Figure 3 illustrates a working model for the “Non-Toxic” complex-induced vacuolation mechanism, as well as a comparison with other toxins (H. pylori VacA and V. cholerae haemolysin) which have been shown to induce cell vacuolation. Serotype C and D botulinum L-TC is internalised into intestinal cells. Nishikawa et al. reported that pre-incubation of HT-29 cells with the O-glycosylation inhibitor d,l-threo-1-phenyl-2-hexa-decanoylamino-3-morpholino-propanol reduces serotype C L-TC internalisation16, suggesting that L-TC may be taken up by cells with an O-linked oligosaccharide derived from glycoproteins, similar to mucin on the cell surface. It was, moreover, demonstrated that the serotype C L-TC binds to mucin derived from the bovine submaxillary gland via sialyl oligosaccharide. The internalisation of L-TC is independent of the presence of BoNT and occurs via binding of HA-33 to a hypothetical receptor on the cell surface. Therefore, in the first step of the vacuolating process, the botulinum “Non-Toxic” complex is internalised into cells via receptor-mediated endocytosis.

Figure 3: Working model for the vacuolation mechanism of the botulinum “Non-Toxic” complex in comparison to the vacuolation mechanisms of other vacuolating toxins. The left panel illustrates the working model of the botulinum “Non-Toxic” complex-induced vacuolating mechanism. The “Non-Toxic” complex enters into the cell in a sialic acid-dependent receptor-mediated manner. The process appears to involve proton pump-related vesicle formation. The “Non-Toxic” complex-containing vesicles generated from late endosomes and lysosomes then mediate the formation and swelling of vacuoles via an unknown mechanism. Central and right panels illustrate the mechanisms of action of VacA17 and V. cholerae haemolysin34. The VacA toxin is inserted into the lipid bilayer of cells and forms an oligomer, which then acts as an anion channel and activates the proton pump. The V. cholerae haemolysin also forms oligomers on the cell membrane. The oligomeric haemolysin forms larger and less specific channels than VacA. Entrance of ions into these channels induces osmotic swelling of the vacuoles in both VacA- and haemolysin-treated cells. Full size image

VacA17 and V. cholerae haemolysin18 are incorporated into lipid bilayers of vesicle membranes, forming polymeric pores and thus acting as channels for anions (central and right panels, Fig. 3). The anions then act as counter ions for the protons that enter into the cells through the proton pump. This process results in osmotic swelling of vacuoles, and the toxin-induced vacuoles thus become acidified. The crystal structure of the serotype C whole HA complex reveals the central pore formed by the trimeric HA-70 proteins10, and the SAXS data reported here indicate that the cell surface is accessible to the central pore. The depth of this pore is, however, insufficient to form a channel on the lipid bilayer of the vesicle and the “Non-Toxic” complex is thus not likely to act as an ion channel. The peripheral regions of the toxin-induced vacuoles were stained with NR; however, stain was not incorporated into the vacuoles, which suggests that the large botulinum “Non-Toxic” complex-induced vacuoles are surrounded by acidified small vesicles. The LysoTracker signals observed also support this model of acidified vesicles surrounding the vacuoles. The induction of vacuole formation by the botulinum “Non-Toxic” complex thus seems to differ from the vacuole induction observed for VacA and V. cholerae haemolysin. The internalised botulinum “Non-Toxic” complex is likely to associate with the late endosome and lysosome, since the induced vacuoles or surrounding vesicles were co-localised with specific proteins for these compartments.

As in the cases of vacuole formation mediated by the VacA19, V. cholerae haemolysin18, CARDS toxin15, and EHEC-Vac19, vacuole formation induced by the botulinum “Non-Toxic” complex was inhibited by bafilomycin A1—a vacuolar-type proton pump inhibitor. Since the “Non-Toxic” complex-induced vacuoles in the cells were partially acidified, however, the vacuole formation observed in this study differs from vacuolation induced by other toxins. Previous reports indicate that bafilomycin A1 inhibits receptor-mediated endocytosis in the rat sinusoidal endothelial cells20,21 and hepatocytes22, indicating that the vascular-type proton pump is essential for the acidification of the endocytic compartment and for the pH gradient between the compartments and the cytoplasm, which in turn is required for receptor-mediated endocytosis. Inhibition of “Non-Toxic” complex-induced vacuolation by bafilomycin A1 may thus result from disruption of endocytosis rather than from the prevention of vacuole swelling by proton pumping. The mechanism of the vacuole swelling observed in this study remains unclear: the acidified vesicles are generated from late endosomes and lysosomes, including “Non-Toxic” complex-induced vacuoles, in an unknown manner.

From the findings reported here, we hypothesise that the serotype C and D botulinum L-TC is a functional hybrid that exhibits both neurotoxicity and vacuolating cytotoxicity resulting from the horizontal transfer of the BoNT-NTNHA gene cluster into a hypothetical VT-producing bacterium (Fig. 4). We thus propose that the VT is an ancestor of current HA proteins. C. botulinum is taxonomically defined as a species based on only one phenotype, i.e., the production of the BoNT, but is genetically related to various other Clostridium species23. Some C. botulinum species furthermore contain multiple BoNT genes with different serotypes24. Various Clostridium species produce BoNT and some strains express combinations of BoNT genes, strongly suggesting that the BoNT gene was derived from a common ancestor and was horizontally transferred between Clostridium species. Among bacteria, the Clostridia group produces the largest number of toxins. Interestingly, the related toxin genes have been identified between distinct Clostridium species, indicating horizontal gene transfers of not only BoNT-related genes between strains23. Doxey et al. demonstrated that the BoNT and NTNHA proteins evolved from a common collagenase-like ancestor via gene duplication25. We recently showed that at least three types of HA-33 genes exist, and that these genes are shuffled among the serotype C and D strains independently of the BoNT serotype26. The duplication, reshuffling, and rearrangement of the genes in the BoNT gene cluster and its neighbour genes may have led to the functional hybridisation resulting in the botulinum toxin complex. Recently, Mansfield et al.27 furthermore identified the gene cluster encoding the BoNT and NTNHA homologs in a non-Clostridium species, Weissella oryzae SG25, which contains no HA genes. These findings may indicate that the horizontal transfer of the BoNT gene always accompanies the NTNHA gene, and that the BoNT/NTNHA complex (M-TC) in serotypes A–D forms the HA complex by chance during the evolution of the bacterium. The serotype E and F strains, however, do not contain the genes for HAs. Instead of the HA gene cluster, these strains possesses the orfx gene cluster, the function of which is unclear28. During the horizontal gene transfer event, the serotype E and F BoNT/NTNHA gene cluster appears to have “jumped” upstream of the orfx gene cluster, whereas the corresponding clusters of other serotypes were inserted downstream of the HA gene cluster.

Figure 4: Model of horizontal BoNT gene transfer of the hybrid botulinum toxin complex. Here, the ancestral BoNT gene cluster is transferred into the clostridial bacterium producing an HA complex with vacuolating toxicity [HA (VT+)], yielding an ancestor of the C. botulinum serotype C and D strains. Meanwhile, the serotype A and B strains also produce the toxin complex with HA components, but these strains do not exhibit vacuolating cytotoxicity. Therefore, the ancestral BoNT gene cluster transfers into the HA complex-producing bacterium [HA (VT−)], whose vacuolating toxicity is lost during its molecular evolution. The BoNT gene cluster transfer into a clostridial bacterium with no HA gene may yield ancestors of the serotype E and F strain. Full size image

Recently, Sugawara et al.29 demonstrated that the serotype C HA complex, but not the serotype B HA complex, affects cell viability of MDCK-I cells. The vacuolating cytotoxicity therefore seems to be restricted to serotypes C and D. Barriers represented by the intestinal wall must be overcome in order for orally ingested BoNT to enter an animal or human body. Free from the “Non-Toxic” complex, BoNT can be transported across the intestinal epithelial cell layer; however, the formation of L-TC with the “Non-Toxic” complex significantly enhances toxin transport through the cell layer5. BoNT may thus have evolved to overcome the intestinal wall barrier via complex formation with VT, a hypothetical ancestor of the botulinum “Non-Toxic” complex or HA complex, through a horizontal gene transfer event (serotypes C and D). Serotype B (in addition to A) L-TC, however, binds directly to E-cadherin and disrupts E-cadherin-mediated cell-cell interactions30. The “Non-Toxic” complex of serotypes A and B may therefore have lost its vacuolating cytotoxicity effects before or after the horizontal transfer of the BoNT gene, and instead evolved to disrupt E-cadherin-mediated cell-cell interactions.

In this study, the novel cytotoxicity of the “Non-Toxic” complex of the botulinum toxin complex was identified. The botulinum toxin complex was hypothesised to be a hybrid of the neurotoxin and “Non-Toxic” cytotoxic protein complex produced by an ancient gene transfer event from the neurotoxin-producing bacterium to the vacuolating toxin-producing bacterium. The “Non-Toxic” complex thus seems to enable the toxin complex to effectively overcome the intestinal barrier against toxin traffic. This hypothesis is supported by the fact that in botulism disease in livestock, animals sometimes exhibit haemorrhage in their intestines and vessels31,32. These symptoms have been thought to be due to enterotoxin concomitantly produced by the serotype C and D C. botulinum strains; however, our findings indicate that the botulinum “Non-Toxic” proteins may also play a role in these symptoms. Further investigation into the cytotoxicity of the “Non-Toxic” complex is required for a greater understanding of this research area to be gained, and such investigations may contribute toward the development of strategies in the prevention of botulism disease. An in-depth knowledge of the toxin traffic mechanisms of the “Non-Toxic” complex would furthermore allow for the development of novel oral drug-delivery systems in which peptide and protein pharmaceuticals are attached to the botulinum toxin complex instead of the BoNT protein and in which the “Non-Toxic” complex is modified to exhibit only moderate cytotoxicity.