Cholesterol-dependent cytolysins, bacterial pore-forming toxins secreted by several pathogenic Gram-positive bacteria, such as Streptococcus spp., Clostridium spp. and Listeria spp., are considered to be important virulence factors. SLO, secreted by Streptococcus pyogenes, has been shown to be important for pathogenicity in a mouse infection model , to accelerate caspase-dependent apoptosis in macrophages , and to induce various cellular immune responses [3-5]. PLY, from Streptococcus pneumoniae, is reported to be associated with pneumococcal pneumonia in mice  and to induce immune responses ([7-12], review in ). PFO, from Clostridium perfringens, contributes to cytotoxicity against macrophages and escape from macrophage phagosomes . LLO, from Listeria monocytogenes, facilitates escape of the bacterium from phagosomes , resulting in cellular internalization of bacteria ; it also causes some immune responses in target cells ([17-20], review in ).
Within the CDC family, the overall molecular structure is highly conserved. CDCs are typically composed of four domains (D1–D4) of molecular weight approximately 50–60 kDa. Domains D1, D2 and D3 contribute to creation of membrane pores by penetration of part of D3 into target cell membranes, whereas domain D4 is associated with receptor(s) recognition on target cell membranes . The exact mechanisms of membrane recognition and pore-formation by CDCs have been elucidated in detail . Briefly, within a CDC molecule (monomer), a loop structure comprising β-strand 5 (β5) and α-strand 1 (α1) in domain D3 is rotated away from β-strand 4 (β4), enabling the exposed edge of the latter to pair with structure β-strand 1 (β1) of another monomer. This facilitates and, through repetition, extends CDC monomer–monomer contacts, resulting in formation of a large, ring-shaped structure or “prepore” [23, 24]. Interaction of D4 and the target cell membrane causes extensive structural change, namely, unfolding in TMH 1 and TMH2 of D3 . These pore-forming mechanisms are thought to be highly conserved in CDCs.
Unlike the secretion mechanism observed with typical CDCs, until recently secretion of PLY had been thought to depend on autolysis induced by autolysin LytA encoded by the lytA gene . However, research findings have suggested that LytA is not responsible for release of PLY ; rather, PLY is exported from the cytoplasm in a PLY D2-dependent manner and localized in the bacterial cell wall . Similarly an N-terminal secretion signal sequence is absent in MLY, a homologue of PLY expressed by some S. mitis strains that is secreted into culture supernatants . It is therefore of interest to investigate associations between the secretion mechanisms of CDCs that lack an important unit for secretion and their pathogenicity. Furthermore, a new type of CDC named LLY, which is a homologue of Sm-hPAF and has an additional-domain in the N-terminal, has been reported from S. mitis strain SK597 . This additional domain has amino-acid sequence similarity with both Anguilla anguilla agglutinin and the glycan binding domain of the family 98 glycoside hydrolases from S. pneumoniae, and functions as an Ley and Leb specific lectin domain . In summary, variations in molecular structure and mode of secretion of CDCs have been elucidated by several recent studies.
There are also variations in CDC receptor recognition. It has been generally accepted that the receptor for CDCs is membrane cholesterol. However, after discovery of the receptor for the human-specific CDC ILY secreted by human oral commensal S. intermedius, this generalized perception had to be changed. The receptor for ILY is the GPI-anchored glycoprotein, huCD59 ; the interaction between ILY and huCD59 forms the basis for its human specificity. Furthermore, it was recently reported that the hemolytic activity of LLY against human erythrocytes is inhibited by anti-huCD59 antibody and also by prepore-locked ILY . In addition, VLY secreted by Gardnerella vaginalis reportedly recognizes huCD59 as its receptor and shows human-specific activity . Regarding the mode of receptor recognition, it has become apparent that there are at least two groups of CDCs: typical CDCs recognizing and binding to membrane cholesterol and atypical CDCs that recognize huCD59.
Previously, Ohkuni et al. reported Sm-hPAF secreted from S. mitis strain Nm-65 isolated from a patient with Kawasaki disease . According to the full-length amino acid sequence encoded by the sm-hpaf gene (GenBank ID: AB051299), Sm-hPAF is the most homologous to LLY, differing from it by only 12 amino acid residues. Interestingly, Sm-hPAF has unique, species-dependent hemolytic activity, humans being the most susceptible species followed by horses, rabbits, rats, sheep and chickens in descending order of susceptibility . Therefore, the demonstration of an additional category of CDCs displaying features intermediate between cholesterol binding (“typical”) CDCs and CDCs recognizing huCD59 suggests that the mode of receptor recognition in CDCs is more diverse than hitherto recognized.
In the present study, we investigated the diversity in mode of receptor recognition of CDCs, focusing on comparisons between Sm-hPAF, which has species-dependent activity but relatively relaxed human-specificity, the more stringently human-specific ILY, and the non-species-specific typical CDCs. We also re-evaluated the human-specificity of VLY.
- Top of page
- MATERIALS AND METHODS
- Supporting Information
Cholesterol-dependent cytolysins are bacterial protein toxins secreted particularly from several species of pathogenic Gram-positive bacteria. This toxin family was formerly called “thiol-activated cytolysins” or “cholesterol-binding cytolysins.” However, because the discovery of ILY introduced a new sub-group into the gene family (huCD59-recognizing cytolysins with dependency on the presence of cholesterol in the target membrane for pore-forming activity), all these cytolysins were termed CDCs. CD59 is a GPI-anchored glycoprotein that helps to confer self-protection from the cytotoxic effect of the membrane attack complex induced by activation of the complement system . Since the demonstration of ILY and its receptor huCD59, several studies on this atypical type of CDC have shown no or very low cholesterol binding [30, 31, 35, 42, 43, 46-51]. A second huCD59-recognizing CDC, named VLY, was subsequently discovered , followed by a report of the recognition of huCD59 by LLY . Consequently a novel CDCs sub-family recognizing huCD59 as the receptor is now established.
To date, two modes of receptor recognition by CDCs have been reported: (i) the more typical cholesterol-binding mode; and (ii) an atypical huCD59-recognizing mode. The latter mode constitutes a sub-family of CDCs that includes ILY, VLY and LLY. As shown in the present study, Sm-hPAF, a homologue of LLY, also belongs to the sub-family with an atypical huCD59-recognizing mode (Fig. 3). This property of Sm-hPAF seems to be responsible for the observed human-preferential activity. However, unlike the more strictly human-specific activity of ILY, Sm-hPAF displays hemolytic activity/cytotoxicity against non-human erythrocytes (34), human cells not expressing huCD59 (Fig. 6a), and rat hepatic cells with no huCD59 (Fig. 6d). This property of the hemolytic activity/cytotoxicity of Sm-hPAF may be partly attributable to their ability to recognize membrane cholesterol, as do the typical CDCs (Figs. 4a, 6k,l and Tables 1, 2). These findings enabled recognition of a new group of Sm-hPAF that possesses receptor recognition properties intermediate between cholesterol-binding typical CDCs and huCD59-recognizing atypical CDCs. Furthermore, the results of the present study suggest that Sm-hPAF preferentially recognizes huCD59 over cholesterol because of (i) the greater susceptibility to Sm-hPAF of huCD59-expressing BRL3A cells compared with the parent, non-human BRL3A cells (Fig. 6d), and (ii) the lesser susceptibility to Sm-hPAF of huCD59-negative U-937 DE-4 compared with huCD59-positive HL60 (Fig. 6a). Quantitative comparison of the affinity of Sm-hPAF to huCD59 with that to cholesterol has not yet been accomplished because of the difficulties in efficient purification of native huCD59 and in reconstitution of the physiological state of huCD59, that is, huCD59 in lipid raft, without cholesterol condition. However, we speculate that the choice by Sm-hPAF between huCD59 and cholesterol receptors might depend on differences in affinity to these molecules on biological membranes. Thus, huCD59-preferential recognition by Sm-hPAF would contribute to their cytotoxicity because of the characteristic of preferentially targeting human cells. Though the precise mechanism of human-preferential, species-dependent hemolytic activity/cytotoxicity of Sm-hPAF is so far unclear, we speculate that, because the primary- and the secondary-structures of the region in huCD59 reported to interact with ILY (30) is not conserved in other CD59 of animal origin, the factor most likely to be responsible for this remarkable property of Sm-hPAF is the difference in affinity between Sm-hPAF and CD59 of various animal origins. Unlike ILY, it seems that the CD59-binding sites of Sm-hPAF and the homologues belonging to this CDC type are not best fit for huCD59; rather, they are somewhat loose, allowing interactions with not only huCD59 but also with other animal CD59s. Therefore Sm-hPAF can also bind to animal CD59s with individual affinity. Attempts at constructing transformants of U-937 DE-4 (huCD59 null cells) with CD59 genes from other animal species in order to further understand the basis for Sm-hPAF species-specificity have so far remained unsuccessful because there has been no detectable expression of CD59 (data not shown). Thus, further study is necessary in order to understand the species-specificity shown by Sm-hPAF.
As to the association of ILY with cholesterol, Dowd et al. have reported that ILY binds to cholesterol-rich POPC liposomes . However, we found no significant association between ILY and cholesterol-containing membranes (Fig. 4b and Table 2); our data strongly suggest that ILY does not specifically associate/interact with membrane cholesterol. Therefore, Dowd et al.'s finding of relatively high concentrations of ILY on cholesterol-rich POPC liposomes may have been attributable to non-specific adsorption rather than reflecting an interaction between ILY and human cells under physiological conditions. Such non-specific adsorption may be facilitated by ILY's highly positive charges (calculated pI: 9.87), which cause non-specific binding as we found during purification of ILY by ion-exchange chromatography .
Recently, Johnson et al. reported that the crystal structure of the ILY-huCD59 complex has two interfaces on huCD59 coordinate ILY monomers . The feature of this interaction may indicate that clustering of huCD59 facilitates binding and ring oligomer (known as “prepore”) formation of ILY. As shown in a previous study, after incubation with human erythrocyte ghost membrane, ILY forms membrane pores that are irreversibly embedded in membranes with SDS-resistance . However, binding of ILY is thought to induce reversible oligomerization that forms “prepores”, the stage prior to formation of membrane-embedded pores of CDCs. We speculate that, supported by the avidity with which ILY molecules associate, ILY binds to huCD59 clusters in lipid rafts with high affinity. However, the binding affinity of ILY to dispersed huCD59 would be significantly weakened by MβCD treatment, which disrupts the lipid raft structure and thus the support for association avidity. This explains why PBS washing only easily washed out ILY bound to MβCD-treated HL60 cells, whereas very little bound ILY was washed out of normal HL60 cells (Fig. 5i). Johnson et al. reported that some amino acid residues involve primary- and secondary-binding sites of ILY for interactions with huCD59 . Among the residues involved in ILY-huCD59 interactions, the conserved residues in huCD59-binding CDC are Y436 and R480. In order to describe the determinant structure/residue for receptor recognition of ILY (Group II) and huCD59-binding CDC (Group III), further information is necessary.
In this study, we suggest a new categorization of the modes of receptor recognition in CDCs (Figs. 9, 10). Group I comprises typical CDCs with high affinity to cholesterol and no or very little affinity to huCD59: most CDCs so far described belong to this group. Group II, ILY being the only group member so far, comprises atypical CDCs with no or only very little affinity to cholesterol and high affinity to huCD59. Group III, a novel group of atypical CDCs with affinity to both cholesterol and huCD59, includes Sm-hPAF, VLY and possibly LLY. The amino acid sequence data for CDCD4s also support this categorization; the amino acid sequence of a signature motif for huCD59-recognizing CDCs (Tyr-X-Tyr-X14-Arg-Ser)  and 11mer region was completely conserved amongst the members of Group III (Fig. 9c). Two of the five amino acids in the CRM (L1), L2 and L3 that are inserted into the membrane surface, are not completely conserved (Fig. 9c). We also observed one of the amino acid variations in L3 in CDCs belonging to Group I (Fig. 9c). However, previously reported data showing that the A464D mutation in ILY does not affect oligomer assembly while blocking the conversion required for pore formation  suggest that this amino acid makes only a small contribution to the interaction with membrane cholesterol and that the amino acid in L3 is not very important in the cholesterol-binding property of CDCs. Another variation has been observed only in LLY; namely, the amino acid Ile in L2 replacing the Val that is present in other CDCs. L3 is located immediately behind this residue , variation in which may also be relatively unimportant for cholesterol binding. Furthermore, it has been reported that only two amino acids of CRM (L1) are essential for cholesterol recognition . Thus, it seems that all CDCs potentially possess the property of cholesterol binding.
These observations invite speculation as to why there is diversity in receptor recognition. Interestingly, previously reported data shows that substitution of the 11mer region in ILYD4 with that of the consensus 11mer of CDCs, ECTGLAWEWWR, makes ILY non-human specific and susceptible to cholesterol inhibition, as seen with typical (i.e., Group I) CDCs . On the other hand, in a review of the interaction between ILY and huCD59, two important functions were suggested : one being initiation of structural change for pore formation and the other positioning of CRM in order to bind to membrane cholesterol and initiate further insertion of short hydrophobic loops into the membrane. Based on these reported findings, although the above consensus sequence in typical CDCs does not affect the interaction between CRM and cholesterol, this region in ILY may inhibit the interaction of CRM on the surface of D4 with the target membrane because the 11mer region is a structure that loops out from the rigid body of D4. It is possible that the interaction of ILY with huCD59 induces conformational changes in the 11mer region that unmask steric hindrance and allow interaction between CRM and membrane cholesterol. In this context, a clear classification of the 11mer region into three patterns representing Groups I–III is desirable (Fig. 9c). In the case of Group III CDCs with 11mer region sequence consensus, CDCs which can directly bind to cholesterol but with an affinity lowered by incomplete hindrance of the region, the inhibitory or masking effect in the interaction of CRM with membrane cholesterol may be less marked than with ILY. However, how this region of Groups II and III participates in the selective interaction with huCD59 is so far unclear. Further research is necessary to identify the basis for the observed species-dependency of Sm-hPAF. This would have to take into account the variations in affinity to CD59 of humans and other animals seen in CDCs of Groups II and III, defined by the complex interactions between the 11mer region and the signature motifs for cholesterol binding and huCD59-recognition. The answer to this question may be found in future studies using in silico analysis, such as molecular modeling of CDCs and their receptors, cholesterol and CD59.