Hemorrhagic activity of Bothrops venoms determined

by two different methods and relationship with proteolytic

activity on gelatin and lethality

Adolfo Rafael de Roodt*, Silvana Litwin, Juan Carlos Vidal1

Instituto Nacional de Produccion de Biologicos—A.N.L.I.S, Dr Carlos G. Malbran, Av. Velez Sarsfield 563, CP 1281 Buenos Aires, Argentina

Received 19 September 2002; accepted 9 December 2002

Abstract

The changes in hemorrhagic activity, proteolytic activity on gelatin and the lethal potency of four Bothrops venoms treated at

different pH values or with EDTA were studied. Venoms from B. alternatus, B. jararaca, B. moojeni and B. neuwiedii of

Argentina were preincubated at pH 5.8, 5.1 or 3.8 or with EDTA and the hemorrhagic activity expressed as size of the

hemorrhagic lesion or as the amount of hemoglobin extracted, the proteolytic activity on gelatin and the lethal potency were

determined. Although the MHDs recorded in rats were 19–56 fold higher than those recorded in mice, the A550 extracted per

gram of hemorrhagic haloes was very similar in rats or mice independent of the venom dose. Inhibition of proteolytic activity

after preincubation at pH 5.1 or 3.8, agrees with the decreased amount of hemoglobin extracted from the hemorrhagic haloes,

and with the increase in mean survival time after the i.p. injection to mice. Preincubation with EDTA resulted in 80% inhibition

of hemorrhagic activity of B. jararaca venom and complete inhibition with the other Bothrops venoms tested. Measurement of

the amount of hemoglobin extracted gives significant information in comparative studies, not available by measurement of the

size of hemorrhagic haloes.

q 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Venoms; Hemorrhage; Enzymatic activity; Bothrops; Snakes

1. Introduction

Envenomation by Bothrops bites is characterized by a

complex series of pathological alterations including hemor-

rhages, which perhaps represent one of the most conspic-

uous toxic activities in bothropic envenoming. After

Bothrops bite the hemorrhage may become systemic and

contribute significantly to the lethal potency of these

venoms. Hemorrhages are principally caused by metallo-

proteinases, enzymes that are responsible for degrading

proteins of extracellular matrix, they also have citotoxic

effect on endothelial cells and acts on components of

the haemostatic system (Kamiguti et al., 1996). Some

metalloproteinases induce hemorrhage by directly affecting

mostly capillary blood vessels cleaving in a highly selective

fashion key peptide bonds of basement membrane com-

ponents affecting the interaction between basement mem-

brane and endothelial cells (Gutierrez and Rucavado, 2000).

This hemorrhagic enzymes contains one single Zn2þ ion

per molecule (Bjarnasson and Fox, 1983, 1988, 1994;

Stocker et al., 1995) and have different molecular weights

due to the presence of additional protein domains and

structurally similar catalytic domain (Bjarnasson and Fox,

1994, 1995; Hite et al., 1992, 1994) which governs the

primary proteolytic specificity (Jia et al., 1996). The crystal

structure (Gomis-Ruth et al., 1993; Gutierrez and Rucavado,

2000; Zhang et al., 1994) show that they have a common

methionine turn below and C-terminal to a helical segment

containing two histidines (His 142 and His 146) of the three

histidines residues (His 152) involved in Zn2þ binding site

(Jia et al., 1996). Treatment at low pH values or with

0041-0101/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.

doi:10.1016/S0041-0101(02)00392-6

Toxicon 41 (2003) 949–958

www.elsevier.com/locate/toxicon

1 In Memoriam.

* Corresponding author. Tel.: þ54-11-4303-1807–11x250; fax:

þ54-11-4303-2492.

E-mail address: aderoodt@uolsinectis.com (A.R. de Roodt).

chelating agents produces structural alterations leading to

the irreversible loss of both, proteolytic and hemorrhagic

activities (Sanchez et al., 1995a).

Hemorrhagins are capable of degrading laminin, fibro-

nectin and Type IV collagen (Bjarnasson et al., 1988;

Baramova et al., 1990; Markland, 1998). In addition, these

enzymes hydrolyze fibrinogen, casein, dimethyl casein,

oxidized insulin B-chain (Mandelbaum et al., 1982;

Bjarnasson and Fox, 1994; Civello et al., 1983; Maruyama

et al., 1992; Sanchez et al., 1995 a,b) and gelatin (Bee et al.,

2001; Markland, 1998).

Hemorrhagic activity of snake venoms can be quantified

by the intradermal injection of a venom in an experimental

animal, which produces a readily observable hemorrhagic

halo on the dermic side of the skin, such as in the rabbit skin

test of Kondo et al. (1960), which have been adapted for rats

and mice (Theakston and Reid, 1983; Gutierrez and Chaves,

1980). This activity can be determined as minimal

hemorrhagic dose, that is represented as the amount of

venom which produces an hemorrhagic halo with an average

of the major perpendicular diameters ½ðd1 þ d2Þ=2� set to

1.0 cm (Theakston and Reid, 1983) or an hemorrhagic halo

of 1.0 cm2 (Gutierrez and Chaves, 1980). There were

performed several assays to study the hemorrhagic activity

like the measurement of hemoglobin in muscle inoculated

with venom (Ownby et al., 1984), the measurement of the

hemorrhagic halo in blood vessels that surround and supply

the chick embryo (Sells et al., 1997) or dog lung (Bjarnasson

and Fox, 1994) or the measurement of hemoglobin

contained in the hemorrhagic haloes in skin (Esmeraldino

et al., 1999; de Roodt et al., 2000). Approximate

quantification of the hemorrhagic activity by measurement

of the average diameter of the hemorrhagic haloes obtained

after the intradermical injection of venom (Kondo et al.,

1960; Oshaka et al., 1966; Theakston and Reid, 1983) is

specific, fast and reproducible. However, problems arise

when the hemorrhagic haloes obtained with several venoms

have similar sizes but differ in color intensity, reflecting

differences in hemoglobin content. In order to avoid this

inconvenient, the measurement of the amount of hemo-

globin in a sample of skeletal muscle injected with venom

(Ownby et al., 1984), scanning the color intensity of the

hemorrhagic haloes (Esmeraldino et al., 1999), or measure-

ment of the amount of hemoglobin extracted from the

excised hemorrhagic haloes (de Roodt et al., 2000) have

been proposed.

Data reported on Bothops venoms indicate that although

measurement of the size of the hemorrhagic haloes or the

amount of hemoglobin extracted from the excised hemor-

rhagic haloes are both functions of the intensity of the

response (i.e. hemorrhage), they give different results when

employed to quantify the same effect (de Roodt et al., 2000).

The fact that the amount of blood leaked as measured by

hemoglobin, appears not to be proportional to the size of the

hemorrhagic haloes rises severe questions on the validity of

measurements employing one or the other parameter. Since

this problem appeared frequently in our comparative studies

of the hemorrhagic activity in the venoms from different

Bothrops species from Argentina, we have compared the

results obtained by measuring the average diameters of the

haloes, the weights of the excised hemorrhagic haloes and

the amount of hemoglobin extracted from these excised

haloes produced after the i.d. injection to mice and rats of

either crude venoms or venom samples in which the

hemorrhagic activity was partially or completely inacti-

vated. Taking into account that treatment of hemorrhagins at

low pH values or with chelating agents like EDTA produces

structural alterations leading to the irreversible loss of both,

proteolytic and hemorrhagic activities (Sanchez et al.,

1995a), we evaluate the proteolytic and hemorrhagic

activities after total or partial inactivation.

These results were employed to determine the dose-

effect curves as well as to analyze the correlation between

hemorrhagic, proteolytic activity on gelatin and lethal

potency among those venoms and to compare the hemor-

rhagic activity by the measurement of the hemorrhagic halo

or the amount of hemoglobin extracted.

2. Materials and methods

2.1. Venoms

Whole venoms were obtained from healthy, adult

specimens of Bothrops (B.) alternatus; B. jararaca; B.

moojeni and B. neuwiedii, kept at the Serpentarium of the

Instituto Nacional de Produccion de Biologicos—A.N.L.I.S.

‘Dr Carlos G. Malbran’. The venoms were collected Petri

dishes, dried in vacuo and kept at 220 8C.

2.2. Animals

Mice (CF-1, 18–22 g) and rats (Whistar, 180–220 g)

were provided by the animal facility of the Instituto

Nacional de Produccion de Biologicos A.N.L.I.S. ‘Dr

Carlos G. Malbran’. They were kept under controlled

environmental conditions with dark/light cycles of 12 h and

received commercial rodent food and water ad libitum.

2.3. Reagents

All the reagents employed were analytical grade. Protein

determinations were performed according to Bradford

(1976) using the Bio-Rad Protein Assay Kit.

2.4. Determination of the activity and composition stability

of venom samples

The instability by autodegradation of Bothrops venoms

is well known (Vidal and Stoppani, 1970; Souza et al.,

2001). In order to establish the range of pH values of interest

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958950

for studies of stability of proteolytic activity on gelatin, two

living specimens of B. alternatus were milked into a pH-Stat

(Radiometer, Copenhagen) titration vessel (0.5–5.0 ml) so

that the tips of the fangs passed through a layer of neutral

liquid paraffin, in order to prevent the exchange of CO2 with

the air. The pH, recorded continuously in the stirred sample

was 5.897 ^ 0.16 and varied by less than 0.01 pH units

during 12 h. Venom samples withdrawn after different

intervals of time were analyzed by SDS–PAGE. No

changes in the electrophoretic profile were observed during

12 h incubation. Once the pH conditions for optimum

stability were established, the dried venoms were dissolved

in 0.15 M NaCl, 20 mM sodium acetate buffer of the desired

pH into a (0.5–5.0 ml) titration vessel at 23 8C. The final pH

was controlled and, if necessary, adjusted in the pH-Stat

(kinetic mode) until a stable reading was obtained for 5 min.

The venom solutions were centrifuged for 10 min at 1200g

and the supernatant was placed in a clean titration vessel

kept at 23 8C. The pH was tested again and the sample was

kept under continuous stirring until use.

2.5. Proteolytic activity on gelatin

Gelatin was chosen as substrate to determine this activity

because in previous reports we found good correlation

between the liquefaction of gelatin and the MHD (cited in de

Roodt et al. (1999)). Gelatin (Sigma, from porcine skin, 300

Bloom) was suspended at 1% in 0.15 M NaCl, heated for 1 h

in a boiling water bath under continuous stirring, and

allowed to cool down up to 25 8C. Five ml were pipetted into

a 0.5–5.0 ml titration vessel of a pH-Stat (Radiometer) and

adjusted to 23 8C and pH 8.0 with 20 mN NaOH under a

current (15 ml/min) of humid argon. After stabilization, the

consumption of NaOH to maintain pH 8.0 was recorded for

10 min. hence, samples of different venom solutions were

added, and the consumption of alkali to maintain pH 8.0 was

recorded for additional 10 min period. The differences in

rates of alkali consumption in the presence and in the

absence of venom measure the rate of reaction. pH 8.0 was

routinely employed in order to improve titration efficiency.

One unit of proteolytic activity was defined as the

consumption of 1.0 nmol NaOH per min. Specific activities

were calculated in units per mg protein.

To determine the proteolytic activity of the venom

treated at different pH values, the proteolytic activity on

gelatin as a function of pH was studied by incubating the

venom solution for 1 h at pH values 5.8 (pH in which

the proteolytic activity is most stable), 5.1 (pH close with

the apparent pK value) and 3.8 (pH in which proteolytic

activity on gelatin is undetectable; see Results section).

Dried venoms (75–80 mg with B. jararaca; B. moojeni or B.

neuwiedii venoms; 63–65 mg with B. alternatus venom)

were dissolved in 1.8 ml of 0.15 M NaCl, 20 mM sodium

acetate buffer of the desired pH in a titration vessel at 23 8C.

The final pH was adjusted, if required in the pH-Stat.

After centrifugation, 1.4 ml of supernatant was transferred

to a clean titration vessel kept at 23 8C, the pH was tested

again and the sample was incubated under continuous

stirring. At different intervals of time 0.2 ml aliquots were

mixed with 0.28 ml of 0.15 M NaCl and the sample was

adjusted to pH 5.8 in the pHStat. The final volume was

adjusted to 0.5 ml, the time recorded and the enzymatic

activity on gelatin was measured using 25–50 ml samples as

described.

To determine the activity of the venom in the presence of

EDTA, Bothrops venoms (0.3 mg/ml) were incubated for

30 min at 23 8C in 0.15 M NaCl containing 3.0–1.0 mM

disodium EDTA and adjusted to pH 7.2. The stability of

proteolytic activity on gelatin was tested with venoms

treated at different pH levels and with venoms preincubated

with EDTA as described above.

2.6. Hemorrhagic activity

With all the Bothrops venoms employed, no significant

differences in hemorrhagic activity were observed with

samples used immediately after dissolution in 0.15 M NaCl

or after 1 h preincubation at pH 5.8.

To determine the MHD, different doses of each venom

(0.5–500 mg) adjusted to pH 5.8 were injected intradermi-

cally in mice, using 3–5 animals per dose level or in rats

using 3–5 points per dose level (Theakston and Reid, 1983).

To determine the inhibition of the hemorrhagic activity

of venom treated at different pH, venom samples were

preincubated at 23 8C for 1 h in 0.15 M NaCl, 20 mM

sodium acetate buffer at pH 5.8, 5.1 and 3.8. After

incubation, the samples were readjusted to pH 5.8 with the

pH-Stat, the final volume was adjusted and the hemorrhagic

activity was assayed by intradermal injection of 0.1 ml

samples in rats. Rats, under light anesthesia with ketamine

(50 mg/kg) were injected intradermically with three samples

(one preincubated at pH 5.8 in the back as control, one

incubated at pH 5.1 and another incubated at pH 3.8 in each

side) of up to 600 mg of each venom, in order to compare in

the same animal the effect of a control sample with that of

samples treated at different pH values.

To determine the inhibition of the hemorrhagic activity

of venom preincubated with EDTA, venom samples

(400 mg) were incubated for 30 min at 23 8C in 0.3 ml of

0.15 M NaCl containing 0.02 to 3.0 mM EDTA adjusted to

pH 7.4. These samples were employed to measure

hemorrhagic activity in rats (n ¼ 5 for each venom) using

a similar scheme. Venom without treatment was injected in

the back and EDTA-treated venom samples injected in the

sides. This methodology was also employed to decrease

individual variability of the response.

Three hours after the injection, the animals were

sacrificed, the skin was removed and the major perpendicu-

lar diameters of the hemorrhagic haloes were measured with

a caliper. The hemorrhagic haloes were immediately cut,

weighed and homogenized for 3.0 min in a Tenbroeker

tissue grinder with 5.0 ml of distilled water as described

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958 951

previously (de Roodt et al., 2000). After centrifugation,

2.0 ml of the supernatants were transferred to glass tubes,

extracted with the same volume of chloroform and 100 ml of

clear aqueous phases (or dilutions) were pipetted in 96-well

plates in which the hemoglobin content was determined by

the absorbance at 550 nm directly measured (Al-Abdulla

et al., 1991) or by measurement of peroxidase activity as

described previously (de Roodt et al., 2000).

The hemorrhagic activity of each venom sample was

expressed (a) as the average diameter ð½d1 þ d2�=2Þ

measured in cm; (b) as the weight of the excised

hemorrhagic haloes (in g ^ SD) or (c) as the amount of

hemoglobin extracted from the excised hemorrhagic haloes

in A550 units per ml (^SD) of the aqueous phases.

2.7. Lethal potency

The lethal potency of samples of the different Bothrops

venoms preincubated at different pH values after readjust-

ment to pH 5.8 or after treatment with EDTA was

determined.

After 1 h preincubation at pH 5.1 or 3.8 as described

above, the Bothrops venom solutions were adjusted to pH

5.8, and their lethal potency were studied on mice by

comparing the mean survival time after the i.p. injection of a

(nominal) dose of 3.0 LD50 (i.p.) of the treated venom

samples and control venoms.

To determine the lethal potency of venoms previously

treated with EDTA, 5.0 mg/ml of venom of B. alternatus

and B. jararaca were incubated for 30 min at 23 8C in

0.15 M NaCl, 0.1 M EDTA, 20 mM Sodium acetate buffer

pH 6.5. The sample was chromatographed in a Sephadex G-

25 column pre-equilibrated with 0.15 M NaCl–20 mM

sodium acetate buffer pH 5.8, in order to eliminate the

excess of EDTA. As controls there were used samples

treated in the same way but in the absence of EDTA. After

measurement of protein content of the samples, mice were

injected i.p. with nominal 3.0 LD50 of treated venom or

venom control.

2.8. Statistics

All data are presented as mean ^ SD. Linear and non-

linear regression analysis as well as tests for statistical

significance were performed by using the combined

Prisma—StatMate software (GraphPad Software, San

Diego, CA).

3. Results

3.1. Proteolytic activity

The specific proteolytic activities (units per mg

protein) of the Bothrops venoms employed were 508 ^ 33

(B. alternatus), 232 ^ 35 (B. neuwiedii); 260 ^ 40

(B. moojeni) and 156 ^ 35 (B. jararaca).

Except for the sample preincubated at pH 5.8, which

changed by less than 10% with time without significant

changes in the SDS–PAGE profiles, the proteolytic activity

on gelatin ða1; a2;…:; anÞ obtained from a venom sample

preincubated at pH values lower than 5.8 for t1; t2;…; tn min

decreased with the time of incubation as an exponential

decay of the form a2=a1 ¼ exp½2k0ðt2 2 t1Þ� and the

apparent first-order constant (k0) for each pH value was

calculated by non-linear regression analysis of the activity

vs. time curves.

The proteolytic activity on gelatin of the Bothrops

venoms studied was significantly inhibited by preincubation

at pH values lower than 5.0 (Table 1, Fig. 1B). For each

fixed pH value, the rate of enzyme inactivation could be

described as an exponential decay, the value of the apparent

first-order constant k 21 (inactivation) increased as the pH

decreased. The decrease in proteolytic activity on gelatin of

B. alternatus, B. neuwiedii and B. moojeni venoms after

incubation at different pH values exhibited a similar profile.

The enzymatic activity decreased smoothly with time at pH

5.8 (k0 , 0.12 h21 with B. alternatus to k0 , 0.135 h21 with

B. moojeni venom). The rates of inactivation increased at pH

5.5 (k0 , 0.18 h21 with B. neuwiedii venom; k0 , 0.32 h21

Table 1

Effect of preincubation at different pH values on hemorrhagic activity of Bothrops venoms

Venoms pH 5.8 pH 5.1 pH 3.8

Average diameter

(cm)

Hemoglobin

(A550)

Average diameter

(cm)

Hemoglobin

(A550)

Average diameter

(cm)

Hemoglobin

(A550)

B. alternatus 0.94 ^ 0.16 0.43 ^ 0.04 0.66 ^ 0.06 0.25 ^ 0.10 0.18 ^ 0.09 0.06 ^ 0.02

B. neuwiedii 1.07 ^ 0.18 0.12 ^ 0.06 0.86 ^ 0.10 0.08 ^ 0.02 0.24 ^ 0.16 0.02 ^ 0.01

B. moojeni 1.13 ^ 0.07 0.23 ^ 0.05 0.99 ^ 0.11 0.14 ^ 0.02 0.54 ^ 0.10 0.04 ^ 0.02

B. jararaca 0.94 ^ 0.43 0.38 ^ 0.02 0.88 ^ 0.08 0.30 ^ 0.03 0.22 ^ 0.02 0.05 ^ 0.03

The hemorrhagic activity of the different venoms treated at different pH values was determined by measuring the average diameters or by

the hemoglobin extracted from the hemorrhagic halo. The diameters are expressed in cm as the mean ^ SD. The hemoglobin extracted is

expressed in A550/ml as the mean ^ SD.

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958952

with B. alternatus and B. moojeni venoms) and at pH 5.0

(k0 , 0.38 h21 with B. neuwiedii venom and k0 , 0.53 with

B. alternatus and B. moojeni venoms). Higher rates of

inactivation were observed at pH values lower than 5.0. The

plots of residual activity after 1 h incubation as a function of

pH (Fig. 1A) were fitted to simple sigmoid curves with pK

values about 5.3. The plots of the logarithm of the residual

activity after 1 h incubation as a function of the pH (Fig. 1B)

exhibited an inflection at pH 5.3 in which the slope changes

from 0 to 1.0 and a second one at pH 4.5, in which the slope

changes from 1.0 to 3.0. The inflection at pH 5.3 may

represent the molecular dissociation constant of the first

ionic species to be protonated as the pH is moved down from

that of maximum stability. Compared to the magnitude of

the known group constants, it is close to that reported for the

imidazolium group of histidine. Further changes in slope in

the logarithmic plot suggest that the ionization of more than

one group may be involved in enzyme stability. The pK

values may be shifted if a conformational change facilitates

the loss of protons or permits the binding of a polyvalent

cation (Tripton and Dixon, 1979).

The rate of inactivation of proteolytic activity on gelatin

with B. jararaca venom changed only poorly in the range

from pH 5.8 (k0 , 0.09 h21) to pH 5.0 (k0 , 0.12 h21). The

rate increased significantly (k0 , 0.42 h21) at pH 4.5 an

increased further at lower pH values. The plot of residual

activity after 1 h preincubation as a function of pH (Fig. 1A)

fitted a simple sigmoidal curve with an apparent pK value

about 4.2. The plot of logarithm of the residual activity after

1 h preincubation as a function of pH (Fig. 1B) showed

again several linear portions with slopes of about 3.0 (from

pH 3.5 to 4.0); about 1.0 (pH 4.0–4.5) and became almost

horizontal (slope zero) from pH 4.5 to 5.8. The inflection at

pH 3.8 may reflect the ionization of an acididic group, like

the active site Glu 143.

No proteolytic activity on gelatin could be detected with

any of the venoms employed in this study after 1 h

incubation at pH 3.8.

On this basis, proteolytic activity on gelatin and

hemorrhagic activity of each venom were compared after

preincubation of the venom samples for 1 h at three pH

values, namely (a) pH 5.8, at which proteolytic activity is

most stable, (b) at pH 5.1, close to the apparent pK values

and (c) at pH 3.8, at which proteolytic activity on gelatin is

undetectable.

Incubation of B. alternatus, B. neuwiedii and B. moojeni

venoms with 1.0 mM EDTA resulted in complete inacti-

vation of proteolytic activity on gelatin. On the other hand,

after incubation of B. jararaca venom with 1.0 mM EDTA

proteolytic activity on gelatin was incompletely inactivated,

and a fraction (17–30%) of the initial proteolytic activity

remained still measurable.

3.2. Hemorrhagic activity

The results, expressed as average diameters of the

hemorrhagic haloes ð½d1 þ d2�=2Þ and as amount of

hemoglobin extracted (A550) are presented in Tables 1 and

3. With all the Bothrops venoms employed, no significant

differences in hemorrhagic activity were observed with

samples used immediately after dissolution in 0.15 M NaCl

or after 1 h preincubation at pH 5.8. The porcentual

inhibitions by preincubation at different pH levels are

presented in Table 2. The MHD found in rats was about

170 ^ 20 mg for B. alternatus, 140 ^ 15 mg for B.

jararaca, 210 ^ 30 mg for B. neuwiedii and 150 ^ 50 mg

for B. moojeni. The MHD in mice of these venoms was

9.0 ^ 0.3 mg for B. alternatus, 2.5 ^ 0.5 mg for B.

jararaca, 17.1 ^ 0.4 mg for B. neuwiedii and 30 ^ 4 for

B. moojeni venom (Table 3).

The values of hemoglobin extracted from hemorrhagic

haloes were very similar for each venom, whereas the

MHDs showed a high variation, ranging the relation MHD

rats / mice from five-fold with B. moojeni venom to 56 with

B. jararaca venom. As shown in Fig. 2 and Table 3, plots of

the ratio [average amount of hemoglobin extracted]/[weight

in grams of the excised hemorrhagic halo] is a constant k,

Fig. 1. (A) Plots of residual proteolytic activity on gelatin of

Bothrops venoms after 1 h preincubation at 23 8C as function of pH.

The data are presented as mean ^ SD (n ¼ 3). (X) B. alternatus,

(P) B. jararaca, (A) B. neuwiedii; (B) plots of the logarithm of

residual activity (as percentage) of Bothrops venoms after 1 h

preincubation at 23 8C as a function of pH. (B) B. alternatus, (P) B.

jararaca, (A) B. neuwiedii. (V) B. moojeni.

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958 953

characteristic of each Bothrops venom and independent of

the venom dose. The numerical values of k (average A550/

ml extracted per gram of excised hemorrhagic haloes) were

3.54 [95% i.c. 3.13–3.95] for B. alternatus, 2.33 [2.14 to

2.52] for B. jararaca, 1.67 [1.24–1.89] for B. moojeni and

1.16 [0.83–1.49] for B. neuwiedii. (i.e. close to those

already published for these venoms in experiments on mice,

see Table 3).

In all the cases the incubation of the venoms at pH 5.1

reduced the hemorrhagic activity. The hemorrhage was

inhibited in over 80% when the residual activity was

determined by measurement of the hemoglobin from the

hemorrhagic halo. When the residual activity was estimated

by measurement of the average diameters of the hemor-

rhagic halo, the inhibition was estimated in the order of 50–

80%. See Table 2.

Treatment of B. alternatus and B. neuwiedii venoms with

0.15 mM EDTA and of B. moojeni venom with 0.3 mM

EDTA resulted in the complete loss of hemorrhagic activity

(measured by both, the average diameter of the hemorrhagic

haloes or the amount of hemoglobin extracted) after the i.d.

injection to rats. On the other hand, incubation of B.

jararaca venom with EDTA up to 0.3 mM inhibited its

hemorrhagic activity after intradermal injection to rats up to

20–26% (by the average diameter of the hemorrhagic

haloes) and up to 70–83% (by measurement of the amount

of hemoglobin extracted from the excised hemorrhagic

haloes).

3.3. Lethal potency

With all the Bothrops venoms employed, no significant

differences in lethal potency were observed with samples

used immediately after dissolution in 0.15 M NaCl or after

1 h preincubation at pH 5.8. The mean survival times were

30–40 min., and all the animals died within 1 h post

injection. Post mortem gross pathological examination

showed generalized hemorrhages in all cases.

Preincubation of Bothrops venoms at pH 5.1 increased

the mean survival time in all cases (Fig. 3). With B.

neuwiedii venom, the mean survival time increased to 1.3 h,

however, all the animals died 2 h after injection. With B.

jararaca venom the mean survival time was about 1.2 h and

15% of the animals survived 24 h after injection. With B.

alternatus and B. moojeni venoms, the mean survival times

were 2.3 and 2.8 h, respectively, and 14–20% of the animals

survived 24 h after the injection. Post-mortem gross

pathological examination again showed generalized

hemorrhages.

Preincubation of Bothrops venoms at pH 3.8 strongly

reduced the lethal potency of all the venoms tested (Fig. 3).

However, except for B. moojeni venom, 16–20% of

Table 2

Inhibition of proteolytic activity on gelatin and hemorrhagic activities using venoms treated at different pH values

Venom Percentage of inhibition of venoms treated at pH 5.1 Percentage of inhibition of venoms treated at pH 3.8

Inhibition of

proteolysis

Inhibition of

hemorrhage

(A550/ml)

Inhibition of

hemorrhage

(diameter)

Inhibition of

proteolysis

Inhibition of

hemorrhage

(A550/ml)

Inhibition of

hemorrhage

(diameter)

B. alternatus 42.0 ^ 2.1 42.0 ^ 16.8 30.0 ^ 2.7 100.0 86.0 ^ 28.7 80.9 ^ 40.1

B. neuwiedii 34.3 ^ 3.2 33.3 ^ 8.3 20.0 ^ 5.0 100.0 83.3 ^ 41.7 77.6 ^ 51.7

B. moojeni 45.0 ^ 5.0 39.0 ^ 5.6 12.0 ^ 1.3 100.0 82.6 ^ 41.3 52.2 ^ 9.7

B. jararaca 18.2 ^ 1.1 21.0 ^ 2.1 6.0 ^ 5.4 100.0 86.8 ^ 52.1 76.6 ^ 7.0

The table indicates the proteolytic activity on gelatin and the hemorrhagic activity of venoms treated at different pH levels. The values in

table represent the percentage of inhibition of the proteolytic activity on gelatin and the hemorrhagic activity determined by measurement of

diameters of the hemorrhagic halo or by the hemoglobin extracted from the hemorrhagic halo. The values are expressed as the mean ^ SD.

Table 3

Hemorrhagic activity in mice and rats determined by both methods

Venom MHD mice MHD rats MHD rats/mice k Rats (A550/ml g21) k Mice (A550/ml g21) k Rats/mice

B. alternatus 9.0 ^ 3.0 170 ^ 20 19 3.54 (3.13–3.95) 3.61 (2.91–4.31) 0.98

B. jararaca 2.5 ^ 0.5 140 ^ 15 56 2.33 (2.14–2.52) 3.19 (2.86–3.53) 0.73

B. moojeni 30.0 ^ 4.0 150 ^ 50 5 1.67 (1.24–1.89) 2.17 (1.97–2.51) 0.78

B. neuwiedii 17.1 ^ 0.4 210 ^ 30 12 1.16 (0.83–1.49) 1.02 (0.95–1.08) 1.13

The table indicates the values of MHD (expressed as mg ^ S.D. of venom) or the hemoglobin extracted from 1 g of hemorrhagic halo

expressed as k (A550/ml g21). The 95% c.i. are indicated into brackets. MHD rats/mice indicates the relation between MHD of rats and mice in

each venom. k rats/mice indicates the relation in hemoglobin extracted per gram of hemorrhagic halo between rats and mice.

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958954

the animals died about 2 h after the injection of B. neuwiedii

venom and a similar percentage of animals died 24–48 h

after the injection of B. alternatusand B. jararaca venoms,

without exhibiting generalized hemorrhages upon post

mortem gross pathological examination.

When mice were inoculated with venom previously

treated with EDTA as described, the survival time was over

12 h to mice injected with B. jararaca venom and over 24 h

to those injected with B. alternatus venom. The controls

died 30–45 min. post injection.

4. Discussion

The inhibition of the proteolytic and hemorrhagic

activities by preincubation of venoms at different pH values

or with EDTA was observed in all the cases, however some

discrepancies were found in the expected values of

inhibition depending of the method used to determine the

hemorrhagic activity (Tables 1 and 2).

The most remarkable observation was that, with all the

Bothrops venoms employed, the rates of inactivation of

proteolytic activity on gelatin at pH 5.1 describes quanti-

tatively the decrease in the amount of hemoglobin extracted

from the excised hemorrhagic haloes (Table 2). In fact, the

pseudo-first order rate constants (k0) for the decay of

proteolytic activity on gelatin at pH 5.1 were about 0.58 h21

(B. alternatus), 0.42 h21 (B. neuwiedii), 0.55 h21 (B.

moojeni) and 0.20 h21 (B. jararaca). Thus, after 1 h

preincubation at pH 5.1, the initial activity will be reduced

Fig. 2. Plot of A550 per gram of excised hemorrhagic haloes (k) as

function of the venom dose in rats. The value of k for each venom is

obtained from the intercept and is expressed as mean [95% c.i.

limits] (n ¼ 5). (B) B. alternatus (k: 3.54 [3.13–3.95]), (P) B.

jararaca (k: 2.33 [2.14–2.52]), (V) B. moojeni (k: 1.57 [1.14–

2.09]), (A) B. neuwiedii (k: 1.16 [0.83–1.49]).

Fig. 3. Effect of preincubation for 1 h at different pH values of Bothrops venoms on the mean survival time of mice after intraperitoneal

injection. Mice (groups of, at least 6 animals per curve) were injected intraperitoneally with samples of B. jararaca (A) B. neuwiedii (B) B.

alternatus (C) and B. moojeni (D) venoms preincubated for 1 h at 23 8C at pH 3.8 (a), 5.1 (b) and 5.8 (c) at a (nominal) dose of 3.0 DL50. Each

curve was obtained by triplicate and the percentage of surviving animals is presented as a function of the time in hours after injection.

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958 955

by 42% (B. alternatus), 34.3% (B. neuwiedii), 45% (B.

moojeni) and 18.2% (B. jararaca), in close agreement with

the degrees of inhibition found experimentally when the

hemoglobin content of the hemorrhagic area were measured

(Table 2). After preincubation at pH 3.8, in agreement with

the complete inhibition of proteolytic activity on gelatin, the

decrease in the amount of hemoglobin extracted from

the hemorrhagic haloes ranged between 80 and 90% with all

the Bothrops venoms studied (Tables 1 and 2).

The effect of preincubation at different pH values can be

interpreted as the result of the distribution of the enzyme(s)

protein(s) into different ionic species. While the distribution

of the enzyme(s) into different ionic species at pH 5.8 is

surely different that that prevailing at pH 8.0 (i.e. the value

at which the enzyme activity is measured), most of the ionic

species are recovered as catalytically active enzyme. At

lower pH values, a significant fraction of the enzymes(s)

undergoes irreversible inactivation. Readjustment to pH 5.8

after incubation will favor the distribution towards poten-

tially catalytically active species, however, the fraction lost

due to irreversible inactivation will not be recovered.

Preincubation of Bothrops venoms at pH 5.1 or 3.8

increased the mean survival time after the i.p. injection to

mice (Fig. 3), and post-mortem examination suggests that

this effect is related to the inhibition in hemorrhagic activity.

The magnitude of these increases ranged from about two-

fold (B. jararaca), 2.5 to 3-fold (B. neuwiedii) up to 4 to 5-

fold (B. alternatus and B. moojeni). This sequence is

consistent with the decrease in the amount of hemoglobin

extracted from the hemorrhagic haloes, rather than with the

decrease in size of the hemorrhagic haloes.

When the venom was treated by EDTA, no hemor-

rhages were detected and the mean survival time of mice

inoculated with B. alternatus or B. jararaca venoms was

over 24 or 12 h, respectively, while the controls died in

30–45 min. Treatment of B. alternatus, B. moojeni and

B. neuwiedii venoms with 1.0 mM EDTA resulted in

complete inactivation of proteolytic activity on gelatin, as

well as in hemorrhagic activity after the intradermal

injection to rats. In contrast, preincubation of B. jararaca

venom with 1.0 mM EDTA inhibited proteolytic activity

on gelatin by about 85%. The remaining hemorrhagic

activity after i.d. injection to rats was about 20% by the

measurement of diameters and about 73% by the amount of

hemoglobin extracted from the hemorrhagic haloes near the

value of inhibition observed in the proteolytic activity of

this venom after EDTA treatment. The results with

B. jararaca venom suggest that the hemorrhagins of this

venom differ in some characteristics with those from the

other venoms studied and/or that other venom components

are involved in the hemorrhagic process on the site of

injection of this venom.

The concordance between the hemorrhage (specially

when it was determined by the measurement of

hemoglobin), the proteolytic activity and the lethal

potency, suggests a relevant participation of hemorrhagic

metalloproteinases (hemorrhagins) in the lethality caused

by Bothrops venoms. Bothropic venoms produce gener-

alized hemorrhages due the procoagulants and thrombin-

like enzymes, which lead to the fibrinogen consumption

(Kamiguti et al., 1996; Mandelbaum et al., 1982).

Metalloproteinases contribute to the haemostatic disturb-

ances acting on some points of the haemostatic system

(Kamiguti et al., 1996; Markland, 1998), by the loss of

the vascular integrity (Bjarnasson and Fox, 1994;

Gutierrez and Rucavado, 2000) and facilitating the

generalization of the venom components destroying the

extracellular matrix and blood vessels at the site of bite

(Anai et al., 2002). In this work we could observe that

when the proteolytic activity of the venoms was inhibited,

there were not observed generalized hemorrhages and the

lethal potency of the venoms diminished, indicating the

importance of this enzymes in the systemic envenoming

by Bothrops snakes.

However, although the hemorrhagic activity seems to be

very important in the lethality caused by bothropic venoms,

the contribution of other components to the lethal potency

resulted evident when no macroscopic hemorrhages were

detected in rats or in mice killed by venom in which the

hemorrhagic activity was inhibited. In fact, except for B.

moojeni venom, 15–20% of the mice injected i.p. with

Bothrops venoms preincubated at pH 3.8 died 2 h after

injection (B. neuwiedii) or about 24 h after injection (B.

alternatus and B. jararaca), without evidences of general-

ized hemorrhages upon post-mortem gross pathological

examination.

The differences between the inhibition of proteolytic

activity and hemorrhagic activity using both methods can be

related with the parameters that we are measuring. The

hemoglobin content in the hemorrhagic halo reflects the

amount of blood leakage from blood vessels after its rupture.

Although the area of the hemorrhagic spot reflects the

hemorrhagic activity of the venom, not necessarily indicates

the amount of red blood cells in the tissue but may reflect the

dispersion of the leakage or extravasation of red blood cells

in this area. This interpretation of the hemorrhagic activity

may reflect changes in vascular permeability that not

necessarily indicate the rupture of blood vessels or

destruction of the extracellular matrix, the principal factors

in the production of hemorrhages. These facts make difficult

the interpretation of the real hemorrhagic potency, specially

in comparative studies and may explain the lack of

concordance between the proteolytic activity and hemor-

rhage determined by both methods.

Having the appropriate control, the simple and repro-

ducible measurement of the size of the hemorrhagic haloes

will be the method of choice. However, some problems can

be found when the comparison between different venoms

has to be done.

Leakage of blood is the inherent effect of hemorrhagic

activity induced by Bothrops venoms, however differences

in potency of the hemorrhagic activity in different venoms

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958956

are likely to occur, given the microheterogeneity among

the hemorrhagic metalloproteinases, and that usually,

multiple forms having significant different potencies (Jia

et al., 1996) are observed even in a single venom

(Bjarnasson and Fox, 1994; Mandelbaum et al., 1982).

This is consistent with the observation of differences in

maximum effect (YMax) among Bothrops venoms in mice

when the amount of hemoglobin extracted was plotted as a

function of the venom dose, that showed big differences in

the hemoglobin content from hemorrhagic areas of a similar

size (de Roodt et al., 2000).

The close agreement between the degree of inhibition of

proteolytic activity on gelatin by preincubation at different

pH values or after treatment with EDTA with the decrease in

hemorrhagic activity expressed as the amount of hemo-

globin extracted, strongly suggest that this method reflects

real differences in potency of hemorrhagic activity (Tables 2

and 3) since some differences are not detected when the

hemorrhagic activity is studied by the size of the hemor-

rhagic haloes. In fact, having the MHD in rats of

B. alternatus (about 170 mg), B. jararaca (about 140 mg),

B. moojeni (about 150 mg) or B. neuwiedii venoms (about

210 mg) there is no way to predict that the amount of

hemoglobin extracted (A550/ml) from a similar hemorrhagic

halo will be about three-fold higher with B. alternatus and

B. jararaca venoms or two-fold higher with B. moojeni

venom than the extracted with B. neuwiedii venom, unless

the A550 be measured (Table 3).

When the differences between the MHD of a snake

venom were compared in different species, we found very

big differences, ranging since 5–56 fold. However, when

the k (A550/ml g21) of the different venoms in rats or mice

were compared, the values of A550/ml g21 found for each

venom were quite similar with differences ranging from 2 to

27% (Table 3). The value of k for each venom was defined

previously (de Roodt et al., 2000) as the slope of the plots of

amount of hemoglobin (A550/ml) extracted as a function of

the weight (in g) of the excised hemorrhagic haloes, so that,

if H is the A550/ml extracted from a hemorrhagic halo

weighing Wgrams, is H ¼ k £ W : Therefore kð¼ H=WÞ is a

constant and seems to be an intrinsic property characteristic

of each venom related with the hemorrhagic potency and

independent on the venom dose (Fig. 2).

Combined with those data on inactivation by EDTA, the

results of pH-stability of the hemorrhagic activity of

Bothrops venoms indicate that the measurement of hemo-

globin well correlates with the activity of proteases that acts

on gelatin and with the lethal potency of the venoms. The

measurement of the hemorrhagic haloes may not be a

reliable parameter to determine the remaining hemorrhagic

activity of a venom after such treatments. Consequently, it

can be assumed that the amount of hemoglobin extracted

from the hemorrhagic haloes produced by Bothrops venoms

rather than their size, seems to reflect more accurately the

potency of their hemorrhagic activity.

Acknowledgements

The authors are very grateful to the anonymous

reviewers by their helpful comments to improve the quality

of the paper.

References

Al-Abdulla, I.H., Sidki, A.M., Candon, J., 1991. An indirect

hemolytic assay for assessing antivenoms. Toxicon 29,

1043–1046.

Anai, K., Sugiki, M., Yoshida, E., Maruyama, M., 2002.

Neutralization of snake venom hemorrhagic metalloproteinases

prevents coagulopathy after subcutaneous injection of Bothrops

jararaca venom in rats. Toxicon 40, 63–68.

Baramova, E.N., Shannon, J.D., Bjarnason, J.B., Fox, J.W., 1990.

Identification of the cleavage sites by a hemorrhagic metallo-

proteinase in Type IV collagen. Matrix 10, 91–97.

Bee, A., Theakston, R.D., Harrison, R.A., Carter, S.D., 2001. Novel

in vitro assay for assessing the hemorrhagic activity of snake

venoms and for demonstration of venom proteinase inhibitors.

Toxicon 39, 1429–1434.

Bjarnasson, J.B., Fox, J.W., 1983. Proteolytic specificity and cobalt

exchange of hemorrhagic toxin e, a zinc protease isolated from

the venom of the western diamondback rattlesnake (Crotalus

atrox). Biochemistry 22, 3770–3778.

Bjarnasson, J.B., Fox, J.W., 1994. Hemorrhagic metalloproteinases

from snake venoms. J. Pharmacol. Therap. 62, 325–372.

Bjarnasson, J.B., Fox, J.W., 1995. Snake venom metallo-endopep-

tidases. In: Barret, A.J., (Ed.), Methods of Enzimology—

Proteolytic Enzymes, vol. 248. Academic Press, New York, pp.

345–368, Part E.

Bjarnasson, J.B., Hamilton, D., Fox, J.W., 1988. Studies on the

mechanism of hemorrhage production by five proteolytic

hemorrhagic toxins from Crotalus atrox venom. Biol. Chem.

Hoppe–Seyler 369, 121–129.

Bradford, M.M., 1976. A rapid and sensitive method for the

quantitation of microgram quantities or protein utilizing the

principle of protein–dye binding. Anal. Biochem. 72, 248–254.

Civello, D.G., Duong, H.L., Geren, C.R., 1983. Isolation and

characterization of a hemorrhagic proteinase from timber

rattlesnake venom. Biochemistry 22, 749–755.

de Roodt, A.R., Dolab, J.A., Segre, L., Simoncini, C., Hajos, S.E.,

Fernandez, T., Dokmetjian, J.C., Litwin, S., Accattoli, C., Vidal,

J.C., 1999. Immunochemical reactivity and neutralizing

capacity of a polyvalent anti-vipera (European) antivenom on

enzymatic and toxic activities in the venoms of crotalids from

argentina. J. Venom. Anim. Toxins 5, 67–83.

de Roodt, A.R., Dolab, J.A., Dokmetjian, J.Ch., Litwin, S., Segre, L.,

Vidal, J.C., 2000. A comparison of different methods to asses

the hemorrhagic activity of Bothrops venom. Toxicon 38,

865–873.

Esmeraldino, L.E., Franco, J.J., Vilela Sampaio, S., 1999. Two new

methods for the quantitation of the hemorrhagic activity induced

by B. jararaca venom. Toxicon 37, 261.

Gomis-ruth, F.X., Kress, L.F., Bode, W., 1993. First structure if a

snake venom metalloproteinase: a prototype for matrix

metalloproteinases collagenases. Eur. Mol. Biol. Org. J. 12,

4151–4157.

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958 957

Gutierrez, J.M., Chaves, F., 1980. Efectos proteolıtico, hemorragico

y mionecrotico de los venenos de serpientes costarricenses de

los Generos Bothrops, Crotalus y Lachesis. Toxicon 18,

315–321.

Gutierrez, J.M., Rucavado, A., 2000. Snake venom metalloprotei-

nases: their role in pathogenesis of local tissue damage.

Biochimie 82, 841–850.

Hite, L.A., Shannon, J.D., Bjarnasson, J.B., Fox, J.W., 1992.

Sequence of a cDNA clone encoding the zinc metalloproteinase

hemorrhagic toxine from Crotalus atrox. Evidence for signal,

zymogen and disintegrin-like structures. Biochemistry 31,

6203–6211.

Hite, L.A., Jia, L.G., Bjarnasson, J.B., Fox, J.W., 1994. cDNA

sequences for four snake venom metalloproteinases: structure,

classification and their relationship to mammalian reproductive

proteins. Arch. Biochem. Biophys. 308, 182–1914.

Jia, L.G., Shimokawa, K.I., Bjarnasson, J.B., Fox, J.W., 1996.

Snake venom metalloproteinases: structure, function and

relationship to the ADAMs family of proteins. Toxicon 34,

1269–1276.

Kamiguti, A.S., Hay, C.R.M., Theakston, R.D.G., Zuzel, M., 1996.

Insights into the mechanism of hemorrhage caused by snake

venom metalloproteinases. Toxicon 34, 627–642.

Kondo, H., Kondo, S., Ikezawa, H., Murata, R., Ohsaka, A., 1960.

Studies on the quantitative method for determination of

hemorrhagic activity of Habu snake venom. Jpn. J. Med. Sci.

Biol. 13, 43–51.

Mandelbaum, F.R., Reichel, A., Assakura, M.T., 1982. Isolation

and characterization of a proteolytic enzyme from the venom of

the snake Bothrops jararaca (jararaca). Toxicon 20, 955–972.

Markland, F.S., 1998. Snake venoms and the hemostatic system.

Toxicon 36, 1749–1800.

Maruyama, M., Sugiki, M., Yoshida, E., Shimaya, K., Mihara, H.,

1992. Broad substrate specificity of snake venom fibrinolytic

enzymes: possible role in hemorrhage. Toxicon 30, 1387–1397.

Oshaka, A., Omori-Satho, T., Kondo, H., Kondo, S., Murata, R.,

1966. Biochemical and pathological aspects of hemorrhagic

principles in snake venoms with special reference to Habu

(Trimeresurus flavoviridis) venom. Mem. Inst. Butantan 33,

193–205.

Ownby, C.L., Colberg, T.R., Odell, G.V., 1984. A new method for

quantitating hemorrhage induced by rattlesnake venoms: ability

of polyvalent antivenom to neutralize hemorrhagic activity.

Toxicon 22, 227–233.

Sanchez, E.F., Cordeiro, M.N., De oliveira, D.E., Juliano, L., Prado,

E.S., Diniz, C.R., 1995a. Proteolytic specificity of two

hemorrhagic factors, LHF-I and LHF-II, isolated from the

venom of the bushmaster snake (Lachesis muta muta). Toxicon

33, 1061–1069.

Sanchez, E.F., Costa, M.I.E., Chavez Olortegui, C., Assakura, M.T.,

Mandelbaum, F.R., Diniz, C.R., 1995b. Characterization of a

hemorrhagic factor, LHF-1, isolated from the Bushmaster snake

(Lachesis muta muta) venom. Toxicon 33, 1653–1667.

Sells, P.G., Richards, A.M., Laing, G.D., Theakston, R.D., 1997.

The use of hens’ eggs as an alternative to the conventional in

vivo rodent assay for antidotes to hemorrhagic venoms. Toxicon

35, 1413–1421.

Souza, J.R., Monteiro, R.Q., Castro, H.C., Zingali, R.B., 2001.

Proteolytic action of Bothrops jararaca venom upon its own

constituents. Toxicon 39, 787–792.

Stocker, W., Grams, F., Baumann, U., Reimemer, P., Gomis-ruth,

F.X., Mckay, D.B., Bode, W., 1995. The metzincins-topological

and sequential relations between the astacins, adamalysins,

serralysins, and matrixins (collagenases) define a superfamily of

zinc-peptidases. Prot. Sci. 4, 823–840.

Theakston, R.D.G., Reid, H.A., 1983. Development of simple

standard assay procedures for the characterization of snake

venoms. Bull. World Health Organization 61, 949–956.

Tripton, K.F., Dixon, H.B.F., 1979. Effects of pH on enzymes. In:

Purich, D.L., (Ed.), Methods in Enzymology—Enzyme kinetics

and mechanism. Part A, vol. 63. Academic Press, New York.

Vidal, J.C., Stoppani, A.O., 1970. Inhibition of phospholipase A by

a naturally occurring peptide in Bothrops venoms. Experientia.

8, 831–832.

Zhang, D., Botos, I., Gomis-ruth, F.X., Doll, R., Blood, C., Njoroge,

F.G., Fox, J.W., Bode, W., Meyer, E., 1994. Structural

interactions of natural and synthetic inhibitors with the venom

metalloproteinase, atrolysin C (Ht-d). Proc. Natl Acad. Sci.

USA 91, 8447–8451.

A.R. de Roodt et al. / Toxicon 41 (2003) 949–958958