The intracellular pathogen Legionella pneumophila can infect and replicate within macrophages of a human host. To establish infection, Legionella require the Dot/Icm secretion system to inject protein substrates directly into the host cell cytoplasm. The mechanism by which substrate proteins are engaged and translocated by the Dot/Icm system is not well understood. Here we show that two cytosolic components of the Dot/Icm secretion machinery, the proteins IcmS and IcmW, play an important role in substrate translocation. Biochemical analysis indicates that IcmS and IcmW form a stable protein complex. In Legionella, the IcmW protein is rapidly degraded in the absence of the IcmS protein. Substrate proteins translocated into mammalian host cells by the Dot/Icm system were identified using the IcmW protein as bait in a yeast two-hybrid screen. It was determined that the IcmS–IcmW complex interacts with these substrates and plays an important role in translocation of these proteins into mammalian cells. These data are consistent with the IcmS–IcmW complex being involved in the recognition and Dot/Icm-dependent translocation of substrate proteins during Legionella infection of host cells.
Legionella pneumophila is a facultative intracellular pathogen that can replicate inside eukaryotic cells (Horwitz and Silverstein, 1980). In the environment, Legionella replicate within protozoan hosts in freshwater reservoirs (Fields, 1996); however, when aerosol droplets are generated from contaminated sources, such as air-conditioning cooling tanks, Legionella can gain access to the human lung (Doebbeling and Wenzel, 1987). In the lung Legionella are ingested by alveolar macrophages, where they then survive and multiply intracellularly (Nash et al., 1984). Legionella infections can result in a severe pneumonia known as Legionnaires’ disease (Fraser et al., 1977; McDade et al., 1977).
Efforts have been made to understand how Dot/Icm substrates control the transport and remodelling of vacuoles containing Legionella. Previous work demonstrated that evasion of lysosome fusion by Legionella occurs independently of host vesicle recruitment and two distinct sets of genes were suggested to be involved in these two processes (Coers et al., 2000; Tilney et al., 2001). In these studies it was shown that evasion of phagosome–lysosome fusion requires the Dot/Icm cytosolic components IcmS and IcmW. Vacuoles containing icmS or icmW mutants of Legionella rapidly acquire late-endosomal and lysosomal markers; however, they still recruit host vesicles to the phagosomal membrane. The IcmS and IcmW proteins were shown to interact with each other in a yeast two-hybrid system, and mutational analysis revealed that icmS and icmW loss-of-function result in indistinguishable intracellular growth phenotypes (Zuckman et al., 1999; Coers et al., 2000). These icmS and icmW mutants have a severe growth defect in primary macrophages; however, in contrast to most other dot/icm mutants, they are capable of attenuated growth in differentiated U937 cells (Zuckman et al., 1999; Coers et al., 2000). Furthermore, the icmS and icmW genes are not required for the formation of pores in the host cell membrane (Zuckman et al., 1999), an activity that requires most of the other Dot/Icm system components (Kirby et al., 1998).
Because the IcmS and IcmW proteins are not required for all Dot/Icm-dependent activities, we had proposed that IcmS and IcmW are components of a protein complex that functions within the bacterial cytosol to facilitate the translocation of a subset of substrate proteins into host cells (Coers et al., 2000). In support of this model, we show here through biochemical analysis that the IcmS and IcmW proteins form a complex within the bacterial cell. Genetic analysis has identified Legionella substrate proteins injected into host cells by the Dot/Icm system based on their interactions with the IcmW protein. These substrate proteins were found to bind and require the IcmS–IcmW complex for efficient translocation into host cells.
IcmW protein levels are severely reduced in the absence of the IcmS protein
Previous data have shown that the IcmS and IcmW proteins are functional when tagged with the M45 epitope (Zuckman et al., 1999; Coers et al., 2000). We therefore used strains producing M45-tagged IcmS or IcmW in order to monitor the steady-state level of these proteins in Legionella. These data show that the IcmWM45 protein was difficult to detect in an icmS mutant of Legionella when a single-copy icmWM45 allele was expressed from the endogenous icmW promoter (Fig. 1A). Additionally, when the IcmWM45 protein was produced ectopically from a multicopy plasmid in an icmS mutant background the levels of the IcmWM45 protein were reduced compared with isogenic Legionella with the wild-type icmS allele (Fig. 1B). Wild-type levels of the IcmW protein were restored to the Legionella icmS mutant upon in trans expression of either wild-type icmS (data not shown) or a fusion protein containing glutathione S-transferase (GST) fused amino-terminally to the icmS coding region (Fig. 1A; GST–IcmS). The IcmW protein is the first product of the icmWX operon. It is unlikely that reduced IcmW protein levels result from a change in message stability or translational attenuation given that IcmX protein levels were unaffected in the Legionella icmS mutant (Fig. 1A). These data indicate that the IcmS protein has a role in controlling steady-state levels of IcmW protein in Legionella. In contrast, IcmSM45 protein levels did not change dramatically in an icmW mutant of Legionella (Fig. 1B), indicating that the IcmW protein is not necessary to maintain steady-state levels of IcmS protein.
The IcmS and IcmW proteins are in a complex with an estimated mass of 85 kDa
Experiments were conducted in order to test biochemically a IcmS–IcmW interaction that was observed previously in yeast two-hybrid studies (Coers et al., 2000). A GST–IcmS fusion protein was produced in a Legionella icmS mutant with a single copy of the icmWM45 gene integrated on the chromosome. Data obtained with this strain show that cellular levels of IcmW protein were restored upon production of GST–IcmS (Fig. 1A). Furthermore, this strain was able to grow in primary macrophages and amoeba (data not shown), indicating that both the GST–IcmS protein and the IcmWM45 protein are functional. To determine whether IcmS and IcmW form a stable protein complex, total protein lysates from this Legionella strain were passed over glutathione agarose and bound proteins were eluted. The IcmWM45 protein was found in the eluate upon release of GST–IcmS from the column (Fig. 1C). The Legionella IcmX and RalF proteins were not detected in the eluate (Fig. 1C and data not shown). The IcmWM45 protein was not found in the eluate when lysates from Legionella producing GST alone were affinity purified in parallel (Fig. 1C). These data indicate that IcmWM45 is binding specifically to a complex containing the IcmS protein.
To further characterize protein complexes that contain the IcmS and IcmW proteins, lysates from Legionella producing both IcmWM45 and IcmSM45 were fractionated by Superdex200 gel filtration. These data show that the IcmS and IcmW proteins co-fractionated with molecules having an apparent mass of ≈85 kDa (Fig. 1D). Because both proteins contain a single M45 epitope tag, the intensity of staining with mAb45 is indicative of the molecular ratio of each protein in this complex. Scanning densitometry studies indicate a 2:1 ratio of IcmS molecules to IcmW molecules in this complex. From these data we conclude that the IcmS and IcmW proteins are contained in a stable complex with a mass of ≈85 kDa.
The IcmS protein prevents rapid degradation of IcmW
Because mutations that eliminate a single component of a protein complex can often affect the stability of other members of the complex, it seemed possible that the stability of the IcmW protein was compromised in the absence of the IcmS protein, which would account for reduced amounts of the IcmW protein in an icmS mutant. To measure protein stability, pulse-chase experiments were conducted to determine the half-life of the IcmW protein in Legionella both in the presence and in the absence of the IcmS protein. These data show that the IcmW protein had a half-life of >40 min when the IcmS protein was present (Fig. 1E; ΔicmW). In contrast, the half-life of the IcmW protein was <5 min in Legionella that lacked the IcmS protein (Fig. 1E; ΔicmS ΔicmW). Thus, the IcmS protein prevents the rapid degradation of IcmW, suggesting that interaction with the IcmS protein greatly enhances the stability of the IcmW protein. Although IcmS is required for IcmW protein stability, IcmS protein levels and the half-life of the IcmS protein were reduced only slightly in a Legionella icmW mutant (Fig. 1E, lower panel), consistent with data indicating IcmS protein levels were not severely altered in an icmW mutant. These data indicate that the IcmW protein is unstable when it is not in a complex containing the IcmS protein.
Identification of IcmW-interacting proteins by yeast two-hybrid analysis
Previous studies support a model whereby the IcmS and IcmW proteins interact directly with a subset of Legionella proteins that are substrates translocated into host cells by the Dot/Icm system (Coers et al., 2000). For this reason, a yeast two-hybrid screen was conducted to identify additional Legionella proteins with an affinity for the IcmW protein. Legionella DNA fragments partially digested with the restriction enzyme Tsp509I were ligated into a series of three vectors derived from the plasmid pJG4-5. These vectors varied in the reading frame in which the EcoRI cloning site had been engineered in relation to the reading frame of the upstream B42-activation domain ensuring an in-frame fusion protein could be made to any Tsp509I site located within a Legionella open reading frame (ORF). The resulting Legionella two-hybrid libraries were screened using IcmW fused to LexA as bait. Fusion junctions from plasmids encoding potential IcmW-interacting proteins (Wips) were sequenced and attention was focused on the Legionella ORFs that were identified multiple times in this yeast two-hybrid screen (Table 1). Of these, five ORFs encode homologues to bacterial proteins that have general house-keeping functions, and were viewed as being unlikely substrates of the Dot/Icm system. Of the three remaining ORFs, two corresponded to genes encoding the SidG and SidH proteins. Given that SidG and SidH had been identified recently as Dot/Icm substrates in an interbacterial protein transfer screen (Luo and Isberg, 2004), their isolation here suggested that our two-hybrid screen was successful in identifying Dot/Icm substrates. The remaining protein, which we termed WipA, was isolated a total of 10 times in two independent libraries, suggesting that this protein is a bone fide IcmW-interacting protein and an attractive candidate for a new substrate protein injected into host cells by the Dot/Icm system.
Table 1. Results from yeast two-hybrid screens for IcmW-interacting proteins.
Legionella protein identified
Position of B42 fusion
Listed are Legionella proteins that were identified multiple times in a yeast two-hybrid screen for IcmW-interacting proteins. Given are the names of the Legionella protein or the closest homologue and the clone numbers of the library plasmid encoding an in-frame fusion between the Legionella protein indicated and the B42 activation domain. Each clone was from one of three libraries that were constructed in vectors derived from pJG4-5 that differed in respect to the location of the EcoRI cloning site in relation to the B42 activation domain (+1, +2, +3). The amino acid position (AA) within the Legionella protein at which the B42 activation domain was fused is indicated as are the blastE-values for the nearest homologue of the Legionella protein.
RNA pol β-subunit
RpoN – sigma 54
ATP synthase γ-chain (Pseudomonas)
Mannose 6-phosphate isomerase (Pseudomonas)
The WipA protein is a Dot/Icm substrate translocated into mammalian host cells
To test whether WipA serves as a substrate for the Dot/Icm transporter, we asked whether this protein can be translocated into the cytoplasm of mammalian host cells during Legionella infection. We used a gene fusion approach, utilizing the calmodulin-dependent adenylate cyclase protein (Cya) of Bordetella pertussis as a reporter. This enzyme has very low activity inside the bacterial cell; however, in the cytoplasm of a eukaryotic host cell it becomes activated by calmodulin, and therefore translocation of a fusion protein can be detected by monitoring the accumulation of cyclic AMP (cAMP) in these cells (Sory and Cornelis, 1994). It has been shown previously that substrate translocation by the Dot/Icm system can be measured quantitatively using Cya fusions to potential substrate proteins (Chen et al., 2004). For our studies, Chinese hamster ovary (CHO) cells producing the FcγRII protein were infected with opsonized Legionella producing a Cya–WipA fusion protein. After 1 h of infection these cells showed a 2.5 log increase in the levels of cAMP compared with uninfected control cells (Fig. 2). This increase in the production of cAMP required the WipA protein, as host cells infected with Legionella expressing the Cya protein alone had cAMP levels that were similar to uninfected cells. Translocation of WipA was found to be dependent on the Dot/Icm system, as infection with a dotA mutant producing the Cya–WipA fusion protein did not result in increased levels of cAMP (Fig. 2). Additional dot and icm mutants were tested for their ability to translocate the Cya–WipA protein into host cells in order to determine the requirements for other components of the Dot/Icm system. These data show that the icmQ, icmR, dotI and icmX mutant strains were unable to translocate the Cya–WipA fusion protein into host cells (Fig. 3), indicating that these components of the Dot/Icm machinery are essential for successful delivery of WipA into host cells. Thus, the Legionella Dot/Icm system can recognize and translocate the WipA protein into mammalian host cells during infection.
WipB is a paralogue of WipA that is translocated into host cells
Database searches failed to detect any other protein with significant amino acid similarity to WipA; however, a search of the Legionella genome database revealed two WipA paralogues. The Legionella paralogues were designated WipB and WipC. At the amino acid level WipA shares 43%/40% similarity, and 34%/26% identity with WipB and WipC respectively. Using the Cya fusion-based translocation assay, it was determined that WipB is also a substrate of the Dot/Icm system, and is translocated into host cells by a Dot/Icm-dependent mechanism to levels that are similar to WipA (Fig. 2). No translocation was detected for WipC, as the cAMP levels measured after infection with Cya–WipC expressing Legionella were similar to levels in control cells. These data indicate that WipA and WipB are substrate proteins that can be translocated into host cells by the Dot/Icm system and suggest that WipC may not be a translocated protein.
The SidG and SidH proteins are translocated into mammalian cells
The SidG and SidH proteins were recently shown to be Dot/Icm substrates that can be transferred between bacteria (Luo and Isberg, 2004). To determine whether these proteins can also be translocated into mammalian cells, Cya–SidG and Cya–SidH fusions were constructed. We found that both SidG and SidH fusions were translocated into host cells during infection. Data showing that cAMP levels were comparable to those seen for the WipA fusion protein (Fig. 2B) indicate that the efficiency of SidG and SidH translocation was comparable to WipA. Although previous data have suggested that IcmS and IcmW remain inside the bacterial cell and are not translocated substrates (Zuckman et al., 1999; Coers et al., 2000), to further test this possibility Cya was fused to the IcmS and IcmW proteins. These data show that Cya–IcmS and Cya–IcmW fusion proteins are not translocated into host cells (Fig. 2B) and provide further evidence in support of IcmS and IcmW functioning from within the bacterial cell.
WipA and WipB are not required for Legionella intracellular multiplication
A wipA wipB double mutant was constructed to determine whether the WipA and WipB proteins were essential for intracellular growth of Legionella. The wipA wipB double mutant strain was used to infect the protozoan host Acanthamoeba castellanii and intracellular growth was compared with that of isogenic wild-type and ΔdotA Legionella strains. The results show no growth defect for the double mutant in these cells (Fig. 4). The number of colony-forming units (cfu) recovered from cells infected with the wipA wipB double mutant over the course of 72 h was similar to the number recovered from cells infected with the wild-type control strain. The ΔdotA mutant served as a negative control and was unable to replicate in these cells. Similar results were obtained when murine bone marrow-derived macrophages from an A/J mouse were used as a host (data not shown). We therefore conclude that the WipA and WipB proteins are not essential for intracellular multiplication of Legionella in these host cells.
The IcmS–IcmW protein complex is important for translocation of Dot/Icm substrates
These results support a model where the IcmS and IcmW proteins form a complex within the bacterial cytoplasm that has the capacity to interact with a subset of substrate proteins. This model predicts that the IcmS and IcmW proteins should play a role in translocation of substrate proteins through the Dot/Icm apparatus. To address this question, translocation of Dot/Icm substrates was analysed in Legionella icmS and icmW mutants. These results show that WipA translocation is severely impaired in both icmS and icmW single mutants, with a residual activity of 10% compared with the translocation efficiency in a wild-type Legionella background (Fig. 5A). The defect in translocation efficiency was more pronounced for WipA in the icmS icmW double mutant. WipB translocation was severely affected by the absence of icmS and icmW singularly, and translocation of WipB was almost completely inhibited in the icmS icmW double mutant. SidG and SidH translocation was also inhibited in these icmS and icmW mutants to similar degrees. To determine whether this defect in translocation efficiency was observed for a Dot/Icm substrate protein that does not interact with either IcmS or IcmW in the yeast two hybrid system (data not shown), the translocation of RalF was examined in the icmS and icmW mutant backgrounds. RalF is an established Dot/Icm substrate (Nagai et al., 2002). These data show that translocation of the RalF protein was not strongly dependent on IcmS and IcmW function. The Cya–RalF fusion protein retained over 50% of its translocation activity, even in mutant Legionella lacking both icmS and icmW (Fig. 5A). Immunoblot analysis showed that protein levels of the substrates tested were not affected by the absence of IcmS or IcmW (data not shown), indicating that the decrease in cAMP production reflects a defect in translocation and is not a consequence of substrate stability within Legionella. To determine whether substrate translocation data correlate with binding of the IcmS–IcmW complex to Dot/Icm substrate proteins, co-purification studies were conducted. It was found that both SidG and WipA were associated with an affinity-purified IcmS–IcmW complex when the proteins were produced together in Escherichia coli (Fig. 5B). The RalF protein was not detected in the eluate containing the affinity-purified IcmS–IcmW complex when produced together in E. coli. These findings support a model where the IcmS–IcmW complex binds to a subset of substrate proteins and facilitates their translocation into host cells by the Dot/Icm apparatus.
The mechanism by which the Legionella Dot/Icm secretion system functions to deliver substrate molecules into host cells is largely unknown. More specifically, it is unclear what components of the Dot/Icm machinery are responsible for interacting with substrates to enable their proper delivery through the apparatus. Previously, a model was proposed where the IcmS and IcmW proteins form a complex capable of binding to substrate molecules, and that this complex then aids in the delivery of substrates into host cells by the Dot/Icm apparatus (Zuckman et al., 1999; Coers et al., 2000). Here we provide direct evidence in support of this model by establishing that IcmS and IcmW form a stable complex, finding genetic interactions between IcmW and Dot/Icm substrate proteins, showing that IcmS and IcmW are required for efficient translocation of substrate proteins into host cells, and showing that the IcmS–IcmW complex can interact biochemically with these substrate proteins.
Our results show that the IcmS and IcmW proteins, which in Legionella are encoded by genes that are located on different regions of the chromosome, interact with each other to form a complex with an approximate mass of 85 kDa. The stoichiometry of IcmS to IcmW within this complex was calculated to be 2:1. We predict that this stable protein complex consists of four molecules of the IcmS protein and two molecules of the IcmW protein, which would yield a complex with a predicted mass of 85.4 kDa. An alternative possibility is that this complex consists of two IcmS molecules and only one IcmW molecule and the remaining mass is contributed by at least one additional protein with a mass of approximately 42 kDa. We were unable to find evidence in support of a third protein being present in this stable complex by SDS-PAGE analysis of Legionella proteins bound to GST–IcmS (data not shown). Additionally, we found that the IcmS–IcmW complex can interact with both SidG and WipA, proteins that are each larger than 42 kDa. These data favour a model where the 85 kDa species is comprised of an IcmS–IcmW complex that does not have substrates bound.
The IcmW protein was used as bait to search for interacting partners in the yeast two-hybrid system to further explore the possibility that the IcmS–IcmW complex has a role in binding to substrates translocated by the Dot/Icm system. Two of the proteins identified several times in this screen were SidG and SidH. These proteins were reported recently in a paper describing a genetic screen aimed at identifying Dot/Icm substrates based on their ability to interact with the Legionella protein DotF and interbacterial protein transfer (Luo and Isberg, 2004). Another protein identified multiple times in our screen was a new protein we called WipA for IcmWinteracting protein A. Clones isolated multiple times in our screen are listed in Table 1. It should be noted that sidH and wipA were the only genes identified that were from independent clones containing different DNA inserts. All of the other proteins in Table 1 represent clones that were derived from the same library and contained the same DNA insert, meaning that they were likely to be siblings. Thus, this yeast two-hybrid screen was 100% successful at identifying substrate proteins injected into host cells by the Dot/Icm system when the filter defining IcmW-interacting partners was made slightly more stringent by considering only genes isolated multiple times from independently derived DNA inserts.
Recent studies have shown that Cya protein fusion technology can be used to investigate translocation of potential substrate proteins into host cells by the Dot/Icm system (Chen et al., 2004). By adapting this approach, we were able to demonstrate that WipA, SidG and SidH are all translocated into mammalian host cells during Legionella infection by a process requiring the Dot/Icm system. These data suggest that the IcmW–IcmS complex can interact directly with Dot/Icm substrate proteins given that these translocated proteins were identified based on their ability to interact with IcmW in the yeast two-hybrid system. Legionella icmS and icmW mutants were used to test whether substrate interactions with the IcmS–IcmW complex facilitates translocation into host cells. These data show that translocation of WipA, WipB, SidG and SidH into host cells is highly dependent on the IcmS and IcmW proteins. These data provide functional evidence in support of the model whereby the IcmS–IcmW complex can engage substrate proteins and assist in their translocation by the Dot/Icm system.
The mechanism by which the IcmS–IcmW complex facilitates the translocation of substrates through the Dot/Icm system is unknown; however, it was determined that IcmS and IcmW are not themselves translocated into host cells during infection. Thus, the translocation process is mediated by the IcmS and IcmW proteins functioning from within the bacterial cell. Previous studies showed that icmS and icmW mutants are transported rapidly to lysosomes even though they possess a Dot/Icm transport apparatus that remains fully capable of inserting pores in the host cell plasma membrane (Zuckman et al., 1999; Coers et al., 2000). This distinct phenotype suggested that the IcmS and IcmW proteins were important for the translocation of a subset of substrate proteins by the Dot/Icm system, suggesting that some substrate proteins may be translocated into host cells in the absence of IcmS and IcmW. In support of this hypothesis, it was determined that translocation of the RalF protein was not highly dependent on the IcmS–IcmW complex. Thus, the requirements for IcmS and IcmW with regards to RalF translocation were different than those for the WipA, WipB, SidG and SidH proteins.
Our data indicate that several Dot/Icm substrate proteins require interaction with the IcmS–IcmW complex for efficient translocation. Interestingly, in the absence of the IcmS–IcmW complex, translocation of many substrate proteins could still be detected at significantly reduced levels and RalF translocation levels remained high. These data suggest that there are additional determinants that facilitate engagement of substrates for translocation by the Dot/Icm system. Given recent data suggesting that Dot/Icm substrate proteins can interact with DotF (Luo and Isberg, 2004), it is possible that substrate recognition and translocation involves cooperative interactions between multiple components of the Dot/Icm secretion machinery. Elimination of a single component of the secretion machinery that mediates substrate recognition could affect the efficiency of protein translocation, but may not abolish translocation completely as long as that substrate protein retains the ability to interact with other components. This might explain why translocation of RalF, a protein that interacts with DotF by two-hybrid analysis (Luo and Isberg, 2004) but does not appear to interact with the IcmS–IcmW complex, is not severely affected in Legionella mutants lacking icmS or icmW. Furthermore, the absence of key interacting partners could explain the low levels of translocation detected for the SidG and SidH proteins in icmS and icmW mutants while the ability of SidG and SidH to still interact with DotF could account for the residual levels of translocation observed.
WipA represents a new substrate protein injected into host cells by the Dot/Icm system. Like many of the recently identified Sid proteins (Luo and Isberg, 2004), there were additional Legionella genes identified that encode WipA paralogues. Using the Cya system it was determined that one of the two WipA paralogues, a protein called WipB, was translocated into host cells by the Dot/Icm system. These data indicate that potentially not all paralogues of bone fide Dot/Icm substrates are recognized and translocated by Legionella.
We found that a Legionella wipA wipB double mutant did not display a measurable intracellular multiplication defect in host cells. The fact that icmS and icmW mutants have a more severe intracellular growth defect than the wipA wipB double mutant is consistent with the IcmS–IcmW complex being important for the translocation of multiple substrates belonging to different families of Legionella proteins. Future studies will focus on biochemical interactions between Legionella Dot/Icm substrate proteins and the IcmS–IcmW complex and the role of the Wip proteins in modulation of host cell function during Legionella infection.
Strains and media
All bacterial strains, yeast strains, plasmids and oligonucleotide primers used in this study have been listed in Table 2. Unless otherwise noted, chemicals were purchased from Sigma. Bacto-agar, tryptone, yeast extract and yeast nitrogen base were purchased from Difco. Legionella strains were grown in AYE broth [1% yeast extract, 1%N-(2-acetamido)-2-aminoethanesulphonic acid (ACES; pH 6.9), 3.3 mM l-cysteine, 0.33 mM Fe(NO3)3] or on charcoal yeast extract (CYE) plates (AYE containing 1.5% bacto-agar, 0.2% activated charcoal) (Feeley et al., 1979). For Legionella, antibiotics were added to the media at the following concentrations: streptomycin 100 µg ml−1, kanamycin 20 µg ml−1, chloramphenicol 10 µg ml−1. For the induction of GST fusion proteins, isopropyl β- d-thiogalactopyranoside (IPTG) was added to 1 mM final concentration. E. coli strains were cultured in LB medium (LB) 1% bacto-tryptone, 0.5% yeast extract, 1% NaCl or on l-agar plates (LB containing 1.5% bacto-agar). Antibiotics were added to the media at the following concentrations: ampicillin 100 µg ml−1, kanamycin 40 µg ml−1, chloramphenicol 25 µg ml−1. Saccharomyces cerevisiae strains were grown at 30°C in YNB media [0.17% yeast nitrogen base, 2% glucose, 0.06% his-ura-trp-leu dropout mix (Bio 101)] or on YNB plates (YNB supplemented with 2% agar). For experiments requiring induction of the GAL promoter, 2% galactose and 1% raffinose were added to the YNB media in place of glucose. When required, the following supplements were added to the YNB media: tryptophan at 0.04 mg ml−1, uracil at 0.02 mg ml−1, leucine at 0.06 mg ml−1 and histidine at 0.02 mg ml−1.
Primary cells and cell lines were cultured at 37°C in 5% CO2. CHO cells were grown in minimal essential medium alpha medium (Gibco) containing 10% heat-inactivated fetal bovine serum (FBS). Bone marrow-derived macrophages were cultured from female A/J mice as described previously (Celada et al., 1984). A. castellanii (ATCC 30234) were cultured routinely at room temperature in ATCC medium 712 (PYG). One hour before and after infection, A. castellanii cultures were maintained at 37°C in a 5% CO2 atmosphere in PYG medium without glucose, proteose peptone and yeast extract (amoeba buffer).
Plasmid and strain construction
To express GST fusion proteins in Legionella, plasmid pGEX-KG with GST fused at the carboxyl-terminal to IcmS was used as template for polymerase chain reaction (PCR) amplification. Primers GEXKG-5′ and GEXKG-3′ were used to amplify the GST–IcmS fusion protein. PCR products were cut with KpnI and HindIII and ligated into the cloning vector pMMB207 cut with the same enzymes. The resulting plasmids, pDMC9, and pDMC11 contain GST–IcmS, and GST alone, respectively, expressed on pMMB207.
To express icmWM45 from the Legionella chromosome, plasmid pDMC32 was constructed by digesting plasmid pCR36 with EcoRI and XbaI, and ligating the fragment containing the tagged gene into pSR47 digested with the same enzymes. The plasmid created contains the icmWM45 allele and the entire icmW promoter region, and was introduced into the Legionella chromosome by recombination as described (Merriam et al., 1997). pDMC32 was then introduced into a ΔicmW strain (CR157) to create CR1111, and into a ΔicmWΔicmS strain (CR503) to create CR875. Strain CR875 was transformed with pDMC9 to create CR876, or with pDMC11 to create CR877. Strain CR1111 was transformed with pDMC9 to create CR1190, or with pDMC11 to create CR1191. Strain CR875 was transformed with pCR43 to create strain CR1188.
The plasmids used for the pulse-chase experiments were constructed by using pMMB207 encoding icmWM45 and icmSM45 as template for PCR. Primers were designed to PCR amplify both of these alleles minus their promoters. To amplify icmWM45 and icmSM45, 5′ primers IcmW-HL1 and IcmS-HL1 were used together with a 3′ primer PMMB207-3′ that overlaps a region downstream of both genes in pMMB207. PCR products were digested with EcoRI and HindIII and ligated into pMMB207 digested with the same enzymes. The resulting plasmids, expressing icmWM45 (pDMC34) and icmSM45 (pDMC36) under strict control of the Ptac promoter, were introduced into Legionella. The plasmid pDMC34 was introduced into an ΔicmW mutant (CR157) and a ΔicmWΔicmS mutant (CR503) creating strains CR1150 and CR1152 respectively. The plasmid pDMC36 was introduced into a ΔicmS mutants (CR393) and a ΔicmWΔicmS mutant (CR503) creating strains CR1154 and CR1156 respectively. The LegionellaΔicmWΔicmS mutant strain CR503 was transformed with pCR36 to create strain CR1121 or with pCR43 to create strain CR1123.
The following sets of primers were used for constructing Cya fusion proteins: SN1 and SN2 for WipA, SN3 and SN4 for WipB, SN5 and SN6 for WipC, SidGcyafwd and SidGcyarev for SidG, SidHcyafwd and SidHcyarev for SidH, IcmWcyafwd and IcmWcyarev for IcmW, IcmScyafwd and IcmScyarev for IcmS. The resulting PCR products were digested with the appropriate enzymes (see Table 2), and ligated to vector pCya digested with the same enzymes. In the case of Cya–SidG, the vector pCya was digested with BamHI/PstI and ligated to the PCR product digested BglII/PstI. The resulting fusion proteins consist of an amino-terminal M45 epitope tag, followed by amino acid residues 2–399 of the B. pertussis CyaA enzyme followed by the indicated Legionella protein. Expression of the Cya fusion proteins is driven by the icmR promoter located upstream. The Cya fusion plasmids were designated pSN2 for the plasmid expressing the Cya–WipA fusion, pSN5 for Cya–WipB, pSN8 for Cya–WipC, pEC35 for Cya–SidG, pEC11 for Cya–SidH, pEC1 for Cya–IcmW and pEC3 for Cya–IcmS.
The wipA gene was inactivated by co-ingegration of plasmid pDMC28 encoding a 500 base pair internal wipA gene fragment. This wipA fragment was amplified using primers PI45K-fwd and PI45K-rev. The fragment was digested with XbaI and SacI and ligated into pSR47 cut with the same enzymes. The wipB deletion strain SN50 was derived from the wild-type L. pneumophila strain CR39 (Lp01) by allelic exchange, as described previously (Merriam et al., 1997). A deletion allele of wipB was constructed using PCR to generate DNA fragments encoding regions of flanking homology that were immediately 5′ to the start codon, and 3′ to the termination codon of the gene. The primers used were SN7 and SN8, and SN9 and SN10 and the two PCR products were joined by recombinant PCR using primers SN7 and SN10. The final product was digested with XbaI and SacI then ligated into the gene replacement vector pSR47S digested similarly, creating plasmid pSN12. The wipA wipB double mutant strain was constructed by integrating plasmid pDMC28 into the genome of strain SN50, to create strain SN53.
Plasmid pEC66 was generated through the ligation of a pET15b (Novagen) vector fragment digested with NdeI and BamHI to a icmW ORF fragment containing 5′NdeI and 3′BamHI sites synthesized with the primers IcmW-5′Nde and IcmW-3′Bam. This vector was digested with BamHI and ligated to an icmS ORF fragment that harboured the −18 to −1 region of the pET15b vector relative to the affinity tag initiator codon. 5′BglII and 3′BamHI sites were generated with the primers IcmSRBSBgl and IcmS-3′Bam.
The pMMB207NT vector was digested with PstI and BamHI and ligated to a wipA ORF fragment generated using primers SN1 and SN2 creating plasmid pSN3, or to a sidG ORF fragment generated with the primers SidGcyafwd and SidGcyarev, creating pEC47.
Affinity purification of IcmWM45 with GST–IcmS in Legionella
Legionella strains having the icmWM45 allele integrated on the chromosome and with plasmids encoding either GST alone or GST–IcmS were grown in 100 ml of cultures of AYE plus chloramphenicol started at an OD600 of 0.1 from plate-grown bacteria that had been incubated for 48 h. Cultures were grown for 12 h. IPTG was added at a final concentration of 1 mM and cells were incubated for an additional 6 h. Cells were pelleted, washed in Tris-buffered saline (TBS) (50 mM Tris pH 8.0, 100 mM NaCl, 1 mM EDTA) and resuspended in 10 ml of lysis buffer (TBS, 0.02 mg ml−1 lysozyme, 0.01 mg ml−1 RNase, 0.002 mg ml−1 DNase, 0.1 mg ml−1 PMSF, 1% NP-40). Cells were disrupted by two passages through a French press at 16 000 psi. Lysates were centrifuged at 16 000 r.p.m. for 20 min and then 20 000 for 30 min in a Beckman JA-20 rotor. Glutathione sepharose beads (250 µl) were washed in TBS and then resuspended in TBS at the same starting volume. Lysates were bound to 250 µl of glutathione sepharose beads in batch for 2–3 h rocking at 4°C. Beads were pelleted and then washed in TBS. Bound proteins were eluted in glutathione elution buffer [50 mM Tris pH 8.0, 10 mM glutathione] in three rounds of 125 µl each. Eluted proteins were concentrated with Centricon centrifugal filter devices (Millipore) that exclude proteins with a molecular weight of less than 10 kDa. Whole-cell lysates and eluted proteins were separated by SDS-PAGE and then probed by immunoblot analysis using monoclonal antibodies against M45 or IcmX and rabbit antibodies specific for RalF.
Gel filtration chromatography
Cultures of Legionella were grown overnight in AYE plus chloramphenicol. Cells were lysed by French press and then cleared by centrifugation, as described above. Dithiothreitol (DTT) (1 mM) was added to the cleared lysate before centrifugation at 100 000 r.p.m. for 30 min in a Sorvall ST120AT2-0220 rotor. Five hundred microlitres were loaded onto a Superdex-200 FPLC column equilibrated with 10 mM Tris pH 7.5, 150 mM NaCl, 1 mM DTT. Fractions (500 µl) were collected. Proteins samples from each fraction were separated by SDS-PAGE and visualized by immunoblot analysis using antibodies against the M45 epitope tag.
Legionella strains with plasmids encoding either icmWM45 or icmSM45 under control of the Ptac promoter were collected from CYE plates containing chloramphenicol after 48 h incubation. Cultures (25 ml) were started at an OD of 0.2 in AYE containing chloramphenicol and grown for 6 h at 37°C. A 1 ml aliquot was taken before IPTG was added. IPTG was added to cultures at 1.0 mM final concentration. Cultures were allowed to induce for 30 min and then were set on ice. To remove the IPTG and begin the chase, cultures were centrifuged at 4°C, washed once with 25 ml of ice-cold AYE plus chloramphenicol, and centrifuged again. Pellets were resuspended in 1 ml of ice-cold AYE and then 23 ml of warm AYE plus chloramphenicol was added. One millilitre was immediately taken for time point T0, and at appropriate time points thereafter. Sampled cells were centrifuged and resuspended in 200 µl of ice-cold phosphate-buffered saline (PBS) and lysed by the addition of 200 µl of 2× Laemmli sample buffer. IcmS and IcmW protein levels were visualized by immunoblot analysis using antibodies against the M45 tag.
Samples in 1× Laemmli Sample Buffer [4% SDS, 20% glycerol, 120 mM Tris pH 6.8, 0.01% bromophenol blue, 720 mM β-mercaptoethanol] were boiled for 5 min and then centrifuged at 4°C for 2–20 min at maximum speed in a microcentrifuge. Proteins were separated by SDS-PAGE then transferred to Immobilon-P membranes (Millipore) in transfer buffer (50 mM Tris, 380 mM glycine, 0.1% SDS, 20% methanol, pH 8.3). The membranes were blocked in Blotto (PBS, 0.1% Tween-20, 5% non-fat dry milk) for 1 h at room temperature. To identify M45 epitope-tagged proteins, the membranes were incubated overnight at 4°C with the monoclonal antibody mAb45 (Obert et al., 1994) diluted 1:20 in Blotto. To identify GST-fused proteins, the membranes were incubated overnight at 4°C with the mouse monoclonal antibody against GST diluted 1:500 in Blotto. To identify IcmX, monoclonal antibodies were used at a 1:10 dilution in Blotto or polyclonal antibodies at 1:10 000 (Matthews and Roy, 2000). After incubation with primary antibody, blots were washed in PBS plus 0.1% Tween-20 and incubated for 1–2 h at room temperature with horseradish peroxidase (HRP)-conjugated goat anti-mouse or anti-rabbit (Zymed) diluted 1:3000 in Blotto. Western Lightning chemiluminescence reagent (PerkinElmer) was used for antibody detection.
Identification of IcmW-interacting proteins
Legionella genomic DNA was isolated and partially digested with restriction enzyme Tsp509I to generate fragments with an average size between 2 kb and 4 kb. The prey vector pJG4-5 encodes the B42 transcriptional activation domain with a 3′EcoRI site that is used to construct two-hybrid libraries. To create two additional reading frames that will allow fusion of the B42 activation domain with proteins encoded by Tsp509I-digested Legionella DNA, linkers were added to pJG4-5 by digesting pJG4-5 with EcoRI, removing overhanging nucleotides with T4 exonuclease and then ligating linkers with the sequences pGGAATTCC (New England Biolabs ♯1020) or pCCGGAATTCCGG (New England Biolabs ♯1070) creating the plasmids pDMC1 and pDMC3 respectively. Legionella chromosomal DNA fragments partially digested with Tsp509I were ligated into the EcoRI site in pJG4-5, pDMC1 and pDMC3 to generate three different libraries. The IcmW protein was used as bait and the DupLEX-A yeast two-hybrid system by OriGene Technologies was used to screen these libraries using the methods recommended by the manufacturer. Yeast clones that grew on medium without leucine and that were found to produce β-galactosidase after plating on medium with X-gal and galactose were further characterized. Library plasmids from these yeast clones were isolated and sequenced to identify the fusion junction between the B42 activation domain and the downstream Legionella ORF. The wipA wipB and wipC sequences were deposited to GenBank (AY693712, AY693713, AY693714).
Translocation of potential substrates into host cells was assayed using the Cya fusion approach (Sory and Cornelis, 1994). A stable CHO cell line producing FcγRII (Joiner et al., 1990) was used, and described previously as a host for Legionella infection (Kagan and Roy, 2002). Cells were plated at 1 × 105 cells per well in a 24-well tissue-culture-treated dish and infected on the next day with the desired Legionella strain carrying a plasmid expressing the Cya fused to the gene of interest. The cells were infected at a multiplicity of infection (moi) of 30, and then centrifuged for 5 min at 150 g to initiate contact and synchronize the infection. Infected cells were incubated for 1 h at 37°C with 5% CO2. Cells were washed three times in ice-cold PBS and lysed in cold buffer containing 50 mM HCl and 0.1% Triton X-100 for 30 min at 4°C. The lysates were boiled for 5 min, and neutralized with 30 mM NaOH. Levels of cAMP were determined using the cAMP Biotrak enzyme immunoassay (EIA) system (Amersham Biosciences).
Intracellular growth assays
Intracellular growth assays were conducted in A. castellanii (ATCC strain 30234). Amoebae were grown in PYG medium in tissue culture flasks at room temperature until confluent. Two days before infection, the media were replaced. The day before infection, amoebae were replated into 24-well dishes. The day of infection, wells were washed two times with PBS and then amoebae buffer was added and dishes were incubated at 37°C with 5% CO2 for 1 h before infecting with Legionella. Host cell monolayers in 24-well dishes were infected with 2.5 × 105Legionella isolated from 2-day-old plate-grown cells. After 1–2 h of infection, each well was washed with PBS three times and then 1 ml of fresh amoebae buffer was added. For time point T0, 1 ml of sterile dH2O was added to lyse host cells. For subsequent time points, T24 h, T48 h, T72 h, supernatant from each well was collected and then the monolayer was lysed with sterile dH2O. Bacteria were collected, serial diluted, and three dilutions were plated on CYE agar. After 3 days of incubation at 37°C, cfu were counted to calculate bacterial cells per millilitre.
Affinity purification of substrates with the IcmS–IcmW complex
Escherichia coli BL21-DE3 was transformed with pEC66, a plasmid that produces an amino-terminal 6X-histidine-tagged (H6) IcmW protein with the IcmS protein encoded immediately downstream. The genes encoding these proteins are transcribed together upon induction of an upstream IPTG-inducible promoter. Plasmids producing either M45-WipA (pSN3), M45-RalF (pM45-ralF) or M45-SidG (pEC47) were introduced into the E. coli strain harbouring pEC66. Bacterial strains harbouring both plasmids were grown in LB broth overnight at 37°C, diluted 1:50 in 500 ml of LB broth and incubated on a shaker at 37°C to an OD600 of 0.8–0.9. IPTG was added to a final concentration of 1 mM and the cultures were incubated for an additional 3 h at 37°C. Bacterial cultures were pelleted by centrifugation and suspended in lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM DTT) plus 1 mM PMSF. Cultures were lysed using a French pressure cell (18 000 psi) and lysates were clarified by centrifugation. Clarified lysates were loaded on a Ni-NTA affinity column, washed sequentially with lysis buffer containing 10 mM, 25 mM and 50 mM imidazole and eluted with lysis buffer containing 0.5 M imidazole. Purification of H6IcmW/IcmS was confirmed by SDS-PAGE and coomassie staining. Co-purification of M45-WipA, M45-RalF or M45-SidG with H6IcmW/IcmS was determined by immunoblot analysis.
We are grateful to Dr Chris Tschudi (Yale University) for his technical assistance with gel filtration chromatograpy. This work was supported by NIH Grant AI41699 (C.R.R.) and The Human Frontiers Science Program (S.N.).