Chapter 152

Pathogenesis of Systemic Disease

PAUL A. GULIG

INTRODUCTION: SYSTEMIC DISEASES CAUSED BY SALMONELLA SPP. AND ESCHERICHIA COLI

In addition to localized infections of the gastrointestinal tract, Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium), other serovars of the genus Salmonella, and Escherichia coli are major causes of systemic disease in both humans and animals. Salmonella spp. can cause life-threatening enteric fever, septicemia, and focal infections (e.g., osteomyelitis and meningitis), which involve invasion of the bacteria into the blood and possibly the reticuloendothelial system and other organs (reviewed in references 37 and 209). Enteric (typhoid) fever in healthy humans is caused by Salmonella serovars adapted to humans, S. typhi or S. paratyphi. Although S. typhimurium usually causes gastroenteritis in humans, this serovar and others containing related virulence plasmids have the potential to cause systemic infection in compromised individuals. In mice, S. typhimurium causes systemic disease and thus provides a murine model for enteric fever. Most disease caused by E. coli is limited to the intestines and urinary bladder; however, certain types of E. coli can cause systemic infections, including pyelonephritis, meningitis, and septicemia (162, 180).

HOW DOES S. TYPHIMURIUM CAUSE SYSTEMIC DISEASE?

This chapter is concerned with the virulence attributes of S. typhimurium and E. coli that contribute to infection outside the intestinal tract. Most discussion is centered on analysis of the murine enteric fever model that utilizes S. typhimurium, whose gross pathogenesis was initially described by Collins and coworkers over 20 years ago (31, 39, 40). After oral inoculation of susceptible mice, the bacteria invade the intestinal mucosa and proliferate in the Peyer’s patches and draining mesenteric lymph nodes. Salmonellae drain through the lymphatics to the thoracic duct into the blood and ultimately infect the spleen and liver. Multiplication of the bacteria in the reticuloendothelial system ultimately results in overwhelming systemic infection and death.

There are four major virulence functions involved in systemic disease: spreading, evasion of defenses, replication, and production of damage. Each of these functions is discussed in terms of specific virulence factors necessary to carry them out. The first step in systemic disease caused by S. typhimurium is infection of the intestinal lamina propria. Salmonellae may invade through M cells into the Peyer’s patches or invade through epithelial cells of the mucosal surface. The mechanisms of spread through intestinal tissues are discussed in chapter 151. In an immunologically naive host lacking specific immunity, the primary defenses to be evaded by salmonellae are complement and phagocytic polymorphonuclear leukocytes (PMNs) and macrophages. Salmonellae must overcome host functions which suppress bacterial growth within tissues. Finally, the damage caused by S. typhimurium infection is caused primarily by the host response to the bacteria. Bacterial lipopolysaccharide (LPS) and peptidoglycan are highly inflammatory. The endotoxic activity of LPS can also stimulate host cells to produce cytokines, resulting in fever and shock. However, less well-described factors such as exotoxins and cytotoxicity to host cells could contribute to damage and are discussed in chapters 151 and 153. This discussion of the virulence factors involved in systemic infection is based in part on genetic analysis of the pathogens. Most of the virulence attributes discussed are encoded on the bacterial chromosome, although serovars of Salmonella, with the exception of S. typhi, that have the potential to cause systemic disease possess related virulence plasmids that are essential for systemic disease (reviewed in reference 76). The genetic map of the S. typhimurium virulence plasmid, sometimes called pSLT, has been added to the S. typhimurium chromosomal map in chapter 110.

Evasion of Host Defenses

Resistance of S. typhimurium to Antibacterial Activities of Complement.

Complement is a major host defense mechanism composed of a system of activation and regulatory proteins and corresponding cell surface receptors (101). Complement is primarily found in blood, but is not normally found at the mucosal surfaces. Therefore, salmonellae and E. coli that have invaded beyond the mucosal membranes encounter complement and must contend with its antibacterial functions of opsonization (facilitated phagocytosis by PMNs or macrophages) and lysis of the bacteria. Complement must be activated by either the classical (antibody-dependent) or the alternative (antibody-independent) pathways. In a nonimmune host, the biochemical interactions between complement and carbohydrates of bacterial surfaces such as LPS generate activated complement component C3b from the C3 component. Activation and deposition of complement on bacteria lead to opsonization by phagocytes with receptors for complement components C3b and C3bi. Gram-negative bacteria can be lysed by the membrane attack complex (MAC), which forms a pore consisting of components C5 to C9, inserting through the bacterial outer and cytoplasmic membranes. Additionally, the C3a and C5a cleavage products recruit host defense cells and stimulate inflammation. However, murine complement is not nearly as active as that of humans.

Resistance to complement involves surface components of the bacterial cells. Many bacteria resist complement activity by having a polysaccharide capsule masking the surface of the bacterial cell and inhibiting the activation of complement. The S. typhi Vi capsular antigen and the E. coli K1 capsule (see How Does E. coli Cause Systemic Disease? below) are examples. S. typhimurium, being unencapsulated, relies on the complete O antigen of the LPS of the outer surface of the outer membrane to resist complement (100). Mutants with incomplete LPS are extremely attenuated relative to their O-antigen-expressing parental strain and are lysed by complement (178). The biosynthesis of LPS is discussed in chapter 69. Because LPS is involved in so many aspects of virulence, including resistance to complement and phagocytes, essentially all of the genes involved in LPS biosynthesis are essential for virulence. There are three aspects to LPS-dependent complement resistance: inhibition of activation, activation at a distance from the outer membrane, and prevention of functional insertion of the MAC (63). In the absence of O antigen, complement is activated and deposited at the surface of the outer membrane (36). The importance of the biochemical composition of the O antigen is demonstrated in the subtle difference between the O-antigen components abequose of S. typhimurium and tyvelose of Salmonella montevideo resulting in differential activation and deposition of C3b (96, 115), leading to differences in phagocytosis in vitro (115, 116) and virulence in the mouse model (117). The length of LPS O antigen affects complement resistance in that the C5 to C9 complex forms preferentially at the outermost end of relatively long O-antigen-containing LPS molecules, so that insertion of the MAC is prevented (103, 104, 105). The MAC is then shed from the bacterial cell without effect.

In addition to the critical roles of LPS in evading the complement system, two outer membrane proteins encoded on the S. typhimurium virulence plasmid possess anticomplement activities. TraT is the surface exclusion protein encoded by the S. typhimurium virulence plasmid, F plasmid, and other related antibiotic resistance (R) plasmids (176). TraT-containing cells bind C9 to the same extent as do non-TraT-containing cells; however, lysis is avoided in an unknown manner. The Rck protein possesses unique anticomplement activity in conferring relatively high levels of complement resistance to E. coli and S. typhimurium with rough LPS (81). Rck inhibits the C5 to C9 MAC from inserting into the membrane so that the MAC is shed from the cell surface, and lysis does not occur (81). However, the complement resistance of virulence plasmid-cured S. typhimurium (75, 76, 189) underscores the effectiveness of the complement resistance functions encoded on the chromosome, most importantly complete O antigen of LPS.

Resistance of S. typhimurium to Phagocytes.

Beyond the mucosal membranes, invasive bacteria encounter phagocytic defenses provided by PMNs or macrophages or both. The interactions of bacteria with phagocytic defenses have been reviewed (147, 211). The means by which bacteria survive the encounter with phagocytes divides the pathogens into the extracellular and intracellular organisms. The extracellular pathogens prevent phagocytosis, usually by a polysaccharide capsule which prevents activation and deposition of complement, as well as presenting a surface to which phagocytes cannot bind, e.g., E. coli K1 (see How Does E. coli Cause Systemic Disease? below). If extracellular pathogens are phagocytosed, they are usually killed by the phagocyte. Salmonellae are intracellular organisms which subvert the antibacterial action of phagocytes after phagocytosis and subsequently survive. There are numerous potential virulence mechanisms involved in intracellular pathogenesis. Some intracellular pathogens escape the phagosome or fused phagosome-lysosome and replicate within the safe environment of the host cell cytoplasm (147) (see chapter 151). Some organisms prevent the fusion of the phagosome and lysosome so that the phagosome does not become as inhospitable as it normally would be (147). Inside the phagosome or phagolysosome, bacteria face both oxidative and nonoxidative antibacterial factors. Successful pathogens in this environment must prevent the synthesis or delivery of the compounds to the phagosome, neutralize the compounds, or, like S. typhimurium, be resistant to their action.

To examine whether macrophages are important in suppressing infection by S. typhimurium, mice have been injected with factors that either occupy macrophages or kill them. Injection of dextran sulfate (83), carrageenan (3), silica (66, 159), and liposome-encapsulated dichloromethylene diphosphate (H. Matsui, T. J. Doyle, and P. A. Gulig, submitted for publication) results in mice becoming more susceptible to infection by S. typhimurium in the early phase (within a week) of infection. In contrast, PMNs do not appear to be involved in suppressing the early stages of systemic salmonella infection. Dunlap et al. reported that almost all of the salmonellae within the spleens of intravenously (i.v.) inoculated mice rapidly became resistant to killing by host defenses (53). Because the protected salmonellae were resistant to gentamicin, an antibiotic commonly used to selectively kill extracellular bacteria in vitro, the salmonellae were presumed to be within host cells. The protected bacteria were observed within PMNs from infected spleens after host cells were immunochemically stained and sorted (52). The lack of importance of PMNs in killing salmonellae was confirmed by Matsui et al., who showed that cyclophosphamide-treated mice (which were depleted of circulating granulocytes by 90%) did not experience increased splenic infection early after oral inoculation (submitted for publication).

  Salmonella-macrophage interactions examined in vitro. Much has been learned from experimental infection of murine macrophages and macrophage-like cell lines in tissue culture. However, the nature of the macrophage being examined and other potential artifacts of in vitro culture must be kept in mind in interpreting data. For example, most macrophage-like cell lines possess some defect in antibacterial function (144). The means of obtaining macrophages from mice can affect the results of in vitro assays. Resident peritoneal macrophages were proficient at killing antibody-opsonized salmonellae; however, thioglycolate-elicited macrophages were highly phagocytic, yet defective in killing (19). In another study (22), salmonellae grew best in the J774 macrophage-like cell line, but were killed most efficiently by resident peritoneal macrophages and survived with intermediate efficiency in splenic and bone marrow-derived macrophages. Proteose peptone-elicited macrophages were defective at killing salmonellae. Since S. typhimurium can replicate very rapidly in tissue culture media, extracellular bacteria are usually killed by the addition of gentamicin to the tissue culture after infection has begun. In such in vitro assays, the number of intracellular bacteria drops by about 10-fold from the initial adherent number (22, 75, 89, 119). However, it is unclear whether this drop represents the killing of extracellular bacteria by gentamicin or the killing of intracellular bacteria by the phagocytes. The number of intracellular salmonellae usually rises over the next 24-h period, demonstrating that the salmonellae are able to replicate intracellularly and resist killing mechanisms of the phagocytes.

Fields et al. (61) examined the genetics of resistance to macrophages by constructing a set of transposon Tn10 insertion mutants of S. typhimurium and screening for killing of mutants by peritoneal macrophages in an in vitro assay. Mutations conferring susceptibility to macrophages and attenuation were in genes involved with LPS biosynthesis and structure; biosynthesis of purines, pyrimidines, and aromatic compounds; resistance to oxidative stress; resistance to serum; motility; and, as determined in a later study, the gene phoP (60), discussed below. Several additional attenuating Tn10 insertions were recently mapped and sequenced (11).

Genetic and biochemical means were recently used to probe the gene expression and growth of salmonellae within phagocytic cells in vitro. Proteins synthesized by salmonellae within J774 macrophage-like cells (23) and within U937 macrophage-like cells (1) were identified using two-dimensional gel electrophoresis. Many proteins were uniquely expressed intracellularly while others were repressed intracellularly; however, differences were noted in the expression of the heat shock proteins known to be expressed by salmonellae under stress conditions. Regulation of gene expression in response to the host is discussed in chapter 154. Two populations of S. typhimurium were identified in U937 cells by measuring acetylation of ribosomal protein L7 as an indication of growth rate—a rapidly growing population and a more slowly growing population (2). The basis for differential growth rates by different bacteria has not been elucidated but could be important in understanding the intracellular pathogenesis of S. typhimurium. It is possible that the salmonellae reside in two different environments within the phagocytes, one being responsible for intracellular killing and the other being more permissive for growth. More thorough understanding of this aspect of the host-pathogen interaction could answer many questions as to the nature of intracellular infection by salmonellae.

Because PMNs are such short-lived cells (surviving only hours in vitro), their analysis in vitro is much more difficult. Baron and Proctor (9) examined murine PMNs in vitro and determined that PMNs were ineffective against virulent S. typhimurium unless specific anti-salmonella antibody was provided. van Dissel and coworkers (201) examined infection of murine PMNs with S. typhimurium and observed results similar to those obtained by others with macrophages: initial killing of the bacteria followed by the survival of resistant bacteria. The means by which salmonellae are opsonized can affect intracellular events such as the subsequent fusion of host cell granules with phagosomes and oxidative burst in human PMNs (102). It is possible that the mechanism by which salmonellae enter a phagocyte determines the outcome of the cellular infection, that is, bactericidal activity versus permissive growth, discussed above. Since S. typhimurium can invade host cells, it could be that active invasion by salmonellae versus phagocytosis by macrophages or PMNs determines the fate of the intracellular salmonellae.

  Evasion of phagocytosis. Although salmonellae act as intracellular pathogens of phagocytes, the bacteria may act like extracellular bacteria in preventing phagocytosis. The first line of defense against phagocytosis is the lack of activation and binding of complement components which could serve as opsonins, as described above. In vitro, opsonized salmonellae appear to be phagocytosed through the CR1 receptor more than the CR3 receptor (92). Rough LPS mutants of S. typhimurium are phagocytosed and killed more efficiently by murine macrophages than are wild-type salmonellae (65), and the specific composition of the O antigen is also important (116).

  Inhibition of phagosome-lysosome fusion. Some evidence suggests that salmonellae can inhibit phagosome-lysosome fusion in macrophages (24, 93). Thorium dioxide-prelabeled lysosomes were inhibited from fusing with phagosomes containing salmonellae by approximately 50%, as observed by electron microscopy (24). Using fluorescence microscopy, Ishibashi and Arai observed that S. typhimurium inhibited fusion of phagosomes with sulforhodamine-labeled lysosomes in murine macrophages (93). Inhibition required viable bacteria capable of protein synthesis in the macrophages. An uncharacterized Tn5 mutant S. typhimurium strain which experienced higher levels of phagosome-lysosome fusion than did the wild-type parent was identified (94). Interestingly, the mutant strain was not attenuated for virulence in mice after intraperitoneal (i.p.) or oral inoculation. Treatment of macrophages with gamma interferon (IFN-γ) increased the frequency of phagosome-lysosome fusion (91, 107). In contrast, it has been reported that S. typhimurium does not inhibit phagosome-lysosome fusion but that the bacteria are resistant to degradation by macrophages (29). Alpuche-Aranda et al. used fluorescent dye markers to prelabel lysosomes and examine host-pathogen interactions by fluorescence microscopy (4). No inhibition of fusion occurred, but salmonellae were found in a unique vacuole called the "spacious phagolysosome." The use of video microscopy identified salmonellae within a vacuole increasing in size with time after the engulfment of the bacteria. The spacious phagolysosome was indistinguishable from macropinosomes; that is, they were formed by the continual fusion of pinosomes with the phagosome. The differences in results pertaining to phagolysosomal fusion most likely are due to the different procedures used by the investigators, including the method of prelabelling lysosomes, fixing cells, and microscopic techniques.

  Survival against phagolysosomal ROI. The oxidative or respiratory burst (sometimes measured experimentally by chemiluminescence) is one of the primary antibacterial functions of macrophages and PMNs and includes the reactive oxidative intermediates (ROI) superoxide, peroxide, hydroxyl radical, and hypochlorous acid (80). Opsonized salmonellae in murine peritoneal macrophages (17) or rough LPS mutant salmonellae in human PMNs (121) caused increased chemiluminescence after phagocytosis. Whether oxidative burst was detected (17, 90) or not in other experiments (204), the response by the phagocytes was not sufficient to kill the salmonellae.

The essential nature of ROI in protection against salmonellosis is seen in the increased incidence of salmonellosis in patients with chronic granulomatous disease, who lack the NADPH dehydrogenase necessary to generate superoxide and other ROI (148). S. typhimurium possesses several enzymes which are capable of neutralizing ROI, including two catalases for peroxide and superoxide dismutase for superoxide (59). These enzymes are regulated as part of the OxyR and SoxRS regulons in response to peroxide and superoxide, respectively. Interestingly, strains with mutations in oxyR and katG (the latter encoding one of the catalases) were not susceptible to killing by PMNs in vitro (163). The alternative sigma factor RpoS is required for production of another catalase, KatE, and RpoS mutants of S. typhimurium are attenuated for virulence in mice. The full spectrum of virulence genes regulated by RpoS is not yet known; however, the spv genes of the virulence plasmid, which are not involved in resistance to ROI, are also expressed directly or indirectly by RpoS (58). The responses of S. typhimurium and E. coli to various stimuli and stresses are discussed elsewhere in this book (see chapters 88, 93, 95, 96, and 106). S. typhimurium recA mutants are sensitive to the oxidative burst of macrophages and are avirulent in mice (25). RecA is a regulatory protein responsible for protecting DNA from damage. Oxidatively damaged DNA may not be repaired in recA mutants, or the RecA protein may be required to express other antioxidant factors of salmonellae. A 59-kDa outer membrane protein was associated with resistance of S. typhimurium to oxidative killing by human PMNs (188). It therefore appears that S. typhimurium does not rely on inhibition of respiratory burst to survive but that the bacteria are resistant to the oxidative products that are formed.

  Resistance to antibacterial proteins. Several nonoxidative defenses are present in human macrophages or PMNs or both. These include cationic proteins called defensins; bactericidal permeability-increasing protein; the enzymes lysozyme, cathepsin G, elastase, azurocidin, and other proteases; and iron-binding lactoferrin (27, 114, 193). However, murine macrophages and PMNs do not produce defensins (114). The importance of complete LPS in resistance to nonoxidative killing is demonstrated by the fact that mutants of S. typhimurium with incomplete LPS are highly susceptible to granular extracts of human PMN lysosomal fractions (175). S. typhimurium phoP mutants are highly attenuated for virulence (60, 68, 137) because of susceptibility to defensins (60, 140). PhoP and PhoQ represent a two-component system for regulation of gene expression in S. typhimurium (73, 137). The identity of the defensin resistance virulence gene(s) regulated by PhoP has not been completely elucidated. The first PhoP-regulated virulence gene identified, pagC (PhoP-activated gene), is not involved in resistance to defensins but is essential for resistance to macrophages (171). PhoP-constitutive (PhoP^c) S. typhimurium strains are attenuated in inducing spacious phagolysosome formation (4) and virulence (139), suggesting that PhoP-repressed genes are involved in this aspect of virulence. However, the phoP(Con) mutation was recently determined to be in the second component, phoQ (138). Groisman et al. performed a systematic genetic analysis of resistance to antimicrobial peptides and identified another locus called sap, whose deduced protein sequences include an ATP-binding cassette transporter and a periplasmic binding protein (74, 164). Additionally, certain polymyxin B-resistant mutants of S. typhimurium are resistant to the cationic antibacterial proteins of human PMNs (181). The DNA sequence of the cloned polymyxin resistance genes revealed a two-component regulatory system previously identified as pmrA (198).

  Resistance to acid. Another potential nonoxidative defense function within the phagolysosome is acidification of the vacuole. The pH of the phagosomal vacuole can drop to as low as 5 (10). This acidic condition could act as a bacteriostatic or bactericidal factor. S. typhimurium can inhibit or delay the acidification of the phagosome in murine macrophages in vitro (5). In addition to being resistant to this vacuolar acidic environment, salmonellae must also contend with the harsher acidic conditions (pH less than 2) of the stomach during digestion. Salmonellae and other enteric pathogens which normally enter the body through the alimentary tract have developed ways to avoid killing by acidic conditions. The genes encoding acid resistance factors are coordinately regulated as part of the acid tolerance response (ATR), reviewed in chapter 96. Two functions involved in ATR include Mg²⁺-dependent proton-translocating ATPase (atp) and regulation of iron (fur). It is easiest to imagine that ATR is primarily involved in resistance to harsh gastric acidity. However, although S. typhimurium mutants in atp or fur are attenuated for virulence of mice by the oral route, neutralization of gastric acidity did not restore virulence (69). Fur^– S. typhimurium retained virulence by the i.p. route, and in vitro, both atp and fur mutants were resistant to murine peritoneal macrophages. Therefore, the attenuation of the mutants remains unexplained. Another ATR component, DNA polymerase I encoded by polA, is required for virulence, but only when combined with other uncharacterized ATR mutations (M. Wilmes-Reisenberg, J. Foster, and R. Curtiss, unpublished data). It is possible that acid treatment, similar to oxidative stress, induces DNA damage and that inhibition of repair mechanisms thereby attenuates virulence.

  Other genes of S. typhimurium involved in macrophage resistance. Analysis of the role of flagella in the virulence of S. typhimurium resulted in the identification of an interesting regulatory gene involved in flagellar biosynthesis, flgM, as a virulence factor. An uncharacterized deletion mutant of S. typhimurium was avirulent in mice because of sensitivity to macrophages (206). The reason for the attenuation of the nonflagellated mutant was a deletion of a gene called mviS which was linked to the flagellar mutation (30). The lack of flagella had no role in the observed attenuation. mviS was later identified as the flgM gene (183) characterized by Hughes and coworkers as being a regulatory protein in S. typhimurium which counteracts the sigma factor FliA by a novel mechanism of action (71). The exact mechanism by which FlgM affects resistance of salmonellae to macrophages is not known. However, the effect appears to be due to the disrupted expression of the fliA gene and not some other pleiotropic effect of FlgM (183).

A hemolysin named salmolysin was identified in S. typhimurium upon plating of a fresh human isolate on blood agar (118). The hemolytic activity had apparently evaded discovery because this protein is poorly expressed by salmonellae in vitro. slyA S. typhimurium was attenuated for virulence when inoculated into mice by oral as well as parenteral routes. The nature of the defect appeared to be in resistance to killing by murine macrophages. Since the virulence phenotype of the slyA gene was unable to be complemented, it is possible that other genes were affected by the mutation.

htrA encodes a heat shock protein which is involved with degradation of aberrant periplasmic proteins. An htrA mutant S. typhimurium strain was attenuated in mice and was able to elicit protective immunity (35). Although the infection by htrA mutant S. typhimurium was consistent with a growth defect, the mutants were also susceptible to oxidative stress and may also be sensitive to killing by phagocytes (99).

Immunosuppression Mediated by S. typhimurium.

Mice infected with a live, attenuated S. typhimurium aroA strain experienced a transient nonresponsiveness to the mitogens LPS and concanavalin A (109) as well as decreased antibody production against heterologous antigens (6). Since immunosuppression was also detected after infection with Listeria monocytogenes (21), it is probably a nonspecific effect in the regulation of inflammation by the host. The salmonella-induced unresponsiveness was mediated by nitric oxide, an immunosuppressive compound produced by macrophages (7, 54). Nauciel and coworkers (157) detected immunosuppression in salmonella-infected mice associated with the H-2 ^b haplotype and Ity^s allele. The lack of response was due to lack of production of and responsiveness to interleukin 2 (IL-2). In other studies, a cell extract of S. typhimurium inhibited mitogen-induced proliferation of lymphocytes in vitro (132), possibly by inhibiting protein kinase C and phosphorylation of tyrosine residues on certain host proteins (131). Hoertt et al. (82) reported that wild-type, but not virulence plasmid-cured, S. typhimurium suppressed mitogen-induced proliferation of spleen cells. However, the roles, if any, of these immunosuppressive effects in pathogenesis of S. typhimurium have not been determined.

Role of Lymphocytes in Early Phases of S. t yphimurium Infection.

The relative roles of T cells and B cells as major factors in protective immunity to salmonellosis are discussed below. However, recent evidence indicates that S. typhimurium interacts with these lymphocytes directly as part of pathogenesis. S. typhimurium can invade human T-cell and B-cell lines in vitro (202), and infection of chickens with S. typhimurium resulted in depletion of lymphocytes from reticuloendothelial organs (79). The mechanism(s) for the lymphoid depletion is not known. αβ-T cells are not important in the early phase of infection with S. typhimurium (88, 125, 158; Matsui et al., submitted); however, these lymphocytes are critical in specific immunity as a protective factor in systemic salmonellosis, as discussed below.

NK cells have traditionally been thought of as defense cells against tumors and viruses. However, NK cells are involved in both short-term, nonspecific defense and specific immunity to S. typhimurium and other bacteria (70). Treatment of salmonella-immunized mice with antiserum to the NK cell surface antigen asialo GM1, which depletes mice of NK cells, reversed the protection by immunization (182). Oral inoculation of mice with S. typhimurium resulted in production of IFN-γ by NK cells in the spleen (173). NK cells are involved in the production of IFN-γ after stimulation with IL-12 and tumor necrosis factor alpha (TNF-α) (20).

γ δ-T cell lymphocytes are also involved in protection against salmonellosis in mice. γδ-T cells are present in epithelial and mucosal tissues found at the reproductive and digestive tracts, skin, and the peritoneal cavity (78). It is believed that these T cells preferentially respond to heat shock proteins, which are antigenically conserved from prokaryotes to humans. Mixter et al. (142) determined that γδ-T cells were necessary for long-term protection against infection by S. typhimurium since the oral 50% lethal dose was significantly decreased in mice pretreated with antibody to the γδ-T cell receptor. Emoto et al. (57) reported that plasmid-cured, but not wild-type, Salmonella choleraesuis elicited γδ-T cells into the liver and peritoneal exudate of i.p.-inoculated mice within a week of infection. The virulence plasmid could therefore have a direct or indirect role in suppressing an early γδ-T cell response. However, an alternative possibility is that a γδ-T cell response indicates the successful check of a bacterial infection by the host and that these T cells are involved in clearing the products of the host-pathogen interaction. Menon, Sleasman, and Gulig recently determined that both wild-type and virulence plasmid-cured S. typhimurium strains elicit similar γδ-T cell responses to the peritoneal cavities of mice (unpublished data). The mechanism by which γδ-T cells interact with salmonellae is not yet known, but these T cells are capable of secreting IFN-γ, as well as other cytokines (192).

Role of Cytokines in Salmonella Infection.

Initial studies of cytokines and their interactions with salmonella infection involved treatment of nonphagocytic cells in cell culture with IFN-γ or TNF-α. These cytokines inhibited the invasiveness of salmonellae into the host cells (49), and IFN-γ increased phagolysosomal fusion in murine macrophages infected in vitro (91). The role of cytokines in salmonellosis in mice has been studied in two ways: inhibiting naturally produced cytokines with antibodies or administering recombinant cytokines. The most studied cytokines are TNF-α and IFN-γ, which can be produced as the result of nonspecific and specific stimulation of appropriate host cells. In mice, injection of purified murine fibroblast IFN (IFN-α and IFN-β) (26) or purified recombinant IFN-γ (72) inhibited oral infection with S. typhimurium, and administration of anti–IFN-γ antibody increased infection (150, 156, 172). The primary antibacterial effect of IFN-γ appeared to be suppression of the bacterial growth rate, as opposed to killing by the host (150, 156). Nitric oxide induced by IFN-γ results in the efflux of nonheme iron from infected macrophages (205) and hence could suppress intracellular bacterial growth. The fact that infection was increased when these cytokines were inhibited is indirect proof that the cytokines are produced as a result of infection with salmonellae. Ramarathinam et al. detected IFN-γ mRNA in Peyer’s patches, mesenteric lymph nodes, and spleens after oral inoculation of BALB/c mice with S. typhimurium (172). Production of IFN-γ in infected mice was related to the Ity genotype of the host, and IFN-γ was produced by NK cells in a TNF-α- and macrophage-dependent manner (174). The availability of transgenic mice unable to synthesize IFN-γ (GKO mice) will no doubt be important in further analysis of the importance and function of this cytokine in salmonella as well as other microbial infections (62).

Administration of TNF-α reduced intestinal infection and bacteremia after oral inoculation with S. typhimurium (48) and helped clear salmonellae from the peritoneal cavity of i.p.-inoculated mice (152). Treatment of mice with rabbit anti–TNF-α antibodies followed by i.v. inoculation with salmonellae had demonstrable effects only on sublethal infection of Ity^r mice; the course of lethal infection was unaltered (126). It appeared that the major role of TNF-α was in suppressing exponential growth of bacteria in tissues of Ity^r mice for the plateau phase of sublethal infection. The earlier net growth rate was unaltered. Different groups reported that TNF-α could not be detected in blood during infection of mice with S. typhimurium (48, 111). The lack of detectable TNF-α activity could have been due to TNF-α inhibitory activity in infected mice (127). In addition to IFN-γ and TNF-α, IL-1 has been reported to increase resistance of mice to challenge with S. typhimurium (141, 146). Monocyte chemotactic and activating factor also can boost resistance to infection by S. typhimurium (154).

The roles of cytokines in protective immunity have been examined by inhibition with antibodies in immunized mice. Inhibition of IFN-γ and TNF-α after immunization with an attenuated S. typhimurium strain inhibited protective immunity that normally would have been conferred by the immunization (130, 151, 156, 194). The idea that macrophages activated by IFN-γ suppress S. typhimurium infection was recently challenged (112, 113). Infection of mice with a sublethal dose of S. typhimurium or administration of IFN-γ stimulated macrophages, but did not increase killing of salmonellae or other bacteria. Conflicting results have been reported (107). Production of cytokines may be related to the genetic susceptibility of mice to infection, with resistant mice producing higher levels of cytokines than do susceptible mice (12, 110).

It is clear that cytokines such as IFN-γ and TNF-α are essential for nonspecific resistance early in infection, as well as protective specific immunity. Progress is being made in differentiating the antibacterial host cell functions affected by these cytokines at the level of bactericidal versus bacteriostatic effects. The most exciting areas for continuing research are elucidating the cellular and molecular mechanisms of action of the cytokines on host cells and the host cell effector functions regulated by cytokines. Undoubtedly, these studies will also reveal mechanisms by which bacteria evade the cytokine-regulated antibacterial functions.

Replication of S. typhimurium within the Host and Bacterial Physiology as a Virulence Attribute

Pathogenic bacteria must be able to replicate within the host after the initiation of infection to achieve sufficient numbers to cause damage. In host cells, salmonellae remain within membrane-enclosed vacuoles, and within macrophages or PMNs salmonellae encounter antibacterial defenses making intracellular replication within phagocytes even more challenging.

Measuring Replication In Vivo.

The analysis of bacterial replication in an animal host is much more complicated than determining the net yield of bacteria from tissues over a period of time. The yield of bacteria represents the sum of growth, death, and movement to the tissue or site. Measuring bacterial replication in vivo has been studied by endowing the bacteria with an unstable genetic element that is lost from one of the two progeny bacteria with each cell division. In this manner, the proportion of the bacterial population maintaining the genetic marker inversely reflects the growth rate. Initially, salmonellae were pseudo-lysogenized with phage P22 in which the nonreplicative bacteriophage served as a genetic marker, and resistance to superinfection by phage P22 provided a measure of the pseudo-lysogenized salmonellae (134). The measured growth rate of S. typhimurium in mouse spleens, as well as the death rate, were very slow. The same procedure was used to show that the growth rate of S. typhimurium was almost 3 h in sensitive mice but 5 h in resistant mice (86). Benjamin et al. used the temperature-sensitive plasmid pHSG422 as the unstable genetic marker and found that Ity^r mice restricted the growth of salmonellae more than did Ity^s mice (14). The mechanisms by which the animal host suppresses bacterial replication and the opposing mechanisms by which the bacteria overcome these defense functions are largely unknown.

Acquisition of Iron.

Interstitial fluid which bathes animal tissues is essentially composed of plasma, so it is relatively rich in nutrients. Mammalian hosts inhibit bacterial replication in this rich environment by limiting critical nutrients such as iron, using transferrin and other iron-binding proteins. To overcome iron-binding defenses, some bacteria secrete siderophores, which scavenge iron from the host (reviewed in references 43 and 167 and chapter 71). S. typhimurium has the potential to produce two different types of siderophores—most commonly enterobactin (also called enterochelin) or aerobactin. However, despite the teleological arguments for the importance of enterobactin in pathogenicity, production of enterobactin was not required for virulence of S. typhimurium in mice (15). This may be due to the fact that sufficient unsequestered iron is present in the intracellular environment where salmonellae replicate in the host.

Prototrophy as a Virulence Factor.

For bacteria to replicate, they must synthesize or acquire all of the components of a bacterial cell. Prototrophic S. typhimurium and E. coli can synthesize a bacterial cell, given a carbon source such as glucose, ammonium salts, and other essential elements. However, in an infected host, auxotrophic bacteria will not replicate and will be attenuated for virulence unless the required nutrients are available at the site of infection. Identifying which auxotrophies cannot be phenotypically complemented by the host therefore identifies steps of metabolism that could be considered virulence factors. Auxotrophic salmonellae therefore have been used as live, attenuated vaccine strains (reviewed in references 28 and 46). Requirements for p-aminobenzoic acid (8), aromatic compounds including the aromatic amino acids, and enterobactin (84) were the first auxotrophies to be examined in mice. AroA^– S. typhimurium strains are deficient at replicating in the murine host, but are not affected in killing by the host (14). Biosynthesis of purines (135), pyrimidines (122), and heme-containing compounds (13) is also essential for complete virulence. hemA S. typhimurium is unable to synthesize glutamyl-tRNA reductase necessary for the synthesis of glutamate-1-semialdehyde and heme-containing molecules including cytochromes and catalase. These bacteria are unable to grow in the presence of oxygen or in mice (13). Curtiss and Kelly determined that cya/crp mutants, lacking adenylate cyclase and the cyclic AMP receptor protein, are attenuated in mice and that Cya^–/Crp^– S. typhimurium mutants have great potential for use as vaccines (45). Although deletion of the genes involved in catabolite repression would be expected to disrupt many metabolic pathways and probably overt virulence factors, the exact mechanism of attenuation by cya/crp mutations is not known; however, it is noted that the bacteria grow very slowly, even in rich media. Although not directly involved in prototrophy, the regulatory system containing OmpR (51) is necessary for complete virulence. Attenuation of ompR mutants is probably due to the aberrant expression of the OmpR-regulated outer membrane proteins OmpC and OmpF (34). Mutations in either ompC or ompF do not significantly attenuate, but mutation of both genes does reduce virulence.

Virulence Plasmid-Mediated Replication.

The spv genes of the S. typhimurium virulence plasmid are involved in replication in the host in an undetermined manner (77). The virulence plasmid is required for the efficient infection of deep tissues such as the spleen and liver of orally inoculated mice (reviewed in reference 76). Early hypotheses of the mechanisms for plasmid-mediated systemic infection included adherence and invasion into host cells, resistance to complement, and resistance to macrophages. However, the virulence plasmid increases the growth rate of salmonellae within the deep tissues without affecting the ability of the bacteria to reach these tissues or to be resistant to killing mechanisms of the host upon arrival there (77). Plasmid-mediated growth probably occurs within host cells, as opposed to the blood or extracellular fluid. Several mechanisms can be hypothesized to explain the increase in growth rate within host cells. The spv genes may help the salmonellae produce or acquire nutrients which improve the growth rate within particular host cells; the plasmid may help salmonellae reach a more permissive environment for rapid growth; or the plasmid may inhibit the host from suppressing intracellular bacterial growth (e.g., through cytokine-mediated mechanisms). Consistent with a role for the spv genes in replication, the alternative sigma factor, RpoS, regulates spv genes and genes expressed during starvation and under stress conditions (58).

Salmonella mviA and Murine Ity.

The mviA locus of S. typhimurium regulates the ability of salmonellae to exploit the susceptibility defect of Ity^s mice (see Host Genes Involved in Resistance to Salmonellosis, below) (16). On the basis of DNA sequence analysis, MviA appears to be the second component of a previously unidentified two-component regulatory system (W. H. Benjamin, Jr., X. Wu, and W. E. Swords, submitted for publication). Salmonellae with the appropriate mviA allele show rapid growth in Ity^s mice (16). The mviA locus has little or no effect on salmonella growth in Ity^r mice. The mviA locus was originally defined using salmonella strains that differed in virulence in Ity^s mice. In the LT2 background, virulence was conferred by a nonfunctional mviA allele, and avirulence was associated with a dominant functional allele. When non-LT2 virulent S. typhimurium strains were examined, however, all had a functional mviA allele. In these strains a nonfunctional mviA allele leads to avirulence (W. E. Swords, A. Giddings, and W. H. Benjamin, Jr., submitted for publication). A second mutation in the rpoS gene of LT2 appears to account for the reversed effects of mviB on virulence in LT2 and non-LT2 salmonella strains (S. Dunkin, W. H. Benjamin, Jr., W. E. Swords, and J. W. Foster, submitted for publication; W. E. Swords and W. H. Benjamin, Jr., submitted for publication). mviA regulates the activity of rpoS posttranscriptionally (Dunkin et al., submitted).

Although growth of bacteria has not received an equal share of attention relative to other virulence functions such as cellular adherence and invasion and evasion of defenses, recent studies of bacterial genetics, host genetics, and host defense functions have demonstrated the importance of bacterial growth in virulence. By identifying host defense functions which suppress bacterial growth in vivo, new avenues may be opened for intervention into the disease process by stimulating bacteriostatic functions.

Damage to the Host in Systemic Salmonellosis

The end result of avoiding host defenses and successful multiplication of S. typhimurium is the production of sufficient numbers of bacteria in deep tissues to result in damage. When salmonellae cause systemic infection, manifested as septicemia or enteric fever, the symptoms are primarily those of gram-negative sepsis, i.e., fever, chills, inflammation, hypotension, and, in severe cases, disseminated intravascular coagulation (reviewed in reference 165). Most of these symptoms are caused by the host response to the inflammatory cell wall components LPS and peptidoglycan. However, the inflammatory response is primarily protective in nature in that host defenses are recruited and stimulated to counter infection. If infection is not held in check, the overwhelming number of bacteria can then stimulate an excessive host response with ensuing damage. For example, LPS (endotoxin) causes host cells to trigger cascades of cytokine secretion and other cellular responses. For a review of endotoxic activity, see reference 145. The most important cells for endotoxin-mediated effects are macrophages. LPS first binds the LPS binding protein (195), and the LPS-LPS binding protein complex then binds to the membrane receptor, CD14, present on certain host cells such as macrophages and PMNs (210). In a manner not completely understood, CD14 delivers the LPS into the host cells, and a series of phosphorylations of host proteins ensues (208). The components of lipid A which cause the biological effects in macrophages involve the novel C₁₄ acyl compounds, their secondary conformations, and phosphate groups (85, 177). It has been proposed that the action of endotoxin stems from the biochemical similarities between lipid A and ceramide, resulting in stimulation of a ceramide-activated protein kinase (106). The end result of the intracellular signalling is expression of cytokines. Of most importance is the production of TNF-α, IL-1, and IL-6. These cytokines are responsible for most of the systemic effects of endotoxin, including fever, and collectively act on the liver to produce the acute-phase reaction.

LPS also directly activates the complement cascade. The production of C5a results in inflammation through the recruitment of PMNs which then are stimulated to secrete granular contents and produce ROI. This inflammatory response can produce localized tissue damage as well as vasodilation. Disseminated intravascular coagulation is caused through the ability of LPS to stimulate Hageman factor (factor XII) of the clotting pathway.

Specific Immunity and Protection against Systemic Salmonellosis

S. typhimurium is a facultative intracellular pathogen of both phagocytes, such as macrophages, and nonphagocytes. The type of immunity that can protect against infection of mice with S. typhimurium has been controversial. Antibodies were believed to be protective against extracellular pathogens, while cell-mediated immunity was believed to be protective against intracellular pathogens. Many arguments centered on which single type of immunity was protective in itself, to the exclusion of other forms of immunity. In reality, though, cell-mediated immunity is not generated in the absence of antibodies, and these immune mechanisms usually work together to protect the host (3, 153). This has been recently demonstrated in passive protection experiments in which T cells and B cells acted synergistically to mediate protection against salmonellosis in mice (129). In discussing factors of protective immunity to systemic salmonella infection, one must keep in mind the model of disease being investigated, i.e., the route of inoculation and the end point of disease. For example, is protection measured as survival from an otherwise lethal dose, an extension of survival even if death ultimately occurs, or the reduction of bacteria recovered from tissues after inoculation? Protective immunity against oral challenge could be very different from that of i.v. or i.p. challenge.

Role of Humoral Antibodies.

There has been considerable confusion and controversy as to the relative role of humoral immunoglobulin G (IgG) in a protective immune response against S. typhimurium. Initially, investigators attempted to examine whether antibodies alone could protect mice from challenge with salmonellae. There are several problems with this approach. First, it would not be expected that antibodies alone would be present in a vaccinated or convalescent host. Second, many researchers attempted to address this question by using active immunization which they dogmatically believed to result exclusively in humoral antibodies. Passive protection experiments, although somewhat artificial, do provide an avenue for controlling which factors or cells are initially present in the recipient. In the case of the mouse model, another shortfall in applicability to humans is that mice are defective in bactericidal activity of complement. Therefore, while S. typhimurium is readily killed by antibodies and complement from humans, in mice complement primarily serves as an opsonic function.

In the mouse model, it appears that the primary role of IgG is to opsonize salmonellae to aid in their clearance by the reticuloendothelial system. If salmonellae are able to replicate within phagocytes and if opsonization with antibodies and complement does not alter the salmonella-phagocyte interaction in a way that alters the ultimate outcome, then antibodies would not be protective. Experiments designed to probe the role of circulating antibodies in murine salmonellosis have generally arrived at this conclusion (38, 55). However, opsonizing antibodies coupled with cell-mediated immunity offer an effective manner of delivering the bacteria to the phagocytes for their ultimate destruction (129). Many investigators have focused on the role of antibodies directed against LPS (41, 179, 191). When survival of the immunized host is used as an end point, conclusions are difficult to draw because antibodies could neutralize endotoxin rather than clear the pathogen. However, several groups have achieved protection by using either vaccines from fractionated salmonella cells or LPS O antigen conjugated to protein (reviewed in reference 191).

Role of Cell-Mediated Responses.

Cell-mediated immunity to intracellular pathogens (108) and S. typhimurium specifically (40, 56, 89) has been reviewed. According to the present understanding, stimulation of specific T helper cells of the Th1 subclass results in the production of cytokines such as IFN-γ that activate macrophages and enable the killing of intracellular pathogens. Much of the earliest work correlated protective immunity with delayed-type hypersensitivity or other immune responses. Conversely, there were negative correlations between protection and immunization that were believed to yield only humoral immunity. However, for analysis of cell-mediated immunity to S. typhimurium, the most definitive results involve passive transfer of lymphocytes or elimination of lymphocytes. Passive immunity with T-cell lines and immune T cells affords some protection against challenge (166). The role of T-cell-mediated immunity in salmonella infection has also been demonstrated by depletion of T cells with antibody treatment either of mice or of cell populations used for passive immunity (47, 128, 133, 155). However, in one study, the stimulation of macrophages by prior sublethal salmonella infection did not result in increased killing of salmonellae (113). NK cells are also an important component of protection early after immunization (182) although NK cells can be stimulated in a T-cell-independent manner via IL-12 (20, 196).

Role of Secretory Immunity.

Since the natural route of infection for S. typhimurium is through the gastrointestinal tract, it is possible that secretory immunity, mediated by secretory IgA (sIgA), could protect against infection. Michetti et al. (136) recently protected mice by using a backpack hybridoma to deliver IgA for passive immunity. The antibody was produced by hybridoma cells implanted into the backs of the mice, from which the sIgA was transported into the intestinal tract. The monoclonal antibody was directed against the O5 epitope of the LPS (187). The IgA protected only against oral challenge, but not i.p. challenge, and the IgA prevented infection of the Peyer’s patches by the salmonellae. This study also demonstrated that secretory immunity could in itself be protective against natural disease. sIgA in the intestines is believed to act by inhibiting adherence or invasion of bacteria and neutralization of toxins. Oral immunization of mice and other animals with live, attenuated S. typhimurium can stimulate all arms of the immune system, including sIgA (28, 46, 184).

Host Genes Involved in Resistance to Salmonellosis

Different strains of mice exhibit different levels of resistance or susceptibility to salmonella infection. As the genetics of mice become more developed, in conjunction with an improved understanding of the virulence mechanisms of S. typhimurium, a clear, but incomplete, picture has emerged for the role of host genes in resistance to salmonella infection (reviewed in references 18 and 186). As is the case for resistance to disease in other human and animal models, resistance of mice to salmonella infection is multigenic (18). The gene with the greatest effect is probably Ity (168), which has a major effect on salmonella pathogenesis during week 1 of infection. Mice can carry either the recessive salmonella-sensitive (Ity^s) or dominant salmonella-resistant (Ity^r) allele. Mice that are Ity^s/s show a high net increase in the numbers of salmonellae in their spleens and livers and die of sepsis before a protective immune response can be produced. This locus appears to be synonymous with the Bcg and Lsh loci which have been described to control the resistance to infection with the intracellular pathogens Mycobacterium bovis BCG and Leishmania donovani in mice (169). By using nonreplicating phage or plasmids to measure growth rate, it has been shown that the major effect of Ity^r is an almost complete cessation of salmonella growth in mice, although there is a small effect of Ity^r on killing of salmonellae (14, 86). In the case of in vitro studies using infected phagocytes, the major effect of these loci has been on killing rather than growth (120, 200, 201). The differences between the in vivo and in vitro results emphasize the importance of in vivo investigations, since complex host-pathogen interactions may not be reproduced adequately in vitro. As discussed above, in response to stimulation with salmonellae, Ity^r mice produce more IFN (12, 172) and other cytokines (110) than do Ity^s mice. The mouse gene responsible for immunity to M. bovis BCG (and presumably salmonellae) has been cloned. The gene, Nramp (natural resistance-associated macrophage protein), produces a 53-kDa protein with sequence homology to "binding-protein-dependent transport system inner membrane component signature" (203).

The Lps locus affects susceptibility of mice during the early phase of infection. Mice homozygous for the Lps^d allele have a more rapid net increase of salmonellae than do mice with the wild-type allele at this locus (160). Macrophages of Lps^d mice fail to respond to endotoxin and therefore do not recruit cellular and inflammatory responses which would normally protect against infection (207). In addition, the Lps^d allele results in greatly increased susceptibly to ascending urinary tract infections with Salmonella strains and E. coli in mice (190). Although an inflammatory response to LPS can be an important cause of shock and death in patients with gram-negative bacterial sepsis, the data from mice indicate that the ability to make this response is clearly a protective adaptation of the host, rather than a virulence adaptation of gram-negative organisms.

In mice that live longer than 1 or 2 weeks postinfection, genes that affect acquired immunity become important. Three loci that affect acquired immune responses to salmonellae are Xid, nu, and H-2. The Xid locus affects the differentiation of B cells and results in aberrant responses to bacterial polysaccharides, including the O antigen of LPS, but relatively normal responses to most proteins. It is assumed that the failure of Xid ^– mice to be able to make antibodies to LPS is responsible for their greater susceptibility to salmonella infections. This supports the importance of anti-LPS antibodies in protective immunity. Mice homozygous at nu fail to develop a functional thymus and therefore lack T-cell-dependent cellular and humoral immune responses. Such mice experience normal net growth of salmonellae during the first few weeks postinfection, but eventually die because they are unable to develop a protective T-cell-mediated immune response that suppresses salmonellae in the liver and spleen (158). This demonstrates the essential nature of T-cell-dependent cellular immunity in protection. The importance of T-cell-dependent immune responses is also demonstrated by the fact that some H-2 haplotypes confer greater resistance late in the course of infection than do others. Immune response genes at this locus are known to be able to have significant effects on T-cell-dependent immune responses (87, 124).

HOW DOES E. COLI CAUSE SYSTEMIC DISEASE?

The major systemic diseases caused by different strains of E. coli include pyelonephritis, meningitis, and sepsis. The organisms causing these disease entities must perform the same general virulence functions described above for S. typhimurium in systemic infection: spreading through tissues, resisting host defenses, multiplying within the host, and causing damage. Pyelonephritis is an ascending extension of urinary tract infection which initiates in the lower urinary tract and bladder. Bacteria causing pyelonephritis usually invade beyond the epithelium and are frequently associated with bacteremia; therefore, this infection is discussed in terms of its systemic nature. Neonatal meningitis, caused primarily by E. coli K1, starts with infection of the mucosa followed by invasion into the bloodstream and infection of the central nervous system. Sepsis caused by E. coli can be the result of pyelonephritis, meningitis, or other predisposing conditions of patients. As discussed below, the virulence factors associated with sepsis of compromised individuals are somewhat different than those of meningitis or pyelonephritis (98, 185). Many of the data concerning potential virulence factors for E. coli pyelonephritis and meningitis are based on epidemiological correlations, which by themselves cannot be considered definitive proof. Although the frequency of a potential virulence attribute in systemic isolates versus fecal isolates may be significantly higher, in many cases fewer than half of systemic disease isolates possess the particular trait. However, a mouse model for pyelonephritis and a neonatal rat model for meningitis have been used to conduct more definitive studies using genetic manipulation of the pathogens.

Spread through Host Tissues

The systemic infections caused by E. coli initiate at a mucosal surface, the urinary tract, the upper respiratory tract, or the intestinal tract. For pyelonephritis the P pili and other adhesins play an integral role in the ascending nature of the infection and damage that occurs in the kidneys, while S pili are associated with meningitis and sepsis (reviewed in chapter 150). Uropathogenic E. coli strains probably spread beyond the renal epithelium as a result of cytotoxicity and damage that occurs at that site (97). However, invasion of epithelial cells by pyelonephritogenic E. coli in vitro has been reported (50). The mechanism by which E. coli K1 invades beyond the epithelium is not known. In any case, the bacteria enter the bloodstream and then gain access to the central nervous system through the blood-brain barrier. Inflammation of the blood-brain barrier enables gram-negative bacteria to gain access to the central nervous system. The focal site of infection in other cases of sepsis can be from an instrument placed in or trauma to the urinary tract or intestinal tract.

Resistance to Host Defenses

The lumens of the urinary and intestinal tracts usually do not possess nonspecific host defenses in the forms of phagocytes or complement. However, the presence of E. coli in the kidneys results in inflammation and infiltration of phagocytes, PMNs in particular. If the bacteria invade beyond the epithelium of the kidney, they will additionally encounter complement in interstitial fluid or blood. Therefore, pyelonephritogenic E. coli must be resistant to these defenses. From the time that meningitis-causing E. coli strains invade beyond the mucosa through the blood to their ultimate site of infection, the meninges and central nervous system, the bacteria must contend with phagocytes and complement. After the bacteria cross the epithelium and become systemic, the pathogenesis of E. coli is very different from that of S. typhimurium. The systemic E. coli strains are extracellular pathogens that avoid phagocytosis primarily by having carbohydrate capsules (slime layers) surrounding their outer membrane. The capsules not only prevent the attachment of phagocytes to the bacteria, but they also prevent the activation and deposition of complement on the bacterial surface (95). The most frequently isolated capsular type of E. coli from septicemia or meningitis is the K1 type, composed of polysialic acid. The biochemical composition of the K1 capsule is identical to that of the capsule of group B Neisseria meningitidis (95). As is the case for many other antiphagocytic capsules, antibodies to the capsules could in theory overcome the virulence factor. However, in addition to acting as a defense against complement and phagocytes, the polysialic acid capsule also mimics the host in that sialic acid is a frequent component of cell surface glycolipids and glycoproteins of host cells (95). Therefore, the host will not produce an immune response to the capsule. The LPS of E. coli K1 appears to contribute to pathogenesis, since only a limited number of O-antigen types are found in systemic disease (185), and certain LPS serotypes retain serum resistance in the absence of capsule (44). O-acetylated O antigen is associated with more severe sepsis by E. coli (64). Similar to the situation for S. typhimurium, LPS-nonresponsive mice (Lps^d) are hypersusceptible to E. coli K1 infection (170) and urinary tract infection (190). The hemolysin produced by many pyelonephritic E. coli strains also has the potential to kill PMNs and inhibit their function (32, 33, 123).

Replication

The major potential virulence factor associated with replication of systemic E. coli is the production of the siderophores enterobactin and/or aerobactin. However, much of the relationship between siderophores and virulence is based on epidemiological correlations between production of siderophores by systemic E. coli isolates versus fecal isolates (199). In some studies, the frequency of enterobactin was higher than that of aerobactin (67). However, not all such studies have arrived at a definite correlation of these siderophores with virulence (197). Aerobactin can be encoded on certain colicin plasmids or on antibiotic resistance plasmids but can also be chromosomally encoded in pathogenic E. coli, as is the usual case for enterobactin (43).

Damage

The presence of gram-negative bacteria within body tissues always has the potential to stimulate damage by inflammation from endotoxin. In pyelonephritis, extracellular generation of ROI by PMNs is associated with severe disease (149). Inflammation of the kidneys can lead to ischemia and necrosis (180). For the systemic isolates of E. coli, additional virulence factors, hemolysin and cytotoxic necrotizing factor, are frequently present (180, 185, 197; reviewed in chapter 153). Both of these toxins can kill host cells, leading to tissue damage (143). In the mouse model of pyelonephritis, antibodies to hemolysins were protective (161). However, when the severity of disease among human patients was compared with the putative virulence attributes of the E. coli isolates from the patients, no clear relationship was found between hemolysin and disease (42).

HOW CAN UNDERSTANDING THE DISEASE PROCESS CONTRIBUTE TO ALLEVIATING DISEASE?

The studies of pathogenesis of systemic diseases caused by S. typhimurium and E. coli outlined in this chapter demonstrate the understanding gained over the past two decades. As indicated throughout, there are still several important areas that require further study. Although this information is interesting and valuable in its own right, it can be put to use in improving medical and veterinary practice. The most obvious manner is in vaccine development. The abilities of S. typhimurium to infect the Peyer’s patches and invade the lymphoid system, to be attenuated, and to express cloned foreign antigens to stimulate all three arms of the immune system have been reviewed (28, 46, 184). S. typhimurium and other related species therefore offer significant promise for development as live, attenuated vaccines for both humans and animals. Furthermore, the elucidation of host-pathogen interactions in systemic disease could offer the promise of improved therapeutic intervention once disease has been initiated. The unraveling of the complex cytokines and host defense cells involved in suppressing bacterial infection will undoubtedly lead to immunotherapy either with, or aimed at inducing the production of, relevant cytokines.

ACKNOWLEDGMENTS

I thank David E. Briles and William H. Benjamin, Jr., for their guidance and comments in the preparation of this work, and I thank Donna Duckworth for her review of the manuscript.