ABSTRACT
Salmonella enterica serovar Typhi causes the systemic disease typhoid fever. After ingestion, it adheres to and invades the host epithelium while evading the host innate immune response, causing little if any inflammation. Conversely, Salmonella enterica serovar Typhimurium causes gastroenteritis in humans and thrives in the inflamed gut. Upon entering the host, S. Typhimurium preferentially colonizes Peyer's patches, a lymphoid organ in which microfold cells (M cells) overlay an arrangement of B cells, T cells, and antigen-presenting cells. Both serovars can adhere to and invade M cells and enterocytes, and it has been assumed that S. Typhi also preferentially targets M cells. In this study, we present data supporting the alternative hypothesis that S. Typhi preferentially targets enterocytes. Using a tissue culture M cell model, we examined S. Typhi strains with a deletion in the stg fimbriae. The stg deletion resulted in increased adherence to M cells and, as expected, decreased adherence to Caco-2 cells. Adherence to M cells could be further enhanced by introduction of the long polar fimbriae (Lpf), which facilitate adherence of S. Typhimurium to M cells. Deletion of stg and/or introduction of lpf enhanced M cell invasion as well, leading to significant increases in secretion of interleukin 8. These results suggest that S. Typhi may preferentially target enterocytes in vivo.
INTRODUCTION
Salmonella enterica strains are responsible for 120 million illnesses and 365,000 deaths each year worldwide (1, 2). With regard to the ability to cause infection, S. enterica can be broadly divided into two classes: generalists, which infect a variety of animals, and host-specific strains infecting only a single host. The generalists, such as Salmonella enterica serovar Typhimurium and Salmonella enterica serovar Enteritidis, cause gastrointestinal disease typified by localized gut inflammation (3). In most cases, the disease is self-limiting in immunocompetent individuals. Generalists, such as S. Typhimurium, elicit a robust host immune response by preferentially targeting the microfold (M) cells present on the lumenal surfaces of the Peyer's patches (PPs) (4–6), while enterocytes are a less favorable invasion target (7, 8). After invasion into the PPs, S. Typhimurium rapidly encounters dendritic cells (DCs) and is phagocytized (9, 10). These Salmonella-containing DCs may interact directly with B cells within the PPs, resulting in IgA switching and production of intestinal IgA (11). In addition, T-cell priming by the Salmonella-containing DCs begins in the PPs and continues in the deeper immunological tissues (e.g., the spleen) (12, 13), resulting in activation of B cells and CD4+ and CD8+ T cells, leading to production of a systemic cell-mediated and humoral immune response (13–15). S. Typhimurium stimulates a strong proinflammatory immune response that assists with effector cell recruitment and DC maturation. During invasion of PPs and intestinal epithelial cells, the host immune system is exposed to numerous pathogen-associated molecular patterns (PAMPs) produced by S. Typhimurium, including flagella, lipopolysaccharide (LPS), and bacterial DNA (16–19). These PAMPs are recognized by their cognate Toll-like receptors (TLRs) and result in the secretion of interleukin 8 (IL-8) and the proinflammatory cytokines IL-1β, IL-6, tumor necrosis factor alpha (TNF-α), and gamma interferon (IFN-γ) (20–23). The production of these cytokines recruits and activates neutrophils, as well as monocytes and DCs (24).
Infection by the human-specific serovar Salmonella enterica serovar Typhi does not result in a proinflammatory cytokine cascade, nor are large numbers of neutrophils or monocytes recruited to the infection site (25–29). Instead, production of proinflammatory cytokines is suppressed (27). Some of this immunological silence can be attributed to the actions of the immunosuppressive Vi capsule (30–32), which masks important TLR ligands, such as lipopolysaccharide and flagella (28, 33). The TviA protein, which regulates Vi production, also downregulates flagella after cellular invasion (34). Further, TviA also downregulates genes encoding the type 3 secretion system (T3SS) located in Salmonella pathogenicity island 1, thereby avoiding activation of NF-κB in epithelial cells (35).
An additional problem in stimulating an immune response to S. Typhi is that the invading bacteria do not target the PPs as efficiently as S. Typhimurium. While it is possible to detect S. Typhi in the PPs shortly after inoculation (36, 37), the long polar fimbriae (Lpf) responsible for M cell recognition and attachment in S. Typhimurium are not present in S. Typhi (38, 39).
When designing S. Typhi-based vaccines, the goal is to achieve high immunogenicity while keeping reactogenicity low (40). Attenuation strategies that result in low reactogenicity are often poorly immunogenic (41). We have been exploring approaches to enhance immunogenicity independently of the attenuation strategy. We hypothesized that we could improve the immunogenicity of S. Typhi strains by enhancing their ability to attach to M cells in the Peyer's patches.
The stg fimbrial operon is present in S. Typhi but not in S. Typhimurium (42). The presence of Stg fimbriae has been demonstrated in Escherichia coli (43), though production of Stg fimbriae in S. Typhi has not been demonstrated. This is an important point, as the gene encoding the chaperone protein, stgC, is a pseudogene in S. Typhi (39, 42). The E. colistgC gene encodes an 841-amino-acid (aa) protein, while in S. Typhi, there is a premature stop at codon 171 (42, 44), resulting in a 170-aa protein. There is also a downstream open reading frame that potentially encodes a 605-aa protein designated StgC′ (44). Thus, the S. Typhi operon is designated stgABCC′D. One ramification of this is that Stg fimbriae may not be produced in S. Typhi. However, there are data supporting the idea that an stg-encoded adhesin plays a role in S. Typhi interactions with host cells. The presence of the stgABCC′D operon in S. Typhi enhances attachment to human epithelial cells (44, 45). S. Typhi mutants carrying a deletion of stgABCC′D exhibit reduced attachment to INT407 (44) and Hep2 (45) monolayers. Conversely, introduction of S. Typhi stgABCC′D into E. coli strain ORN172 enhances its ability to bind to INT407 cells (44), while ectopic expression of the S. Typhi stg operon in S. Typhimurium increases its attachment to polarized HT-29 cells (45). Thus, we infer that an stg-encoded fimbria or adhesion is produced in spite of the apparent lack of a functional chaperone and that this adhesin promotes attachment to enterocytes in the gut. Stg fimbriae/adhesin may also have a role in the interaction of S. Typhi with macrophages, though whether it inhibits (44) or enhances (45) phagocytosis is unclear.
The S. Typhimurium Lpf, not present in S. Typhi, enhance colonization of Peyer's patches via interactions with M cells (6). Lpf also play a role in the early stages of biofilm formation on host epithelial cells (46), which could explain their involvement in intestinal persistence (47). Synthesis of Lpf is regulated by an on-off switch mechanism (phase variation) to avoid host immune responses (48).
S. Typhi migrates through polarized C2BBe cells (a subclone of Caco-2 cells) more efficiently than S. Typhimurium (49). In addition, adherence and invasion of human-derived Int-407 by S. Typhi are 3- to 10-fold more efficient than those by S. Typhimurium (8). The proficiency of S. Typhi in adhering to enterocytes is due, at least in part, to the presence of the Stg fimbriae/adhesin (44, 45). Further, it is also likely that S. Typhimurium is more proficient than S. Typhi at colonizing Peyer's patches due, in part, to the presence of Lpf. Based on these observations, we propose that after oral ingestion, enterocytes are the preferred target of invasion by S. Typhi, while M cells are the preferred targets of S. Typhimurium.
As a strategy to redirect S. Typhi to Peyer's patches, we reasoned that modifying S. Typhi to express the lpfABCDE operon would enhance attachment to M cells. We also deleted the stg operon as a way to reduce adherence of S. Typhi to enterocytes. In the current study, we compared the adherence and invasion capabilities of S. Typhi with or without production of Stg fimbriae and/or Lpf in polarized Caco-2 cultures and in a previously described M cell model system in which Raji B cells were cocultured with Caco-2 cells to drive the formation of M-like cells (50, 51). Our results confirm that Stg facilitates adherence to epithelial cells. Our findings also show that adherence of S. Typhi to M-like cells is significantly enhanced when the stg operon is deleted and that introduction of Lpf in S. Typhi enhances adherence and invasion of M-like cells. Finally, we show that an S. Typhi Δstg strain producing Lpf elicits levels of interleukin 8 secretion in M-like cells similar to those of S. Typhimurium.
RESULTS
Tissue culture models.The Caco-2/Raji B cell coculture model has been previously described (50, 51). We validated the model in our hands (see the supplemental material), demonstrating that the polarized cells react with an anti-ZO-1 antibody, indicating the presence of tight junctions (see Fig. S1A and D in the supplemental material). We also probed for the presence or absence of cell markers, including those associated with enterocytes (UEA binding) (see Fig. S1B and E in the supplemental material) or M cells (increased galectin-9 production) (see Fig. S1C and F in the supplemental material) (52). Our results indicated that the M-like cells in our coculture system show reduced affinity for UEA and increased galectin-9 production, consistent with the expected M cell phenotype.
Cloning of stgABCC′D and lpfABCDE.The stgABCC′D and lpfABCDE operons were cloned by PCR into the low-copy-number plasmid pWSK29, as described in Materials and Methods. The DNA sequence of each operon was confirmed by DNA sequence analysis. We were unable to clearly demonstrate production of Stg fimbriae and Lpf in S. Typhi directly by Western blotting or electron microscopy (data not shown). However, we were able to use an indirect method. Mice were immunized 4 times, alternating between oral and intranasal (i.n.) immunizations, with either ISP1820, ISP1820 Δstg, ISP1820 Lpf+, or ISP1820 Δstg Lpf+. Sera were collected from each group of mice 7 weeks after the initial inoculation, and titers against recombinant LpfA (rLpfA) and rStgA were determined. Although titers against both of these proteins were low, our results showed that mice inoculated with strains capable of producing StgA mounted an anti-rStgA serum IgG response that was 5- to 10-fold greater than that in mice inoculated with the Δstg strain (see Fig. S2 in the supplemental material). Conversely, mice immunized with strains capable of producing LpfA had elevated serum anti-rLpfA titers. Taken together, these results indicate that the cloned operons were capable of directing production of stg or lpf operon-encoded proteins in S. Typhi. Thus, it is likely that strains carrying the cloned fimbrial operons produce the expected fimbriae or adhesin.
The presence of Stg fimbriae/adhesin reduces adherence to M cells.We examined S. Typhi adherence to polarized Caco-2 monolayers and Caco-2/Raji B cell cocultures (M-like cells). The S. Typhi wild-type strains ISP1820 and Ty2 adhered to Caco-2 cells in greater numbers than S. Typhimurium (P < 0.0001), while adherence of S. Typhimurium to M-like cells was greater than that of the S. Typhi wild-type strains (Fig. 1A) (P ≤ 0.02). Both S. Typhi wild-type strains showed a strong preference for binding to Caco-2 cells over M-like cells (P ≤ 0.01). Deletion of stg reversed this trend. As expected, deletion of stg reduced adherence to Caco-2 cells (P < 0.0001). However, adherence to M-like cells by the stg mutants was significantly increased compared to ISP1820 and Ty2 (Fig. 1A) (P ≤ 0.02). Finally, the Δstg mutant showed a strong preference for binding to M-like cells compared to Caco-2 cells (P ≤ 0.01). Introduction of a plasmid-borne copy of stg into the ISP1820-derived Δstg mutant resulted in an adherence profile similar to that of the Stg+ parent, ISP1820 (Fig. 1A). Finally, expression of stg in S. Typhimurium resulted in increased adherence to Caco-2 cells, as seen previously (45), and a noticeable but nonsignificant decrease in adherence to M-like cells.
S. Typhi adherence to Caco-2 and M-like cells. All the Salmonella strains carried either the empty vector pWSK29 or a derivative containing the indicated fimbrial operon. (A) Salmonella cells were added to tissue culture wells at an MOI of 10 and allowed to incubate for 1 h. The percentages of the inoculum associated with Caco-2 or M-like cells after 1 h of incubation are shown. Significant differences (P ≤ 0.004) in adherence to M-like and Caco-2 cells for each strain are indicated by brackets. Significant differences (P ≤ 0.0133) between mutant and parental-strain adherence to M-like cells (#) or Caco-2 cells (*) are indicated. The data are expressed as geometric means and standard errors of the mean. (B) The indicated Salmonella strains were added to tissue culture wells at an MOI of 100 and allowed to incubate for 1 h. Shown is immunocytochemistry of Caco-2 cells with (a and b) or without (c and d) coculture of Raji B cells. The slides were stained with anti-Salmonella CSA-1 (a and c) and a merge of all stains (b and d) with anti-Salmonella CSA-1 (green), anti-ZO-1 (yellow), anti-galectin-9 (red), and DAPI (4′,6-diamidino-2-phenylindole) (blue). Bars, 25 μm. Plasmid pKR012 carries the lpfABCDE operon in vector pWSK29.
Adherence preferences for key strains were confirmed by confocal microscopy (Fig. 1B). Note that the confocal data were obtained using a multiplicity of infection (MOI) of 100 and that the data in Fig. 1A were obtained using an MOI of 10. However, the trends observed in Fig. 1A were similar using an MOI of 100 (see Fig. S3 in the supplemental material).
The presence of Lpf reduces adherence to Caco-2 cells.Since lpf has been implicated in binding to M cells, we examined the impact of introducing a plasmid-borne copy of lpfABCDE into S. Typhi. Addition of lpf enhanced adherence of ISP1820 to M cells (Fig. 1A), as expected, although this difference was not significant (P = 0.104). Surprisingly, the presence of plasmid-borne Lpf resulted in a slight, nonsignificant (P = 0.96) decrease in the adherence of the Δstg mutant to M cells. Interestingly, we observed a similar decrease in adherence when the lpf clone was introduced into S. Typhimurium. Finally, introduction of lpf reduced adherence of ISP1820 to Caco-2 cells (P < 0.0001). These quantitative results are supported by the confocal data (Fig. 1B).
Lpf enhance S. Typhi invasion of M cells.We examined the impact of Stg fimbriae and Lpf on invasion. Deletion of stg resulted in decreased invasion of Caco-2 cells and increased invasion of M-like cells for both Ty2 and ISP1820 (P < 0.0001) (Fig. 2), a phenotype that was reversed by complementation with a plasmid copy of stg. Introduction of lpf into ISP1820 or ISP1820 Δstg resulted in a significant increase in invasion of M-like cells (P ≤ 0.003). Introduction of lpf into the ISP1820 Δstg strain resulted in a significant increase in M-like-cell invasion compared to ISP1820 Δstg (P < 0.0001). This was also greater than the level of invasion seen with S. Typhimurium (P < 0.0001). These results suggest that, in this setting, Lpf actively promote M cell invasion.
Salmonella invasion of polarized Caco-2 and M-like cells. All the Salmonella strains carried either the empty vector pWSK29 or a derivative containing the indicated fimbrial operon. Salmonella strains were added to tissue culture wells at an MOI of 10. The ability to invade Caco-2 cells was different from the ability to invade M-like cells for all the strains except S. Typhimurium plus lpf (P ≤ 0.0007). Significant (P < 0.0001) differences between mutant and parental-strain invasion of M-like cells (#) or Caco-2 cells (*) are indicated. The data are expressed as geometric means and standard errors of the mean.
Impact of Stg fimbriae and Lpf on IL-8 secretion.The Salmonella strains were incubated with M-like cells for 4 h, at which time the basolateral medium was assayed for IL-8. We monitored lactate dehydrogenase (LDH) release as a measure of cell death. In all experiments, LDH levels were low, averaging 5% or less of the amount of LDH detected in lysed cells (data not shown). For ISP1820, there was little IL-8 secretion (Fig. 3). Introduction of Lpf resulted in a slight increase in IL-8 secretion, as did deletion of the stg operon, although neither increase was significant. However, introduction of Lpf into the Δstg S. Typhi strain resulted in a roughly 10-fold increase in IL-8 levels compared to the parent strain (P = 0.003). The secretion profile of the Δstg Lpf+S. Typhi strain was similar to the profile observed for S. Typhimurium (Fig. 3), indicating that the presence or absence of these fimbriae can impact host responses to invasion.
Basal secretion of IL-8 by polarized M-like cells 4 h after inoculation with S. Typhi or S. Typhimurium. All the strains had levels of IL-8 significantly (P ≤ 0.008) different from those of the phorbol myristate acetate (PMA) control. *, significant (P ≤ 0.003) differences from S. Typhimurium. Brackets indicate differences between strains (P ≤ 0.012). The data are expressed as geometric means and standard errors of the mean.
DISCUSSION
Peyer's patches have been referred to as the “immune sensors of the gut” (53). They are lymphoid follicles surrounded by a follicle-associated epithelium that includes M cells. The M cells are situated above B cell follicles, T cells, dendritic cells, and macrophages. Peyer's patches are required for the induction of mucosal antibody responses against Salmonella (54). Bypassing the PPs during infection has been shown to dramatically reduce anti-Salmonella mucosal immune responses (54–56). Mouse studies using an S. Typhimurium invA mutant defective in colonization of Peyer's patches showed that such colonization is an important driver of mucosal IgA responses (55). S. Typhi evades the host immune system by downregulating or masking its TLR and NOD agonists, including LPS, flagella, and T3SS components and effectors (28, 34, 35). We hypothesize that S. Typhi also preferentially avoids Peyer's patches during the initial interactions with the host. This would comprise an additional strategy for avoiding the host immune system.
While it is clear that S. Typhi can adhere to and invade M cells in the mouse (36, 37), it has not been possible to quantify how frequently this occurs versus the frequency at which it invades enterocytes. This question is not easily addressed because (i) the surface area covered by enterocytes is large compared to the tight clusters of M cells in Peyer's patches and (ii) addressing this question is technically challenging, since S. Typhi is host restricted to humans. However, there are data on how S. Typhimurium targets Peyer's patches that may shed light on the issue.
Unlike S. Typhi, S. Typhimurium elicits an inflammatory response in the human gut. One advantage of an inflamed gut for S. Typhimurium is the generation of nitrate, which it uses as a terminal electron acceptor, fueling a Salmonella bloom in the lumen through anaerobic respiration (57). Prior to invasion, nitrate is constitutively produced in the ileum, generating a nitrate gradient into the intestinal lumen (58). S. Typhimurium follows this gradient toward the Peyer's patches, where it colonizes the associated mucus layer and invades. This chemotactic behavior requires the nitrate sensor Tsr, a methyl-accepting chemotaxis protein. There are no data on how S. Typhi senses host epithelial cells lining the ileum, but S. Typhi only transiently colonizes the gut and does not rely on creating a nitrate-rich environment. In this regard, it is interesting that tsr is a pseudogene in S. Typhi (38, 39). Thus, it is possible that S. Typhi uses a different mechanism and may, in fact, migrate toward enterocytes instead of M cells. To our knowledge, our study provides the first evidence that S. Typhi displays a preference for adhering to enterocytes over M cells, while as expected, S. Typhimurium preferentially adheres to M cells (Fig. 1).
Despite the severely truncated chaperone protein, the stg operon appears to encode a functional adhesin. Strains carrying the stg operon are known to enhance S. Typhi adherence to a number of epithelial cell lines, including INT-407 cells (44), HEp2 cells, and polarized HT-29 cells (45). We have expanded these findings to include polarized Caco-2 cells. Deletion of stg significantly reduced adherence to polarized Caco-2 cells (Fig. 1A), and complementing the mutation with a plasmid-borne copy of stg restored adherence (Fig. 1A). The quantitative data were supported by our confocal-microscopy data, in which fewer S. Typhi cells were observed on the Caco-2 monolayers after deletion of stg (Fig. 1B).
Surprisingly, the stg mutant was better able to adhere to M cells than the parent strain in the quantitative assays (Fig. 1A; see Fig. S3 in the supplemental material). Complementation of the stg mutant resulted in an adherence profile similar to that of the Stg+ parent (Fig. 1A). The results with confocal microscopy were consistent with those of the quantitative assays (Fig. 1B). In fact, the confocal-microscopy data indicated that S. Typhi Δstg adhered to M cells as well as S. Typhimurium. Taken together, these results suggest that the presence of the Stg fimbriae/adhesin interferes with the adherence of S. Typhi to M cells and indicate that S. Typhi produces an adhesin that facilitates adherence to M cells when Stg is absent. The nature of this M cell adhesin is unknown. Any of the 11 additional fimbrial operons in S. Typhi (42) could potentially perform this function, though many are considered putative and some, like the stg operon, carry pseudogenes. Stg interference with M cell binding could be due to steric hindrance of the M cell fimbriae/adhesin or occur via inverse regulation of stg operon and M cell adhesin operon expression.
Introduction of the lpf operon into the S. Typhi wild-type strain interfered with its ability to adhere to Caco-2 cells (Fig. 1A and B). However, in both the wild type and the stg mutant, lpf increased invasion of M-like cells (P ≤ 0.003) (Fig. 2). We speculate that this was driven by Lpf-mediated intimate contact, allowing the type 3 secretion system of Salmonella pathogenicity island 1 to initiate docking. In S. Typhimurium, type 1 fimbriae have been shown to account for intimate contact prior to SPI-1-mediated docking to HeLa cells (59).
Invasion of polarized epithelial cells by S. Typhimurium induces expression and production of the chemokine IL-8, while S. Typhi is able to suppress this response (30). In our study, we examined the amount of IL-8 secreted into the basal chamber after infection of M cells with S. Typhi at 4 h. Our results were consistent with previous work using a variety of cell types, including Caco-2 and T84 epithelial cells, HEK cells, THP-1 monocytes, and human colonic tissue explants (30, 34). S. Typhimurium induced over 10-fold-greater levels of IL-8 than S. Typhi in the M-cell model (Fig. 3). The lower levels of IL-8 produced by cells infected with S. Typhi have been shown to be caused by immune avoidance due to several virulence traits, including Vi antigen (60) and the downregulation of flagellin (30, 34) and T3SS effectors (35) by TviA, the regulator of Vi antigen synthesis. Infection with the Δstg mutant resulted in higher IL-8 levels than infection with the parent strain, ISP1820, although these levels were still much lower than for S. Typhimurium χ3761 (Fig. 3). Strikingly, infection with the Δstg mutant carrying the lpf operon resulted in IL-8 levels indistinguishable from those of S. Typhimurium strain χ3761. These results suggest that when S. Typhi Δstg invades M cells, it is less able to suppress IL-8 secretion than the Stg+ parent. These studies spotlight the role that fimbriae play in Salmonella pathogenesis and suggest that Stg fimbriae play a supporting role in immune avoidance by S. Typhi.
As stated previously, the goal of this project was to redirect S. Typhi toward M cells and away from enterocytes in the gut to enhance immunogenicity. The current study demonstrates that deletion of stg and introduction of lpf enhance adherence and invasion of M-like cells while reducing adherence and invasion of Caco-2 cells, our model enterocyte system. The Stg- Lpf+S. Typhi strain was also better able to stimulate the innate immune system, as demonstrated by the increase in IL-8 secretion (Fig. 3). Taken together, these data suggest that the Stg- Lpf+S. Typhi approach may enhance immune responses to S. Typhi vaccines. While these results are promising, our tissue culture data using immortalized cell lines cannot be considered definitive, since intestinal enterocytes and M cells may behave differently from our modeled systems. In addition, host complexities, including the microbiome and the presence of chemotactic signals in the gut, may also influence the outcome (58). More work needs to be done to understand whether this strategy will be effective in vivo. However, our study represents an important first step in achieving that goal.
MATERIALS AND METHODS
Bacterial strains, growth media, and DNA manipulations.The bacterial strains used in this study are described in Table 1 and Table S1 in the supplemental material. Bacteria were routinely grown in LB broth or agar plates (61) unless otherwise indicated. When needed, antibiotics were added to growth media as follows: kanamycin, 50 μg/ml; ampicillin, 100 μg/ml; and chloramphenicol, 25 μg/ml. The high-fidelity Phusion polymerase (New England BioLabs, Ipswich, MA) was used for all cloning done by PCR.
Strains and plasmids used in this study
Construction of an S. Typhi strain with a Δstg deletion.DNA sequences upstream and downstream of the stg operon were cloned using primer pairs 3663_for/3663_rev and glmS_for/glmS_rev, respectively (Table 2). We engineered various restriction sites into the primers to facilitate plasmid construction. S. Typhi Ty2 chromosomal DNA was used as the template. The two fragments were digested with EcoRI and ligated. The resulting fragment was digested with XbaI and NotI and cloned into the suicide plasmid pMEG-375 digested with the same enzymes to yield plasmid pKR005. Plasmid pKR005 was moved into the E. coli donor strain MGN617. The resulting strain was conjugated to S. Typhi ISP1820. Selection for transconjugants was achieved by plating onto LB plates containing chloramphenicol. Transconjugants were grown in LB broth and plated onto LB plates without NaCl containing 7% sucrose (62). Isolates with a deletion in the stg operon were identified by PCR. One isolate was designated RAZ025. The point of deletion and 200 bp of the surrounding upstream and downstream regions were confirmed by DNA sequence analysis. Production of Vi antigen by S. Typhi strains was confirmed by slide agglutination, and production of complete lipopolysaccharide by all Salmonella strains was confirmed using silver-stained gels, as previously described (63). All S. Typhi strains were confirmed to have similar growth rates in LB broth.
Primers used in this study
Cloning of fimbrial operons.The lpfABCDE genes were cloned by PCR using the primer pair Lpf728_for/Lpf6821_rev and S. Typhimurium χ3761 chromosomal DNA as the template. The resulting 6,112-bp fragment was ligated to plasmid pCR-BLUNT-Topo and transformed into E. coli to yield plasmid pKR009. The sequence of the entire operon was confirmed by DNA sequence analysis. The lpf operon was subcloned into the low-copy-number plasmid pWSK29 using BamHI and XhoI to yield plasmid pKR012. Plasmid pKR012 was then moved into various S. Typhi strains for further analysis.
The stgABCC′D operon was cloned using a similar strategy. The plasmid pair glmS_2474_for_Eco/3663_8425_for_Xho and S. Typhi Ty2 template DNA were used. The resulting 6,313-bp fragment was then cloned into pCR-BLUNT-Topo to yield plasmid pKR017. After confirmation by DNA sequence analysis, plasmid pKR017 was digested with EcoRI and XhoI. The resulting stg fragment was then purified and ligated to pWSK29 digested with the same enzymes to yield plasmid pKR022. Plasmid pKR022 was used for complementation studies.
Tissue culture.The human colon carcinoma cell line Caco-2 (ATCC HTB-37) and human Burkitt's lymphoma cell line Raji (ATCC CCL-86) were obtained from the American Type Culture Collection (ATCC) (Manassas, VA). The Caco-2 cells for this experiment originated from frozen stocks no less than 6 months old with an early passage number from the original ATCC stock. Our stock of Caco-2 cells was authenticated by observing phenotypic characteristics of enterocytic differentiation, which the cells express upon reaching confluence. The Raji Burkitt's lymphoma cell line stock was received directly from ATCC. The cell line was grown up, and early passages were frozen immediately for future use.
Caco-2 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 4.5 g/liter glucose (Corning, Manassas, VA) containing 4 mM l-glutamine, 1% sodium pyruvate, 1% nonessential amino acids (NEEA), 100 U/ml penicillin, 100 μg/ml streptomycin, 20% heat-inactivated fetal calf serum. Raji cells were cultured in RPMI 1640 medium with l-glutamine (2 mM) (Corning, Manassas, VA) containing 100 U/ml penicillin, 100 μg/ml streptomycin, 10% heat-inactivated fetal calf serum.
M cell coculture model.Caco-2 cells were maintained and grown in the upper compartments of 6.5-mm by 3-μm polycarbonate Transwell inserts (Corning, Manassas VA), seeding 3 × 105 cells/insert. Monolayer confluence was monitored by measuring transepithelial electrical resistances (TER) using a Millicell ERS-2 voltohmmeter (Millipore, Temecula, CA). Caco-2 cells were grown for 14 days on permeable supports to achieve tight and fully differentiated monolayers with a TER of ∼420 Ω cm2 (52). To confirm the polarized status of the monolayers, we monitored UEA-1 lectin binding sites and expression of the tight-junction protein ZO-1 by confocal laser scanning microscopy (CLSM) (50). Raji B cells (5 × 105) were resuspended in RPMI-DMEM (1:2) and added to the basolateral chambers of 14-day-old Caco-2 monolayers, and cocultures were maintained for 4 days. To confirm the presence of M-like cells, expression of galectin-9 was monitored by CLSM. Corresponding monocultures of Caco-2 cells on matched filter supports were used as controls. The integrity of the cell monolayers was measured by TER before infections were performed.
M cell coculture infection assays.On the day of infection, all Transwell inserts were washed out with Hanks balanced salt solution (HBSS) and replaced with antibiotic-free medium on both sides of the chamber 45 min prior to infection. Using static starter cultures, the strains were grown at 37°C statically for 18 h in LB medium and then pelleted and resuspended in phosphate-buffered saline (PBS) at a concentration of 1.5 × 109 CFU/ml. For every experiment, bacteria were added at an MOI of 100 or 10, which was approximately 3 × 107 CFU or 3 × 106 CFU, respectively, to approximately 3 × 105 Caco-2 or M-like cells and incubated for 1 to 2 h at 37°C, 5% CO2.
In experiments using an MOI of 100, plates were centrifuged for 5 min at room temperature at 1,000 × g. For all other experiments, which used an MOI of 10, this step was omitted. For attachment, M cells and Caco-2 monolayers were washed two times with sterile PBS and lysed with 0.1% sodium deoxycholate (SDC). For invasion, the medium in the upper chamber was removed and replaced with DMEM containing 100 μg/ml gentamicin. The plates were incubated for an additional 1 h at 37°C, 5% CO2. Wells were washed two times with PBS and lysed with 0.1% SDC. Harvested lysate from attachment-and-invasion collections was diluted and plated onto LB agar plates containing ampicillin. For IL-8 cytokine analysis, infections were done using the same procedure we used for the invasion assay, except that the cells were incubated for 3 h after the addition of gentamicin (the 4-h time point). IL-8 levels in cell culture supernatants at 4 h were analyzed using the Human IL-8 ELISA Ready-Set-Go! kit (Affymetrix eBioscience, San Diego, CA) according to the manufacturer's instructions. To determine cytotoxic effects of S. Typhi on M cells and Caco-2 monolayers, LDH was measured in the same supernatants used for measuring the cytokine IL-8, using a Pierce LDH cytotoxicity assay kit (Rockford, IL) according to the manufacturer's instructions. Supernatants from cells in the absence of S. Typhi or treated with 0.1% SDC were used as negative and positive controls separately. All experiments were performed in a minimum of triplicate sets.
Immunocytochemistry.M cells and Caco-2 monolayers were fixed with 2% paraformaldehyde in PBS for 20 min at 25°C on a shaker and then permeabilized with 2% paraformaldehyde, 1% Tween 20 in PBS for 20 min at 25°C on a shaker and finally washed 3 times with 1% bovine serum albumin (BSA) in PBS. The primary antibodies were rabbit anti-galectin-9 (Abcam, Cambridge, MA), used at 1:500; Alexa 594-conjugated anti-ZO-1 (Molecular Probes), used at 5 μg/ml; fluorescein-labeled anti-Salmonella CSA-1 (KPL, Gaithersburg, MD), used at 50 μg/ml; and fluorescein-conjugated UEA-I (Vector Laboratories), used at 10 μg/ml. The secondary antibodies were goat anti-rabbit IgG Fc conjugated to Alexa Fluor 568 (Abcam, Cambridge, MA) and goat anti-rabbit IgG Fc conjugated to Alexa Fluor 647 (Abcam, Cambridge, MA), used at 2 μg/ml. Fixed and labeled samples were mounted on slides using Vectashield. Images were obtained using a Leica DM2500 confocal laser scanning microscope. Brightness and contrast were adjusted.
Statistical analysis.Data are expressed as geometric means and standard errors of the mean and were evaluated with one- or two-way analysis of variance (ANOVA) (Prism version 7.0) for multiple comparisons among groups. Two-way ANOVA was used for analyzing adherence and invasion data sets. One-way ANOVA was used for analyzing the IL-8 enzyme-linked immunosorbent assay (ELISA) data sets. A P value of <0.05 was considered statistically significant.
ACKNOWLEDGMENTS
We thank Karen Brenneman and Stephen Forbes for valuable discussions during the planning stages of this work. We thank our undergraduate students Kyle Horace, Andrea Segerstrom, Nyja Brown, and Miranda Yousif for their enthusiastic assistance at various stages of the project. We also thank Salvatore Oddo for graciously granting us access to his confocal microscope for these studies.
The research was supported by R21 grant AI119697 from the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
K.L.R. and A.M.G. conceived and designed experiments. A.M.G., S.W., and K.L.R. performed the experiments. A.M.G. and K.L.R. analyzed the data. K.L.R. and A.M.G. wrote the paper.
FOOTNOTES
- Received 4 March 2017.
- Returned for modification 29 March 2017.
- Accepted 14 June 2017.
- Accepted manuscript posted online 19 June 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00172-17 .
- Copyright © 2017 American Society for Microbiology.
