Evaluation of the Role of Shiga and Shiga-Like Toxins in Mediating Direct Damage to Human Vascular Endothelial Cells

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Uniformed Services University of the Health Sciences U.S. Department of Defense

1991

Evaluation of the Role of Shiga and Shiga-like Toxins in Mediating Direct Damage to Human Vascular Endothelial Cells

Vernon L. Tesh Uniformed Services University of the Health Sciences

James E. Samuel Biocarb, Inc.

Liyanage P. Perera Uniformed Services University of the Health Sciences

John B. Sharefkin Uniformed Services University of the Health Sciences

Alison D. O'Brien Uniformed Services University of the Health Sciences, [email protected]

Follow this and additional works at: https://digitalcommons.unl.edu/usuhs

Part of the Medicine and Health Sciences Commons

Tesh, Vernon L.; Samuel, James E.; Perera, Liyanage P.; Sharefkin, John B.; and O'Brien, Alison D., "Evaluation of the Role of Shiga and Shiga-like Toxins in Mediating Direct Damage to Human Vascular Endothelial Cells" (1991). Uniformed Services University of the Health Sciences. 113. https://digitalcommons.unl.edu/usuhs/113

This Article is brought to you for free and open access by the U.S. Department of Defense at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Uniformed Services University of the Health Sciences by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. 344

Evaluationof the Role of Shiga and Shiga-like Toxins in Mediating Direct Damage to Human Vascular Endothelial Cells

Vernon L. Tesh, James E. Samuel,* Liyanage P. Perera, Departmentsof Microbiologyand Surgery, UniformedServices University John B. Sharefkin, and Alison D. O'Brien of the Health Sciences, Bethesda, Maryland

Infectionwith Shiga toxin- andShiga-like toxin-producing strains of Shigelladysenteriae and Escherichiacolif respectively,can progressto the hemolytic-uremicsyndrome. It has been hy- pothesizedthat circulatingShiga toxin, Shiga-liketoxins, and endotoxins may contribute to the diseaseby directlydamaging glomerular endothelial cells. The effectsof these toxins on HeLa, Vero,and human vascular endothelial cells (EC)were examined. Confluent EC weresensitive to Shigatoxin butwere at least 106-fold less sensitiveto the toxinsthan were Vero cells. Shigatoxin was the predominantcytotoxic factor. Lipopolysaccharides were not cytotoxicand did not aug- mentShiga toxin-mediated toxicity. Lower doses of Shigatoxin causedcytotoxicity when coin- cubatedwith tumornecrosis factor. The relativeresistance of EC to Shigatoxin and Shiga-like toxins may be due to reducedtoxin binding,as low levels of globotriaosylceramide(Gb3), the toxin-specificreceptor, were found in EC membranes.

Inflammatorybacterial colitis caused by infection with from the 28S rRNA component of the 60S eukaryoticribo- Shigella species continues to be a major cause of morbidity some complex [8, 9]. The B subunit mediatesbinding of the and mortality in underdeveloped countries, especially holotoxin to specificglycolipid receptorsin mammaliancell among young children [1]. Of particularinterest is Shigella membranes [10-13]. Although the precise role of Shiga dysenteriaetype 1, which is capable of causingepidemic out- toxin and the SLTsin the pathogenesisof bacillarydysentery breaksof dysentery.In addition, it has recently been shown and hemorrhagiccolitis, respectively, is not fully under- that some strainsof Escherichiacoli, particularlyE. coli sero- stood, there is now convincing evidence to suggest that the type 0157:H7, can cause hemorrhagiccolitis [2, 3]. Collec- toxins are importantvirulence factors in the developmentof tively, these E. coli strains are referredto as enterohemor- bloody, edematous vascular lesions of the colon [14]. The rhagicE. coli (EHEC). S. dysenteriaetype 1 and EHECshare toxins may also participatein the direct killing of colonic the characteristicof producinghigh levels of protein toxins epithelial cells and may provokefluid secretionand diarrhea called Shigatoxin or Shiga-liketoxins (SLTs or Verotoxins), in the host [7]. respectively.EHEC may produce two antigenicallydistinct Perhaps the most serious sequela of colitis caused by S. toxins, designated SLT-I and SLT-II. SLT-I is essentially dysenteriaetype 1 or EHEC is the progressionof the disease identicalto Shigatoxin, and its cytotoxic activityfor Vero or to the hemolytic-uremicsyndrome (HUS), which is charac- HeLa cells can be neutralizedwith anti-Shigaantibodies [4, terized by acute renal failure, thromboticmicroangiopathy, 5]. SLT-II is 567chomologous to Shiga toxin/SLT-I at the and thrombocytopenia[15]. Histopathologicstudies of the deduced amino acid level, and its cytotoxicityis not blocked kidneys of HUS patientshave shown profoundalterations in by anti-Shiga toxin antibodies [5, 6]. the glomeruli.More specifically,glomerular endothelial cells Shiga toxin and SLTs are holotoxins that consist of a sin- (EC) appearswollen, and there is abundantfibrin deposition gle 32- to 33-kDa A subunit in associationwith a pentamer and inflammatorycell influx in the lumina of the glomeruli of 7.7-kDa B subunits [7]. The A subunit is a specific N-gly- [16]. These observations,as well as the criticalrole of EC in cosidase that selectively cleaves a single adenine residue maintainingnormal blood flow [ 17], have led to the concept that systemic Shiga toxin or SLTs may specificallytarget the glomerularEC for damageand therebycontribute to the de- of HUS. In of this et al. Received 16 revised 13 1991. velopment support hypothesis,Obrig August 1990; February showed that toxin are for The opinions or assertionsherein are the privateones of the authorsand [18] Shiga preparations cytotoxic are not to be construedas official or reflectingthe views of the Department human umbilical vein EC (HUVEC) in vitro. of Defense or of the Uniformed ServicesUniversity of the Health Sciences. In the here, we assessed the Grant National Institutesof Health experiments reported cyto- support: (AI-20148). toxic of crude toxin, SLT-II,and Reprintsor correspondence:Dr. Alison O'Brien,Department of Microbi- potential Shiga affinity-pu- ology, Uniformed Services University of the Health Sciences, 4301 Jones rified Shiga toxin for confluent human saphenous vein EC BridgeRd., Bethesda,MD 20814. (HSVEC) and HUVEC monolayers. In addition, we com- * Presentaddress: Biocarb, Inc., Gaithersburg,Maryland. pared Shiga toxin-mediated EC and Vero cell cytotoxicity The Journalof InfectiousDiseases 1991;164:344-52 and examined the role of and re- Thisarticle is in the publicdomain. lipopolysaccharides(LPS) 0022-1899/91/6403-0017 combinanthuman tumornecrosis factor-a in directEC cyto- JID 1991;164 (August) ShigaToxin and SLT-II-MediatedEC Toxicity 345 toxicity. Finally, we analyzed toxin binding to intact EC and IL). The stock preparationswere then diluted in growth medium quantitated levels of toxin receptor in EC membranes. as needed. Tissue culture. Vero and HeLa cells were maintained in Ea- gle's MEM (Flow Laboratories, McLean, VA) supplemented with \07c fetal bovine serum, 10 mM L-glutamine, 50 units/ml Materials and Methods penicillin, and 50 Mg/mlstreptomycin. EC were harvested from adult human saphenous veins (HSV) obtained during coronary Bacterial strains and S. 1 strain plasmids. dysenteriae type bypass operations as previously described [26]. Vein segments 3818T levels of toxin and was isolated from produces high Shiga (~2.5 cm) were transported to the laboratory in chilled blood a with severe in Central America E. coli patient dysentery [19]. with 100 units/ml preservative-freeheparin (Sigma). The HSV 0157:H7 strain 933 both SLT-I and II and was iso- - produces were cannulated and flushed with cold Ca"^ and Mg^ -free lated from an outbreak of colitis in the United hemorrhagic Dulbecco's PBS (PBS-CMF; GIBCO, Grand Island, NY). HSV States [2, Toxins used in this were derived from E. coli 5]. study were ligated after filling to slight distension with 0.17c CLS II DH5a Research Laboratories, (Bethesda Gaithersburg, MD) Freehold, transformed with recombinant the cloned collagenase (Worthington Diagnostic Systems, NJ) plasmids containing and 0.5^ bovine serum albumin (BSA) in PBS-CMF and im- toxin genes. E. coli harbors the plasmid that (pLPSH3) pBR328 mersed in Hanks' balanced salt solution (GIBCO) for 15 min at contains the entire Shiga toxin operon on a 5.9-kb Bglll-Sall 370C. HSV were flushed with 10 ml of cold complete medium insert subcloned from pNAS 10 [4,20]. E. coli (pLP32) harborsa (CM) composed of Medium 199 (GIBCO) with 207cfetal bovine Bluescribe vector (pBS[?] phagemid kit; Stratagene, La Jolla, serum (Hyclone Laboratories, Logan, UT), 100 pg/ml heparin CA) with a 3.0-kb Sphl-Kpnl fragment from pNN76 encoding (Sigma), 100 pg/ml L-glutamine, 50 units/ml penicillin, 50 pg/ SLT-II [21, 22]. Clones producing high levels of toxin were ml streptomycin (GIBCO), and 2 pg/ml amphotericin B (Fungi- maintained under BL3 -f- EK1 containment [23]. zone; GIBCO). Toxin preparations. LPS were prepared from S. dysenteriae The vein contents were centrifuged (200 g, 1 min, 40C), and 3818T and E. coli 933 by the hot aqueous-phenol method of the cell pellet was resuspended in 1.0 ml of CM Westphal and Jann [24]. The pooled aqueous phases were dia- supplemented with 20 /d/ml EC-mitogen extract prepared from bovine retina lyzed against nine changes of double-distilled, deionized HjO, using a modification of a described method concentrated 10-fold (Amicon, Danvers, MA), and lyophilized. previously [27, 28]. The cells were placed in wells of 24-well (Falcon; LPS preparationscontained ^ Mg/mlprotein (BCA protein as- single plates Becton Dickinson Labware,Oxnard, coated with 1 say; Pierce Chemical, Rockford, IL). E. coli Ol 11:B4 LPS was CA) pg/cm2 human fibronectin and incubated at 370C in humidified 57c purchased from Sigma Chemical (St. Louis). Concentrated C02. The other wells were filled with sterile PBS-CMF to mini- crude Shiga toxin and SLT-II were prepared from E. coli mize in Cells were fed other with (pLPSH3) and E. coli (pLP32), respectively. The strains were changes osmolality. every day fresh CM 10 pl/ml bovine retinal extract grown at 370C in 50 ml of Luria broth (LB) containing 250 containing (CM-BRE). When cultures became confluent, HSVEC were de- Mg/ml ampicillin for 24 or 48 h with constant agitation. The primary tached 0.05% EDTA, washed in CM with bacteria were pelleted by centrifugation, washed, and resus- using trypsin-0.02% 107cserum, and in CM-BRE. pended in 5.0 ml of LB. The bacteria were disrupted by sonica- resuspended Subsequent passages of HSVEC were on 25-cm2 tissue culture tion and the lysates cleared by centrifugation at 12,000 g for 15 plated gelatin-coated flasks or 96-well microtiter All were done min. Toxins in the clarified lysates were concentrated by precipi- plates. experiments with HSVEC two to five times in vitro. tation with 60(3dsaturated ammonium sulfate. The precipitates passaged HSVEC were characterized were resuspended in 5.0 ml of PBS and dialyzed overnight by inverted-phase microscopic against PBS. Toxins prepared in this manner reproducibly con- observation of cell morphology and immunoenzymatic staining ~ tained 108-109 CDso/ml when assessed by the Vero cell cyto- for factor VHI-related antigen (Endo-RAP I.D.; Endotech, In- toxicity assay. Purified Shiga toxin was prepared from this crude dianapolis). PrimaryHUVEC were the gift of D. Silverman (De- starting material as described [22, 25]. Briefly, crude toxin was partment of Microbiology, University of Maryland School of passed over a DEAE-Sepharose column equilibrated with 0.05 Medicine, Baltimore). HUVEC were passaged and maintained M TRIS buffer and eluted with a gradient of 0-1.0 M NaCl in as described above. 0.05 M TRIS buffer. Fractions cytotoxic for Vero cells were Cytotoxicity assays. The cytotoxicity of toxin preparations pooled and passed over a chromatofocusing column. Cytotoxic for Vero cells was determined as described [29]. Vero cells were ~ fractions were pooled, dialyzed, and passed over an antitoxin plated at 104 cells/well in 96-well plates and incubated over- affinity column. Shiga toxin was eluted with 0.05 M glycine, night at 370C in 57cCO2. Toxin preparations were serially di- dialyzed against PBS, and stored at 40C until used. Purified luted in complete Eagle's MEM medium, and 100 pi of the Shiga toxin preparations contained k0. 1 ng/ml endotoxin as dilutions was transferredto the cells. Incubation was continued assessed by limulus amoebocyte lysate gelation (Endotect; for 48 h. The cells were fixed with formalin and stained with Schwarz/Mann Biotech, Cleveland). 0.137c crystal violet. Absorbance was measured at 620 nm. A Recombinant human tumor necrosis factor-a (rhTNFa). 50%cytotoxic dose (CD5o) represents the reciprocal dilution of rhTNFa (lot NBP-802A; 2.7 X 107 units/mg) was the gift of A. toxin necessary to kill 50^ of the Vero cells in a well. The cyto- Creasey (Cetus, Emeryville, CA). Before use in cytotoxicity as- toxic potency of toxin preparations is defined as the number of says, the lyophilized cytokine was resuspended to 260 pg/ml in CD50 per milliliter of toxin preparation. Cytotoxicity of toxin sterile pyrogen-free water (Travenol Laboratories, Deerfield, preparations for EC was assessed by hemocytometer counts or 346 Tesh et al. JID 1991;164 (August) by spectrophotometry. EC were plated on the interior 24 wells The cells were resuspended in 500 pi of a 1:100 dilution of of gelatin-coated 96-well plates at ~2 X 104cells per well. Exte- primary antibody and agitated for 45 min. Primary antibodies rior wells were filled with PBS-CMF. When the cells reached used were 4F7, monoclonal IgG 1 directed against the A subunit confluence, fresh CM-BRE containing the toxin(s) to be assayed of Shiga toxin [33]; 13C4, monoclonal IgG 1 directed against the were added to triplicate wells in a total volume of 200 pi. Incu- B subunit of Shiga toxin [34]; and BC5BB12, monoclonal IgGl bation was continued at 370C in 57c C02. Cells that subse- directed against the B subunit of SLT-II [35]. Unbound anti- quently detached from the gelatin substratum failed to exclude body was removed by three washes with diluent buffer, and sec- trypan blue, or after extensive washing, did not grow when re- ondary antibody (500 pi of a 1:100 dilution of a fluorescein cultured on gelatin-coated plates. isothiocyanate-conjugated goat F[ab']2anti-mouse IgG; Cappel was added to the cells. Viable cells were stained for counting in a hemocytometer by Laboratories, Westchester, PA) After unbound was washed the cells were replacing the growth medium with 60 pi of 0.17ccrystal violet in secondary antibody out, treated with 100 of a fluorescent 0.1 M citric acid (pH 7.26). After the cells were incubated with pi nonspecific membrane the dye at 370C for 20 min, the wells were scraped with sterile probe, l,r-dioctadecyl-3,3,3' ,3'-tetramethyl indocarbocyanine in 17cethanol and serum-free Molec- pipette tips, and 15 /d of cell suspension was transferredto the perchlorate(5 pg/ml EBSS; ular at 370C for 10 min. hemocytometer. Stained nuclei were then counted. Cells were Probes, Eugene, OR) After extensive with cold serum-free the cells counted using a microtiter plate reader according to the method washing EBSS, were transferredto flat-bottomed 96-well and of Brasaemle and Attie [30]. Cells were washed with warm PBS- plates centrifuged for 10 min at 40C. The were scanned CMF, and 100 pi of methanol per well was added for 15 min at (200 g) plates immediately the Meridian ACAS 470 with a 5-W laser room temperature. The methanol was aspirated from the wells, using equipped argon tuned to 488 nm. Laser was to 200 mW and scan the plates were air dried, and 100 pi of 0.17c crystal violet was power adjusted to 357c. The collimated first-orderbeam was added to each well for 5 min. After the dye was removed and the strength adjusted directed a standard dichroic filter cube. for a plates dried, each well was washed with HjO, and the stained through (See [32] more detailed of laser cells were solubilized with 100 pi of 27c(wt/vol) sodium deoxy- description parameters.)Qualitative analysis of toxin was done conver- cholate. The plates were heated in a microwave oven (Samsung Shiga binding by microcomputer sion of fluorescent emissions into a scan based MW5510; setting 2) for 1 min, and agitated (mini-orbital 15-pseudocolor on fluorescence shaker, setting 6; Bellco Glass, Vineland, NJ) for 10 min. Absor- intensity. To- bance was read at 570 nm. Cell counts were correlated by linear Thin-layer chromatographyof toxin-bindingglycolipids. tal extracts from HSVEC, HUVEC, Vero, and HeLa cells regression analysis of hemocytometer cgunts of stained nuclei lipid were to on aluminum-backed silica against absorbance measurements of triplicate serial twofold di- subjected chromatography 60; Merck AG, lutions of EC. This assay produced a linear standard curve be- gel thin-layer chromatography plates (silica gel Darmstadt, with CHCl3-CH3OH-0.257o KC1 tween 1.5 X 103 and 6 X 104 cells per well (mean R value, .92). FRG) aqueous Data were as of which is defined (5:4:1) as previously described [22, 36]. Purified glycolipids expressed percentage viability, served as as of cells H-toxin ? of control controls. After chromatography, the chromatography (number background/number were coated with cells ? X 100. absorbance was de- plates 0.17c polyisobutylmethacrylate (Poly- background) Background air and with rived from blank wells. Control cells were incubated science, Warrington, PA), dried, sprayed TBS-BSA staining M 0.15 M with CM-BRE (0.1 TRIS-HC1 [pH 7.4], NaCl, 17cBSA). The plates only. were overlaid with ~105 toxin in TBS-BSA. Neutralization Dilutions of toxins were made in CD507ml Shiga assay. ap- After incubation at 40C, the were washed media in 96-well and 100 plates extensively propriategrowth plates (100 pl/well), with PBS and overlaid with MAb 13C4 diluted in TBS-BSA. of toxin monoclonal was pi anti-Shiga antibody (MAb 13C4) Goat anti-mouse labeled with 125Iwas used as added to each well. The were incubated at 370C IgG secondary plates agitated, The were washed, air dried, and bind- in 57c for 1 and at 40C. Toxin-antitoxin antibody. plates specific CO2 h, kept overnight was visualized Relative amounts of mixtures were then added to Vero cells or and ing by autoradiography. (100 pi) EC, were estimated from areas of den- was as described above. An toxin-binding glycolipids peak cytotoxicity determined isotype- sitometric scans obtained with a densitometer Shi- matched MAb directed the B subunit of cholera toxin (CS9000U; against madzu, Kyoto, Japan) and compared with serial dilutions of the [31] did not affect cytotoxicity in this assay. control glycolipid globotriaosylceramide (Gb3). Autoradio- toxin to whole cells. A modification of an Analysis of binding graphs of several exposure times for each thin layer chromatogra- immunofluorescence a scan- protocol using computer-assisted phy plate were scanned to ensure the peak areas were within the laser instrument Meridian Oke- ning (ACAS 470; Instruments, linear range of the densitometer. mos, was devised to toxin to EC MI) compare Shiga binding Statistical analysis. The significance of differences between versus Vero and HeLa cells Cells were and [20, 32]. trypsinized means of samples was determined by Student's two-tailed t test. to ~5 X 105 in cold diluent buffer adjusted cells/ml (phenol P < .025 was considered significant. red-free Earle's balanced salt solution [EBSS; GIBCO] containing 10^. fetal bovine serum and 0.1^ Na^). Concentrated ~ crude Shiga toxin (50 pi; 109Vero cell CD507ml)was added to Results 450 pi of cells in Eppendorf tubes and agitated at 40C for 60 min. Unless otherwise noted, all binding and washing steps were Role of Shiga toxin, SLT-II, and LPS in direct EC cytotoxic- done at 40C. Unbound toxin was removed by three washes with ity. To assess the cytotoxic potential of Shiga toxin for cold diluent buffer. HSVEC, various dilutions of crude Shiga toxin preparations JID 1991;164 (August) ShigaToxin and SLT-II-MediatedEC Toxicity 347

ing of the cells (figure 1A). Cytotoxicity occurred within the first 24 h of incubation, and EC viability remained reduced relative to control cells over 72 h. When HUVEC were used as target cells, the kinetics and dose response of Shiga toxin- mediated toxicity were essentially identical to those of HSVEC (data not shown). Treatment of both EC types with crude SLT-II preparations killed the cells in a manner similar to crude Shiga toxin (data not shown). Finally, treatment of both EC types with affinity-purified Shiga toxin resulted in killing that was similar to that observed with crude toxin preparations (figure 1B). Dilutions of the toxins were made based on cytotoxicity for Vero cells. From these data, we estimated that 1 EC CD50 was roughly equivalent to 107 ~ Vero cell CD50, or 1.4 X 10~8 M Shiga toxin. These results suggest, therefore, that while confluent human EC derived from large veins are sensitive to Shiga toxin, they are at least 106-fold less susceptible than Vero cells. The observation that purified Shiga toxin killed EC in a time (days) manner similar to crude toxin preparations suggested that Shiga toxin was the primary mediator of cytotoxicity for EC. However, the effects of small amounts of endotoxin present in the toxin preparations could not be discounted, especially in light of recent evidence that nanogram quantities of endotoxins can have profound effects on EC function [37-39]. To more fully assess the role of Shiga toxin in EC cytotoxicity, we incubated the cells with varying amounts of affinity- purified Shiga toxin in the presence or absence of anti-Shiga toxin MAb (table 1). Even at the highest toxin concentration tested, cytotoxicity was completely neutralized by antitoxin antibody. Antibody alone had no effect on cell viability, and the addition of an irrelevant, isotype-matched MAb did not block Shiga toxin cell killing (data not shown). The observation that MAb directed against Shiga toxin neutralizes cytotoxicity provides further evidence in support of the concept that, in vitro, Shiga toxin is the predominant mediator of direct EC cytotoxicity. In light of clinical evidence suggesting that patients with HUS are frequently endotoxemic [40] as well as histopatho-

time (days) Table 1. Neutralizationof endothelialcell (EC) cytotoxicitywith Figure 1. Effectof crude and purifiedShiga toxin on endothelial anti-Shiga toxin monoclonal antibody (MAb). cell viability. Confluent human saphenous vein endothelial cell (HSVEC) monolayerswere incubated with dilutions of (A) crude Durationof EC exposure to toxin or (B) affinity-purifiedShiga toxin. At the indicated time points, the numbers of viable, adherent cells were determined and com- 24 h 72 h pared to HSVECcultured in the absence of toxin. Toxin dilutions Toxin dose were made on the basis of Vero cell 50^ cytotoxic doses (CD50): in CD50Zml* No. MAb With MAb No. MAb With MAb S, 109;A, 107;I, 105;T, 103(all CD5o).Error bars indicate SD. 109 42.2 ?17.3 104.6 ?3.2 20.0 ? 5.0 96.7 ? 8.6 107 52.7 ?6.0 100.4 ?6.5 50.1 ?4.9 110.0 ?2.3 were incubated with confluent EC monolayers, and the num- 105 91.6 ?4.9 109.1 ?3.2 100.1 ?8.8 106.6 ?3.0 ber of viable cells attached to wells remaining gelatin-coated NOTE. ConfluentEC wereincubated for 24 or 72 h withthe indicateddoses was determined by hemocytometer counting or spectropho- of Shigatoxin, in thepresence or absenceof anti-Shigatoxin MAb. The number of The substratum-adherentcells wasdetermined as describedin Materialsand Methods. tometry. incubation of HSVEC with the highest dilu- Dataare of viableEC ? SD. * percentages tions of crude Shiga toxin resulted in a dose-dependent kill- Numberof Verocell CD50fm\. 348 Tesh et al. JID 1991; 164 (August)

tial step in target cell intoxication and death [5, 43, 44]. Therefore, reduced binding of Shiga toxin and SLT-II to EC is one possible mechanism to explain the relative insensitivity of EC to the toxins. To assess toxin binding to intact EC and to compare binding to control cells, we devised an indirect immunofluorescence assay using a computer-assisted laser scanner that con- verted fluorescence intensity into a pseudocolor spectrum. To monitor toxin binding, EC, Vero, and HeLa cells were treated with equivalent doses of Shiga toxin or SLT-II in suspension, and toxin binding was detected with antitoxin antibodies and fluorescein isothiocyanate-conjugated secondary antibody. The cells were then treated with a nonspecific, fluorescent membrane intercalator to outline the cells Figure 2. Effect of Shiga toxin (ShT) versus lipopolysaccharides and allow comparison of levels of toxin-specific fluorescence on endothelial cell that was determinedafter (LPS) (EC) viability within the same field. As shown in figure 3A, in any given confluent cells were incubated 48 h with ~ 1 EC 50%cytotoxic field, virtually every Vero (left panel) or HeLa cell (right dose (1 EC CD50^ 107 Vero cell CD5o)of purifiedShT only; 10 was stained after treatment with toxin. Mg/mlLPS derived from Escherichia coli 0111 :B4, E. coli 0157:H7 panel) intensely Shiga strain933, or Shigelladysenteriae 3818T; or mixturesof 1 EC CD50 Although every cell showed toxin-specific fluorescence, the = ShT and LPS. Errorbars SD. Differencesbetween samples were intensity varied between cells (i.e., not every cell appeared to Student's two-tailed t test; P < .025 was analyzed by significant. bind toxin equivalently). The fluorescence intensity and binding pattern of SLT-II for Vero cells was similar to that seen with Shiga toxin (data not shown). Since essentially logic data suggesting that endotoxins may contribute to the every cell bound Shiga toxin, the binding pattern produced pathologic changes in the kidney characteristic of HUS [41], by the nonspecific membrane probe was nearly identical to we queried whether LPS might play a direct role on in vitro toxin-specific fluorescence (not shown). In contrast to Vero EC killing. We demonstrated that purified LPS derived from or HeLa cells, when EC were treated with an equivalent dose S. dysenteriae type 1 and EHEC strains were not directly of Shiga toxin or SLT-II, much lower levels of fluorescence cytotoxic for confluent human EC monolayers (figure 2) or were detected (figure 3B, C, left panels), indicating reduced Vero cells (data not shown) at doses as high as 10 pgjml. toxin binding in comparison to Vero and HeLa cells. As with Furthermore, the coincubation of EC with 10-fold dilutions the control cells, however, there appeared to be differences of LPS and a constant dose of affinity-purified Shiga toxin in the amounts of toxin bound to each individual EC. Some (~1 EC CD50) did not have a statistically significant effect cells showed very low levels of fluorescence; others showed a on EC cytotoxicity in comparison with treatment with Shiga more intense, punctate binding pattern. These data suggest toxin alone (P ^ .025). Thus, while the presence of endotox- that if toxin binding and cell killing are directly correlated, ins in vivo may be an important factor in the pathogenesis of then EC may be heterogeneous in terms of sensitivity to HUS, purified LPS alone does not appear to be a major deter- Shiga toxin. Use of the fluorescent membrane probe (figure minant of direct EC cytotoxicity in vitro. 3B, C, right panels) clearly showed the outline of the EC LPS elicits the synthesis and secretion of TNFa by mono- cytes and macrophages. TNFa may have pleiotropic effects Table 2. Augmentationof endothelialcell cytotoxicityby recom- in vivo, and this cytokine has emerged as a central mediator binant human tumor necrosis factor (TNF). in initiating and regulating the complex cascade of events that results in a procoagulant and proinflammatory state in %viability the vasculature [42]. However, the contributory effect, if Toxin dose No TNF TNF any, of TNFa in Shiga toxin-mediated direct EC cytotoxic- (CDso/ml)* is not clear. We therefore coincubated EC with dilutions ity 107 56.7 37.4 of Shiga toxin and rhTNFa and assessed EC cytotoxicity. In 105 85.6 60.4 ~ the presence of 10 ng/ml rhTNFa, we detected an 100- 103 97.7 81.4 fold reduction in the Shiga toxin EC CD50 (table 2). 10' 105.0 90.3 0 92.1 Comparative analysis of Shiga toxin and SLT-II binding to intact cells. toxin and SLTs bind to Shiga specific glycolipid NOTE. Confluentendothelial cells wereincubated in triplicatefor 48 h with receptors in the membranes of Vero or HeLa cells [10-13]. the indicateddoses of Shigatoxin in presenceor absenceof rhTNFa.The number of substratum-adherentcells wasdetermined as describedin Materialsand Meth- Antibody neutralization, competitive binding, and holo- ods. Data meanvalues of three * represent separateexperiments. toxin dissolution studies have shown binding to be an essen- Numberof Verocell CDso/ml. JID 1991;164(August) ShigaToxin and SLT-II-MediatedEC Toxicity 349

[A [rBs

ColorValues - 4095 I- 3839 - 3583 - 3327 - 3071 - 2015 - 2559 - 2303 - 2047 - 1791 FD - 1535 'C - 1279 - 1023 - 767 - 511 - 255 - 0

10 pm

Figure 3. Comparativeanalysis of Shiga toxin and Shiga-liketoxin type II (SLT-II)binding to intact cells. Equivalentnumbers of Vero cells, HeLa cells, and endothelialcells (EC) were treatedwith 109Vero cell 50%cytotoxic doses of crude toxin preparationsat 40C in the presenceof NaN3. Bound toxins were detectedwith anti-toxinmonoclonal antibodiesand fluorescein-conjugatedsecondary antibody. The cells were outlined by simultaneousstaining with a nonspecificfluorescent membrane probe. Fluorescencewas measuredusing a Meridian ACAS 470 laser scannerin which fluorescenceintensity was convertedto the 15 color scale at right.Moving up the scale, colors represent increasingfluorescence intensity. A, Vero and HeLa cells (left and rightpanels, respectively)treated with Shigatoxin. B, EC treatedwith Shigatoxin (left) and membraneprobe (right). C, EC treatedwith SLT-II(left) and membraneprobe (right). D, EC treatedwith Escherichia coli DH5a concentratedsonicated lysate, (left) and membraneprobe (right). present in the fields. As a control, concentrated sonicated tated the relative amounts of toxin-binding glycolipids by lysates were prepared from E. coli DH5a in a manner analo- direct densitometric scanning of the autoradiographs. HU- gous to that used to concentrate Shiga toxin (see Materials VEC and HSVEC contained 0.06 and 0.03 nM Gb3/mg of and Methods). The control lysates showed no binding to EC cells, respectively. In contrast, Vero and HeLa cells con- (figure 3D, left). In addition, the deletion of primary or sec- tained 80 and 25 nM Gb3/mg of cells, respectively. Thus, the ondary antibody from the assay showed that nonspecific diminished toxin-binding glycolipid content of EC com- background fluorescence was very low in this procedure pared with Vero and HeLa cells directly correlated with re- (data not shown). Therefore, the fluorescence detectable in duced Shiga toxin sensitivity and toxin binding. Shiga toxin-treated cells is specifically due to toxin binding. These data that one reason EC are less sensitive than suggest Discussion Vero or HeLa cells to the cytotoxic effect of Shiga toxin may be low levels of toxin binding to the cells. The experimentsreported here were designedto examine Comparative quantitative analysis of toxin-specific glyco- the role of crude or purifiedShiga toxin, SLT-II,LPS, and lipid receptors in EC. Toxin-binding glycolipids present in recombinantcytokines in killing confluent human vascular Vero and HeLa cells and EC were detected by the direct endothelial cells and to compare the EC cytotoxicity with binding of saturating amounts of Shiga toxin to thin-layer that of other cell lines. The incubation of EC with purified chromatograms of serial dilutions of membrane total glyco- Shiga toxin resultedin a reductionin the numbersof viable lipid extracts (figure 4, lanes 6-15). Glycolipid standards cells remainingattached to gelatin-coateddishes, and this were used to compare the glycolipids present in each cell line activitywas neutralizedby anti-Shiga toxin MAb, suggesting (figure 4, lanes 1-5, 16). In contrast to Vero and HeLa cells, that Shiga toxin was the primarymediator of cytotoxicityin the amount of toxin-binding glycolipids detected in EC vitro. These resultssupport and extend the findingsof Obrig membrane extracts prepared from equivalent cell wet et al. [18], who demonstratedthat rabbit polyclonal anti- weights was greatly reduced. Bound toxin was detected with Shiga toxin antibodies,as well as heat denaturation,totally Shiga toxin-specific MAb and 125I-labeledgoat anti-mouse neutralizethe EC cytotoxic potential of Shigatoxin prepara- IgG antibody. Since the amount of toxin-specific glycolipids tions. These investigatorsalso showed, however, that Shiga present in membrane extracts was previously shown to corre- toxin, to a dose of 10-7 M, had a delayed (>48 h) cytotoxic late directly with the amount of bound toxin [45], we quanti- effecton confluentHUVEC [18]. In contrastto theirstudies, 350 Tesh et al. JID 1991;164(August)

Figure 4. Quantitationof toxin- specific glycolipid receptor in human saphenous vein endothelial cells (HSVEC), human umbilical vein endothelial cells 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 (HUVEC), Vero cells, and HeLa cells. Shiga toxin was bound to glycolipidssep- arated by thin-layerchromatogra- - CDH phy, detected with anti-Shiga toxin monoclonal antibody and '25I-labeledgoat anti-mouse - Gb3 IgG antibody,and visualizedby autora- -Gb4 diography. By lane: 1-5, 100, 50, 10, 5, and 1 ng globotriaosylceramide (Gb3), respectively;6, 7, glycolipids from I and 3 mg of Vero cells, respectively;8, 9, glycolipids from 1 and 3 mg of HeLa cells, respectively; 10-12, glycolipidsfrom 1, 3, and 10 mg of HSVEC,respectively; 13-15, glycolipids from 1, - ORIGIN 3, and 10 mg of HUVEC, respectively; 16, 1 gg each of standardgly- colipids lactosylceramide(CDH), Gb3, and globotetraosylceramide (Gb4). All cell masses are ex- pressedaccording to wet weight. we detected a more rapid loss of EC viability, occurring Several lines of evidence suggest that endotoxins may con- within the first 24 h of incubation with toxin. These differ- tribute to the development of HUS. Bertani et al. [41] dem- ences in kinetics may be due to differences in the metabolic onstrated that when rabbits were infused with E. coli activity of the EC used in the cytotoxicity assays. Obrig et al. 0111:B4 LPS, marked damage to glomerular EC became [18] cultured different concentrations of HUVEC at 370C apparent within 5 min. Changes in EC morphology preceded for 12-24 h, and these cells were termed confluent or non- polymorphonuclear cell infiltration and fibrin deposition. confluent. We seeded EC, monitored growth using an in- These changes were consistent with those seen in the kidneys verted-phase microscope, and allowed the cells to replicate of HUS patients [16]. Barrett et al. [48] showed that a single to a confluent state over 3-5 days. Since Shiga toxin acts by sublethal dose of endotoxin given to rabbits 3 days after the inhibiting protein synthesis, differences in target cell meta- initiation of continuous SLT-II infusion enhances the lethal bolic activity may have drastic effects on Shiga toxin sensi- effect of the toxin. In the in vitro cytotoxicity assays reported tivity. here, however, purified LPS derived from S. dysenteriae type The amount of toxin added to EC was based on the cyto- 1 or EHEC strains was not directly cytotoxic for EC at doses toxicity of the toxin preparations for Vero cells, a cell line as high as 10 Ag/ml. This finding is in accordance with earlier that has been used to characterize toxin activity and toxin-re- studies showing that HUVEC are refractory to both LPS-me- ceptor interaction [7]. While EC are susceptible to Shiga diated direct cytotoxicity and changes in EC function at toxin, they are at least 106-fold less sensitive than Vero cells. doses as high as 100 ,g/ml [49, 50]. In addition, we have On the basis of the cytotoxic potential of Shiga toxin for Vero shown that the addition of anti-Shiga toxin MAb to crude cells (1 CD50 - 1 pg of toxin), we estimate that 1 EC CD50 is toxin preparations neutralizes cytotoxicity, and the coincu- - 1.4 X 10-8 M or ~ 10 gg of toxin. As previously reported bation of EC with LPS and purified Shiga toxin neither en- for other cytotoxicity assays using Shiga toxin and SLTs [46, hanced nor inhibited EC cytotoxicity. However, the correla- 47], we noted a shallow dose response curve, that is, incubation of our in vitro cytotoxicity data with animal models of tion of EC with concentrations of Shiga toxin that differ 100- HUS should be made with caution. The capacity of endotox- fold produced only slight changes in percentages of viability ins to mediate alterations in cellular functions, particularly (compare 109vs. 10' CD50, figure 1). In addition, we isolated through the elicitation of endogenous cytokines, is now well viable EC, even when the cells were incubated with - 10 EC documented [42, 51]. Thus, the contributory effect of endo- CD50 of Shiga toxin. Although previous studies have clearly toxins in HUS may be mediated through changes in EC or demonstrated differential cytotoxicity between confluent inflammatory cell functions that have yet to be fully eluci- and nonconfluent HUVEC [18], our data suggest that even dated. within a population of confluent EC, there may be heteroge- Our data suggest that in the presence of the cytokine neity in terms of susceptibility to Shiga toxin. TNFa, the Shiga toxin EC CD50 may be reduced from JID 1991; 164 (August) Shiga Toxin and SLT-II-Mediated EC Toxicity 351

~ lO"8M to as low as ~ lO"10M (table 2). Although free EC may include differentialresponses to Shigatoxin or LPS. Shiga toxin or SLTs in human sera have not been quanti- There is one reportthat endotoxins are directlycytotoxic for tated, it is not unreasonableto speculate that this lower dose human glomerularendothelial cells [56]. Furtherstudies on of toxin may be in the physiologicallyrelevant range [52]. In the effect of Shiga toxin and endotoxin on human glomeru- addition, Shiga toxin may exacerbate any cytotoxic effects lar cells are clearlywarranted and are currentlyin progressin mediated directly by cytokines. rhTNFa at 20 ng/ml or re- our laboratory. combinant human interleukin-10 at 100 pg/ml have been demonstratednot to be directly cytotoxic for HUVEC [53]. However, in the presence of the protein synthesis inhibitor Acknowledgments or the RNA inhibitor cycloheximide synthesis actinomycin We thankStefanie Vogel for helpfuldiscussions; David Sil- D, HUVEC have been shown to be sensitized to the cyto- verman,Abla Creasey,and the OperatingRoom staffof Wash- toxic effects of these cytokines [53]. Therefore,by inhibiting ingtonAdventist Hospital for assistance in obtainingtissues and de novo protein synthesis, Shiga toxin and SLTs may act to reagents;Wei Du, Pin Yu Perera,and ThomasSellner for tech- block the protective mechanism(s) necessary to overcome nical assistance;and ElizabethLeach for art work. cytokine-mediateddirect cytotoxicity. We used an indirect immunofluorescenceassay to monitor the binding of Shiga toxin to EC. Toxin-specificbinding References to EC was reduced relativeto Vero and HeLa cells. We also 1. Bennish ML, HarrisRJ, WojtyniakBJ, StruelensM. Death in shigello- noted cell-to-cell variationsin toxin binding when Vero or sis: incidence and risk factors in hospitalized patients. J Infect Dis HeLa cells or EC were treatedwith equivalentdoses of Shiga 1990;161:500-6. toxin. Again, this suggeststhat the cells may consist of hetero- 2. Riley LW, Remis RS, Helgerson SD, et al. Hemorrhagiccolitis asso- in terms of the to bind ciated with a rare Escherichia coli serotype. N Engl J Med geneous populations capacity Shiga 1983;308:681-5. toxin. Toxin studies toxin-sensitive and binding using Shiga 3. Pai CH, Gordon R, Sims HV, BryanLE. Sporadiccases of hemorrhagic resistantHeLa cell lines have shown that 10-folddifferences colitis associatedwith EscherichiacoliO\51:H7. Clinical, epidemio- in toxin binding sites can produce 109-fold variability in logic, and bacteriologicfeatures. Ann InternMed 1984;101:738-42. toxin sensitivity [46, 47]. Thus, the minute differencesin 4. Strockbine NA, Jackson MP, Sung LM, Holmes RK, O'Brien AD. fluorescencewe observed cells in the toxin Cloning and sequencing of the genes for Shiga toxin from Shigella among binding 1. J Bacteriol 1988; 170:1116-22. differences in dysenteriaetype experiments may represent profound Shiga 5. Strockbine NA, Marques LRM, Newland JW, Williams Smith H, toxin sensitivity. To more directly compare differences in Holmes RK, O'BrienAD. Two toxin-convertingphages from Esche- toxin binding, we examined the glycolipid components of richia coli 0157:H7 strain 933 encode antigenically distinct toxins the cell membranes.Both EC types contained ~ 1000-fold with similarbiological activities. Infect Immun 1986;53:135-40. less than did Vero cells. Thus, the relative 6. Jackson MP, Neill RJ, O'Brien AD, Holmes RK, Newland JW. Nu- Gb3 insensitivity cleotide and of the structural for of human EC derivedfrom veins be corre- sequence analysis comparison genes large may directly Shiga-liketoxin I and Shiga-liketoxin II encoded by bacteriophages lated with reduced numbersof toxin-specificmembrane re- from Escherichiacoli 933. FEMS Microbiol Lett 1987;44:109-14. ceptors,and our data supportearlier studies showinga corre- 7. O'BrienAD, Holmes RK. Shiga and Shiga-liketoxins. Microbiol Rev lation of Shiga toxin sensitivitywith membraneGb3 content 1987;51:206-20. K. 47, 8. Endo Y, Tsurugi K, Yutsudo T, Takeda Y, OgasawaraT, Igarashi [45, 54]. Site of action of a Vero toxin from Escherichiacoli 0157:H7 Ourdata the relative of human (VT2) demonstrating insensitivity and of Shiga toxin on eukaryoticribosomes. RNA iV-glycosidaseac- vascularEC to Shiga toxin and the paucity of toxin-specific tivity of the toxins. Eur J Biochem 1988;171:45-50. membranereceptors suggest that the use of these cells may 9. Saxena SK, O'BrienAD, AckermanEJ. Shiga toxin, Shiga-liketoxin II not representthe most appropriatein vitro model of HUS. variant,and ricin are all single-siteRNA N-glycosidasesof 28S RNA studies have shown, however, that when microinjected into Xenopus oocytes. J Biol Chem 1989; Histopathologic glomeru- 264:596-601. lar EC is a hallmark of HUS and damage [16, 41]. Boyd 10. LindbergAA, SchultzJE, Westling M, et al. Identificationof the recep- Lingwood [45] showed that Gb3 is the predominantneutral tor glycolipid for Shigatoxin producedby Shigelladysenteriae type 1. glycolipid in human renal tissue; their values were ~2- to In: Lark D, ed. Protein-carbohydrateinteractions in biological sys- 30-fold greaterthan the amount of Gb3 we detected in EC. tems. London: Academic Press, 1986:439-46. the amount of in a tissue sam- 11. Jacewicz M, Clausen H, Nudelman E, Donohue-Rolfe A, Keusch GT. Although glycolipid complex of diarrhea.XI. Isolation of a toxin- not reflectthe relativeamount in a Pathogenesis Shigella Shigella ple may accurately partic- bindingglycolipid from rabbitjejunum and HeLacells and its identi- ular cell type, the data suggest that EC derived from large fication as globotriaosylceramide.J Exp Med 1986; 163:1391-404. veins may differin their toxin-receptorcontent fromglomer- 12. LingwoodCA, Law H, RichardsonS, et al. Glycolipid bindingof puri- ular endothelial cells. Finally, it is important to note that fied and recombinantEscherichia coli producedverotoxin in vitro. J human vascularEC and EC are and Biol Chem 1987;262:8834-9. glomerular functionally 13. Waddell Head Petric Cohen A, C. distinct cells. Glomerular EC are T, S, M, Lingwood Globotriosyl morphologically highly ceramide is specificallyrecognized by the Escherichiacoli verocyto- fenestratedand coated with a sialoglycoprotein-richanionic toxin 2. Biochem Biophys Res Commun 1987;152:674-9. glycocalix [55]. The structuraland functional differencesin 14. Fontaine A, Arondel J, SansonettiPJ. Role of Shigatoxin in the patho- 352 Tesh et al. JID 1991; 164 (August)

genesis of bacillary dysentery, studied by using a tox mutant of acterization of Shiga-liketoxin II and production of toxin-specific Shigella dysenteriae1. Infect Immun 1988;56:3099-109. monoclonal antibodies.Infect Immun 1988;56:1926-33. 15. Cleary TG. Cytotoxin-producingEscherichia coli and the hemolytic 36. Magnani JL, Smith DF, Ginsburg V. Detection of gangliosides that uremic syndrome.Pediatr Clin North Am 1988;35:485-99. bind choleratoxin: directbinding of ,25I-labeledtoxin to thin-layered 16. RichardsonSE, KarmaliMA, Becker LE, Smith CR. The histopathol- chromatograms.Anal Biochem 1980;108:399-402. ogy of the hemolytic uremic syndrome associated with verocyto- 37. Libby P, OrdovasJM, Auger KR, Robbins AH, BirinyiLK, Dinarello toxin-producing Escherichia coli infections. Hum Pathol 1988; CA. Endotoxin and tumor necrosis factor induce interleukin-1gene 19:1102-8. expression in adult human vascular endothelial cells. Am J Pathol 17. Gerlach H, Esposito C, Stern DM. Modulation of endothelial hemo- 1986;124:179-85. static properties:an active role in the host response.Annu Rev Med 38. SchleimerRP, RutledgeBK. Culturedhuman vascular endothelial cells 1990;41:15-24. acquireadhesiveness for neutrophilsafter stimulationwith interleu- 18. ObrigTG, Del Vecchio PJ, BrownJE, et al. Direct cytotoxic action of kin 1, endotoxin, and tumor-promotingphorbol diesters. J Immunol Shiga toxin on human vascular endothelial cells. Infect Immun 1986;136:649-54. 1988;56:2373-8. 39. Loppnow H, Libby P. Adult human vascularendothelial cells express 19. Mata LJ, GangarosaEJ, Caceres A, PereraDR, Mejicanos ML. Epi- the IL-6 gene differentiallyin responseto LPSor IL-l. Cell Immunol demic Shigabacillus dysentery in CentralAmerica. I. Etiologicinves- 1989;122:493-503. tigationsin Guatemala.J Infect Dis 1969;122:170-80. 40. Koster F, Levin J, Walker L, et al. Hemolytic-uremicsyndrome after 20. PereraLP, SamuelJE, Holmes RK, O'BrienAD. Identificationof three shigellosis. Relation to endotoxemia and circulatingimmune com- amino acid residues in the B subunit of Shiga toxin and Shiga-like plexes. N Engl J Med 1978;298:927-33. toxin type II that are essential for holotoxin activity. J Bacteriol 41. BertaniT, Abbate M, Zoja C, Corna D, Remuzzi G. Sequence of glo- 1991;173:1151-60. merularchanges in experimentalendotoxemia: a possible model of 21. Newland JW, StrockbineNA, Neill RJ. Cloning of genes for produc- hemolytic uremic syndrome.Nephron 1989;53:330-7. tion of Escherichia coli Shiga-like toxin type II. Infect Immun 42. LarrickJW, KunkelSL. The role of tumor necrosisfactor and interleu- 1987;55:2675-80. kin 1 in the immunoinflammatoryresponse. Pharm Res 1988; 22. Samuel JE, PereraLP, Ward S, O'BrienAD, GinsburgV, KrivanHC. 5:129-39. Comparisonof the glycolipid receptorspecificities of Shiga-liketoxin 43. Fuchs G, MobassalehM, Donohue-Rolfe A, MontgomeryRK, Grand type II and Shiga-like toxin type II variants. Infect Immun RJ, Keusch GT. Pathogenesisof Shigella diarrhea:rabbit intestinal 1990;58:611-8. binding site for Shigella toxin. Infect Immun 1987;53:372-7. 23. Recombinant DNA researchguidelines. Fed Register 1986;51:16972. 44. Donohue-RolfeA, JacewiczM, KeuschGT. Isolationand characteriza- 24. Westphal O, Jann K. Bacterial lipopolysaccharides.Extraction with tion of functionalShiga toxin subunitsand renaturedholotoxin. Mol phenol-waterand furtherapplications of the procedure.In: Whistler Microbiol 1989;3:1231-6. Vol 5. New York: RC, ed. Carbohydratechemistry. Academic Press, 45. Boyd B, Lingwood C. Verotoxin receptor glycolipid in human renal 1965:83-91. tissue. Nephron 1989;51:207-10. 25. O'Brien LaVeckGD. Purificationand characterizationof AD, Shigella 46. Eiklid K, Olsnes S. Interactionof Shigellashigae cytotoxin with recep- 1-like toxin Escherichiacoli. Infect Immun dysenteriae producedby torson sensitiveand insensitivecells. J Recept Res 1980; 1:199-213. 1983;40:675-83. 47. Jacewicz M, Feldman HA, Donohue-Rolfe A, BalsubramanianKA, 26. WatkinsMT, SharefkinJB, et al. Adult human MaciagTM, saphenous Keusch GT. Pathogenesis of Shigella diarrhea. XIV. Analysis of vein endothelial cells: assessment of their for reproductivecapacity Shiga toxin receptors on cloned HeLa cells. J Infect Dis endothelial seeding of vascularprostheses. J SurgRes 1984;36:588- 1989;159:881-9. 96. 48. BarrettTJ, Potter ME, Wachsmuth IK. Bacterialendotoxin both en- 27. Glaser BA, D'Amore PA, Michels RG, Patz A, FenslauA. Demonstra- hances and inhibits the toxicity of Shiga-liketoxin II in rabbitsand tion of vasoproliferativeactivity from mammalianretina. J Cell Biol mice. Infect Immun 1989;57:3434-7. 1980;84:298-304. 49. HarlanJM, HarkerLA, StrikerGE, Weaver LJ. Effectsof lipopolysac- 28. SharefkinJB, Fairchild KD, Albus RA, Cruess DF, Rich NM. The charides on human endothelial cells in culture. Thromb Res cytotoxic effect of surgicalglove powder particleson adult human 1983;29:15-26. vascular endothelial cell cultures: for clinical uses of implications 50. Harlan JM, Harker LA, Reidy MA, Gajdusek CM, Schwartz SM, tissue culture J Res 1986;41:463-72. techniques. Surg StrikerGE. Lipopolysaccharide-mediatedbovine endothelialcell in- 29. MK, JM. Quantitativemicroliter Gentry Dalrymple cytotoxicityassay jury in vitro. Lab Invest 1983;48:269-74. for Shigella toxin. J Clin Microbiol 1980;12:361-6. 51. Morrison DC, Ryan JL. Endotoxins and disease mechanisms. Annu 30. Brasaemle DL, Attie AD. MicroELISAreader of fixed, quantitation Rev Med 1987;38:417-32. stained, solubilized cells in microtitre dishes. BioTechniques 52. BS, TG, TG. Recent advances in 1988;6:418-9. Kaplan Cleary Obrig understanding the pathogenesis of the hemolytic uremic syndromes. Pediatr 31. Holmes RK, Twiddy EM. Characterizationof monoclonal antibodies Nephrol 1990;4:276-83. that react with unique and cross-linkingdeterminants of cholera en- 53. PohlmanTH, HarlanJM. Humanendothelial cell to terotoxin and its subunits. Infect Immun 1983;42:914-23. response lipopolysaccharide,interleukin-1, and tumor necrosis factor is 32. Hogan MM, Perera PY, Vogel SN. Examination of macrophagecell regulatedby Cell Immunol 1989;119:41-52. surface antigen regulation by rIFN-7 and IFN-a//? utilizing digital protein synthesis. 54. Cohen A, GE, Williams BRG, CA. Roles of imaging by a novel laser detection system. Anchored cell analysis Hannigan Lingwood glo- and in verotoxin and af- station (ACAS) 470. J Immunol Methods 1989;123:9-18. botriosyl- galabiosylceramide binding high interferon J Biol Chem 33. Griffin DE, Gentry MK, Brown JE. Isolation and characterizationof finity receptor. 1987;262:17088-91. monoclonal antibodies to Shiga toxin. Infect Immun 1983;41: 55. Bulger RE, Hebert SC. Structural-functionalrelationships in the kid- 430-3. ney. In: SchrierRW, Gottschalk CW, eds. Diseases of the kidney. 34. StrockbineNA, MarquesLRM, Holmes RK, O'BrienAD. Characteriza- Boston: Little, Brown, 1988:3-63. tion of monoclonal antibodies against Shiga-liketoxin from Esche- 56. Raghu G, StrikerLJ, StrikerGE. Lipopolysaccharide-mediatedinjury richia coli. Infect Immun 1985;50:695-700. to cultured human glomerularendothelial cells. Clin Immunol Im- 35. Downes FP, BarrettTJ, Green JH, et al. Affinitypurification and char- munopathol 1986;38:275-81.