Shiga Toxin Pathogenesis: Kidney Complications and Renal Failure

Abstract

The kidneys are the major organs affected in diarrhea-associated hemolytic uremic syndrome (D⁺HUS). The pathophysiology of renal disease in D⁺HUS is largely the result of the interaction between bacterial virulence factors such as Shiga toxin and lipopolysaccharide and host cells in the kidney and in the blood circulation. This chapter describes in detail the current knowledge of how these bacterial toxins may lead to kidney disease and renal failure. The toxin receptors expressed by specific blood and resident renal cell types are also discussed as are the actions of the toxins on these cells.

This review provides a detailed view of the current knowledge of Shiga toxin (Stx) actions and host responses that comprise the disease known as diarrhea-associated hemolytic uremic syndrome, D⁺HUS. Stx2 is the predominant form of Stx produced by clinical isolates of STEC and is considered to be the form of Stx that causes D⁺HUS in humans. Expression of the Stx receptor, glycosphingolipid Gb3, or a close analog, is required for Stx interaction with eukaryotic cells. Stx targets specific cell types in the kidney and blood circulation causing the cells to either become activated, elicit proinflammatory mediators, or apoptose. The primary target of Stx is thought to be endothelial cells, however, additional Stx-sensitive cell types exist in the kidney that are required for normal filtration of the blood (endothelial, podocyte) and fluid balance (tubules). Damage to these cells helps explain how Stx causes proteinuria, dehydration, and renal failure. Stx-sensitive platelets in the blood appear to be central to the renal vascular coagulation and thrombosis of D⁺HUS. Recent evidence indicates that bacterial lipopolysaccharide (LPS) is required for platelet activation in D⁺HUS prior to their interaction with Stx. The role of specific cytokines and chemokines in the kidneys and blood circulation remains to be determined for D⁺HUS. Limited data support the concept that transport of Stx from the colon to the kidneys may involve neutrophils. Although successful delineation of the pathophysiology of D⁺HUS in humans will continue to depend on animal models of D⁺HUS, none of the animal models currently available for D⁺HUS completely recapitulates the disease in humans.

1 Shiga Toxin Interaction With Blood Cells in the Circulation

Stx-producing Escherichia coli (STEC) are non-invasive. Stx is released in the intestine and after passing through the intestinal mucosal barrier the toxin circulates in the bloodstream before reaching its target organs. The main target organ is the kidney. How the toxin circulates and whether it is free in the circulation or cell-bound has been addressed in numerous studies. In patients with hemolytic uremic syndrome minimal amounts of Stx were found in the serum (Brigotti et al. 2011) but in mice infected with E. coli O157:H7 Stx was detected in serum 2–5 days after inoculation (Kurioka et al. 1998). The presence of Stx in the circulation in mice but not in humans may reflect the longer time span that elapses from infection to clinical presentation in patients. In addition to free toxin, which may no longer be present in the circulation when HUS develops, studies have shown that toxin may bind to blood cells and thus reach its target organs. Stx is not cytotoxic for neutrophils, monocytes, and certain B lymphocytes (Cohen et al. 1990; van Setten et al. 1996; Liu et al. 1999a; Brigotti et al. 2008). Platelets and red blood cells (RBCs) are presumed to be less sensitive to the cytotoxic effects of Stx as they lack nuclei. Thus binding of the toxin to these blood cells will not destroy the cells.

1.1 Platelets

Low platelet counts are a hallmark of HUS and lower levels are associated with worse renal prognosis (Robson et al. 1988). Mice inoculated with STEC developed thrombocytopenia which was also demonstrated in mice injected with Stx2 and LPS (Keepers et al. 2006; Calderon Toledo et al. 2008). Thrombocytopenia may be induced by multiple mechanisms such as consumption of platelets in microthrombi on the surface of damaged endothelium (Zoja et al. 2010), activation by circulating chemokines or other circulatory platelet aggregating factors and by a direct interaction between STEC virulence factors and platelets leading to their aggregation and deposition on endothelial cells (Karpman et al. 2001; Ghosh et al. 2004; Guessous et al. 2005b; Ståhl et al. 2006; Zoja et al. 2010).

Stx circulates in vivo bound to platelets during HUS (Ståhl et al. 2006). The toxin binds to activated platelets (Ghosh et al. 2004) via the globotriaosylceramide (Gb3) receptor and an alternative glycosphingolipid receptor termed band 0.03 (Cooling et al. 1998; Ghosh et al. 2004). Stx undergoes endocytosis and exerts an activating effect on platelets which readily aggregate on endothelial cells (Karpman et al. 2001). There is, to date, no evidence that Stx is transferred from platelets to cells in target organs.

1.2 Monocytes

Stx1 binds to monocytes via a Gb₃ receptor that differs somewhat from the receptor on endothelial cells. Binding was enhanced in the presence of LPS and did not inhibit protein synthesis (van Setten et al. 1996). On the contrary, Stx induced the synthesis and release of cytokines IL-1β, IL-6, IL-8, and TNF-α from human monocytes in vitro (van Setten et al. 1996). Similar results were obtained using a monocytic cell line THP-1 which, upon stimulation with Stx2, released IL-8, macrophage-derived chemokine (MDC), and regulated upon activation, normal T-cell expressed and secreted (RANTES) capable of activating platelets (Guessous et al. 2005a). A role in the prothrombotic manifestations occurring during HUS was suggested by the finding that Stx1 and Stx2 could induce the expression of tissue factor on monocytes (Murata et al. 2006; Ståhl et al. 2009). Tissue factor expression was increased when Stx2 was co-incubated with LPS and particularly when monocytes were in complex with platelets (Ståhl et al. 2009). Patients with STEC-associated HUS were found to have Stx2 on platelet-monocyte complexes. Furthermore, Stx2 induced the release of monocytic microparticles bearing tissue factor and tissue factor expressing-microparticles were detected in patients (Ståhl et al. 2009). Tissue factor-bearing microparticles may fuse with platelets and thus induce a pro-thrombotic process (Del Conde et al. 2005).

Although Stx can bind to and activate monocytes, and induce the formation of platelet-monocyte complexes with prothrombotic properties, monocytes have not been shown to assist in the circulatory transfer of toxin to target cells (Geelen et al. 2007a, b). All the same, biopsies from patients with HUS have demonstrated that monocytes infiltrate the kidneys most probably recruited by increased levels of monocyte chemoattractant protein-1 (van Setten et al. 1998).

1.3 Neutrophils

Stx was detected on the surface of neutrophils from HUS patients (Te Loo et al. 2001; Tazzari et al. 2004; Ståhl et al. 2009; Brigotti et al. 2011). In addition, Stx2 was detected on the surface of neutrophil-platelet complexes in whole blood from HUS patients (Ståhl et al. 2009). Binding of Stx to neutrophils has been debated (Geelen et al. 2007a), especially as the Stx receptor on these cells has still not been identified (van Setten et al. 1996; Geelen et al. 2007a; Arfilli et al. 2010). Toxin bound preferentially to mature PMNs (Brigotti et al. 2008). In vitro studies have shown that Stx could transfer from the surface of neutrophils to human endothelial cells suggesting that neutrophils could serve as a carrier for the toxin until the target organ was reached (Te Loo et al. 2000; Brigotti et al. 2010).

Neutrophil counts are high during HUS and higher levels, as well as elevated IL-8 levels, correlated with poor prognosis (Fitzpatrick et al. 1992; Robson et al. 1992; Fernandez et al. 2007). High PMN levels could be partially attributed to decreased spontaneous neutrophil apoptosis during HUS (Fernandez et al. 2007). Stx2 may promote neutrophilia in mice by triggering the release of cells of myeloid lineage from the bone marrow and by accelerating proliferation of PMN progenitors (Fernandez et al. 2006). Stx2 was also shown to impair neutrophil migration in mice (Fernandez et al. 2006). Neutrophils are activated during STEC infection. Circulating neutrophils degranulate thus releasing proteases and reactive oxygen species (Fernandez et al. 2005). In vitro studies have shown that Stx2 activates neutrophils, particularly those in complex with platelets (Ståhl et al. 2009). Furthermore, leukocytes adhere to the endothelium in the presence of Stx, a process promoted by endothelial expression of fractalkine, MCP-1, IL-8, and heparan sulfate proteoglycans (Zoja et al. 2002; Geelen et al. 2008; Zanchi et al. 2008).

In addition to Stx, E. coli O157 may also secrete StcE, a metalloprotease that cleaves mucin demonstrated to increase neutrophil oxidative burst and cell adhesion, thus impairing neutrophil migration (Szabady et al. 2009). Taken together, the interactions between Stx or StcE and neutrophils could explain neutrophil activation, degranulation, and impaired migration leading to increased tissue destruction at sites of neutrophil infiltration. Furthermore, the possibility that neutrophils bear Stx on their membrane during HUS would enable transfer of the toxin to other target cells more prone to a cytotoxic reaction.

1.4 Red Blood Cells

Fragmented RBCs occur during HUS in which non-immune hemolytic anemia occurs. It is assumed that RBCs fragment mechanically while passing through partially occluded capillaries. One study also reported that RBC membranes were stiffened by oxidative damage (Turi et al. 1994). Although the exact mechanism by which hemolysis occurs is, as yet unclear, the RBC fragmentation can be reproduced in animal models inoculated with E. coli O157:H7 (Karpman et al. 1997) or injected with Stx (Taylor et al. 1999).

Hemolysin is one of the virulence factors expressed by E. coli O157:H7 encoded in a large plasmid (pO157) (Schmidt et al. 1994). The hemolysin is a pore-forming cytolysin. E. coli 0157:H7 hemolysin has been shown to lyse human RBCs in vitro (Bauer and Welch 1996), but this phenomenon most probably does not occur in vivo as bacteremia does not occur during STEC infection and STEC hemolysin is bound to the bacterial cell membrane (Schmidt et al. 1994; Bauer and Welch 1996).

Studies have investigated whether the expression of different P blood group antigens on RBCs may influence the risk of developing HUS. The P blood group includes three different antigens: P, P1, and Pk, giving rise to five different phenotypes P1, P2, P1k, P2k, and p. The Pk antigen is the precursor of the P antigen as well as the Gb3 receptor for Stx. Thus the Gb3 receptor for Stx is present on all RBCs that possess the Pk antigen (all but the rare p phenotype) (Bitzan et al. 1994; Spitalnik and Spitalnik 1995). Studies have addressed whether expression of the P1 blood group could be protective reducing the risk of developing HUS based on the assumption that P1 expression on RBCs would bind Stx and thereby reduce the amount of circulating toxin. Thus individuals with RBCs negative for P1 or with lower amounts of Gb3 would be at higher risk of developing HUS (Taylor et al. 1990; Newburg et al. 1993). These findings however, could not be confirmed by others (Jelacic et al. 2002).

2 Gb3 Expression in the Kidney

Specific and high-affinity binding of Stx to its receptor is required for toxin action in the disease process. The receptor for Stx1 and Stx2 is the glycosphingolipid Gb3 that is expressed in kidneys and some other tissues of humans and animals (McCluer et al. 1981; Cohen et al. 1987; Lindberg et al. 1987; Lingwood et al. 1987; Boyd and Lingwood 1989; Obrig et al. 1993; Lingwood 1996; Hughes et al. 2002; Garcia et al. 2006). Gb3 is synthesized within golgi of cells, then transported to and inserted into the plasma membrane outer leaflet where the trisaccharide faces outward for specific recognition by the B-subunits of Stx (Lingwood 1993). The importance of Gb3 for Stx action was revealed by the total absence of Stx effects in mice lacking Gb3 and, in vitro, by concomitant reconstitution of Gb3 and Stx-sensitivity in cells lacking Gb3 (Waddell et al. 1990; Okuda et al. 2006). Recent advances in glycolipid biology indicate that Gb3 is localized to lipid rafts rich in cholesterol (Hakomori 2000; Falguieres et al. 2001; Falguieres et al. 2006; Muthing et al. 2009). Binding of Stx to Gb3 localized to lipid rafts is important for action of Stx in eukaryotic cells (Hoey et al. 2003; Smith et al. 2006a; Khan et al. 2009; Lingwood et al. 2010a). Human kidney contains a series of Gb3 subspecies which differ in their ceramide hydrocarbon chain length (C16–C24) and degree of hydroxylation (Shayman and Radin 1991; Muthing et al. 2009). Recent data support the concept that not all Gb3 in the plasma membrane is bioavailable (Johannes and Romer 2010; Lingwood et al. 2010a, b; Mahfoud et al. 2010; Betz et al. 2011).

2.1 Glomerular Gb3

It is generally accepted that the renal endothelium is a primary target of Stx in STEC-associated hemolytic uremic syndrome, also known as D⁺HUS, in which the D⁺ stand for diarrhea (Zoja et al. 2001; Karch et al. 2005; Ahn et al. 2009). Changes in the renal endothelium typically include swelling and detachment from the basement membrane (Habib 1992). Human glomerular microvascular endothelial cells have been shown to be sensitive to sub-nanomolar Stx, in vitro (Obrig et al. 1993; van Setten et al. 1997a, b). This is in contrast to human large vessel umbilical vein endothelial cells that are relatively refractory to Stx action (Obrig et al. 1988, 1993; Tesh et al. 1991). This differential sensitivity to Stx was demonstrated to be due to a 50-fold higher expression of Gb3 by the glomerular endothelial cells (Obrig et al. 1993). As described below in more detail, human podocytes and mesangial cells are also sensitive to Stx (Simon et al. 1998; Psotka et al. 2009). Human podocytes and mesangial cells express Gb3 (Robinson et al. 1995; Psotka et al. 2009).

Animal models for D⁺HUS should preferably reflect the Gb3 expression pattern of human kidneys. However, an analysis of Gb3 location in kidneys of such animals remains incomplete although the mouse, rat, pig, rabbit, dog, ferret, and baboon have been utilized as models of D⁺HUS (Taylor et al. 1999; Rutjes et al. 2002; Melton-Celsa and O’Brien 2003; Caprioli et al. 2010). None of these models completely recapitulates the D⁺HUS renal disease of humans. Presently, the baboon model of D⁺HUS appears best in this regard following intravenous administration of purified Stx2 (Taylor et al. 1999; Siegler et al. 2003; Stearns-Kurosawa et al. 2010). However, details of location and quantification of Gb3 in baboon kidney have yet to be reported.

2.2 Extraglomerular Gb3

There is a need for a thorough assessment of Gb3 expression along the different segments of the nephron in human and animal kidneys. Most of the available data for Gb3 expression in tubules are derived from isolated propagated cell types of human and animal kidneys. Numerous studies have demonstrated the significant Stx-sensitivity of tubule cells, but little information is presented about Gb3, per se, in these cells in culture (Hughes et al. 1998a; Kodama et al. 1999; Liu et al. 1999b; Williams et al. 1999; Kaneko et al. 2001; Nestoridi et al. 2005a; Paixao-Cavalcante et al. 2009). Analysis of intact renal tissues from human and animal sources provided evidence that Gb3 is expressed by renal tubules (Lingwood 1994; Rutjes et al. 2002; Ergonul et al. 2003a; Winter et al. 2004; Silberstein et al. 2008). Direct binding of Stx to frozen sections of murine kidney revealed Gb3 in proximal, distal, and collecting duct cells (Tesh et al. 1993; Rutjes et al. 2002). In human kidney sections, Stx binding was localized to distal and collecting duct tubule epithelium (Lingwood 1994). A more detailed examination of human and murine renal tissue performed with co-localization of Gb3 with renal tubule cell-type specific markers indicated that Gb3 is expressed by proximal and collecting duct tubules in both human and murine kidney (Psotka et al. 2009) (Obata and Obrig, unpublished data). Interstitial microvascular endothelial cells also express Gb3. The effects of Stx on renal tubule cells are described in more detail in the following section.

3 Stx Interaction With Cell Types of the Kidney

Much information is now available regarding Stx interaction with resident cells of the kidney (Karpman et al. 2010; Obrig 2010). The fact that Stx interacts with many cell types of the kidney makes difficult the assignment of their relative importance for development of renal disease in D⁺HUS.

3.1 Glomerular Cells

As described above, human glomerular podocyte, endothelial, and mesangial cells express Gb3. This, in part, explains the changes observed in D⁺HUS glomeruli which are likely due to Stx. Interaction of Stx with these cell types is described next in more detail. Stx may target Gb3-positive glomeruli of the pediatric versus adult kidney, an observation that could explain the higher incidence of D⁺HUS among the pediatric population (Lingwood 1994; Chaisri et al. 2001). However, this concept needs to be validated.

3.1.1 Endothelial Cells

Although direct effects of Stx were first described with large vessel endothelial cells, these cells were orders of magnitude less sensitive to concentrations of Stx than microvascular endothelial cells (Obrig et al. 1988, 1993; Louise and Obrig 1994; Ohmi et al. 1998). However, it was noted that only 7% of 55 human umbilical vein endothelial cell cultures, each derived from a different individual, were truly sensitive to the cytotoxic action of Stx (Kaye et al. 1993). The topic of Stx-endothelial interaction has been reviewed recently (Zoja et al. 2001; Muthing et al. 2009; Petruzziello et al. 2009; Karpman et al. 2010; Obrig 2010). It should be noted that the generally accepted pathological description of endothelial damage in D⁺HUS is swollen and detached endothelium (Habib 1992; Zoja et al. 2001). While this level of damage appears less severe than cell death, endothelial cell viability is known to be dependent on attachment to basement membrane (Yannariello-Brown et al. 1988; Hoch et al. 1989; Lalka et al. 1989; Smith et al. 1989; Grant et al. 1990). Given this information, it seems likely that most of the damage to glomerular endothelium is due to direct action of Stx in D⁺HUS. However, there are many additional host response factors in the kidney during D⁺HUS, some of which are known to interact with endothelium. The health of endothelium is dependent on other cells such as renal glomerular podocytes that produce vascular endothelial cell growth factor (VEGF) essential for endothelial cells (Eremina et al. 2008; Sison et al. 2010). The fact remains that Stx interacts with many different cell types eliciting a myriad of responses (O’Loughlin and Robins-Browne 2001; Karpman et al. 2010; Obrig 2010). Cytokines elicited by bacterial LPS during D⁺HUS such as TNF-alpha and IL-1beta, and LPS itself, can induce Gb3 synthesis and increase Stx-sensitivity of endothelial cells (Louise and Obrig 1991, 1992; van de Kar et al. 1992, 1993; Kaye et al. 1993; van Setten et al. 1997a, b; Stricklett et al. 2005).

Responses of endothelial cells to Stx are diverse. Stx was shown to elicit release of chemokines which may be important for development of some aspects of D⁺HUS (Zoja et al. 2002; Guessous et al. 2005b). Stx was also shown to decrease prostacyclin synthesis by endothelial cells (Karch et al. 1988). In addition, Stx caused microvascular endothelial cells from human brain or renal glomeruli to apoptose (Kaneko et al. 2001; Pijpers et al. 2001; Ergonul et al. 2003b; Fujii et al. 2008; Psotka et al. 2009). In other cases, Stx activated endothelial cells for increased leukocyte adherence (Zoja et al. 2002; Geelen et al. 2008). As would be expected, some of the direct effects of Stx on endothelial cells result in changes in physiology related to coagulation and thrombosis (Louise and Obrig 1994; van de Kar et al. 1994; Kaye and Obrig 1995; Karpman et al. 2001; Nolasco et al. 2005; Te Loo et al. 2006; Huang et al. 2010). The interaction of Stx and TNF with human renal endothelial cells was shown to elicit tissue factor expression (Nestoridi et al. 2005b). Whether some of these latter events are related to complement activation typical of other thrombotic microangiopathies remains an interesting, but unanswered question (Moake 2009; Thurman et al. 2009; Caprioli et al. 2010; Zipfel et al. 2010). In summary, ample evidence exists demonstrating the central role of endothelial cells in D⁺HUS.

3.1.2 Podocytes

Podocytes or visceral epithelial cells are an integral part of the kidney filtration barrier presenting a slit-diaphragm barrier for blood components (Hirschberg et al. 2008; Fogo 2009; Peti-Peterdi and Sipos 2010). Podocytes develop extended foot processes which wrap around the glomerular capillaries and are separated from glomerular endothelium by a basement membrane and glycocalyx. Changes in podocyte physiology are contributing factors in some important renal diseases (Coward et al. 2005; Tryggvason et al. 2006; Henao et al. 2007; Marshall 2007; Collino et al. 2008; D’Agati 2008; Thorner et al. 2008; Quaggin 2009; Clement et al. 2010). Damaged podocytes result in faulty filtration of the blood into the urinary space resulting in increased proteinuria.

Given that podocyte–endothelium interactions are important to both cell types, how does this relate to D⁺HUS? One example is that podocytes secrete VEGF that is essential to the nearby endothelial cells (Sison et al. 2010). A Stx-induced decrease in VEGF production by podocytes may add to the proteinuria (Shankland 2006; Eremina et al. 2008; Izzedine et al. 2010). In addition, Stx has been shown to increase endothelin-1 in podocytes that may be deleterious to podocytes (Morigi et al. 2006). An interesting phenomenon is that both CNS neurons and renal podocytes express Gb3, are sensitive to Stx, and exhibit glutamatergic signalling. Podocyte glutamine signalling is important for the function of the filtration barrier (Giardino et al. 2009). While it is has been demonstrated that Stx increases glutamine release in murine neurons, changes in glutamine release have not been studied in glomerular podocytes (Obata et al. 2008; Obata 2010; Obata and Obrig 2010). Important to this review is that human podocyte and glomerular endothelial cells are sensitive to picomolar concentrations of Stx (Obrig et al. 1993; Psotka et al. 2009). Detailed analyses of Stx effects on glomerular podocytes and endothelial cells have not been reported in animal models of D⁺HUS. Stx was detected bound to podocytes in the kidney tissue of an infant who died of D⁺HUS (Chaisri et al. 2001). This is in agreement with an earlier report showing that glomeruli from an infant, but not an adult kidney bound Stx (Lingwood 1994). Unfortunately, murine podocyte and endothelial cells do not express Gb3 and are insensitive to Stx, thus placing limits on the use of the murine model of D⁺HUS (Rutjes et al. 2002; Psotka et al. 2009). This concept appears to be true for other animal models of D⁺HUS including the New Zealand White rabbit.

3.1.3 Mesangial Cells

Expansion of glomerular mesangium occurs in D⁺HUS (Shigematsu et al. 1976). Stx binds to and has multiple direct effects on human mesangial cells, in vitro (Robinson et al. 1995; Simon et al. 1998). Although Stx enters these cells to inhibit protein synthesis, mesangial cells are not killed by the toxin (van Setten et al. 1997a, b; Simon et al. 1998). The total importance of Stx interaction with mesangial cells is largely unknown for D⁺HUS as very few reports currently exist on this topic. It also appears that basic science knowledge on the normal interaction between mesangial cells and glomerular endothelium or podocytes has lagged behind other areas of kidney research (Vaughan and Quaggin 2008; Picken 2009; Schlondorff and Banas 2009). However, it seems very likely that products elicited in mesangial cells by Stx will be of importance for a complete understanding of D⁺HUS pathophysiology.

3.2 Extraglomerular Cells

Sensitivity of renal cells to Stx requires interaction with Gb3 Stx receptors expressed on the plasma membrane. As alluded to above, the distribution of Gb3 among the different tubular epithelial cells of the nephron has not been fully described for human or animal kidneys. However, there is good evidence that some of these specialized cells express Gb3 and are sensitive to Stx (Obrig 2010). The following is a more detailed description of these data. A significant number of immortalized cell lines now exist for in vitro studies of Stx action in the different specialized cells of the nephron (Bens and Vandewalle 2008). At present, it is not clear which animal model of D⁺HUS may best represent extraglomerular pathophysiology of D⁺HUS in humans (Caprioli et al. 2011).

3.2.1 Proximal Tubule Cells

Proximal tubule cells are the most studied of the renal tubular cell types for D⁺HUS. Human proximal tubule cells are sensitive to picomolar concentrations of Stx, in vitro (Hughes et al. 1998a; Sood et al. 2001; Fuller et al. 2011). This high degree of sensitivity to Stx was attributed to ample Gb3 production by human proximal tubule cells, in vitro (Hughes et al. 2002). Interestingly, the StxB (binding) subunit alone also exhibited some responses in this cell type, in vitro (Creydt et al. 2006). Stx induced tissue factor activity on the surface of human proximal tubule cells (Nestoridi et al. 2005a). As there is a role for complement-induced dysfunction of proximal tubules (Buelli et al. 2009), this may also contribute to renal disease in D⁺HUS. Low concentrations of Stx induced an increase in TNF-alpha mRNA and release of TNF-alpha from human proximal tubule cells, in vitro (Hughes et al. 1998b). A similar effect was observed in these cells for an increase in IL-1 beta, while Stx synergized with LPS for an increase in IL-6 mRNA (Hughes et al. 1998b). Stx inhibited water absorption by proximal tubule cells in an in vitro assay (Silberstein et al. 2008). Stx caused apoptosis of human proximal tubule cells that could be reversed by silencing the proapototic factor Bak (Karpman et al. 1998; Kodama et al. 1999; Wilson et al. 2005). Finally, it remains to be determined in animal models of D⁺HUS whether Stx-induced damage to the filtration barrier results in proteinuria which, in turn, leads to indirect damage to proximal tubules (Zoja et al. 2003; Li et al. 2010; Wu et al. 2010).

3.2.2 Distal Tubule Cells

Mice injected with Stx exhibit specificity for toxin interaction with distal tubules (Rutjes et al. 2002). In some cases the affected cells are described only as cortical or medullary, so interpretation of these data is limited in scope. However, Stx binding to distal tubules has been reported in human renal tissue (Lingwood 1994).

3.2.3 Collecting Duct Cells

Stx targets collecting ducts of mice (Rutjes et al. 2002; Psotka et al. 2009). Renal collecting ducts of mice apoptose in response to intraperitoneally administered Stx (Psotka et al. 2009). Rats given Stx respond with an increased urinary aquaporin-2 level suggesting damage to the AQP2-producing collecting duct cells (Sugatani et al. 2002). Polyuria was reported in both mice and rats given Stx indicating damaged collecting duct cells leading to decreased water reabsorption (Sugatani et al. 2002; Psotka et al. 2009).

3.2.4 Loop of Henle Cells

Very little is known about Stx interaction with epithelial cells of the descending and ascending Loop of Henle in human or animal kidneys. Future in vivo studies of Gb3 co-localization with cell type specific markers will be necessary. Using this approach, preliminary results indicate that anti-Gb3 antibody co-localizes with anti-Tamm Horsfall protein antibody in sections of the murine kidney suggesting that the thick ascending Loop of Henle expresses Gb3 (Obata and Obrig, unpublished data). We also observed co-localization of anti-Gb3 and anti-aquaporin 1 antibodies in medullary sections of human kidney indicating that Gb3 is expressed in the thin descending limb of the Loop of Henle.

In summary, compelling evidence is available supporting a key role of renal microvascular endothelial cells in D⁺HUS. However, the fact that human glomerular podocytes are also very sensitive to Stx requires a rethinking of how Stx action leads to renal failure in D⁺HUS. In a similar tone, direct effects of Stx on human proximal and collecting duct tubules may explain the dehydration observed in D⁺HUS patients. Thus, ample evidence now provides for a rational scheme of how Stx causes renal failure in D⁺HUS.

4 LPS Interaction With Cells of the Blood and Resident Renal Cells

4.1 Platelets

LPS is a component of the outer membrane of Gram-negative bacteria which may be released into the circulation leading to endotoxemia and endotoxic shock. It has potent biological activity activating the innate immune response leading to cytokine release (Aderem and Ulevitch 2000). Various studies have shown that LPS may induce platelet activation and aggregation both in vitro (Wachowicz et al. 1998; Saluk-Juszczak et al. 1999) and in vivo (Itoh et al. 1996). This effect may result in consumptive thrombocytopenia (Cicala et al. 1997). Resting platelets must be primed with LPS before interaction with Stx (Viisoreanu et al. 2000; Ståhl et al. 2009). O157LPS binds to platelets via a complex receptor composed of TLR-4 and CD62 (P-selectin) and thus activates them. LPS was also detected on the surface of platelets from HUS patients suggesting that it may activate platelets in the circulation (Ståhl et al. 2006). Thus, as mentioned above, both Stx and O157LPS (Ståhl et al. 2006) are capable of activating platelets and co-stimulation with both virulence factors simultaneously has an additive effect on the formation of platelet-leukocyte complexes expressing tissue factor in whole blood (Karpman et al. 2001; Ghosh et al. 2004; Ståhl et al. 2006, 2009). This effect was further enhanced at increasing shear rates. Co-stimulation of whole blood with Stx and O157LPS also induced an enhanced release of platelets-derived microparticles and tissue factor-coated microparticles, more than each stimulant alone (Ståhl et al. 2009). Thus LPS appears to trigger several mechanisms leading to a prothrombotic state. O157LPS is at least as potent as other forms of LPS in this respect, if not more (Ståhl et al. 2009).

4.2 Resident Renal Cells

LPS may be important for the development of D⁺HUS. Although the amount of LPS in the blood circulation of D⁺HUS patients remains undocumented, one would expect a lower level of LPS in D⁺HUS than in systemic bacterial infections because D⁺HUS is not a bacteremic disease (Tarr et al. 2005; Ahn et al. 2009). The primary receptor for bacterial LPS is toll-like receptor 4 (TLR4) (Beutler and Poltorak 2000; Knotek et al. 2001; Takeuchi and Akira 2001; Yamamoto and Akira 2010). LPS is a more active proinflammatory agent than Stx, and TLR4 receptors are expressed on specific cell types of the kidney. Fortunately, a considerable amount is known about the TLRs in kidneys (Vandewalle 2008; Mkaddem et al. 2010; Pulskens et al. 2010; Batsford et al. 2011; Goncalves et al. 2011). Thus, LPS would be expected to be active in the pathophysiology of D⁺HUS. LPS modulates Stx action in cells through induction of Gb3 (Louise and Obrig 1992; Hughes et al. 1998a; 2000; Clayton et al. 2005). In animal models of D⁺HUS, administration of LPS either before, during, or after Stx significantly influences the effects of Stx (Barrett et al. 1989; Palermo et al. 2000; Siegler et al. 2001; Clayton et al. 2005; Keepers et al. 2006, 2007). Mice lacking an adequate response to LPS due to a mutation in, or deficiency of, the TLR4 receptor exhibit an increased response to Stx most probably due to decreased bacterial clearance in the gut (Karpman 1997, Calderon Toledo 2008). The action of LPS in animal models of D⁺HUS is discussed in more detail below.

5 Inflammatory Responses in the Kidney

D⁺HUS exhibits an inflammatory component. Cytokines and chemokines are increased in the blood and kidneys during D⁺HUS. The role of these proinflammatory agents in D⁺HUS remains to be delineated. The following is a review of data from humans and from animal models of D⁺HUS derived from many different laboratories documenting the presence of such agents in D⁺HUS.

5.1 Cytokines

The role of TNF-alpha in D⁺HUS is controversial. Several reports agree that TNF-alpha can induce Gb3 in some eukaryotic cells, sensitizing the cells to Stx (van de Kar et al. 1992; Louise et al. 1997; van Setten et al. 1997a, b). However, data from a murine model of D⁺HUS suggest that TNF-alpha is not an essential contributor to the disease (Wolski et al. 2002). In humans with D⁺HUS, TNF-alpha and IL-6 were higher in the urine than in the blood circulation, suggesting localized production of the cytokines (Karpman et al. 1995). These cytokines were not present in the blood and urine of healthy individuals (Karpman et al. 1995). A similar scenario was described for urinary IL-8 in D⁺HUS pediatric patients (Inward et al. 1997). Stx directly induced increased release of TNF-alpha from human renal proximal tubule cells, in vitro (Hughes et al. 1998b).

5.2 Chemokines

There is growing evidence of an increase in chemokine synthesis in the kidneys during D⁺HUS. In a murine model of D⁺HUS, LPS induces C–X–C and C–C chemokines by proximal tubule cells (Keepers et al. 2007; Roche et al. 2007). In homogenates of whole kidney from mice injected with LPS, these chemokines are also increased. In these cases, Stx stabilizes the half-life of chemokine mRNA, further enhancing the total amount of chemokine produced. Chemokines including MCP-1 and IL-8 are increased in urine of D⁺HUS patients (van Setten et al. 1998). It is reasonable to expect that individual chemokines may be derived from more than one cell type in D⁺HUS. However, it is clear that inflammatory cells such as monocytes and PMNs are attracted into the kidneys during D⁺HUS (Keepers et al. 2006, 2007; Roche et al. 2007). Their relative importance to renal damage in D⁺HUS remains to be determined. In the murine model of D⁺HUS, it was determined that renal fibrin deposition and the lethal effect of Stx were not affected by the elimination of monocyte/macrophage cell type from these animals (Obrig. et al., unpublished data). Renal chemokines in D⁺HUS may act as secondary activators of platelets, along with thrombin or ADP, for renal thrombosis common to this disease (Gear et al. 2001; Gear and Camerini 2003; Guessous et al. 2005b).

5.3 Nitric Oxide

Nitric oxide (NO) is known to exhibit antiplatelet and renal vasodilatory activity (Martin et al. 1986; Radomski et al. 1987). NO was protective in mice against Stx-induced renal toxicity (Dran et al. 2002). However, NO bioavailability was reduced in baboons in response to Stx (Siegler et al. 2005). The role of NO in humans with D⁺HUS is not clear. The fact that Stx inhibited NO production by microvascular endothelial cells, in vitro may, in part explain the interaction between Stx and NO in D⁺HUS (Te Loo et al. 2006).

6 Thrombosis and Fibrinolysis in D+HUS

A major feature of D⁺HUS is the appearance of thrombi in the renal microvasculature. This process involves platelets and endothelial cells. More research is needed to identify additional cell types and inflammatory mediators which also have a role in the thrombosis of D⁺HUS. The relative potential of therapeutics for D⁺HUS based on targets of thrombosis is made less attractive due to the risk of resulting hemorrhages within the CNS.

6.1 Platelet Activation

Endothelial cells are damaged during HUS. Upon exposure of the subendothelium platelets bind to subendothelial matrix proteins including von Willebrand factor (VWF), fibrinogen and collagen resulting in formation of platelet thrombi. Formation of microthrombi will compromise the blood flow in the microvasculature of the kidney. The initial event in this process is binding of VWF to the platelet membrane receptor glycoprotein 1b (GP1b) allowing a conformational change to occur in the platelet exposing the GPIIb/IIIa receptor on the platelet membrane (Savage et al. 1992). VWF and fibrinogen bind to GPIIb/IIIa which will link one platelet to another, thus forming a clot. Activated platelets release granular contents, such as ADP, thromboxane A2 and thrombin, promoting further activation.

Platelets in HUS appear to be degranulated (Fong and Kaplan 1982; Sassetti et al. 1999) and circulatory levels of β-thromboglobulin (Appiani et al. 1982) and soluble P-selectin (Katayama et al. 1993) are increased. VWF may be secreted from both endothelial cells and platelets and levels are elevated during HUS enabling enhanced formation of thrombi (van de Kar et al. 1994). A decreased multimer size presumably reflects enhanced proteolysis (Tsai et al. 2001) although the activity of ADAMTS13, the vWF cleaving protease, was found to be normal (Tsai et al. 2001). In vitro experiments have shown that Stx induces the release of ultra-large VWF multimers from human endothelial cells and delayed cleavage of VWF by ADAMTS13 (Nolasco et al. 2005), thus promoting the formation of platelet strings attached by ultra-large VWF on endothelial cells. The importance of VWF for promoting platelet aggregation on endothelial cells was illustrated by blocking VWF-platelet specific receptors (Morigi et al. 2001).

There is no consumption of plasma coagulation factors during HUS but prothrombotic markers such as prothrombin 1 + 2 and thrombin–antithrombin III complex are elevated (Nevard et al. 1997; Van Geet et al. 1998) occurring even before patients develop HUS (Chandler et al. 2002). HUS patients were found to have elevated levels of tissue factor (Kamitsuji et al. 2000) as well as circulating platelet-leukocyte complexes and platelet-derived microparticles, both with deposits of tissue factor and complement on their surface (Ståhl et al. 2009, 2011). These changes were specific for the acute phase of disease, decreasing upon remission. In vitro experiments showed that Stx induced the formation of complexes between platelets and leukocytes, mostly platelets and monocytes. In addition, the toxin induced the release of tissue factor and C3 as well as C9-bearing microparticles, mainly from platelets (Ståhl et al. 2009; Ståhl et al. 2011). Microparticles are procoagulant due to membrane expression of phosphatidylserine (Mallat et al. 2000), an effect that is further enhanced when they are coated with tissue factor. Tissue factor is a transmembrane glycoprotein receptor for coagulation factor VII, leading to conversion of factor X into factor Xa in the extrinsic coagulation pathway (Rao et al. 1986). This will ultimately lead to thrombin generation, resulting in thrombus formation and further platelet activation.

Stx exerts other indirect effects on platelets such as inducing the release of platelet-stimulatory cytokines from monocytes described above (Guessous et al. 2005a). When co-administrated with O157LPS, and especially under high shear rates, the toxin induces pronounced platelet activation (Ståhl et al. 2009). Taken together, platelet activation leading to thrombus formation may occur as a direct effect of bacterial toxin and/or LPS as well as an indirect effect via stimulatory cytokines, in addition to the platelet aggregating events occurring during extensive endothelial cell damage.

6.2 Endothelial Activation

The pathological lesion seen in kidneys during HUS is termed thrombotic microangiopathy and includes the presence of thrombi in glomerular capillaries as well as pronounced endothelial cell swelling and detachment due to subendothelial swelling (Benz and Amann 2009). Damaged endothelium promotes a pro-thrombotic state. Endothelial cell damage presumably occurs before patients develop HUS, as markers of endothelial cell activation (thrombomodulin, soluble vascular cell adhesion molecule and E-selectin) in HUS patients were not found to be higher than in dialysis patients (Nevard et al. 1999).

Stx1 binds to glomerular endothelial cells via the Gb3 receptor and induces a cytotoxic effect (Obrig et al. 1993; van Setten et al. 1997a, b). Stx1 also leads to platelet aggregation on human endothelium (Karpman et al. 2001) particularly on microvascular endothelial cells under conditions of high shear stress (Morigi et al. 2001). Glomerular microvascular endothelial cells pretreated with TNF-α expressed reduced levels of thrombomodulin after incubation with Stx (Fernandez et al. 2003). As thrombomodulin has anti-thrombogenic properties this may promote thrombus formation. Endothelial beta-3-integrin subunit, vitronectin receptor, P-selectin, and PECAM-1 are all involved in Stx-induced platelet aggregation on endothelial cells (Morigi et al. 2001).

In addition to Stx, neutrophils may promote endothelial cell injury during HUS (Forsyth et al. 1989). Stx induces leukocyte adherence to the endothelium (Morigi et al. 1995) as well as their transmigration under flow (Zoja et al. 2002). Activated neutrophils degranulate releasing reactive oxygen species as well as proteases (Fernandez et al. 2005) which may contribute to local cellular injury.

6.3 Fibrinolysis

Under physiological conditions endothelial cells present an antithrombotic surface due to expression of heparan sulphate and thrombomodulin. Heparan sulphate binds antithrombin III which promotes inactivation of the intrinsic coagulation cascade. Thrombomodulin is a glycoprotein receptor on the endothelial surface which binds thrombin thus activating protein C. Activated protein C and its cofactor, protein S, inactivate factors Va and VIIIa (Lammle and Griffin 1985). Endothelial cells also produce inhibitors of platelet aggregation such as prostacyclin and tissue plasminogen activator (t-PA) (Pearson 2000). Fibrin clots have binding sites for plasminogen and t-PA. t-PA promotes conversion of plasminogen into plasmin. When the fibrinolytic system is activated fibrin is degraded to degradation products, D dimers, by plasmin.

Increased levels of D-dimers, t-PA, t-PA-plasminogen-activator inhibitor type 1 (PAI-1) complex were demonstrated in the early stage of STEC infection even before HUS developed further increasing after the development of HUS (van de Kar et al. 1994; Chandler et al. 2002). Impaired fibrinolysis was demonstrated in patients with STEC-induced HUS (Nevard et al. 1997) which may be due to increased PAI-1 in the circulation (Bergstein et al. 1992). In summary, coagulative aberrations occurring during HUS promote a prothrombotic and hypofibrinolytic state.

7 Development of Acute Renal Failure in D+HUS

The kidney is the main target organ in STEC-mediated HUS. Renal biopsies are not routinely carried out as the diagnosis is made on a clinical basis and patients are usually thrombocytopenic. Histopathological features during STEC-associated HUS are the presence of microthrombi in glomerular capillaries, extensive endothelial damage with occlusion of capillary lumina, deposition of fibrin, mesangiolysis, and mesangial expansion (Shigematsu et al. 1976) as well as extensive tubular apoptosis (Karpman et al. 1998). Severe cases develop acute cortical necrosis affecting most cells in the renal cortex. Bacterial virulence factors may affect the glomerular endothelium, podocytes, mesangial cells as well as tubular cells, as described above. Damage to glomerular endothelium will initiate thrombus formation resulting in multiple microthrombi in glomerular capillaries. Damage to tubular cells will result in electrolyte disturbances, acidosis and decreased urine production. Damaged tubuli may have a secondary deleterious effect on the glomerulus in the same nephron, but the converse is also possible, ischemic injury to the glomerulus will eventually lead to tubular damage within the nephron. Stx was demonstrated in a limited number of human kidneys from patients with HUS, in both glomeruli and tubuli (Uchida et al. 1999; Chaisri et al. 2001). As bacterial toxin may affect both tubular and glomerular cells it is unclear whether the toxin targets certain renal cells preferentially or whether the various cells are affected simultaneously.

In addition to the noxious effect of Stx, presumably in combination with LPS, heme proteins, released during hemolysis, may also have a cytotoxic effect on renal tubular cells and microvascular endothelial cells. This effect was enhanced by Stx in vitro (Bitzan et al. 2004). Recent studies have also raised the possibility that complement-mediated renal injury may occur in STEC-associated HUS. At presentation, patients exhibited activation of the alternative pathway of complement in the circulation (Robson et al. 1992; Thurman et al. 2009; Ståhl et al. 2011). Interestingly, in vitro studies have shown that Stx could induce complement activation and deposition on microvascular endothelial cells under flow (Zoja et al. 2010) as well as activate the alternative pathway in serum (Orth et al. 2009). Complement-mediated renal injury could thus enhance the direct effect of bacterial toxins. Stx may also inhibit the regulatory effect of factor H (Orth et al. 2009) although it is unclear whether this occurs in vivo.

8 Experimental Therapeutics for Renal Disease in D+HUS

Therapeutics are not currently available for the treatment of D⁺HUS. Standard supportive care includes maintenance of electrolytes and fluid balance as well as dialysis in more severe cases. However, the early development and preclinical testing of potential therapeutics has taken place in recent years.

Vaccines are effective to reduce STEC colonization of commercial bovine species (Potter et al. 2004; McNeilly et al. 2010). Vaccines for STEC in humans are less advanced in development and face the test of financial feasibility by pharmaceutical companies. Humanized anti-Stx antibodies have entered the clinical testing phase (Tzipori et al. 2004; Akiyoshi et al. 2005; Dowling et al. 2005; Smith et al. 2006b). Phase I clinical testing of a humanized anti-Stx monoclonal antibody in adult and pediatric subjects demonstrated safety at doses up to 3mg/kg (Lopez et al. 2010). Passive immunization of animals with anti-Stx neutralizing antibodies has demonstrated a moderate degree of protection against challenge with either oral STEC or parenteral Stx (Sauter et al. 2008; Mohawk et al. 2010). The important question to be answered here is whether Stx remains in the blood circulation for a prolonged period of time in D⁺HUS patients after the hemorrhagic colitis stage to be available for neutralization by the anti-Stx antibodies. This question remains to be answered. Stx receptor mimics (Lingwood and Mylvaganam 2003; Miura et al. 2006; Watanabe-Takahashi et al. 2010) or inhibitors of Stx enzymatic activity (Wahome et al. 2010) represent different approaches to Stx neutralization in the blood circulation. These agents remain to be approved for clinical testing.

A number of therapeutic strategies have been proposed for late-stage intervention of D⁺HUS which target Stx action in the kidneys. Cell permeable small molecules which interfere with Stx trafficking may be effective inhibitors of Stx action (Nishikawa et al. 2006; Sandvig et al. 2009). Others have identified Stx-initiated signal-transduction events in cells as potential targets for therapeutic intervention (Jandhyala et al. 2008). Finally, deleterious host responses to Stx in renal tissue are also potential targets for therapeutics. Examples of these include biological events which lead to coagulation, thrombosis, and changes in the filtration barrier in the kidney. The earlier literature is replete with superficial clinical tests for these targets, none of which showed results sufficient for further testing in properly designed clinical trials. However, some preclinical studies with animal models of D⁺HUS have shown promise for late-stage intervention at the kidney level (Nishikawa et al. 2006; Warnier et al. 2006; Roche et al. 2007; Psotka et al. 2009; Jeong et al. 2010).

Development of therapeutics for D⁺HUS will depend on measurements of surrogate biomarkers of the disease. These biomarkers are released from damaged cells into the blood circulation or into urine (Rosner 2009). To date, biomarkers specific to D⁺HUS are yet to be reported. Urinary neutrophils gelatinase-associated lipocalin (Ngal) has been detected in urine of D⁺HUS patients (Trachtman et al. 2006). However, Ngal is less specific as a biomarker of renal injury (Bolignano et al. 2010; Devarajan 2010; Viau et al. 2010). Future application of the Ngal gene reporter mouse to D⁺HUS could lead to data more helpful for this application (Paragas et al. 2011). A group of different biomarkers specific to D⁺HUS may be required to achieve an acceptable level of clinical utility.