Infection with verocytotoxin (VT; Shiga-like toxin)-producing Escherichia coli (VTEC), usually serotype O157, is the major cause of hemolytic uremic syndrome (HUS). Children and the older population appear to be the most vulnerable to this type of infection. VT-1 and VT-2, part of the family of Shiga toxins, are implicated in the pathogenesis of the vascular complications of VTEC infection, including glomerular thrombotic microangiopathy (TMA), which is the hallmark of HUS. However, pathogenic details are incomplete, and research is hampered by the paucity of laboratory models of the condition.
In order to gain cellular uptake, VT requires the receptor Gb3 (CD77 or the blood group Pk), which is expressed in human gut and kidney1. The different expression of Gb3 across species is the likely reason why commonly available laboratory animals do not develop HUS after VT exposure. We have overcome this restriction by using the plant toxin ricin. Ricin, like VT, intoxicates cells by inhibition of protein synthesis and has identical enzymatic activity in cleaving adenine from ribosomal RNA at a point in which aminoacyl RNA is assembled2. However, unlike VT, ricin gains cell entry through a variety of galactose-bearing surface receptors using clathrin-dependent and -independent pathways3. Virtually all mammalian cells have a surface galactose-containing glycolipid or glycoprotein, and therefore, ricin receptors are universally distributed. The clathrin-dependent endocytic pathway is generally regarded as essential for cell viability, whereas the clathrin-independent pathway is thought to be widely distributed but not ubiquitous. We have confirmed that on vero cells, the two toxins have similar effects4.
In VTEC-induced HUS, there is evidence of an acute inflammatory response with neutrophilia5, raised concentrations of C-reactive protein, and plasma elastase6. The magnitude of these responses predicts adverse outcome. Raised concentrations of the cytokines tumor necrosis factor-
(TNF-), interleukin (IL)-1
, IL-6, and the chemokines IL-87,8,9,10 and monocyte chemotactic protein-1 (MCP-1)11 have been reported. Although the cause for this is unknown, it may be relevant that VT superinduces the production of cytokines from a variety of cells12,13. Ricin also causes the superinduction of proinflammatory cytokines from human14 and rat monocytes (abstract; BAVARI S et al, Toxicologist 12:1126, 1992).
We report that ricin administered to rats, with or without simultaneous injection of lipopolysaccharide (LPS), rapidly and reliably produced glomerular TMA, which closely resembled that found in children with diarrhea-associated HUS. The animals exhibited oliguric renal failure, hemolysis, and thrombocytopenia. The model was accompanied by an inflammatory response with a marked increase in the plasma concentrations of inflammatory cytokines and a glomerular infiltrate of macrophages.
METHODS
Experimental protocol
Young adult male rats (Sprague-Dawley strain, weight range 260 to 300 g) were lightly anesthetized with ether and flexible polypropylene cannulae (Portex PP25) implanted in a tail artery and vein. The animals were placed in restraining cages and were allowed to recover from the anesthetic. An initial blood sample was taken from the arterial cannula, and the animals received, via the venous cannula, a bolus of (0.5 ml) 0.7% NaCl solution (control group) or a similar volume of NaCl solution containing LPS (E. coli 0111:B4, 10 
g/100 g body wt; Sigma, Poole, Dorset, UK), ricin, a gift from Professor J.M. Lord (University of Warwick, Warwickshire, UK; 6.7 g/100 g body wt), or a combination of LPS and ricin at these doses.
An infusion of 0.7% NaCl was commenced, via the tail vein, at 6 ml/hr. This was continued for six hours or until the animals were volume expanded by more than 20 ml. Urine was collected by spontaneous voiding, and blood samples (2 ml) were taken at three and five hours. At a minimum of seven hours post-toxin administration, a terminal 2 ml blood sample was obtained, and the animals were anesthetized via the tail vein (sodium pentobarbitone; Sagatal, Rhone Merieux, Dublin, Ireland). After the abdomen was opened, the renal vein was punctured, and the animal was perfusion fixed by flushing the circulation, via the tail artery, with 2 
10 ml 0.7% NaCl followed by 10 ml formol saline. The kidneys were excised for histologic examination, together with samples of lung, liver, heart, muscle, gut, spleen, pancreas, and brain.
A separate series of experiments was performed to measure the effects of the toxins on glomerular filtration rate (inulin clearance), uncomplicated by the effects of the large blood sample volumes mentioned earlier in this article. In these experiments, the protocol was similar, except that the initial boluses were given i.p. instead of i.v. and the infusate included inulin (1.5% wt/vol). Blood samples (0.5 ml) were taken from the arterial line for inulin assay at two-hour intervals. When not being used for sampling, the arterial line was used to monitor blood pressure. Urine was collected every two hours by spontaneous voiding. The effects of the toxins were identical (as assessed by blood pressure, body temperature, and histology) whether administration was i.p. or i.v.; therefore, where appropriate, the data from the two series have been pooled. The urine volumes were determined, and the urine osmolality was measured prior to inulin assay.
Hematological parameters were measured in the hematology laboratory at the Birmingham Children's Hospital using a Coulter counter.
Hemolysis
Plasma was obtained from treated rats at eight hours. This was diluted 10-fold with 0.7% NaCl, and the absorbance was measured at 555 nm. The results were compared with standards to determine the degree of hemolysis. To examine the effect of hemolysis alone, 0.5 ml of blood was removed from untreated rats, hemolyzed, and administered to animals. This represented 3% hemolysis. The hemolyzed blood was given as three aliquots at three, four, and five hours into a six-hour infusion of 0.7% saline and inulin.
Histology
After perfusion fixation, the kidneys were removed, bisected longitudinally, and placed in formol saline for 24 hours. These were embedded in paraffin, sectioned, and stained by hematoxylin and eosin and periodic acid-methenamine silver. Sections were also stained for fibrin using an immunoperoxidase method. Endogenous peroxidase was blocked with hydrogen peroxide 0.5% in methanol. Sections were exposed to goat antirat fibrinogen (Organon Teknika Ltd., Cambridge, UK) at 1:100. Donkey antisheep peroxidase-conjugate (1:100) and tetra-amino-biphenyl hydrochloride (diaminobenzidine) were used to visualize binding and sections counter-stained with Mayer's hemalum. Glomeruli showing thrombi by fibrinogen staining were counted and expressed as a percentage (N = 100). Tissues were stained for the presence of macrophages using an anti-ED1 antibody (Serotec, Ltd., Oxford, UK). The mouse antirat ED1 was used at 1:50 and visualized using rabbit antimouse (1:25), alkaline phosphatase antialkaline phosphatase (1:50), and the substrate fast red. The number of ED1-positive cells were counted in 100 glomeruli.
Fluorescent staining
Frozen sections of rat kidney were air dried and blocked with 0.1% gelatin in phosphate-buffered saline (PBS). Five g/ml of ricin-fluorescein isothiocyanate (FITC; Sigma) was applied for one hour at room temperature in a humid environment in the dark. After washing with PBS, sections were fixed with 2% formaldehyde/PBS for 20 minutes and were then mounted in 75% glycerol in 10 mM Tris-buffered saline, pH 8.6.
Electron microscopy
A few small pieces were taken from the cortex after perfusion, placed in glutaraldehyde 2.5% in PBS, and then fixed further in osmium tetroxide. These were embedded in Araldite, sectioned, stained with uranyl acetate and lead citrate, and examined using an electron microscope.
Cytokine/chemokine assays
Levels of plasma cytokines were measured by enzyme-linked immunosorbent assay (Biosource International, Lifescreen, Ltd., Hertfordshire, UK).
RESULTS
The exposure of rats to either saline or LPS alone did not induce glomerular thrombosis. However, when animals were treated with ricin or ricin plus LPS, glomerular thrombosis was marked Figure 1a, and pyknotic cell nuclei were seen in endocapillary sites Figure 1b. Thrombosis was confined to the glomerular microcirculation with occasional extension into the afferent arteriole. The percentage of glomeruli showing thrombi by fibrinogen staining was control 2.67 
1.2, LPS 1.86 0.46, ricin 94.57 3.0, and LPS + ricin 86.33 4.26 (mean SEM, N = 4). Thrombi were not seen in lung, liver, heart, muscle, gut, spleen, pancreas, or brain of affected animals.
Figure 1.
(A) Glomeruli from rat treated with lipopolysaccharide (LPS) and ricin stained with an antifibrinogen antibody that has specifically detected thrombosis in glomerular capillaries (1600). (B) Glomerulus from rat treated with ricin showing pyknotic nuclei in an endocapillary distribution (2400). (C) Frozen section of cortex treated with ricin-fluorescein isothiocyanate (FITC). The strongest binding is to distal tubules, particularly their basement membrane, and there is also binding to glomerular and intertubular capillaries, walls of arterioles, and the brush border of proximal tubules (1600). Publication of this figure in color was made possible by grants from Fujisawa Ltd. and Roche Pharmaceuticals UK.
By electron microscopy, glomerular capillaries contained fibrin thrombi interspersed with platelets and areas of endothelial detachment from the basement membrane Figure 2. Rats treated with ricin had an increased number of ED1-positive cells in their glomeruli. The number of ED1-positive cells per glomerulus was control 0.59 0.1, LPS 0.74 0.21, ricin 3.53 0.27 (P < 0.005), and LPS + ricin 1.37 0.35 (mean SEM, N = 4). No apparent increase in ED1 staining was observed in any of the other organs examined.
Figure 2.
(A) Glomerular capillary loop showing fibrin thrombus and platelets (4700). (B) A capillary showing detachment of endothelial cells from basement membrane and early blebbing of the nucleus (4700). Publication of this figure in color was made possible by grants from Fujisawa Ltd. and Roche Pharmaceuticals UK.
Ricin-FITC when applied to frozen sections of rat kidney bound most strongly to distal tubules, particularly their basement membrane, and also to glomerular and intertubular capillaries, walls of arterioles, and the brush border of proximal tubules Figure 1c.
Decreases in blood pressure were observed only in ricin- and ricin plus LPS-treated rats Figure 3a. This commenced at five hours and continued to the end point of the experiment. The glomerular filtration rate (GFR) decreased progressively in these animals from three to seven hours but not in the rats receiving LPS alone Figure 3b.
Figure 3.
(A) Alterations in blood pressure over time in LPS (
), ricin (
), and LPS plus ricin (
) treated rats. (B) The glomerular filtration rate at four and seven hours in LPS (), ricin plus LPS (
), and ricin () treated rats. Data are mean SEM; N = 3; *P < 0.05, ***P < 0.001.
The administration of ricin with or without LPS induced thrombocytopenia and anemia Table 1. Schistocytes were not evident on a blood film (data not shown); however, there was visible hemolysis in plasma, calculated to be 3% at eight hours and visible hemoglobinuria from three hours. LPS alone did not have this effect. In separate experiments on normal animals, intravenous administration of hemolyzed blood elicited transient hemoglobinuria; however, glomerular filtration rate and urine flow remained normal, and histologic examination showed no evidence of glomerular thrombosis. Thus, the hemolysis induced by ricin did not itself promote the lesions of this model. Ricin alone caused an increase in peripheral blood total leukocyte count, whereas ricin with LPS together caused a significant decrease.
Ricin alone gave an early and sustained increase in MCP-1 and TNF-, whereas IL-1, IL-6, and interferon-
(IFN-) increased more slowly but were significant by seven hours Figure 4. LPS alone gave an early and transient rise in TNF-, MCP-1, and IFN- with a decrease from three hours. Synergism was observed between LPS and ricin in the responses of IL-1, IL-6, and IFN-, whereas the effect on TNF- and MCP-1 appeared to be additive. The terminal concentrations were similar when LPS was added, while with saline there was no increase in any of the cytokines measured.
Figure 4.
Plasma concentrations of (A) tumor necrosis factor-
, (B) interleukin-1
, (C) interleukin-6, (D) macrophage chemotactic protein-1, and (E) interferon-
. Data are mean SEM, N = 3; *P < 0.05 by the Mann–Whitney test. Symbols are: () ricin; () LPS + ricin; () LPS; (
) control.
DISCUSSION
The need for a model with which to explore the pathogenesis of VTEC-induced HUS is widely recognized (report of World Health Organization Working Group Meeting on Shiga-like Toxin-producing E. coli, with an emphasis on zoonotic aspects, 1994). However, the models described so far are varied, perhaps because the distribution of VT receptors is different between species. For example, an injection of VT in the rabbit causes focal endothelial lesions in the central nervous system, a site where Gb3 has been identified16, whereas VTEC infection in mice has been shown to give rise to an interstitial renal lesion without TMA17. Given that the distribution of Gb3 in primates may more closely resemble humans, Siegler et al administered Shiga toxin to baboons (abstract; Siegler et al, J Am Soc Nephrol 7:1834, 1996). This gave rise to gut and renal lesions, but glomerular thrombosis was scanty. Another possible model is "Alabama Rot," a naturally occurring VTEC-related disease of greyhounds. In this, dogs develop bloody diarrhea, edema, ulcerating skin lesions, and renal failure. Histologically, the kidney shows extensive glomerular TMA18. To bypasses the species restriction of VT receptors, we have used the ribotoxin ricin, which has identical protein synthesis inhibitory effects to VT2.
The pathogenesis of glomerular TMA is unknown. It has been suggested that the predilection for glomerular thrombosis in humans relates to the glomerular expression of Gb31 and direct vascular toxicity. The ricin model throws doubt on this concept. Ricin-FITC binds to a number of different structures in the renal cortex Figure 1c, illustrating that receptor distribution does not account for the specificity of the glomerular TMA; for example, there is obvious binding of ricin to arteriolar endothelium, but thrombi were rarely found in this site. It must be remembered, however, that applying ricin to sections is an artificial condition, as in vivo exposure would be via the blood stream. Ricin is delivered to liver or lung during first pass after intraperitoneal or intravenous administration, respectively. Previous studies have shown the toxin to be widely distributed, the majority being found in the liver (46%), spleen (9.9%), and muscle (13%), with less than 2% recoverable in the kidney19. It is therefore of interest that within the time frame of the experiment, ricin primarily gives rise to glomerular thrombosis. As in human HUS, extensive fibrin thrombi fill the capillaries that appear engorged and occluded. The obstruction to the microcirculation in the model is sufficient to explain renal failure. Occasionally, we saw thrombus extending into the afferent artery, as recently described in a postmortem study in children20. Although our model has the characteristic clinical triad of HUS, oliguric renal failure, thrombocytopenia, and hemolysis, fragmented erythrocytes were not seen. Very little is known about the mechanism that produces schistocytes in HUS, making it difficult to comment on this difference.
Children with VTEC infection and diarrhea-associated HUS have been shown to have an increase in the cytokines TNF-, IL-1 and IL-6, and the chemokines IL-8 and MCP-1 in plasma and/or urine. Similarly, in our model, there was a marked increase in these inflammatory mediators, the concentrations far exceeding those induced by LPS. The synergistic response observed between ricin and LPS for IL-1, IL-6, and IFN- indicates that different signaling pathways are involved. In that the shallow rise in IFN- occurred late in the course of the experiment after glomerular thrombosis had begun, it is unlikely that this cytokine participates in the model. The pattern of cytokines MCP-1, > TNF-, > IL-1 indicates the importance of monocytes. Verocytotoxin is known to cause superinduction of human monocytes to secrete TNF-, IL-1, IL-6, and IL-8 over a prolonged period in vitro13, and ricin has been shown to behave similarly on human peripheral blood mononuclear cells14. We have also demonstrated an infiltration of monocytes/macrophages into the glomeruli of the model.
There are similarities between our model and others that give rise to glomerular thrombosis such as the generalized Shwartzman reaction and accelerated nephrotoxic serum nephritis. These reactions are leukocyte and complement dependent21 and involve proinflammatory cytokines22,23, up-regulation of adhesion molecules24, and tissue factor25. The thrombotic lesion can be prevented by combined the administration of TNF- receptor antagonist and soluble IL-1 receptor in the nephrotoxic serum nephritis model26. In the Shwartzman reaction, the challenge dose of LPS can be by replaced by TNF- and IL-122, whereas anti-TNF- antibody administered prior to the priming injection gives partial protection against fibrin deposition27. The similarity of the proinflammatory cytokine response in all three models of glomerular capillary thrombosis suggests a common pathogenic pathway. The ricin model has the potential to elucidate these mechanisms, as well as providing a framework for investigating therapeutic strategies for HUS.
References
| 1. | Lingwood CA. Verotoxin-binding in human renal sections. Nephron 1994; 66: 21−28. | PubMed | ChemPort | |
| 2. | Endo Y & Tsurgi K. RNA N-glycoside activity of ricin A-chain: Mechanism of action of the toxic lectin ricin on eukaryotic ribosomes. J Biol Chem 1987; 363: 812−813. |
| 3. | Sandvig K & Van Deurs B. Endocytosis, intracellular transport, and cytotoxic action of Shiga toxin and ricin. Physiol Rev 1996; 76: 949−966. | PubMed | ISI | ChemPort | |
| 4. | Williams JM, Lea N, Lord JM, Roberts LM, Milford DV & Taylor CM. Comparison of ribosome-inactivating proteins in the induction of apoptosis. Toxicol Lett 1997; 91: 121−127. | Article | PubMed | ChemPort | |
| 5. | Walters MS, Matthei IU, Kay R, Dillon MJ & Barratt TM. The polymorphonuclear leukocyte count in childhood haemolytic uraemic syndrome. Pediatr Nephrol 1989; 3: 120−134. |
| 6. | Milford DV, Staten J, MacGreggor I, Dawes J, Taylor CM & Hill FG. Prognostic markers in diarrhoea-associated haemolytic uraemic syndrome: Initial neutrophil count, human neutrophil elastase and von Willebrand factor antigen. Nephrol Dial Transplant 1991; 6: 232−237. | PubMed | ChemPort | |
| 7. | Fitzpatrick MM, Shah V, Trompeter RS, Dillon MJ & Barratt TM. Interleukin-8 and polymorphoneutrophil leukocyte activation in hemolytic uremic syndrome of childhood. Kidney Int 1992; 42: 951−956. | PubMed | ChemPort | |
| 8. | van de Kar NCAJ, Sauerwein RW, Demacker PNM, Grau GE, van Hinsbergh VWM & Monnens LAH. Plasma cytokine levels in hemolytic uremic syndrome. Nephron 1995; 71: 309−313. | PubMed | ChemPort | |
| 9. | Karpman D, Andreasson A, Thysell H, Kaplan BS & Svanborg C. Cytokines in the childhood hemolytic uremic syndrome and thrombotic thrombocytopenic purpura. Pediatr Nephrol 1995; 9: 694−699. | Article | PubMed | ChemPort | |
| 10. | Inward CD, Varagunam M, Adu D, Milford DV & Taylor CM. Cytokines in haemolytic uraemic syndrome associated with verocytotoxin-producing Escherichia coli infection. Arch Dis Child 1997; 77: 145−147. | PubMed | ChemPort | |
| 11. | van Setten PA, van Hinsberg VWM, van den Heuvel LPWJ, Preyers F, Dijkman HBPM, Assman KJM, van der Velden TJAM & Monnens LAH. Monocyte chemoattractant protein-1 and interleukin-8 levels in urine and serum of patents with hemolytic uremic syndrome. Pediatr Res 1998; 43: 759−767. | PubMed | ChemPort | |
| 12. | Tesh VL, Ramegowda B & Samuel JE. Purified Shiga-like toxins induce expression of proinflammatory cytokines from murine peritoneal macrophages. Infect Immun 1994; 62: 5085−5094. | PubMed | ChemPort | |
| 13. | van Setten PA, Monnens LAH, Verstraten RGG, van den Heuval LPWJ & van Hinsberg VWM. Effects of verocytotoxin-1 on nonadherent human monocytes: Binding characteristics, protein synthesis and induction of cytokine release. Blood 1996; 88: 174−183. | PubMed | ChemPort | |
| 14. | Licastro F, Morini MC, Bolognesi A & Stirpe F. Ricin induces the production of tumour necrosis factor- and interleukin-1 by human peripheral blood mononuclear cells. Biochem J 1993; 294: 517−520. | PubMed | ChemPort | |
| 15. | Lillie LE, Temple NJ & Florence LZ. Reference values for young normal Sprague-Dawley rats: Weight gain, hematology and clinical chemistry. Hum Exp Toxicol 1996; 15: 612−616. | PubMed | ChemPort | |
| 16. | Richardson SE, Rotman TA, Jay V, Smith CR, Becker LE, Petric M, Olivieri NF & Karmali MA. Experimental verocytotoxemia in rabbits. Infect Immun 1992; 60: 4154−4167. | PubMed | ChemPort | |
| 17. | Karpman D, Connell H, Svensson M, Scheutz F, Alm P & Svanborg C. The role of lipopolysaccharide and Shiga-like toxin in a mouse model of Escherichia coli O157:H7 infection. J Infect Dis 1997; 175: 611−620. | PubMed | ChemPort | |
| 18. | Carpenter JL, Andelman NC, Moore FM & King NW. Idiopathic cutaneous and renal glomerular vasculopathy of greyhounds. Vet Pathol 1988; 25: 401−407. | PubMed | ChemPort | |
| 19. | Ramsden CS, Drayson MT & Bell EB. The toxicity, distribution and excretion of ricin holotoxin in rats. Toxicology 1989; 55: 161−171. | Article | PubMed | ChemPort | |
| 20. | Inward CD, Howie AJ, Fitzpatrick MM, Rafaat F, Milford DV & Taylor CM. Renal histopathology in fatal cases of diarrhoea-associated haemolytic uraemic syndrome. Pediatr Nephrol 1997; 11: 556−559. | Article | PubMed | ChemPort | |
| 21. | Vedanarayanan VV, Kaplan BS & Fong JSC. Neutrophil function in an experimental model of hemolytic uremic syndrome. Pediatr Res 1987; 21: 252−257. | PubMed | ChemPort | |
| 22. | Ozmen L, Pericin M, Hakimi J, Chizzonite RA, Wysocka M, Trinchieri G, Gately M & Garotta G. Interleukin 12, interferon  and tumor necrosis factor are the key cytokines of the generalised Shwartzman reaction. J Exp Med 1994; 180: 907−915. | Article | PubMed | ChemPort | |
| 23. | Movat HZ, Burrowes CE, Cybulsky MI & Dinarello CA. Acute inflammation and a Schwartzman-like reaction induced by interleukin 1 and tumor necrosis factor: Synergistic action of the cytokines in the induction of inflammation and microvascular injury. Am J Pathol 1987; 129: 463−476. | PubMed | ChemPort | |
| 24. | Koide N, Abe K, Narita K, Kato Y, Sugiyama T, Yoshida T & Yokochi T. Expression of intercellular adhesion molecule-1 (ICAM-1) on vascular endothelial cells and renal tubular cells in the generalized Shwartzman reaction as an experimental disseminated intravascular coagulation model. FEMS Immunol Med Microbiol 1997; 18: 67−74. | PubMed | ChemPort | |
| 25. | Pober JS & Cotran RS. Cytokines and endothelial cell biology. Physiol Rev 1990; 70: 427−451. | PubMed | ISI | ChemPort | |
| 26. | Karkar AM, Tam FWK, Steinkasserer A, Kurrle R, Langner K, Scallon BJ, Meager A & Rees AJ. Modulation of antibody-mediated glomerular injury in vivo by IL-1ra, soluble IL-1 receptor and soluble TNF receptor. Kidney Int 1995; 48: 1738−1746. | PubMed | ChemPort | |
| 27. | Ueda Y, Lee KH, Ito S, Saito K, Niesen N, Brentjens JR & Gans ROB. In vivo neutralization of tumor necrosis factor attenuates the generalized Shwartzman reaction in the rabbit. J Endotox Res 1996; 3: 67−75. | ChemPort | |
Acknowledgments
This work was supported by the Medical Research Council (UK) grant number G 96 28976, Action Research grant S/P/3170, and an Allen and Hanbury's 1998 travel award. Publication of Figure 1 in color was made possible by grants from Fujisawa Ltd. and Roche Pharmaceuticals UK. The authors acknowledge the technical assistance of Mrs. M.C. Williams and generous gift of ricin from Professor J.M. Lord (University of Warwick). Some of this work was presented in part at The Renal Association autumn meeting 1997.
