Vi polysaccharide of Salmonella typhi targets the prohibitin family of molecules in intestinal epithelial cells and suppresses early inflammatory responses

Abstract

Vi capsular polysaccharide (Vi) was first identified as a virulence antigen of Salmonella typhi, the causative agent of typhoid fever in humans; it renders S. typhi resistant to phagocytosis and the action of serum complement. However, the role of Vi during the infection of intestinal epithelium with S. typhi is not completely understood. We show here that Vi can interact with a model human intestinal epithelial cell line, Caco-2, through a cell-surface-associated molecular complex containing two major proteins of 30 and 35 kDa and a minor protein of ≈68 kDa. The two major proteins were identified as the putative tumor suppressor molecule, prohibitin, and its closely related homolog, B cell receptor-associated protein 37. These two proteins were enriched in lipid rafts, and Vi readily associated with these membrane microdomains. Engagement of Caco-2 cells with Vi inhibited their ability to produce an inflammatory response upon infection with Vi^– S. typhi. Consistent with this effect, infection of Caco-2 cells with Vi⁺ S. typhi produced less IL-8 compared with Vi^– S. typhi. Cells treated with Vi showed reduced extracellular signal-regulated kinase phosphorylation in response to infection with Vi^– S. typhi or stimulation with phorbol 12-myristate 13-acetate, suggesting that the mitogen-activated protein kinase pathway might be a target for Vi-mediated inhibition of inflammatory responses. These findings reveal a crucial role for Vi in the modulation of early inflammatory responses during infection with S. typhi. This kind of a modulation could play a significant role in the establishment of infection by S. typhi.

Salmonella typhi is a facultative intracellular pathogen that causes typhoid in humans. The bacterium enters the body orally after the intake of contaminated food or water, and, upon reaching the small intestine, adheres to and invades the specialized M cells and enterocytes. The pathogen is translocated to the intestinal submucosa and subsequently disseminates throughout the reticuloendothelial system. S. typhi can be isolated from spleen, liver, bone marrow, and gall bladder during typhoid fever (1). The host–pathogen interactions during infection with this bacterium remain incompletely characterized in the small intestine. Most of our current understanding of how this organism interacts with host cells comes largely from studies carried out with Salmonella typhimurium, which in mice causes an analogous disease commonly referred to as murine typhoid (2). However, although these studies have considerably increased our understanding of how pathogenic Salmonella species in general invade host cells or induce cellular cytotoxicity, the reasons for the host specificity exhibited by S. typhi and many other Salmonella serovars are not well understood. Unlike S. typhi, S. typhimurium causes enteritis in humans, which is characterized by self-limited fever and diarrhea and, in some cases, dysentery. These symptoms are rarely observed during infection with S. typhi (3). These different manifestations may be produced by one or more unidentified molecules not conserved between these two closely related Salmonella species. These molecules may also determine the quality and magnitude of inflammatory and immune responses during infection with these two pathogens. One of the crucial distinctions between S. typhi and S. typhimurium is the outer capsule, Vi capsular polysaccharide (Vi); it is absent in the mouse pathogen (4). Vi is a polymer of α-1→4-galacturonic acid with an N-acetyl at position C-2 and variable O-acetylation at C-3 (5). The virulence of S. typhi correlates with the expression of this molecule, because >98% S. typhi isolates derived from the blood of typhoid patients are encapsulated (6, 7). Studies carried out in human volunteers have shown that Vi⁺ strains of S. typhi are more virulent than Vi^– S. typhi (8). The expression of Vi is associated with resistance of S. typhi to the action of anti-O antibody and to phagocytosis and complement-mediated killing. The latter two actions can be initiated by antibodies to Vi (4, 9). Vi has been extensively studied for its potential as a vaccine against infection with S. typhi, and currently it is one of the vaccines available for use in humans (10–15). Vi has also been shown to enhance survival of S. typhi in cultured macrophages in vitro (16). Thus, studies carried out so far have suggested an important role for this molecule in protecting S. typhi against multiple host defense mechanisms. However, its interaction with intestinal epithelial cells, the first cell type that S. typhi is believed to infect during typhoid fever, has not been investigated.

In the present study, we demonstrate that Vi can interact with intestinal epithelial cells through a specific cell-surface-associated recognition complex containing tumor suppressor protein, prohibitin, and prohibitin-related molecule B cell receptor-associated protein 37 (BAP-37), as major constituents. The data presented here show that Vi can down-regulate early inflammatory responses from these cells. The involvement of the prohibitin family of molecules in S. typhi–host cell interaction has significant implications for the pathogenesis of typhoid fever.

Materials and Methods

Cells and Reagents. S. typhi Vi⁺ and Vi^– isolates from clinically confirmed typhoid patients were kindly provided by Geeta Mehta (Lady Hardinge Medical College, New Delhi). The expression of Vi and other antigens (O and H) on these isolates was further ascertained by reactivity with anti-S. typhi monoclonal antibodies, which has been described (refs. 17–19 and data not shown). Bacteria were grown in LB at 37°C. The human intestinal epithelial cell line Caco-2 was obtained from the American Type Culture Collection and was maintained in RPMI medium 1640 containing 10% FCS (RPMI medium 10) at 37°C in a humidified CO₂ (5%) incubator. Vi derived from S. typhi was obtained from Aventis (Connaught, India) and was dialysed against PBS before being used in cellular studies. LPS derived from S. typhi and polygalacturonic acid were purchased from Sigma, rabbit anti-prohibitin antibodies were obtained from Neo Markers (Fremont, CA), and anti-extracellular signal-regulated kinase (ERK) and anti-phosphoERK antibodies were from Cell Signaling Technology (Beverly, MA).

Binding of Vi to Cells. The binding of Vi to a human intestinal epithelial cell line, Caco-2, was analyzed by flow cytometry. Briefly, cells were incubated at 4°C with varying concentrations of Vi in PBS containing 1% BSA (PBS–BSA; 50 mM phosphate/150 mM NaCl, pH 7.4), washed, and incubated with an anti-Vi monoclonal antibody [P2B1G2/A9 (10 μg/ml PBS–BSA) (19)] or an isotype-matched control antibody for 1 h at 4°C. Cells were washed and mixed with FITC-labeled anti-mouse Ig antibody (The Jackson Laboratory) for 1 h at 4°C. Control cells were incubated only with the anti-Vi monoclonal antibody and FITC anti-mouse Ig. Five thousand cells were analyzed in a flow cytometer (Becton Dickinson). In some experiments, binding of Vi to cells was carried out in the presence of polygalacturonic acid or LPS. The internalization of the polysaccharide was also studied by flow cytometry (see Supporting Materials and Methods, which is published as supporting information on the PNAS web site).

Identification of Vi-Binding Molecules and Determination of Amino Acid Sequence. The Vi-interacting molecules in Caco-2 cells were identified by immunoprecipitation as described in ref. 20 with slight modifications. Briefly, cells were surface-biotinylated by the method of Meier et al. (21) and lysed in TKM buffer (50 mM Tris·HCl, pH 7.4/25 mM KCl/5 mM MgCl₂/1 mM EDTA/0.02% NaN₃/mixture of protease inhibitors) containing 1% Triton X-100. The lysate was centrifuged at 15,000 × g for 20 min, and the supernatant was used in immunoprecipitation experiments. Anti-Vi monoclonal antibody was mixed with protein G-Sepharose beads for 45 min at room temperature, washed, and incubated with Vi or culture supernatant from Vi⁺ S. typhi. The beads were washed and incubated with biotinylated Caco-2 cell lysate at 4°C for 1–2 h. After being washed extensively with the lysis buffer, the beads were boiled with Laemmeli sample buffer (nonreducing) and run in a 10% SDS/polyacrylamide gel (PAG). The proteins were transferred to a nitrocellulose (NC) membrane and incubated with horseradish peroxidase (HRP)-labeled avidin. The NC sheet was developed by using enhanced chemiluminescence reagent (ECL; Amersham Pharmacia). Control immunoprecipitations had either LB or culture supernatant from Vi^– S. typhi in place of Vi or an isotype-matched control antibody instead of anti-Vi monoclonal antibody.

To determine the amino acid sequences of Vi-binding molecules, proteins were isolated from Caco-2 cells by bulk immunoprecipitation (from 10⁹ cells), run in a SDS/PAG and stained with Coomassie brilliant blue. The bands corresponding to 30 kDa and 35 kDa were subjected to mass spectrometric analysis at the Biomolecular Research Facility, University of Virginia Health System (funded by the University of Virginia Pratt Committee, Charlottesville).

Immunoblot with Anti-Prohibitin Antibodies. The proteins immunoprecipitated with Vi from Caco-2 cells by the method described above were transferred to a NC membrane. The NC sheet was blocked with 1% milk protein in PBS and incubated with rabbit anti-prohibitin antibodies. Subsequently, the NC membrane was treated with HRP-labeled anti-rabbit IgG (The Jackson Laboratory), and, after extensive washing, the blot was developed by using ECL. The reactivity with anti-prohibitin antibodies was also carried out with lipid rafts prepared from Caco-2 cells by using the method described by Xavier et al. (22).

Localization of Vi and Prohibitin in Caco-2 Cells by Confocal Microscopy. Caco-2 cells were grown on cover slips in a 24-well tissue culture plate. Cells were washed with serum-free RPMI medium 1640 and incubated with Vi for 1 h at 4°C. After being washed with PBS–BSA, the cells were incubated with biotinylated anti-Vi monoclonal antibody [prepared as described by Meier et al. (21)] followed by phycoerythrin-labeled avidin. Cells were washed and permeabilized with chilled methanol containing 0.01% Triton X-100 for 5 min at 4°C. Subsequently, cells were washed with PBS–BSA and incubated with rabbit anti-prohibitin antibodies followed by FITC-labeled anti-rabbit IgG. The cover slips were washed, mounted with anti-fade reagents (Molecular Probes), and observed in an Zeiss LSM 510 confocal microscope with an oil immersion objective.

Analysis of Effect of Stimulation with Vi on Inflammatory Responses in Caco-2 Cells. Caco-2 monolayers grown in 75-cm² tissue culture flasks were trypsinized (21 mM trypsin/0.68 mM EDTA/150 mM NaCl), washed with serum-free RPMI medium 1640, and incubated with various concentrations of Vi or LPS for 1 h at 37°C under serum-free conditions. Afterward, cells were infected with Vi^– S. typhi at a multiplicity of infection (moi) of 10 for 1 h. Gentamycin was added at 200 μg/ml, and the cells were incubated for another 5 h at 37°C, at which time culture supernatants were collected and assayed for IL-8 by ELISA (Pharmingen). The effect of Vi on IL-8 secretion induced by phorbol 12-myristate 13-acetate (PMA) was also investigated either in the absence or presence of anti-Vi monoclonal antibodies.

To understand the nature of intracellular signals modulated during stimulation of cells with Vi, phosphorylation of ERK was analyzed in Caco-2 cells. Briefly, freshly harvested cells were incubated with Vi and then infected with Vi^– S. typhi or activated with PMA for different time periods. Cells were washed and lysed with TKM buffer containing 1% Triton X-100. The lysates were spun at 10,000 × g, and the supernatants were run in a 10% SDS/PAG. Proteins were transferred to NC and blotted with anti-phosphoERK antibodies. The blot was developed by using ECL. Subsequently, the NC sheet was incubated with a solution containing 0.1 M acetic acid and 0.15 M NaCl to strip anti-phosphoERK antibodies and reblotted with anti-ERK antibodies.

Detection of Flagellin During Infection of Cells with S. typhi. Flagellin is considered to be the major proinflammatory determinant of pathogenic Salmonella (23). It is recognized by Toll-like receptor 5 (24) and induces secretion of IL-8 and other inflammatory mediators from intestinal epithelial cells (23). To detect flagellin released by Vi⁺ and Vi^– S. typhi during their interaction with Caco-2 cells, an immunoprecipitation analysis was carried out by using an anti-S. typhi flagellin monoclonal antibody (ref. 18 and Supporting Materials and Methods).

Results

Vi Binds to Human Intestinal Epithelial Cells. To study the interaction of Vi with human intestinal epithelial cells, binding of this molecule with a model intestinal epithelial cell line, Caco-2, was analyzed by flow cytometry with an anti-Vi monoclonal antibody as a probe. The polysaccharide showed a dose-dependent binding with these cells (Fig. 1a). The binding was also seen with the culture supernatant derived from Vi⁺ S. typhi but not with the culture supernatant from Vi^– S. typhi (Fig. 1b). The interaction of Vi with Caco-2 cells was not inhibited when cells were preincubated with polygalacturonic acid, which differs from Vi in the absence of N- and O-acetyl groups at positions C-2 and C-3, respectively, suggesting that acetyl groups in the polysaccharide might be crucial for its binding to these cells. The specificity of Vi–Caco-2 interaction was also demonstrated by the inability of LPS to inhibit binding of Vi to Caco-2 cells (Fig. 1c).

Fig. 1.

Binding of Vi to a human intestinal epithelial cell line Caco-2 analyzed by flow cytometry. (a) Cells were incubated at 4°C with 0.1, 1, and 10 μg/ml Vi, followed by incubation with anti-Vi monoclonal antibody and FITC-anti-mouse Ig and analyzed in a flow cytometer. Control cells (shaded curve) were incubated only with anti-Vi antibody and FITC-anti-mouse Ig. (b) Binding of Caco-2 cells with Vi⁺ S. typhi or Vi^– S. typhi culture supernatants (CS). Cells were incubated with the culture supernatants at 4°C for 1 h, followed by incubation with anti-Vi antibody and FITC anti-mouse Ig. (c) Interaction of Vi with Caco-2 cells was not inhibited by polygalacturonic acid (PGUA) or LPS. Cells were mixed with polygalacturonic acid (500 μg/ml) or LPS (100 μg/ml) for 1 h at 4°C, washed, and incubated with Vi (10 μg/ml). Binding of Vi to cells was analyzed as described for a.

To investigate whether binding was followed by internalization, surface staining of Vi was analyzed in cells that had been incubated with the polysaccharide at 4°C and then transferred to 37°C for various time periods. There was a gradual decrease in cell-surface-associated fluorescence as cells were shifted from 4°C to 37°C (Fig. 6a, which is published as supporting information on the PNAS web site), suggesting a time-dependent internalization of the polysaccharide. Internalization was confirmed by intracellular staining for Vi in permeabilized cells (Fig. 6b).

The ability to bind to intestinal epithelial cells was not a function associated only with soluble Vi. The binding of Vi⁺ S. typhi to Caco-2 cells at 4°C could be blocked by Vi in a dose-dependent fashion (Fig. 7, which is published as supporting information on the PNAS web site). This blockade was not seen with LPS. Furthermore, Vi did not inhibit the binding of Vi^– S. typhi to cells (data not shown). These results show that Vi associated with the surface of S. typhi was also capable of interacting with Caco-2 cells.

Interaction of Vi with Cells Is Mediated Through a Cell-Surface-Associated Recognition Complex Containing the Prohibitin Family of Molecules. Having established that both soluble and pathogen-associated Vi could bind to intestinal epithelial cells, we next investigated whether this interaction was mediated through specific recognition molecules by immunoprecipitation with surface-biotinylated Caco-2 cells. Vi was able to pull down a molecular complex containing two major proteins of ≈30 kDa and 35 kDa and a minor protein of ≈68 kDa (Fig. 2a, lane 1). A similar profile was obtained when the culture supernatant from Vi⁺ S. typhi was used as the source of Vi (Fig. 2a, lane 2). The bands were not seen with culture supernatant derived from Vi^– S. typhi or when an isotype-matched control antibody was used in immunoprecipitation instead of anti-Vi antibody (Fig. 2a, lane 3). The Vi-interacting molecular complex could also be immunoprecipitated from a human monocytic cell line, U937 (data not shown). Vi did not bind to prohibitin or BAP-37 after these molecules were electrophoresed and transferred to NC, which suggests that the two proteins need to be in their native form and associated with each other to interact with Vi or that the interaction of Vi with the proteins in the complex is indirect. Analysis of binding with purified prohibitin molecules should clarify this issue.

Fig. 2.

Vi interacts with a cell-surface-associated molecular complex in Caco-2 cells. (a) Cells were surface-biotinylated with N-hydroxysuccinimidebiotin, lysed with TKM buffer containing 1% Triton X-100, and centrifuged at 15,000 × g. The supernatant was incubated with protein G-Sepharose beads preloaded with anti-Vi monoclonal antibody and Vi (either purified polysaccharide or culture supernatant derived from Vi⁺ S. typhi). The beads were washed, boiled with Laemmli sample buffer (nonreducing), and electrophoresed in a 10% SDS/PAG. Proteins were transferred to a NC sheet and incubated with HRP-labeled extravidin. The blot was developed by using ECL. Lane 1, lysate incubated with protein G-Sepharose preloaded with anti-Vi monoclonal antibody and purified Vi; lane 2, as in lane 1, except that culture supernatant from Vi⁺ S. typhi was used in place of purified Vi; lane 3, as in lane 1, except that culture supernatant from Vi^– S. typhi was used instead of purified Vi. (b) Immunoblot with anti-prohibitin antibodies. Vi-binding proteins were immunoprecipitated by using culture supernatant from Vi⁺ S. typhi as described for a, run in a 10% SDS/PAG and transferred to NC. The NC membrane was blotted with rabbit anti-prohibitin antibodies followed by HRP-anti-rabbit Ig. The blot was developed by using ECL. Controls (represented by lane 1) consisted of protein G-Sepharose incubated with the cell lysate, Caco-2 lysate incubated with protein G beads loaded with anti-Vi monoclonal antibody, or protein G beads incubated with anti-Vi monoclonal antibody, LB broth, and the cell lysate. Lane 1, control; lane 2, Vi-immunoprecipitated complex from Caco-2 cells; lanes 3–5, Caco-2 lysate derived from 0.35 × 10⁵, 0.7 × 10⁵, and 1.4 × 10⁵ cells, respectively, directly blotted with anti-prohibitin antibodies.

The identities of the 30- and 35-kDa bands were determined by amino acid sequence analysis with mass spectrometry. These two proteins, isolated from Caco-2 cells by bulk immunoprecipitation, contained many peptide sequences corresponding to a recently identified putative tumor suppressor protein, prohibitin, and its closely related, higher-molecular-mass homolog, BAP-37 (25–27). The mass spectrometric data were confirmed by reactivity with antibodies generated against the 30-kDa recombinant rat prohibitin. These antibodies reacted strongly with the 30-kDa prohibitin band that was immunoprecipitated with Vi and also recognized the band representing BAP-37. Prohibitin and BAP-37 share a very high degree of homology in their central regions (28), which is why anti-prohibitin antibodies showed crossreactivity with BAP-37. The antibodies also reacted, although weakly, with the 68-kDa molecule present in the immunoprecipitated complex (Fig. 2b), suggesting that this molecule might also share some similarity with the prohibitin family of molecules. The exact identity of this protein remains to be established. A polyubiquitinated prohibitin of ≈65 kDa has been recently reported by Thompson et al. (29) in the mammalian sperm. It is therefore possible that our 68-kDa band represents a similar molecule.

Vi-Binding Molecules Are Enriched in Lipid Rafts. To gain insights into the localization of Vi-binding molecules in the membrane, detergent-insoluble membrane domains or lipid rafts prepared from Caco-2 cells were probed with anti-prohibitin antibodies. As can be seen from Fig. 3a, prohibitin and BAP-37 were enriched in the rafts represented by fractions 3–5. Our attempts to detect a lipid raft marker, such as caveolin or ganglioside GM-1, in these domains were unsuccessful, which may be because there are very low levels of these molecules in Caco-2 cells as has been reported (30). However, lipid rafts prepared under identical conditions from MDCK or NIH 3T3 cells readily showed expression of caveolin in the detergent-insoluble membrane domains. Similarly, human T cell lymphoma line, Jurkat, showed ganglioside GM-1 in the raft fraction (data not shown). When lipid rafts were prepared from Caco-2 cells that had been incubated with Vi at 4°C, the polysaccharide was found associated with these membrane microdomains (Fig. 3b Left, fractions 4 and 5). The rafts prepared from control cells, as expected, did not show any staining for Vi (Fig. 3b Right); low-level binding seen with fractions 10–12 derived from these cells was nonspecific reactivity. Importantly, immunofluorescence analysis with confocal microscopy showed that membrane-associated prohibitin colocalized with Vi in Caco-2 cells that had been incubated with Vi (Fig. 3c).

Fig. 3.

Prohibitin and BAP-37 are enriched in lipid rafts. (a) Caco-2 cells were lysed with TKM buffer containing 1% Triton X-100 for 30 min at 4°C. The suspension was centrifuged at 500 × g to remove unlysed cells and cell debris. The supernatant (1 ml), mixed with an equal volume of 85% sucrose, was placed at the bottom of an ultracentrifuge tube and overlaid with 6 ml of 35% sucrose and 4 ml of 5% sucrose. The gradient was centrifuged at 200,000 × g for 16 h. Twelve 1-ml fractions were collected, run in a 12.5% SDS/PAG, transferred to NC, and blotted with anti-prohibitin antibodies as outlined for Fig. 2b. (b) Vi bound to lipid rafts prepared from Caco-2 cells. Cells were incubated with Vi at 4°C for 1 h and washed with cold PBS to remove unbound Vi. Subsequently, lipid rafts were prepared as described for a. (Left) Five microliters of each fraction were dotted onto a NC sheet and incubated with anti-Vi monoclonal antibody followed by HRP-labeled anti-mouse Ig. The blot was developed by using ECL. (Right) Control fractions were from cells incubated with PBS instead of Vi. (c) Vi and prohibitins colocalize in the Caco-2 cell membrane. Cells were incubated with Vi, followed by biotinylated anti-Vi monoclonal antibody and PE-labeled streptavidin. Subsequently, cells were permeabilized with chilled methanol containing 0.01% Triton X-100 at 4°C for 5 min, washed, and incubated with rabbit anti-prohibitin antibodies. After incubating with FITC-anti-rabbit Ig, cells were washed, placed on a coverslip, and observed in a confocal microscope. (A) Staining for Vi. (B) Staining for prohibitin. (C) Merge of A and B.

Engagement of Caco-2 Cells with Vi Suppresses Early Inflammatory Responses. Lipid rafts are believed to be the sites in the membrane where many cellular signaling events are initiated (31). The enrichment of Vi-binding proteins in these domains suggested that engagement of cells with Vi might modulate intracellular signaling events. We investigated this possible modulation by analyzing early inflammatory responses in cells treated with the polysaccharide. As can be seen from Fig. 4a, Caco-2 cells incubated with Vi secreted significantly reduced levels of IL-8 upon infection with Vi^– S. typhi. The inhibition was dose-dependent and specific to Vi because LPS did not induce any significant suppression, and it was best seen under serum-free conditions with freshly detached Caco-2 cells or cells that had been plated at lower cell densities for shorter time periods. The effect was not due to the blockade of the binding of Vi^– S. typhi to Caco-2 cells by Vi because a comparable number of bacteria bound to Caco-2 in the presence or absence of Vi (data not shown). The reduced IL-8 secretion in cells preincubated with Vi was also seen in response to stimulation with PMA (Fig. 4b) and could be reversed by anti-Vi antibodies (Fig. 8, which is published as supporting information on the PNAS web site). The degree of inhibition was reduced when stimulation with PMA was carried out in the presence of serum (Fig. 9, which is published as supporting information on the PNAS web site). Consistent with the effect of Vi on IL-8 secretion, infection of Caco-2 with Vi⁺ S. typhi produced significantly lower levels of IL-8 as compared with Vi^– S. typhi (Fig. 10a, which is published as supporting information on the PNAS web site). These lower levels of IL-8 could not be ascribed to the reduced amount of flagellin released by Vi⁺ bacteria during infection of Caco-2 cells. In fact, Vi⁺ S. typhi at 100 moi produced significantly more flagellin than Vi^– S. typhi at 10 moi, but the latter still induced more IL-8 secretion from Caco-2 cells (Fig. 10).

Fig. 4.

Stimulation with Vi inhibits the ability of Caco-2 cells to secrete IL-8 in response to infection with Vi^– S. typhi or activation with PMA. (a) Freshly detached Caco-2 cells (4 × 10⁴) or cells plated at 37°C for 4 h were incubated with Vi or LPS diluted in serum-free medium for 1 h at 37°C, after which the cells were infected with Vi^– S. typhi at 10 moi. One hour later, unbound bacteria were removed and cells were incubated for another 5 h in serum-free medium containing 200 μg/ml gentamycin. The culture supernatants were collected and analyzed for IL-8 by ELISA. (b) Freshly detached cells were incubated with Vi or LPS in serum-free medium and stimulated 1 h later with 25 ng/ml PMA in serum-free RPMI medium 1640. After 5 h at 37°C, IL-8 was determined in the culture supernatants by ELISA. Dotted lines represent IL-8 levels from Caco-2 in the absence of any stimulation.

Vi Targets Mitogen-Activated Protein (MAP) Kinase Pathway. To understand the mechanism by which Vi mediates inhibition of IL-8 secretion, the effect of activation with the polysaccharide on the phosphorylation of ERK was investigated. Cells were incubated with Vi and subsequently activated with PMA. As can be seen from Fig. 5a, although cells activated with PMA showed sustained ERK phosphorylation that looked biphasic, cells prestimulated with Vi and then activated with PMA had slightly increased ERK phosphorylation at earlier time points, but, at later time points, it was significantly reduced. On the other hand, cells incubated with Vi and subsequently infected with Vi^– S. typhi showed reduced ERK phosphorylation at all time points (Fig. 5b), indicating that there might be subtle differences in the way Vi affected signaling in Caco-2 cells in response to the two stimuli. Nevertheless, the data suggested that the MAP kinase pathway might be one of the targets that is modulated by Vi to dampen early inflammatory responses. The importance of ERK in IL-8 secretion was revealed by reduced secretion of this chemokine in Caco-2 cells infected with S. typhi in the presence of a specific MAP kinase kinase inhibitor PD98059 (Fig. 11, which is published as supporting information on the PNAS web site).

Fig. 5.

Stimulation with Vi modulates MAP kinase pathway in Caco-2 cells. Freshly detached cells (1 × 10⁶) were incubated with 12.5 μg/ml Vi for 1 h in serum-free medium and stimulated with 100 ng/ml PMA for different time periods (a) or infected with Vi^– S. typhi at 50 moi (b). Cells were lysed with TKM buffer containing 0.5% Triton X-100, and the supernatants were run in a 12.5% SDS/PAG and transferred to a NC membrane. The NC sheet was blotted with anti-phosphoERK (p-ERK) antibodies and developed by using ECL. Subsequently, the NC sheet was incubated with a low-pH solution (0.1 M acetic acid containing 0.15 M NaCl) to strip the bound antibodies and blotted with anti-ERK antibodies. (a) Lanes 1–3, Caco-2 cells incubated with medium for 5, 15, and 60 min, respectively; lanes 4–6, cells activated with PMA for 5, 15, and 60 min, respectively; lane 7, cells incubated with Vi for 60 min; lanes 8–10, cells incubated with Vi for 60 min followed by activation with PMA for 5, 15, and 60 min, respectively. (b) Lanes 1–3, Caco-2 cells incubated with RPMI medium 1640 for 5, 15, and 60 min, respectively; lanes 4–6, cells infected with Vi^– S. typhi for 5, 15, and 60 min, respectively; lanes 7–9, lanes treated with Vi for 1 h followed by infection with Vi^– S. typhi for 5, 15, and 60 min, respectively.

Discussion

S. typhi causes typhoid fever in humans. In this study, we have investigated the interaction of Vi capsular polysaccharide of S. typhi with human intestinal epithelial cells. Vi is absent in the closely related Salmonella serovar S. typhimurium, which has been used as a model in mice to understand S. typhi pathogenesis. S. typhimurium does not, however, produce typhoid-like manifestations in humans. Our results show that Vi can interact with a model human intestinal epithelial cell line, Caco-2, through a specific cell-surface-associated recognition complex containing prohibitin and its closely related homolog, BAP-37, as major components. Prohibitin is a putative tumor suppressor molecule that regulates mammalian cell cycle by repressing the E2F family of transcription factors (32–37). Prohibitin and its related members are abundant in mitochondria (38–41) but have also been localized in the cell membrane (27) and the nucleus (36). In the present study, both prohibitin and BAP-37 were found enriched in lipid rafts, sites that are believed to regulate many intracellular signaling events (31). Significantly, Vi also associated with these membrane microdomains, suggesting that the interaction of cells with Vi might bring about modulation of cellular signaling.

The regulation of cell cycle by prohibitin is believed to be mediated through the MAP kinase pathway (33). Considering that this pathway also plays a crucial role in producing inflammatory responses during infection with Salmonella and many other pathogens (42, 43), the question we asked in this study was whether engagement of cells with Vi could modulate inflammatory responses during infection with S. typhi. Our results show that treatment of intestinal epithelial cells with Vi can significantly inhibit IL-8 secretion in response to infection with Vi^– S. typhi. This inhibitory effect was also revealed by reduced IL-8 secretion from Caco-2 cells infected with Vi⁺ S. typhi as compared with those infected with Vi^– S. typhi. The reduced IL-8 secretion was seen even when Vi⁺ bacteria produced sufficient amounts of flagellin during interaction with Caco-2 cells, which suggests that the engagement of cells with Vi might interfere with Toll-like receptor 5 signaling. As mentioned earlier, flagellin is believed to be the major proinflammatory determinant of pathogenic Salmonella (23). Interestingly, Vi-mediated inhibition of IL-8 secretion was best seen with freshly detached Caco-2 cells or cells that had been plated at lower cell densities for shorter durations (not grown to confluency). We believe that this finding may be due to sequestration of Vi-interacting receptor complex in the tight junctions in highly confluent cultures.

The investigation of the mechanism by which Vi mediates suppression of IL-8 secretion from Caco-2 cells revealed that the MAP kinase pathway might be a target of this modulation. Incubation of cells with Vi resulted in a significant reduction in the induction or sustenance of ERK phosphorylation upon infection with Vi^– S. typhi or stimulation with PMA. The significance of this pathway in IL-8 secretion after infection with S. typhi was also revealed by reduced chemokine secretion from Caco-2 cells in the presence of MAP kinase kinase inhibitor PD98059. Remarkably, serum, which has been shown to reverse prohibitin-mediated repression of E2F transcriptional activity, also reversed polysaccharide-mediated suppression of IL-8 secretion in our study. The reversal of prohibitin-mediated effects with serum is believed to be mediated through activation of the MAP kinase pathway (33).

Taken together, our findings reveal a role for Vi in down-regulating early inflammatory responses from intestinal epithelial cells during infection with S. typhi. In vivo, this down-regulation may be mediated by Vi, which is released in abundance by S. typhi during in vitro growth and has also been reported in the serum and urine samples of typhoid patients (44, 45) or by S. typhi-associated Vi. S. typhi, unlike S. typhimurium, does not evoke a strong inflammatory response in the gut (1). The interaction reported here, therefore, represents a significant distinction between these two pathogens and could well be one of the primary reasons responsible for the diminished inflammatory response seen during early stages of infection with S. typhi. We have also seen that incubation of Caco-2 cells with Vi brings about perturbation of tight junctions (unpublished data), indicating that Vi might also modulate other cellular functions. Future studies should focus on understanding the mechanism by which the interaction of Vi with the prohibitin-containing recognition complex modulates the MAP kinase pathway. Identification of targets upstream of ERK would not only provide important insights into Vi-mediated suppression of inflammatory responses during typhoid fever but might also help unravel the role of membrane-associated prohibitin and its related members in intracellular signaling. Vi can also interact with a human monocytic cell line through a molecular complex similar to the one reported here with intestinal epithelial cells (data not shown), suggesting that it might modulate cellular responses in mononuclear phagocytes as well. The role of Vi in macrophages needs to be further investigated.