Properties of Volkensin , a Toxic Lectin from Adenia voZkensii *

Volkensin, a highly toxic protein from the roots of Adenia volkensii (kilyambiti, kinoria), was purified by affinity chromatography on acid-treated Sepharose 6B. The toxin is a glycoprotein (M, 62,000, neutral sugar content 5.74%) consisting of an A subunit (M, 29,000) and of a B subunit (Mr 36,000) linked by disulfide and noncovalent bond(s). The amino acid, amino sugar, and neutral sugar composition of the protein were determined. Volkensin is a galactosespecific lectin and is a potent inhibitor of eukaryotic protein synthesis in whole cells as well as in a cell-free system (a rabbit reticulocyte lysate). The inhibitory and the lectin activities are functions of the A and B subunits, respectively. Volkensin  can  be  included amongst he ricin-like toxins and  resembles  most closely modeccin, the toxin of Adenia digitata.

determination were obtained from the same sources as in previous work (Barbieri et d., 1983). N a 9 was obtained from Amersham International, Amersham, Bucks, United Kingdom. Iodogen was purchased from Pierce Eurochemie, Beijerland, Holland.
Cells-The Vero cells, HeLa S3 cells, and baby hamster kidney cells used have been cultivated at the Norsk Hydro's Institute for Cancer Research, Oslo, for years (see . The modeccin-resistant variant was isolated from mutagenized Vero cells by growth in medium containing modeccin. ' Purifiation of Volkensin-The toxin was extracted from A. uolkensii roots, precipitated with ammonium sulfate, and purified by affinity chromatography on acid-treated Sepharose 6B (Ersson et al., 1973) as described previously (Barbieri et al., 1984) except that dialysis of the redissolved ammonium sulfate precipitate was replaced by gel filtration on a Sephadex G-25 column. This procedure was more rapid and removed some colored material present in the extract.
Volkensin was also isolated from A. volkensii seeds that were ground with an Ultra-Turrax apparatus (Jankel & Kunkel, Staufen, West Germany) with 0.14 M NaCl containing 5 mM sodium phosphate buffer, pH 7.2 (8 ml/g of seeds). After overnight stirring at 4 "C, the extract was centrifuged at 20,000 X g for 30 min, and the clear supernatant was applied to a column of acid-treated Sepharose 6B which, after washing, was eluted with 0.2 M galactose in phosphatebuffered NaCl solution.
Throughout all operations, the greatest care was taken to avoid any contact of the body with the roots or with the extract at any stage. This was not only to avoid poisoning, but also because we observed that one person involved in the work developed an allergy to this material.
Radioiodinution-Volkensin was labeled with '%I essentially by the method of Fraker and Speck (1978) as described by Olsnes et ~l . (1982), except that borate buffer was replaced by a universal buffer NaOH).

Properties of Volknsin
M, values were estimated also by gel filtration through a column (80 cm X 1.6 cm) of Bio-Gel P-150, equilibrated with 0.3 M NaCl containing 5 mM sodium phosphate buffer, pH 7.2, eluted at the rate of 12 ml/h, at 4 "C. The column was calibrated with the following markers; chymotrypsinogen (M, 25,000), ovalbumin, bovine serum albumin (all from Pharmacia Fine Chemicals, Uppsala, Sweden), and glucose 6-P dehydrogenase (104,000) (from Boehringer Mannheim GmbH, Mannheim, West Germany). The isoelectric point and the amino acid, amino sugar, and neutral sugar composition of the toxin were determined as described by Falasca et al. (1982).
Protein was determined by the method of Lowry et al. (1951) with bovine serum albumin as a standard or spectrophotometrically (Kalb and Bernlohr, 1977).
Cell Culture-Cells were propagated as monolayer cultures in minimum essential medium containing 10% fetal calf serum. The day before the experiments, the cells were seeded into 24-well disposable trays as earlier described (Sandvig and Olsnes, 1982a).
Protein Synthesis-This was measured in a lysate of rabbit reticulocytes or in intact cells from the incorporation of ~-[~'C]leucine or ~-[~H]leucine, respectively, as described by Sargiacomo et al. (1983) or by Sandvig and Olsnes (1982a). Details are given in the legends to the appropriate figures and tables. The ID50 was calculated by linear regression analysis. Poly(U)-directed polyphenylalanine synthesis by ESP' :richia coli ribosomes was measured as described by GrisB-Miron e' al. (1981).
Hemagglutinating Activity-Blood was collected and the erythrocytes were separated and washed as described previously (Barbieri et al., 1983) and were used either fresh or trypsinized (Lis and Sharon, 1972) and fixed with glutaraldehyde (Turner and Liener, 1975). Hemagglutinating activity was determined in Greiner microtiter plates. Each well contained, in a final volume of 100 pl, 50 pl of a 1% erythrocyte suspension and serial dilutions of the toxin and, when appropriate, of the sugars tested for inhibitory activity.
Toxicity Experiments-The toxicity of volkensin was evaluated as described by Barbieri et al. (1984) using male Swiss mice weighing 20 g and male Sprague-Dawley rats weighing 200 g, receiving food and water ad libitum. LDbo values at 48 h and at 14 days were calculated by the method of Spearman Kaerber as described by Finney (1964).

RESULTS
Chemical Properties of Volkensin-Volkensin was isolated from the roots of A. volkensii by affinity chromatography on acid-treated Sepharose 6B, a procedure commonly used to purify other galactose-specific lectins. The results of a typical preparation are given in Table I. The purified toxin inhibited protein synthesis and agglutinated erythrocytes (see below). After gel filtration on Sephacryl S-200, both activities emerged together in a sharp protein peak (Fig. 1). A single protein peak with apparent M, 62,000 was obtained also by gel filtration on Bio-Gel P-150 (results not shown). Volkensin was also isolated from the seeds of A. volkensii by the same procedure.
When volkensin was analyzed by electrophoresis on cellulose acetate under nondenaturing conditions at pH 4.5 and 9.25, a single major band and a faint minor band were observed. The minor band had a higher mobility at pH 4.5 and a lower mobility at pH 9.25 than the major band (results not shown). On SDS-polyacrylamide gel electrophoresis, volkensin migrated corresponding to a M, 62,000, and after treatment with 2-mercaptoethanol, two major bands with mobility corresponding to M, 36,000 and 29,000 and a fainter heavy band with apparent M, 37,400 were seen (see Fig. 5

below).
On isoelectric focusing, volkensin showed two intense bands at pH 8.2 and 7.8 and three faint bands at pH 8.5, 8.1, and 7.5 (results not shown).
The amino acid, amino sugar, and neutral sugar contents of volkensin are given in Table 11, together with the corresponding values for modeccin. It is clear that the composition of the two toxins is very similar, the main differences being that volkensin contains more half-cystine residues and more than twice as much sugar as modeccin due to a higher content of mannose and galactose. It should be considered, however, that the galactose detected could represent traces of that used to elute the toxins.
The similarities between volkensin and modeccin prompted us to examine whether the two toxins were also immunologically related. When analyzed by immunodiffusion in agarose, volkensin reacted with anti-modeccin serum, giving rise to a single precipitation line which formed a spur with that of modeccin ( Fig. 2) as evidence of partial immunological identity.
Inhibition of Protein Synthesis: Cell-free System-Volkensin inhibited protein synthesis by a rabbit reticulocyte lysate, with an ID5,, of 5 pg/ml (8.4 X lo-' M). The inhibitory effect was abolished when volkensin was heated at 70 "C for 20 min, whereas it was greatly enhanced when the toxin was reduced with 2-mercaptoethanol (IDm = 22 ng/ml (3.7 X loT1' M)).
The concentration of ribosomes in the assay mixture was 5 X M, as determined from the ribosomes recovered by centrifugation of the lysate and measured as described by Stirpe et al. (1981). Assuming a complete recovery of completely active ribosomes, intact and reduced volkensin inactivated ribosomes at a rate of 3 and 357/min, respectively.

TABLE I
Purification of volkensin The purification procedure is described in the text. Protein synthesis was measured with reaction mixtures containing, in a final volume of 62.5 pl, 10 mM Tris/HCl buffer, pH 7.4, 100 mM ammonium acetate, 2 mM magnesium acetate, 1 mM ATP, 0.2 mM GTP, 15 mM phosphocreatine, 3 pg of creatine kinase, 0.05 mM amino acids (minus leucine), 0.19 FCi of ~-['~C]leucine, and 25 pl of a rabbit reticulocyte lysate prepared as described by Allen and Schweet (1962). Incubation was at 28 "C for 5 min. The reaction was arrested with 1 ml of 0.1 M KOH, and the radioactivity incorporated into protein was measured as described by Gasperi-Campani et al. (1978). One unit of activity is defined as the amount giving 50% inhibition of protein synthesis in 1 ml. Hemagglutinating activity was estimated as described in the text. One unit of activity is defined as the minimal amount giving visible agglutination of trypsinized rabbit erythrocytes fixed with glutaraldehyde, in 1 ml of the assay mixture. Results refer to a preparation from 100 g of roots (slightly air-dried, corresponding to 200 g, approximately, of fresh roots).  were determined in the legend to Table I.

Inhibition of protein synthesis Hemagglutinating activity
The toxin at a concentration up to 120 pg/ml did not affect polyphenylalanine polymerization by E. coli ribosomes (results not shown).
Cells-Volkensin also strongly inhibited protein synthesis by various animal cells (Fig. 3A). In this case, the inhibitory activity was decreased when the toxin was reduced. Thus, the ID50 values for Vero cells were 18 pg/ml (3 X M) and 1.5 ng/ml (2.5 x lo-" M) for the native and the reduced toxin, respectively. Even with high concentrations of the toxin, protein synthesis started to decline only after a certain time lag (Fig. 3B). Volkensin inhibited protein synthesis at concentrations 10 times lower than those required for modeccin. Anti-modeccin serum protected cells also against volkensin (Fig. 3C), and this was also the case with NH&l or monensin ( Fig. 30), which both protect cells against modeccin (Sandvig and Olsnes, 1982b). Cell mutants resistant to modeccin were equally resistant to volkensin (Fig. 3E). It is clear that volkensin and modeccin act on cells in similar, but not identical, manners.
Hemagglutinating Activity-Volkensin agglutinated human erythrocytes without any specificity for any particular blood group and the erythrocytes of any animal species examined except sheep (Table 111). The latter were agglutinated by high concentrations of volkensin if they were pretreated with trypsin, a treatment which greatly enhanced the agglutination of the erythrocytes of all species except horse. In all cases except cat erythrocytes, the hemagglutinating activity of volkensin was consistently twice that of modeccin. In the present experiments, the latter had an activity somewhat higher than that previously observed (Gasperi-Campani et al., 1979).
The effect of various sugars on the hemagglutinating activity of volkensin is shown in Table IV. Only D-galactose and sugars with the same configuration inhibited hemagglutination. The inhibitory sugars had an identical effect on the agglutination of human erythrocytes (group 0, Rh+) by volkensin and modeccin (results not shown).
Separation of Subunits-On SDS-polyacrylamide gel elec-  (Barbieri et al., 1980), respectively, were assumed for volkensin and modeccin. a Determined as cysteic acid and methionine sulfone (Moore, 1963). * Hydrolysis with p-toluenesulfonic acid (Allen and Neuberger, 1975  trophoresis, volkensin appears constituted by two subunits (see above). On the assumption that they were analogous to the A and B subunits of ricin (Olsnes and Pihl, 1973), their separation was attempted by exploiting the galactose-binding

FIG. 3. Effect of volkensin on protein synthesis by cells. Volkensin and modeccin were added to cells
growing in 24-well microtiter plates (5 X lo4 cells/well) in minimal essential medium with 10% fetal calf serum.
After 18 h at 37 "C, the medium was removed, and leucine-free medium containing 21 mM Hepes buffer, pH 7.7, instead of bicarbonate, 2 pCi/ml ~-[~H]leucine, and no serum was added. After incubation at 37 "C in an atmosphere of air for the time period indicated, the cells were washed twice with 5% (w/v) trichloroacetic acid and dissolved with 0.2 ml of 1 M KOH, and the radioactivity incorporated was measured as described by Sandvig and Olsnes (1982a). capacity of the B subunit. When the reduced toxin was applied to a column of acid-treated Sepharose 6B, only a small amount of protein emerged unretained, with electrophoretic mobility corresponding to that of the smaller subunit. The same protein inhibited cell-free protein synthesis and did not agglutinate erythrocytes (results not shown). The bulk of the reduced toxin applied to the column was retained, could be eluted with galactose, and showed the presence of both subunits (results not shown). When the reduced toxin was applied to a column of untreated Sepharose 4B, two protein peaks were not retained, and a third one was eluted with galactose ( Fig. 4). On SDS-polyacrylamide gel electrophoresis, peak 1 showed a single band, corresponding to the smaller subunit, peak 2 contained both subunits, the heavier being much more abundant, and peak 3 contained the heavier subunit only (Fig. 5).
The lighter subunit had the properties of an A chain and the heavier subunit those of a B chain, as assessed from their effects on protein synthesis and from their hemagglutinating activity (Table V). (The slight inhibitory activity on protein synthesis exerted by both isolated subunits and the slight hemagglutinating activity of the lighter subunit are probably due to contaminating traces of the intact toxin.) The inhibitory activity of the A subunit on cell-free protein synthesis was less than that of the reduced toxin. This was probably due to partial inactivation of the A subunit after separation. Indeed, the inhibitory activity of the A subunit solution was almost completely lost after storage for 2 weeks at 4 "C under sterile conditions in the presence of 1% 2-mercaptoethanol and was not restored by the addition of the B chain (results not shown). From these results it seems that the two subunits were still linked together after the reduction of volkensin, a finding apparently not consistent with the reduced effect on protein synthesis by cells. A difference between the chromatographies and the experiments with cells was that in the former case the toxin was at a higher concentration and at a lower temperature, which could favor the association of the subunits by forces other than disulfide bond(s). That this was the case was shown by experiments in which reduced and nonreduced volkensin labeled with lZ5I was added to cells. On SDS-

TABLE IV
Inhibition of kmagglutinuting activity of volkensin by sugars Hemagglutination was determined as described in the legend to Table I. Volkensin was added at a final concentration of 2 pg/ml. The following sugars did not affect hemagglutination: all, a all, D-Alt, D-Ara, 2-deoxy-2-amino-~-Gal, L-Gal, D-G~c, L-G~c, D-Gu~, D-Ido, L-Rib, L-Xyl (aU at 100 mM concentration), stachyose (at 10 mM concentration), and p-nitrophenyl-a-D-galactopyranoside, p-nitrophenyl-P-D-galactopyranoside (at 1 mM concentration). polyacrylamide gel electrophoresis, both subunits were visible in the Triton X-100 extract of the cells incubated with the native toxin, whereas the heavier subunit and traces only of the lighter one were present in the cells incubated with reduced volkensin (Fig. 6). This indicates that at low concentration and high temperature the subunits are separated after reduction of the toxin and corhrms that the larger subunit only binds to cells, thus corresponding to the B chain of the other toxins. Toxicity to Animak-Animals poisoned with volkensin always died not less than 7 h after administration. Those receiving low doses died after several days. Therefore, an acute and a delayed LDW were calculated. The acute LD50 at Tris/HCl buffer, pH 8.5, containing 2% (v/v) 2-mercaptoethanol, was incubated at 37 "C for 2 h. Some precipitate formed during the incubation was removed by centrifugation, and the clear solution was applied to a Sepharose 4B column (10 cm X 0.7 cm) at 4 "C previously equilibrated with buffer/mercaptoethanol. After washing, the column was eluted with 0.2 M galactose in the same solution (arrow). Fractions of 20 drops were collected.

Minimal
48 h was 0.32 pg/kg of body weight (95% confidence limits 0.28-0.37) and 1.73 pg/kg (0.79-3.79) for rats and mice, respectively. The delayed LDm estimated at 14 days was 0.061 pg/Kg (0.045-0.083) for rats and 1.38 pg/kg (0.56-3.36) for mice. Thus, volkensin proved to be a very potent toxin, particularly for rats, for which the LDm was approximately 10-fold lower than that of modeccin (Gasperi-Campani et d., 1978). Rats poisoned with higher doses died between 7 and 12 h after administration. They appeared normal for 1-2 h after poisoning and then they became progressively sedated until death, which occurred suddenly, preceded by shortlasting seizures. The rats which received lower doses and died after several days often showed ascites and wax-like peritoneal fat, a sign of pancreatic lesion.

DISCUSSION
Our results demonstrate that volkensin is a toxic protein which inhibits eukaryotic protein synthesis both in celIs and in cell-free systems, without affecting protein synthesis by E. coli ribosomes. The toxin is a galactose-specific lectin since it 1) binds to Sepharose and can be eluted with galactose, and 2) has hemagglutinating activity which can be inhibited by galactose and structurally related sugars.
Like ricin and related toxins, volkensin consists of an A and a B subunit, with protein-inhibitory and sugar-binding capacity, respectively. These subunits are disulfide-linked and presumably held together by noncovalent bonds, since they cannot be easily separated after reduction with 2-mercaptoethanol, as was the case also with modeccin (Barbieri et al., 1980). The hemagglutinating activity suggests that either the B subunit has two sugar-binding sites, as reported for ricin (Villafranca and Robertus, 1981), or that the toxin may form multivalent aggregates, as observed with viscumin .
Volkensin appears to be similar to the other plant toxins which inhibit protein synthesis, and especially to modeccin. The latter finding is not surprising, since the two toxins are present in taxonomically related plants, both being members of the Passifloraceae. Thus, the two toxins are immunologi-

TABLE V
Properties of the isolated subunits of volkensin Protein synthesis by cells was assayed with AKR-a cells incubated in the presence of the toxin (or of its subunits) for 24 h, followed by a further 24 h with 1 pCi of [3H]leucine, with the toxin still present. Other details were as described by Thorpe et al. (1984). The assays with the lysate system were as described in the legend to Table I, except that the lysate was prepared as described by Pelham and Jackson (1976). Hemagglutinating activity was determined as described in the legend to Table I  cally cross-reacting, as are other lectins from plants belonging to the same family (Leguminosae (Howard et al., 1979) and Solanaceae (Kilpatrick et al., 1980)). The homology, although not identity, in their immunologic behavior is consistent with their similar, but not identical, amino acid composition. On isoelectrophocusing, both volkensin and modeccin show several bands, presumably iso-forms, although those of modeccin have lower PI values (Olsnes et al., 1978). They also show identical sugar specificity, and the experiments with modeccin-resistant cell mutants suggest that they bind to the same receptors on the cell membrane. The fact that NH&l and monensin protect against both toxins suggests that volkensin and modeccin are taken up by cells by the same mechanism.
On the other hand, volkensin 1) has a much higher sugar content than modeccin, and 2) appears to be a more potent inhibitor than modeccin of protein synthesis in whole cells and, after reduction, it also inhibits protein synthesis by a reticulocyte lysate more strongly than modeccin. Furthermore, volkensin (3) is more toxic to animals and (4) has a higher hemagglutinating activity than modeccin.
The toxicity of volkensin to animals and particularly to rats is very high, and volkensin is in fact one of the most potent toxins of plant origin (see review by . This and the lesions observed in poisoned animals suggest that the toxicity of A. volkemii roots may be due to volkensin rather than to the cyanogenetic glycoside they contain. The latter may contribute to the toxicity, especially if large amounts of roots are ingested, but cannot account for the delayed deaths and for the lesions observed in poisoned rats (Kamau, 1975).