Impact of the Lipopolysaccharide Chemotype of Salmonella Enterica Serovar Typhimurium on Virulence in Gnotobiotic Piglets

Salmonella Typhimurium is an enteric pathogen that causes acute and chronic infections in humans and animals. One-week-old germ-free piglets were orally colonized/infected with the Salmonella Typhimurium LT2 strain or its isogenic rough ΔrfaL, ΔrfaG or ΔrfaC mutants with exactly defined lipopolysaccharide (LPS) defects. After 24 h, the piglets were euthanized and the colonization of the small intestine, translocations into the mesenteric lymph nodes, liver, spleen, lungs, and bacteremia, along with changes in the ileum histology, and transcription levels of the tight junction proteins claudin-1, claudin-2, and occludin were all assessed. Additionally, transcription levels of IL-8, TNF-α, and IL-10 in the terminal ileum, and their local and systemic protein levels were evaluated. Wild-type Salmonella Typhimurium showed the highest translocation, histopathological changes, upregulation of claudins and downregulation of occludin, transcription of the cytokines, intestinal IL-8 and TNF-α levels, and systemic TNF-α and IL-10 levels. Depending on the extent of the incompleteness of the LPS, the levels of the respective elements decreased, or no changes were observed at all in the piglets colonized/infected with Δrfa mutants. Intestinal IL-10 and systemic IL-8 levels were not detected in any piglet groups. This study provided foundational data on the gnotobiotic piglet response to colonization/infection with the exactly defined rough Salmonella Typhimurium LT2 isogenic mutants.

tachycardia, tachypnea, and tremor. The piglets infected with the ∆rfaL mutant showed similar signs. Weaker and delayed signs were expressed in the ∆rfaG-infected group. The piglets inoculated with the ∆rfaC Salmonella Typhimurium mutant less expressed clinical signs of salmonellosis than the ∆rfaG piglets.

Salmonella Colonization of the Small Intestine and Its Translocation
Wild-type Salmonella Typhimurium colonized the small intestine at the highest degree, with a median count greater than 10 10 CFU/g in the ileum ( Figure 1A). ∆rfaL and ∆rfaG mutants colonized the intestine at a lower density with medians around 10 8 CFU/g, and this decrease was statistically significant in ∆rfaL but not in the ∆rfaG mutant. Colonization ability of the ∆rfaC mutant was 6-log lower in comparison to the wild-type Salmonella Typhimurium, and 4-log lower as compared to the ∆rfaG mutant. Wild-type Salmonella Typhimurium was found in all tested organs of all piglets-mesenteric lymph nodes ( Figure 1B), liver ( Figure 1C), spleen ( Figure 1D), and lungs ( Figure 1E), and caused bacteremia ( Figure 1F). Salmonella Typhimurium counts in mesenteric lymph nodes (MLN) ( Figure 1B) are directly proportional to the completeness of the LPS. Translocations of ∆rfaG, and ∆rfaC mutants, but not ∆rfaL, were significantly lowered. Three out of 8 ∆rfaG and 2 out of 6 ∆rfaC, respectively, were found in the MLN of these groups. The liver translocation ( Figure 1C) was detected in all WT-infected piglets, but only in 6, 5, and 0 out of 8 piglets infected with the ∆rfaL, ∆rfaG, and ∆rfaC mutants, respectively. Similarly, the wild-type Salmonella Typhimurium infected the spleen of all piglets ( Figure 1D), while ∆rfaL, ∆rfaG, and ∆rfaC mutants were detected in only 6, 5 and 2 out of 8 infected piglets, respectively. All the mutants displayed low levels of lung translocation ( Figure 1E). The Kruskal-Wallis test with Dunn's post-hoc multiple comparisons test was used, and statistical differences p < 0.05 are denoted with different letters above the groups. The same letter above the groups indicates no statistically significant differences.

Histopathological Changes in the Ileum
The ileum of the germ-free piglets showed long villi containing abundant vacuolated enterocytes and no signs of inflammation (Figure 2 A,B). In contrast, the ileum of the piglets infected with wild-type Salmonella Typhimurium ( Figure 2C,D) was affected by acute inflammation characterized by villus atrophy, submucosal edema, vessel dilation, exudate in the lumen, the presence of neutrophils in the lamina propria, hyperemia, erosion of the epithelial layer, and peritonitis, as it is summarized in the histological score ( Figure 2K). The majority of these changes, expressed at lower levels, were also found in the ∆rfaL group ( Figure 2E,F,K). Hemorrhage was absent in the WT and other groups. Inflammatory changes in the ∆rfaG ( Figure 2G,H) and ∆rfaC ( Figure 2I,J) groups were indiscernible-with only negligible villus atrophy, vessel dilation, neutrophils in the lamina propria, and submucosal edema. Using our scoring system, the total histological score for WT, ∆rfaL, ∆rfaG, and ∆rfaC groups were evaluated as 8.2, 5.8, 1.9, and 1.9, respectively ( Figure 2K).

Figure 2.
Representative hematoxylin and eosin-stained cross sections of the ileum in the gnotobiotic piglets and a histological score. The ileum of one-week-old gnotobiotic piglets: germ-free piglets (GF; A,B), piglets colonized/infected with Salmonella Typhimurium strain LT2 for 24 hrs (WT; C,D) or its isogenic mutants ∆rfaL (E,F), ∆rfaG (G,H) or ∆rfaC (I,J). Bars represent 1 mm (A,C,E,G,I) and 500 µm (B,D,F,H,J) cross sections, respectively. Histological scores from the ileum of six piglets per group are depicted (K).

Expression of Tight Junction Proteins Claudin-1, Claudin-2, and Occludin in the Ileum
The transcription of claudin-1 increased after the infection of the germ-free piglets with the wild-type Salmonella Typhimurium and the ∆rfaL mutant ( Figure 3A). Statistical differences were observed between the WT, and the germ-free, ∆rfaG, and ∆rfaC groups. Claudin-2 transcription was increased in the WT group, and statistical differences were found between this group and the ∆rfaC group, but not other piglets ( Figure 3B). Occludin ( Figure 3C) showed the opposite trend than the claudins ( Figure 3A,B), i.e., its transcription in the ileum of wild-type and ∆rfaL Salmonella Typhimurium-infected piglets was statistically significantly lower than in germ-free piglets. Comparable relations to other groups were found in the ∆rfaL mutant-infected piglets. However, no statistical differences were found between WT or ∆rfaL groups, or between the ∆rfaG or ∆rfaC groups. Claudin-1 (A), claudin-2 (B), and occludin (C) mRNA were normalized to β-actin and cyclophilin A. The relative expressions (fold change) were evaluated in the germ-free piglets (GF) and the piglets colonized/infected with wild-type Salmonella Typhimurium (WT), or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants. One-way analysis of variance (ANOVA) with Tukey's multiple comparisons post-hoc test was used to compare differences among the groups. The values are presented as mean + SEM. Statistical differences p < 0.05 are denoted with different letters above the columns, and the same letter shown above the column indicates no statistically significant differences. Six samples in each group were analyzed.  Figure 4C) in the ileum, as compared to the control germ-free piglets. Only in the case of IL-8 was this induction statistically significant. IL-8 ( Figure 4D) and TNF-α ( Figure 4E) levels were significantly induced by wild-type Salmonella Typhimurium, and its ∆rfaL mutant, in the ileum. No IL-10 levels in the ileum ( Figure 4F) or IL-8 levels in plasma ( Figure 4G) were found. TNF-α in plasma ( Figure 4H) was induced by wild-type Salmonella Typhimurium and was statistically different from other groups. Similarly, IL-10 was increased in plasma ( Figure 4I) and was statistically different in the ∆rfaC and germ-free groups, but not in the ∆rfaL and ∆rfaG groups. , and IL-10 (C) in the terminal ileum tissue of the germ-free piglets (GF) and the piglets colonized/infected with wild-type Salmonella Typhimurium (WT), or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants, were normalized to β-actin and cyclophilin A. The protein levels of the same cytokines-IL-8 (D), TNF-α (E), and IL-10 (F)-in the ileum lavage and the plasma (G,H,I, respectively) were estimated. Kruskal-Wallis test with Dunn's multiple comparison post-hoc test was used to compare the groups. The transcriptions (A-C) were measured in six piglets per group, and the protein levels (D-I) in eight piglets per group. The values are presented as mean + SEM. Statistical differences p < 0.05 are denoted with different letters above the columns and the same letter shown above the column indicates no statistically significant differences.

Discussion
Salmonella translocates beyond the intestinal barrier to the mesenteric lymph nodes (MLN) via lymphatic vessels, while less virulent strains can be trapped and destroyed [33]. Highly virulent Salmonella can spread via lymph vessels to the liver, spleen and other organs, or it can reach these sites via the blood [34]. Salmonella Typhimurium strain LT2 is avirulent for conventional piglets [30], but the germ-free piglets infected with this strain die 36-48 hrs after the infection (unpublished results). While wild-type Salmonella Typhimurium LT2 translocated to all observed organs and caused bacteremia in our experiments, its ∆rf a mutants with incompletely synthetized LPS showed lowered or neglected translocation and did not cause bacteremia. These findings corresponded to the findings in conventional mice [12].
Wild-type Salmonella Typhimurium LT2 and the ∆rfaL mutant induced devastating histopathological changes in the ileum of the full-term gnotobiotic piglets, but with no hemorrhages, unlike the previous study performed in preterm hysterectomy-derived gnotobiotic piglets [28]. The mutants with more incomplete LPS chains (∆rfaG and ∆rfaC) showed a reduced detrimental effect, as did other rough Salmonella strains of different serovars [14,33].
The intestinal epithelium, the single cell layer of epithelial cells in the gastrointestinal tract, is important for both nutrient uptake and protection of the host against bacterial translocation.
The barrier function is dependent on tight junction proteins such as claudins and occludin, which join adjacent enterocytes in the apical region. This barrier can be disturbed with enteric infections [35]. Claudin-1 belongs to the group of barrier-forming claudins, and claudin-2 to the pore-forming claudins [36]. The increase in claudin-1 transcription in the ileum of the wild-type Salmonella Typhimurium-infected gnotobiotic piglets probably indicates the body's attempt to seal the intestinal barrier to protect the piglets against excessive loss of electrolytes via diarrhea. The scope of claudin-1 induction gradually decreased in piglets infected with ∆rfa mutants, with the decreasing completeness of their lipopolysaccharide chains in the direction of ∆rfaL > ∆rfaG > ∆rfaC, corroborating with decreasing severity of the infection as evaluated by clinical signs. In contrast to our finding, the conventional piglet ileum infected with Salmonella Infantis showed decreased claudin-1 protein [37]. This discrepancy should be explained by the attempt to seal the intestinal barrier from the beginning of infection, but by manifestation of its devastating effect five days post infection. Claudin-2 was increased in the ST group too. We suppose that these changes of claudin-2 with seemingly opposite function than claudin-1 [36] complete the attempt to maintain the functionality of damaged intestinal barrier. In contrast, transcription of occludin decreased in wild-type Salmonella Typhimurium and ∆rfaL mutant-infected piglets. Although occludin is also a tight junction protein and is related to the transfer of large biomolecules and cells [38], its function is much less understood than that of the claudins [36]. We were unaware of a precise mechanism of action by Salmonella Typhimurium that resulted in occludin's expression opposite to that reported for claudins, but Salmonella Typhimurium alters the tight junctions in the epithelial barrier that result in increased bacterial translocation and induction of neutrophil transepithelial migration [39]. Similarly, neutrophil recruitment into inflamed intestine was found in Salmonella Typhimurium-induced colitis in streptomycin-treated C57Bl/6 mice [40]. Our newly obtained data are in agreement with previous findings in preterm gnotobiotic piglets colonized with probiotic Lactobacillus rhamnosus GG, and challenged with virulent Salmonella Typhimurium [41]. Other authors found that the expression of occludin protein in the conventional Salmonella Infantis-infected piglets was not influenced five days after challenge [37].
Inflammatory changes are governed by inflammatory mediators, mainly by cytokines. Salmonella Typhimurium can disrupt tight junctions, either by direct contact with epithelium or via cytokine induction [42]. To monitor piglet response to Salmonella Typhimurium challenge, we used three cytokines with different activities-chemotactic cytokine (chemokine) interleukin IL-8, a pro-inflammatory cytokine TNF-α, and a regulatory cytokine IL-10 [43]. IL-8 attracts neutrophils into inflammatory site [44], TNF-α is a pluripotent cytokine that is a potent mediator of inflammation inducible by LPS [45], and IL-10 regulates immune response to prevent an excessive inflammatory reaction [46]. Gene transcripts of all three cytokines increased in the terminal ileum of the piglets infected with wild-type Salmonella Typhimurium. Expression of these cytokines was also partially induced by the ∆rfaL mutant, while infection of piglets with ∆rfaG and ∆rfaC did not induce expression of these cytokines above the levels observed in the non-infected controls. These results, together with the above-mentioned translocations and histological changes, indicate high virulence and the stimulatory effect of wild-type Salmonella Typhimurium and decreasing virulence with the decreasing completeness of the LPS chains. This is in agreement with in vitro observations in which the ∆rfaC mutant manifested lower protein secretion and invasion compared to wild-type Salmonella Typhimurium. So the total virulence of the Salmonella enterica mutants with affected LPS might be a sum of different structure, different length and other affected processes like protein secretion and motility [16].
Only a few data are available that characterize the presence of inflammatory cytokines in intestinal content during enteric infections. Jeong et al. found significantly elevated IL-8, TNF-α, and IL-10 levels in the jejunum and ileum of gnotobiotic piglets infected with Shigella dysenteriae at 24 h post-infection [47]. Our finding of increased levels of IL-8 in the distal small intestine corresponds to the presence of neutrophils in the ileum of both wild-type Salmonella Typhimurium and ∆rfaL mutant-infected piglets. We propose that high local levels of IL-8, together with high levels of TNF-α in the intestine, in the case of wild-type Salmonella Typhimurium, exceeded normal physiological levels and contributed to the damage of the intestine itself. Lower levels of IL-8 and TNF-α in the ∆rfaL group may correspond to the protective chemoattraction of neutrophils to the inflammatory site [14] as a mechanism of protection by rough Salmonella enterica strains against subsequent infection with virulent Salmonella Typhimurium [14,15,33]. Less understandable is the absence of secreted IL-10 in the ileum of all groups, since its transcription was highly induced in wild-type Salmonella Typhimurium and ∆rfaL mutant-infected piglets. A hypothesis on the possible immaturity of the intestine and absence of cells able to secrete IL-10, or the strong degrading effect of protease-rich surroundings can be excluded because IL-10 was found in the intestine of gnotobiotic piglets of the same age infected with E. coli O55 [48]. We propose that the secretion of IL-10 in the intestine might be delayed, and measurable later than 24 hrs post-infection. This delay is probably the reason why local IL-10 did not play a role in the regulation of the inflammatory response and did not ameliorate a cytokine storm and its deleterious consequences [10].
Increased mRNA levels for IL-8 and TNF-α in different parts of the intestine, in the conventional pig infected with Salmonella Typhimurium DT104, corresponded to the concurring serum values and showed significant increases with the highest levels at 1 dpi for IL-8 and 2 dpi for TNF-α [49]. Other conventional pigs infected with Salmonella Typhimurium X4232 showed increased serum levels of IL-8, TNF-α and IL-10 two days post-infection [50]. Systemic IL-8 was not found in blood plasma of the gnotobiotic piglets infected with Salmonella Typhimurium strain LT2 or colonized with probiotic E. coli Nissle O55 or avirulent E. coli O86. However, high systemic IL-8 levels were found in the piglets with E. coli O55 infection [48,51]. In our present study, we did not detect IL-8 in the plasma of the piglets infected with wild-type Salmonella Typhimurium or any of the ∆rfa mutants. This likely supports our supposition that longer time would be needed for the development of detrimental consequences of the infection with the virulent Salmonella Typhimurium, as we concluded in the case of IL-10 in the intestinal content.
Both TNF-α and IL-10 were highly increased in the plasma of wild-type Salmonella Typhimuriuminfected piglets. High levels of systemic IL-10 and TNF-α indicated a poor prognosis of survival in preterm infants [52] and adult patients [53]. Similarly, in the gnotobiotic piglets, both IL-10 and TNF-α were found in E. coli O55 infected piglets ante finem [48]. High levels of IL-10 were detected in the piglets infected with wild-type Salmonella, and in lower levels in the plasma of the piglets infected with the ∆rfa mutants. They were absent, however, in the plasma of the control germ-free piglets. This further contrasted with the absence of IL-10 in the intestinal lavages, and the source of the plasma IL-10 is currently unknown.

Conclusions
In this study, we characterized the course of infection of germ-free piglets with wild-type and ∆rfa mutants of Salmonella Typhimurium. Any defect in the LPS structure correlated with reduced virulence. However, the absence of O-antigen in the ∆rfaL mutant resulted in reduced, but still detectable, residual virulence. The absence of an outer core of LPS in both ∆rfaG and ∆rfaC mutants resulted in complete avirulence. This also means that additional defects in the structure of the inner core of LPS present in ∆rfaC resulted in no further reduction in virulence, even in an extremely sensitive model of colostrum-deprived germ-free piglets. These data will be used to plan future studies in gnotobiotic piglets colonized/infected for a longer period with wild-type Salmonella Typhimurium LT2 and its ∆rfa mutants, to determine suitable candidates for the induction of innate and specific immune response and protection of the piglets against subsequent infection with wild-type Salmonella Typhimurium.

Ethical Statement
All experiments with animals were approved by the Animal Care and Use Committee of the Czech Academy of Sciences of the Czech Republic (protocol #63/2015; date of approval: 6 September, 2015).

Salmonella Typhimurium LT2 Strain and Its Isogenic ∆rfa Mutants
The wild-type LT2 strain of Salmonella enterica serovar Typhimurium (WT) and its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants with various completeness of LPS chain ( Figure 5) were obtained from a collection of microorganisms of the Institute of Microbiology (Novy Hradek, Czech Republic). The mutants with different LPS chemotype were prepared by the same methods as described elsewhere [17] (Appendix A).  [7,11]. The complete lipopolysaccharide (LPS) chain corresponds to the wild-type Salmonella Typhimurium (WT). The incompleteness of the LPS chain and corresponding chemotypes of the ∆rfaC, ∆rfaG, and ∆rfaL mutants are depicted.
The bacterial inocula were prepared by the cultivation of bacteria on meat peptone agar slopes (blood agar base; Oxoid, Basingstoke, UK) at 37 • C overnight. The bacteria were taken from the agar slopes and resuspended in PBS to an approximate density of 5 × 10 8 colony forming units CFU/mL. The number of CFU estimated by spectrophotometry at 600 nm was verified by a cultivation method on Luria-Bertani agar (Difco Laboratories, Detroit, MI, USA) at 37 • C for 24 h. The growth of the wild-type Salmonella Typhimurium and its ∆rfa mutants on the agar was comparable.

Gnotobiotic Piglets
Miniature Minnesota-derived germ-free piglets were obtained by hysterectomy as described elsewhere [28]. Briefly, hysterectomies were performed on the 112 th day of gestation under isoflurane (Isoflurane; Piramal Healthcare UK, Morpeth, UK) anesthesia. The total number of 40 gnotobiotic piglets were divided into five groups with eight piglets per group. Each piglet group was created from three independent hysterectomies. The piglet groups were reared separately in positive-pressure sterile fiberglass isolators with heated floors. They were fed to satiety 6-7 times a day with an autoclave-sterilized, condensed cow's milk-based formula (Mlekarna Hlinsko, Hlinsko, Czech Republic), using a bottle with a nipple. The following were sampled and cultivated for the presence of aerobic and anaerobic bacteria, and mold: amniotic membranes, umbilical cords, meconium, mouth and isolator surface smears after hysterectomy and later mouth, isolator surface smears, and stool twice a week. Additionally, Gram-stained rectal swabs were inspected under a light microscope.

Colonization/Infection of the Germ-Free Piglets
One-week-old germ-free piglets were orally administered 1 × 10 8 CFU of Salmonella Typhimurium (WT) or its isogenic LPS deletion mutants (∆rfaL, ∆rfaG, and ∆rfaC) in 5 mL of milk diet. The control germ-free piglets received 5 mL of milk without Salmonella. Each group of piglets was created from three hysterectomies. Twenty-four h after the challenge, the piglets were euthanized, bled out via cardiac puncture under isoflurane anesthesia, and samples were collected.

Clinical Signs
The piglets were observed for the appearance of fever, anorexia, somnolence, and diarrhea.

Bacterial Colonization and Translocation
Samples of peripheral blood were cultivated both undiluted and as serial log-dilutions with PBS. Ileum lavage was obtained by cutting off a 40-cm segment of a terminal part of the small intestine containing the whole ileum and part of the jejunum, filling it with 2 mL of Dulbecco's PBS (Life Technologies, Carlsbad, CA, USA), followed by gentle kneading, and rinsing. The vortexed intestinal lavages were diluted logarithmically in PBS. One g of each type of tissue (mesenteric lymph nodes, the liver, spleen, and lungs, respectively) was homogenized in 4 mL of deionized water in a glass piston homogenizer and serially diluted in PBS, to allow for separate analyses. The serial dilutions were aerobically cultivated on plates with LB agar (Difco Laboratories) at 37 • C for 24 h. The CFU were counted from dishes containing less than 200 colonies.

Blood Plasma and Intestinal Lavages
Citrated blood was spun at 1200× g for 10 min at 8 • C, and protease inhibitor cocktail (Roche Diagnostics, Manheim, Germany) was added to the collected plasma. The ileum lavages with added protease inhibitor cocktail (Roche Diagnostic) were spun at 2500× g for 30 min at 8 • C and supernatants were filtered through 0.2 µm nitrocellulose filters (Sartorius, Goettingen, Germany). Both the plasma and the lavage supernatants were immediately frozen and stored at −45 • C until the cytokines were measured.

Isolation of Total RNA and Reverse Transcription
1-2 mm-thin cross sections of the terminal ileum were placed in RNAlater (Qiagen, Hilden, Germany), and stored at −20 • C until RNA purification was performed. A Teflon piston homogenizer (Institute of Microbiology of the CAS, Novy Hradek, Czech Republic) and a Spin Tissue RNA Mini Kit (Stratec Molecular, Berlin, Germany) were used for total RNA purification according to the manufacturer's instructions. Five hundred ng of the total RNA, with ratio absorbance at 260 and 280 nm ≥2.0 as measured in 10 mM Tris-HCl buffer (pH 7.5), were used for reverse transcription with the QuantiTect Reverse Transcription kit (Qiagen). The RNA was incubated at 42 • C for 2 min with a genomic DNA wipeout buffer. The synthesis of cDNA with a mixture of random hexamers and oligo d(T) primers was performed at 42 • C for 20 min, and then heated at 95 • C for 3 min. 180 µl of PCR quality water (Life Technologies, Carlsbad, CA, USA) was added to 20 µl of the synthetized cDNA mixture to prepare the PCR template, which was stored at −25 • C until Real-Time PCR was performed.

Statistical Analysis
The counts of wild-type Salmonella Typhimurium and its isogenic ∆rfa mutants in the small intestine, their translocations into different organs, and the cytokine transcription and protein levels were compared among the groups with a Kruskal-Wallis multiple comparisons test with a post-hoc Dunn's test. One-way analysis of variance (ANOVA) with Tukey's post-hoc test was used in the evaluation of differences in the tight junction protein transcription levels. The statistical comparisons at p < 0.05 and graphs were processed by GraphPad 6 software (GraphPad Software, La Jolla, CA, USA).