Response to Shiga toxin-1, with and without lipopolysaccharide, in a primate model of hemolytic uremic syndrome.

Shiga toxin (Stx) and lipopolysaccharide (LPS) both participate in the pathogenesis of post-diarrheal hemolytic uremic syndrome (HUS), yet little is known about the factors that modulate the host response to these toxins. We have previously shown that the baboon develops HUS if 100 ng/kg of purified Stx-1 is administered rapidly as a single bolus, but not if it is given as four 25-ng/kg doses every 12 h. We therefore used this baboon model to study the response to small intravenous doses of Stx-1, with and without the co-administration of LPS. The co-administration of two 1-mg/kg doses of LPS (given at 0 and 24 h) and four 25-ng/kg doses of Stx-1 (given at 0, 12, 24, and 36 h) resulted in HUS, but the administration of either toxin separately did not. The development of HUS was associated with a rise in urinary, but not plasma concentrations of TNF, and a rise in both urinary and plasma concentrations of IL-6 and IL-8. We speculate that LPS is not required for disease expression in the human, but that it can augment the response to otherwise subtoxic amounts of Stx and this augmentation may be mediated by LPS-induced cytokine release.

Shiga toxin-induced tumor necrosis factor alpha expression: requirement for toxin enzymatic activity and monocyte protein kinase C and protein tyrosine kinases.

Infections with Shiga toxin (Stx)-producing bacteria cause bloody diarrhea which may progress to life-threatening complications, including acute renal failure and neurological abnormalities. The precise mechanism of disease progression is unclear, although evidence suggests that the localized production of the host proinflammatory cytokines tumor necrosis factor alpha (TNF-alpha) and interleukin-1 may exacerbate toxin-mediated vascular damage. Purified Stxs have been demonstrated to elicit proinflammatory cytokine synthesis from human peripheral blood mononuclear cells and monocytic cell lines in vitro. To understand toxin-monocyte interactions required for cytokine synthesis, we have treated differentiated THP-1 cells with purified wild-type toxins, enzymatic mutants, or B subunits and measured TNF-alpha production. Our data suggest that A subunit enzymatic activity is essential for cytokine production. THP-1 cells were treated with a series of protein kinase C (PKC), PKA, and protein tyrosine kinase inhibitors to examine the role of intracellular signaling molecules in Stx-mediated cytokine production. Treatment of cells with PKC and tyrosine kinase inhibitors blocked TNF-alpha secretion by Stx-stimulated THP-1 cells. Stx treatment directly activated PKC, which occurred at a point upstream of transcriptional activation of the gene encoding TNF-alpha.

Interaction of Shiga toxins with human brain microvascular endothelial cells: cytokines as sensitizing agents.

Neurologic abnormalities are among the most serious extraintestinal complications of infection with Shiga toxin (Stx)-producing bacteria. Histopathologic examination of tissues from patients with extraintestinal sequelae suggested that Stxs damage endothelial cells. It is shown here that human brain microvascular endothelial cells (HBMECs) are relatively resistant to purified Stxs (50% cytotoxic doses [CD50s] >/=10 microgram/mL). Pretreatment of HBMECs with tumor necrosis factor (TNF)-alpha, interleukin (IL)-1beta, n-butyric acid, or a cAMP analogue resulted in a 103- to 104-fold decrease in CD50 values and a 2- to 4-fold increase in fluoresceinated Stx binding to HBMECs. Treatment of HBMECs with lipopolysaccharides did not significantly alter cytotoxicity or toxin binding. TNF-alpha and IL-1beta treatment was associated with the increased HBMEC expression of the toxin-binding glycolipid globotriaosylceramide. HBMECs did not produce IL-1beta and produced only trace amounts of TNF-alpha when stimulated with purified Stx1 in vitro.

Characterization of the baboon responses to Shiga-like toxin: descriptive study of a new primate model of toxic responses to Stx-1.

The baboon response to intravenous infusion of Shiga toxin 1 (Stx-1) varied from acute renal failure, proteinuria, hyperkalemia, and melena with minimal perturbation of host inflammatory and hemostatic systems (high-dose group, 2.0 microg/kg; n = 5) to renal failure with hematuria, proteinuria, thrombocytopenia, schistocytosis, anemia, and melena (low-dose group, 0.05 to 0.2 microg/kg; n = 8). Both groups exhibited renal shutdown and died in 57 hours or less. Both groups produced urine that was positive for tumor necrosis factor and interleukin-6 although neither of these cytokines was detectable (

Shiga toxin type 1 activates tumor necrosis factor-alpha gene transcription and nuclear translocation of the transcriptional activators nuclear factor-kappaB and activator protein-1.

Shiga toxins (Stxs) produced by Shigella dysenteriae 1 and Escherichia coli have been implicated in the pathogenesis of bloody diarrhea, acute renal failure, and neurologic abnormalities. The pathologic hallmark of Stx-mediated tissue damage is the development of vascular lesions in which endothelial cells are swollen and detached from underlying basement membranes. However, in vitro studies using human vascular endothelial cells demonstrated minimal Stx-induced cytopathic effects, unless the target cells were also incubated with the proinflammatory cytokines tumor necrosis factor-alpha (TNF-alpha) or interleukin-1beta (IL-1bta). These cytokines have been shown to upregulate the expression of the Stx-binding membrane glycolipid globotriaosylceramide (Gb3). We show here that purified Stx1 induces TNF secretion by a human monocytic cell line, THP-1, in a dose- and time-dependent manner. Treatment of cells with both lipopolysaccharides (LPS) and Stx1 results in augmented TNF production. Treatment with the nontoxic Gb3-binding subunit of Stx1 or with an anti-Gb3 monoclonal antibody did not trigger TNF production. Northern blot analyses show that Stx1 causes increased TNF-alpha production through transcriptional activation. Increased levels of TNF-alpha mRNA are preceded by the nuclear translocation of the transcriptional activators NF-kappaB and AP-1 and the loss of cytoplasmic IkappaB-alpha. These data are the first to show that, in addition to direct cytotoxicity, Stxs possess cellular signaling capabilities sufficient to induce the synthesis of cytokines that may be necessary for target cell sensitization and the development of vascular lesions.

Virulence of enterohemorrhagic Escherichia coli: role of molecular crosstalk.

Many bacterial exotoxins, originally defined by cytopathic effects, may also possess additional biological activities. The capacity of exotoxins to elicit the synthesis and secretion of pro- or anti-inflammatory cytokines may be as important as their direct toxic effects in pathogenesis. One example of such 'molecular crosstalk' occurs between Shiga toxins and the cytokines made in response to these toxins during the development of disease.

Differentiation-associated toxin receptor modulation, cytokine production, and sensitivity to Shiga-like toxins in human monocytes and monocytic cell lines.

Infections with Shiga toxin-producing Shigella dysenteriae type 1 or Shiga-like toxin (SLT)-producing Escherichia coli cause bloody diarrhea and are associated with an increased risk of acute renal failure and severe neurological complications. Histopathological examination of human and animal tissues suggests that the target cells for toxin action are vascular endothelial cells. Proinflammatory cytokines regulate endothelial cell membrane expression of the glycolipid globotriaosylceramide (Gb(3)) which serves as the toxin receptor, suggesting that the host response to the toxins or other bacterial products may contribute to pathogenesis by regulating target cell sensitivity to the toxins. We examined the effects of purified SLTs on human peripheral blood monocytes (PBMn) and two monocytic cell lines. Undifferentiated THP-1 cells were sensitive to SLTs. Treatment of the cells with a number of differentiation factors resulted in increased toxin resistance which was associated with decreased toxin receptor expression. U-937 cells, irrespective of maturation state, and PBMn were resistant to the toxins. U-937 cells expressed low levels of GB(3), and toxin receptor expression was not altered during differentiation. Treatment of monocytic cells with tumor necrosis factor alpha (TNF-alpha) did not markedly increase sensitivity or alter toxin receptor expression. Undifferentiated monocytic cells failed to synthesize TNF and interleukin 1beta when treated with sublethal concentrations of SLT type I (SLT-I), whereas cells treated with 12-0-tetradecanoylphorbol-13-acetate acquired the ability to produce cytokines when stimulated with SLT-I. When stimulated with SLT-I, U-937 cells produced lower levels of TNF than PBMn and THP-1 cells did.

Purified Shiga-like toxins induce expression of proinflammatory cytokines from murine peritoneal macrophages.

Infections with Shiga toxin-producing Shigella dysenteriae type 1 and Shiga-like toxin (SLT)-producing Escherichia coli cause outbreaks of bloody diarrhea in which patients are at risk for developing life-threatening complications involving the renal and central nervous systems. Histopathology studies and in vitro experiments suggested that the toxins damage toxin receptor-expressing endothelial cells (EC) lining glomerular and central nervous system capillaries. In the presence of inducible host factors (cytokines), EC sensitivity to SLT toxicity was increased approximately 1 million-fold. We hypothesized that to manifest the vascular lesions characteristic of infection with toxin-producing bacteria, two signals were needed: systemic toxins and elevated proinflammatory cytokines (tumor necrosis factor alpha [TNF-alpha], interleukin 1 [IL-1], and IL-6). Human EC do not secrete these cytokines when stimulated with SLTs in vitro, suggesting that additional cells may be involved in pathogenesis. Therefore, we carried out comparative analyses of the capacity of purified (endotoxin-free) SLTs and lipopolysaccharides (LPS) to induce cytokine mRNA and proteins from murine macrophages. The cells were essentially refractory to SLT cytotoxicity, expressing low to undetectable levels of toxin receptor. SLTs and LPS induced TNF activity and IL-6 expression from macrophages, although dose response and kinetics of cytokine induction differed. LPS was a more effective inducing agent than SLTs. SLT-I-induced TNF activity and IL-6 expression were delayed compared with induction mediated by LPS. IL-1 alpha production required approximately 24 h of exposure to SLTs or LPS. Macrophages from LPS-hyporesponsive C3H/HeJ mice produced low levels of TNF activity when treated with SLT-I, suggesting that LPS and SLTs may utilize separate signaling pathways for cytokine induction.

Mutation of flgM attenuates virulence of Salmonella typhimurium, and mutation of fliA represses the attenuated phenotype.

Salmonella typhimurium ST39 exhibits reduced virulence in mice and decreased survival in mouse macrophages compared with the parent strain SL3201. Strain ST39 is nonmotile, carries an indeterminate deletion in and near the flgB operon, and is defective in the mviS (mouse virulence Salmonella) locus. In flagellum-defective strains, the flgM gene product of S. typhimurium negatively regulates flagellar genes by inhibiting the activity of FliA, the flagellin-specific sigma factor. In this study, flgM of wild-type S. typhimurium LT2 was found to complement the mviS defect in ST39 for virulence in mice and for enhanced survival in macrophages. Transduction of flgM::Tn10dCm into the parent strain SL3201 resulted in attenuation of mouse virulence and decreased survival in macrophages. However, a flgM-fliA double mutant was fully virulent in mice and survived in macrophages at wild-type levels. Thus, the absolute level of FliA activity appears to affect the virulence of S. typhimurium SL3201 in mice. DNA hybridization studies showed that flgM-related sequences were present in species other than Salmonella typhimurium and that sequences related to that of fliA were common among members of the family Enterobacteriaceae. Our results demonstrate that flgM and fliA, two genes previously shown to regulate flagellar operons, are also involved in the regulation of expression of virulence of S. typhimurium and that this system may not be unique to the genus Salmonella.

Comparison of the relative toxicities of Shiga-like toxins type I and type II for mice.

In earlier studies using a streptomycin-treated mouse model of infection caused by enterohemorrhagic Escherichia coli (EHEC), animals fed Shiga-like toxin type II (SLT-II)-producing strains developed acute renal cortical necrosis and died, while mice fed Shiga-like toxin type I (SLT-I)-producing clones did not die (E. A. Wadolkowski, L. M. Sung, J. A. Burris, J. E. Samuel, and A. D. O'Brien, Infect. Immun. 58:3959-3965, 1990). To examine the bases for the differences we noted between the two toxins in the murine infection model, we injected mice with purified toxins and carried out histopathological examinations. Despite the genetic and structural similarities between the two toxins, SLT-II had a 50% lethal dose (LD50) which was approximately 400 times lower than that of SLT-I when injected intravenously or intraperitoneally into mice. Histopathologic examination of toxin-injected mice revealed that detectable damage was limited to renal cortical tubule epithelial cells. Passive administration of anti-SLT-II antibodies protected mice from SLT-II-mediated kidney damage and death. Immunofluorescence staining of normal murine kidney sections incubated with purified SLT-I or SLT-II demonstrated that both toxins bound to cortical tubule and medullary duct epithelial cells. Compared with SLT-I, SLT-II was more heat and pH stable, suggesting that SLT-II is a relatively more stable macromolecule. Although both toxins bound to globotriaosylceramide, SLT-I bound with a higher affinity in a solid-phase binding assay. Differences in enzymatic activity between the two toxins were not detected. These data suggest that structural/functional differences between the two toxins, possibly involving holotoxin stability and/or receptor affinity, may contribute to the differential LD50s in mice.

The pathogenic mechanisms of Shiga toxin and the Shiga-like toxins.

It is now well documented that some enteric bacteria which cause diarrhoeal and/or dysenteric disease produce, at high levels, one or more of a family of protein toxins referred to as Shiga toxin and Shiga-like toxins (SLTs; alternatively called verocytotoxins or VTs). Within the past few years, there have been considerable advancements made in our understanding of the biochemistry and molecular biology of Shiga toxin and SLTs. However, the precise role of the toxins in mediating colonic disease, as well as their contribution to the development of extra-intestinal sequelae (e.g. the haemolytic uraemic syndrome and neurological disorders), remain less clear. In this MicroReview, we will briefly summarize recent progress in Shiga toxin- and SLT-related research and present evidence supporting the concept that these toxins contribute to pathogenesis by directly damaging vascular endothelial cells, thereby disrupting the homeostatic properties of these cells. We will also discuss data which suggest that toxin-mediated damage in the kidney may not be limited to glomerular endothelial cells but may include tubular epithelial cells. Thus, the role of the toxins in renal disease may not be limited to the glomeruli, as was initially hypothesized when the association of infection with toxin-producing strains and the development of acute renal failure was established.

Evaluation of the role of Shiga and Shiga-like toxins in mediating direct damage to human vascular endothelial cells.

Infection with Shiga toxin- and Shiga-like toxin-producing strains of Shigella dysenteriae and Escherichia coli, respectively, can progress to the hemolytic-uremic syndrome. It has been hypothesized that circulating Shiga toxin, Shiga-like toxins, and endotoxins may contribute to the disease by directly damaging glomerular endothelial cells. The effects of these toxins on HeLa, Vero, and human vascular endothelial cells (EC) were examined. Confluent EC were sensitive to Shiga toxin but were at least 10(6)-fold less sensitive to the toxins than were Vero cells. Shiga toxin was the predominant cytotoxic factor. Lipopolysaccharides were not cytotoxic and did not augment Shiga toxin-mediated toxicity. Lower doses of Shiga toxin caused cytotoxicity when coincubated with tumor necrosis factor. The relative resistance of EC to Shiga toxin and Shiga-like toxins may be due to reduced toxin binding, as low levels of globotriaosylceramide (Gb3), the toxin-specific receptor, were found in EC membranes.

The physical-chemical characterization and biologic activity of serum released lipopolysaccharides.

As a result of the incubation of Escherichia coli in normal human serum, a finite fraction of LPS is released from the bacterial membrane. Approximately half of the LPS released by the action of serum (S-LPS) exists in association with serum proteins in a lower m.w. form than that manifest in phenol-water extracted LPS preparations. The two major LPS-serum protein complexes have apparent Mr of 68 and 32 kDa. The LPS subunit heterogeneity of S-LPS, however, does not appear to differ significantly from LPS retained on the bacteria after serum treatment, or from LPS derived by lysis of whole cells. The biologic activities of S-LPS and phenol-water extracted LPS examined in these studies, differed significantly. In contrast to phenol-water extracted LPS, S-LPS was 1) reduced in lethal toxicity for sensitized mice; 2) reduced in Limulus reactivity; 3) a more potent murine splenocyte mitogen; 4) reduced in the capacity to elicit extracellular, but not membrane-associated IL-1; and 5) reduced in the ability to mediate TNF production. These data suggest that humoral "detoxification" of LPS may involve, in part, the formation of LPS-serum protein complexes with reduced capacities to elicit extracellular cytokine production, whereas the immunomodulatory effects of LPS appear to be enhanced.

The interaction of Escherichia coli with normal human serum: factors affecting the capacity of serum to mediate lipopolysaccharide release.

We previously demonstrated that incubation of E. coli in normal human serum (NHS) resulted in the release of a finite fraction (approximately 30%) of LPS from the bacterial outer membrane. In experiments reported here, we examined factors which may enhance or diminish the capacity of NHS to mediate this limited LPS release. Both the susceptibility to serum killing and LPS release were dependent on growth phase. Optimal killing and release coincided with the midlogarithmic growth phase. The composition of LPS subunits in the outer membrane appeared to influence serum-mediated LPS release. Serum treated E. coli enriched for Rc-chemotype LPS released less LPS from their outer membrane than the wildtype 'smooth' bacteria during exponential growth. LPS fractions released by NHS or EDTA appeared to a large degree to overlap, suggesting that NHS-mediated LPS release may involve the action of a serum chelator. A serum-resistant mutant failed to release LPS in either NHS or EDTA. This latter observation suggests that LPS release may be a relevant event in serum killing. We did not detect any modulation of LPS release when E. coli were pre-incubated with a series of antibiotics prior to treatment with NHS.

The interaction of Escherichia coli with normal human serum: the kinetics of serum-mediated lipopolysaccharide release and its dissociation from bacterial killing.

We have examined the killing of E. coli and kinetics of lipopolysaccharide (LPS) release after the exposure of the bacteria to normal human serum (NHS) and sera deficient in complement components, or with inactivated complement components. LPS of the galactose epimerase-deficient strain E. coli J5 were specifically radiolabeled by growing the bacteria in a medium containing [3H]galactose. Exposure of the washed bacteria to NHS resulted in a significant reduction (greater than 99%) in viability within 15 min and the concomitant release of radiolabeled LPS. However, maximal release of LPS was consistently 30% of the total radiolabel incorporated into the LPS molecules. The amount of tritium-labeled LPS released was shown to be directly proportional to the concentration of bacteria exposed to NHS, suggesting that release of LPS was not limited by the availability of some critical serum component(s). The consumption of complement in NHS by incubation with E. coli was demonstrated by decreased alternative and classical pathway-specific hemolytic activity. The use of Factor D-depleted and VEM-treated human sera demonstrated that, with these bacteria, both the alternative and classical pathways of complement contribute to bacterial killing and release of LPS. It is noteworthy that, in VEM-treated and Factor D-depleted sera, the rate of killing and the kinetics of LPS release were somewhat slower as compared to control serum. Bacterial killing in C7-depleted and C9-deficient human sera was minimal. Neither killing nor LPS release occurred in heat-inactivated (56 degrees C, 30 min) human serum. The amount of [3H]LPS released by C9-deficient serum was qualitatively similar to the amount released by the action of NHS. Tritium-labeled LPS was not released in C7-depleted serum. These data indicate that bacterial killing can be dissociated from LPS release, and suggest that, whereas LPS release may be necessary for the bactericidal effects of serum complement, it is probably not sufficient to effect killing. Furthermore, a significant fraction of LPS can be removed from the outer membrane of the bacteria without an apparent affect on viability.

Immunologic activity of lipopolysaccharides released from macrophages after the uptake of intact E. coli in vitro.

Lipopolysaccharides (LPS) have been isolated from culture supernatants and from cell lysates after the in vitro phagocytosis of E. coli by murine macrophages. By using E. coli radiolabeled specifically in the LPS component with [3H]galactose, our studies have shown that the macrophage-"processed" LPS is enhanced with respect to its immunostimulatory activity in comparison with control phenol-water-extracted LPS. As assessed by its ability to induce interleukin 1 production in naive macrophages or proliferation in cultures of murine splenocytes, the macrophage-processed LPS is between 10- and 100-fold greater in specific activity. Evidence is presented for both structural and chemical alterations in the LPS macromolecule.