SALMONELLA INTERACTIONS WITH HOST CELLS: Type III Secretion at Work

Transcription

1 Annu. Rev. Cell Dev. Biol :53 86 Copyright c 2001 by Annual Reviews. All rights reserved SALMONELLA INTERACTIONS WITH HOST CELLS: Type III Secretion at Work Jorge E. Galán Section of Microbial Pathogenesis, Boyer Center for Molecular Medicine, Yale University School of Medicine, New Haven, Connecticut 06536; jorge.galan@yale.edu Key Words bacterial pathogenesis, protein secretion, actin cytoskeleton, Rho GTPases, vesicular trafficking Abstract The bacterial pathogen Salmonella enterica has evolved a very sophisticated functional interface with its vertebrate hosts.at the center of this interface is a specialized organelle, the type III secretion system, that directs the translocation of bacterial proteins into the host cell. Salmonella spp. encode two such systems that deliver a remarkable array of bacterial proteins capable of modulating a variety of cellular functions, including actin cytoskeleton dynamics, nuclear responses, and endocytic trafficking. Many of these bacterial proteins operate by faithful mimicry of host proteins, in some cases representing the result of extensive molecular tinkering and convergent evolution. The coordinated action of these type III secreted proteins secures the replication and survival of the bacteria avoiding overt damage to the host. The study of this remarkable pathogen is not only illuminating general paradigms in microbial pathogenesis but is also providing valuable insight into host cell functions. CONTENTS INTRODUCTION TYPE III SECRETION AND SALMONELLA PATHOGENICITY General Properties of Type III Secretion Systems Two Type III Secretions Systems at the Center of Salmonella-Host Cell Interactions INTERACTION OF SALMONELLA WITH HOST CELLS THROUGH THE SPI-1 TYPE III SECRETION SYSTEM Modulation of the Actin Cytoskeleton by the SPI-1 TTSS Effector Proteins Stimulation of Nuclear Responses, Polymorphonuclear Leukocyte Infiltration, and Chloride Secretion by the SPI-1 TTSS Effector Proteins: The Path to Diarrhea Macrophage Cytotoxicity Mediated by the SPI-1 TTSS Effector Proteins /01/ $

2 54 GALÁN INTERACTION OF SALMONELLA WITH HOST CELLS THROUGH THE SPI-2 TYPE III SECRETION SYSTEM Phenotypes Associated with the SPI-2 TTSS Modulation of Intracellular Trafficking by SPI-2 TTSS Effector Proteins CROSS-TALK BETWEEN SPI-1 AND SPI-2 TTSSs? HOST MIMICRY: A PREVAILING THEME IN TTSS EFFECTOR PROTEINS SUMMARY AND PERSPECTIVES INTRODUCTION From an evolutionary and ecological standpoint, microbial pathogens may be classified in two broad categories: those that infect the host of interest (e.g., humans) only accidentally and those that do so as part of their normal life style. Accidental in this context does not imply a particular frequency of the event but rather it implies that the accidental pathogen can replicate in another niche to perpetuate itself in nature. The distinction, although admittedly arbitrary, is important as it may predict the terms of a given pathogen/host interaction. Accidental pathogens have not co-evolved with the host they occasionally infect, and therefore often enough they are lethal. In contrast, infections with pathogens that have sustained a long-standing association with their hosts are most often self-limiting or go unnoticed. Therefore, the functional interfaces between these pathogens and their hosts are characterized more by their refinement than by their potential for harm. Indeed, close examination of these pathogens not only reveals important insight into the mechanisms of pathogenesis but also contributes to the understanding of the inner workings of the host cell. Salmonella enterica (which encompasses hundreds of serovars such as Typhimurium, Typhi, Enteritidis, etc) are enteropathogenic bacteria capable of causing a wide range of illnesses ranging from mild food poisoning to life-threatening systemic infections. The type of disease caused by these bacteria is dependent on the serovar of the infecting Salmonella as well as the species and immunological status of the infected host. As species, Salmonella enterica can infect a variety of vertebrate animals ranging from cold-blooded vertebrates such as lizards and turtles to warm-blooded animals such as chickens and humans. Salmonella enterica has sustained a coexistence with vertebrate hosts, perhaps going as far back as 10 million years. As a result of this association, Salmonella has evolved extremely sophisticated mechanisms to engage vertebrate hosts. When examined at the cellular and molecular levels, this functional interface reveals an impressive array of determinants designed to engage the innate immune system, to sense the host s environment, or to modulate a variety of cellular processes ranging from actin cytoskeleton dynamics to endocytic trafficking. The overall picture emerging from this close examination is perhaps one of balance, self restraint, and sophistication rather than one of uncontrolled hostility.

3 TYPE III SECRETION 55 A central strategy evolved by Salmonella to interact with host cells involves a specialized organelle termed the type III secretion system (TTSS) (Galán 1999, Galán & Collmer 1999, Galán & Zhou 2000). This system mediates the transfer of bacterial proteins capable of modulating cellular functions into the host cell. Salmonella encodes two such systems that operate in an independent, yet apparently coordinated, fashion (Galán 1999, Hensel 2000). This review focuses on the interaction of Salmonella with host cells through the activity of these specialized organelles. It is not intended to be a comprehensive review of all articles published in this fast-moving field. Rather, I attempt to convey a view of the field, with emphasis on the issues that are best understood and that are more directly related to the cell biology of Salmonella infections. There are a number of excellent reviews covering other aspects of type III secretion and Salmonella pathogenesis (Cornelis & Van Gijsegem 2000, Darwin & Miller 1999b, Hueck 1998, Kingsley & Baumler 2000, Lucas & Lee 2000, Mittrucker & Kaufmann 2000, Ohl & Miller 2001, Wallis & Galyov 2000). TYPE III SECRETION AND SALMONELLA PATHOGENICITY General Properties of Type III Secretion Systems Type III secretion systems (TTSS) are specialized organelles whose central function is the delivery of bacterial protein into eukaryotic cells (Cornelis & Van Gijsegem 2000, Galán & Collmer 1999, Hueck 1998). These systems are evolutionarily related to the flagellar export apparatus and are present not only in several species of bacteria pathogenic for animals but also in bacteria pathogenic for plants or in symbionts for plants or insects. Composed of more than 20 proteins, type III systems are among the most complex protein secretion systems known in bacteria. This complexity is likely related to the temporal and spatial constraints that are central to the activity of these systems. Indeed, TTSSs are highly regulated, both at the transcriptional and post-transcriptional level (Lucas & Lee 2000). Such regulation is essential for the orderly delivery of bacterial proteins that must exert their function in a temporal and spatially coordinated manner. These systems have presumably evolved not just to deliver proteins into eukaryotic cells but also to deliver these effector proteins at the right time to the right place. Other general features distinguish TTSSs: (a) absence of a cleavable signal peptide characteristic of proteins secreted via the sec-dependent pathway, (b) requirement of host cell contact for the full activation of the secretion pathway, and (c) presence of customized accessory proteins that aid the secretion and translocation of many (but not all) of the secreted proteins. Although the mechanisms of type III protein secretion are not well understood, recent studies have begun to provide important insight into the function of this complex system. A subset of structural components of the TTSSs has been shown to form a supramolecular structure termed the needle complex (Kubori et al. 1998, 2000) (Figure 1). This structure spans both the inner and outer

4 56 GALÁN Figure 1 Needle complex of the Salmonella SPI-1 type III secretion system. (A) electron micrograph of osmotically shocked S. typhimurium showing the needle complex on the bacterial envelope. (B) Electron micrograph of purified needle complex (Kubori et al. 1998). (C) Schematic representation of the Salmonella needle complex and its components. Scale bar: 100 nm. membranes of the bacterial envelope and resembles the flagellar hook basal body complex. The needle complex is composed of two pairs of membrane-localized rings joined by a hollow cylindrical structure that serve as the base for an externally localized needle-like protruding structure. Although this structure has been visualized only in Salmonella (Kubori et al. 1998) and Shigella (Blocker et al. 1999, Tamano et al. 2000), the conservation of its core components in other microorganisms suggests that it is a common attribute of all TTSSs. It has been established that the base of the needle complex of one of the Salmonella TTSSs (see below) is composed of three proteins, InvG, PrgH, and PrgK (Kimbrough & Miller 2000; Kubori et al. 1998, 2000). InvG, a homolog of the secretin family of proteins, makes up the outer rings, whereas PrgH and PrgK form the inner rings and connecting cylindrical substructure. Apparently, a single protein, PrgI, forms the needle structure (Kubori et al. 2000). The assembly of the needle complex occurs in a step-wise fashion that is initiated by the sec-dependent secretion of the base components, which are assembled into a multi-ring structure (Sukhan et al. 2001) (Figure 2). Assembly of the needle portion of the needle complex, however, requires the function of the type III secretion apparatus and is carefully coordinated by another protein, InvJ, which controls the needle length (Kubori et al. 2000). Proteins destined to be transported through TTSSs carry multiple signals that route them to the secretion apparatus and eventually to the host cell (Cheng & Schneewind 2000, Cornelis & Van Gijsegem 2000). The nature of the secretion

5 TYPE III SECRETION 57 Figure 2 Model for the assembly pathway of the needle complex. The first step in the assembly process is the sec-dependent export of PrgH, PrgK, InvG, and the accessory protein InvH. Upon secretion through the inner membrane, these proteins may form intermediate structures that eventually lead to the assembly of the complete and stable base substructure which, in conjunction with the export machinery (composed of InvA, InvC, SpaP, SpaQ, SpaR, and SpaS), functions as a type III secretion machinery. This incomplete type III secretion machinery recognizes only a limited number of substrates, which are required for the assembly of the needle portion of the needle complex. No bacterial translocases or TTSS effectors can be secreted until completion of the assembly of the needle substructure. Once the entire needle complex is assembled, the TTSS machinery switches specificity to secrete bacterial effector proteins. This specificity switch is governed by InvJ. A loss-of-function mutation in this protein results in a machinery unable to secrete any other protein but PrgI, the main subunit of the needle substructure, and PrgJ, a putative minor component or scaffold of this substructure. As a result, the invj mutant strain exhibits needle complexes with grossly elongated needles (adapted from Sukhan et al. 2001).

6 58 GALÁN signal appears to differ among secreted proteins and has been the subject of some controversy (Lloyd et al. 2001). Nevertheless, work mostly carried out in the bacterial pathogen Yersinia has defined different secretion signals (Cheng & Schneewind 2000, Cornelis & Van Gijsegem 2000). These signals may involve the messenger RNA of the secreted protein, an amino-terminal domain in the polypeptide chain of the secreted protein, or a cognate-specific associated protein (chaperone) that binds to a discrete domain of the secreted protein. Discussion of the proposed mechanisms of substrate recognition is beyond the scope of this article and the reader is referred to other reviews (Cheng & Schneewind 2000, Cornelis & Van Gijsegem 2000, Galán and Collmer 1999). Nevertheless, it is likely that these different secretion signals provide the molecular basis for the temporal regulation of the secretion of different proteins. It has been demonstrated that there is a built-in hierarchy in the type III secretion process (Kubori et al. 2000). For example, PrgI, the protein that makes up the needle portion of the Salmonella needle complex, is a substrate of the TTSS itself. Only after completion of the needle assembly can the TTSS engage substrates other than PrgI. Such a reprogramming of the type III secretion apparatus depends on the function of another protein, InvJ. Therefore, mutations in invj result in the inability of the apparatus to switch specificity, which results in extremely long needles (Kubori et al. 2000). The same principle may operate for subsequent reprogramming of the secretion apparatus to engage different secreted proteins that may need to act at different times during the infection cycle (see below). As is discussed below, the battery of effector proteins ultimately delivered by these systems into host cells includes a variety of enzymes capable of modulating cellular functions. Two Type III Secretions Systems at the Center of Salmonella-Host Cell Interactions Salmonella enterica encodes two TTSSs located in discrete regions of its chromosome (pathogenicity islands) that are essential for pathogenicity. One of these systems is located at centisome 63 of the chromosome within the pathogenicity island 1 (SPI-1) (Galán 1999). The other system is encoded within the pathogenicity island 2 (SPI-2) at centisome 31 (Hensel 2000). Both systems were most likely acquired by horizontal gene transfer as suggested by their G + C content, which significantly deviates from that of the rest of the bacterial chromosome. Although some cross-talk between these two systems has been reported, it appears that their function is largely independent and is exerted at different stages of infection (see below). While the SPI-1 encoded TTSS is required for the initial interaction of Salmonella with intestinal epithelial cells, the SPI-2 encoded system is required for systemic infection. Consistent with their functions at different stages of the pathogenic cycle, the expression of these two systems also occurs at different stages of infection. The SPI-1 encoded system is expressed by Salmonella within the intestinal lumen, whereas the SPI-2-encoded system is expressed only after Salmonella has gained access to host cells.

7 TYPE III SECRETION 59 INTERACTION OF SALMONELLA WITH HOST CELLS THROUGH THE SPI-1 TYPE III SECRETION SYSTEM The Salmonella SPI-1 encoded TTSS was originally identified during a search for Salmonella genes involved in mediating bacterial entry into non-phagocytic cells (Galán & Curtiss 1989). Subsequent analysis established that this system is encoded within a pathogenicity island (Mills et al. 1995) most likely acquired by Salmonella some million years ago, in an event that probably marked the beginning of the close association of these bacteria with vertebrate hosts. The nucleotide composition of coding sequences of this pathogenicity island, combined with the presence of remnants of mobile genetic elements in its immediate vicinity, strongly suggests that this system was acquired by horizontal gene transfer. The SPI-1 TTSS is present in all Salmonella serovars including the ancient Salmonella agona (Ochman & Groisman 1996). SPI-1 TTSS is strikingly similar to a recently identified TTSS in Escherichia coli 0157:H7, the causative agent of hemolytic uremic syndrome, associated with a number of outbreaks due to tainted beef products (Perna 2001). The conservation of the components of these TTSSs is remarkable and includes the gene order of the locus. This observation strongly argues for a common ancestor for these two systems. However, the proteins that are destined to travel through these two different systems are not conserved. Therefore, if these two systems have a common origin, they have been adapted to carry out specific functions in each bacterial species. At least 19 polypeptides are secreted by the SPI-1 TTSS of Salmonella typhimurium (Galán & Collmer 1999) (Table 1). Four, SpaO, InvJ, PrgI and PrgJ, are part of the complex of proteins involved in protein secretion and/or the assembly of the needle complex (Collazo & Galán 1996; Collazo et al. 1995; Kimbrough & Miller 2000; Kubori et al. 1998, 2000). Three proteins, SipB, SipC, and SipD, are required for protein translocation across the eukaryotic cell membrane (Collazo & Galán 1997). This does not exclude the possibility that some of these proteins (e.g., SipB and SipC) may carry out additional functions (see below). The remaining proteins carry out effector functions (or are predicted to do so) within the host cell and are therefore known as effector proteins. The specific function of some of these proteins is discussed below.unlike the secreted proteins involved in secretion or translocation, which are encoded within SPI-1, the majority of the effector proteins are encoded elsewhere in the Salmonella chromosome in mobile genetic elements or pathogenicity islands (Bakshi et al. 2000; Galyov et al. 1997; Hardt et al. 1998b; Hong & Miller 1998; Stender et al. 2000;Wood et al. 1998, 1996). In at least one case, an effector protein is encoded within the genome of a functional bacteriophage (Hardt et al. 1998b, Mirold et al. 1999). Not all Salmonella strains carry the same battery of effector proteins (Hardt & Galán 1997, Hardt et al. 1998b). Therefore, it appears that there is plasticity in the set of proteins assembled by a given Salmonella to be delivered by the SPI-1 TTSS. Such plasticity may be aided by placing the genes encoding effector proteins within mobile genetic elements and away from the location of the core components of the TTSS. Whether the

8 60 GALÁN TABLE 1 Salmonella proteins secreted through the SPI-1 type III secretion system Secreted protein Proposed function References InvJ Required for needle complex assembly and protein (Collazo et al. 1995, Kubori secretion; controls length of the needle component et al. 2000) of TTSS complex SpaO Required for needle complex assembly and protein (Collazo & Galán 1996, secretion Sukhan et al. 2001) PrgI Component of the needle portion of the TTSS (Kubori et al. 2000) complex PrgJ Required for the assembly of the needle portion of (Sukhan et al. 2001) the TTSS complex SipA Binds actin, diminishes its critical concentration, (Zhou et al. 1999a,b) stabilizes F-actin and increases the bundling activity of T-plastin Alternative function: stimulation of PMN (Lee et al. 2000) transmigration SipB TTSS effector protein translocation through (Collazo & Galán, 1997) eukaryotic cell membranes Alternative function: binding and activation of (Hersh et al. 1999) caspase 1 SipC TTSS effector protein translocation through (Collazo & Galán 1997) eukaryotic cell membranes Alternative function: actin nucleation and bundling (Hayward & Koronakis, 1999) SipD TTSS effector protein translocation through (Collazo & Galán, 1997) eukaryotic cell membranes SptP Amino terminus: GTPase activating protein activity (Fu & Galán 1999, Kaniga towards Cdc42 and Rac; carboxy terminus: et al. 1996, Stebbins & Galán, 2000) tyrosine phosphatase activity Reverses cellular changes stimulated by other Salmonella effectors AvrA Sequence homology to YopJ which has been (Hardt & Galán 1997, proposed to be an ubiquitin-like protein proteinase; Schesser et al. 2000) singificance unknown SopE Exchange factor for Cdc42 and Rac; stimulates actin (Hardt et al. 1998a, Rudolph cytoskelecton rearrangements and nuclear responses et al. 1999) SopE2 Exchange factor for Cdc42 and Rac; stimulates actin (Bakshi et al. 2000, Stender cytoskeleton rearrangements and nuclear responses; et al. 2000) 70% sequence similarity to SopE SopA Stimulates PMN transmigration; mechanisms (Wood et al. 2000) unknown SopB Inositol phosphatase; stimulates actin cytoskeleton (Norris et al. 1998, Zhou reorganization, nuclear responses and chloride et al. 2001) secretion SopD Promotes fluid accumulation in an intestinal loop (Jones et al. 1998) model of infection; mechanism unknown SlrP Contains leucine-rich repeats; required for mouse (Tsolis et al. 1999b) virulence; mechanism unknown SspH1 Contains leucine-rich repeats; required for cow (Miao et al. 1999) virulence; mechanism unknown

9 TYPE III SECRETION 61 profile of effector proteins presented to the host contributes in any way to host specificity remains to be determined. In this context, it is interesting to note that at least one effector protein, AvrA, is absent from two Salmonella serovars with very narrow host range (i e., S. choleraesuis and S. typhi) (Hardt & Galán 1997). More intriguing, a homolog of this protein has been implicated in determining host range in the plant pathogenic bacteria Xanthomonas campestris (Whalen et al. 1993). Another homolog, the Yersinia TTSS secreted protein YopJ, has been recently reported to exert its function as a ubiquitin-like proteinase (Orth et al. 2000). The significance of this finding for AvrA function is unclear because it has been shown that, despite the sequence similarity, AvrA and YopJ do not appear to exert a similar activity (Schesser et al. 2000). It is becoming increasingly clear that different effector proteins modulate different cellular processes and therefore are most likely involved in various stages of bacterial infection. For example, a subset of bacterial effectors modulates the actin cytoskeleton to facilitate bacterial entry into non-phagocytic cells, an event that occurs very rapidly and in the early stages of the infection process (Fu & Galán 1998, 1999; Hardt et al. 1998a; Zhou et al. 2001, 1999a,b). However, the majority of the proteins secreted by the SPI-1 TTSS do not appear to be required for these early events and therefore are most likely involved in events that may follow bacterial internalization. The SPI-1 TTSS continues to be active after bacteria internalization and intracellularly located bacteria can deliver type III secreted proteins into the host cell cytosol through the membrane of the enclosing phagosome (Collazo & Galán 1997). Therefore, it is possible that Salmonella reprograms its SPI-1 TTSS machinery to change the battery of proteins delivered to the host cell subsequent to bacterial entry. Such reprogramming may occur as a consequence of changes in the pattern of effector-protein gene expression and/or through posttranscriptional regulatory mechanisms. Indeed, TTSS-associated effector proteins are differentially regulated. For example, InvF, a SPI-1 encoded transcriptional activator protein, specifically regulates the expression of some, but not all, effector proteins (Darwin & Miller 1999a, Eichelberg & Galán 1999, Kaniga et al. 1994). It is therefore possible that different regulators control the expression of different sets of effector proteins, providing the basis for at least some reprogramming of the type III secretion machinery. Modulation of the Actin Cytoskeleton by the SPI-1 TTSS Effector Proteins Salmonella enterica has evolved the ability to enter cells that are normally nonphagocytic such as those that line the intestinal epithelium (Giannella et al. 1973, Takeuchi 1967). This is an important step for pathogenicity because it presumably helps Salmonella in avoiding host defense mechanisms, breaching the intestinal epithelium to reach deeper tissues, reprogramming gene expression, and gaining access to a niche that may be more permissive for its replication. The events surrounding Salmonella internalization into intestinal epithelial cells of an

10 62 GALÁN infected animal were carefully documented by Takeuchi more than 30 years ago (Takeuchi 1967). He observed that after reaching close proximity to the brush border of the intestinal epithelium, Salmonella induced pronounced changes in the plasma membrane that resulted in the disruption of the microvilli. Shortly after, Salmonella could be seen inside the intestinal epithelial cells in a membranebound compartment. Notably, he observed that the changes in the brush border of intestinal cells induced by the incoming bacteria were reversible. Thus, following bacterial internalization, the brush border of the intestinal epithelium regained its normal architecture despite the presence of many internalized bacteria. More than 30 years after this landmark description, the molecular bases for these observations are beginning to be understood in some detail. It is known now that the changes observed by Takeuchi are the consequence of profound actin cytoskeleton rearrangements stimulated by Salmonella at the point of bacteria/host cell contact (Figure 3). These changes are the result of three coordinated steps: (a) stimulation of host signal transduction pathways to promote actin cytoskeleton rearrangements, (b) direct modulation of actin dynamics, and (c) downregulation of the bacterial-stimulated signaling pathways. SIGNAL TRANSDUCTION PATHWAYS LEADING TO SALMONELLA-INDUCED ACTIN CY- TOSKELETON REARRANGEMENTS Contact of Salmonella with host cells results in marked actin cytoskeleton rearrangements, as well as the recruitment of several cytoskeletal-associated proteins to the site of bacterial entry (Finlay & Ruschkowski 1991, Ginocchio et al. 1992). The changes in the actin cytoskeleton and in the plasma membrane of infected cells resemble membrane ruffles induced by growth factors or activated oncogenes and are accompanied by macropinocytosis (Francis et al. 1993, Galán et al. 1992, Garcia-del Portillo & Finlay 1994). These cellular responses lead to the internalization of the bacteria in a membrane-bound compartment. Drugs that interfere with actin dynamics, such as cytochalasin, effectively block bacterial entry (Buckholm 1984), consistent with the requirement for intact actin cytoskeleton to mediate bacterial internalization. Work over the last few years has shown that the function of the Rho-family GTPases Cdc42 and Rac1 is central to the cellular responses leading to the actin cytoskeleton rearrangements stimulated by Salmonella (Chen et al. 1996a). Expression of dominant-negative mutants of Cdc42 (Cdc42N17) and to lesser extent Rac1 (Rac1N17) abolished bacterial internalization. Furthermore, bacterial infection was shown to result in the activation of these GTPases and their recruitment to the cell membrane (LM Chen & JE Galán, unpublished data). Rho GTPases exert their function through a variety of downstream cellular targets (Hall 1998, Van Aelst & D Souza-Schorey 1997). However, the Cdc42 and Rac1 effector targets that mediate Salmonella-stimulated actin cytoskeleton rearrangements have not been identified. Studies have identified mutations in Rho GTPases that are impaired in binding to specific downstream effectors and therefore are differentially defective in various downstream signaling pathways (Lamarche et al. 1996). The use of such effector-loop mutants of Cdc42 allowed the identification of a specific mutant, Cdc42C40, that exerts a dominant interfering effect on S. typhimurium-

11 TYPE III SECRETION 63 induced signaling, which leads to actin rearrangements (Chen et al. 1999). This mutant is unable to bind to a conserved 16-amino acid domain, termed CRIB or p21-binding domain (PBD), present in several Cdc42 downstream effector proteins such as PAK, the Wiskott-Aldrich syndrome protein (WASP), and MSE55 (Burbelo et al. 1995). These results suggest that a CRIB-containing Cdc42 effector protein is essential for Salmonella entry. Studies have demonstrated that one of these proteins, the p21-activated kinase (PAK), is not required for bacterial entry despite its demonstrated involvement in actin dynamics (Chen et al. 1999). A good candidate is WASP because it has been implicated in Cdc42-mediated actin polymerization through its activity on the Arp2-3 complex (Snapper & Rosen 1999). However, the involvement of either of these proteins in Salmonella entry has not been demonstrated. Other events potentially involved in bacterial-induced actin rearrangements include calcium and phosphoinositide fluxes and the stimulation of tyrosine kinases and phospholipase A 2 (PLA 2 ) (Pace et al. 1993, Rosenshine et al. 1994, Ruschkowski et al. 1992). Salmonella infection leads to rapid calcium fluxes, and calcium chelators interfere with bacterial internalization (Ginocchio et al. 1992, Pace et al. 1993). Calcium fluxes appear to be the result of the influx of extracellular calcium, although the mechanisms leading to such an influx are not understood (Pace et al. 1993). The involvement of PLA 2 is supported by the increased levels of icosenoides detected immediately after Salmonella infection of intestinal cells, in particular leukotriene D4 (Pace et al. 1993). Although inhibitors of PLA 2 interfere with bacterial internalization, the mechanisms by which this enzyme is activated and the role of the generated second messengers in bacterial entry are not understood. It is possible that Salmonella-induced activation of Rac-1 leads to the stimulation of PLA 2 activity because this enzyme has been shown to be a downstream effector of this GTPase (Peppelenbosch et al. 1995). Activation of PLA 2 may then lead to calcium fluxes (Pace et al. 1993). However, this possibility needs further investigation. Salmonella encodes a tyrosine phosphatase, SptP, that is delivered into host cells via the SPI-1 TTSS and exerts its function on proteins that are tyrosine phosphorylated upon bacterial infection (see below) (Kaniga et al. 1996; S. Murli, R. Watson &J.E.Galán, submitted). The observation that SptP functions to reverse the cellular changes stimulated by Salmonella argues in favor of an important role for tyrosine kinases in the bacterially induced cellular responses. Indeed, Salmonella infection of host cells leads to the tyrosine phosphorylation of several proteins including the EGF receptor (Galán et al. 1992, Rosenshine et al. 1994). However, the role of tyrosine kinases in Salmonella signaling has been the subject of contradictory reports (Galán et al. 1992, Rosenshine et al. 1994). More studies, including the identification of the specific substrates of the SptP tyrosine phosphatase, will be required to clarify the role of tyrosine kinases in Salmonella-induced signaling. ACTIVATION OF Cdc 42 AND Rac BY SALMONELLASPI-1 TTSS EFFECTOR PROTEINS The strict dependence on the SPI-1 TTSS for Salmonella s ability to stimulate Cdc42- and Rac1-dependent actin cytoskeleton rearrangements indicates that proteins

12 64 GALÁN delivered into the cell via this system are involved, directly or indirectly, in the activation of these Rho GTPases (Chen et al. 1996a). The search for cellular proteins capable of interacting with one of the SPI-1 TTSS effector proteins, SopE, identified Cdc42 and Rac1 (Hardt et al. 1998a). Further biochemical studies established that SopE is a potent guanine nucleotide exchange factor for these GTPases (Hardt et al. 1998a, Rudolph et al. 1999). Consistent with SopE s ability to activate these G proteins, its transient transfection or microinjection into host cells results in the stimulation of marked actin cytoskeleton rearrangements resembling the changes induced by Salmonella infection (Hardt et al. 1998a). These studies, therefore, established that Salmonella is capable of directly engaging key intracellular regulatory molecules to modulate cellular functions. A closely related protein, SopE2, was recently identified in Salmonella and was found to have similar properties to SopE (Bakshi et al. 2000, Stender et al. 2000). A Salmonella strain lacking SopE and SopE2 was still able to stimulate actin cytoskeleton rearrangements (Zhou et al. 2001). This finding indicated the existence of an alternative mechanism by which Salmonella stimulates actin cytoskeleton rearrangements that does not require the activity of these two activators of Rho GTPases. This alternative mechanism involves SopB (SigD), another effector protein of the SPI-1 TTSS (Galyov et al. 1997, Hong & Miller 1998). A Salmonella strain simultaneously carrying loss-of-function mutations in the genes that encode these three effector proteins was unable to stimulate actin cytoskeleton rearrangements (Zhou et al. 2001). SopB is an inositol phosphatase also involved in the stimulation of chloride secretion by Salmonella (see below) (Norris et al. 1998). Although it was originally reported that SopB can cleave in vitro nearly every phosphate in InsP 5 (Norris et al. 1998), more recent studies have shown that SopB exhibits a more restricted activity consistent with that of a 3-phosphatase (Zhou et al. 2001). Indeed, Salmonella-infection of intestinal epithelial cells leads to an accumulation of (1,4,5,6)P 4 and a depletion of (1,3,4,5,6)P 5, which is consistent with its proposed in vitro 3-phosphatase activity (Eckmann et al. 1997, Norris et al. 1998, Zhou et al. 2001). SopB by itself was shown to stimulate actin cytoskeleton rearrangements and mediate bacterial entry (Zhou et al. 2001). These activities are dependent on SopB s phosphatase activity and require Cdc42 (but not Rac1) since expression of a dominant-negative mutant of Cdc42 abolished both SopBinduced actin cytoskeleton rearrangements and bacterial entry. Because SopB lacks exchange activity on Rho GTPases (Zhou et al. 2001), its Cdc42-activating function is most likely the result of changes in phosphoinositide metabolism. Therefore, Salmonella encodes three type III effectors that are capable of stimulating Cdc42 and Rac1, either directly (SopE and SopE2) or indirectly (SopB). The stimulation of these GTPases triggers a cascade of cellular events that lead to actin cytoskeleton reorganization and eventually bacterial internalization. ROLE OF SALMONELLA ACTIN-BINDING PROTEINS IN BACTERIAL ENTRY The actin cytoskeleton is highly dynamic and tightly regulated by several actin-binding proteins, which control the transition of actin from its monomeric form (G-actin) to its multimeric filamentous form (F-actin) (Chen et al. 2000, Cooper & Schafer 2000).

13 TYPE III SECRETION 65 In addition, F-actin is organized in a variety of supramolecular structures whose architecture and stability are also controlled by actin-binding proteins. The concentration of free G-actin in the cell is kept below the level required for its polymerization (critical concentration), thus polymerization requires the function of actin-nucleating proteins. Salmonella has evolved specific mechanisms to modulate these various regulatory steps in order to redirect the actin cytoskeleton to mediate bacterial entry. At least two SPI-1 TTSS secreted proteins have been reported to bind actin and directly modulate actin dynamics. One is SipA, which is required for efficient bacterial entry into host cells (Zhou et al. 1999a). SipA exerts its actin-modulating function through a carboxyl-terminal actin-binding domain. SipA does not nucleate actin and does not bind G-actin (Zhou et al. 1999a). However, SipA significantly reduces the critical concentration of actin and increases the stability of F-actin (Zhou et al. 1999a). Furthermore, SipA causes a conformational change in the F-actin filaments, which when viewed under the electron microscope appear straightened and devoid of their characteristic pitch (Zhou et al. 1999b). This effect may be responsible for the marked increase in the bundling activity of plastin (fimbrin) in the presence of SipA (Zhou et al. 1999b). The precise mechanisms by which SipA modulates actin dynamics are not known. A model has been proposed in which SipA binds two consecutive actin monomers across the double-stranded right-handed helix in both of its faces (Mitra et al. 2000) (Figure 4). The elongated form of SipA (calculated at 95 Å) would allow it to span both actin monomers ensuring the rigidity and stabilization of the filament (Mitra et al. 2000). This model is consistent with the experimentally observed molar ratio of 1:1 of SipA to actin monomer (Zhou et al. 1999a) (in the proposed model, the predicted molar ratio would be n-1:n, which for a long filament would essentially be 1:1). Therefore, SipA may help the initiation of actin polymerization at the site of Salmonella entry by lowering the critical concentration and may increase the stability of actin bundles that drive and support the growth of membrane ruffles and filopodia. These activities may ultimately result in increased efficiency of bacterial internalization. Consistent with this prediction, a Salmonella sipa null mutant exhibits reduced ability to stimulate actin cytoskeletal rearrangements and bacterial internalization (Zhou et al. 1999a). In addition, the cytoskeletal changes induced by the sipa null mutant are distinctly more diffused, contrasting with the more profuse and localized changes induced by wild-type strains. Another SPI-1 TTSS-secreted protein, SipC, has been reported to nucleate and bundle actin in vitro (Hayward & Koronakis 1999). The bundling and nucleating activities were mapped to different domains of SipC. A discrete 120-amino acid domain located at its amino terminus was shown to be sufficient for actin bundling, whereas a 209-amino acid carboxyl-terminal domain was shown to contain the actin-nucleating function (Hayward & Koronakis 1999). The contribution of these activities to Salmonella entry into cells and pathogenicity has not been investigated. The experiments required to address the in vivo contribution of the actin-modulating activities of SipC are hampered by the fact that SipC is required for the translocation of all SPI-1 TTSS effector molecules into host cells (Collazo

14 66 GALÁN Figure 4 Model for SipA activity. (A) Actin monomers are drawn as rounded squares with a groove indicating the nucleotide-binding site. The rigid actin-binding domain of SipA (SipA ), which is 95 Å in length and is predicted to consist of two antiparallel helices, is drawn as an elongated rod. SipA spans two adjacent actin monomers across the double-stranded helix both in the front (dark) and in the back (light gray ) faces. (B) SipA may decrease the actin critical concentration for filament assembly by shifting the equilibrium toward the trimeric actin nucleus. This may occur by multiple mechanisms. SipA may directly bind the trimeric actin nucleus (1) or may bind a G-actin dimer first (2) with addition of either G actin (2a) or another SipA molecule (2b) (adapted from Mitra et al. 2000). & Galán 1997). Thus it will be necessary to dissect the actin-binding and effectortranslocation functions of SipC in order to generate a mutation that will selectively abolish each of these functions. The availability of this mutant would then allow the investigation of the potential contribution of the SipC actin-modulating activities to Salmonella entry. A Salmonella mutant defective in the three activators of Rho GTPases, SopE, SopE2, and SopB, is unable to induce actin cytoskeleton rearrangements despite the fact that this mutant is fully competent for the delivery of SipC (Zhou et al. 2001). These results indicate that the putative actin-modulating activities of SipC are not sufficient to mediate bacterial entry and may not represent an alternative entry pathway. ACTIVE REVERSION OF THE CELLULAR RESPONSES STIMULATED BY SALMONELLA: A CENTRAL ROLE FOR SptP As discussed above, the actin cytoskeleton changes induced by Salmonella are reversible. Therefore, shortly after bacterial invasion, the infected cells regain their normal architecture despite the presence of numerous

15 TYPE III SECRETION 67 bacteria in membrane-bound compartments within the cell (Takeuchi 1967). Remarkably, Salmonella actively participates in this reversion process through the activity of the SPI-1 TTSS effector protein SptP (Fu & Galán 1999). A Salmonella strain carrying a loss-of-function mutation in SptP is fully competent for the stimulation of actin cytoskeleton rearrangements and bacterial entry. However, cells infected with this mutant fail to regain the normal architecture of their cytoskeleton (Fu & Galán 1999). SptP is composed of two functionally independent domains: (a) an aminoterminal domain homologous to the Yersinia YopE protein and to the aminoterminal half of the Pseudomonas aeruginosa ExoS proteins and (b) a carboxyl-terminal domain homologous to the Yersinia YopH protein and several eukaryotic tyrosine phosphatases (Kaniga et al. 1996). The amino-terminal domain of SptP possesses GTPase-activating protein (GAP) activity toward Cdc42 and Rac but not toward other related GTPases such as Rho (Fu & Galán 1999). GAPs exert their function by dramatically stimulating the intrinsic nucleotide hydrolyzing activity of G proteins and therefore effectively switching these regulatory molecules to the inactive (GDP-bound) conformation. Therefore, SptP effectively reverses the SopE/SopE2/SopB-mediated activation of Cd42 and Rac-1 in a remarkable yin and yang. The role of the carboxyl-terminal domain of SptP is less well understood. Similar to its homolog YopH, SptP exhibits potent tyrosine phosphatase activity (Kaniga et al. 1996). However, the identity of the tyrosine phosphorylated proteins that serve as its cellular substrates is not known. Unlike YopH, which disrupts focal adhesions by dephosphorylating p130 cas and the focal adhesion kinase, SptP exerts its effect on proteins that are tyrosine phosphorylated upon bacterial infection (S. Murli, R. Watson & J. E. Galán, submitted). This is consistent with its role as a downregulator of cellular responses stimulated by the bacteria. Whether the tyrosine phosphatase activity of SptP is synergistic with its GAP domain in reversing the actin cytoskeleton rearrangements is not known. A Salmonella strain expressing a phosphatase-defective mutant of SptP that retains wild-type GAP function is capable of reversing the actin cytoskeleton rearrangements stimulated by bacterial infection (Fu & Galán 1999). However, this strain exhibits increased and persistent MAP kinase activation in comparison with wild-type (S. Murli, R. Watson & J. E. Galán, submitted), which suggests that the tyrosine phosphatase domain of SptP may be required for the downregulation of the nuclear responses that follow the stimulation of Cdc42 and Rac by Salmonella (see below). These results also argue for the involvement of tyrosine kinases in Salmonella-induced signaling. Stimulation of Nuclear Responses, Polymorphonuclear Leukocyte Infiltration, and Chloride Secretion by the SPI-1 TTSS Effector Proteins: The Path to Diarrhea A characteristic feature of Salmonella-induced pathology is the induction of an inflammatory response in the intestinal epithelium that leads to the infiltration of

16 68 GALÁN polymorphonuclear leukocytes (PMNs) (Day et al. 1978, McGovern & Slavutin 1979, Rout et al. 1974, Takeuchi et al. 1968). It is believed that this inflammatory response plays a central role in the generation of the typical inflammatory diarrhea that characterizes Salmonella infections (Giannella 1979). The induction of such inflammatory responses is likely to be from, at least in part, the production of cytokines or other pro-inflammatory molecules such as arachidonic acid metabolites. Consistent with this notion, several studies have shown that Salmonella spp. can elicit the production of a variety of cytokines such as IL-8, TNFα, interferon γ, MCP-1, and GM-CSF (Arnold et al. 1993, Eckmann et al. 1993, Hess et al. 1989, Jung et al. 1995, McCormick et al. 1993), as well as peptidoleukotrienes (Pace et al. 1993). The Salmonella SPI-1 TTSS plays a central role in the stimulation of nuclear responses and pro-inflammatory cytokine production (Chen et al. 1996a, Hobbie et al. 1997). These responses are the direct consequence of the stimulation of the MAP kinase pathways, Erk, p38 and Jnk, that follows the activation of the Rho GTPases Cdc42 and Rac1 by the SPI-1 TTSS effector proteins SopE, SopE2, and SopB (Chen et al. 1996a, Hardt et al. 1998a, Hobbie et al. 1997, Zhou et al. 2001). Inhibition of Cdc42, p38, or Mek results in the abrogation of pro-inflammatory cytokine production (Hobbie et al. 1997; S. Murli, R. Watson & J. E. Galán, submitted). Therefore, the stimulation of nuclear responses is a different manifestation of the same events that lead to bacterial entry. Indeed, both responses are the consequence of the activation of Cdc42 and Rac and therefore require the same bacterial effectors (Chen et al. 1996a, Hardt et al. 1998a, Hobbie et al. 1997, Zhou et al. 2001). The commonality in the mechanisms of stimulation of these two cellular responses has led to the erroneous conclusion that bacterial entry (bacterial invasion) per se stimulates pro-inflammatory cytokine production. Although the stimulation of both nuclear and morphological responses requires Cdc42 and Rac, the actual downstream effectors that ultimately mediate these responses are different. For example, the activation of the MAP kinase pathways is the consequence of the Cdc42 and Rac-mediated activation of the p21-activated kinase (PAK) (Chen et al. 1999). Indeed, Salmonella infection of intestinal epithelial cells leads to a rapid and pronounced activation of PAK that precedes the activation of Jnk (Chen et al. 1999). However, PAK is not required for Salmonella-induced actin cytoskeleton rearrangements (see above). Therefore, as observed with other agonists of small G proteins, the bifurcation of the signals stimulated by Salmonella to produce different outputs occurs at the levels of Cdc42 and Rac. The SPI-1 TTSS is not the only mechanism by which Salmonella can stimulate pro-inflammatory cytokine production. Indeed, like any other bacteria, Salmonella encodes several molecules (e.g., lipopolysaccharide, lipoproteins, flagella, etc) collectively recognized as pathogen-associated molecular patterns (PAMPs) that are capable of stimulating the innate immune system (Medzhitov & Janeway 2000a,b). However, stimulation of pro-inflammatory cytokine production in a SPI-1 TTSS-independent manner requires the presence of cognate functional Tolllike receptors for the different bacterial PAMPs in the infected cell. Although not

17 TYPE III SECRETION 69 specifically investigated, it is likely that the apical side of the normal intact epithelium may not display functional Toll-like receptors to prevent potentially harmful responses to the bacterial flora that bathe the intestinal mucosa and express a panoply of PAMPs. Therefore, when interacting with an intact epithelium, the SPI-1 TTSS may well be the only mechanism that Salmonella possesses to stimulate pro-inflammatory cytokine production and initiate the establishment of the inflammatory response that eventually leads to pathology. The SPI-1 TTSS is capable of delivering bacterial effectors through the apical side of intact polarized epithelial cells thereby stimulating Rho GTPases, MAP kinases, and eventually the production of pro-inflammatory cytokines. After the integrity of the intestinal epithelium is compromised, stimulators of the innate immune system are likely to contribute to the amplification of the host response. Indeed, Salmonella causes the disruption of tight junctions of the intestinal epithelium (Jepson et al. 1995, 2000), most likely as a consequence of the activation of Rho GTPases by the SPI-1 TTSS (Gopalakrishnan et al. 1998, Joberty et al. 2000). Such disruption may lead to the exposure of Toll-like receptors to bacterial ligands subsequently amplifying the inflammatory response. At this stage, the ability of Salmonella to kill macrophages with the subsequent release of IL-1β, which is also dependent on the SPI-1 TTSS (see below) (Chen et al. 1996b; Hersh et al. 1999; Monack et al. 1996, 2000; van Der Velden et al. 2000), may contribute to the inflammatory response. This model is consistent with the absolute requirement of the SPI-1 TTSS to establish intestinal infection in animal models of Salmonella-induced diarrhea (Wallis & Galyov 2000). In addition to the stimulators of Rho GTPase function, other SPI-1 TTSS effector proteins have been implicated in the stimulation of PMN transmigration and the induction of fluid accumulation and diarrhea. These include SopA, SopD, and SipA. SopA and SopD were shown to be required for the induction of fluid accumulation in a ligated intestinal loop model of infection (Jones et al. 1998). A Salmonella strain carrying a mutation in sopa was defective in the stimulation of fluid accumulation in this assay (Wood et al. 2000). Similarly, SopD was shown to synergize with SopB (see below) to induce fluid accumulation in the same model system (Jones et al. 1998). The mechanisms by which these two effector proteins exert their function are not known. The SPI-1 TTSS effector protein SipA has been reported to induce PMN transmigration in an in vitro model system (Lee et al. 2000). This function appears to be independent of SipA s actin-binding activity (see above). Surprisingly, the SipA-stimulating activity did not require the translocation functions of TTSS since exogenous application of purified SipA to cells resulted in PMN transmigration. In this context, SipA is the only TTSS effector protein identified so far in any bacteria that appears not to have a requirement for the translocation functions of this system to carry out at least part of its activities. However, these results are difficult to reconcile with the observation that a Salmonella strain carrying a mutation in another effector protein gene, sopa, was unable to stimulate PMN transmigration in an identical assay (Wood et al. 2000) despite the fact that this strain apparently

18 70 GALÁN secretes identical amounts of SipA to that secreted by wild-type. More studies are required to clarify the PMN transmigration stimulatory activity of SipA and the lack of a requirement of TTSS-mediated translocation functions for its activity. In addition to the induction of nuclear responses and the related functions of pro-inflammatory cytokine production and stimulation of PMN transmigration, Salmonella induces other cellular changes that can apparently modulate chloride secretion directly thereby contributing to diarrhea. The modulation of chloride secretion appears to be a direct consequence of products of inositol phosphate metabolism, in particular (1,4,5,6)P 4 (Eckmann et al. 1997). Consistent with this model, infection of intestinal cells with Salmonella leads to the rapid accumulation of (1,4,5,6)P 4 (Eckmann et al. 1997, Norris et al. 1998, Zhou et al. 2001). It has been postulated that (1,4,5,6)P 4 exerts its function by antagonizing the closure of chloride channels by a poorly understood mechanism (Eckmann et al. 1997). Accumulation of (1,4,5,6)P 4 requires the activity of the inositol phosphatase SopB (see above) because a Salmonella sopb mutant strain is impaired in its ability to flux this inositol phosphate to cause fluid accumulation in a ligated ileal loop model of infection (Norris et al. 1998, Zhou et al. 2001). Interestingly, the production of (1,4,5,6)P 4 does not depend exclusively on SopB because the ability of Salmonella to stimulate inositol phosphate metabolism is reduced but not eliminated by the deletion of the sopb gene (Zhou et al. 2001). The Rho GTPase guanine nucleotide exchange factor SopE has also been implicated in the stimulation of inositol phosphate metabolism as a sope-defective mutant strain of S. typhimurium was less active than the wild-type at hydrolyzing InsP 5 and InsP 6 and at generating lower inositol phosphates (Zhou et al. 2001). Unlike SopB, SopE does not exhibit intrinsic phosphatase activity so it must promote inositol phosphate fluxes by activating an endogenous cellular inositol phosphate phosphatase. A sopb, sope double mutant strain was unable to promote any significant changes in inositol phosphate turnover in host cells, which indicates that these two proteins are responsible for most of the inositol phosphate fluxes resulting from bacterial infection (Zhou et al. 2001). The available data indicate that the pathway to Salmonella-induced diarrhea involves a cascade of cellular events orchestrated, directly or indirectly, by the SPI-1 TTSS effector proteins. The following model is proposed to delineate these events (Figure 5). The initial interaction of Salmonella with an intact intestinal epithelium leads to the activation of Cdc42 and Rac1 by SopE, SopE2, and SopB. The activation of these Rho GTPases results in the stimulation of the MAP kinase pathways (Erk, p38, and Jnk), which leads to the activation of the transcription factors AP1 and NF-κB. These transcription factors direct the production of proinflammatory cytokines and other factors, which stimulate PMN transmigration and initiate the establishment of the inflammatory response that eventually leads to diarrhea. The Salmonella-induced responses also result in depolarization of the intestinal epithelial cells and compromise of the intestinal epithelial barrier. These conditions may result in the availability of Toll-like receptors for the engagement of Salmonella-associated stimulators of the innate immune responses (LPS, lipoprotein, flagellin, etc), which may contribute to the amplification of inflammation. The

19 TYPE III SECRETION 71 ability of Salmonella to kill macrophages releasing IL-1β may also enhance the inflammatory response. The indirect modulation of chloride channels by the SPI- TTSS effectors such as SopB and SopE may also contribute to the development of diarrhea. The model outlined above may approximate the events that lead to intestinal pathology. However, it is important to keep in mind that in the overwhelming majority of cases, Salmonella infections are subclinical and do not lead to overt pathology. This is an important issue when trying to understand the function and role of different TTSS effector proteins in Salmonella-host interactions. By excessively focusing on pathology, undoubtedly the reason why we studied these bacteria, we may have missed the true contribution of these effector proteins as meant by the evolutionary process that led to their design. Macrophage Cytotoxicity Mediated by the SPI-1 TTSS Effector Proteins Salmonella spp. are cytotoxic for macrophages (Chen et al. 1996b, Monack et al. 1996, van Der Velden et al. 2000). When macrophages are infected with Salmonella that has been grown under conditions that allow the expression of the SPI-1 TTSS, death is very rapid (<60 min) (Chen et al. 1996b). Although some of the infected macrophages exhibit features of apoptosis (Chen et al. 1996b, Monack et al. 1996), a significant proportion of the dying macrophages do not (Brennan & Cookson 2000, Chen et al. 1996b, Watson et al. 2000). It is now clear that macrophage cell death induced by Salmonella is much more complicated that initially thought and that most likely more than one mechanism of cell death is triggered by Salmonella via the SPI-1 TTSS. In addition, it is also evident that Salmonella can induce macrophage cell death by mechanisms independent of SPI-1 TTSS most likely involving the SPI-2 TTSS (van Der Velden et al. 2000). This complexity has resulted in significant confusion in the field due to, at times, the contradictory nature of some reports in the literature. Some of the confusion is likely related to different growth conditions and different multiplicities of infection utilized in the different experimental setups. It is conceivable that this could lead to the preponderance of one mechanism of death over another. Furthermore, macrophages, unlike intestinal epithelial cells, are responsive to a variety of Toll-like receptor-dependent stimulators of the innate immune system such as LPS and lipoprotein. The sensitivity of macrophages to these stimuli has also contributed to confusion. It has been often overlooked that it is difficult to distinguish signaling pathways stimulated by these stimulators from those specifically stimulated by the SPI-1 TTSS effector proteins (Procyk et al. 1999a,b). However, amidst this confusion, the emerging theme is that of different mechanisms of cell death, which most likely occur simultaneously. The use of knockout mice and different bacterial mutant strains is beginning to provide a more clear picture of this complex interaction. The SPI-1 TTSS secreted protein SipB has been reported to bind and activate caspase-1, and it was proposed that this mechanism is responsible for the induction of apoptosis in macrophages (Hersh et al. 1999). Consistent with this model,

20 72 GALÁN

21 TYPE III SECRETION 73 macrophages obtained from caspase-1 knockout mice were shown to be resistant to Salmonella-induced cytotoxicity (Hersh et al. 1999). In this context, SipB would act in an analogous manner to the highly related IpaB protein from Shigella, for which a similar biochemical activity has been proposed (Y Chen et al. 1996). However, recent reports have shown that macrophages from caspase-1 knockout mice are susceptible to Salmonella-induced cytotoxicity, albeit with delayed kinetics (Jesenberger et al. 2000). Although the reasons for the discrepancy between these reports are not completely clear, differences in the experimental setup, such as multiplicity and time of infection and/or growth conditions, may explain the different results. The mechanisms by which caspase-1 may trigger apoptosis are not understood, as this caspase is not generally considered to be involved in the most common apoptotic pathways. It is possible that caspase-1 may be required for the very rapid death induced by Salmonella, which may be morphologically closer to necrosis than apoptosis (Brennan & Cookson 2000). It is clear that caspase-1 plays an important role in Salmonella pathogenicity because caspase-1 knockout mice are more resistant to Salmonella infection (Monack et al. 2000). However, the role Figure 5 Model for Salmonella-induced enteropathology. Upon interaction with the brush border of the intact intestinal epithelium, Salmonella delivers a battery of effector proteins through its SPI-1 type III secretion system (TTSS). Among these effectors are the guanine nucleotide exchange factors SopE and SopE2 and the inositol phosphatase SopB, which activate the Rho-family GTPases Cdc42 and Rac. Activation of these GTPases leads to a series of downstream events that result in actin cytoskeleton rearrangements, MAP kinase (Erk, Jnk/p38) activation, and the destabilization of tight junctions. The actin cytoskeleton rearrangements, further modulated by the actin-binding protein SipA, result in bacterial uptake. The stimulation of Erk, Jnk, and p38 results in nuclear responses leading to the production of a variety of chemokines, which attract polymorphonuclear leukocytes (PMNs) to the site. The destabilization of tight junctions opens a paracellular pathway for bacterial penetration and allows PMN transmigration. The depolarization of the intestinal epithelium may expose Toll receptors, presumably located in the basolateral side. Consequently, these receptors can be stimulated by a variety of bacterial products such LPS, lipoprotein, and flagellin (generally known as PAMPs), which will further amplify the inflammatory response. The ability of Salmonella to kill macrophages releasing IL-1β may also contribute to this process. SopB, through its inositol phosphatase activity, causes the accumulation of Ins(1,4,5,6)P 4, which stimulates Cl- secretion. Two SPI-1 TTSS effectors, SopA and SopD, contribute to intestinal pathology via unknown mechanisms. In most cases, infection with Salmonella is self-limiting. In these cases, the activity of the GTPaseactivating protein SptP, restoring of the integrity of the intestinal cell by reversing the activation of Cdc42 and Rac, may be crucial. Once inside the cell, expression of SPI-2 TTSS is stimulated. Through its effector proteins, the SPI-2 TTSS mediates the building of an intracellular niche permissive for Salmonella intracellular growth. See text for details.

22 74 GALÁN of caspase-1 in the activation of the pro-inflammatory cytokines IL-1β and IL-18 may be just as important in determining the outcome of Salmonella infections as its potential role as an inducer of macrophage apoptosis. A role for caspase-2 in Salmonella-induced apoptosis has been proposed given that an inhibitor of this caspase appears to interfere with bacterial cytotoxicity (Jesenberger et al. 2000). These results, however, are unclear since macrophages from caspase-2 knockout mice were shown to be fully susceptible to Salmonella-induced cytotoxicity (Jesenberger et al. 2000). The role of SPI-1 TTSS effector proteins other than SipB in macrophage cytotoxicity has not been investigated. The role of SipB itself in this process needs further clarification. Is the caspase-binding activity of SipB the trigger for apoptosis? Alternatively, is its role as a translocator of other SPI-1 TTSS effector proteins and/or its putative pore-forming ability the central contribution to this phenomenon? If SipB by itself induces macrophage cell death, is this death due to apoptosis or necrosis? Although microinjection of SipB into macrophages did result in cell death (Hersh et al. 1999), the experimental approach used in those studies (i.e., dye exclusion) is not able to distinguish between different types of cell death. The clarification of the role of SipB in macrophage cytotoxicity will require the isolation of a SipB mutant unable to bind caspase-1 but still able to function as a translocator of SPI-1 TTSS effector proteins. Such a mutant was recently isolated in the highly related Shigella protein IpaB (Guichon et al. 2001). A Shigella strain expressing a IpaB mutant unable to bind caspase-1 was not impaired in its ability to induce macrophage apoptosis (Guichon et al. 2001). These results suggest that the postulated caspase-1-activating activity of IpaB and, by analogy SipB, may not be the main contribution of these proteins to the induction of apoptosis. Experiments using macrophages from different knockout mice and Salmonella strains carrying mutations in different SPI-1 TTSS effector proteins may help in dissecting the apparently different types of macrophage cell death induced by these bacteria. Such experiments may prove fruitful not only for the understanding of Salmonella pathogenesis but also for the dissection of potentially different forms of cell death. INTERACTION OF SALMONELLA WITH HOST CELLS THROUGH THE SPI-2 TYPE III SECRETION SYSTEM The SPI-2 type III secretion was independently identified during signature tagging mutagenesis of S. typhimurium (Hensel et al. 1995, Shea et al. 1996) and during a search for sequences that were present in S. enterica but absent from E. coli K12 (Ochman et al. 1996). SPI-2 is located within a 40-kb pathogenicity island located at centisome 31 of the Salmonella chromosome immediately adjacent to a trna gene. Unlike SPI-1, SPI-2 is absent from Salmonella bongori, which is thought to be a phylogenetically older serovar of Salmonella enterica (Ochman & Groisman 1996). This observation suggests that SPI-2 was presumably acquired subsequently to the acquisition of SPI-1. Amino acid sequence analysis of the SPI-1 and SPI-2 TTSS components indicate that these two systems did not arise

23 TYPE III SECRETION 75 by gene duplication, but rather are the result of two independent acquisition events. In fact, phylogenetic analysis indicates that the SPI-2 TTSS is among the most distantly related systems to the SPI-1 TTSS. The identification of proteins that are secreted via the SPI-2 TTSS has been hampered by the fact that this system is not normally expressed under standard laboratory conditions but is only expressed after Salmonella entry into host cells (Cirillo et al. 1998, Pfeifer et al. 1999). Nonetheless, a number of proteins that are substrates of the SPI-2 TTSS have been identified by a variety of approaches (Beuzon et al. 1999, Hensel 2000, Miao et al. 1999, Uchiya et al. 1999). The biochemical and biological activities of these different effector proteins have only recently begun to be probed. A subset of the putative secreted proteins (e.g., SseB, SseC, and SseD) is believed to be required for effector protein translocation into the host cell based on their amino acid sequence similarity to proteins that carry out this function in other TTSSs. Putative effector proteins include SpiC, SifA, and a family of leucine-rich repeat proteins with homologs in other bacterial species (Beuzon et al. 2000, Guy et al. 2000, Uchiya et al. 1999). The specific functions of some of these effector proteins are discussed below. A subset of SPI-2 TTSS substrates possess a 100-amino acid amino-terminal domain that shares striking sequence similarity (Beuzon et al. 2000, Guy et al. 2000, Miao et al. 1999). It has been postulated that this domain contains secretion and targeting signals for the SPI-2 TTSS. Thus it is possible that such a domain represents a binding site for a common chaperone. Phenotypes Associated with the SPI-2 TTSS It is widely accepted that SPI-2 TTSS is required for systemic infection (Hensel 2000). In support of this notion, mutations in SPI-2 TTSS result in a very significant attenuation of Salmonella virulence in a mouse model of infection regardless of the inoculation route (Ochman et al. 1996, Shea et al. 1996) and a significant defect in its ability to induce enteropathology in a cow enteritis model of infection (Tsolis et al. 1999a). Although SPI-2 TTSS mutants of S. dublin were defective at inducing intestinal pathology in a similar model of infection, the defect was not as pronounced as that exhibited by a SPI-1 TTSS mutant. Consistent with its involvement in systemic infection, expression of the SPI-2 TTSS occurs only after Salmonella has gained access into cells (Cirillo et al. 1998, Pfeifer et al. 1999). Therefore, the intracellular environment seems to be the place of action of the SPI-2 TTSS. Indeed, the phenotypes thus far associated with this TTSS are consistent with this notion. Mutations that disrupt the SPI-2 TTSS exhibit a defect in intracellular growth in macrophages, as well as in epithelial cells (Cirillo et al. 1998, Hensel et al. 1998, Ochman et al. 1996). The intracellular growth defect observed with these mutants is approximately 10-fold, which is more modest than the defect observed with other Salmonella mutations. Considering the rather strong effect of SPI-2 mutations on virulence, these results suggest that phenotypes other than intracellular growth may be mediated by SPI-2 function. In fact, the SPI-2 TTSS was recently implicated in the induction of apoptosis in macrophages (van

24 76 GALÁN Der Velden et al. 2000) and the avoidance of NADPH oxidase dependent killing (Vazquez-Torres et al. 2000). It was proposed that SPI-2 mediates the latter effect by preventing the phagocytic NADPH oxidase from trafficking to the Salmonellacontaining vacuole. Consistent with this hypothesis, NADPH oxidase was shown to co-localize with vacuoles containing SPI-2 mutants but was absent from vacuoles containing wild-type Salmonella (Vazquez-Torres et al. 2000). It is possible that the ability of Salmonella to grow intracellularly and to inhibit NADPH oxidase function is mediated by similar SPI-2 TTSS effector proteins that may interfere with cellular trafficking (see below). Modulation of Intracellular Trafficking by SPI-2 TTSS Effector Proteins Work from several laboratories indicates that Salmonella traffics along an unusual endocytic pathway, which results in the establishment of a compartment suitable for its growth (Garcia-del Portillo & Finlay 1995; Garcia-del Portillo et al. 1992, 1993, 1994; Meresse et al. 1999; Mills & Finlay 1998; Rathman et al. 1997; Steele- Mortimer et al. 2000). In non-phagocytic cells, this compartment takes the form of an unusual filamentous tubular network where Salmonella resides and replicates (Garcia-del Portillo et al. 1993). The ability of Salmonella to form this unusual compartment has been correlated with the ability of Salmonella to grow intracellularly (Garcia-del Portillo et al. 1993). Mounting evidence indicates that the unusual trafficking route followed by Salmonella, as well as its ability to induce the formation of this unusual vesicular compartment, are dependent on the function of the SPI-2 TTSS (Beuzon et al. 2000, Guy et al. 2000, Uchiya et al. 1999). Consistent with this notion, mutations that disrupt this TTSS result in bacteria that traffic along a more conventional pathway and are defective for intracellular growth. Two SPI-2 TTSS effector proteins have been implicated in this process. One of these proteins is SpiC, which is encoded within SPI-2 in the immediate proximity to genes that encode components of the secretion apparatus (Uchiya et al. 1999). Consistent with its postulated involvement in trafficking, SpiC was shown to be translocated to the cell cytosol. A Salmonella mutant defective in spic was shown to traffic along a more conventional pathway apparently leading to lysozomes (Uchiya et al. 1999). Wild-type Salmonella appears to have a more general effect on trafficking since in infected cells, the traffic of vesicles devoid of Salmonella was also shown to occur along an unconventional pathway. Transient expression of SpiC resulted in the inhibition of the endocytosis of the transferrin receptor, and purified SpiC inhibited membrane fusion in a cell-free system. These results suggest that SpiC, by itself, is capable of interfering with cellular trafficking. The mechanisms by which SpiC exerts its function, however, are not understood. Another SPI-2 TTSS effector protein implicated in altering the trafficking route of Salmonella is SifA (Guy et al. 2000, Stein et al. 1996). Originally identified as a protein involved in the formation of the Salmonella-induced filamentous vesicular structures (Stein et al. 1996), SifA was recently shown to be a substrate of the

25 TYPE III SECRETION 77 SPI-2 TTSS (Guy et al. 2000). Unlike SpiC, however, it appears that SifA is not translocated to the cell cytosol; instead, it has been postulated to exert its function at the membrane of the Salmonella-containing vesicles. A Salmonella strain carrying a loss-of-function mutation in sifa apparently breaks out of the endocytic vacuole and reaches the cell cytosol (Guy et al. 2000). Although the implications of this observation are unclear, it suggests that sifa is required for the maintenance of the Salmonella-containing vacuolar compartment rather than in the redirection of Salmonella to an unusual trafficking pathway. SpiC and SifA are most likely just the tip of the iceberg of what is likely to be a genetically complex phenotype involving several effector proteins. More studies will be required to unravel the specific contribution of each of these effector proteins and to establish their mechanism of action. CROSS-TALK BETWEEN SPI-1 AND SPI-2 TTSSs? Several lines of evidence indicate that the SPI-1 and SPI-2 TTSSs work independently. In general, mutations affecting the function of one system do not significantly affect the phenotypes mediated by the other system. In addition, their pattern of expression appears to be significantly different: while the SPI-1 TTSS is expressed during the initial interaction of Salmonella with the intestinal epithelium, the SPI-2 TTSS is expressed only after Salmonella has reached an intracellular compartment. This is consistent with a relay model in which SPI-1 TTSS first mediates the entry of Salmonella into cells thus allowing the expression of the SPI-2 TTSS, which controls events related to bacterial intracellular trafficking and growth. Despite this clear functional independence, the relay model postulates the possibility of a functional coordination between the two systems. In this context, it is possible that effectors needing to exert their function late during the functional half-life of the SPI-1 TTSS (and by definition early in the half-life of the SPI-2 TTSS) may be targeted to both these systems. Indeed, at least two effectors are candidates to fulfill this role. SspH1 has been shown to be targets of both systems and SopD has been shown to be secreted through the SPI-1 TTSS (Jones et al. 1998) but also exhibits structural features at its amino terminus present in SPI-2 secreted proteins (Miao et al. 1999). In addition, certain mutations in SPI-2 influence the expression of SPI-1 genes (Deiwick et al. 1998) suggesting the existence of a mechanism that coordinates the temporal expression patterns of these two systems. HOST MIMICRY: A PREVAILING THEME IN TTSS EFFECTOR PROTEINS The identification and mechanistic understanding of the function of several Salmonella SPI-1 TTSS effector proteins have revealed what is likely to be a dominant theme not only for this specific TTSS but, most likely, for TTSSs in other

26 78 GALÁN bacteria. This theme is one of modulation of cellular functions by bacterial products that mimic host proteins (Stebbins & Galán 2000). This strategy stands in sharp contrast to mechanisms that involve bacterial products that, although capable of modulating cellular responses, have no counterpart in the host. For example, many bacterial pathogens secrete toxins that modulate Rho GTPase function (Aktories 1997). However, most of these toxins do so by introducing covalent modifications in their target cellular proteins, including glucosylation, deamidation, ADP-ribosylation, etc, which involve enzymes with no obvious eukaryotic counterpart (Aktories 1997). Often, this type of modification leads to overt and irreversible toxicity, a strategy not well-suited for the interaction of more adapted pathogens that have sustained a long-standing close association with their hosts. In contrast to this strategy, Salmonella has evolved sophisticated mechanisms for the reversible activation of Rho GTPases that involve host mimicry. Thus SopE activates Cdc42 and Rac by working as a guanine nucleotide exchange factor, whereas SptP reverses this activation by functioning as a GAP (Fu & Galán 1999, Hardt et al. 1998b). At the molecular level, host mimicry may take two forms. One form is characterized by the utilization of direct homologs of host proteins that have been subverted for the pathogen s needs. The best example in Salmonella is that of the tyrosine phosphatase domain of SptP (Kaniga et al. 1996, Stebbins & Galán 2000). This domain exhibits amino acid sequence similarity to eukaryotic tyrosine phosphatases and the Yersinia TTSS effector protein YopH. Consistent with its specific role in Salmonella interactions, the SptP tyrosine phosphatase has unique surface properties (Stebbins & Galán 2000). In this case, however, the host mimicry can be easily inferred from the sequence homology. It can be hypothesized that Salmonella may have horizontally acquired this gene and subsequently subverted it to fit its own requirements. The other form of mimicry is less obvious as it has been sculpted by convergent evolution, which has resulted in molecules that do not share obvious sequence similarity to any known host protein. An example of this type of mimicry is the amino-terminal domain of SptP, which functionally mimics eukaryotic GAP proteins despite the lack of apparent sequence similarity to this family of proteins (Fu & Galán 1999, Stebbins & Galán 2000). Remarkably, the crystal structure of SptP reveals that, although not obvious at the sequence level, this protein shares the most fundamental elements that eukaryotic GAPs present to their host cell targets (Stebbins & Galán 2000) (Figure 6). Therefore, despite the differences in the general architecture, the actual surface presented by SptP to its cellular targets has much in common with its eukaryotic counterparts. Thus like host GAPs, SptP interacts extensively with similar residues in the regulatory Switch I and Switch II regions of the GTPase. Indeed, the Switch I and II regions of Rac-1 acquired a similar conformation when bound to SptP as when bound to cellular GAPs (Figure 6). Members of the Ras/Rho superfamily of small GTPases lack a catalytic arginine that is essential for GTP hydrolysis. Consequently, all GAPs for these GTPases insert an arginine in the catalytic site, which

27 TYPE III SECRETION 79 drastically increases the intrinsic GTPase activity of the target GTPase (Scheffzek et al. 1998). In analogy to host GAPs, SptP inserts an arginine residue into the active site of Rac1 in a manner that is nearly identical to host enzymes (Stebbins & Galán 2000). Unlike host GAPs, however, the critical arginine in SptP is not located in a flexible loop, but instead extends from an α helix. Thus SptP mimics host GAPs with a different tertiary structure but still achieves a precise chemical mimicry of the host proteins. It is likely that both types of host mimicry will be found in other TTSSs effector proteins once their structures are solved. Several TTSSs effectors do not share any apparent sequence similarity to host proteins despite the fact that functionally they behave like eukaryotic protein counterparts. Examples of this are SopE, which functions as an exchange factor for Rho GTPases (Hardt et al. 1998a), and SipA, which is an actin-binding protein that modulates actin dynamics (Zhou et al. 1999a). These proteins are candidates to be host mimics that have perhaps evolved by convergent evolution. Other TTSSs effectors do show sequence similarity to host cell counterparts, potentially representing additional examples of genes that may have been acquired from an eukaryotic host and later adapted for pathogenic functions. An example of this category is SopB, an inositol phosphate phosphatase that contains the highly conserved P-loop motif, xcxxgxxr(t/s)g, found in tyrosine and dual-specificity protein phosphatases, as well as in the PTENlike family of lipid phosphatases (Norris et al. 1998). However, SopB possesses no other detectable homology to these enzymes, an indication of a significant diversion as a result of the evolutionary process that led to its adaptation for its function in Salmonella. Undoubtedly, the solution of the structure of these effector proteins promises to illuminate important aspects of microbial pathogen evolution, as well as provide a unique view at bacterial enzymes with eukaryotic counterparts. SUMMARY AND PERSPECTIVES The last few years have seen remarkable progress in the understanding of the interaction of Salmonella with its host. We have learned that through its type III secretion systems, this bacterium engages the cell in a remarkable biochemical interaction that is better characterized by its refinement than by its harmful consequences. Salmonella has shaped a remarkably sophisticated functional interface with its host, the details of which we are just beginning to decipher. Although we understand this pathogen better than perhaps any other, it is also evident that we have only seen the tip of the iceberg. The next few years promise to be as exciting as the last few. Most of the TTSS-associated genes involved in Salmonella host interactions are now in hand. The challenge remains in understanding their function. The study of molecular and functional details of the Salmonella-host cell interactions will not only lead to new therapeutic and prevention avenues but will also continue to teach us about the inner workings of the cell.

28 80 GALÁN ACKNOWLEDGMENTS I thank members of the Galán laboratory for critical review of this manuscript. I also thank present and past members of my laboratory who have contributed to some of the work discussed in this article. Work in my laboratory is supported by Public Health Service Grants AI30492, GM52543 and AI46953 from the National Institutes of Health. Visit the Annual Reviews home page at LITERATURE CITED Aktories K Bacterial toxins that target Rho proteins. J. Clin. Invest. 99: Arnold JW, Niesel DW, Annable CR, Hess CB, Asuncion M, et al Tumor necrosis factor-alpha mediates the early pathology in Salmonella infection of the gastrointestinal tract. Microb. Pathog. 14: Bakshi CS, Singh VP, Wood MW, Jones PW, Wallis TS, Galyov EE Identification of SopE2: a Salmonella secreted protein which is highly homologous to SopE and involved in bacterial invasion of epithelial cells. J. Bacteriol. 182: Beuzon CR, Banks G, Deiwick J, Hensel M, Holden DW ph-dependent secretion of SseB, a product of the SPI-2 type III secretion system of Salmonella typhimurium. Mol, Microbiol, 33: Beuzon CR, Meresse S, Unsworth KE, Ruiz- Albert J, Garvis S Waterman SR, et al Salmonella maintains the integrity of its intracellular vacuole through the action of SifA. EMBO J. 19: Blocker A, Gounon P, Larquet E, Niebuhr K, Cabiaux V, et al The tripartite type III secreton of Shigella flexneri inserts IpaB and IpaC into host membranes. J. Cell Biol. 147: Brennan MA, Cookson BT Salmonella induces macrophage death by caspase-1-dependent necrosis. Mol. Microbiol. 38:31 40 Buckholm G Effect of cytochalasin B and dihidrocytochalasin B on invasiveness of enteroinvasive bacteria in Hep-2 cell cultures. Acta Path. Microbiol. Immun. Scand. 92: Burbelo PD, Drechsel D, Hall A A conserved binding motif defines numerous candidate target proteins for both Cdc42 and Rac GTPases. J. Biol. Chem. 270: Chen H, Bernstein BW, Bamburg JR Regulating actin-filament dynamics in vivo. Trends Biochem. Sci. 25:19 23 Chen LM, Bagrodia S, Cerione RA, Galán JE Requirement of p21 activated kinase (PAk) for Salmonella typhimurium-induced nuclear responses. J. Exp. Med. 189: Chen LM, Hobbie S, Galán JE. 1996a. Requirement of CDC42 for Salmonella typhimurium-induced cytoskeletal reorganization and nuclear responses in cultured cell. Science 274: Chen LM, Kaniga K, Galán JE. 1996b. Salmonella spp. are cytotoxic for cultured macrophages. Mol. Microbiol. 21: Chen Y, Smith MR, Thirumalai K, Zychlinsky A A bacterial invasin induces macrophage apoptosis by binding directly to ICE. EMBO J. 15: Cheng LW, Schneewind O Type III machines of gram-negative bacteria: delivering the goods. Trends Microbiol. 8: Cirillo DM, Valdivia RH, Monack DM, Falkow S Macrophage-dependent induction of the Salmonella pathogenicity island 2 type III secretion system and its role in intracellular survival. Mol. Microbiol. 30: Collazo C, Galán JE Requirement of ex-

29 TYPE III SECRETION 81 ported proteins for secretion through the invasion-associated Type III system in Salmonella typhimurium. Infect. Immun. 64: Collazo C, Galán JE The invasion-associated type III system of Salmonella typhimurium directs the translocation of Sip proteins into the host cell. Mol. Microbiol. 24: Collazo CM, Zierler MK, Galán JE Functional analysis of the Salmonella typhimurium invasion genes invi and invj and identification of a target of the protein secretion apparatus encoded in the inv locus. Mol. Microbiol. 15:25 38 Cooper JA, Schafer DA Control of actin assembly and disassembly at filament ends. Curr. Opin. Cell Biol. 12: Cornelis GR, Van Gijsegem F Assembly and function of type III secretory systems. Annu. Rev. Microbiol. 54: Darwin KH, Miller VL. 1999a. InvF is required for expression of genes encoding proteins secreted by the SPI1 type III secretion apparatus in Salmonella typhimurium. J. Bacteriol. 181: Darwin KH, Miller VL. 1999b. Molecular basis of the interaction of Salmonella with the intestinal mucosa. Clin. Microbiol. Rev. 12: Day DW, Mandal BK, Morrson BC The rectal biopsy appearances in Salmonella colitis. Histopathology 2: Deiwick J, Nikolaus T, Shea JE, Gleeson C, Holden DW, Hensel M Mutations in Salmonella pathogenicity island 2 (SPI2) genes affecting transcription of SPI1 genes and resistance to antimicrobial agents. J. Bacteriol. 180: Eckmann L, Kagnoff MF, Fierer J Epithelial cells secrete the chemokine interleukin-8 in response to bacterial entry. Infect. Immun. 61: Eckmann L, Rudolf MT, Ptasznik A, Schultz C, Jiang T, et al D-myo-inositol 1,4,5, 6 tetrakisphosphate produced in human intestinal epithelial cells in response to Salmonella invasion inhibits phosphoinositide 3-kinase signaling pathways. Proc. Natl. Acad. Sci. USA 94: Eichelberg K, Galán JE Differential regulation of Salmonella typhimurium type III secreted proteins by pathogenicity island 1 (SPI-1)-encoded transcriptional activators InvF and hila. Infect. Immun. 67: Finlay BB, Ruschkowski S Cytoskeletal rearrangements accompanying Salmonella entry into epithelial cells. J. Cell. Sci. 99: Francis CL, Ryan TA, Jones BD, Smith SJ, Falkow S Ruffles induced by Salmonella and other stimuli direct macropinocytosis of bacteria. Nature 364: Fu Y, Galán JE The Salmonella spp. protein tyrosine phosphatase SptP is translocated into host cells and disrupts the host-cell cytoskeleton. Mol. Microbiol. 27: Fu Y, Galán JE A Salmonella protein antagonizes Rac-1 and Cdc42 to mediate hostcell recovery after bacterial invasion. Nature 401: Galán JE Interaction of Salmonella with host cells through the centrisome 63 type III secretion system. Curr. Opin. Microbiol. 2:46 50 Galán JE, Collmer A Type III secretion machines: bacterial devices for protein delivery into host cells. Science 284: Galán JE, Curtiss R III Cloning and molecular characterization of genes whose products allow Salmonella typhimurium to penetrate tissue culture cells. Proc. Natl. Acad. Sci. USA 86: Galán JE, Pace J, Hayman MJ Involvement of the epidermal growth factor receptor in the invasion of the epithelial cells by Salmonella typhimurium. Nature 357: Galán JE, Zhou D Striking a balance: modulation of the actin cytoskeleton by Salmonella. Proc. Natl. Acad. Sci. USA 97: Galyov EE Wood MW, Rosqvist R, Mullan PB, Watson PR, et al A secreted effector protein of Salmonella dublin is translocated into eukaryotic cells and mediates in-

30 82 GALÁN flammation and fluid secretion in infected ileal mucosa. Mol. Microbiol. 25: Garcia-del Portillo F, Finlay BB Salmonella invasion of nonphagocytic cells induces formation of macropinosomes in the host cell. Infect. Immun. 62: Garcia-del Portillo F, Finlay BB Targeting of Salmonella typhimurium to vesicles containing lysosomal membrane glycoproteins bypasses compartments with mannose 6-phosphate receptors. J. Cell Biol. 129:81 97 Garcia-del Portillo F, Foster JW, Maguire ME, Finlay BB Characterization of the micro-environment of Salmonella typhimurium-containing vacuoles within MDCK epithelial cells. Mol. Microbiol. 6: Garcia-del Portillo F, Picciarelli MG, Jefferies WA, Finlay BB Salmonella typhimurium induces selective aggregation and internalization of host cell surface proteins during invasion of epithelial cells. J. Cell Sci. 107: Garcia-del Portillo F, Zwick MB, Leung KY, Finlay BB Salmonella induces the formation of filamentous structures containing lysosomal membrane glycoproteins in epithelial cells. Proc. Natl. Acad. Sci. USA 90: Giannella RA Importance of the intestinal inflammatory reaction in Salmonellamediated intestinal secretion. Infect. Immun. 23: Giannella RA, Washington O, Gemski P, Formal SB Invasion of HeLa cells by S. typhimurium: a study of invasiveness of Salmonella. J. Infect. Dis. 128:69 75 Ginocchio C, Olmsted SB, Wells CL, Galán JE Contact with epithelial cells induces the formation of surface appendages on Salmonella typhimurium. Cell 76: Ginocchio C, Pace J, Galán JE Identification and molecular characterization of a Salmonella typhimurium gene involved in triggering the internalization of Salmonellae into cultured epithelial cells. Proc. Natl. Acad. Sci. USA 89: Gopalakrishnan S, Raman N, Atkinson SJ, Marrs JA Rho GTPase signaling regulates tight junction assembly and protects tight junctions during ATP depletion. Am. J. Physiol. Cell Physiol. 275:C798 C809 Guichon A, Hersh D, Smith MR, Zychlinsky A Structure-function analysis of the Shigella virulence factor IpaB. J. Bacteriol. 183: Guy RL, Gonias LA, Stein MA Aggregation of host endosomes by Salmonella requires SPI2 translocation of SseFG and involves SpvR and the fms-aroe intragenic region. Mol. Microbiol. 37: Hall A Rho GTPases and the actin cytoskeleton. Science 279: Hardt W-D, Chen L-M, Schuebel KE, Bustelo XR, Galán JE. 1998a. Salmonella typhimurium encodes an activator of Rho GTPases that induces membrane ruffling and nuclear responses in host cells. Cell 93: Hardt W-D, Galán JE A secreted Salmonella protein with homology to an avirulence determinant of plant pathogenic bacteria. Proc. Natl. Acad. Sci. USA 94: Hardt W-D, Urlaub H, Galán JE. 1998b. A target of the centisome 63 type III protein secretion system of Salmonella typhimurium is encoded by a cryptic bacteriophage. Proc. Natl. Acad. Sci. USA 95: Hayward RD, Koronakis V Direct nucleation and bundling of actin by the SipC protein of invasive Salmonella. EMBO J. 18: Hensel M Salmonella pathogenicity island 2. Mol. Microbiol. 36: Hensel M, Shea JE, Gleeson C, Jones MD, Dalton E, Holden DW Simultaneous identification of bacterial virulence genes by negative selection. Science 269:400 3 Hensel M, Shea JE, Waterman SR, Mundy R, Nikolaus T, et al Genes encoding putative effector proteins of the type III secretion system of Salmonella pathogenicity island 2 are required for bacterial virulence and proliferation in macrophages. Mol. Microbiol. 30: Hersh D, Monack DM, Smith MR, Ghori N, Falkow S, Zychlinsky A The Salm-

31 TYPE III SECRETION 83 onella invasin SipB induces macrophage apoptosis by binding to caspase-1. Proc. Natl. Acad. Sci. USA 96: Hess CB, Niessel DW, Klimpel GR The induction of interferon production in fibroblasts by invasive bacteria: a comparison of Salmonella and Shigella species. Microb. Pathog. 7: Hobbie S, Chen LM, Davis R, Galán JE Involvement of the mitogen-activated protein kinase pathways in the nuclear responses and cytokine production induced by Salmonella typhimurium in cultured intestinal cells. J. Immunol. 159: Hong KH, Miller VL Identification of a novel Salmonella invasion locus homologous to Shigella ipgde. J. Bacteriol. 180: Hueck CJ Type III protein secretion systems in bacterial pathogens of animals and plants. Microbiol. Mol. Biol. Rev. 62: Jepson MA, Collares-Buzato CB, Clark MA, Hirst BH, Simmons NL Rapid disruption of epithelial barrier function by Salmonella typhimurium is associated with structural modification of intercellular junctions. Infect. Immun. 63: Jepson MA, Schlecht HB, Collares-Buzato CB Localization of dysfunctional tight junctions in Salmonella enterica serovar Typhimurium-infected epithelial layers. Infect. Immun. 68: Jesenberger V, Procyk KJ, Yuan J, Reipert S, Baccarini M Salmonella-induced caspase-2 activation in macrophages. A novel mechanism in pathogen-mediated apoptosis. J. Exp. Med. 192: Joberty G, Petersen C, Gao L, Macara IG The cell-polarity protein Par6 links Par3 and atypical protein kinase C to Cdc42. Nat. Cell Biol. 2: Jones MA, Wood MW, Mullan PB, Watson PR, Wallis TS, Galyov EE Secreted effector proteins of Salmonella dublin act in concert to induce enteritis. Infect. Immun. 66: Jung HC, Eckmann L, Yang S-K, Panja A, Fierer J, et al A distinct array of proinflammatory cytokines is expressed in human colon epithelial cells in response to bacterial invasion. J. Clin. Invest. 95:55 65 Kaniga K, Bossio JC, Galán JE The Salmonella typhimurium invasion genes invf and invg encode homologues to the PulD and AraC family of proteins. Mol. Microbiol. 13: Kaniga K, Uralil J, Bliska JB, Galán JE A secreted tyrosine phosphatase with modular effector domains encoded by the bacterial pathogen Salmonella typhimurium. Mol. Microbiol. 21: Kimbrough TG, Miller SI Contribution of Salmonella typhimurium type III secretion components to needle complex formation. Proc. Natl. Acad. Sci. USA 97: Kingsley RA, Baumler AJ Host adaptation and the emergence of infectious disease: the Salmonella paradigm. Mol. Microbiol. 36: Kubori T, Matsushima Y, Nakamura D, Uralil J, Lara-Tejero M, et al Supramolecular structure of the Salmonella typhimurium type III protein secretion system. Science 280:602 5 Kubori T, Sukhan A, Aizawa S-I, Galán JE Molecular characterization and assembly of the needle complex of the Salmonella typhimurium type III protein secretion system. Proc. Natl. Acad. Sci. USA 97: Lamarche N, Tapon N, Stowers L, Burbelo PD, Aspenstrom P, et al Rac and Cdc42 induce actin polymerization and G1 cell cycle progression independently of p65pak and the JNK/SAPK MAP kinase cascade. Cell 87: Lee CA, Silva M, Siber AM, Kelly AJ, Galyov E, McCormick BA A secreted Salmonella protein induces a pro-inflammatory response in epithelial cells, which promotes neutrophil migration. Proc. Natl. Acad. Sci. USA 97: Lloyd SA, Norman M, Rosqvist R, Wolf-

32 84 GALÁN Watz H Yersinia YopE is targeted for type III secretion by N-terminal, not mrna, signals. Mol. Microbiol. 39: Lucas RL, Lee CA Unravelling the mysteries of virulence gene regulation in Salmonella typhimurium. Mol. Microbiol. 36: McCormick BA, Colgan SP, Delp-Archer C, Miller SI, Madara JL Salmonella typhimurium attachment to human intestinal epithelial monolayers: transcellular signalling to subepithelial neutrophils. J. Cell Biol. 123: McGovern VJ, Slavutin LJ Pathology of Salmonella colitis. Am. J. Surg. Pathol. 3: Medzhitov R, Janeway C. 2000a. Innate immunity. N. Engl. J. Med. 343: Medzhitov R, Janeway C. 2000b. The toll receptor family and microbial recognition. Trends Microbiol. 8: Meresse S, Steele-Mortimer O, Finlay BB, Gorvel JP The rab7 GTPase controls the maturation of Salmonella typhimuriumcontaining vacuoles in HeLa cells. EMBO J. 18: Miao EA, Scherer CA, Tsolis RM, Kingsley RA, Adams LG, et al Salmonella typhimurium leucine-rich repeat proteins are targeted to the SPI1 and SPI2 type III secretion systems. Mol. Microbiol. 34: Mills DB, Bajaj V, Lee CA A 40 kilobase chromosomal fragment encoding Salmonella typhimurium invasion genes is absent from the corresponding region of the Escherichia coli K-12 chromosome. Mol. Microbiol. 15: Mills SD, Finlay BB Isolation and characterization of Salmonella typhimurium and Yersinia pseudotuberculosis-containing phagosomes from infected mouse macrophages: Y. pseudotuberculosis traffics to terminal lysosomes where they are degraded. Eur. J. Cell Biol. 77:35 47 Mirold S, Rabsch W, Rohde M, Stender S, Tschape H, et al Isolation of a temperate bacteriophage encoding the type III effector protein SopE from an epidemic Salmonella typhimurium strain. Proc. Natl. Acad. Sci. USA 96: Mitra K, Zhou D, Galán JE Biophysical characterization of SipA, an actin-binding protein from Salmonella enterica. FEBS Lett. 482:81 84 Mittrucker HW, Kaufmann SH Immune response to infection with Salmonella typhimurium in mice. J. Leukoc. Biol. 67: Monack DM, Hersh D, Ghori N, Bouley D, Zychlinsky A, Falkow S Salmonella exploits caspase-1 to colonize Peyer s patches in a murine typhoid model. J. Exp. Med. 192: Monack DM, Raupach B, Hromockyj AE, Falkow S Salmonella typhimurium invasion induces apoptosis in infected macrophages. Proc. Natl. Acad. Sci. USA 93: Norris FA, Wilson MP, Wallis TS, Galyov EE, Majerus PW SopB, a protein required for virulence of Salmonella dublin, is an inositol phosphate phosphatase. Proc. Natl. Acad. Sci. USA 95: Ochman H, Groisman EA Distribution of pathogenicity islands in Salmonella spp. Infect. Immun. 64: Ochman H, Soncini FC, Solomon F, Groisman EA Identification of a pathogenicity island required for Salmonella survival in host cells. Proc. Natl. Acad. Sci. USA 93: Ohl ME, Miller SI Salmonella: a model for bacterial pathogenesis. Annu. Rev. Med. 52: Orth K, Xu Z, Mudgett MB, Bao ZQ, Palmer LE, et al Disruption of signaling by Yersinia effector YopJ, a ubiquitin-like protein protease. Science 290: Pace J, Hayman MJ, Galán JE Signal transduction and invasion of epithelial cells by Salmonella typhimurium. Cell 72: Peppelenbosch MP, Qiu RG, de Vries-Smits AM, Tertoolen LG, de Laat SW, et al Rac mediates growth factor-induced arachidonic acid release. Cell 81:849 56

33 TYPE III SECRETION 85 Perna NT Genome sequence of enterohaemorrhagic Escherichia coli O157:H7. Nature 409: Pfeifer CG, Marcus SL, Steele-Mortimer O, Knodler LA, Finlay BB Salmonella typhimurium virulence genes are induced upon bacterial invasion into phagocytic and nonphagocytic cells. Infect. Immun. 67: Procyk KJ, Kovarik P, von Gabain A, Baccarini M. 1999a. Salmonella typhimurium and lipopolysaccharide stimulate extracellularly regulated kinase activation in macrophages by a mechanism involving phosphatidylinositol 3-kinase and phospholipase D as novel intermediates. Infect. Immun. 67: Procyk KJ, Rippo MR, Testi R, Hoffmann F, Parker PJ, Baccarini M. 1999b. Distinct mechanisms target stress and extracellular signal-activated kinase 1 and Jun N-terminal kinase during infection of macrophages with Salmonella. J. Immunol. 163: Rathman M, Barker LP, Falkow S The unique trafficking pattern of Salmonella typhimurium-containing phagosomes in murine macrophages is independent of the mechanism of bacterial entry. Infect. Immun. 65: Rosenshine I, Ruschkowski S, Foubister V, Finlay BB Salmonella typhimurium invasion of epithelial cells: role of the induced host cell tyrosine protein phosphorylation. Infect. Immun. 62: Rout WR, Formal SB, Dammin GJ, Giannella RA Pathophysiology of Salmonella diarrhoea in the Rhesus monkey: intestinal transport, morphological and bacteriological studies. Gastroenterology 67:59 70 Rudolph MG, Weise C, Mirold S, Hillenbrand B, Bader B, et al Biochemical analysis of SopE from Salmonella typhimurium, a highly efficient guanosine nucleotide exchange factor for RhoGTPases. J. Biol. Chem. 274: Ruschkowski S, Rosenshine I, Finlay BB Salmonella typhimurium induces an inositol phosphate flux in infected epithelial cells. FEMS Lett. 74: Scheffzek K, Ahmadian MR, Wittinghofer A GTPase-activating proteins: helping hands to complement an active site. Trends Biochem. 23: Schesser K, Dukuzumuremyi JM, Cilio C, Borg S, Wallis TS, et al The Salmonella YopJ-homologue AvrA does not possess YopJ-like activity. Microb. Pathog. 28:59 70 Shea JE, Hensel M, Gleeson C, Holden DW Identification of a virulence locus encoding a second type III secretion system in Salmonella typhimurium. Proc. Natl. Acad. Sci. USA, 93: Snapper SB, Rosen FS The Wiskott- Aldrich syndrome protein (WASP): roles in signaling and cytoskeletal organization. Annu. Rev. Immunol. 17: Stebbins CE, Galán JE Modulation of host signaling by a bacterial mimic. structure of the Salmonella effector SptP bound to Rac1. Mol. Cell 6: Steele-Mortimer O, St-Louis M, Olivier M, Finlay BB Vacuole acidification is not required for survival of Salmonella enterica serovar Typhimurium within cultured macrophages and epithelial cells. Infect. Immun. 68: Stein MA, Leung KY Zwick M, Garcia-del Portillo F, Finlay BB Identification of a Salmonella virulence gene required for formation of filamentous structures containing lysosomal membrane glycoproteins within epithelial cells. Mol. Microbiol. 20: Stender S, Friebel A, Linder S, Rohde M, Mirold S, Hardt W-D Identification of SopE2 from Salmonella typhimurium, a conserved guanine nucleotide exchange factor for Cdc42 of the host cell. Mol. Microbiol. 36: Sukhan A, Kubori T, Wilson J, Galán JE Genetic analysis of assembly of the Salmonella typhimurium type III-associated needle complex. J. Bacteriol. 183: Takeuchi A Electron microscopic studies of experimental Salmonella infection.

34 86 GALÁN 1. Penetration into the intestinal epithelium by Salmonella typhimurium. Am. J. Pathol. 50: Takeuchi A, Formal SB, Spring H Experimental acute colitis in the Rhesus monkey following peroral infection with Shigella flexneri. Am. J. Pathol. 52: Tamano K, Aizawa S, Katayama E, Nonaka T, Imajoh-Ohmi S, et al Supramolecular structure of the Shigella type III secretion machinery: the needle part is changeable in length and essential for delivery of effectors. EMBO J. 19: Tsolis RM, Adams LG, Ficht TA, Baumler AJ. 1999a. Contribution of Salmonella typhimurium virulence factors to diarrheal disease in calves. Infect. Immun. 67: Tsolis RM, Townsend SM, Miao EA, Miller SI, Ficht TA, et al. 1999b. Identification of a putative Salmonella enterica serotype typhimurium host range factor with homology to IpaH and YopM by signature-tagged mutagenesis. Infect. Immun. 67: Uchiya K, Barbieri MA, Funato K, Shah AH, Stahl PD, Groisman EA A Salmonella virulence protein that inhibits cellular trafficking. EMBO J. 18: Van Aelst L, D Souza-Schorey C Rho GTPases and signaling networks. Genes Dev. 11: van Der Velden AW, Lindgren SW, Worley MJ, Heffron F Salmonella pathogenicity island 1-independent induction of apoptosis in infected macrophages by Salmonella enterica serotype typhimurium. Infect. Immun. 68: Vazquez-Torres A, Xu Y, Jones-Carson J, Holden DW, Lucia SM, et al Salmonella pathogenicity island 2-dependent evasion of the phagocyte NADPH oxidase. Science 287: Wallis TS, Galyov EE Molecular basis of Salmonella-induced enteritis. Mol. Microbiol. 36: Watson PR, Gautier AV, Paulin SM, Bland AP, Jones PW, Wallis TS Salmonella enterica serovars Typhimurium and Dublin can lyse macrophages by a mechanism distinct from apoptosis. Infect. Immun. 68: Whalen MC, Wang JF, Carland FM, Heiskell ME, Dahlbeck D, et al Avirulence gene avrrxv from Xanthomonas campestris pv. vesicatoria specifies resistance on tomato line Hawaii Mol. Plant Microbe Interact. 6: Wood MW, Jones MA, Watson PR, Hedges S, Wallis TS, Galyov EE Identification of a pathogenicity island required for Salmonella enteropathogenicity. Mol. Microbiol. 29: Wood MW, Jones MA, Watson PR, Siber AM, McCormick BA, et al The secreted effector protein of Salmonella dublin, SopA, is translocated into eukaryotic cells and influences the induction of enteritis. Cell. Microbiol. 2: Wood MW, Rosqvist R, Mullan PB, Edwards MH, Galyov EE SopE, a secreted protein of Salmonella dublin, is translocated into the target eukaryotic cell via a sip-dependent mechanism and promotes bacterial entry. Mol. Microbiol. 22: Zhou D, Chen L-M, Hernandez L, Shears SB, Galán JE A Salmonella inositol polyphosphatase acts in conjunction with other bacterial effectors to promote host-cell actin cytoskeleton rearrangements and bacterial internalization. Mol. Microbiol. 39: Zhou D, Mooseker M, Galán JE. 1999a. Role of the S. typhimurium actin-binding protein SipA in bacterial internalization. Science 283: Zhou D, Mooseker MS, Galán JE. 1999b. An invasion-associated Salmonella protein modulates the actin-bundling activity of plastin. Proc. Natl. Acad. Sci. USA 96:

35 Figure 3 Interaction of Salmonella with epithelial cells. (A) Electron micrograph of a polarized intestinal epithelial cell infected with S. typhimurium (Ginocchio et al. 1994). Note the disruption of the microvilli at the point of bacterial/cell contact. Scale bar: 8 µm. (B) Photomicrograph showing actin cytoskeleton rearrangements (arrows) stimulated by S. typhimurium in epithelial cells (Zhou et al. 2001). Scale bar: 20 µm.

36 Figure 6 Comparisons of GAP enzymes and their interactions with Rho GTPases. Structures of transition state complexes between small GTPases and their cognate GAPs. (A) The three GAPs are seen to engage the small G proteins in a similar fashion, primarily contacting the Switch I (orange), Switch II (red), and nucleotide regions (green) of these proteins. (B) Alignment of the active sites of these complexes with the nucleotide and catalytic arginine. Despite employing a similar chemistry to the host GAPs, SptP (blue) presents the key catalytic arginine from a completely different protein architecture (Stebbins & Galan 2000).