Intracellular Voyeurism: Examining the Modulation of Host Cell Activities by Salmonella enterica Serovar Typhimurium

JASON SZETO¹ AND JOHN H. BRUMELL^1,2*

[SECTION EDITOR: BRETT FINLAY]

Posted July 25, 2005

Infection, Immunity, Injury, and Repair Program, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8,¹ and Department of Medical Genetics and Microbiology, University of Toronto, #4388, Medical Sciences Building, 1 King's College Circle, University of Toronto, Toronto, Ontario, Canada M5S 1A8²

*Corresponding author. Phone: (416) 813-7654 ext. 3555, fax: (416) 813-5028, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

Salmonella spp. can infect host cells by gaining entry through phagocytosis or by inducing host cell membrane ruffling that facilitates bacterial uptake. With its wide host range, Salmonella enterica serovar Typhimurium has proven to be an important model organism for studying intracellular bacterial pathogenesis. Upon entry into host cells, serovar Typhimurium typically resides within a membrane-bound compartment termed the Salmonella-containing vacuole (SCV). This specialized compartment is essential for proper bacterial replication and survival. From the SCV, serovar Typhimurium can inject several effector proteins that subvert many normal host cell systems, including endocytic trafficking, cytoskeletal rearrangements, lipid signaling and distribution, and innate and adaptive host defenses. The study of these intracellular events has been made possible through the use of various imaging techniques, ranging from classic methods of transmission electron microscopy to advanced live-cell fluorescence confocal microscopy. In addition, DNA microarrays have now been used to provide a "snapshot" of global gene expression in serovar Typhimurium residing within the infected host cell. Here, we describe key aspects of Salmonella-induced subversion of host cell activities, providing examples of imaging that have been used to elucidate these events.

INTRODUCTION

Salmonellae are facultative intracellular pathogens responsible for several diseases. Salmonella enterica serovar Typhimurium is able to infect a broad host range, causing gastroenteritis in humans and a systemic disease in mice that resembles typhoid fever (210). This organism must remodel or subvert various host cell systems to create an intracellular environment that is amenable to its survival. As such, serovar Typhimurium has proven to be an important model organism for studying intracellular pathogenicity.

Many Salmonella virulence genes have been identified, and it is estimated that at least 4% of the serovar Typhimurium genome is required for full virulence of this pathogen in mice (14). A large number of these genes are contained within large gene clusters termed Salmonella pathogenicity islands (SPIs) (83, 96, 97, 141). Similar to other pathogenicity islands, SPIs possess a lower G + C content than the rest of the chromosome. This suggests that SPIs consist of foreign DNA acquired by horizontal transfer from phages, transposons, or other sources (96, 141). Central to the virulence of serovar Typhimurium are type III secretion systems (TTSS), which are specialized needlelike structures that span both the inner and outer bacterial membranes. With these delivery systems, serovar Typhimurium injects an arsenal of virulence proteins (called effectors) into the host cell cytosol to modulate specific host cell processes (Table 1). In this review, we will focus on some of the key aspects of Salmonella intracellular pathogenicity, in particular, the modulation of host cell processes by type-III-secreted effector proteins.

Table 1Proteins secreted by serovar Typhimurium type III secretion systems

SEROVAR TYPHIMURIUM TYPE III SECRETION SYSTEMS

Serovar Typhimurium employs two TTSS, encoded separately in SPI-1 and SPI-2 (67, 82, 100). Genes within SPI-1, located at centisome 63, are needed for invasion of epithelial cells and are expressed in growth conditions that mimic the extracellular environment typical of the intestinal lumen, including high osmolarity and slight alkalinity (2, 67, 69). Typical of other TTSS, some of the proteins secreted by the SPI-1 TTSS (e.g., PrgI) serve a structural role and form part of the syringelike secretion apparatus itself (68). Detailed electron micrographs show that this complex consists of a cylindrical base that resides in the bacterial envelope, from which a needlelike projection originates (Fig. 1) (126, 143). The reader is also directed to other reviews for more comprehensive discussions of TTSS in Salmonella and in other bacteria (38, 44, 68, 106).

Fig. 1Electron micrograph of type III secretion system needle complexes isolated from serovar Typhimurium. Note the wider base of the complex, involved in association with the bacterial cell wall, and the needlelike projection composed of PrgI protein through which effectors are translocated. Bar = 30 nm.

Source: Reproduced from Marlovits et al. (2004), Science306:1040--1042, with permission. Copyright (2004) AAAS.

To permit the transfer of bacterial proteins into the host, a subset of secreted proteins (e.g., SipB and SipC) act to form a pore, or translocon, in the host cell membrane (31). Other SPI-1 TTSS-secreted proteins, termed effectors, are subsequently injected into the host cell cytosol through the bacterial secretion apparatus and translocon. These effectors can modulate a variety of host cell responses that facilitate bacterial entry into the host, including actin remodeling, membrane ruffling, and phosphoinositide metabolism (see below) (19, 53, 70). Similarly, components of the translocon can also serve as effectors to modulate host cell machinery on their delivery into the host cell. Invasion by Salmonella is therefore quite distinct from classical, receptor-mediated phagocytosis and can be more appropriately described as a "forced feeding" of the host cell (25).

Some TTSS-secreted proteins also require specific chaperones to ensure their proper delivery into the host cell. Chaperones may achieve this through several mechanisms, such as stabilizing their respective TTSS substrates in the bacterial cytosol, optimizing the presentation of secretion signals on these TTSS substrates, or even regulating the expression of TTSS proteins and other virulence genes (58, 175). It has also been proposed that TTSS chaperones may help establish a hierarchy of secretion to ensure that temporal control of TTSS components and effectors is maintained (9). The multifunctional nature of TTSS chaperones is illustrated by the serovar Typhimurium SicA protein, which acts as a chaperone for the secreted translocon components SipB and SipC (212). In a serovar Typhimurium sicA mutant, SipB and SipC are unstable and degraded, indicating a role for SicA in providing stability to its protein substrates (212). SicA may also act to partition SipB and SipC in the bacterial cytosol, thus preventing their premature association. Furthermore, this chaperone appears to function with the InvF transcriptional regulator to activate several virulence genes, including sipB, sipC, sigD, sptP, and sicA itself (47, 48).

Most of the serovar Typhimurium population that enters host cells resides within a membrane-bound compartment termed the Salmonella-containing vacuole (SCV) (54, 123, 207). SPI-2 TTSS genes, located at centisome 31, are expressed inside the SCV and are essential for intracellular replication and systemic infection of mice (see below for a more detailed discussion of SPI-2 gene expression) (39, 99, 171, 172, 179, 192). While expression of genes encoded within SPI-1 is downregulated after invasion, SPI-1 and SPI-2 are not mutually exclusive, and there is evidence that some interaction exists between the two. Since SPI-2 genes are not induced until several hours after invasion, certain SPI-1 effectors may be needed to establish a permissive intracellular niche for serovar Typhimurium. Indeed, although usually associated with host cell invasion, the SPI-1 TTSS is required for maximal intracellular replication (197). Moreover, some effectors, such as SopD, SseK, and SspH1, appear to be secreted by both SPI-1 and SPI-2 TTSS (22, 130, 157).

Recently, the SPI-1 effector SigD has been shown to persist within infected cells at time points long after the initial invasion event (>12 h) (52). SigD was detected at these late time points in RAW 264.7 macrophages infected with wild-type serovar Typhimurium by using immunoblotting methods and by monitoring Akt serine/threonine kinase activity and downstream iNOS production, both of which are stimulated by this effector (52, 198). SigD transcript levels, as well as transcripts of the SPI-1 TTSS structural gene prgI and SPI-1 regulator invF, were detectable in serovar Typhimurium several hours after invasion. Hence, SPI-1 effectors, and presumably the SPI-1 TTSS, can modulate host cell activities even after the induction of SPI-2 genes (52). Studies have also demonstrated that certain SPI-2 gene mutations result in abrogation of SPI-1 gene expression, including those encoding the translocon protein SipC and the SPI-1 transcriptional activator HilA (50, 98). Hence, while serovar Typhimurium SPI-1 and SPI-2 TTSS are often associated with distinct purposes, a temporally coordinated functional overlap may exist between the two secretion systems.

While some TTSS effectors are encoded within their respective SPI-1 and SPI-2 pathogenicity islands, many are encoded elsewhere on the chromosome, including other SPIs (e.g., SPI-5), or within lysogenized phages (15, 120, 141, 172, 222). There is no conserved amino acid sequence or motif common to all TTSS effectors that would direct their translocation (106). However, a subset of effectors that are members of the Salmonella Translocated Effector (STE) family (Table 1) have been shown to contain a conserved N-terminal region within the first 150 amino acids that serves as a translocation signal for SPI-1 and SPI-2 effector delivery into the host cell (22, 156). Other factors that regulate TTSS effector translocation may involve mRNA signals within the encoded effector genes themselves and/or involvement of dedicated chaperones (58, 79, 175). Serovar Typhimurium can also secrete the plasmid-encoded, actin-depolymerizing SpvB toxin in a SPI-1 and SPI-2 TTSS-independent manner, suggesting yet another route by which this pathogen can introduce virulence proteins into the host cytosol (80).

SEROVAR TYPHIMURIUM INVASION OF HOST CELLS

Serovar Typhimurium has a broad host range and can infect both phagocytic and nonphagocytic cells, thus serving as a prototypical model for Salmonella invasion and intracellular parasitism (23). Typically, this pathogen enters the host by oral ingestion and breaches the intestinal epithelia to promote systemic disease. Like many other pathogens, a major route of entry across the intestinal barrier appears to occur through M cells. These specialized epithelial cells line the intestinal lumen at Peyer’s patches through which antigens traverse to the basolateral side and are delivered to antigen-presenting cells (115, 167, 196). The absence or poor development of protective mucous and glycocalyx layers may also make M cells attractive portals for pathogen entry (196). By interacting with macrophages and dendritic cells found underneath M-cell layers, serovar Typhimurium can be transported to other locations such as the liver, spleen, and bone marrow to cause systemic infection (59, 145, 183, 189, 204, 217).

Noninvasive serovar Typhimurium SPI-1 mutants can still breach the intestinal barrier in an M-cell-independent manner, a process that requires CD18-expressing phagocytes (217). Dendritic cells may facilitate this mode of entry since they are able to extend dendrites between absorptive epithelial cells without disrupting the epithelial barrier itself. As a result, dendrites sampling the intestinal lumen for antigens provide another conduit for serovar Typhimurium entry (167, 182, 204). Furthermore, Salmonella may traverse the intestinal barrier by direct invasion of columnar absorptive cells (137, 153).

Invasion of nonphagocytic cells by serovar Typhimurium in vitro is a SPI-1-dependent process (40, 69). SPI-1 is essential for promoting intestinal infection and subsequent gastroenteritis since serovar Typhimurium SPI-1 mutants are attenuated for virulence when delivered orally to mice. However, these mutants retain virulence when administered intraperitoneally, indicating a role for SPI-1 in the invasion and traversal of gastrointestinal epithelial cells by serovar Typhimurium, but not in the systemic phases of disease (69).

Entry of serovar Typhimurium into epithelial cells is characterized by dramatic host cell surface membrane ruffling that drives bacterial uptake (Fig. 2) (68, 70). To date, the contributions of six serovar Typhimurium SPI-1 TTSS effectors involved in invasion have been examined. These effectors include SopE, SopE2, SopB/SigD (herein referred to as SigD), SptP, SipA, and SipC (68). These effectors have been shown to localize to the host cell plasma membrane ( 32). However, SipC is the only member that contains a potential sequence, specifically a central hydrophobic region, that might allow its insertion into host cell membranes (190). The remaining effectors likely associate with the host cell membrane by interacting with resident membrane-associated components.

Fig. 2Transmission electron micrograph showing cell surface membrane ruffling in Caco-2 epithelial cells that facilitates serovar Typhimurium internalization. Arrow indicates serovar Typhimurium in nascent Salmonella-containing vacuoles (SCV) after entry into the host cell.

Source: Image courtesy of B. Brett Finlay, University of British Columbia.

SopE, SopE2, and SigD are involved in remodeling host actin near the cell surface to produce the characteristic membrane ruffles that precede bacterial entry (Fig. 3). SopE and SopE2 function as guanine nucleotide exchange factors that activate important regulators of actin cytoskeleton assembly, namely, the Rho family GTPases Cdc42 and Rac-1 (3, 63, 89, 201). In turn, downstream effector proteins of the GTPases, such as N-WASP and WAVE, are recruited to membrane ruffles to mediate Arp2/3-dependent actin nucleation (214). Consistent with their role in promoting membrane ruffling, both SopE and SopE2 have been localized along the tips and length of filopodial extensions in fibroblasts infected with serovar Typhimurium or in fibroblasts transfected with genes encoding either effector (32). While the exact mechanism of Rho GTPase activation by Salmonella effectors is not clear, structural studies have provided some insight into this process. The crystal structure of SopE complexed with Cdc42 has recently been obtained and showed that conformational changes in the switch I and II regions of Cdc42 are induced by SopE binding, which may promote guanine nucleotide exchange and activation (30). Similarly, nuclear magnetic resolution (NMR) studies have indicated that SopE2 may also induce conformational changes in the nucleotide-binding region of Cdc42 to promote nucleotide exchange (226).

Fig. 3Confocal micrograph depicting the accumulation of polymerized actin (green) at the site of serovar Typhimurium (red) contact with a mouse embryonic fibroblast cell.

In contrast to SopE and SopE2, SigD is not a nucleotide-exchange factor but can still activate Cdc42 and stimulate actin assembly by an unknown mechanism (234). SigD possesses phosphoinositide phosphatase activity (142, 169, 234) and it is increasingly apparent that phosphoinositides play critical roles in modulating actin assembly and actin-regulatory proteins (206, 229). Using a chimeric protein composed of the pleckstrin homology domain of phospholipase Cδ and green fluorescent protein (GFP), Terebiznik et al. (208) visualized phosphatidylinositol-4,5-bisphosphate [PtdIns(4,5)P₂] in serovar Typhimurium-infected cells. This phosphoinositide was found to be enriched at the outermost regions of membrane ruffles but not at their bases where plasmalemmal invagination occurred (Movie 1). SigD was found to be responsible for the elimination of PtdIns(4,5)P₂ at these invagination sites and for promoting membrane fission to facilitate bacterial entry, indicating that the hydrolysis of PtdIns(4,5)P₂ is required for Salmonella entry (208). This is consistent with previous studies suggesting a general requirement for PtdIns(4,5)P₂ depletion at sites of phagocytosis and endocytosis (12, 118).

Two SPI-1 effectors, SipA and SipC, appear to have cooperative roles in nucleating and stabilizing actin polymers required for membrane ruffling during bacterial invasion (93, 150). By cross-linking actin monomers, it is proposed that SipA reduces the critical concentration of actin needed for polymerization and stabilizes F actin (136, 160, 237). SipA also complexes with and promotes the actin-bundling activity of T plastin at the site of bacterial–host contact (235). Furthermore, SipA has recently been shown to interfere with the actin-depolymerizing proteins ADF/cofilin and gelsolin (149). Thus serovar Typhimurium translocates effectors into the host cell that not only trigger actin polymerization but also counteract the major F-actin-destabilizing proteins (70).

Serovar Typhimurium can also reverse the actin rearrangements that it induces (70). The SPI-1 TTSS effector SptP functions as a GTPase-activating protein for Cdc42 and Rac-1, converting the latter two proteins into inactive GDP-bound forms. As a result, normal actin architecture is restored at the site of bacterial entry and membrane ruffling ceases (64, 65). To help temporally regulate actin assembly and disassembly during invasion, serovar Typhimurium has exploited the host cell ubiquitin-mediated proteasome degradation pathway. While SopE is rapidly ubiquitinated and degraded when introduced into host cells, SptP is much more stable (125). This difference in stability is determined by the N-terminal secretion and translocation domain found in each protein. The half-life of SopE was significantly increased when fused to this N-terminal domain of SptP, while the opposite was observed with a SptP fusion to the secretion and translocation domain of SopE (125). Hence, the persistence of SptP compared with the relatively short-lived SopE likely ensures that SopE-induced actin rearrangements at the site of serovar Typhimurium entry can be restored to normal.

Our understanding of Salmonella invasion continues to evolve, in part, because of a recent focus on this process using polarized epithelial cells. One study determined that Rac activation, not Cdc42, was required for invasion of the apical surface of polarized MDCK cells (45). By contrast, during basolateral invasion both Rac and Cdc42 were activated but neither was required for bacterial entry (45). Thus, invasion of the apical and basolateral surface occurs by distinct mechanisms. Another recent study demonstrated a role for the SPI-1 effectors SopA and SopD in the invasion process (181). Baumler and colleagues complemented a noninvasive mutant (lacking sipAsopABDE2) with plasmids expressing different SPI-1 effectors (181). They demonstrated for the first time that SopA and SopD contribute to Salmonella invasion; however, the apparent role of SopA and SopD could only be detected in polarized epithelial cells.

SEROVAR TYPHIMURIUM GENE EXPRESSION WITHIN HOST CELLS

After internalization, Salmonella cells reside within a membrane-bound compartment termed the Salmonella-containing vacuole (Fig. 4) (54, 123, 207). Here, induction of SPI-2 gene expression ensues, which is essential for intracellular growth of serovar Typhimurium (39, 123, 222). Expression of SPI-2 genes requires the SsrA-SsrB two-component regulatory system which is, in turn, activated by the OmpR-EnvZ two-component system (39, 76, 133). Various studies have shown that environmental changes, such as low pH and osmolarity, and low concentrations of Mg²⁺, Ca²⁺, or phosphate, can activate SPI-2 gene expression (49, 76, 133, 155). These conditions may reflect those encountered by serovar Typhimurium within the nascent SCV. Some of the environmental factors that induce SPI-2 genes (e.g., pH and osmolarity) are dissimilar to those that stimulate SPI-1 genes, likely ensuring that the correct TTSS is activated at the proper time (2, 49, 76, 133, 155).

Fig. 4Transmission electron micrograph of serovar Typhimurium-infected Caco-2 epithelial cell shows internalized bacteria residing within compartments termed Salmonella-containing vacuoles (SCVs) (arrow).

Source: Image courtesy of B. Brett Finlay, University of British Columbia.

Aside from SPI-2 gene expression, many other genes must also be activated to allow intracellular growth of Salmonella. The PhoP-PhoQ two-component regulatory system responds to low Mg²⁺ and acts as a master regulator of at least 40 serovar Typhimurium genes necessary for intracellular survival (1, 72, 81). PhoP-PhoQ signaling activates mgtC gene expression, which is involved in Mg²⁺ acquisition and is required for intracellular replication and virulence in mice (10). Genes that confer resistance to antimicrobial peptides produced by the host cell, as well as genes that modify bacterial lipopolysaccharide (LPS), are also under the control of the PhoP-PhoQ two-component system (57, 86, 193, 194). Furthermore, the PhoP-PhoQ regulon appears to play a role in inhibiting fusion of lysosomes with the SCV, thus contributing to the intracellular survival of Salmonella (78).

Using techniques such as in vivo expression technology (IVET) and random chromosomal insertion of promoterless reporter genes, several studies have attempted to determine which serovar Typhimurium genes may be upregulated after host cell invasion (95, 179). Not surprisingly, most genes identified in these studies are located in pathogenicity islands. Recently, the complete transcriptional profile of serovar Typhimurium following macrophage infection has been obtained by using DNA microarray analyses, representing the first global survey of bacterial gene expression within a host cell system (56). Of the 4,451 coding sequences examined, 919 exhibited significant changes in expression within a macrophage cell line, with 384 genes upregulated and 535 genes downregulated (56). As expected, SPI-2 genes were induced, while SPI-1 gene expression was reduced after invasion. In addition, genes involved in LPS biosynthesis and those encoding flagellin and fimbriae were downregulated, possibly to avoid recognition of serovar Typhimurium surface structures by the immune system (56). By examining the gene-expression profiles it was possible to deduce that the SCV is likely low in Mg²⁺, Fe²⁺, and phosphate, but high in potassium and Fe³⁺ (56). Serovar Typhimurium does not seem to be starved of amino acids since genes responsible for amino acid biosynthesis were not upregulated. Evidence of using the Entner-Doudoroff pathway and gluconate as a carbon source in the SCV was found (56). Significantly, 408 genes of unknown function had altered expression upon invasion of macrophages, further highlighting the complexity of the serovar Typhimurium response to the host cell environment (56).

DYNAMIC INTERACTIONS OF SALMONELLA-CONTAINING VACUOLES WITH THE ENDOCYTIC PATHWAY

One of the challenges of elucidating the mechanisms of serovar Typhimurium pathogenesis has been the heterogeneous nature of its intracellular niche in host cells examined in vitro. Studies have indicated that there are at least two distinct populations of intracellular serovar Typhimurium, namely static (or slow growing) and rapidly growing cells (8, 29). Also, it has recently been found that a small but significant fraction of internalized serovar Typhimurium are not present within SCVs and do not colocalize with typical SCV markers such as LAMP-1 (6, 27). Studies in dendritic cells have revealed that a certain proportion of intracellular serovar Typhimurium do not express SPI-2 genes (111). Since SPI-2-encoded proteins are important for intracellular survival (123, 222), this differential expression of SPI-2 in a given population of serovar Typhimurium may account for the observed intracellular heterogeneities. However, most internalized serovar Typhimurium cells reside in SCVs and replicate within these compartments. This population has been the subject of most studies and will be discussed in the following sections.

Unlike classic phagosomes, Salmonella growth and replication is permitted within the SCV (74, 123, 189). During the early stages of its development, the nascent SCV closely resembles a phagosome as it acquires protein markers typically associated with phagosome maturation. These include the GTPase Rab5, implicated in directing early endocytic vesicle trafficking, transferrin receptor, phosphatidylinositol-3-phosphate [PtdIns(3)P], and early endosome antigen 1 protein (EEA1) (166, 176, 191, 199, 221). Similar to phagosomes, these markers remain on the SCV for only a brief period after infection. The rapid recycling of these early endosomal markers from the SCV suggests that the initial dynamics of SCV maturation closely resemble those of nascent phagosomes.

It is established that the dissociation of early endosomal markers from the SCV is followed by the acquisition of a subset of proteins typically associated with late-endocytic and lysosomal compartments. These include Rab7, vacuolar ATPase (vATPase), CD63, and lysosomal glycoproteins (Lgps) LAMP-1 and LAMP-2 (26, 73, 151, 199). While there has been one report of fusion occurring between the SCV and lysosomes (173), it is now generally accepted that survival of bacteria within the SCV is attributed to its uncoupling, or segregation, from the normal route of the host endosomal pathway (20, 28, 73, 92, 108). SCV-lysosomal fusion appears to be impaired in infected macrophages (28, 46, 92, 108). Furthermore, there is a delay in the interaction between the SCV and late-endocytic compartments in epithelial cells (26). Evidence of impaired lysosomal fusion with the SCV is the reduced recruitment of the lysosomal hydrolase cathepsin D and mannose 6-phosphate receptor (MPR; responsible for trafficking hydrolases to the lysosome) on the SCV during the first few hours (~3 h) after infection (73, 151).

What dictates the appearance of a specific subset of late-endosomal/lysosomal markers on the SCV is not known. It has been shown that serovar Typhimurium can promote the clustering of Rab7- and Lgp-enriched, but cathepsin D- and MPR-deficient, vesicles around the SCV within the first hour of invasion ( 151). Hence, it seems that serovar Typhimurium can divert the normal trafficking of specific vesicles to provide a source of Rab7 and LAMP proteins, but not cathepsin D and MPR, to the SCV. While the nature and origin of such vesicles is currently unknown, they may represent intermediate compartments that normally connect late endosomes to lysosomes in the late-endocytic pathway (151). The SPI-2 effector SpiC has been shown to interfere with global intracellular trafficking, including endosome–endosome and phagosome–lysosome fusion events (213), and may play a role in regulating the interactions that occur between the SCV and host cell vesicle traffic.

While its recruitment is impaired at early times after infection, cathepsin D is eventually localized to the SCV at later time points (>3 h), suggesting fusion of the SCV with another subset of late-endocytic compartment may occur with delayed kinetics (26). Similarly, the late-endosome-specific lipid lysobisphosphatidic acid (LBPA) is also acquired by the SCV at late time points after infection (26). Why serovar Typhimurium would permit hydrolytic components such as cathepsin D to be recruited to SCVs is unclear. However, analyses of purified SCV compartments have indicated they preferentially retain unprocessed, inactive forms of cathepsin D (92, 144, 158).

The delayed interaction between the SCV and late-endocytic compartments is intriguing since the GTPase Rab7, which normally modulates transport between late endosomes and lysosomes, is also localized to the SCV. SCV-associated Rab7 still appears to be functional based on its cycling dynamics, as demonstrated by fluorescent recovery after photobleaching (FRAP) experiments (Fig. 5) (144). Expression of a dominant negative Rab7 slows LAMP-1 acquisition by the SCV but does not impair the recruitment of earlier endosomal markers, further implicating the importance of active Rab7 in directing SCV maturation (151).

Fig. 5Functional Rab7 is recruited to the SCV as determined by fluorescent recovery after photobleaching (FRAP). HeLa cells expressing Rab7-GFP were infected with serovar Typhimurium, and the kinetics of Rab7 recruitment to the SCV was measured by using FRAP. Rectangle and circle indicate the positions of SCV and late-endocytic structures, respectively, that were selected for photobleaching. Note that Rab7-GFP fluorescence recovery is similar between both regions subjected to photobleaching, indicating active Rab7-GFP cycling on the SCV. Bar = 10 μm.

Source: Reproduced from Marsman et al. (2004), Mol. Biol. Cell15:2954--2964 (published online before print as 10.1091/mbc.E03-08-0614), with permission of the American Society for Cell Biology.

SALMONELLA -INDUCED FILAMENTS AND THE SCV

Several hours (>5 h) after its formation, the SCV undergoes SPI-2-dependent extensive tubulation to produce structures termed Salmonella-induced filaments (Sifs) that emanate from the bacterial compartment (5, 18, 75, 200). These unique structures likely represent the delayed fusion of the SCV with late-endosomal compartments, since several late-endosomal markers are found on Sifs, including LAMP-1, LAMP-2, Rab7, LBPA, and cathepsin D (Fig. 6 and 7) (18, 26, 75, 90). Sifs are more easily visualized in nonphagocytic cell lines, such as epithelial cells. It is likely that the tubular-vesicular nature of lysosomal structures normally found in macrophages may mask Sif structures in these particular cell lines. However, Sif-like structures have recently been observed in gamma interferon (IFN-γ)-primed murine macrophages infected with serovar Typhimurium expressing a HA-tagged SPI-2 TTSS effector, PipB2 (124). Hence, the Sif phenotype appears to be common to both nonphagocytic and phagocytic cells.

Fig. 6Lysosomal glycoprotein LAMP-1 is present on Salmonella-induced filaments (Sifs) that form at later time points (>5 h) after infection. (A) Henle cells infected with wild-type serovar Typhimurium exhibit long filamentous LAMP-1⁺ structures (green) emanating from the bacteria (red). (B) Cells infected with a serovar Typhimurium SPI-2 mutant do not exhibit the Sif phenotype.

Source:

Fig. 7Cathepsin D is recruited to Sifs and SCVs at late time points after infection. HeLa cells were infected with serovar Typhimurium, fixed 6 h after infection, and costained for LAMP-2 (A; green) and cathepsin D (B; red). Note cathepsin D localizing along Sif structures indicated by LAMP-2 staining. Arrowheads indicate intracellular bacteria residing in SCVs.

Source:

Sifs are associated with serovar Typhimurium undergoing rapid intracellular replication (8). Sif formation occurs in a SPI-2-dependent manner and requires the SPI-2 sifA gene (200). Deletion of sifA renders serovar Typhimurium incapable of Sif formation and destabilizes the SCV, leading to bacterial release into the cytosol (5, 200). Hence, Sifs may play a role in maintaining the integrity of the SCV, perhaps by providing a thoroughfare along which late-endosomal membrane materials can be recruited to maintain the bacterial vacuole. Expression of SifA-GFP alone in uninfected cells is sufficient to induce the filamentation and vacuolation of host cell structures containing LAMP-1 (Fig. 8) (13, 24). This suggests SifA may be responsible for directing extensive fusion of late-endocytic compartments with the SCV to promote filamentous Sif formation (105). The SCV remains stable around sifA mutants that lack the sseJ effector gene. SseJ is predicted to have acyltransferase/lipase activity and may be involved in modifying and/or remodeling the SCV in conjunction with SifA (186).

Fig. 8SifA-GFP alone is sufficient to induce Sif-like structures in uninfected epithelial cells. HeLa cells were transfected with a vector encoding SifA-GFP. After 16 to 20 h, cells were fixed, immunostained for LAMP-1 (A; red), and analyzed by confocal microscopy. While predominantly cytosolic, the SifA-GFP (B; green) also localizes to filamentous LAMP-1⁺ structures.

Source:

The importance of an intact SCV is apparent by observing the fate of bacteria released into the host cell cytosol upon SCV destabilization. In macrophages or dendritic cells, cytosolic serovar Typhimurium do not replicate and, at least in macrophages, are killed instead (6, 178, 200). In stark contrast, cytosolic bacteria replicate rapidly in epithelial cells (Movie 2) (6, 27). It has recently been observed that cytosolic serovar Typhimurium becomes ubiquitinated upon entry into the cytosol of several cell types, including macrophages and nonphagocytic cells (Fig. 9). Proteasomes were also observed to localize on the surface of cytosolic serovar Typhimurium in macrophages but not in epithelial cells (177). Pharmacologic proteasome inhibition enhanced bacterial replication in macrophages; however, this treatment also appeared to destabilize the SCV and, hence, the role of the proteasome in restricting bacterial growth in the cytosol is unclear (177). Proteasome targeting to serovar Typhimurium in the cytosol of macrophages may play a role in presentation of bacterial peptides on major histocompatibility complex class I (MHC Class I) molecules, an event required for immunity to these bacteria (138, 219). Because macrophages are the host cell niche for serovar Typhimurium during systemic disease, it seems crucial that the integrity of the SCV be maintained to avoid the bactericidal environment of the macrophage cytosol.

Fig. 9Serovar Typhimurium colocalize with ubiquitinated proteins in the cytosol of mammalian cells. (Left) RAW 264.7 cells were fixed 2 h after infection by invasive serovar Typhimurium and stained for LAMP-1 (red) and bacteria (blue). (Right) Ubiquitinated proteins stained with monoclonal antibody (FK2) that recognizes mono- and polyubiquitinated proteins, but not free ubiquitin. Arrow indicates cytosolic bacteria. Arrowhead indicates bacteria residing in LAMP-1⁺ SCV. Bar = 5 μm.

Source: Reproduced from Perrin et al. (2004), Curr. Biol.14:806--811, with permission from Elsevier.

In addition to SifA, other effectors, including PipB, PipB2, SifB, SopD2, SseF, SseG, and SseJ (Table 1), have been shown to localize to Sifs and/or SCVs and may modulate their biogenesis (21, 61, 101, 123, 124, 128, 186, 188). Similar to SifA, at least three of these effectors, SseF, SseG, and SopD2, appear to have direct roles in promoting Sif formation (87, 114, 128). For example, deletion of sseF or sseG was shown to arrest Sif formation in serovar Typhimurium-infected cells at an apparently intermediate stage, characterized by the discontinuous localization of LAMP-1 along tubules termed "pseudo-Sifs" (128). Recently, SopD2 was shown to act with SifA to promote the extension of Sifs from the SCV (114). Similar to sseF and sseG deletions, deletion of sopD2 in serovar Typhimurium also disrupted Sif formation, concomitant with an increase in the number of pseudo-Sifs (Fig. 10) (114).

Fig. 10Deletion of sopD2 impairs Sif formation. HeLa cells were infected with wild-type serovar Typhimurium (A) or an isogenic mutant bearing a disruption in the sopD2 gene (B). Both strains carried a plasmid expressing SifA with two internal HA epitope tags. Infected cells were fixed 8 h after infection, coimmunostained for LAMP-2 and HA epitope, and analyzed by confocal microscopy. HA-tagged SifA protein was localized predominantly on Sifs in wild-type infected cells (A, arrows). In cells infected with the sopD2 mutant, SifA was also localized to filamentous structures exhibiting punctate rather than continuous distribution of LAMP-2 (B, arrows), representing a phenotype known as "pseudo-Sifs." (C) Cells were infected as described in panels A and B, and the number of Sifs (filled boxes) and pseudo-Sifs (open boxes) formed was determined. Bar = 10 μm.

Source:

In contrast to these proteins, the SPI-2 effector SseJ and virulence plasmid-encoded SpvB may have roles in negatively regulating Sif formation. Serovar Typhimurium mutants lacking one or both of these genes can induce the formation of significantly more Sifs in epithelial cells (8). However, both proteins are not required for the decline in Sifs typically observed in serovar Typhimurium-infected epithelial cells after 8 to 10 h following invasion (8). Why Sif prevalence decreases at later time points after infection is not clear; however, the involvement of several positive and negative regulatory effectors in the Sif phenotype suggests Sif dynamics are an important aspect of serovar Typhimurium pathogenesis.

CHOLESTEROL, LIPIDS, AND THE SALMONELLA-CONTAINING VACUOLE

Significant amounts of cholesterol are redistributed to the SCV, with as much as 20 to 40% of total cellular cholesterol eventually accumulating on these vacuoles (26, 34). Cholesterol recruitment to the SCV at early and later time points after infection depends on SPI-1 and SPI-2, respectively (34, 77). In addition to cholesterol, glycosylphosphatidylinositol (GPI)-anchored CD55 and glycosphingolipid GM1 are also found on SCVs (34). Cholesterol distribution on the SCV is uneven and "patchy" (26, 34), suggesting that lipid microdomains may form on this vacuole to promote signaling and/or recruiting events that promote SCV maturation (34). Consistent with this, the SPI-2 effectors PipB and PipB2 have been detected in lipid rafts isolated from infected cells, suggesting serovar Typhimurium may exploit such domains to modulate various host activities, such as signal transduction, cytoskeletal remodeling, and membrane and protein trafficking (123, 124).

Modulation of host phosphoinositide metabolism also appears to be essential for serovar Typhimurium growth in host cells (53). While the inositol phosphatase SigD is implicated in facilitating serovar Typhimurium entry (see above) (208), this SPI-1 effector also acts to maintain high levels of PtdIns(3)P on the SCV membrane after bacterial entry (101). In the absence of sopB (gene encoding SigD), serovar Typhimurium exhibited decreased intracellular growth and delayed SCV maturation, suggesting PtdIns(3)P is required to establish a permissive environment for growth in the SCV (101). Elevated levels of PtdIns(3)P on the SCV may contribute to the fusion of the SCV with other bacteria-free, PtdIns(3)P-containing vesicles observed to form near the site of invasion, thus forming more spacious vesicles in which the bacteria reside (101).

THE SALMONELLA-CONTAINING VACUOLE AND THE SECRETORY PATHWAY

Extensive evidence shows that the SCV is derived from the plasma membrane and subsequently interacts selectively with specific components of the endocytic pathway (20, 105, 151, 199). However, recent studies provide some evidence that the SCV, as well as phagosomes in general, may form from the fusion of the endoplasmic reticulum (ER) with the plasma membrane to facilitate particle engulfment (66). By using immunoelectron microscopy, ER-specific glucose-6-phosphatase has been detected on vacuoles surrounding serovar Typhimurium (66). Furthermore, contacts between the ER and SCV have been observed (66). Other proteins associated with the secretory pathway, such as calnexin, calreticulin, and Sec61, have also been detected by using Western blotting on isolated phagosomes formed from internalized latex beads, suggesting this is a general paradigm for phagocytosis (66). As it is established that numerous endocytic markers characterize maturation of the SCV, the contribution of the ER and secretory pathway toward SCV formation will be an interesting topic for further study.

The late SCV is also directed to and surrounded by the Golgi network in a process that depends on the SPI-2 effector SseG (Movie 3) (188). Serovar Typhimurium sseG mutants that are not targeted to the Golgi apparatus are inhibited in intracellular growth (188). The association of SCVs with the Golgi was observed only in epithelial cell lines such as HeLa and INT 407 and not in macrophages; however, it remains possible that SCV–Golgi interactions still occur in phagocytic cells, albeit in a less obvious manner (188). The exploitation of both endocytic and secretory pathways during Salmonella pathogenesis has led to the proposal that the former provides a source of membrane to maintain the SCV, while the latter may provide a source of nutrition for the bacteria (188).

INTERNALIZED SEROVAR TYPHIMURIUM AND INTERACTION WITH HOST CYTOSKELETAL STRUCTURES

In addition to cell-surface-associated actin rearrangements that facilitate serovar Typhimurium invasion, modulation of actin dynamics along endocytic networks is also important for intracellular bacterial survival. Several hours (4 to 8 h) after invasion, a network of filamentous actin is induced to form de novo around the SCV in a SPI-2 TTSS-dependent process (152). Treatment with actin-depolymerizing agents such as latrunculin B or cytochalasin D destabilized the SCV and significantly decreased intracellular replication of wild-type serovar Typhimurium in RAW and murine-elicited peritoneal macrophages, highlighting the importance of vacuole-associated actin polymerization (VAP) in pathogenesis (152). It is possible that VAP provides a spatial and temporal barrier from host factors that are detrimental to Salmonella replication. Surrounding actin networks may also provide a route along which selective host factors are delivered to the SCV. Currently, it is not known what factor(s) trigger the accumulation of actin around the SCV. The remodeling of actin by internalized serovar Typhimurium does not depend on Rho family GTPases, nor the Arp2/3 complex, in stark contrast to invading bacteria at the host cell surface, suggesting a novel mechanism for inducing actin polymerization in vivo (214).

Recently, the SPI-2 effectors SspH2 and SseI were observed to colocalize with polymerized actin networks surrounding the SCV on transfection. These effectors were shown to interact with the actin-binding proteins filamin (SspH2) and profilin (SspH2 and SseI), suggesting their potential roles in the VAP phenotype; however, VAP could still occur in cells infected with serovar Typhimurium sspH2 and/or sseI mutants (154). VAP structures are not detected at later time points (18 h) after infection (18). In vitro, SspH2 can inhibit actin polymerization; thus, it has been proposed that SspH2 may actually inhibit actin polymerization at a later time after infection, acting with the virulence plasmid-encoded ADP-ribosylating factor SpvB ( 134, 209) to limit VAP formation (123).

Serovar Typhimurium can also modulate microtubule assembly. The SPI-2 effectors SseF and SseG colocalize with and induce the bundling of the host cell microtubule network (129). Furthermore, these effectors appear to aggregate endosomal vesicles along microtubules; hence, it is possible that manipulation of the microtubule network provides internalized bacteria with access to membrane compartments that help maintain the SCV, or that redirect innate host cell antimicrobial factors away from bacteria (129).

Microtubules also play a role in the Sif phenotype. Sifs colocalize with microtubules, and the addition of the microtubule-disrupting drug nocodazole blocks Sif formation (Fig. 11) (18). This indicates that microtubules act as a scaffold on which Sifs are assembled (18, 75). While Sifs also colocalize with actin filaments, disruption of the actin network with cytochalasin D does not prevent their formation (18, 152).

Fig. 11Sifs form along microtubules. HeLa cells were infected with wild-type serovar Typhimurium for 18 h, fixed and stained for tubulin (red) and LAMP-1 (green). Yellow indicates colocalization of LAMP-1⁺ filamentous structures (Sifs) with the microtubule network.

Source:

Microtubule motors have recently been implicated in SCV and Sif maturation. During its development, the SCV gradually moves centripetally toward the microtubule-organizing center of the host cell, while Sifs elongate from the SCV in a centrifugal manner. A meshwork of microtubules and the microtubule-associated motors dynein and kinesin localize to the SCV (Fig. 12) (85, 91, 144). The Rab7-interacting lysosomal protein (RILP) is also recruited to the early SCV in a Rab7-dependent manner (85, 91, 144). RILP is involved in coupling Rab7-containing compartments to the dynein/dynactin complex responsible for centripetal migration of intracellular cargo (33, 117). In contrast to the SCV, Sifs that extend from this compartment contain significantly reduced RILP and dynein, despite the presence of active Rab7 on these filamentous structures (Fig. 13) (91). The serovar Typhimurium effector SifA, an important contributor to Sif formation, was shown to interact with Rab7, leading to the proposal that SifA facilitates the centrifugal expansion of Sifs by uncoupling Sif-associated Rab7 from RILP. The resulting RILP-free Sifs would be detached from the centripetal dynein motor complex and would be free to expand outward. Sif elongation also requires centrifugal kinesin motors, which facilitate movement toward the cell periphery (91).

Fig. 12The microtubule motors kinesin and dynein are recruited to the vicinity of intracellular serovar Typhimurium in HeLa cells. Cells were infected with wild-type serovar Typhimurium (green) for 10 h. Red signal in merged images shows distribution of kinesin or dynein, as indicated. Host cell perimeters are outlined by dotted line.

Source: Reproduced from Guo et al. (1997), Science276:250--253, with permission of the publisher.

Fig. 13Sifs contain Rab7 but are deficient in Rab7-interacting lysosomal protein (RILP). HeLa cells were transfected with Rab7-GFP (A, C; green) or with RILP-GFP (D, F; green) 2 h after infection by serovar Typhimurium. Cells were incubated another 12 to 16 h, fixed, and stained for LAMP-1. Arrows point to Sifs, identified as filamentous, LAMP-positive structures. Solid arrows correspond to the magnified inset in each panel. Note that Rab7-GFP is localized to Sifs (A, C) but that RILP-GFP is not (D, F).

Source: Reproduced from Harrison et al. (2004), Mol. Biol. Cell15:3146--3154 (published online before print as 10.1091/mbc.E04-02-0092), with permission of the American Society for Cell Biology.

MODULATION OF HOST CELLULAR DEFENSES BY SEROVAR TYPHIMURIUM

To promote survival in its intracellular niche, serovar Typhimurium has developed mechanisms to evade bactericidal reactive oxygen species (ROS) and reactive nitrogen intermediates (RNI) generated by the host cell (35, 71, 218). Both immunofluorescence and electron microscopy have shown that NADPH oxidase activity, as well as the NADPH oxidase subunits p22^phox, p47^phox, and cytochrome b₅₅₈, are significantly reduced on the SCV of wild-type serovar Typhimurium compared with vacuoles containing SPI-2 mutants (Fig. 14) (71, 218). Similarly, inducible nitric oxide synthase (iNOS) does not localize to the SCV of wild-type bacteria (35). Since serovar Typhimurium SPI-2 mutants are unable to avoid the delivery of ROS and RNI to their vacuoles, it suggests that SPI-2-dependent effectors modify the intracellular trafficking of host bactericidal components (35, 71, 218). This involvement of SPI-2 is further highlighted by the fact that SPI-2 mutant bacteria regain their virulence in mice that lack NADPH oxidase activity (218). Currently, it is not known which secreted SPI-2 effector(s) are involved in subverting ROS or RNI delivery to the SCV.

Fig. 14Significant NADPH oxidase activity is not present on the SCV. Peritoneal macrophages from periodate-elicited C57BL/6 mice were infected with wild-type (A) or SPI-2 mutant (B) serovar Typhimurium. NADPH oxidase activity was detected as cerium perhydroxide precipitate. Note the dark precipitate localized on the SCV surrounding SPI-2 mutant (B) but not wild-type serovar Typhimurium (A).

Source: Reproduced from Vazquez-Torres et al. (2000), Science287:1655--1658, with permission of the publisher.

Another host factor implicated in controlling bacterial infection is the natural resistance-associated macrophage protein 1 (Nramp1), also known as solute carrier 11a1 (SLC11A1), a divalent metal transporter associated with lysosomal compartments in macrophages (84, 220). Nramp1^+/+ mice are resistant to serovar Typhimurium infection; however, the mechanism for this is not clear (220). Nramp1 may confer resistance to serovar Typhimurium by actively depleting the SCV of Fe²⁺, Mn²⁺, or other essential divalent cations, or by promoting SCV fusion with the normal endocytic pathway (46, 109, 110, 232). Note that serovar Typhimurium still exhibit SPI-2 dependent replication in Nramp1^+/+ mice, albeit not to the high levels observed in susceptible Nramp1^–/–animals . In fact, serovar Typhimurium can persist within mesenteric lymph node macrophages of Nramp1^+/+ mice for at least a year (161). This ability to chronically infect a host may, in part, be the result of upregulated SPI-2-associated gene expression in the presence of Nramp1 (232). Because the SPI-2 TTSS, or its translocated effectors, is required for replication in Nramp1^+/+ mice, it is conceivable that SPI-2 acts to counter the detrimental effects of Nramp1 and promote persistent infection (232).

In addition to affecting host cytoskeletal rearrangements and endocytic trafficking, serovar Typhimurium can modulate host gene expression, particularly those involved in cytokine production. Infection of cultured intestinal epithelial cells has been shown to activate ERK, c-Jun NH₂-terminal kinases (JNK), and p38 protein kinase pathways (36, 89, 104, 174, 184). These pathways can activate NF-κB, an important transcriptional factor of proinflammatory cytokines such as interleukin-8 (IL-8) (104). The interaction of serovar Typhimurium with polarized epithelial cells induces the basolateral secretion of IL-8 (55, 146). Serovar Typhimurium attachment to these polarized cells also stimulates the release of other chemoattractants on the apical surface (146, 147, 148). As a result, transepithelial polymorphonuclear neutrophil (PMN) migration proceeds toward the apical membrane.

While this localized inflammatory response induced by serovar Typhimurium may facilitate its ability to invade and spread within the host (163), the influx of immune cells to sites of inflammation may be detrimental to the pathogenic cycle as well. As such, several serovar Typhimurium effectors, namely AvrA and SptP (secreted by SPI-1) and SspH1(secreted by both SPI-1 and SPI-2 TTSS) (Table 1), appear to have roles in downmodulating IL-8 production after invasion (43, 88). AvrA inhibits the NF-κB signaling pathway by blocking the pathway downstream of IKK activation and preventing the p65 subunit of NF-κB from entering into the nucleus (43). SspH1 appears to act by localizing to the host cell nucleus after its translocation where it presumably inhibits NF-κB-dependent gene expression (88). SptP, normally involved in reversing actin polymerizations at the host cell surface, can also attenuate IL-8 production (88). However, SptP has been localized to the plasma membrane by using both immunofluorescence and cell fractionation studies (32), suggesting that this effector disrupts IL-8 gene expression by affecting signaling events at the host cell surface.

The presentation of peptides by MHC class I and class II molecules forms an integral part of the host T-cell immune response. To counter this, intracellular bacteria have developed several strategies to modulate or impair antigen presentation. These strategies include inhibiting the maturation of phagosomal compartments to limit bacterial degradation and impairing synthesis of MHC class I and II molecules (139). Studies have shown that expression of serovar Typhimurium phoP-regulated genes can impair the processing and MHC class II presentation of antigens expressed by the bacteria (225). In the absence of phoP, serovar Typhimurium are more efficiently presented by MHC class II molecules (225). Currently it is not known which of the phoP-regulated genes in serovar Typhimurium are involved in modulating antigen presentation by host cells.

Serovar Typhimurium can also downregulate the translocation of MHC class II molecules to the host cell surface (159). This process does not affect overall class II biosynthesis and peptide loading and, as a result, class II molecules accumulate within the host cell in LAMP-1-positive vacuoles that are distinct from the SCV. SifA may have a role in this process, since mutant sifA bacteria do not inhibit cell surface presentation of MHC class II molecules (159).

A serovar Typhimurium gene involved in disrupting peptide presentation by MHC class I molecules has also been identified. Inactivation of this gene, yej, was shown to increase MHC class I presentation in murine macrophages and to induce CD8+ T-cell responses against serovar Typhimurium in mouse models (180). Hence, serovar Typhimurium has developed several ways to avoid eliciting CD4+ and CD8+ T-cell responses by restricting the presentation of Salmonella-derived epitopes. While the exact mechanisms of restricting class I and class II presentation are not known, they will undoubtedly provide interesting avenues of future study.

SALMONELLA -INDUCED CELL DEATH

During the systemic phase of disease, serovar Typhimurium relies primarily on phagocytic cells such as macrophages and dendritic cells to provide replicative niches and/or vehicles for dissemination (183, 189, 224). While essential for systemic infection, macrophages also undergo a cell death process triggered by serovar Typhimurium infection. Both SPI-1 and SPI-2 TTSSs are able to induce this event at different times by using different mechanisms in vitro (107). SPI-1-mediated cell death does not require bacterial internalization, provided the SPI-1 TTSS genes are expressed optimally (37).

The exact nature of Salmonella-induced cytotoxicity has been the subject of considerable discussion (11, 107, 121). Macrophages undergoing Salmonella-induced death exhibit features that are typical of both apoptosis and necrosis. Classical apoptotic changes, including chromatin condensation and fragmentation, cytoplasmic vacuolation, and membrane blebbing have been observed in infected macrophages (Fig. 15) (37, 165). Others have observed dying macrophage morphologies more consistent with necrosis, including disrupted nuclear and plasma membranes and a lack of DNA condensation or fragmentation (17, 223). These distinctly different observations may be a reflection of the increasingly blurred lines of distinction between apoptosis and necrosis (121). Furthermore, these differences may have resulted from the different experimental conditions used in each study (121).

Fig. 15Serovar Typhimurium induces cell death in macrophages. RAW 264.7 macrophages were infected with wild-type serovar Typhimurium (A) or were left uninfected (B). Cells were processed for transmission electron microscopy 2 h after infection. Note the aggregation of chromatin and vacuolization of the cytosol, which is indicative of apoptotic cell death, in the macrophage infected with wild-type serovar Typhimurium (A).

Source: Reproduced from Monack et al. (1996), Proc. Natl. Acad. Sci. USA93:9833--9838. Copyright (1996) National Academy of Sciences.

What is apparent is that caspase-1, belonging to a class of caspases implicated in inflammatory responses, is involved in Salmonella-induced cell death. Loss of caspase-1 activity confers increased macrophage resistance to serovar Typhimurium-induced cytotoxicity (17, 102, 103). Furthermore, caspase-1-knockout mice have significantly greater resistance against serovar Typhimurium than wild-type mice do (163). The SPI-1 translocon component/effector SipB has been shown to bind and activate caspase-1, resulting in the processing and release of IL-1β and IL-18 accompanying macrophage cell death (103, 170). Dendritic cells also exhibit Salmonella-induced cell death that depends on bacterial SipB and host cell caspase-1 (216). In addition, the cytotoxic effects of serovar Typhimurium depend on the activation of Cdc42 and Rac-1, which are normally stimulated upon bacterial invasion (see above); however, the mechanism(s) by which these Rho-GTPases induce/signal host cell death is not clear (60).

While IL-1β and IL-18 are not required for the actual process of cell death itself (164), the caspase-1-mediated release of these cytokines may have a proinflammatory role that supports the spread of serovar Typhimurium within the host. One interesting proposal is that IL-1β and IL-18 contribute to inflammation and disruption of the intestinal barrier to facilitate secondary bacterial invasion (112, 121). It has also been proposed that sites of Salmonella-induced inflammation and cell death could attract other phagocytic cells, including dendritic cells, which may be used by serovar Typhimurium to further disseminate within the host (112). The antigenic material released from macrophages that succumb to Salmonella-induced cell death can be processed and presented by bystander dendritic cells (230). Hence, it is possible that the host cell immune system can benefit from the death of macrophages by virtue of released Salmonella antigens that can be more efficiently presented to T cells by dendritic cells (204, 230).

Despite the importance of caspase-1 in mediating rapid macrophage death, even caspase-1-deficient macrophages eventually succumb to serovar Typhimurium infection (102, 113). This death pathway also requires SipB; however, caspase-2 activation is also involved ( 113). In addition, caspase-1-deficient macrophages may succumb to autophagy-mediated, or type II, cell death upon serovar Typhimurium infection. Again, SipB is involved in this process, which has been observed to localize to mitochondria where it appears to disrupt organelle integrity (102). Furthermore, SipB can induce the formation of and localize within multilamellar structures that resemble autophagosomes surrounding, or in close association with, mitochondria in transfected epithelial cells (Fig. 16). Because caspase-1-deficient macrophages still succumb to Salmonella-induced death in the presence of a pan-caspase inhibitor, it appears that autophagy itself is the main contributor to cytotoxicity in these cells (102).

Fig. 16The serovar Typhimurium SPI-1 effector SipB is able to induce formation of multilamellar structures that are near or surrounding mitochondria. COS-2 cells transfected with SipB-GFP were fixed and subjected to immunoelectron microscopy with rabbit anti-GFP antisera and gold-labeled anti-rabbit secondary antibody. (A) SipB-GFP localized to multimembrane-layered structures near mitochondria. (B and C) Mitochondria (B and C, right side) appeared to be fusing with SifB-containing multilamellar structures (B and C, left side). Arrows and box indicate regions where apparent membrane fusion is occurring. (D) Mitochondria surrounded by SipB-induced multilamellar structure. Bars = 250 nm.

Source: Reproduced from Hernandez et al. (2003), J. Cell Biol.163:1123-1131, with permission of The Rockefeller University Press.

Activation of caspase-3 is well characterized to initiate the classic apoptotic pathway; however, it has been reported that Salmonella-induced cell death is caspase-3 independent (17). Recently, SPI-1 proteins were found to be continuously expressed in a serovar Typhimurium ▵Lon strain residing in macrophages, resulting in increased caspase-3-mediated apoptosis compared with the wild-type control (205). Hence, Lon protease may play a role in suppressing the induction of the caspase-3 cascade by serovar Typhimurium, thus preventing excessive macrophage apoptosis that might otherwise elicit a heightened host immune response, in particular, through bystander dendritic cells (205).

Serovar Typhimurium can also kill macrophages in the absence of an active SPI-1 TTSS. However, this killing is delayed, beginning up to 12 h after infection, and occurs in a SPI-2-dependent manner (162, 215). Although the effector(s) that may be involved are currently not known, SPI-2-dependent cytotoxicity is mediated in part by caspase-1 (162, 215). SPI-2-mediated cell death also results in the release of IL-1β, which may induce inflammation and promote dissemination of the pathogen at systemic sites of infection such as the liver and spleen (162).

Other mechanisms for inducing cell death have also been described for serovar Typhimurium. Sequence analysis of another Salmonella pathogenicity island, SPI-4, indicates it may encode a type I (ATP-binding cassette) secretion system (227). Such systems are often associated with the export of cytotoxins (7). Because several cytotoxins are encoded by various Salmonella species (135, 140, 187), it is possible that their secretion is mediated by SPI-4; however, studies must be done to confirm this. Microarray analysis of human macrophage gene responses to invading wild-type and phoP-deficient serovar Typhimurium have also revealed a role for the bacterial phoP/phoQ regulatory system in promoting host cell death (51). Again, little is known about which serovar Typhimurium phoP/phoQ -regulated gene(s) may be involved in cytotoxicity. Finally, expression of the Salmonella ADP-ribosylating protein SpvB is also able to induce apoptosis in macrophages (131); however, there is no evidence that SpvB is secreted by a TTSS. Thus, it is apparent that serovar Typhimurium can trigger host cell death by multiple routes. The benefits that arise from host cell killing, however, remain to be determined.

CONCLUDING REMARKS

It is now clearly established that serovar Typhimurium invasion and intracellular parasitism involves an intricate series of spatially and temporally regulated events. This demands that serovar Typhimurium be able to engage specific host cell machinery during every point of the infection cycle, from initial contact with the host cell to replication within the SCV. This continuous interaction with the host cell has likely contributed to the extensive arsenal that serovar Typhimurium now possesses, including two type III secretion systems, a range of ammunition in the form of TTSS effectors, and a complex genetic regulatory network that coordinates the expression of hundreds of virulence factors.

Studies of serovar Typhimurium intracellular parasitism have provided valuable insight into the inner workings and failures of the host cell in dealing with a highly successful pathogen. Not surprisingly, these same studies have spawned many more questions than answers regarding Salmonella-induced modulation of host cell systems. Despite identifying many secreted Salmonella proteins and noting their localization and the phenotypes they induce, we are only beginning to elucidate the molecular mechanisms of action for a select few TTSS effectors. Much research has centered around SifA, the effector protein essential for Sif formation, which provides SCV stability and determines the fate of internalized serovar Typhimurium. While current proposals favor that Sifs are involved in recruiting host membrane compartments for maintenance of the SCV, the actual function of the Sif structures themselves is still unclear.

Thus far, the maturation of the SCV has also been perplexing. This compartment selectively acquires only a subset of late-endosomal markers. The origins of these markers in the endocytic pathway is not known but may reflect the existence of intermediate fusogenic steps, in particular, in later stages of endosome maturation. Identification of the complete proteome of the SCV would be especially useful; however, technical challenges have limited the success of this endeavor to date (158).

Several reports have also implicated the importance of the phoP/phoQ regulatory system in serovar Typhimurium pathogenesis. Since this two-component system regulates several dozen serovar Typhimurium genes, the identification of specific phoP/phoQ-regulated genes implicated in promoting host colonization by this pathogen would provide a much clearer picture of how these bacteria usurp the host cell. The identification by microarray analyses of several hundred serovar Typhimurium genes upregulated within the macrophage has also provided many candidate pathogenesis genes of which little, or nothing, is known.

Questions abound regarding how serovar Typhimurium deals with host immune defenses. How do the bacteria avoid ROS and RNI? What allows the bacteria to modulate MHC class I and II presentation? Does Salmonella-induced macrophage cell death benefit the pathogen or the host? These and many other important questions await investigation in our attempts to unravel the mechanisms responsible for successful intracellular pathogenesis by serovar Typhimurium and other Salmonella serovars. Ultimately, it is pertinent to confirm the observations made during in vitro cell culture experiments in animal disease models. In doing so, it will be possible to determine which of the many Salmonella-induced phenotypes in host cells are truly relevant in promoting disease.

ACKNOWLEDGMENTS

Our work is supported by grant funding and a New Investigator Award from the Canadian Institutes of Health Research (to J.H.B.). Infrastructure for the Brumell laboratory was provided by a New Opportunities Fund from the Canadian Foundation for Innovation and the Ontario Innovation Trust. J.H.B. is a recipient of the Premier’s Research Excellence Award from the Ontario Ministry of Economic Development and Trade. J.S. is supported by a CAG/CIHR/Solvay Pharma postdoctoral fellowship administered by the Canadian Association of Gastroenterology.

We thank Dr. B. Brett Finlay and members of the Brumell laboratory for critically reading the manuscript.