TEXT

Top Abstract Text References

Enteric infection by Shiga toxin-producing enterohemorrhagic Escherichia coli (EHEC) is associated with bloody diarrhea and, in some cases, systemic complications including hemolytic uremic syndrome (HUS). Although the triad of hemolytic anemia, acute renal failure, and thrombocytopenia define HUS, central nervous system (CNS) involvement is often a major complication (8, 32, 34). Signs of severe CNS involvement are associated with high mortality (41) and, because modern therapy compensates for renal failure, with the majority of deaths due to HUS (22).

During EHEC infection, Shiga toxin is thought to move into the blood through the inflamed intestinal mucosa. The thrombotic microangiopathic damage characteristic of HUS is thought to result from direct cytotoxic effects of circulating Shiga toxin on the vascular endothelium (17, 36). Shiga toxin, with one central A subunit surrounded by five B subunits, inhibits protein synthesis in eukaryotic cells through the action of the A subunit on the ribosome. The B subunits of the toxin bind specifically to the galactosyl-1,4-galactose (Gal1, 4Gal) linkage found in a globotriaosylceramide (Gb₃) located on the surface of endothelial cells. Thus, expression of Gb₃, the receptor for Shiga toxin, in endothelial cells is thought to relate to their susceptibility to Shiga toxin.

EHEC infection may also result in absorption of lipopolysaccharide, an endotoxin known to stimulate production of tumor necrosis factor alpha (TNF-) and other inflammatory cytokines. Shiga toxin itself may stimulate local production of cytokines in some tissues (35). TNF- and Shiga toxin synergistically affect human umbilical vein endothelial cells (HUVEC) (21). In cultured HUVEC, TNF- increases the concentration of Gb₃ (26, 37) along with the cells' sensitivity to Shiga toxin (18).

Cultured human renal endothelial cells, which are microvascular in origin, were reported in 1993 to contain appreciably more glycolipid than macrovascular HUVEC and to be far more sensitive to Shiga toxin injury than were cultured HUVEC (26). However, a reexamination of this phenomenon with highly purified and characterized glomerular endothelial cells indicated that glomerular endothelial cells are closer to HUVEC in glycolipid content, sensitivity to toxin, and response to TNF- than was originally reported (38). The earlier preparations may have included other renal cell populations with high Gb₃ content and, therefore, elevated sensitivity to Shiga toxin (10, 11, 33).

The renal microangiopathy characteristic of HUS is thought to result directly from Shiga toxin damage to kidney cells and to be a direct consequence of high Gb₃ levels. This microangiopathy is thought to lead to coagulation within the glomerulus and renal failure. Hemolytic anemia and thrombocytopenia are also thought to result from endothelial microangiopathy. In contrast, there is no consensus on the pathogenic mechanisms of brain involvement in HUS; CNS damage has not been linked directly to cerebral coagulopathy (27). We hypothesize that a common pathogenic event in both kidney and CNS is Shiga toxin damage to endothelial cells, magnified by upregulation of cell surface Gb₃ in response to TNF-.

Toxin binding can be affected by factors other than the total amount of Gb₃, including the type of fatty acyl moieties contained within Gb₃ (i.e., the Gb₃ species) (29) and the characteristics of the microenvironment in which the glycolipid is anchored (2, 3, 40). Toxin binding is also different in different endothelial cell types. For example, Jacewicz et al. have reported that transformed human intestinal microvascular endothelial cells are more sensitive to Shiga toxin than are macrovascular endothelial cells from human saphenous vein (14). In contrast, human brain microvascular endothelial cells are more resistant to Shiga toxin, but their sensitivity increases significantly in response to cytokines (31).

Human brain microvascular endothelial cells, which are essential constituents of the blood-brain barrier, form a network of complex tight junctions while permitting asymmetric transport and vesicular transcytosis (16, 28). Lacking fenestrations, these specialized cells, strategically located at the interface between blood and brain, are actively involved in maintaining homeostasis of the CNS.

In HUS, endothelial damage by Shiga toxin could result in CNS pathology as a consequence of loss of homeostasis, direct nerve cell damage by Shiga toxin crossing the blood-brain barrier, and/or production of toxic mediators by endothelial cells. Pathologic events could occur at toxin levels that cause endothelial cell dysfunction or individual cell death without triggering frank widespread endothelial necrosis and intravascular thrombosis. Although vascular occlusions due to thrombosis might sometimes cause cerebral ischemia, the critical common event in CNS involvement during HUS could be damage that occurs early and at low Shiga toxin levels.

To test this hypothesis on the neuropathology of HUS, the effects of TNF- and Shiga toxin on human brain endothelial cells (HBEC) were studied and compared with results obtained in the more commonly used HUVEC model. Specifically, we investigated whether the increased sensitivity caused by TNF- in microvascular HBEC is mediated by elevated Gb₃ levels, as it is in macrovascular HUVEC. Using both cell types, we also determined whether normal fatty acyl (NFA) or hydroxylated fatty acyl (HFA) species of Gb₃ are more strongly associated with increased sensitivity.

HUVEC were isolated from one to four human umbilical vein segments by collagenase digestion as described previously (7). HBEC were obtained from fresh autopsied brains and processed as described previously (5, 39). Once the cultures were established, their cerebral endothelial nature was confirmed morphologically and by the presence of factor VIII antigen. The cells in these experiments were between passages 2 and 3.

Shiga toxin was purified from a sonic extract of Shigella dysenteriae serotype 1 strain 60R (30). Purity was confirmed with silver staining, and the specific activity was confirmed in a HeLa cell assay system.

Confluent HUVEC were treated with Shiga toxin alone or in combination with either TNF- or interleukin-4 (IL-4). Shiga toxin, TNF-, and IL-4 were diluted in the appropriate growth media before their addition to cell cultures. Cell cultures were examined for morphological changes using an inverted phase-contrast microscope.

For tests of Shiga toxin binding to HUVEC, purified toxin was iodinated with ¹²⁵I-labeled sodium iodide (New England Nuclear, Boston, Mass.) by the solid-phase lactoperoxidase and glucose oxidase technique (Enzymobead, Bio-Rad, Hercules, Calif.). The radioiodinated toxin was separated from the reaction products on a Bio-Gel (Bio-Rad) P-6 column. Binding to HUVEC was quantitated as previously described (15). Briefly, cells were treated for 72 h with recombinant human TNF- (500 IU/ml) or vehicle alone and harvested nonenzymatically. Cells (2.5 × 10⁴) were resuspended in 100 µl of medium 199 containing 1% bovine serum albumin with radioiodinated Shiga toxin (4 × 10⁸ to 10⁹ M) with or without 100-fold excess unlabeled toxin. After incubation for 2 h at 4°C, samples were layered over oil [0.2 ml of dibutylphthalate-bis(2-ethylhexyl)phthalate 1.5:1 (vol/vol); Kodak, Rochester, N.Y.] and centrifuged at 12,000 × g for 15 s. Cell pellets (bound fraction) were recovered by cutting off the tube bottom; pellets and supernatants (free fraction) were counted in a  counter (Beckman 5500B). Each of two experiments was performed in triplicate. Data reduction was performed by linear regression analyses, and significance was tested by the Wilcoxon signed ranks test.

Concentrations of Gb₃ species were determined by high-performance liquid chromatography (HPLC). Cultured cells (2 million to 20 million cells) were harvested, rinsed, resuspended in 0.6 ml of distilled water, sonicated, extracted overnight by the addition of 4 volumes of methanol and then 8 volumes of chloroform, and filtered. The filtrate was analyzed as described previously (24). The perbenzoylated glycolipids were separated by HPLC on a silica column with a hexane-dioxane gradient (1 to 23%) and detected at 229 nm; the peaks were integrated by Dynamax software (Rainin [now Varian Corporation], Walnut Creek, Calif.) and then identified and quantified by comparison with authentic standard glycolipids. This HPLC method resolved each glycolipid type into a peak containing short-chain and any HFA moieties and another peak containing NFA moieties.

For measurement of protein synthesis, HUVEC were incubated with and without cytokine for 48 h, whereupon all media were supplemented with [¹⁴C]leucine (0.5 Ci/ml, 1,000 Ci/mmol; New England Nuclear, Boston, Mass.); at 48 h, cells were also supplemented with Shiga toxin and/or additional cytokine. At 72 h, all cells were incubated for 3 h at 37°C with unlabeled medium, whereupon incorporation of label into protein was measured to determine inhibition of protein synthesis in HUVEC. HBEC were likewise treated with or without TNF- (100 IU/ml) for 48 h and an additional 24 h with logarithmic dilutions of Shiga toxin. The cells were incubated in a methionine-cysteine-free medium with [³⁵S]Met-Cys (specific activity, >1,000 Ci/mmol; ICN Pharmaceutical, Irvine, Calif.) for 3 h at 37°C. At the end of the incubation period, the cells were rinsed and solubilized in detergent, and labeled proteins were precipitated and counted in a scintillation counter (6).

The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was used to assess cell viability (9, 23). Results were expressed as percent reduction of MTT relative to controls.

Effect of TNF- on HUVEC Gb₃. TNF- exerted differential effects on the upregulation of NFA glycolipids and HFA glycolipids. The HFA glycolipid levels were low and were not upregulated by treatment with cytokines. NFA glycolipid levels were initially higher than HFA glycolipid levels; they were significantly increased by treatment with proinflammatory cytokines such as IL-1 and IL-6 (not shown), and especially by TNF-, but not by others, exemplified by IL-4 (Table 1). Treatment with TNF- caused a general elevation of NFA glycolipids and a threefold elevation of NFA-Gb₃. Furthermore, the ratio of NFA-Gb₃ to lactosylceramide increased significantly with TNF- treatment; the loss of lactosylceramide concurrent with increased Gb₃ is consistent with increased activity of the galactose-1,4-galactosyltransferase responsible for the synthesis of NFA-Gb₃ from NFA-lactosylceramide, although other mechanisms are also possible.

TNF--mediated sensitivity to Shiga toxin in HUVEC. This increase in cellular NFA-Gb₃ levels was accompanied by increased binding of ¹²⁵I-labeled Shiga toxin to the TNF--treated HUVEC (Fig. 1), supporting our hypothesis that the NFA-Gb₃ upregulated by TNF- is a functional receptor for Shiga toxin in these cells. HUVEC treated for 72 h with TNF- (500 IU/ml) demonstrated the ability to bind increasing quantities of toxin as the amount of radiolabeled toxin was raised; naive HUVEC showed essentially no binding of radiolabeled Shiga toxin (P = 0.01). The specificity of this binding was determined by the ability of excess unlabeled toxin to displace bound labeled toxin.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   Shiga toxin binding to HUVEC treated with TNF-. HUVEC, pretreated with recombinant human TNF- (500 IU/ml) or vehicle alone, were incubated with 125I-labeled Shiga toxin (1 × 109 to 4 × 108 M) in the absence or presence of 106 M unlabeled toxin. Specific binding is the difference between total binding (in the absence of competing unlabeled toxin) and nonspecific binding (in the presence of excess competing unlabeled toxin). Each value represents the mean (± standard error) of triplicate determinations; these results are representative of two similar experiments. Cells treated with TNF- show significant, specific, dose-dependent binding by Shiga toxin, while the untreated cells did not show specific binding (P = 0.01), indicating that TNF- induces the expression of Shiga toxin receptors in these cells.

View this table: [in this window] [in a new window]   TABLE 2.   Effects of TNF- and Shiga toxin treatment on HUVEC

Effect of TNF- on HBEC Gb₃. Confluent cultures of HBEC were exposed for 72 h to concentrations of human TNF- varying from 5 to 1,000 IU/ml, and levels of Gb₃ were measured. As in HUVEC, the predominant species of Gb₃ were NFA-Gb₃, and only the NFA species changed in response to TNF-. Untreated HBEC contain NFA-Gb₃ at 20 amols/cell; TNF- increased the NFA-Gb₃ content in a dose-dependent manner, up to a level of 120 amols/cell at 500-IU/ml TNF- treatment (Table 3). In a replicate experiment with primary HBEC from a different source, the maximum increase in Gb₃ again occurred at 500 IU of TNF- per ml. Thus, the basal levels of NFA-Gb₃ and their upregulation by TNF- are very similar to those observed in HUVEC.

View this table: [in this window] [in a new window]   TABLE 3.   Gb3 expression in TNF--treated HBEC

TNF--mediated sensitivity to Shiga toxin in HBEC. As with HUVEC, the sensitivity of HBEC cultures to toxin increased in parallel with the increases in NFA-Gb₃ levels. Confluent HBEC, treated with TNF- at 100 IU/ml and exposed to 10⁷ to 10¹² M Shiga toxin for 24 h, displayed a dose-dependent decrease in protein synthesis (Fig. 2A) and a corresponding decrease in cell viability as measured by reduction of MTT (Fig. 2B). At 10⁸ M Shiga toxin, the level of protein synthesis was reduced by 25% and cell viability was 84% of untreated controls.

View larger version (61K): [in this window] [in a new window]   FIG. 2.   TNF- treatment increases the sensitivity of HBEC to multiple concentrations of Shiga toxin. HBEC were treated with TNF- (0 or 100 IU/ml) for 48 h, after which cultures from each group were treated with serial logarithmic dilutions of Shiga toxin from 107 to 1012 M for an additional 24 h. (A) Protein synthesis was determined by [35S]Met-Cys incorporation. (B) Cell viability was determined by MTT assay. Data are means ± standard deviations of triplicate samples at each concentration of Shiga toxin.

8

8

8

Conclusion. EHEC contains distinct phage genes that encode two major classes of Shiga toxin: Shiga toxin 1 and Shiga toxin 2. The Shiga toxins all contain an enzymatically active A subunit that inhibits protein synthesis at the ribosome, and five B subunits that each bind to glycosphingolipids whose carbohydrate structures contain a Gal1, 4Gal moiety. Gb₃ contains Gal1, 4Gal as its nonreducing terminus, and it is thought to be the major functional receptor for both Shiga toxin 1 and the forms of Shiga toxin 2 associated with human disease. The specific mechanism whereby toxin binding and inhibition of protein synthesis results in the pathology associated with HUS is not clear and may vary among endothelia from different tissues. Proposed mechanisms include direct or indirect injury to vascular endothelial cells (4) and, possibly, apoptosis (13).