Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
The oxidative burst produced by the NADPH oxidase (Phox) is an essential weapon used by host cells to eradicate engulfed pathogens. In Salmonella typhimurium, oxidative stress resistance has been previously proposed to be mediated by the pathogenicity island 2 type III secretion system (T3SS-2), periplasmic superoxide dismutases and cytoplasmic catalases/peroxidases. Here, we fused an OxyR-dependent promoter to the gfp to build the ahpC-gfp transcriptional fusion. This reporter was used to monitor hydrogen peroxide levels as sensed by Salmonella during the course of an infection. We showed that the expression of this fusion was under the exclusive control of reactive oxygen species produced by the host. The ahpC-gfp expression was noticeably modified in the absence of bacterial periplasmic superoxide dismutases or cytoplasmic catalases/peroxidases. Surprisingly, inactivation of the T3SS-2 had no effect on the ahpC-gfp expression. All together, these results led to a model in which Salmonella resistance relies on its arsenal of detoxifying enzymes to cope with Phox-mediated oxidative stress.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Salmonella enterica serovar Typhimurium is a facultative intracellular bacterium that causes a wide range of infections in mammals, ranging from self-limiting gastroenteritis to severe systemic diseases. Among the key points leading to the establishment of infection is the ability of the pathogen to survive and proliferate within host cells, and more precisely inside a membrane-bound compartment called the Salmonella-containing vacuole (SCV) (Haraga et al., 2008). The host makes use of various killing mechanisms to eradicate Salmonella, e.g. synthesis of antimicrobial peptides and production of reactive oxygen species (ROS).
The generic term ROS covers a series of highly reactive molecules such as the superoxide anion (O2.−), hydrogen peroxide (H2O2) or the hydroxyl radical (HO.). The phagocyte NADPH oxidase (Phox) has been shown to be one of the most powerful enzymatic complexes producing the superoxide anion O2.− (Miller and Britigan, 1997; Babior, 1999). Such a molecule plays a major antibacterial role (Vazquez-Torres et al., 2000a; Huang and Brumell, 2009). Humans who are genetically defective in O2.− production develop chronic granulomatous disease, an illness characterized by recurrent life-threatening bacterial and fungal infections (Winkelstein et al., 2000; van den Berg et al., 2009). O2.− being short-lived, it is rapidly dismutated into H2O2, which can diffuse across membranes. Ultimately, H2O2 is converted to HO. in the presence of Fe2+ during the Fenton reaction. All together, these ROS can kill bacterial cells by oxidizing and damaging proteins and nucleic acids (Storz and Imlay, 1999).
Salmonella expresses an arsenal of detoxifying enzymes, which differ by their cellular location and their substrate specificity. Two periplasmic Cu/Zn superoxide dismutases, SodCI and SodCII, contribute to the dismutation of superoxide O2.− into H2O2 and O2 (Fang et al., 1999). SodCI but not SodCII was found to play a role during infection of mice by Salmonella (Uzzau et al., 2002; Krishnakumar et al., 2004). Recently, we have shown that three catalases and two peroxidases are involved in H2O2 degradation within the bacterial cytoplasm (Hébrard et al., 2009). A third peroxidase, Tpx, was also found to scavenge H2O2 and facilitate intracellular growth in macrophages (Horst et al., 2010). Collectively, these enzymes contribute to bacterial survival during infection and to virulence in mice. Moreover, the type III secretion system (T3SS) encoded by the Salmonella pathogenicity island 2 (SPI-2) has been proposed to exclude the membrane component of the Phox, flavocytochrome b558, from the SCV membrane (Gallois et al., 2001). Thereby, this exclusion mechanism would prevent the assembly of the Phox complex and the intraphagosomal production of the superoxide anion (Vazquez-Torres et al., 2000b; Gallois et al., 2001). Such a mechanism would require the secretion of an effector targeted to the Phox, which remains to be identified. Thus, a comprehensive picture could be that Salmonella relies on two ways to cope with oxidative stress: metabolizing ROS produced by the host and inhibiting the Phox assembly. The first way could consist in the dismutation of O2.− into H2O mediated by periplasmic superoxide dismutases and cytoplasmic catalases and peroxidases. The second way might be T3SS-2-dependent and would avoid O2.− production.
Experimentally, a well-defined picture of the oxidative burst has always been difficult to obtain as the chemical probes used to quantify ROS are lacking both in substrate specificity and subcellular location definition (Vazquez-Torres et al., 2000a; Amatore et al., 2007). Moreover, the precise quantification of ROS is a real challenge because their production and degradation rates are very rapid, many enzymes being dedicated to metabolize such molecules (i.e. Sod for O2.− or catalases for H2O2). In the present paper, we have built a transcriptional fusion to monitor the oxidative stress level. Using dedicated genetic backgrounds, we monitored the expression of the fusion to investigate the contribution of cytoplasmic, periplasmic and secreted defences, and we have identified the exclusive source of ROS production in macrophages. Our results allowed the characterization of the oxidative burst produced by the host and of its perception by Salmonella during the course of infection. They also assigned a major role to Salmonella detoxifying enzymes in protection against oxidative damage.
Construction and characterization of transcriptional fusion dedicated to oxidative stress
The ahpC gene is a member of the OxyR regulon, and its expression was found to be highly induced in presence of H2O2 (Zheng et al., 2001). Therefore, the pLA40 plasmid was generated by fusing the ahpC promoter region to the gfpmut3a gene, and a Salmonella wild-type strain was transformed with this vector. Increasing H2O2 levels were added to this strain, and GFP synthesis was monitored over time. In the absence of hydrogen peroxide, the fluorescence level remained stable, whereas addition of 2 to 10 µM H2O2 led to a dose-dependent induction of 2.4- to 4.5-fold respectively (Figs 1A and S1). The maximal level of fluorescence appeared between 15 and 30 min after stress, and the fluorescence level decreased steadily between 30 and 180 min (Fig. 1A). A dose-dependent induction of the fusion was also observed when the bacteria were submitted to different amounts of paraquat, a superoxide generator (Fig. S2). To find out more about the dynamics and the detection range of this reporter, bacteria carrying the ahpC-gfp fusion were submitted to different doses of H2O2 every 20 min. This led to a gradual increase in the fluorescence level (Fig. 1B). Repeated stresses also displayed a dose-dependent induction of the reporter: 2, 4, 6 and 8 µM of H2O2 yielded 4.9-, 9.2-, 11- and 14.7-fold induction after 150 min, respectively, compared with non-treated bacteria (Fig. 1B). As a control, an oxyR mutant was transformed with the pLA40 plasmid, and the resulting strain was submitted to 10 and 20 µM H2O2 (Fig. 1C). The fluorescence level of this strain did not change in the presence of H2O2, which confirmed that the fusion is OxyR-dependent.
Next, we wanted to know if intracellular Salmonella senses H2O2 within its SCV. Bone marrow-derived macrophages (BMM) from C57BL/6 mice were infected with Salmonella strains transformed with the pLA40 plasmid. Intracellular bacteria were extracted after different times of infection, and the fluorescence level of bacteria was monitored by flow cytometry (Fig. 1D and E). Single peak GFP histograms were recorded for Salmonella extracted from macrophages, indicating that the entire population of intracellular bacteria was responding to the oxidative environment sensed within the vacuole (Fig. 1D). The mean GFP levels of a wild-type strain increased significantly up to 90 min after infection, then stalled for the next 3 h, and finally decreased slightly (Fig. 1E). Conversely, the fluorescence in an oxyR mutant remained at the basal level as expected (Fig. 1E). These results indicated that Salmonella sensed H2O2 during BMM infection. Finally, to estimate the H2O2 level to which intracellular bacteria were exposed, we compared by flow cytometry the fluorescence of a wild-type strain carrying the ahpC-gfp fusion during BMM infection with an identical strain submitted to 10 µM H2O2 in culture medium (Fig. 1F). The GFP level reached 3 h post infection within BMM was similar to the fluorescence level measured 30 min after a treatment with 10 µM H2O2 (Fig. 1F). All together, these data demonstrate that the ahpC-gfp fusion is responding to H2O2 in a dose-dependent manner and represents a relevant tool to monitor oxidative stress as perceived by Salmonella under infection conditions.
The oxidative burst and involvement of Salmonella detoxifying enzymes
To investigate the contribution of the Phox in H2O2 perception by Salmonella, BMM prepared from C57BL/6 and congenic gp91 phox−/− mice were infected with a wild-type strain carrying the ahpC-gfp fusion. In contrast with our observations in BMM derived from C57BL/6 (Figs 1E, F and 2A), the ahpC-gfp expression remained at the background level during the course of infection of gp91 phox−/− macrophages (Fig. 2A). These data indicated that H2O2 perception by intracellular Salmonella is strictly dependent on the function of the host NADPH oxidase.
Next, to monitor the oxidative stress generated in the animal, C57BL/6 and gp91 phox−/− mice were infected with bacteria transformed with the pLA40 plasmid. Forty-eight hours after infection, spleens were removed, intracellular bacteria extracted, and the fluorescence monitored by flow cytometry. Whereas the GFP levels did not statistically differ between injected bacteria (LB) and bacteria recovered from gp91 phox−/− spleen mice, the fluorescence increased significantly in bacteria isolated from C57BL/6 mice (Fig. 2B). This indicates that the Phox complex also mediates the oxidative stress sensed by Salmonella in the mouse model.
Because Phox produces O2.− and intracellular bacteria sense H2O2, we predicted that periplasmic Sod would be required to convert O2.− into H2O2. Therefore, the pLA40 plasmid was transformed into a sodCI sodCII mutant, and mouse RAW 264.7 macrophages were infected by the resulting strain. Compared with the wild type, the fluorescence levels in the sodCI sodCII mutant were 1.46- and 2.09-fold lower at 90 and 180 min post infection respectively (Fig. 2C). Similar results were obtained using BMM (data not shown). These data indicated that a significant fraction of the O2.− produced by the Phox was converted into H2O2 by periplasmic Sod.
Finally, to evaluate the contribution of catalases and peroxidases to oxidative stress resistance, a mutant lacking four H2O2-degrading enzymes (katE katG katN ahpCF) was built. This strain did not present any defect in proliferation in macrophages, mainly because a fifth peroxidase is still present (Hébrard et al., 2009). This mutant was transformed with the pLA40 plasmid, and the resulting strain was used to infect RAW264.7 macrophages. Three hours post infection, the GFP level was found to be 4.1-fold higher in a katE katG katN ahpCF mutant than in a wild-type strain (Fig. 2D). These results confirmed that Salmonella catalases and peroxidase are required for degrading cytoplasmic H2O2 deriving from the Phox activity.
Alteration of the T3SS-2 does not modify the expression of the ahpC-gfp fusion
As an inhibition of the Phox assembly has been proposed to be mediated by the T3SS-2 (Vazquez-Torres et al., 2000b; Gallois et al., 2001), an increase in the expression of the ahpC-gfp fusion would be expected in a T3SS-2 deficient strain. To test this hypothesis, a ssaV mutant was transformed with the pLA40 plasmid. Such mutant was chosen because SsaV was found to be a major component of the needle-like organelle, and it was shown that the ssaV mutant lost its ability to secrete T3SS-2 effectors (Klein and Jones, 2001; Nikolaus et al., 2001). RAW 264.7 macrophages were infected, and the fluorescence of intracellular bacteria was monitored by flow cytometry after different times of infection. Fluorescence levels of the wild-type and ssaV strains were found to be identical during the infection course: they were maximal between 90 and 180 min post infection, and then decreased (Fig. 3A). Similar profiles were obtained in a T3SS-1 mutant (prgH), and in a T3SS-2 T3SS-1 mutant (ssaV prgH) (Fig. 3A), PrgH being a membrane protein of the needle complex forming the T3SS-1 (Miller, 1991). We asked whether this unexpected result could be a consequence of a rapid killing of the ssaV mutant by microbicidal activities of activated macrophages. We tested the survival of wild-type and ssaV strains in BMM prepared from C57BL/6 and gp91 phox−/− mice. Identical survival rates were recorded for both bacteria strains at 4, 6 and 8 h post infection and in both types of macrophages (Fig. S3). These results are in agreement with a recent study showing that T3SS-2 does not have a major influence on Salmonella resistance to killing (Helaine et al., 2010). Next, we measured the oxidative stress sensed by Salmonella in the animal. C57BL/6 mice were infected by various strains transformed with the pLA40 plasmid. Intracellular bacteria were extracted from spleens after 2 days and analysed by flow cytometry. We observed identical GFP levels in ssaV and ssaV prgH mutant strains, and a lower GFP level in these mutants as compared with the wild-type Salmonella strain (Fig. 3B). Taken together, these results show that alteration of the T3SSs does not increase the level of cytoplasmic H2O2 sensed by intracellular Salmonella in macrophages and mice.
Role of the T3SS-2 in oxidative stress adaptation
It has been previously shown, by using a comparative survival test, that Salmonella mutants lacking a functional T3SS-2 were able to cause lethal infection of gp91 phox−/− but not of congenic C57BL/6 mice (Vazquez-Torres et al., 2000b). To assay the role of the T3SS-2 in the control of the oxidative burst, we compared the virulence of wild-type and ssaV Salmonella strains in gp91 phox−/− and C57BL/6 mice by performing mixed infection. Mice were injected intraperitoneally with 105 cfu of a 1:1 mix of both strains. Bacteria were recovered from mouse spleens after 2 days, and the competitive index was determined (Beuzon and Holden, 2001). We found that the ssaV mutant was highly attenuated in both mouse strains. Moreover, no significant difference was found for the competitive index in C57BL/6 and gp91 phox−/− mice (0.014 ± 0.004 versus 0.019 ± 0.007, P = 0.26) (Fig. 3C). This indicated that T3SS-2 supports Salmonella replication regardless of whether respiratory burst occurs or not. Spleens of gp91 phox−/− mice contained 50- to 100-fold more of either bacterial cell than those of C57BL/6 mice (Fig. 3D). Strikingly, the ssaV load in gp91 phox−/− mice was comparable or even higher than the load of wild-type Salmonella in C57BL/6 mice.
We then compared the infectious process occurring in both mice strains by localizing Salmonella in splenic tissue and cells. Microscopic observation of immunostained sections revealed that bacteria were essentially localized in the red pulp of the spleen both in gp91 phox−/− and C57BL/6 mice (Fig. 4). However, more infectious foci containing more bacteria were observed in gp91 phox−/− mice (Fig. 4). Next, we tested by flow cytometry whether Salmonella targeted the same cell repertoires in the two mouse strains. Splenocytes were immunostained with a combination of fluorescent antibodies allowing the identification of monocytes, neutrophils, eosinophils, dendritic cells and B cells. As shown in Table 1, the cell population profiles were not significantly different in uninfected C57BL/6 and gp91 phox−/− mice, and after Salmonella infection. As previously observed (Richter-Dahlfors et al., 1997), infected spleens were characterized by an increased percentage of neutrophils. Compared with C57BL/6 mice, we found by flow cytometry an average of six times more infected splenocytes in gp91 phox−/− (Table 2) whereas their spleens contained 50- to 100-fold more bacteria (Fig. 3D). This suggests that infected splenocytes contain approximately 10 times more bacteria in gp91 phox−/− than in C57BL/6 mice and confirms our microscopic observation of the size of infectious foci (Fig. 4). In both mouse strains, we identified about 85% of infected splenocytes, indicating that Salmonella targeted identical repertoires of spleen cells. Monocytes, eosinophils and dendritic cells represented approximately 21% and 11% of infected splenocytes in C57BL/6 and gp91 phox−/− respectively. In agreement with a previous report (Geddes et al., 2007), bacteria were for the most part localized in neutrophils and B lymphocytes. However, while Salmonella targeted mainly neutrophils in C57BL/6 mice, B lymphocytes represented the main fraction of infected cells in gp91 phox−/− mice (Table 2). We observed, in both mouse strains, direct and inverted correlations between the percentages of infected splenocytes and the extent of infected B cells and neutrophils respectively (Fig. S4). This indicates that the larger proportion of infected B lymphocytes observed in gp91 phox−/− mice is a direct consequence of the higher bacterial load. Altogether, these results indicate that the bacterial localization and the profiles of infected splenocytes are similar in mice producing an oxidative burst or not, and that T3SS-2 increases Salmonella virulence in mice independently of a functional NADPH oxidase.
Table 1. Analysis of spleen cells of uninfected and infected C57BL/6 and congenic gp91 phox−/− mice.
% of spleen cells
C57BL/6 n = 3
gp91 phox−/−n = 4
C57BL/6 n = 5
gp91 phox−/−n = 9
2.4 ± 0.3
2.8 ± 1.1
5.7 ± 1.4
4.9 ± 2.1
1 ± 0.4
1.6 ± 0.4
1.6 ± 0.3
1 ± 0.3
Eosinophils and monocytes CX3CR1+
0.8 ± 0.1
0.9 ± 0.3
1.7 ± 0.5
0.78 ± 0.3
2.4 ± 0.3
2.4 ± 0.3
1.4 ± 0.1
1.5 ± 0.1
52.7 ± 3.5
45.7 ± 4.6
52.5 ± 1.7
53.6 ± 3.5
40.6 ± 4.3
46.6 ± 6.6
37.04 ± 3.6
38.3 ± 3.1
Table 2. Identification of infected spleen cells of C57BL/6 and congenic gp91 phox−/− mice.
1.1 ± 0.4 and 6.3 ± 6% splenocytes were infected in C57BL/6 and gp91phox−/− respectively. Between 3100 and 58 600 (mean = 16 800) splenocytes were analysed in C57BL/6 mice; between 1500 and 127 000 (mean = 32 600) in gp91phox−/− mice.
46.2 ± 3.5
26.6 ± 17.1
10.6 ± 2
5.1 ± 3.7
Eosinophils and monocytes CX3CR1+
9.9 ± 3.2
3.5 ± 3.7
1 ± 0.1
2.3 ± 0.9
16.1 ± 4.4
49.1 ± 19.1
16 ± 2.8
13.3 ± 8
In the present study, we made use of a transcriptional fusion to monitor the oxidative stress sensed by Salmonella during infection. We showed that Salmonella rapidly perceived oxidative burst originating from the host phagocyte NADPH oxidase in macrophages. Bacterial periplasmic Sod contributed to the dismutation of a significant fraction of the host released O2.− into H2O2, which was eventually degraded by catalases and peroxidases within the bacterial cytoplasm. We showed that wild-type and ssaV strains were exposed to identical levels of H2O2 in macrophages and during the process of mouse infection. We also observed that the T3SS-2 contributed similarly to Salmonella virulence in C57BL/6 and gp91 phox−/− mice. Taken together, these data assign a paramount importance to bacterial detoxifying enzymes in the degradation of host-released ROS, whereas the contribution of the T3SS-2 remains an open issue.
Using the ahpC-gfp fusion presents multiple advantages: its specificity to a dedicated ROS (H2O2), its sensitivity, and its location in the bacterial cytoplasm. This reporter can be used in different conditions (culture media, host cells, animal model) and can be monitored using various methods (fluorimeter, flow cytometry, quantitative microscopy). Moreover, this sensor is of particular interest in the field of oxidative stress as a balance rapidly occurs between ROS production and degradation. Previous studies took advantage of OxyR sensitivity to monitor low H2O2 concentrations, e.g. by fusing OxyR-controlled promoters to reporter genes (Deretic et al., 1995; Ochsner et al., 2000; Pulliainen et al., 2008) or by inserting a fluorescent protein into the OxyR regulatory domain to produce a sensor detecting intracellular H2O2 in eukaryotic cells (Belousov et al., 2006). The use of different genetic backgrounds, either in the host or in the pathogen, allowed us to identify the contribution of (i) host factors to oxidative stress production, and (ii) bacterial enzymes to ROS degradation. Interestingly, the ahpC-gfp reporter was not induced in gp91 phox−/− macrophages and mice, which demonstrates that the oxidative burst sensed by intracellular bacteria is exclusively produced by the phagocyte NADPH oxidase. Moreover, the lack of induction of this reporter in the gp91 phox−/− background shows that endogenous H2O2 produced by the bacterial metabolism is negligible compared with the exogenous Phox-dependent H2O2. Our data are also consistent with a previous report showing that the H2O2 inducible oxyS promoter was activated within S. typhimurium 30 minutes post infection (Schlosser-Silverman et al., 2000).
All together, these results allowed us to propose a schematic model for oxidative stress control by Salmonella (Fig. 5).
In agreement with previous studies (Korshunov and Imlay, 2006), we make the hypothesis that three pathways could arise for the elimination of periplasmic O2.−. One is mediated by periplasmic SodCI and SodCII, another is non-enzymatic or so-called autodismutation, and the third could be mediated by the reduction of an unknown target biomolecule. These three independent mechanisms would lead to the production of periplasmic H2O2. One hour post infection, we showed that the fluorescence levels of the wild-type and the mutant strains were similar (Fig. 2C), indicating that during the early stages of infection, the absence of periplasmic Sod did not reduce the amount of H2O2 that was formed. In the context of the hypothesis above, this suggests that the H2O2 perceived at this stage comes from autodismutation and biomolecule reduction. Between 1 and 6 h post infection, the fluorescence levels in the wild-type strain were found to be higher compared with the sodCI sodCII mutant (Fig. 2C). This indicates that in the sodCI sodCI mutant, lower amounts of H2O2 would be produce by autodismutation and reduction of the ‘unknown’ target biomolecule. To account for this limitation, one might postulate that the reducing power of the biomolecule could be saturated.
Next, H2O2 can efficiently move to the bacterial cytoplasm as H2O2 flux was shown to be rapid with a high permeability coefficient of the membrane (Seaver and Imlay, 2001). We have previously demonstrated that cytoplasmic catalases and peroxidases constitute a second line of defence against H2O2 (Hébrard et al., 2009). This is supported by the 4.1-fold increase of the ahpC-gfp expression upon deletion of the katE katG katN ahpCF genes (Fig. 2D). Taken together, these data strengthen the role of periplasmic Sod and cytoplasmic catalases and peroxidases in the degradation of ROS, and strongly suggest that the cytoplasm is not experiencing damaging oxidative stress during the infection process.
Surprisingly, our results failed to assign any role for the T3SSs in inhibiting the host enzymatic production of the oxidative burst. Indeed, we observed that the ahpC-gfp expression remained the same whether T3SSs were functional or not. Note that before infection, bacterial cultures were grown in minimal medium containing glycerol as a carbon source. In such medium, the SPI-2 genes were found to be highly expressed (data not shown). This observation rules out the possibility that a suboptimal level of ssaV expression was responsible for the lack of difference between the wild-type and ssaV mutant. Previously, Vazquez-Torres and coll. reported that gp91 phox−/− mice were susceptible to infection by T3SS-2 mutants while C57BL/6 mice were not (Vazquez-Torres et al., 2000b). They concluded that SPI-2 genes were not required for virulence in the absence of a phagocyte respiratory burst. Bacterial loads produced by wild-type Salmonella in C57BL/6 and by the ssaV strain in gp91 phox−/− mice were comparable (Fig. 3D), thus confirming the importance of the NADPH oxidase in the innate immune control of Salmonella infections. However, our measure of the competitive index indicated that the T3SS-2 plays the same central role in virulence, whether or not the host NADPH oxidase was functional (Fig. 3C). A more puzzling observation is the intracellular distribution of the NADPH oxidase complex, which appears excluded from the SCV membrane in a T3SS-2-dependant manner (Vazquez-Torres et al., 2000b; Gallois et al., 2001). Because the putative T3SS-2 effector, which targets the Phox, is likely to be translocated in small amounts as compared with the NADPH oxidase, one cannot exclude that a significant part of the enzymatic complex is still targeted to the SCV membrane. This would explain why wild-type and T3SS-2 mutant strains experienced indistinguishable oxidative stresses as measured by the expression of the ahpC-gfp reporter.
We reported that monocytes represented less than 20% of infected splenocytes, while Salmonella were mainly localized in neutrophils and B lymphocytes. These results are in good agreement with another flow cytometric study showing that neutrophils and B cells represent approximately 50–60 and 10% of infected spleen cells (Geddes et al., 2007). Previous confocal microscopy studies showed that Salmonella resides in CD18-positive phagocytes of the liver (Richter-Dahlfors et al., 1997) and Mac-1 (Cd11b/CD18) expressing phagocytes of the spleen (Matsui et al., 2000). CD18 is a leukocyte-specific integrin that is expressed and necessary for normal neutrophil functions (Volz, 1993). In 2001, a flow cytometric analysis of infected splenocytes found that a pool of anti-macrophage antibodies stained a large part of infected cells (Salcedo et al., 2001). This pool contained a F4/80 antibody that is also expressed by GR-1+/CD11B+ spleen cells and thus labels monocytes and neutrophils (Geddes et al., 2007). The same study reported that, by confocal microscopy, 20–30% of infected splenocytes were expressing the class A macrophage scavenger receptor (MSR-A). Based on these results, which were essentially performed by microscopy, it is commonly admitted that Salmonella resides and replicates in monocytes/macrophages. However, these studies did not search for the presence of Salmonella in neutrophils using specific markers. Moreover, in light of the knowledge for the specificity of antibodies that have been used, these results can be interpreted in a different way and are not necessarily contradictory with ours. We also observed in both mouse strains a direct correlation between the percentage of infected splenocytes and the fraction of infected B lymphocytes. Even though this phenotype was detectable in C57BL/6 mice (Fig. S4), it became more visible in gp91 phox−/− mice having a high bacterial load. These surprising results will deserve further analyses to understand how cells populations targeted by Salmonella evolve as the infection progresses. Identical repertoires of cells were infected in C57BL/6 and gp91 phox−/− mouse strains. This indicates that bacterial colonization of spleen tissues is similar whether mice express or not a functional NADPH oxidase and, importantly, validates the gp91 phox−/− mouse strain as a model to study systemic Salmonella infection.
It would now be of particular interest to understand the molecular basis of ROS-based antibacterial effects, i.e. identify the biomolecules targeted by ROS as well as their subcellular location. Periplasmic SodCI and SodCII were found to combat phagocytic O2.−, and this activity was proposed to rely on the protection of extracytoplasmic molecules (Craig and Slauch, 2009). Such targets have not yet been identified as biomolecules directly damaged by O2.− are quite limited (Craig and Slauch, 2009). It might be interesting to consider this issue in light of the assumption we made earlier, i.e. a target biomolecule could be involved in the production of periplasmic H2O2 during the early stage of the infection. Moreover, we showed that SodCI and SodCII contributed to O2.− dismutation into H2O2, but we also observed that Salmonella still sensed H2O2 in a sodCI sodCII mutant. Moreover, the ahpC-gfp expression was found to increase noticeably in the absence of catalases and peroxidases. These results indicate that H2O2 from an exogenous origin accumulates within Salmonella cytoplasm, and highlight the importance of catalases and peroxidases in oxidative stress adaptation. In their absence, proteins would be oxidized and/or damaged, and OH. would be produced, leading to DNA lesions. Potential cytoplasmic targets are various: from genomic DNA and nucleotides, to iron sulphur clusters and proteins bearing amino acids sensitive to oxidation. In the E. coli katE katG ahpC mutant, the H2O2 level produced by aerobic metabolism was found to be high enough to damage DNA (Park et al., 2005). Thus, the identification of biomolecules targeted by phagocytic superoxide represents a major challenge and would lead to a better understanding of this killing mechanism.
The phagocytic oxidative burst is a highly efficient killing mechanism against fungi and bacteria, and thus, an essential element of the innate immune response. Likewise, Listeria monocytogenes is confronted with severe oxidative stress in host cells, leading to the activation of various resistance mechanisms. Among them, the expression of kat and sod genes was found to be highly upregulated during infection (Camejo et al., 2009). In Mycobacterium tuberculosis, a recent study revealed that infection of murine macrophages with a katG mutant increased phagosomal ROS levels (Miller et al., 2010). Moreover, this mutant induced apoptosis in macrophages, assigning to KatG a role for M. tuberculosis persistence within host cells (Miller et al., 2010). In Salmonella, we have shown that five cytoplasmic enzymes are involved in H2O2 degradation (Hébrard et al., 2009). Thus, the presence of a detoxifying arsenal could be considered as a progressive adaptation shared by various intracellular pathogens to eradicate host-produced superoxide.
Bacterial strains and growth conditions
The bacterial strains used in this study are described in Table 3, and oligonucleotide sequences are listed in Table S1. Strains were routinely grown at 37°C in minimal medium (M9, glycerol 0.2%, MgSO4 1 mM, CaCl2 200 µM, thiamine 1 µg ml−1, casamino acids 1 mg ml−1). Ampicillin was added when necessary at 50 µg ml−1. Deletions of various genes and concomitant insertion of an antibiotic resistance cassette was carried out using Lambda-Red-mediated recombination (Datsenko and Wanner, 2000). Mutations were moved to 12023 wild-type strain by P22 transductions.
pFPV25 derivative carrying the ahpC-gfp promoter (Apr)
The cloning vector used was pFPV25, carrying promotorless gfpmut3a gene (Valdivia and Falkow, 1997). The insert carrying 300 bp upstream aphC start codon was PCR-amplified from S. typhimurium 12023 by using the forward primer 5′-CCCTCTAGAGTAATGTAGAGCGCAACACTT-3′ and the reverse primer 5′-CCCCATATGTACTTCCTCCGTGTTTTCGTT-3′. PCR products were digested using XbaI and NdeI, cloned into pFPV25 vector to generate the pLA40 plasmid. The insert was verified by DNA sequencing.
Bacterial infection of macrophages
RAW 264.7 macrophages were seeded at a density of 4 × 105 cells per well in 6-well tissue culture plates containing DMEM with 10% fetal calf serum (FCS) (HyClone). Bone marrow cells were isolated from femurs of 8- to 10-week-old C57BL/6 female mice and differentiated into macrophages as previously described (de Chastellier et al., 1993). Macrophages were supplemented with IFN-γ (10 U ml−1, ImmunoTools) 24 h before use. Bacteria were cultured overnight at 37°C with shaking, and opsonized in DMEM containing FCS and normal mouse serum (10%, Perbio) for 30 min. The macrophages were activated with 0.2 µM phorbol 12-myristate 13-acetate (PMA) (Sigma Aldrich) before infection. Bacteria were added to the monolayers at a m.o.i. ≈ 70:1, centrifuged at 500 g for 5 min at 4°C, and incubated for 30 min at 37°C in 5% CO2. The macrophages were washed three times, and incubated with DMEM containing FCS and 100 µg ml−1 gentamicin for 90 min, after which the gentamicin concentration was decreased to 10 µg ml−1 for the remainder of the experiment.
Flow cytometry analysis of bacteria extracted from macrophages or spleen
Infected macrophages were washed twice with PBS, lysed with 0.1% Triton X-100 in PBS and immediately fixed in two volumes of 3% paraformaldehyde for 1 h. Large debris and nuclei were removed by centrifugation for 5 min at 200 g and bacteria were pelleted at 20 000 g for 10 min. Spleens were removed from infected mice and homogenized on a 70 µm sieve in 4 ml 0.1% Tx-100. Debris were removed by centrifugation for 5 min at 200 g and bacteria were pelleted at 20 000 g for 10 min. Pellets were resuspended in 0.5 ml 3% paraformaldehyde and fixed for 1 h. Bacteria were pelleted again. Bacteria extracted from macrophages or spleen were resuspended in 0.5 ml of 10 mM NH4Cl in PBS and immunolabelled with a mouse anti-Salmonella 1E6 and a donkey anti-mouse PE, both diluted 1:1000. For flow cytometric analysis, bacteria were gated in FL2 and analysed for the expression of GFP in FL1. Data were acquired with a FACScalibur (BD Biosciences) and analysed with the FlowJo software (Tree Star Inc, Ashland, OR).
Eight- to 10-week-old C57/B6 mice were inoculated intraperitonealy with equal amounts of two bacterial strains for a total of 105 bacteria per mouse. The spleens were harvested 48 h after inoculation and homogenized. Bacteria were recovered and enumerated after plating a dilution series onto LB agar and LB agar with the appropriate antibiotics. Competitive indexes (CI) were determined for each mouse (Beuzon and Holden, 2001). The CI is defined as the ratio between the mutant and wild-type strains within the output (bacteria recovered from the mouse after infection) divided by their ratios within the input (initial inoculum). A one-sample t-test was used to determine whether the CI was significantly different. All statistical analyses were performed by using Prism (GraphPad, San Diego, CA, USA). The two-tailed P-value was calculated.
Immunofluorescence staining and confocal microscopy
Spleens were fixed with 3.2% paraformaldehyde for 1 h, washed in PBS, infused overnight in 35% sucrose, and frozen in Tissue–Tek OCT compound (Electron Microscopy Sciences, Hatfield, PA, USA). After permeabilization with 0.5% saponin for 5 min and unspecific binding site blockade with 2% bovine serum albumin, 1% FCS, and 1% donkey or goat serum for 30 min, 8 µm cryostat tissue sections was labelled overnight at 4°C with primary antibodies or control antibodies followed by incubation for 1 h at room temperature with secondary antibodies. Slides were mounted in Prolong Gold (Invitrogen) and observed with a Zeiss LSM 510 confocal microscope (Carl Zeiss, Jena,Germany).
Animal experimentation was conducted in strict accordance with good animal practice as defined by the French animal welfare bodies (Law 87–848 dated 19 October 1987 modified by Decree 2001-464 and Decree 2001-131 relative to European Convention, EEC Directive 86/609). All animal work was approved by the Direction Départementale des Services Vétérinaires des Bouches du Rhône (authorization number 13.118 to S.M.).
Flow cytometry analysis of hematopoietic spleen cells
Spleens were removed from infected and non-infected mice and cut in pieces before digestion for 20 min at room temperature with a mixture of Collagenase type 2 (Serlabo) and of DNase I (Sigma). Cell suspensions were treated for 5 min with EDTA and filtered trough cell strainers. Red blood cells were lysed with TRIS-buffered ammonium chloride. Before staining, cells were preincubated at 4°C for 10 min with the 2.4G2 antibody to block Fc receptors. Single-cell suspensions were stained for 20 min with combinations of FITC-, PE-, PercP-Cy5-5, APC-, Alexa700-, PE-Cy7-, Pacific Blue-conjugated antibodies. Data were acquired with a FACSCanto II flow cytometer (BD Biosciences) and analysed with the FlowJo software (Tree Star Inc, Ashland, OR, USA). Anti-CD11b (M1/70), and anti-CD11c (HL3), anti-Ly6C (AC21), anti-Ly6G (1A8), anti-CD4 (L3T4), anti-CD8 (53–6.7) were purchased from BD Biosciences Pharmingen. Anti-CD5 (53.7-3), anti-B220 (RA3-6B2), anti-NKp46 (29A1.4) and anti-MHCII (M5/114) were purchased from eBioscience.
Thanks are due to the members of the FB and JPG groups for fruitful discussions. This work was funded by the ANR (Programme M.I.E. ‘Salmo-sensor’, Grant ANR-08-MIEN-025-01), the CNRS, and the Université de la Méditerranée. We thank S. Altuvia (The Hebrew University, Israel) and L. Bossi (CGM, France) for the kind gifts of mutant strains. gp91 phox−/− mice were generously provided by S. Amigorena and A. Savina (Institut Curie, Paris). We thank Rebecca Stevens (GAFL, INRA, Avignon) for critical reading of the manuscript, Patrice Moreau (LCB, Marseille) for intuitive discussions, and the mouse functional genomics platform of the Marseille-Nice Genopole for immunohistochemistry with support from the INCA PROCAN program. MH was supported by the Ministère de la Recherche and the Fondation pour la Recherche Médicale (FRM). JV was funded by the Fondation pour la Recherche Médicale (FRM).