|Home | About | Journals | Submit | Contact Us | Français|
The virulence of Salmonella enterica critically depends on the functions of two type III secretion systems (T3SS), with the Salmonella pathogenicity island 1 (SPI1)-encoded T3SS required for host cell invasion and the SPI2-T3SS enabling Salmonella to proliferate within host cells. A further T3SS is required for the assembly of the flagella. Most serovars of Salmonella also possess a lipopolysaccharide with a complex O-antigen (OAg) structure. The number of OAg units attached to the core polysaccharide varies between 16 and more than 100 repeats, with a trimodal distribution. This work investigated the correlation of the OAg length with the functions of the SPI1-T3SS and the SPI2-T3SS. We observed that the number of repeats of OAg units had no effect on bacterial motility. The interaction of Salmonella with epithelial cells was altered if the OAg structure was changed by mutations in regulators of OAg. Strains defective in synthesis of very long or long and very long OAg species showed increased translocation of a SPI1-T3SS effector protein and increased invasion. Invasion of a strain entirely lacking OAg was increased, but this mutant strain also showed increased adhesion. In contrast, translocation of a SPI2-T3SS effector protein and intracellular replication were not affected by modification of the OAg length. Mutant strains lacking the entire OAg or long and very long OAg were highly susceptible to complement killing. These observations indicate that the architecture of the outer membrane of Salmonella is balanced to permit sufficient T3SS function but also to confer optimal protection against antimicrobial defense mechanisms.
Salmonella enterica is a remarkable pathogen with strategies for adaptation to different lifestyles in the environment as well as within various host organisms. The requirements for life within the host can vary dramatically, for example, after transition from extracellular life within the intestine to an intracellular life within a special organelle formed inside infected host cells (reviewed in reference 23). The presence of an outer membrane is an important structural feature that enables commensal as well as pathogenic bacteria to adapt to the intestine and to resist bile salts and various molecules of the host innate immune system (44). Of specific importance is the lipopolysaccharide (LPS), the major constituent of the outer leaflet of the outer membrane. LPS is composed of (i) the lipid A portion, consisting of acyl chains linked to phosphorylated N-acetylglucosamine; (ii) the inner and outer core moieties, consisting of rather conserved sugars; and (iii) a highly variable O antigen (OAg). The LPS of S. enterica serovar Typhimurium and several other serovars has special characteristics, with extreme heterogeneity in the length of the OAg repeats. LPS species with a short OAg (S-OAg) consisting of about 16 repeats of OAg units can be found. In addition, LPS species with long and very long OAg (L-OAg and VL-OAg, respectively) are present, containing about 35 and more than 100 repeats, respectively, of the OAg units. The synthesis of L- and VL-OAg in S. enterica serovar Typhimurium is regulated, and WzzST and WzzfepE, respectively, have been characterized as the corresponding regulators (5, 40; reviewed in reference 41).
In contrast to the commensal intestinal flora, S. enterica has the ability to breach intestinal barriers, to invade host cells, and to enter into an intracellular lifestyle. These hallmarks of the pathogenesis of Salmonella critically depend on the functions of two type III secretion systems (T3SS) (31, 46). T3SS are complex molecular machines that mediate the contact-dependent translocation of sets of effector proteins (reviewed in reference 18). Translocation requires the insertion of a membrane pore, termed a translocon, in the target membrane and the assembly of a needle-like oligomeric structure that links the translocon to the T3SS apparatus in the bacterial envelope (reviewed in reference 13). The T3SS encoded by Salmonella pathogenicity island 1 (SPI1) is active in extracellular bacteria and mediates the internalization of Salmonella into nonphagocytic cells by injection of effector proteins that manipulate the host cell actin cytoskeleton (46). The second T3SS is encoded by SPI2 and mediates the translocation of a second set of effector proteins by bacteria residing inside the Salmonella-containing vacuole, a specialized compartment that allows the intracellular proliferation of bacteria (31). Although both the SPI1-T3SS and the SPI2-T3SS are likely to have structural similarities, the numbers of active T3SS and their subcellular localization appear to be different (11, 30, 48).
The flagellar assembly system is considered the third T3SS in S. enterica (34). Flagellum-mediated motility can be considered a basic function of a bacterial cell. However, a contribution of motility to the interaction with host cells during pathogenesis has been demonstrated (50, 51).
Previous work indicates that the function of the T3SS required for invasion of Shigella flexneri is dependent on the modification of the structure of the LPS (52). We assume that similar functional constraints also apply to the function of the Salmonella T3SS for invasion and intracellular proliferation. The reduction of LPS length by glucosylation, as identified in S. flexneri, is not found in S. enterica. We questioned if alternative mechanisms exist that allow the function of LPS in mediating protection against antimicrobial function of the host as well as the proper functions of the SPI1-T3SS, the SPI2-T3SS, and the flagellar assembly system.
We investigated the effects of reduced numbers of repeats of OAg on the invasion and intracellular replication of Salmonella. Our data demonstrate that OAg length is critical for the function of the SPI1-T3SS and for bacterial invasion. In contrast, no influence of OAg length was observed on the functions of the SPI2-T3SS and the flagellar assembly system.
Routinely, 3-ml bacterial cultures were grown aerobically in glass test tubes at 37°C in a roller drum (TC-7; New Brunswick) with agitation at 30 rpm. Antibiotics were used at the following concentrations: carbenicillin or kanamycin, 50 μg ml−1; and tetracycline, 20 μg ml−1.
Bacterial strains used in this study are described in Table Table1.1. The one-step inactivation method (14) was used for the generation of strains deleted for wzzST, wzzfepE, or wzzST and wzzfepE in the background of wild-type (WT) Salmonella and for the deletion of waaL in the background of strain MvP1156 (sipA::sipAM45), MvP1281 (sipA::sipAM45::tem1), or MvP661 (sseJ200::luc FRT). After mutagenesis, the aph resistance cassettes were subsequently deleted by FLP-mediated recombination. Primers used for deletion and controls are listed in Table Table22.
For analyses of secretion and translocation, the SPI1 effector SipA-M45 from Salmonella 12023-derived strain MvP1156 and SipAM45-TEM1 from SL1344-derived strain M1104 (47) or the SPI2 effector SseJ200-Luc (20) was transferred into WT Salmonella 12023 and invC, ssaV, wzzST, wzzfepE, and wzzST wzzfepE mutants, using P22 transduction according to standard methods (35).
Plasmids carrying WT alleles of the deleted genes under the control of their own promoters were generated by PCR amplification of wzzST or wzzfepE, using primers listed in Table Table2.2. The PCR products were purified using a nucleotide removal kit (Qiagen); digested by BamHI/XbaI or EcoRI/BamHI, for wzzfepE or wzzST, respectively; and cloned into pWSK29 for the generation of p3388 and p3389, for expression of wzzfepE and wzzST, respectively. For complementation of the wzzST wzzfepE double mutant, wzzfepE was excised from p3388 by digestion with BamHI/NotI, gel purified using a gel extraction kit (Qiagen), and subcloned into BamHI/NotI-digested p3389.
For inducible expression of wzzST and wzzfepE, these genes were amplified by PCR using primers listed in Table Table2.2. A nucleotide removal kit (Qiagen) was used for purification of the PCR product; the purified wzzST fragment was digested with EcoRI and XbaI and cloned into EcoRI/XbaI-digested vector pBAD myc hisB. Subsequently, the wzzfepE fragment was digested by PstI and EcoRI and subcloned into the pBAD myc hisB-wzzST plasmid. The resulting plasmid, p3456, carries wzzfepE upstream of wzzST, with both under the control of the PBAD promoter.
Sensitivity to polymyxin B was tested as described before, with slight modifications (15). Bacterial overnight cultures were diluted 1:31 in fresh LB with appropriate antibiotics and were grown at 160 rpm and 37°C for 3.5 h to late logarithmic phase (SPI1-inducing conditions). Cultures were washed twice with tryptone saline (0.5% Bacto tryptone [Difco], 0.9% NaCl) and diluted to 1 × 104 bacteria per ml in tryptone saline with or without various concentrations of polymyxin B (Sigma). The assay mixtures were incubated for 1 h at 37°C in a thermomixer (Eppendorf, Hamburg, Germany) at 550 rpm. Following the incubation period, the samples were serially diluted in tryptone saline and plated on Mueller-Hinton (MH) plates to determine numbers of viable bacteria. The survival rates of bacteria exposed to polymyxin B were compared to those of controls without polymyxin B in order to normalize all samples, and the percent survival of WT bacteria was calculated.
To determine the MIC of polymyxin B, Etest strips (bioMérieux, Deutschland GmbH, Nürtingen, Germany) were used. Overnight cultures of various strains in BBL MH II broth (Becton Dickinson GmbH, Heidelberg, Germany) were adjusted to a 0.5 McFarland standard in fresh medium. Bacteria were spread on dry MH plates by use of a sterile cotton swab according to the manufacturer's recommendation. The Etest strips were applied in the middle of the agar surface, and the plates were inoculated overnight at 37°C.
For analyses of the sensitivities of various Salmonella strains to serum complement, bacteria were grown under SPI1-inducing conditions as described above, washed twice in phosphate-buffered saline containing 5 mM MgCl2 (PBSM), and diluted in PBSM to 1 × 108 CFU ml−1. Normal human sera (NHS) of four healthy volunteers were pooled and diluted to 50% in PBSM. Heat-inactivated serum (HIS) was prepared by incubation for 30 min at 56°C and dilution to 50% in PBSM. To test serum sensitivity, 50 μl of the bacterial suspension was mixed with an equal volume of NHS or HIS and incubated for 60 min at 37°C as described by Bliska and Falkow (9). Serial dilutions of the samples were prepared in PBSM and plated onto MH plates to determine the number of viable bacteria. The percent survival of the WT strain was calculated as follows: 100 × [mean (CFU ml−1 in NHS/CFU ml−1 in HIS) of mutant]/[mean (CFU ml−1 in NHS/CFU ml−1 in HIS) of WT].
The murine monocyte cell line RAW264.7 and the nonpolarized epithelial cell line HeLa were cultured in Dulbecco's modified Eagle's medium (DMEM) containing a high glucose concentration (4.5 g/liter), l-glutamine, and sodium pyruvate (PAA, Pasching, Austria) and supplemented with 10% fetal calf serum (Sigma-Aldrich, Seelze, Germany) as well as 2 mM Glutamax (Invitrogen, Karlsruhe, Germany). The canine kidney epithelial cell line MDCK was cultured in minimum essential medium (MEM; PAA) supplemented with penicillin-streptomycin (PAA) as described previously (19). All cell lines were maintained at 37°C in an atmosphere containing 5% CO2 and 90% humidity.
Bacteria grown on LB agar were inoculated into 3 ml of LB medium and grown for 16 h at 37°C. Subsequently, bacteria were subcultured (1:31 from LB overnight culture) for 3.5 h at 37°C. One hundred fifty microliters of bacterial culture was harvested by centrifugation for 3 min at 18,000 × g and resuspended in lysis buffer (2% sodium dodecyl sulfate [SDS], 4% β-mercaptoethanol, 10% glycerol, 1 M Tris-HCl, pH 6.8, and bromophenol blue) according to the optical density at 600 nm (OD600) (50 μl of lysis buffer per 1.6 OD600 units). Lysates were boiled at 100°C for 10 min, and subsequently, 2 U Turbo DNase was added to the chilled samples for 15 min at room temperature (RT). For protein digestion, samples were treated with 50 μg proteinase K (MBI-Fermentas) at 60°C for 1 h. LPS preparations were separated in 15% polyacrylamide gels, using a Laemmli buffer system at 25 mA (2 gels) for 1.5 h according to standard protocols (4), and were visualized by silver staining using the procedure described by Hitchcock and Brown (24).
The swimming and swarming behaviors of WT Salmonella and various mutants were analyzed using semisolid plates as described elsewhere (29). Briefly, the surfaces of swim plates containing 0.25% Difco Bacto agar and 0.5% glucose in LB were carefully pierced in the center with 0.2 μl of bacterial liquid overnight cultures. Swarm plates containing 0.5% Difco Bacto agar and 0.5% glucose in LB were inoculated with 0.5 μl of bacterial overnight cultures dropped on the surface of the plate. After incubation at 37°C for 5 to 6 h (swimming) or 12 to 13 h (swarming), the plates were photographed with a Canon EOS 450D camera.
MDCK cells were seeded in 24-well plates (Greiner Bio-One) at a density of 1 × 105 cells/well and allowed to polarize for 5 to 6 days, reaching a density of about 5 × 105 cells per well. Three hours prior to infection, culture MEM was replaced by fresh MEM without penicillin-streptomycin. HeLa cells were seeded in 24-well plates at a density of 5 × 105 cells per well 1 day before infection and allowed to duplicate. For infection under SPI1-inducing conditions, bacteria were subcultured 1:31 from an overnight culture in LB with appropriate antibiotics and incubated for 3.5 h at 37°C. The bacterial cultures were diluted in PBS to an OD600 of 0.2 (equivalent to 4 × 108 bacteria ml−1). A master mix of the inoculum of approximately 2.5 × 106 bacteria per well (multiplicity of infection [MOI], 5) for MDCK infection or 1 × 106 bacteria per well (MOI, 10) for HeLa cell infection was prepared in MEM or DMEM, and 300 μl was added to each well. For the adhesion assay, the inoculum and the cells were prechilled on ice for 15 min in order to inhibit invasion. The infection procedure was performed on ice, with further incubation at 4°C for 30 min. Subsequently, the infected monolayer was washed twice with prechilled PBS and lysed with 500 μl per well of 0.5% sodium deoxycholate (for MDCK cells) or 0.1% Triton X-100 (for HeLa cells) for 15 min at 4°C. For the invasion assay, infection was performed for 30 min at 37°C, and subsequently, plates were washed twice with PBS. In order to allow bacterial invasion, cells were incubated for an additional 1 h at 37°C in MEM or DMEM containing 100 μg ml−1 gentamicin to kill extracellular bacteria. The cells were lysed with 500 μl per well of 0.5% sodium deoxycholate (for MDCK cells) or 0.1% Triton X-100 (for HeLa cells) for 5 min at 37°C. Serial dilutions of the inoculum and the lysates were plated onto MH plates to determine the number of intracellular bacteria. The percentages of adherent (number at 30 min versus number in inoculum × 100) and invading (number at 1 h versus number in inoculum × 100) bacteria were calculated.
For arabinose induction of strains containing p3456 prior to infection, overnight cultures were diluted 1:50 in fresh medium and incubated at 37°C for 1.5 h (OD600 of approximately 0.6 to 0.8) in a shaking air incubator at 150 rpm. The cells were collected by centrifugation in a microcentrifuge for 10 min at 5,000 rpm and diluted 1:3 in LB containing 0.2% arabinose (PBAD induction) or 0.2% glucose (PBAD repression) as described by Khlebnikov et al. (28). Bacteria were grown for an additional 3.5 h at 37°C in order to obtain SPI1-inducing conditions.
RAW264.7 macrophages were seeded at a density of 2 × 105 cells per well into 24-well plates 24 h prior to infection and then allowed to duplicate. Bacteria were grown overnight in LB with appropriate antibiotics and stored at 4°C until use. Immediately prior to infection, the OD600 of the bacterial culture was measured and the bacteria were diluted in PBS to an OD600 of 0.2 (equivalent to 4 × 108 bacteria ml−1). A master mix of the inoculum of approximately 4 × 105 bacteria per well (MOI = 1) was prepared in DMEM, and 300 μl was added to RAW264.7 cells. Infection was synchronized by centrifugation at 500 × g for 5 min, and macrophages were allowed to internalize bacteria for 25 min at 37°C. The cells were washed twice with PBS, and 500 μl/well DMEM containing 100 μg ml−1 gentamicin was added for 1 h to kill remaining extracellular bacteria. For the rest of the experiment, 500 μl/well DMEM containing gentamicin at 15 μg ml−1 was applied to the infected cells. At 2 h and 16 h postinfection, the macrophages were washed twice with PBS and lysed for 10 min at RT with 500 μl per well of 0.1% Triton X-100. Serial dilutions of the inoculum and the lysates were plated onto MH plates to enumerate the intracellular CFU. The percentage of phagocytosis (value at 2 h versus value for inoculum × 100) and x-fold intracellular replication (value at 16 h versus value at 2 h) were calculated.
Analyses of the translocation of SPI1-encoded proteins were conducted as described by Schlumberger et al. (48). Briefly, 1 day prior to infection, 6 × 103 HeLa cells were seeded in 96-well half-area μclear plates (Greiner Bio-One), and if indicated, 50 μM cytochalasin D (Sigma-Aldrich) was added 30 min prior to infection to inhibit actin polymerization and Salmonella invasion. Cells were washed twice and infected with SPI1-induced Salmonella strains expressing a chromosomal SipAM45-TEM1 fusion at an MOI of 100 for 30 min at 37°C (17). Cells were washed twice at room temperature with Hanks balanced salt solution containing 10% fetal calf serum and 100 μg ml−1 gentamicin. The infected cells were loaded with 1 μg ml−1 CCF2-AM (Invitrogen) according to the manufacturer's recommendation and incubated for another 3 h at RT. The fluorescence signals were detected via a Victor3 plate reader (PerkinElmer), with excitation at 405 nm and emission at 460 nm (blue fluorescence) as well as at 535 nm (green fluorescence). Background signals were subtracted, and the translocation of SipAM45-TEM1 was quantified by calculating the blue/green ratio of the background-corrected data.
Analyses of SPI2 protein translocation in RAW264.7 cells infected with Salmonella harboring an SseJ-Luc fusion protein were performed as described before (20). Briefly, 8 × 105 RAW264.7 macrophages seeded in 24-well plates were infected for 30 min with Salmonella strains expressing the SseJ-Luc fusion at an MOI of 10. Infected cells were washed twice with PBS, and gentamicin at 100 μg ml−1 was added for 1 h and replaced by gentamicin at 15 μg ml−1 for the remaining experiment. At 15 h postinfection, macrophages were washed twice with PBS and lysed with 100 μl of eukaryotic lysis buffer (Roche). Ten microliters of the lysate was used to perform serial dilution series to determine the number of viable intracellular bacteria. The residual lysate was centrifuged, and aliquots of the supernatant were applied to 96-well microtiter plates (Microfluor; Dynatech). Luciferase activities were quantified using a TopCount instrument (PerkinElmer) and expressed in relative light units. The amount of translocated SseJ-Luc was calculated as the number of relative light units per intracellular bacterium.
For analyses of in vitro expression and secretion of either SPI1- or SPI2-encoded proteins in the background of the various OAg mutants, samples were prepared as follows. For induction of SPI1-encoded proteins, bacteria expressing a chromosomal sipA::M45 fusion were cultured overnight in LB with appropriate antibiotics, diluted 1:31 in fresh medium, and subcultured for another 3.5 h at 37°C. Samples of 2.5 ml of the bacterial cultures were harvested by centrifugation. Induction of SPI2 proteins was done according to the method of Coombes et al. (12). Briefly, Salmonella cells grown overnight in LB were washed twice in low-phosphate, low-magnesium medium (LPM) at pH 5.8 and diluted 1:50 in fresh LPM, pH 5.8. Cultures were grown for 7 h with shaking at 180 rpm, and 1.8 ml of culture was harvested by centrifugation.
The culture supernatants were filtered through a 0.2-μm filter (0.2-μm Acrodisc syringe filter with PES membrane; Pall Life Sciences), and 100 ng purified p39 (a gift from Gerhard Groer, Institute for Microbiology, Erlangen, Germany) was added as a precipitation control. Proteins were recovered by precipitation with trichloroacetic acid (TCA) (final concentration, 10% [wt/vol]) overnight at 4°C and centrifugation for 45 min at 18,000 × g. The pellet was washed twice with 1 ml cold acetone and recovered by centrifugation at 18,000 × g for 30 min. The air-dried final protein pellet of the supernatant as well as the bacterial cell pellet was solubilized with a volume of SDS sample buffer (50 mM Tris-HCl, pH 6.8, 4% SDS, 2% β-mercaptoethanol, 12.5% glycerol, and 0.01% bromophenol blue) adjusted according to the OD600 of the original culture.
Samples were separated by SDS-polyacrylamide gel electrophoresis on a Tris-Tricine system (45). For Western blot analysis, proteins were transferred onto 0.2-μm nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany) in Towbin buffer, using discontinuous semidry immunoblotting, as previously described (7).
The SPI1 protein SipA-M45 was detected via a monoclonal antibody against the M45 tag, and the SPI2 protein SseB was detected via antisera against recombinant SseB. Anti-DnaK antibody was used to control for equal loading of bacterial lysates as well as for release of cytosolic protein into the culture supernatant. Anti-p39 (a gift from Gerhard Groer) was used to detect purified p39 added to culture supernatants as a precipitation control.
IR800 and IR680 (gifts from Reinhard Voll) were used as secondary antibodies, and an Odyssey infrared imaging system (Li-Cor) was employed to visualize the infrared signals. The intensities of the protein bands were quantified by the Odyssey software, and the values were expressed as ratios of protein to loading control or protein to precipitation control and plotted on a graph.
Prism 4.0 software was used to calculate statistical significance. Student's t test was used for comparison of two groups, while the one-sample t test was used for comparison of the mean of a sample to a known value (see Fig. Fig.66 and Fig. 7A and B).
The biosynthesis of the LPS of S. enterica serovar Typhimurium has been studied in detail, and gene functions for the addition of the OAg and for the regulation of the OAg chain length are known from previous studies (8, 33, 38, 39). To generate a set of isogenic mutant strains with defined defects in the OAg, wzzfepE and wzzST were deleted, as regulators of VL-OAg and L-OAg, respectively. Furthermore, a double mutant in wzzfepE and wzzST was generated that expressed only S-OAg. Finally, mutant strains were generated in waaL (previous designation, rfaL), which is required for the addition of the OAg chain to the outer core. The strains were grown in liquid media and processed for analyses of the LPS (Fig. (Fig.1).1). LPS profiles indicated that all mutant strains had the expected phenotypes with respect to OAg length. The waaL mutant strain entirely lacked OAg, the wzzfepE wzzST double mutant expressed only S-OAg, and the wzzST and wzzfepE strains were devoid of L-OAg and VL-OAg species, respectively. To restore their functions, the WT alleles of wzzST and/or wzzfepE were cloned into low-copy-number plasmids under the control of their native promoters. Analysis of the mutant strains harboring the complementation plasmids indicated fully restored OAg length of the LPS (Fig. (Fig.1).1). We also analyzed growth rates of the various strains and did not observe differences in growth in rich medium (LB broth) or in minimal medium (PCN at pH 7.4 or 5.8) (data not shown).
For further functional characterization of the set of mutant strains, we determined the sensitivities of the various strains to polymyxin B, as a representative antimicrobial peptide, and to human serum complement (Fig. (Fig.2).2). The routinely applied Etest for polymyxin B indicated that the MIC for the WT strains, all mutant strains, and complemented mutant strains was 500 ng ml−1 polymyxin B (data not shown). As a more sensitive approach, the sensitivity toward polymyxin B was investigated in liquid medium (Fig. (Fig.2A).2A). The strains lacking L-OAg, VL-OAg, or both L-OAg and VL-OAg exhibited a small increase in their sensitivity to polymyxin. In contrast, the waaL strain lacking the entire OAg showed a highly increased sensitivity, with about 5% survival after exposure to 500 ng ml−1 polymyxin B. The contribution of OAg length to complement resistance was more dramatic (Fig. (Fig.2B).2B). A strain lacking VL-OAg was not affected in survival in serum. In accordance with a previous study (39), we observed that the mutant lacking L-OAg was about 100-fold more sensitive to complement-mediated lysis. Interestingly, we observed that the wzzST wzzfepE mutant strain as well as the waaL mutant lacking the entire OAg showed similar highly increased (about 1,000-fold) sensitivities to complement killing.
Collectively, these data indicate that OAg is an important factor for the protection of Salmonella against polymyxin B and serum complement. With respect to the different modal lengths, L-OAg and VL-OAg have only minor contributions to resistance against polymyxin but an important function for resistance of Salmonella against serum complement.
We investigated if variations of the OAg length affect the function of the flagellar assembly system and compared the motilities of WT Salmonella and various mutant strains (Fig. (Fig.3).3). A mutation resulting in the loss of the entire OAg led to a highly reduced motility that was comparable to that for a flagellar defect caused by a deletion of the main flagellar subunit. As a simple test for the assembly of flagella, we performed slide agglutination with Hi-H1,2 antisera against flagellar protein. Agglutination was observed for the WT and all mutant strains with defects in OAg or T3SS. No agglutination was detected for a mutant strain defective in fliGHI (Fig. (Fig.33).
During the interaction of Salmonella with the intestinal mucosa, adhesion to epithelial cells and host cell invasion are important virulence functions. We investigated the effect of OAg on this interaction (Fig. (Fig.4).4). Since our previous work indicated a profound difference in the invasion of nonpolarized and polarized epithelial cells (19), cell culture models of HeLa cells (nonpolarized) and MDCK cells (polarized) were used. The ability to adhere to HeLa cells or MDCK cells was not affected by the lack of L-OAg or VL-OAg and was slightly increased for a strain lacking both L-OAg and VL-OAg. The mutant strain lacking the entire OAg showed highly increased (about fivefold) rates of adhesion. The SPI1-mediated invasion was quantified by a gentamicin protection assay. A strain defective in the SPI1-T3SS was dramatically reduced in triggering its entry into HeLa cells as well as into MDCK cells. The wzzST strain showed slightly reduced invasion, while the wzzfepE strain had threefold higher rates of invasion. The most dramatic increase in invasiveness was observed for the wzzST wzzfepE strain. This strain invaded HeLa and MDCK cells at 6.8-fold and 2.7-fold higher rates, respectively, than that of WT Salmonella. The presence of complementing plasmids again reduced the rate of invasion to the level of the WT strain. The waaL strain was also fivefold more invasive in HeLa cells, and this increase is similar to the increase in adhesion of this strain. Although increased adhesion to MDCK cells was observed for the waaL strain, the invasion of the polarized cells was similar to that of the WT strain.
Since invasion was increased in the absence of L-OAg and VL-OAg, we next investigated if invasion would be affected by overexpression of these LPS species. To test this hypothesis, a plasmid for the arabinose-inducible expression of wzzST and wzzfepE was generated and introduced into the WT and wzzST wzzfepE double mutant strains. We observed that invasion of the mutant strain was higher than that of the WT strain after growth under noninducing conditions. In contrast, induction of wzzST wzzfepE expression by growth in arabinose-containing medium resulted in 3.6-fold decreased invasion. This result further demonstrates the correlation between OAg chain length and host cell invasion by Salmonella.
Our data show that adhesion to host cells is largely independent of the length of the OAg and that only a complete lack of the OAg increases binding to epithelial cells. The invasion of host cells is strongly affected by OAg length. The presence of L-OAg and VL-OAg appears to interfere with efficient entry into epithelial cells, but this effect is distinct from an altered adhesion to host cell surfaces. We did not observe major differences in the effect of OAg length during interaction with nonpolarized and polarized cells, but the effect on invasiveness was more pronounced for nonpolarized cells.
We analyzed the effect of alterations of OAg length on phagocytic uptake and intracellular replication of Salmonella in the macrophage-like cell line RAW264.7 (Fig. (Fig.5).5). The phagocytic uptake of ssaV, wzzST, and wzzfepE strains was not significantly different from that of WT Salmonella. The uptake of the wzzST wzzfepE double mutant was slightly higher, and the waaL strain was phagocytosed at about threefold higher rates (Fig. (Fig.5A5A).
The intracellular fate of Salmonella is critically dependent on the function of the SPI2-encoded T3SS and its effector proteins. Thus, we used a strain defective in the SPI2-T3SS as a control with reduced intracellular proliferation. As expected, the SPI2 strain showed highly reduced intracellular proliferation. In contrast, the intracellular replication of all mutant strains with altered OAg lengths was similar to that of the WT strain (Fig. (Fig.5B).5B). These data indicate that the OAg length is not critical for the intracellular fate of Salmonella in macrophages.
The observation that OAg length influences host cell invasion but not intracellular proliferation prompted us to investigate the functions of SPI1-T3SS and SPI2-T3SS in strains expressing various OAg lengths. To analyze if the number of OAg repeats affects the function of the SPI1-T3SS or the SPI2-T3SS, we quantified the secretion and translocation of selected effector proteins of either T3SS.
For the quantification of secretion by the SPI1-T3SS, we monitored SipA, a substrate protein with direct effector function. The synthesis of SipA under SPI1-inducing conditions was similar for the WT and for strains with altered OAg lengths (Fig. (Fig.6A).6A). The secretion of SipA into the culture supernatant was similar for all mutant strains except for the wzzST strain, which exhibited a slight secretion defect of SipA (Fig. (Fig.6B).6B). Only the invC strain, as expected, lacked secretion of SipA. To quantify translocation by the SPI1-T3SS, a fusion protein consisting of SipA and the TEM β-lactamase as a reporter was used. As previously described, the translocation of this hybrid protein can be followed by the conversion of a TEM substrate by a fluorescence resonance energy transfer-based assay (47). We observed that SipA translocation, corresponding to TEM activity, was very low for the invC and wzzST strains (Fig. (Fig.6C).6C). Translocation by the waaL strain was similar to that by the WT, the wzzfepE strain showed a >2-fold higher translocation rate, and about 4.5-fold higher reporter activities were detected for the wzzST wzzfepE double mutant strain. The very low level of translocation of the SipA-TEM reporter by the wzzST strain was in contrast to the minor effect of this mutation on invasion (Fig. (Fig.4B)4B) but correlates with the reduced SipA secretion in vitro (Fig. (Fig.6B).6B). To test if the presence of the SipA-TEM reporter protein affects invasion, we compared the invasion by wzzST strains with and without the sipA::tem fusion (data not shown). Our data indicate that the presence of the reporter fusion results in reduced invasion. This interference was observed only for the wzzST strain, not for other mutant strains harboring the reporter construct.
For analysis of secretion by the SPI2-T3SS, we monitored SseB, a component of the translocon. The synthesis of SseB was affected neither by any of the alterations in the OAg nor by a defect in ssaV, encoding an essential subunit of the SPI2-T3SS (Fig. (Fig.7A).7A). No secretion of SseB into the culture supernatant was observed for the ssaV strain (Fig. (Fig.7B).7B). The secretion of SseB by the wzzST, wzzfepE, and wzzST wzzfepE strains was comparable to that of the WT strain, while secretion by the waaL strain was about fourfold lower (Fig. (Fig.7B).7B). None of the strains released detectable amounts of the cytoplasmic protein DnaK, indicating that none of the OAg defects disturbed the integrity of the cell envelope. To quantify translocation by the SPI2-T3SS, we used a luciferase fusion to SseJ, an effector protein that is translocated in large amounts by intracellular Salmonella (22, 42). The use of a chromosomal fusion between SseJ and luciferase has been described previously (20). In order to compensate for any differences in the degree of intracellular replication of the various strains, the number of intracellular bacteria was determined and the luciferase activity was normalized to the number of intracellular bacteria. In accordance with our previous observations, we found that a mutation in ssaV resulted in a severe reduction of luciferase activity. The luciferase activities determined for lysates of cells infected with strains with altered OAg lengths were similar to that for WT-infected host cells. These results show that the function of the SPI2-T3SS of intracellular Salmonella is not affected by the length of the OAg.
The LPS constitutes a ubiquitous component of the cell envelopes of the large group of gram-negative bacteria. Despite its wide distribution, the LPS is highly variable in structure, and the most variability can be observed in the chemical structure and length of the OAg of the LPS. In this study, we systematically investigated the correlation between the OAg length of the LPS of Salmonella enterica serovar Typhimurium and the functions of the three T3SS of this pathogen.
The effect of OAg defects on complement resistance was in line with previous studies (21, 27). Early studies did not indicate a role of the OAg in resistance to antimicrobial peptides, and this resistance is thought to be independent of OAg but dependent on the lipid A portion of LPS. The functions of the T3SS were affected to different degrees by variations of the OAg length. Our data from Etest assays confirmed these results, but analyses in liquid culture suggested increase polymyxin B-mediated killing of wzzfep wzzST and waaL strains. Similar observations have been made for Erwinia carotovora strains with defects in OAg (6). No change in flagellum-mediated motility was observed for mutants with truncated OAg, and only the waaL strain lacking the entire OAg was highly reduced in motility. This correlates with previous reports on the effect of a waaL mutation on the motility of Pseudomonas aeruginosa (1) or Erwinia amylovora (6). The SPI2-T3SS is essential for intracellular replication of Salmonella in cultured RAW264.7 macrophages. Intracellular replication was not affected by removal of L-OAg and VL-OAg or by the entire lack of the OAg. This finding is in accord with the intracellular function of the SPI2-T3SS, since translocation of a representative SPI2-T3SS effector was not affected by any of the modifications of the OAg length investigated here. If Salmonella is residing in the Salmonella-containing vacuole and closely surrounded by a vacuolar membrane, the presence of OAg would not affect the function of the SPI2-T3SS needle or the insertion of the translocon due to steric hindrance.
In contrast, we found that the OAg length affected extracellular Salmonella in a rather different manner. The adhesion to and invasion of epithelial cells were strongly influenced by truncations of the OAg. This finding may indicate that WT Salmonella synthesizes OAg that negatively affects invasion. However, the protection mediated by the long and very long forms of OAg is likely more important for adaptation of intestinal colonization, or invasion is already very efficient, so that a compromised invasion may be tolerated.
Currently, alternative roles of Salmonella invasion are being discussed. Ackermann et al. (2) proposed that invasion by a proportion of the intestinal Salmonella organisms is important to trigger an inflammatory response that affects the competing microbial flora. This effect requires invasion by only a certain subpopulation of Salmonella and would be compatible with a suboptimal invasiveness. The work of Gerlach et al. (19) revealed that the interaction between Salmonella and polarized epithelial cells is more complex than initially thought. The presence of the apical brush borders appears to be an efficient barrier, and Salmonella requires the function of the SPI4-encoded large adhesin SiiE to efficiently adhere to and subsequently invade polarized cells. The divergent roles of OAg species of Salmonella enterica serovar Typhimurium in motility, host cell invasion, and intracellular replication are also summarized in Fig. Fig.88.
In contrast to observations by Murray et al. (38), we could not detect an influence of OAg length on uptake by macrophages. In our setting, only mutant strains lacking the OAg were taken up in significantly larger amounts by phagocytic cells. An experimental difference may be the competition invasion/uptake assay performed by Murray et al., in contrast to the conventional phagocytosis assay performed in our study. We also failed to observe the reduction of OAg length in intracellular Salmonella (data not shown) that has been reported previously (32) and is thought to be dependent on the PmrA/PmrB and RcsC/YojN/RcsB regulators (16). It is likely that modifications resulting in OAg truncation occur late in the intracellular life cycle (>16 h postinfection). The function of the SPI2-T3SS in host cell modification is more important in the initial intracellular phase (2 to 6 h postinfection). Any interference with SPI2-T3SS function by OAg would be critical in this phase.
In contrast to Salmonella, Shigella possesses only two modal forms of the OAg. The synthesis of S-OAg chains with a modal length of 11 to 17 repeat units is mediated by WzzSF. This function is required for sufficient expression of the cell surface protein IcsA, which mediates F-actin comet tail formation and cell-to-cell spread (43). Shigella OAg is also thought to partially mask IcsA (37). The VL-OAg, containing more than 90 repeat units, is determined by WzzpHS2 (CldpHS-2), which is encoded on the pHS-2 plasmid. This OAg type is required for serum resistance and enhances the Sereny reaction in mice (26, 36). Previous work (52) demonstrated that a modification of the OAg in Shigella flexneri is required for the optimal balance between invasion and resistance against antimicrobial factors. This modification was mediated by glucosylation of OAg. Mutant strains in gtr deficient in LPS glucosylation are highly attenuated in virulence in a rabbit ligated ileal loop model. It was also found that gtr mutants show an about 10-fold reduced invasion compared to the WT strain with glucosylated OAg, and ultrastructural analyses indicated that the needle structures assembled by the Shigella mxi/spa T3SS are less accessible at the cell surface of a gtr strain. Glucosylation of OAg has been reported for S. enterica serovar Typhimurium as well (10), but the resulting O12-2 OAg variant appears to be important mainly for immune evasion, and the effect on OAg length has not been investigated.
Our data show that the functional constraints on OAg are similar in Shigella and Salmonella, but both species evolved different mechanisms to modify the OAg to balance between resistance against antimicrobial effectors and host cell invasion. In both organisms, the resistance against antimicrobial effectors appears most important, and a reduction of the efficiency of invasion is tolerated.
It was recently shown that changes in the OAg structure of P. aeruginosa also affect the secretion of T3SS effectors, such as exoS, exoT, or pcrV, the cytotoxic effect mediated by Pseudomonas, and the severity of disease in an animal model (3). In contrast, we did not detect altered expression levels or amounts of secreted effector proteins as a consequence of OAg defects, but we saw altered characteristics of adhesion, invasion, and effector proteins in the case of OAg-deficient Salmonella.
At present, it is not known if the different LPS species are homogenously distributed in the outer membranes of S. enterica and related species or if membrane microdomains exist that are enriched for S-OAg, L-OAg, or VL-OAg. Such a distribution could minimize the interference of the OAg with the functions of T3SS that may be located in regions with smaller amounts of L-OAg and VL-OAg. Unfortunately, the inability to directly visualize LPS by electron microscopy hampers the analysis of the distribution of LPS species within the outer membrane.
In conclusion, our study demonstrates that the presence of three length variants of OAg of LPS allows S. enterica serovar Typhimurium to balance between defensive and offensive virulence functions required for adaptation to various niches within the eukaryotic host. A major challenge for future research will be the visualization of the bacterial cell surface with approaches that maintain the integrity of the carbohydrates.
This work was supported by the Elitenetzwerk Bayern as part of the Lead Structures of Cell Function graduate school and by grant HE1964/8-3 of the Deutsche Forschungsgemeinschaft.
We thank Barbara Bodendorfer for excellent technical support and Gerhard Groer for providing antiserum against p39.
Editor: A. J. Bäumler
Published ahead of print on 21 September 2009.