ABSTRACT
Salmonella enterica serovar Enteritidis is a common cause of foodborne illness in the United States. The bacterium can be transmitted to humans via contaminated chicken meat and eggs, and virulence in humans requires type III secretion system 1 (TTSS-1), encoded on Salmonella pathogenicity island 1 (SPI-1). Chickens often carry S. Enteritidis subclinically, obscuring the role of SPI-1 in facilitating bacterial colonization. To evaluate the role of SPI-1 in the infection of chicks by Salmonella, we created and utilized strains harboring a stable fluorescent reporter fusion designed to quantify SPI-1 expression within the intestinal tracts of animals. Using mutants unable to express TTSS-1, we demonstrated the important role of the secretion system in facilitating bacterial colonization. We further showed that coinoculation of an SPI-1 mutant with the wild-type strain increased the number of mutant organisms in intestinal tissue and contents, suggesting that the wild type rescues the mutant. Our results support the hypothesis that SPI-1 facilitates S. Enteritidis colonization of the chicken and make SPI-1 an attractive target in preventing Salmonella carriage and colonization in chickens to reduce contamination of poultry meat and eggs by this foodborne pathogen.
INTRODUCTION
Salmonella enterica is a ubiquitous bacterium capable of colonizing mammals, amphibians, birds, and plants (1). S. enterica causes disease in animals, including humans, livestock, and laboratory mice (2). In the United States, S. enterica serotypes Enteritidis and Typhimurium are the most common causes of foodborne Salmonella infection in humans (3). Despite efforts to curb transmission, overall Salmonella infection prevalence has not changed in 50 years (3). Though S. Typhimurium and S. Enteritidis infections cause acute gastroenteritis in humans (3), the bacteria colonize the intestines of chickens without causing disease (4). This subclinical colonization poses a challenge for the prevention of foodborne transmission of Salmonella, as colonized poultry are difficult to identify.
The basis of the difference in the clinical consequences of S. Typhimurium and S. Enteritidis infection between mammals and birds is elusive. One possible explanation for this difference is the differential expression of virulence genes and corresponding host inflammation. In mammals, the inflammation that characterizes Salmonella gastrointestinal infection is caused by the secretion of effectors dependent on the expression of Salmonella pathogenicity island 1 (SPI-1) (2, 5), a horizontally acquired island whose gene products include the structural and effector proteins of type III secretion system 1 (TTSS-1) (6). When expressed, TTSS-1 injects effector proteins into host intestinal epithelial cells. These effectors induce bacterial uptake and elicit a potent host neutrophilic inflammatory response that is conducive to bacterial outgrowth in the intestinal lumen (7–10). This process is critical for successful colonization of the mammalian host; bacterial mutants unable to express TTSS-1 or its effectors are unable to colonize the intestine efficiently and are attenuated in their virulence (8).
The expression of SPI-1 is energetically burdensome (11), and Salmonella employs two strategies to maximize intestinal proliferation in the face of costly gene expression. The first strategy is biphasic expression of SPI-1: only a subpopulation of bacteria express genes on SPI-1, leaving a nonexpressing population to proliferate in the intestines (12). In a second strategy, individual bacteria within an infected mammalian host acquire mutations that prevent SPI-1 expression and are sometimes called “defectors” (13). Both strategies allow non-SPI-1-expressing bacteria to capitalize on the inflammatory environment created by neighboring SPI-1-expressing bacteria in the mammalian intestine.
Salmonella (serovars Typhimurium and Enteritidis) colonizes the chick intestine without clinical signs; thus, the role of SPI-1 in colonization of poultry intestines is unclear. Although hilA mutants of S. Enteritidis (which express SPI-1 genes poorly) exhibit reduced colonization of the intestines, spleen, and liver in chicks infected at 1 day posthatch (14, 15), genetic-screening approaches have failed to identify the role of the major transcriptional activators of SPI-1, hilA and hilD, in colonization of the chicken intestine (16). Furthermore, SPI-1 expression is significantly reduced at 42°C (the body temperature of a chicken) compared to expression at 37°C (17). The importance of SPI-1 expression in the chicken is further complicated by the fact that both hilA and hilD regulate the expression of genes outside of SPI-1.
Given these apparently contradictory data, we evaluated the requirement for SPI-1 expression in S. Enteritidis infection of chickens. We used fluorescent reporters to identify the anatomic location of SPI-1-expressing organisms and to evaluate the level of SPI-1 expression in the chicken intestine. We found that SPI-1 expression is required in the chicken and promotes mild intestinal inflammation, consistent with the subclinical nature of disease in the species. We further showed that SPI-1-nonexpressing populations take advantage of the environment created by SPI-1-expressing bacteria. These results implicate SPI-1 as an important virulence target to prevent colonization of chickens and to ultimately curtail foodborne transmission of Salmonella.
RESULTS
Chromosomally encoded fluorescent proteins permit SPI-1 quantitation in Salmonella.While SPI-1 is a critical determinant of Salmonella infection in mammals, its role in establishing colonization in poultry is unclear. We sought to characterize SPI-1 expression during Salmonella infection of the chicken and to compare this expression to the dynamics of SPI-1 expression in mammalian infection. We first employed a fluorescent reporter strain of Salmonella Typhimurium that expresses blue fluorescent protein (BFP) constitutively (inserted at phoN, which has been shown to be nonessential for in vivo virulence [18]), and green fluorescent protein (GFP) as a reporter of SPI-1 expression through its insertion within the SPI-1 sic-sip operon under the control of the sicA promoter (see Fig. S1A in the supplemental material). A unique advantage of our system is that both fluorescent proteins are encoded chromosomally, and thus, long-term infection models can be conducted without reporter loss. Our strain behaved as expected in vitro: BFP was expressed under all conditions tested, and GFP was expressed biphasically (Fig. S1B and C) and without affecting the expression of adjacent genes. Our reporter strain responded as expected to conditions that affect SPI-1 expression, exhibiting a reduced proportion of GFP-positive Salmonella bacteria upon treatment with butyrate, nonanoate, and cholate, which represent three recognized classes of SPI-1 inhibitors: short-chain fatty acids, long-chain fatty acids, and bile acids (Fig. 1) (19–23).
Fluorescent reporter showing reduction in the population of SPI-1-expressing bacteria upon treatment with repressors. A strain expressing BFP constitutively and GFP as a reporter for SPI-1 expression was grown with shaking in MOPS (morpholinepropanesulfonic acid)-buffered LB broth alone or with butyrate, cholate, or nonanoate at the indicated concentration. After 16.5 h, the bacteria were fixed and analyzed by flow cytometry. This experiment was performed with 3 biological replicates. Error bars indicate standard errors of the mean.
Fluorescent Salmonella Typhimurium enables SPI-1 quantitation in the murine intestine.To first demonstrate that our reporter strain retained virulence and to characterize the expression of SPI-1 in the mouse intestinal tract, a well-defined animal model, we performed a competition assay between the fluorescent reporter and the otherwise isogenic wild-type (WT) S. Typhimurium ATCC 14028s in mice. Streptomycin-treated C57BL/6 mice were orally inoculated with an equal mixture of the two strains, and bacterial colonization of the ilea, ceca, liver, and spleen was quantified at 2 days postinfection, a time by which robust inflammation is observed in this model (24–26). We found that our reporter strain retained full virulence, colonizing both the intestine and systemic sites similarly to the wild-type organism (Fig. 2A); invasion of the spleen and liver was, in fact, slightly higher for the fluorescent reporter. Overall, the insertion of BFP and GFP genes did not perturb S. Typhimurium virulence using a well-established model of animal infection.
A fluorescent strain is fully virulent. A streptomycin-resistant strain of S. Typhimurium containing ΔphoN::BFP-cat and sicA→GFP was compared to its streptomycin-resistant parent strain for virulence in streptomycin-treated C57/BL6 mice. (A) Two days after infection, mice were euthanized, and organ homogenates (spleen and liver) or contents (ileum and cecum) were plated for CFU enumeration. The circles represent competitive indices between strains for individual animals. The bars represent means (±standard errors of the mean) for each group. This experiment was performed with 5 biological replicates. The asterisks indicate a significant (P < 0.05) difference between the wild-type and fluorescent strains. (B) Ileal and cecal contents of mice coinoculated with the WT and the fluorescent reporter strain were analyzed by flow cytometry. Fluorescent Salmonella bacteria were identified by their constitutive BFP expression; then, the percent SPI-1 expression of that population was assessed by GFP quantitation. The circles represent the SPI-1-positive fluorescent bacterial population for individual animals, while the bars signify means for each organ.
We also characterized SPI-1 expression by the fluorescent strain in the murine intestine. A portion of intestinal contents from both the ilea and ceca was fixed at the time of collection and subsequently analyzed by flow cytometry. Salmonellae were identified by their constitutive expression of BFP, and the BFP-positive population was analyzed for expression of GFP to gauge SPI-1 expression. We found that the fluorescence of BFP in the reporter strain was greater than the fluorescence background of intestinal contents from uninfected animals (data not shown), and BFP-positive populations could be readily observed from the ceca, where they accounted for an average of 18.6% of the total events analyzed (see Fig. S2 in the supplemental material). However, when we interrogated this population for SPI-1 expression using GFP, we observed a very low proportion of GFP-positive cells (average, 4.1%), indicating that SPI-1 expression was low in the organ (Fig. 2B). In contrast, we were able to detect a BFP-positive population in mouse ileal contents in 4 of the 5 animals analyzed (see Fig. S3 in the supplemental material). The ability to detect the fluorescently labeled bacteria was dependent on the number of bacteria available for analysis in the sample (ceca contained on average of 1.0 × 109 CFU/organ, while ilea contained on average of 2.5 × 106 CFU/organ), rather than a reflection of technical limitations in the fluorescent detection of the BFP-labeled bacteria. For the samples that yielded detectable BFP-positive populations, these presumed Salmonella strains comprised less than 1.0% of the total events, demonstrating that they were a small proportion of detectable events in the debris of the ileal lumen. When we analyzed these bacteria, they exhibited distinctly biphasic expression of GFP, with the GFP-positive population averaging 13.5% (Fig. 2B). These findings show that our system is capable of detecting small proportions of Salmonella bacteria in animals, and they indicate expression of SPI-1 in the mouse ileum and are consistent with established models of infection (12, 27).
Fluorescent Salmonella Enteritidis enables SPI-1 quantitation in the intestines of chicks.Having determined that our fluorescent proteins did not interfere with virulence, we moved these constructs into S. Enteritidis strain CDC_2010K_0968, a clinical isolate associated with the 2010 egg outbreak of Salmonella infection in the United States (28, 29). The strain behaved as expected in vitro (see Fig. S4 in the supplemental material) and was used to orally inoculate white leghorn chicks at 4 days posthatch. The chicks were euthanized at either 2 or 9 days postinoculation, and colonization was assessed by serial dilution and plating (Fig. 3). Colonization of the ceca was similar at days 2 and 9 postinfection.
Fluorescent Salmonella colonizes chick intestines and organs. Four-day-old chicks were inoculated orally with a strain of S. Enteritidis containing constitutively expressed BFP and an inducible GFP reporter of SPI-1. At 2 and 9 days postinoculation, bacterial colonization of the intestines (ileal content [A], cecal content [B], and colonic content [C]) and spleen (D) was assessed. The circles represent data points for individual animals. The bars signify means (±standard errors of the mean) for each group. This experiment was performed with 3 to 5 biological replicates.
We also quantified the expression of SPI-1 in the intestines of infected chickens, using the same techniques as for the murine experiments. Salmonella populations were identified from all ceca and 7 out of 8 colons of infected chicks. Similar to mice, Salmonella recovery from infected ilea was inconsistent; in only 1 of 8 samples analyzed could a distinct BFP-positive population be identified (see Fig. S5 and S11 in the supplemental material). Again, our ability to detect bacteria corresponded to the absolute bacterial numbers in the samples (ceca contained on average of 2.5 × 108 CFU/organ, while colons contained on average of 5.4 × 107 CFU/organ and ilea an average of 1.7 × 107 CFU/organ). As the ceca and colons yielded sufficient bacterial populations for analysis, SPI-1 quantitation was conducted. Expression of SPI-1 in the intestinal tract varied greatly between animals, but our analysis demonstrated a detectable proportion of Salmonella bacteria to be expressing SPI-1 at both intestinal sites (Fig. 4). On day 2 postinfection, all the populations analyzed from both the ceca and colons exhibited SPI-1 expression in less than 15% of the population. Fluorescence-activated cell sorter (FACS) histograms from each of the organs (see Fig. S6 and S7 in the supplemental material) exemplify inherent variation of the samples and also demonstrate the biphasic expression that characterizes SPI-1 expression. By day 9 postinfection, average SPI-1 expression was increased in both organs, though variation among animals was high. GFP histograms of some samples show increased fluorescence intensity of one population rather than the formation of two distinct populations (see Fig. S12 and S13 in the supplemental material). It is unclear whether this shift in single-population fluorescence indicates a departure from biphasic expression or if it is an intermediate state of biphasic induction.
SPI-1 is variably expressed in the chicken intestine. The intestinal contents of chicks infected with a fluorescent reporter strain of S. Enteritidis were analyzed by FACS. Salmonella bacteria were initially identified by their constitutive BFP expression, and then, the SPI-1 expression of that population was assessed by GFP quantitation. The circles represent the SPI-1-positive bacterial population for individual animals, while the bars represent means (±standard errors of the mean) for each group. This experiment was performed with 3 to 5 biological replicates.
SPI-1 expression facilitates colonization of the chick intestine at 9 days postinfection.After observing that SPI-1 is expressed in the intestines of infected birds, we asked whether its expression plays an essential role in establishing colonization. To assess the effects of invasion during the most acute stages of infection, we first measured numbers of a ΔhilD mutant and wild-type strains recovered 2 days postinfection. hilD encodes a transcriptional activator that tightly regulates SPI-1 expression, and mutants of hilD are nearly devoid of SPI-1 expression in vitro. When chicks were infected with either the wild-type or ΔhilD strain, no defect in intestinal colonization was observed for the ΔhilD mutant at 2 days postinoculation (Fig. 5A, B, and C), while we observed a significant reduction in the splenic bacterial load (Fig. 5D). These data suggest that the ΔhilD strain has a reduced ability to colonize systemic sites as early as 2 days postinfection. Histopathologic analysis of cecal tissue from these animals revealed that birds inoculated with wild-type S. Enteritidis had a mild inflammatory response relative to birds inoculated with the ΔhilD mutant (Fig. 6; see Fig. S17 in the supplemental material). At 9 days postinfection, however, the ΔhilD mutant exhibited reduced colonization of the ceca and colon (Fig. 5F and G). We compared this mutant to a mutant with deletion of the entire pathogenicity island (ΔSPI-1) and found their phenotypes to be comparable, suggesting that the loss of hilD faithfully recapitulates that of SPI-1. Both mutants were also recovered in lower numbers from the spleens of infected animals (Fig. 5H). Our observations expand existing work by quantifying SPI-1 expression in the chicken intestine and suggest that SPI-1 expression does influence the ability of S. Enteritidis to colonize the chick intestine when measured at later time points (14, 15).
SPI-1 contributes to colonization on day 9, but not day 2, postinoculation. Four-day-old chicks were inoculated orally with a nalidixic acid-resistant strain of S. Enteritidis (WT), its ΔhilD derivative (ΔhilD), or a mutant of the entire SPI-1 (ΔSPI1). At 2 days (A to D) or 9 days (E to H) postinoculation, bacterial colonization of the intestines (ileal content [A and E], cecal content [B and F], and colonic content [C and G]) and spleen (D and H) was assessed. The circles represent data points for individual animals whose organ plating yielded countable colonies, while the triangles represent data points below the limit of detection for that organ. The bars represent means (±standard errors of the mean) for each group. This experiment was performed with 4 to 7 biological replicates. The asterisks indicate a significant (P < 0.05) difference between the WT and ΔhilD populations or the WT and ΔSPI-1 populations, as indicated.
Wild-type Salmonella produces subclinical inflammation by day 2 postinoculation. Chicks were mock inoculated (Uninfected) or inoculated with either a nalidixic acid-resistant strain of S. Enteritidis (Wild Type) or its ΔhilD derivative (ΔhilD). At 2 days postinoculation, cecal-tissue samples were collected and fixed, and inflammation was scored based on the eight criteria shown. The bar segments indicate the means of each criterion for that group. This experiment was performed with 3 biological replicates for each group.
SPI-1-expressing salmonellae support colonization by Salmonella mutants unable to express SPI-1.In mice, SPI-1 expression by a subpopulation promotes an environment conducive to outgrowth by strains not expressing SPI-1 (13). We asked whether an analogous situation occurs in the chick intestine and tested this idea by coinoculating the wild-type strain with the ΔSPI-1 mutant at a 1:1 ratio. In these experiments, the ΔSPI-1 mutant colonized intestinal sites normally when the wild type was present (Fig. 7A and B), reversing the colonization defect exhibited by the SPI-1 mutant when it is inoculated individually (Fig. 5F). In the same infections, the SPI-1 mutant still colonized systemic sites, spleen, and liver, but at a reduced level relative to the wild type, similar to what occurs in individual infections (Fig. 5H and 7B).
Coinoculation with the wild type enables SPI-1 mutants to colonize the intestinal tract. Four-day-old chicks were inoculated orally with a mixture of a nalidixic acid-resistant strain of S. Enteritidis (WT) and a derivative mutant of the entire SPI-1 (ΔSPI-1). Inocula were mixed at either a 1:1 (A and B) or 1:4 (C and D) ratio. At 9 days postinoculation, bacterial colonization of the intestines (A and C) and spleen and liver (B and D) was assessed, and competitive indices between the two strains were calculated. The circles indicate competitive indices for individual animals whose organ plating yielded countable colonies, while the triangles represent data points for which one strain was below the limit of detection for that organ. The bars represent means (±standard errors of the mean) for each group. This experiment was performed with 5 biological replicates. The asterisks indicate a significant (P < 0.05) difference between the wild-type and ΔSPI-1 populations.
Because by flow cytometry analysis only a small percentage (16.7% and 7.5% from ceca and colon, respectively) of Salmonella strains express SPI-1 in the intestine at day 9 postinfection, only a small population of SPI-1-expressing organisms may be needed to create favorable growth conditions for Salmonella in the intestine. To test this idea, we reduced the number of potential SPI-1-expressing organisms even further by coinoculating wild-type bacteria with a ΔSPI-1 mutant at a ratio of 1:4. In this experiment, the wild type and ΔSPI-1 mutant were recovered from the intestine at a ratio of 1:5 (Fig. 7C). These data suggest that the presence of SPI-1-competent organisms improves colonization by organisms that lack SPI-1. Overall colonization of the ceca of these chicks reached 9.6 × 108 CFU/cecum total Salmonella organisms, consistent with that observed for wild-type infections alone (3.4 × 108 CFU/cecum). These results show that in the chicken, only 20% of a Salmonella population needs to have the ability to express SPI-1 in order to support colonization by mutants lacking SPI-1.
DISCUSSION
We employed novel fluorescent reporter strains of Salmonella to monitor SPI-1 expression in the intestines of chicks and compared our observations to the well-characterized murine infection. In this work, we show that SPI-1 is important for Salmonella to reach systemic organs (the spleen and liver), and also that SPI-1 aids intestinal colonization. Importantly, the work described here reconciles seemingly contradictory results of past studies in the field. Our findings, for example, support previous studies demonstrating that mutants of hilA, an SPI-1 regulator, colonize the chicken intestine poorly (14, 15). Previous genetic screens in which mutants were pooled, however, failed to identify similar SPI-1 regulators as important to intestinal colonization (16). Our finding that SPI-1-expressing bacteria rescue the growth of those unable to express SPI-1 neatly explains this discrepancy: when even a minority of the Salmonella population is capable of invasion and the inflammatory response that it evokes, invasion-deficient bacteria benefit and are thus able to survive and proliferate within the animal intestine. Indeed, only a minority of SPI-1-expressing bacteria are required to support the growth of the Salmonella population as a whole, in accordance with the merely mild inflammatory response elicited in chickens and the lack of clinical disease in the species.
Previous work has also suggested that HilD activity is reduced at 42°C, the normal body temperature of chicks, resulting in a 40-fold repression of regulated genes, including a reduction in the expression of SPI-1 (17). We show unequivocally here that SPI-1 expression does occur in vivo at this temperature, with populations of Salmonella expressing SPI-1 identified throughout the intestines of live chicks. Furthermore, the proportion of SPI-1 expression we observed (12.6%) agrees with previous data detailing the proportion of SPI-1 expressed by S. Typhimurium in the mouse (Fig. 2) (12). Thus, the chick intestine provides conditions that permit SPI-1 expression, even though the consequences of infection are very different from those in mammals and SPI-1 expression is, in turn, required for the efficient colonization of the intestinal lumen.
We noted considerable variation in SPI-1 expression in the chick intestine. This variation included both the proportion of the Salmonella population expressing SPI-1 and whether that expression was monophasic or biphasic. These variations may represent interanimal variability, where the host environment differs enough among individual chicks to produce different bacterial outcomes. This variation may also represent heritable changes in the bacterial population, as within-host bacterial evolution can occur even over this short time of infection (2 or 9 days) (13, 30). Despite inherent variation, we observed measurable SPI-1 expression in both the ceca and colons and found that this expression facilitated Salmonella proliferation.
When we infected chicks with wild-type bacteria and ΔSPI-1 mutants together, but in different ratios, we determined that ΔSPI-1 mutants colonized as well as the wild type in the intestine. This finding suggests that in birds inoculated at 4 days posthatch, an age when nontyphoidal Salmonella strains establish robust colonization of the cecum, the wild-type strain supports growth of the mutant. However, in previous work by Dieye et al. (36), ΔSPI-1 mutants of S. Typhimurium exhibited a colonization defect in the intestine in the presence of the wild-type organism in chicks inoculated at a later time, 7 days posthatch. Differences in experimental approach, including the inoculated serotype, may contribute to the different findings in our results and earlier work. It is, however, possible that the benefit of SPI-1 expression, and the subsequent invasion of the intestinal epithelium, is manifested only transiently in chicks soon after hatching. Such an effect thus might occur prior to the establishment of a complete intestinal microbiota and could explain the resistance of older chickens to Salmonella infection. Additional studies will be required to determine the conditions within the intestine of the chicken that predispose it to colonization.
As SPI-1 aids Salmonella colonization of both birds and mammals, it represents an attractive target for inhibition. Treatments that reduce SPI-1 expression do not inhibit bacterial growth, as is the case for traditional antibiotics. Quite the opposite; inhibition of SPI-1 reduces the metabolic burden, and consequently, bacteria are less likely to develop resistance to inhibitors of SPI-1 expression (11). This strategy of targeting virulence rather than viability is considered a promising approach to the development of alternative treatments that minimize the potential for development of resistance (31). This tactic represents a useful approach to curbing the foodborne transmission of Salmonella.
MATERIALS AND METHODS
Bacterial strains and construction.Table 1 describes the strains used in this study. For initial fluorescent-strain construction and in vitro studies, Salmonella Typhimurium 14028s (ATCC) was used. phoN was chosen for integration of the constitutive fluorescent protein, as disruptions in the gene do not affect bacterial virulence (18, 32). Construction of ΔphoN::BFP-cat (BFP expressed under the control of a constitutive promoter in the place of the phoN gene, followed by the cat gene conferring chloramphenicol resistance) and sicA→gfp-cat (GFP expressed under the control of the sicA promoter, followed by the cat gene conferring chloramphenicol resistance) is described in the supplemental material. To create one strain bearing both fluorescent proteins, the cat resistance gene was removed from the chromosome of the sicA→gfp-cat construct by recombination (33), allowing subsequent P22 transduction of ΔphoN::BFP-cat. For mouse studies, fluorescent fusions were moved similarly by transduction into a spontaneous streptomycin mutant of S. Typhimurium 14028s (31). An isogenic ΔphoN::kan mutant was also constructed and used as the wild-type strain in mouse competition. Salmonella Enteritidis strain MD15 (a gift from Devendra Shah [28, 29]) was the background for all strain variants employed in chicken studies. To generate a nalidixic acid-resistant isolate, 100 μl of a log-phase culture was plated on LB agar containing 100 μg/ml nalidixic acid. Following overnight incubation, a spontaneously resistant colony was isolated and purified. For other gene deletions or fluorescent-protein integration, all constructs were generated in S. Typhimurium 14028s and then transduced into S. Enteritidis by phage P22.
Strains used in this study
Mouse propagation and infection.All mouse experiments were conducted in accordance with IACUC protocol 2012-0074 at Cornell University. Female C57BL/6 mice 5 to 7 weeks of age were obtained from Jackson Laboratory. The mice were given 20 mg of streptomycin orally 24 h prior to infection by Salmonella. For infection, Salmonella was grown with shaking overnight in LB broth and then washed and resuspended in phosphate-buffered saline (PBS) at an approximate density of 2 × 1010 CFU/ml. Each mouse received a 50-μl inoculum orally. At 2 days postinfection, the mice were euthanized by CO2 asphyxiation, and tissues were processed for colonization and FACS as described below.
Chicken propagation and infection.All chick experiments were conducted in accordance with IACUC protocol 2012-0074 at Cornell University or protocol 2013-0263 at Texas A&M University. White leghorn eggs were obtained from Charles River or the Cornell poultry farm. The eggs were set and hatched according to suppliers’ instructions. Briefly, the eggs were incubated in an egg incubator (CQF Manufacturing Co.) at 38°C and 60 to 65% humidity for 21 days. The eggs were periodically rotated during the first 18 days and then moved to the hatching tray for the last 3 days prehatch. Within 24 h of hatching, the chicks were moved to brooders kept at 32 to 35°C. The chicks were fed a diet of antibiotic-free feed (Harlan Teklad, Madison, WI) ad libitum. At 4 days posthatch, the chicks were infected with Salmonella. For this inoculum, a single colony of Salmonella was inoculated into LB broth and grown with shaking overnight. The culture was then centrifuged at 4,000 × g for 15 min, the supernatant was removed, and the bacterial pellet was resuspended in one-fifth the original volume of PBS. Each chick was given 100 μl by oral gavage (approximately 1 × 108 to 2 × 108 CFU/chick), and serial dilutions of inocula were plated for back-calculating inoculum density.
Bacterial enumeration and fixation.At 2 or 9 days postinoculation, chicks were euthanized by CO2 asphyxiation, and CFU enumeration was conducted to determine bacterial colonization of the chicks. Spleens and livers were removed, weighed, and homogenized in 1 or 3 ml PBS. Serial dilutions of homogenates were plated for CFU enumeration. For intestines, sections of the ilea, ceca, or colons were isolated and weighed. These sections were cut longitudinally and vortexed in two consecutive washes with 5 ml PBS to liberate the contents. The washes were combined, and serial dilutions were plated for CFU enumeration. CFU counts were normalized to the weight of the excised tissue and inoculum input and then multiplied by an arbitrary factor of 109 to yield data points above the x axis before being log transformed. For competitive indices, the number of CFU per organ for the mutant was divided by that of the wild type and then log transformed. When plating yielded no colonies, the limit of detection (which varies slightly between samples, depending on the organ weight) is reported and indicated in each figure.
Bacterial fixation and analysis by FACS.At the time of intestinal content isolation, a portion of the intestinal washes was filtered through a 5 μM nylon filter (Target2), removing debris while allowing bacteria to pass through. The filtrate was centrifuged at 4,000 × g for 10 min at 4°C, and the supernatant was removed. The pellet containing bacteria was resuspended in 4% paraformaldehyde in 1× PBS and turned at 4°C for 30 min to fix bacteria. The bacteria were centrifuged as before, the fix buffer was removed, and samples were resuspended in PBS and kept at 4°C until analysis by FACS. The fixed, filtered intestinal contents were analyzed on either a BD FacsAria Custom or BD Fortessa. The Salmonella population was identified based on BFP fluorescence (excitation, 407 nm; emission, 450 nm with 50-nm bandpass), and GFP signal (excitation, 488 nm; emission, 530 nm with 30-nm bandpass) for the population was measured as a readout for SPI-1 expression. Gates for BFP- and GFP-positive versus -negative populations were based on in vitro culture controls and background fluorescence of intestinal material from uninfected animals (see Fig. S4, S8 to S10, and S14 to S16 in the supplemental material). Analysis of results was conducted with FCS Express 5 software.
Histopathology.At the time of necropsy, a section of cecum was isolated from each chick and immediately preserved in 10% neutral-buffered formaldehyde fixative. Tissues were embedded and sectioned, and a board-certified veterinary pathologist blinded with regard to the experimental design scored inflammation based on 8 criteria (described in Fig. S17). Total histopathology scores (rather than individual criteria) were compared in statistical analyses.
Statistics.JMP 11 Pro was used for all statistical analyses. For single-strain inoculations, CFU counts were normalized to the tissue weight and inoculum inputs and then log transformed and compared by either Student's t test (for comparisons between only two data sets) or Dunnett’s method (for comparison of at least two experimental groups to a control data set). For in vivo coinoculations of wild-type and ΔSPI-1 mutant strains, data were normalized to the inoculum input and then log transformed and compared by matched-pairs analysis.
ACKNOWLEDGMENTS
This work was supported by grant no. 2014-67015-21697 from the NIH/USDA-NIFA Dual Purpose with Dual Benefit Program (to C.A.). This material is based upon work that is supported by the National Institute of Food and Agriculture, U.S. Department of Agriculture, under award number 2016-67012-25184 (to C.R.E.).
FOOTNOTES
- Received 28 June 2018.
- Returned for modification 16 July 2018.
- Accepted 24 October 2018.
- Accepted manuscript posted online 5 November 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00503-18.
- Copyright © 2018 American Society for Microbiology.
