria (Fig. 7A). In contrast, SseJ-2HA expressed in the same strain was also detected in the host cell membrane fraction (Fig. 7A), consistent with previous findings (Kuhle and Hensel, 2002).
The similar phenotypes of the ssaM and spiC mutant strains, and the failure to obtain evidence that SsaM is a secreted or translocated protein, led us to re-examine the question of translocation of SpiC. For these experiments, we used a rabbit polyclonal anti-SpiC antibody, which is capable of detecting SpiC when expressed from a plasmid (Yu et al., 2002), and an anti-HA antibody, which detects SpiC-2HA when expressed from a plasmid (Fig. 7). Bacterial strains carried the spiC mutation and also expressed SseJ-2HA from the chromosome, to serve as a positive control. Both plasmid-expressed SpiC and SpiC-2HA were shown to be functional by the fact that they restored the secretion of translocon proteins and Sif formation to the spiC mutant strain (data not shown). A total of 11 separate experiments were performed in attempts to detect translocation of SpiC or SpiC-2HA by immunoblotting after infection of macrophages. These involved the use of both RAW264.7 cells and J774A.1 macrophages, analysing at different times after uptake of bacteria, and fractionating cells either with or without the use of detergents. These experiments are summarized in Table 1, and representative results are shown in Fig. 7B and C. In all of these experiments, SpiC and SpiC-2HA were only found in the fractions containing bacteria. In contrast, SseJ-2HA expressed from the same cells was readily detected in the fraction containing host cell membranes, consistent with a previous report (Kuhle and Hensel, 2002). To assess the sensitivity of detection of SpiC-2HA, a series of dilutions of the fraction containing bacteria shown in Fig. 7C was analysed by immunoblotting. The anti-HA antibody detected SpiC when the fraction was diluted 20-fold (Fig. 7D). Because equivalent proportions of each fraction were analysed in Fig. 7C, this result implies that if only 5% of total SpiC were translocated into the host cell, it would have been detected in these experiments.
Table 1. Summary of fractionation assays to detect translocation of SpiC and SpiC-2HA.
. The spiC mutant carrying plasmid pspiC or pspiC-2HA was used to infect either J774A.1 or RAW 264.7 macrophages. At the time points shown, cells were harvested and subjected to fractionation, SDS-PAGE and immunoblotting.
. Macrophages were disrupted physically and separated into host cytosolic, host membrane and bacteria-containing fractions by centrifugation.
. Macrophages were extracted sequentially with saponin and Triton X-100 to fractionate proteins into soluble macrophage cytosol, host membranes and the pellet-containing bacterial cells.
6 h, 7 h
7 h, 8 h
7 h, 8 h
We also attempted to detect SpiC-2HA by confocal immunofluorescence microscopy, after infection of cells with the spiC mutant expressing SpiC-2HA from the plasmid, and treatment of fixed, infected cells with either Triton X-100 (to permeabilize both host and bacterial cell membranes) or saponin (to permeabilize only host cell membranes; see Fig. 2), at 25°C. SpiC-2HA was detected after exposure of cells to Triton X-100, but not saponin (Fig. 7E). Furthermore, SpiC-2HA colocalized entirely with bacteria in cells permeabilized with Triton X-100 (Fig. 7E). Therefore, these experiments provide no evidence that SpiC or SpiC-2HA are translocated into host cells.
SsaM interacts with SpiC
In view of the similar phenotypes of spiC and ssaM mutant strains, the requirement for both proteins for secretion of translocon components, and the lack of evidence from our experiments that SpiC and SsaM are translocated, we hypothesized that they might function together within the bacterial cell, possibly to control the secretion of translocon proteins. Such a role has been proposed for InvE of the SPI-1 TTSS, which interacts with a complex involving SipB, SipC and their chaperone SicA (Kubori and Galán, 2002).
To investigate possible protein–protein interactions, we performed pull-down assays using a GST-SpiC fusion protein. We first confirmed that GST-SpiC was functional by showing that its expression from a plasmid in the spiC mutant strain could restore Sif formation and virulence in mice (data not shown). Lysates from wild-type S. typhimurium expressing GST-SpiC were incubated with glutathione-coated agarose beads, and after extensive washing, bound proteins were released, separated by SDS-PAGE and probed with antibodies against SseB, SseA (a chaperone for SseB and SseD; Ruiz-Albert et al., 2003; Zurawski and Stein, 2003), SseC and SseD. There was no evidence for any interaction between SpiC and these proteins (Fig. 8A, and data not shown).
Pull-down experiments were also performed to test for possible interactions between SpiC and SsaM. GST-SpiC and SsaM-2HA were expressed in S. typhimurium, and lysates were incubated with glutathione-coated beads. SsaM-2HA was pulled down by GST-SpiC but not by GST alone (Fig. 8A). However, in these experiments, a greater quantity of GST bound to beads than GST-SpiC (Fig. 8A). To rule out the possibility that the lower amount of bound GST-SpiC might allow SsaM-2HA to bind non-specifically to the beads, lysate from cells expressing SsaM-2HA was mixed with glutathione-coated beads in a mock pull-down assay. No detectable SsaM-2HA could be recovered from these beads (data not shown).
The last 18 amino acids of SsaM are predicted with low probability to form a coiled-coil structure (Lupas, 1997). To determine whether this is important for the interaction between SsaM and SpiC, this region was deleted, and the truncated protein was fused to a double HA tag. The resulting protein (SsaM104-2HA) failed to interact with GST-SpiC, when tested by pull-down assay (Fig. 8B).
The interaction between SsaM and SpiC was also tested by co-immune precipitation (Co-IP). Lysates from different strains expressing GST-SpiC, or GST-SpiC together with either SsaM-2HA or SsaM104−2HA were mixed with anti-HA antibody and immunoprecipitated. Protein(s) binding to the antibody were released from protein G-coated beads, separated by SDS-PAGE, transferred to PVDF membrane and probed with either an anti-HA antibody or an anti-GST antibody. The results showed that SsaM-2HA but not SsaM104-2HA interacted with GST-SpiC (Fig. 8C).
Finally, the functional importance of the last 18 amino acids of SsaM was investigated by expressing SsaM104-2HA in the ssaM mutant, and by testing the capacity of this strain to form Sifs in infected Hela cells. The truncated protein was unable to restore Sif formation (data not shown), showing that this region is essential for SsaM function.
In this study, we investigated the function of SsaM, a small protein encoded by one of the four major operons that constitute the TTSS locus within SPI-2. SsaM appears to be located exclusively in the bacterial cell, and is essential for in vitro secretion of the translocon components SseB, SseC and SseD. This explains the attenuated virulence of the ssaM mutant, its failure to form Sifs and to translocate the effector protein SseJ into infected cells. However, SsaM is not required for secretion of either SseJ or PipB in vitro, indicating that it is not an essential component of the secretion apparatus. Indeed, in the absence of SsaM, bacterial cells oversecreted both effector proteins in vitro. Analysis of a spiC mutant strain showed that it also displays these phenotypes, suggesting that SsaM and SpiC might be involved in the same physiological activity. This was supported by further experiments which showed that they interact within the bacterial cell, and that this interaction requires the C-terminal 18 amino acids of SsaM.
The fact that ssaM and spiC mutant strains have similar phenotypes in vitro raised the question of whether SsaM is translocated into the host cell or not. Using a variety of approaches, we found no evidence for translocation of SsaM. This led us to re-examine the question of translocation of SpiC. Two groups have reported that SpiC is a translocated effector of the SPI-2 TTSS. This is based on: (i) immunofluorescence microscopy of infected macrophages using an anti-SpiC antibody (Uchiya et al., 1999), (ii) Western blot analysis of infected macrophages after fractionation after physical disruption (Uchiya et al., 1999), (iii) Western blot analysis of infected macrophages after fractionation after solubilization with digitonin or octyl glucoside (Lee et al., 2002) and (iv) assaying adenylate cyclase activity in macrophages infected with strains expressing a SpiC-CyaA fusion protein (Shotland et al., 2003). On the other hand, our group and two others (Freeman et al., 2002; Hansen-Wester et al., 2002; Yu et al., 2002) have failed to detect secretion of unmodified or epitope-tagged SpiC in vitro. In the present work, we carried out several experiments, using unmodified SpiC and SpiC-HA, in an attempt to demonstrate translocation of these proteins into macrophages. Despite the fact that we could readily detect these proteins by immunoblotting and immunofluorescence, we obtained no evidence for their translocation at different times after infection of either J774A.1 or RAW264.7 macrophages. It is possible that translocation of SpiC was not detected because neither the anti-SpiC antibody nor the anti-HA antibody was sufficiently sensitive to detect translocated protein. However, this is unlikely given the ease of detection of both proteins in the fractions containing bacterial cells by immunoblotting. In the study by Uchiya et al. (1999), a substantial proportion of total SpiC was detected in the macrophage cytosol fraction by 6 h after infection. In the study by Lee et al. (2002), all of the detectable SpiC was found in the macrophage cytosol fraction, although the time point after infection when the assay was performed was not stated (Lee et al., 2002). Our immunoblot analysis of diluted samples of the bacteria-containing pellet from infected macrophages showed that, had 5% or less of total SpiC protein been translocated, it would have been detectable. The presence of the double HA tag (20 amino acids in total) did not appear to affect SpiC function because the plasmid expressing the fusion protein fully rescued Sif formation and secretion of translocon components in vitro. The functionality of the protein expressed from the complementing plasmid also argues against the possibility that a failure to detect translocation of SpiC might result from suboptimal levels of expressed protein (Shotland et al., 2003). Furthermore, the presence of a double HA tag at the C-termini of the SPI-2 effectors SopD2, PipB, PipB2 (Knodler et al., 2002; 2003; Brumell et al., 2003) and SseJ (this work) does not block their translocation into host cells.
We also failed to detect translocated SpiC-2HA in infected macrophages by fluorescence microscopy. Collectively, our experiments have failed to support published work indicating that SpiC is a translocated effector of the SPI-2 TTSS. However, this failure does not prove that the protein is not translocated, and further efforts will be required to resolve the controversy over the localization and function(s) of SpiC.
We found that mutation of spiC or ssaM led to oversecretion of PipB and SseJ fusion proteins, whereas Freeman et al. (2002) reported that a spiC mutant secreted SseJ at an level equivalent to that seen by wild-type bacteria. This apparent inconsistency might be explained by the fact that Freeman et al. obtained secreted proteins by precipitation of culture supernatants, whereas we recovered proteins from the plastic surface of the tube containing the bacteria. The oversecretion of PipB and SseJ did not appear to be a direct result of increased activity of pipB and sseJ promoters, as measured using transcriptional fusions between pipB and sseJ promoters to gfp. The oversecretion phenotype therefore probably results from increased translational activity of the respective mRNAs, increased mRNA stability or post-translational modification.
SpiC and SsaM appear to represent a class of type III secretion proteins that can distinguish effector proteins from translocon proteins, and which are necessary for the ordered secretion of these two classes of secreted proteins through the TTSS. SsaM has weak similarity to orf12 of the LEE of EPEC, and an orf12 mutant strain in the homologous LEE of Citrobacter rodentium was reported recently to be defective for secretion of both translocators and effectors. However, strains carrying mutations in two other genes of the C. rodentium LEE (sepL and rorf6) failed to secrete translocon components but displayed enhanced secretion of two effectors, Tir and NleA (Deng et al., 2004). A mutation in the invE gene of S. typhimurium, which encodes a component of the SPI-1 TTSS, drastically reduced the secretion of translocon components SipB, SipC and SipD, but did not prevent secretion of effector proteins into culture supernatants (Kubori and Galán, 2002). InvE was shown to interact with SipB and SipC, but only in the presence of their chaperone, SicA. It was proposed that InvE, which is localized within the bacterial cell, is involved in the recognition of translocon protein complexes and has a role in establishing the secretion hierarchy (Kubori and Galán, 2002). SsaM and SpiC might function in a similar way, and require the presence of additional protein(s) to interact. If so, this would explain our failure to detect any interaction between SsaM and SpiC by two-hybrid analysis (our unpubl. results).
For a TTSS to function optimally, one might expect that translocon proteins must be distinguished from and secreted before effectors, otherwise a proportion of secreted effectors would not be able to gain access to their targets within the host cell cytosol or membranes through the translocon pore. It is possible that complexes involving SpiC and SsaM of SPI-2, InvE, SipB, SipC, SicA of SPI-1, and SepL and rORF6 of the C. rodentium and EPEC LEE are representative of a general class of proteins which act as molecular ‘gatekeepers’, ensuring that translocon proteins are secreted before effectors. It will be interesting to examine the composition of these complexes in more detail and to determine their precise roles in TTSS function.
Bacterial strains and growth conditions
The S. typhimurium strains used in this study are listed in Table 2. Bacteria were grown in Luria–Bertani (LB) medium supplemented with carbenicillin (50 µg ml−1), kanamycin (50 µg ml−1), chloramphenicol (30 µg ml−1), nalidixic acid (100 µg ml−1) or tetracycline (25 µg ml−1), for strains resistant to these antibiotics (Ampr, Kmr, Cmr, Nalr and Tetr respectively). To induce SPI-2 gene expression and SPI-2-dependent secretion, bacteria were grown in MgM-MES, containing 170 mM 2-[N-morpholino] ethane-sulphonic acid at pH 5.0, 5 mM KCl, 7.5 mM (NH4)2SO4, 0.5 mM K2SO4, 1 mM KH2PO4, 8 µM MgCl2, 38 mM glycerol and 0.1% casamino acids (Beuzón et al., 1999) with the corresponding antibiotics when appropriate. Bacteria were grown at 37°C overnight with aeration.
Table 2. S. typhimurium strains used in this study.
Strain HH215, containing a non-polar deletion mutation in ssaM, was obtained by allelic exchange as follows: a 2089 bp DNA fragment containing the ssaM gene and the corresponding flanking regions was recovered after BglII and NsiI digestion of lambda clone λ7 (Shea et al., 1996), and subcloned in BamHI/PstI-digested pKS+. The resulting plasmid was used as template for a PCR with primers ssaMF and ssaMB. These and all other primers are listed in TableS1 in Supplementary material. ssaMF and ssaMB were used to delete codons encoding amino acids 10–103 of SsaM. The PCR product containing NsiI sites at each terminus was digested with NsiI and ligated, resulting in pKS-ΔssaM. Plasmid pUC18-K2 (Ménard et al., 1993) was digested by EcoRI and BamHI, and blunt-ended with the Klenow fragment of DNA polymerase. The 850 bp DNA fragment containing the Kmr gene aphA-3 was gel-purified, and ligated into pKS-ΔssaM which had been digested with NsiI and blunt-ended by T4 DNA polymerase. The 2655 bp DNA fragment of ΔssaM::aphA-3 was released from the above construct by digestion with EcoRI and XbaI, and inserted into the same sites of plasmid pGP704 (Miller and Mekalanos, 1988) to generate plasmid pGP704-ΔssaM::aphA-3. The resulting plasmid was transferred by conjugation from E. coli S17-1λpir to S. typhimurium 12023 Nalr, and exconjugants were selected as previously described (Shea et al., 1999).
To tag the chromosomal copy of sseJ with DNA encoding a double 2HA epitope, the λ Red recombination system was used as described previously (Uzzau et al., 2001). Briefly, one primer (sseJHAcat1) contained the 3′-terminal sequence (without the stop codon) of sseJ followed by a sequence encoding an HA epitope, and another primer (sseJHAcat3) corresponds to a chromosomal region downstream from sseJ. Primers were used to amplify the chloramphenicol acetyltransferase (cat) gene from pSU314. The PCR product was purified and electroporated into competent cells of S. typhimurium 12023 (pKD46). Transformants (HH216) were selected on LB-Cm plates. The expression and translocation of SseJ-2HA by HH216 were confirmed by immunoblot analysis and immunofluorescence microscopy. The modified sseJ was transduced into different strains of S. typhimurium by phage P22 as described previously (Davis et al., 1980). When necessary, the FRT-flanked antibiotic resistance cassette was removed after transformation with pCP20, as described (Datsenko and Wanner, 2000). Mutation of ssaC in HH218 (to create HH220) and HH221 (to create HH223) was achieved following the method described by Datsenko and Wanner (2000), using ssaCcat1 and ssaCcat2 as primers and pKD3 as template. HH219 and HH222 were constructed by transducing prgH020::TnlacZY from EE656 (Bajaj et al., 1996) to HH218 and HH221.
The complementing plasmid pssaM is a derivative of pACYC184 (Chang and Cohen, 1978) carrying the ssaM gene under the control of a constitutive promoter. A DNA fragment including the complete ORF of ssaM was amplified by PCR from 12023 genomic DNA using primers SMF and SMR. The PCR product, containing terminal EcoRV and SalI sites, was digested and ligated into the EcoRV and SalI sites of pACYC184 within the Tetr gene, generating pssaM. Plasmids pssaM-2HA and pssaM104-2HA are derivatives of pssaM. PCR products amplified from pSU315 (Uzzau et al., 2001) by using primer ACYCsite2 together with either primer ssaMHA or primer ssaM104HA were transferred into electrocompetent cells of E. coli DH5α carrying pssaM and pKD46, to tag ssaM with a sequence encoding double HA at the C-terminus, or to delete the last 18 amino acids of SsaM and tag the truncated protein with the double HA epitope. To generate pspiC-2HA, electrocompetent cells of E. coli DH5α carrying pspiC and pKD46 (Yu et al., 2002) were elecroporated with PCR product amplified from pSU315 using primers ACYCsite2 and spiCHA. To fuse SpiC to the C-terminus of GST, a DNA fragment encoding the second amino acid to the last amino acid of SpiC was amplified by PCR from genomic DNA of S. typhimurium strain 12023, using primers GSTSpiC and ssaByr. The PCR product, containing terminal EcoRI and XhoI sites, was digested and ligated into vector pGEX-KG (Pharmacia), generating pgstspiC. Plasmids psseJpro and ppipBpro are derivatives of pFPV25, a vector carrying a promoterless gfpmut3A (Valdivia and Falkow, 1997). The PCR products of the promoter region of sseJ (using primers sseJpt1 and sseJpt2) or pipB (using primers pipBpt1 and pipBpt2) amplified from genomic DNA of S. typhimurium strain 12023, were ligated into BamHI and EcoRI sites of pFPV25 to fuse the promoter regions to the gfp gene. All plasmids, including pspiC (Yu et al., 2002) and pACBC-2HA (Knodler et al., 2002), were transferred into Salmonella strains by electroporation. Plasmids constructed as part of this study were verified by DNA sequencing.
Female BALB/c mice (20–25 g) were used for Competitive Index studies (CIs). At least three mice were inoculated i.p. with a mixture of two strains comprising ≈5 × 104 colony forming units of each strain in physiological saline, and the CIs were determined from spleen homogenates 48 h post inoculation as described previously (Beuzón et al., 2001).
Antibodies and reagents
The mouse monoclonal anti-HA antibody HA.11 (MMS-101P, Covance) was used at a dilution of 1:1000 for immunofluorescence microscopy and immunoblot analysis. The mouse monoclonal anti-human LAMP1 antibody H4A3 developed by J.T. August and J.E.K. Hildreth, and the mouse monoclonal antibody against β-tubulin (E7) developed by M. Klymkowsky, 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 both used at a dilution of 1:1000. Goat anti-Salmonella polyclonal antibody CSA-1 was purchased from Kirkegaard and Perry Laboratories and was used at a dilution of 1:200. Rabbit polyclonal anti-SseB (Beuzón et al., 1999), anti-SseC, anti-SseD (Nikolaus et al., 2001), anti-SpiC (Yu et al., 2002) and anti-SseA (Ruiz-Albert et al., 2003) antibodies were used as described. Rabbit polyclonal anti-Sec61β antibody was provided by S. Méresse (Centre d’Immunologie de Marseille-Luminy, Marseille, France) and was used at a dilution of 1:10 000. Rabbit anti-RecA antibody (1:10 000, a gift from Kenji Adzuma, The Rockefeller University, NY, USA), rabbit anti-Salmonella i-H serum (1:500, Murex Biotech Limited) and rabbit anti-GST antibody (1:50 000, Covance) were also used as primary antibodies for immunoblot analysis. Texas Red sulphonyl chloride (TRSC)-conjugated donkey anti-mouse antibody and cyanine 2 (Cy2)-conjugated donkey anti-goat antibody were purchased from Jackson Immunoresearch Laboratories, and used at a dilution of 1:400. Horseradish peroxidase (HRP)-conjugated donkey anti-rabbit and sheep anti-mouse antibodies were purchased from Amersham Life Sciences and used at a dilution of 1:10 000.
RAW 264.7 murine macrophage-like cells were obtained from ECACC (ECACC 91062702). HeLa (clone HtTA1) human epithelioid and J774A.1 murine macrophages were kindly provided by S. Méresse (Centre d’Immunologie de Marseille-Luminy, Marseille, France) and D. Young (Imperial College, London, UK) respectively. Cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS) and 2 mM glutamine at 37°C in 5% CO2.
Bacterial infection of cultured cells, immunofluorescence 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 subsequently incubated in medium containing gentamicin to kill extracellular bacteria. For immunofluorescence, cells were fixed in paraformaldehyde, then either labelled in the presence of 0.1% saponin or permeabilized with 0.1% Triton X-100 at room temperature for 15 min, then washed and labelled in PBS. Cells were mounted in Mowiol and analysed using a confocal laser scanning microscope (LSM 510, Zeiss) as described previously (Beuzón et al., 2000). For enumeration of intracellular bacteria, macrophages were washed three times with PBS, lysed with 0.1% Triton X-100 for 10 min and a dilution series was plated onto LB agar.
Preparation of protein fractions from bacteria grown in vitro
Bacterial cell densities were determined by measurement of the OD600 after overnight growth. To ensure that protein from equal numbers of cells was analysed, in all experiments protein samples were adjusted to OD600 values such that a volume corresponding to 10 ml of a culture of OD600 0.6 was taken up in 100 µl of protein-denaturing buffer for gel electrophoresis. The secreted, cell wall-associated and total bacterial fractions were prepared as described before (Yu et al., 2002).
Subcellular fractionation of infected macrophages
J774A.1 or RAW264.7 cells were infected with bacteria for 6–10 h as described by Uchiya et al. (1999). Macrophages were harvested from six-well plates, and washed once with PBS before physical disruption or detergent solubilization, and fractionation. For the physical disruption method, cells were broken by passage through a 27 G needle several times, until more than 80% of the cells were lysed (Beuzón et al., 2002). Cytosolic and membrane fractions were separated from bacteria and large organelles as described (Uchiya et al., 1999). For the detergent solubilization method, cytosolic and membrane fractions were extracted by sequential treatment with 0.2% saponin and 0.1% Triton X-100 at 0°C, as described (Kuhle and Hensel, 2002). The detergent-resistant fraction contains intact bacteria and host cell membranes.
Pull-down and immunoprecipitation assays
Volumes corresponding to 20 ml of a bacterial culture with an OD600 of 1.0 after overnight growth of S. typhimurium in MgM-MES at pH 5.0 were used to perform pull-down and immunoprecipitation assays. Bacteria were collected by centrifugation. For the GST pull-down assays, the B-PERTM GST spin purification kit (Pierce) was used with a slight modification to the manufacturer's instructions. Bacteria were resuspended in 1 ml of B-PERTM Reagent and mixed by vortexing. The homogeneous mixture was shaken gently at room temperature for 10 min, before the soluble fraction was separated from the insoluble fraction by centrifugation at 18 000 g for 15 min. The supernatant was mixed with 50 µl of immobilized glutathione beads. After gentle shaking for 2 h at 4°C, the resin was pelleted by centrifugation at 500 g for 4 min, and washed four times with 700 µl of wash buffer. The resin-bound proteins were released by boiling the resin in 50 µl of SDS-PAGE sample buffer.
To immunoprecipitate HA-tagged proteins, bacteria were resuspended in a solution comprising 1 ml of 50 mM glucose, 10 mM EDTA, 4 mg ml−1 lysozyme, 25 mM Tris·Cl (pH 8.0) and incubated for 5 min at room temperature to generate spheroplasts. The spheroplasts were resuspended into 1.5 ml of lysis buffer (50 mM Tris·Cl pH 8.0, 100 mM NaCl, 1% Triton X-100, 0.05% SDS, 1 mM PMSF), and incubated on ice for 30 min with occasional mixing. The lysate was centrifuged for 10 min at 10 000 g, and the supernatant was transferred into a fresh tube and incubated with 50 µl of protein G-immobilized 4% beaded agarose (Sigma) for 1 h to pre-clear the lysate. The pre-cleared supernatant was incubated with 10 µl of mouse anti-HA antibody (MMS-101P, Covance) for 2 h at 4°C to form antibody–antigen complexes, then 50 µl of protein G-immobilized 4% beaded agarose was added to the reaction and incubated for 2 h at 4°C. The beads were collected by centrifugation at 500 g for 4 min, and washed four times with 700 µl of lysis buffer before boiling in 50 µl of SDS-PAGE sample buffer.
Polyacrylamide gel electrophoresis and immunoblot analysis of proteins
Protein fractions were dissolved in the appropriate volume of protein-denaturing buffer (Beuzón et al., 1999) and held at 100°C for 5 min. Proteins were immediately separated on 12% SDS-polyacrylamide gels (Laemmli, 1970). For immunoblot analysis, proteins were transferred to Immobilon-P (PVDF) membranes (Millipore) as described previously and examined using the ECL detection system under conditions recommended by the manufacturer (Amersham Life Science). Incubation of membranes with primary antibodies was followed by incubation with HRP-conjugated anti-rabbit or anti-mouse as secondary antibodies.
Flow cytometric analysis
Bacterial strains carrying the plasmids psseJpro or ppipBpro were grown overnight in MgM-MES at pH 5.0, and used for flow cytometric analysis. A total of 105 cells were analysed on a FACS Calibur cytometer (Becton Dickinson), and GFP was detected at 525 nm in the FL1 channel. Data were analysed with CellQuest software.
We are grateful to Kate Unsworth, Christoph Tang and Suzana Salcedo for critical review of the manuscript, to Leigh Knodler for providing pACBC-2HA and Junkal Garmendia for help with the flow cytometry. This work was supported by grants from the Medical Research Council (UK) and The Wellcome Trust to D.W.H.