Introduction
This paper is based in part on a keynote presentation given (by REI) to the Bacterial Pathogenesis Section at the 2014 Conference of Research Workers in Animal Diseases.
Salmonella enterica is a gram negative, non-lactose fermenting, facultative anaerobic rod in the family Enterobacteriaceae. It is a frequent cause of diarrhea and can be found in most species of animals. S. enterica is a facultative intracellular bacterium. It has an intracellular lifestyle in animals and this allows S. enterica to evade many of the host's innate and adaptive immune clearance mechanisms (Broz et al., Reference Broz, Ohlson and Monack2012). S. enterica can be found colonizing the ileum, cecum, and colon of pigs. Studies on the pathogenesis of diarrheal disease caused by S. enterica have shown that this microbe invades M-cells in the ileum of the small intestines (Carter and Collins, Reference Carter and Collins1974; Hohmann et al., Reference Hohmann, Schmidt and Rowley1978; Jones et al., Reference Jones, Ghori and Falkow1994; Penheiter et al., Reference Penheiter, Mathur, Giles, Fahlen and Jones1997) as well as epithelial cells (Takeuchi, Reference Takeuchi1967). S. enterica also is frequently found in mesenteric lymph nodes and spleens. In severe cases of salmonellosis, S. enterica can enter the blood stream and spread systemically causing sepsis. There are more than 2500 known serovars of S. enterica. Host adapted serovars, like the swine pathogen S. enterica serovar Choleraesuis, are most likely to cause systemic infections and cause enteric fever. To invade epithelial cells and maintain its intracellular lifestyle, S. enterica has a well-coordinated set of virulence genes many of which are involved in the synthesis of type three secretion systems (TTSS) (Troisfontaines and Cornelis, Reference Troisfontaines and Cornelis2005; Coburn et al., Reference Coburn, Sekirov and Finlay2007). TTSSs are known for their abilities to inject effector proteins into target cells. For example, one TTSS encoded in the Salmonella pathogenicity island 1 (SPI1) (Lostroh and Lee, Reference Lostroh and Lee2001) mediates the initial invasion of S. enterica in to epithelial cells and is essential for disease.
Other host-adapted serovars of S. enterica, including Typhi (human) and Dublin (cattle), cause severe disseminated disease. The non-host adapted serovars of S. enterica Typhimurium and Enteritidis are the two most common serovars causing food borne illnesses (CDC, 2013). The Centers for Disease Control has estimated that S. enterica is responsible for over 1 million cases of food-borne illness per year in the USA resulting in over 19,000 hospitalizations and 378 deaths (Scallan et al., Reference Scallan, Hoekstra, Angulo, Tauxe, Widdowson, Roy, Jones and Griffin2011). S. enterica has been ranked as the leading cause of food borne disease as measured by the combined cost of illness and Quality Adjusted Life-Year (Batz et al., Reference Batz, Hoffmann and Morris2011). It has been estimated that the economic losses due to salmonellosis in humans in the USA exceeds $3.5 billion per year (Batz et al., Reference Batz, Hoffmann and Morris2011). Pork is frequently associated with food borne illnesses caused by S. enterica in humans (Bryan, Reference Bryan1988; Bean and Griffin, Reference Bean and Griffin1990; EFSA, 2012; Brendan et al., Reference Brendan, Patricia, Dana, Kelly and Shua2013).
It is well known that the prevalence of S. enterica in most livestock species is high including pigs, cattle, and chickens and in most cases these animals are healthy carriers. Estimates from the National Animal Health Monitoring System (NAHMS) (Anonymous, 2009) put the farm prevalence of S. enterica in pigs at 50% in the USA. Other studies have shown that the prevalence of S. enterica is even higher (Bahnson et al., Reference Bahnson, Damman, Isaacson, Miller, Weigel and Troutt2006a). In naturally infected animals, most never experience overt disease. It is well known that naïve pigs can be infected with S. enterica early in life and consequently become long-term carriers of this microbe (Williams and Newell, Reference Williams and Newell1970). Wood et al. (Reference Wood, Pospischil and Rose1989) demonstrated that pigs orally challenged with S. enterica serovar Typhimurium (S. Typhimurium) shortly after weaning became long-term carriers of the challenge strain but when pigs were 28 weeks of age, only low numbers of the challenge strain could be found in their feces. Wood et al. (Reference Wood, Pospischil and Rose1989) also looked for internal reservoirs of the challenge S. Typhimurium strain and found that it was most consistently found in the tonsils. However, detection of the challenge strain was sporadic and currently the location of the animal's natural reservoir remains unknown. Most experimentally challenged pigs do experience diarrhea, though the disease usually is mild and self-limiting. However, most become persistently infected (carriers) yet remain clinically healthy. While S. enterica is generally thought to be a pathogen, the emerging picture suggests that S. enterica may be a component of the normal intestinal microbiota. Based on the broad distribution of S. enterica among animals and the apparent lack of salmonellosis in these animals we suggest that S. enterica can be both a pathogen and a commensal microbe.
The goal of this paper is to review our data describing S. enterica persistence in swine and to then summarize a unique phenotypic phase variation process that may explain how S. enterica can be a pathogen and a commensal. In addition, we will describe data demonstrating an interaction between S. enterica and another pig pathogen, Lawsonia intracellularis, how L. intracellularis may contribute to the development of a carrier state, and describe the alteration to the composition of the intestinal microbiota when pigs carry S. enterica.
Additional evidence that pigs can be persistent carriers of S. Typhimurium
S. Typhimurium is the most common serovar found in pigs that causes food borne salmonellosis (CDC, 2013). We wanted to confirm that this serovar could persistently infect pigs. We orally challenged pigs and followed the number of pigs shedding the challenge strain and the amount of cells actually being shed. In addition, we wanted to know whether stressing pigs could increase shedding of the challenge strain. To do this we challenged pigs with a nalidixic acid-resistant mutant of S. Typhimurium strain 798 after weaning (4-weeks of age) and collected fecal samples at various intervals to both detect, which pigs were shedding the challenge strain and in what quantities (Isaacson et al., Reference Isaacson, Firkins, Weigel, Zuckermann and Dipietro1999b). A two-step enrichment protocol was used for detection (Bahnson et al., Reference Bahnson, Damman, Isaacson, Miller, Weigel and Troutt2006a, Reference Bahnson, Fedorka-Cray, Ladely and Mateus-Pinillab) and a three tube most probable number analysis test was used to quantify the challenge strain in feces. The final stage of the culturing process was by plating the enrichments on XLT4 agar containing nalidixic acid. One-week post challenge S. Typhimurium strain 798 was detected in the feces of approximately 75% of the pigs (group size was 16 pigs) (Fig. 1). However, over time the number of pigs actively shedding the challenge strain decreased. By 4 months post challenge only 4% of pigs were shown to be shedding S. Typhimurium strain 798. Likewise, the quantity of S. Typhimurium strain 798 in feces decreased over time. One-week post challenge we detected approximately 109 S. Typhimurium cells per gram of feces. But at 4 months post challenge the quantity of with S. Typhimurium strain 798 in feces decreased to a mean concentration of less than 1 colony forming unit per gram of feces (Fig. 2). These data demonstrated that there was a slow but continual clearance of S. Typhimurium from the pig feces. However, based on the work of Williams and Newell (Williams and Newell, Reference Williams and Newell1970) we wondered whether the challenge organism was really cleared from these animals or if it was only cleared from their feces? Therefore, we stressed the pigs by subjecting them to feed withdrawal (0, 6, 12 or 24 h prior to transport) followed by a 2-h ride in a truck (group size was 15 pigs per group). After the truck ride, the pigs were necropsied and contents of their ceca were cultured to detect the infecting strain of S. Typhimurium (Fig. 3). The pigs that did not receive the feed withdrawal process showed a slight increase in the prevalence of S. Typhimurium strain 798. But pigs that were not fed for 24 h exhibited a large increase in shedding: almost 80% of the pigs were now positive for S. Typhimurium strain 798. The interpretation of that data was that the majority of pigs in the study had remained colonized with S. Typhimurium strain 798 even though they were not actively shedding this microbe in their feces 4 months post challenge. The stress associated with feed withdrawal and transporting them was sufficient to cause them to return to a shedding state. The exact compartment(s) where the challenge strain was hiding remains unknown. Based on the work of Wood et al., a likely reservoir was the tonsils (Wood et al., Reference Wood, Pospischil and Rose1989). The reason that cecal contents were collected at necropsy rather than feces was because pigs that had been in the 24 feed withdrawal group did not have feces that could be collected from their rectums. Whether cecal contents and feces represent an essentially identical sample is not known and thus could have biased the outcome. We know that feces and colonic tissues are not the same in terms of physical properties (nutrient availability, water, pH, etc.). There are three hypotheses that might explain the increased shedding after feed withdrawal and transport. The first is that the challenge strain of S. Typhimurium strain 798 was present in all pigs and feed withdrawal coupled with the stress of transportation caused a reactivation of the challenge strain and increased shedding. The second hypothesis is that we would have detected the challenge organisms in cecal contents prior to feed withdrawal and transport had we assayed them. Thus, the difference was due to sample differences. The third hypothesis is that relatively few pigs actually shed the S. Typhimurium challenge strain at the time of feed withdrawal, but that during transport stress those few pigs that were shedding the challenge strain provided the inoculum for the other pigs that began to shed S. Typhimurium after transport. This conclusion is consistent with the work of Hurd et al. (Reference Hurd, Mckean, Griffith, Wesley and Rostagno2002) where they noted that commercial pigs held in lairage prior to slaughter began to shed serotypes of S. enterica not seen in those animals during on farm sampling. Their assumption was that the pigs held in lairage pens picked up additional strains of S. enterica in this new environment and rapidly became shedders of these new strains. Supporting evidence that a brief exposure to S. enterica in the environment could rapidly result in shedding is based on the work of Fedorka-Cray et al. (Reference Fedorka-Cray, Kelley, Stabel, Gray and Laufer1995). They demonstrated that S. Typhimurium could be detected in the intestines of pigs within 3 h post challenge by the intranasal route in pigs that had their esophagus surgically ligated. Presumably the S. Typhimurium gained entry in to the blood stream via the respiratory tract and then homed to the gut lumen. Thus, shedding of S. Typhimurium strain 798 within the 2 h transport time frame is plausible. However, an important difference between our experiment and the one using esophagotomized pigs is that the inoculum used in the esophagotomized pigs was (109), which is quite high and it is unlikely that a similar concentration of S. Typhimurium would be found in the truck during transport. In addition, we found that the increase in the number of pigs positive for the challenge strain of S. Typhimurium strain 798 correlated with the length of feed withdrawal suggesting that the degree of stress related to duration of feed withdrawal is an important contributor in the conversion of apparent S. Typhimurium negative pigs to S. Typhimurium positive pigs. It is likely that the resumption of shedding of S. enterica by carriers is due to many factors including stress induced as well as exposure to S. enterica contamination in different environments.
The number of pigs actively shedding the challenge strain was determined at the indicated times using feces collected from pigs orally challenged with a nalidixic acid resistant S. Typhimurium strain 798 using a double enrichment (tetrathionate broth and Rappaport-Vassiliadis R10 Broth) followed by plating on XLT-4 agar containing nalidixic acid.
Quantitative measurement of a nalidixic acid-resistant S. Typhimurium strain 798 in pig's feces at the indicated times post oral challenge. Quantitative measurement utilized a most probable number (MPN) protocol.
Percentage of pigs actively shedding the nalidixic acid-resistant S. Typhimurium strain 798 approximately 4 months post challenged. Prior to sample collection pigs were subjected to feed withdrawal (0, 6, 12 or 24 h) and then loaded onto a truck with separate partitions for each feed withdrawal group and transported for 2 h prior to collection of cecal contents. Challenge strain positive pigs were determined after using a double enrichment (tetrathionate broth and Rappaport-Vassiliadis R10 Broth) followed by plating on XLT-4 agar containing nalidixic acid.
S. Typhimurium can grow in two phenotypic phases
In our studies to understand how S. Typhimurium was able to enter a carrier state within pigs, we found that one strain, S. Typhimurium strain 798, was able to grow in one of two phenotypes (Isaacson and Kinsel, Reference Isaacson and Kinsel1992). S. Typhimurium strain 798 was initially isolated from a pig with diarrhea. But subsequent work with this strain showed that it could cause persistent non-clinical infections of pigs (Wood et al., Reference Wood, Pospischil and Rose1989). Initially, the observation that S. Typhimurium strain 798 grew with two phenotypes was based on the ability of S. Typhimurium strain 798 to grow in two different and distinct colony morphologies when grown on blood agar plates: small, non-mucoid, cohesive colonies compared with larger, mucoid, non-cohesive colonies. Cells from each colony type were tested for their abilities to attach to enterocytes isolated from baby pigs. The larger colonies were shown to be highly adhesive while the smaller colonies did not attach to enterocytes (Isaacson and Kinsel, Reference Isaacson and Kinsel1992). When colonies of either phenotype were picked and re-plated, a small number of colonies would arise that appeared to be of the other phenotype. The larger the numbers of total colonies on the plate the more likely that colonies of the other phenotype would be present. When these cells were tested for adhesion to enterocytes, they either became adhesive (if they originated from the non-adhesive phenotype) or become non-adhesive (if they originated from the adhesive phenotype). Repeating this process always resulted in the same result indicating that the two phenotypes could at low frequency transition to the other phenotype.
Further characterization of cells in the two phenotypes revealed many other differences, which are summarized in Table 1. Based on their adhesive phenotype to porcine enterocytes we designated the phenotypes as non-adhesive phenotype or adhesive phenotype. Data demonstrated that type 1 fimbria was responsible for adhesion to enterocytes and type 1 fimbriae were only produced by cells in the adhesive phenotype (Isaacson and Kinsel, Reference Isaacson and Kinsel1992; Althouse et al., Reference Althouse, Patterson, Fedorka-Cray and Isaacson2003). Further investigations of cells in the adhesive phenotype showed that they also were highly invasive and once inside a mammalian cell were able to survive (Isaacson and Kinsel, Reference Isaacson and Kinsel1992). This is a clear link to virulence and the intracellular lifestyle of S. enterica. Cells in the non-adhesive phenotype were much less invasive and once internalized were rapidly killed. In studying the differences between cells of the two phenotypes, we noted that all traits associated with a phenotype were coordinately regulated. That is all were either expressed or not expressed based on their phenotype and when a phenotypic variant was selected by plating on agar, all of the traits were either expressed or not expressed consistent with the phenomenon called phase variation (reviewed by (Henderson et al., Reference Henderson, Owen and Nataro1999)).
Some characteristics of the two phenotypes of S. Typhimurium strain 798
Because we knew that some of the co-expressed genes associated with the adhesive phenotype would encode membrane-associated proteins (e.g. type 1 fimbriae), we used the transposon TnphoA to create a library of mutants that were in genes that encode membrane proteins. TnphoA encodes alkaline phosphatase (PhoA). If TnphoA inserts into a gene causing an in-frame gene fusion and if the protein encoded by that gene is exported from the cell by virtue of having a leader peptide, an active extra-cellular alkaline phosphatase will be produced. If cells are then plated on an agar medium containing 5-bromo-4-chloro-3-indolyl phosphate (XP), a substrate for alkaline phosphatase, colonies will be blue. Using the TnphoA library, we showed that the short O-antigen trait was related to differential expression of the gene rfaL, which is an O-antigen ligase (Kwan and Isaacson, Reference Kwan and Isaacson1998). Expression of the PhoA::RfaL fusion protein was shown to be phase variable. In addition, cells expressing the PhoA::RfaL fusion protein retained all features of the adhesive phenotype except the long O-antigen. Conversely, cells that did not express the PhoA::RfaL fusion protein retained the traits associated with the non-adhesive phenotype.
Because of the diverse number of differences between the two phenotypes, we decided to determine the complete repertoire of coordinately controlled genes in the two phenotypic phases. To do this, RNAseq was used to compare global gene expression in cells of both phenotypes (Patterson et al., Reference Patterson, Borewicz, Johnson, Xu and Isaacson2012). Eighty-three genes were shown to be up regulated in cells in the adhesive phenotype, while 31 genes were shown to be down regulated. Many of the up-regulated genes encode known virulence factors. In particular, all genes in the Salmonella pathogenicity island SPI1 were shown to be up regulated. The genes in this pathogenicity island are known to encode a TTSS that is involved in invasion of epithelial cells including enterocytes and M-cells (Wallis and Galyov, Reference Wallis and Galyov2000; Lostroh and Lee, Reference Lostroh and Lee2001). In addition to the SPI1 genes, the genes that encode the metabolic pathway for propanediol utilization were up-regulated. The propanediol pathway has been shown to be important for survival in macrophages (Klumpp and Fuchs, Reference Klumpp and Fuchs2007). Several genes encoded in SPI2 (sopA, sopB, sopD, and sopE) also were up regulated. These genes are part of a second TTSS, one that is important in intracellular survival (Hensel, Reference Hensel2000).
Measuring the rate of phenotypic phase variation
When the two phenotypes of S. Typhimurium strain 798 were discovered, it was noted that cells could switch between the phenotypes. A rough estimate of the rate of transition was between 10−2 and 10−4 per cell per generation (Isaacson and Kinsel, Reference Isaacson and Kinsel1992). This rate was considered to be too frequent to be caused by a mutation and was more similar to the process of phase variation that is known to control expression of E. coli type 1 pili (Brinton, Reference Brinton1959; Eisenstein, Reference Eisenstein1981). During our genetic analysis of S. Typhimurium strain 798 we noticed that when cells of either phenotype were plated on Evans blue-uranine agar (EBU) colonies from cells in the non-adhesive phenotype were blue while colonies from cells in the adhesive phase were white. Furthermore, colonies of either phenotype yielded small numbers of colonies of the opposite color on EBU plates similar to the colony variations we previously saw on blood agar plates. Based on analysis of several traits, we confirmed that these colonies represented phase variants. Using EBU agar as a differential medium to more effectively identify colonies of the two phenotypes, we were able to more accurately estimate the frequency of phase variation from the non-adhesive to the adhesive phenotype (Patterson et al., Reference Patterson, Borewicz, Johnson, Xu and Isaacson2012). The rate was estimated to be 1.7 × 10−4 per cell per generation. However, while we were able to find the occasional colony that varied from the adhesive to the non-adhesive phenotype, the rate was too low on EBU agar plates to accurately measure the rate of phase variation. The limit of detection using EBU plates was approximately 10−5 per cell per generation and the actual rate was presumed to be lower.
In our studies to characterize the adhesive and non-adhesive phenotypes, we found that cells of the non-adhesive phenotype, when inoculated into Luria broth containing no NaCl had a very long lag phase (2.5–7.5 h). However, cells in the adhesive phenotype exhibited normal growth characteristics with a short lag phase. We reasoned that a colony containing cells from the adhesive phase would contain a small fraction of cells that had varied to the non-adhesive phase even though the number was too low to accurately estimate the number. If such a colony was inoculated into LB broth containing no NaCl and the antibiotic ampicillin so that any actively growing adhesive phase cells would be killed by ampicillin while any non-adhesive phase cells in the extended lag phase would remain viable. Using this ampicillin enrichment procedure, and incubating the culture for 2 h we were able to increase the sensitivity of the assay to measure phase variation and in doing so accurately measured the rate of phase variation from adhesive to non-adhesive phases at 1.5 × 10−6. Thus, in broth conditions (LB with no NaCl), S. Typhimurium strain 798 had a rate of phase variation from non-adhesive to adhesive phase that was about 100-times more frequent than the other direction. Because we propose that phenotypic phase variation is important for S. Typhimurium in animals it is possible that the rates could be quite different in pigs compared with broth cultures.
What controls phase variation in S. Typhimurium strain 798?
In an attempt to understand what controls phase variation, a library of transposon (Tn10d-Cam) mutants was created in cells of the adhesive phenotype and then transduced into non-adhesive phase cells using the bacteriophage P22 (Elliott and Roth, Reference Elliott and Roth1988). The goal was to identify mutants that had altered rates of phase variation compared with wild type cells. The transductants were plated on LB agar plates. Individual colonies were picked and plated on EBU plates. Colonies that appeared to phase vary to light-colored (adhesive phase) more frequently were identified. Insertion mutants were identified that had increased rates of phase variation. Several colonies exhibited a 10-fold increase in the rate of phase variation from the non-adhesive phenotype to adhesive phenotype. To confirm that the insertion mutation was the cause of the increased rate of phase variation and not a spurious result, the Tn10d-Cam insertion was transduced again into non-mutant, non-adhesive phase cells. All transductants demonstrated the 10-fold increase in the rate of phase variation from non-adhesive to adhesive phenotype. Sequence data from these insertion mutations indicated the transposon insertion sites were just upstream of the gene lrhA.
To confirm the association between phase variation and LrhA, a second genetic screen was used to identify mutations with alterations in phase variation. The screen used a novel two-color identification scheme. A fimA::TnphoA mutant that had been previously created was one part of the screen (Althouse et al., Reference Althouse, Patterson, Fedorka-Cray and Isaacson2003). FimA is the major subunit of type 1 fimbriae that we have shown to be involved in adhesion to pig enterocytes and is expressed as part of the adhesive phenotype (Althouse et al., Reference Althouse, Patterson, Fedorka-Cray and Isaacson2003). The fimA::TnphoA fusion was transduced into adhesive phase S. Typhimurium strain 798 cells using bacteriophage P22. Cells growing on agar medium containing XP yielded blue colonies if the PhoA::FimA fusion was expressed. A second mutation was introduced into either sigE or pefB using adhesive phase cells using the transposon Tn5lacZY. These two genes were selected because they encode known virulence factors and were expressed at high levels in cells of the adhesive phase but were not encoded near the type 1 fimbrial gene cluster. Rather they are encoded in SPI1. Cells expressing either sigE or pefB grew as red colonies when plated on agar containing 6-chloro-3-indolyl-β-D-galactoside (Red-Gal) because of the expression of β-galactosidase. Double mutants containing PhoA::fimA and either sigE::lacZY or pefB::lacZY were plated on agar containing XP and Red-Gal. If both gene fusions were expressed the colonies were purple. If neither of the two gene fusions were expressed, the colonies were white. The double mutants were subsequently mutagenized with the transposon Tn10d-Cam to simultaneously knockout expression of both fimA and SigE or pefB. A Tn10d-Cam transposon library was screened for white colonies. White colonies represented either knockouts of the phase variation process or were simply phase variants. Using EBU plates, we were able to identify colonies that contained phase variants because they would have the low percentage mix of blue and white colonies. Phase variants were eliminated. We then cloned and sequenced what appeared to be mutations that knocked out two genes simultaneously (Tn10d-Cam mutants) and were not phase variants and all were found to be in the gene lrhA. This is the same gene mutagenized in the non-adhesive phase cells that resulted in an increase in phase variation.
Using DNA sequencing, the sites of the transposon insertions again were shown to be located just upstream of the lrhA gene. Since the transposon insertions were located upstream of the start site (320 bp in two insertion sites and 1010 bp in two other insertion sites) we wondered if the mutation increased or decreased expression of lrhA. Previous studies in Salmonella and Escherichia coli identified insertions in similar sites upstream of lrhA (Cunning, and Elliott, Reference Cunning and Elliott1999). These mutations actually increased the expression of lrhA because they likely knocked out repression of lrhA. Defined S. Typhimurium mutants that were known to either increase or knock out expression of lrhA were obtained (Cunning and Elliott, Reference Cunning and Elliott1999). The defined mutations were moved into non-adhesive and adhesive phase cells and the rates of phase variation were calculated for each of the mutants. Strain 7637-8, which over expresses lrhA, was found to phase vary from non-adhesive phase cells to adhesive phase cells on EBU plates at a rate of 9.29 × 10−4 per cell per generation. This is almost 10-fold higher than the rate compared with non-adhesive phase cells (1.70 × 10−4 per cell per generation, P = 0.0003) and was similar to the rate of phase variation of the hyper variable mutants obtained in the first screen. The rate of phase variation for the null mutation in lrhA (7638-8) was 1.37 × 10−4 per cell per generation. This was not statistically different from wild type non-adhesive phase cells (P = 0.233). Mutants that over expressed lrhA were assayed to determine if they showed altered rates of phase variation. No changes were detected. Hence we concluded that over production of LrhA led to an increased rate of phase variation from non-adhesive to adhesive phenotype but did not alter phase variation from the adhesive phenotype to the non-adhesive phenotype.
LrhA has been shown to result in the degradation of the alternate RNA polymerase sigma factor, RpoS, in E. coli and possibly in Salmonella (Cunning and Elliott, Reference Cunning and Elliott1999). Therefore, we looked at the effect of an rpoS null mutation (Brown and Elliott, Reference Brown and Elliott1996) on the rate of phase variation. Strain 6168-8 (non-adhesive/rpoS::Tn10d-Cam) did increase the rate of phase variation from blue to light-colored colonies (non-adhesive phase to the adhesive phase) by approximately 10-fold. Strain 6168-9 (adhesive/rpoS::Tn10d-Cam) did not exhibit a change in the rate of phase variation from adhesive to non-adhesive phase cells. Oddly, and in contradiction to this result, based on our RNAseq data rpoS is expressed at an approximately 10-fold higher rate in non-adhesive phase cells compared with adhesive phase cells. However, if RpoS is not involved in phase variation from adhesive to non-adhesive phase cells the level of expression of rpoS in adhesive phase cells is irrelevant. Thus, as the data shows, LrhA reduces RpoS in non-adhesive phase cells and does so through a regulatory effect.
Is S. Typhimurium strain 798 unique in its ability to phase vary?
Since strain 798 is a clinical isolate capable of persistently infecting pigs but is not highly virulent in the BalbC mouse model of disease (oral LD50 ~ 109, Isaacson et al., Reference Isaacson, Argyilan, Kwan, Patterson and Yoshinaga1999a; Althouse et al., Reference Althouse, Patterson, Fedorka-Cray and Isaacson2003), we wanted to know whether other S. Typhimurium isolates undergo phenotypic phase variation and in particular if more highly virulent strains of S. Typhimurium were capable of phase variation between the adhesive and non-adhesive phenotypes. Two high virulence strains, SL1344 (Hoiseth and Stocker, Reference Hoiseth and Stocker1981; Garcia-Quintanilla and Casadesus, Reference Garcia-Quintanilla and Casadesus2011) and 14028 (ATCC® 14028™) were plated on EBU agar to see if blue colonies were present. These strains are normally light-colored on EBU. We used the ampicillin enrichment procedure to increase the likelihood of finding blue colonies (Patterson et al., Reference Patterson, Borewicz, Johnson, Xu and Isaacson2012). After enrichment blue colonies were isolated from SL1344 and 14028. The rates of phase variation from the adhesive phenotype to the non-adhesive phenotype were calculated and both were statistically the same for both strains compared with S. Typhimurium strain 798 (Table 2). We used blue colonies from both non-adhesive phase S. Typhimurium strain 14028 and SL1344 to measure the rates of phase variation back to the adhesive phase. Both strains reverted to the adhesive phase almost 100-fold more frequently compared to S. Typhimurium strain 798 (Table 2). Thus, phenotypic phase variation discovered in S. Typhimurium strain 798 also occurred in two additional and highly virulent strains. It is unknown if other serotypes also undergo a similar phenotypic phase variation process.
Rate of phenotype phase variation for three strains of S. Typhimurium
How phenotypic phase variation may contribute to the development of the S. enterica carrier pig
Based on the recognition that S. Typhimurium cells in the adhesive phenotype also produce an array of important virulence genes while cells in the non-adhesive phenotype do not, it is logical to believe that the adhesive phenotype cells are virulent while non-adhesive phenotype cells are not. Since S. Typhimurium cells appear to be able to transition from one phenotype to the other, we would assume that highly virulent strains would have a larger proportion of cells in the adhesive or virulent phenotype. When the two high virulence strains of S. Typhimurium (SL1344 and 14028) were analyzed for rates of phenotypic phase variation, the rates were consistent with these strains being most likely to be found in the adhesive or virulent phenotype (Table 2). That is, the rate of phase variation was four orders of magnitude higher going from the non-adhesive phenotype to the adhesive phenotype. The LD50 values for these two strains in the BalbC mouse model are between 104 and 105 (Hoiseth and Stocker, Reference Hoiseth and Stocker1981). On the other hand, based on the measured rates of phenotypic phase variation, S. Typhimurium strain 798 is less likely to be in the adhesive or virulent phase compared with S. Typhimurium strains SL1344 or 14028. S. Typhimurium strain 798 also is much less virulent having a mouse LD50 > 109. We would expect that if all three S. Typhimurium strains were used to challenge pigs, that the two high virulence strains would cause typical salmonellosis, perhaps severe disease, while we already know that S. Typhimurium strain 798 causes a milder disease. Because of the differences in virulence, S. Typhimurium strain 798 might be more likely to cause asymptomatic, persistent infections. Our experience with this strain is that it indeed does cause mild diarrhea that lasts no longer than 24 h and then causes persistent but asymptomatic infections of pigs (Isaacson et al., Reference Isaacson, Firkins, Weigel, Zuckermann and Dipietro1999b).
A model of how phase variation contributes to disease is shown in Fig. 4. Cells in the adhesive phenotype produce an array of virulence factors and would be expected to attach to M-cells and villus absorptive enterocytes in the intestines and then invade into these intestinal cells. S. Typhimurium in the M-cells would escape into the sub epithelial layer and enter dendritic cells, macrophages, and/or neutrophils but because of the expression of virulence factors encoded in SPI2 (and other virulence genes), would survive and grow intracellularly. These phagocytes would then migrate to local mesenteric lymph nodes where the S. Typhimurium would eventually be controlled. S. Typhimurium within the enterocytes also would proliferate and ultimately spread to other enterocytes, stimulate an inflammatory response and thus be exposed to inflammatory neutrophils in the intestinal tissues and the gut lumen, which has been shown to enhance growth of S. Typhimurium (Stecher et al., Reference Stecher, Robbiani, Walker, Westendorf, Barthel, Kremer, Chaffron, Macpherson, Buer, Parkhill, Dougan, Von Mering and Hardt2007; Winter et al., Reference Winter, Thiennimitr, Winter, Butler, Huseby, Crawford, Russell, Bevins, Adams, Tsolis, Roth and Baumler2010).
Model of the interactions of S. enterica (red rod shaped figures) with enterocytes in the intestinal tract (light blue), their brush borders, and M-cells (light green). The two cartoons represent the expected interactions that occur when the S. enterica are in the two phenotypic phases.
On the other hand, S. Typhimurium cells in the non-adhesive phenotype would not attach to enterocytes because, among other things, they do not produce type 1 fimbriae (Althouse et al., Reference Althouse, Patterson, Fedorka-Cray and Isaacson2003). These cells would still enter M-cells because M-cells are constantly sampling the environment in the lumen and because these cells also produce other adhesins, but would ultimately be killed by resident dendritic cells or macrophages found under the M-cells. S. Typhimurium cells remaining in the lumen of the intestines would be cleared by peristalsis and eliminated in feces.
Based on this information, we hypothesize that the development and maintenance of a persistent carrier state is related to the proportion of S. Typhimurium cells in a population in the two phenotypes based on rates of phenotypic phase variation. If the rate of phenotypic phase variation from the non-adhesive and adhesive phenotypes was modulated such that there was a sufficient number of adhesive phase cells to maintain a mucosal and/or intracellular population but was too low to cause disease, a carrier state would develop. After a pig is exposed to S. Typhimurium, cells in the non-adhesive phase would be rapidly cleared from the lumen of the intestines because they did not attach to intestinal cells, or be killed by phagocytic cells. The remaining S. Typhimurium cells would be those in the adhesive phenotype. Because phenotypic phase variation is a dynamic process, the remaining cells in the adhesive phenotypic phase would continually spin off cells in the non-adhesive phenotype. The actual number of cells that varied to the non-adhesive phenotype is related to the actual rate of phase variation in vivo. However, since non-adhesive phase cells would be rapidly cleared from the lumen or killed intracellularly after the initial exposure period, the only non-adhesive phase cells present would be those derived from the remaining adhesive phase cells and these too would be rapidly cleared or killed. Thus, depending on the actual in vivo rates of phenotypic phase variation it would be possible to maintain a small but consistent population of S. Typhimurium in pigs. This population would not be large enough to cause disease and thus pigs would remain asymptomatic but remain infected. It should be noted that while the hypothesis stated is related to pig growth, we have not measured rates of phase variation in vivo to confirm the hypothesis.
S. enterica shedding is enhanced by co-infection with Lawsonia intracellularis
Other factors that may contribute to the levels of S. enterica shedding prevalence, shedding levels, and persistence include interactions of S. enterica with other members of the intestinal microflora. For example, recent work by Beloeil et al. (Reference Beloeil, Fravalo, Fablet, Jolly, Eveno, Hascoet, Chauvin, Salvat and Madec2004) proposed that co-infection of pigs with pathogens such as Lawsonia intracellularis or porcine respiratory and reproductive syndrome virus might predispose them to shed S. enterica. In their study, 105 French pork production farms were sampled and a statistically significant association (Odds Ratio 3.2, 90% Confidence Interval 1.4 to 7.2) between infections with L. intracellularis and carriage of S. enterica was found. That work is consistent with the hypothesis that L. intracellularis interacts with S. enterica and/or other members of the gut microbiome and that these interactions lead to increased colonization and shedding of S. enterica. L. intracellularis is the cause of porcine proliferative enteritis.
To experimentally determine if L. intracellularis had any effects on the levels of S. enterica in pigs, we performed a challenge study. Groups of 5 week old pigs were challenged with L. intracellularis, S. Typhimurium strain 798, or both. A fourth group of pigs served as non-challenged controls. At various times, pigs were euthanized and the quantity of S. Typhimurium in the jejunum, ileum, cecum, and colon of pigs was determined using a most probable number analysis. The results are shown in Fig. 5. When the pigs were 7 weeks of age (1 week post challenge with S. Typhimurium and 2 weeks post challenge with L. intracellularis) the S. Typhimurium challenge strain was found in all pigs that had been challenged regardless of tissue site. The challenge strain was not found in any of the pigs that had not been challenged with S. Typhimurium strain 798. In jejunal and ileal tissues from pigs challenged with the nalidixic acid-resistant S. Typhimurium strain 798, the challenge strain was found at the highest levels in 7-week old pigs. Over the next 4 weeks the tissue levels dropped to ≤2 log10 regardless of whether the pigs also had been challenged with L. intracellularis. However, the culture results from the cecal and colonic tissues showed that tissues from pigs co-challenged with L. intracellularis did not show decreases in the overall concentration of the S. Typhimurium challenge strain and were statistically significantly higher at 11 weeks of age compared with pigs challenged only with S. Typhimurium (P < 0.05). The cecal levels of S. Typhimurium in the animals co-challenged with S. Typhimurium and L. intracellularis were 4 log10 higher (P < 0.05) compared with those only challenged with only S. Typhimurium while the colonic levels were 3 log10 higher (P < 0.05). In pigs challenged only with S. Typhimurium, a consistent decrease in the level of S. Typhimurium in the cecum or colon was observed and by 11 weeks of age all pig tissues contained ≤2 log10 S Typhimurium strain 798 per gram of tissue. This level of colonization is what we typically see after pigs are challenged with this strain of S. Typhimurium (Isaacson et al., Reference Isaacson, Firkins, Weigel, Zuckermann and Dipietro1999b).
Shedding at the indicated times of the nalidixic acid resistant S. Typhimurium strain 798 in pigs challenged only with this strain (■) or also challenged with L. intracellularis (•). Intestinal tissue samples were collected from the jejunum, ileum, cecum, and colon of pigs, the luminal surfaces scraped and the number of S. Typhimurium determined using most probable number (MPN) protocol employing a double enrichment (tetrathionate broth and Rappaport-Vassiliadis R10 Broth) followed by plating on XLT-4 agar containing nalidixic acid.
We also looked at the levels of L. intracellularis in the same tissues. Co-infection with S. Typhimurium strain 798 had no effect on the concentration of L intracellularis in any of the intestinal tissues.
The concept of pathogen super shedders has long been postulated and for Escherichia coli O157:H7, super shedders have been identified (Matthews et al., Reference Matthews, Low, Gally, Pearce, Mellor, Heesterbeek, Chase-Topping, Naylor, Shaw, Reid, Gunn and Woolhouse2006). However, for S. enterica the identification of super shedders has not occurred. Huang et al. (Reference Huang, Uthe, Bearson, Demirkale, Nettleton, Knetter, Christian, Ramer-Tait, Wannemuehler and Tuggle2011) showed that within a population of pigs orally challenged with S. Typhimurium, high and low shedders could be identified. However, this is unlikely to be the identification of super shedders because high shedding status is only for a short period of time (approximately 2 weeks). The observations that L. intracellularis can promote prolonged shedding of S. Typhimurium and at higher levels might represent a situation where a super shedder is created.
Does S. enterica change the composition of the microflora in the colon of pigs?
Pigs experimentally challenged orally with S. Typhimurium strain 798 have been shown to have altered compositions of their gut microbiota (the bacterial microbiome) (Borewicz et al., Reference Borewicz, Kim, Singer, Gebhart, Sreevatsan, Johnson and Isaacson2015). Commercial pigs that are naturally infected with S. enterica and that are shedding this organism also exhibit altered compositions of the fecal microbiomes (Borewicz et al., Reference Borewicz, Kim, Singer, Gebhart, Sreevatsan, Johnson and Isaacson2015). The challenged and naturally infected pigs at 10 weeks of age showed similar shifts in the microbiota composition, with significant changes in the relative abundance of Barnesiella, Pseudobutyrivibrio, Prevotella, Lactobacillus, Anaerobacter, Roseburia, Fastidiosipila, Campylobacter, and Succinivibrio. Because there were similar shifts in microbiome composition between experimentally challenged pigs and those naturally exposed to ‘presumably’ low doses of S. enterica, we hypothesize that the shifts were due to the presence of S. enterica and not an effect of a microbial composition shift making the pigs more susceptible to S. enterica. This further suggests that while it has been assumed that one function for normal microflora is to exclude pathogens, in the case of S. enterica this may not hold true. One of the microbial shifts observed in older experimentally challenged pigs is the increased levels of Akkermansia. While this observation was observed in a sample of colon from one pig at 22 weeks of age (Borewicz et al., Reference Borewicz, Kim, Singer, Gebhart, Sreevatsan, Johnson and Isaacson2015), it is known that Akkermansia does contribute to infection with S. enterica presumably by degrading mucin in the gut and increasing gut inflammation (Ganesh et al., Reference Ganesh, Klopfleisch, Loh and Blaut2013).
Acknowledgments
This work was supported by research grants #2002-35201-12542, #2007-35212-18046, and #2015-67017-23076 from the United States Department of Agriculture, National Institute for Food and Agriculture (NIFA) and by a grant from the University of Minnesota, College of Veterinary Medicine, Emerging Infectious Diseases Signature Program.
