Tissue-Specific Salmonella Typhimurium Gene Expression during Persistence in Pigs

Salmonellosis caused by Salmonella Typhimurium is one of the most important bacterial zoonotic diseases. The bacterium persists in pigs resulting in asymptomatic ‘carrier pigs’, generating a major source for Salmonella contamination of pork. Until now, very little is known concerning the mechanisms used by Salmonella Typhimurium during persistence in pigs. Using in vivo expression technology (IVET), a promoter-trap method based on ΔpurA attenuation of the parent strain, we identified 37 Salmonella Typhimurium genes that were expressed 3 weeks post oral inoculation in the tonsils, ileum and ileocaecal lymph nodes of pigs. Several genes were expressed in all three analyzed organs, while other genes were only expressed in one or two organs. Subsequently, the identified IVET transformants were pooled and reintroduced in pigs to detect tissue-specific gene expression patterns. We found that efp and rpoZ were specifically expressed in the ileocaecal lymph nodes during Salmonella peristence in pigs. Furthermore, we compared the persistence ability of substitution mutants for the IVET-identified genes sifB and STM4067 to that of the wild type in a mixed infection model. The ΔSTM4067::kanR was significantly attenuated in the ileum contents, caecum and caecum contents and faeces of pigs 3 weeks post inoculation, while deletion of the SPI-2 effector gene sifB did not affect Salmonella Typhimurium persistence. Although our list of identified genes is not exhaustive, we found that efp and rpoZ were specifically expressed in the ileocaecal lymph nodes of pigs and we identified STM4067 as a factor involved in Salmonella persistence in pigs. To our knowledge, our study is the first to identify Salmonella Typhimurium genes expressed during persistence in pigs.


Introduction
Non-typhoidal salmonellosis is one of the most important bacterial zoonotic diseases, yearly resulting in an estimated 155,000 deaths worldwide [1]. In European countries, Salmonella enterica subspecies enterica serovar Typhimurium (Salmonella Typhimurium) is the serovar most frequently isolated from slaughter pigs [2]. Porcine carcass contamination with Salmonella Typhimurium can largely be attributed to persistently infected pigs [3]. Transmission of Salmonella Typhimurium between pigs occurs mainly via the faecal-oral route. After ingestion by the pig, the bacterium will preferentially colonize its tonsils and ileum, where it adheres to the intestinal epithelium. This is followed by invasion and subsequent migration of the Salmonella bacterium to the underlying lymphoid tissues, like the ileocaecal lymph nodes, resulting in so called 'carrier status pigs' [4].
In the past, Salmonella interactions with hosts were largely examined in murine and avian models, while Salmonella Typhimurium behaviour in pigs is only poorly investigated. Due to the different and often hostile environments Salmonella must combat to successfully colonize one host, it is expected that the bacterium is equipped with a broad range of survival strategies, each one adapted to a certain biological niche. Therefore the bacterial genes involved in Salmonella survival in the tonsils might markedly differ from those essential for colonization of and persistence in for example the lymph nodes or the ileum in pigs [5].
The last decades, new techniques have been developed that allow high-throughput screening for genes important for microbial colonization and persistence in animal models [6,7]. Recently, using signature-tagged mutagenesis (STM), Salmonella genes were identified that were specifically expressed in the deleterious gastric environment of pigs [8]. Furthermore, genes involved in Salmonella virulence in pigs shortly after oral inoculation have been identified [9,10]. However, to date very little is known about the mechanisms employed by Salmonella Typhimurium during the persistent phase of infection of pigs. Because increasing evidence demonstrates that Salmonella Typhimurium behaves markedly different in various hosts, a thorough understanding of Salmonella pathogenesis in pigs is required to develop successful intervention strategies to cope with Salmonella infection of pigs.
Using the genome-wide approach of IVET, our goal was to identify genes specifically induced in porcine tonsils, ileum and ileocaecal lymph nodes at 3 weeks post oral inoculation as an indication of genes involved in Salmonella persistence in these 3 different organs of pigs.

Ethics statement
All animal experiments were approved by the ethical committee of the Faculty of Veterinary Medicine, Ghent University (EC2008/074; EC2009/021; EC2010/005 and EC2010/158 respectively).

Bacterial strains and growth conditions
Salmonella Typhimurium strain 112910a phage type 120/ad, isolated from a pig stool sample, was used as the wild type strain (WT). A spontaneous nalidixic acid resistant derivative of the wild type strain was used in the mixed inoculation in vivo experiments. Salmonella Typhimurium substitution mutants DpurA::kanR, DsifB::kanR and DSTM4067::kanR were constructed according to the one-step inactivation method described by [11] and slightly modified for use in Salmonella Typhimurium as described before [12]. Primers used to create the gene-specific linear PCR fragments are given as Supporting Information S1. For construction of the IVET pool, the DpurA deletion mutant was obtained from the DpurA::kanR substitution mutant, by excision of the kanamycin resistance cassette [11].
For oral inoculation of piglets, bacteria were cultured aerobically for 16 h at 37uC in 5 ml Luria-Bertani broth (LB; Sigma-Aldrich Chemie Gmbh, Steinheim, Germany) with the appropriate antibiotic(s), if required, included at the following concentrations: 20 mg/ml for nalidixic acid (Sigma-Aldrich); 100 mg/ml for kanamycin (Sigma-Aldrich) and 50 mg/ml for ampicillin (Sigma-Aldrich). The DpurA strain was grown in LB enriched with 1.35% adenine and 0.337% thiamine. The bacterial 16 h cultures were then centrifuged at 2300 g for 10 min at 4uC and pellets were washed in Hank's buffered salt solution (HBSS; Gibco Life Technologies, Paisley, Scotland). After subsequent centrifugation at 2300 g for 10 min at 4uC, pellets were resuspended in 5 ml HBSS and diluted to the appropriate concentration for oral inoculation. The actual number of viable Salmonella colony forming units (CFU) per ml inoculum was determined by plating 10-fold dilutions on brilliant green agar (BGA; Lab M Limited, Lancashire, UK), supplemented with the appropriate antibiotic(s) for selective growth of the inoculated strain.

Construction of IVET pool
The pIVET1 plasmid is a derivate of the suicide vector pGP704 and contains a promoterless synthetic operon of purA coupled to lacZY, preceded by a BglII restriction enzyme site. This plasmid also possesses a mobilization site (mob), which results in wide bacterial host transfer through conjugation and an ampicillin resistance gene (bla). The IVET fusion pool was constructed as earlier described for Salmonella enterica serovar Enteritidis [13]. In short, Salmonella Typhimurium genomic DNA was purified and subsequently digested with the Sau3AI restriction enzyme (New England Biolabs, Ipswich, England), resulting in a library of 1-4 kb overlapping genomic DNA fragments. These fragments were then cloned in the BglII (New England Biolabs) site of the pIVET1 plasmid, upstream to promoterless wild type copies of purA and lacZY. The plasmids were transferred to Escherichia coli DH5alpir by electroporation and approximately 100.000 different clones were pooled. pIVET1 fusion plasmids were then isolated using the Plasmid Midi Kit (Qiagen, Venlo, The Netherlands) and electroporated in the conjugative Escherichia coli SM10 lpir. Again, 100.000 different E. coli SM10 lpir clones were pooled. Fusion plasmids were then mobilized into the Salmonella Typhimurium DpurA strain by conjugation using the Escherichia coli SM10 lpir as a donor strain. Integration of the pIVET1 plasmid in the chromosome resulted in a single cross over and therefore did not disrupt the wild type locus of the gene. In this way, Salmonella Typhimurium DpurA transformants were obtained, in which the native promoter drove the purA-lacZY fusion, while the cloned promoter drove the expression of the wild type gene. Correct integration of the pIVET1 fusion plasmid in the genome was assessed by growth on MacConkey agar plates with 1% lactose, nalidixic acid, ampicillin, thiamine and adenine. The diversity of the IVET library was randomly tested by repeatedly purifying and subsequently sequencing colonies from MacConkey agar plates after conjugation between E. coli and DpurA Salmonella Typhimurium. Approximately 15.000 different IVET transformants were purified and pooled, resulting in the final IVET bank that was stored at 270uC until further use.

Promoter identification
The sequences that were cloned at the 59 site of the purA-lacZY fusion were identified by a modification of the PCR-based method of [14]. Genomic DNA of IVET fusion strains of interest was isolated using the QIAamp DNA Mini Kit (Qiagen), and completely digested with the NlaIII restriction enzyme (New England Biolabs). These DNA fragments were ligated to a 'Y-linker' using T4 DNA ligase (New England Biolabs). The linker sequences used to synthesize the 'Y-linker' are given as supplementary information. Prior to adding linker 1 to linker 2, the latter was phosphorylated at the 59 end using T4 polynucleotide kinase (New England Biolabs). The mixture of both linkers was heated to 95uC and slowly cooled down to room temperature, to form the 'Y-linker'. After ligation of the 'Y-linker' to the genomic DNA fragments, a PCR was performed using a purA primer, binding the purA sequence, and a primer binding the 'Ylinker' sequence (see supplementary information for primer sequences). The sequence between the purA and 'Y-linker' primer was thus sequenced. DNA sequences of IVET fusion strains that resulted in a band on gel electrophoresis were sequenced using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, CA 94404, USA) according to the manufacturer's guidelines. Sample sequences were determined using a 3100 Genetic Analyzer (Applied Biosystems). Identification of sequences of the cloned Salmonella promoters was done by BLAST analysis using the genome sequence of Salmonella Typhimurium LT2 (http://blast.ncbi.nlm.nih.gov/).

In vivo experiments
Four-week-old piglets (commercially closed line based on Landrace) from a serologically Salmonella negative breeding herd (according to the Belgian Salmonella monitoring program) were used in the in vivo trials. The piglets arrived at the facility 5 days before they were inoculated. Prior to inoculation, the Salmonellafree status of the multiple piglets used in the in vivo experiments was tested serologically and bacteriologically. For the serological test, blood samples were collected from all piglets and serum was isolated and subsequently analyzed using a commercially available Salmonella antibody test kit (IDEXX, Hoofddorp, The Netherlands). Faeces were collected and enriched in tetrathionate broth overnight at 37uC. This suspension was then plated on BGA, incubated overnight at 37uC and visually checked for growth of salmonellae. Only piglets that were serologically negative and negative at faecal sampling were used. The animals were housed in separate isolation units at 25uC under natural day-night rhythm in HEPA-filtered stables, with ad libitum feed and water.
A. Salmonella Typhimurium DpurA behaviour in piglets. To assess the suitability of a Salmonella Typhimurium purA mutant (DpurA::kanR) for use in the current IVET approach, we first determined whether this mutant was significantly attenuated compared to the wild type. For this purpose, 6 Salmonella-free piglets were orally inoculated with a mixture of approximately 2610 7 CFU Salmonella Typhimurium wild type and 2610 7 CFU Salmonella Typhimurium DpurA::kanR. Eight days post oral inoculation, the piglets were humanely euthanized and samples of the palatine tonsils, lymph nodes (ileocaecal, mesenteric and colonic), ileum and contents, colon and contents, caecum and contents, spleen, liver and faeces (1 sample per organ) were collected and the number of Salmonella bacteria was determined in each sample.
Prior to further processing, tissue samples were rinsed and cut to small pieces. Tissue and contents samples were then weighed and 10% (w/v) suspensions were made in buffered peptone water (BPW; Oxoid, Basingstoke, UK) after which the tissue material was homogenized with a Colworth stomacher 400 (Seward and House, London, UK). The homogenized samples were examined for the presence of bacteria by plating 10-fold dilutions on BGA with nalidixic acid or kanamycin for selective growth of the wild type Salmonella Typhimurium and the DpurA::kanR substitution mutant respectively. BGA plates were incubated for 16 h at 37uC. When negative at direct plating, samples were pre-enriched for 16 h in BPW at 37uC and enriched for 16 h at 37uC in tetrathionate broth (Merck KGaA, Darmstadt, Germany) and then again plated on BGA with antibiotics. Samples that were negative after direct plating but positive after enrichment were presumed to contain 50 CFU Salmonella per gram tissue (detection limit for direct plating). Samples that remained negative after enrichment were presumed to contain 0 CFU Salmonella per gram tissue.

B. Screening for in vivo induced Salmonella
Typhimurium genes in pigs. For IVET screening of Salmonella Typhimurium genes specifically induced during persistence in piglets, 8 four-week old Salmonella-free piglets were orally inoculated with approximately 2610 8 CFU of the Salmonella Typhimurium IVET pool. Three weeks post inoculation, all piglets were humanely euthanized and samples of the palatine tonsils, ileocaecal lymph nodes and ileum were collected and bacteriologically analyzed as described above. For monitoring the transcriptional activity of the isolated IVET transformants in vitro, tissue homogenates were plated on MacConkey lactose agar with ampicillin, nalidixic acid and adenine/thiamine. This allowed detection of bacterial strains containing promoters expressed in vivo in the tonsils, lymph nodes and ileum (purA expression) and not in vitro. Fusion strains containing promoters induced in vitro showed red colonies (high-level lacZY expression) on MacConkey agar, whereas fusion strains carrying promoters inactive in vitro (low-level lacZY expression) displayed white to pink colonies on MacConkey medium. Because we were interested in genes that were specifically induced in vivo and not in vitro, approximately 500 colonies with low-level in vitro lacZY expression were picked up and purified prior to sequencing.

C. Reintroduction of identified IVET transformants in
piglets to verify tissue-specific gene expression. To verify if the identified promoters from the initial IVET screening exhibited tissue-specific expression patterns, the 32 IVET transformants (corresponding to the 37 identified genes listed in Table 2) isolated and identified in the initial screening were pooled and reintroduced in 6 four-week old piglets. To assure that every single IVET transformant was equally present in the inoculum, each transformant was grown separately and in triplicate in 5 ml LB with the appropriate additives for 16 h at 37uC on a shaker and the optical density of each culture was then measured at a wavelength of 650 nm (OD 650 ) using a microplate ELISA reader (Multiscan MS, Thermo Labsystems, Helsinki, Finland). One ml of each 16 h culture was pooled, washed in HBSS and the bacterial pellet was diluted to a concentration of 5610 8 CFU per ml HBSS. Subsequently, piglets were orally administered 2 ml of this inoculum, corresponding to approximately 1610 9 CFU of the IVET transformants pool. Each IVET transformant was thus theoretically represented by circa 3610 7 CFU. Twenty days post oral inoculation, all piglets were humanely euthanized and samples of tonsils, ileocaecal lymph nodes and ileum were collected. Tissue samples were processed as described above and plated on MacConkey agar with the appropriate additives. Plates were incubated overnight at 37uC. From each tonsil and lymph node sample, the cloned promoters in 24 IVET colonies were identified by sequencing as described above, except for the tonsil of 1 piglet and the lymph nodes of another piglet, from which only 18 and 20 colonies were sequenced, respectively. For the ileum samples, all isolated colonies were sequenced. If no IVET transformants were isolated from the ileum after direct plating, samples were preenriched and enriched as described above, and plated on MacConkey agar with the appropriate additives. Subsequently, 10 colonies per enriched ileum sample were sequenced.

D. Experimental inoculation of piglets with Salmonella
Typhimurium DsifB and DSTM4067 strains. Following the IVET screening in pigs and based on the literature, 2 in vivo induced genes were selected of which substitution mutants were constructed. These mutants were then tested for attenuation compared to the wild type strain in a mixed infection experiment. For this in vivo trial, 14 piglets were randomly divided in 2 groups of 6 animals and 1 negative control group of 3 animals. Animals of the first two groups were orally inoculated with a mixture of approximately 2610 7 CFU Salmonella Typhimurium wild type and 2610 7 CFU Salmonella Typhimurium DsifB::kanR or DSTM4067::kanR, respectively, while the negative control pigs were administered 2 ml PBS. Three weeks post oral inoculation, the piglets were humanely euthanized and samples of the palatine tonsils, ileum and contents, ileocaecal lymph nodes, caecum and contents and faeces were collected and bacteriologically analyzed as described above for the wild type/DpurA::kanR mixed infection experiment.

Statistical analysis
For the WT versus DpurA::kanR, DsifB::kanR and DSTM4067::kanR mixed infection experiments, a non-paramatric Mann-Whitney U test was performed to determine whether the log value of the WT/mutant ratios of the samples was significantly different from the log value of the WT/mutant ratios of the inocula before and after oral inoculation. The OD 650 values of the 16 h cultures of IVET transformants were statistically evaluated using a one-way ANOVA. Tissue-specific gene expression patterns between tonsils and ileocaecal lymph nodes were analyzed for each gene using a McNemar's test with tonsils and lymph nodes as paired groups with two possible outcomes (expressed or not). Statistical analyses were performed with the SPSS Statistics 17.0 software (SPSS Inc., Chicago, USA). Differences with a p-value #0.05 were considered statistically significant; differences with a value 0.05,p#0.1 were considered borderline significant.

Results and Discussion
The Salmonella Typhimurium DpurA::kanR strain is severely attenuated in pigs The wild type Salmonella Typhimurium was recovered from all organs and organ contents from all piglets, except for the ileum contents and faeces from 1 animal, while the purA::kanR substitution mutant was found in the tonsils of only 1 piglet and the faeces of another, eight days post oral inoculation ( Table 1). The WT and DpurA::kanR CFU isolated from the tonsils and the faeces were therefore not statistically different (p = 0.182 and p = 0.317 resp.). However, in all other analyzed samples the number of WT bacteria significantly differed from the number of DpurA::kanR (p = 0.046 for each sample), except for the ileum contents in which the difference between both strains was statistically borderline (p = 0.096). These data clearly indicate that purA expression is required for Salmonella Typhimurium survival and persistence in pigs. This finding allowed us to use the pIVET1 system based on purA attenuation for the identification of Salmonella genes expressed during persistence in piglets.

IVET screening for in vivo induced genes
To screen for Salmonella Typhimurium genes that are specifically induced in porcine tissues during the persistent phase of infection, we constructed an IVET transformants pool covering the major part of the Salmonella Typhimurium genome [14]. We purified and subsequently sequenced 394 colonies from the tonsils, ileum and ileocaecal lymph nodes resulting in the identification of 37 different Salmonella Typhimurium genes that were induced 3 weeks post oral inoculation of piglets with the IVET pool and that might be essential for Salmonella persistence in pigs. These genes encode proteins that belong to diverse functional groups. Of these genes, 5 genes were identified in all 3 analyzed tissues. Furthermore, respectively 7, 4 and 12 genes were identified in the tonsils, ileum and ileocaecal lymph nodes only. Finally, several genes were identified from 2 different organs. Six cloned fragments contained 2 different genes. Since it can not be determined which gene(s) is (are) expressed in these fragments, the genes are discussed separately in the paragraphs below. The majority of the genes was expressed in only 1 piglet. This might be a consequence of the small number of piglets (n = 8) that was used in the IVET study, while we screened approximately 15.000 different IVET transformants. It is possible that after inoculation not all transformants had an equal chance to successfully infect piglets, creating a bottleneck. An overview of the identified genes, categorized according to their function and the tissue(s) in which they were expressed, is given in Table 2.

Chaperones
Four IVET identified genes encode chaperones or heat-shock proteins (hsp's) that guide correct folding of newly synthesized proteins and that protect the cellular components under stressful conditions. The htpG gene is the only chaperone encoding gene that was expressed in tonsils, ileum and ileocaecal lymph nodes. HtpG is an hsp90 homologue that is more than 2-fold expressed in Salmonella Typhimurium in response to the bactericidal/permeability increasing protein (BPI) from human neutrophils [15]. It is possible that the expression of htpG is induced in Salmonella Typhimurium following exposure to antimicrobial peptides present in the tonsils, ileum and/ or ileocaecal lymph nodes of pigs. Such potential candidate is the palate, lung and nasal epithelium clone or SPLUNC1, a BPIhomologue secreted by the epithelial cells of the nasal cavity and the respiratory tract in pigs, although this protein has not yet been detected in porcine tonsils [16]. dnaK was expressed in the lymph nodes of 3 piglets and its gene product forms a chaperone machinery with co-chaperones DnaJ and GrpE, that protect from low pH and heat stress by binding non-specifically to unfolded polypeptides to produce or restore functional proteins. This gene was shown to be involved in the acid stress response of Salmonella Typhimurium in the gastric environment of pigs [8].
Two other chaperones were identified from the tonsils. MopA is involved in ceftriaxone resistance in Salmonella Typhimurium, is similar to the E. coli chaperonin GroEL and is transcribed extensively during heat recovery in Salmonella Enteritidis [17]. cpbA encodes a 'curved DNA binding protein', a molecular hsp40 chaperone, homologous to DnaJ and involved in bacterial responses to environmental stress. The expression of both genes is regulated by the virulence-associated bacterial-transcriptional regulator SlyA, involved in systemic but not enteric salmonellosis, resistance of Salmonella Typhimurium to oxidative stress and destruction of M cells in mice [18].

LPS, amino acid and protein biosynthesis
Several genes were identified that play a role in biosynthesis pathways. RfaE encodes the gene for ADP-L-glycero-D-mannoheptose, a component of the Salmonella Typhimurium lipopolysaccharide (LPS) inner-core. A DrfaE Salmonella Typhimurium has an incomplete LPS and exhibits a lower growth rate, higher temperature sensitivity and decreased invasiveness of epithelial cells [19]. Salmonella Typhimurium strains defective in LPS are less resistant to the antimicrobials baicalin and novobiocin [20], rendering it possible that rfaE is expressed in the ileocaecal lymph nodes by Salmonella Typhimurium as a response to certain host antimicrobial proteins. AroK catalyzes the production of shiki- mate-3-phosphate from shikimate in the aromatic amino acid biosynthesis pathway. In E. coli, AroK is involved in resistance to the extended-spectrum penicillin mecillinam [21] and might play a role in the response of Salmonella Typhimurium to (yet unknown) porcine antimicrobial peptides. The among bacteria highly conserved elongation factor P (encoded by efp) was identified in the tonsil and ileocaecal lymph nodes. Efp is homologous to the eukaryotic translation initiation factor eIF5A and is essential for protein synthesis and for growth and viability in E. coli [22]. The exoribonuclease T encoding rnt gene was identified in the 3 investigated organs and is involved in tRNA processing and thus in protein biosynthesis. It is the most important exoribonuclease in E. coli [23]. The 30S ribosomal protein S21 encoding rpsU gene was identified in the tonsils and is essential for proper function of the 30S ribosomal subunit during mRNA translation [24]. The genes encoding RNA polymerase subunit omega (rpoZ) and RNA polymerase sigma-54 factor (rpoN) were expressed in the ileocaecal lymph nodes. RpoZ restores denatured RNA polymerase in vitro, recruits the rest of the RNA polymerase core enzyme and is involved in relA gene expression. RelA plays a key role in the 'stringent response' due to amino acid and carbon starvation in Salmonella Typhimurium by the production of the ppGpp alarmone [25]. Since the stringent response in Salmonella Typhimurium and Salmonella Gallinarum is induced during infection of mice and chicks respectively [27], this stress response might also play a role during persistence of Salmonella Typhimurium in pigs. The rpoN encoded sigma factor (s 54 ) recognizes especially promoters of genes important for survival and adaptation under unfavourable conditions [26,27].

Surface structure and OMP assembly
Flagellin type 1 and type 2 encoding genes fljB and fliC, respectively, were identified from the tonsils. Flagellin is the main component of bacterial flagellae and only 1 of both genes is expressed at any given time in Salmonella Typhimurium, resulting in so called 'phase variation'. It is generally believed that motility through flagellae mediates the initial interaction between bacterium and host. However, the contribution of flagellae to Salmonella Typhimurium virulence is strongly host-dependent. Flagellae contribute to invasion of the mucosa in the calf intestinal model [28], while neither flagellae, nor synthesis of the flagellar export machinery are necessary for pathogenicity in mice [29]. Nevertheless, a blocked fljB expression leads to attenuation of Salmonella Typhimurium in vivo in mice [30]. Furthermore, flagellin is required for the production of a variety of cytokines and inflammatory responses. No data about the role of flagellae in Salmonella Typhimurium persistence in pigs is available to date.
The gene encoding b-barrel integral protein BamA (yaeT) was expressed in the lymph nodes. The 'b-barrel assembly machine' or Bam anchors outer membrane proteins (OMPs), especially porins, into the outer membrane of Gram-negative bacteria. OMPs function as the interface between the bacterium and the environment and are therefore major virulence factors in these bacteria [31]. Expression of OMPs under iron limitation, oxidative stress and anaerobic conditions was shown for Salmonella Typhi [32]. The OMP AsmA was identified from the ileocaecal lymph nodes and was shown to be essential for Salmonella Typhimurium survival in ex vivo swine stomach contents [8]. A Salmonella Typhimurium DasmA is attenuated in virulence in mice, shows enhanced bile resistance and is defective in invasion of nonphagocytic HeLa cells. Therefore, the presence of AsmA in the outer membrane of Salmonella Typhimurium might be required for invasion in pigs [33].

DNA replication and repair
The IVET screening allowed the identification of several genes that play a role in DNA replication and/or repair. As components of the primosome that creates RNA primers for DNA polymerase III on single stranded DNA, DnaC and DnaT are central proteins in DNA replication initiation. DnaT is involved in primosome assembly and DnaC associates with DnaB to form the DnaB-DnaC helicase complex that will unwind the DNA duplex prior to DNA replication [34]. gyrB encodes the B subunit of DNA gyrase, an essential enzyme in prokaryotic replication and transcription. DNA gyrase is a type II topoisomerase that unwinds the parental DNA duplex prior to replication and transcription [35]. It is a critical target for antimicrobial chemotherapy and it is therefore possible that gyrB is expressed as a response to gyrase inactivation by (unknown) host antimicrobial peptides. The ribonucleoside diphosphate reductase encoding nrdB is part of the nrdAB operon that is involved in DNA replication and repair by reducing ribonucleotides to deoxyribonucleotides. The E. coli nrdA promoter was found to respond strongly to DNA damaging chemicals and is suggested to function as a biosensor for DNA damage [36]. It is possible that this operon has a comparable function in Salmonella Typhimurium and that the operon is expressed upon DNA damage induced in the hostile porcine environment.

Quorum sensing and anaerobic growth
Three genes were identified that play a role in anaerobic growth. The 'fumarate nitrate reduction transcriptional regulator' (Fnr) is a global regulator that plays a major role in bacterial switching from aerobic to anaerobic growth by regulation of a set of 311 known genes in Salmonella Typhimurium [37]. Fnr is only active in the absence of oxygen and among the 311 known genes under its regulation in Salmonella Typhimurium are reductases that use nitrate, nitrite and fumarate instead of oxygen as terminal electron acceptors [38]. Furthermore, Fnr is involved in regulation of pflC and pflD, encoding the 'pyruvate formate lyase activase II' and 'putative formate acetyltransferase 2' respectively, that were also identified from in IVET screening. Pyruvate formate lyase catalyzes the reversible reaction pyruvate + CoA « acetyl2CoA + formate, a key reaction in the glucose fermentation route. It is likely that, due to the low oxygen tension in the ileocaecal lymph nodes, Salmonella Typhimurium is dependent on energy production routes that do not require oxygen for its survival in these organs.
One gene involved in Salmonella Typhimurium quorum sensing was identified from our screening. LsrR (previously YdeW) is a transcriptional repressor of the lsr operon that encodes a transport apparatus for auto-inducer 2 (AI-2) quorum sensing signals into the bacterial cell [39]. After entry in the cell, AI-2 is phosphorylated and will inhibit the LsrR repressor. The fact that lsrR is expressed in tonsils of pigs, might suggest that the lsr operon is of little importance for Salmonella Typhimurium persistence in porcine tonsils. In Salmonella Typhimurium, the luxS gene is directly involved in AI-2 production and deletion mutants lacking luxS are severely attenuated in SPI-1 gene expression. SPI-1 genes are of major importance for Salmonella Typhimurium invasion of epithelial cells in many animal hosts including pigs and do not play a role in persistence of porcine tonsils [5,40], which might explain why the lsr operon repressor LsrR is expressed in the tonsils.

Metabolism
Several genes were identified that have a known role in metabolism. yadF encodes a carbonic anhydrase, a zinc metalloenzyme that catalyzes the interconversion of CO 2 and H 2 CO 3 . In E. coli, YadF becomes an essential enzyme for growth when the bacterium needs more H 2 CO 3 [41] and the protein was already shown to induce virulence genes in Vibrio cholerae and several fungi [42]. Since Salmonella Typhimurium requires HCO 3 2 for the biosynthesis of pyrimidines, yadF expression might be induced due to a higher requirement of pyrimidines for survival in the host. As an ABC transporter, ArtP is part of the periplasmic arginine transport system encoded by the artPIQMJ operon, which transports arginine from the periplasmic space to the cytosol crossing the inner membrane [43]. All three mammalian forms of reactive oxygen species (ROS) are assumed to be formed by the same biochemical pathway that starts with the oxidation of Larginine [44]. Since host produced ROS play a role in the protection against microbial pathogens like Salmonella, the arginine transport system might be expressed in Salmonella Typhimurium to deplete the arginine in the host environment, preventing the formation of antimicrobial ROS and thus playing a role in virulence. The lipopolysaccharide transport protein encoding yhbG (or lptB), expressed in the lymph nodes, is an ABC transporter located at the cytoplasmic side of the inner membrane and is essential for E. coli viability [45]. The Lpt machinery transports LPS from the cytosol to the outer membrane and is induced under stressful conditions when cell envelope integrity is affected. It is likely that the latter is the fact when Salmonella Typhimurium resides in the hostile porcine tissues and that expression of the Lpt machinery is essential to restore the cell envelope. Expression of lysS, encoding lysyl-tRNA-synthetase, was upregulated in the tonsils and couples L-lysine to free tRNA for protein synthesis. Since lysine is an essential amino acid for all animals, it is added to the feed of piglets. In contrast to pigs, Salmonella Typhimurium is able to produce lysine but also imports exogenous lysine via the lysine/cadaverine antiporter encoded by the cad locus, earlier shown to be involved in E. coli virulence. It is possible that upregulation of lysS, together with the exogenous lysine from the pig feed, plays a yet unknown role in Salmonella Typhimurium virulence in pigs. The gene encoding the 'suppressor of copper sensitivity' protein ScsA was expressed in the ileum. ScsA is a copper binding protein that possibly functions as a peroxidase, by preventing formation of free hydroxyl radicals resulting from the reaction of copper with hydrogen peroxide [46]. Copper sulphate has been extensively used as growth promoter in pig diets because of its selective antimicrobial effect on gut microflora and was also present in the feed that we administered to our pigs. Therefore, expression of the copper binding protein ScsA by Salmonella Typhimurium in the ileum of these pigs might protect the bacterium from the copper supplement's antimicrobial properties. As a member of the kdg regulon, the kdgK encoded 'ketodeoxygluconokinase' plays a role in the carbohydrate metabolism by phosphorylating 2-keto-3-deoxygluconate (KDG). The citG2 gene encodes a 'triphosphoribosyl-dephospho-CoA synthase' that might be essential for the proper formation of the citrate lyase enzyme that catalyzes the cleavage of citrate to acetate and oxaloacetate, representing the initial step in all known bacterial citrate fermentation pathways [47]. Yet, no role in Salmonella virulence for these 2 genes has been established.

Pathogenicity island genes
Salmonella Typhimurium possesses several pathogenicity islands (SPIs) that encode different virulence factors. Because many of these SPI-encoded genes are expressed on most laboratory media and we only identified IVET transformants that showed no in vitro activity (see Materials and Methods), it is not surprising that only 1 known SPI encoded gene was identified 3 weeks post inoculation of piglets. SifB is a SPI-2 effector, encoded outside the genomic SPI-2 region, that localizes to the Salmonella containing vacuole (SCV) and Salmonella induced filaments (Sifs) [48]. A sifB insertion mutant was found defective in colonization of calves and not chicks [49] and a sifB deletion mutant was not attenuated in mice after intraperitoneal inoculation. SPI-2 and its effectors are especially involved in intracellular survival and since we recently showed that Salmonella Typhimurium resides largely extracellularly in porcine tonsils [50], the upregulation of sifB in tonsils is rather unexpected. We compared the persistence capacity of Salmonella Typhimurium DsifB::kanR to that of the WT in a pig mixed infection experiment. Indeed, no significant differences in persistence capacity between the mutant and the WT could be detected in any of the examined samples (Figure 1). This is in accordance with [51], who found that deletion of the SPI-2 regulating ssrA/B operon did not significantly attenuate Salmonella Typhimurium for colonization in pigs compared to the wild type.

Genes with putative and unknown function
Several Salmonella Typhimurium genes were identified with a putative or yet unknown function. YbjP is a 'putative lipoprotein' that is induced during the stationary phase and by acivicin and that is regulated by Lrp and RpoS [52]. Both regulators of ybjP play a known role in Salmonella Typhimurium virulence by regulation of various virulence genes and therefore it is possible that YbjP also exhibits a role in Salmonella Typhimurium virulence. For example, RpoS controls a regulon of genes required for protection against external stresses. Interestingly, Lrp was expressed in Salmonella Enteritidis after inoculation of chicks [13]. yggE encodes a 'putative periplasmic immunogenic protein' that is 2 to 3 times upregulated after UV-radiation and thermal stress in E. coli. Its promoter region has consensus sequences for RpoS and overexpression of RpoS results in enhanced yggE expression. YggE closely interacts with the cell membrane to maintain the cell's rigidity and intactness [53]. Yet, no function for YggE in Salmonella Typhimurium virulence is known, although [54] found yggE upregulated during Salmonella Typhimurium swarming and described the gene as a putative SPI-2 gene. The 'putative ligase' YgfA was expressed in the tonsils, ileum and lymph nodes in our IVET screening. Interestingly, the same gene was upregulated 3 weeks after Salmonella Enteritidis inoculation of chicks [13], suggesting a yet unknown role for Salmonella YgfA in persistence in different hosts. As expected from sequence homology to E. coli, via IVET identified STM3020 encodes a putative LysR-type transcriptional regulator. The continuously increasing LysR family of transcriptional regulators is highly conserved among bacteria and is involved in a wide range of hostmicrobe interactions [55]. Possibly, the STM3020 gene product is a yet unspecified transcriptional regulator involved in Salmonella Typhimurium colonization of and persistence in pigs. In the lymph nodes, the citX2 gene was expressed encoding a 'putative cytoplasmic protein' with a yet unknown function In Salmonella Typhimurium. STM4067 is part of the STM4065-4066-4067 operon of which STM4066 has aminoimidazole riboside (AIR) kinase activity, phosphorylating AIRs and satisfying the thiamine requirement of pur mutant strains [56]. Because our IVET library was constructed in a DpurA Salmonella Typhimurium, which lacks the enzyme adenylosuccinate synthase, required for synthesis of adenosine 59-monophosphate (AMP). For that reason, a bias to the upregulation and subsequent identification of genes involved in AMP synthesis, or more generally the purine biosynthesis pathway of Salmonella Typhimurium, from the IVET in vivo screening might be expected. To rule out this bias, the Salmonella Typhimurium DSTM4067::kanR was tested in a mixed infection experiment with the WT. The DSTM4067::kanR was attenuated in the ileum and ileum contents, the caecum and caecum contents and the faeces and these differences were significant in the ileum contents (p = 0.04), the caecum (p = 0.033) and caecum contents (p = 0.033) and the faeces (p = 0.032; Figure 1). The fact that DSTM4067::kanR was attenuated in the pigs' intestines compared to the WT proved that our IVET selection actually resulted in the identification of Salmonella genes involved in persistence in pigs.
Salmonella Typhimurium specifically expresses rpoZ and efp in porcine lymph nodes The average OD 650 between the separate IVET transformant 16 h cultures did not significantly differ (p.0.05), so each transformant was equally present in the final inoculum and, consequently, had an equal chance to establish an infection in pigs.
First, we compared the Salmonella Typhimurium gene expression pattern between tonsils and ileocaecal lymph nodes. The majority of the identified genes showed no tissue-specific expression pattern (p.0.05) and we were unable to find genes specifically expressed in the tonsils, although artP/ybjP was only isolated from the tonsils (and not from the lymph nodes or ileum) with a frequency of 2/6 (p = 0.227).
However, efp, encoding the elongation factor P, and rpoZ, encoding the RNA polymerase omega subunit, were expressed in the lymph nodes of 5/6 pigs and in the tonsils of 1/6 pigs, and these frequencies were significantly different (p = 0.04). It can thus be assumed that both genes are specifically expressed in ileocaecal lymph nodes during Salmonella Typhimurium persistence in pigs. STM3020 was expressed in 4 of 6 lymph nodes and in 1 of 6 tonsils (p = 0.121). The comparison of isolation frequency of transformants between tonsils and lymph nodes is graphically presented in Figure 2.
The results of the ileum could not be statistically analyzed, because too few colonies were isolated at direct plating (as expected in the used infection model), and because enrichment of samples might bias the result of a sample towards 1 specific transformant. However, interestingly, the genes efp, rpoZ and STM3020 are also expressed at higher frequencies in the ileum than in the tonsils: 4/6, 3/6 and 2/6, respectively. It is possible that expression of these genes is required for colonization of the ileum and subsequent penetration of and persistence in the underlying lymphoid tissues, suggesting that Salmonella Typhimurium exploits different genes for tonsil and intestinal colonization, as proposed earlier [5].
The chaperone Hsp60 encoding gene dnaK was expressed in the lymph nodes of all 6 piglets and in 3 of 6 examined tonsils and ileum samples and this difference in frequency was borderline significant (p = 0.091). In addition to this, the shikimate kinase encoding aroK was isolated at equal frequencies from both the tonsils and the lymph nodes (4/6) and was also expressed in 3/6 ileum samples. These results suggest a general role for both genes in Salmonella persistence of porcine tissues.
The lysyl tRNA synthetase encoding lysS was expressed in 3 of 6 lymph nodes and not in the tonsils (p = 0.091). Interestingly, after modification onto highly-conserved lysine residues by the enzymes Figure 1. Salmonella Typhimurium wild type and DsifB or DSTM4067 colonization of pigs. Recovery of bacteria from various organs of 2 groups of 6 piglets orally inoculated with an equal mixture of the wild type Salmonella Typhimurium and DsifB or DSTM4067 Salmonella respectively. The log 10 value of the ratio of CFU per gram sample of the wild type and DsifB or DSTM4067 mutants is given as the mean 6 standard deviation. A non-paramatric Mann-Whitney U test was performed to determine whether the log value of the WT/mutant ratios of the samples was significantly different from the log value of the WT/mutant ratios of the inocula. An asterisk (*) indicates that the output ratio was significantly different (p#0.05) from that in the inoculum. doi:10.1371/journal.pone.0024120.g001 YjeK and PoxA, of which the latter is a paralog of lysyl tRNA synthetase, efp was shown to contribute to Salmonella enterica virulence in a mouse model [57]. The expression of lysS and efp suggests that the family of lysyl tRNA synthetases, and the proteins that they modify, might be involved in Salmonella virulence in porcine lymph nodes, by regulating the synthesis and/or activation of a limited subset of proteins. Indeed, recently a role for Salmonella Typhimurium poxA in colonization of porcine tissues was established, providing additional evidence for an important role for lysyl tRNA synthetases in Salmonella pathogenesis in pigs [58].
Eighteen different IVET transformants, corresponding to 21 different gene fragments, were isolated from the 3 examined organs (see Supporting Information S2 for an overview per organ). Among others, the gene encoding STM4067, that we identified as a factor involved in Salmonella persistence in the intestines of pigs, could not be recovered from the reintroduction experiment. This shows that the 'bottleneck' in the original IVET screening was to a certain extent but not completely circumvented in the reintroduction experiment. Furthermore, isolation of low numbers of bacteria and, consequently, an increased variation between pigs are inherent to the infection model we use to study Salmonella persistence in pigs, compared to invasion and colonization studies in the early phase of Salmonella infection [59]. So there is a reasonable chance of missing several IVET transformants, even when pigs are inoculated with a limited pool.
Our results show that conclusions about tissue-specific gene expression patterns from an initial IVET screening must be drawn very carefully. When the lymph-node specific expressed genes rpoZ and efp are considered, rpoZ was identified from only the ileocaecal lymph nodes in the initial IVET screening. However, efp was identified from the lymph nodes and from the tonsils in the initial screening, but remained absent from the latter in the subsequent reintroduction experiment. In contrast, lysS that was expressed only in the tonsils in the initial screening, was expressed in 3 lymph nodes but not in the tonsils in the second IVET experiment. Furthermore, aroK and dnaK that are probably involved in overall Salmonella persistence in pigs, were only expressed in porcine lymph nodes in the initial screening ( Figure 2 and Table 2).
In general, our results emphasize that for proper interpretation of tissue-specific gene expression patterns using IVET screening, a subsequent infection of the host animal that is subject of study with a limited number of IVET transformants is required, to minimize bottleneck effects and to get an indication of tissue-specificity.

Conclusion
Using the IVET screening, we were able to identify 37 Salmonella Typhimurium genes that were expressed 3 weeks post inoculation of pigs. To our knowledge, this study is the first to identify genes that might play a role in Salmonella Typhimurium persistence in pigs. After Salmonella entry, the porcine immune system will fight the bacterium by establishing a hostile environment. It can thus be expected that the expression of Salmonella genes encoding proteins that protect the bacterium against stressful conditions like antimicrobial proteins, cell wall degradation, lack of oxygen, etc. is upregulated during Salmonella Typhimurium persistence in pigs. Interestingly, there is little or no overlap between the genes expressed in our study and the genes previously identified by an IVET [10] and STM [9] approach, respectively 2 and 3 days post oral inoculation of pigs with Salmonella Typhimurium. This finding indicates that different sets of Salmonella genes could be involved in colonization (2-3 days pi) of and persistence (20 days pi) in pigs.
A possibility existed that the total number of different transformants exceeded a bottleneck above which each pool member no longer had an equal chance of establishing itself within the IVET population. To minimize these bottleneck effects and to find tissue-specific gene expression patterns, we reintroduced the pool of different IVET transformants identified in the initial IVET screening in 6 Salmonella-free piglets. We found that efp and rpoZ were specifically expressed in porcine lymph nodes during Salmonella persistence in pigs. Genes that were expressed at higher frequencies in the lymph nodes than in the tonsils, also appeared to be more frequently expressed in the ileum, suggesting that comparable sets of Salmonella genes are involved in colonization of the ileum and lymph nodes. Furthermore, dnaK and aroK were found to play a general role in Salmonella persistence in pigs. Although the results obtained from the IVET reintroduction experiment give a more reliable indication of tissue-specific Salmonella gene expression patterns than the initial screening, extrapolation of such expression patterns should still be drawn very carefully, among others due to the limited number of piglets used.
It was also possible that the inoculation dose applied in our initial screening experiment was insufficient for some of the IVET transformants to initiate infection in pigs. To assess a role in Salmonella pathogenicity for IVET identified genes, we constructed Salmonella substitution mutants for 2 genes that were expressed in the initial IVET screening. We identified STM4067 in a subsequent mixed infection experiment as a factor for intestinal Salmonella Typhimurium persistence in pigs. Finally, we must emphasize that although the IVET identified genes are expressed during Salmonella Typhimurium persistence in pigs, these genes are not per se essential for persistence, as becomes clear from the mixed infection experiment with the WT and the DsifB::kanR.

Supporting Information
Supporting Information S1 Primers used in this study. Primers used in this study to create the Salmonella Typhimurium DpurA, DsifB and DSTM4067 substitution mutants and Y-linker component sequences and purAand 'Y-linker'-primer sequences used for sequencing IVET fusion strains. (DOC) Supporting Information S2 Overview of identified IVET transformants per organ. Identification of IVET transformants recovered from the tonsils, ileocaecal lymph nodes and ileum in the reintroduction experiment, in which 6 piglets were orally inoculated with approximately 5610 8 CFU of a pool of 32 IVET transformants. (DOC)