Toll-Like Receptor 4 Signaling in the Ileum and Colon of Gnotobiotic Piglets Infected with Salmonella Typhimurium or Its Isogenic ∆rfa Mutants

Salmonella Typhimurium is a Gram-negative bacterium that causes enterocolitis in humans and pigs. Lipopolysaccharide (LPS) is a component of the outer leaflet of Gram-negative bacteria that provokes endotoxin shock. LPS can be synthesized completely or incompletely and creates S (smooth) or R (rough) chemotypes. Toll-like receptors (TLR) 2, 4, and 9 initiate an inflammatory reaction to combat bacterial infections. We associated/challenged one-week-old gnotobiotic piglets with wild-type S. Typhimurium with S chemotype or its isogenic ∆rfa mutants with R chemotype LPS. The wild-type S. Typhimurium induced TLR2 and TLR4 mRNA expression but not TLR9 mRNA expression in the ileum and colon of one-week-old gnotobiotic piglets 24 h after challenge. The TLR2 and TLR4 stimulatory effects of the S. Typhimurium ∆rfa mutants were related to the completeness of their LPS chain. The transcription of IL-12/23 p40, IFN-γ, and IL-6 in the intestine and the intestinal and plasmatic levels of IL-12/23 p40 and IL-6 but not IFN-γ were related to the activation of TLR2 and TLR4 signaling pathways. The avirulent S. Typhimurium ∆rfa mutants are potentially useful for modulation of the TLR2 and TLR4 signaling pathways to protect the immunocompromised gnotobiotic piglets against subsequent infection with the virulent S. Typhimurium.


Introduction
Lipopolysaccharide (LPS) is the major component of the outer membrane of Gram-negative bacteria. It composes hydrophobic domain lipid A, a core oligosaccharide, and an O-antigen [1,2]. LPS can be synthesized in complete smooth (S) or incomplete rough (R) forms (chemotypes) [1][2][3]. Wild-type Gram-negative bacteria (S chemotype) contains all three regions: (i) the O-polysaccharide chain, which is made up of repeating oligosaccharide units, (ii) the core oligosaccharide, and (iii) the lipid A which harbors the endotoxic activity of the entire molecule [1,4,5]. R chemotype characterizes incompletely synthesized LPS due to biosynthetic defects. The ∆rfa mutants lack the O-specific chain, and the core-oligosaccharides are presented in various degrees of completeness. The descending completeness classifies them into five main R chemotypes [3,6]. However, LPS of wild-type bacteria is

Virulence of Wild-Type Salmonella Typhimurium Strain LT2 for Germ-Free Piglets
Four one-week-old germ-free piglets that were orally infected with 1 × 10 8 CFUs of wild-type Salmonella Typhimurium LT2 strain died or were ante finem euthanized 36-48 h after the infection. Their germ-free counterparts thrived.

Colonization of Germ-Free Piglets with Wild-Type Salmonella Typhimurium and Its Isogenic ∆rfa Mutants
All germ-free (GF) piglets and piglets mono-associated for 24 h with wild-type S. Typhimurium (WT) or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants with variously truncated LPS survived the 24 h experimental period. The piglets infected with wild-type S. Typhimurium showed signs of salmonellosis (diarrhea, anorexia, sleepiness, and fever). The piglets associated with ∆rfa S. Typhimurium mutants showed intermediate signs of salmonellosis between WT and their absence in GF piglets that were less obvious in the piglets challenged with the ∆rfa mutant S. Typhimurium with highly truncated LPS chain. While it was possible to see this to a lesser extent than in the WT group in the ∆rfaL group, it was absent in the piglets challenged with ∆rfaG and ∆rfaC mutants.

Virulence of Wild-Type Salmonella Typhimurium Strain LT2 for Germ-Free Piglets
Four one-week-old germ-free piglets that were orally infected with 1 × 10 8 CFUs of wild-type Salmonella Typhimurium LT2 strain died or were ante finem euthanized 36-48 h after the infection. Their germ-free counterparts thrived.

Colonization of Germ-Free Piglets with Wild-Type Salmonella Typhimurium and Its Isogenic ∆rfa Mutants
All germ-free (GF) piglets and piglets mono-associated for 24 h with wild-type S. Typhimurium (WT) or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants with variously truncated LPS survived the 24 h experimental period. The piglets infected with wild-type S. Typhimurium showed signs of salmonellosis (diarrhea, anorexia, sleepiness, and fever). The piglets associated with ∆rfa S. Typhimurium mutants showed intermediate signs of salmonellosis between WT and their absence in GF piglets that were less obvious in the piglets challenged with the ∆rfa mutant S. Typhimurium with highly truncated LPS chain. While it was possible to see this to a lesser extent than in the WT group in the ∆rfaL group, it was absent in the piglets challenged with ∆rfaG and ∆rfaC mutants.

Relative Expression of TLR4, MD-2, CD14, LBP, TLR2, TLR9, MyD88, and TRIF mRNA in the Ileum
Relative TLR4 mRNA expression was induced by wild-type and ∆rfaL S. Typhimurium, but a statistically significant increase was found in the case of the WT group only (Figure 2A).  Relative TLR4 mRNA expression was induced by wild-type and ∆rfaL S. Typhimurium, but a statistically significant increase was found in the case of the WT group only (Figure 2A).
The WT piglets showed a significant increase in MD-2 ( Figure 2B). CD14 was statistically significantly increased in the WT and ∆rfaL groups and also differences between these groups were statistically significant ( Figure 2C). The induced LBP expression with wild-type S. Typhimurium was significantly higher from the GF and ∆rfaC groups but not from the ∆rfaL and ∆rfaG ones ( Figure 2D). Both wild-type and ∆rfaL S. Typhimurium significantly induced TLR2 expression ( Figure 2E). In contrast, wild-type S. Typhimurium and its ∆rfa mutants decreased TLR9 expression in comparison to the GF piglets, but this decrease was significant in the ∆rfaC group only ( Figure 2F). MyD88 expression was statistically significantly induced with wild-type S. Typhimurium only ( Figure 2G). The opposite trend was found in TRIF expression, but without any statistical significance ( Figure 3H).  relative mRNA expression (fold change) in the ileum. The differences were evaluated in control germfree piglets (GF) and piglets mono-associated with wild-type S. Typhimurium (WT) or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants. The values are presented as mean + SEM. Statistical differences were calculated by one-way ANOVA with Tukey's multiple comparison post-hoc test, and p-values < 0.05 are denoted with different letters above the columns. Six samples in each group were analyzed.
The WT piglets showed a significant increase in MD-2 ( Figure 2B). CD14 was statistically significantly increased in the WT and ∆rfaL groups and also differences between these groups were statistically significant ( Figure 2C). The induced LBP expression with wild-type S. Typhimurium was significantly higher from the GF and ∆rfaC groups but not from the ∆rfaL and ∆rfaG ones ( Figure 2D). Both wild-type and ∆rfaL S. Typhimurium significantly induced TLR2 expression ( Figure 2E). In contrast, wild-type S. Typhimurium and its ∆rfa mutants decreased TLR9 expression in comparison to the GF piglets, but this decrease was significant in the ∆rfaC group only ( Figure 2F). MyD88 expression was statistically significantly induced with wild-type S. Typhimurium only ( Figure 2G). The opposite trend was found in TRIF expression, but without any statistical significance ( Figure 3H).

Relative Expression of TLR4, MD-2, CD14, and LBP mRNA in the Colon
The wild-type and ∆rfaL S. Typhimurium statistically significantly upregulated TLR4 expression in the colon of the gnotobiotic piglets ( Figure 3A).  In contrast, neither wild-type nor mutant S. Typhimurium influenced the expression of MD-2 ( Figure 3B). Similar profiles to TLR4 expression with statistically significant upregulations by wildtype and ∆rfaL S. Typhimurium were in found CD14 ( Figure 3C) and LBP ( Figure 3D). Wild-type S. Typhimurium and its ∆rfaL mutant significantly increased the expression of TLR2 in the colon ( Figure  3E). In contrast, wild-type S. Typhimurium downregulated TLR9 expression, but this downregulation was not statistically significant ( Figure 3F). Wild-type, ∆rfaL, and ∆rfaG upregulated MyD88 expression, but only in the mutants were these upregulations statistically significant (Figure The differences were evaluated in the germ-free piglets (GF) and piglets mono-associated with wild-type S. Typhimurium (WT) or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants. The values are presented as mean + SEM. Statistical differences were calculated by one-way ANOVA with Tukey's multiple comparison post-hoc test, and p-values < 0.05 are denoted with different letters above the columns. Six samples in each group were analyzed.

Relative Expression of TLR4, MD-2, CD14, and LBP mRNA in the Colon
The wild-type and ∆rfaL S. Typhimurium statistically significantly upregulated TLR4 expression in the colon of the gnotobiotic piglets ( Figure 3A).
In contrast, neither wild-type nor mutant S. Typhimurium influenced the expression of MD-2 ( Figure 3B). Similar profiles to TLR4 expression with statistically significant upregulations by wild-type and ∆rfaL S. Typhimurium were in found CD14 ( Figure 3C) and LBP ( Figure 3D). Wild-type S. Typhimurium and its ∆rfaL mutant significantly increased the expression of TLR2 in the colon ( Figure 3E). In contrast, wild-type S. Typhimurium downregulated TLR9 expression, but this downregulation was not statistically significant ( Figure 3F). Wild-type, ∆rfaL, and ∆rfaG upregulated MyD88 expression, but only in the mutants were these upregulations statistically significant ( Figure 3G). TRIF showed a similar profile to TLR9 expression, but without any significant differences ( Figure 3H).  The wild-type S. Typhimurium induced a statistically significant increase in all observed cytokines in the ileum ( Figure 4A-C) and IFN-γ in the colon ( Figure 4F). The ∆rfaL mutant induced a statistically significant increase in IL-12/23 p40 and IFN-γ mRNA expression in the colon ( Figure  4D,F). Neither ∆rfaG nor ∆rfaC mutants influenced the mRNA expression of any cytokine in the ileum and colon ( Figure 4A-F).

Local and Systemic Levels of Inflammatory Cytokines IL-12/23 p40, IL-6, and IFN-γ
Wild-type S. Typhimurium induced increased statistically significantly ileal ( Figure Figure 5F) in the WT piglet group but a possible increase induced by its isogenic ∆rfa mutants was not statistically significant. No local or systemic levels of IFN-γ were detected, and this is the reason that IFN-γ-corresponding graphs are not included in Figure 5.  Figure 5F) in the WT piglet group but a possible increase induced by its isogenic ∆rfa mutants was not statistically significant. No local or systemic levels of IFN-γ were detected, and this is the reason that IFN-γ-corresponding graphs are not included in Figure 5.

Systemic Levels of C-Reactive Protein
Both WT and ∆rfaL groups showed statistically significantly increased levels of C-reactive protein (CRP) in blood plasma (Table 1). Moreover, the wild-type S. Typhimurium-induced CRP levels were significantly higher than its ∆rfaL mutant.

Discussion
Studies and vaccine applications against Salmonella or other pathogens are commonly targeted toward the production of specific antibodies [49][50][51]. In this study, we focused on the initial phase of the defense against infection that comprises a sense of infectious agents and the consequent production of inflammatory mediators. The inflammatory cytokines participate in innate response but also control the development of more effective and specific but delayed adaptive immune The differences were evaluated in the germ-free piglets (GF) and the piglets mono-associated with wild-type S. Typhimurium (WT) or its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants. The values are presented as median, maximal, and minimal values. Statistical differences were calculated by the Kruskal-Wallis test with Dunn's multiple comparisons post-hoc test, and p-values < 0.05 are denoted with different letters above the columns. Six samples in each group were analyzed.

Systemic Levels of C-Reactive Protein
Both WT and ∆rfaL groups showed statistically significantly increased levels of C-reactive protein (CRP) in blood plasma (Table 1). Moreover, the wild-type S. Typhimurium-induced CRP levels were significantly higher than its ∆rfaL mutant.

Discussion
Studies and vaccine applications against Salmonella or other pathogens are commonly targeted toward the production of specific antibodies [49][50][51]. In this study, we focused on the initial phase of the defense against infection that comprises a sense of infectious agents and the consequent production of inflammatory mediators. The inflammatory cytokines participate in innate response but also control the development of more effective and specific but delayed adaptive immune responses [52]. In our experiments, we used a well-defined laboratory LT2 strain of Salmonella Typhimurium [53]. This strain showed limited virulence for one-week-old conventional piglets but was avirulent for six-week-old conventional piglets [54]. The absence of the microbiota in germ-free animals determines their lowered colonization resistance [41] and, along with the absence of maternal immunoglobulins and cells [47], sensitizes the colostrum-deprived germ-free piglets to infection with S. Typhimurium strain LT2 [42,55]. However, mutations can modify its virulence [56,57]. In contrast to well-known virulence for the germ-free piglets, the lethality of S. Typhimurium LT2 for them was not established. Therefore, we associated one-week-old germ-free piglets with wild-type S. Typhimurium LT2 strain to clarify this result. The spontaneous death or euthanasia of the piglets ante finem occurred within 36-48 h after infection. This knowledge predetermined the 24 h duration of the following experiments, and we expected that all piglets would be alive for the whole 24 h experimental period.
Humans and pigs show similar sensitivity to LPS [58], but rodents are much less sensitive [59]. The interaction of LPS with cells of various lineages results in the formation and release of different inflammatory mediators [21,60]. An intense inflammatory reaction to LPS may provoke endotoxin shock and death [14,61]. R mutants of serovars Infantis and Typhimurium induced an inflammatory response that protected gnotobiotic piglets against subsequent infection with virulent S. Typhimurium serovars F98 [44] and LT2 [62]. However, these mutants resulted from spontaneous mutations with non-defined defects. A similar protective effect showed association of the gnotobiotic piglets with serum-sensitive probiotic E. coli Nissle 1917 with R chemotype of LPS [63]. In our experiments, we used isogenic ∆rfa S. Typhimurium strain LT2 mutants with defined deletions to modify LPS [64]. We expected that changes in LPS completeness in ∆rfa mutants can influence their virulence and ability to induce an inflammatory reaction. The ability of the ∆rfa mutants to colonize the sterile intestine of the one-week-old germ-free piglets and induce histopathological changes, modulation of the tight junction proteins claudin-1 and claudin-2 and occludin expression, as well as induction of inflammatory cytokines IL-8, IL-10, and TNF-α was verified elsewhere [64]. However, downstream signaling was never studied in experiments with non-defined Salmonella R mutants [44,62] or with defined ones [64]. Therefore, this work covers the analysis of bacterial recognition by TLR2, TLR4, and TLR9. Induced inflammatory mediators with different activities serve as biomarkers for the diagnosis of infection and sepsis [60,[65][66][67][68][69].
The distal ileum and colon are the main sites of Salmonella versus host interaction [37]. Relative constitutive expression of TLR2 and TLR4 mRNA was higher in the colon than in the ileum of weaned uninfected conventional pigs, but TLR9 showed the reverse relationship [70]. Fourteen-day infection with S. Typhimurium in these pigs increased the relative expression of TLR4 and TLR9 mRNA in the ileum. In contrast, the infection decreased TLR9 mRNA expression in the colon. Other six-week-old conventional piglets were orally infected with 3 × 10 9 CFU of S. Typhimurium. In the distal ileum, TLR2 and TLR4 relative mRNA expression significantly increased within 24 h, but TLR9 mRNA was not influenced [71]. Other authors described expressed mRNA of TLR2, TLR4, and TLR9 in the intestine of the conventional piglets orally infected with S. Typhimurium. They confirmed variously induced TLRs in the jejunum, ileum, and colon [72].
We found that infection with the wild-type S. Typhimurium upregulated mRNA in the terminal ileum in all analyzed members of the TLR4 signaling pathway-TLR4, MD-2, CD14, and LBP. In the case of the mutant Salmonella, the highest induction showed the ∆rfaL mutant, but it statistically significantly differed from the unstimulated control only exceptionally. Both other mutants with less complete LPS chains failed to induce any changes. This indicates that the lowering of LPS completeness decreased the ability of the mutant S. Typhimurium to trigger TLR4 signaling as a starting point of the inflammatory process. Conventional mice orally infected with 1 × 10 9 CFU of S. Typhimurium wild-type, ∆rfaL and ∆rfaG mutants proved that any truncation of the LPS chain decreased the ability of the mutant to translocate into mesenteric lymph nodes [50]. This finding is in agreement with our previous results obtained in the gnotobiotic piglets [64]. However, our results are in contrast with the results of other authors working with conventional mice infected with S. Typhimurium. In their experiments, the R mutant S. Typhimurium was superior in the upregulation of TLR4 expression to wild-type S. Typhimurium and was not dependent on the support of LBP and CD14 [7]. In contrast to sequential differences in TLRs among species, their function is similar [73]. This discrepancy between the immunostimulatory effect of R mutants in the gnotobiotic piglets and conventional mice could have several reasons. Mice show much lower sensitivity to LPS than pigs. This is probably due to the presence of various factors in murine plasma and their participation in LPS induced signaling [59,73]. S. Typhimurium causes enterocolitis in pigs but typhoid fever-like illness in mice. The streptomycin-induced suppression of microbiota diversity in conventional mice results in the change of the illness from typhoid fever-like to enterocolitis [36,37]. This points to the importance of microbiota in recognizing PAMPs and the possible role of quorum sensing for the expression of bacterial virulence factors [36,74].
MD-2, which creates a complex with TLR4, also enables TLR2 to recognize LPS and enhances TLR2-mediated responses to both Gram-positive and Gram-negative bacteria by recognizing peptidoglycan, lipoteichoic acid components, and LPS [75]. It was found that Gram-negative bacteria-induced innate response in conventional mice also took place via TLR2, but its contribution was neglected [76]. Highly upregulated expression of TLR2 mRNA dependent on the completeness of LPS attests to the participation of TLR2 in the discrimination of Salmonella. This was probably due to the cooperation of MD-2 and CD14, which are both crucial molecules in the TLR4 signaling pathway [77]. The TLR2 mRNA can be also upregulated indirectly, e.g., by a nuclear protein HMGB1 (high mobility group box1) that is an intrinsic ligand of TLR2 [78]. It was released from damaged enterocytes of the wild-type and ∆rfaL S. Typhimurium-infected gnotobiotic piglets [63,64].
MyD88 is the common adaptor protein in TLR2 and TLR4 signaling pathways, but TRIF participates in TLR4 and TLR9 signaling pathways [22,23]. The MyD88 mRNA upregulation is in concordance with the upregulation of TLR2 and TLR4 mRNA expression. The LPS completeness influences the activation of TLR4 signaling via the MyD88 and TRIF pathways [79]. In our case, we conclude that the primary signaling pathway was that which took place through MyD88.
CD14 is a membrane protein found in myeloid cells. A cleaved CD14 is present in serum in its soluble form. It participates in the monomerization of LPS molecule and its better recognition by the TLR4/MD-2 complex. The soluble CD14 enables epithelial, endothelial, and other cells that do not express membrane CD14 to sense LPS and respond to it [80]. It was proven that antibody-mediated neutralization of systemic CD14 modulates the TLR4 signaling pathway induced by E. coli-derived LPS or virulent E. coli in a pig model of sepsis. This modulation resulted in the induction of low levels of inflammatory cytokines IL1β, IL-6, IL-8, and TNF-α and decreased activation of neutrophils [69]. CD14 is a co-receptor of TLR4, encompassing both TLR2 [25,77] and TLR9 [26]. This overlapping of CD14 in TLR2, TLR4, and TLR9-driven signaling pathways led us to include both TLR2 and TLR9 in our analyses primarily targeted toward TLR4.
Around 200 inflammatory biomarkers have been identified for the evaluation of sepsis. Among the most frequently used are procalcitonin, C-reactive protein (CRP), interleukin (IL)-6, IL-8, IL-10, IL-12, and tumor necrosis factor (TNF)-α [60,81]. Levels of inflammatory cytokines indicate whether they participate in physiological functions or mediate a deleterious effect on hosts [14,16]. The gnotobiotic piglets infected with necrotoxigenic E. coli O55 that thrived showed low or undetectable levels of intestinal and systemic inflammatory cytokines 24 h post-infection. In contrast, the piglets with distinctly expressed clinical signs of infection or piglets ante finem had high levels [82]. In this work, three inflammatory cytokines with different action, IL-8, IL-12/23 p40, and IL-6, and CRP were taken to indicate inflammatory processes or sepsis. IL-12/23 p40 is a shared subunit of IL-12 and IL-23 that forms dimers with p35 and p19 subunits, respectively [83]. IL-12/23 p40 is upregulated by microbial stimuli [84] and both IL-12 and IL-23 play an essential role in the intestinal inflammation and host resistance in Salmonella infections [65,66]. Moreover, IL-12 was approved as a relevant biomarker of sepsis in the infectious pig model [85]. Upregulation of IL-12/23 p40 mRNA and its higher levels in wild-type S. Typhimurium and the ∆rfaL mutant confirmed higher virulence of this S. Typhimurium compared to S. Typhimurium ∆rfaG and ∆rfaC mutants with lower completeness of their LPS chain. The leading producers of IL-12 and IL-23 are antigen-presenting cells [83]. IL-12 controls differentiation of naive T-cells into IFN-γ producing Th1 cells and induces IFN-γ production by NK cells and innate lymphoid cells [86,87]. The produced IFN-γ promotes phagocytosis by macrophages and the destruction of phagocytosed microbes by free radicals in phagosomes [66,88]. Regarding IFN-γ, we found highly expressed IFN-γ mRNA in the piglets infected with the wild-type S. Typhimurium, but these values decreased with decreasing degrees of LPS completeness. In contrast to this unequivocal effect on IFN-γ mRNA expression, no local or systemic IFN-γ levels were found in the intestine or blood plasma of any of the S. Typhimurium challenged piglets. Cells that sense infection produce one cytokine set that induces lymphocytes to produce another cytokine set that activates the effector response. While IL-12, IL-23, and IL-6 belong to the first cytokine set, IFN-γ belongs to the second set [52]. This may be the reason that, 24 h after challenge, IFN-γ clearly shows upregulation of its mRNA, but this time may be too short for the appearance of IFN-γ secreted protein on local and systemic levels.
Finally, we measured IL-6 as the third inflammatory cytokine. In clinical studies, according to Sepsis-3 criteria [89], the diagnostic and prognostic value of IL-6 was superior to other compared markers for both sepsis and septic shock [67]. Both local and systemic IL-6 levels in the S. Typhimurium-challenged gnotobiotic piglets were related to the completeness of LPS. The pro-inflammatory cytokine IL-6 is closely related to acute-phase proteins, including their diagnostically frequently used CRP that is synthesized in the liver [90]. CRP served in our study as "a classical" inflammatory marker that has been used by clinicians for decades [91]. The synthesis of CRP in the liver [92] was the reason that we did not measure its intestinal levels, as in the case of cytokines that are produced by different cell populations, including intestinal epithelial cells [20,93].

Conclusions
The wild-type Salmonella Typhimurium LT2 strain with complete LPS chain highly induced TLR2 and TLR4 mRNA expression but not TLR9 mRNA expression in the ileum and colon of one-week-old gnotobiotic piglets 24 h after challenge. The TLR2 and TLR4 stimulatory effects of the S. Typhimurium ∆rfa mutants related to the completeness of their LPS chain (wild-type > ∆rfaL > ∆rfaG > ∆rfaC) [64]. TLR4 signaling was transduced through MyD88 but TRIF played no role or a marginal role only. The transcription of inflammatory cytokines IL-12/23 p40, IFN-γ, and IL-6 in the intestine and the intestinal and plasmatic levels of IL-12/23 p40 and IL-6 but not IFN-γ were related to the activation of TLR2 and TLR4 signaling pathways. The plasmatic levels of CRP correlated with the cytokine levels. Future long-term studies are needed to verify the avirulence of the S. Typhimurium LT2 strain ∆rfa mutants and their potential usefulness in modulation of the TLR2 and TLR4 signaling pathways for the protection of immunocompromised gnotobiotic piglets against subsequent infection with the virulent wild-type S. Typhimurium LT2 strain.

Ethical Statement
The Animal Care and Use Committee of the Czech Academy of Sciences approved all experiments with animals (protocol #63/2015; 6 November 2015).

Bacterial Cultures
Wild-type Salmonella enterica serovar Typhimurium strain LT2 and its isogenic ∆rfaL, ∆rfaG, and ∆rfaC mutants with different LPS chemotypes were obtained from the collection of microorganisms of the Laboratory of Gnotobiology of the Institute of Microbiology of the Czech Academy of Sciences. The mutant Salmonella strains were constructed as described previously [64]. The completeness of their LPS decreased in the order of wild-type > ∆rfaL > ∆rfaG > ∆rfaC. Wild-type Salmonella has LPS of S chemotype, and all ∆rfa mutants have their LPS of R chemotypes [64].
The bacterial suspensions of wild-type S. Typhimurium strain LT2 and its isogenic ∆rfa mutants were prepared by cultivation on meat peptone agar slopes (blood agar base; Oxoid, Basingstoke, UK) at 37 • C overnight. The bacteria were scraped from the agar and resuspended in PBS to approximately 5 × 10 8 colony forming units (CFUs)/mL, as measured by spectrophotometry at 600 nm. The CFU counts were verified by a cultivation method on Luria-Bertani agar (Difco Laboratories, Detroit, MI, USA) at 37 • C for 24 h.

Gnotobiotic Piglets
Germ-free piglets of miniature Minnesota-derived pig breed [94] (Animal Research Institute, Kostelec nad Orlici, Czechia) were obtained by hysterectomy under isoflurane anesthesia at expected term, reared in positive-pressure fiber-glass isolators, fed via nipple by cow's milk-based formula (Mlekarna Hlinsko, Hlinsko, Czechia), and microbiologically checked, as described in detail elsewhere [45]. Each piglet group was created from three independent hysterectomies.

Virulence of Wild Type Salmonella Typhimurium Strain LT2 for Germ-Free Piglets
In total, 1 × 10 8 CFUs of wild-type S. Typhimurium were orally administered in 5 mL of milk diet to one-week-old germ-free piglets (n = 4). The piglets were derived from two independent hysterectomies.

Challenge of the Germ-Free Piglets with Wild-Type and ∆rfa Mutant Salmonella Typhimurium
One-week-old germ-free piglets were orally infected with 1 × 10 8 CFUs of wild-type (WT), ∆rfaL, ∆rfaG, or ∆rfaC S. Typhimurium in 5 mL of milk diet. The control germ-free piglets (GF) received 5 mL of milk only. Each piglet group created six piglets that were derived from 3 independent hysterectomies.

Tissue Sample Collections
Twenty-four hours after the challenge, we euthanized the piglets by cardiac puncture exsanguination under isoflurane inhalation anesthesia (Piramal Healthcare UK). Then, 1-2 mm thick cross-sections of the terminal ileum and transversal colon without lumen contents were put into RNAlater (Qiagen, Hilden, Germany) and stored at −20 • C until total RNA purification was performed.

Immunohistochemistry Detection of TLR4 in the Colon
The transverse colon was embedded in Tissue-Tek (Sakura, Tokyo, Japan), snap-frozen in isopentane, cooled in liquid nitrogen vapor, and stored at −70 • C. Then, 5-µm acetone-fixed cryosections were cut on a cryostat CM 1860 UV (Leica Microsystems, Wetzlar, Germany) and put on SuperFrost/Plus slides (Thermo Fisher Scientific, Darmstadt, Germany) and were kept at −40 • C until labeling. The sections were incubated with 10% normal rabbit serum (Life Technologies, Carlsbad, CA, USA) in a humid chamber for one h at RT. Labeling by anti-TLR4 rabbit polyclonal antibodies (Novus Biologicals, Centennial, CO, USA) was performed overnight at 4 • C. The sections were incubated with secondary antibody, peroxidase-conjugated F(ab')2-goat anti-rabbit IgG (H+L) (Life Technologies, Carlsbad, CA, USA) for 2 h at RT. The TLR4 localization was visualized by an incubation with AEC substrate kit (Sigma-Aldrich, St. Louis, MO, USA) and examined under an Olympus BX 40 microscope with Olympus Camedia C-2000 digital camera (Olympus, Tokyo, Japan). Control sections without primary antibody were treated in the same way.

Isolation of Total RNA and Reverse Transcription
The total RNA was purified and was reversely transcribed, as we described previously [64]. Briefly, 1-2 mm thick cross-sections of the ileum and colon stored at −20 • C in RNAlater (Sigma-Aldrich) were homogenized by a Teflon piston homogenizer (Institute of Microbiology, Novy Hradek, Czechia) in 1.5 mL Eppendorf tubes and purified by the Spin Tissue RNA Mini Kit (Stratec Molecular, Berlin, Germany) according to the manufacturer's instructions. The purity and concentration of the RNA in 10 mM Tris-HCl pH 7.5 buffer were evaluated at 260 and 280 nm, and samples with ratio absorbances A 260 /A 280 ≥ 2.0 were used for cDNA synthesis. Five hundred ng of the total RNA was reverse transcribed with the QuantiTect Reverse Transcription kit (Qiagen) according to the manufacturer's recommendation. The synthesized cDNA mixture was diluted 1/10 with PCR quality water (Life Technologies, Carlsbad, CA, USA), and these PCR templates were stored at −25 • C until real-time PCR was performed.

Real-Time PCR
Real-time PCR was performed as we described previously [45], and the used a locked nucleic acid (LNA) probe-based real-time PCR systems are listed in Table 2.  9 TTCCTGTGTCCATGAAAACTTC AGGTACCAGTGGCCCTGAAT 77 IL-6 9 CAAAGCCACCACCCCTAAC TCCACTCGTTCTGTGACTGC 40 INF-γ 10 TGGAAAGAGGAGAGTGACAAAAA GAATGGCCTGGTTATCTTTGA 21 Briefly, two µL PCR template was added to 18 µL of the PCR master mix (FastStart Universal Probe Master mix; Roche Diagnostics, Darmsted, Germany) containing 500 nM each of the forward and reverse primers (Generi-Biotech, Hradec Kralove, Czechia) and 100 nM LNA (lock nucleic acid) probe (Universal ProbeLibrary; Roche Diagnostics). Ten minutes' initial heating at 95 • C was followed by 45 cycles at 95 • C for 15 s and 60 • C for 60 s. Samples were incubated and measured in duplicates on an iQ cycler with iQ5 Optical System Software 1.0 (Bio-Rad, Hercules, CA, USA). Cq for genes of interest were normalized to β-actin and cyclophilin A, and the relative mRNA fold change expressions were calculated by GenEx 6.1 software (MultiD Analyses AB, Gothenburg, Sweden) according to the 2 −CT method [95].

Statistical Analysis
One-way analysis of variance (ANOVA) with Tukey's post-hoc test was used in the evaluation of differences in the mRNA expressions and CRP levels. Kruskal-Wallis test with Dunn's post-hoc test was used in the evaluation of differences in the cytokine protein expressions. The different letters indicate statistically significant differences at p < 0.05. The same letter indicates no statistically significant differences. The statistical comparisons and graphs were processed by GraphPad Prism 6 software (GraphPad Software, San Diego, CA, USA).