Many plant lectins have the ability to induce cellular toxicity, likely due to their role in self-defense against invading organisms such as pathogenic bacteria or insects [68,69]. This toxicity arises from their capacity to cross-link surface carbohydrate-containing molecules on target cells, influencing the signaling pathways within those cells. Moreover, when incorporated into protein toxins, lectin domains can function as cell-binding modules to exert potent toxicity. A well-known example of a toxic lectin is ricin, which was the first discovered toxic lectin derived from the castor bean (Ricinus communis) [70]. Ricin consists of an A chain with N-glycosidase activity that inactivates eukaryotic ribosomes and a B chain that serves as a lectin subunit. The B chain assists in delivering the A chain to the cytosol by binding to cell surface glycans [71,72]. Similar toxins with enzymatic “A” subunits and receptor-binding “B” subunits (lectin subunits) are referred to as AB toxins and are also produced by various pathogenic bacteria [73]. Another type of toxic lectin includes pore-forming proteins, which form ion-permeable pores in the target cell membrane after binding to cell surface glycans through the lectin domains. These examples demonstrate the important roles that lectins often play in the actions of protein toxins. It has also been revealed that venomous marine animals produce various toxins containing lectin domains, suggesting a close association between their carbohydrate-binding activity and their toxic actions.

The venom of the globiferous pedicellariae of the venomous sea urchin T. pileolus contains several toxic proteins that exert various biological effects [74]. Among these proteins, the galactose-specific lectin SUL-I exhibits various activities, such as chemotactic activity on guinea pig neutrophils and mitogenic activity on murine splenocytes [75,76]. The amino acid sequence of SUL-I indicates that it belongs to the L-rhamnose-binding lectins (RBLs), many of which have been found in fish eggs [77]. SUL-I is composed of 284 residues (30.5 kDa) with three repetitive sequence regions, each showing similarity to CRDs of rhamnose-binding lectins (RBLs) [78]. The three-dimensional structure of the rSUL-I/L-rhamnose complex was determined by X-ray crystallographic analysis with a resolution of 1.8 Å (Figure 1) [79]. The overall structure of rSUL-I consists of three distinctive CRDs, which share structural similarities with the CRDs of CSL3, the rhamnose-binding lectin from chum salmon (Oncorhynchus keta) eggs [80]. The bound L-rhamnose molecules are primarily recognized by rSUL-I through hydrogen bonds between their 2-, 3-, and 4-hydroxy groups and Asp, Asn, and Glu residues in the binding sites. When interactions of rSUL-I with oligosaccharides were examined, a higher affinity was observed for galactosylated (asialylated) N-glycans, suggesting that the potential target carbohydrate structures are galactose-terminated N-glycans. While the CRDs of rSUL-I adopt similar folds compared to those of CSL3, a significant difference was found around the variable loop in the carbohydrate-binding sites, which may be related to the difference in binding specificities for oligosaccharides between these lectins. SUL-I has carbohydrate-binding sites in its three domains, all oriented towards the same side of the protein. This structure could be advantageous for cross-linking specific membrane glycoproteins containing galactose-terminated carbohydrate chains, thereby triggering cellular responses.

CEL-III is a galactose-specific lectin from C. echinata, which also contains the previously mentioned C-type lectins CEL-I and CEL-IV. Although CEL-III shows Ca²⁺-dependent carbohydrate binding, it belongs to the R-(ricin)-type lectin family, characterized by two β-trefoil folds [81]. Interestingly, it exhibits strong hemolytic activity and cytotoxicity [82]. Hemolysis induced by this lectin is prominent in the alkaline region above pH 7 in the presence of Ca²⁺ and can be inhibited by galactose and carbohydrates containing galactose at nonreducing ends. These observations suggest that the Ca²⁺-dependent galactose binding ability is a prerequisite for its hemolytic action. CEL-III mediates hemolysis by forming pores in the target erythrocytes, relying on its galactose-specific lectin activity. Structural analyses of CEL-III have revealed that it consists of two R-type lectin domains (domains 1 and 2) in the N-terminal two-thirds and a C-terminal one-third with a β-sheet-rich novel domain (domain 3) (Figure 2) [83,84]. Domain 1 contains two carbohydrate-binding sites, while domain 2 contains three. The binding of galactose-related carbohydrates is mediated by coordinate bonds with Ca²⁺ ions bound in the carbohydrate-binding sites, along with hydrogen bond networks involving nearby amino acid residues. This mechanism shares similarities with the carbohydrate-recognition mechanism of C-type CRDs, despite the lack of homology in the amino acid sequences between CEL-III and C-type CRDs.

The crystal structure of the soluble oligomer of CEL-III, induced by the binding of lactulose, a galactose-containing disaccharide, revealed that CEL-III heptamerizes upon binding to galactose-containing carbohydrates (Figure 3). This binding leads to the formation of a β-barrel structure (25 Å) composed of the C-terminal β-sheet-rich domain [84]. When these conformational changes occur on the target cell surface, the β-barrel forms a membrane pore, allowing the passage of small molecules and ions across the cell membrane. This drastic change in the tertiary structure of domain 3, from a soluble form to a membrane-penetrating β-barrel, is driven by the formation of numerous hydrogen bonds within the β-barrel and hydrophobic interactions between the hydrophobic face of the β-barrel and the nonpolar interior of the membrane.

Oligomerization of CEL-III can be triggered by β-galactoside-containing disaccharides, including lactose (Galβ1-4Glc), lactulose (Galβ1-4Fru), and N-acetyllactosamine (Galβ1-4GlcNAc), but not by galactose, N-acetylgalactosamine, and melibiose (Galα1-6Glc) [85]. This suggests that heptamerization of CEL-III on the target cell surface is induced by cell surface receptors containing β-galactoside linkages, particularly glycosphingolipids such as lactosyl ceramide, which has been demonstrated to be an effective receptor for membrane pore formation in artificial lipid vesicles [86]. It appears that the drastic conformational change of CEL-III is triggered by the structural changes in the binding of the specific carbohydrate into domains 1 and 2, which are then transmitted to domain 3, leading to the opening of the interface with domain 3 [84].

The presence of β-trefoil fold domains containing QXW motifs [87] has been observed in a wide range of proteins, including lectins, toxins, carbohydrate-related enzymes, and immune cell surface receptors, which function as carbohydrate-binding modules. Structural analysis of the hemolytic lectin LSL, isolated from the mushroom Laetiporus sulphureus, has revealed that it also comprises an N-terminal carbohydrate-binding domain with a β-trefoil fold and a C-terminal pore-forming domain [88]. Notably, the pore-forming domain of this lectin exhibits remarkable structural similarity to those found in pore-forming toxins of pathogenic bacteria such as Aeromonas hydrophila and Clostridium perfringens [89]. Based on these findings, it can be inferred that LSL, similar to CEL-III, exhibits hemolytic activity by binding to cell surface glycans through its carbohydrate-binding domain and subsequently inserting a pore-forming domain into the cell membrane. Although the pore-forming domain of LSL does not share structural similarity with domain 3 of CEL-III, both domains are characterized by elongated β-sheets, which likely contribute to a favorable structure for penetrating the cell membrane.

Some fish species possess stinging spines that contain protein toxins; among them, a toxin called natterin has been isolated from the spines of Thalassophryne nattereri [65]. Additionally, a homologous protein Ij-natterin has been found to be expressed in the spine venom of Inimicus japonicus [66]. Natterin and Ij-natterin consist of an N-terminal lectin domain, which is homologous to the oyster lectin CGL1 (DM9 domain protein) [57], and a C-terminal aerolysin-like pore-forming domain (Figure 4A). Although the specific carbohydrate-binding activity of the lectin domain has not been reported to date, its amino acid sequence similarity with CGL1 suggests that it may have mannose-specific lectin activity.

Proteins that have C-terminal domains homologous to those of natterin, called natterin-like proteins, are also widely distributed in fish skin [90,91]. The crystal structure of Dln1, one of the natterin-like proteins from zebrafish (Danio rerio), has revealed that Dln1 consists of an N-terminal jacalin-like lectin domain and a C-terminal aerolysin-like domain (Figure 4B) [92]. The lectin domain was found to bind mannobiose (Manα1-2Man and Manα1-3Man), indicating that Dln1 exhibits an affinity for high-mannose glycans. Furthermore, electron microscopic analysis revealed that upon binding to high-mannose glycans, Dln1 undergoes significant conformational changes, leading to the formation of octameric pores in the lipid membrane.

Although natterin and natterin-like proteins possess C-terminal aerolysin-like domains in common, they differ in N-terminal lectin domains; the former shows a CGL1-like DM9 domain fold, while the latter has a jacalin-like domain. Considering the large conformational change during the pore-forming process of Dln1, it seems highly likely that natterin (and Ij-natterin) also induce significant conformational changes upon binding to target cell surface glycans, leading to oligomerization and pore formation. Similar large conformational changes after binding to cell surface carbohydrate chains are also observed in the course of pore formation by the hemolytic lectin CEL-III (Figure 3). The lectin domains of these pore-forming proteins not only play roles in binding to specific carbohydrate chains on the target cell but also trigger subsequent conformational changes leading to the formation of oligomers with transmembrane pores.