Host Transmission of Salmonella enterica Serovar Typhimurium Is Controlled by Virulence Factors and Indigenous Intestinal Microbiota

Bacterial strains and growth.The S. enterica serovar Typhimurium strains used in this study were derived from wild-type strain SL1344 (51). The isogenic, nonpolar sipB::Cm (TL156) (29), ssaV::Kan (TL355) (20), hha::Kan::lux (SMB500) (7), and hha::Cm::lux (CK162) (7) mutants were described previously. For competitive index experiments, single isolates of Salmonella serovar Typhimurium were collected from feces and P22 phage transduced to create the indicated mutants. For all infections, strains were grown at 37°C in Luria-Bertani (LB) broth or on LB agar plates containing streptomycin (200 μg/ml), chloramphenicol (10 μg/ml), and kanamycin (40 μg/ml). For mouse inoculation, an overnight culture of bacteria was diluted into phosphate-buffered saline (PBS) to obtain the desired density.

Animal infections.Female 129X1/SvJ mice (Jackson Laboratories, Bar Harbor, ME) were 7 to 9 weeks old at the beginning of each experiment. Orogastric inoculation was performed as described previously using a dose of 10⁸ CFU in 200 μl of PBS (38). Competitive index experiments were performed as previously described (29). Bedding, food (ProLab 3000 RMH; Purina Mills, Inc., St. Louis, MO), and water were changed every 7 days, and access to food and water was unlimited. Transmission experiments were performed by infecting mice and waiting 5 days before these mice were added to cages containing uninfected mice. At the indicated times, mice were treated with an antibiotic by instilling 5 mg of streptomycin sulfate (MP Biomedicals, Aurora, OH) or neomycin sulfate (Sigma, St. Louis, MO) dissolved in 200 μl H₂O via oral gavage. All animal experiments were performed in accordance with the recommendations of the Stanford University Institutional Animal Care and Use Committee.

The methods used for determination of colonization levels and serum collection have been described previously (38). Intestinal washing was performed by removing the entire small intestine and injecting 5 ml of sterile PBS containing Complete protease inhibitor cocktail (2 tablets/100 ml; Roche Diagnostics, Mannheim, Germany) into the proximal end (33). The wash liquid was collected in a sterile petri dish before transfer to a 15-ml Falcon tube and stored on ice. The intestinal contents were separated from the immunoglobulin A (IgA) fraction by centrifugation for 30 min at 3,000 rpm at 4°C. The intestinal wash preparations were stored at −80°C.

Immunosuppression.Dexamethasone was used to immunosuppress mice and allow reactivation of Salmonella serovar Typhimurium as previously described (48). At the indicated time, dexamethasone (dexamethasone 21-phosphate disodium salt; Sigma) was dissolved at a concentration of 5 mg/liter in the drinking water. Mice drink on average 3 ml of water per day (4), and therefore it was estimated that each mouse ingested 0.015 mg of dexamethasone per day.

Feces collection and culture.To collect fresh fecal pellets, individual mice were placed into isolation containers until four to six pellets were excreted. The mice were previously marked so individual mice could be followed for the duration of the experiment. For CFU determination, pellets were collected and weighed. Fecal pellets were homogenized in 1 ml of sterile PBS, and serial dilutions were plated on LB agar containing streptomycin. For fecal IgA enzyme-linked immunosorbent assays (ELISAs), fecal pellets were weighed and PBS containing Complete proteinase inhibitor cocktail (2 tablets/100 ml; Roche Diagnostics, Mannheim, Germany) was added (1 ml per 100 mg of feces). Samples were homogenized and cleared by centrifugation for 30 min at 3,000 rpm at 4°C. The supernatant was collected and stored at −80°C. Feces from the cecum (cecal contents) and colon (colon contents) were harvested by gently squeezing the luminal contents from the tissue into a sterile collection tube. For bacterial community analysis, fecal pellets were weighed and stored at −80°C prior to DNA extraction.

IgA and IgG titer determination.The serum IgG response to Salmonella serovar Typhimurium was determined by an ELISA using sonicated-killed Salmonella serovar Typhimurium as the coating antigen as previously described (38). The titer was defined as the highest dilution of serum that gave an optical density reading above the background value. The intestinal IgA response to Salmonella serovar Typhimurium was determined using the same ELISA method except that the titer was expressed as an absolute optical density value.

Ex vivo gentamicin protection assay.Two-centimeter segments of proximal colon just distal to the cecum were aseptically removed and longitudinally cut in half with a razor blade. Luminal contents were separated from the colon by gently scraping the mucosa to remove the luminal contents. The luminal contents and colon were separately incubated at 37°C in Dulbecco's modified Eagle's medium or Dulbecco's modified Eagle's medium containing 100 μg/μl gentamicin for 90 min. After the incubation period, samples were washed three times in sterile PBS, homogenized in 1 ml of PBS, serially diluted, and then plated on LB medium plates containing streptomycin.

Histology and immunohistochemistry of intestinal sections.Tissues were prepared for histology and immunohistochemistry as previously described (38). For immunohistochemistry, frozen sections were incubated with anti-Salmonella serovar Typhimurium polyclonal rabbit antiserum (diluted 1:1,000) in combination with monoclonal antibodies against CD11c raised in a hamster and biotinylated (Pharmingen 553800; 1:250) and Ly6G (Gr-1) raised in a rat (catalog no. 14-5931-81; ebioscience) in PBS containing 3% bovine serum albumin and 0.2% saponin. The secondary antibodies anti-rabbit Alexa488 (Molecular Probes), anti-hamster Alexa594-streptavidin (Molecular Probes), and anti-rat IgG-Alexa594 were diluted 1:250. ToTo-3 (diluted 1:2,000; Molecular Probes) was used to stain host cell nuclei and bacterial nuclei, and Phalloidin 647 (diluted 1:50; Molecular Probes) was used to stain actin. Coverslips were mounted over antiquench (Vector Laboratories) and sealed.

DNA extraction and Q-PCR.DNA was extracted from mouse cecal contents, colonic contents, and fecal pellets using a QIAamp stool DNA mini kit (Qiagen, Valencia, CA). Quantitative real-time PCR (Q-PCR) was performed using universal bacterial forward primer 8FM (5′-AGAGTTTGATCMTGGCTCAG-3′) (adapted from the primer described in references 26 and 27), reverse primer Bact515R (5′-TTACCGCGGCKGCTGGCAC-3′) (adapted from the primer described in references 26 and 27), and TaqMan probe Bact338K (5′-6-carboxyfluorescein-CCAKACTCCTACGGGAGGCAGCAG-6-carboxytetramethylrhodamine-3′) (adapted from the primer described in reference 1) to determine the total number of bacterial rRNA gene copies. Salmonella serovar Typhimurium-specific Q-PCR was performed using S. enterica-specific forward primer ttr-6F (5′CTCACCAGGAGATTACAACATGG-3′) (34), reverse primer ttr-4R (5′-AGCTCAGACCAAAAGTGACCATC-3′) (34), and TaqMan probe ttr-5 (5′-6-carboxyfluorescein-CACCGACGGCGAGACCGACTTT-6-carboxytetramethylrhodamine-3′) (adapted from the primer described in reference 34). The reaction conditions used for Q-PCR have been previously described (5). Reactions were carried out with an ABI Prism 7900HT sequence detection system (ABI). Tenfold serial dilutions of plasmids containing the 16S rRNA gene of Escherichia coli or the ttr gene of Salmonella serovar Typhimurium served as standards.

Experimental model to study transmission of Salmonella among mammalian hosts.Oral challenge of 129X1/SvJ (Nramp1^+/+) mice with Salmonella serovar Typhimurium can be accomplished both by oral gavage and by ingestion of infected bread. The 50% infective dose (ID₅₀) of Salmonella serovar Typhimurium strain SL1344 via oral gavage is 2 × 10⁶ CFU, while the ID₅₀ for ingestion is 5 × 10⁵ CFU as determined by using the Reed-Muench formulation (46). For the experiments reported here we used oral gavage to ensure a uniform inoculum for each animal. When either infection route is used, mice challenged with Salmonella serovar Typhimurium develop a subacute infection that lasts at least 30 to 40 days postinfection. The infected mice shed bacteria in their feces consistently during this time (Fig. 1a). Subsequently, serum antibodies specific to Salmonella serovar Typhimurium appear, and concomitantly the level of bacteria excreted is reduced. Nevertheless, in many animals persistent infection can last for up to 12 months and sporadic bacterial excretion is common during this time (38).

• Open in new tab
• Download powerpoint

FIG. 1.

Transmission of Salmonella serovar Typhimurium between hosts results in a systemic infection. Four orogastrically infected donor mice shedding Salmonella serovar Typhimurium were housed with one uninfected mouse. (a) Fecal shedding was monitored every 1 to 3 days by culturing fresh fecal pellets from the previously uninfected mouse (asterisks) and donor mice (gray circles) for Salmonella serovar Typhimurium. (b) Feces from the newly infected mouse (asterisks) and orogastrically infected mice (gray circles) were monitored for the induction of Salmonella serovar Typhimurium-specific IgA during infection. (c) Levels of Salmonella serovar Typhimurium within the cecum, Peyer's patches (PP), mLN, spleen, and liver were determined for the originally uninfected mouse (asterisks) and the orogastrically infected mice (gray circles). The detection limit for each tissue is indicated by the dashed line. (d) Titers of Salmonella serovar Typhimurium-specific IgG in the serum of mice, showing comparable levels in orogastrically (gray bars) and transmission-infected (black bar) mice.

We investigated whether this Salmonella infection model could be adapted to study host-to-host transmission of bacteria. Four infected donor mice (5 days postinfection) were placed in a cage with a single uninfected mouse for 28 days (Fig. 1). The previously uninfected recipient mouse began shedding Salmonella serovar Typhimurium within 1 to 2 days after exposure (Fig. 1a). In all instances that we tested (Table 1), the levels of Salmonella serovar Typhimurium shedding and the anti-Salmonella serovar Typhimurium fecal IgA levels for exposed, infected recipient mice were essentially identical to the levels for the donor mice infected with laboratory-grown cultures administered by oral gavage or ingestion (Fig. 1a and b). The recipient mice were also consistently colonized at mucosal and systemic sites at levels comparable to the levels in donor mice at 28 days postinfection (Fig. 1c). We also observed that the levels of Salmonella serovar Typhimurium excreted in the feces correlated with the levels of bacteria in the large intestine but not with the levels in the small intestine (data not shown). Additionally, Salmonella serovar Typhimurium-specific IgG antibodies were found in the serum of recipient mice at levels comparable to those in infected donor mice (Fig. 1d). Thus, Salmonella serovar Typhimurium-infected mice can rapidly transmit bacteria to naïve uninfected recipient cagemates, which in turn become persistently infected at mucosal and systemic sites. Animals infected by exposure are indistinguishable from animals deliberately infected with Salmonella serovar Typhimurium by oral gavage or ingestion.

“Supershedder” mice are the major reservoir of infection.We next characterized some of the features of the transmission model by altering the ratio of infected donor mice to uninfected recipient mice within a cage. Salmonella serovar Typhimurium transmission to uninfected mice was very efficient (Table 1). For example, when a single infected donor mouse was placed into a cage with four uninfected recipient mice, transmission to 48% (21/44) of the uninfected mice occurred (Table 1). During these experiments we noted that there was a cage-specific pattern of transmission such that all uninfected mice in five of the experimental cages (n = 20) became infected, whereas only one uninfected animal became infected in six other cages (n = 24) (Table 1). In addition to this transmission variability, we noted that the shedding levels varied significantly among individual mice over time whether they were infected directly by oral gavage or following exposure to an infected donor animal.

To monitor the levels and pattern of fecal shedding more closely, 45 mice distributed in groups of five in nine cages were challenged by oral gavage with 10⁸ CFU of Salmonella serovar Typhimurium. Fecal shedding was monitored every 1 to 3 days. The Salmonella serovar Typhimurium shedding patterns for individual mice could be roughly sorted into three groups based on the peak shedding level (determined by the highest levels of shedding observed for three consecutive sampling times) over the course of 30 days postinfection: supershedders (>10⁸ CFU/g feces), moderate shedders (10⁴ to 10⁸ CFU/g feces), and low shedders (<10⁴ CFU/g feces). The majority of Salmonella serovar Typhimurium-infected mice that we monitored were low or moderate shedders (71%). The rest of the infected animals were supershedders (27%) (Fig. 2a and b).

• Open in new tab
• Download powerpoint

FIG. 2.

Transmission of Salmonella serovar Typhimurium between hosts requires high levels of shedding which is dependent on SPI1 and SPI2. (a and b) Mice were infected orogastrically with 10⁸ CFU of either wild-type (wt) (n = 45), sipB::Cm (n = 30), or ssaV::Kan (n = 30) Salmonella serovar Typhimurium (Stm), and the peak fecal shedding levels (a) and duration of fecal shedding (b) were monitored. (c) Summary of single-donor transmission experiments in which one infected mouse was housed with four uninfected mice for 1 to 58 days and shedding was monitored. The peak level of shedding from donor mice was plotted against the percentage of uninfected mice that became infected. The shedding ranges and numbers of donor mice are indicated below the x axis. (d) A supershedder (donor) (black line) was added to a cage with four uninfected mice (gray lines) for 16 h and then removed. Shedding from the uninfected mice was monitored for 19 days. Previously uninfected mice were colonized in the liver, spleen, and mLN. The dashed line indicates the supershedding level.

We tested the hypothesis that Salmonella serovar Typhimurium transmissibility was directly related to the shedding level. We performed experiments to determine if, as expected, the animals shedding higher levels of bacteria were the most efficient at infecting their naïve cagemates. Indeed, the peak shedding level of each donor mouse plotted against the percentage of uninfected mice that became infected with Salmonella serovar Typhimurium revealed a direct correlation between the level of bacterial shedding and the transmission level (Fig. 2c). If an animal characterized as a supershedder was placed in a cage containing naïve animals, all of the exposed animals became infected in as few as 16 h (Fig. 2d). Salmonella serovar Typhimurium was not transmitted from animals characterized as low or moderate shedders even after naïve cagemates were exposed to them for 30 days (data not shown). Therefore, it is apparently necessary for donor mice to shed Salmonella serovar Typhimurium at levels of ≥10⁸ CFU/g (i.e., to be “supershedders”) in order for them to effectively and efficiently act as a reservoir of infection for uninfected recipient mice. Animals exhibiting the supershedder phenotype were first observed between days 5 and 20 postinfection and continued to show this phenotype for up to 40 days postinfection, when the experiments were typically terminated.

In summary, infection of 129X1/SvJ mice resulted in a high level of shedding (>10⁸ CFU/g feces) in a subset of the animals (“supershedders”). This phenotype is essential for the rapid (in as few as 16 h) and efficient transmission of Salmonella serovar Typhimurium to uninfected recipient mice. We subsequently focused on the supershedder phenomenon and the relative contributions of the infecting microbe, the host, and other factors that contributed to this essential aspect of bacterial transmission from host to host.

Supershedder phenotype requires virulence genes encoded within SPI1 and SPI2.SPI1 and SPI2 have been implicated in the gastrointestinal phase of infection in mice (6, 17). We tested whether mice infected with an SPI1 (sipB::Cm) (29) or SPI2 (ssaV::Kan) (20) mutant could become supershedders and whether these mutants were transmissible to other animals. The sipB::Cm and ssaV::Kan mutants are deficient for the translocation of all type III secretion effector proteins from SPI1 and SPI2, respectively. Only 3% (1/30) of the mice infected with sipB::Cm and none (0/30) of the mice infected with ssaV::Kan shed >10⁸ CFU Salmonella serovar Typhimurium/g feces (Fig. 2a). Furthermore, the shedding patterns for mice infected with the wild-type, sipB::Cm, and ssaV::Kan Salmonella serovar Typhimurium strains were distinct (Fig. 3a, b, and c); all mice infected with wild-type Salmonella serovar Typhimurium were shedding at a detectable level at day 20 postinfection, whereas only 43% of mice infected with sipB::Cm and 10% of mice infected with ssaV::Kan were shedding viable bacteria by day 20 postinfection (Fig. 2b). In addition, 40% of mice infected with sipB::Cm shed at moderate levels, whereas all of the mice infected with ssaV::Kan shed only at low levels. Mice infected with either the sipB::Cm or ssaV::Kan mutant did not develop colitis and were not able to serve as reservoirs of transmission (data not shown). We also noted that mice infected with the sipB::Cm and ssaV::Kan mutants had lower Salmonella serovar Typhimurium colonization levels in the mesenteric lymph nodes (mLN), liver, and spleen than mice infected with wild-type Salmonella serovar Typhimurium (Fig. 3d). Thus, functional SPI1 and SPI2 are required by Salmonella serovar Typhimurium for efficient colonization of animals, and while bacteria may persist at low levels for a significant period of time at both mucosal and systemic sites, SPI1 and SPI2 mutants are not likely transmissible in the model system that we describe here.

• Open in new tab
• Download powerpoint

FIG. 3.

SPI1 and SPI2 contribute to high levels of shedding and systemic colonization. (a to c) Representative shedding patterns for mice infected orogastrically with 10⁸ CFU of wild-type (n = 5) (a), sipB::Cm (n = 5) (b), or ssaV::Kan (n = 5) (c) Salmonella serovar Typhimurium. The level of detection was ∼10 CFU of Salmonella/g of feces. The dashed lines at 10⁴ and 10⁸ CFU/g indicate the shedding categories. (d) Systemic colonization levels for the mice infected for 30 days with wild-type (wt), sipB::Cm, and ssaV::Kan Salmonella serovar Typhimurium. The colonization levels for individual mice are indicated by circles, and the geometric mean for each group is indicated by a solid line. The detection limit for each tissue is indicated by a dashed line. The P values obtained with a Mann-Whitney test comparing the geometric means for the wild-type and sipB::Cm Salmonella serovar Typhimurium strains are as follows: 0.3095 for the liver; 0.0952 for the spleen; and 0.0556 for the mLN. The P values obtained with a Mann-Whitney test comparing the geometric means for the wild-type and ssaV::Kan Salmonella serovar Typhimurium strains are as follows: 0.0079 for the liver; 0.0556 for the spleen; and 0.0177 for the mLN. The data are representative of three experiments.

Supershedder mice exhibit significant colonic inflammation.The gastrointestinal tracts of persistently infected mice were examined using standard histopathological methods to determine what effects, if any, were associated with persistent Salmonella serovar Typhimurium colonization and the supershedder phenotype. The cecum and colon of supershedder mice consistently displayed moderate to severe mucosal to transmural mixed inflammation, focal to regionally extensive mucosal necrosis, and mucosal collapse, with adjacent glandular hyperplasia (Fig. 4b and e), compared to the cecum and colon of uninfected mice (Fig. 4c and f). Lesions in the ileum of the supershedder mice were less severe, ranging from no obvious pathology to moderate mixed inflammatory infiltrates in the lamina propria (data not shown). In contrast to samples from supershedders, most cecal and colon tissue samples from moderate-shedder mice showed only mild to moderate inflammation in the lamina propria, with no necrosis or glandular destruction (Fig. 4a and d). It is notable that the ceca from 2 of 10 moderate-shedder mice also displayed severe inflammation comparable to that seen in supershedder ceca, although the colons of these mice displayed only mild inflammation. The ileal histopathology of moderate-shedder mice was comparable to that seen in the supershedders (data not shown). Thus, the supershedder phenotype is associated with severe inflammation of the cecum and colon.

• Open in new tab
• Download powerpoint

FIG. 4.

Colonization by Salmonella serovar Typhimurium leads to moderate to severe colitis in supershedder mice. (a and b) Ceca of moderate-shedder mice (a) and supershedder mice (b) infected with Salmonella serovar Typhimurium for 20 days. (c) Cecum of an uninfected mouse shown for comparison. (d and e) Colons of moderate-shedder mice (d) and supershedder mice (e) infected with Salmonella serovar Typhimurium for 20 days. (f) Colon of an uninfected mouse shown for comparison. All sections were stained with hematoxylin and eosin. Scale bars = 100 μm.

Salmonella predominantly occupies an extracellular niche within the lumen and a minor intracellular niche within the colonic mucosa of supershedders.Salmonella serovar Typhimurium can invade host cells and replicate intracellularly during infection (16, 22). Since mice which were supershedders showed signs of acute inflammatory disease, we investigated whether the bacteria shed in mice which were persistently infected were extracellular or intracellular in the lumen and colon using an ex vivo gentamicin protection assay (see Materials and Methods). We found that the majority of Salmonella serovar Typhimurium within the colons of supershedders (>99%) and moderate shedders (>99%) were considered extracellular by this assay and, therefore, associated with the luminal contents (Fig. 5). Yet, 0.035% (7 × 10⁴ CFU/g colon) of the Salmonella serovar Typhimurium found in the colons of supershedder mice was protected from gentamicin (Fig. 5) and might constitute a small intracellular subpopulation within the colonic tissue.

• Open in new tab
• Download powerpoint

FIG. 5.

Distribution of Salmonella serovar Typhimurium within colons of infected mice: levels of Salmonella serovar Typhimurium in the luminal contents and colons of supershedder (n = 4) and moderate-shedder (n = 4) mice after incubation in the absence (gray bars) and presence (open bar) of gentamicin. The detection limit was 10² CFU/g. The data are representative of three experiments.

To support this idea, we performed scanning confocal immunofluorescence microscopy with infected colons and luminal contents that were stained with an anti-Salmonella serovar Typhimurium antibody in combination with a variety of phagocyte-specific antibodies (CD18, CD11b, CD11c, and Ly6G). The colonic mucosa of moderate shedders and supershedders were compared to that of uninfected mice, and the results showed that there was increased staining for CD18 and CD11b, indicating that there was an inflammatory infiltrate (data not shown). The colonic mucosa of uninfected and moderate-shedder mice contained low levels of dendritic cells (CD11c⁺) (Fig. 6a and b) but was not stained for the neutrophil marker Ly6G⁺ (Fig. 6d and e). In contrast, the colons of supershedders contained large aggregates of dendritic cells within the lamina propria and submucosa (Fig. 6c), suggesting that the inflammatory lesions observed within the colons of supershedders were at least partially the result of dendritic cell migration to the site of infection. In addition, low levels of neutrophils were observed within the colons of supershedders (Fig. 6f). Moreover, Salmonella serovar Typhimurium was routinely observed within dendritic cells (Fig. 6g and h), as we found that 36 of 44 of the intracellular Salmonella serovar Typhimurium cells were within CD11b⁻ CD11c⁺ cells and 8 of 44 cells were within CD11b⁺ CD11c⁺ cells in the colons of supershedder mice.

• Open in new tab
• Download powerpoint

FIG. 6.

Colons of supershedder mice contain intracellular (CD11c⁺ dendritic cells) and extracellular luminal Salmonella serovar Typhimurium. (a to f) Immunofluorescent photomicrographs of serial sections of colons from uninfected (a and d), moderate-shedder (b and e), and supershedder (c and f) mice. Host actin (blue) was stained with Phalloidin, and host leukocytes (red) were stained with CD11c (a, b, and c) and Gr-1 (d, e, and f) antibodies. Scale bars = 100 μm. (g and h) Immunofluorescent photomicrographs of supershedder colon stained for CD11c (red), Salmonella (green), and host nuclear DNA (blue). Scale bars = 100 μm (g) and 10 μm (h). (h) Magnified section of panel g, showing Salmonella serovar Typhimurium within a CD11c⁺ host cell. The XZ (above panel h) and YZ (to the right of panel h) images are sections from a three-dimensional reconstruction of the image in panel g. (i to k) Immunofluorescent photomicrographs of colonic contents from uninfected (i), moderate-shedder (j), and supershedder (k) mice stained for Gr-1 (red), Salmonella (green), and DNA (host cell or bacterial) (blue). Scale bars = 20 μm. The images are representative of the results of a microscopy analysis of tissues from five different mice.

Since the majority of Salmonella serovar Typhimurium cells reside within the lumen of the colon in all persistently infected animals, we also stained the luminal contents with anti-CD18, -CD11b, -CD11c, and -Ly6G antibodies. The luminal contents of the supershedder mice did not contain CD18-, CD11b-, and CD11c-positive cells (data not shown) but were dominated by neutrophils, as determined by their characteristic morphology and staining by the Ly6G⁺ antbody (Fig. 6k). In contrast, neutrophils were not observed in the luminal contents from noninfected and moderate-shedder mice (Fig. 6i and j). Thus, the colitis induced in the supershedder mice was associated with an influx of neutrophils into the lumen of the colon, as well as severe transmural inflammation characterized in part by dendritic cell infiltration. In addition, supershedder animals harbored a small Salmonella serovar Typhimurium population within the dendritic cells of the colonic mucosa.

Salmonella cultured from supershedder mice does not have heritable differences in virulence compared to Salmonella cultured from moderate-shedder animals.In our experiments we employed a single Salmonella serovar Typhimurium strain, SL1344, and homogenic variants of this strain. We presumed therefore that we were introducing a genetically homogeneous bacterial population into mice during oral challenge. Infection, colonization, and the subsequent shedding of bacteria may reflect selective pressures that promote the emergence and enrichment of genetic variants with altered virulence that may have important implications for supershedding. To test this idea, mice were infected orally with 10⁸ CFU of an inoculum containing an equal mixture of Salmonella serovar Typhimurium isolated from a supershedder marked with Kan^r (designated SS) and Salmonella serovar Typhimurium isolated from a moderate shedder marked with Cm^r (designated MS) (antibiotic resistance cassettes had no effect on virulence). At 10 days postinfection, these mice were sacrificed to determine the bacterial loads in the cecum, colon, mLN, liver, and spleen. The competitive index values indicate that in general, equal numbers of Salmonella serovar Typhimurium SS and Salmonella serovar Typhimurium MS were present in the mLN, livers, and spleens of infected animals (data not shown). However, in the large intestine of infected mice one strain or the other generally dominated the Salmonella serovar Typhimurium population. This was quite possibly due to a strong “bottleneck” effect previously observed during bacterial intestinal colonization (8). This experiment was repeated three times with different Salmonella serovar Typhimurium isolates, and the same results were obtained. Subsequently, competition experiments with the same two strains were performed by infecting mice by the intraperitoneal route to avoid the intestinal bottleneck. When this route of infection was used, Salmonella serovar Typhimurium strains isolated from supershedders and moderate shedders were equally virulent (data not shown). In addition, Salmonella serovar Typhimurium isolated from the feces of a supershedder had an ID₅₀ identical to that of the laboratory-grown wild-type strain (data not shown). Thus, there was no obvious measurable heritable change in the bacteria that emerged from the supershedder animals compared with the the bacteria that emerged from the moderate shedders. Of course, this is not to say that there are not important transient population changes that may be operative during infection that lead to increased shedding by some animals.

Immunosuppression does not induce the supershedder phenotype.We employed an inbred mouse strain in all of our transmission studies. Therefore, it seemed unlikely that a heritable trait was responsible for differences in Salmonella serovar Typhimurium fecal shedding. So, why did some mice develop into highly infectious supershedders and act as a reservoir for transmission while others did not? In light of our failure to detect any obvious difference in the general genetic properties of the infecting microorganisms, we considered the possibility that Salmonella serovar Typhimurium might benefit from the physiological stress induced by social hierarchy among mice in a cage that is known to lead to an immunosuppressed state in some animals but not in others (28, 41). Therefore, we treated mice with dexamethasone, a corticosteroid immunosuppressant, and fecal shedding was monitored to determine if deliberate suppression of the immune response induced the supershedder phenotype. In the first series of experiments, dexamethasone (5 mg/liter) was added to the drinking water of mice 1 day before infection with Salmonella serovar Typhimurium. The frequencies of supershedders in treated mice and untreated mice were monitored for 10 to 14 days postinfection. We found that 22% (2/9) of the mice treated with dexamethasone and 30% (3/9) of the untreated mice developed into supershedders (Fig. 7a) even though the levels of Salmonella serovar Typhimurium at systemic sites were significantly higher in immunosupressed mice (Fig. 7b). These results suggest that immunosuppression with dexamethasone did not induce the development of the supershedder phenotype during the first 2 weeks of infection.

• Open in new tab
• Download powerpoint

FIG. 7.

Immunosuppression does not induce the supershedder phenotype in mice. (a) Mice were treated with dexamethasone (black bars) (n = 9) or were not treated (gray bars) (n = 9) and then were infected with Salmonella serovar Typhimurium, and shedding was monitored for 10 to 14 days to identify the peak shedding range. (b) Systemic colonization for infected mice that were treated with dexamethasone (DXM) (from the experiment whose results are shown in panel a) or were not treated (from the experiment whose results are shown in panel a). The colonization levels of individual mice are indicated by black (dexamethasone treated) or gray (untreated) circles, and the geometric means are indicated by solid lines. The detection limit for Salmonella serovar Typhimurium for each tissue is indicated by a dashed line. The levels of Salmonella serovar Typhimurium for mice treated with dexamethasone are significantly higher that the levels of Salmonella serovar Typhimurium for untreated mice as determined by the Mann-Whitney test (for the liver, P = 0.0286; for the spleen, P = 0.0286; for the mLN, P = 0.0286). (c) Mice persistently infected with Salmonella serovar Typhimurium for 110 days were treated with dexamethasone (arrow), and shedding was monitored for 21 days. (d) Systemic colonization in persistently infected mice that were treated with dexamethasone (from the experiment whose results are shown in panel c) or were not treated (data not shown). The levels of Salmonella serovar Typhimurium for mice treated with dexamethasone are significantly higher than the levels of Salmonella serovar Typhimurium for untreated mice as determined by the Mann-Whitney test (for the liver, P = 0.0079; for the spleen, P = 0.0079; for the mLN, P = 0.0079). The data are representative of two experiments.

In another experiment, mice (n = 10) that were persistently infected for 110 days but were not presently shedding detectable levels of Salmonella serovar Typhimurium (negative fecal cultures for five consecutive days prior to treatment) were treated with dexamethasone. Despite a general modest increase in shedding over the 21-day treatment period, all mice remained low-level shedders (Fig. 7c) (fecal levels correlated with intestinal levels in untreated and dexamethasone-treated mice [data not shown]) even though these mice exhibited significantly higher levels of Salmonella serovar Typhimurium in the liver, spleen, and mLN than untreated controls (Fig. 7d). These results indicate that immunosuppression with dexamethasone resulted in reactivation of a systemic Salmonella serovar Typhimurium infection. In summary, immunosuppression of mice during the subacute or persistent stage of infection with dexamethasone did not induce the supershedder phenotype.

Indigenous intestinal microbiota controls the development of the supershedder phenotype.Persistent infection and subsequent bacterial shedding take place in the context of a complex competitive environment. Therefore, we considered the possible role of the intestinal microbiota in controlling the development of the supershedder phenotype. Might colonization resistance (53) play a role? Did a supershedder animal result from Salmonella serovar Typhimurium outcompeting the indigenous microbiota and replicating to high levels in some mice with a “susceptible” indigenous microbial community? During the course of our experiments, we noted that infection of mice with a streptomycin-resistant Salmonella serovar Typhimurium strain 24 h after streptomycin treatment resulted in all mice becoming supershedders by 24 h postinfection (data not shown), as observed in mice with the C57BL/6 genetic background (2). We also treated mice that were shedding very low levels of Salmonella serovar Typhimurium during the subacute phase of infection (day 23 postinfection) with a single dose of streptomycin (5 mg). By 48 h posttreatment all of these mice were shedding Salmonella serovar Typhimurium at supershedder levels, which remained elevated for the duration of the experiment (Fig. 8a). The levels of transmission to naïve mice and the colonic pathology of these mice were comparable to those of supershedders that developed in the absence of streptomycin.

• Open in new tab
• Download powerpoint

FIG. 8.

Induction of the supershedder phenotype following a reduction in the size of the intestinal bacterial community due to streptomycin treatment. (a) Shedding patterns of mice (n = 5) that were infected for 23 days and then treated orally with 5 mg of streptomycin (arrow). (b) Q-PCR of total bacterial rRNA gene (open bars) and Salmonella serovar Typhimurium DNA (gray bars) from feces of mice prior to and after streptomycin treatment of mice that were shedding low levels of Salmonella serovar Typhimurium. The number of CFU of Salmonella serovar Typhimurium (black bars) was determined by culturing organisms from feces to compare the quantification methods. The number of CFU was not determined for the 6-h time point. The dashed line at 10⁸ CFU/g indicates the supershedder level. The data are representative of three experiments.

To monitor the shift in the intestinal bacterial community during the induction of the supershedder phenotype, we performed Q-PCR with feces from mice before and after streptomycin treatment using broad-range bacterial 16S rRNA primers (26, 27) and S. enterica-specific primers (34). As expected, there was a dramatic reduction in the bacterial 16S rRNA gene copy number 6 h after streptomycin treatment, and by 24 h posttreatment the total bacterial 16S rRNA gene copy number was reduced by 88% (Fig. 8b). By 24 to 48 h posttreatment the total number of bacterial 16S rRNA gene copies returned to pretreatment levels (Fig. 8b). During this same time period, the Salmonella serovar Typhimurium numbers increased 10,000-fold to supershedder levels as measured by Q-PCR and CFU determination (Fig. 8b). By 48 h posttreatment, Salmonella serovar Typhimurium comprised ∼5% of the total bacterial population within the feces of streptomycin-induced supershedders (Fig. 8b). Thus, we demonstrated that the supershedder phenotype was induced in low-shedder mice by reducing the competitive intestinal microbiota with streptomycin.

We were also able to show that persistently infected mice that were not shedding detectable levels of Salmonella serovar Typhimurium could be reactivated into supershedders by treatment with streptomycin. For example, we treated mice (n = 10) that had been infected for 125 days with a single dose of streptomycin, and individual mice were then housed in separate cages. Streptomycin-treated mice rapidly began shedding (8 of 10 mice), and the levels reached supershedder levels by 3 to 6 days posttreatment (Fig. 9a). In two mice that were treated with streptomycin the infection was not reactivated, nor were these mice colonized with Salmonella serovar Typhimurium in any of the systemic tissues examined, suggesting that they had completely cleared the infection. The colons of the streptomycin-reactivated supershedder mice displayed severe transmural inflammation (data not shown), indicating that streptomycin reactivation of chronic infection results in supershedders that display pathology similar to that seen in supershedders during the subacute phase. In contrast, the control mice that were treated with PBS did not begin shedding detectable levels of Salmonella serovar Typhimurium (Fig. 9b) and displayed no signs of intestinal inflammation (data not shown). Upon termination of the experiment, mice reactivated with streptomycin contained significantly higher levels of Salmonella serovar Typhimurium in the liver, spleen, and mLN (Fig. 9c) and displayed elevated serum levels of Salmonella serovar Typhimurium-specific antibodies (Fig. 9d) compared to control mice, indicating that streptomycin reactivation of the supershedder phenotype in persistently infected mice resulted in an acute systemic infection.

• Open in new tab
• Download powerpoint

FIG. 9.

Reactivation of the supershedder phenotype in chronically infected mice after the size of the indigenous microbiota was reduced by streptomycin treatment. (a) Mice chronically infected for 125 days and treated with streptomycin (arrow) (n = 10). (b) Mice chronically infected for 125 days and mock treated with PBS (arrow) (n = 10). (c) Colonization levels for various organs from mice that were treated with streptomycin or PBS. Black (streptomycin) and gray (PBS) circles indicate the levels for individual mice, and the geometric means are indicated by solid lines. The detection limit for Salmonella serovar Typhimurium for each tissue is indicated by a dashed line. The levels of Salmonella serovar Typhimurium for mice treated with streptomycin are significantly higher than the levels of Salmonella serovar Typhimurium for mock-treated mice as determined by the Mann-Whitney test (for the cecum, P = 0.029; for the colon, P = 0.029; for the feces; P = 0.029; for the liver, P = 0.029; for the spleen, P = 0.029; for the mLN, P = 0.029). (d) Levels of serum anti-Salmonella serovar Typhimurium IgG for mice treated with streptomycin or PBS. The positive control (pos.) was a mouse orally infected with 10⁸ CFU Salmonella serovar Typhimurium for 30 days, and the negative control (neg.) was an uninfected mouse.

Antibiotic-induced alterations in the intestinal microbiota result in a lasting predisposition to the supershedder phenotype.While reducing the intestinal microbiota with a single dose of streptomycin during infection uniformly resulted in animals that exhibited a supershedder phenotype, these conditions selectively promoted Salmonella serovar Typhimurium growth and replication within the intestinal tract. To more accurately determine if antibiotic alterations to the indigenous microbiota predispose animals to the supershedder phenotype, mice were treated with 5 mg of neomycin by oral gavage, and groups of mice were infected posttreatment. Neomycin treatment reduced the overall number of intestinal bacteria by 95% at 24 h posttreatment, but the numbers returned to normal levels by 48 h posttreatment and remained stable (data not shown). Furthermore, the composition of the culturable microbiota was altered in treated mice as the levels of E. coli and neomycin-resistant enterococci increased 100- and 500-fold, respectively, after neomycin treatment (data not shown). On day 4 (n = 4) and day 7 (n = 4) posttreatment groups of mice were infected with 10⁸ CFU of Salmonella serovar Typhimurium by oral gavage. As a control, a group of untreated mice (n = 4) were infected with the same dose. Shedding was monitored at regular intervals to determine the levels of Salmonella serovar Typhimuriun in the feces. As expected, one of the four untreated mice became a Salmonella serovar Typhimurium supershedder by day 8 postinfection (Fig. 10a). When mice were infected 4 days after neomycin administration, three of four mice were shedding >10⁷ CFU/g feces at 24 h postinfection (Fig. 10b), a level >1,000-fold higher than the levels shed by untreated mice at the same time (average, 4 × 10⁴ CFU/g; statistically different [P < 0.05] based on a Mann-Whitney test). The same three mice (75% of the mice in the cage) began shedding at supershedder levels on day 4 postinfection (Fig. 10b). The effects of neomycin treatment were still present on day 7 posttreatment as two of four mice infected at this time shed high levels of Salmonella serovar Typhimurium at 24 h postinfection (>10⁸CFU/g feces; P < 0.05 compared to untreated controls based on a Mann-Whitney test) and all four mice were supershedders by day 5 postinfection (Fig. 10c). Thus, after a single treatment with neomycin seven of the eight mice were predisposed to the supershedder phenotype for several days, even though the overall abundance of intestinal bacteria returned to normal levels within 48 h. Hence, alterations in one of the host's protective innate defense systems can significantly influence Salmonella serovar Typhimurium disease and transmission.

• Open in new tab
• Download powerpoint

FIG. 10.

Antibiotic treatment induces a lasting predisposition to the supershedder phenotype. (a) Shedding from mice infected with 10⁸ CFU of Salmonella serovar Typhimurium by oral gavage (n = 4). (b) Shedding from mice infected with 10⁸ CFU of Salmonella serovar Typhimurium by oral gavage 4 days after neomycin treatment (n = 4). (c) Shedding from mice infected with 10⁸ CFU of Salmonella serovar Typhimurium by oral gavage 7 days after neomycin treatment (n = 4). The dashed lines at 10⁸ CFU/g indicate the supershedder level. The time of neomycin treatment on day 0 and the time of Salmonella serovar Typhimurium infection are indicated by arrows. The black lines indicate data for mice that are supershedders, whereas the data for moderate shedders are indicated by gray lines.