Intestinal Innate Immunity and the Pathogenesis of Salmonella Enteritis

Abstract

Acute gastroenteritis caused by Salmonella typhimurium infection is a clinical problem with significant public health impact. The availability of several experimental models of this condition has allowed detailed investigation of the cellular and molecular interactions involved in its pathogenesis. Such studies have shed light on the roles played by bacterial virulence factors and host innate immune mechanisms in the development of intestinal inflammation.

Introduction and Overview

The Gram-negative bacterial pathogen Salmonella typhimurium is one of the most frequent causes of acute gastroenteritis in humans, and is responsible for significant morbidity, and some mortality, worldwide (1). It is also one of the few microorganisms that has been used in a well-documented attempt at causing widespread illness for political purposes (2). This incident, involving a religious cult in Oregon, illustrates the potential use of Salmonella in bioterrorism, and provides additional motivation for understanding the pathogenesis of the diseases caused by this organism.

Salmonella enteritis is the outcome of the acute intestinal inflammation that results from the triggering of host innate immune responses by the pathogen. These responses are protective, and help to clear the infection over the course of 3–4 d in most cases. However, the inflammation causes disturbances of gastrointestinal function that manifest clinically as abdominal pain, nausea, vomiting, and diarrhea (3). Related serotypes of Salmonella, S. typhi and S. paratyphi, cause the systemic illness typhoid fever rather than gastroenteritis. The pathogenesis of this disease has been reviewed elsewhere (4) and will not be considered further here.

S. typhimurium usually infects the host orally following ingestion of contaminated water or food. Spread of the infection is facilitated by the ability of the organism to survive for several days in groundwater, pond water, or seawater, and for months in contaminated food material like eggs and oysters (4–7). A key aspect of pathogenesis is the active bacterial invasion of intestinal epithelial cells (IECs), particularly those that overlie the Peyer's patches of the small intestine. This process is mediated by Salmonella effector proteins that are introduced into the host cells through a specialized secretory apparatus, and is accompanied by the production of several epithelium-derived proinflammatory molecules (8). Once the bacteria have crossed the epithelial monolayer, they interact with resident lamina propria macrophages, and with circulating monocytes and neutrophils that are recruited to the site of infection by the initial invasion of IECs. These cells also respond by secreting cytokines, chemokines, and other soluble mediators that amplify the inflammation. The recruited phagocytes, along with local defense mechanisms, help in the elimination of the bacteria, while other cells such as dendritic cells and lymphocytes initiate the steps leading to adaptive immunity.

In this review, we will discuss what is currently known about the cellular and molecular processes involved in the intestinal innate immune response to S. typhimurium and how they contribute to the pathogenesis of acute enteritis. We will focus mainly on the interactions between the bacteria and epithelial cells, macrophages, and neutrophils, because these are the cells that participate in the earliest stages of the response. Before proceeding, however, we will digress briefly to consider the experimental models that have been used to study Salmonella enteritis.

Experimental Models of Salmonella-Induced Intestinal Inflammation

Apart from volunteer studies conducted to determine the minimal pathogenic dose (9), there have been few, if any, investigations of Salmonella pathogenesis in humans at the cellular and molecular level. Researchers have had to rely instead on the use of model systems that are amenable to detailed dissection. In vitro tissue culture studies, involving the infection of cultured cells of various types, including polarized monolayers of the T84 human colonic cell line, have yielded important insights into the functions of the pathogen and host molecules required for cell invasion (10). Such studies have also provided information on the cellular inflammatory mediators that are released in response to infection. However, the reductionist nature of this approach places a significant limitation on understanding the complex interplay of molecules and cells that occurs in vivo. Accordingly, investigators have turned to animal models of salmonellosis. Cattle are natural hosts for S. typhimurium, and infected calves develop an acute gastroenteritis that resembles the human disease both clinically and histopathologically (11). Infection of whole animals, or of isolated calf intestinal loops, has thus proved quite useful in examining the role of various Salmonella molecules involved in the induction of intestinal inflammation. In general, these studies have confirmed observations made in tissue culture. Unfortunately, the difficulties of large animal husbandry, together with high cost and the inability to manipulate host cellular and immunologic factors, have made the bovine model usable only in restricted circumstances. Salmonella enteritis can also be modeled in Rhesus monkeys, but this approach suffers from disadvantages similar to infection in cattle and has not been used extensively (12,13). A recent advance in the field was made when it was discovered that oral streptomycin pre-treatment of C57BL/6 mice followed by oral infection with a streptomycin-resistant strain of S. typhimurium resulted in a robust and reproducible large intestinal inflammation (14). This enteritis, which is most marked in the cecum, has several of the pathologic hallmarks of acute Salmonella gastroenteritis in humans, including neutrophil and mononuclear infiltration, and epithelial damage. The major difference from the human disease is the lack of a significant secretory response, and the infected mice rarely, if ever, develop diarrhea. Despite this shortcoming, this model is convenient, relatively inexpensive, and allows the detailed study and manipulation of both bacterial and host factors involved in disease pathogenesis. The ability to use genetically engineered mouse strains deficient in specific aspects of the immune response is a particular advantage of this experimental system. Its usefulness has already been illustrated by several informative studies (15–20) and is likely to make it the model of choice for future investigations.

There is, in general, a satisfying concordance of findings from the various models used to study Salmonella infection experimentally. There are discordant results, however, that could be related to variations in methodology, bacterial strain, host cell type, or real differences in biology, and we will attempt to point out these discrepancies where appropriate.

Salmonella Epithelial Interactions: The Initiation of the Inflammatory Response

Colonization of the intestine is an essential pre-requisite to Salmonella infection. It takes only 100–1000 colony forming units of S. typhimurium to cause disease in humans (9). These bacteria have to survive and proliferate in a microenvironment that is usually hostile to incoming invaders because of the presence of large numbers of commensal microorganisms and the action of host immune defenses. Successful colonization depends on the abil ity to resist mucosal anti-microbial mechanisms, compete with other bacteria for space and nutrients, and adhere to the apical surface of IECs. The importance of this step of the infection process is suggested by the fact that without the streptomycin pre-treatment step that is routinely used in the mouse model of Salmonella enteritis, intestinal colonization by the pathogen is insufficient to cause intestinal inflammation (14). The antibiotic is presumed to alter the commensal flora and create a niche in which Salmonella can survive. The existence of Salmonella mutants that have abnormal intestinal colonization characteristics also points to specific molecular interactions that must occur for this process to be successful (21–24). Further study of the bacterial and host factors involved in intestinal colonization and their role in enteritis may suggest new ways to interrupt pathogenesis at a very early stage. It should be mentioned that the oral infecting dose required to cause disease in humans is much lower than in mice (9,14), so caution is required when extrapolating from results obtained with the mouse model.

Contact between infecting Salmonella and the intestinal epithelium initiates a series of events that represents, in effect, a battle between the host and the pathogen. On the bacterial side, mechanisms for invasion and intracellular survival are activated, while on the epithelial side several defensive measures are deployed, including local production of anti-microbial factors, and the recruitment of back-up troops in the form of phagocytes and other cells of the immune system. Analysis of the molecular details of these interactions is an active area of research, one that has produced a wealth of information on both Salmonella biology and mucosal immunity. Broadly speaking, there are two main pathways by which anti-microbial inflammatory responses are activated during Salmonella–epithelial interactions. One is a by-product of bacterial invasion, while the other involves specific sensing of microbial components by host pattern recognition receptors (PRRs).

Epithelial Invasion and the Activation of IEC Pro-inflammatory Signals

The ability to actively invade IECs is an important aspect of Salmonella virulence. It depends on the concerted action of a battery of bacterial proteins, one key component of which is a specialized secretory apparatus, a type III secretion system (TTSS), encoded by genes located on Salmonella pathogenicity island 1 (SPI1) (8). The SPI1 TTSS is a multi-subunit needle-like structure that protrudes from the bacterial surface and penetrates the plasma membrane of the cell that is being invaded. Several bacterial proteins, many of them also encoded by SPI1 genes, are actively secreted through the needle apparatus into the host cytosol with the assistance of a translo-case complex made up of the three Salmonella proteins, SipB, SipC, and SipD. The main function of the translocated proteins, which are sometimes called effector proteins, is to induce rearrangements of the host actin cytoskeleton at the site of cell–bacterial interaction, an activity in which the bacterial gua-nine nucleotide exchange factors (GEFs) SopE and SopE2, as well as the actin binding proteins SipA and SipC, play important roles. The cytoskeletal alterations lead to membrane ruffling and the formation of lamellapodial extensions that engulf the bacterium and ultimately encompass it within a membrane-bound intra-cellular compartment known as the Salmonella-containing vacuole (SCV). The highly regulated activity of Salmonella SptP, a dual function phosphatase and GTPase activating protein (GAP), restores the basal state of the actin cytoskeleton and plasma membrane (25).

Some of the Salmonella effectors that are involved in cell invasion also activate intra-cellular signals that result in changes in IEC gene expression and physiology. SopE and SopE2 activate the cellular Rho GTPases Cdc42 and Rac1 to induce actin polymerization, but these interactions also result in the activation of signaling pathways leading to the nuclear translocation of the transcription factor NF-κB and the resultant upregulation of genes involved in immune and inflammatory responses (26). One outcome is the baso-lateral secretion of various soluble mediators such as IL-8 that recruit neutrophils and other cells from the circulation into the intestinal lamina propria. Experiments carried out in our laboratory suggest that signals triggered by SopE2 act at a step of NF-κB activation that is downstream of the induced degradation of the inhibitor of this transcription factor, IκBα, and that such signals may interact synergistically with those initiated by Toll-like receptor (TLR) 5 (see below) (27).

Neutrophils recruited to the intestinal lamina propria are induced to migrate across the epithelium into the gut lumen in response to the effects of hepoxilin A3. This prostanoid chemoattractant is secreted apically by IECs as a result of cellular interactions with the Salmonella SipA protein (28). Interestingly, this function of SipA does not appear to require translocation into the IEC cytosol, and, instead, may involve stimulation of a cell surface receptor that is yet to be characterized. Activation of protein kinase C is a key step downstream of this putative receptor in SipA-induced production of hepoxilin A3 (29). Because migration of neutrophils across the epithelium disrupts the integrity of the mono-layer, contributes to IEC dysfunction, and gives rise to the microabscesses that are a characteristic pathologic feature (30,31), further study of the effects of SipA is likely to lead to important insights into the pathogenesis of Salmonella enteritis.

The Salmonella effector SopB is an inositol phosphatase that increases inositol flux in IEC tissue culture models (32). It has also been shown to induce chloride secretion in such models (32,33) and may contribute thereby to the pathogenesis of diarrhea. In addition, this protein activates the phosphatidylinositol 3-kinase/Akt kinase signaling pathway in epithelial cells (34). We demonstrated recently that this pathway can also be activated in a SopB-independent manner, and that it may have an anti-inflammatory function by inhibiting IL-8 production (35). Such attenuation of IEC inflammatory responses may be necessary for Salmonella to obtain a foothold in the gut. Additional anti-inflammatory effects are provided by AvrA, a SPI1 effector that inhibits the activation of MAP kinase-dependent signals, and by an uncharacterized bacterial factor that appears to act by altering neddylation of the enzyme complex that ubiquitinates IκBα (36,37).

The in vivo role of the various SPI1 effectors has been examined in both the bovine and murine models of enteritis (15,38,39). These studies have confirmed the importance of SipA, SopB, SopE, and SopE2 in intestinal inflammation, but exactly how these proteins contribute to disease pathogenesis, and, specifically, whether their effects are primarily on cell invasion or on activation of pro-inflammatory signaling pathways, are issues that await clarification. Experiments with the bovine model have also implicated the SopA and SopD effectors in the pathogenesis of gastroenteritis, but very little is known about the molecular function of these proteins in inflammation other than that they are involved in invasion of polarized epithelial cells (38–40). The putative anti-inflammatory function of AvrA has not been substantiated in vivo (38). Table 1 summarizes what is currently known about the role of Salmonella effector proteins in the activation of IEC responses and in the pathogenesis of gastroenteritis, and Fig. 1 illustrates some of these concepts.

Table 1.

Summary of the Functions of SPI1 Effector Proteins and Their Roles in Intestinal Inflammation

Effector protein	Molecular function	Role in inflammation	Reference
SipA	Actin crosslinking		98
	Hepoxillin A3 induction	PMN migration/ diarrhea in calves, murine colitis	15,28
			38
SipB	Caspase binding, translocase	Diarrhea in calves	11,93
SipC	Actin bundling, nucleation, translocase		99
SipD	Translocase		100
SopA	Unknown	Diarrhea in calves	38
SopB	Inositol phosphatase	Diarrhea in calves	15,32
			38
SopD	Unknown	Diarrhea in calves	38,101
SopE	GEF for Cdc42/Rac1	Murine colitis	15,102
SopE2	GEF for Cdc42/Rac1	Diarrhea in calves, murine colitis	15,38
			103,104
SptP	Phosphatase and GTPase activation	None	38,105
AvrA	Inhibition of NF-κB activation	None	38,106

Fig. 1.

Initial interactions between Salmonella and IECs. Several effector proteins are translocated into the IEC cytosol via the SPI1 TTSS. Some of these proteins are involved in the actin remodeling and plasma membrane protrusions that are required for cell invasion, and also activate intracellular signals. SipA may also act from an extracellular location to induce hepoxilin A3 secretion. Flagellin is transcytosed from the apical surface of the epithelium by vesicular transport, and upon reaching the basolateral aspect is able to activate TLR5. The combined effect of the signals activated by the effector proteins and flagellin/TLR5 interactions results in changes in gene expression and the secretion of inflammatory mediators from either the basolateral or apical surface.

Epithelial PRRs in the Response to Salmonella

The cells of the innate immune system use relatively invariant, non-clonotypic, germ-line encoded receptors to recognize conserved molecules that are expressed by broad groups of microorganisms. An important subset of such PRRs is made up of a family of structurally related transmembrane proteins known as the TLRs (41). There are 11 TLRs that are involved in sensing and responding to various microbial components such as lipopolysaccharide (LPS), lipopeptides, and DNA, and there has been a great deal of recent interest in elucidating the role of these receptors in epithelial responses. TLR5, the receptor for monomeric bacterial flagellin, is expressed on the basolateral aspect of primary intestinal epithelium or polarized IEC lines, a location that is presumed to prevent inappropriate activation by commensal bacteria (42,43). Pathogenic flagellated bacteria have special virulence mechanisms that result in flagellin crossing the epithelial monolayer, or the bacteria may themselves invade or damage the epithelium and come into contact with TLR5. Apical exposure of T84 cells to Salmonella, for instance, results in the vesicle-mediated transcytosis of flagellin to the basolateral aspect of the monolayer, a process that depends on molecules encoded by a second Salmonella pathogenicity island SPI2 (44). The polarized localization of TLR5 thus allows it to respond selectively to the presence of flagellin on the basolateral aspect of the epithelium, circumstances that would indicate infection with a presumptive pathogen. Flagellin is one of the few TLR ligands that has been demonstrated to associate physically with its receptor (45). The signals transduced by TLR5 following flagellin binding are a major stimulus for the epithelial expression of IL-8, implying an important role in neutrophil recruitment to the intestine (42,43). Studies carried out recently in the mouse enteritis model support the importance of TLRs in the intestinal inflammatory response by demonstrating that Salmonella-induced intestinal inflammation was significantly attenuated in mice lacking MyD88, an adaptor protein that is required for signal transduction downstream of most TLRs (17). It is not clear, however, whether the requirement for MyD88 is in IECs or in sub-epithelial cells such as macrophages. These studies also raised some questions about the role of flagellin–TLR5 interactions, and suggested that the attenuated enteritis caused by aflagellate mutants of Salmonella may have more to do with decreased motility than a failure to activate TLR5 (16). Definitive resolution of this issue will require investigations in TLR5-deficient mice. See note added in proof.

Recent work has demonstrated that the production of monomeric flagellin, the ligand for TLR5, may depend on exposure of Salmonella to lysophospholipids secreted by IECs (46). Because IEC production of the lysophospholipids requires contact with the bacteria, bidirectional cross-talk between host and pathogen appears to be involved in the activation of TLR5. The in vivo significance of these intriguing tissue culture observations is yet to be confirmed.

Other TLRs, such as TLR2, TLR3, TLR4, and TLR9, have been reported to be expressed on IECs (47,48). However, there is some controversy regarding the precise role of these receptors in the intestinal epithelium, and their in vivo contribution to the activation of IEC inflammatory responses in infectious enteritis has yet to be convincingly demonstrated.

Other Factors Involved in Bacterial Epithelial Interactions

Apart from the SPI1 TTSS and flagellin, other bacterial structures may contribute to the Salmonella–IEC interaction. The Salmonella genome contains a number of fimbrial operons that carry out crucial roles during the intestinal phase of infection. Fimbrial adhesins are involved in attachment to and invasion of IECs and also act as targets of the adaptive immune response (49). The csgBA Salmonella mutant, which carries defects in one of the fimbrial operons, is impaired in eliciting fluid accumulation and induction of Groα mRNA in bovine ligated ileal loops, possibly because of a failure to activate TLR2 (50).

Epithelial Anti-microbial Peptides in the Control of Salmonella Infection

In addition to the production of inflammatory mediators involved in recruiting circulating phagocytic cells to the site of infection, IECs also play an important role in killing bacterial invaders via the secretion of antimicrobial peptides. Paneth cells, one of the lineages that differentiate from epithelial precursors, contribute significantly to this function. These cells, which are located at the bottom of small intestinal crypts, express a number of short cationic peptides, the α-defensins or cryptdins, that are able to disrupt the membranes of a number of microorganisms, including Salmonella (51). The defensins are secreted apically into the crypt lumen when Paneth cells are exposed to intact bacteria or bacterial components such as LPS (52). The importance of defensins in protection against enteropathogens is illustrated by the observation that transgenic mice expressing human defensin 5 exhibited a dramatic increase in resistance to Salmonella infection (53).

IECs express members of another class of defensive peptide, the cathelicidins, in which a cathelin domain is linked to a peptide with anti-microbial activity (54). Several cathelecidins have been identified in mammals, but of these humans express only one, LL-37 (hCAP18). LL-37/hCAP18 is expressed in the epithelial cells located at the surface and upper crypts of the normal human colon (55). In the mouse, cathelicidin-related anti-microbial peptide (CRAMP) has been shown to inhibit Salmonella replication in vivo and in vitro, and to induce the formation of long filamentous bacteria (56).

Developmental Changes in Epithelial Function and the Intestinal Response to Salmonella Infection

Systemic spread of Salmonella during intestinal infection in normal adult individuals is minimized by the acute intestinal inflammatory response. However, young infants, particularly those under 3 mo age, are prone to extra-intestinal dissemination of the bacteria, leading to complications such as sepsis, osteomyelitis, and meningitis. We observed recently that this age-dependent variation in the response to Salmonella infection also occurs in the mouse model (19). One- to 2-wk-old pre-weaned mouse pups had significantly less intestinal inflammation, based on a number of assays, than 5–6-wk-old adults. Correspondingly, the younger animals had larger numbers of Salmonella in the mesenteric lymph nodes and spleen than the adults. Using this model, we found that the intestinal expression of a number of genes involved in anti-microbial and inflammatory responses was significantly upregulated with age. The expression of many of these genes, including chemokines such as CXCL9 and CXCL10, and regulators of intracellular infection such as LRG-47 and IIGTP, are under the control of interferon (IFN) γ. Indeed, we found that the age-dependent increase in expression of such genes did not occur in IFNγ-deficient mice. Correspondingly, the response of adult IFNγ-deficient animals to Salmonella infection resembled that of wild-type pups, with lower levels of intestinal inflammation and higher numbers of tissue bacteria at systemic sites. Based on these observations, we have suggested that IFNγ plays an important role in the maturation of intestinal innate immunity, and in regulating Salmonella-induced intestinal inflammation.

Salmonella Beyond the Epithelium: Amplification of the Inflammatory Response

Once Salmonella has breached the intestinal epithelial barrier, either by transcytosis across the IEC or by paracellular movement through breaks in the epithelium, the bacteria encounter cells of the immune system that are resident in the lamina propria, such as tissue macrophages, dendritic cells, and mast cells, as well as newly recruited neutrophils and monocytes. Resident lamina propria macrophages of the intestine, unlike their newly recruited counterparts, are generally considered to be poorly responsive to bacteria because of low levels of expression of the TLR accessory molecule CD14 and the influence of TGFβ in the local microenvironment (57). Nevertheless, they are avidly phagocytic and have been shown to kill internalized Salmonella very effectively (58). Mast cells, which are present in large numbers in the gastrointestinal tract, play an important role in bacterial infection by their ability to rapidly release several inflammatory mediators from pre-formed granule stores (59). However, recent experiments with a low-dose intraperitoneal infection model suggest that these cells are not critical to protection against Salmonella (60). It is not clear whether this conclusion is also applicable to enteric infection. Dendritic cells are key to the development of adaptive immune responses during Salmonella infection and may actively take up the bacteria from the lumen even before epithelial invasion (61). The interactions of these cells with Salmonella have been reviewed in detail elsewhere (62).

Influx of neutrophils and monocytes into the intestinal lamina propria and epithelial layer is one of the characteristic histopatho-logic features of Salmonella enteritis. The migration of neutrophils to the site of infection depends on specific chemoattractants such as IL-8, GROα, GROγ, and hepoxilin A3 that are produced by infected IECs and other cells in the lamina propria (63). The exact chemoattractants involved in recruiting monocyte/macrophages have not been delineated, although epithelial expression of CCL2 has been implicated in both human tissue culture systems and in vivo mouse models of salmonellosis (64,65). Both cell types play important roles in phagocytosing and killing Salmonella and in the production of various mediators such as IL-1 and TNFα that amplify the events set in motion by IECs. Some of the molecules secreted by these cells contribute significantly to the tissue damage that is the inevitable accompaniment of the inflammatory response. The molecular details of these processes are discussed below, and are depicted in Fig. 2.

Fig. 2.

Interactions between Salmonella and phagocytic cells. Phagocytic cells such as macrophages recognize and respond to the bacteria through cell surface TLRs, which sense microbial components such as LPS, and intracellular NLRs, which sense flagellin translocated into the cytosol through the SPI1 TTSS. These interactions lead to the secretion of inflammatory cytokines and the induction of apoptosis.

Phagocyte Anti-microbial Mechanisms

Bacteria phagocytosed by macrophages and neutrophils are subjected to intracellular killing mechanisms that include trafficking to the degradative environment of the lysosome, the generation of toxic reactive oxygen and nitrogen intermediates (ROI and RNI) by phagocyte oxidase (phox) and inducible nitric oxide synthase (iNOS), respectively, exposure to bactericidal peptides, and micronutrient deprivation. As a successful intracellular pathogen, Salmonella has evolved strategies to resist some of these anti-microbial defenses, allowing it to survive for extended periods inside macrophages both in vitro and in vivo (66). Genes encoded by SPI2 play an important role in intra-macrophage survival, largely by modulating the characteristics of the SCV so that fusion with the lysosome, as well as recruitment of phox and iNOS, are prevented (67). Certain SPI2 defective strains of Salmonella, specifically those with mutations in the sseD or ssaT genes, cause an attenuated form of enteritis in both mouse and bovine models, suggesting that the ability to survive within macrophages influences the inflammatory response (11,17,68). However, inactivation of another SPI2 gene, ssrA, does not result in alteration of fluid secretion in a rabbit ligated ileal loop model (69).

Intracellular survival and multiplication of Salmonella are also regulated by the Nramp1 gene, which encodes a macrophage-specific divalent cation transporter that is believed to pump iron out of the SCV, thereby depriving the bacteria of an essential nutrient (70). Although there are no published studies that directly address the role of Nramp1 in Salmonella-induced intestinal inflammation, it is reasonable, based on the effects of SPI2 muta tions, that changes in the number of intracellular bacteria would influence the enteritis. Furthermore, production of the macrophage anti-inflammatory protein, secretory leukocyte protease inhibitor (SLPI), which inhibits TLR activation, is dependent on Nramp1 (71,72). We have shown recently that another macrophage iron transporter, ferroportin (FPN), influences the intracellular growth of Salmonella, probably by reducing iron concentrations within the SCV (73). We are currently in the process of analyzing the effects of altered FPN expression on enteritis in the mouse model.

Phagocyte Inflammatory Mediators

One of the important functions of macrophages newly recruited to the gut during Salmonella infection is the secretion of soluble mediators that help to amplify the inflammatory response initiated by the IECs. Two mediators that play important roles in this process are TNFα and IL-1β (74,75). TNFα promotes the recruitment of additional leukocytes, including neutrophils, by inducing chemokine and adhesion molecule expression on surrounding cells. In addition, it enhances phox- and iNOS-independent microbicidal functions of macrophages, thus aiding in the killing of phagocytosed bacteria. Not surprisingly, mice deficient in TNF receptor I are impaired in their ability to control the growth of several intracellular pathogens including Salmonella. IL-1β has effects similar to TNFα and also acts on the liver to induce production of acute phase reactants and on the hypothalamus to influence body temperature regulation.

The anti-microbial armamentarium of neutrophils includes a number of proteases that are stored in granules (76). Upon appropriate stimulation, these granules fuse with the phagolysosome and discharge their contents into the environment of the internalized bacteria. However, some of the granules also undergo exocytosis, resulting in the release of the proteases outside the cell, with consequent degradation of extracellular matrix components and damage to epithelial cells (31). The importance of this aspect of the inflammatory response is highlighted by the observation that inhibition of neutrophil elastase results in significant amelioration of dextran sulfate sodium (DSS)–induced colitis in mice (77). Interestingly, some neutrophil proteases also have anti-inflammatory effects. Neutrophil elastase and cathepsin G degrade flagellin from several kinds of bacteria, including Salmonella, thereby inhibiting flagellin-induced inflammatory responses (78). It will be important and informative to determine the role of neutrophil proteases in Salmonella enteritis in vivo.

One of the consequences of the Salmonella-induced inflammatory response is the induction of cell death. Intestinal epithelial cells may die from the direct effects of the infection or in response to various inflammatory mediators. Neutrophils are relatively short-lived cells, die once they have fulfilled their anti-bacterial function, and are phagocytosed by macrophages. Macrophages themselves are induced to undergo apoptosis as a result of the actions of the Salmonella SipB effector protein (discussed below) or in response to stimulation by LPS (79). Cell death may facilitate the spread of Salmonella, but may also lead to the generation of additional inflammatory stimuli as a result of the release of intracellular contents such as uric acid (80).

Phagocyte Sensing of Bacteria

The activation of phagocyte anti-microbial and inflammatory mechanisms occurs in response to microbial triggers that are recognized by specific cellular sensors. The TLRs constitute one important class of such sensors, and both macrophages and neutrophils express several of these receptors (81). Work from our laboratory, as well those of others, has shown that TLR2 and TLR4 play important roles in inducing the production of TNFα and other inflammatory cytokines by macrophages during Salmonella infection (82–85). TLR signals also lead to increased expression of IL-1β mRNA and intracellular protein (74).

In addition to the TLRs, which are expressed on the cell surface, recent work has demonstrated the existence of PRRs in the cytosol. This intracellular sensing system consists of a large family of structurally related proteins that have been given the name NLRs, for NACHT (a domain common to this family)-leucine rich repeats (86). The pattern of expression of these proteins varies, but several NLRs are expressed in macrophages and neutrophils. Their roles in innate immunity are the subject of active, on-going investigations. One important function of a subset of these proteins that has already come to light is the activation of caspase 1. This protease is essential for the proteolytic maturation and secretion of IL-1β and the related cytokine IL-18, and is also involved in the induction of apoptosis. Various members of the NLR family are required for inducing IL-1β processing and secretion in response to a number of endogenous and exogenous triggers, including ATP, uric acid crystals, Gram-negative and Gram-positive bacteria, bacterial toxins, and bacterial RNA (87–90). Several laboratories have also reported that Salmonella-induced IL-β secretion depends on at least two of the NLR proteins, ASC and IPAF (89,91,92). IPAF has a direct activating interaction with caspase 1, while ASC functions as an adaptor that links other NLRs with the protease. The specific Salmonella component involved in IPAF-dependent induction of IL-1β secretion was found to be soluble flagellin (91,92), and it was demonstrated that the entry of flagellin into the cytosol required the SPI1 TTSS (91). This requirement would help to explain earlier observations indicating a role for SipB in caspase 1 activation and apoptosis, although the significance of the direct binding of SipB to caspase 1 needs further clarification (93). It is not yet clear whether the flagellin is translocated through the needle complex of the SPI1 TTSS, or whether it leaks from the phagosome via the pore made by this structure. The exact roles played by IPAF and ASC in flagellin sensing are also matters of speculation. It is tempting to think that the leucine-rich repeat domain of IPAF may bind to flagellin directly, but there is no evidence yet to support this idea. It has been suggested that ASC may act to stabilize the IPAF–caspase 1 interaction, but it is also possible that it could be part of a separate NLR complex required for the response to cytosolic flagellin. Recent experiments have confirmed the protective role of Salmonella-induced caspase 1 activation in vivo (94), in contrast to earlier observations suggesting that caspase 1 activity facilitated colonization of Peyer's patches by the bacteria (95).

Experiments in cultured macrophages have revealed an additional mechanism for recognizing bacteria in the cytosol (96). These studies demonstrated that although the majority of intracellular Salmonella reside within the SCV, a small population of the bacteria escape into the cytosol. Such bacteria become marked by the accumulation of polyubiquitinated proteins on their surface and associate with the proteasome. Inhibition of proteasomal activity enhanced intracellular growth of Salmonella, suggesting that ubiquitination of cytosolic bacteria may be an important antimicrobial defense mechanism. Elucidation of the molecular pathway that leads to tagging of bacteria with ubiquitin, which may involve the NLRs or other types of intracellular PRRs, will be an important line of further investigation.

Why have cells evolved special systems for responding to bacteria in the cytosol when it would appear that the TLRs would be adequate for bacterial sensing? NLRs may serve simply as a redundant back-up system, but there may be an advantage to recognizing the presence of bacterial components specifically in the cytoplasm. Pathogens are more likely to end up in the cytosolic compartment, e.g., both Listeria and Shigella can escape from the phagosome into the cytoplasm. Pathogens are also more likely to have virulence mechanisms such as the SPI1 TTSS of Salmonella or the type IV secretory system of Helicobacter pylori that are able to facilitate the movement of bacterial molecules into the cytosol. Therefore, triggering of the NLRs and the consequent release of IL-1β would indicate a potentially more “dangerous” situation than TLR activation alone. The induction of apoptosis by NLRs may represent an extreme response to such a situation by helping to eliminate pathogen-infected cells. Clearly, this is a growth area in inflammation research, and the next few years (if not months) will no doubt reveal the molecular details of NLR function, and the role played by these proteins in a variety of inflammatory processes, including Salmonella enteritis.

The Role of IFNγ in the Intestinal Response to Salmonella

IFNγ plays a very important role in protection against several intracellular pathogens, including Salmonella. This cytokine is typically produced by T cells of the Th1 pheno-type and is required for activating various macrophage microbicidal functions. Humans with defects in Th1 development, or in the production of or response to IFNγ, are susceptible to severe mycobacterial and Salmonella infections (97). IFNγ can also be produced early in the course of infection from cells of the innate immune system such as macrophages and natural killer (NK) cells, as well as intra-epithelial lymphocytes and memory CD8+ T cells, and can have a major impact on subsequent immunologic events (62). We have shown recently that IFNγ has a significant influence on Salmonella enteritis in the mouse model (19). The intestinal inflammation induced by Salmonella in adult IFNγ-deficient mice was significantly attenuated, as indicated by fewer histopathological abnormalities and lower levels of inflammatory mediator expression. Some of these effects could be related to the influence of IFNγ on maturation of intestinal innate immune defenses, but direct regulation of IEC and macrophage function with respect to cytokine/ chemokine expression and anti-microbial mechanisms also probably plays a role.

Conclusion

The intestinal inflammation induced by S. typhimurium is an integral part of the innate immune response to the organism in which IECs, macrophages, and neutrophils play key roles. It serves to protect the host against spread of the infection, and in cases where the response is effective, the bacteria are confined to the gut and ultimately eliminated. When it is poorly developed, as in young infants or pre-weaned mouse pups, the risk of systemic dissemination of the bacteria is increased. Does the enteritis serve the pathogen in any way? The fact that several of the SPI1 TTSS effector proteins activate IEC inflammatory responses raises this question, and it is possible that the induction of diarrhea and the consequent spread of the organism to the environment may facilitate its transmission to other hosts.

Salmonella enteritis in humans is generally self-limited and is not usually treated except in infants less than 3 mo of age. With increased understanding of the cells and molecules involved in the pathogenesis of this condition, opportunities to minimize even the transient morbidity associated with the infection will become available through the development of novel therapeutic and prophylactic interventions. Targeting cytokines such as TNFα and IL-1β, or transcription factors such as NF-κB, is a potential approach to controlling the inflammation, one that can now be tested fairly easily using the mouse model.

One issue that we have not touched on is the resolution of Salmonella-induced intestinal inflammation. To our knowledge, there have been no studies on how healing occurs in this condition. This is likely to be an interesting and fruitful line of investigation, but it is hampered by the fact that Salmonella infection in the C57BL/6 strain that is typically used in the murine enteritis model results in death in 5–7 d in a large number of animals. Mouse strains such as 129/SvJ that carry the resistant allele of Nramp1 are able to survive the acute infection (66), and it would be worthwhile to analyze the course of Salmonella enteritis in these strains. Insights into the processes involved in the restitution of normal intestinal structure and function could lead to the identification of additional molecular targets for therapeutic manipulation, and could prove beneficial for the control of both Salmonella-induced and other types of intestinal inflammation.