Introduction

Shiga toxins comprise a group of structurally and functionally related type 2 ribosome-inactivating proteins (RIPs) that are produced in several types of bacteria. The name of Shiga toxins refers to a Japanese bacteriologist, Kiyoshi Shiga, in honor of his research on isolation and studies of a gram-negative bacteria named Bacillus dysenteriae, from the stools of patients suffering from dysentery during a large epidemic in the nineteenth century. The bacterium was renamed as Shigella dysenteriae for his discovery, and the toxin released was named Shiga toxin (Stx). The toxin was firstly described as a neurotoxin, as the extract of bacterial lysates containing Stx caused limb paralysis when injected intravenously to rabbits, until the 1970s when researchers revealed the relevance of Stx to gastrointestinal toxicity (Keusch et al. 1981).

A few years later, research revealed that some strains of Escherichia coli (E. coli) produced toxins that killed Vero cells in in vitro studies. They named the toxins as verotoxins and those E. coli strains as verotoxin-producing E. coli (VTEC) (Konowalchuk et al. 1978). Soon, it was found that the verotoxins behaved like Stx to produce similar symptoms like hemorrhagic colitis (HC) and hemolytic uremic syndrome (HUS). Application of antibodies against Stx also worked to neutralize the verotoxins, indicating that Stx and verotoxins have a remarkable identity; thus, the verotoxins were re-designated as Shiga-like toxins (SLTs), and the VTEC strain was renamed as Shiga-like toxin-producing E. coli (STEC). As two very different bacterial species produced such a similar toxin, it was believed that there was a transfer of genes to the bacteria. This was proved by the identification of lambdoid bacteriophages that encode Shiga toxins. The Stx-phages incorporate their genome into host bacterial cells and produce the toxin. Stx-phages could target not only S. dysenteriae and E. coli but also some commensal bacteria in the gut to yield more Shiga toxins (Cornick et al. 2006), increasing the severity of symptoms in the gastrointestinal areas produced by the toxins during infection.

Stx-phages have a high degree of heterogeneity. They infect different strains of E. coli and produce variants of Shiga toxins. Two major types of Shiga toxins produced from STEC were identified and named Shiga toxin 1 (Stx-1) and Shiga toxin 2 (Stx-2), respectively. Together with the original Stx, they are the main members of Shiga toxins. Stx-1 shows remarkable homology in amino acid sequence with Stx found in S. dysenteriae. Stx-1 can be neutralized by antibodies raised against Stx, as they only differ in only one amino acid residue (Thr in Stx; Ser in Stx-1 in the 45th residue of the A chain). Stx-2, on the other hand, shows only around 55 % similarity in sequence with Stx, and antibodies of Stx cannot work on Stx-2 (Calderwood et al. 1987).

Structure of Shiga toxins

Although Stx and Stx-1 show obvious variation in their sequences with Stx-2, these Shiga toxins share the same structure. They are typical members of the RIP family. RIPs are produced in various plants, bacteria, and fungi (Jiménez et al. 2015; Stirpe 2013). There are three major types of RIPs. Type 1 RIPs (e.g., saporin) consist of a single ∼30-kDa polypeptide chain exhibiting N-glucosidase activity that depurinates conserved adenine residue in the α-sarcin/ricin loop (SRL) of 28S ribosomal RNA, bringing about inactivation of ribosomes (Wiese et al. 2013). Type 2 RIPs (e.g., ricin) consist of an A chain similar to type 1 RIPs and also one or more B chains that aids/aid binding and entrance of the toxin to the target cells (Fang et al. 2012). Type 3 RIPs are atypical ones with small molecular sizes (Pan et al. 2013). Shiga toxins belong to type 2 RIPs. They own the AB5 structure, that is, a single A subunit non-covalently linked to five identical B subunits (Ng et al. 2010). The 32-kDa A subunit consists of the 27.5-kDa A1 fragment and 4.5-kDa A2 fragment that are linked with a disulfide bond between the residues 242 and 261. The five identical 7.7-kDa B subunits form a pentameric ring, surrounding the A subunit near the C-terminus, in the region of the A2 fragment (Fraser et al. 1994). Some physical and chemical properties of Stx, Stx-1, and Stx-2 are described in Table 1.

Table 1 Physical and chemical properties of Stx, Stx-1, and Stx-2
Full size table

It is interesting that the bacteria-derived Shiga toxins share similar basic structures with some plant-originated RIPs like ricin. The catalytic A chain of ricin has a molecular weight (32 kDa) similar to that of Shiga toxins. Their amino acid sequences show a certain degree of similarity (23 %) though they originate from totally different species. Crystal structures of the A chains of Shiga toxins and ricin also show similar patterns of protein folds (Fraser et al. 1994; Katzin et al. 1991). Several amino acids at their active sites are conservative with each other, providing evidence for their common N-glucosidase activity. On the other hand, ricin and Stx possess B subunits with totally different structures. The B subunit of ricin is a single 34-kDa galactose-specific lectin chain, while the 7.7-kDa Stx B subunits form a pentameric structure, which possesses a carbohydrate binding specificity distinct from ricin (Brigotti 2012). This accounts for the differences of their target cells.

Crystallographic studies on Shiga toxins clearly described the structures of the subunits of Shiga toxins and their connections. Each of the B chains forms an α-helix, and those from the five B chains arrange themselves to form a ring, surrounding an α-helix at the C-terminus of the A chain (at the A2 fragment), forming a non-covalent bridge between the A and B chains. The B chains form anti-parallel β-sheets outside their α-helixes to encircle the region of A-B chain interaction (Fraser et al. 1994). The orientation of the A chain with respect to the B chains in Stx and Stx-1 is different from that in Stx-2. Besides, the C-terminus of the A chain of Stx-2 is more extended and structured when compared with those of Stx and Stx-1 (Fraser et al. 2004). It extends through the ring formed by the B chains, which is thought to interfere with receptor binding, but no evidence is available to show interactions of the Stx-2 A chain with Gb3 receptors.

The active site for the N-glycosidase activity of Shiga toxins lies at the A1 fragment. Stx-2 have their active sites accessible in the crystal structure (Fraser et al. 1994), and Stx2 can cleave the adenine substrate in the presence of adenosine in its crystallized form. On the other hand, those of Stx and Stx-1 are blocked by their A2 fragments, keeping them inactive in their native form. To free the active site and achieve the active form of Shiga toxin, the A2 fragment can be cleaved from the A1 fragment by the protease furin. The enzyme recognizes a specific sequence motif (Arg–X–X–Arg) in the protein loop formed by the disulfide bond between the residues 242 and 261 of the A chain. Furin cleaves the A chain of Stx and Stx-1 at the site between Arg251 and Met252 and cleaves Stx-2 at the site between Arg250 and Ala251 (Fagerquist and Sultan 2010). Cells without furin are also detected with cleavage of the A chain, though at a lower efficiency. It shows that furin is not absolutely necessary for the toxicity of Shiga toxins, and other routes are available for activation of the toxin. The A1 and A2 fragments are still connected through the disulfide bond after furin cleavage, until the bond gets reduced in the endoplasmic reticulum (ER) lumen (Garred et al. 1997).

Each of the B subunits owns three carbohydrate binding sites. Each Shiga toxin molecule holds up to 15 sites for interaction with carbohydrate moieties, enabling a high affinity of Shiga toxins. Shiga toxins were firstly found to specifically bind to carbohydrates with terminal Galα1–4Gal, and later the trisaccharide moiety (Galα1–4Galβ1–4Glc ceramide) of globotriaosylceramide (Gb3) receptor was identified as the major target of the toxin (Ling et al. 1998). One exception of the specificity is a variant of Stx-2, the Stx-2e, which is specific in globotetraosylceramide (Gb4) binding instead. Other Shiga toxins also can interact with Gb4, albeit with a low affinity (Müthing et al. 2012).

Production of Shiga toxins

The original Stx produced by S. dysenteriae is believed to be encoded by its own gene. Stx-1 and Stx-2 are produced in bacteria infected with Stx-phages. E. coli (STEC) is the major source of Shiga toxins correlated with Stx-phages, but Stx-phages also can infect other types of bacteria, such as Citrobacter freundii, Enterobacter cloacae, and Shigella sonnei, for the production of Shiga toxins. Stx-phages are defined by the presence of the Shiga toxin operon and also a group of mobile genetic elements that can be incorporated into the specific sites of the host genome (O’Brien et al. 1984). Bacterial cells can be infected by multiple Stx-phages, thus producing variants of Shiga toxins. For example, the E. coli O157:H7 strain 933 that caused the first outbreak of HC in the USA was able to produce both Stx-1 and Stx-2 (Strockbine et al. 1986). The cells may also be infected with other types of phages, allowing gene recombination inside the cells, enhancing the genetic diversity of the phages.

When Stx-phages infect the bacterial cells, they choose between two developmental fates. They may grow lytically, cause damage, and kill the host cell. They may also grow in a lysogenic state, where the genes of the phage are inserted to the bacterial chromosomes, and replicate with them (Gamage et al. 2004). Although Stx-phages have a generally highly mosaic construction of genes, they always have their Stx gene at the late region of the genome, at a region downstream of anti-terminator gene Q and upstream of the lysis cassette. The expression of the Stx gene is controlled by promoter PR. These genes are not expressed in the lysogenic state, until a signal is generated to induce the lytic cycle. In lysogenic states, repressor DNA binding blocks the early promoters PL and PR. During initiation of the lytic cycle, the repressor is inactivated, and a transcription process on the promoters PL and PR is induced. PL leads to N gene transcription followed by translation, and N protein allows the transcription process of RNA polymerase on PR to extend through the terminators to express the Q gene. The Q protein binds DNA at a site overlapping PR′, blocking the terminator and extending the expression through the operon including the Stx genes. Expression of Shiga toxins through the lytic cycle removes the need for the development of a secretion system of the toxin, as the host cells will eventually lyse and release the Shiga toxins (Fogg et al. 2012). However, a secretion system for releasing Stx-2 from viable E. coli cells has been detected.

The decision for the phage genome to shift from the lysogenic state to the lytic cycle depends on the phage-coded cI repressor protein (Serra-Moreno et al. 2008), and inactivation of this repressor protein is usually initiated by the SOS response, i.e., a ubiquitous DNA damage response of the host cell. The response can be triggered when the host cell is susceptible to antibiotics or reactive oxygen species released from leukocytes.

In patients suffering from STEC infections, when the STEC is under stress from the attack of leukocytes, or from administration of antibiotics (McGannon et al. 2010), the lytic cycle can be induced, leading to expression of Stx-1 and/or Stx-2, and cell lysis causes the toxin to be released to the gut, leading to HC and HUS. The Stx-phages formed in the STEC cells are also released and may infect the commensal gut bacteria, leading to further toxin production, increasing the severity of symptoms.

Binding of Shiga toxin to target cells

To induce toxicity in target cells, Shiga toxins have to bind to specific receptors on the surface of target cells, enter the cell through endocytosis, be transported through the Golgi network and ER, and released to the cytosol, in order to approach the ribosomes to exert N-glycosidase activity, causing ribosome inactivation.

The first step is binding to target cells. The binding properties of Shiga toxins are quite specific. The B chains of Shiga toxins are highly specific to the trisaccharide moiety (Galα1–4Galβ1–4Glc ceramide) of the Gb3 receptor (Ling et al. 1998). They also interact with Gb4 receptors but to a lesser extent. One exception, the Stx-2e, shows especially high affinity toward Gb4 receptors instead (Müthing et al. 2012). The importance of the toxicity of Shiga toxin is evidenced by generation of Shiga toxin-responsive cells from non-Gb3-expressing cell lines by addition of exogenous Gb3 onto the cell surface using liposomes (Waddell et al. 1990). Furthermore, deletion of the Gb3 synthase gene in mice, which is responsible for production of Gb3 from lactosylceramide, resulted in complete resistance of the mice to the toxin. This indicated that the Gb3 receptor is the major cellular target for Shiga toxin toxicity in most mammals.

Gb3 is expressed in a few types of cells in humans. They are found on epithelial, endothelial, mesangial, and glomerular cells of the kidney, microvascular endothelial cells lining the brain and intestine, as well as some subsets of B lymphocytes (Lingwood 1999). The distribution of Gb3 receptors explains the involvement of the gut, kidney, and brain in HC and HUS.

Gb3 is the main target for Shiga toxins, but Stx-1 was also found to interact with human neutrophils, which do not express Gb3 or Gb4. It was firstly thought to be another possible route for the initiation of Shiga toxin activity. Interaction with the cells took place at the A subunit of Stx-1 instead of the carbohydrate binding domains of the B chains, and internalization of the toxin was not induced. It was also found that the Toll-like receptor 4 (TLR4) is the cellular receptor responsible for the interaction (Brigotti et al. 2013). This interaction is probably initiated by the neutrophils, which is the route for the human immune system to recognize the A chain of Shiga toxins as the pathogenic antigen. This helps to trigger the downstream signaling pathways in the neutrophils to stimulate further immune responses.

The presence of Gb3 receptors generally determines the responsiveness of the cells to Shiga toxins. However, the strength of toxicity exerted on the cells is not only determined by interactions of Gb3 with the toxin. Stx-1 was shown with higher binding affinity to Gb3 than Stx-2 (Itoh et al. 2001). However, the in vivo toxicity of Stx-2 was found to be much higher than that of Stx-1 in animal models (Tesh et al. 1993). The LD50 of Stx-2 was 400-fold lower than that of Stx-1 in mice. Application of four doses (25 ng/kg) of Stx-2 caused the development of HUS in baboons, but the same dose of Stx-1 did not cause damage (Siegler et al. 2001). Epidemiological studies also revealed that the majority of casualties from HUS were associated with infections of Stx-2-producing E. coli strains. The Gb3 receptors only account for part of the toxicity of Shiga toxins. The effectiveness of transport mechanisms of the toxin, including entrance from the cell surface into intracellular areas, movement along the Golgi network and ER, and release of the toxin into the cytosol, play an important role in the cytotoxicity of Shiga toxin.

Entrance of Shiga toxin into target cells

After binding with Gb3 receptors, Shiga toxins are transported into the cells through endocytosis. Clathrin-dependent endocytosis plays a role in the intake of Shiga toxins. Shiga toxins have the ability to induce phosphorylation and activation of the kinase Syk, followed by phosphorylation of the clathrin heavy chain, inducing formation of a complex between clathrin and Syk, forming clathrin-coated pits that carry the toxin (Lauvrak et al. 2006). The B chain of the toxin alone is enough to trigger Gb3 receptor binding, as well as endocytosis of the B chain. However, increasing the amount of B chains alone could not give the same effect to enhance the rate of clathrin-coated pit formation as that from increasing the amount of the total Shiga toxin. Although the exact mechanism is unknown, the A chain of Shiga toxins should contribute to interactions with other cellular membrane proteins to enhance the toxin uptake.

Clathrin-dependent endocytosis machinery involves a number of components, such as clathrin heavy chain, dynamin, epsin, or eps15 (Utskarpen et al. 2010). Inhibition of some of these components to interfere with the machinery can reduce the intake of Shiga toxins. Blockage of clathrin heavy chain by small interfering RNA (siRNA) lowered ∼40 % of Shiga toxin endocytosis, while clathrin-dependent endocytosis was reduced by ∼80 %, as indicated by the clathrin pathway marker transferrin. Similarly, mutations of epsin and eps15 lowered 40–50 % of Shiga toxin endocytosis, while transferrin was reduced by 70 % (Popoff et al. 2007). Blocking clathrin-dependent endocytosis cannot completely block endocytosis of Shiga toxin. It takes place in a clathrin-independent manner. Clathrin-independent endocytosis can compensate for the loss of clathrin function after longer treatments. After treatment with Stx B chain for 10 min, endocytosis was ∼30 % lower in cells with clathrin blocked by siRNA compared with the control. After treatment with Stx B chain for 40 min, the cells with siRNA could attain an equal level of endocytosis as the control.

Clathrin-independent pathways are mainly classified into dynamin-dependent or dynamin-independent, according to the involvement of dynamin in membrane scission. Clathrin-dependent pathways are also dynamin-dependent. Depending on the molecular components they require, the endocytosis pathways are sub-divided into different categories, such as caveolae-mediated and RhoA-regulated in dynamin-dependent pathways (Römer et al. 2007) and Cdc42-regulated and Arf6-regulated in dynamin-independent pathways (Malyukova et al. 2009). Similar to those clathrin-related components, blockage of other molecular targets like RhoA, Cdc42 partially blocks Stx endocytosis. With multiple endocytosis pathways and the properties of compensation by one other, it is very difficult to completely block the intake of Stx.

Retrograde transport—from endosomes to Golgi

The toxin-carrying endosomes require retrograde transportation to the Golgi network and ER before the toxin can be released to the cytosol. Transport machineries are required to sort the endosomes to the Golgi network; otherwise, they may mature into late endosomes and will be degraded by lysozymes.

To achieve effective sorting, the endosomes can form a coat-like retromer complex. The retromer comprises a number of components: Vps26, Vps29, Vps35, and sorting nexins (SNXs) (Cullen and Korswagen 2011). The Vps form the cargo recognition subcomplex, and the SNXs contain a BAR domain for sensing and inducing curvature. Switching between the sensing and inducing modes is involved in the formation of endosomal tubules. The complex is coated with clathrin, together with the clathrin adaptor epsinR. RME-8 is a DnaJ protein that interacts with SNX1 of the retromer, and it is linked with Hsc70, a clathrin-uncoating ATPase, that is involved in uncoating of clathrin from endosome tubules. Many other factors are required for fusion of the endosome with the Golgi network, including the tethering of golgin97, golgin245, GCC185, and SNARE complexes (syntaxin6, syntaxin16, Vti1a, Vamp3/4 and syntaxin5, Ykt6, GS15, GS28) (Sandvig et al. 2013). Some are involved in microtubule regulation, while others are for docking onto the Golgi network, etc. The process of transportation is highly complicated and regulated by many factors (Mukhopadhyay and Linstedt 2013). The main concern about Stx on retrograde transport is how Stx takes part in mediating itself to run through this transport program, instead of being degraded in lysosomes. Two major pathways have been identified with Stx involvement. They are protein-based sorting and membrane lipid sorting, respectively.

In the protein-based sorting pathway, the B chain of Stx was found to interact with GPP130, a dimeric transmembrane protein that cycles between sorting endosomes and Golgi network (Mukhopadhyay et al. 2013). GPP130 interacts with COG3P, and knockdown of COG3P causes accumulation of GPP130 in COG complex-dependent vesicles, leading to GPP130 degradation. Also, treating cells with a small amount of manganese (Mn2+) ions diverts the GPP130 to lysosomes leading to degradation. In both cases, interactions of Stx with GPP130 did not take place. The endosomes carrying Stx were missorted and degraded in lysosomes. Furthermore, mutations on GPP130 at the binding site of Stx did not cause GPP130 degradation, but failure of interaction of Stx with GPP130 led to Stx degradation. The presence of GPP130 in the endosomes was not enough for the sorting, but Stx B chain-GPP130 interaction was required for successful sorting. These observations indicated the importance of GPP130 as the crucial trafficking receptor of Stx to perform retrograde transport toward the Golgi.

After internalization, Stx B chains interact with GPP130 receptors in the endosomes. Whether the Stx is detached from Gb3 for the interaction or the B chains interact with both receptors at the same time is still unknown. Yet, the Gb3 receptors are evidenced to take part in retrograde sorting. This may be a reason why Stx chooses the Gb3 receptor as its specific receptor. Gb3 receptors are involved in the control of organization of its surrounding lipids. In the endosomes, the Gb3 receptors are recovered with cholesterol-rich detergent-resistant membrane (DRM) preparations. Depletion of cholesterol in the cells blocks the retrograde transport, indicating the importance of the lipid structure for the sorting. Proper Gb3 receptor isoforms are also required for successful transport (Sandvig et al. 1996). It was shown by sensitization of Gb3-expressing human epidermoid carcinoma A431 cells toward Stx by butyric acid. The A431 cells express Gb3 receptors, and Stx endocytosis is available, but retrograde transport is not allowed, so that the cells are not sensitive to Stx toxicity. When the cells were treated with butyric acid, the composition of Gb3 fatty acid chains was altered in such a way that the Gb3 receptors express longer lipid chains (more C24:1 chains from C18:1 chains), and Stx was found to be successfully transported into the Golgi and exhibited cytotoxicity to the cells. Meanwhile, Stx binding on the Gb3 receptors at the cell surface was not affected. Fumonisin B1, PDMP, and PPMP were able to inhibit glycosphingolipid synthesis. Addition of these compounds with butyric acid on A431 cells blocked the effect of butyric acid sensitization on Stx, as the building of longer Gb3 lipid chains was abolished.

Stx also contributes to retrograde transport by association with the lipid raft (Tam et al. 2008). In Gb3-expressing monocyte-derived cells, no association of Stx B chains with the DRM was detected, and those cells were not sensitive to Stx toxicity. Similarly, in Gb3-expressing bovine intestine epithelial cells, endocytosis of Stx was detected, but the toxin could not associate with the lipid microdomains, and the cells have the toxin degraded in the lysosomes and are insensitive to it. Meanwhile, in the Stx-sensitive HeLa cells, the toxin was detected to associate with the DRM throughout the retrograde pathway, and destabilization of the DRM with cholesterol extraction interfered with Stx transportation.

Retrograde transport—from Golgi to ER

Retrograde transport of most components from Golgi to ER depends on the COPI vesicle coat complex (Mukhopadhyay and Linstedt 2013). In this model, the transmembrane KDEL receptor recognizes the KDEL motif present in the contents in the Golgi lumen being transported. Receptor binding leads to activation of tyrosine kinase Src and in turn interacts with cytosolic COPI, forming a COPI coating vesicle that buds off from the Golgi network and moves toward the ER. However, Stx does not express a KDEL domain that is recognizable by the receptor (Mukhopadhyay and Linstedt 2013). The transport of Stx through Golgi is COPI-independent. Instead, the transport of Stx showed relationships with the small GTPase protein Rab6 (Starr et al. 2010). Rab6 has two isoforms, Rab6a and Rab6a′, and only Rab6a′ was found to mediate the transport of Stx. Rab6 is involved in interactions with retrograde Golgi tether complexes, conserved oligomeric Golgi (COG), and Zeste White 10 (ZW10)/RINT-1. Rab6 and Rab33b are functionally overlapping regulators for homeostasis of the organization of Golgi network (Starr et al. 2010). Depleting Rab33b caused inhibition of GTP-Rab6-induced redistribution of Golgi enzymes and transportation of Shiga toxin to the ER, but not vice versa, indicating that Rab33b acted sequentially after Rab6 in the Stx transport process.

The transportation of Stx-carrying vesicles from Golgi to ER depends on an actin-based network of microtubules. The movement is mediated by the motor protein myosin II, together with small GTPase Cdc42 and N-WASP (Luna et al. 2002). Transportation of Stx to ER can be inhibited when the actin network of microtubules is disrupted.

Release of Shiga toxin to the cytosol

To exert the toxicity, the Stx A chain has to be translocated to the cytosol to approach cellular ribosomes. As described before, the A chain of Stx consists of the A1 fragment that possesses N-glycosidase activity and the A2 fragment that is linked to the B subunits. The endoprotease furin cleaves the A chain into the two fragments in endosomes and the Golgi network, and the A1 and A2 fragments are still connected by the disulfide bond. When the toxin reaches the ER, the disulfide bond will be reduced, releasing the Stx A1 chain in the ER lumen (Garred et al. 1997).

The Stx A1 chain is translocated to the cytosol through the ER-associated protein degradation pathway (ERAD). After dissociation of the A1 chain from the A2 fragment, the hydrophobic C-terminal domain of the A1 chain is exposed and is recognized as a misfolded peptide domain by the ER export mechanism.

The A1 chain forms a complex with ER chaperone components including HEDJ, BiP, and GRP94 (LaPointe et al. 2005; Yu and Haslam 2005). Then, the complex interacts with Sec61 translocon. The translocon acts as a channel through the ER membrane. The protein being translocated is unfolded and passes through the central pore of the translocon to be released to the cytosol. Knockdown of Sec61B could protect the cells from Stx’s toxicity, indicating that the transportation mechanism through Sec61 translocon should take place for the Stx A1 chain to approach the cytosol. In any case, only a small proportion of the Stx A1 chain is translocated into the cytosol. In the study on furin-containing Vero cells, under continuous exposure of the toxin, only 4 % of the A1 chain is translocated into the cytosol of the Vero cells, but the small proportion of the A1 chain is enough to cause damage to the cells.

Interaction with ribosomes

When the Stx A1 chain gets into the cytosol, it exhibits its ribosome-inactivating activity with a mechanism which is the same as many other RIPs like ricin. That involves exerting its N-glycosidase activity for irreversible depurination of a universally conserved adenine residue in the α-sarcin/ricin loop (SRL) of the 28S ribosomal RNA (rRNA), leading to blockage of binding with elongation factor EF-2 and inhibition of translation. Similar to ricin, Stx has a much higher efficacy of depurination toward ribosomes over naked 28S rRNA. It indicates that the structure of the ribosome also has an important role in Stx toxicity.

The P-protein stalk showed interactions with the Stx A1 chain (McCluskey et al. 2008). The P0, P1, and P2 proteins of the P-protein stalk are involved in Stx A1 chain interactions. Deletion of 17 amino acids at the C-terminus of P1 and P2 proteins blocks interactions of the Stx A1 chain with the ribosomal protein stalk, implying that the C-terminal domains of P1 and P2 proteins are required for Stx interactions. Deletion of the C-terminus of P0 does not block Stx binding with the protein stalk. The P0 protein plays its role to connect with the P1 and P2 proteins instead of direct connection with Stx, and deletion of P1 and P2 binding sites on P0 can affect Stx’s depurinating activity. Interestingly, both Stx and Stx-1 A1 chain require an intact P0 protein to deliver the toxin to the target site of 28 rRNA, but the Stx-2 A1 chain does not. Without the attachment to P0 protein, the cytosolic P1 and P2 proteins are able to interact with the Stx-2 A1 chain and deliver the chain to the target rRNA, so the Stx-2’s depurinating activity is not affected by a defective P0 protein (Chiou et al. 2011).

The hydrophobic and cationic surface of the Stx A1 chain is responsible for interactions with the C-terminal domain of P2. A group of arginine residues (Arg172, 176, 179, 188) are involved, as mutations of these amino acids can affect P2 binding. Although Stx-1 and Stx-2 A1 chains have a certain degree of heterogeneity, those arginine residues are conserved, implying that both types of Shiga toxins share similar mechanisms to interact with the P proteins. For interactions with the ribosome, Stx A1 chains are concentrated on the ribosomal surface by non-specific electrostatic interactions. Then, the toxin chain interacts with the C-terminal domain of P proteins, in which the binding sites are at the side opposite to the active site for the N-glycosidase activity. The P proteins with a flexible C-terminal domain show a conformational change that delivers the Stx A1 chain toward the 28S rRNA in the ribosome (Li et al. 2010). The active site of the Stx A1 chain comes into contact with the SRL of 28S rRNA, and depurination proceeds.

Toxicity to cells

The activities of Stx bring several consequences to the cells. In most cases, the fate is the triggering of apoptosis. One of the routes is activation of the ribotoxic stress response. Once 28S rRNA of the ribosome is cleaved by Stx, the ribosome is rendered inactive, and conformational changes in the intoxicated toxin take place. The stress elicited by the inactivated ribosomes leads to activation of mitogen-activated protein kinase (MAPK) signaling pathways (Ikeda et al. 2000).

Stx-induced ribotoxic stress was found to induce the p38 MAPK pathway, c-Jun N-terminal kinase (JNK) pathway, as well as extracellular signal-regulated kinase (ERK) pathway (Cameron et al. 2003; Cherla et al. 2006; Foster and Tesh 2002). Initiation of these pathways triggers the kinase cascades that leads to control in cell proliferation and initiation of apoptosis. Apoptosis through Stx-1-induced ribotoxic stress was first shown in human epithelial HCT-8 cells, in which caspase 3 activation and DNA fragmentation were detected (Smith et al. 2003). Inhibition of the p38 MAPK pathway could reduce Stx-1-induced apoptosis on HCT-8 cells. Besides, inhibition of zipper sterile-α-motif kinase (ZAK), which acts as a transducer of the ribotoxic stress signal to activate the MAPK pathway, also yielded some protection on HCT-8 from Stx-2 exposure. Although ZAK inhibition limited caspase 3 activation, DNA fragmentation was not blocked. It indicated Shiga toxins can also induce apoptosis through some routes other than MAPK pathways.

Other than apoptosis, the ribotoxic stress may cause regulation in cytokine production in the cells (Cherla et al. 2006; Foster and Tesh 2002), and the types of cytokines and the extent of expression affected vary in different cells. Shiga toxins could induce secretion of IL-1β, IL-6, and IL-8 in human proximal tubular cells and glomerular epithelial cells. In human intestinal epithelial T84 cells and Caco2 cells, treatment with Shiga toxins led to an elevation in interleukin-8 (IL-8) secretion, while the non-toxic Stx-1 mutant could not produce the same effect. It was suggested that the toxin brings about stabilization of IL-8 mRNA, which causes prolonged translation that yields IL-8. Some immune cells are also molecular targets of Stx. Ribotoxic stress was induced in peripheral blood monocyte THP-1 cells after exposure to Stx-1. The expression level of a number of cytokines, including IL-1α, IL-1β, IL-6, IL-8, and tumor necrosis factor α (TNF-α), showed significant upregulation (Cherla et al. 2006; Foster and Tesh 2002). The expression of TNF-α showed dependence on the JNK and p38 signaling pathways, in which blockage of these pathways in turn limited TNF-α induction. Instead, the ERK signaling pathway seemed not to affect TNF-α expression. Apart from cytokine synthesis, prolonged exposure to Stx-1 also ensued in apoptosis of THP-1 cells. However, cytokine induction is not necessarily accompanied by apoptosis. Stx-1 caused IL-4 induction in bovine peripheral and intraepithelial lymphocytes (IEL), without affecting IL-2, IL-8, and IL-10, and Stx-1 did not induce apoptosis in these cells.

Other than elevating ribotoxic stress through damaging the ribosomes, Shiga toxins are also able to induce ER stress and trigger unfolded protein response (UPR). UPR is a quality control mechanism in the ER to ensure the proteins are properly folded before exporting from the ER. When unfolded or misfolded proteins accumulate in the ER, UPR will be triggered to create signals that lead to temporary attenuation of protein synthesis, providing time for correction. Under prolonged or severe ER stress, the cell is incapable of restoring protein synthesis to normal, and cell death signal will be generated through activation of the sensor proteins including protein kinase-like ER kinase (PERK), inositol-requiring ER to nucleus signal kinase-1 (IRE-1), and activating transcription factor-6 (ATF6) (Tabas and Ron 2011). PERK will lead to activation of eIF-2, causing C/EBP-homologous protein (CHOP) expression to be upregulated. This transcription factor upregulates expression of pro-apoptotic factors such as death receptor 5 (DR5) and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and downregulates the survival factor Bcl-2, leading to activation of the caspase cascade. IRE-1 will be activated and upregulated, causing TNF receptor-associated factor 2 (TRAF2) activation, which in turn triggers the JNK signaling pathway (Lee et al. 2008). ATF6 will also be activated, which acts as a transcription factor to regulate gene expression. Besides, ER stress causes Ca2+ efflux from the ER lumen, leading to activation of pro-caspase 12, in turn initiating the caspase cascade.

Inside the ER lumen, the A1 fragment of Shiga toxin is cleaved and released from the A2 and B subunits. The A1 fragments will be temporarily unfolded in order to be translocated to the cytosol through the Sec61 translocons. These fragments will be regarded as misfolded proteins and recognized by the UPR sensors (Tesh 2012). Also, as only a small proportion of A1 fragments is translocated, the remainder accumulates in the ER lumen, in turn building up the ER stress. In THP-1 cells, Stx-1 treatment could induce not only ribotoxic stress but also ER stress. The Stx-1-treated cells were detected with Ca2+ efflux from the ER lumen, together with caspase activation. Also, the cells showed upregulation of CHOP, as well as upregulation of pro-apoptotic factors and downregulation of survival factors, which eventually led to apoptosis. Interestingly, treatment with Stx-1 B chain alone could activate IRE-1 signaling, indicating the Stx-1 B subunits may also contribute to UPR, but the B chain alone was unable to trigger apoptosis.

Other than Stx-induced apoptosis, formation of necrotic cells was also detectable upon Stx treatments. In various studies of STEC infection in human, mouse, pig, and other animal models, Stx-induced necrosis was detected in brain endothelial cells (Kausche et al. 1992), kidney endothelial cells (Wadolkowski et al. 1990), as well as some monocytes (Methiyapun et al. 1984). The necrotic cells showed noticeable formation of cytoplasmic debris and nuclear fragments. Some of them showed loss of adherence and may express acute swelling to death (Matise et al. 2000). Both apoptosis and necrosis could result from Stx’s toxicity. Some cell types preferentially underwent necrosis over apoptosis while some did not (Matise et al. 2000).

Toxicity in humans

Shiga toxin intoxification in humans is mainly caused by ingestion of STEC-contaminated food or water, causing bacterial infections in the gut. STECs do not express Shiga toxins at their lysogenic state until they enter the lytic cycle (Gamage et al. 2004). Without storing the toxins inside the bacterial cells, the level of toxins is very low at the early stages of STEC infections, and Shiga toxins are still not involved here. At this stage, the bacteria play their role in colonization and adhesion to the intestinal cells. STECs interact with the intestinal mucosa and produce the adhesin intimin that acts on the intestinal enterocytes, allowing adhesion of the bacteria onto the epithelial wall (Zoja et al. 2010). When the bacteria come in contact with the epithelial cells, they produce and release effectors that cause rearrangement of epithelial cellular actin, forming pedestals near the site of contact, thus allowing the bacterial products to cross the epithelial cell barrier and enter the circulatory system. The infectious bacterial lesions cause derangements in the intestinal lining, disrupting the structure of the villi, affecting absorption in the gut, and eventually leading to watery diarrhea (Zoja et al. 2010).

Within a few days, Shiga toxins are produced and released from the STEC. Although the enterocytes do not express Gb3 receptors and are not intoxicated by the toxins, the infectious activities from the bacteria have disrupted the intestinal wall; hence, the toxins can be translocated through the loosened junctions of the epithelia, reaching the lamina propria, and be absorbed in the circulatory system (Schüller 2011). The Shiga toxins reach the Gb3-expressing intestinal endothelial cells, causing cell damage and cell death. Damage to the intestinal endothelium causes mucosal and submucosal edema, hemorrhage, and bloody diarrhea. The onset of HC with these symptoms can be a sign for the development of HUS after a few days.

The STEC-induced HUS is characterized by the presence of thrombotic microvascular lesions in the target organs, in which Gb3-expressing cells allow intoxication by Shiga toxins. The toxins attack the Gb3-expressing kidney epithelial, endothelial, mesangial, and glomerular cells; intestinal microvascular endothelial cells; and brain microvascular endothelial cells (Lingwood 1999), causing severe damage to the kidney, gut, and the brain, and may eventually result in death.

The intestinal endothelium is located near the source of infection; thus, it does not require a long journey through the blood stream for the Shiga toxins to exert their toxicity. To develop HUS in the kidney and the brain, Shiga toxins can be carried to the target through some carriers in the blood stream. The toxins interact with red blood cells, platelets, and some monocytes presenting with Gb3 on the membrane. The TLR4 on the neutrophils also interact with the A chains of Shiga toxins (Brigotti et al. 2013). When the neutrophil-bound toxin reaches the Gb3-expressing endothelial cells in the kidneys and the brain, the high affinity between the B chain and Gb3 receptors allows the toxin to dock onto the endothelial cells.

The mediation of cytokine production also contributes to the toxicity of Shiga toxins. At high toxin dosages to the endothelial cells, cell death is triggered, while at lower, non-lethal concentrations, the ribotoxic stress induced by the toxin causes regulations on IL-1α, IL-1β, IL-6, IL-8, and TNF-α production, inducing pro-inflammatory responses and favoring recruitment of leukocytes. The Gb3-expressing monocytes being attracted are also susceptible to Shiga toxin binding. The monocytes also undergo regulations in the production of cytokines such as IL-1 and TNF-α, and the cytokines further enhance Gb3 receptor expression on endothelial cells, making them more susceptible to the toxins. The continuous Shiga toxin-induced cytokine production brings about excess pro-inflammatory effects, causing severe harm at the endothelial lesions.

Shiga toxins are directly involved in platelet activation and aggregation. This in turn reduces thromboresistance in endothelial cells, leading to microvascular thrombosis (Karpman et al. 2001).

Shiga toxins also directly interfere with the complement system of the patient (Keir 2015). The system is a part of the innate immune defense involved in the removal of foreign or apoptotic cells. It comprises the pathogen-induced classical pathway and lectin pathway, and the alternative pathway (AP) with a tick-over mechanism. The system is mediated by activation of complement C3 convertase that cleaves the complement component C3. C3 is cleaved into C3a and C3b. C3b is readily decomposed in blood but can also bind onto the cell surface as a signal, which in turn triggers an amplification loop that favors further C3 decomposition, as well as leads to formation of membrane attack complexes. Some patients suffering from STEC-induced HUS were found to have C3 reduction (Keir 2015). This is caused by the ability of Shiga toxin to induce upregulation of the membrane adhesion molecule P-selectin, which binds C3b with a high affinity. C3b binding around the endothelial cells favors activation of AP. C3a is an anaphylatoxin that induces degranulation of endothelial cells, triggering a local inflammatory response. Shiga toxins also bind with Factor H (Keir 2015), which protects the host cells from complement system activities. Shiga toxin binding prevents Factor H from binding onto the surface of endothelial cells, and the cells become susceptible to damages caused by AP activation.

Shiga toxins contribute to HUS development through multiple effects on the endothelial cells and other cells like monocytes. The major mechanisms entail pro-inflammatory effects, pro-apoptotic effects, and pro-thrombotic effects. At lower Shiga toxin dosages, the Gb3-expressing cells in the kidneys and the brain are stimulated with regulation of cytokines that promote local inflammatory responses; the immune cells responsive to the toxin further amplify the response. The oversensitized inflammatory response causes harm to the areas nearby. At higher toxin dosages, the ribotoxic stress and ER stress exerted on the cells drive the cells to death. The pro-thrombotic effects not only cause thrombocytopenia but also cause microvascular thrombosis that disturbs circulation to the affected areas. As many cell types around the kidneys are Gb3-expressing, the kidneys are highly susceptible to the toxins, and kidney failure is one of the major consequences from Shiga toxin-induced HUS (Majowicz et al. 2014). Dialysis is often required for patients suffering from acute kidney failure. Some of the patients may show permanent kidney damage accompanied by proteinuria, hypertension, and/or reduced glomerular filtration rate, even after they have recovered from HUS. As the brain endothelial cells are also Gb3-presenting, the central nervous system (CNS) is also threatened by Shiga toxins (Obata 2010). The proportion of patients developing HUS in the CNS is less common than that of patients developing HUS in the kidneys, but CNS damage to the patients can have serious complications. It involves a wide range of symptoms, such as apnea, coma, seizures, cortical blindness, and hemiparesis, and may even cause sudden death. The damage may also cause long-term effects even after recovery, in which there are studies suggesting some children who have recovered from HUS show abnormal test findings in some parts of the neurocognitive profile, yet long-term follow-up investigations on those patients are still required to provide more evidence on the neurological interferences.

Strategies to counter Shiga toxin toxicity in humans

Shiga toxins can exert devastating damage to the human body. When infected with STEC, application of antibiotics is usually not recommended. They kill the bacteria, but the lytic cycle of the bacteria is also triggered for stimulation of Shiga toxin production, adding to the severity of HUS complications. However, there are reports demonstrating that some of the antibiotics may be safe to use in STEC infections. Early treatments of fosfomycin on the first 2 days of STEC infections were found to prevent the development of HUS (Ikeda et al. 1999). Treating the Stx-1 and Stx-2-producing E. coli O157:H7 strain with fosfomycin, panipenem, ceftazidime, and aztreonam did not induce expression of stx genes and had little effect on the induction of Stx production, while norfloxacin greatly upregulated the stx genes (Ichinohe et al. 2009). Fosfomycin acts to inhibit UDP-N-acetylglucosamine pyruvate transferase, leading to inhibition of cell wall synthesis and induction of cell lysis. On the other hand, norfloxacin functions by inhibiting DNA gyrase, which inhibits separation of DNA and cell division. Differences in their mechanisms probably distinguish between the different responses from STECs. Ciprofloxacin is a quinolone antibiotic found to induce oxidative stress on STECs and result in abrupt production of Stx-2. Exogenous anti-oxidants like tiron counteracted the oxidative stress and toxin production was consequently greatly reduced, indicating that reactive oxygen species may take a role in induction of Stx production (Angel Villegas et al. 2015). Although treatment with antibiotics on STEC can be controversial, choosing the appropriate antibiotic should be beneficial to the patients. Clarifying the mechanisms of Stx-induction and mode of actions of antibiotics should be helpful to make the right decision.

Specific antibodies for Stx-1 and Stx-2 also act against the toxicity. Humans are able to produce antibodies against Shiga toxins upon encounter. These antibodies have been isolated and identified from the STEC-infected patients (Griffin et al. 1983; Strockbine et al. 1985). However, production of antibodies from the humoral immunity response usually takes a couple of days, which is not sufficient for the human body to counteract the Shiga toxins immediately. Instead, antibodies can be raised safely from non-toxic synthetic peptides of Shiga toxins, separated A and B subunits, or heat-inactivated toxins that are unable to exert cytotoxicity. Various groups have generated monoclonal and polyclonal antibodies against subunits A and B of Stx-1 and Stx-2, from animal models, as well as human/mouse chimeric antibodies and fully humanized antibodies (Melton-Celsa et al. 2012). Many of them were able to neutralize the native toxin, suppressing the cytotoxicity of the Shiga toxin-targetable cell lines (Harari et al. 1988). By application of a Stx-2A-specific monoclonal antibody, the antibody-bound Stx-2 was detected to be internalized into Vero cells, but the toxin was blocked from retrograde transportation and accumulated in endosomes. The toxin did not reach the cytoplasm and the cells were protected from the toxicity of Stx-2 (Smith et al. 2009). The antibodies also limited the toxic effects on STEC-infected animal models (Jeong et al. 2010). In animal models infected with STEC that produces both Stx-1 and Stx-2, monoclonal antibodies for both toxins have to be administered to protect the animals from Shiga toxin toxicity (Melton-Celsa et al. 2015). Some of those monoclonal human/mouse chimeric and humanized antibodies were tested to be safe in humans in several clinical trials (Bitzan et al. 2009; López et al. 2010), allowing them to be a viable choice for treatment of patients with STEC infections.

Another approach to minimize the risk of Shiga toxins on humans is to control the growth of STEC sources. Cattle is the main reservoir of E. coli O157. Consumption of contaminated uncooked food, or contaminated water with livestock wastes is a big menace in human infections. Controlling E. coli in the cattle is one route to reduce the risk for human exposure to the bacteria. Colonization of bacteria takes place through the type III secreted protein (TTSP) system with involvement of a number of components including intimin, translocated intimin receptor, EspA, EspB, and EspD, making them potential targets in vaccine development (Moxley 2004). Siderophore receptor and porin (SRP) vaccine is another approach to act against E. coli membrane, interfering with their iron transport and colonization (Thomson et al. 2009). A number of vaccines have been developed to control E. coli O157 in cattle, and two of them developed in Canada and the USA have been approved and commercialized (Matthews et al. 2013). Vaccination of cattle leads to good control of E. coli O157. However, this incurred extra cost without actual benefit to the farmers. This explains why the farmers are reluctant to apply vaccination.

In some other food sources like plant materials, vaccination against STEC is not applicable. Other methods of sterilization can be applied to minimize the risk of STEC contamination. High-pressure processing of fresh strawberry puree was able to limit the survival of both O157 and non-O157 E. coli strains. A 350-MPa-pressure treatment for over 5 min reduced the count of STECs to a non-detectable level (Hsu et al. 2014). Ultraviolet (UV) treatment is another strategy for bacterial reduction in food. UV-C treatment at a dose of 74 mJ/cm2 significantly reduced the population of O157 E. coli present on the surface of fresh apricots. During storage, the bacterial population was also much lower on the surface of apricots after UV-C treatment (Yun et al. 2013). Cooking is not recommended for most fruits as some of their nutrients may get destroyed from heating. These non-thermal treatments preserve the nutrients while reducing the risk of STEC contaminations.

Application of Shiga toxins

Shiga toxins specifically bind Gb3-expressing cells. The B subunits act as the delivery tool for Gb3 recognition, and the A subunit exerts cytotoxic activities. Recombinant B subunits alone have limited toxicity, while their capability for Gb3 recognition is kept, allowing them to be applied as a tool for the detection of Gb3-expressing cells.

Many reports showed that there is a high occurrence of Gb3 overexpression in various types of tumor cells, such as breast cancer (LaCasse et al. 1999), ovarian cancer (Arab et al. 1997), prostate cancer (Maak et al. 2011), and testicular cancer (Ohyama et al. 1990), as well as in the tumor vasculature (Johansson et al. 2009). The presence of Gb3 overexpression at the region normally lacking in Gb3 receptors reflects the disorder, and these tissues can be imaged through binding with radioactively labeled or fluorescence-labeled Shiga toxin B subunits (Engedal et al. 2011). The gamma emitter 99mTc is one of the most used isotopes for labeling, and the signal can be detected through positron emission tomography. However, the B subunit also binds other normal Gb3-repressing cells that generate noise signals. It would be a challenge for tumor detection if we have no idea about the location of the tumor.

The Gb3-binding capability of Shiga toxin B subunits also allows targeted therapy toward Gb3-overexpressing tumors. The B subunits coupled with cytotoxic compounds, such as topoisomerase I inhibitor SN38 and benzodiazepine RO5-, were produced for drug delivery toward tumor cells (El Alaoui et al. 2007, 2008). The holotoxin has also been tested to treat tumors in mouse xenograft models bearing human malignant meningioma (Salhia et al. 2002), renal carcinoma (Ishitoya et al. 2004), etc. Intratumoral injection of Stx-1 triggered apoptosis in the tumor cells and also cells of the vasculature, and no marked side effects were observed in the mouse models. However, the usage of Stx on cancer treatment in humans is still hampered. The Gb3-targeting capability of Stx allows a strong inhibitory activity against Gb3-expressing tumor cells but also causes damage to Gb3-presenting normal cells, raising concerns that it may give rise to side effects similar to HUS. Yet, it was believed that the Stx toxicity raised through injection of Stx alone should be much lower than that raised from STEC infections, as STEC also produces other bacterial factors, such as lipopolysaccharides, that can enhance the efficiency of Stx intoxification, through regulation of cytokines, and stimulation of expression of Gb3 receptors (Palermo et al. 2009). Without other virulence factors, the activity of Stx toward Gb3-overexpressing tumor cells should be more pronounced than the normal endothelial cells. The dose of Stx being administered can also be well-controlled to limit the adverse effects.

Usage of Stx B subunit-drug conjugates is the other major strategy for Stx B-targeted therapy, but these drugs may raise the same problem as the holotoxin, as the cytotoxic components may also attack and kill the normal cells. Strategies that limit the side effects from the drug conjugates lie in the choice of the drug. Phototoxic drugs can be coupled to the Stx B subunit for delivery to the tumor cells, and local illumination can be applied around the tumor to exert the cytotoxicity (Amessou et al. 2008). Another way is to generate a drug that preferentially kills tumor cells. The Stx B subunit coupled with topoisomerase I inhibitor SN38 delivered the drug to Gb3-expressing cells, while the drug killed rapid-growing tumor cells more readily. Furthermore, the pro-drug was used for conjugation with the Stx B subunit, which would be activated after retrograde transport to the ER, thus probably being more damaging to the tumor cells that are more active in such transportation (El Alaoui et al. 2007).

Although in vitro and in vivo animal model studies supported the anti-tumor potentials of Stx and Stx B subunit-drug conjugates, a number of problems have to be solved for human use. The toxicity of Stx or drug conjugates has to be well addressed to avoid severe side effects, and the half-life of the drug in the human body needs to be clarified. Also, Stx and Stx B subunits are proteins, which can stimulate the immune response, leading to inactivation of the drugs, limiting their anti-tumor effects.

Conclusion

Shiga toxins are a group of RIPs highly toxic to humans. They possess catalytic domains which inactivate cellular ribosomes and induce cellular damages. The main cause of Shiga toxin intoxication in humans is ingestion of STEC-contaminated food or water. The toxin requires a series of steps before the toxin can exert its toxicity on the host. The toxin has to allow the STEC to infect the intestine of the host and generate paths for the translocation of the toxin into the circulatory system. The toxin also needs help from carriers in the blood to convey them to the target cells. When the toxin arrives at the target cells, it needs to interact with the specific receptors (mainly Gb3) on the cell surface and depends on complicated cellular pathways and mechanisms for endocytosis into the cell, retrograde transport to Golgi and ER, and release to the cytosol. The toxin also requires appropriate binding with the ribosomal structure to exert its ribosome-inactivating activity. Disturbance of these steps of Shiga toxin transport often protects the host from toxicity, and it provides insights for researchers to investigate the therapeutic strategies against Shiga toxins. Nevertheless, the best way to tackle Shiga toxins is to exercise due care about the food quality and ensure the food and water are clean and/or well-cooked/boiled, in order to avoid intake of live STEC that can threaten our health.