The botulinum toxin complex meets E-cadherin on the way to its destination

Abstract

Botulinum neurotoxin (BoNT) causes the disease botulism, which is characterized by flaccid paralysis, in humans and animals. The metalloprotease activity of BoNT inhibits neurotransmitter release at neuro-muscular junctions. In most cases, poisoning occurs when BoNT is ingested. Therefore, BoNT must pass through the epithelial barrier of the gastrointestinal tract to enter the systemic circulation and reach the target site. BoNT forms large protein complexes by associating with non-toxic components referred to as non-toxic non-hemagglutinin (NTNH) and hemagglutinin (HA). These proteins protect BoNT from the low pH and proteases in the digestive tract. We recently determined that HA has an unexpected function of disrupting the intercellular epithelial barrier by directly binding to E-cadherin. HA binds to E-cadherin and disrupts its function in a species-specific manner, and this interaction is essential to disrupt tight junctions. This activity is thought to facilitate the absorption of BoNT through the paracellular route of the intestinal epithelium in susceptible species.

BoNT is the causative agent of botulism, which is divided into seven serotypes, designated A–G (BoNT/A-BoNT/G). Types A, B, E and F cause botulism in both humans and animals. In contrast, types C and D cause botulism mainly in animals and are rarely associated with clinically significant disease in humans.1 BoNTs target peripheral nerve endings where they proteolytically cleave SNARE proteins to inhibit neurotransmitter release.2 BoNTs are produced as large protein complexes and the composition of these complexes differs among serotypes.3,4 Types B–D form two types of complexes referred to as the 12S and 16S toxins, in which non-toxic components attach to BoNT. The 12S toxin consists of BoNT and the NTNH, while the 16S toxin is comprised of BoNT, NTNH and HA, which itself is made up of three proteins, HA1, HA2 and HA3. In addition to the 12S and 16S toxins, BoNT/A forms the 19S toxin, which is thought to be a dimer of the 16S toxin. In the 1970s, Sakaguchi's group found that the 12S toxin had higher oral toxicity than BoNT and that the 16S toxin had higher toxicity than the 12S toxin.5 In addition, they showed that BoNT is more stable in digestive fluids as a higher molecular weight complex.6 They concluded that the non-toxic components protect BoNT from the low pH and enzymes in the digestive tract after oral ingestion and thereby contribute to the oral toxicity of the BoNT complex.3

HA, as indicated by its name, has hemagglutination activity through its carbohydrate-binding activity. Although the BoNT complex was determined to have hemagglutination activity approximately 60 years ago, the significance of this activity in the pathogenesis of botulism was unknown.7 In the late 1990s, Fujinaga et al. reported that the HA in type C 16S toxin binds to the intestinal epithelium by recognizing carbohydrates.8 HA was shown to facilitate toxin absorption through the intestinal epithelium in guinea pigs and is considered to contribute to the high oral toxicity of the 16S toxin. Subsequently, HAs from other serotypes of the botulinum toxin complex have been shown to bind to intestinal epithelial cells.9–11 Furthermore, it was reported that the type C 16S toxin, but not the 12S toxin or BoNT alone, induces massive internalization of the toxin complex and subsequent transport of BoNT into the Golgi in HT-29 cells, a human colon carcinoma cell line.12,13 One of the simplest conclusions from these observations is that the binding properties of HA in the BoNT complex induce the transport of BoNT across the intestinal epithelial barrier via endocytosis and vesicular trafficking, a process termed transcytosis. In contrast, it was reported that BoNT alone traverses the intestinal epithelial barrier via transcytosis in cultured epithelial cells and that the non-toxic components do not facilitate BoNT transcytosis.14,15 Thus, the role and importance of the adhesive properties of HA in the passage of BoNT across the intestinal epithelium remain controversial.

In addition to the ability of HA to adhere to the intestinal epithelium, we found that type B HA disrupts the intestinal epithelial intercellular barrier.16 HA compromises the localization of tight and adherens junction proteins, such as occludin, ZO-1, β-catenin and E-cadherin, at cell-to-cell boundaries without affecting the viability of cultured epithelial cells. Furthermore, we showed that HA facilitates the paracellular transport of 12S toxins in an in situ loop assay. In addition to type B, type A HA also exhibited this barrier-disrupting activity. Type C HA also disrupted the epithelial barrier in cultured cells derived from certain species, but this phenotype appeared to result from cytotoxicity and differed from the activities of types A and B.17 Recently, we determined that epithelial cadherin (E-cadherin) is a target molecule of type B HA.18 E-cadherin is a canonical member of the cadherin superfamily of adhesion molecules and mediates calcium-dependent cell-to-cell adhesion.19 In epithelial cells, E-cadherin is present in adherens junctions and is essential for the function of tight junctions.20 HA directly binds to E-cadherin and disrupts E-cadherin-mediated cell-to-cell adhesion. All three HA subunits were needed for full binding to E-cadherin, although a weak binding ability was observed in a complex consisting of HA2 and HA3. The carbo-hydrate-binding activity of HA is not required for this interaction. Although HA binds to human, bovine and mouse E-cadherin, it does not bind the rat or chicken homologs of E-cadherin. HA does not interact with other members of the classic cadherin family, such as N- and VE-cadherin. Expression of rat, but not mouse, E-cadherin rescues MDCK I cells, which are susceptible to HA action, from HA-induced disruptions in tight junctions, indicating that the HA-E-cadherin interaction is essential to disrupt tight junctions.

Several pathogens have been shown to target E-cadherin to invade the host epithelial barrier. The Listeria monocytogenes surface protein, internalin A, directly binds E-cadherin, and this interaction allows the bacterium to cross the intestinal barrier, which is a critical step in the pathogenesis of listeriosis.21,22 The pathogenic fungus Candida albicans also exploits E-cadherin to internalize into mammalian cells through direct interactions between the surface adhesin Als3 on the fungus and E-cadherin or N-cadherin on the target cell.23 These two pathogens use E-cadherin to internalize into host cells. Meanwhile, Bacteroides fragilis, Porphyromonas gingivalis and C. albicans secrete proteases that cleave E-cadherin and thereby disrupt and invade the intercellular epithelial barrier.24–29 Botulinum HA appears to breach the epithelial barrier through a mechanism that resembles the latter case, although botulinum HA exerts its activity without proteolytically cleaving E-cadherin. Even the 12S toxin alone, which is devoid of HA, can be absorbed from the intestinal epithelium; however, HA-mediated disruption of the intestinal epithelial barrier might further facilitate the influx and dissemination of BoNT, instead of bacteria or fungi in the above examples, into the systemic circulation through the paracellular route, thereby contributing to the high oral toxicity of the 16S toxin.

There are three major questions about the barrier-disrupting activity of HA. First, how does apically localized HA interact with E-cadherin, which exclusively resides on the basolateral surface of epithelial cells? Because the action of HA was inhibited in the presence of an anti-HA antibody in the basolateral chamber of Transwell, an in vitro epithelial model, the HA target site is on the basolateral surface, and HA must be transported from the apical to basolateral surface to co-localize with E-cadherin.16,18 Transcytosis appears to mediate the transport of HA, at least in vitro. We found that Caco-2 cells are susceptible to apically administered HA, and HA was internalized from the apical surface in Caco-2 cells but not MDCK-I cells, which are not susceptible to apically applied HA. Thus, transcytosis seems to be a prerequisite for the action of HA. It has been reported that the nontoxic components are not dissociated from BoNT in the intestinal fluid; therefore, it is possible that HA is transcytosed as a whole toxin complex in the intestine.6 Alternatively, the BoNT complex may dissociate in the intracellular compartment, and HA might be transported through its specific trafficking pathway.13 Thus, this first trafficking mechanism of HA remains unclear.

A second question is how does HA disrupt E-cadherin-mediated cell-to-cell adhesion and the intercellular barrier? We showed that the residue 20 of E-cadherin is critically involved in the interaction with HA.18 This residue is located in the most distal of five tandemly repeated extracellular cadherin domains (EC1), and near the cadherin trans-dimer interface.30 Therefore, it is conceivable that HA binding to E-cadherin sterically hinders E-cadherin trans-dimer formation. Using bacterially expressed EC proteins, we found that EC1 alone is insufficient and EC2, which is adjacent to EC1, is also required for the interaction with HA.18 Thus, HA seems to recognize several residues in the EC1 and EC2 domains of E-cadherin. The mechanism by which HA disrupts E-cadherin-mediated cell adhesion appears to resemble that of anti-E-cadherin function-blocking antibodies because both directly bind to the E-cadherin extracellular region. However, anti-cadherin function-blocking antibodies only partially disrupt E-cadherin-mediated cell adhesion and intercellular junctions, while HA causes a complete loss of junctional integrity in affected cells.31,32 Thus, HA must have molecular properties that explain its remarkable ability to disrupt barriers. One possible explanation for this function is that HA recognizes E-cadherin through several different sites within the EC1 and EC2 domains and that these binding sites represent a much broader molecular surface than the binding sites of anti-E-cadherin function-blocking antibodies. This may enhance the steric hindrance of E-cadherin homophilic binding. Alternatively, HA might function as a trivalent complex, while IgG is divalent. In a structural model of the type D 16S toxin, HA3 forms a trimer and an HA1 and HA2 complex is attached to each HA3 molecule.33 This model proposes that HA functions in a trivalent manner if one HA3 is associated with an HA1/HA2 sub-complex in a single functional unit. Finally, we cannot exclude the possibility that HA interacts with other molecule(s) within intercellular junctional complexes to mediate its function(s). These possibilities are not mutually exclusive, and additional studies are needed to address these hypotheses. Additionally, the events downstream of HA binding to E-cadherin remain to be elucidated.

Finally, is the barrier-disrupting activity involved in the pathogenesis of food-borne botulism? We observed that the action of HA was species-specific and interestingly, this species specificity appeared to correlate with the epidemiology of botulism. For example, type B HA did not interact with chicken E-cadherin, which is consistent with the rare reports of avian botulism caused by BoNT/B. Furthermore, birds were experimentally resistant to the BoNT/B complex, especially when administered orally.34,35 These findings indicate that the HA-E-cadherin interaction is significant. However, the relevance of the action of HA in the pathogenesis of food-borne botulism remains to be elucidated and is an area of future study.

Abbreviations

BoNT	=	botulinum neurotoxin
EC	=	extracellular cadherin
HA	=	hemagglutinin
NTNH	=	non-toxic non-hemagglutinin
TJ	=	tight junction