Comparison of the Effects of Shiga-like Toxin 1 on Cytokine- and Butyrate-Treated Human Umbilical and Saphenous Vein Endothelial Cells

Gerald T. Keusch 0 1 2 David W. K. Acheson 0 1 2 Leonie Aaldering 0 1 2 John Erban 0 1 2 Mary S. Jacewicz 0 1 2 0 The Journal of Infectious Diseases 1996; 173:1164-70 1996 by The University of Chicago. All rights reserved. 0022-1899/96/7305 -00 15$0 1.00 1 Received 26 June 1995; revised 6 November 1995. Presented in part: 2nd International Symposium and Workshop on Verocyto toxin (Shiga-like toxin)-producing Escherichia coli infections , Bergamo, Italy, 27-30 June , 1994. Financial support: National Institutes of Health (AI-07329 and AI-14242, to G.T.K., and DK-34928, to A. V. Kane [New England Medical Center Gastroenterology Research Center Core Microbiology Laboratory, Boston], who purified Stx toxins used in this study) and Rockefeller Foundation (Health Sciences for the Tropics Program, to G.T.K.). tute , 750 Washington St., Box 041, Boston, MA 02111 2 Tupper Research Institute, Division of Geographic Medicine and Infectious Diseases, New England Medical Center , Boston, Massachusetts To examine the reported heterogeneity of endothelial cells to Shiga-like toxin 1 (Stxl), the responses of human umbilical (HUVEC) and saphenous (HSVEC) vein endothelial cells to cytokines, butyrate, and toxin were compared. Untreated HSVEC were generally more susceptible than were HUVEC to Stxl; pretreatment of either cell with lipopolysaccharide, interleukin-lfJ, or tumor necrosis factor-a enhanced Stxl toxicity. Dexamethasone alone increased total globotriaosylceramide (Gb3) content and toxin binding but inhibited cytokine-enhanced cytotoxicity, whereas the differentiation agent, sodium butyrate, increased both Gb3 content and cytotoxicity responses to Stxl, most prominently in HSVEC. Stxl toxicity directly correlated with the release of von Willebrand factor from HSVEC but not from HUVEC. Thus, HUVEC and HSVEC exhibit distinctive responses to Stxl, cytokines, and butyrate. This suggests the need for caution in extrapolating from in vitro studies utilizing one endothelial cell type to in vivo events during pathogenesis of Stxmediated thrombotic microangiopathies. - Infections due to organisms producing cytotoxins of the Shiga toxin family, including Shigella dysenteriae type I, which produces Shiga toxin (Stx), the prototype Shiga family toxin, and certain Escherichia coli serotypes, which produce Shiga-like toxin 1 or 2 (Stxl or 2, also known as SLT-I or -II), are associated with bloody diarrhea and systemic compli cations, including hemolytic-uremic syndrome and thrombotic thrombocytopenic purpura [I]. These complications are charac terized by a thrombotic microangiopathy and hemolytic anemia [2] thought to be secondary to endothelial cell damage initiated by systemic toxin, perhaps in concert with lipopolysaccharide (LPS) endotoxin [3]. There is little direct evidence to support this hypothesis, as circulating and tissue-fixed toxin have not been detected in patients. However, Shiga family toxins are reported to be di rectly cytotoxic in vitro to human umbilical vein endothelial cells (HUVEC) [4] and human glomerular endothelial cells [5]. HUVEC are not very sensitive to toxin, as isolated, but can be induced to express the toxin glycolipid receptor, globotriao sylceramide (Gb3), and become sensitized to Stxl by preincu bation with LPS or the LPS-inducible cytokines interleukin (IL)-1(3 and tumor necrosis factor (TNF)-a [6, 7]. In contrast, human glomerular cells appear to constitutively express higher levels of Gb3, are not induced by cytokines, and are consider ably more responsive to Stxl than are HUVEC [5]. A major concern regarding the use of HUVEC as a model for studying the pathogenesis of thrombotic microangiopathies is that they are derived from large blood vessels rather than from arteriolar or capillary endothelial cells, the sites of pathol ogy [2]. There is abundant evidence that the properties of endo thelial cells from different vascular beds differ significantly from one another [8-I 0]. That this heterogeneity may typify the response to Stx as well is suggested by reported differences in the response of HUVEC and human glomerular endothelial cells [4, 5]. However, microvascular endothelial cells are more difficult to obtain and maintain, The present study was designed to further assess the hetero geneity of readily available endothelial cells by comparing the responses of HUVEC and another large-vein endothelial cell population, human saphenous vein endothelial cells (HSVEC), to Stx 1 and inducers of the toxin receptor in other cells, includ ing cytokines and butyrate. Materials and Methods Toxin preparation, cytotoxicity, and binding assay. Stx1 was produced and purified from sonic lysates of E. coli HB 101 lysoge nized with bacteriophage H19B [11], and it was labeled with 1251 by a modification of the chloramine T method, which does not alter the specific activity of the toxin effect on protein synthesis [12]. Stx2e was purified from E. coli HBI0l(pDLW5) (provided by A. O'Brien, Uniformed Services University, Bethesda, MD) as previously described [13]. Specific radioactivity of labeled toxin preparations ranged from 20,000 to 40,000 cpm/ng of toxin protein for each toxin in different iodinations. Stxl Effects on Endothelial Cells 1165 Cytotoxicity (0/0 inhibition of leucine incorporation) HUVEC Toxin cytotoxic activity, defined as the inhibition of protein synthesis due to the toxin, was assessed by measuring [3H]leucine incorporation into trichloroacetic acid (TCA)-precipitable protein, as previously described [14], and is reported as the percentage of leucine incorporation compared with that of control cells not ex posed to toxin. In brief, endothelial cells were treated for 3- 24 h with IO-fold dilutions of toxin in medium, followed by 30 min in medium containing 1 J.lCi/iOO J.lL [3H]leucine. TCA-precipitable protein was harvested, and radioactivity was determined. In some .experiments, however, we estimated the concentration of toxin causing 50% inhibition of leucine incorporation directly from the dose-response curve of leucine incorporation and designated this the ICso. Binding of 12sI-labeled StxI was always measured follow ing exposure of cells to labeled toxin for 1 h at 4C, as previously described [14]. Endothelial cells. HUVEC were harvested from cannulated human umbilical cord veins by pronase digestion and maintained in McCoy's 5-A medium (modified) containing 20% fetal bovine serum (FBS), 100 J.lg/mL heparin, 50 J.lg/mL endothelial cell growth factor, and antibiotics [15]. HSVEC (provided by A. King, Division of Nephrology, New England Medical Center) were re leased from saphenous vein segments, which had been discarded after coronary bypass operations, by collagenase treatment and maintained in medium-I99 with 10% FBS and additives, as de scribed above [16]. Cells were grown in gelatin-coated flasks and used between passages 2-4. In all experiments, triplicate endothe lial cell monolayers were used, and the mean value ( 1 SD) of the (...truncated)