Analysis of Cells Targeted by Salmonella Type III Secretion In Vivo

The type III secretion systems (TTSS) encoded in Salmonella pathogenicity island-1 and -2 (SPI-1 and -2) are virulence factors required for specific phases of Salmonella infection in animal hosts. However, the host cell types targeted by the TTSS have not been determined. To investigate this, we have constructed translational fusions between the ß-lactamase reporter and a broad array of TTSS effectors secreted via SPI-1, SPI-2, or both. Secretion of the fusion protein to a host cell was determined by cleavage of a specific fluorescent substrate. In cultured cells, secretion of all six effectors could be observed. However, two to four days following i.p. infection of mice, only effectors secreted by SPI-2 were detected in spleen cells. The cells targeted were identified via staining with nine different cell surface markers followed by FACS analysis as well as by conventional cytological methods. The targeted cells include B and T lymphocytes, neutrophils, monocytes, and dendritic cells, but not mature macrophages. To further investigate replication in these various cell types, Salmonella derivatives were constructed that express a red fluorescent protein. Bacteria could be seen in each of the cell types above; however, most viable bacteria were present in neutrophils. We find that Salmonella is capable of targeting most phagocytic and non-phagocytic cells in the spleen but has a surprisingly high preference for neutrophils. These findings suggest that Salmonella specifically target splenic neutrophils presumably to attenuate their microbicidal functions, thereby promoting intracellular survival and replication in the mouse.


Introduction
The innate and adaptive immune systems of the host present a formidable barrier to infection. To overcome the multi-faceted defenses, microbial pathogens have evolved equally complex mechanisms that are only partially understood. One of these mechanisms is the type III secretion system (TTSS) found in many Gram-negative bacterial pathogens. These are sophisticated secretion devices that inject specific proteins (called effectors) directly into the host cell cytoplasm. Various cell culture models are used to study effectors but the cell types targeted by the Salmonella TTSS during the course of infection have not been studied.
Salmonella enterica serovar Typhimurium (referred to as S. typhimurium hereafter) has two TTSSs that are expressed under different conditions and required for distinct aspects of infection [1][2][3]. Effectors secreted by the Salmonella pathogenicity island-1 TTSS (SPI-1 TTSS) are associated with the invasion of intestinal epithelial cells and enhanced intestinal inflammation in infected hosts [4][5][6]. The Salmonella pathogenicity island-2 TTSS (SPI-2 TTSS) is required for intracellular survival during the systemic phase of infection [7][8][9][10][11], but it also enhances inflammation during the enteric phase [12,13]. In previous work, effectors could be placed into three categories; those secreted via SPI-1 TTSS only, those secreted by SPI-2 TTSS only, or those secreted by both [14,15]. Additional roles for SPI-1 and SPI-2 are still being found. For example, Lawley et al. found that components of the SPI-1 TTSS are required for persistence in a chronic infection model in 129X1/SvJ mice [16].
Whether Salmonella persists or kills its host is determined by several factors such as the route of administration, the strain of Salmonella, and the strain of mouse infected. Acute infection is usually studied in Balb/c or C57BL/6 mice because they are as sensitive to S. typhimurium infection as humans are to S. enterica serovar Typhi. In acute mouse infection Salmonella moves rapidly to the two filtering organs, the spleen and liver, and within those organs is found in macrophages, neutrophils, and dendritic cells [17][18][19][20][21][22]. Macrophages are considered the primary reservoir of Salmonella because survival within macrophages is an essential virulence mechanism [23]. However, the specific cell types targeted by SPI-1 TTSS and SPI-2 TTSS in vivo have not been identified.
In this study, mice were infected i.p. with strains of S. typhimurium expressing different effector-ß-lactamase (Bla) fusions. This reporter system allows detection of secreted effectors by detecting cleavage of coumarin cephalosporin fluorescein (CCF2-AM) [24,25]. This dye is cell permeant but is rapidly cleaved by endogenous cytoplasmic esterases resulting in a charge moiety that is concentrated within the cell. When cleaved by ß-lactamase the dye no-longer displays fluorescence resonance energy transfer (FRET) altering the emission spectrum from green to blue. Three-laser FACS analysis allowed us to simultaneously identify multiple cell surface markers and CCF2-AM fluorescence thus identifying targeted cells more accurately than was possible in a previous study with Yersinia [26]. Our results show that only the SPI-2 TTSS is active in the mouse spleen and that many cell types are infected with viable bacteria.

Effector-Bla Fusions Are Secreted into Cultured Cells
Six effector proteins were selected because they display different secretion patterns in cultured cell lines and are directly involved in virulence [4,15,16,[27][28][29][30]. These secreted proteins were classified into three groups based on their secretion patterns in cultured cells [15]: SPI-1 only, SPI-2 only, and SPI-1/2. The first group, referred to as S1, includes two proteins, both encoded within Salmonella pathogenicity island 1: SipA and SptP (although SptP is secreted at low levels through SPI-2 TTSS). The second group, S2, consists of SteC and SseJ. The last group, S1/2, includes SteA and SlrP secreted via both SPI-1 and SPI-2 TTSS.
We generated chromosomally encoded effector-Bla fusions in S. typhimurium strain 14028 to the six TTSS effectors listed above and tested them to verify the appropriate secretion patterns in cultured cells. Translational fusions to each of these effectors were expressed from their natural location on the chromosome in order to maintain their native expression patterns. In addition, each fusion was tested to ensure that the derivative retained virulence. For initial experiments, we used cultured cells and infection conditions under which only SPI-1 or SPI-2 TTSS was expressed. The six Bla fusion strains were tested for secretion following infection of HeLa cells for 2 hours and the J774 macrophage-like cell line for 10 hours using either SPI-1 or SPI-2 TTSS inducing conditions, respectively. We found that the expression of the Bla fusions had no effect on the course of infection in these cells lines and that intracellular bacteria were present at all time points investigated. Following the infection, the cells were loaded with CCF2-AM (Invitrogen). Under SPI-1 TTSS inducing conditions, microscopic analysis revealed blue cells among the HeLa cells infected with the S1 and S1/2 fusion strains indicating CCF2-AM cleavage by Bla ( Figure 1A). No blue cells were observed in HeLa cells infected with the S2 fusions or in control infections with 14028 or 14028 harboring pWKS30, a plasmid expressing wild-type ß-lactamase protein that is not secreted via TTSS [31]. When SPI-1-dependent infection of HeLa cells was extended to 8 and 20 hours, secretion of S2 effectors was detected ( Figure 1D), indicating that the SPI-2 TTSS is active within epithelial cells at these times. Using SPI-2 inducing conditions, blue cells could be seen in J774s infected with the S1/2 and S2 Bla fusions ( Figure  1B). A low level of secretion was observed in J774s infected with the SptP-Bla strain. No blue cells were observed in J774s infected with SipA-Bla or control infections. No secretion was observed from strains in which the secretion apparatus had been inactivated ( Figure 1C). These results confirmed that secretion by the six Bla fusion strains could be detected in cultured cells under the appropriate infection conditions.

Detection of Secretion Was Optimal 2 Days following Infection
To test secretion in an in vivo model, C57BL/6 mice were inoculated via i.p. infection with S. typhimurium strains expressing the effector-Bla fusions. Spleens from infected mice were then removed at 24-hour intervals, homogenized, and loaded with CCF2-AM. FACS analysis was performed to detect CCF2-AM cleavage. We were only able to detect secretion in mice when the bacterial load in the spleen was greater than ;3 3 10 7 and therefore could not detect secretion within the first 24 hours of infection. If the bacterial load exceeded 1 3 10 8 , the mice became septic resulting in overt splenic cell death making the analysis questionable. Despite this narrow window we could reproducibly recover between 3 3 10 7 and 1 3 10 8 bacteria from the spleen 2 days following infection ( Figure S1A) and therefore performed all studies using these conditions. By counting viable bacteria recovered from the spleen of infected mice, we confirmed that none of the effector-Bla fusion strains exhibited any defect in their ability to colonize and replicate within the spleen. It is unclear why no secretion could be detected at 24 hours post-infection.

Red Fluorescent Bacteria Detected within Secretion-Positive ''Blue'' Cells
To determine if cells staining blue with CCF2 also contain live Salmonella we expressed Tomato red fluorescent protein in each Bla-fusion derivative [32]. Tomato is a red fluorescent protein that has an emission spectrum that does not overlap the other fluorescent dyes used in this work. Splenic cells were scanned by eye for those that fluoresce blue and then examined for the presence of red bacteria ( Figure 2). Only cells expressing Bla-fusions to SteA, SlrP, SteC, and SseJ gave positive results. Furthermore, every blue cell identified also contained red-fluorescent bacteria (.100 cells). 3-D reconstruction confirmed that the bacteria associated with these cells were intracellular (Video S1).

Author Summary
Bacteria of the Salmonella genus are important human pathogens and a leading cause of food-borne illness. Salmonella species' ability to cause disease relies on the activities of two sophisticated molecular syringes that allow the bacteria to pump proteins into cells that they infect. The activities of these syringes have been studied extensively in cells grown under laboratory conditions and shown to be essential for the infectious process in animal models. However, the specific cells within infected organs that are targeted by these syringes have not been identified. In this work we describe the specific spleen cells targeted by Salmonella in the mouse. We find that Salmonella is capable of targeting most cell types using their molecular syringes. Quite surprisingly, we find that Salmonella mostly targets neutrophils, a cell type not thought to be associated with live Salmonella in host tissues. These findings challenge our current views of Salmonella infection and may lead to new insight for treating the disease. 72 or 96 hours post inoculation, we performed the majority of our analysis after 48 hours. Secretion was detected only in mice infected with strains harboring S1/2 and S2 effector fusions (SteA, SlrP, SteC and SseJ; Figure 3A and 3B). No secretion was detected in mice infected with SipA and SptP effector (S1) fusions. Therefore, the in vivo secretion we observed is likely due to the expression of the SPI-2 TTSS. This finding corresponds well with the virulence data, indicating that SPI-2 but not SPI-1 TTSS is required for spleen colonization and morbidity in this acute mouse infection model ( Figure S1B and S1C, Protocol S1). In this same FACS analysis it was not possible to detect internalized bacteria expressing Tomato because the appropriate excitation wavelength was not available. However, bacteria could be detected microscopically following cell sorting of blue fluorescent cells (see below).

Identification of Spleen Cells Containing ß-Lactamase Reporter Fusions
To identify the cell types that were targeted by S. typhimurium, spleen cells were loaded with CCF2-AM, then stained with several fluorescently conjugated antibodies and S. typhimurium 14028 strains expressing Bla fusions to SipA, SptP, SlrP, SteA, SteC, and SseJ; WT 14028, or 14028 harboring pWKS30 were used to infect HeLa cells for 2 hours using SPI-1-inducing conditions and J774s for 10 hours using SPI-2-inducing conditions. Following the infections, cells were loaded with CCF2-AM and visualized for green and blue fluorescence by microscopy. Green fluorescence indicates CCF2-AM was loaded and the presence of blue cells is evidence of secretion. SPI-1 TTSS-and SPI-2 TTSS-dependent secretion detected by FACS analysis (C). HeLa cells and J774s were infected as described above with the six effector-Bla fusions that were expressed in WT 14028, invA::cat and ssaK::cat backgrounds, mutations in structural components of SPI-1 and SPI-2 TTSS, respectively. FACS analysis was performed on CCF2-AM loaded cells to determine the percentage of blue cells (positive for secretion). Background percentage of blue cells was set using WT 14028-infected cells. At least 10,000 cells were analyzed for each sample. Each bar represents the mean percentage of blue cells from triplicate samples and the error bars are 6 one standard error of the mean. FACS analysis was also performed on HeLa cells infected with each Bla fusion strain for 1, 8, or 20 hours using SPI-1 inducing conditions (D). doi:10.1371/journal.ppat.0030196.g001 analyzed by FACS. Beta-lactamase fusions to SipA or SptP were not analyzed because secretion of these effectors could not be detected in splenic cells ( Figure 3). Using FACS analysis, we were able to simultaneously detect four cellsurface markers, as well as blue and green CCF2-AM fluorescence. Each sample was split into two groups and stained with a total of eight fluorescent antibodies. One group was stained with antibodies that recognize cell surface markers specific for T cells and B cells (CD4, CD8, CD3, and CD19) and the second group was stained with antibodies to distinguish macrophages, neutrophils, monocytes and dendritic cells (CD11b, CD11c, GR-1, and F4/80). Parallel samples were stained with propidium iodide to stain dead cells and determine their light scattering properties enabling exclusion of . 90% of dead cells from our analysis. Dead cells do not load with CCF2-AM because retention of this dye requires active cell processes ( [25]; Figure S2). Therefore the contribution of autofluorescence from dead cells was minimized. Table 1 summarizes the percentage of each spleen cell type that was observed in infected and uninfected mice as well as the marker designations used to define each cell type. The percentage of lymphocytes, mature macrophages, monocytes and neutrophils in infected mice, as determined by FACS, corresponded well with the percentages observed by analyzing Wright-Giemsa stained slides. Monocytes found in tissues are considered macrophages. However, we distinguish mature macrophages from monocytes by the presence of tingible bodies on Wright-Giemsa stains as well as by the presence of cell surface markers. Monocytes express intermediate to high levels of both GR-1 and CD11b on their surface whereas mature macrophages do not (our unpublished observation; [33][34][35]). Table 2 shows the percentage of blue spleen cells represented by specific cell types from mice infected with effector-Bla fusion expressing Salmonella strains. There was no statistically significant difference in the cell types targeted by each effector. Secretion could be detected in all cell populations analyzed including CD4 T cells, CD8 T cells, B cells, dendritic cells, monocytes and neutrophils but not mature macrophages. Most of the secretion (;73% of blue cells) was found in GR-1þ/CD11bþ monocytes and neutrophils. Thus, there were several surprising results that warranted additional investigation.
The absence of mature macrophages as targets of type III secretion was puzzling hence we carried out additional controls to verify this conclusion. Spleen cell suspensions were infected with the effector-Bla fusion strains in vitro for 1 hour using SPI-1 TTSS inducing conditions, loaded with CCF2-AM and analyzed by FACS. This control confirmed CCF2-AM cleavage could be detected in all cell types including mature macrophages ( Figure S3).
These results differ somewhat from previous work based on distinguishing infected cell types with F4/80 [18]. We found that F4/80 is expressed on most GR-1þ/CD11bþ cells regardless of CD11c expression during infection ( Figure 4B). This indicates that F4/80 cannot distinguish splenic monocytes from neutrophils. However, in uninfected mice, the majority of the spleen cells expressing F4/80 are indeed monocytes and macrophages (Table 1 and Figure 4A). CD115 is also used to distinguish monocytes in uninfected animals. However following infection we detected CD115 on both neutrophils and monocytes ( Figure S4). Therefore it appears that infection itself alters the expression pattern of many cell surface proteins including those that are most often used to distinguish monocytes, macrophages and neutrophils.

GR-1þ/CD11bþ Cells Contain Viable Bacteria
The GR-1þ/CD11bþ splenic neutrophils and monocytes increased dramatically following infection (Table 1, Figure  4A). These are infiltrating professional phagocytic cells that play a crucial role in innate immunity by directly killing the invading microbe. We wished to determine whether secretion to these cells correlated with the presence of viable intracellular bacteria or whether the cleaved CCF2-AM was by residual ß-lactamase, left over after the Salmonella bacteria were killed. GR-1þ/CD11bþ cells were FACS sorted, lysed, and plated on LB agar to determine the number of CFU present as compared to the number of CFU from total splenic cells. GR-1þ/CD11bþ cells consistently contained most of the total intracellular CFU recovered from the spleen of infected mice ( Figure 4D, ;90% on average). Separation of the cells into neutrophils and monocytes revealed that ;70% of the intracellular CFUs were associated with neutrophil enriched population while ;20% were associated with a monocyteenriched population. Thus, in spite of the fact that neutrophils are very microbicidal, most bacteria were isolated from these cells. We also wished to distinguish between Salmonella actively growing within cells versus simply not yet killed. One way to do this is to examine the cells in which cleavage of CCF2 has taken place for the presence of multiple intact bacteria, as described in the next section.

FACS Sorted Blue Cells Contain Intracellular Salmonella
To determine if cells in which CCF2 had been cleaved contained intracellular bacteria we simultaneously expressed SteA-Bla and Tomato fluorescent protein in the same Salmonella derivative. Following infection of C57BL/6 mice, splenic cells were treated with CCF2, and blue fluorescent cells were isolated via FACS sorting. We confirmed by Wright-Giemsa staining that the blue cells were mostly neutrophils and that the remaining cells were a mixture of monocytes and lymphocytes. A fraction of the sorted cells were examined for the presence of intracellular bacteria by fluorescence microscopy. In addition, samples were stained with the nuclear stain, DRAQ5 to aid in distinguishing cell type. Every secretion positive cell that we visualized contained redfluorescent bacteria that appeared to be intact ( Figure 5 and supplemental videos). Most of these cells had nuclei with a multi-lobed appearance suggesting that they were neutrophils (see Videos S2-S5 for visualization of nuclei). Nearly all neutrophils contained multiple bacteria most likely resulting from intracellular replication. The number of infected cells was always ,2% of total splenocytes. The low ratio of bacteria to phagocytes indicates a low probability that the same cell could encounter multiple bacteria. Therefore it is unlikely that the presence of multiple bacteria within neutrophils resulted from sequential phagocytic events. Thus, the data suggests that Salmonella is replicating within GR-1þ/ CD11bþ neutrophils.

Salmonella Internalization by B and T Cells Is SPI-1 Independent
In tissue culture experiments, SPI-1 TTSS is required for cell invasion in non-professional phagocytic cells. We there-fore wished to determine if SPI-1 TTSS were required for internalization by B and T cells and whether SPI-2 TTSS dependent secretion in these cells is associated with intracellular Salmonella. To test if uptake by B and T cells occurred via SPI-1 mediated cell invasion the same FACS analysis as above was repeated but in a strain containing a mutation in the SPI-1 structural apparatus (invA::cat). Mice were infected with this strain or the wild-type control containing both SteA-Bla and Tomato fluorescent protein. Two days after infection cells expressing ß-lactamase and CD19þ (B cells) or CD3þ (T cells) were sorted, stained with DRAQ5, and analyzed by fluorescence microscopy. As expected, sorted cells had nuclear morphology typical of lymphocytes. These cells contained red-fluorescent bacteria for both the SPI-1 mutation and the control ( Figure 6). FACS analysis also revealed that secretion into T and B cells occurred at the same frequency in the absence of invA (Table S1). These results indicate that Salmonella infects non-phagocytic T and B cells in a SPI-1 TTSS independent manner.

Discussion
This is the first report detailing the cell types targeted by Salmonella's TTSS during a course of infection. Within the spleen a variety of cell types including neutrophils, monocytes, and B and T lymphocytes all showed ß-lactamase cleavage of CCF2-AM demonstrating secretion. Not only were the cells targeted by type III secretion but intact bacteria were also observed in each of these cells. The presence of viable replicating bacteria within monocytes is expected but the presence within neutrophils suggests that Salmonella must be defending itself successfully against microbicidal factors present in these cells.
Salmonella encodes two different type III secretion systems that play distinct roles by secreting different virulence factors at specific steps during infection. In this work six different secreted effectors were chosen because their secretion patterns were diverse. Two were secreted by SPI-1 TTSS, two secreted by SPI-2 TTSS and two secreted by either apparatus. Although SPI-1 plays a role in a chronic mouse   infection [16] it had no effect in our acute infection model. Only SPI-2 TTSS dependent secretion was observed in the mouse spleen after 48 hours of infection. However, we cannot rule out the possibility that SPI-1 TTSS-dependent secretion occurs below our limits of detection, at later time points of infection, in different organs, or into the extracellular space. In a recent report the cell types targeted by Yersinia type III secretion were examined with findings that differed from those reported here. These investigators found variations in the cellular targets for secretion of different effectors [26]. In our studies there was no statistical difference in secretion pattern for any of the effectors examined.

Salmonella Targets Non-Phagocytic Lymphocytes
We were surprised to detect secretion of SteA-Bla to splenic B (CD3-and CD19þ) and T cells (CD3þ and CD4þ or CD8þ). This is not a consequence of auto-fluorescence for several reasons. No blue CCF2-AM fluorescence was observed in B and T cells from mice infected with wild type Salmonella (not expressing a Bla fusion). Additionally, dead cells that can exhibit auto-fluorescence were eliminated from analysis using propidium iodide staining to set live cell gates and furthermore live cells were selected by their ability to retain CCF2-AM. Finally, intact Salmonella were observed within purified B and T cells. As these are not professional phagocytic cells our first hypothesis was that SPI-1 secreted effectors were stimulating engulfment of the bacteria. However, the percentage of each targeted cell type was unchanged when FACS analysis was carried out with a SPI-1 mutant suggesting that this was not true. Furthermore, using competitive index as a sensitive way of gauging virulence, there was no statistical difference in colonization between the wild type Salmonella and a SPI-1 mutant ( Figure S1C). By targeting B and T cells Salmonella may influence the development of an adaptive immune response. In fact, infection of B and T cells may account for the course of infection of typhoid fever in man. In typhoid fever, Salmonella spreads from primary sites of infection to the bone marrow as well as to other sites [36]. B cells that have been primed by exposure to antigen return to the bone marrow to become blast cells and produce antibody but may be carrying Salmonella to the bone marrow as cargo. This may explain why Salmonella is a common cause of osteomyelitis.

Salmonella Is Not Found in Mature Splenic Macrophages
Salmonella was observed within splenic monocytes but not mature macrophages 48 hours following i.p. infection. Monocytes are immature macrophages that are morphologically distinct from mature macrophages based on Wright-Giemsa staining because they lack ingested material that form tingible bodies typical of mature histiocytes. In addition, monocytes express intermediate to high levels of both GR-1 and CD11b whereas macrophages do not (our unpublished observation; [33][34][35]). Macrophages were detected among the splenic cells in our studies and could be infected once the cells were isolated ( Figure S3). Thus the result is puzzling given the fact that macrophages are professional phagocytes and considered one of the hallmarks of the innate immune response to infection. One possibility is that monocytes are more phagocytic than mature macrophages. In fact a previous study concluded that Salmonella targets a specific subset of highly phagocytic macrophages [18]. Another possible explanation is that macrophages are being killed by Salmonella shortly after infection. This explanation would also explain the decrease of mature macrophages in the spleen by day 2 of infection (Table 1). Interestingly, Rydstrom and Wick found that splenic monocytes kill Salmonella more efficiently than neutrophils [22]. Thus the lack of Salmonella in mature macrophages may result from killing by the precursor cell. However this is difficult to reconcile with the fact that Salmonella resides in macrophages at later times of infection [21] and that the rate of Salmonella killing in vivo is very low [37,38]. Richter-Dalfors et al found that Salmonella resides almost exclusively in liver macrophages 5 days post oral inoculation [21] thus it is possible that mature macrophages may simply leave the spleen for other sites.

Expression of Cell Surface Proteins Changes on Infection
Pathogen associated molecular patterns are detected by the host's innate immune system and result in an immediate response to the infecting organism. The consequence of this is up-regulation of cytokines, chemokines, their receptors, and adhesion molecules. Classically, specific cell surface proteins have been used to distinguish cell types. The drawback of relying on the same cell surface proteins in uninfected and infected hosts is evident in Table 1 and from our other results. This study has the advantage of using two different approaches to distinguish cell types; by fluorescent antibody staining to a variety of cell surface proteins followed by FACS analysis and verification by traditional cytological differentiation after cell sorting. From these two approaches it is clear that cells express a different spectrum of surface proteins after infection than in an uninfected mouse including those that are commonly used to distinguish cell type. As one example, cell markers that were thought to be unique to specific classes of cells such as F4/80 (for macrophages) were found on CD11bþ/GR-1þ neutrophils and monocytes.

Salmonella Survives and Replicates within Neutrophils
The current model for Salmonella systemic infection is that Salmonella reside and replicate within macrophages. However, Dunlap et al. found that most Salmonella are within neutrophils 24 hours after i.v. inoculation [20]. Rydstrom and Wick also found Salmonella within neutrophils in the Peyer's patches and mesenteric lymph nodes 4 days post oral inoculation [22]. Yet the role of neutrophils in controlling Salmonella infection is still controversial. One approach to assess their role in vivo is to deplete neutrophils using specific antibodies and then compare course of infection to an untreated mouse [39][40][41][42]. However, depletion studies based on anti-GR-1 deplete not only neutrophils but also monocytes, consequently it is difficult to assess the role of neutrophils [22,35,43]. In vitro, neutrophils kill Salmonella only in the first two hours followed by intracellular bacterial replication [44][45][46] precisely the same pattern as observed for primary macrophages [47][48][49][50].
Neutrophils are thought to be a dead end for an intracellular pathogen because they are short-lived and highly microbicidal. They kill bacteria by both oxidative mechanisms (NADPH oxidase and inducible nitric oxide synthetase) and non-oxidative mechanisms (defensins to depolarize the bacterial membrane, neutrophil elastase to destroy components of the type III secretion system, and a protein net to capture and destroy bacteria) [51][52][53][54]. However, our data indicates that Salmonella not only resides within neutrophils it also replicates within these cells. Our findings challenge the current paradigm that Salmonella evades neutrophil killing in vivo by hiding within macrophages. Rather Salmonella actively targets neutrophils with type III secretion and subverts them to promote an intracellular niche.

Materials and Methods
Bacterial strains and cell culture. ATCC S. typhimurium 14028 was the parental strain used in all assays. LB media supplemented with antibiotics was used unless otherwise stated. Antibiotics were used at the following concentrations: kanamycin 60 lg/ml, carbenicillin 100 lg/ml. 14028 derivatives and plasmids used in this study are listed in Table S2. J774s and HeLa cells were obtained from ATCC and grown in DMEM supplemented with 10% FBS. HeLa cells and J774s were infected using SPI-1 TTSS and SPI-2 TTSS conditions, respectively, as previously described [15]. For SPI-1 inducing conditions log phase cultures were prepared by sub-culturing overnight cultures 1:33 then growing for three hours with aeration. These cultures were then washed and diluted before infecting HeLa cells at a MOI of 50. For SPI-2 inducing conditions, overnight cultures were washed with PBS then diluted and added to J774s at an MOI of 250. Gentamicin was added to culture medium as previously described [15] to kill extracellular bacteria. To prepare samples for FACS analysis, infected HeLa cells were trypsonized, and infected J774 cells were scraped to prepare cell suspensions, then 10 7 cells/ml were loaded with .125X solution of CCF2-AM (Invitrogen) for 2 hours at room temperature following the manufacturer's recommendations. To prepare samples of J774 or HeLa cells for microscopy, cells were seeded in Lab-Tek II chamber coverglass slides (Nalge Nunc International), infected, and then 10 7 adherent cells were loaded with 1 ml of 1X CCF2-AM solution for 2 hours at room temperature.
Generation of effector-Bla fusions and protein expression analysis. The mini-Tn5-bla transposon was created by replacing the cyaA' gene in mini-Tn5-cycler [15] with codons 70-792 of the ß-lactamase gene from pBluescriptSK. This transposon was then used as a template for PCR using primers with extensions specific for the target effector. PCR products were then used to transform 14028 using the k-red PCR based recombination technique [55] (See Table S3 for a list of primers used). Constructs were confirmed by PCR and by sequencing the PCR products. Each fusion was then transduced into WT 14028, MJW1301 (ssak::cat) and MJW1835 (invA::cat) backgrounds using P22 phage as previously described [56].
Preparation of spleen cells. All mouse studies were approved by the Oregon Health and Science University institutional animal care and use committee (animal protocol #A085). C57BL/6 mice were inoculated via i.p. injection with ;2 3 10 5 CFU of the desired S. typhimurium strain. For microscopy, mice were infected with 14028 derivatives carrying plasmid pWKS30-Tomato. Tomato fluorescent protein was obtained from the Tsien laboratory [32]. pWKS30-Tomato was generated by cloning a Shine-Dalgarno sequence and the first half of the tdTomato fluorescent protein into pWKS30. After 2 days of infection, spleens were removed and homogenized using sterile glass slides. The spleen cell suspensions were then digested with collagenase 0.5 mg/ml and DNAse 50 lg/ml in RPMI at 37 8C for 25 minutes. Red blood cells in the spleen cell suspensions were lysed by incubation in a hypotonic red blood cell lysis buffer on ice for 5 minutes. Debris and clumps were removed by straining though a 70lm cell strainer (BD Falcon). The procedure described above for cultured cells was used to load CCF2-AM. To prepare samples for microscopy, FACS sorted spleen cells were stained with DRAQ5 (Alexis Biochemicals) at a 1:1000 dilution then seeded in Lab-Tek II chamber coverglass slides (Nalge Nunc International).
FACS analysis and sorting. To prepare samples for FACS cells loaded with CCF2-AM as described above were first treated with Fc receptor blocking antibodies (see Table S4 for a list of anti-bodies used) for 15 minutes at 4 8C, followed by staining with fluorescently conjugated antibodies at 4 8C for 15 minutes. FACS analysis was performed using an LSRII (Becton Dickinson) FACS analysis machine equipped with 488, 633, and 405-nm lasers. Four cell-surface markers could be simultaneously analyzed along with CCF2-AM fluorescence. Cells were simultaneously stained for four lymphoid specific markers or four myeloid specific markers. The lymphoid specific markers that were analyzed are: CD3, CD4, CD8, and CD19. Myeloid specific markers were: CD11c, CD11b, GR-1, and F4/80 (Table S4). Appropriate isotype control antibodies were used to determine the levels of background staining. Parallel samples were stained with propidium iodide to determine the settings for a live cell gate based on light scatter properties. FlowJo (Tree Star) software was used to analyze the FACS data.
FACS sorting was performed using a FACSVantage SE (Becton Dickinson), with the Digital Vantage option, using the 633, 488 nm, and 405 nm lasers. Spleen cells from infected mice were prepared as described above. Then the cells were stained with antibodies specific myeloid cell surface antigens (CD11c, CD11b, GR-1, and F4/80) or loaded with CCF2-AM and stained with antibodies specific for B and T cell antigens (CD19, CD3). Cells of the desired populations, as described in the results section, were sorted into RPMI media containing gentamicin 100 lg/ml. Reanalysis of sorted cells determined that .99% purity was achieved.
Cell pathology. Spleen samples from infected mice were prepared for Wright-Giemsa staining as follows. 1 ml of RPMI containing ;1 3 10 4 -5 3 10 4 cells were cytospun onto slides using a Cytospin 2 (Shandon) at 1,500 RPM for 5 minutes. After air-dry fixing, samples were processed for Wright-Giemsa staining [57,58]. Samples were visualized by microscopy and scored for differentials, counting at least 300 cells per sample. Microscopy pictures of Wright-Giemsa stained slides were captured at 1003 magnification using a Microphot-FX (Nikon) microscope equipped with a Magnafire (Optronics) camera and using Magnafire (Optronics) software.
Microscopy on infected cells. Cells prepared as described above were visualized using 603 or 403 oil-immersion lenses along with emission filter sets for blue (457 nm) and green (528 nm) fluorescence by CCF2-AM, Tomato fluorescent protein (617 nm), and DRAQ5 (685 nm). A UV laser with a DAPI excitation filter was used for CCF2-AM excitation, a 568 nm laser was used for Tomato excitation, and a 647nm laser was used for DRAQ5 excitation. z-sections (0.2 lm) were captured at a resolution of 1,024 by 1,024 pixels. Images were acquired by Aurelie Snyder of the OHSU-MMI Research Core Facility (http://www.ohsu.edu/core/) with an Applied Precision DeltaVision image restoration system. This includes an API chassis with a precision motorized XYZ stage, a Nikon TE200 inverted fluorescence microscope with standard filter sets, halogen illumination with an API light homogenizer, a CH350L camera (500 kHz, 12-bit, 2 Mp, KAF 1400 GL, 1,317 3 1,035, liquid cooled), and DeltaVision software. Deconvolution using the iterative constrained algorithm of Sedat and Agard, and additional image processing were performed using Softworx Explorer Suite (Applied Precision) image processing software.
Determining percentage of intracellular bacteria in various cell populations. To determine intracellular CFUs, non-specifically sorted spleen cells (viable cell gate based on light scatter properties) or specifically sorted spleen cell populations were incubated in RPMI containing gentamicin 100 lg/ml ;2 hours. The cells were then lysed in PBS with 1% Triton X-100 (Sigma) and serial dilutions were plated on LB agar to determine the number of CFU of S. typhimurium present per 10 5 host cells. To determine the percentage of intracellular bacteria present in different cell populations, the following formula was used: [(CFU/10 5 specifically sorted spleen cells) 3 F1] / [(CFU/10 5 nonspecifically sorted spleen cells) 3 F2] 3 100. Where F1 and F2 are the fractions that a specifically sorted cell population or a nonspecifically sorted cell population, respectively, represent out of the total spleen cell population. For example, if the CD11bþ/GR-1 Hi/ CD11c Lo (neutrophils) cell population represented 6% (F1 ¼ 0.06) of the total spleen cells and contained ;40 CFU / 10 5 cells and if in the same mouse, the non-specifically sorted spleen cells contain ;10 CFU / 10 5 cells and represent 33.4% (F2 ¼ 0.334) of the total spleen cells. Then using the formula above we would calculate: (40 3 0.06) / (10 3 0.334) 3 100 ¼ 72%. Figure S1. Analysis of Mice Infected i.p. with an Inoculum of 2 3 10 5 The number of bacteria present in the spleen at 24-h intervals following S. typhimurium 14028 infection was determined using at least three mice per time point (A). At 72 hours only nine of 14 mice survived and at 96 h only three out of seven mice survived. The graph shows the log of the number of bacteria present 6 one standard error of the mean. Mouse survival assays comparing WT 14028, invA::cat and ssaK::cat were performed using five mice per group. (B). The percentage of mice surviving at 24-h intervals following inoculation is shown. 4/5 mice infected with ssaK::cat survived for 28 days (not shown). Competitive infections were performed mixing WT 14028, invA::cat or ssaK::cat with MA6054 at a 1:1 ratio and by mixing ssaK::cat with invA::cat expressed in the same background as MA6054 at a 1:1 ratio (C). 2 days following infection, the number of both strains of bacteria present in the spleen was determined and used to calculate the competitive index (CI) (see Protocol S1). The graph shows the mean CI from five mice 6 one standard error of the mean. The * indicates that the Student's t-test returned a value of p , 0.0001 when compared to the control infection.