Chapter 151

Molecular and Cellular Bases of Salmonella and Shigella Interactions with Host Cells

JORGE E. GALÁN and PHILIPPE J. SANSONETTI

INTRODUCTION

Salmonella and Shigella spp. have evolved different strategies to subvert normal host cellular functions. These strategies allow these bacteria to gain access to and survive and replicate within host cells as well as, in the case of Shigella spp., to spread to neighboring cells without an extracellular phase. Reaching an intracellular compartment may allow these microorganisms to gain access to more favorable environments in which they can replicate and/or avoid host defense mechanisms. In this chapter, we will give an overview of the interactions of Salmonella and Shigella spp. with host cells, focusing on those interactions that occur at early stages of the pathogenic cycle involving the intestinal epithelium. Other aspects of the pathogenesis of Salmonella spp. are covered in chapter 152. Readers are also referred to several reviews on these subjects (22, 55, 71, 75, 146, 190).

Experimental Systems To Assay Salmonella and Shigella Interactions with Host Cells

A number of in vitro and in vivo systems have been used to study the interaction of salmonellae and shigellae with host cells. Although these assays are a simplification of the undoubtedly more complex events that occur in the infected host, they have been tremendously useful in the study of a variety of pathogenic mechanisms. Furthermore, a number of studies have demonstrated that there is a good correlation between these in vitro models and natural infections (reviewed in reference 157).

Infection of Semiconfluent Epithelial Cells.

Infection of semiconfluent cultured epithelial cells has been extensively used to study several aspects of the interactions of Shigella and Salmonella spp. with host cells. Pioneering work by Gerber and Watkins (82), LaBrec et al. (125), Hale et al. (95, 96), and Giannella et al. (84) helped to establish these systems, which were later refined by improving their quantitative aspects (168). These assays, in combination with a variety of bacterial genetic techniques, have allowed the identification of a number of genes involved in different aspects of the interactions of these bacteria with their host cells. In addition, they have been instrumental in the identification of host cell functions required for bacterial internalization.

Infection of Polarized Epithelial Cells.

The interface between the intestinal epithelium and the lumen consists of the apical surface of enterocytes, which is characterized by the presence of microvilli that form the brush border. The microvilli contain bundles of filaments of polymerized actin cross-linked by plastin and villin and laterally bound to the membrane by a calmodulin–110-kDa protein complex (137, 159, 160). These structural features are absent from nondifferentiated cells; therefore, a more faithful reconstruction of the epithelial barrier can be achieved with polarized epithelial cells, which can form impermeable monolayers with fully developed tight junctions, leading to measurable transepithelial electrical resistance. A number of cell lines capable of differentiating into a polarized monolayer, including Madin-Darby canine kidney (MDCK), human Caco-2, and T84 colon carcinoma cell lines, have been used in studies of Shigella and Salmonella spp. (57, 161). These monolayers have also been used in model systems aimed at reconstructing some components of the intestinal barrier (149, 172, 173).

Sereny Test.

The Sereny test has been used to assess the capacity of shigellae to elicit an acute keratoconjunctivitis in guinea pigs (207). Although its relevance may be questioned since corneal and conjunctival cells are not involved in the natural disease, this test remains the "gold standard" for assessment of Shigella invasiveness.

Ligated Loops.

Ligated intestinal loops have been used to study the interaction of both Salmonella and Shigella spp. with the intestinal epithelium (83, 92, 172, 176, 181). This model has allowed the specific evaluation of fluid accumulation due to toxin production, the analysis of the tissue destruction and inflammatory responses that follow infection, and the specific interaction of these organisms with Peyer’s patches (111, 123). Organ culture systems have also been used as a model of Salmonella gastroenteritis (9).

Animal Models of Infection.

A variety of animal models have been used in the study of Salmonella and Shigella infections. The Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium) mouse model of infection is particularly useful in the study of invasive, typhoid-like salmonellosis, but it is less adequate for the study of Salmonella gastroenteritis (30, 103). Better animal models for Salmonella gastroenteritis include rats (176), pigs (181), or rhesus monkeys (119). There is no good animal model for Salmonella typhi infections, although a mouse model that requires the administration of a high number of organisms in the presence of hog gastrin mucin has been used. Rhesus monkeys have long been used for the study of shigellosis (214). When infected orally or intragastrically, these animals develop bacillary dysentery within 3 days (191). Characteristics of the colonic and rectal invasive process can be evaluated by endoscopy and confirmed by histopathological studies of intestinal biopsies. More recently, a mouse model of Shigella infection has also been developed (140).

MOLECULAR AND CELLULAR BASES OF SHIGELLA PATHOGENESIS

Introduction

Shigellosis, or bacillary dysentery, is endemic throughout the world. As poor sanitation and low hygiene standards account for this enteric disease, it is particularly prevalent in developing areas, where children represent the major target. The prevalent species in these areas is Shigella flexneri, while Shigella sonnei, Shigella boydii, and the closely related enteroinvasive Escherichia coli are less frequently isolated. S. sonnei is by far the most prevalent species found in Western countries, although rarely, Shigella dysenteriae 1 (Shiga bacillus) may cause deadly epidemics. The clinical symptoms of shigellosis may range from mild diarrhea to severe dysentery with permanent passage of blood and mucoid stools. Although usually self-limiting, shigellosis may be life threatening in infants and children, causing severe sepsis, intestinal occlusion and/or perforation, dehydration, and subsequent chronic malnutrition. Among the several million cases that occur each year, approximately 600,000 are fatal. Shigellosis is extremely contagious, since as few as 10 microorganisms may cause clinical disease and 500 microorganisms routinely cause disease in 50% of healthy volunteers (46). The pathogenesis of shigellosis is a very complex process in which the invasion of the human colonic mucosa is an essential step (125). The small intestine is not generally affected except at the very early stages of infection, in which transient watery diarrhea seems to reflect active jejunal secretion (184). The pathogenesis of this process has not yet been elucidated, although production of an enterotoxin may play an important role. Invasion of the colonic mucosa remains the major issue in the pathogenesis of shigellosis, and therefore this chapter will concentrate mainly on the molecular and cellular bases of this process. The key features of shigellosis are the interplay between invasion of colonic epithelial and other cells of the intestinal barrier (i.e., resident macrophages) (214, 229) and the induction of an acute inflammatory response (142). The resulting lesions are characterized by an acute inflammatory rectocolitis leading to the desquamation of the epithelium as well as to purulent necrosis of the intestinal tissues. The various aspects of the pathogenesis of shigellosis have been considered in several reviews (94, 146, 189, 190).

Overview of Earlier Work

Pioneering studies undertaken in the 1960s exploited the close genetic relatedness that exists between E. coli and Shigella spp. to provide the first genetic analysis of Shigella virulence. Chromosomal sequences were transferred by conjugation from E. coli K-12 Hfr donors to S. flexneri 2a recipients. A transconjugant Shigella strain that inherited the xyl-rha-arg region of the E. coli chromosome (i.e., min 80 to 90) was positive in the Sereny test but did not cause disease when fed to rhesus monkeys (53, 64). The actual marker that affected virulence was not identified. Similar experiments showed that a second chromosomal locus located between lac and gal was essential for virulence (62). A third chromosomal region necessary for virulence was identified by screening His⁺ hybrids following conjugative mating between an E. coli K-12 Hfr donor and a Shigella recipient. Transconjugants that lost expression of group- and type-specific S. flexneri somatic antigens could neither cause a positive Sereny test nor invade the intestinal mucosa of experimental animals (63). Thus, genes encoding the biosynthesis of the O antigen in shigellae were mapped close to the his locus (min 44) and correspond to the rfb locus found in E. coli and S. typhimurium. An additional locus encoding the type specificity of the O antigen was found at min 7 in the S. flexneri chromosome (61). S. flexneri Hfr donors were also used to achieve conjugative transfer of chromosomal sequences into E. coli K-12. These experiments confirmed similar gene orders in the E. coli and Shigella chromosomal maps. However, all attempts to confer Shigella virulence properties to E. coli K-12 were unsuccessful (63, 65, 206). The subsequent demonstration that Shigella invasion genes were located on an extrachromosomal element explained this failure (195). These early studies have been essential for the present understanding of the genetic bases of Shigella pathogenesis.

The irreversible generation of noninvasive variants of Shigella strains suggested that extrachromosomal genes were responsible for the invasive phenotype. Initial work in S. sonnei showed that a 180-kb plasmid was present in smooth invasive form I strains but was absent from rough noninvasive form II strains that occur at high frequency in culture. Transfer of the 180-kb plasmid to the latter strains restored both invasiveness and form I antigen expression (124, 192, 197). A 220-kb plasmid was subsequently shown to confer invasive properties to S. flexneri (196). This observation was then extended to other species of Shigella as well as to enteroinvasive E. coli. These plasmids were subsequently shown to be functionally interchangeable between species and serotypes, indicating that they are probably derived from a common ancestor (99, 193, 194). Replicon typing studies demonstrated that all virulence plasmids tested belonged to the incompatibility group F II (208). The systematic transfer into E. coli of the S. flexneri virulence plasmid as well as chromosomal markers finally demonstrated a major role of the virulence plasmid in controlling invasion of cells and tissues (195).

Molecular Genetic Bases of the Invasive Phenotype

After a step of weak adhesion, shigellae are internalized by an actin microfilament-dependent process initially described as bacterium-directed phagocytosis (95, 96). Like Salmonella entry, Shigella entry is accompanied by major cytoskeletal rearrangements, in a process resembling membrane ruffling (35). A few minutes after entry, the Shigella bacterium lyses its phagocytic vacuole and gains access to the cell cytoplasm, where it grows at a rate of about 40 min per generation. A plasmid-mediated contact hemolytic activity expressed only by invasive Shigella isolates likely accounts for this lysis (199).

Basic Components.

Initial studies with S. flexneri and S. sonnei minicell-producing mutants identified 15 to 20 proteins encoded by the 220-kb virulence plasmid (98). Among these polypeptides, four were found to be the predominant antigens recognized by the sera of convalescent patients or monkeys and therefore were named IpaA (70 kDa), IpaB (62 kDa), IpaC (43 kDa), and IpaD (38 kDa), for invasion plasmid antigens (27, 97, 167). The entry genes were further characterized by cosmid cloning and transposon mutagenesis. A library of cosmids containing large fragments of the pWR100 plasmid from an S. flexneri serotype 5 isolate was introduced into a plasmidless derivative of the same strain. Recombinants were then screened for the capacity to enter HeLa cells. A common region of 37 kb that directed the expression of the IpaA, IpaB, IpaC, and IpaD proteins was identified (143). The invasion region of S. sonnei was subsequently identified by a similar strategy (117). An alternative approach relied on the isolation of transposon Tn5 insertions in pMYSH6000, the virulence plasmid of S. flexneri 2a (203). These experiments led to the identification of a 30-kb region of this plasmid which was required for invasion (201). Efforts from several laboratories have now generated the complete nucleotide sequence of the invasion region. This segment encodes 33 genes organized in two regions that are transcribed in opposite orientation (Fig. 1) (1, 4, 5, 6, 7, 11, 13, 27, 28, 200, 202, 223, 224, 226). One region, shown transcribed from right to left in Fig. 1, comprises seven genes that encode the following proteins: IcsB (57 kDa), IpgB (21 kDa), IpgC (17 kDa), IpaB (62 kDa), IpaC (43 kDa), IpaD (38 kDa), and IpaA (70 kDa). The transcriptional organization of this locus shows that in addition to a promoter located upstream of icsB, another promoter preceding ipgB is necessary for full expression of the ipa genes. Furthermore, a weaker promoter is present upstream of ipaD (5, 14, 200, 224). Nonpolar mutagenesis of the ipa genes defined IpaB, IpaC, and IpaD as effectors of S. flexneri entry into epithelial cells (102, 151, 152). Low levels of these three proteins are secreted into the supernatant of bacterial cultures despite their lack of a typical signal peptide. Upon contact with epithelial cells, their secretion is induced and a complex that contains at least IpaB and IpaC is formed, which most probably interacts with the eukaryotic cell surface to induce the major cytoskeletal rearrangements that lead to entry (153, 154). This hypothesis is consistent with recent reports indicating that the Yersinia YopE protein is released upon contact with epithelial cells (182) and that S. typhimurium expresses surface appendages (i.e., invasomes) upon contact with epithelial cells in vitro (see below) (86). Although no specific function has so far been attributed to the respective IpaB, IpaC, and IpaD proteins, sequence analysis suggests that IpaB is a pore-forming protein (93). Whether pore formation per se is sufficient to induce entry via formation of an ion channel or reflects IpaB capacity to achieve internalization of another bacterial protein (Ipa?) is still unknown but under investigation. Unlike salmonellae, shigellae do not cause calcium fluxes upon contact with epithelial cells (33, 170). In addition, IpaB has been shown to be involved in lysis of the phagocytic vacuole, thus accounting for escape of shigellae into the host cell cytoplasm (102, 199, 234). No direct role in the entry process has so far been demonstrated for IpaA or IpgB. In contrast, mutations in ipgC rendered shigellae deficient for entry into epithelial cells. It has recently been shown that IpgC, a 17-kDa cytoplasmic protein, acts as a chaperone which keeps IpaB and IpaC from forming a molecular complex that would otherwise be rapidly degraded by proteases inside the bacterial cytoplasm (154).

Fig. 1Invasion regions of Shigella (A) and Salmonella (B) spp.

Source:

The mxi and spa Secretion Genes.

Secretion of IpaB and IpaC into the culture medium was first demonstrated by Andrews et al. (10). Subsequent analysis indicated that at least 10 proteins, including the Ipas, are present in the cultured supernatant of wild-type Shigella strains (6, 153). The secretion of these proteins is dependent on the mxi (membrane expression of invasion plasmid antigens) and spa (surface presentation of Ipa antigens) loci (Fig. 1), which consequently are required for bacterial invasion (6, 7, 10, 202, 226). The mxi-spa region contains 24 genes organized in two operons, one comprising ipgD through spa15 and the other comprising the remaining genes starting from spa47 (Fig. 1) (4, 10, 202, 216). Sequence homology as well as preliminary functional analysis has given some indication as to the localization of some of the Spa and Mxi proteins. IpgF, MxiD, MxiJ, and MxiM have typical signal sequences indicating that they must be located beyond the cytoplasmic membrane. In fact, MxiJ and MxiM are lipoproteins and therefore are likely to be anchored in the outer membrane by their N-terminal lipid moieties (6). MxiJ is homologous to YscJ, a secretion factor of the Yersinia Yop proteins. The C-terminal portion of MxiD shows similarity to the C-terminal domain of PulD from Klebsiella oxytoca (39). This sequence may be necessary for the targeting of this protein to the outer membrane (7). Sequence analysis predicts that MxiA may contain an N-terminal domain composed of six transmembrane segments presumably anchored in the inner membrane and a C-terminal domain located in the cytoplasm (10, 11). Similarly, the presence of internal hydrophobic segments in Spa9, Spa15, Spa24, Spa29, and Spa40 suggests that these proteins may also be located in the inner membrane (202, 226). Spa47 belongs to a family of proteins that have homology to the F₀F₁ ATPases (44, 226). The Salmonella homolog of this protein, InvC, has been shown to have ATPase activity (49) (see below), suggesting that Spa47 may act as the energizer of the secretory apparatus. It therefore appears that shigellae make use of a complex apparatus in order to secrete Ipa proteins to trigger entry into cells. Such an apparatus probably represents more than a simple channel that allows direct passage of invasins from the cytoplasmic compartment to the extracellular medium. It must also be able to recognize a component of the cell surface and respond to its presence by an energy-requiring process that activates a channel which ultimately allows the highly controlled translocation of these proteins. It is becoming increasingly clear that this type of translocation system is present in several animal pathogens, including Yersinia spp. (155), Salmonella spp.(see next section), and enteropathogenic E. coli (109), as well as in a number of plant pathogens such as Erwinia carotovora (163), Xanthomonas campestris (54), and Pseudomonas solanacearum (108) (Table 1). Such widespread distribution has led to its consideration as an independent protein secretion, or type III, system. Interestingly, some of the components of this apparatus are also part of the flagellar assembly machinery of enterobacteria and Bacillus subtilis (2, 20, 21, 29, 228).

Table 1Protein homologs in type III secretion systems in various mammalian and plant pathogens

Regulation of the Entry Phenotype

The most characteristic environmental cue that affects Shigella entry into epithelial cells is growth temperature. The invasive phenotype requires growth at 37°C (144). This temperature-dependent expression of the invasion phenotype is under the control of a cascade of regulatory events that involve at least one chromosomal and two plasmid genes (Fig. 2) (145). At the top of the regulatory cascade is the chromosomal gene virR, which acts as a negative regulator (147). Mutations in this gene render shigellae able to enter into cultured epithelial cells at the nonpermissive temperature. virR maps at min 27 in the Shigella chromosome (105) and is allelic to a number of regulatory loci in enterobacteria, such as osmZ, drdX, bglY, and pilG (43, 89, 118, 148). These allelic genes encode H-NS, an abundant histone-like protein which binds DNA with high affinity, inducing its compaction and increasing its thermal stability (52, 69, 107, 174, 210). There is not yet a clear molecular model to explain the role of H-NS in regulating invasion gene expression in S. flexneri. It is likely that this protein induces local changes in DNA supercoiling which in turn may negatively modulate the expression of certain promoters (107). One of the targets for the negative regulation by H-NS has been recently identified (218). Such a target is the promoter of virB, which encodes a positive transcriptional activator of the ipa and mxi-spa operons (1, 28) (Fig. 1). VirB is a 36-kDa polypeptide which belongs to a family of DNA binding proteins involved in plasmid partitioning such as ParB of phage P1 and SopB of plasmid F. Transposon insertions in virB (143, 201) and in its equivalent in S. sonnei, invE (230), led to a noninvasive phenotype because of the lack of expression of the ipa, mxi, and spa genes. The precise binding site of VirB on these promoter sequences has not yet been characterized.

Fig. 2Scheme of the regulation of expression of the Shigella invasion phenotype.

Source:

Expression of the virB gene, on the other hand, is positively regulated by virF, which encodes a protein belonging to the AraC family of transcription activators. This gene is located about 40 kb apart from the entry region on the virulence plasmid of S. flexneri 2a (186, 187). Mutations in virF abolished expression of virB, thereby resulting in the absence of expression of other invasion genes (1, 188, 216). Activation of virB by VirF requires a DNA sequence located 110 bp upstream from the virB transcription start site (218). Unlike virB, virF is expressed at 30°C, although overexpression of virF at 30°C does not derepress the transcription of virB. This is most likely due to binding of H-NS to the virB promoter at 30°C, thus preventing its activation by VirF (218).

The expression of Shigella invasion genes is also influenced by osmolarity. Growth of shigellae under osmotic conditions equivalent to those of extracellular body fluids increased the expression of an mxi::lacZ gene fusion. Mutations in the OmpB osmoregulatory locus (ompR-envZ) significantly decreased the expression of this reporter gene and severely diminished the invasion phenotype, as measured in tissue culture and Sereny assays (16). In addition, mutations in ompC, another target of the OmpB regulatory locus, rendered shigellae defective in the tissue culture plaque assay and in the Sereny test (18). The exact function of this protein in the invasive process remains to be established. Thus, the Shigella invasive process appears to be finely tuned at the transcriptional level by both chromosomal and plasmid genes (Fig. 2).

Movement of S. flexneri within and between Epithelial Cells

Movement of S. flexneri within Epithelial Cells.

Subsequent to entry into cells, S. flexneri gains access to the cytosol by lysing the phagocytic vacuole. This step is closely correlated with the production of the IpaB, IpaC, and IpaD proteins (102, 199). Reaching the cell cytoplasm facilitates the growth of these organisms and opens the way to their intracellular movement, a spectacular phenotype observed for the first time by Ogawa et al. in 1968 (169). The rapid movement of S. flexneri inside cells was unexpected, since this pathogen does not express any flagellar structure. Makino et al. later demonstrated that virG (icsA), a plasmid gene distinct from the entry locus, was necessary for permanent reinfection of adjacent cells (139). Bacterial movement requires participation of the host cell cytoskeleton, since treatment of infected cells with cytochalasin D, which prevents actin polymerization, blocks both intracellular movement and cell-to-cell spread (171). Immediately after escape into the cytoplasm, short filaments of actin accumulate at one extremity of the bacterium, as demonstrated by labeling of infected cells with NBD-phalloidin (17). The actin filaments become bundled, possibly by the actin-binding protein plastin (177), and form an actin-containing tail several micrometers in length that trails the bacterial body as it moves through the cytoplasm. IcsA (VirG) is found at the distal pole of the dividing bacteria, with maximum expression on the mother cell immediately before cell division is completed. Although the mechanism of polarization of this protein on the bacterial surface is not yet known, this phenomenon certainly accounts for the uneven nucleation of actin which ultimately leads to bacterial movement. Subsequent to its export to the bacterial surface via a spa- or mxi-independent mechanism, IcsA is cleaved at a consensus site (SSRRASS) between amino acids 754 and 760. Cleavage releases a 95-kDa species to the extracellular milieu (88). The cleavage site is phosphorylated in vitro in the presence of cyclic AMP-dependent kinases; therefore, a potential role for intracellular phosphorylation in controlling IcsA function has been suggested (41). Removal of the SSRRASS consensus sequence impairs cleavage of the protein and increases the intracellular motility of the pathogen, most likely as a result of the unrestricted accumulation of IcsA on the bacterial surface. Introduction of the ompT gene encoding an E. coli surface protease decreases the amount of cell-associated IcsA, since most of the protein is converted to the 95-kDa secreted species (165). A temperate phage carrying ompT, which is located at min 12 on the chromosome of many E. coli strains, is absent from Shigella spp. This explains the Kcp phenotype and therefore the previously mapped Shigella kcpA virulence locus actually corresponds to the lack of the ompT gene (62, 165). Another plasmid gene, virK, also affects the amount of cell-associated IcsA, although the precise function of this gene has not been characterized (164). VirG (IcsA) is therefore the major player in the intra- and intercellular movement, exerting its function by inducing actin polymerization in a spatially and temporally controlled manner.

Movement of S. flexneri between Epithelial Cells.

The actual significance of Shigella intercellular spread in vivo is not clear, although it most likely allows the bacterium to gain access to neighboring cells without an extracellular phase. Alternatively, this phenotype may simply be a manifestation of the host cell response aimed at eliminating the invading bacteria. However, the observation that Shigella icsA mutants are severely impaired in tissue colonization suggests that intercellular spread is important in this step of the pathogenic cycle (139, 189).

A large number of the invading bacteria appear to associate with the actin cables anchored to the structure of intermediate junctions (221). These bacteria seem to move along the actin cables according to a process described in chicken embryo fibroblasts as an Olm (organelle-like movement) phenotype (222). The genetic and molecular bases of this alternative motility process have not yet been identified. Bacteria present in this particular area of the cell use the cell adhesion molecules cadherins. In a mouse fibroblast cell line that does not produce cadherins (S180 cells), shigellae are unable to pass from cell to cell. After transfection with a cDNA encoding L-CAM, these cells become epithelialized, form junction-like structures, and acquire the capacity to transport shigellae from cell to cell (198). This finding indicates that the bacterium, pushed forward by the IcsA-mediated actin polymerization process, is able to engage the inner face of the cytoplasmic membrane in the area of the intermediate junctions and to mobilize the major components of this apparatus in order to form an efficient protrusion which is ultimately endocytosed by the next cell. Bacteria within the tip of the protrusion subsequently lyse the two membranes in which they are entrapped (i.e., the membrane of the protrusion and the membrane of the newly infected cell). Inactivation of the icsB gene, which is located in the invasion locus of the virulence plasmid (Fig. 1), prevents shigellae from lysing the two membranes, although it does not affect the ability of these organisms to gain initial access to and move through the cytoplasm and to form apparently normal protrusions. Therefore, icsB mutants remain trapped inside growing vacuoles surrounded by two membranes (5). The biochemical function of IcsB, a 56-kDa protein, has not yet been characterized. Mutations in the Listeria monocytogenes plcB gene, which encodes a lecithinase, confer a phenotype very similar to that conferred by Shigella icsB (220). However, no sequence similarity between these two gene products has been detected, and IcsB has nondemonstrable lecithinase activity.

S. flexneri and Macrophages

Macrophages infected with S. flexneri rapidly die (34) as a result of the bacterium-induced activation of programmed cell death (i.e., apoptosis) (235). If macrophages are preincubated with lipopolysaccharide (LPS), their apoptotic death is preceded by the release of a substantial amount of mature interleukin-1β. Shigella ipaB mutants are unable to induce macrophage killing, indicating that IpaB function is required for this phenotype (234). How this phenomenon fits in the scheme of shigellosis will be considered below.

Orphan Plasmid Genes of S. flexneri

A number of plasmid genes of unknown function have been identified. Some of these genes (ipgB, ipgD, ipgF, and spa15) are located in the invasion region of the virulence plasmid. However, their inactivation does not seem to affect the entry phenotype (4, 202; R. Ménard et al., unpublished data). These provisionally cryptic genes may turn out to encode invasive or regulatory functions that are too subtle to be detected by the current virulence assays. Another gene of unknown function is ipaH, which was identified by screening a λgt11 recombinant library with a serum raised against a preparation enriched for proteins encoded by the virulence plasmid (27). Multiple copies of this gene have been found not only on the virulence plasmid but also on the chromosome, although it is not known whether all of these copies are expressed (100). Two adjacent copies, ipaH _7.8 and ipaH _4.5, are located between the invasion region and the virG (icsA) locus. The deduced amino acid sequences of these two ipaH genes revealed the presence of repeated LPX motifs (225) reminiscent of the sequence of YopM, a virulence protein from Yersinia pestis which is known to inhibit thrombin-induced platelet activation (130, 131). No equivalent phenotype has yet been demonstrated in shigellae.

Chromosomal Genes Involved in Shigella Virulence

In addition to the already described virR, a number of chromosomal virulence genes that are involved in different aspects of Shigella pathogenesis have been characterized.

Invasion Genes.

An extensive mutagenesis analysis identified another chromosomal locus, vacB, involved in bacterial invasion. Mutations in this gene, which is located immediately downstream of purA, resulted in a 10-fold decrease in Shigella entry into cells and the inability of the bacteria to spread from cell to cell (217). These phenotypes appear to be correlated with a decreased level of the IpaB and IcsA (VirG) proteins in this mutant. Although the precise function of VacB is unknown, it may act at the posttranslational level on the stability of these proteins.

Aerobactin.

S. flexneri makes use of the hydroxamate siderophore aerobactin and its 76-kDa outer membrane receptor for scavenging and transporting Fe³⁺. These proteins are encoded by the iucABCD and iutA genes, respectively, which are located at min 83 on the chromosome (40, 90, 127). Aerobactin synthesis and transport genes have been detected in all Shigella species except S. dysenteriae 1 and some isolates of S. sonnei (127). The role of aerobactin in the virulence of S. flexneri was investigated by a variety of assays. Aerobactin-defective mutants were not affected in their capacity to invade and kill host cells, although they were defective in the Sereny test and in the rabbit ligated loop assay, suggesting that aerobactin production is important for bacterial growth within tissues (126, 166).

Superoxide Dismutase and Catalase.

S. flexneri sodB mutants were extremely sensitive to phagocytes in vitro. In addition, they were negative in the Sereny test and in the rabbit ligated loop infection assay. These results indicate that the superoxide dismutase plays a major role in Shigella virulence. In contrast, kat mutants were not significantly affected in these assays, suggesting that catalases may not be essential for virulence (68).

Somatic Antigens.

Details on the genetic and biochemical aspects of LPS biosynthesis in the various Shigella species will not be covered in this chapter; the reader is referred to recent comprehensive reviews of this topic (24, 205). The S. flexneri loci necessary for the synthesis of LPS are the chromosomal equivalent of the rfb and rfa loci of E. coli. They are located near the his locus, whereas the genes encoding the type-specific antigen are located near the pro (min 6) and the lac (min 8) loci (61). Temperate phages play a crucial role in determining this serotype specificity. They encode proteins required for the glycosylation or acetylation of the O polysaccharide which give rise to immunodominant epitopes determining the serotype (209). Plasmid genes are required for LPS biosynthesis in S. sonnei (124) and in S. dysenteriae 1 (231, 232). The large virulence plasmid of S. sonnei carries the genes required for biosynthesis of the form 1 antigen, which is the unique somatic antigen of this species. In S. dysenteriae 1, a 9-kb plasmid encodes one or more functions for O-polysaccharide biosynthesis. Currently available data suggest that the rfb locus encodes a function essential for the somatic antigen biosynthesis, which is a rate-limiting step in O-antigen biosynthesis.

Shiga Toxin.

S. dysenteriae 1 is the only Shigella species that produces Shiga toxin. The mechanisms of action of this toxin are described in detail in chapter 153. The role of Shiga toxin in the pathogenesis of shigellosis was investigated in the Sereny tests and in macaque monkey experimental infections. S. dysenteriae 1 strains carrying a mutation in the stx locus did not show significant difference in the Sereny test, and their capacity to invade epithelial cells remained unaltered (60). When administered intragastrically to macaque monkeys, the toxin-deficient strain was as invasive as the wild type, although it caused significantly less vascular damage in infected colonic tissues. These experiments suggested that Shiga toxin may account for the increased severity of infections caused by S. dysenteriae 1 compared with other Shigella serotypes and species that do not produce this toxin.

An Integrated Model of Shigella Pathogenesis

On the basis of available data, the following model for the pathogenesis of shigellosis, encompassing the various Shigella-host interactions, can be presented (Fig. 3). How do shigellae gain access to the basolateral pole of the enterocytes to invade the intestinal epithelium? Current evidence indicates that the M cells, which constitute part of the dome of the lymphoid follicles associated with the intestinal mucosa, represent the major route of access of shigellae to the mucosa (172, 191). The initial target therefore is probably the lymphoid structures associated with the mucosa. Resident macrophages present in these structures likely constitute the first line of defense. Shigella-induced apoptotic killing of infected macrophages may have two important consequences. One is the release of invading bacteria in subepithelial tissues, thus allowing direct basolateral invasion. The other consequence is more complex and involves the release of interleukin-1 from apoptotic macrophages. Release of this cytokine, which may be considered an early warning signal emitted by the infected host, induces a very early inflammatory response. This inflammatory response likely provokes destabilization of the epithelial architecture, thus facilitating further entry of luminal bacteria. Two pieces of experimental data support this model. Apical infection of T84 human colonic epithelial cells grown on filters is not followed by Shigella invasion of the monolayer. However, if human polymorphonuclear cells are added to the basal side of the epithelial layer, bacterium-induced migration of polymorphonuclear cells takes place, which in turn opens a paracellular pathway for shigellae to gain access to the epithelial monolayer (173). Shigella infection of rabbit ligated loops from animals previously treated with an anti-CD18 monoclonal does not lead to significant mucosal invasion and tissue destruction (172). These results indicate that if migration of circulating phagocytes is blocked by antagonizing the adhesion of the Mac1 receptor either to vascular cells or to epithelial cells, no evidence of experimental shigellosis is observed. These experiments suggest that shigellosis may be primarily an acute inflammatory bowel disease subsequently amplified by the development of epithelial invasion by the pathogen.

Fig. 3Schematic model of the Shigella interactions with the intestinal epithelium.

Source:

MOLECULAR AND CELLULAR BASES OF SALMONELLA INTERACTION WITH NONPHAGOCYTIC HOST CELLS

Introduction

Salmonella species can cause a variety of diseases ranging from localized gastroenteritis to more systemic illnesses such as typhoid fever (104). Although certain Salmonella serovars are highly adapted to a specific host (e.g., S. typhi and S. gallinarum are adapted to humans and poultry, respectively) or preferentially infect a particular species (e.g., S. choleraesuis and S. dublin usually infect swine and bovine, respectively), there are a number of Salmonella serotypes (e.g., S. enteritidis) that can infect a wide range of hosts. The type of illness that these organisms may cause depends not only on the serotype of the infecting Salmonella organism but also on the species and/or immunological status of the infected host. Most infections result from the ingestion of contaminated food or water (31, 32, 132). The ingested microorganisms proceed to the intestinal tract and breach the intestinal wall to reach the lamina propria. At this site, salmonellae may replicate and establish a local infection, or they may be ingested by macrophages, which may disseminate these organisms to deeper tissues such as regional lymph nodes, liver, and spleen to establish a systemic infection (104). Although salmonellae are largely considered facultative intracellular pathogens, an alternative view questioning this notion has been proposed (106).

Entry Site

The primary site at which salmonellae breach the intestinal epithelium has not been precisely defined and it is a matter of some controversy. However, it is clear that the site of entry is dependent on the serotype of the infecting Salmonella organism as well as on the species of the infected host (25, 30, 103, 150, 176, 181). In humans, the distal part of the small bowel has been most frequently suggested as the primary site of infection for nontyphoidal salmonellae and although this is often cited in textbooks, there is surprisingly little direct evidence in the literature to support this hypothesis (3). In other animals, the involvement of other segments of the intestinal track have been frequently observed (25, 119, 180, 181). Some controversy has also surrounded the potential role of the M cells of the intestinal epithelium in Salmonella invasion through the intestinal barrier. Experiments with mouse ligated intestinal loops showed that shortly after infection, S. typhimurium exclusively associated with M cells, leading to their destruction and the subsequent spreading of the infecting microorganisms to adjacent cells (111). The correlation of these finding with natural infections is less clear, since in such a short-term experimental setting, the lack of glycocalyx on the surface of M cells may make these cells more readily accessible to the infecting microorganisms. However, noninvasive strains of S. typhimurium could not be found associated with M cells within comparable infection times, arguing in favor of a more specific role for M cells in the Salmonella mouse infection process (111). It is clear that unlike shigellae, salmonellae can breach the intestinal barrier through the cells of the columnar epithelium of the small and large intestines (56, 57, 213). It has been argued that the latter type of cells may well constitute the main port of entry for nontyphoidal salmonellae since they vastly outnumber M cells (149, 219). The histopathological findings of natural infections are consistent with this notion (38, 150, 183, 215).

Interaction with the Intestinal Epithelium

Pioneering work by Takeuchi and colleagues, using electron microscopy and a guinea pig model of infection, gave the first morphological details of the interaction of salmonellae with the intestinal epithelium (213, 214). They showed that shortly after coming into close contact with the host cell, salmonellae induced profound changes in the brush border, characterized by the denaturation of microvilli at the point of contact between the bacteria and the host cell membrane. It is now known that the changes observed by Takeuchi et al. ressemble the membrane ruffles induced by a variety of stimuli such as growth factors, mitogens, or oncogene stimulation (12, 115). The alterations induced by salmonellae on the brush border of infected guinea pigs were transient, since after bacterial entry, the normal architecture of the microvilli was regained and internalized organisms were observed in membrane-bound compartments. More recent studies have shown that the interaction of S. typhi and S. typhimurium with M cells also leads to membrane ruffling (111, 123, 175). However, the changes in the architecture of the M cells are not transient, since they ultimately lead to destruction of the cells (111, 123). These findings suggest that the interaction of salmonellae with the intestinal epithelium may not be the same in the different hosts, which in part may explain differences in the clinical manifestations of Salmonella infections.

One of the hallmarks of the pathology of nontyphoidal Salmonella infections is the massive infiltration of neutrophils in both the large and small intestines (104, 119, 150, 215). Although the mechanisms by which salmonellae induce this pathology are not completely understood, recent studies indicate that the interaction of salmonellae with intestinal epithelial cells triggers a cascade of events that results in the transepithelial migration of neutrophils (38, 149). The precise details of this response are poorly understood, although it appears that bacterial protein synthesis subsequent to contact with epithelial cells is required to induce this phenomenon. Contact of salmonellae with cultured epithelial cells induces the production of the proinflammatory cytokine interleukin-8 (48, 149). However, this cytokine does not appear to be directly responsible for neutrophil transepithelial migration; rather, it may play an important role in attracting these cells to the site of infection (149).

There are conflicting results regarding the ability of salmonellae to alter the integrity of polarized epithelial cell monolayers (57, 149). However, if salmonellae are capable of depolarizing the intestinal epithelium, this ability may facilitate the penetration of the intestinal barrier perhaps by opening a paracellular entry pathway.

Although the ability to gain access to the cells of the intestinal epithelium is an essential feature of all pathogenic salmonellae, access to other, less well characterized intracellular compartments may be also important, particularly at later stages of the pathogenic cycle (37, 45, 133, 227). For example, S. typhimurium can enter and survive within human B and T cells, which could play a role in the dissemination of infection (227). Furthermore, salmonellae can also gain access to nonphagocytic cells of the liver and spleen, which may constitute a "safe site" for replication during the early phases of systemic infection (45, 133).

The Intracellular Environment

Unlike shigellae, internalized salmonellae remain in a membrane-bound compartment throughout the intracellular life cycle. After a lag period of approximately 4 h, salmonellae can replicate in nonphagocytic cells, a process that appears to be required for virulence, since mutants defective for intracellular replication were avirulent in a mouse model of infection (70). The characteristics of this membrane-bound compartment have been recently examined (79, 80, 81). S. typhimurium triggered the capping of several cytoskeleton-associated proteins such as the fibronectin receptor, CD44, β2-microglobulin, and class I major histocompatibility complex heavy chain, although only the latter two were seen associated with the bacterium-containing vacuoles. These results indicate that sorting mechanisms at the cell surface preclude the internalization of some cytoskeletal associated proteins. At later times, the composition of the Salmonella-containing vacuole changes, the amounts of class I major histocompatibility complex and β2-microglobulin decrease, and lysosomal markers such as lysosomal glycoprotein and lysosomal acid phosphatase begin to appear, although the mannose-6-phosphate receptor is never observed in these vacuoles (80). These observations suggest that salmonellae bypass the late endosomal pathway, proceeding directly to a lysosomal compartment which remains isolated from other endocytic compartments, since fluid-phase markers seldom colocalized with Salmonella-containing vacuoles. Four to six hours after internalization, intracellular bacteria induce the formation of tubular structures containing lysosomal markers that are connected with the Salmonella-containing vacuole (81). These structures may facilitate intracellular replication, since Salmonella mutants defective for intracellular growth were also defective in triggering the formation of the tubular structures. The intracellular environment of this specialized compartment has been indirectly characterized by using Salmonella reporter genes which are stimulated by specific signals (79). These studies suggest that the Salmonella vacuole contains lysine and oxygen and low concentrations of Mg²⁺ and Fe²⁺ and has a mildly acidic pH.

Signal Transduction Pathways Triggered by S. typhimurium in Cultured Mammalian Cells

A number of studies have clearly demonstrated that the host cell plays an active role in Salmonella interactions with host cells. For example, addition of cytochalasins B and D, which prevent actin microfilament polymerization, effectively prevents bacterial entry, indicating that an intact host cell cytoskeleton is necessary for the internalization process (26, 122). In addition, the production of interleukin-8 and an unidentified transcellular chemotactic factor(s) by the host cell is required for the Salmonella-induced neutrophil transmigration across the intestinal epithelium (47, 149). These host cell responses are most likely the result of signal transduction pathways triggered by salmonellae at the cell surface.

Salmonella entry is accompanied by an increase in the levels of intracellular free calcium (170) and a profound rearrangement of the host cell cytoskeleton, with the accumulation of a number of cytoskeletal proteins such as actin, α-actinin, talin, and ezrin around the incoming bacteria (58). These events lead to membrane ruffling, micropinocytosis, and subsequent internalization of the bacteria (56, 78, 213). The ability to induce membrane ruffling appears to be an intrinsic part of the Salmonella entry mechanisms, since mutants that fail to induce these changes are unable to gain access to cultured cells (76, 87). More recent studies have begun to provide some insights into the host cell signal transduction pathways that lead to membrane ruffling and bacterial uptake (77, 113, 170, 185; L.-M. Chen, J. Pace, and J. E. Galán, unpublished data). Some of these pathways are also likely responsible for the production of proinflammatory cytokines that follows the Salmonella interaction with a number of host cells. The understanding of these events is clearly incomplete and is complicated by the fact that salmonellae can evoke different signaling events in various cell types (reviewed in reference 71). For example, contact of S. typhimurium with Henle-407 cells results in the stimulation of the epidermal growth factor receptor (77). Such stimulation leads to a signaling cascade including the activation of the mitogen-activated protein kinase and one of its substrates, phospholipase A₂ (PLA₂) (170). PLA₂ activity generates arachidonic acid, which is ultimately converted to different leukotrienes by a variety of enzymes, including 5-lipooxygenase. Salmonella infection of Henle-407 cells leads to the production of leukotriene D₄, which directly or indirectly triggers the influx of calcium required for bacterial entry. Interruption of any of the steps leading to the production of leukotriene D₄ (i.e., blocking the activities of PLA₂ or 5-lipooxygenase) effectively prevents bacterial uptake into these cells. How calcium fluxes and arachidonic acid metabolites are involved in the stimulation of membrane ruffling is poorly understood. Salmonellae can enter into virtually any cell line, many of which do not express the epidermal growth factor receptor (66, 77, 113). Therefore, it is not surprising that salmonellae can stimulate more than one signal transduction pathway to promote entry into mammalian cells. For example, in human epithelioid HeLa cells or in mouse B82 fibroblasts, salmonellae cause an increase in the concentration of inositol phosphate as a result of the activation of phospholipase Cγ which is essential for bacterial internalization (185; Chen et al., unpublished data). Thus, inositol triphosphate-mediated Ca²⁺ release from intracellular stores may be necessary for bacterial entry into these cells. In contrast, phospholipase Cγ does not appear to play a role in Salmonella entry into Henle-407 cells.

Despite the differences in the signaling events leading to bacterial uptake in various cell lines, it is likely that common cellular effectors exist, since the ultimate outcomes of the signaling process appear to be similar in all cells (i.e., cytoskeletal rearrangements leading to membrane ruffling and bacterial internalization). Wild-type (but not Inv^–) Salmonella strains mobilize Ca²⁺ and activate mitogen-activated protein kinase in several cell lines (Chen et al., unpublished data). Host cell protein tyrosine phosphorylation is also required for Salmonella entry, since tyrosine kinase inhibitors blocked bacterial uptake into various cell lines (Chen et al., unpublished data). However, it is not understood how all of these second messangers interplay to induce the host cell cytoskeletal changes that lead to membrane ruffling or nuclear responses with subsequent production of proinflammatory cytokines. Clearly, more studies will be required to fully understand the complexity of the signal transduction pathways evoked by salmonellae that lead to bacterial uptake and cytokine production.

Genetic Bases of Salmonella Entry

The complexity of the molecular genetic bases of Salmonella entry into host cells is reflected by the large number of genetic loci required for bacterial entry. Most of the determinants of entry appear to be encoded in a discrete region of the Salmonella chromosome at centisome (cs) 63. However, a number of loci located in other regions of the Salmonella chromosome have also been implicated in bacterial entry. The functional relationship, if any, among these different loci is poorly understood.

Role of LPS, Motility, and Type I Fimbriae in Salmonella Entry.

A number of genetic loci encoding well-characterized determinants such as flagella, type I fimbriae, or the surface LPS have been implicated in Salmonella entry into cultured cells (51, 59, 112, 114, 120, 134, 162). It is not clear whether these determinants simply facilitate productive contact between the bacteria and the host cell or whether they play a more direct role in bacterial entry. Mutations in chemotaxis genes such as cheA, cheR, cheW, and cheY, which confer a smooth-swimming phenotype to salmonellae, rendered these strains more invasive than wild-type strains. In contrast, mutations in cheB, which result in a tumbly phenotype, rendered these organisms deficient in entry (112, 120, 134). Furthermore, nonmotile strains can regain wild-type levels of entry if a mild centrifugal force is applied during the internalization process (112). S. typhimurium strains lacking type I fimbriae or flagella retained virulence in a mouse model of infection, although nonflagellated, nonfimbriated double mutants exhibited significantly higher 50% lethal dose values (135, 136). These results favor a secondary role for these surface determinants in the cell entry process. However, Liu et al. reported that centrifugation did not reverse the entry deficiency of S. typhi Fla^– strains (134). This finding may reflect alternate mechanisms of entry of different Salmonella species or serovars. The role of LPS in bacterial entry also appears to depend on the Salmonella serotype, since rough strains of S. typhimurium are not affected in their ability to enter into cultured mammalian cells (121), while rough mutants of S. typhi (162) or S. choleraesuis (59) are deficient for entry. Further studies will be required to establish if this molecule is involved in evoking signaling events in the host cells that may aid the entry process.

Bacterial Entry and the cs 63Region of the Salmonella Chromosome.

A variety of genetic approaches have also identified bona fide genetic loci required for Salmonella entry. Interestingly, the majority of these loci are clustered at cs 63 of the Salmonella chromosome.

  The inv / spa loci. Characterization of an S. typhimurium strain defective for entry led to the identification of a locus, inv, which is absolutely required for bacterial internalization into cultured nonphagocytic cells (72). This locus is also required for mouse virulence, as strains carrying mutations in inv had an increased 50% lethal dose value when administered orally into BALB/c mice (72). A very extensive survey of Salmonella isolates comprising over 100 serovars indicated that the inv locus was present in all virulent isolates examined (74, 179). Introduction of an inv mutation in several Salmonella strains belonging to various serotypes rendered these strains severely deficient for entry into cultured epithelial cells, indicating the this locus is not only present but also functional (74, 87). Subsequent nucleotide sequence and functional analyses of this region have identified at least 14 genes in the following order: invH, invF, invG, invE, invA, invB, invC, invI, invJ, spaO, spaP, spaQ, spaR, spaS (8, 36, 49, 76, 87, 91, 116; C. Collazo and J. Galán, unpublished data) (Fig. 1). All of these genes except invB are required for entry of S. typhimurium into cultured epithelial cells (49). Mutations in all these genes except invH did not have any measurable effect on the ability of these organisms to attach to cultured mammalian cells, indicating that attachment and entry are genetically separate events in salmonellae (8). Sequence homology analyses of the predicted gene products indicate that this region encodes a dedicated sec-independent type III protein secretion machinery, homologous to that encoded in the mxiA and spa loci of shigellae and other animal and plant pathogens (see above and Table 1). Although the organization of the Salmonella genes and the Shigella homologs is remarkably conserved (Fig. 1), significant differences are observed in genes that encode products that are presumably not components of the protein secretion apparatus itself. For example, the invI and invJ gene products have no significant homology to the Shigella counterparts Spa13 and Spa32, although they encode proteins of similar sizes. In addition, a homolog of invH, which is located immediately upstream of invF, has not yet been dentified in shigellae. Mutations in invH rendered salmonellae defective in entry, presumably as a result of the effect of this gene on bacterial adhesion (8). Interestingly, the phenotype of the invH mutants was more pronounced in host-adapted Salmonella serovars such as S. typhi or S. gallinarum.

The similarity between the protein secretion systems in the different microorganisms goes beyond the protein sequence homology, since complementation between homologous genes from different species has been observed. For example, mutations in the S. typhimurium genes invA and spaP can be complemented by the cognate S. flexneri genes mxiA and spa24, respectively (87, 94). However, the Yersinia enterocolitica lcrD gene failed to complement a mutation in the cognate S. typhimurium gene, invA. Chimeric proteins consisting of the N-terminal half of LcrD and the C-terminal half of InvA successfully complemented an invA mutant, indicating that despite functional conservation among these export systems, they have been tailored to export specific determinants in each pathogen (85). In addition, these data also indicate that determinants of specificity are located in the C terminus of the InvA family of proteins, which includes Shigella MxiA (see above). In fact, there is little similarity among the primary sequences of the targets of these export systems so far identified in a variety of pathogenic microorganisms. These include the Yersinia outer proteins, or Yops (212), the Harpins of plant pathogenic bacteria such as Erwinia amylovora and P. solanacearum (23, 101, 233), and the invasion protein antigens of shigellae (Ipa proteins) (see above) (190). All of these secreted proteins lack typical signal sequences and are found loosely associated with the bacterial cell envelope or in the culture supernatant of the respective microorganisms.

Recently, a target of the protein secretion apparatus encoded by the inv locus was identified (36). This protein, termed InvJ, is also encoded by the inv locus and is found at low levels in the culture supernatant of invasion-competent S. typhimurium. The export of InvJ is dependent on the secretion system encoded in the inv and spa loci, since mutations in invG, invA, invC, spaO, spaP, spaQ, spaR, and spaS effectively prevented InvJ translocation to the culture supernatants (36; Collazo and Galán, unpublished data). InvJ does not show similarity to any of the targets of the related export systems encoded in Shigella and Yersinia spp. However, this protein shows homology to EaeB, a secreted protein of enteropathogenic E. coli (42). EaeB is involved in triggering the signaling pathways that lead to the formation of the attaching and effacing lesions typical of infections by these microorganisms. Interestingly, the secretion of this protein is dependent on a type III secretion system similar to that encoded in the Salmonella inv locus (109).

Salmonella entry is an energy-requiring process, since dead or inactivated bacteria are unable to enter into cultured epithelial cells (67, 121, 138). Energy is probably required to export invasion determinants through the inv- and spa-encoded type III secretion system. Most likely, the energizer of this system is InvC, since sequence analysis revealed that this protein belongs to the F₀F₁ ATPase family of proteins, members of which are found in all type III secretion systems (49). The involvement of InvC as the energizer of the secretion system is further supported by the fact that the purified protein exhibited significant ATPase activity. Furthermore, a site-directed mutant with a single amino acid substitution in the putative nucleotide binding site of this protein failed to hydrolyze ATP and also failed to complement an S. typhimurium invC mutant (49).

It is clear, then, that Salmonella entry into host cells requires a dedicated protein secretion system. This secretion apparatus is most likely involved in the delivery of invasion determinants to the host cell and/or in the surface assembly of an invasion organelle (see below).

  Contact-dependent assembly of an invasion organelle (the invasome). High-resolution low-voltage scanning electron microscopic examination of the interaction of salmonellae with cultured epithelial cells showed that contact with epithelial cells results in the transient formation of appendages (termed invasomes) on the surface of S. typhimurium (Fig. 4) (86). The appendages are approximately 60 nm in diameter and 0.3 to 1.0 μm in length and therefore are shorter and thicker than other well-characterized surface structures such as flagella, type I fimbriae, or bundle-forming pili. Appendages were observed as early as 10 min after bacterial infection, although they were never observed on the surface of S. typhimurium grown in L broth under conditions that render these organisms competent for entry. Furthermore, appendages were not observed on the surface of bacteria associated with host cell membrane ruffles, although bacteria that were associated with the host cell surface but had not initiated the internalization process always displayed the invasomes on their surfaces. It appears that the contact-dependent assembly of the invasome is transient and that these structures are later shed from the surface of salmonellae concomitant with triggering the host cell responses that lead to membrane ruffling and subsequent bacterial uptake. Formation of the surface appendages does not require de novo protein synthesis, since addition of chloramphenicol at levels that immediately block protein synthesis does not prevent the formation of the invasome. This finding is consistent with the observation that addition of chloramphenicol does not block the entry of invasion-competent salmonellae (138). Assembly of the invasome is dependent on the protein secretion apparatus encoded by the inv locus, since these structures were not present in strains of S. typhimurium carrying mutations in invG and invC. A mutation in invE prevented the shedding of these structures but did affect not their contact-dependent assembly. In this mutant, however, the appendages appeared longer than those seen on the surface of wild-type S. typhimurium. invG, invC, and invE mutants are severely defective for entry, indicating that the transient assembly of the invasome is required for bacterial internalization. Although the composition of this invasion organelle is not known, it is very likely that the targets of the inv- or spa-encoded type III secretion system are components of the invasomes. Consistent with this notion is the finding that contact with cultured Henle-407 cells dramatically increased the secretion of InvJ to the extracellular milieu (M. Zierler and J. Galán, in press). This effect required live cells, since Henle-407 cells fixed with glutaraldehyde prior to infection did not enhance the secretion of InvJ. S. typhimurium strains carrying mutations in invC and invG did not secrete InvJ even in the presence of cultured Henle-407 cells. In contrast, invE mutants did not prevent the secretion of InvJ, a finding consistent with the phenotype of this mutant under a low-voltage scanning electron microscope. These data indicate that secretion through the type III system encoded in the inv locus requires activating signals and further supports the notion that S. typhimurium entry into cultured cells is the result of a cross talk between the bacteria and the host cell. It remains to be seen whether InvJ is a component of the invasion organelle.

Fig. 4High-resolution scanning electron micrograph of S. typhimurium interacting with cultured epithelial cells depicting the transient assembly of the invasomes. (B) Shortly after infection (15 min) S. typhimurium can be seen in close association with the apical microvilli of cultured MDCK cells, and invasomes are apparent on the surface of these bacteria. (A) In contrast, wild-type organisms not exposed to cultured epithelial cells do not exhibit these surface appendages, although flagella are readily apparent. (C) At later times after infection (30 min), a bacterium is seen associated with membrane ruffles lacking the surface appendages.

Source:

  The hil, prgH, and orgA loci. Other approaches have identified additional invasion genes located in the min 59 region. The ability of salmonellae to gain access to host cells is modulated by the bacterial growth state and a variety of environmental conditions, such as oxygen tension and osmolarity, which are known to alter the levels of DNA superhelicity (see below). On the basis of these observations, Lee et al. designed a strategy to isolate mutants of S. typhimurium able to gain access to cultured epithelial cells when grown under nonpermisive conditions (129). This approach led to the identification of three classes of mutants. One class affected a number of che genes and conferred smooth-swimming behavior, which presumably increases the frequency of productive contact between the bacteria and the host cell. The second class consisted of mutations in the promoter of rho, a transcriptional termination factor. Loss of Rho alters gene expression and therefore presumably alters the expression of genes encoding invasion factors. The third class of mutations was in a locus termed hil (hyperinvasive locus), which is located at min 59 on the Salmonella chromosome. Introduction of a deletion of the hil region rendered S. typhimurium severely deficient for entry, indicating that this locus is required for bacterial internalization. Presumably, hil may encode a regulatory factor or a product which is rate limiting for invasion of organisms grown under nonpermissive conditions. The isolation of oxygen-regulated genes led to the identification of orgA, a gene closely linked to hil which is also required for entry (110). The identification of genes whose expression was regulated by the PhoP-PhoQ response regulator led to the isolation of another locus required for entry (15). This locus, prgH, is closely linked to orgA and hil and is negatively regulated by the PhoP-PhoQ two-component system, which is required for survival within macrophages. This finding established a link in the regulation of two different and yet related events: entry and intracellular survival (156).

  The cs 63 region is another example of a "pathogenicity island." The G+C content of the genes encoded in the cs 63 region of the Salmonella chromosome is very low in comparison with the overall average for Salmonella chromosomal DNA (8, 36, 49, 76, 87, 91, 116, 141). Therefore, it has been proposed that salmonellae may have acquired these genes by horizontal transmission from another organism (76, 87). The horizontal acquisition of this genetic information as a block is supported by indications that the 40-kb region of the Salmonella chromosome is not present in the closely related bacterium E. coli and therefore represents another example of a pathogenicity island (74, 158). This region of the chromosome has been reported to be unstable in certain Salmonella serotypes such as S. seftenberg and S. lichtfield, leading to the frequent occurrence of deletions. The instability of this region has been ascribed to the presence of IS3-like elements immediately adjacent to one of the borders of this pathogenicity island (8; C. Ginocchio, K. Rahn, R. C. Clarke, and J. Galán, unpublished data).

Other Entry Loci.

Although the cs 63 region of the chromosome is essential for Salmonella entry, a variety of approaches have identified a number of invasion loci that lie outside this region. One approach identified S. typhi genes that, when cloned into the noninvasive strain E. coli HB101, rendered the organism capable of entering cultured epithelial cells (50). This region was shown to contain a minimum of four separate loci required for invasion. A homologous region from the S. typhimurium chromosome was unable to confer invasiveness to E. coli HB101, indicating that there are differences in the invasion mechanisms of these two microorganisms. Since these genes are located near the cs 63 region of the Salmonella chromosome, it is tempting to hypothesize that they are functionally related to the entry genes encoded in this region. However, this relationship is uncertain, since this locus is located in a region of the chromosome that is not unique to salmonellae (interspersed among the recA gene and genes involved in sorbitol metabolism) and therefore their proximity to the inv, spa, hil, orgA, and prgH loci may not be functionally significant.

Using TnphoA mutagenesis, Stone et al. screened S. enteritidis for alkaline phosphatase-positive colonies that were defective in entry and/or attachment (211). The mutants were classified into different groups on the basis of their relative ability to enter HEp-2, CHO, and MDCK cells in culture and were mapped to nine different loci on the Salmonella chromosome. The presence of a large number of loci having different effects in different cell lines suggests that entry into host cells by salmonellae is a multifactorial process and/or that these organisms encode alternative entry pathways. The multifactorial nature of Salmonella entry is also supported by the work of Betts and Finlay, which found four additional loci in S. typhimurium that were important for entry into Caco-2 cells (19). However, a definitive proof of the existence of alternative entry pathways awaits the characterization of these various mutants.

Regulation of the Invasion Phenotype

The invasion phenotype of salmonellae appears to be highly regulated. The assembly of the invasome may require the coordinated regulation of the expression of the genes encoding the structural components themselves as well as the assembly apparatus. Although the regulatory mechanisms governing this assembly process are poorly understood, evidence indicating that there are probably several regulatory networks modulating the expression of genes required for Salmonella entry is beginning to accumulate. Interestingly, some of the regulatory cues that regulate Salmonella entry appear to be remarkably similar to those that regulate Shigella invasion (see above). For example, the expression of the inv locus is regulated by osmolarity, with optimal expression achieved under osmotic conditions (300 mM NaCl) equivalent to that of the extracellular body fluids (73). However, in contrast to results for shigellae, mutations in ompR had no effect on inv gene expression (73). Expression of the inv genes is also influenced by changes in the levels of DNA superhelicity. Mutations in topA, which drastically alters the levels of DNA superhelicity, severely affected invA expression and rendered S. typhimurium unable to enter cultured epithelial cells (73). However, mutations in osmZ (HN-S), the Salmonella equivalent of Shigella virR, had no significant effect on inv gene expression and bacterial entry (K. Kaniga and J. Galán, unpublished data). Consistent with this finding, temperature, which is the main regulatory signal in shigellae and exerts its influence through HN-S (VirR), plays no role in regulating Salmonella invasion or inv gene expression (73). Salmonellae enter cultured epithelial cells equally well when grown at 30 or 37°C (J. Galán, unpublished data). An additional level of regulation of inv gene expression may involve InvF, a protein required for Salmonella entry with significant homology to the AraC family of positive transcriptional activators (116). However, the putative regulatory target(s) for InvF has not yet been identified. Interestingly, InvF is homologous to two Shigella proteins required for invasion, VirF and MxiE (see above). Although the regulatory functions of VirF are well established, the potential regulatory role of MxiE has not yet been investigated. Other conditions known to regulate the invasion phenotype are the growth state (128) and the oxygen tension (51, 204), which have also been shown to alter the level of DNA superhelicity (178). The hil locus is predicted to have regulatory functions dependent on these cues, since it was isolated by identifying Salmonella mutants capable of entering cultured epithelial cells grown under nonpermissive conditions (129). The identification of an invasion locus, prgH, negatively regulated by the PhoP-PhoQ two-component system indicates that this regulator is also involved in the regulation of the invasion phenotype, therefore linking the regulatory mechanisms of bacterial entry and intracellular survival (15). The contact-dependent activation of the type III secretion system does not require de novo protein synthesis, which suggests the existence of yet another, even more complex regulatory layer (86). More work will be required to understand the complexity of the regulation of the entry phenotype.

CONCLUSIONS

Despite the distinct clinical pattern of salmonellosis and shigellosis, common features are beginning to emerge when the early steps of host cell invasion by Salmonella and Shigella spp. are compared. Microorganisms of both genera are able to induce membrane ruffles on epithelial cells, thereby entering via a micropinocytic event. No common patterns are yet emerging in the signaling cascades triggered by these microorganisms, but since these events are not completely understood, it is premature to draw conclusions as to similarities and differences. The absence of Ca²⁺ fluxes in the Shigella signaling pathways and the inability of Salmonella spp. to induce apoptosis in infected macrophages are indications that significant differences between microorganisms of these two genera may be found. Both salmonellae and shigellae express supramolecular structures on their surface while establishing contact with epithelial cell surface. These structures are likely to consist of a presynthesized pool of invasion-associated proteins that are rapidly released upon contact with host cells through a specialized type III secretory apparatus which is functionally and structurally highly related in species of the two genera. Nothing is presently known about the cell-associated molecules that may activate the secretory apparatus. Invasins have been identified in shigellae, and a putative candidate has been identified in salmonellae. However, meaningful comparison of effector molecules awaits the identification of all putative effectors in both shigellae and salmonellae. Similarities and differences are apparent in the various cues that regulate expression of invasion-associated molecules in both shigellae and salmonellae. While osmolarity and changes in DNA superhelicity may influence the expression of the invasion phenotype in both shigellae and salmonellae, the roles of H-NS and OmpR in this context appear to differ. Another potential similarity in the regulatory mechanisms of shigellae and salmonellae may involve members of the AraC family of transcription regulators. Finally, the G+C contents of the genes encoding entry functions in both salmonellae and shigellae are significantly different from the average of other chromosomal genes. This observation, in conjunction with their locations in the vicinity of insertion sequences in salmonellae or in a plasmid in shigellae, suggests that these organisms may have acquired this genetic information by horizontal transmission from another source. It is likely that comparative studies on the molecular bases of Shigella and Salmonella invasion of epithelial cells will soon provide fascinating examples of convergence and divergence in virulence strategies of closely related pathogens.