Toxicon, Vol. 19, No. 5, pp. 667-675, 1981 . 0041-0101/81/050667-08 502.

00/0
Printed in Greal Britain. © 1981 Pergamon Press Ltd.

ISOLATION AND CHARACTERIZATION OF A TOXIC

PROTEIN FROM CANAVALIA ENSIFORMIS (JACK BEAN)
SEEDS, DISTINCT FROM CONCANAVALIN A*

CELIA R. CARLINIt and JORGE A. GUIMARÂES$

Department of Physiology, Universidade Federal Fluminense, Caixa Postal 183, 24210, Niteroi,
Rio de Janeiro, Brasil

(Accepted for publication 26 March 1981)

CELIA R. CARLINI änd IORC)E A . GUIMARÀFS . Isolation and characterization of a toxic protein from
Canavalia ensijormis (jack bean) seeds, distinct from concanavalin A. Toxicon 19, 667-675, 1981 . A
toxic protein present in the crude extract of Canavalia ensiformis seeds induces within 24 hr
dyspncea, ataxia, hypothermia, coma, tonic convulsions and death in mice injected i.p. with
100-200 mg of protein per kg. The toxin was separated from concanavalin A by affinity adsorption of
the latter on Sephadex G-100, and further purified by sequential fractionation by (a) removal of
polysaccharides with 30'ßo v/v ethanol ; (b) ammonium sulfate precipitation at 0.35-0.55 saturation
and (c) DEAE-cellulose chromatography . The purified toxin, with an LD SO of 2-5 mg of protein/kg
mouse, is unstable . Gel-filtration of the purified toxin on Bio-Gel P-200 showed a single protein peak
with a molecular weight corresponding to 88,000 daltons. Polyacrylamide gel electrophoresis of this
material indicated the presence of two small contaminants . A central convulsive effect of the toxin can
be proposed since no effects were found on isolated pharmacological preparations . The name
CANATOXIN is suggested for this new toxic protein .

INTRODUCTION
THE TOXICITY to animals ofthe extract from jack beans (Canavalia ensiformis' seeds) has been
recognized for many years (ASSMANN, 1911). This toxicity was attributed by SUMNER and
HOWELL (1936) to concanavalin A (Con A) since their purified lectin preparation injected
into rabbits caused toxic ef%cts, considered a consequence of the hemagglutinating action of
Con A. It is known, however, that the Con A preparation obtained by SUMNER and HOWELL'S
method was inhomogeneous (OLSON arid LIENER, 1967), thus the toxicity of their Con A
could be due to some protein contaminant.
An association between toxicity and the presence of lectins has been suggested for other
plant seeds : Ricinus communis and A6rus precatorius (cf. SHARON and Lls, 1972), Glicina max
(LIENER arid PALLANSCH, 1952 ; SAMBETH et al., 1967) and Phaseolus vulgaris ( .ÎAFFÉ, 1960 ;
HAMAGUCHI et a1.,1977) . In a few cases the lectin activity and the toxicity have been separated
(TAKAHASHI et al., 1962 ; OLSNES and PIHL,1973 ; OLSNES et al ., 1974 ; STEAD et al., 1966 ; LIN
et al., 1978).

* Work subsidized by Conselho Nacional de Pesquisas, Brazil. A summary of this work was presented at the V'h
Symposium on Brazilian Medicinal Plants, Säo Paulo, SP, September 1978 .
t Fellowship from Fundaçäo de Amparo à Pesquisa do Estado de Säo Paulo (FAPESP), Säo Paulo, SP and
$ Conselho Nacional de Pesquisas (CNPq), Brazil.

667
668 CELIA R . CARLINI and JORGE A. GUIMARÂES

Concerning the different biological activities of Con A preparations, SHARON and LIS
(1972) considered it difficult to conclude whether these activities reside in only one molecular
species. No data is available describing the separation ofthese principles . The present report
describes the partial characterization and isolation from Canavalia ensiformis seeds of a toxic
protein distinct from concanavalin A.
MATERIALS AND METHODS
Adult female Swiss mice (25-35 g) and male weanling Wistar rats (100-200 g) were used . Mature Canavalia
ensiformis seeds were a kind gift from Dr . E . D . de Lucas, Ministério da Agricultura, Brazil. The chemical reagents
were obtained commercially : concanavalin A, shell fish glycogen, chicken egg albumin, bovine serum albumin,
ammonium sulfate (Sigma Chemical Company, USA) ; Tris (tri-hydroxilmethylaminomethane), Comassie Brilliant
Blue G-250, Bio-Gel P-200 and chemicals for polyacrylamide gel electrophoresis (Bio-Rad Laboratories Inc .,
U.S.A .) ; Sephadex G-100 (Pharmacia Fine Chemicals, Sweden) ; DEAE-cellulose (Whatman, U.S.A .) ; native
human hemoglobin was a gift from Dr. J . D . Macarini, Dept . Physiology, UFF, Niteroi, Brazil . Other reagents were
of analytical grade .
Protein determination of preparations and fractions was calculated either by absorbance at 280 nm (AZeo) in an
1 cm cuvette (one A Zeo unit is equivalent to ca. 1 mg of purified protein/ml) or using Comassie Brilliant Blue G-250
as a specific dye, according to the method of SPecroR (1978).
Polyacrylamide Disc Gel Electrophoresis was carried out by the method of Dnvis (1964) . Aliquots containing
50-250 ug of protein were applied to 7% polyacrylamide gel tubes (0.6 x 9 .5 cm) . After running the samples for 4 hr
at 4 mA/tube, in 2 mM Tris and 30 mM glycine buffer (pH 8 .9), the gels were fixed and stained with 0.1 ~ amido black
in 7~ acetic acid . Destaining was performed electrophoretically.
Lethality was estimated by calculating the I,D SO of toxic preparations and fractions injected i.p . The ~~~SO,
expressed in AZeo units of protein per kg of mouse body weight, was calculated by the method of LITCHFIELD and
Wttcoxotr (1949) using 6-8 animals per dose, causing zero to 100 death in 24 hr. When only small quantities of
toxin were available, the minimum lethal dose (MLD, AZao protein unit/kg) was used as an index of toxic activity . In
this case, 3 doses of toxin (2-4 animals/dose) were tested and the smallest dose causing death was considered the
MLD .
Concanavalin A was assayed by a slight modification of the method described by PoRezT and Go~usTe~x (1968).
An aliquot of the sample was mixed with 501 of 1 M sodium phosphate and 3 M NaCI buffer (pH 7 .0) and the
volume made up to 980 PI with distilled water . After addition of 20 ~1 of an oyster glycogen solution (20 mg/ml), the
test tube was immediately agitated in a Vortex apparatus. The absorbance at 420 nm was read after 60 min of
incubation at room temperature and compared with a standard curve of Con A (10-150 Ng) assayed in the same
conditions.
Molecular weight determination was made by the method Of ANDREWS (1964). The elution volume of the toxin
obtained by gel-filtration on a Bio-Gel P-200 column was plotted together with those obtained for standard proteins
gel-filtered in the same conditions.
Conductivity was measured at 25°C with a Radiometer conductometer and expressed as ~ or mSiemens. Salt
concentrations were estimated from conductivity measurements using standard curves .

RESULTS
Lethality and chemical nature of the toxin
A crude extract (Prep A, see below) prepared from mature Canavalia ensiformis' seeds,
when injected i.p. in mice causes dyspnoea, ataxia, hypothermia, coma, tonic convulsions and
death, which occurs between 0.5 and 24 hr (the time between injection and death is dose-
dependent). When compared with the i.p. injection, the toxinis equally active either by i.m. or
s.c. administration, however, i.v. injection elicited convulsions in a time 6 to 20-fold shorter.
On a weight basis, rats were about 5-fold more susceptible than mice. The toxin was also
lethal and induced seizure in animals under anaesthesia (urethane or sodium pentobarbital)
or artificial respiration. No signs of pathological alterations could be found upon
macroscopic examination of organs and tissues.
The toxic protein does not contain cyanide residues, is not dialysable, is heat-labile, soluble
in water and diluted salt solutions and unstable below pH 6.0.

Separationfrom Concanavalin A (Con A)

In order to exclude the possibility that the toxic activity exhibited by C. ensiformis' extracts
 A Toxic Protein from Canavalia ensiformis 669

could be due to Con A, as previously suggested (SUMMER and HOWELL, 1936 ; SHnltorr and
Lls, 1972 ; Bltowty and Hu1vT, 1978), this lectin was removed from the crude preparation by
affinity chromatography on Sephadex G-100 according to the method described by OLSON
and LIENER (1967).
Figure 1 shows that ca . 90~ of the total protein was not retained by the gel, while Con A
was eluted in a single fraction with 0.02 M glycine-HCL buffer, pH 2.0. The unretained
material recovered as peak A contained most of the toxin, while the Con A fraction (peak C)
was completely atoxic. Furthermore, commercial Con A given i.p . to mice (up to 200 mg/kg
of body weight) did not reproduce any of the effects observed with the toxin. This dose of Con
A is about 5 times the level of the lectin present in the crude extract of C. ensiformis (SUIv1NER
and HowELL,1936) . The data clearly indicate that the toxin and Con A are different proteins.
The name canatoxin is tentatively given to this new toxic principle.

Purification of canatoxin
The following steps, performed at 4°C, give high reproducibility (Table 1), either using
small (10 g) or large (500 g) quantities of seeds.
Step A. Crude extract-the seeds were swollen for 4 hr in 25 mM Tris-HCl buffer, pH 7.5,
in the proportion of 1 g seeds/10 ml buffer . After removal of shells, the seeds were
homogenized in the Waring blender containing fresh, ice-cold buffer solution in a proportion
to give a 10% w/v final extract. The homogenates were filtered through cheese cloth and the
filtrate's pH immediately adjusted from ca . 6.0 to 7.5 with 5 N NaOH . Aliquots of the
resultant suspensions, called Prep A, were dialysed for protein determination and toxicity
assay.

..
é
0m
N
a
2
W
H

lY
a

VOLUME,mI

FIG. 1. SEPARATION OF CANATOXIN FROM CON A.

Column :1 .2 x 24.0 cm, packed with Sephadex G-100, equilibrated in 10 mM Tris-CIbuffer, pH 7.4 ;
Sample : 10.0 ml of crude seed extract prepared according to OL.soN and LIENER (1967), total protein
content ca. 215 A28o units. Elution : peaks A and B, equilibrating buffer ; peak C, lowpH buffer . 2.0 ml
fractions were collected over 150~I of 1 M Tris-Cl buffer, pH 8.0. Flow rate of 6 .0 ml/hr. The symbol
(") indicates protein exhibiting toxic activity .
670 CELIA R . CARLINI and JORGE A . GUIMARÂES

TABLE 1. CANATOXIN PURIFICATION

Lusot Azao/g seed Yield

Step* n x f S.D . x±S .D . Purification °j

A 7 122 ± 30 340 ± 26 1 .0 100

B 7 25±6 27±3 4.8 38
C 9 6±1 5 .7±0.7 19 .9 33
D 3 4±1 .5 1 .7±0.4 30 .5 16

* See text for description of steps A, B, C and D.

t LD Sa given in A zBO units of protein/kg mouse body weight .

Step B. Ethanol treatment-in order to clarify the suspensions from starch-like substances,
ice-cold ethanol (96%) was slowly added to Prep A samples, making a final alcohol
concentration of 30~ v/v . The mixtures were maintained under stirring during addition of
ethanol and immediately centrifuged at 8000 g for 15 min. The precipitates were discarded
and the clear yellowish supernatants were dialysed against the extracting buffer . The
dialysed, alcohol-free solutions were designated Prep B.
Step C. Ammonium sulfate fractionation-to Prep B, maintained under stirring, solid
ammonium sulfate was slowly added to give 0.35 saturation . The precipitates formed were
separated by centrifugation at 8000 g for 15 min and discarded . To the supernatants, more
salt was added to make 0 .55 saturation and the precipitates, collected by centrifugation, were
dissolved in ca. 20 ml of25 mM Tris-Cl buffer, pH 7.5. The samples were dialysed against the
same buffer until complete removal of the salt. The resultant solutions were called Prep C.
Step D. DEAE-cellulose chromatography-as illustrated in Fig . 2, a sample of Prep C,
containing 5 mg protein/ml, was chromatographed in a DEAE-cellulose column . The
protein content as well as the toxicity of the fractions were determined . Unretained material
accounts for 66~ of the total protein and was completely devoid oftoxic activity . The active
toxin was eluted between 0.12 and 0.17 M NaCI in a peak of about 10~ of the total protein,
with an LD so ranging from 2 to 5 mg of protein/kg.
Since the Unretained protein could be separated from the adsorbed toxin and since the
elution of the active toxin could be achieved in a sharp salt gradient, the DEAE-cellulose
chromatography on columns could be replaced by a faster procedure using a batch
adsorption step. In these cases, the samples, prepared as mentioned, were stirred with the
resin, in the given proportion, for 2 hr. The mixtures were then filtered through a fritted-glass
Buchner funnel, and the resin washed similarly as in the column method until complete
removal of non-adsorbed proteins (65-75%). The retained proteins (Prep D) were then
eluted in a step-wise procedure with the equilibrating buffer containing 0.30 M NaCI
(conductivity, 20 mSiemens).
Step E. Gel-filtration on Bio-Gel P-200-an aliquot (84 mg) of Prep D was filtered on a
Bio-Gel P-200 column according to the conditions described in Fig. 3. Although a sharp
protein peak containing the toxic activity has been obtained, polyacrylamide disc gel
electrophoresis showed (Fig. 3, insert) that small amounts of contaminants were still present
on the main active fraction (n' 44) . Lethality determination on the gel-filtered material could
not be done due to instability (see below) ofthe toxin . Thus, purification data at this step were
not included in Table 1 .
There was a parallelism between the symptoms observed in the animals injected with the
crude extract and purified preparations at all steps. When tested in doses up to 20-fold that of
 A Toxic Protein from Canavalia ensiformis 67 1

FIG . 3. GEL-FILTRATION OF PREP D ON BIO-GEL P-20O.

Column : 2.0 x 100.0 cm, packed with Bio-Gel P-200, equilibrated in 25 mM Tris-HCl buffer, pH 7.5 .
Sample : 4.0 ml of Prep D containing 84 mg of protein was dissolved in the equilibrating buffer. This
material was introduced in the column immediately after 2 ml of 1 M NaCI in the same butler.
Elution : a flow rate of 10.5 ml equilibrating buffer/hr was maintained constant with a peristaltic
pump. Fractions of 3.0 ml were collected. INSERT : 7% polyacrylamide disc gel electrophoresis of
Prep D (254 pg, left) and fraction n' 44 (75 Kg, right) obtained from the Bio-Gel P-200 column
described above. The symbol (*) indicates protein exhibiting toxic activity .
 A Toxic Protein from Canavalia ensiformis 67 3

û
0
u
~x
O

FIG. 2. DEAF-CELLULOSE CHROMATOGRAPHY .

Column : 1.2 x 30.Ocm, packed with DEAE-cellulose and equilibrated in 25mM Tris-CI and
100 mM NaCI buffer, pH 7.7., conductivity 7.5 mSiemens. Sample : 90 ml (515 mg protein) of Prep C
was diluted in the equilibrating buffer to 103 ml and introduced in the column . The column was
washed, with 10 timesits own volume, with theequilibrating buffer for removal of unretained proteins
(66%). Elution: (a) NaCI linear gradient (0.1-0 .3 M) in Tris-buffer; (b) 0 .5 M NaCI to remove the
proteins still bound to the resin. The toxic fractions, representing 10% of the total protein introduced
into the column,were pooled (toxic activity expressed as reciprocal of the minimum lethal dose, MLD).

the LD SO, neither the crude nor the purified toxin had any effect on the following isolated
pharmacological preparations : isolated atrium, phrenic nerve-diaphragm or jejunum (rat),
vas deferens (guinea-pig), rectus abominis muscle (toad). Furthermore, it was observed that
canatoxin does not cause hemolysis. It was also observed that rats injected i.p . or i.v. with the
toxin (Prep C) showed typical electroencephalographic alterations, with epileptic-like
activity registered on cerebral neocortex and dorsal hippocampus with local bipolar
electrodes (CARLINI et al., unpublished results).

Molecular weight determination

A sample of Prep C was gel-filtered on a Bio-Gel P-200 column using the same conditions
described for Fig. 3. The elution volumes of the toxin and standard proteins were plotted
(ANnREws,1964). As illustrated in Fig. 4, a molecular weight of 88,000 daltons was estimated
for canatoxin.

Instability of canatoxin
With time, purified canatoxin loses toxicity. The time-course inactivation was inversely
proportional to the state of purification, and the Lv SO for the highest purified toxin
preparation (Step E) could not be determined. At this stage, toxicity was completely lost,
usually in about 24 hr, both when frozen or at 4°C.
 674 CELIA R. CARLINI and JORGE A. GUIMARÂES

20

a0

t 9
. 8
~r
3 7
6
_~o
5
U
N
p 4
E
3:

80 ~ 100 ~ 120 ~ 140

volume, ml

FICi. 4. MOLECULAR WEIGHT DETERMINATION.

The same column andchromatographicconditions described for Fig. 3 were used for gel-filtration of
standard proteins : A, bovine serum albumin dimer, 138,000 dallons ; B, bovine serum albumin
monomer, 69,000 dallons ; C, human hemoglobin, 63,500 dallons ; D, egg albumin, 45,000 dallons and
an aliquot of `canatoxin' (3 ml preparation C containing 148 mg protein). The elution volumes of the
standard proteins and that obtained for canatoxin were plotted according to the procedure of
ANDREWS (1964).

DISCUSSION

The present findings confirmed the existence of a toxic protein in Canavalia ensiformis
seeds, as suggested a long time ago by ASSMANN (1911) and more recently by SHARON and LIs
(1972) and BROWN and HUNT (1978). The toxin was probably a contaminant of the Con A
preparation obtained by SUMNER and IIOWI:LL'S method, since we separated it from the lectin
by an affinity chromatography procedure as described by OLSON and LIENER (1967) .
Furthermore, Con A by itself in doses up to 200 mg/kg was atoxic to mice. This amount of
Con A is equivalent to about 5-fold the lectin content in a crude extract which is able to kill
100 of the experimental animals. Further purification showed that the toxic protein has a
molecular weight of 88,000 daltons. It displays instability, which is probably responsible for
the low yield found in the isolation procedure. Thus, the LD SO estimated for the purified toxin
(up to Step D), which ranged between 2 and 5 mg of protein/kg, may actually be much smaller.
A parallelism ofsymptoms was observed with the crude extract and the purified preparations,
thus indicating that the toxic activity is due to a single principle. Furthermore, dialysis of the
crude or purified toxin did not affect the toxicity .
The toxin induced several symptoms in experimental animals either given i.p. or i.v. ; the
difference being only the time between the injection and the appearance of effects. No
pharmacological actions were found either when crude or purified toxin was tested on
several isolated organ preparations . In addition, no macroscopic alterations of organs or
 A Toxic Protein from Canavalia ensiformis 67 5

tissues could be found in dead animals. This and other effects such as the short time from
injection to death are very different from the actions of ricin and abrin (OLSNES et al., 1974 ;
WALLER et al., 1966), thus excluding a common mechanism for the three toxins . Recently,
Roos et al. (1980) described a cytostatic effect of proteins, different from Con A, obtained
from Canavalia ensiformis extract . Any similarity between canatoxin and these proteins
remains to be shown. Convulsions that precede death with canatoxin appear to be associated
with severe central nervous system alterations, as indicated by hypothermia and elec-
troencephalographic changes (CARLINI et al., unpublished results) . This point raises the
question of how the toxic protein reaches the central nervous system . The mechanism may be
similar to that of tetanus toxin (PRlce et al., 1975) and other proteins (SCHWAB et al., 1979)
which are actively transported toward the central nervous system. It is too early to speculate
on the mechanism of action of canatoxin since it is possible that a fragment rather than the
entire protein is active in inducing toxic effects.
REFERENCES
ANDREWS, P. (1964) Estimation of the molecular weight by gel-filtration . Biochem. J . 91, 222.
ASSMANN, F. (1911) Beiträge zur Kenntnis pflanzlicher Agglutinine. Arch. für die gesamnte Physiol. 137, 489 .
BROWN, J. C . and HuNr, R . C. (1978) Leetins . Int . Reu . Cytol . 52, 277 .
DAV~S, B . J. (1964) Disc electrophoresis . II . Method and application to human serum proteins. Ann . N. Y . Acad . Sci.
121, 404 .
HAMAGUCHI, Y., YAGi, N., NISHINO, A., M(KH~ZUKI, T., MIZUKAMI, T. and MmosHi, M. (1977) The isolation and
characterization of a lethal protein from kintoki beans (Phaseolus vulgaris) . J . Nutr. Sci . Vitaminol. 23, 525.
JAFFÉ, W. G. (1960) Ueber Phytotoxine aus Bohnen (Phaseolus vulgaris). Arzneimittel-Forsch 12, 1012 .
LIENER, I . E. and PALLANSCH, M . J . (1952~Purification of a toxic substance from defatted soy bean flour . J . 6iol.
Chem. 197, 29.
LiN, J. Y., Hou, M . J . and CFIEN, Y. C. (1978) Isolation of toxic and non-toxic lectins from the bitter pear melon
Mormodica charantia Linn . Toxicon 16, 653 .
LITCHFIELD, J. T., JR and W~LCOXON, F. (1949) A simplified method for evaluating dose-effect experiments. J.
pharmacol. exp. Ther. 96, 99 .
OLSON, M . O . J. and LIENER, I. E. (1967) Some physical and chemical properties of concanavalin A, the
phytohemagglutinin of the jack bean . Biochemistry 6, 105 .
OLSNES, S . and PiHL, A . (1973) Isolation and properties of abrin : a toxic protein inhibiting protein synthesis. Eur . J .
Biochem. 35, 179.
OLSNES, S., REFSNES, K . and PIFIL, A. (1974) Mechanism of action of the toxic lectins abrin and ricin . Nature (Lond. )
249, 627 .
PoREZT, R . D . and GoLDSretN, I. J. (1968) A study of turbidity as it relates to concanavalin A-glycan interaction .
Immunology 14, 165.
PRICE, D . L ., GRIFFIN, J ., YOUNG, A., PECK, K. and STOCKS, A. (1975) Tetanus toxin : direct evidence for retrograde
intraaxonal transport . Science, N .Y . 188, 945 .
RODS, O., KONZ, W., DANIEL, H., WALDECK, K. and JENNEWEIN, H . M . (1980) Isolation of cytostatically active
proteins from seeds of Ricinus communis, Abrus precatorius and Canavalia ensiformis. Arzneimittel-
Forschung/Drug Research 30, 759 .
SAMBETH, W., NESHEIM, M. C. and SERAFIN, J. A. (1967) Separation of soybean whey into fractions with different
biological activities for chicks and rats. J . Nutrition 92, 479 .
SCHWA& M. E ., SUDA, K . and THOENEN, H. (1979) Selective retrograde transsynaptic transfer of a protein, tetanus
toxin, subsequent to its retrograde axonal transport . J . Cell Biol. 82, 798.
SHARON, N. and LIS, H. (1972) Lectins : cell-agglutinating and sugar-specific proteins . Science, N .Y. 177, 949.
SPECTOR, T. (1978) Refinement of the Comassie Blue method of protein quantitation. A simple and linear
spectrophotometric assay for 0.5-50 ug of protein. Anal. Biochem. 86, 142.
STEAD, R . H., DE MUELENAERE, :H . J. H. and QutcKE, G. V . (1966) Trypsin inhibition, hemagglutination and
intraperitoneal toxicity in extracts of Phaseolus vulgaris and Glicina max . Arch . Biophys . Biochem. 113, 703 .
SuntNeR, J. B. and HoweLL, S. F. (1936) Identification of the hemagglutinin of the jack bean with concanavalin A . J .
Bacteriol . 32, 227 .
TAKAHASHI, T., FUNATSU, G. and FUNATSU, M . (1962) Biochemical studies on castor bean hemagglutinin . II .
Hemagglutinin separated from crystalline ricin and its molecular weight. J. Biochem. Tokyo 52, 50 .
WALLER, G. R., EBNER, K. E ., SCROGGS, B . R ., DASGUPI'A, B: R. and CORCORAN, J. B. (1966) Studies on the tOX1C
action of ricin . Proc. Soc. exp . Biol. Med. 121, 685 .