Effects of TNFα on verocytotoxin cytotoxicity in purified human glomerular microvascular endothelial cells

Kidney International, Vol. 51(1997), pp.

1245—1256

Effects of TNFa on verocytotoxin cytotoxicity in purified human glomerular microvascular endothelial cells PETRA A. VAN SEYFEN, VICTOR W.M. VAN HINSBERGH, THEA J.A.N. VAN DER VELDEN, NIcoLE C.A.J. VAN DE KAR, MARIO VERMEER, JOHN D. MAHAN, KAREL J.M. ASSMANN, LAMBERTUS P.W.J. VAN DEN HEUVEL, and LEO A.H. MONNENS Department of Pediatrics, University Hospital, Nmegen, and Gaubius Laboratoty TNO-PG, Leiden, The Netherlands; Department of Pediatrics, The Ohio State University, Columbus, Ohio, USA; and Department of Pathology, University Hospital, Nijmegen, The Netherlands

Efl'ects of TNFa and verocytotoxin cytotoxicity in human glomerular microvascular endothelial cells. In the pathogenesis of the hemolytic uremic syndrome (HUS), endothelial damage of glomeruli and arterioles of the kidney appears to play a central role. Previous studies have shown that verocytotoxin-i (VT-i) cytotoxicity on human vein endothelial cells require additional stimuli, in particular the inflammatory mediator tumor necrosis factor alpha (TNFa). In this study the effects of VT on human glomerular microvascular endothelial cells (GMVEC) were examined. A reproducible method was developed for the isolation and purification of large numbers of highly purified GMVEC. The obtained GMVEC were over 99% pure; their endothelial origin was demonstrated by the expression of the endothelial antigens von Willebrand factor, EN-4, PECAM-I and V,E-cadherin. Upon stimulation with TNFs the cells expressed the endothelial-specific adhesion molecule E-selectin. A limited number of fenestral structures was observed by scanning electron microscopy (SEM), suggesting glomerular origin of the endothelial cells. Cytotoxicity of VT-I

to GMVEC was evaluated by determination of the number of viable adherent cells and by assay of overall protein synthesis after exposure to varying concentrations of VT-i. In non-stimulated GMVEC, cytotoxicity of VT-i was inversely related to the degree and duration of confluence, subconfluent cells being the most sensitive. In highly confluent GMVEC, VT cytotoxicity required pre-exposure of the cells to the inflammatory mediator TNFa, which induced an increase in the number of VT receptors on GMVEC. Thin layer chromatography of extracted glycolipids from the GMVEC showed binding of VT-i to globotriaosylceramide (Gb3), known to be the functional receptor for VT. There were no major differences in protein synthesis inhibition with equal concentrations VT-I and VT-2. In conclusion, in this study we provide a reproducible method to isolate, purify and culture well characterized human GMVEC on a routine basis. In vitro studies with these GMVEC demonstrate that VT cytotoxicity depends on the degree of confluence and the additional preexposure to the inflammatory mediator TNFa. These observations provide further

insight into the complex events that may occur in glomeruli in the pathogenesis of HUS.

The hemolytic uremic syndrome (HUS) is characterized by renal failure, thrombocytopenia and hemolytic anemia [I]. Endothelial damage of glomeruli and kidney arterioles appears to play

a pivotal role in the pathogenesis of HUS. Histopathological studies of the kidney in HUS patients reveal swollen and detached

endothelial cells and deposits of fibrin, all within the glomerulus [21. In severe cases of HUS endothelial cell damage is not limited

to the kidney, but extends to other organs, such as brain and

pancreas. Infection with a verocytotoxin (VT) producing E. coli has been implicated in the etiology of the epidemic form of HUS [31. In vitro studies with umbilical vein and adult vein endothelial cells indicate that endothelial cells become more susceptible to VT if prestimulated with inflammatory mediators [4—7]. Induction

of the globotriaosylceramide (Gb3) receptor, known to be the functional receptor for VT, proved to be the mechanism by which inflammatory mediators induce increased susceptibility to VT [6].

A number of observations have indicated morphological, biochemical and functional heterogeneity between macro- and mlcrovessel endothelial cells, and even between microvascular endothelial cells derived from different organs [8]. Therefore, to test

if non-stimulated glomerular microvascular endothelial cells (GMVEC) are more susceptible to verocytotoxin-1 (VT-I) than their venous counterparts, it would be desirable to directly study

GMVEC. The isolation and culture of human renal [9]

and

glomerular [10—12] MVEC has been reported, but currently no procedure is available to culture human GMVEC on a routine basis. Here, we report on a reproducible method to isolate, purify and culture large numbers of GMVEC that are free of contaminating cells, and on the characterization of these cells. Further-

more, we describe the effects of TNFcr and the degree of confluence on VT cytotoxicity for human GMVEC. We found that VT-I cytotoxicity for highly confluent GMVEC requires pre-exposure of the cells to the inflammatory mediator TNFa, which is similar to previous findings on umbilical and adult vein endothelial cells [4—7]. These data seem to contrast to observations by Obrig et al [13], who showed that VT-i had an identical

toxic effect on TNFa-stimulated and non-stimulated renal MVEC. To explain these apparent differences, we analyzed the effects of VT-i on GMVEC during maintenance in a confluent culture, and demonstrate that the sensitivity of GMVEC for VT-i is markedly reduced in the post-confluent state. Methods

Received for publication August 12, 1996 and in revised form November 8, 1996 Accepted for publication November 12, 1996

© 1997 by the International Society of Nephrology

Isolation of glomeruli

Studies were performed with human kidneys not suitable for transplantation because of anatomical anomalies, technical or 1245

1246

van Setten et al: Effects of verotoxin and TNFa on GMVEC

immunological reasons. Written consent was obtained from the Medical Ethical Review Board of the University Hospital Nijmegen, The Netherlands. In general, the procedure was started as soon as possible after explantation of the kidney, usually within 24

and maintained at 37°C with 5% CO2 atmosphere. The medium was changed every two to three days and cells were passaged with trypsin/EDTA onto gelatin-coated wells at a split ratio of 1:2 or 1:3. In general it was necessary to repeat the immunomagnetic

to 72 hours. Glomeruli were isolated under sterile conditions by separation technique once or twice or to perform additional dissecting the cortex followed by a gradual sieving procedure manual weeding to obtain pure cultures of GMVEC. In case the epithelial cells seemed to overgrow the endothelial (sieves with opening sizes 2000 jxm to 53 jim; Endecotts, London, UK). Depending on the age of the donor, the glomeruli were cells in the primary culture from morphological point of view, the collected on top of the screens with opening sizes of 180, 125, 108, protocol as described above was preceded by a negative immuno90 or 53 jim and digested with 0.1% (wt/vol) collagenase type 2 magnetic separation technique using an epithelial-cell specific CLS (Worthington, NJ, USA) in M199 (Biowithaker, MD, USA) antibody (CD 26; Pharmingen, San Diego, CA, USA) to remove for two hours at 37°C. Following digestion glomerular remnants the contaminating epithelial cells. were shaken vigorously to loosen up the glomeruli and liberate the

endothelial cells. Subsequently the glomerular remnants were washed, centrifuged and resuspended in complete media that

Characterization

consisted of M199 supplemented with 10% (vol/vol) newborn calf Primary and serially passaged cultures were characterized using serum (NBCS; GIBCO, Grand Island, NY, USA), 10% (vol/vol) three criteria. human serum (HS; local blood bank), 2 mmol/liter glutamine (1.) Morphological criteria. Outgrowth from collagenase treated

(ICN Biomedicals, Zoetermeer, The Netherlands), 100 IU/ml glomerular remnants and subcultures were examined with a penicillin/0.1 mg/mi streptomycin and 5 U/ml heparin (Leo Phar- phase-contrast microscope. Photomicrographs were made at varmaceuticals, Weesp, The Netherlands) and 150 mg/liter crude ious passages using Technical Pan films. (2.) Immunological criteria. The presence of endothelial cellpreparation of endothelial cell growth factor (prepared from bovine brains as described by Maciag et al [141) and plated on specific antigens was determined by performing indirect immunogelatin (1%; Fluka BioChemika, Buchs, Switzerland) coated wells fluorescence studies with a panel of endothelial cell-specific monoclonal antibodies (MoAb). GMVEC were seeded on gela(Costar Corp., Cambridge, MA, USA). tin-coated glass-coverslips and grown until confluence. Cells were fixed in 80% (vol/vol) aceton for 10 minutes at 4°C and rinsed with phosphate-buffered saline (PBS). The presence of von Willebrand In general the glomerular remnants attached within one to two factor was demonstrated by indirect immunofluorescence staining days, after which outgrowth of predominantly cells of epithelial with rabbit anti-human von Willebrand antiserum [1:80 dilution in and endothelial origin was noted. Daily monitoring of proliferat- 10% (vol/vol) pig serum; DAKO; Glostrup, Denmarki and FITCing cells determined the exact timing of separating the endothelial swine anti-rabbit Ig (1:20 in PBS; DAKO). The presence of other and epithelial cells, usually within three to eight days after plating endothelial-cell specific antigens such as PECAM-1, EN-4 antigen of the glomerular remnants. The purification procedure consisted and V,E-cadherin was investigated by incubating the fixed cells Primary culture and purification

basically of three steps. First, selective trypsinization was per- with anti-PECAM-1 MoAb (10 jig/mI; CLB, Amsterdam, The formed, allowing endothelial cells to detach and leaving the Netherlands), MoAb EN-4 [15] (10 jig/mi; Sanbio, Uden, The epithelial cells attached to the plate. Selectivity in cell detachment Netherlands) and anti-V,E-cadherin MoAb [16] (1:500; provided was reached by incubating the cells with trypsin/EDTA (0.5 g by Dr. Dejana, Mario Negri Institute, Milan, Italy) in PBS/10% trypsin 1:250/0.2 g EDTA in 1 liter of Modified Puck's Saline A; goat serum. After a 30-minute incubation period at room temperGibco) at room temperature for a short incubation period, the ature the cells were washed twice with PBS and incubated with duration of which was determined during visual inspection of the FITC-goat anti-mouse Ig (1:20; DAKO). After a 30-minute cell layer by phase-contrast microscopy (about 60 seconds). Sec- incubation time at room temperature, cells were washed twice ond, to relieve glomerular remnants, which proved to be a source with PBS and mounted with Vectashield (Burlingame, CA, USA) of contaminating cells when subcultured, trypsinized cells and under a coverslip and investigated using a fluorescence microglomerular remnants were filtrated through a 38 jim sieve. The scope. Similar immunofluorescent assays were carried out using filtrate was collected, washed and pelleted by centrifugation MoAb to control for contamination with mesangial-cells (anti-a(200 X g). Third, the pellet was resuspended in Hanks' balanced smooth muscle actin, clone 1A4; Sigma, Zwijndrecht, The Nethsalts (ICN Biomedicals)/10% (vol/vol) fetal calf serum (FCS; erlands) and epithelial cells (monoclonal anti-cytokeratin 8, M20; GIBCO) and incubated with a monoclonal antibody against provided by Dr. G. van Muijen, University Hospital Nijmegen, PECAM-1 (CLB, Amsterdam, The Netherlands) for 30 minutes The Netherlands [17]). The expression of the endothelial-cell on ice. Excess of unbound anti-PECAM-1 monoclonal was re- specific adhesion molecule E-selectin upon stimulation with moved by washing for three times. Subsequently, an immunomag- TNFa was determined by ceil-ELISA. GMVEC were grown until netic separation technique with dynabeads coated goat anti- confluence in 96-well tissue culture plastic plates (Costar Corp.) mouse antibodies (Dynal A.S; Oslo, Norway) and a magnetic and subsequently exposed to TNFa (10 ng/ml) during a four-hour particle collector (Dynal A.S) was performed to selectively collect incubation period. Cells were rinsed once with M199 and fixed in glomerular endothelial cells. In general the number of beads used 0.025% glutaraldehyde. Following fixation wells were incubated was calculated according to the following formula: (total number with MoAb against E-selectin (ENA-2; gift of Dr. J. Leeuwen-

of cells) X (% estimated endothelial cells) X 3. The use of too berg, University Hospital Maastricht, The Netherlands [18]) many beads per cell reduced the attachment of the endothelial diluted in M199/5% FCS at a concentration of 10 jig/mI during 60 cells. PECAM-1 positive cells were plated on gelatin-coated wells minutes at 37°C. The cells were washed twice with PBS/Tween

van Setten et a!: Effects of verotoxin and TNFc on GMVEC

and incubated with peroxidase labeled, goat anti-mouse second antibody (1:1000; DAKO) for 30 minutes at 37°C. After washes with PBS/Tween o-phenylenediamine-peroxide was added. The reaction was stopped by the addition of 2.5 mol/liter H2S04 (50

1247

room temperature and counting radioactivity in a gamma-

counter. Non-specific binding was assessed in parallel incubation by determining 1251-VT-I binding in the presence of a 100-fold excess of unlabeled VT-i. Cellular specific binding was calculated x1/well). Absorbency was read at 492 nm on a Titretek Multiscan. by subtracting the non-specific binding from the cellular binding Absorbency with second antibody alone was subtracted from all of l25\T as determined in the absence of unlabeled VT-i. All absorbency values expressed. Expression of E-selectin was also determinations were done in duplicate. Data were analyzed using analyzed by performing immunofluorescence studies and FACS the method described by Scatchard [201. To investigate whether analysis as described above with minor adjustments, in that the Gb3 is also the functional receptor for VT in GMVEC, TNFacells were incubated with the primaly and secondary antibodies stimulated GMVEC were incubated with a MoAb against Gb3 diluted in PBS plus (PBS plus Mg2 and Ca2) and cells were only [21] (MoAb anti-CD77; clone #38-13; Biodesign International, fixed after incubation with primary and secondary antibodies. Kennebunk, ME, USA) one hour before and during assaying the (3.) Ultrastructure. Evaluation of human GMVEC by scanning binding of 1251-VT-1. electron microscopy was performed by the osmium-thiocarbohy-Glycolipid extraction and thin layer chromatography (TLC) drazide-osmium technique. GMVEC grown on gelatin coated glass-coverslips were rinsed with Mi99 and fixed for two hours at Highly confluent GMVEC were cultured in six-well plates (20

room temperature with 2.5% glutaraldehyde in 0.1 mol/liter cm2) for 24 hours with and without TNFa (10 nglml). Subsecacodylate buffer. The cells were then rinsed carefully with 0.1 quently the cells were washed with PBS and trypsinized, after mol/liter cacodylate/6.8% sucrose pH 7.38 and postflxed with 1% 0s04 in the same cacodylate buffer. Subsequently the cells were rinsed with aquabidest and immersed in 1% filtrated thiocarbohydrazide-solution (TCH). Following immersion, alternate fixa-

tion in 0s04 and immersion in TCH and fixation in 0s04 were repeated after which the cells were selectively dehydrated in graded alcohol solutions and critical point dried. Cells were visualized with a SEM Jeol 1200 EX/2 at an accelerating voltage of 50 Ky. Cytotoxicily assay

which glycolipids were extracted and separated by methods earlier described by Lingwood et al [22]. Following thin layer chromatography, the silica gel TLC plate was incubated with 50 ml, 1.5

nmol/liter 1251-VT-i in 1% BSA and 0.05% Tween-20 in PBS during four hours at 4°C. The plate was washed extensively with PBS supplemented with 1% BSA and 0.05% Tween-20, air dried and analyzed by a Fuji BAS 1000 phosphor-imager. Protein synthesis

Protein synthesis was determined by assaying the incorporation of 35S-labeled methionine (0.25 jxCi/ml complete medium) in

GMVEC were cultured in complete medium on gelatin coated 35S-proteins during an 8 to 24 hours incubation period with 24-well plates and grown until confluence. Five days after reaching different concentrations of VT-i and VT-2. After incubation, the confluence (highly confluent) cells were preincubated with or cells were washed and 35S-labeled cellular proteins were precipiwithout TNFa (10 ng/ml) for 24 hours. The next day the medium tated by adding 10% trichloroacetic acid. Precipitated proteins was aspirated and new medium with different concentrations of were dissolved in 0.3 ml 0.3 mol/liter NaOH, 60 sl 1.5 mol/liter VT-I and VT-2 (provided by Prof Karmali; Hospital for Sick HC1 was added and precipitated radioactivity was counted in a Children, Toronto, Canada) was added. VT-i and VT-2 were liquid scintillation counter. diluted in medium with 20% FCS instead of 10% NBCS and 10% HS, since previous studies have indicated that NBCS and HS may Results have neutralizing activity against VT-i and VT-2. After 8 and 24 Isolation and punfi cation of human GMVEC hours the cells were washed with PBS, released with trypsin/ EDTA and subsequently viable, Trypan blue excluding cells were Incubation of pure populations of human glomeruli, which were counted with a hemocytometer. In order to study the influence of isolated by a gradual sieving procedure with collagenase for two confluence, cells at different degrees of confluence were preincu- hours at 37°C followed by vigorously shaking, resulted in small bated with TNFx and tested for VT-i cytotoxicity as described clumps consisting of glomerular remnants. These remnants atabove. tached to gelatin-coated plates within one to two days. Following attachment, outgrowth of primarily endothelial and epithelial cells lodination of VT-i and the binding of J25j VT-i to human was noted (Fig. IA), the latter easily overgrowing the GMVEC. GMVEC As judged by phase-contrast microscopy the outgrowing endotheVT-i was radiolabeled with Na-'251 according to the lodogen hal cells demonstrated a slightly elongated morphology with a procedure [19]. The radioactive activity of '251-VT-1 ranged from nucleus containing one or two nucleoli and a dark perinuclear 10 to 12 xCi4tg of protein. GMVEC were grown in 24-well plates cytoplasm area that was surrounded by a bright halo. Epithelial until they reached confluence. After five days the cells were cells displayed the epithelial cell specific cobblestone pattern of

preincubated with or without inflammatory mediators for 24 densely packed polygonal cells. Besides the tendency of the hours at 37°C. Subsequently the GMVEC were washed with M199

contaminating epithelial cells to overgrow the endothelial cells,

medium/0.i% human serum albumin (HSA) and incubated for

these cells seemed to inhibit the growth of endothelial cells,

three hours with 0.25 to 32 nmol/liter 25I-VT-1 in M199 medium/ 0.1% HSA at 0°C. After three hours the incubation medium was

probably by producing growth inhibiting factor(s). Therefore, it was crucial to remove epithelial cells and purify the endothelial cell populations. The purification procedure consisted basically of three steps: selective detachment of most of the endothelial cells by trypsin-EDTA treatment; removal of glomerular remnants by

aspirated and the cells were washed for five times in Mi99 medium/0.1% HSA. Cell-associated 1251-VT-1 was determined by

solubilizing the cells in 400 jsl 1 mol/liter sodium hydroxide at

1248

van Setten et al: Effects of verotoxin and TNFa on GMVEC

passages 3 to 4 (Fig. 1C). Subculturing of these GMVEC with of split ratio of 1:3 or 1:6 has been successfully accomplished to passage 10 without significant changes in morphology and function. Identification and characterization of human GMVEC

Upon examination by phase contrast microscopy, confluent monolayers of human GMVEC exhibited a typical contact-inhib-

ited morphology (Fig. 1C). They were not contaminated by

'y_:%I4 •

-

spindle shaped and/or overcrossing mesangial cells [23], or by epithelial cells, which display a vety regular, polygonal cobblestone morphology [24]. The endothelial and glomerular nature of the cells was confirmed by the expression of endothelial-specific antigens and by their ultrastructure. Indirect immunofluorescence microscopy revealed the presence of von Willebrand factor in a distinct granular pattern (Fig. 2A). Furthermore, the endothelial specific antigens EN-4, PECAM-1 and V,E-cadherin antigen were present at the regions of intercelp

•

lular contacts (Figs. 2 B, C, E). When the GMVEC were subcultured for more than 10 passages, the cells remained positive

for all these markers. The cultured human GMVEC showed no immunoreactivity to the anti-cytokeratin 8 [17] antibody or to the anti-a smooth-muscle actin antibody, excluding the contaminating epithelial and mesangial cells, respectively (data not shown). Upon stimulation with TNFa (10 ng/ml) for four hours, expression of the endothelial cell specific protein E-selectin was induced

_•w-

on human GMVEC, whereas unstimulated controls did not 0.55 and OD, 0.15 0.02, express E-selectin (OD, 3.24 respectively, as determined by cell-ELISA; mean SD; N = 3). Similar results were obtained when E-selectin expression was determined by indirect immunofluorescence or FACS analysis (data not shown).

At the ultrastructural level, the cultured human GMVEC displayed small pores with a diameter of about 100 nm, as revealed by scanning electron microscopy (Fig. 3). These struc-

tures were not observed in human umbilical and iliac vein endothelial cells (data not shown), and probably represent fenestrae. The persistent presence of these fenestral structures in GMVEC in Vitro, although to a lesser extent when quantified per

square micrometer in vivo, points not only to the glomerular endothelial origin, but also to a certain degree of differentiation of

the cultured GMVEC. Fig. 1. A. Phase-contrast microscopy of primaiy culture of human GMVEC.

Human glomeruli were isolated by a gradual sieving procedure and vigorously treated with collagenase. Subsequently the glomerular remnants were plated on gelatin coated plates that attached within Ito 2 days. Following attachment, outgrowth of primarily endothelial (>) and epithehal (*) cells was noted. B. Phase-contrast microscopy of GMVEC in the first passage after performing the initial purification procedure. Predominantly cells of endothehial origin surrounded by immunomagnetic beads are visible. C. Phase-contrast microscopy of a confluent monolayer of purified third passage human GMVEC (magnification X75).

Effect of VT-I on viabilily of human GMVEC To evaluate whether GMVEC are sensitive to the toxic effect of VT-i, cell viability was estimated by counting viable adherent cells with a hemocytometer after an 8 to 24 hour exposure to various

concentrations of VT-i. VT-I at concentrations ranging from 1 fmol/hiter to 10 nmol/hiter did not significantly affect cell viability of three different populations of highly confluent human GMVEC when cultured under basal conditions over a 24-hour time period.

At the highest dose tested, a 24 hour incubation with VT-I (10 nmol/hiter) caused 0%, 1% and 6% cytotoxicity, respectively, in filtration; and selective collection of GMVEC by binding to these three different populations of highly confluent GMVEC. anti-PECAM-1 MoAb and immunomagnetic separation tech- Similar to previous observations from our group, studying the nique (Fig. 1B). The immunomagnetic separation with anti- effects of VT-i on human umbilical vein and human femoral vein PECAM-1 MoAb was repeated once or twice and additional endothelial cells (HUVEC) [6], human GMVEC cultured and manual weeding of contaminating cells was performed at passages treated under identical conditions became susceptible to the 2 to 3. This resulted in highly purified cultures of GMVEC by cytotoxic effect of VT-i after preincubation with the inflammatory

1249

van Setten et al: Effects of verotoxin and TNFa on GMVEC

Fig. 2. Endothelial-specific markers were visualized by indirect immunofluorescence microscopy as described in the Methods section. A. von Willebrand

factor antigen. B. EN-4 antigen. C. PECAM-1 antigen. D. Negative control lacking primary antibody (showing an immunomagnetic bead). E. V,E cadherin antigen. The magnification for A to E was 350-fold.

mediator TNFs (10 ng/mI). In TNFa-stimulated GMVEC, VT-i

a marked effect on VT-I cytotoxicity in highly confluent cells,

incubation at 10 nmol/iiter for 24 hours resulted in the three different populations of GMVEC studied, in respectively 80%,

while it increased VT-I cytotoxicity in subconfluent cells only to a limited extent. Similar results were obtained with GMVEC of two

other donors (data not shown). 76% and 90% cytotoxicities. Figure 4 shows the time and concentration dependent effect of VT-i during an 8- to 24-hour incubation period on cell-viability of TNFa increases the number of VT-i receptors on human GMVEC

unstimulated and TNFa-stimulated GMVEC isolated from one Binding of '251-VT-1 to human GMVEC was saturable (Fig. representative donor out of three. The threshold dose to exert a cytotoxic effect on TNFa-stimulated GMVEC was 0.1 pmol/Iiter. 6A) and specific; 25I-VT-1 binding was entirely (>95%) disNo cytotoxicity was noticed when VT-i was heat-inactivated (30 placed by a 100-fold excess of unlabeled VT-i. Pretreatment of mm at 90°C) prior to addition to the culture medium indicating a GMVEC with TNFa resulted in an increase in '251-VT-1 binding specific cytotoxic effect of VT-i.

(Fig. 6A). Scatchard plot analysis revealed that one type of

Susceptibility of the human GMVEC did not only depend on the additional stimulation with inflammatory mediators but also on the degree of confluence of the GMVEC. Figure 5 shows the influence of confluence of a representative population of GMVEC on the susceptibility for VT-I. Cell-viability of subconfluent and confluent GMVEC was affected under basal conditions by VT-I whereas highly confluent GMVEC were resistant to the cytotoxic effect of VT-I. Pre-exposure of GMVEC to TNFa had

binding site is involved in VT-i binding to human GMVEC (Fig.

6B). Furthermore, it showed that 24 hours of preincubation of human confluent GMVEC with TNFa caused a threefold increase in the number of specific binding, while the apparent affinity of VT-I did not change significantly (1.5 nmol/liter and 1.7 nmol/

liter, respectively). In addition to TNFa, interleukin-la and bacterial lipopolysaccharide induced an increase in specific 1251..

VT-I binding to GMVEC (1.6

0.2-fold and 1.7

0.4-fold

van Setten et al: Effects of verotoxin and TNFs on GMVEC

1250

Fig. 3. Scanning electron microscopy of cultured human GMJ/EC. Cells were used after six passages; notice the small pores, approximately

uniform in size and with a diameter of about 100 nm, suggestive for fenestral structures. The arrow (") indicates a representative fenestral structure (magnification 15000-fold).

[25], lymphocytes [26] and human kidney [271, is also the receptor

10

x

for VT-i on human GMVEC, TNFa-stimulated GMVEC were incubated with a MoAb against Gb3 one hour before and during 8

assaying the binding of 1251-VT-1 to human GMVEC. The anti-Gb3 MoAb inhibited the binding of 1251-VT-1 to TNFsstimulated GMVEC in a concentration dependent way, but, probably because of the moderate affinity of the MoAb, no complete inhibition was reached at the concentrations used

c'J

E

0 U) ci)

61

C-)

C a) ci)

0.3 and 27.7% 0.6 inhibition at 10 and 20% (vol/vol) antibody solution, respectively).

4

(18.3

2

To further characterize the nature of the receptor for VT-I present on human GMVEC neutral glycolipid extracts were prepared from both unstimulated and TNFa-stimulated human GMVEC and thin layer chromatography of these extracts was

-o Cci

ci) cci

>

—

0 0

8

16

24

Time, hours Fig. 4. Time- and dose-dependent cytotoxicity of VT-I for human GMVEC. Human GMVEC of one representative donor (sixth passage) out of three were grown to confluence and maintained for five days in the confluent state. Subsequently, the cells were preincubated for 24 hours without (open symbols) or with TNFa (10 ng/ml; closed symbols) after which the cells were washed and incubated with either control medium (the same

culture medium without addition; , S) or VT-I (10 nmol/Iiter; A, A), VT-I (0.1 nmol/l; 0, S) for 8 and 24 hours. After the indicated periods of time, the number of viable adherent cells was counted with a hemocytometer. Similar results were obtained in two other experiments with two different donors.

performed. After incubation of these thin layer chromatograms

with 1251-VT-1 and extensive washing, the bound '251-VT-1 was

detected by exposure to a phosphor-imager. The radiolabeled VT-l bound strongly to Gb3 and not to Gb4 when a standard preparation of neutral glycolipids was tested (Fig. 7, lane C). In unstimulated and TNFs-stimulated human GMVEC binding of '251VT1 occurred predominantly to Gb3 (Fig. 7, lanes A, B). This pattern was consistently found with different preparations of

human GMVEC obtained from two different donors. The increase in 1251-VT-1 binding to the glycolipid extracts of TNFastimulated human GMVEC indicates that the synthesis of VT-i receptors is increased rather than that it only reflects a redistri-

bution of VT-i receptors from intracellular stores towards the plasma membrane.

increase, respectively); interleukin-6 (0.9

kin-4 (0.9

0.07) and interleu-

0.02) did not affect VT-i binding to GMVEC.

Characterization of the ligand to which VT-i hinds on human GMVEC To investigate whether globotriaosylceramide, which has been

demonstrated to be the receptor for VT-i on endothelial cells

Effect of VT-I and VT-2 on protein synthesis

VTs may exert their cytotoxic effect by inhibiting the elongation factor-i dependent binding of aminoacyl-tRNA to the ribosomcs [28], a crucial step in the process of eukaryotic protein synthesis. To investigate whether VT cytotoxicity in human GMVEC was

1251

van Setten et a!: Effects of verotoxin and TNFcc on GMVEC

AB

A

A

C 60

10

x

8

c'J

E

E

.0

6

ci)

0

40

-o

C

ci) ci)

4

a)

2

.0

C

0

.0

> 20

cci

LC)

c'J

0

3

0

6

9

12

0

Days after trypsinization

AB

B

() 0

36

0.20

0.30

125IVT1 nM

B 1.50

75

/

0 U)

0

C

100

E

a)

24

12

0

x

E1-

50

C

1.00

a)

a)

a)

U

25

U-

a) -o

>

0

ci)

0

3

6

9

0.50

12

Days after trypsinization Fig. 5. VT-I cytotoxicüy is related to the degree of confluence of human GMVEC. Human GMVEC of one representative donor out of three were grown to confluence, trypsinized and seeded 1:6 onto gelatin coated 24 wells plates. Forty-eight hours before the indicated time points after trypsinization (as indicated on the x-axis) GMVEC were preincubated with either control medium (same culture medium without addition; open symbols) or TNFa (10 nglml; closed symbols) for 24 hours. Subsequently the cells were washed and exposed to either control medium or VT-I (I nmol/liter) for 24 hours, after which the number of viable adherent cells

0.00 0.00

0.10

B (mol/ceII)x 10-17 Fig. 6. TNFa increases the specific binding of'251-l/T-I to human GMJ/EC. A. Human GMVEC of one representative donor were grown to "just

reaching confluence" and pre-treated with either control medium (LI) or TNFu (10 nglml; U) for 24 hours. Subsequently the cells were incubated with increasing concentrations of '251-VT-l (0.25 to 32 nmol/liter) for three hours at 4°C, after which specific binding of '251-VT-i to GMVEC

was counted with a hemocytometer. GMVEC indicated by A were was calculated as indicated in the Methods section. B. Scatchard plot categorized subconfluent (<50% confluent), by B "just reaching confluence" and by C "highly confluent" for respectively two, four and seven days. A. Effect of VT-i on the viability of non-stimulated-(open symbols) and TNFa-treated GMVEC (closed symbols). Symbols are: (V, V) control cells; (0, S) VT-i treated cells. B. Effect of VT-i on the number of viable adherent cells expressed as percentage of control (calculated by dividing the number of viable cells after VT-I treatment by the number of viable adherent cells exposed to control medium X 100%). Symbols are: (LI) unstimulated GMVEC; (U) TNFa-prestimulated GMVEC. Similar results were obtained in two other experiments with two different donors.

based on inhibition of protein synthesis, newly synthesized pro-

analysis of 251-VT-i binding to human GMVEC preincubated with either control medium (0) or TNFa (10 ng/ml; S) for 24 hours.

synthesis of highly confluent GMVEC over a 24-hour incubation period, whereas it inhibited protein synthesis in subconfluent and confluent cells (Fig. 8). When the highly confluent human GM-

VEC were preincubated with TNFs for 24 hours, the protein synthesis rate was inhibited similarly to subconfluent cells. Table I summarizes the results of protein synthesis inhibition by VT-i in GMVEC differing in degree of confluence (N = 3). TNFa alone did not decrease overall protein synthesis.

teins were determined by assaying the incorporation of 35SWhen the effect of VT-i and VT-2 on protein synthesis methionine into total cellular proteins. In agreement with cell- inhibition on TNFa-stimulated GMVEC of two different donors viability data, VT-i alone did not or only slightly affect protein were compared, VT-i was slightly more potent in inhibiting

/v II

van Setten et a!: Effects of verotoxin and TNFa on GMVEC

1252

A

— Front

40

- Gbl

-Gb2 —Gb3

20

II

-Gb4 -Gb5

A

C

B

/ /.s____.

0,S

0

0

3

6

9

12

—Origin Days after trypsinization

A

B

C

B

D

100

A B ——

C

Fig. 7. '251-VT-i binding to glycolipids extracted from human GMVEC.

Glycolipids were extracted from "highly confluent" human GMVEC, separated by TLC and assayed for '251-VT-1 binding and visualized by a phosphor imager. Lanes A and B. Glycolipid extracts of 20 cm2 human GMVEC of one representative donor (lane A, control GMVEC; lane B, TNFa-treated 10 nglml GMVEC). Lane C. Standard neutral glycolipids, 2

75 4-.

0

pg of each glycolipid. Lane D. Orcinol staining of standard neutral

'.0 0'..

glycosphingolipids (Gbl, galactosylceramide; Gb2, lactosylceramide; Gb3, globotriaosylceramide; Gb4, globotetraosylceramide; Gb5, Forssman pentasaccharide).

S

.

50

D'

25

III,.•..—_-—.____—.___—_.

ath

"

U,

protein synthesis than VT-2, at all concentrations tested (Fig. 9). The lowest dose to inhibit overall protein synthesis was 1 pmol/ liter for both VT-i and VT-2. No effect of VT-2 was observed on the protein synthesis of non-stimulated highly confluent GMVEC.

0 0

3

6

9

12

Discussion Days after trypsinization VT producing E. coli play an important role in the etiology of the epidemic form of HUS. In this most common form of HUS in Fig. 8. The protein synthesis inhibitory effect of VT-i is related to the degree

young children, the hallmark of the pathogenetic mechanism is endothelial cell damage, most predominantly displayed by the glomerular capillaries. In this study we have described a modified technique for the reproducible culture of large numbers of highly purified human GMVEC that express a variety of endothelial-cell specific antigens. Furthermore, we have demonstrated that susceptibility for the cytotoxic effect of VT-i of these highly purified human GMVEC depended on the degree and duration of confluence and additional preexposure to inflammatory mediators. Pre-exposure to inflammatory mediators resulted in an increase in specific binding sites for VT-I accompanied by a rise in cellular amount of Gb3, known to be the functional receptor for VT-I. Like other eukaryotic cells possessing the Gb3 receptor, cytotoxicity was associated with protein synthesis inhibition. The cytotoxic effect of VT on HUVEC and the presence of the functional receptor Gb3 on HUVEC have been well described [4—7]. Additional stimuli such as the inflammatory mediators TNFcr and IL-la and LPS seemed to be crucial for VT to cause cytotoxicity to HUVEC. Because VT appears to have a predilec-

of confluence of human GMVEC. The experimental design is comparable to Figure 5. The incorporation of 35S-methionine in proteins by GMVEC is shown at different degrees of confluence (A, B and C), either cultured

under basal conditions (open symbols) or pre-treated with TNFa (10 nglml; closed symbols) before exposing the cells to either control medium (same culture medium without addition) or VT-i (1 nmol/liter). A. Effect

of VT-i on the incorporation of "S-methionine in proteins of nonprestimulated GMVEC (V, 0) and TNFcs-pretreated GMVEC (V, •). Symbols are: (V, 7) control cells; (S, 0) VT-I treated cells. B. Effect of VT-i on the incorporation of 35S-methionine in proteins expressed as percentage of control (35S-methionine incorporation in proteins of VT-l treated cells by 35S-methionine incorporation in proteins of control cells >< 100%). Symbols are: (LI) unstimulated GMVEC; (U) TNFa-prestimulated GMVEC. Similar results were obtained in two other experiments with two different donors.

tion for the glomerular microvascular endothelium in vivo, the interaction of VTs with human glomerular microvascular endothelial cells was studied in vitro. We investigated whether GMVEC display a higher basal susceptibility for VT than other

1253

van Setten et al: Effects of verotoxin and TNFa on GMVEC

Table 1. VT-i inhibitory effect on overall protein synthesis is related to the degree of confluence of human GMVEC

Donor 1 Without

TNFa Subconfluent Confluent Highly confluent (2 days) Highly confluent (4 days) Highly confluent (7 days)

54 73 87 92 87

Incorporation of 35S-methionine in proteins % of control Donor 3 Donor 2 With Without With Without TNFa TNFu TNFa TNFa 35 28 24 33 36

5 10 39 71 88

2 8 25

33 40

51 69 71 82 87

With

TNFa 30 32 46 48

37

Experimental design as described in the legend of Figure 8. The data of three independent experiments are given. The incorporation of 35S-methionine in proteins is expressed as % of control [35S-methionine incorporation in proteins of VT-I (I nmol/liter) treated cells divided by 35S-methionine incorporation in proteins of control cells x 100%].

standard techniques the cultured cells have been characterized as pure populations of glomerular microvascular endothelial cells. The combined presence of the endothelial specific markers von Willebrand factor, V,E cadherin and EN-4 antigen and PECAM-I (which is only encountered on endothelial cells, leukocytes and platelets), demonstrates the endothelial nature of our GMVEC. Upon inflammatory activation the endothelial cell specific leukocyte adhesion molecule E-selectin was also expressed. This agrees well with observations in vivo [29—33] showing an induction of

() 4— 0 (1)

C

I)

0

0

E-selectin in glomeruli during inflammation. The presence of fenestral structures suggests that one of the specific properties of the glomerular endothelium is preserved in culture. Suboptimal expression of fenestrae in these human GMVEC may be due to the fact that they are cultured under static conditions. The need

Cl)

U,

C,,

for dynamic flow to induce the expression of fenestrae in cultured 0.0001

0.1

glomerular endothelial cells was suggested by Ott, Olson and

VT concentration, nmol/Iiter

fenestrae in bovine glomerular endothelial cells cultured under

0.001

0.01

Fig. 9. EJfrct of VT-I and VT-2 on protein synthesis by "highly confluent"

human GMVEC. Human GMVEC were grown to confluence and maintained confluent for five days. Subsequently the cells were pre-treated without (open symbols) or with TNFa (10 nglml; closed symbols). After 24 hours of preincubation, the cells were washed and incubated with different concentrations (0.1 pmollliter to I nmol/liter) of VT-I (0, •) and VT-2 (A, A) for 24 hours. Incorporation of 35S-methionine in proteins was determined as described in the Methods section. Data are expressed as the

Ballermann [341 who, in contrast to our findings, did not observe

static conditions, but were able to induce the expression of fenestrae by shear forces.

Several papers have reported that HUVEC cultured under basal conditions are resistant to the cytotoxic effect of VT-i [4—7].

When prestimulated with inflammatory mediators HUVEC become susceptible to the cytotoxic effect of VT-I presumedly by an up-regulation of the receptors for VT on the cell-surface [6]. mean of two different experiments with two different donors (% of Recently Obrig et al [13] have demonstrated that cell viability and control). protein synthesis of unstimulated renal endothelial cells was affected by VT with no alteration in sensitivity for VT when pre-exposed to TNFn or LPS. Furthermore, Obrig et al described endothelial ceJis or need the local exposure to inflammatory high basal levels of VT receptor in renal endothelial cells reaching mediators to up-regulate Gb3 receptors and become suscep- concentrations approximately 50 times higher than in HUVEC. tible to the cytotoxic effect of VT, as seen in venous endothelial These data suggest a higher relative sensitivity of renal endothelial cells. cells versus HUVEC and a possible explanation for localized Since Striker et al [101 and Green et al [12] described their involvement of the kidney. In our study, cytotoxic and protein methods to isolate and purify human glomerular endothelial cells, synthesis inhibitory effects of VT were found only on GMVEC very few reports deal with glomerular endothelial cells in vitro, that were subconfluent, just reaching confluence or pre-exposed reflecting difficulties in the process of isolating and purifying large to the inflammatory mediator TNFa, but not in the highly numbers of glomerular microvascular endothelial cells devoid of confluent GMVEC. Because Obrig et al have used cells one day contaminating mesangial- and epithelial cells. In this report we after seeding in wells, it is likely that they studied the effect of VT have described an adapted procedure that is reproducible, rela- on subconfluent or confluent renal endothelial cells. tively simple to perform and yields high numbers of GMVEC. It Several papers have reported that subconfluent cells display a

appeared to be crucial to avoid any contamination early in the process of purification because a minute contamination of epithehal or mesangial cells causes inhibition of GMVEC proliferation

and rapid overgrowth of the contaminating cells. By several

significant higher binding capacity of radiolabeled VT-I and a higher susceptibility to the cytotoxic effect of VT than confluent cells [6, 35, 36]. Most probably this higher susceptibility of

subcorifluent and confluent cells is related to the cell-cycle

1254

van Setten et al: Ejfrcts of verotoxin and TNFo on GMVEC

dependent expression of Gb3 on the cell-surface compounded inside the glomerular capillaries such as mesangial [52, 53] and with increased Gb3 turnover [36]. Our observation that the degree epithelial [54] cells need further investigation. It remains to be of confluence of GMVEC affects susceptibility for VT corre- clarified for what reason inflammatory mediators are predomisponds with these findings. Since the half life of endothelial cells nantly produced inside the glomerular capillaries. in the normal mature human body is approximately a hundred to Although our in vitro data strongly suggest that local availability thousand days [37] and autoradiographic analysis in glomeruli of of inflammatory mediators is essential for VT to result in a rats demonstrated a constant rate of cell renewal of about 1% cytotoxic response, these data should be interpreted with some [38], we believe that from pathophysiological point of view, data reserve. Since mainly children are affected by infection with a VT on highly confluent cells are most appropriate in the study of the producing E. co/i, it remains to be investigated whether GMVEC pathogenetic mechanisms of HUS. isolated from kidneys of juvenile origin display an increased Infections with VT producing E. co/i most commonly involve sensitivity towards VT cytotoxicity when compared to GMVEC isolates producing either VT2 alone or VT-2 with YT1. The isolated from kidneys of adult origin. Of interest is the observation structural genes of the two toxins share 58% overall nucleotide of Lingwood [55] that FITC labeled VT binds to human renal and 56% projected amino acid sequence homologies [39]. The sections of infant < two years and not to the glomerulus of the toxins show similar physical properties such as toxin binding to adult. It should be noted, however, that Lingwood for the most Gb3 and inhibition of protein synthesis by the same mechanism, part examined kidney sections of minimal lesion nephrotic synhowever, differences in biological activity have been observed. In drome patients, where locally released cytokines may induce the vitro studies comparing the cytotoxic effects of VT-i versus VT-2 Gb3 receptor [56—581 resulting in the observed enhanced VT on HUVEC [40], Vero [41], HeLa [41] and Daudi cells [42] binding. We conclude from this study with highly purified GMVEC, demonstrated reduced susceptibility to VT2. While performing studies with VT1 and VT2 hybrid toxins [43] it was found that the showing glomerular microvascular endothelial cell specific feamodulation in VT susceptibility is due to differences in the B tures, that VT cytotoxicity depends on the degree of confluence subunit. Furthermore, Head, Karmali and Lingwood's study pro- and the additional pre-exposure to inflammatory mediators. The vided evidence for reduced binding affinity of VT2 for Gb3 when

latter are presumedly locally produced by monocytes and not only

compared to VT1 [43]. This observation corresponds to our up-regulate the receptor for VT, but also exert a variety of finding concerning the slightly reduced cytotoxicity of VT2 in pro-inflammatory and coagulatory effects on the endothelium, GMVEC, but is in contrast with a recent study by Louise and resulting in a cascade of events of mutually influencing factors in Obrig [40] suggesting an increased susceptibility of renal endo- the end leading to glomerular capillary damage, as observed in thelial cells for VT2. In vivo studies in mice [44] have demon- HUS patients. strated enhanced lethality of Shiga-like toxin-2 when compared to

Shiga-like toxin-i (toxins closely related to VT-2 and VT-i, respectively), suggesting that differences in properties other than direct target cell lysis are responsible for the observed heightened sensitivity of animals to VT-2. Since in vitro data described in this report do not provide strong evidence for an intrinsic enhanced basal susceptibility of GMVEC to VT as a possible explanation for the preferential glomerular microvascular endothelial cell damage observed in HUS patients, the local availability of inflammatory mediators in glomeruli may contribute to local susceptibility for the cytotoxic effect of VT. A local role for inflammatory mediators was also suggested by Hard et a! [45] reporting that Shiga-like toxin-i injection into transgenic

mice, bearing a chioramphenicol acetyltransferase (CAT) reporter gene coupled to a TNFa promoter, induced CAT activity (reflecting stimulation of the TNF promoter) in the kidney but not in other tissues. Monocytes and macrophages, known to produce and release inflammatory mediators may be the source of locally

produced glomerular inflammatory mediators. Of interest are recent in vitro studies of Tesh, Ramegowda and Samuel [46] and Van Setten et a! [47], which describe the interaction of VT with murine peritoneal macrophages and human monocytes, respectively. Both cells upon VT binding do not display protein synthesis

inhibition but synthesize and release inflammatory mediators, suggesting a key role for monocytes in the process of endothelial cell damage in HUS. In vivo data supporting involvement of inflammatory mediators in the pathogenesis of HUS are provided by several papers demonstrating elevated concentrations of inflammatory mediators in plasma and urine among HUS patients [48—51]. Other possible local sources of cytokine production

Acknowledgments This study was supported by grants from the Termeulen Fonds (The Netherlands) and the Dutch Kidney Foundation, grant number C94.1344.

We thank H. Dijkman (University Hospital Nijmegen, Department of Pathology, The Netherlands) for his assistance in performing the scanning electron microscopy.

Reprint requests to Prof Dr. L.A.H. Monnens, University Hospital Nijmegen, Department of Pediatric Nephrology, Geert Grooteplein 20, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands.

References 1. F0NG JS, DE CHADAREVIAN JP, KAPLAN BS: Hemolytic uremic

syndrome. Current concepts and management. Pediatr Gun N Am 29:835—856, 1982 2. RICHARDSON SE, KARMALI MA, BECKER LE, SMITH CR: The histo-

pathology of the hemolytic uremic syndrome associated with verocytotoxin-producing Escherichia coli infections. Hum Pathol 19:1102— 1108, 1988 3. KARMALI MA, PETRIC M, LIM C, FLEMING DC, ARBUS GS, LI0R H:

The association between idiopathic hemolytic syndrome and infection by verocytotoxin producing Escherichia coli. J Infect Dis 15 1:775—782, 1985

4. TvsH VL, SAMUEL JE, PERERA LP, SHAREFKIN JB, O'BRIEN AD:

Evaluation of the role of Shiga and Shiga-like toxins in mediating

direct damage to human vascular endothelial cells. I Infect DEs 164:344—352, 1991

5. Loijis GB, OBRIG TG: Shiga toxin-associated hemolytic-uremic syndrome: Combined cytotoxic effects of Shiga toxin, interleukin-1 beta, and tumor necrosis factor alpha on human vascular endothelial cells in vitro. Infect Immun 59:4173—4179, 1991

van Setten et a!: Effects of verotoxin and TNFs on GMVEC 6. VAN DE KAR NC, MONNENS LA, KARMALI MA, VAN HINSBERGFI VW:

Tumor necrosis factor and interleukin-1 induce expression of the verocytotoxin receptor globotriaosylceramide on human endothelial cells: Implications for the pathogenesis of the hemolytic uremic syndrome. Blood 80:2755—2764, 1992 7. KAYE SA, LOUISE CB, BOYD B, LiNowooD CA, OBRIG TG: Shiga toxin-associated hemolytic uremic syndrome: Interleukin-1 beta enhancement of Shiga toxin cytotoxicity toward human vascular endothelial cells in vitro. Infect Immun 61:3886—3891, 1993 8. HEWETF PW, MURRAY JC: Human microvessel endothelial cells:

Isolation, culture and characterization. In Vitro Cell Dev Biol Anim 29A:823—830, 1993

9. GIBBS VC, Woou DM, GAROVOY MR: The response of cultured human kidney capillary endothelium to immunologic stimuli. Hum Immunol 14:259—269, 1985

1255

long-lived antibody response to dysentcric disease. mt Immunol 2:1—8, 1990 27. BOYD B, LINGW000 C: Verotoxin receptor glycolipid in human renal tissue. Nephron 51:207—210, 1989 28. OBRIG TG, MORAN TP, BROWN JE: The mode of action of Shiga toxin

on peptide elongation of eukaryotic protein synthesis. Biochem J 244:287—294, 1987

29. HANCOCK WW, COTRAN RS: Induction of activation antigens on human glomerular endothelium by interleukin 1, interferon-y and endotoxin. (abstract) Kidney Int 31:322, 1987 30. CoTtwt RS, POBER JS: Effects of cytokines on vascular endothelium: Their role in vascular and immune injury. Kidney mt 35:969—975, 1989 31. REDL H, DINGES HP, BUURMAN WA, VAN DER LINDEN CJ, POBER JS,

COTRAN RS, SCHLAG G: Expression of endothelial leukocyte adhesion

molecule-I in septic but not traumatic/hypovolemic shock in the

G, JOHNSON A, LUCHTEL D, Ross R, STRIKER U: Isolation, charac-

baboon. Am J Pathol 139:461—466, 1991 32. NIKOLIC-PATERSON DJ, Yu Y, ATKINS RC: Regulation of glomerular

terization, and propagation in vitro of human glomerular endothelial

yearn-i and elam-1 gene expression by inflammatory stimuli. (ab-

10. STRIKER GE, SODERLAND C, BOWEN-POPE DF, GOWN AM, SCHMER

cells. J Exp Med 160:323—328, 1984

11. BALLERMANN BJ: Regulation of bovine glomerular endothelial cell growth in vitro. Am J Physiol 256:C182—C189, 1989 12. GREEN DF, HWANG KH, RYAN US, BOURGOIGNIE JJ: Culture of

stract) JAm Soc Nephrol 3:639, 1992 33. BRISCOE DM, COTRAN RS: Role of leukocyte-endothelial cell adhe-

sion molecules in renal inflammation: In vitro and in vivo studies.

endothelial cells from baboon and human glomeruli. Kidney mt

Kidney mt 44:(Suppl 42):S27—S34, 1993 34. Orr MJ, OLSON JL, BALLERMANN BJ: Phenotypic differences between

41:1506—1516, 1992 13. OBRIG TG, LOUISE CH, LINGw0OD CA, BOYD B, BARLEY-MALONEY

glornerular capillary (GE) and aortic (AE) endothelial cells in vitro. (abstract) JAm Soc Nephrol 4:564, 1993

L, DANIEL TO: Endothelial heterogeneity in Shiga toxin receptors and responses. J Biol Chem 268:15484—15488, 1993 14. MAClAG T, CERUNDOLO J, ILSLEY S, KELLEY PR, FORAND R: An endothelial cell growth factor from bovine hypothalamus: identification and partial characterization. Proc NatI Acad Sci USA 76:5674— 5678, 1979 15. CUI YC, TAI PC, GAVFER KC, MASON DY, SPRY CJ: A vascular

35. OBRIG TG, DEL-VECCHIO PJ, BROWN JE, MORAN TP, Rowui'tn BM,

endothelial cell antigen with restricted distribution in human foetal, adult and malignant tissues. Immunology 49:183—189, 1983 16. LAMPUGNANI MG, RESNATI M, RAITERI M, PIGOTF R, PISACANE A,

HOUEN G, RUCO LP, DEJANA E: A novel endothelial-specific membrane protein is a marker of cell-cell contacts. J Cell Biol 118:1511— 1522, 1992 17. VAN MUIJEN GN, RUITER DJ, WARNAAR SO: Coexpression of inter-

mediate filament polypeptides in human fetal and adult tissues. Lab Invest 57:359—369, 1987

18. LEEUWENBERG JF, JEUNHOMME TM, BUURMAN WA: Role of ELAM-1 in adhesion of monocytes to activated human endothelial cells. Scand J Immunol 35:335—341, 1992 19. SALACINSKI PR, MCLEAN C, SYKES JE, CLEMENT-JONES VV, LOWRY

PJ: lodination of proteins, glycoproteins, and peptides using a solidphase oxidizing agent, 1,3,4,6-tetrachloro-3 alpha,6 aipha-diphenyl glycoluril (Lodogen). Anal Biochem 117:136—146, 1981

20. SCATCHARD G: The attractions of proteins for small molecules and ions. Ann NYAcad Sci 51:660—672, 1946 21. NUDELMAN E, KANNAGI R, HAKOMORI 5, PARSONS M, LIPIN5KI M,

WIELS J, FELLOUS M, TURSZ T: A glycolipid antigen associated with

Burkitt lymphoma defined by a monoclonal antibody. Science 220: 509—511, 1983 22. LINGw00D CA, LAW H, RICHARDSON 5, PETRIC M, BRUNTON JL,

DE-GRANDIS 5, KARMALI M: Glycolipid binding of purified and recombinant Escherichia coli produced verotoxin in vitro. J Biol Chem 262:8834—8839, 1987 23. STERZEL RB, LOVEEr DH, FOELLMER HG, PERFETFO M, BIEMESDER-

FER D, KASHGARIAN M: Mesangial cell hillocks. Nodular foci of exaggerated growth of cells and matrix in prolonged culture. Am J Pathol 125:130—140, 1986 24. DETRISAC CJ, SENS MA, GARVIN AJ, SPICER SS, SENS DA: Tissue

culture of human kidney epithelial cells of proximal tubule origin. Kidney mt 25:383—390, 1984 25. MORAN T, LINGWOOD C, BRANCA A, DEL VECCHIO P, BROWN J,

OBRIG T: Analysis of shiga toxin receptors on human vascular endothelial cells. Abstract Annual Meeting of the American Society of Microbiology 46:B93, 1989

JUDGE TK, ROTHMAN SW: Direct cytotoxic action of Shiga toxin on human vascular endothelial cells. Infect Immun 56:2373—2378, 1988 36. PUDYMAITIS A, LINGWOOD CA: Susceptibility to verotoxin as a function of the cell cycle. J Cell Physiol 150:632—639, 1992 37. HOBSON B, DENEKAMP J: Endothelial proliferation in tumours and normal tissues: Continuous labelling studies. BrJ Cancer 49:405—413, 1984 38. PABsT R, STERZEL RB: Cell renewal of glomerular cell types in normal rats. An autoradiographic analysis. Kidney list 24:626-631, 1983 39. JACKSON MP, NEWLAND JW, HOLMES RK, O'BRIEN AD: Nucleotide

sequence analysis of the structural genes for Shiga-like toxin I encoded by bacteriophage 933J from Escherichia coli. Microb Pathog 2:147—153, 1987

40. LOUISE CB, OBRIG TG: Specific interaction of Escherichia coli 0157:

H7-derived Shiga-like toxin II with human renal endothelial cells. JlnfectDis 172:1397—1401, 1995 41. STROCKBINE NA, MARQUES LR, NEWLAND JW, SMITH HW, HOLMES

RK, O'BRIEN AD: Two toxin-converting phages from Escherichia coli O157:H7 strain 933 encode antigenically distinct toxins with similar biologic activities. Infect Immun 53:135—140, 1986 42. WADDELL T, HEj.o S, PETRIC M, COHEN A, LINGWOOD C: Globotrio-

syl ceramide is specifically recognized by the Escherichia coli verocytotoxin 2. Biochem Biophys Res Commun 152:674—679, 1988 43. Hixr SC, KARMALI MA, LINGWOOD CA: Preparation of VT1 and VT2 hybrid toxins from their purified dissociated subunits. Evidence for B subunit modulation of a subunit function. J Biol C/scm 266:3617— 3621, 1991 44. TESH VL, BURRIS JA, OWENS JW, GORDON VM, WADOLKOWSKI EA,

O'BRIEN AD, SAMUEL JE: Comparison of the relative toxicities of Shiga-like toxins type I and type II for mice. Infect Immun 61:3392— 3402, 1993 45. HAREL Y, SILVA M, GIR0IR B, WEINBERG A, CLEARY TB, BEUTLER B:

A reporter transgene indicates renal-specific induction of tumor necrosis factor (TNF) by shiga-like toxin. Possible involvement of TNF in hemolytic uremic syndrome. J am Invest 92:21 10—21 16, 1993 46. TESIJ VL, RAMEGOWDA B, SAMUEL JE: Purified Shiga-like toxins induce expression of proinflammatory cytokines from murine peritoneal macrophages. Infect Immun 62:5085—5094, 1994 47. VAN SEn-EN PA, MONNENS LAFI, VERSTRATEN RGG, VAN DEN HEUVEL LPWJ, VAN HINSBERGH VWM: Effects of verocytotoxin-1 on

nonadherent human monocytes: Binding characteristics, protein synthesis, and induction of cytokine release. Blood 88:174—183, 1996 48. FITZPATRICK MM, SHAH V, TROMPETER RS, DILLON MJ, BARRATI-

26. COHEN A, MADRID-MARINA V, ESTROV Z, FREEDMAN MH, LING-

TM: Interleukin-8 and polymorphoneutrophi! leucocyte activation in

WOOD CA, DOSCH HM: Expression of glycolipid receptors to Shiga-

hemolytic uremic syndrome of childhood. Kidney Int 42:951—956, 1992 49. SIEGLER RL, EDWIN SS, CHRISTOFFERSON RD, MITCHELL MD:

like toxin on human B lymphocytes: A mechanism for the failure of

1256

van Setten Ct aL Effects of verotoxin and TNFa on GMVEC

Plasma and urinary cytokines in childhood hemolytic uremic syndrome. (abstract) JAm Soc Nephrol 2:274, 1991 50. KARPMAN D, ANDREASSON A, TIIYSELL H, KAPLAN BS, SVANBORG C:

Cytokines in childhood hemolytic uremic syndrome and thrombotic thrombocytopenic purpura. Pediatr Nephrol 9:694— 699, 1995 51. VAN DE KAR NCAJ, SAUERWEIN RW, DEMACKER PNM, GRAU GE, VAN HINSBERGH VWM, MONNENS LAH: Plasma cytokine levels in

hemolytic uremic syndrome. Nephron 71:309—313, 1995 52. ABBOVr F, TAM FW, RYAN JJ, REvs AJ: Human mesangial cells

synthesize interleukin 1 alpha but not interleukin 1 beta, interleukin 1 receptor antagonist, or tumor necrosis factor. Nephrol Dial Transplant 7:997—1001, 1992 53. ABBoTT F, RYAN JJ, CESKA M, MATSUSWMA K, SARRAF CE, REvS AJ:

Interleukin-i beta stimulates human mesangial cells to synthesize and release interleukins-6 and -8. Kidney In! 40:597—605, 1991

54. VON ASMUTH EJ, DENTENER MA, CESKA M, BUURMAN WA: IL-6,

IL-S and TNF production by cytokine and lipopolysaccharide-stimulated human renal cortical epithelial cells in vitro. Eur ('ytokine Netw 5:301—310, 1994

55. LiNowooD CA: Verotoxin-binding in human renal sections. Nephron 66:21—28, 1994 56. BUSTOS C, GONZALEZ E, MULEY R, ALONSO JL, EGIDO J: Increase of

tumour necrosis factor alpha synthesis and gene expression in periph-

eral blood mononuclear cells of children with idiopathic nephrotic syndrome. EurJ Clin Invest 24:799—805, 1994 57. TAKEMURA T, YosH1oi K, MURAKAMI K, AKANO N, OKADA M, AYA N, MAKI S: Cellular localization of inflammatory cytokines in human

glomerulonephritis. Virchows Arch 424:459—464, 1994 58. SAXENA 5, MITrAL A, ANDAL A: Pattern of interleukins in minimal-

change nephrotic syndrome of childhood. Nephron 65:56—61, 1993