Multifaceted Interactions of Bacterial Toxins With the Gastrointestinal Mucosa

MR Popoff


Future Microbiol. 2011;6(7):763-797. 

In This Article

Intracellularly Active Enterotoxins

Many bacterial toxins have the ability to enter target cells, most often by using physiological endocytosis pathways, and to modify a specific intracellular target (Table 3). According to the nature of the target and the type of modification, intracellular active toxins cause a dramatic alteration of cellular functions, such as protein synthesis, cell homeostasis, cell cycle progression, vesicular trafficking or actin cytoskeleton. Enterotoxins (cholera toxin [CT] or STb), which selectively upregulate fluid secretion are the most potent diarrheagenic toxins. Disorganization of the actin cytoskeleton induces alterations of intestinal barrier functions, leading to passage of fluids into the intestinal lumen, but to a lesser degree. Intracellularly active toxins usually display a unique enzymatic activity, but they can stimulate several cell signaling pathways, which often results in intestinal cell necrosis, leading to a mid-to-severe inflammatory response. The inflammatory process, which can also be triggered by other toxin-induced cell signaling, contributes to the damage of the intestinal mucosa.

Enterotoxins Modifying Cell Homeostasis

CT & Heat-labile Enterotoxins Cholera is a serious epidemic disease characterized by severe diarrhea and dehydratation, caused principally by CT. Other members of the cholera toxin family are the E. coli heat-labile enterotoxins (LT)-I and-II. The CT gene is localized to filamentous bacteriophage DNA, and can be chromosomally integrated or replicated as a plasmid.[89] By contrast, the heat-labile enterotoxin genes are located on plasmids (LT-I) or are integrated into the chromosome (LT-II). CT and LT subunits are exported across the bacterial membrane by the type II secretion pathways, and assemble in the periplasm. In V. cholerae, CT is actively secreted through the outer membrane, while the release of LT-I depends on cell lysis.[90–92]

Similar to Shiga toxin (Stx), CT and LTs consist of an A-subunit (28 kDa), and five B-subunits (11 kDa each) assembled in a pentamer (AB5 structure). The A-subunit is proteolytically activated by a V. cholerae endopeptidase into two components A1 (22 kDa) and A2 (5.5 kDa), which remain linked by a disulfide bridge. The C-terminal part of A2 extends through the central pore of the B pentamer, and is linked noncovalently to the B-subunits.

Cholera toxin, via the B-subunits, binds with high affinity to GM1 ganglioside, which is distributed in the detergent-resistant membrane fraction of enterocytes. GM1 directs the toxin into lipid rafts, which facilitates toxin entry into noncoated vesicles, but also into clathrin-coated vesicles. Alternative entry pathways have also been observed, including Arf6- and dynamin-dependent or -independent routes. Whatever the entry mechanism, CT is directed to the Golgi via early and recycling endosomes, and then in caveolin-containing intermediate vesicles in a Rab7- and Rab9-dependent manner. Then, CT migrates from the Golgi to the ER via coatomer I-coated vesicles. The C-terminal sequence of the A2 fragment contains an ER-retention sequence (KDEL), which recognizes the receptor Erd2p and directs the retrograde Golgi–ER trafficking of CT. However, B-subunits lacking an ER retention signal or CT mutated on the KDEL motif are also transported to the ER, via an unknown mechanism. Dissociation of the A-subunits from the B-pentamer in the Golgi or ER is still under discussion. There is some evidence that the A1-subunit is released from the A2/B complex through a protein disulfide isomerase in the ER. Then, A1 translocates into the cytosol by the Sec61 secretion channel in the ER membrane. The Sec61 complex usually transports newly synthesized proteins from cytosol into the ER, and also misfolded proteins from the ER to proteosomal degradation in the cytosol, a process known as the ER-associated degradation. The A1-subunit uses the retrotranslocation Sec61 pathway, but escapes from the proteasomal degradation, probably owing to its poor content in lysines, which are sites of ubiquitination and proteasomal targeting. The A1 fragment is responsible for the intracellular enzymatic activities of the toxin in the presence of the membrane Arf factor, including NAD hydrolysis in ADP ribose and nicotinamide, and transfer of ADP ribose to Arg-187 of the α-subunit of the stimulatory protein Gsα, leading to a continuous stimulation of adenylyl cyclase and elevated intracellular cAMP. The increased cAMP levels lead to an activation of protein kinase A, which, subsequently, phophorylates numerous substrates in the cell including CFTR. This results in active Cl secretion by intestinal crypt cells and a decrease of NaCl-coupled absorption by villus cells, which is accompanied by a massive osmotic efflux of water into the intestinal lumen (Figure 4). This results in a severe secretory and noninflammatory diarrhea. Cholera patients might lose up to 80 l diarrhea. In contrast to V. cholerae, LT-producing E. coli cause milder diarrheic disease. Several reasons might explain this difference in disease severity: different intestinal colonization and toxin secretion between V. cholerae and E. coli (LT accumulates in the periplasm whereas CT is secreted), production of additional enterotoxins by V. cholerae (ZOT, cholix and ACE), greater stability between A- and B-subunits of CT compared with LT due to a different amino acid segment in the A2 subunit in the two toxins, and stimulation of 5-HT release from enterochromaffin cells by CT but not by LT, which contributes to intestinal secretion.[90,92–95]

E. coli STb STb is a small-peptide, heat-stable enterotoxin produced by certain E. coli strains, mainly enterotoxigenic E. coli from pigs, more rarely from other animal species (e.g., chicken, cattle, ferret, marmot and dog), and occasionally from humans. This toxin is frequently associated with diarrhea in pigs. STb forms a distinct class of enterotoxins from STa. In contrast to STa, STb is insoluble in methanol. STb is synthesized as a 71-amino acid precursor, containing an N-terminal signal peptide (23 amino acids). The mature protein (5.2 kDa) forms two intracellular disulfide bonds. STb in monomeric or oligomeric (hexamer/heptamer) forms bind to a specific receptor on the cell membrane, which is a glycosphingolipid termed sulfatide, present on intestinal cells of pigs and humans. The mechanism of STb-mediated fluid secretion is still a matter of debate. STb is internalized into cells by a yet undefined actin cytoskeleton-dependent process, activates an intracellular target (heterotrimeric protein Gαi3) and forms pores through the plasma membrane leading to Ca2+ influx. This results in activation of several intracellular signaling pathways, such as opening of ion channels, protein kinase C and subsequent CFTR stimulation, activation of phospholipases A2 and C with release of arachidonic acid, and formation of prostaglandins E2 and 5-HT, leading to an efflux of water and electrolytes from intestinal cells.[96–99]

Enterotoxins Modifying the Actin Cytoskeleton & Intercellular Junctions

Actin is one of the most abundant molecules in eukaryotic cells, and participates in numerous and essential functions, such as maintaining the polarized architecture of epithelial cells, cell motility and migration, motility of intracellular organelles, endocytosis, exocytosis, phagocytosis, cytokinesis and control of intercellular junctions. Actin is in a dynamic equilibrium between monomeric and filamental forms. Actin polymerization is a highly spatially and temporally coordinated process, in which Rho GTPases have an essential regulatory role. Rho GTPases are active when bound to GTP and inactive in the GDP form, and control extrinsic or intrinsic signals to downstream pathways involved in polymerization/depolymerization of actin filaments to adapt cellular responses involving the actin cytoskeleton. Bacterial toxins modifying the actin cytoskeleton essentially act at two levels of the actin-regulation network, actin monomers and Rho GTPases, probably because they represent the most critical regulatory and structural molecules. The main consequence of alteration of the actin cytoskeleton is a loss of intestinal barrier integrity, leading to the leakage of fluid and electrolytes (diarrhea) and lesions of the intestinal mucosa (necrotic and/or hemorrhagic enteritis). Thereby, C. difficile toxins, clostridial binary toxins, cytotoxic necrotizing factors and E. coli plasmid-encoded toxins (PETs) are directly involved in diarrhea frequently accompanied by necrotic lesions of the intestinal mucosa. In addition, C. difficile toxins induce strong intestinal inflammation, which is particularly prominent in the pseudomembranous colitis.

C. difficile ToxinsC. difficile toxins A and B (TcdA and TcdB) are members of the clostridial glucosylating toxin family, which also encompasses Clostridium sordellii lethal toxin and hemorrhagic toxin, and Clostridium novyi α toxin. C. difficile is the etiological agent of pseudomembranous colitis and approximately 30% of cases of postantibiotic diarrhea, which are the most frequent nosocomial intestinal diseases.[100] TcdA, which experimentally induces necrotic and hemorrhagic intestinal lesions, was considered as the main virulence factor.[101,102] However, using genetically modified C. difficile strains and the hamster disease model, TcdB was found to be the essential virulence factor.[103] However, a recent study with isogenic tcdA or tcdB mutations of C. difficile shows that both toxins, TcdA and TcdB, are critical to induce C. difficile pathological effects.[104] Since C. difficile strains producing both TcdA and TcdB or only TcdB cause enteric disease in humans, TcdB might also be an important enterotoxin.[101,102] Both TcdB and TcdA participate in the alteration of the intestinal barrier and in the recruitment of inflammatory cells, which are abundant in the lesions.

TcdA and TcdB are single-chained proteins with molecular masses of 307 and 270 kDa, respectively. containing several functional domains, which mediate their intracellular activity. TcdA and TcdB C-terminal domains contain multiple repeated sequences and are involved in cell surface receptor recognition. A trisaccharide (Gal-α1–3Gal-β1–4GlcNac) has been found to be the motif recognized by TcdA. The central domain contains hydrophobic sequences that are thought to mediate the translocation of the toxin across the membrane and the enzymatic and cytotoxic activity (DxD motif) of the toxins is supported by the N-terminus. The N-terminal domain (amino acids 1–543 in TcdB) is delivered into the cytosol by an autoproteolytic process stimulated by inositol hexakisphosphate. A cysteine protease domain has been identified close to the cutting site in TcdB (amino acids 544–767), which is conserved in all clostridial glucosylating toxins.[102,105]

Clostridial glucosylating toxins enter cells by receptor-mediated endocytosis. The cytotoxic effects are blocked by endosomal and lysosomal acidification inhibitors (monensin, bafilomycin A1 and ammonium chloride), and the inhibiting effects can be bypassed by an extracellular acidic pulse. This indicates that the clostridial glucosylating toxins translocate from early endosomes upon acidification. At low pH, TcdB induces channel formation in cell membranes and artificial lipid bilayers, and shows an increase in hydrophobicity.[106,107] This is thought to involve a conformational change and insertion of the toxin into the membrane mediated by the hydrophobic segment of the central domain (amino acids 965–1128 in TcdB). The N-terminal domain is then delivered into the cytosol by an autoproteolytic activity stimulated by inositol hexakisphosphate.[108–110]

Clostridial glucosylating toxins catalyze the glucosylation of 21-kDa G-proteins using UDP-glucose as the sugar donor (with the exception of α-novyi toxin, which preferentially uses UDP-N-acetylglucosamine). The toxins transfer the glucose or N-acetylglucosamine moiety to the acceptor amino acid Thr37 of Rho or Thr35 of Rac, Cdc42 and Ras proteins.[111,112] Rho complexed to GDP dissociation inhibitor is not a substrate for glucosylation, and modified Rho does not bind to GDP dissociation inhibitor.[113]

The conserved glucosylated Thr (Thr37/35) is located in switch I of Rho/RasGTPases. Thr37/35 is involved in the coordination of Mg2+ and the subsequent binding of the β- and γ-phosphates of GTP. The hydroxyl group of Thr37/35 is exposed at the surface of the molecule in its GDP-bound form, which is the only accessible substrate for glucosylating toxins. Crystal structure analysis of Ras modified by C. sordellii lethal toxin shows that glucosylation prevents the formation of the GTP conformation in the effector loop of Ras, which is required for the interaction with the effector Raf.[114] Similar results were found when RhoA glucosylation by ToxB was studied.[115] Hence, glucosylation of RhoGTPases by C. difficile toxins block these proteins in their inactive conformation preventing their interaction with downstream effectors such as those involved in actin filament polymerization (Figure 5). In addition, it has been shown that glucosylation of GTPase by the toxins reduces the intrinsic GTPase activity, completely inhibits GAP-stimulated GTP hydrolysis,[116] and leads to accumulation of the GTP-bound form of Rho at the membrane.[113]

Figure 5.

Representative enterotoxins modifying the actin cytoskeleton.
Clostridial glucosylating toxins, such as Clostridium difficile toxin A, enter cells by receptor-mediated endocytosis and deliver the N-terminal catalytic domain into the cytosol from acidified early endosomes via an autoproteolytic process. Then, they inactivate Rho GTPases by glucosylation of Thr35/37 located in switch I leading to actin depolymerization and subsequent alteration of intercellular junctions. Clostridial binary toxins also enter cells by receptor-mediated endocytosis. The enzymatic component is transported into the cytosol through a pore formed by the binding components in the endosomal membrane, and modifies the actin monomer by ADP ribosylation, impairing actin filament polymerization. This results in actin cytoskeleton depolymerization and alteration of intercellular junctions. Bacteroides fragilis enterotoxin or fragylisin exerts a protease activity on E-cadherin at the cell surface and thus alters the actin cytoskeleton and intercellular junctions.
ADPR: ADP-ribose.

The modification of Rho proteins by the clostridial glucosylating toxins induces cell rounding, the loss of actin stress fibers, reorganization of cortical actin, disruption of the intercellular junctions and increased cell barrier permeability. TcdA and TcdB disrupt apical and basal actin filaments and subsequently disorganize the ultrastructure and component distribution (ZO-1, ZO-2, occludin and claudin) of tight junctions, whereas E-cadherin junctions show little alteration.[117,118] By contrast, C. sordellii lethal toxin, which only modifies Rac among the Rho proteins, alters the permeability of intestinal cell monolayers causing a redistribution of E-cadherin, whereas tight junctions are not significantly affected.[119] The signaling pathway downstream of Rho protein inactivation (mainly Rac) by clostridial glucosylating toxins, leading to actin filament depolymerization includes dephosphorylation of paxillin, one of the major of focal adhesion molecules, and disorganization of focal adhesion complexes.[120,121] In addition, TcdA disrupts the tubulin filament network by deacetylation of monomeric tubulin.[122]

Besides their effect on the actin cytoskeleton and intercellular junctions, TcdB and TcdA trigger cell apoptosis or necrosis as a consequence of Rho glucosylation and RhoB overexpression.[102] Cell necrosis, which results in release of cell constituents and, subsequently, in an inflammatory response, seems critical in C. difficile-associated disease. Inactivation of Rho proteins also impairs other cellular functions such as endocytosis, exocytosis, NADPH oxidase regulation and transcriptional activation mediated by JNK and/or p38.[123]

E. coli Cytotoxic Necrotizing Factor The cytotoxic necrotizing factor (CNF) is produced by some pathogenic strains of E. coli and consists of two variants, CNF1 and CNF2. CNF1 is synthesized by human strains mainly isolated from urinary tract infections and neonatal meningitis, whereas CNF2 is primarily produced by strains of animal origin that are often involved in enteritis and septicemia of calves.[124]

E. coli CNF1 and CNF2 are highly related proteins (86% sequence identity) of approximately 110 kDa, and are released from E. coli into the local environment, bind to laminin receptors on the cell surface and possibly other receptors.[125] The N-terminal domain (amino acids 135–164), but also a C-terminal part (amino acids 683–730), are involved in the recognition of the cell receptor.[126]

Structurally, the CNF toxins contain three main functional regions: an N-terminal domain (amino acids 1–299) involved in binding to a cell surface receptor, a central domain (amino acids 299–720) containing two hydrophobic regions that likely enable translocation of toxin across the cell membrane, and the C-terminal (residues 720–1014) catalytic domain possessing deamidase activity, which retains an original protein fold consisting of a single compact domain with a central β-sandwich surrounded by helices and a catalytic triad (Val, Cys and His), located in a deep pocket.[124,127] CNF enters cells via a nonclathrin pathway, which is also dynamin and intersectin independent.[128] The catalytic domain translocates from late endosomes into the cytosol upon an acidic pH. A serine protease cleavage site has been identified upstream of the C-terminal domain (amino acid 532–544). The catalytic domain of CNF inserted in the membrane of acidifed endosome is cut, probably by cell protease, and then released into the cytosol.[129] CNF preferentially targets Rho and catalyzes deamidation of Gln63 to Glu, but Rac and Cdc42 are also modified at an equivalent residue (Gln61), which is located in the GTPase domain of RhoGTPases. Deamidation of Gln63 (Gln61) into Glu impairs binding of a water molecule required for GTP hydrolysis; therefore, CNF blocks Rho GTPases into a biologically active form linked to GTP.[125] CNF–Rho GTPase activation is only transient, modified GTPases are then directed to ubiquitination and proteasomal degradation.[125] Transient activation of Rho GTPase results in stimulation of downstream signaling pathways leading to multinucleation, increased actin polymerization, reorganization of stress fibers, membrane ruffles, as well as increased phagocytic activity in human epithelial cells. In intestinal cell monolayers, CNF1 decreases the transepithelial electrical resistance and enhances paracellular permeability from the basal-to-apical surface. Concomitantly, tight-junction proteins are redistributed in endosomal caveolar compartments, but microvillus F-actin and its binding protein, villin, disappear.[130,131] CNF-dependent activation of RhoA seems to be the main signaling pathway leading to increased paracellular permability, whereas activation of Rac and Cdc42 contribute to stabilization of epithelial barrier functions.[132] Moreover, modification of the actin cytoskeleton by CNF influences transcriptional signaling leading to upregulation of MAPK and NF-κB pathways with release of proinflammatory cytokines, such as IL-8, and induction of an inflammatory response.[125]

Clostridial Binary Toxins The clostridial binary toxins have a common structure, consisting of two independent protein components that are not covalently linked, one being the binding component (100 kDa), and the other the enzymatic component (45 kDa). Both components are required for biological activity. Two families can be distinguished. The ι-family encompasses ι-toxin produced by C. perfringens type E, Clostridium spiroforme toxin and C. difficile ADP-ribosyltransferase (CDT) synthesized by some strains of C. difficile. The C2 family corresponds to the C2 toxins from C. botulinum C and D. At the amino acid sequence level, the components of the ι-family are highly related (80–85% identity), whereas C2 toxin shows 31–40% identity with the ι-family proteins. Components from ι and C2 families show no immunological crossreaction and no functional crosscomplementation (reviewed in[133]). The binary toxin-producing Clostridium species are involved in necrotizing enteritis and diarrhea in animals and, occasionally, in humans (discussed later).

The binding component, which is structurally related to the protective antigen of B. anthracis toxin,[134] binds to the surface of the target cell, and is essential for the uptake of the toxin into the cell. For this, the binding component has to be activated by protease cleavage. In solution, the binding components of ι and C2 toxins (Ib and C2-II, respectively) can be processed by trypsin or α-chymotrypsin. However, unprocessed Ib and C2-II can also bind to the cell surface receptor, but do not mediate the entry of the enzymatic component.[135] The processed binding component recognizes specific cell membrane receptors, heptamerize in lipid rafts and form small ion permeable channels that trap the enzymatic component into endocytic vesicles.[136] Alternatively, binding and enzymatic components might assemble in solution prior to binding to the cell surface receptor.[137] There is evidence that ι and C2 toxins are endocytosed via a Rho-/dynamin-dependent and clathrin-independent pathway,[138] although a contradictory result suggests a clathrin-dependent route.[139] The enzymatic component is subsequently translocated into the cytosol upon acidification from early endosomes.[140–142] In contrast to C2 toxin, ι-toxin also requires a membrane potential gradient to translocate the enzymatic component into the cytosol.[143] Moreover, host cell chaperone proteins, such as heat-shock protein 90, and peptidyl-prolyl cis/trans isomerases, such as cyclophilins, facilitate the translocation and refolding of the enzymatic component in the cytosol.[144–146] C2 toxin endocytosis also requires the activation of phosphatidylinositol 3-kinase (PI3K), phosphoinositide-dependent kinase 1 and Akt signaling pathway by a currently undefined mechanism.[147]

The enzymatic component catalyzes the ADP-ribosylation of actin monomers at Arg-177 but not of polymerized F-actin, since Arg177 is located in the actin–actin binding site. While toxins of the ι-family modify all actin isoforms, including cellular and muscular isoforms, C2 toxins only interact with cytoplasmic and smooth muscle α-actin. The cumbersome ADP-ribose at the actin-binding site prevents the nucleation and polymerization of ADP-ribosylated actin monomers. Moreover, ADP-ribosylated actin acts as a capping protein; it binds to the barbed end of the actin filament and inhibits the further addition of unmodified actin monomers. Actin filaments depolymerize at the pointed end and the released actin monomers are immediately ADP-ribosylated (Figure 5). In addition, ADP-ribosylation inhibits the intrinsic ATPase activity of actin. Cell microinjection of ADP-ribosylated actin monomers induces the same effect as C2 or ι-toxin. This results in a complete disassembly of the actin filament and accumulation of actin monomers.[148,149] Moreover, clostridial binary toxins induce a reorganization of the peripheral microtubule network with microtubule-based protrusions at the surface of intestinal epithelial cells.[150] As a consequence, cells become rounded, detach from the surface and die. Studies with epithelial and endothelial cells have shown that clostridial ADP-ribosylating toxins alter the tight and adherens junctions resulting in a loss of cell barrier function leading to a passive efflux of fluids in the intestinal lumen.[65,149,151,152] Finally, cells die by a delayed caspase 8- and 9-dependent apoptotic process, which depends on cell type.[153]

Clostridial binary toxins are responsible for intestinal dysfunction, since Clostridium species, which only produce a binary toxin as their major toxin, are the causative agents of enteritis. Thereby, C. perfringens E causes enterotoxemia in calves and other young animals. C. spiroforme is responsible for enteritis and death in rabbits and rarely in humans. C. botulinum C2 toxin induces intestinal hemorrhagic lesions in birds.[65] The role of CDT in the pathogenesis of C. difficile is still questionable. The recent emerging epidemic C. difficile strain 027 produces CDT in addition to TcdA and TcdB and seems to be associated with severe cases of C. difficile-associated disease.[154–158] Rare C. difficile strains only produce CDT in the absence of the glucosylating toxins and their clinical relevance is currently unclear.[159]

Plasmid-encoded Toxin Plasmid-encoded toxin is produced by EAggEC, which is associated with persistent diarrhea in developing and industrialized countries, including in HIV-patients, traveller's diarrhea, food/water outbreaks of diarrhea, and sporadic cases of diarrhea. PET is a 140 kDa protein from the autotransporter class of proteins, which contain a self-secretion system and is secreted in the external medium through a type V secretion system. PET consists of a N-terminal signal sequence, a C-terminal translocation unit (also called the β-domain) able to form a β-barrel in the outer membrane, and a passenger domain (108 kDa) which is the mature active domain in the external medium. After binding to a specific receptor on the plasma membrane, PET enters cells via clathrin-dependent endocytosis, and migrates from early endosomes to the Golgi apparatus and then to the endoplasmic reticulum, where it is translocated into the cytosol via the Sec61p secretion system. Thereby, PET trafficking in cells is similar to that of CT and Stx. The main cellular effects induced by PET consist of actin cytoskeleton alteration with loss of actin stress fibers, disorganization of focal contacts, cell rounding and increased permeability of intestinal cell monolayers. PET is a serine protease, which cleaves α- and β-spectrin from erythroid cells and fodrin, a related protein in epithelial cells, which is an actin- and calmodulin-binding protein coupled to sodium channels. The proposed model of PET activity includes degradation of α-fodrin and subsequent actin cytoskeleton disruption and impairment of Na+ entry into cells, resulting in the diarrheal pathogenesis.[160,161]

Enterotoxins Directly Altering Cell Viability

Intracellularly active toxins can target molecules that have a pivotal role in cell survival and therefore compromise the cell viability very efficiently. These, toxins act at the DNA level, causing cell cycle arrest (cytolethal distending toxins), at the translational level (rRNA and elongation factor), inducing an inhibition of protein synthesis (Stx and cholix toxin) or at the mitochondrial level triggering programmed cell death (VacA). Such effects in enterotcytes lead to necrotic lesions of the intestinal mucosa accompanied by a more-or-less severe inflammatory response. However, these toxins can act in other cell types, such as lymphocytes and, thus, induce an immunosuppression facilitating the local development of the pathogenic bacteria, or at a distance causing lesions in other organs.

Shiga Toxin Shiga toxin (Stx), Shiga-like toxins, verotoxins or verocytotoxins are expressed by several enteric pathogens, including Shigella dysenteriae and enterohemorrhagic E. coli. S. dysenteriae produces Stx, and enterohemorrhagic E. coli Stx1 (one amino acid difference from Stx), Stx2 (56% amino acid sequence identity with Stx1) and genetic variants of the two toxin types. This group of toxins play an important role in the pathogenesis of a number of severe complications, such as hemorrhagic colitis, the hemolytic uremic syndrome and CNS complications. Shigellosis is mostly a pediatric disease ranging from watery diarrhea to acute inflammatory colitis. Food intoxication with Stx-producing E. coli strains, most commonly serotype O157:H7, results in hemorrhagic enteritis and extraintestinal complications, including renal failure, neurological syndrome and sometimes death.[162]

Stxs are structurally related to CT, and are composed of a catalytically active subunit (A-subunit) and a receptor-recognition subunit (B-subunit). The B-subunit that recognizes the cell surface receptor globotriosyl ceramide Gb3, consists of five B-fragments that form a symmetrical ring-like structure in solution. The catalytic domain is located in the A-subunit, which is activated by proteolytic cleavage leading to two fragments (A1 and A2) that are linked together by a disulfide bridge. Stx assembles in the periplasm and is released by phage-mediated bacterial lysis.[162]

Several studies have previously shown that Stx enters the cell by a clathrin-dependent pathway, and is then transported directly from early/recycling endosomes to the Golgi apparatus and then to the ER.[163] However, a clathrin-independent mechanism has also been described involving lipid rafts.[164–166] Association of the receptor Gb3 to lipid rafts is crucial for the intracellular transport of B-subunits. Hydroxylation, chain length, degree of unsaturation of Gb3 fatty acids, as well as the membrane characteristics, such as cholesterol levels, may influence the lateral mobility and presentation of the polar head on the cell surface, which is recognized by the toxin. Interestingly, binding of B-subunits to Gb3, preferentially to long-chain, unsaturated Gb3 isoforms, induces lipid reorganization characterized by a lipid compaction that favors negative membrane curvature and a tubule formation-dependent endocytosis.[167,168] Thereby, Stx, once bound to its receptor, induces plasma membrane invagination without the help of clathrin, and endocytic vesicles are then processed by cellular factors, including dynamin, actin and membrane cholesterol. Transport from early endosomes to the Golgi requires clathrin and retromer, which is comprised of a curvature subunit (nexins) and a cargo recognition subunit (three vacuolar protein sorting-associated proteins). Unlike CT, Stx is transported from the Golgi to the ER in a COPI-independent manner. Proteolytic activation of the catalytic A1 fragment probably occurs in the trans-Golgi network and/or in the ER by the action of furin and, to a lesser extent, by other cellular proteases. The disulfide bond linking the A1 fragment to the A2–B complex is reduced in the ER. Subsequent retrotranslocation of Stx A1 fragment into the cytosol through the Sec61 translocon is similar to that of CT.[162]

The A1 fragment is released into the cytosol and inactivates the 60S subunit of host cell ribosomes by cleaving the N-glycosidic bond of adenosine4324 of the 28S rRNA of the 60S subunit. This induces a dramatic inhibition of cellular protein synthesis (Figure 4). Stx-dependent inhibition of protein synthesis induces a stress response, known as the ribotoxic stress response, leading to activation of MAPK cascades (JNK, p38, and ERK) and subsequent apoptosis. Signaling pathways from Stx-mediated MAPK activation to apoptosis include alteration of the balance of pro- and anti-apoptotic Bcl-2 proteins, but remain to be clearly defined. Apoptosis might also result from interaction of B-subunits with the Gb3 receptor in a caspase 1- and 3-dependent manner.[169] Stx binding to receptor also triggers additional signaling pathways, including activation of Src family kinases, leading to various cellular effects, such as actin cytoskeleton remodeling or cytokine release. Cellular effects induced by Stxs depend on the cell types. In the initial steps of intestinal infection, Shigella enter enterocytes and damage the cells, generating ulcerations of the mucosa and an inflammatory response. Stx is not the essential virulence factor, since human enterocytes express a low level of Gb3. However, in Gb3 negative enterocytes, Stx can be transported by transcytosis or a paracellular route to the submucosa. Stx interacts with endothelial cells of microvessels inducing apoptosis and damage to the microcirculation, which exacerbate the mucosal lesions. In addition, Stx induces the release of proinflammatory cytokines (e.g., IL-8 and TNF) from monocytes or macrophages, as well as of reactive oxygen metabolites from polymorphonuclear leukocytes (PMNs), which contribute to the tissue injury. Stx delays apoptosis of PMNs and can use these cells for its transport through the circulation to other organs, such as kidneys. Renal tubular epithelial cells and glomerular endothelial cells express Gb3 and are susceptible to Stx-mediated apoptosis leading to kidney failure.[162,169]

Cholix Toxin Cholix toxin is an additional toxin produced by V. cholerae. Genome analysis of non-O1, non-O129 strains have revealed a gene encoding a putative toxin related at the amino acid sequence level (32% identity) to exotoxin A of Pseudomonas aeruginosa. Cholix toxin gene is widely distributed among V. cholerae strain. Cholix toxin retains a three structural domain organization, similar to exotoxin A, corresponding to the three main steps of activity (binding to cell surface receptor, translocation into the cytosol, and enzymatic modification of an intracellular target). As exotoxin A, cholix toxin (71 kDa) enters eukaryotic cells via low-density lipoprotein receptor-related protein and probably other membrane receptors, and then catalyzes ADP-ribosylation of eukaryotic elongation factor 2, resulting in protein synthesis arrest and cell death (Figure 4).[170] The role of cholix toxin in the pathogenesis of cholera is not yet fully understood.

Cytolethal Distending Toxins Cytolethal distending toxins (CDTs) belong to a family of toxins that cause irreversible cell cycle arrest, alteration of epithelial barrier integrity and, ultimately, death of the target cells. CDT was first described in 1987 when certain strains of E. coli were found to cause cytopathic effects that were distinct from those induced by E. coli toxins, such as LT, ST, verotoxin and hemolysin. Cells that are sensitive to CDT first increase in size (three- to five-fold), followed by a slowly developing cell distention, which finally leads to cell death. Apart from E. coli, CDTs are produced by a wide variety of Gram-negative bacteria, including Shigella, Hemophilus ducreyi, Actinobacilus actinomycetemcomitans, H. pylori and Campylobacter.[171] In E. coli, it has been shown that CDT is encoded by three adjacent or slightly overlapping genes: cdtA, cdtB and cdtC, all of which are required for the activity of the toxin. While CdtB contains the enzymatic activity, CdtA and CdtC are required for the translocation of CdtB into the target cell. Internalization of CDT occurs via endocytosis mediated by clathrin-coated pits. The toxin has been shown to traffic through the Golgi apparatus into the cytosol and then the nucleus. The proposed mechanism of action of CDTs is not yet fully elucidated. However, it has been reported that the toxin blocks cells in the G2 phase of the cell cycle by preventing dephosphorylation of the inactive form of cdc2. In addition, CdtBs possess DNase I activity that causes dsDNA breaks (Figure 4).[172–175]

VacA The vacuolating cytotoxin (VacA) is one of the most important virulence factors produced by Helicobacter pylori, a causative agent of severe gastric diseases such as gastritis, peptic ulcers, gastric adenoma and lymphoma. VacA is a multifunctional toxin that is involved in initial bacterial colonization, persistence of H. pylori in the stomach, and contributes to pathological alterations. VacA has been shown to induce large cytoplasmic vacuoles in cultured cells and apoptosis in gastric epithelial and parietal cells.[176]

VacA is released from bacteria by the autotransporter type V secretion system. Cleavage of the secreted VacA protein (95 kDa) results in an N-terminal 34–37 kDa (p34) and a C-terminal 58 kDa (p58) fragment that remain associated with each other. The p58 fragment mediates VacA monomer binding to the target cell.[176] Several receptors have been identified including sphingomyelin, protein tyrosine phosphatase on gastric cells, and β2 integrin on T lymphocytes.[177,178] This results in VacA oligomerization in the membrane and the formation of anion-selective channels that release bicarbonate, chloride and urea from the cell cytosol.[179] Release of urea, which is the substrate of urease, an enzyme produced by H. pylori, contributes to buffering the acidic pH of the gastric mucus layer, a condition required for H. pylori colonization. In addition, VacA increases the paracellular permeability of epithelial cell monolayers, resulting in the release of small molecules, such as ions, sugars and amino acids, by a currently unidentified mechanism, but apparently not dependent on toxin entry into the cell.[176] Moreover, VacA reduces acid production from gastric parietal cells, which contributes to the development of chronic atrophic gastritis. This effect is attributed to a VacA-dependent influx of Ca2+, probably through channel formation in the plasma membrane, accompanied by a subsequent calpain activation, mobilization of internal Ca2+, and specific proteolytic cleavage of ezrin, an actin-binding protein from the ezrin/radixin/moesin family. Proteolyzed ezrin is redistributed from the apical membrane of parietal cells to the cytosol, resulting in a rearrangement of actin filaments in apical microvilli and inhibition of recruitment of H,K-ATPase to the apical membrane leading to hypochlorhydria. Calpain activation also causes necrosis and cell parietal loss, characteristic of chronic atrophic gastritis.[180]

VacA toxin channels are also internalized and transported to the late endosomal compartments where they change the anion permeability, leading to an enhancement of vacuolar ATPase proton pump activity. VacA internalization is not yet fully understood, and involves a clathrin-independent and a Cdc42-dependent pinocytic mechanism.[181] p34 fragment of VacA targets mitochondria through its N-terminal part, which binds to the outer membrane TOM complex, usually used for importing of endogeneous mitochondrial proteins, and forms heptameric ion channels. This results in release of cytochrome c, reduction of the mitochondrial transmembrane potential, activation of caspase 3, DNA fragmentation and cell apoptosis.[182] In gastric epithelial cells, VacA induces programmed necrosis, which is characterized by release of lactate dehydrogenase (LDH) and HMGB1 and decrease in intracellular ATP in a caspase-independent manner resulting from activation of poly(ADP–ribose) polymerase (PARP). In contrast to apoptosis, cell necrosis, which results in membrane lysis and release of intracellular compounds, efficiently triggers an inflammatory response.[183] In addition, VacA modulates various cellular functions, such as activation of MAPK, p38 and ERK1/2 and PI3K/Akt signaling pathways, which are not dependent on toxin internalization, but likely result from interaction of VacA binding to cell surface receptor, leading, according to the cell type, to an inflammatory response (as in gastric epithelial cells) or localized immunosuppresion (inhibition of T lymphocytes and phagocytes).[176,178] Thus, VacA has multiple roles, directly involved in alteration of the gastric mucosa integrity, but also implicated in the control of the immune system by inducing proinflammatory effects and local immunosuppression, permitting H. pylori colonization and persistence in the stomach mucosa and contributing to the development of peptic ulceration or gastric cancer.