Commensal Bacteria Increase Invasion of Intestinal Epithelium by Salmonella enterica Serovar Typhi

Abstract

The intestinal microflora consists of a heterogeneous population of microorganisms and has many effects on the health status of its human host. Here, it is shown that the products of certain strains of bacteria normally present in the intestinal microflora are able to trigger redistribution of the cystic fibrosis transmembrane conductance regulator (CFTR) protein in epithelial cells. CFTR is used by Salmonella enterica serovar Typhi as a receptor on epithelial cells which mediate the translocation of this microorganism to the gastric submucosa. Serovar Typhi-epithelial cell adhesion and CFTR-dependent invasion by serovar Typhi of epithelial cells were increased following commensal-mediated CFTR redistribution. These data suggest that commensal microorganisms present in the intestinal lumen can affect the efficiency of serovar Typhi invasion of the intestinal submucosa. This could be a key factor influencing host susceptibility to typhoid fever.

The cystic fibrosis transmembrane conductance regulator (CFTR) is used by Salmonella enterica serovar Typhi as a receptor on intestinal epithelial cells (18). Cell surface expression of the CFTR protein by intestinal epithelium is increased during serovar Typhi infection (12). This increase is brought about by a redistribution of preformed CFTR protein from intracellular stores to the epithelial cell plasma membrane. Increased membrane expression of CFTR is correlated with enhanced CFTR-dependent entry of serovar Typhi into epithelial cells.

In vivo, serovar Typhi must establish infection in the presence of a complex population of commensal microorganisms that range in numbers from 10⁸ CFU per ml in the small intestine to 10¹¹ to 10¹² CFU per ml in the large intestine (9, 20). Serovar Typhi is probably introduced into this large number of commensal organisms in relatively small numbers during most natural cases of infection with serovar Typhi. The ingested serovar Typhi bacteria transit through the intestinal lumen, with each bacterial cell likely having very limited contact time with each epithelial cell. In this scenario, in which serovar Typhi bacteria are far outnumbered by commensal microbes and in which the commensal microbes are in contact with the epithelium for a longer time than are serovar Typhi bacteria, it is possible that commensal-mediated effects on CFTR trafficking have a greater impact on serovar Typhi invasion than does serovar Typhi-mediated CFTR trafficking. Therefore, an objective of this study was to determine whether any commensal bacteria normally present in the intestinal microflora also possess the ability to mobilize CFTR to the epithelial cell plasma membrane, and if so, what effect this trafficking has on serovar Typhi invasion of epithelial cells.

Water extracts of commensal bacteria are able to trigger redistribution of CFTR protein to the plasma membrane.

Direct comparison of the abilities of various intestinal commensals to mobilize CFTR would be complicated by the diverse requirements of and tolerances of these commensals for molecular oxygen. To avert these difficulties, sterile water extracts were prepared from each commensal strain (3, 12), and the extracts were tested for their ability to stimulate redistribution of CFTR protein in epithelial cells. Workers in this laboratory have previously demonstrated that mobilization of CFTR to the plasma membrane by serovar Typhi does not require live bacteria and that it can be induced by sterile water extracts of this bacterium (12). MDCK(green fluorescent protein [GFP]-CFTR) cells, expressing a fusion of human CFTR and GFP, were seeded into glass-bottom culture dishes (MatTek, Ashland, Mass.). Cells at 50 to 70% confluence were incubated for 1 h at 37°C with 5 μg of bacterial extract/ml, washed with ice-cold phosphate-buffered saline, and examined on an Axiovert S100 microscope (Carl Zeiss, Inc., Thornwood, N.Y.) with a Bio-Rad (Hercules, Calif.) MRC 1024 krypton-argon laser. GFP-CFTR was scored as being mobilized to the plasma membrane (12) if GFP fluorescence was concentrated at the periphery of the cell in all cross-sectional Z-sections observed, and all such scores were verified by a second operator who was kept unaware of the identity of the samples.

MDCK(GFP-CFTR) cells treated with certain commensal extracts contained a greater percentage of cells with plasma membrane-localized CFTR (Fig. 1). The activity was also strain dependent, as Enterococcus faecalis strain E1 triggered CFTR redistribution, while strain E2 did not. Moreover, E. faecalis preparation A, a mixture of unrelated E. faecalis strains, showed a level of CFTR-modulatory activity that was intermediate between those of strains E1 and E2. The level of CFTR redistribution triggered by extract from strains E1 and E2 was retested in ligated mouse intestinal loops to determine if these extracts performed similarly with primary intestinal epithelial cells. Ligated intestinal loops (10) from C57BL/6 mice were injected with 5 μg of sterile extract from E. faecalis strain E1 or E. faecalis strain E2 and incubated for 1 h at 37°C to allow redistribution of CFTR protein (12). After this incubation, intestinal enterocytes were removed from the intestine by gentle lavage with phosphate-buffered saline containing 1% bovine serum albumin and 2% normal goat serum (to block against nonspecific staining). Indirect immunofluorescent staining for cell surface CFTR expression and flow cytometry were both then carried out as described previously (12). The results showed that 8.4% of enterocytes from an intestine which had been treated with extract from strain E1 expressed a high level of cell surface CFTR (relative CFTR expression determined by comparison to control intestine and corrected for nonspecific staining observed with an irrelevant isotype- and species-matched control antibody) compared to only 0.8% of enterocytes in E2 extract-treated samples. Given that in an intact intestine, CFTR redistribution is first detected only in enterocytes at the villus tip, but not in those on the villus neck or crypt region (12), a finding of 8.4% positive enterocytes is within the range of expected values for a positive result.

FIG. 1.

Mobilization of GFP-CFTR fusion protein to the epithelial cell plasma membrane in response to sterile extracts of commensal microorganisms. The horizontal axis indicates the species of bacteria from which extract was derived. The vertical axis indicates the percentage of epithelial cells in which CFTR protein was accumulated on the plasma membrane. Bacterial strains for extract preparation were serovar Typhi H193, E. coli DH5α, B. fragilis 9343, B. vulgatus 8482, B. thetaiotaomicron 29741, and B. ovatus 8483. E. faecalis bacterial strains are indicated. The A column indicates a mixture of E. faecalis strains.

Species of Bacteroides varied in activity, with Bacteroides thetaiotaomicron and Bacteroides ovatus possessing a high level of activity and Bacteroides fragilis and Bacteroides vulgatus lacking the activity (Fig. 1). In addition, one of two clinical isolates of Clostridium difficile (strain 4268) showed CFTR-modulatory activity, while the other (strain 4269) did not (data not shown).

CFTR modulation by commensal bacterial extracts is correlated with enhanced serovar Typhi-epithelium interactions.

Previous work (12, 18) has demonstrated that the availability of CFTR on the epithelial plasma membrane correlates positively with serovar Typhi translocation to the gastric submucosa. This is likely due to the increased availability of epithelial CFTR for binding by its serovar Typhi-based ligand, lipopolysaccharide (13). Therefore, I determined whether commensal-stimulated redistribution of the CFTR receptor increased serovar Typhi-epithelium adhesion. MDCK(GFP-CFTR) cells were treated with CFTR-mobilizing extracts (serovar Typhi strain H193 and E. faecalis strain E1) or CFTR-nonmobilizing extracts (Escherichia coli strain DH5α and E. faecalis strain E2). The extract-treated MDCK cells were then incubated on ice for 1 h with heat-killed, fluorescently labeled (Syto-17; Molecular Probes, Inc., Eugene, Oreg.) serovar Typhi (strain Ty2). In this experiment, the bacterial extract added to the epithelial cells represents the only experimental source of CFTR-mobilizing activity, since the activity present in the fluorescently labeled serovar Typhi is inactivated during heat killing of the bacteria (12). Nonadherent serovar Typhi cells were removed from the MDCK cells with three washes of Tris-buffered saline, and serovar Typhi cells that remained adherent were fixed to the MDCK cells by 15 min of incubation with 2% glutaraldehyde at room temperature.

Extracts that were incapable of mobilizing CFTR to the plasma membrane had no effect on serovar Typhi-epithelium adhesion (Fig. 2A and D). In contrast, pretreatment of MDCK(GFP-CFTR) cells with CFTR-modulatory extracts enhanced the adhesion of serovar Typhi to the MDCK cells (Fig. 2B and C). The increase in serovar Typhi-epithelium adhesion triggered by extract from strain E1 was competitively inhibited by a CFTR-specific monoclonal antibody (CF3; see reference 22) that binds to the serovar Typhi binding site on the first predicted extracellular loop of CFTR (18), but adhesion was not inhibited by an isotype-matched monoclonal antibody (CF8) that binds an irrelevant, intracellular epitope on the CFTR protein (data not shown).

FIG. 2.

Adhesion of heat-killed, Syto 17-stained serovar Typhi (strain Ty2) to extract-treated MDCK(GFP-CFTR) cells. Extracts were from E. coli DH5α (A), serovar Typhi H193 (B), E. faecalis E1 (C), and E. faecalis E2 (D). Images are at a focal plane at the top of the epithelial cells, rather than at a cross-section through the epithelial cells, in order to visualize serovar Typhi bacteria that are adherent to the upper surface of the epithelial cells. As a result, CFTR redistribution to the periphery of the cell (the plasma membrane) is not evident. Magnification, ×640.

CFTR modulation by commensal extracts is correlated with enhanced epithelial invasion by serovar Typhi.

Redistribution of CFTR protein to the plasma membrane of epithelial cells results in enhanced invasion of serovar Typhi, which uses CFTR as a receptor (12). The effect of commensal extracts on CFTR localization and serovar Typhi-epithelium adhesion led me to hypothesize that serovar Typhi invasion of epithelial cells and of the gastric submucosa would also be enhanced by the commensal extracts. To test this hypothesis, 1.5-cm-long ligated intestinal loops (10) from C57BL/6 mice were injected with 5 μg of sterile extract from E. faecalis strain E1 or E. faecalis strain E2 and incubated for 1 h at 37°C to allow redistribution of CFTR protein (12). A sterile extract of serovar Typhi strain H193, which is known to trigger CFTR redistribution (12), was used as a positive control, and extract from E. coli strain DH5α, which does not trigger redistribution of CFTR, was used as a negative control. A total of 5 × 10⁶ CFU of live serovar Typhi (strain Ty2) was then injected into the lumens of the extract-treated intestinal loops, and the loops were incubated for 2 h at 37°C. Lumenal serovar Typhi was killed by gentamicin treatment (300 μg of gentamicin/ml for 1 h at 37°C). Invaded and translocated serovar Typhi bacteria were then enumerated by epithelial cell lysis (tissue was homogenized in a solution of 0.1% Triton X-100 and incubated for 30 min at 37°C) followed by serial dilution and culture of the lysate on MacConkey agar. In some samples, the number of invading serovar Typhi bacteria was below the lower limit of detection (100 CFU). In these cases, nonparametric tools were used for statistical analysis.

Pretreatment of the intestinal loops with CFTR-modulatory extract of E. faecalis strain E1 significantly enhanced serovar Typhi invasion of the intestinal submucosa (Fig. 3A). In contrast, extract of E. faecalis strain E2, which is negative for CFTR-modulatory activity, had no effect on serovar Typhi invasion. Serovar Typhi invasion of the submucosa was enhanced by the positive control, i.e., CFTR-modulatory extract of serovar Typhi, but not by the negative control extract of E. coli (Fig. 3A), as previously reported (12). Similar results were obtained with the T84 human colonic carcinoma cell line (Fig. 3B). The experiment conducted with T84 cells also included samples treated with extract from E. faecalis preparation A, which is a mixture of E. faecalis strains. Extract from preparation A enhanced invasion by serovar Typhi to a degree that was intermediate to that observed with strains E1 and E2 (Fig. 3B). This finding is in agreement with the intermediate level of CFTR redistribution triggered by extract from preparation A (Fig. 1).

FIG. 3.

Effect of extract pretreatment on serovar Typhi invasion. (A) Translocation of serovar Typhi strain Ty2 to the submucosa of extract-pretreated intestine. Each bar indicates the average number of translocated CFU of triplicate intestinal loops. Groups were compared by using the Kruskal-Wallis nonparametric analysis of variance (ANOVA), and pairwise comparisons were made by the Dunn posthoc test. *, significantly different from values obtained when no extract was used (P ≤ 0.002). Invasion by serovar Typhi strain Ty2 (B) and by serovar Typhimurium strain LT2 (C) into extract-pretreated T84 cells. Each bar indicates the average number of invaded CFU in quadruplicate samples. Error bars indicate standard deviation. Groups were compared by ANOVA, and pairwise comparisons were made by the Fisher probable least-significant difference test. *, significantly different from E. coli extract-treated samples (P ≤ 0.0001). (D) Invasion by serovar Typhi strain Ty2 into extract-pretreated T84 cells in the presence of CFTR-specific monoclonal antibodies CF3 and CF8 (−, no antibody added). Each bar indicates the average number of invaded CFU in triplicate samples. Error bars indicate standard deviation. Groups were compared by ANOVA, and pairwise comparisons were made by the Fisher probable least-significant difference test. §, significantly different from E. coli extract-treated samples (P ≤ 0.001).

Treatment of epithelial cells with crude bacterial extracts likely has numerous effects on the physiology of the epithelial cells in addition to that of triggering redistribution of the CFTR protein. To demonstrate that enhanced invasion by serovar Typhi following extract treatment was indeed due to CFTR redistribution and not due to a nonspecific or general stimulation of the epithelial cells, two sets of control experiments were performed (Fig. 3C and D). In the first (Fig. 3C), invasion by serovar Typhi (which is known to use CFTR as a receptor) was compared to that by serovar Typhimurium, which invades epithelial cells in a CFTR-independent manner (18). In the second (Fig. 3D), the CFTR-specific component of serovar Typhi invasion was competitively inhibited (18) through the addition of CFTR-specific monoclonal antibody. Pretreatment of the epithelial cells with bacterial extracts had no significant effect on invasion by serovar Typhimurium (Fig. 3C). Thus, mobilization of CFTR to the epithelial cell plasma membrane only enhanced invasion by Salmonella bacteria that use CFTR as an epithelial cell-based receptor. Additionally, while treatment of T84 cells with extract from E. faecalis strain E1 significantly enhanced invasion by serovar Typhi, this increase in invasion was negated by antibody CF3 (Fig. 3D), which binds to the serovar Typhi binding site on CFTR (18). E1 extract enhancement of serovar Typhi invasion was not affected by the isotype- and species-matched control antibody CF8, which binds to an intracellular epitope on the CFTR protein.

The commensal microbes which constitute the normal intestinal microflora have previously been shown to have numerous beneficial effects on the health of their human hosts, including production of vitamins (7), conversion of toxic metabolites (16), maintenance of the architecture (11) and barrier function (5) of the intestinal epithelium, and regulation of the mucosal immune system (14). It is also commonly hypothesized that commensals benefit the host by competing with potential pathogens, depriving them of required space and nutrients. The reports of these beneficial effects have led to the practice of probiosis, the intentional consumption of live commensal microbes to bring about their colonization and to obtain their beneficial effects.

The data reported here demonstrate for the first time another mechanism by which commensal microorganisms may influence the health of the human host, i.e., by triggering epithelial cell trafficking of a protein that serves as a receptor for pathogenic bacteria. This mechanism differs from those reported previously in two important aspects. First, while previously studied mechanisms of probiotic action focus on a protective effect for the host against infectious disease, the commensal-mediated mobilization of CFTR reported here may result in increased susceptibility of the human host to disease. Secondly, most protective probiotic effects affect the human host in a way that is independent of a bacterial pathogen's species. For example, the act of increasing the barrier function of the intestinal epithelium would be expected to offer protection against numerous pathogenic species, as would increasing competition for vital nutrients. In contrast, the data reported here suggest that the physiology of intestinal epithelial cells can be modified by commensal microorganisms in such a way that the epithelium is rendered selectively susceptible to invasion by a specific bacterial pathogen (serovar Typhi but not serovar Typhimurium).

S. enterica serovar Typhi causes typhoid or enteric fever and initiates infection by entering the enterocytes and specialized M cells of the small intestine. The mechanism by which serovar Typhi enters enterocytes depends upon the delivery of bacterial components into the host cell via the bacterial type III secretion apparatus (4). Since assembly of the type III apparatus has been shown to be stimulated by bacterium-epithelium contact (6), early interactions between serovar Typhi and the epithelial cell play an important role in the initiation of infection.

CFTR is used by S. enterica serovar Typhi as a receptor to facilitate its invasion of epithelial cells and its translocation to the gastric submucosa (18). This activity is in addition to CFTR's normal physiological role as a regulator of epithelial cell secretion of potassium, chloride, and sodium ions (reviewed in reference 19). Mutations in CFTR that diminish its expression dramatically reduce the efficiency of serovar Typhi invasion of epithelial cells, as do experimental conditions that render CFTR unavailable for interaction with its serovar Typhi-based ligand, lipopolysaccharide (13, 18).

CFTR protein is expressed on the apical plasma membrane of intestinal epithelium. However, at any given time, a large portion of the CFTR in each epithelial cell is stored in subapical vesicles rather than residing on the plasma membrane (23). Workers in this laboratory have previously reported that vesicular stores of CFTR protein can be stimulated to accumulate on the plasma membrane in response to serovar Typhi infection (12). The CFTR that accumulates on the plasma membrane serves as a receptor for serovar Typhi and mediates enhanced invasion by this bacterial pathogen.

The bacterial factor that stimulates trafficking of the CFTR protein from intracellular stores to the plasma membrane has not yet been identified. Preliminary analysis suggests that it is a protein (12) and is >100 kDa in size (unpublished observations). While the identity of the CFTR-modulatory activity is not yet known, the fact that it has been detected in examples of both gram-negative (12, 24) and gram-positive (this study) species suggests that there may be more than one substance associated with this biological activity. It is possible that innate mechanisms for detecting pathogen-associated molecule patterns are involved in this process, since such pathways are able to detect the products of both gram-positive and gram-negative species (for a review, see reference 2). Identification of this bacterial factor will allow wider genetic screening of commensal and pathogenic bacteria for possession of the factor, and it will make possible the characterization of its expression and how its expression is affected by environmental stimuli. Additionally, identification of the epithelial cell-based receptor that initiates the CFTR-modulatory event will further our understanding of the communication between commensal bacteria and the intestinal epithelium, and such identification may suggest novel mechanisms for the control of physiological functions such as CFTR-regulated ion conductance (8).

In this report, I have shown that the CFTR protein, which serves as an epithelial cell receptor for serovar Typhi, is mobilized to the epithelial cell plasma membrane in response to the products of some microorganisms that are normally present in the intestinal microflora. This mobilization of CFTR enhances invasion and translocation of serovar Typhi. These results suggest that the combination of commensal microorganisms present in the small intestine can influence the susceptibility of the intestinal epithelium to invasion by serovar Typhi. The population of normal flora in a given individual is not static but rather changes with age (15), alterations in diet (17, 21), and the clinical application of antibiotics (1). The findings reported here are relevant to these and other scenarios in which the population of commensals in the intestine is perturbed.