Passage of Salmonella through polarized epithelial cells: role of the host and bacterium

Diseases caused by Salmonella species (salmonellosis) pose major health problems throughout the world (Cohen & Tauxe, 1986). However, the mechanisms used by Salmonella to cause disease are not well characterized (Stocker & Makela, 1986; Finlay & Falkow, 1988a). Salmonella bacteria are considered intracellular parasites and it is in the small intestine that Salmonella species begin to penetrate the intestinal mucosa (Rubin & Weinstein, 1977). The bacteria interact with tips of epithelial microvilli and cause the microvilli to disappear before entering membrane-bound inclusions in the epithelial cells (Takeuchi, 1967). Once inside, the bacteria can pass through these cells by a process we have called transcytosis (Finlay et al. 1988a), thereby gaining access to the host.

Several workers have used in vitro tissue culture monolayers to examine bacterial interactions with eucaryotic cells, including bacterial invasion (or entry) (for examples, see Giannella et al. 1973; Devenish & Schiemann, 1981). These systems have the advantage of consistency and are much easier to use than primary isolates of eucaryotic cells. We have used tissue culture models to study Salmonella interactions with eucaryotic cells (Finlay & Falkow, 1988a,b,c; Finlay et al. 1988a,b). The cell lines we have used extensively are MDCK (Madin Darby canine kidney epithelial cells), HEp-2 (human larynx epithelial cells), CACO2 (human intestinal epithelial cells) and CHO (Chinese hamster ovary fibroblast cells). Salmonella species can invade all of these cell lines in a manner similar to that described above for entry into the intestinal epithelium. Use of various inhibitors and mutant cell lines allowed us to demonstrate that endosome acidification is not required for bacterial entry or intracellular replication (Finlay & Falkow, 1988c; Finlay et al. 19886).

Although the growth of tissue culture cells on plastic supports permits the study of bacterial invasion, it does not enable the investigator to examine subsequent intracellular events such as migration through the cell and exit to the opposite surface, processes essential to Salmonella penetration of the intestinal epithelium. To facilitate study of these processes, we used polarized epithelial monolayers of MDCK epithelial cells grown on filters with 3 μm pores (Finlay et al. 1988a). Other workers have demonstrated that these cells form a barrier impermeable to molecules as small as ions (reviewed by Simons & Fuller, 1985). Cells grown in this manner have defined apical (top) and basolateral (bottom) surfaces separated by tight junctions, mimicking an epithelial barrier. When added to the apical surface, S. choleraesuis cells pass through (transcytose) this barrier and enter the medium bathing the basolateral surface by 4 h, reaching a maximal rate of 14 bacteria/MDCK cell per hour after nine hours. Although other Salmonella species such as S. typhimurium and S. enteritidis behave nearly identically to S. choleraesuis, noninvasive Escherichia coli added to the same samples do not penetrate these barriers. Basolateral infection was about 100 times less efficient. Interestingly, S. choleraesuis bound much more efficiently to the apical surface than the basolateral surface, perhaps indicating a concentration of bacterial receptors on the apical surface. By measuring electrical resistance across these monolayers, we found that Salmonella disrupted the epithelial tight junctions, causing a loss in resistance across the monolayer. This epithelial model system has allowed us to begin to address most of the events involved in epithelial penetration by Salmonella, not just initial entry steps.

We have previously shown that cytochalasins B and D, agents which disrupt microfilament formation, inhibit Salmonella internalization (Finlay & Falkow, 1988b,c). Another invasive bacterium Shigella flexneri, which also requires microfilaments for internalization, is surrounded by polymerized actin filaments as it invades (Clerc & Sansonetti, 1987). We used the fluorescent stain phallicidin to visualize the role of microfilaments in Salmonella invasion. Phallicidin binds polymerized F actin but does not stain G actin monomers. When we treated MDCK epithelial monolayers with phallicidin, various microfilament structures could be observed. When these monolayers were infected with the highly invasive and pathogenic Salmonella, S. choleraesuis, many small, brightly staining ‘balls’ were apparent in the infected monolayers (Fig. 1). Although infected monolayers contained many of these structures, uninfected monolayers exhibited only a few. By counterstaining with anti-bacterial antibodies and rhodamine, we found that there were single bacteria inside nearly all of these structures (unpublished data). That such structures are similar to those observed with Shigella (Clerc & Sansonetti, 1987) suggests that the role of microfilaments in invasion for these two organisms may be equivalent.

The presence of these regions of concentrated microfilaments in uninfected monolayers suggests that this process occurs in normal epithelial cells, albeit infrequently. This process is similar to phagocytosis in macrophages, which requires localized actin polymerization at the site of internalization. It would seem that these bacteria can take advantage of this pre-existing process to gain entry into host cells, perhaps by interacting with a specific receptor on the epithelial cell surface which is normally internalized by this process.

When a Salmonella interacts with an epithelial cell, it causes microvilli to disappear (Takeuchi, 1967; Finlay et al. 1988a). This is illustrated in Fig. 2A. As these organisms enter epithelial cells, they are enclosed by a membrane-bound vesicle within which they remain (Fig. 2B). As time progresses, the bacteria multiply within these vacuoles (Fig. 2C), with a generation time of approximately 50min (Finlay & Falkow, 1988c), and eventually fill the entire cytoplasmic space with Salmonellae.

If MDCK cells are grown on filters with 3 pores, the bacteria can enter these cells and proceed through to the opposite surface (Fig. 2D; Finlay et al. 1988a). In previous studies we did not determine how efficient bacterial transcytosis was compared to bacteria remaining within the cell or escaping to the same surface from which they invaded. To answer this question, S. choleraesuis bacteria were radiolabelled with [³⁵S]methionine, added to the apical surface of polarized MDCK monolayers grown on filters (Finlay et al. 1988a), and incubated for 4h. After incubation, the medium was replaced and allowed to incubate for given time periods before counting radioactivity in the monolayer and media from both sides. These results are presented in Table 1. After 4h incubation and washing, most of the remaining counts were associated with the monolayer. However, even after 8 h, the majority (90 %) of bacteria remained in association with the monolayer with 8-7 % of the counts in the apical medium and only 1·3 % in the basolateral medium.

These data suggest that although Salmonella enters polarized epithelial cells, the bacteria are not efficiently transported through the monolayer, and most remain internalized within the MDCK cells. Bacterial exit to the apical surface (the surface to which they were added) is more efficient than passage through MDCK cells to the basolateral surface. Nevertheless, because Salmonella replicates within these cells, many bacteria still appear in the basolateral medium (Finlay et al. 1988a). In vivo, the successful penetration of a few microorganisms through the intestinal barrier is probably sufficient to cause disease. Thus it appears that Salmonella penetration through epithelial cells does not occur by a specific transport mechanism (Mostov & Simister, 1985), but a small fraction of intracellular bacteria still escape through the basolateral surface.

We previously determined that Salmonella association with MDCK cells required bacterial RNA and protein synthesis (Finlay et al. 1988a). Other workers have found that 5. typhimurium adherence to rat intestinal epithelial cells requires bacterial protein synthesis (Lindquist et al. 1987). More recently we have learned that bacterial adherence and invasion require both RNA and protein synthesis (Finlay et al. unpublished). These results suggest that Salmonella species must synthesize proteins to interact with epithelial cells.

We attempted to identify these induced proteins by pulse-labelling S. choleraesuis with [³⁵S]methionine in the presence and absence of epithelial cells (Fig. 3). Bacteria adhering to MDCK monolayers which had been fixed in 2% glutaraldehyde made several proteins not synthesized by non-adhering bacteria in the supernatant of the same sample (Fig. 3). Additionally, some proteins were ‘shut off’ by bound bacteria. S. typhimurium, another invasive Salmonella, made identical induced proteins (data not shown). Induction of these proteins required about 4 h incubation with epithelial cells before detectable amounts were synthesized (Fig. 3A). After about 10h the synthesis of these proteins decreased. However if bacteria were labelled at 4h and then incubated for another 6 to 8 h, these labelled proteins were still present (data not shown), suggesting that even after 10 h these proteins are present in the bacteria, but their synthesis is stopped.

As an alternative to using fixed MDCK monolayers to induce protein synthesis, we treated viable MDCK and HEp-2 (human epithelial) monolayers with cycloheximide and emitine to inhibit eucaryotic protein synthesis. This treatment did not affect bacterial adherence or invasion, and identical induced bacterial proteins were observed by pulse labelling (data not shown), indicating that numerous Salmonella proteins are induced in the presence of epithelial cells.

We attempted to determine the molecular nature of the entity on the epithelial cell surface which induces specific bacterial proteins. MDCK polarized monolayers were treated with trypsin, periodic acid (which oxidizes sugar residues), and neuraminidase (which removes sialic acid residues) before fixation and bacterial addition. As shown in Fig. 3B, any of these treatments completely eliminated the synthesis of the induced proteins. We also found that these treatments inhibited bacterial adherence and invasion (Finlay et al. unpublished). These data indicate that glycoprotein-like structures are required on the epithelial cell surface for stimulating induction of specific proteins which are required for bacterial adherence and invasion.

The interactions of Salmonella species with epithelial cells are complex. Host cells appear to contribute to bacterial adherence and invasion by providing glycoproteinlike receptors and active microfilament polymerization to internalize bacteria. Host cell protein synthesis is not required, as demonstrated by the use of cycloheximide and emitine, which did not alter bacterial adherence or invasion levels. Bacteria were also capable of adhering to glutaraldehyde-fixed monolayers at levels similar to viable monolayers, indicating that epithelial cell receptors are present on the surface of uninfected cells. However, once inside, Salmonella does not appear to utilize any pre-existing pathway for intracellular transport through the epithelial cell. Instead, most of the bacteria remain within the cell, with only a few exiting to either epithelial cell surface.

Salmonella adherence and invasion of epithelial cells is an active event. These bacteria must synthesize RNA and proteins in order to adhere to the epithelial cell surface and invade. This is in contrast to other invasive enteric bacteria such as Yersinia spp., which can adhere to epithelial cells on ice or invade in the presence of bacterial RNA and protein synthesis inhibitors (Isberg et al. 1987; Finlay et al. unpublished data). Salmonella spp. presumably interact with a glycoprotein-like structure(s) on the epithelial cell surface, which then triggers the synthesis of several bacterial proteins required for bacterial adherence and invasion. Salmonella adherence to eukaryotic cells is biphasic, with an initial weak interaction which is easily disrupted followed by strong adherence (Jones et al. 1981). It is the second phase of adherence that is blocked by bacterial protein synthesis inhibitors (Lindquist et al. 1987). We propose that the first phase of adherence is the initial interaction required to stimulate the production of bacterial proteins essential for the stronger attachment phase which occurs later and is associated with invasion.

The stimulation of several bacterial proteins and inhibition of synthesis of others is suggestive of a coordinately regulated event, stimulated by epithelial cell surfaces. Bacteria have many systems which respond to environmental signals and alter the synthesis of bacterial proteins to adjust to these new environments (Gottesman, 1984). A bacterium passing from the small intestine into eukaryotic cells would enter a markedly different environment. We hypothesize that the alterations in Salmonella proteins which we observe are required for the bacteria to successfully enter and survive in an intracellular environment, and these changes are probably a coordinately regulated event. As we begin to characterize the bacterial genes involved in adherence and invasion (Finlay et al. 19886), we will gain further insight into the regulation and nature of these genes and their products.

This work was supported by grant Al 18719 from the National Institute of Health to SF. BBF is a recipient of a post-doctorate Fellowship from the Alberta Heritage Foundation for Medical Research.