Bovine lymphocytes express functional receptors for Escherichia coli Shiga toxin 1

Article available online at http://www.idealibrary.com on1 Microbial Pathogenesis 2002; 33: 251±264

doi:10.1006/mpat.2002.0527

Bovine lymphocytes express functionalreceptors for Escherichia coli Shiga toxin 1Ivonne Stamma, M. Wuhrerb, R. Geyerb, G. Baljera & Ch. Mengea*

aInstitut fuÈr Hygiene und Infektionskrankheiten der Tiere der Justus-Liebig-UniversitaÈt, Giessen andbBiochemisches Institut am Klinikum der Justus-Liebig-UniversitaÈt Giessen, D-35392 Giessen, Germany

(Received April 19, 2002; accepted in revised form August 13, 2002)

Interactions of Shiga toxins (Stxs) and immune cells contribute to the pathogenesis of diseases dueto Stx-producing Escherichia coli (STEC) infections in humans and facilitate the persistence ofinfection in asymptomatically infected cattle. Our recent ®ndings that bovine B and T lymphocytesexpress Gb3/CD77, the human Stx-receptor, prompted us to determine whether the bovinehomologue also mediates binding and internalization of Stx1. In fact, Stx1 holotoxin and recombi-nant B subunit (rStxB1) bound to stimulated bovine peripheral blood mononuclear cells, especiallyto those subpopulations (B cells, BoCD8� T cells) that are highly sensitive to Stx1. Competition andHPTLC-binding studies con®rmed that Stx1 binds to bovine Gb3, but different receptor isoformswith varying af®nities for rStxB1 were expressed during the course of lymphocyte activation. At leastone of these isoforms mediated toxin uptake. An anti-StxB1 mouse monoclonal antibody, used as amodel for bovine serum antibodies speci®c for Stx1, modulated rather than generally preventedrStxB1 binding to and internalization by the receptors. The presence of functional Stx1-receptors onbovine lymphocytes explains the immunomodulatory effect of Stx1 observed in cattle at a molecularlevel. Furthermore, expression of such receptors by bovine but not human T cells enlightens thebackground for the differential outcome of STEC infections in cattle and man, i.e., persistentinfection and development of disease, respectively. & 2002 Elsevier Science Ltd. All rights reserved.

Keywords: EHEC, STEC, Shiga toxin 1, receptor, bovine lymphocytes.

Introduction

Enterohemorrhagic Escherichia coli (EHEC), asubtype of the Shiga toxin-producing E. coli(STEC), are the cause of epidemics and sporadiccases of hemorrhagic colitis in humans thatcan progress to life-threatening diseases as thehemolytic uremic syndrome (HUS) [1]. The main

0882±4010/02/$ ± See front matter

*Author for correspondence: E-mail:[email protected]

virulence factor of EHEC are the Shiga toxins(Stxs), potent cytotoxins of the ribosome-inactivating type [2]. Upon absorption fromthe intestine [3] Stxs damage renal endothelialcells resulting in a thrombotic microangiopathy,the histological hallmark of HUS [4]. However,in recent years compelling evidences arose thatcells other than endothelial cells i.e., monocytesand granulocytes [5, 6] must be involved in theinitiation of this syndrome [7]. To explain theinconsistencies in our current model of HUSpathogenesis, Heyderman et al. [8] further

& 2002 Elsevier Science Ltd. All rights reserved.

252 I. Stamm et al.

speculated that Stxs may modulate intestinalimmune cell/epithelial cell interactions result-ing in activated T cells, which then target thekidney.

In contrast to man, high percentages of rumi-nants shed STEC without any clinical symptoms[9±11]. Cattle, sheep, and goats are thereforeregarded as `tolerant' carriers of these bacteria[12], albeit the remarkable persistence of STECinfections in these animals correlates with theability of the bacteria to synthesize Stx [13].Owing to ®ndings that Stxs affect bovine Band T lymphocyte functions in vitro and ex vivo[14±16] we concluded that the persistence ofbovine STEC infections has its roots in a Stx-induced immunomodulation, which may provea suitable target for measures to reduce theSTEC prevalence in cattle. If the hypothesis ofHeyderman et al. [8] holds true, that T cellactivation is essential for the pathogenesis ofHUS in man, an impact of Stxs on Tcell functionswould diametrically contribute to the courseof bovine and human STEC infections, i.e.,asymptomatic persistence and development ofdisease, respectively.

A reasonable explanation may be the expres-sion of Stx-receptors by bovine but not humanT cells. The eukaryotic cell surface receptor forthe B subunit of Stxs is globotriaosylceramide(Gb3, syn. CD77) [17]. Gb3/CD77 mediatesendocytosis of its ligand gaining Stxs access tothe cytosol where the A-subunit is enzymaticallyactive [18]. In the human immune system Gb3/CD77 expression is limited to B cells while Tcellslack Stx-receptors [19]. In contrast, we recentlyshowed that Gb3/CD77 is equally synthesizedand surface expressed by bovine B and T cells[16]. However, functional studies revealed thatbovine lymphocytes' sensitivity to Stx1 is notstrictly correlated to the level of Gb3/CD77 sur-face expression: the cells are only transientlysensitive in an early phase of activation charac-terized by a low to moderate Gb3/CD77 expres-sion [20]. The broad cellular distribution of Gb3/CD77 also includes BoCD4� T cells, the prolif-eration of which is only marginally affectedby Stx1 [15]. It has to be considered that highaf®nity binding of Stxs to cells depends onseveral features of Gb3/CD77 such as lengthand saturation of the fatty acid incorporatedin different isotypes of the molecule [21].Binding of Stx to cell surface protein of Gb3/CD77-de®cient Vero cell clones has also beendescribed [22]. In order to understand an effectof Stx1 unique for the bovine immune system at

a molecular level it became considerably im-portant to prove whether Gb3/CD77 detected onbovine lymphocytes in fact represents a func-tional Stx1 receptor. This brought us to performbinding, competition, and internalization stu-dies with stimulated bovine peripheral bloodmononuclear cells (PBMC) in vitro.

(Part of this work was presented at theConference on Pathogenicity and Virulence ofVTEC, Verocytotoxigenic E. coli in Europe, EUConcerted Action Project, Liege, Belgium, 1999.)

Results

Stx1 and rStxB1 binding to bovinelymphocytes

Freshly isolated PBMC were unable to binddetectable amounts of Stx1 holotoxin or rStxB1subunit (data not shown). However, stimulationof PBMC with Phytohemagglutinin P (PHA-P;respective cells are referred to as stimulatedPBMC below) rendered a prominent proportionof cells able to bind Stx1 and rStxB1 [Fig. 1(a)].All lymphocyte subpopulations investigatedcould bind rStxB1, but the binding pattern ref-lected the subpopulations' susceptibility to theproliferation inhibitory effect of Stx1: about halfof the BoCD8� cells and two thirds of theBoCD21� cells, the main targets of Stx1 in vitro[15], bound rStxB1 in large amounts, whereasonly about one third of BoCD4� cells and onequarter of WC1� cells (i.e., the majority of gdTcells in bovine peripheral blood), respectively,did so [Fig. 1(b)].

When neutral and acidic glycosphingolipids(GSL) were separated from PBMC before (day 0)and after (day 4) stimulation and analysed byHPTLC-overlay, rStxB1 bound speci®cally to theceramide tri-hexoside bands of the standard andto two distinctive bands of the day 4 neutral GSL[Fig. 2(b)], which we previously structurallydetermined as Gal(a1-4)Gal(1-4)Glc(1-1)cera-mide (Gb3) [16]. No binding could be observedto day 0 neutral GSL ± due to the low amountof Gb3 detected at this time point [16] ± or toacidic GSL.

Double staining experiments yielded rStxB1-Gb3/CD77 double positive cells ®nally provingthat rStxB1 is able to bind bovine Gb3/CD77even in its physiological environment withina membrane (Fig. 3). However, binding ofrStxB1 was not closely linked to binding of theanti-CD77 mab 38.13. A prominent population

WC1

BoCD4 BoCD8

BoCD21

(a) (b)

binding of rStxB1

log green fluorescence

even

ts18.54 % 17.38 %

26.18 %

24.02 % 13.72 %

29.28 %

10.38 %

20.52 %

20.02 %

32.71 %

2.41 %6.22 %

Figure 1. Binding of Stx1 and rStxB1 by bovine PBMC after cultivation for 4 days in the presence of PHA-P(5 mg/ml). (a) Representative ¯ow cytometric histograms illustrating the binding of Stx1 (thin solid line;50,000 CD50/ml, quanti®ed on Vero cells) and rStxB1 (thick solid line; 30 mg/ml) to the blast cell populationof bovine PBMC. Grey shaded histogram denotes the corresponding antibody control. Histograms are fromone representative of twelve PBMC preparations. (b) PBMC subpopulation binding of FITC-labelled rStxB1(10 mg/ml). Representative two-colour dot plots are shown from one out of four PBMC preparations.Percentage of cells from the blast cell-gated population are indicated in the quadrants.

Shiga toxin 1 receptors on bovine lymphocytes 253

of anti-CD77 negative cells that bound rStxB1could be detected at the beginning of theincubation period on day 2. In turn, a populationof anti-CD77 positive cells unable to bind rStxB1appeared later on and its percentage constantlyincreased during the following days of incuba-tion (Fig. 3).

Binding of rStxB1 to Gb3/CD77 isoformsexpressed by bovine lymphocytes

In subsequent competition studies by using theanti-CD77 mab 38.13 and rStxB1 as competitorsand stimulated PBMC (day 4) as targets, con-centrations of rStxB1 above 3.75mg/ml were ableto prevent subsequent binding of anti-CD77 asconcluded from the reduction in the percentageof anti-CD77 positive cells [Fig. 4(b)]. It is remark-able that the percentage of anti-CD77 positivecells was almost identical when the cells wereincubated ®rst with either mab 38.13 or withrStxB1. Obviously, at this time a major portion ofbovine PBMC expressed an isoform of Gb3 withan af®nity for rStxB1 that was much higher than

that for mab 38.13. On the other hand, even at thehighest concentration of rStxB1 tested, anti-CD77positive cells could still be detected, pointing toa second population of cells expressing Gb3

molecules with very high af®nity for mab 38.13.As shown in Figure 4(a), the percentage of cellspositive for binding of rStxB1 decreased at con-centrations of rStxB1 lower than 7.5mg/ml. Inter-estingly, in the range from 1.875 to 3.75mg/mlthe percentage of cells positive for rStxB1-binding was in¯uenced by the order the ligandswere given to the cells. Within this range, pre-incubation of the cells with mab 38.13 preventedbinding of rStxB1 indicating a third populationof cells expressing Gb3 isoforms that bound mab38.13 and rStxB1 with almost similar af®nity.

Effect of rStxB1 on bovine lymphocytetransformation and proliferation

When PBMC were incubated in the presence ofrStxB1 (10mg/ml) for 4 days, the B subunit wasable to exert a suppressive effect on lymphocyteproliferation on its own (Fig. 5) comparable to

S Day 0 Day 4 Day 0 Day 4 S Day 0 Day 4 Day 0 Day 4 Neutral Acidic Neutral Acidic

CMH-

CDH-

CTH-

CTetH-

(a) (b)

Figure 2. rStxB1 binding to bovine PBMC gly-cosphingolipids (GSL). The total neutral and acidicGSL fractions of bovine PBMC freshly after prepara-tion (day 0) and after stimulation with PHA-P(5 mg/ml; day 4) were spotted onto HPTLC plates.The amount of PBMC GSL per lane correspondedto 2 mg (a) and 0.5 mg (b) PBMC dry weight,respectively. GSL were resolved by using therunning solvent chloroform:methanol:water, 65:25:4(by volume) and visualized chemically with orcinol/H2SO4-staining (a) or by HPTLC-overlay withrStxB1 (b). The standard (S) of CMH-CTetHcorresponded to globo-series ceramide mono-, di-,tri- and tetrahexosides, respectively.

Figure 3. Gb3/CD77 expression and rStxB1 binding bstimulated with PHA-P (5 mg/ml) for 2±6 days. At the endCD77 and FITC-labelled rStxB1 (10 mg/ml) as depicted ithe blast cell-gated population are indicated in the quresults are shown from one out of four PBMC preparati

254 I. Stamm et al.

that of the holotoxin in the nanogram range[15, 20]. Addition of rStxB1 to stimulated PBMCfrom the beginning of the cultivation reducedthe percentage of viable BoCD8� and BoCD21�

blast cells remaining on day 4 and also signi®-cantly reduced the percentage of Gb3/CD77�

blast cells. The latter effect could even be detectedat rStxB1 concentrations as low as 20 ng/ml[Fig. 6(a)]. To exclude that this effect was simplydue to capping of Gb3/CD77 molecules theconcentration threshold for the detection ofrStxB1-binding to stimulated PBMC (day 4)was determined. In fact, rStxB1-binding wasundetectable below 937.5 ng/ml [Fig. 6(b)] clear-ly arguing against a capping phenomenon at20 ng/ml.

Internalization of Stx1 and rStxB1 bybovine lymphocytes

Internalization studies proved that a minority ofthe Gb3/CD77 isoforms of bovine PBMC waseven able to mediate uptake of the toxin. Whenstimulated PBMC (day 4) were loaded withtoxin at 4�C and incubated for further 30 min at37�C surface ¯uorescence was decreased com-pared to cells that were kept at 4�C throughout.

y bovine PBMC. Flow cytometry analysis of PBMCof the incubation period cells were labelled with anti-

n the lower part of the ®gure. Percentage of cells fromadrants. Representative two-colour ¯ow cytometry

ons.

Figure 4. Competition of rStxB1 and anti-CD77 for binding to bovine PBMC. Graphs represent data obtainedby ¯ow cytometry analysis of PBMC stimulated with PHA-P (5 mg/ml) for 4 days. At the end of the incubationperiod cells were labelled with anti-CD77 (15 mg/ml) and rStxB1 as depicted in the lower part of the ®gure.Graphs represent mean, max and min of the data obtained from blast cells of four PBMC preparations.


In this course mean ¯uorescence intensities were0.732+0.12 and 0.719+0.0817 (P4 0.05) for cellslabelled with Stx1 holotoxin and 3.605+0.62and 2.465+0.571 (P� 0.01) for cells labelledwith rStxB1 after incubation at 4�C and 37�C,

respectively (mean+SD of 4 PBMC prepara-tions). Incubation at 37�C also slightly reducedthe percentage of Stx1 and rStxB1-binding cells,but differences missed statistical signi®cance(Fig. 7, bar A, B; data not shown for Stx1).

Figure 5. Effect of rStxB1 on transformation andproliferation of bovine PBMC. PBMC were stimu-lated with PHA-P (5 mg/ml) for 4 days in thepresence (black bars) or absence (grey bars) of rStxB1(10 mg/ml) and analysed by ¯ow cytometry at the endof the incubation period as depicted in the lowerpart of the ®gure. Lymphocyte subpopulations wereidenti®ed by immunophenotyping. Data analysis wasperformed by calculating the percentage of viableblast cells. Bars represent the mean+SD of the dataobtained from four PBMC preparations. Comparisonof cells incubated with and without rStxB1 wasperformed by the paired t-test. Horizontal bracketswith centered asterisks enclose bars that aresigni®cantly different (P5 0.05).

256 I. Stamm et al.

Effect of anti-Stx on binding andinternalization of rStxB1

High proportions of cattle harbour Stx1-neutralizing antibodies in their sera [23]. Indeed,19 out of 20 plasma samples from blood donorsincluded in the present and previous studies [20]contained Stx1-neutralizing antibodies with atleast 1 neutralizing unit/ml (data not shown).Non-parametric correlation analysis revealedthat there is no signi®cant correlation betweenantibody titers and the level of the suppressiveeffect of Stx1 on the number of BoCD8�,BoCD21�, and Gb3/CD77� blast cells in PBMCcultures obtained from these animals (Spearmanrank correlation coef®cients were 0.145, 0.073,and ÿ0.049, respectively; critical value� 0.447at a two-sided signi®cance level of 5%).

Because PBMC are separated from other bloodcomponents including immunoglobulins, these®ndings stress that the sensitivity of lympho-cytes towards the effect of Stx1 is an inherentfeature of cattle independent of a certain animals'immune status.

However, possible implications of the host'simmune response became apparent by the dicho-tomous effect of anti-StxB1 mab 13C4 [24] usedas a model to study the role of Stx1-speci®cserum immunoglobulins. Preincubation of up to50,000 CD50/ml Stx1 holotoxin with anti-StxB1(45mg/ml) ± shown to neutralize the effect ofStx1 on bovine lymphocytes [15, 20] ± abrogatedbinding of the holotoxin (data not shown). Incontrast, preincubation of rStxB1 with anti-StxB1(1.5 mg/ml) only partially neutralized the effectof rStxB1 on the percentage of Gb3/CD77� cellsremaining in PBMC cultures after 4 days[Fig.6(a)].Moreover,anti-StxB1evensigni®cantly(P� 0.01) enhanced the percentage of stimulatedPBMC (day 4) able to bind rStxB1 and rStxB1/anti-StxB1-complexes at least at intermediateconcentration ratios [Fig. 6(b)]. The idea thatrStxB1/anti-StxB1-complexes retain their bind-ing capability to cells was ®nally con®rmed,because such complexes could still be detectedon a prominent number of PBMC after preincu-bation of 30mg/ml rStxB1 with 45mg/ml of anti-StxB1 (Fig. 7, bar C). Moreover, these complexescould even be internalized at 37�C (Fig. 7, bar D).Controls included in every set of experiments, i.e.cells that were only incubated with anti-StxB1and anti-mouse conjugate, always gave negativeresults proving that even the rStxB1/anti-StxB1-complexes bound via Gb3/CD77 and not via Fcreceptors. Anti-StxB1 completely blocked inter-nalization in this model only when given to thecells after they had already bound rStxB1 (Fig. 7,bar E) or Stx1 holotoxin (data not shown).

Discussion

Most STEC infections in ruminants are charac-terized by the lack of clinical symptoms, but thepersistence of enteral infection with prolongedfecal shedding [13] is of critical importance forhuman food safety. We are currently investigat-ing the hypothesis that a Stx1-induced immuno-modulation fundamentally contributes to thepersistence of bovine STEC infection. Con®rm-ing our previous assumption that Stx1 directlyacts on bovine lymphocytes [20] we show here

Figure 6. Effect of rStxB1 on Gb3/CD77 expression by bovine PBMC (a) and determination of rStxB1-bindingas a function of rStxB1 concentration (b). Flow cytometry analysis of PBMC stimulated with PHA-P (5 mg/ml)for 4 days. PBMC were incubated in the presence of rStxB1 (10 mg/ml) and the presence and absence of anti-StxB1 (1.5 mg/ml) as depicted in the lower part of the ®gure. Data analysis was performed by calculating thepercentage of viable blast cells. Graphs represent the mean+SD (a) and mean, maximum and minimum (b) ofthe data obtained from four PBMC preparations each. Two-way ANOVA revealed signi®cances for the curvespresented in graph B (P� 0.01 for anti-StxB1 and P� 0.001 for the concentration of rStxB1).


that these cells are in fact capable of bindingStx1 holotoxin as well as recombinant B sub-unit (rStxB1). Competition and HPTLC-bindingstudies demonstrated that the binding site

for Stx1 is a glycosphingolipid which we recent-ly biochemically characterized as Gb3, thusrepresenting the bovine homologue of thehuman Stx-receptor CD77 [16]. Thereby, cattle

Figure 7. Binding and internalization of rStxB1 and rStxB1/anti-StxB1-complexes by bovine PBMC. Flowcytometry analysis of PBMC stimulated with PHA-P (5 mg/ml) for 4 days. At the end of the incubation periodcells were labelled with rStxB1 (30 mg/ml) after preincubation with (C, D) or without (A, B) anti-StxB1 (45 mg/ml), and kept either on ice throughout or warmed to 37�C for 30 min as depicted in the left-handed part of the®gure. In addition, cells were labelled with rStxB1 followed by incubation with anti-StxB1 prior to warming to37�C (E). Bars represent the mean+SD of four PBMC preparations. Data were analysed by the paired t-test.Horizontal brackets with centered asterisks enclose bars that are signi®cantly different [P� 0.05 (*), P� 0.01(**), P� 0.001 (***); n.s.�not signi®cantly different (P4 0.05)].

258 I. Stamm et al.

is the ®rst species reported to date that expressesfunctional Stx-receptors on the surface of T cells.Because bovine intraepithelial lymphocytes, the®rst immune cells which gain contact to the toxin,predominantly are T cells and equally expressGb3/CD77 in vivo [25], the direct targeting ofT cells by Stx1 must be of particular importancefor the colonization of the bovine intestineby STEC.

Bovine lymphocytes' CD77-expression was notcongruenttorStxB1-bindingbythecells,althoughrStxB1 and anti-CD77 both speci®cally recognizethe terminal galabiose in Gb3/CD77 [17, 26].Competition studies revealed three subtypes ofcells expressing receptors with varying af®nitiesfor the two ligands. The fact that rStxB1-bindingcells mainly appeared at the beginning of theincubation period suggests that only at an earlytime point the cells exhibit a reasonable sensi-tivity against Stx1. This idea is supported by ourprevious ®ndings that Gb3/CD77 expression bybovine lymphocytes parallels the activation ofthe cells [16]. Nevertheless, Stx1 only affects cells

between the Gb3/CD77low and the Gb3/CD77moderate state of activation while cells thatreached the Gb3/CD77high state before experienc-ing Stx1 become refractory again [20]. Com-parative fatty acid analysis of Gb3/CD77obtained from bovine PBMC prior to (day 0)and after (day 4) stimulation had shown thatstimulation altered the fatty acid pattern withhexadecanoic acid dominating on day 0 and fattyacids with more than 20 carbon atoms prevailingon day 4 [16]. Pellizzari et al. [21] found Gb3/CD77 molecules in human kidneys which con-tained hexadecanoic acid as the major compon-ent having a consistently higher af®nity for Stx1than other ceramide trihexosides. Given thatbovine lymphocytes express high af®nity Stx1-receptors only during early states of activation,STEC by secreting Stx1 prevent the onset of animmune response rather than down regulate anestablished one. This may be of particularbene®t for the bacteria because low amounts ofStx are suf®cient to target small numbers ofsensitive cells.

Figure 8. Proposed model for the pathwaysinvolved in the suppressive effect of Stx1 on bovinelymphocytes. For details see text.


The biochemical target of the Stx1 holotoxinis the ribosome [2], and internalization fromcoated pits after clustering of randomly distri-buted Stx binding sites [18] is a prerequisite forthe toxin to exert its toxicity [27]. To prove theability of bovine lymphocytes to internalize Stxwe applied a method previously describedfor Stx1 [18, 19] and quanti®ed the amount ofimmunologically detectable Stx1 and rStxB1 onthe cellular surface after incubation of the cells at37�C by ¯ow cytometry. Our ®ndings that cellslabelled by anti-StxB1 mab after warming to37�C exhibited a lower ¯uorescence intensitycompared to cells kept at 4�C strongly suggestthe existence of a toxin translocation mechanismlinked to surface glycolipids. However, only aminor portion of surface-bound Stx was inter-nalized by bovine PBMC. Moreover, rStxB1 canbe detected on the cellular surface throughoutwhen the cells are incubated at 37�C in thepresence of rStxB1 for days (data not shown). Asuitable explanation for that is the existence oftwo populations of receptors as predicted byLindberg et al. [17] and meanwhile con®rmed bystudies on the aglycone modulation of humanGb3/CD77 receptor function [28]: one isoform ofbovine Gb3/CD77 binds toxin only and anotherisoform, possibly present in smaller numbers,mediates toxin uptake.

Interestingly, the isolated B subunit on its ownexhibited an effect similar to that of Stx1 inbovine PBMC cultures. Although the B subunithas no inhibitory effect on protein synthesisitself, a series of recent studies indicates thatit interferes with cell signalling [29±32]. Never-theless, the fact that anti-StxB1 mab in someinstances is able to facilitate rStxB1-binding tobovine lymphocytes but fails to equally enhancethe biological effect of rStxB1 and Stx1 standsagainst the hypothesis of a signal solely derivedfrom binding of the protein. Because eventhe Stx1 B subunit can exert cytotoxicity whenexpressed inside transfected cells [33] it may bethat internalization into bovine lymphocytes isadditionally required. With respect to the courseof events during bovine STEC infections it mustbe stressed, however, that a 50,000-fold higherconcentration of the B subunit was requiredcompared to incubation with the holotoxin(10 mg/ml vs. 200 pg/ml of Stx1 holotoxin) [15],and toxin concentrations in the microgram rangeare unlikely to occur in situ. In other in vitrosystems inhibition of protein synthesis byStx1 holotoxin sensitizes cells for a separatecell surface-derived signal induced by minute

concentrations of the B subunit that then leads toapoptosis [32, 34]. Apoptosis does not occur inStx1-treated bovine PBMC cultures [15], butbefore inhibiting protein synthesis Stx1 ®rstdamages 28S rRNA and induces a `ribotoxicstress response' resulting in an induction ofmRNAs of primary response genes [35]. We thusassume that signals originating from the surfaceafter Gb3/CD77-binding of the B subunit in vivo(or the B pentamer of the holotoxin) act in concertwith the induction of primary response genesafter internalization of the Stx1 holotoxin andthereby contribute to the suppressive effect Stx1exerts on bovine immune cells (Fig. 8).

Active or passive immunization against Stxs,prior to experimental inoculation with STEC,prevents the systemic complications of STECinfection in animal models [36±38] and Stx-speci®c antibodies prevent humans fromdeveloping HUS [39, 40]. However, it is still amatter of discussion in literature whether theprotective effect of these antibodies is based onthe capability to prevent binding or to prevent

260 I. Stamm et al.

internalization of the toxins. Nakao et al. [41]showed that a neutralizing monoclonal antibodyto Stx2 blocks receptor binding and Lindberget al. [17] as well as Eiklid and Olsnes [42]observed that cells cannot be rescued by add-ition of antiserum after toxin binding. In contrast,Sandvig et al. [18] were able to partially protectcells by addition of antiserum even 15 min aftertoxin binding. A similar instance was reportedfor human microvascular endothelial cells [43].In our study an anti-StxB1 mab, used as a modelfor bovine serum antibodies speci®c for Stx1 [23],prevented rStxB1 binding at high antibody-antigen ratios (data not shown) and also effect-ively blocked internalization of cell-boundrStxB1 under certain conditions (see Fig. 7,bar E). High titers of Stx-speci®c antibodies asreported for cattle should therefore be highlyprotective [23, 44, 45]. There is some circumstan-tial evidence, however, that these antibodies'effects are not limited to neutralization but alsomay in¯uence the way the toxin binds to cells.While at least a prominent portion of rStxB1/anti-StxB1-complexes formed in solution duringpreincubation retained their ability to bind toand translocate into bovine lymphocytes (seeFig. 7, bars C and D), rStxB1 molecules that wererecognized by anti-StxB1 after cell surface bind-ing did not internalize during 30 min at 37�C(see Fig. 7, bar E). Because Gb3/CD77 exists indifferent isoforms and a single Stx1 B subunitharbours three independent Gb3/CD77 bindingsites [46] these dichotomous effects of anti-StxB1may result from the usage of an alternative bind-ing site of rStxB1 or the usage of a differentreceptor isoform or both. Anti-StxB1 inducedalterations in Stx1/receptor interactions mayconsequently modify the cellular effects of thetoxin because the three receptor binding sites ofStx1 are linked to different biological activities[47]. Furthermore, the ability of anti-StxB1 atsubinhibitory concentrations to promote rStxB1binding to bovine lymphocytes increases mask-ing of surface Gb3/CD77 molecules. This willprofoundly in¯uence cellular interactionsbetween immune cells as well as interactions ofimmune cells with other tissue cells in situ [48].As a consequence ± although speci®c antibodiesare able to neutralize Stx1 in principle ± theiractivity fundamentally depends on the antibody-antigen ratios within the tissues and the timepoint (before or after receptor binding) the anti-bodies bind to the toxin. Hence, the detection ofStx-speci®c antibodies in sera of cows must notnecessarily result in an ef®cient protection

against immunomodulatory effects of Stx1. Infact, serological responses in cattle do not corre-late with elimination of STEC infection [45] orprotection against reinfection [14, 45].

Bovine intestinal epithelial cells express recep-tors for Stx1 [49; Menge, unpublished data] butthe same holds true for some human intestinalcell lines [50] which are able to translocatesigni®cant amounts of Stx1 [3]. Because bovineendothelial cells also resemble their humancounterpart in their susceptibility to Stx1 [51],expression of functional Stx1-receptors by bovinebut not human T cells is the only fundamentalspecies difference discovered to date andthus probably represents the missing link toexplain the differential outcome of STECinfections ± persistent infection or disease ± inthese species. Detailed investigations of T cell/epithelial cell interactions in the presence andabsence of Stx1 are currently under way in ourlaboratory and will help to understand theunderlying mechanisms.

Materials and Methods

Toxin puri®cation

Stx1 was produced from the bovine STEC1strain 2403 (rough, Hÿ) [10] and puri®ed by aprocedure that was described previously [15]. Atthe end of the puri®cation process, toxin pre-parations were passed through Detoxi-GelTM

columns (Pierce, Old-Beijerland, Holland) toreduce contaminations with endotoxin. TheStx1 preparation contained 50,000 CD50 of Stx1(see below) and 0.85 ng of Endotoxin per ml asdetermined by the Limulus amoebocyte lysateassay.

Puri®cation of B subunit protein

Recombinant StxB1 (rStxB1) was puri®ed fromE. coli DH5a [pSU108] [52] by the method ofNiebuhr [53] with slight modi®cations. Bacteriawere grown at 30�C overnight in Luria brothsupplemented with ampicillin (50mg/ml),diluted to an optical density at 560 nm (OD560)of 0.4 and incubated further until the OD560

reached 0.8. The culture was induced at 42�C for4 h, and the cells were harvested by centrifuga-tion (82006g for 15 min). The pellet was washed


twice with 10 mM Tris±HCl (pH 8.0), resus-pended in a solution containing 25% sucrose(wt/vol), 1 mM Na2EDTA, and 10 mM Tris±HCl(pH 8.0), and gently shaken at 30�C for 10 min.The cells were harvested and immediatelyresuspended in ice-cold distilled water (osmoticshock treatment) and shaken gently at 4�C for10 min. After centrifugation, the supernatantfraction (periplasmic fraction) was collectedand diluted in the same volume of 20 mM Tris-HCl (pH 7.6), followed by another centrifugationstep (100,0006g, 1.5 h) immediately before fur-ther puri®cation.

Puri®cation of B subunit was performed byuse of a fast protein liquid chromatography sys-tem (FPLC1; Amersham Pharmacia, Freiburg,Germany). The periplasmic fraction was appliedto a MonoQ HR 5/5 column (AmershamPharmacia) equilibrated with 20 mM Tris±HCl(pH 7.6). Elution was carried out with a gradientfrom 0 to 1 M NaCl in the same buffer. The Bsubunit was eluted at a salt concentration of50 to 90 mM. Afterwards, the preparationwas supplemented with aprotinin (50 KIE/ml)(Trasysol1, Bayer, Leverkusen, Germany) and5% glycerol (v/v), and passed through a Detoxi-GelTM column (Pierce) to reduce endotoxincontaminants. Further analysis of the B subunitpreparation was carried out by subsequent SDS-PAGE and Western blot techniques. Proteinconcentration was determined with BCA proteinassay

1 (Pierce, Old Beijerland, The Netherlands)according to the instructions of the manufactur-er. The rStxB1 preparation used in the presentstudy contained 330mg/ml of rStxB1 and 0.92 ngEndotoxin per ml.

Cytotoxicity assay

The cytotoxic activities of toxin preparationswere determined on Vero cells (ATCC CRL 1587)by the method of Gentry and Dalrymple [54]with minor modi®cations [15]. Cellular metabolicactivity was assessed by MTT reduction assayas described previously [15, 55]. CD50 was cal-culated from dose response curves geometri-cally as the reciprocal of the toxin dilutioncausing 50% reduction in cellular metabolicactivity. In studies determining the neutralizingactivity of plasma samples obtained from theblood donors, medium was supplemented withplasma that had been collected from the upperphase of the Ficoll preparation and heated (56�C,

30 min). Neutralizing activity of plasma sampleswas calculated as: 6neutralizing units/ml�reduction of CD50/ml by 10x [23].

Cell preparation and stimulation

Blood samples were taken from healthy cows(Holstein6german black pied) from the dairyherd of the farm of the Justus-Liebig-University.Samples were diluted 1:1 with Ca2�-Mg2�-freePBS and layered onto Ficoll-Paque1 (AmershamPharmacia) as described by Bùyum [56]. Aftercentrifugation (8006g, 20�C, 45 min), cells wererecovered from the Ficoll buffer interface. Con-taminating erythrocytes were removed by incu-bating the cell suspension with lysis buffer(8.26 g NH4Cl, 1.09 g NaHCO3, 0.037 g Na3EDTAad 1000 ml A. dest.) at RT for 5 min. The cellswere washed twice with PBS and resuspendedat 56106 cells/ml in modi®ed cell culturemedium (RPMI 1640 supplemented with10% fetal calf serum, 3 mM 2-mercaptoethanol,and Phytohemagglutinin P (PHA-P; Sigma,Taufkirchen, Germany) at a ®nal concentrationof 5 mg/ml). The cell suspension was subse-quently added to 96-well ¯at-bottomed micro-titer plates (Nunc, Wiesbaden, Germany) with50ml per well. The plates had been preparedwith 50ml per well of dilution series oftoxin preparations generated with 0.15 M NaClin triplicate plus 50 ml of cell culture medium. Inneutralization studies, medium was addition-ally supplemented with puri®ed anti-StxB1(mAb 13C4) [24]. Plates were incubated at 37�Cunder 5% CO2 environment.

Immunophenotyping and ¯owcytometry analysis

At the end of the cultivation period, cellswere thoroughly resuspended and transferredto V-shaped microtiter plates (Greiner,Frickenhausen, Germany) for immunolabellingas described previously [55, 57]. In short, thecells were centrifuged (1506g, 4�C, 7 min) andresuspended in 50ml of cell culture medium as anegative control or with supernatant of hybri-doma cell lines (IL-A11 for BoCD4, IL-A105for BoCD8, IL-A65 for BoCD21 and IL-A29 forWC1). Alternatively, the cells were resuspendedwith 25 ml of rat IgM (1 mg/ml, 1:50 in PBS;

262 I. Stamm et al.

Camon, Wiesbaden, Germany) as a negativecontrol or with anti-human CD77 antibody(1:10 in PBS; Beckman-Coulter, Krefeld,Germany). Upon incubation (20 min, 4�C), thecells were washed once and resuspended witheither 50 ml of anti-mouse IgG PE-conjugate(Sigma) diluted 1:100 in PBS or anti-rat IgMFITC- or PE-conjugate (Dianova, Hamburg,Germany) 1:200 diluted in PBS containing2 mg/ml propidium iodide (Sigma). Followinganother 20 min on ice, the cells were washedtwice and analysed with an EPICS ELITE1

Analyser (Beckman±Coulter). Five thousandevents were aquired from each sample. Dataanalysis was performed by using the ELITE 4.01software provided by the manufacturer. Elec-tronic gates were set according to the negativecontrol included in each test series de®ning lessthan 2% of the cells as positive. Populations ofenlarged lymphoblast cells were de®ned accord-ing to its light scatter characteristics as described[16, 55] and analysed separately.

Binding and internalization studies

Cells were transferred to V-shaped microtiterplates and incubated with 50ml of Stx1(50,000 CD50/ml) or rStxB1 (either 30mg/ml orvarying concentrations as indicated) for 30 minon ice [55]. In the case of neutralization studies,Stx1 and rStxB1, respectively, had been preincu-bated with anti-StxB1 for 1 h. After incubationcells were washed once. [In some studiesFITC-conjugated rStxB1 (kindly provided byC. A. Lingwood, The Hospital for Sick Children,Department of Microbiology, Toronto, Canada)was used (10 mg/ml). The following stepswere omitted in this case.] For internalizationstudies the cells were next incubated at 37�C for30 min to allow uptake of the bound proteins.Afterwards the cells were washed once andresuspended in 50 ml of anti-StxB1 (45 mg/ml),followed by another incubation on ice for 30 min.Cellswerewashedagainandresuspendedin50mlof anti-mouse IgG FITC-conjugate (Dianova)diluted 1:400 in PBS and containing 2 mg/mlpropidium iodide (Sigma). Following another30 min on ice, the cells were washed twice andanalysed by ¯ow cytometry analysis as describedabove. In some instances cells were additionallyimmunolabelled as described above prior to thelast incubation step.

Isolation and puri®cation of neutralglycolipids

PBMC of day 0 (the day of the preparation of thecells) and day 4 (500 million cells each) werewashed several times with PBS and lyophilizedyielding dry-weights of 21 (day 0) and 41 mg(day 4). Acidic fraction and neutral fractionglycolipidswereisolatedandanalysedbyHPTLCand orcinol/H2SO4-staining as described else-where [16, 58].

Binding of rStxB1 to glycolipids onHPTLC plate

After glycolipid separation by HPTLC, the air-dried plates were dipped three times in a hexanesolution of 0.5% Plexigum (Aldrich, Steinheim,Germany) for 60 s and then dried under warmair. The plates were blocked by incubation with2% bovine serum albumin (w/v) in PBS supple-mented with 0.5% Tween 201 (v/v) for 1 h at RT.After washing, the plates were overlayed for afurther 2 h with rStxB1 (5mg/ml) diluted in thesame buffer. The plates were washed six timesand incubated with anti-StxB1 (3mg/ml) for 1 h.After washing six times the plates were over-layed with anti-mouse Ig conjugated to alkalinephosphatase (Dako, Hamburg, Germany) diluted1:200 in buffer for 1 h. The plates were washedagain three times, and bound antibody wasvisualized with the 5-bromo-4-chloro-3-indolyl-phosphate and nitroblue tetrazoliumchloridesubstrate system (Biomol, Hamburg, Germany).

Statistical analysis

Data were analysed statistically by the pairedt-test, the Wilcoxon signed rank test, two-wayanalysis of variance (ANOVA) and the Spearmanrank order correlation by using SigmaStat2.0 software (1992; SPSS Inc., Chicago, IL, USA).Signi®cant differences were separated at P�0.001 (***), P� 0.01 (**), and P� 0.05 (*).

Acknowledgements

Kirsten Niebuhr (Gesellschaft fuÈ r BiotechnologischeForschung Braunschweig) is acknowledged for thekind donation of E. coli-K12-DH5a [pSU108], CliffordLingwood for supplying FITC-labelled rStxB1. We


further thank J. Naessens at ILRI, Nairobi, Kenya forgenerously supplying hybridoma cell lines producingantibodies to bovine leukocyte antigens. This workwas supported by the German Research Council(Sonderforschungsbereich 535, projects A11 and Z1).

References

1 Karmali MA. Infection by Verocytotoxin-producingEscherichia coli. Clin Microbiol Rev 1989; 2: 15±38.

2 Endo Y, Tsurugi K, Yutsudo T, Takeda Y, Ogasawara T,Igarashi K. Site of action of a Vero toxin (VT2) fromEscherichia coli O157:H7 and of Shiga toxin on eukaryoticribosomes. RNA N-glycosidase activity of the toxins.Eur J Biochem 1988; 171: 45±50.

3 Acheson DWK, Moore R, De Breucker S, Lincicome L,Jacewicz M, Skutelsky E, Keusch GT. Translocation ofShiga toxin across polarized intestinal cells in tissueculture. Infect Immun 1996; 64: 3294±300.

4 van de Kar NC, Monnens LAH, Karmali MA, vanHinsbergh VWM. Tumor Necrosis Factor and Inter-leukin-1 induce expression of the Verocytotoxinreceptor globotriaosylceramide on human endothelialcells: implications for the pathogenesis of the hemolyticuremic syndrome. Blood 1992; 80: 2755±64.

5 van Setten PA, Monnens LA, Verstraten RG, van denHeuvel LP, van Hinsbergh VW. Effects of verocyto-toxin-1 on nonadherent human monocytes: bindingcharacteristics, protein synthesis, and induction of cyto-kine release. Blood 1996; 88: 174±83.

6 te Loo DM, Monnens LA, van Der Velden TJ,Vermeer MA, Preyers F, Demacker PN, van DenHeuvel LP, van Hinsbergh VW. Binding and transferof verocytotoxin by polymorphonuclear leukocytes inhemolytic uremic syndrome. Blood 2000; 95: 3396±402.

7 O'Loughlin EV, Robins-Browne RM. Effect of Shigatoxin and Shiga-like toxins on eukaryotic cells. MicrobesInfect 2001; 3: 493±507.

8 Heyderman RS, Soriani M, Hirst TR. Is immune cellactivation the missing link in the pathogenesis ofpost-diarrhoeal HUS? Trends Microbiol 2001; 9: 262±6.

9 Heuvelink AE, van den Biggelaar FLAM, deBoer E, Herbes RG, Melchers WJ, Huis in 't Veld JH,Monnens LA. Isolation and characterization of Vero-cytotoxin-producing Escherichia coli O157 strains fromDutch cattle and sheep. J Clin Microbiol 1998; 36: 878±82.

10 Wieler LH, Bauerfeind R, Baljer G. Characterization ofShiga-like toxin producing Escherichia coli (SLTEC) iso-lated from calves with and without diarrhoea. Int J MedMicrobiol Virol Parasitol Infect Dis 1992; 276: 243±53.

11 Wieler LH, Schwanitz A, Vieler E, Busse B, Steinruck H,Kaper JB, Baljer G. Virulence properties of Shiga toxin-producing Escherichia coli (STEC) strains of serogroupO118, a major group of STEC pathogens in calves.J Clin Microbiol 1998; 36: 1604±7

12 Pruimboom-Brees IM, Morgan TW, Ackermann MR,Nystrom ED, Samuel JE, Cornick NA, Moon HW. Cattlelack vascular receptors for Escherichia coli O157:H7 Shigatoxins. Proc Natl Acad Sci USA 2000; 97: 10325±9.

13 Cornick NA, Booher SL, Casey TA, Moon HW. Persis-tent colonization of sheep by Escherichia coli O157:H7

and other E. coli pathotypes. Appl Environ Microbiol2000; 66: 4926±34.

14 Hoffman M, Casey T, Bosworth B. Bovine immuneresponse to Escherichia coli O157, abstr. In: Abstractsof the 3rd International Symposium and Workshop onShiga Toxin (Verocytotoxin)-producing Escherichia coliinfections 1997: V67/VIII, p. 117.

15 Menge C, Wieler LH, Schlapp T, Baljer G. Shiga toxin 1from Escherichia coli blocks activation and proliferationof bovine lymphocyte subpopulations in vitro. InfectImmun 1999; 67: 2209±17.

16 Menge C, Stamm I, Wuhrer M, Geyer R, Wieler LH,Baljer G. Globotriaosylceramide (Gb3/CD77) is synthe-sized and surface expressed by bovine lymphocytesupon activation in vitro. Vet Immunol Immunopathol2001; 83: 19±36.

17 Lindberg AA, Brown JE, Stromberg N, Westling-Ryd M,Schultz JE, Karlsson KA. Identi®cation of the carbo-hydrate receptor for Shiga toxin produced by Shigelladysenteriae type 1. J Biol Chem 1987; 262: 1779±85.

18 Sandvig K, Olsnes S, Brown JE, Petersen OW, vanDeurs B. Endocytosis from coated pits of Shiga toxin:a glycolipid-binding protein from Shigella dysenteriae 1.J Cell Biol 1989; 108: 1331±43.

19 Cohen A, Madrid-Marina V, Estrov Z, Freedman MH,Lingwood CA, Dosch HM. Expression of glycolipidreceptors to Shiga-like Toxin on human B lymphocytes:a mechanism for the failure of long-lived antibodyresponse to dysenteric disease. Int Immunol 1990; 2: 1±8.

20 Menge C,Stamm I,Wieler LH,Baljer G. Shiga toxin 1 fromEscherichia coli affects Gb3/CD77� bovine lymphocytesindependent of Interleukin-2, Tumor Necrosis Factor-alpha, and Interferon-alpha. Toxicology 2002; in press.

21 Pellizzari A, Pang H, Lingwood CA. Binding of vero-cytotoxin 1 to its receptor is in¯uenced by differencesin receptor fatty acid content. Biochemistry 1992; 31:1363±70.

22 Devenish J, Gyles C, LaMarre J. Binding of Escherichiacoli verotoxins to cell surface protein on wild-typeand globotriaosylceramide-de®cient Vero cells. Can JMicrobiol 1998, 44: 28±34.

23 Pirro F, Wieler LH, Failing K, Bauerfeind R, Baljer G.Neutralizing antibodies against Shiga-like toxins fromEscherichia coli in colostra and sera of cattle. Vet Microbiol1995; 43: 131±41.

24 Strockbine NA, Marques LRM, Holmes RK, O'Brien AD.Characterization of monoclonal antibodies againstShiga-Like toxin from Escherichia coli. Infect Immun1985; 50: 695±700.

25 Blessenohl M, Menge C, Baljer G. Bovine ileal intra-epithelial lymphocytes represent target cells for Shigatoxin 1. In: Abstracts of the 4th International Symposiumand Workshop on Shiga Toxin (Verocytotoxin)-producing Escherichia coli infections 2000: abstr. 304p. 131.

26 Wiels J, Fellous M, Tursz T. Monoclonal antibody againsta Burkitt lymphoma-associated antigen. Proc Natl AcadSci USA 1981; 78: 6485±88.

27 Jacewicz M, Keusch GT. Pathogenesis of Shigella diar-rhea. VIII. Evidence for a translocation step in the cyto-toxic action of Shiga toxin. J Infect Dis 1983; 148: 844±54.

28 Lingwood CA.Aglyconemodulation ofglycolipid recep-tor function. Glycoconj J 1996; 13: 495±503.

29 Arab S, Lingwood CA. Intracellular targeting of theendoplasmic reticulum/nuclear envelope by retro-grade transport may determine cell hypersensitivity

264 I. Stamm et al.

to verotoxin via globotriaosyl ceramide fatty acid iso-form traf®c. J Cell Physiol 1998; 177: 646±60.

30 Jones NL, Islur A, Haq R, Mascarenhas M,Karmali MA, Perdue MH, Zanke BW, Sherman PM.Escherichia coli Shiga toxins induce apoptosis inepithelial cells that is regulated by the Bcl-2family. Am J Physiol Gastrointest Liver Physiol 2000;278: G811±9.

31 Katagiri YU, Mori T, Nakajima H, Katagiri C, Taguchi T,Takeda T, Kiyokawa N, Fujimoto J. Activation of Srcfamily kinase yes induced by Shiga toxin binding toglobotriaosyl ceramide (Gb3/CD77) in low density,detergent-insoluble microdomains. J Biol Chem 1999;274: 35278±82.

32 Mangeney M, Lingwood CA, Taga S, Caillou B, Tursz T,Wiels J. Apoptosis induced in Burkitt's lymphoma cellsvia Gb3/CD77, a glycolipid antigen. Cancer Res 1993; 53:5314±9.

33 Nakagawa I, Nakata M, Kawabata S, Hamada S.Regulated expression of the Shiga toxin B gene inducesapoptosis in mammalian ®broblastic cells. Mol Microbiol1999; 33: 1190±9.

34 Katagiri YU, Kiyokawa N, Fujimoto J. The effect of Shigatoxin binding to globotriaosylceramide in rafts ofhuman kidney cells and Burkitt's lymphoma Ramoscells. Trends Glycosci Glycotech 2001; 13: 281±290.

35 Thorpe CM, Hurley BP, Lincicome LL, Jacewicz MS,Keusch GT, Acheson DW. Shiga toxins stimulate secre-tion of interleukin-8 from intestinal epithelial cells. InfectImmun 1999; 67: 5985±93.

36 Donohue-Rolfe A, Kondova I, Mukherjee J, Chios K,Hutto D, Tzipori S. Antibody-based protection of gnoto-biotic piglets infected with Escherichia coli O157:H7against systemic complications associated with Shigatoxin 2. Infect Immun 1999; 67: 3645±8.

37 MacLeod DL, Gyles CL. Immunization of pigs with apuri®ed Shiga-like toxin II variant toxoid. Vet Microbiol1991; 29: 309±18.

38 Wadolkowski EA, Sung LM, Burris JA, Samuel JE,O'Brien AD. Acute renal tubular necrosis and deathof mice orally infected with Escherichia coli strains thatproduce Shiga-like toxin type II. Infect Immun 1990; 58:3959±65.

39 Karmali MA, Petric M, Winkler M, Bielaszewska M,Brunton J, van de Kar N, Morooka T, Nair GB, Richard-son SE, Arbus GS. Enzyme-linked immunosorbentassay for detection of immunoglobulin G antibodiesto Escherichia coli Vero cytotoxin 1. J Clin Microbiol1994; 32: 1457±63.

40 Levine MM, McEwen J, Losonsky G, Reymann M,Harari I, Brown JE, Taylor DN, Donohue-Rolfe A,Cohen D, Bennish M. Antibodies to shiga holotoxinand to two synthetic peptides of the B subunit in seraof patients with Shigella dysenteriae 1 dysentery. J ClinMicrobiol 1992; 30: 1636±41.

41 Nakao H, Kiyokawa N, Fujimoto J, Yamasaki S, TakedaT. Monoclonal antibody to Shiga toxin 2 which blocksreceptor binding and neutralizes cytotoxicity. InfectImmun 1999; 67: 5717±22.

42 Eiklid K, Olsnes S. Interaction of Shigella shigae cytotoxinwith receptors on sensitive and insensitive cells. J RecRes 1980; 1: 199±213

43 Jacewicz MS, Acheson DW, Binion DG, West GA,Lincicome LL, Fiocchi C, Keusch GT. Responses of

human intestinal microvascular endothelial cells toShiga toxins 1 and 2 and pathogenesis of hemorrhagiccolitis. Infect Immun 1999; 67: 1439±44.

44 Borman-Eby HC, McEwen SA, Clarke RC, McNab WB,Rahn K, Valdivieso-Garcia A. The seroprevalence ofverocytotoxin-producing Escherichia coli in Ontariodairy cows and associations with production and man-agement. Prev Vet Med 1993; 15: 261±74.

45 Johnson RP, Cray WC Jr, Johnson ST. Serum antibodyresponses of cattle following experimental infectionwith Escherichia coli O157:H7. Infect Immun 1996; 64:1879±83.

46 Bast DJ, Banerjee L, Clark C, Read RJ, Brunton JL. Theidenti®cation of three biologically relevant globotriaosylceramide receptor binding sites on the Verotoxin 1 Bsubunit. Mol Microbiol 1999; 32: 953±60.

47 Wolski VM, Soltyk AM, Brunton JL. Mouse toxicity andcytokine release by verotoxin 1 B subunit mutants. InfectImmun 2001; 69: 579±83.

48 Maloney MD, Lingwood CA. CD19 has a potential CD77(globotriaosyl ceramide)-binding site with sequencesimilarity to verotoxin B-subunits: implications ofmolecular mimicry for B cell adhesion and enterohe-morrhagic Escherichia coli pathogenesis. J Exp Med1994; 180: 191±201.

49 Hoey DE, Currie C, Else RW, Nutikka A, Lingwood CA,Gally DL, Smith DGE. Expression of receptors for vero-toxin 1 from Escherichia coli O157 on bovine intestinalepithelium. J Med Microbiol 2002; 51: 143±9.

50 Holgersson J, Jovall PA, Breimer ME. Glycosphingoli-pids of human large intestine: detailed structural char-acterization with special reference to blood groupcompounds and bacterial receptor structures. J Biochem(Tokyo) 1991; 110: 120±31.

51 Bitzan MM, Wang Y, Lin J, Marsden PA. Verotoxin andricin have novel effects on preproendothelin-1 expres-sion but fail to modify nitric oxide synthase (ecNOS)expression and NO production in vascular endo-thelium. J Clin Invest 1998; 101: 372±82.

52 Su GF, Brahmbhatt HN, Wehland J, Rohde M,Timmis KN. Construction of stable LamB-Shiga toxinB subunit hybrids: analysis of expression in Salmonellatyphimurium aroA strains and stimulation of B subunit-speci®c mucosal and serum antibody responses. InfectImmun 1992; 60: 3345±59.

53 Niebuhr K. Biochemische und immunologischeCharakterisierung der rekombinanten B-Untereinheitdes Shigatoxins. Diploma thesis. 1991; TechnischeUniversitaÈt Braunschweig.

54 Gentry MK, Dalrymple JM. Quantitative microtiter cyto-assay for Shigella toxin. J Clin Microbiol 1980; 12: 361±6.

55 Menge C. Protocols to study effects of Shiga toxin onmononuclear leukocytes. Methods Mol Med 2003; 73:275±289.

56 Bùyum A. Isolation of lymphocytes, granulocytes andmacrophages. Scand J Immunol 1976; Suppl.5 5: 9±15.

57 Menge C, Neufeld B, Hirt W, Bauerfeind R, Baljer G,Wieler LH. Phenotypical characterization of peripheralblood leucocytes in the newborn calf. Zentralbl Veteri-narmed [B] 1999; 46: 559±65.

58 Dennis RD, Lochnit G, Geyer R. Strategies for prelimin-ary characterization of novel amphoteric glycosphingo-lipids. In: Hounsell EF, Ed. Glycoanalysis Protocols.Totowa, NJ: Humana Press, 1998: 197±212.

Documents

Bovine lymphocytes express functional receptors for Escherichia coli Shiga toxin 1