+Corresponding author. Fax: +55-85-288.9818. E-mail: vramos@ufc.br
⁺⁺Research fellow CNPq

Received 6 July 1998
Accepted 18 November 1998

Key words: Abrus pulchellus - protein-carbohydrate interaction - plant toxin

A highly toxic protein, which exhibits similar effects to ribosome inactivating proteins (RIPs type II), isolated from Abrus pulchellus seeds kills mice in a few hours after injection in the peritoneal cavity (MV Ramos et al. 1998a Toxicon 36: 477-484). It is well documented in the literature that some toxins mediate their incursion by recognizing glycoconjugates on the cell surface membrane before translation. Plant and microbial toxins have been studied in detail mainly to understand their intrinsic toxic pathway. However, the toxin recognition mechanism by membrane receptors and its translation through the membrane is poorly investigated, excepting abrin, the most investigated plant toxin. It has been determined that the initial recognition step must be mediated by carbohydrate-protein interaction. The ribosome inactivating protein from the seeds of Viscum album is a typical protein of this class and its carbohydrate-binding specificity was carefully investigated (H Debray et al. 1994 Glycoconjugate J 11: 550-557). Although the RIPs type II are commonly specific for galactose residues, more complex carbohydrates are distinctly recognized.

Thus, to well define the carbohydrate moiety that interacts with a toxin, a number of carbohydrates (monosaccharides, di- tri- and oligosaccharides) and complex glycans from glycoproteins should be tested. In this respect, we have investigated the carbohydrate-binding specificity of a toxin from A. pulchellus seeds by the combination of haemagglutination inhibition assay and affinity chromatography using immobilized glycoproteins.

The toxin was isolated following the procedure previously described (Ramos et al. 1998a loc. cit.) and the haemagglutinating activity induced by the toxin was studied using rabbit and human ORh⁺ erythrocytes. Carbohydrates, glycoproteins and immobilized glycoproteins columns were obtained from Sigma Chemical Co. Haemagglutination inhibition assays were performed in U-bottomed plates as formally described (RA Moreira & JC Perrone 1977 Plant Physiol 59: 783-787). Affinity chromatography on avidin-Sepharose, fetuin-agarose and mucin-Sepharose columns with the purified toxin were carried out in the same conditions described in the Figure.

The haemagglutinating activity induced by the A. pulchellus seed toxin was not inhibited by a large range of mannose/glucose monosaccharides and their derivatives. Carbohydrates as D-idose, D(+)xylose, D(+)ribose and most of the glycoproteins tested were not recognized by the toxin (Table I). Although the RIPs type II are commonly described as galactose-specific proteins, some galactose derivatives are more potent inhibitors, like methylated forms or lactose (b -D-Gal-[1® 4]-a -D-Glc) (Table II). Interestingly, the p-nitrophenyl groups attached to C1 of galactose ring did not increase the inhibitory potency. In general, for plant carbohydrate-specific proteins, such as true lectins, it has been shown that the hydrophobic interactions established between amino acid residues in the vicinity of the carbohydrate-binding site and the p-nitrophenyl group, enhances the affinity of the complex (MV Ramos et al. 1996 Rev Bras Fisiol Veg 8: 193-199). Remarkably, the inhibition patterns obtained using rabbit erythrocytes differ from those with human ORh⁺ cells. This may be due to the differential distribution of potential receptors on the cell surface membrane in each erythrocyte, or their accessibility to the toxin binding-site. Therefore, the use of two cell lines is a strategy to better determine the inhibitory effect of sugar and glycoproteins. Regarding the inhibitory effect of glycoproteins, only asialofetuin (possessing a triantennary N-linked glycan), glycophorin (O-linked and N-linked glycans), human lactotrans-ferrin (biantennary N-linked glycan, usually fucosylated) and porcine thyroglobulin (possessing highly heterogeneous N-linked glycans) were recognized by A. pulchellus toxin. Glycophorin was only active when tested against human ORh⁺ erythrocytes. With the excepting of glycophorin which presents both O- and N-linked glycan structures, these glycoproteins are characterized by the presence of complex glycan N-acetyllactosamine type of high heterogeneity, branched to asparagine residues. However, glycophorin a glycoprotein largely present in human erythrocytes membranes possesses a well defined glycan structure O-linked to serine residues, composed of Ser-GalNAc-[b 1® 3Gal-a 2® 3NANA]a 2® 6NANA which may be a receptor for the toxin in human erythrocytes. The interaction with human lactotransferrin, asialofetuin and porcine thyroglobulin glycan structures of high heterogeneity may explain why the plant toxin RIPs can invade almost all types of Eukariotic cells of unrelated organisms, systems or organs.

Affinity chromatography on fetuin-agarose and porcine stomach mucin-Sepharose did not reproduce the result from inhibition tests (Fig.).

The toxin interacted with the immobilized glycoproteins although inhibition tests failed to reveal the interactions. According to the glycan structure of the desyalylated fetuin (asialofetuin), the presence of N-acetylneuraminic acid in the terminal non-reducing extremity of fetuin could mask the recognition when the inhibition process between the free glycoprotein and a membrane receptor is carried on. However, the toxin strongly interacted with immobilized fetuin. Avidin and porcine stomach mucin columns were also good matrix to fix the toxin. These results suggest that differences in the spatial conformation of a glycan chain in glycoprotein solution may significantly differ from that of immobilized glycoproteins. The high flexibility of these structures in glycoproteins may interfere on the recognition process. In fact, conflicting results concerning lectin-receptor interaction was previously discussed (ME Etzler 1994 Glycocon-jugate J 11: 395-399). Although the direct interaction of the toxin with either matrix (agarose or Sepharose) could occur, in our extensive work with different carbohydrate-binding proteins, it has never been observed that a protein interacts with the matrix in the presence of an attached ligand. A typical example of this, is the galactose-specific lectin from Artocarpus incisa seeds which binds strongly to agarose-galactose column but not to agarose-fetuin or Sepharose-mucin (MV Ramos et al. 1998b Physiol Mol Biol Plants 4: 157-163). Furthermore, when a macromolecule (glycoprotein) is attached, the accessibility of a protein (the toxin in this case) to the main carbohydrate-chain (oligosaccharide) composing the matrix is highly reduced.

Although the haemagglutinating activity is not a powerful technique to define minor differences in affinity between molecular partners, the recognition of complex N-linked antennary glycan chain and O-linked shorter carbohydrate moiety (as in glycophorin) covalently attached to different glycoproteins, strongly suggests that the receptors for plant RIPs may be widespread in nature. Firstly, this toxin, and others previously studied, all interact with N-linked and O-linked glycan structures. Secondly, it is well established that these glycan structures occur on a large number of cell surface glycoproteins. Therefore, we can suggest that these structures may be present in different cells surfaces. Accordingly, RIPs type II usually invade all Eukariotic cells disregarding their origin.

Copyright 1999 Fundacao Oswaldo Cruz - Fiocruz

The following images related to this document are available:

Photo images