differential tissue targeting and pathogenesis of verotoxins 1 and 2 in the mouse animal model

Kidney International, Vol. 62 (2002), pp. 832–845

CELL BIOLOGY – IMMUNOLOGY – PATHOLOGY

Differential tissue targeting and pathogenesis of verotoxins1 and 2 in the mouse animal model

NIELS W.P. RUTJES,1 BETH A. BINNINGTON, CHARLES R. SMITH, MARK D. MALONEY, andCLIFFORD A. LINGWOOD

Division of Infection, Immunity, Injury and Repair, Research Institute, Hospital for Sick Children, Toronto, Ontario, Canada;Department of Pediatric Laboratory Medicine, Hospital for Sick Children, Toronto, Canada; Spelman College, Atlanta,Georgia; and Departments of Biochemistry and Laboratory Medicine & Pathobiology, University of Toronto, Toronto,Ontario, Canada

ing a different lipid composition. AdamantylGb3, a soluble Gb3Differential tissue targeting and pathogenesis of verotoxins 1analogue, competed effectively for Gb3 binding by VT1 andand 2 in the mouse animal model.VT2 in vitro. However, the effect in the mouse model (onlyBackground. Both verotoxin (VT) 1 and VT2 share the samemeasured against VT2, due to the lower LD50, a concentrationreceptor, globotriaosyl ceramide (Gb3). Although VT1 is slightlyrequired for 50% lethality) was to increase, rather than reduce,more cytotoxic in vitro and binds Gb3 with higher affinity,

VT2 is more toxic in mice and may be associated with greater pathology and further reduce the VT2 serum clearance rate.pathology in human infections. In this study we have compared Additional renal pathology was seen in VT2 � adamantylGb3-the biodistribution of iodine 125 (125I)-VT1 and 125I-VT2 versus treated mice.pathology in the mouse. Conclusions. The lung is a preferential (Gb3) “sink” for VT1,

Methods. 125I-VT1 whole-body autoradiography defined the which explains the relatively slower clearance of VT2 and sub-tissues targeted. VT1 and VT2 tissue distribution, clearance, sequent increased VT2 renal targeting and VT2 mortality inand tissue binding sites were compared. The effect of a soluble this animal model.receptor analogue, adamantylGb3, on VT2/Gb3 binding and invivo pathology was assessed.

Results. 125I-VT1 autoradiography identified the lungs andThe hemolytic uremic syndrome (HUS) remains thenasal turbinates as major, previously unrecognized, targets,

while kidney cortex and the bone marrow of the spine, long leading cause of acute pediatric renal failure [1]. The syn-bones, and ribs were also significant targets. VT2 did not target drome comprises acute renal failure, thrombocytopenia,the lung, but accumulated in the kidney to a greater extent and microangiopathic hemolytic anemia [2]. Most oftenthan VT1. The serum half-life of VT1 was 2.7 minutes with

HUS is a complication of gastrointestinal infection with90% clearance at 5 minutes, while that of VT2 was 3.9 minutesverotoxin (VT) producing enterohemorrhagic Esche-with only 40% clearance at 5 minutes. The extensive binding

of VT1, but not VT2, within the lung correlated with induced richia coli (EHEC) [3, 4]. Several serotypes of EHEClung disease. Extensive hemorrhage into alveoli, edema, alveo- bacteria have been implicated in disease, but E. colilitis and neutrophil margination was seen only after VT1 treat-

O157:H7 is the predominant EHEC serotype associatedment. VT1 targeted lung capillary endothelial cells. Identicalwith HUS in North America and Europe [5, 6]. EHECtissue binding sites (subsets of proximal/distal tubules and col-

lecting ducts) for VT1 and VT2 were detected by toxin overlay bacteria elaborate either or both of two serologicallyof serial frozen kidney sections. Glucosuria was found to be a distinct VTs: VT1 and VT2. VTs show a high degree ofnew marker of VT1- and VT2-induced renal pathology and homology with Shiga toxin, produced by Shigella dysen-positive predictor of outcome in the mouse, consistent with

teriae 1, and hence are also referred to as Shiga toxin 1VT-staining of proximal tubules. Lung Gb3 migrated on thin-and Shiga toxin 2, respectively.layer chromatography (TLC) faster than kidney Gb3, suggest-

After synthesis in the intestine, the toxins may reachthe systemic circulation by transcellular translocation1 Current address: Faculty of Medicine, University of Groningen, Gron-

ingen, The Netherlands through intestinal epithelial cells [7, 8] aided by neutro-phil migration in the opposite direction [9]. Recent inves-

Key words: biodistribution, lung and renal pathology, globotriaosyltigations in vitro [10] and in vivo [11] have shown thatceramide, glycolipid analogue, adamantylGb3, glucosuria, Shiga-like

toxin. VTs bind to monocytes without inducing cytotoxicity.This is postulated to be a possible mechanism of trans-

Received for publication December 11, 2001port in the circulation. Upon reaching the target organs,and in revised form March 8, 2002

Accepted for publication March 27, 2002 VT1 and VT2 bind to a specific receptor, globotriaosylceramide (Gb3, also termed CD77 or the pk blood group 2002 by the International Society of Nephrology

832

Selective VT1 lung targeting and decreased renal pathology 833

antigen) [12]. Various cells in the human body express stability. These authors also noted that VT1-treated micedied in the absence of excessive renal pathology, sug-this receptor, but most common complications of EHECgesting that the fatal effects lay in a different organinfection are hemorrhagic colitis, HUS, and neurologicalsystem. These studies also showed that the Gb3 bindingdysfunction [13].affinity for VT2 was, like VT2c [47], approximately 10-VTs are A:B5 subunit toxins, consisting of five recep-fold lower than that of VT1 [46].tor-binding subunits (B subunit) and one enzymatically

In rabbits, VT1 is more lethal than VT2. Brain lesionsactive subunit (A subunit). Upon binding to Gb3 via thecan be induced by VT2, but not VT1, whereas VT1-B-pentamer, the toxin is internalized via clathrin-coatedinduced spinal cord lesions can be seen [48, 49]. VT2,pits and caveoli-mediated endocytosis [14–16]. Once in-although generally less cytotoxic than VT1 in vitroternalized, the toxin undergoes a process of retrograde[50, 51] shows increased cytotoxicity for cultured humantransport to the Golgi and/or endoplasmic reticulumrenal microvascular endothelial cells [52]. The A subunit[17, 18]. The mechanistic paradigm is that the A subunitof VT2, but not VT1, was recently found to contain ais cleaved into an A1 and an A2 fragment and the A1sequence capable of binding anti-apoptotic partners thatfragment then targets the 28S RNA portion of 60S ribo-prevent Bcl-2 action, which may provide a second mecha-somal subunits in the cytosol and by so doing disruptsnism for VT2-induced cell killing [53].ribosomal function of protein production resulting in cell

In this study we expanded on the mouse model of VTdeath. While inhibition of cellular protein synthesis isby defining the biodistribution and clearance characteris-clearly established, VT appears to have diverse intercel-tics of VT1 and VT2. This data show significant differ-lular actions. VT induces apoptosis, a process requiringences in tissue targeting and clearance of VT1 and VT2,protein synthesis, in many cell types [19–25] and canwhich could explain the distinct toxicities observed ininduce specific gene expression [26–31]. VT can depuri-mice and, perhaps, relate also to the differential pathol-nate DNA as well as RNA [32, 33]. Intracellular tar-ogy in humans.geting of the holotoxin to the endoplasmic reticulum and

VT1 and VT2 binding to the receptor Gb3 plays thenucleus has been associated with high VT sensitivityprimary role in determining tissue targeting and suscepti-[18, 34]. Thus, the exact mechanism or combination ofbility in rabbits [54, 55]. Prophylactic attempts to com-mechanisms by which VTs kill cells remains speculativepete with insoluble Gb3 analogues within the gastrointes-[35, 36].tinal tract have proven ambiguous [56], possibly becauseIn human kidneys, Gb3 is expressed by endothelial cellsVT may be transcytosed directly across the intestinal(and epithelial cells [37]) located primarily in the pediat-epithelium from the site of bacterial attachment and isric glomerulus and, to a lesser extent, in the tubules [38].receptor independent [9]. A different approach would

Glomerular Gb3 expression is reduced or lost in the adult be to use soluble Gb3 analogues to prevent systemic VTkidney [38], despite the fact that the total level of renal tissue targeting. Analysis of the serum Gb3 levels in VTGb3 is increased [39]. The ontologically regulated glo- producing E. coli-infected patients who did or did notmerular expression of Gb3 may explain the fact that HUS develop HUS showed that the level was higher in themainly occurs in young children [38]. This has been re- non-HUS patients [57], suggesting that soluble Gb3 ana-cently supported by the detection of VT2 in glomeruli of logues could indeed, prove an effective prophylactic.pediatric, but not geriatric, VT producing E. coli-induced High-affinity VT/Gb3 binding requires the lipid moietyHUS cases [40]. At autopsy, glomerular lesions are seen. [58] and, if administered to compete in such a format,Changes noted include glomerular endothelial cell swell- would likely partition into cells to provide additional VTing, thromboses, congested glomeruli, arteriolar throm- receptors, rather than fewer. The weakly VT-bindingboses, which are common at the hilum of glomeruli and galactose (�1-4) galactose disaccharide from Gb3 canare sometimes seen proximally, including in interlobular be appropriately multimerized to provide a structurearteries, and in the most severe cases, cortical infarction. capable of cross-linking (and hence inactivating) VT BThese changes are assumed to be responsible for the subunits to effectively prevent VT cytotoxicity [59, 60]acute renal failure observed [41]. in vitro. Our approach has been to develop monomeric

Epidemiological studies have suggested that patients soluble Gb3 analogues that retain high-affinity VT bind-infected with EHEC producing VT1 in combination with ing [61]. We have now tested the ability of such a speciesVT2 or VT2 alone were more prone to develop serious to alter the biodistribution and toxicity of VT2 in thecomplications like HUS than patients infected with mouse model.EHEC producing VT1 alone [42–45]. Tesh et al showeda clear difference between in vivo toxicities of VT1 and

METHODSVT2 in mice [46]. Their studies showed a 400-fold lowerReagentsLD50 for VT2 and that VT2 caused greater renal pathol-

ogy than VT1. It was thought that the difference in toxic- Monoclonal antibody (mAb) PH1, specific for the VT1B subunit, has been previously described [62]. PH1 wasity could partially be explained by a difference in toxin

Selective VT1 lung targeting and decreased renal pathology834

purified from culture supernatant using Protein A-sepha- tabolism and Pharmacokinetics Division, Lawrence, KS,USA). Mice were pretreated with oral potassium iodiderose chromatography. Rabbit polyclonal antiserum against

the purified VT1 B subunit was prepared, as previously 24 hours prior to injection of 125I-VT1 to block thyroiduptake of any free 125I. 125I-VT1 (1.12 �g) was adminis-described [63]. Rabbit antiserum raised against VT2e

and reactive with VT2 was a generous gift of Dr. Carlton tered intravenously through the tail vein and the animalswere sacrificed by CO2 inhalation 15 minutes (N � 2)Gyles, University of Guelph, Guelph, Canada. Antibody

against the facilitative glucose transporter GLUT-2, se- or 45 minutes (N � 3) later. Animals were frozen inhexane/dry ice and embedded in carboxymethyl cellu-lectively expressed in the S1 segment of proximal tubules

[64, 65], was obtained from Alpha Diagnostic Interna- lose. Sagittal cryosections (40 �M) using a Richert-JungPolycut S cryomictrotome (Leica Microsystems, Inc.,tional, Inc. (San Antonio, TX, USA). Biotinylated Dol-

ichos biflorus agglutinin (which binds �-N-acetyl galac- Bannockburn, IL, USA) were collected on adhesive tape,dried, and exposed to Kodak SB-5 x-ray film. Hematoxy-tosamine, highly expressed in collecting ducts [66]) and

peanut agglutinin (which binds terminal �-galactose and lin and eosin (H&E) stained sections used for autoradi-ography were used for tissue assignment.galactosyl (�-1,3) N-acetylgalactosamine; distal tubules)

were purchased from Vector Laboratories (Burlingame, Quantitative tissue distribution. Mice were injected viathe tail vein with 2 �g/kg 125I-VT1 or 125I-VT2 (about 50CA, USA).ng/mouse) diluted in 200 �L saline. In these studies, no

Toxins allowance for the difference in LD50 values for thesetoxins was made and pathologic-induced changes forVT1 was purified from strain JB28 [67], as previously

described [68]. VT2 was purified from E. coli R82pJES1- VT2 might have occurred. Pilot experiments were per-formed to examine toxin distribution at 1, 2, or 8 hours20DH5a, kindly provided by Dr. J. Samuel (Texas A&M

University, College Station, TX, USA). Nucro-Technics, post injection and the effects of perfusion prior to tissuecollection on the biodistribution data obtained. For finalInc. (Toronto, Ontario, Canada) analyzed purified toxins

for endotoxin using the Limulus amoebocyte lysate clearance/biodistribution studies, groups of mice (N � 8)were injected with toxin and a 50 �L blood sample wasmethod. VT1 had �0.1 EU/�g (�10 pg/�g) while VT2

had �12.5 EU/�g (�1.25 ng/�g). These lipopolysaccha- obtained from the lateral saphenous vein at 5 and 30minutes post injection. At 60 minutes post injection, miceride (LPS) levels are far below those shown to cause

synergistic or protective effects on VT2 pathology [69]. were anesthetized (0.1 mL/g 3.6% chloral hydrate intra-peritoneally) and a 200 �L blood sample was obtainedPurified toxins were labeled with iodine 125 (125I) using

iodogen, as described previously [70]. The protein con- from the orbital plexus. Upon taking the last blood sam-ple, mice were sacrificed immediately by cervical disloca-centration of the isolated iodinated VTs was determined

using the bicinchoninic acid (BCA) protein assay (Pierce tion and tissues were removed for analyses of radioactiv-ity using a gamma counter.Chemical Co., Rockford, IL, USA). Greater than 97%

of recovered radioactive material was protein-bound asEffect of adamantylGb3 on VT2 biodistributiondetermined by trichloroacetic acid precipitation. Specific

activities were typically 8 to 12 � 106 cpm/�g for VT1 The semi-synthetic, soluble Gb3 derivative, N-ada-mantylGb3 (adaGb3), was prepared from human renaland VT2. To verify that radiolabeling of the toxins did

not affect activity, a Vero cell cytotoxicity assay was Gb3, as previously described [61]. Two groups of mice(N � 8) were injected intravenously with a mixture ofperformed to compare labeled and unlabeled toxin. Io-

dinated VT1 was heat-inactivated by treatment at 90 to 4 mg/kg adaGb3 and 2 �g/kg 125I-VT2 (representing about100 �g (100 nmol) of adaGb3 and 50 ng VT2 (0.7 pmol)100�C for 30 minutes. Binding of denatured toxin to

Vero cells was reduced by 90% compared to native per mouse) in 200 �L. The procedures followed matchedthose used in the biodistribution studies discussed above.125I-VT1.

Mice Clinicopathologic studies with VT2 challenged mice

The LD50 of VT2 in mice is 400-fold lower than thatThe mice were 12- to 16-week-old female BALB/cmice (weight range, 20 to 30 grams) that were purchased of VT1 [46]. Therefore, VT2 was chosen for in vivo

protection studies using adaGb3. Initial studies showedfrom Charles River, Canada. Male mice were used inthe whole-body autoradiography studies. The mice were that a dose of 2 ng/mouse (80 ng/kg) caused development

of fatal disease in 100% of mice 80 to 105 hours post-maintained under a 12-hour light-dark cycle and fed withstandard diet and water ad libitum. injection. This dose represents 1.5 to 4 times the reported

LD50 in Balb/c mice [69, 71].Biodistribution The ability of adaGb3 to alter the VT2 disease course

with respect to clinical, biochemical, and histologic symp-Whole-body autoradiography. Whole-body autoradi-ography was performed by Oread Biosafety, Inc. (Me- toms was investigated. In all cases, mice were injected


intravenously through the tail vein (N � 5) for each tavidin peroxidase system. Sections were counterstainedwith hematoxylin, dehydrated, and mounted. Serial sec-group. Group 1 received 2 ng (80 ng/kg) VT2. Group 2

received a mixture of 2 ng VT2 and 100 �g (4 mg/kg) tions were also stained with PAS, anti-GLUT-2, biotinyl-ated Dolichos biflorus agglutinin, peanut agglutinin, oradaGb3. Group 3 was injected intravenously with 4 mg/kg

adaGb3 alone. A fourth group of mice (N � 3) was 1% goat serum/PBS as antibody control.injected intravenously with sterile saline alone. AdaGb3-

In situ immunodetection of targeted VTVT2 mixtures were pre-incubated for 20 to 30 minutesat room temperature prior to injection. Mice were injected through the tail vein with 10, 50 or

100 �g of VT1 or VT2 in 200 �L of saline. After 1 hour,In experiments involving blood collection, 24 hoursprior to injection, a 200 �L blood sample was taken from the animals were sacrificed by cervical dislocation and

organs were collected and divided. Half was placed inthe lateral saphenous vein. At 72 hours post injection,mice were anesthetized with chloral hydrate (3.6% 0.01 O.C.T compound and snap-frozen in liquid nitrogen for

preparation of frozen sections. The remainder was placedmL/g intraperitoneally) and a 300 �L blood sample wastaken from the orbital plexus. After obtaining the second in neutral buffered formalin for a minimum of 16 hours

before being paraffin-embedded and sectioned.blood sample, mice were immediately sacrificed by cervi-cal dislocation and tissues were processed as described VT was detected in the frozen sections essentially as

described above for VT staining of frozen kidney sec-under Histopathology below.All blood samples were collected with heparinized tions, with the following two exceptions. The frozen sec-

tions were fixed with 100% acetone for 10 minutes priorcapillary tubes. Blood was centrifuged immediately uponcollection, plasma and cells were separated, and plasma to assay rather then dried overnight, and the VT1 binding

step was omitted. Tissues from mice injected with salinewas stored at –20�C until analysis. To monitor for signsof renal dysfunction, plasma samples were analyzed for rather than VT served as controls for non-specific bind-

ing of anti-toxin antibodies.blood urea nitrogen (BUN) and creatinine at the Depart-ment of Pediatric Laboratory Medicine, Hospital for Sick Paraffin sections were de-paraffinized by immersion in

xylene, followed by baths of decreasing ethanol contentChildren, in Toronto, Ontario, Canada. Blood glucoselevels were measured using a One Touch Basic blood (100%, 95%, 70% ethanol in water), and finally, water.

After de-waxing, antigen retrieval was required for de-glucose monitoring system (Lifescan Canada, Ltd., Bur-naby, B.C., Canada). Urine samples were collected upon tection of VT in kidney sections. Antigen unmasking

was performed by heating the de-waxed sections in 0.01spontaneous voiding. Using Albustix (Bayer, Inc., To-ronto, Ontario, Canada), urine was analyzed for glucose, mol/L citrate buffer pH 6, in a pressure cooker [73].protein, pH, specific gravity, erythrocytes, and leukocytes.

VT TLC overlayHistopathology Glycosphingolipids (GSLs) were extracted from lungs

and kidneys (pooled from 5 animals) as previously de-Animals treated with VT1, VT2, VT2 � adaGb3,adaGb3, or saline were sacrificed by cervical dislocation. scribed [39] and equivalent aliquots according to tissue

weight were separated by TLC and overlaid with eitherKidney, lung, brain, spleen, and liver were fixed in neu-tral-buffered formalin and then paraffin-embedded and VT1 or VT2 [74]. Toxin binding was visualized using ECL

detection (Amersham Biosciences, Piscataway, NJ,sectioned. All sections were stained with H&E and, inaddition, kidney sections were stained with periodic acid- USA).Schiff (PAS) to identify the brush-border of proximal

Statistical analysistubules.Results are expressed as the mean standard deviation.

VT staining of frozen kidney sections

A fresh tissue sample was placed in Tissue-Tek O.C.T.RESULTS

compound (Miles, Inc., Diagnostics Division, Elkhart,Whole-body autoradiography afterIN, USA) and snap-frozen in liquid nitrogen. Serial 6 �m125I-VT1 administration.frozen sections were cut and stored at –70�C. The immu-

noperoxidase VT staining was performed, as described Several illustrative sagittal sections are shown in Fig-ure 1. No additional organs were labeled in other sections.[72]. Sections were dried overnight and then blocked

separately for endogenous peroxidase, avidin, and biotin. VT1 labeling of the kidney cortex, lung, and the nasalturbinates were the most prominent features of the auto-The sections were incubated with 50 ng/mL of VT1 or

VT2 diluted in 1% goat serum/phosphate-buffered saline radiogram (Fig. 1). Labeling of the bone marrow of thespine, long bones, and ribs was also extensive. The labeling(PBS) for 30 minutes. After washing, polyclonal anti-

VT1 or anti-VT2 was added and bound antibody was in the lung could be seen to be non-uniform since bronchi-oles were not labeled and foci within the lung could bedetected with a biotinylated goat anti-rabbit IgG-step-


Fig. 2. Serum clearance of 125I-VT1 and 125I-VT2. Approximately 50 ng125I-labeled toxin was injected into mice intravenously (N � 8) andcounts remaining in circulation over the next hour were measured. Theeffect of mixing the semi-synthetic soluble Gb3 derivative N-ada-mantylGB3 (adaGb3) with 125I-VT2 prior to injection was also deter-mined. (�), 125I-VT1; (�), 125I-VT2; (�), 125I-VT2 � adaGb3.

counts injected. To compare clearances of both VT1 andVT2, the amount of toxin as measured in the bloodsamples was expressed as a percentage of injected countsrecovered in 50 �L blood (Fig. 2). Within 5 minutes,90% of injected VT1 but only 40% of injected VT2 wascleared. After 60 minutes there were sixfold less countscirculating for VT1 than for VT2. The estimated half-lives for VT1 and VT2 were 2.7 and 3.9 minutes, respec-tively, based on the decrease in counts between timepoints 0 and 5 minutes.

Biodistribution. All measurements were at 1 hour postinjection. Both groups of mice injected with 125I-VT1 orVT2 showed tissue localization of the toxins primarilyin the lungs and nasal turbinates (Fig. 3) consistent withthe autoradiography study (Fig. 1). Toxin targeting to these

Fig. 1. Autoradiography of 125I-VT1 injected mouse. Sample sagittal organs has not been previously reported. However, VT1sections from several mice are shown with labeled tissues indicated. showed a 10-fold higher lung targeting than VT2. VT2The lower panel highlights the non-uniform labeling in the lung seen

preferentially targeted kidneys (2.7-fold that of VT1),on reduced exposure.whereas VT1 spleen targeting was twofold that of VT2.Heat inactivation, for the most part, eliminated tissuetargetting (Fig. 3). Increased levels of the inativated VT1

distinguished on less exposure (Fig. 1, lower panel). The in the liver and kidney probably reflect clearance of thespleen was significantly labeled, but less than lung. The denatured protein.liver was barely labeled above the background. No label- The biodistribution of 125I-VT1 was also compared ining of the gastrointestinal tract, central nervous system, male and female outbred CD1 mice. Pulmonary and renalbrain stem, spinal chord, testes, lymph nodes, thymus, re- VT1 accumulation was �40% higher in male than femaleproductive tract, pancreas, blood, skeletal/cardiac muscle, animals, consistent with the greater (renal) Gb3 contentdermis, or epidermis was seen. in males [76], but otherwise, distribution was not signifi-

cantly different from BALB/c mice (data not shown).Tissue distribution of 125I-VT1 and 125I-VT2

VT1 and VT2 immunolocalizationClearance. Blood samples were taken at three differenttime points. Because blood was not collected before 5 min- Kidney. VT1 and VT2 were immunolocalized in fixed/utes, the data for zero time points were calculated from paraffin-embedded or frozen renal sections from mice 1

hour after 10, 50 or 100 �g of VT1 or VT2 had beenthe estimated blood volume for each mouse [75] and the


Fig. 3. Biodistribution of 125I-VT1 and 125I-VT2. Approximately 50 ng 125I-labeled toxin was injected into mice intravenously (N � 8). After 1 hour,the toxin distribution to various tissues was determined and expressed as the percentage of injected counts per minute recovered per mg of tissue.(�), heat-denatured VT1; (�), VT2; ( ), VT1.

injected (Fig. 4). The VT1 and VT2 immunostaining structures labeled by both VT1 and VT2 as collectingducts (Fig. 5e), while immunostaining with anti-GLUT-2patterns were similar. Immunostaining was clear in the[64] identified others as proximal convoluted tubulesfrozen sections, even at the 10 �g dose, but morphology(Fig. 5f). Some distal tubules labeled with peanut aggluti-was better preserved in the fixed, embedded sample.nin [77] were labeled with VT1 or VT2, but most wereClear staining of distinct tubule subpopulations was seennot (Fig. 5g).for both toxins. No glomerular staining was seen (Fig.

Lung. Frozen-section immunostaining of VT1 accu-4 A to C). In the paraffin sections, it was clear that VT-mulated in the lung following intravenous injection showstargeted tubules were dilated and the epithelial cellsareas of some intra-alveolar capillaries, and patches offlattened, even after this short period (Fig. 4 C, D). Thispneumocytes within the alveoli are targeted (Fig. 6).effect was dose dependent. Immunodetection of targetedDistinct, sometimes discontinuous, subpopulations ofVT1 in the unmasked paraffin sections was adequate atendothelial cells within some capillaries were labeled.100 �g but poor at 50 �g (and not detectable at 10 �g).Epithelial cells were, for the most part, VT1-negative.Nevertheless, the dismorphic tubules could be identifiedVT could not be detected in the paraffin sections perhaps

in the H&E section and the corresponding stained tubule due to a failure to unmask the antigen in this tissueidentified (for the 50 and 100 �g dosage). Dilated tubules with the conditions used. Gb3 was detected in untreatedwere less apparent in H&E stained sections at the 10 �g mouse lung frozen sections by VT1 overlay, but the mor-dosage (not shown). phology was not sufficiently retained to allow identifica-

By overlay of (untreated) mouse kidney frozen section of cell targets.tions, both VT1 and VT2 were found to bind exactly the

Pathologysame tubular structures (Fig. 5). Neither toxin boundto glomeruli (Fig. 5 a, b). Labeling serial sections with Kidney. Mice injected with 1.5 to 4 LD50 (VT1 or

VT2) showed kidney pathology, characteristic of tubularthe lectin D. biflorus [66] identified some of the renal


Fig. 4. Immunolocalization of kidney-tar-geted VT1 and VT2. Frozen (A, B) or fixed,paraffin-embedded (C, D) sections of kidneywere prepared from mice 1 hour after injec-tion with either 50 �g VT1 or VT2 and stainedwith antiserum against VT1 (A, C) or VT2(B). A similar subpopulation of tubules arelabeled by each toxin but neither bound toglomeruli (arrows). In fixed sections, tubulestargeted by VT1 (C) were identified in hema-toxylin and eosin (H&E)-stained serial sec-tions as dilated and containing flattened epi-thelial cells (D).

necrosis as previously reported [46] by day 3. Glomerular some edematous material was lost during tissue prepara-tion/staining (Fig. 7L). Edema of larger blood vesselsmorphology appeared normal. A significant subpopula-was evident (Fig. 7K). Margination of neutrophils intion of proximal and especially distal tubules were dilatedvenules was extensive (Fig. 7H) and infiltration of typein both VT1- and VT2-treated mice. Medullary tubularII pneumocytes was evident. In contrast, in VT2-treateddilation was more widespread after VT1 treatment. In-animals, there was some slight edema and neutrophilterstitial hemorrhage was seen for all VT-treated mice.margination evident, but lung histology was, for the mostNeutrophil migration to sites of tubular injury was notpart, indistinguishable from controls (Fig. 7 B, E, H com-seen. Apoptosis, as monitored by terminal deoxy trans-pared with C, F, I).ferase uridine triphosphate nick end-labeling (TUNEL)

Other. Active acute phagocytosis was present in focistaining, was not increased in the kidneys of VT-treatedin the spleen after either VT2 or VT1 treatment. TUNELmice (not shown). Glucose was detected in the urine ofstaining revealed increased apoptosis in spleens of VT1VT-treated mice after 3 to 4 days, before the onset ofor VT2 but not in saline-treated mice (not shown). Also,any visible symptoms (Table 1) and was present untiliron pigment from increased red cell turnover was evi-the time of death. Glucosuria was an excellent predictordent in the spleen of toxin-treated mice.of mortality (Table 1). Co-administration of the neu-

In the liver, lipid vacuoles were seen in VT2-treatedtralizing anti-VT1 mAb PH1 [62] with VT1 completelyanimals but were less apparent for VT1-treated animals

protected mice from pathology and development of glu-(not shown).

cosuria (Table 2). VT-treated mice did not become an-uric, but did develop a urine-concentrating defect (Ta- Tissue Gb3 contentble 1). Using urinalysis sticks, no changes in urinary The Gb3 content of glycolipid extracts of lung andprotein concentration, pH, or specific gravity was mea- kidney was visualized by VT1 and VT2 TLC overlaysured. but erythrocytes and leukocytes were sometimes (Fig. 8). The mouse renal Gb3 migrated more slowly thandetected in urine of moribund mice. that from lung. VT2/Gb3 binding was significantly less

Lung. Lung sections of 2 LD50 [78] VT1-treated mice than that of VT1. A dose-dependent binding to humanshowed significant hemorrhage (Fig. 7 A, E, J) and marked renal Gb3 standard was seen, and both toxins efficientlyalveolitis was apparent (Fig. 7 E, H). The lungs had a recognized adaGb3.congested appearance (Fig. 7H) with thickened edema-

Treatment with a soluble Gb3 analogtous alveolar walls showing numerous entrapped erythro-cytes (Fig. 7L). Edema within alveoli was also seen (Fig. The water-soluble Gb3 derivative, adaGb3 [61], was

effective to prevent VT1 and VT2 binding to immobi-7L). The edema was often discontinuous, suggesting that


Table 1. Biochemical parametersa of VT2-treated mice

Blood glucose Plasma urea Plasma creatinine Clinicalb disease Colorc

Treatment Glucosuria mmol/L mmol/L lmol/L at 72 hours of urine

Baseline � 4.720.6 6.431.27 21.22.4 � ��(N � 32) (N � 43) (N � 39)

2 ng VT2 � (0/5) 4.840.84 9.922.0 27.03.24 � ��2 ng VT2 � adaGb3 � (5/5) 4.500.43 24.64.0 55.05.66 � (5/5) �adaGb3 � (0/5) 4.760.65 6.280.63 22.42.6 � ��4 ng VT2 � (0/5)d 5.680.26 16.27.9 45.326.1 � (1/5)4 ng VT2 � adaGb3 � (4/5)e 4.220.85 24.65.8 69.35.8 � (4/6)adaGb3 � (0/5) 5.040.84 6.81.0 22.24.6 � (0/5)

a Parameters monitored at 72 hours post injectionb Ruffled fur, inactivec VT-treated mice developed urine-concentrating defect and when moribund had nearly colorless urine (�) in contrast to untreated mice (��)d Glucosuria was detected at 96 hourse No urine obtained from one of five mice

Table 2. Clinical parameters of mice treated with VT1PH1monoclonal antibodies (mAb)

Time of Urine glucose Clinical ColorTreatment deatha hour mmol/L disease of urine

1.2 �g VT1 58.5 nd � nd1.2 �g VT1 58.5 14–28 � �1.2 �g VT1 68 14 � �1.2 �g VT1 � PH1 68 � � ��1.2 �g VT1 � PH1 68 � � ��1.2 �g VT1 � PH1 68 � � ��

a VT1-treated mice were sacrificed when moribund and healthy PH1-protectedanimals were sacrificed for analysis of histopathology

lized Gb3 as monitored by receptor enzyme-linked immu-nosorbent assay (RELISA) (Fig. 9). The IC50 for VT1was �10 �mol/L [61] while that for VT2 was higher (�50�mol/L). Due to the much lower LD50 in mice, we usedVT2 to determine whether this protection in vitro couldbe effective in vivo. We mixed 500 �mol/L adaGb3 withapproximately 1 LD50 VT2 and injected it into the ani-mals. Surprisingly, mice co-injected with VT2 � adaGb3

were found to die more rapidly than those injected withVT2 alone (96 hours compared with 72 hours). Indeed,since a near-LD50 dose was administered, in some cases,

Fig. 7. Comparison of VT1 and VT2 induced lung pathology. Micethe mice treated with VT2 alone survived (Table 1).(N � 3) were injected with 2 LD50 of VT1 [102] or VT2, or with saline.Glucose was detected in the urine of VT2 � adaGb3- Toxin injected mice were sacrificed when moribund and saline-treated

treated mice earlier than for mice given VT2 alone (Ta- mice at corresponding time points. Hematoxylin and eosin (H&E) stain-ing of lung of VT1 (A), VT2 (B), and saline (C ) treated mice; adjacentble 1). AdaGb3 alone at this dosage had no obvious toxicblood vessel and alveoli in VT1 (D), VT2 (E ) and saline (F ) treatedeffect in mice and renal histology was normal (Fig. 10 A,mice; alveoli of VT1 (G), VT2 (H ) and saline (I ) treated animals. In

D, G). In the VT2 � adaGb3-treated mice, many tubular VT1-treated mice, alveolar hemorrhage (J ), blood vessel edema (K ),and alveolar edema (L) were widespread.epithelial cells were swollen. This was seen in the cortex,

but particularly in the medulla, such that many tubulelumina were occluded (Fig. 10 C, F). Evidence of epithe-lial cell detachment from the basement membrane is

brain pathology was seen for VT1-, VT2- , or VT2 �seen in both VT2- and VT2 � adaGb3-treated mice. TheadaGb3-treated mice.epithelial morphology of the affected tubules was more

Comparison of the biodistribution of 125I-VT2 with anddysplastic, and more granular eosinophilic spheres in thewithout adaGb3 showed that, in most tissues, VT2 tissuetubular lumina (Fig. 10 H, I), typical of acute tubular

necrosis, were seen for VT2 � adaGb3-treated mice. No targeting was marginally increased, rather than de-


Fig. 10. Effect of AdamantylGb3 (adaGb3)on VT2-induced renal pathology. Mice (N �5) were injected intravenously with adaGb3,VT2, or a mixture of VT2 and adaGb3. Ani-mals were sacrificed at 72 hours post injectionand hematoxylin and eosin (H&E) stainedparaffin sections of renal tissue were inspectedfor pathology. (A, D, G), adaGb3-treatedmice; (B, E, H), VT2-treated mice; (C, F, I ),VT2 � adaGb3-treated mice. (A to C), cortex;(D to F), medulla; (G to I), tubules at highpower. Glomerular morphology appears unaf-fected by VT2 (B). The tubular dilation (Band E) are reversed in the VT2 � adaGb3

mice (C, D). The epithelial cell narrowing (Eand H) after VT2 treatment is also reversed,but the epithelial cells of many tubules nowbecome enlarged such that their lumina areoccluded (F). The epithelial cell layer of tu-bules of VT2-treated mice was often dismor-phic (H) and the frequency of eosinophilicnuclei (in dying cells) increased after VT2 �adaGb3 treatment (I, arrows)

creased, in the presence of adaGb3 (results not shown).This was consistent with the finding that the serum half-life for 125I-VT2 was increased to 8.7 minutes in the pres-ence of adaGb3 (Fig. 1).

DISCUSSION

VT-induced cell cytotoxicity [79, 80] and likely in vivopathology [38] are mediated by binding to its glycolipidreceptor, globotriaosyl ceramide [12]. Can this bindingin any way explain the increased sensitivity of mice (andhumans?) to VT2? The binding of VTs to Gb3 is differen-

section corresponding to A and B stained with non-immune serum.Fig. 6. Immunolocalization of lung-targeted VT1. Frozen sections oflung were prepared from mice 1 hour after injection with 50 �g VT1. Discontinuous staining of endothelial cells lining a capillary is indicated

in (B). Continuous staining of endothelial cells within a capillary is(A), Hematoxylin and eosin (H&E) stain; (B), corresponding serialsection stained with anti-VT1; (C ), anti-VT1 immunostaining; (D), lung shown in (C). Some staining of pneumocytes is also evident.


�

Fig. 5. Comparison of VT1 and VT2 binding sites in frozen kidney sections. Serial frozen sections of kidney were stained with VT1 (a, c) or VT2(b, d), the lectin Dolichos biflorus (e), anti-GLUT2 ( f), peanut agglutinin (g) and periodic acid-Schiff (PAS; h). Both toxins label the same collectingducts and proximal tubules. Comparison staining of the equivalent areas is shown (c to h). A representative glomerulus unstained by VT1 or VT2 isshown (a and b). Four collecting ducts labeled with D. biflorus and VT1, VT2 are indicated (e). The arrowhead indicates a D. biflorus-reactivecollecting duct, not labeled with peanut agglutinin (g). Proximal tubules reactive with VT1, VT2 and anti-GLUT2 are indicated (f). Anti-VT2 renalstaining after in vivo VT2 administration ( j) is compared to PNA staining (i) of serial sections. A VT2-negative, PNA-positive distal tubule is marked* (i), while a PNA-negative, VT2-positive proximal tubule is marked * ( j). Distal tubules labeled by both PNA and VT2 are indicated by arrows.

tially modulated by the lipid content of the glycosphingo- place. Glucose appeared in the urine approximately 3to 4 days after VT injection and prior to the onset oflipid (GSL) [58, 81, 82] and also by the phospholipid

content of the membrane in which it is contained [83]. any visible symptoms. Acute pancreatitis as a cause ofglucosuria was excluded by showing that there was noDue to the extensive repertoire of such modulation,

Gb3-mediated tissue targeting of these VTs subtypes increase of the blood glucose over time. Using Albustixto detect urinary glucose proved an easy and inexpensivecould be distinct and is likely complex. Although ourway to monitor disease development in mice. Urinaryoriginal studies compared the Gb3 binding of VT1 to thatglucose was seen for VT1- or VT2-treated animals andof VT2c [82], rather than VT2, VT1 and VT2 receptoran absolute correlation with eventual death was seen.binding is distinguishable [84]. Although the differentCo-administration of mAb anti-VT1 protected micemembers of the VT family all bind Gb3, this recognition isagainst VT1-induced mortality and development of glu-slightly different in each case, since removal of individualcosuria. VT2 mortality was exacerbated by adaGb3, andhydroxyl groups within the terminal galabiose moietyurinary glucose was detected earlier in such mice. How-has a differential effect on the binding of each toxinever, increasing VT above a lethal dosage did not in-variant [74].crease the level of glucosuria.Whole-body autoradiography has identified new tissue

The murine renal VT distribution does not correlatetargets for VT1, specifically the nasal turbinates and thewith the pathology observed in human pediatric HUSlung. No binding to neural, reproductive, or gastrointesti-where mainly glomeruli are affected by the toxins [4, 41,nal tissue was seen. Binding within the spleen and bone88] or with VT binding in pediatric renal sections [38, 89].marrow confirms the VT propensity for lymphoid cellsGlucosuria has not been reported in humans and this[85–87].may prove another distinction between this model andOur results show a decreased clearance rate and anHUS. However, HUS patients have been shown to haveincreased binding in kidneys for VT2 when comparedurinary markers of proximal tubule pathology early into VT1, which was primarily recovered in the lungs.the course of disease [90] and evidence of tubular cell

Renal targeting death [91, 92]. Several studies have reported that cul-tured proximal tubule cells are highly sensitive to VT1VT1 and VT2 have the same pattern of binding in

the kidney, determined either by immunolocalization of [23, 93, 94]. The distribution of VT in the murine kidneyis similar to that seen in overlays of sections of adultkidney-targeted VT or by frozen-section VT overlay,

both targeting proximal and distal tubules and collecting human kidneys [38], which, on an epidemiological basis,are less susceptible to VT-induced HUS.ducts. By both approaches, no labeling of glomeruli was

seen. Detection of in vivo renal-targeted VT1 shows that In situ detection of in vivo kidney-targeted VT bettermimics the normal pathophysiology of systemic VT, astoxin binding to tubular subpopulations has an immedi-

ate effect to dilate the targeted tubule and (as a conse- compared to overlays with VTs on untreated sections.In these sections, VT1 and VT2 were found in similarquence?) flatten the epithelial cells. No overt pathology

in these cells could be seen. tubule subpopulations. This means that not only can VTbind to collecting ducts and distal and proximal tubules,Frozen serial section overlays showed precisely that

the same renal structures were reactive with both toxins. it can access these structures post glomerular filtration.The fact that in situ immunostained toxins labeled tu-Thus, although differences in Gb3-mediated tissue tar-

geting are seen with respect to the lung, Gb3-mediated bules that were dilated and contained flattened epithelialcells indicates that VT can have an immediate effect ontargeting in the kidney sections is not discriminatory.

However, not every proximal/distal tubule was bound renal fluid balance. The cell flattening may compromisewater reabsorption, resulting in local tubular dilation.by the toxins, which means that there is a difference in

Gb3 expression among tubules. VT binding in the over- The later renal pathology indicates that the tubular dila-tion becomes more extensive, consistent with the glucos-lays was, for the most part, coincident with GLUT2,

indicating binding to the S1 and S2 segment of the convo- uria, a new predictor of outcome in this animal model.We found lipid inclusions in the liver of VT2- (3 of 3),luted proximal tubule. It is in these segments of the

nephron where re-uptake of filtered glucose mainly takes but less in VT1- (1of 3) treated mice. This suggests that


Fig. 8. VT1 / VT2 thin-layer chromatography(TLC) overlay of renal and lung glycolipids.Glycolipids were isolated from mouse lungand kidney and an aliquot representing 20 mgof tissue was separated by TLC and visualizedby orcinol (A). VT binding to Gb3 was de-tected by TLC overlay with either VT1 (B)or VT2 (C) as described [74]. Lane 1, adaGb3;lanes 2, 3, standard glycolipid mixture at 0.25,0.5 �g each; lane 4, mouse renal glycolipids;lane 5, mouse lung glycolipids. Glycolipidstandards are, from the top of the plate, gluco-sylceramide, galactosylceramide, lactosylcera-mide, Gb3, Gb4, Forssman glycolipid.

hemorrhage may initiate the more generalized pathol-ogy. Since VT binding is restricted to pulmonary capillar-ies, high local VT1 levels must accumulate to accountfor the biodistribution observed. Extensive localizedhemorrhage and edema were evident. Margination ofpolymorphonuclear neutrophils was seen throughout thelung, indicating an active inflammatory response. Thehigh level of VT1 absorbed by the lungs could explainthe difference in LD50 for the toxins. The lungs couldfunction as an absorbing buffer, which permits less toxinto reach the kidneys where it could otherwise havecaused more life-threatening pathology. However, wedid not assess the kidneys of mice given sub-lethal dosesof VT1.

Fig. 9. AdamantylGb3 (adaGb3)inhibition of VT/Gb3 binding. VT1 orA similar effect may be relevant to HUS in humans.VT2 were mixed with the indicated concentration of adaGb3 and tested

for residual binding to immobilized Gb3 in a RELISA [61]. (�), VT1; Although the renal Gb3 increases with age [39], suscepti-(�), VT2. bility to VT-induced HUS is greatest in the very young

when the receptor is expressed in the glomerulus [38].Thus, the increased expression of tubular Gb3 as a func-tion of age might, to some extent, prevent toxin reabsorp-VT2 induces a more severe, generalized metabolic stresstion by providing a similar, less crucial VT “sink.”such that lipid stores are mobilized for breakdown.

VT receptor isoformsLung targetingThe Gb3 species from kidney and lung clearly showSome degree of pulmonary symptoms has been re-

differential migration on TLC, indicating differencesported in a minor fraction of HUS patients [95]. VT1,within the lipid moiety, which may affect VT bindingbut not VT2, was detected in lung sections obtained from[99]. Mouse renal Gb3 has been reported to contain hy-a patient who died of HUS [22], despite the fact thatdroxylated fatty acids [76], which would be consistentboth VT1 and VT2 were detected in the kidney of thewith its reduced TLC mobility. Overall, VT2 binds Gb3same patient [96]. Reported pulmonary histopathologyless effectively than VT1 [84] and this was reflected inincludes microthrombosis, capillary congestion in the al-the TLC overlay both for renal and lung Gb3. This proce-veolar septa [97], and hemorrhage [98]. Mechanisms re-dure, however, is semiquantitative and not optimum forsponsible for this pathology are thought to be the samedistinguishing the relative binding of Gb3 isoforms [82].as mechanisms causing angiopathy at the renal level.Kinetic binding analysis and Gb3 characterization hasOur biodistribution study shows that the lungs are ayet be performed to determine whether differential bind-major target for both toxins, although 10-fold more VT1ing to Gb3 lipid isoforms could provide the basis of thethan VT2 was bound.differential tissue recognition by VT1 and VT2.When detecting injected VT1 in the lung sections, we

found the toxin to be localized primarily in capillariesReceptor analogue treatmentin some of the alveolar septa. Based on the extent of

It is clear that our first attempts to prevent VT/Gb3VT1-induced pathology, we expected the lung distribu-tion to be more widespread. However, VT1-induced binding in vivo have failed. Despite the fact that adaGb3


dren, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada.inhibited VT2/Gb3 binding in vitro, co-administrationE-mail: [email protected]

exacerbated, rather than attenuated, VT2-induced dis-ease. This was particularly evident when less than 1 LD50 REFERENCESVT2 was given, when all the VT2-treated mice survived,

1. Arbus GS: Association of verotoxin-producing E. coli and vero-but all the VT2 � adaGb3 mice died. This correlatestoxin with hemolytic uremic syndrome. Kidney Int 51(Suppl 58):with evidence of “novel” renal pathology, despite the fact S91–S96, 1997

that some of the VT2-alone pathologies were reduced 2. Kaplan BS, Cleary TG, Obrig TG: Recent advances in under-standing the pathogenesis of the hemolytic uremic syndromes.by adaGb3 co-administration. The additional pathologicPediatr Nephrol 4:276–283, 1990features seen in the presence of adaGb3 may result from 3. Karmali MA: Infection by Verocytotoxin-producing Escherichia

acceleration of the VT2-induced pathology, rather than coli. Clin Microbiol Rev 2:15–38, 19894. Ray PE, Liu XH: Pathogenesis of Shiga toxin-induced hemolyticnew lesions per se. The unusual properties of adaGb3

uremic syndrome. Pediatr Nephrol 16:823–839, 2001include that despite its water solubility, the TLC mobility5. Griffin PM: Escherichia coli O157:H7 and other enterohemor-

is not greatly affected and VT binding is effectively moni- rhagic Escherichia coli, in Infections of the Gastrointestinal Tract,edited by Blaser MJ, Smith PD, Ravdin JI, et al, New York,tored by TLC overlay. Such solubility properties mightRaven Press, Ltd., 1995, pp 739–761result in complex effects in vivo. Our in vivo results are

6. Karch H, Bielaszewska M, Bitzan M, et al: Epidemiology andsimilar to studies attempting to use lyso-Gb3 to protect diagnosis of Shiga toxin-producing Escherichia coli infections.

Diag Microbiol Infect Dis 34:229–243, 1999mice from VT2, which found symptoms were increased7. Hurley BP, Jacewicz M, Thorpe CM, et al: Shiga toxins 1 andrather than alleviated by this treatment [100].

2 translocate differently across polarized intestinal epithelial cells.Tissue Gb3 is heterogeneous within its lipid moiety and Infect Immun 67:6670–6677, 1999

8. Acheson DWK, Moore R, De Breucker S, et al: Translocationmembrane environment and therefore may bind VTsof Shiga toxin across polarized intestinal cells in tissue culture.differentially [81]. We would interpret our results inInfect Immun 64:3294–3300, 1996

terms of inhibition of lower affinity VT2/Gb3 interactions 9. Hurley BP, Thorpe CM, Acheson DW: Shiga toxin translocation(and hence increasing the serum half-life as observed) to across intestinal epithelial cells is enhanced by neutrophil transmi-

gration. Infect Immun 69:6148–6155, 2001allow more VT2 to bind to high-affinity (renal) receptors,10. te Loo DM, Monnens LAH, van de Velden TJAM, et al: Bindingthus increasing/altering pathology. The organization of and transfer of verocytotoxin by polymorphonuclear leukocytes

Gb3 at the plasma membrane into lipid microdomains in hemolytic uremic syndrome. Blood 95:3396–3402, 200011. te Loo DM, Heuvelink AE, de Boer E, et al: Vero cytotoxinhas recently been found to be important in determining

binding to polymorphonuclear leukocytes among householdsVT1 sensitivity [101]. Such organization may affect the with children with hemolytic uremic syndrome. J Infect Disrelative Gb3-binding affinity of VT1 (and VT2). 184:446–450, 2001

12. Lingwood CA: Role of verotoxin receptors in pathogenesis.At specific Gb3 concentrations, evidence for coopera-Trends Microbiol 4:147–153, 1996tive VT1-receptor binding has been found [102]. If such 13. Griffin PM, Ostroff SM, Tauxe RV, et al: Illnesses associated

binding also occurs for VT2, an alternative explanation with Escherichia coli 0157:H7 infections: A broad clinical spec-trum. Ann Int Med 109:705–712, 1988for the effect of adaGb3 might be that binding of adaGb3

14. Sandvig K, Olnes S, Brown J, et al: Endocytosis from coatedto a proportion of the Gb3 binding sites within the B pits of Shiga toxin: A glycolipid-binding protein from Shigellasubunit pentamer cooperatively increases the Gb3 bind- dysenteriae 1. J Cell Biol 108:1331–1343, 1989

15. Khine AA, Lingwood CA: Capping and receptor mediated endo-ing affinity of the residual sites to increase VT2 pathogen-cytosis of cell bound verotoxin (Shiga-like toxin) 1; Chemicalesis. However, this explanation would not be consistent identification of an amino acid in the B subunit necessary for

with the increased serum half-life seen for VT2 in the efficient receptor glycolipid binding and cellular internalization.J Cell Physiol 161:319–332, 1994presence of adaGb3.

16. Schapiro F, Lingwood CA, Furuya W, et al: pH-independentWhile these results are disappointing, they indicate targeting of glycolipids to the Golgi complex. Am J Physiol 274:that adamantyl glycolipids are active in an in vivo system 319–332, 1998

17. Sandvig K, Garred Ø, Prydz K, et al: Retrograde transport ofto reveal further complexity of VT2-receptor targeting.endocytosed Shiga toxin to the endoplasmic reticulum. NatureThese studies must also serve as a caution in the develop- 358:510–512, 1992

ment of receptor analogues for clinical use. 18. Arab S, Lingwood C: Intracellular targeting of the endoplasmicreticulum/nuclear envelope by retrograde transport may deter-mine cell hypersensitivity to verotoxin: Sodium butyrate or selec-ACKNOWLEDGMENTStion of drug resistance may induce nuclear toxin targeting viaglobotriaosyl ceramide fatty acid isoform traffic. J Cell PhysiolThis study was supported by CIHR grant # MT13073 (to C.A.L)177:646–660, 1998and by NIH/MBRS grant # GM08241 and NIH/RIMI grant # RR11598

19. Inward CD, Williams J, Chant I, et al: Verocytotoxin-1 induces(M.D.M). The authors are indebted to Dr. Ernest Cutz, Departmentapoptosis in vero cells. J Infect 30:213–218, 1995of Pediatric Laboratory Medicine, Hospital for Sick Children, for evalu-

20. Arab S, Murakami M, Dirks P, et al: Verotoxins inhibit theation of the lung histopathology. We thank Ms. Lily Morikawa andgrowth of and induce apoptosis in human astrocytoma cells.Mr. Davin Chark from the Hospital for Sick Children for preparingJ Neuro Oncol 40:137–150, 1998and immunostaining some histological sections and Ms. Bonnie Welsh

21. Taguchi T, Uchida H, Kiyokawa N, et al: Verotoxins inducefor assistance in animal procedures.apoptosis in human renal tubular epithelium derived. Kidney Int53:1681–1688, 1998Reprint requests to Clifford A. Lingwood, Ph.D., Division of Infection,

Immunity, Injury and Repair, Research Institute, Hospital for Sick Chil- 22. Uchida H, Kiyokawa N, Taguchi T, et al: Shiga oxins induce


apoptosis in pulmonary epithelium-derived cells. J Infect Dis 180: cial reference to production of Vero cytotoxins VT1. Epidem Inf99:613–624, 19871902–1911, 1999

43. Bitzan M, Karch H, Altrogge H, et al: Hemolytic-uremic syn-23. Williams J, Boyd B, Nutikka A, et al: A comparison of thedrome associated with a variant Shiga-like cytotoxin of Esche-effects of verocytotoxin-1 on primary human renal cell cultures.richia coli 0111. Pediatr Infect Dis J 7:128–132, 1988Toxicol Lett 105:47–57, 1999

44. Bitzan M, Ludwig K, Klemt M, et al: The role of Escherichia24. Gordon J, Challa A, Levens JM, et al: CD40 ligand, Bcl-2, andcoli O 157 infections in the classical (enteropathic) haemolyticBcl-xl spare group I Burkett lymphoma cells from CD77-directeduraemic syndrome: Results of a central European, multicentrekilling via verotoxin-1 B chain but fail to protect against thestudy. Epidemiol Infect 110:183–196, 1993holotoxin. Cell Death Differ 7:785–794, 2000

45. Nakao H, Takeda T: Escherichia coli Shiga toxin. J Nat Toxins25. Jones NL, Islur A, Haq R, et al: Escherichia coli Shiga toxins9:299–313, 2000induce apoptosis in epithelial cells that is regulated by the Bcl-2

46. Tesh V, Burris J, Owens J, et al: Comparison of the relativefamily. Am J Physiol Gastrointest Liver Physiol 278:G811–G819,toxicities of Shiga-like toxins type I and type II for mice. Infect2000Immun 61:3392–3402, 199326. Tesh VL, Ramegowda B, Samuel JE: Purified Shiga-like toxins

47. Head S, Karmali M, Lingwood CA: Preparation of VT1 andinduce expression of proinflammatory cytokines from murineVT2 hybrid toxins from their purified dissociated subunits: Evi-peritoneal macrophages. Infect Immun 62:5085–5094, 1994dence for B subunit modulation of A subunit function. J Biol27. Ramegowda B, Tesh VL: Differentiation-associated toxin recep-Chem 266:3617–3621, 1991tor modulation, cytokine production, and sensitivity to Shiga-like

48. Fujii J, Kinoshita Y, Kita T, et al: Magnetic resonance imagingtoxins in human monocytes and monocytic cell lines. Infect Immunand histopathological study of brain lesions in rabbits given intra-64:1173–1180, 1996venous verotoxin 2. Infect Immun 64:5053–5060, 199628. Bitzan MM, Wang Y, Lin J, et al: Verotoxin and ricin have novel

49. Fujii J, Kinoshita Y, Yutsudo T, et al: Toxicity of shiga toxin 1 ineffects on preproendothelin-1 expression but fail to modify nitricthe central nervous system of rabbits. Infect Immun 69:6545–6548,oxide synthase (ecNOS) expression and NO production in vascu-2001lar endothelium. J Clin Invest 101:372–382, 1998

50. Yutsudo T, Honda T, Miwatani T, et al: Physicochemical charac-29. Kojima S, Yanagihara I, Kono G, Sugahara T: Nasu H et al:terization of A and B subunits of Shiga toxin and reconstitution ofmkp-1 encoding mitogen-activated protein kinase phosphatase 1,holotoxin from isolate subunits. Microbiol Immunol 31:189–197,a verotoxin 1 responsive gene, detected by differential display1987reverse transcription-PCR in CaCo-2 cells. Infect Immun 68:2791–

51. Kiyokawa N, Taguchi T, Mori T, et al: Induction of apoptosis2796, 2000in normal human renal tubular epithelial cells by Escherichia coli30. Ishii H, Takada K, Higuchi T, et al: Verotoxin-1 induces tissueShiga toxins 1 and 2. J Infect Dis 178:178–184, 1998factor expression in human umbilical vein endothelial cells

52. Louise C, Obrig T: Specific interaction of Escherichia colithrough activation of NF-�B/Rel and AP-1. Thromb Haemost 84:0157:H7-derived Shiga-like toxin II with human renal endothelial712–721, 2000cells. J Infect Dis 172:1397–1401, 199531. Thorpe CM, Smith WE, Hurley BP, et al: Shiga toxins induce, 53. Suzuki A, Doi H, Matsuzawa F, et al: Bcl-2 antiapoptotic proteinsuperinduce, and stabilize a variety of C–X-C chemokine mRNAs mediates verotoxin II-induced cell death: possible association be-

in intestinal epithelial cells, resulting in increased chemokine ex- tween Bcl-2 and tissue failure by E. coli 0157:H7. Genes Dev 14:pression. Infect Immun 69:6140–6147, 2001 1734–1740, 2000

32. Brigotti M, Accorsi P, Carnicelli D, et al: Shiga toxin 1: damage 54. Zoja J, Corna D, Farina C, et al: Verotoxin glycolipid receptorsto DNA in vitro. Toxicon 39:341–348, 2001 determine the localization of microangiopathic process in rabbits

33. Brigotti M, Carnicelli D, Accorsi P, et al: 4-aminopyrazolo challenged with verotoxin-1. J Lab Clin Med 120:229–238, 1992[3,4-d] pyrimidine (4-APP) as a novel inhibitor of the RNA and 55. Bast DJ, Brunton JL, Karmali MA, et al: Toxicity and immuno-DNA depurination induced by Shiga toxin 1. Nucleic Acids Res genicity of a verotoxin 1 mutant with reduced globotriaosylcera-28:2383–2388, 2000 mide receptor binding in rabbits. Infect Immun 65:2019–2028,

34. Facchini LM, Lingwood CA: A verotoxin 1 B subunit-lambda 1997cro chimeric protein specifically binds both DNA and globotriao- 56. Armstrong GD, McLaine PN, Rowe PC: Clinical trials of syns-sylceramide (Gb3) to effect nuclear targeting of exogenous DNA orb-Pk in preventing hemolytic uremic syndrome, in Escherichiain Gb3 positive cells. Exp Cell Res 269:117–129, 2001 coli O157:H7 and other Shiga toxin-producing E. coli strains,

35. van Setten PA, van Hinsbergh VW, Van der Heuvel LP, et edited by Kaper JB, O’Brien AD, Washington, D.C., Americanal: Verocytotoxin inhibits mitogenesis and protein synthesis in Society for Microbiology, 1998, pp 374–384purified human glomerular mesangial cells without affecting cell 57. Watarai S, Yokota K, Kishimoto T, et al: Relationship betweenviability: evidence for two distinct mechanisms. J Am Soc Nephrol susceptibility of hemolytic-uremic syndrome and levels of globo-8:1877–1888, 1997 triaosylceramide in human sera. J Clin Microbiol 39:798–800, 2001

36. Simon M, Cleary T, Hernandez J, et al: Shiga toxin 1 elicits 58. Boyd B, Zhiuyan Z, Magnusson G, et al: Lipid modulationdiverse biological responses in mesangial cells. Kidney Int of glycolipid receptor function: Presentation of galactose �1-454:1117–1127, 1998 galactose disaccharide for verotoxin binding in natural and syn-

37. Hughes A, Stricklett P, Schmid D, et al: Cytotoxic effect of thetic glycolipids. Eur J Biochem 223:873–878, 1994Shiga toxin-1 on human glomerular epithelial cells. Kidney Int 59. Dohi H, Nishida Y, Mizuno M, et al: Synthesis of an artificial57:2350–2359, 2000 glycoconjugate polymer carrying Pk-antigenic trisaccharide and

38. Lingwood CA: Verotoxin-binding in human renal sections. Neph- its potent neutralization activity against Shiga-like toxin. Bioorgron 66:21–28, 1994 Med Chem 7:2053–2062, 1999

39. Boyd B, Lingwood CA: Verotoxin receptor glycolipid in human 60. Kitov PI, Sadowska JM, Mulvey G, et al: Shiga-like toxins arerenal tissue. Nephron 51:207–210, 1989 neutralized by tailored multivalent carbohydrate ligands. Nature

40. Chaisri U, Nagata M, Kurazono H, et al: Localization of Shiga 403:669–672, 2000toxins of enterohaemorrhagic Escherichia coli in kidneys of paedi- 61. Mylvaganam M, Lingwood C: Adamantyl globotriaosyl cera-atric and geriatric patients with fatal haemolytic uraemic syn- mide—A monovalent soluble glycolipid mimic which inhibits ver-drome. Microb Pathog 31:59–67, 2001 otoxin binding to its glycolipid receptor. Biochem Biophys Res

41. Richardson SE, Karmali MA, Becker LE, et al: The histopathol- Commun 257:391–394, 1999ogy of the hemolytic uremic syndrome associated with verocyto- 62. Boulanger J, Petric M, Lingwood CA, et al: Neutralizationtoxin-producing Escherichia coli Infections. Hum Pathol 19:1102– receptor-based immunoassay (NeutrELISA) for detection of neu-1108, 1988 tralizing antibodies to Escherichia coli verocytotoxin 1. J Clin

42. Scotland SM, Willshaw GA, Smith HR, et al: Properties of Micro 28:2830–2833, 199063. Boyd B, Richardson S, Gariepy J: Serological responses to thestrains of Escherichia coli belonging to serogroup 0157 with spe-


B subunit of Shiga-like toxin 1 and its peptide fragments indicate toxin 2c preferentially recognize different globotriaosyl ceramidefatty acid homologues. J Biol Chem 269:11138–11146, 1994that the B subunit is a vaccine candidate to counter the action

83. Arab S, Lingwood CA: Influence of phospholipid chain length onof the toxin. Infect Immun 59:750–757, 1991verotoxin/globotriaosyl ceramide binding in model membranes:64. Dominguez J, Camp K, Maianu L, et al: Glucose transportersComparison of a surface bilayer film and liposomes. Glycoconjof rat proximal tubule: differential expression and subcellularJ 13:159–166, 1996distribution. Am J Physiol 262:F807–812, 1992

84. Nakajima H, Kiyokawa N, Katagiri YU, et al: Kinetic analysis65. Wallner EI, Wada J, Tramonti G, et al: Status of glucose trans-of binding between Shiga toxin and receptor glycolipid Gb3Cerporters in the mammalian kidney and renal development. Renby surface plasmon resonance. J Biol Chem 276:42915–42922,Fail 23:301–310, 2001200166. Laitinen L, Virtanen I, Saxen L: Changes in the glycosylation

85. Cohen A, Madrid-Marina V, Estrov Z, et al: Expression ofpattern during embryonic development of mouse kidney as re-glycolipid receptors to Shiga-like toxin on human B lymphocytes:vealed with lectin conjugates. J Histochem Cytochem 35:55–65, a mechanism for the failure of long-lived antibody response to

1987 dysenteric disease. Int Immunol 2:1–8, 199067. Huang A, DeGrandis S, Friesen J, et al: Cloning and expression 86. Menge C, Wieler LH, Schlap T, et al: Shiga toxin 1 from Esche-

of the genes specifying Shiga-like toxin production in Escherichia richia coli blocks activation and proliferation of bovine lympho-coli H19. J Bact 166:375–379, 1986 cyte subpopulations in vitro. Infect Immun 67:2209–2217, 1999

68. Lingwood CA, Law H, Richardson S, et al: Glycolipid binding 87. Sugatani J, Igarashi T, Shimura M, et al: Disorders in the im-of purified and recombinant Escherichia coli-produced verotoxin mune responses of T- and B-cells in mice administered intrave-in vitro. J Biol Chem 262:8834–8839, 1987 nous verotoxin 2. Life Sci 67:1059–1072, 2000

69. Palermo M, Alves-Rosa F, Rubel C, et al: Pretreatment of mice 88. Inward C, Howie J, Fitzpatrick M, et al: Renal histopathology inwith lipopolysaccharide (LPS) or IL-1beta exerts dose-dependent fatal cases of diarrhoea-associated haemolytic uraemic syndrome.opposite effects on Shiga toxin-2 lethality. Clin Exp Immunol Pediatr Nephrol 11:556–559, 1997

89. Lingwood CA, Boyd B, Nutikka A: Analysis of interactions119:77–83, 2000between glycosphingolipids and microbial toxins, in Methods in70. Boyd B, Tyrrell G, Maloney M, et al: Alteration of the glycolipidEnzymology, edited by Merrill A, Hannun Y, London, Aca-binding specificity of the pig edema toxin from globotetraosyl todemic Press, 2000, pp 459–473globotriaosyl ceramide alters in vivo tissue targeting and results

90. Takeda T, Dohi S, Igarashi T, et al: Impairment by verotoxinin a VT1-like disease in pigs. J Exp Med 177:1745–1753, 1993of tubular function contributes to the renal damage seen in haemo-71. Suzuki K, Tateda K, Matsumoto T, et al: Effects of interactionlytic uraemic syndrome. J Infect 27:339–341, 1993between Escherichia coli verotoxin and lipopolysaccharide on

91. Karpman D, Hakansson A, Perez MT, et al: Apoptosis of renalcytokine induction and lethality in mice. J Med Microbiol 49:905–cortical cells in the hemolytic-uremic syndrome: In vivo and in910, 2000vitro studies. Infect Immun 66:636–644, 199872. Nutikka A, Binnington-Boyd B, Lingwood C: Methods for the

92. Kaneko K, Kiyokawa N, Ohtomo Y, et al: Apoptosis of renalidentification of host receptors for verotoxin, in Methods in Molec-tubular cells in Shiga-toxin-mediated hemolytic uremic syndrome.ular Medicine, edited by Philpot D, Ebel F, Totowa, NY, Humana Nephron 87:182–185, 2001

Press (in press) 93. Hughes AK, Stricklett PK, Kohan DE: Cytotoxic effect of73. Pileri SA, Roncador G, Ceccarelli C, et al: Antigen retrieval Shiga toxin-1 on human proximal tubule cells. Kidney Int 54:648–

techniques in immunohistochemistry: comparison of different 649, 1998methods. J Pathol 183:116–123, 1997 94. Kodama T, Nagayama K, Yamada K, et al: Induction of apoptosis

74. Nyholm PG, Magnusson G, Zheng Z, et al: Two distinct binding in human renal proximal tubular epithelial cells by Escherichiasites for globotriaosyl ceramide on verotoxins: Molecular model- coli verocytotoxin 1 in vitro. Med Microbiol Immunol 188:73–78,ing and confirmation by analogue studies and a new glycolipid 1999receptor for all verotoxins. Chem Biol 3:263–275, 1996 95. Siegler R, Loghman-Adham M, Timmons O: Acute respiratory

75. Diehl KH, Hull R, Morton D, et al: A good practice guide to failure in the hemolytic uremic syndrome. Clin Pediatr 34:660–662,the administration of substances and removal of blood, including 1995

96. Uchida H, Kiyokawa N, Horie H, et al: The detection of Shigaroutes and volumes. J Appl Toxicol 21:15–23, 2001toxins in the kidney of a patient with hemolytic uremic syndrome.76. McCluer R, Deutsch C, Gross S: Testosterone-responsivePediatr Res 45:133–137, 1999mouse kidney glycosphingolipids: Developmental and inbred strain

97. Upadhyaya K, Barwick K, Fishaut M, et al: The importanceeffects. Endocrinol 113:251–258, 1983of nonrenal involvement in hemolytic uremic syndrome. Pediatr77. Engel U, Breborowicz D, Bog-Hansen T, et al: Lectin staining65:115–120, 1980of renal tubules in normal kidney. APMIS 105:31–34, 1997

98. Green J, Brenner B, Gery R, et al: Adult hemolytic uremic78. Wolski VM, Soltyk AM, Brunton JL: Mouse toxicity and cyto-syndrome associated with nonimmune deposit crescentic glomer-kine release by verotoxin 1 B subunit mutants. Infect Immun 69:ulonephritis and alveolar hemorrhage. Am J Med Sci 296:121–125,579–583, 2001198879. Waddell T, Cohen A, Lingwood CA: Induction of verotoxin

99. Lingwood CA: A glycone modulation of glycolipid receptor func-sensitivity in receptor deficient cell lines using the receptor glyco-tion. Glycoconj J 13:495–503, 1996lipid globotriosyl ceramide. Proc Natl Acad Sci USA 87:7898– 100. Takenaga M, Igarashi R, Higaki M, et al: Effect of a soluble

7901, 1990 pseudo-receptor on verotoxin 2-induced toxicity. J Infect Chemo-80. Jacewicz M, Feldman HA, Donohue-Rolfe A, et al: Pathogene- ther 6:21–25, 2000

sis of Shigella diarrhea. XIV. Analysis of Shiga toxin receptors 101. Falguieres T, Mallard F, Baron C, et al: Targeting of Shigaon cloned HeLa cells. J Infect Dis 159:881–889, 1989 toxin b-subunit to retrograde transport route in association with

81. Pellizzari A, Pang H, Lingwood CA: Binding of verocytotoxin detergent-resistant membranes. Mol Biol Cell 12:2453–2468, 20011 to its receptor is influenced by differences in receptor fatty acid 102. Peter M, Lingwood C: Apparent cooperativity in multivalentcontent. Biochem 31:1363–1370, 1992 verotoxin globotriaosyl ceramide binding: Kinetic and saturation

82. Kiarash A, Boyd B, Lingwood CA: Glycosphingolipid receptor binding experiments with radiolabelled verotoxin [125I]-VT1. Bio-chim Biophys Acta 1501:116–124, 2000function is modified by fatty acid content: Verotoxin 1 and Vero-

differential tissue targeting and pathogenesis of verotoxins 1 and 2 in the mouse animal model

Documents