Beth A. McCormick^1,2, Samuel I. Miller³, and James L. Madara^1,2.

¹ Departments of Pathology, Brigham and Women's Hospital, and Harvard Medical School, Boston, MA 02115

² The Harvard Digestive Diseases Center

³ Departments of Medicine and Microbiology, University of Washington, Seattle, WA 98195.

Received 06/19/96; Accepted 07/09/96; On-line 08/01/96

3. THE NATURE OF SALMONELLA-INTESTINAL INTERACTIONS

3.1 Salmonella and environmental cues

Salmonella spp. are facultative intracellular pathogens that cause a variety of diseases in both humans and animals, which range from a self-limiting enterocolitis (food poisoning) to more systemic illnesses such as typhoid fever. The type of diseases caused by these organisms depends not only on the serovar or species of the infecting bacteria but also on the species of the infected host. Some serotypes such as S. typhi are host adapted, in this case for humans, while others, such as S. typhimurium and S. enteritidis, can cause disease in a large variety of hosts (11). S. typhimurium, for example, will specifically cause self-limited gastroenteritis in immuno-competant humans, while in mice it will cause a severe systemic illness, much like typhoid fever. Salmonella infection is initiated when bacteria enter a host via contaminated food or water. Following passage through the stomach, the organisms move into the gastrointestinal tract of the host, and upon reaching the distal ileum, establish contact with a cellular target within the intestinal mucosa. As a result of such associations Salmonella are able to initiate passage through the intestinal epithelium where they can gain access to the reticuloendothelial system, thus providing an avenue for the dissemination to the lymph nodes, spleen, liver, and blood (12).

Once in the gastrointestinal microenvi-ronment, the microorganism may interact with the apical membranes of columnar intestinal epithelia (7) or with specialized cells, termed M cells, which lie over the Peyer's patches (7, 13-14). The relative contributions of these interactions to the pathogenesis of disease is uncertain. In mice, evidence suggests that early entry of S. typhimurium appears to be via transepithelial transport by M cells to the Peyer's patches (12-13). M cells represent a minor constituent of the epithelial surface (far less than 1%) (15), and are specialized epithelial cells that appear to be designed for taking up large particles and, in addition, are believed to be important in antigen sampling. Some pathogens which are able to associate with and translocate across columnar intestinal epithelia use the M cell pathway to enter the host (for instance reovirus) (16). Moreover, enterocytes can also be invaded by S. typhimurium, providing an additional portal of entry (7). Columnar epithelial cells of the intestine constitute the major portion of surface area (15), are known to bind Salmonella and internalize it, and are the site at which neutrophil transmigration in response to such surface colonization is known to occur (7-10). Thus, it appears that while a number of microorganisms are able to enter the host through the M cells, invasion of enterocytes seems to be a less restricted more prevalent route. Although M cell and Salmonella associations play a role in host immunity to this organism, the bulk of procaryotic/eucaryotic interactions in primary colonization of the intestine by Salmonella likely occurs over the general columnar epithelial surface as suggested by studies of Takeuchi (7).

Once within the gastrointestinal microenvironment, the bacterium is exposed to extremes of temperature and pH, oxygen tension, bile salts, digestive enzymes, and a multitude of diverse, competing microorganisms. Such distinct, and seemingly hostile environments, are not only tolerated by the bacteria but importantly, serve as environmental signals for the microbe to initiate transcription of genes specifically adapted for host-microbe interactions. Thus, it is not suprising that the expression of S. typhimurium virulence factors important to interactions with epithelia is influenced by various environmental stimuli including oxygen (17), osmolarity (2, 18), and growth phase (1, 3, 18); conditions known to have effects on the level of DNA superhelicity (2). For example, previous reports have indicated that Salmonella adherence to and subsequent invasion into cultured epithelial cells was greatest either during the late logarithmic phase of growth, presumably due to oxygen limitation, or when the bacteria were grown either anaerobically or incubated with cells under anaerobic conditions (1, 3). Furthermore, an assay in which a short bacterium-cell interaction period was used, also concluded that only Salmonella grown under low oxygen conditions, but not bacteria from stationary phase cultures, elicited rapid changes in cell morphology, internal actin filament rearrangement, and cell entry(17). Recent work (5), however, has subsequently demonstrated that invasion of S. typhimurium by epithelial cells could be reduced during utilization of carbohydrates and that the repression of cell association by certain carbohydrates (i.e. glucose) was greater during aerobic growth of the bacteria. Thus, this study suggests that previous reports of greater cell invasion by S. typhimurium during anaerobic growth may have risen from the use of media containing carbohydrates which were found to be more repressive during aerobic growth of the bacterium (5). Nonetheless, depletion of a preferred carbohydrate substrate in the presence of other potential carbon energy sources represents a nutrient limitation that can stimulate a responsive change and adds to the record of environmental stimuli that control the capability of Salmonellae to associate with and invade epithelial cells in vitro (5).

S. typhimurium virulence can also be controlled, at least in part, by the global regulatory network, PhoP/PhoQ (19-22). Such two component regulators are members of a family of environmental sensors (PhoQ) and transcriptional activators (PhoP) that are required for the expression of genes termed pag (phoP-activated-genes) (23-25). pag are transcriptionally activated within acidified macrophage phagosomes several hours after phagocytosis and are required for intracellular survival (20, 24, 26). In addition to the ability to transcriptionally activate pag, PhoP/PhoQ can repress the synthesis of proteins encoded by genes designated prg (phoP repressed-genes) (19 21). prg products are likely to play important roles in S. typhimurium signaling to eucaryotic cells which include; (i) induction of macrophage generalized membrane ruffling, macro-pinocytosis and internalization of bacteria within spacious phagosomes (27-28); (ii) induction of bacterial mediated endocytosis (BME) by epithelial cells (11); and (iii) induction of polymorphonuclear leukocyte transmigration across polarized epithelial cell monolayers (29) (see below). Recent evidence also indicates that Mg²⁺ is an extracellular environmental signal that controls the PhoP/PhoQ regulon (30).

3.2 Salmonella pathogenesis and host cell invasion

The ability to penetrate the cells of the intestinal epithelium is an essential step in the pathogenic cycle of the enteric pathogen Salmonella (7). The Salmonella invasion process is thought to involve the interaction of determinants on the surface of the bacteria with the host cell, triggering an event that resembles macropinocytosis, referred to as bacterial mediated endocytosis. Observations by Takeuchi (7) first provided the groundwork for understanding the sequence of events that leads to the entry of Salmonella into intestinal epithelial cells. Via electron microscopic studies, he was able to demonstrated that the microvilli of the intestinal epithelium underwent dramatic changes after Salmonella came into close proximity to the brush border. Such changes, exemplified by membrane ruffles, localized to the point of bacterial contact and were transient, since after internalization of the bacterium the microvilli recovered their normal, preinfected appearance. Neither the molecular basis for this phenomenon, which appears to be unique to Salmonella and Shigella , nor the significance of this event, are completely understood. Our current understanding is that the ability of Salmonella to induce formation of membrane ruffles is critical for entry since mutants unable to induce these changes are severely impeded in their ability to enter cultured mammalian cells (31, 32). Salmonella are then subsequently internalized via membrane bound vacuoles formed from such membrane ruffles. This process is termed macropinocytosis, and results in the formation of spacious phagosomes (26-28, 33-34). The appearance of membrane ruffles on the surface is accompanied by profound cytoskeletal rearran-gements at the point of bacterial- host cell contact, which require a number of cytoskeletal proteins, including actin, alpha-actinin, talin, tubulin, tropo-mysin, and ezrin (35). Although the significance of the recruitment of several of these proteins is unclear, actin is likely to play a role in the formation of membrane ruffles since inhibitors of actin microfilament function blocks Salmonella entry (36-37). Other organisms have also been shown to induce cytoskeletal alterations. Listeria monocytogenes, for example, continues to generate interest by virtue of its ability to induce locomotion through directional actin assembly within the host cells (38-41). Previous results have demonstrated that actin-rich rocket tails trailing behind motile bacteria became anchored in the cytoplasm (38, 42-43). Such rocket tails create physically confining boundaries that control the direction of bacterial movement. Yersinia spp. can also influence host cytoskeletal proteins. Entry of enteropathogenic Yersinia into cultured mammalian cells has been described as parasite-specified phagocytosis (44), in which movement of the host cytoskeleton in response to signals is sent from a transmembrane receptor that is recognized by bacterially encoded ligands. For example, previous results have indicated that actin, and actin associated proteins, such as filamin and talin, accumulate around the entering bacterium (45). Yersinia invasin is responsible for such activity, and this observation is consistent with the finding that invasin recognizes multiple beta 1 integrins (46).

S. typhimurium infection of cultured epithelial cells is also accompanied by a marked increase in [Ca²⁺]i flux (32, 47), an event which is most likely necessary for internalization and may play a role in the formation of membrane ruffles. Other cellular responses include tyrosine phosphorylation of a number of host proteins along with the epidermal growth factor receptor and the initiation of signal transduction pathways which ultimately lead to the activation of phospholipase A2, and production of arachidonate metabolites (47). However, Francis et al (28) found that invasive Salmonella elicited characteristic ruffles in Swiss 3T3 fibroblast and NR-6 cells, a 3T3 derivative that does not express the EGF receptor. Moreover, invasive Salmonella also elicited membrane ruffling in rat basophilic leukemia cells, despite failure of EGF to elicit ruffling in these cells (28). Thus, this study suggests that the EGFR is not required for S. typhimurium invasion, although there may be alternative (parallel) pathways that cause either direct or indirect phosphorylation of the EGF receptor during Salmonella-induced ruffling. Nonetheless, it is interesting to note that the events of membrane ruffling, cytoskeletal rearrangements, and Ca²⁺ influxes also occur as part of a global cellular response to mitogens, oncogene expression, and growth factors (48-50) and are associated with enhanced pinocytosis (48). Thus, the formation of membrane ruffles comprises the macropinocytotic machinery mediating pinocytosis, and is subverted by Salmonella so as to enter into diverse mammalian cells, perhaps reflecting the existence of a common pathway for ruffle induction.

3.3 Salmonella entry into epithelial cells is determined by genetically defined factors

Animal cells in culture as well as transformed human cell lines have become popular in vitro models for studying attachment to and invasion into epithelial cells (6, 51-58). Although, genetic approaches have been used to identify the Salmonella factors that directly interact with the epithelial cells and facilitate invasion, such studies have indicated that the genetics controlling these processes are complex and involve multiple chromosomal loci (59-62). For example, S. typhi genes have been identified which, when cloned into Escherichia coli K-12, allow this normally non-invasive bacterium to enter cultured cells (63). In addition, Stone et al., (62) has identified a number of TnphoA mutants of S. enteritidis that render these organisms defective for invasion into cultured epithelial cells. Such mutants mapped to 9 different loci on the chromosome and affected entry to different degrees. Using a similar technique, Betts and Finlay (59) isolated transposon mutants from S. typhimurium in 4 distinct loci that also rendered these organisms deficient for entry into epithelial cells. Together these studies imply that Salmonella may encode alternative entry pathways since it has not been determined whether these different loci are functionally related. Moreover Lee et al., developed a strategy which selected for mutants of S. typhimurium which were competent for cell entry into epithelial cells only under non-physiologic conditions (i.e. during non microaerophilic growth) (61). Such a strategy identified the hil locus which is essential to bacterial entry into cultured epithelial cells, and which presumably encodes a regulatory factor required for proper expression of entry determinants. The PhoP-repressed locus prgH was also previously identified as being important for signaling epithelial cells to endocytose S. typhimurium. Characterization of prgH revealed that it is an operon of four genes prgHIJK, strongly linked to the hil locus (19). Synthesis of the transcript was repressed in bacteria that activate PhoP/PhoQ, thus indicating that PhoP/PhoQ regulates prgHIJK by transcriptional repression. Another successful strategy for the detection of Salmonella invasion genes, was developed by Galan and Curtiss who identified a S. typhimurium genetic locus, inv, based on its ability to complement a non-invasive strain of S. typhimurium (60). Subsequent analysis of this locus has identified at least 14 genes, which upon mutation affected the ability of Salmonella spp. to enter cultured epithelial cells, without affecting the ability of these cells to attach, suggesting that attachment and entry are genetically separate events in Salmonella. These genes are apparently arranged in the same transcriptional unit and were mapped to 59 min on the Salmonella chromosome (31) near the hil (64) and prgHIJK loci (19). Subsequent to the discovery of inv, a new assemblage of genes responsible for invasion properties of Salmonella were identified which were remarkably similar in order, arrangement and sequence to the gene cluster controlling the presentation of surface antigens (spa) on the virulence plasmid of Shigella (46). In Salmonella, this chromosomally encoded complex, also called spa, consists of over 12 overlapping or adjoining genes with the inv locus, suggesting a single transcription unit (inv/spa complex) (65).

To date some 25 invasion genes, as shown in Figure 1, have been found to be clustered near minute 63 of the S. typhimurium chromosome (19, 31, 61-63, 65), and 40 kb of unique DNA may be necessary for entry of Salmonella into mammalian cells. Interestingly, the finding that a non-invasive spa mutant of Salmonella could be rescued by the corresponding Shigella homologue, provided the initial evidence that spa, perhaps, promotes equivalent functions in Shigella and Salmonella (Figure 1). Presumably, this gene cluster has been acquired independently by each genus yet displays motifs used by diverse antigen export systems including those required for flagellar assembly and protein secretion (65). For example, a number of predicted inv gene products have been identified and are similar to proteins thought to be involved in export and assembly of bacterial flagellar components (65-68). Among these homologues are proteins involved in flagellar assembly in E. coli (FlhA, Flil, FliJ, and FliN), Bacillus subtilis (FliP and flaA locus), as well as in bacteriophage assembly (Protein IV and Pf3). Furthermore, recent observations have indicated that 12 of these genes, invG, invE, invA, invB, invC/spaL, spaM, spaN, spaO, spaP, spaQ, spaR, and spaS, have the identical gene order and significant sequence similarity to the Shigella mxi and spa genes (31, 65-70). The mxi and spa genes are encoded on the large Shigella virulence plasmid and are required for export of Ipa proteins (invasion protein antigens) which facilitate Shigella entry into mammalian cells (68-69, 71-73). Additionally, the role of the prgHIJK operon in BME was supported further by the finding that some of its predicted gene products of this locus were similar to S. flexneri secretion determinants that are essential for epithelial cell invasion. For example, the prgI, prgJ, and prgK predicted gene products of S. typhimurium were recently found to be similar to the MxiH, and MxiI, and MxiJ proteins, respectively, of S. flexneri (74). The relationship of such genetic organization of the invasion genes between S. typhimurium and S. flexneri is depicted in Figure 1. Such Salmonellae genes also show sequence similarity to several genes that encode proteins involved in the surface presentation and/or secretion of a variety of molecules in a number of other organisms including, Yersinia (LcrD, LcrE, and YscA), Klebsiella (PulD), Aeromonas hydrophila (ExeD), and Xanthomonas campestris (PefD). The significance of these findings is that these homologies indicate that Salmonella may externalize invasion proteins by a mechanism that is functionally similar to that involved in flagellar export and assembly, such that Salmonella, like Yersinia, may assemble a supramolecular structure on its surface, in order to induce its internalization into mammalian cells.

Figure 1. Similarities between the genetic organization of the invasion genes clusters from Salmonella typhimurium and Shigella flexneri. This map shows the relative positions and the transcriptional directions of the genes illustrated, as indicated by the position of the arrow. Gene clusters which are conserved both in sequence and in gene order are indicated by the following key: inv/spa;mxi/spa (open/white bars), prgIJK;mxiHIJ (diagonal stripped bars), and ipa;sip (stippling). Black bars indicate genes with no homologues within the respective regions. The map organization was compiled from information published by the following investigators (65, 69, 74, 84-85, 88).

3.4 Salmonella and secreted invasion determinants

To this end, in studies recently performed by Miller and colleagues (75), analysis of the culture supernatants from wild-type S. typhimurium demonstrated that at least 25 polypeptides larger than 14 kDa were detected. In contrast, prgH1:TnphoA, phoP-constitutive, and hil deletion mutants had significant defects in their supernatant protein profiles. These results suggest that PhoP/PhoQ regulates extracellular transport of proteins by transcriptional repression of secretory determinants and that secreted proteins may be involved in signaling the epithelial cells to endocytose bacteria (69). Because of the similarity between predicted gene products from the prgHIJK operon and gene products required for protein secretion in other bacterial species, an analysis of proteins present in culture supernatants of S. typhimurium was performed (75). This analysis suggested that PhoP/PhoQ could control protein secretion, at least in part, by repressing prgHIJK, whose product could form part of a secretion machinery. Since the strains with altered Ssp profiles were impaired in signaling epithelial cells, this report suggests that Ssp are involved in signaling such cells to initiate BME. The possibility that Salmonella proteins form an apparatus assembled on the cell surface that is necessary in order to signal eucaryotic cells was suggested by the work of Galan and colleagues (67), who demonstrated that S. typhimurium forms a novel surface structure which is lost as the organism enters membrane ruffles of epithelial cells. The release of this apparatus from the cell surface during growth in culture could result in the detection of these proteins in the supernatant. Specifically, contact between S. typhimurium and epithelial cells resulted in the formation of appendages (invasomes) on the surface of the bacteria, which did not require de novo protein synthesis, and was a transient event (67). Such, appendages were immediately shed or retracted before or subsequent to signaling the host, since S. typhimurium associated with membrane ruffles did not exhibit these surface structures. Moreover, such surface structures were not seen on organisms unexposed to the host cells, and S. typhimurium mutants defective in the transient formation of these surface appendages were unable to enter into cultured epithelial cells, suggesting that these structures are required for bacterial internalization. As a consequence of this interaction, appendages are specifically assembled on the surface of the Salmonella, a process which appears to be required for subsequent triggering of the host cell signal-transduction pathways that lead to membrane ruffling and the internalization of these organisms (31, 47, 67). Thus, such intimate interactions represent a phenomenal example of reciprocal biochemical signaling between a pathogen and the host, itself, and emphasizes the notion that Salmonella can sense environmental signals from the host cell resulting in the transient assembly of a surface organelle on the bacterium.

3.5 Salmonella and the protein secretion apparatus

Salmonella entry (internalization) into host cells has recently been found to require the function of a dedicated, sec-independent, type III protein secretion system encoded in the inv and spa loci located at minute 59 on the Salmonella chromosome (76). Presumably, such a translocation apparatus would actively participate in the host cell contact-dependent assembly of a supramolecular structure, presumably required for the presentation and/or delivery of invasion determinants to the target cell. Evidence to support this notion is exemplified by the fact that invasome assembly, itself, requires a functioning type III secretion system. The type III secretion systems are usually encoded by genes that are clustered together on the chromosome (Salmonella ssp.(76)) or on large plasmids (Shigella, Yersinia (77, 78)). One protein common to all of the type III systems is an ATPase which presumably energizes the transport system. In Salmonella spp. this is InvC, which is homologous to spa47 from Shigella (73) and yscN of Yersinia spp. (79-80). Another common component of the type III systems is an outer membrane-associated translocase that is homologous to PulD from K. oxytoca (81). This protein is InvG in Salmonella spp. (70), MxiD in Shigella spp. (69) and YscC in Yersinia spp. (82). Similar type III secretion systems are also required for the virulence phenotype of other pathogenic bacteria including Yersinia spp., Shigella spp., and enteropathogenic E. coli, as well as a number of plant pathogens from the Xanthomonas, Pseudomonas, Aeromonas, and Erwinia genera (83-84). Such a secretion system is distinct from both the type I (sec-independent) protein secretion system exemplified by the export of the E. coli heamolysin, and the type II (sec-dependent) general secretory pathway of gram negative bacteria exemplified by the secretion of pullulanase of Klebsiella oxytoca.

Several proteins have currently been identified whose secretion into the culture supernatant of S. typhimurium is dependent on the type III secretion system (85), and in addition have determined to be potential components of the invasome structure (67). InvJ was the first identified target of the protein secretion apparatus, and exhibited significant sequence similarity to the EaeB protein of enteropathogenic E. coli. Initial observations found that 30% of the product of invJ(spaN) was recovered in the culture supernatant of wild-type S. typhimurium but not from that of invG or invC mutants (86). Moreover, mutations in invC or invG prevented the assembly of the invasome and dramatically reduced Salmonella entry into the host cells, indicating that there is a close correlation between invasome assembly and the internalization process. Although the function of InvJ is yet to be determined, it is thought to be a candidate for either a structural component of the invasome appendage, or alternatively, InvJ secretion may precede invasome formation, serving as a signal to assemble the appendage (87). Other genes encoding secretion proteins have been identified and include sipB, sipC, sipD, and sipA (85-86). Such genes encode polypeptides that have significant sequence homology to the IpaB, IpaC, and IpaA proteins of Shigella spp., respectively (85-86, 88), and are themselves targets of the type III secretion system (Figure 1).