Proc. Natl. Acad. Sci.
USA
Vol. 77, No. 11, pp. 6496-6500, November 1980
Biochemistry
Three-dimensional structure of a protein from scorpion venom: A
new structural class of neurotoxins
(sodium channels/membranes/crystal structure/protein conformation/amino acid sequence)
JUAN C. FONTECILLA-CAMPS*, ROBERT J. ALMASSYt*, FRED L. SUDDATHtt§, DEAN D. WATTS, AND
CHARLES E. BUGGt*§II
Departments of Microbiology and §Biochemistry, tComprehensive Cancer Center and tInstitute of Dental Research, University of Alabama, University Station,
Birmingham, Alabama 35294; and IDepartment of Biochemistry, School of Medicine, Creighton University, Omaha, Nebraska 68178
Communicated by Alexander Rich, August 6,1980
ABSTRACT The three-dimensional crystal structure of
variant-3 toxin from the scorpion Centruroides sculpturatus
Ewing has been determined at 3 A resolution. Phases were ob-
tained by use of K2PtCL4 and K2IrCl derivatives. The most
prominent secondary structural features are two and a half turns
of a-helix and a three-strand stretch of antiparallel a-sheet,
which runs parallel to the a-helix. The helix is connected to the
middle strand of the a-sheet by two disulfide bridges; a third
disulfide bridge is located nearby. Several loops extend out of
this dense core of secondary structure. The largest loop is joined
to the COOH terminus of the molecule by the fourth disulfide
bridge. The overall shape of the molecule resembles a right-hand
fist: the a-helix runs along the knuckles of the fist; the a-sheet
lies along the second and third joints of the fingers; the thumb
is defined by two short loops that are composed of residues
16-21 and residues 41-46; the wrist corresponds to the COOH-
terminal stretch of residues 52-65 and a loop composed of
residues 5-14; and the second joint of the little finger is near the
NH2 terminus of the molecule. The a-carbon backbone displays
a large flat surface that lies along the second joints of the fingers
and the heel of the hand in the fist model. Several of the con-
served residues in the scorpion neurotoxins are clustered on this
surface, which may play a role in interactions of scorpion toxins
with sodium channels of excitable membranes.
Scorpion venoms contain sets of small basic proteins that are
responsible for the neurotoxic activities of the venoms. These
toxins display considerable variations in amino acid composi-
tion, but they generally consist of 60-70 amino acids and are
crosslinked by four disulfide bridges. The partial or complete
primary structures of about 25 toxins, from several different
species of scorpions, have been determined (1). The amino acid
sequences display a number of common features, including the
same general locations of the eight cysteine residues, similar
disulfide bridging patterns, and the location of several invariant
or conserved residues. Thus, the available chemical data suggest
that all of the scorpion toxins probably have similar three-
dimensional structures.
The general effects of whole venoms or purified toxins in-
jected in mammals are those expected from the massive release
of neurotransmitters induced by depolarization of the nerve
endings (2-4). Despite structural similarities among the various
scorpion toxins, these proteins display wide variations in activity
(5, 6), and it is not clear if all of the scorpion toxins follow the
same general mechanism of action (7). Several of the toxins have
been shown to prolong the action potentials of excitable
membranes by blocking the inactivation of sodium channels
(8-10), an effect similar to that produced by sea anemone toxins
(11). However, the scorpion toxin and sea anemone toxin
binding sites appear to be distinct from those occupied by other
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be hereby marked "ad-
vertisement" in accordance with 18 U. S. C. §1734 solely to indicate
this fact.
6496
neurotoxins such as tetrodotoxin and saxitoxin, which block
sodium channels, or veratridine and batrachotoxin, which cause
a persistent activation of sodium channels (12).
Scorpion toxins are of considerable interest as probes for
studying the properties of sodium channels. In order to obtain
information about the three-dimensional structures of these
proteins, we have initiated a series of crystallographic studies
of toxins from the venom of Centruroides scuipturatus Ewing,
a scorpion that is native to the Arizona desert. In this paper we
describe the crystal structure at 3 A resolution of the variant-3
toxin, which is one of those for which the amino acid sequence
is available (Fig. 1).
METHODS
Variant-3 scorpion toxin was prepared as described (13). The
crystallization technique was the same as outlined (14) except
that the seeded solution was equilibrated by vapor diffusion
against unbuffered 66% (vol/vol) 2-methyl-2,4-pentanediol in
water. The crystals, which grow as large prisms, belong to the
orthorhombic space group P212121 with a = 51.89 A, b 41.75
A, and c = 28.23 A; there is one molecule of scorpion toxin per
asymmetric unit.
Heavy-atom derivatives were prepared by soaking the
crystals in 75% (vol/vol) 2-methyl-2,4-pentanediol that con-
tained 0.02 M Na cacodylate (pH = 5.8) and the heavy-atom
compound. These crystals then were mounted in glass capil-
laries in the usual way and examined by use of a Rigaku rotating
anode generator and a precession camera. The hOl zone was
usually recorded to determine the success of the substitution.
Of 30 heavy-atom compounds assayed under about 70 different
conditions, only 2, K2PtCl4 and K2IrCl6, produced clear in-
tensity changes without significant unit cell changes. The Pt
derivative was obtained by soaking crystals for 5 days in 0.1 mM
K2PtCl4; the Ir derivative was prepared by 16-hr soaks in a so-
lution 50% saturated with K2IrCl6.
Data for native and derivative crystals were collected with
a Picker FACS-I diffractometer at room temperature using an
w step-scan procedure and nickel-filtered CuKa radiation. The
reflections were divided in 20 shells containing 100 reflections
each and were collected from high to low resolution. Friedel
mates were generated during data collection and were collected
at negative 20 values in groups of 10 reflections, alternating with
their unique counterparts. The w scan width was typically
0.6°-1.0° and was selected so that, for most reflections, half of
the scan width contained the peak and the other half corre-
sponded to background. A typical scan step size was 0.02°-
0.03 .
The crystals were always mounted with the c crystallographic
1I To whom correspondence should be addressed.
 Biochemistry: Fontecilla-Camps et al. Proc. Natl. Acad. Sci. USA 77 (1980) 6497

sured at 400-reflection intervals. During the collection of data

5 S)wAi
FIG. 1. Amino acid sequence for the variant-3 toxin from Ce,n-
truroides scuipturatus Ewing. This primary structure is the or
originally reported (13), except for residues 25-29 which are base
on new sequence data and COOH-terminal residues 64 and 65 whicIh
are assigned from the electron density maps. Blackened portiorns
represent cysteine residues in disulfide bridges.
axis parallel to the capillary axis. Absorption corrections werre
applied according to the method of North et al. (15). In orde'r
to monitor and correct for decomposition effects, 18 standardd
reflections were measured periodically; 6 of these were meaL-
sured every 200 reflections, and the remaining 12 were mea
A composition effects. The data for the derivatives were scaled
250
200 +
r4 150
4 tailed models for the heavy-atom distributions were obtained
100-
504
6 5 3 rangement of heavy atoms. The Pt and Ir parameters were
0.0
.
to 3 A resolution, the intensities of the standard reflections de-
creased an average of 13%, 33%, and 15% for the native crystal,
Pt derivative, and Ir derivative, respectively.
A unique octant of data, plus the complete set of Friedel
mates, was collected from the native protein and from the two
derivatives. Native data and data from the Pt derivative were
measured out to Bragg spacings of 3 A, and the Ir derivative
data were measured out to a spacing of 3.1 A. The integrated
intensities were obtained in a two-step process during which
several of the shells were processed together. Peak positions
were determined by the method of Tickle (16). In the first pass
through the data, peak widths were obtained for the stronger
reflections by using the methods of Lehman and Larsen (17).
These peak widths were then expressed as a function of the
diffractometer angles 0 and X. In the second pass through the
data, all reflections were assigned peak widths that were cal-
culated from the (X, X function. The steps outside the peaks were
used to determine the background level for each individual
reflection.
The overall quality of the data sets was evaluated by com-
paring the intensities of Friedel mates. We computed the reli-
ability index:
Rsym = E I(h) -I(S)
s Em ,(I(h) + I([))
in which the sums included all Friedel mates for the native data
and those Friedel mates from the centric zones for the Pt and
Ir derivatives. The resultant Rsym values were 0.024 for the
native data, 0.020 for the Pt data, and 0.027 for the Ir data. The
standard reflections were used to correct the data sets for de-
to the native data by means of relative-Wilson plots (18), fol-
lowed by equating Z2Fp2 to .FPH2 for the individual shells of
data.
The Pt and Ir parameters were determined and refined by
standard methods (18). The heavy-atom sites were located by
use of isomorphous difference Patterson maps, and more de-
from cross-difference Fourier maps. Cross-difference Fourier
maps were also used to bring the Pt and Ir coordinates to a
common origin and, in conjunction with the use of anomalous
dispersion effects, to establish the correct enantiomeric ar-

0.005 0.010 0.015 0.020 0.025 0.030 refined by the phase-refinement method, which involves al-
ternate cycles of phase calculations and least-squares minimi-
B zation of lack-of-closure errors. The final model for the Pt
250+
distribution includes two sites which are about 3 A apart and
are within 3 A of the disulfide bridge between residues 12 and
200+ 65; this appears to be the only exposed disulfide bridge in the
molecule. Cross-difference Fourier maps of the Ir derivative
150+ displayed a single elongated column of electron density which
extended about 5-6 A in the c direction. We represented this
electron density by a 4-site Ir model. These Ir sites form several
100+ close contacts with three symmetry-related molecules of the
scorpion toxin. The center of the Ir distribution is about 6 A
50 +
from the c-amino group of lysine-13 and about 4 A from the
terminal hydroxyl group of tyrosine-58 of one molecule; about
6 5 4d 3A)3 5 A from the hydroxyl group of serine-9 from a second mole-
0 L
a
cule; and about 5 A from the c-amino group of lysine-18 from
0.0 0.005 0.010 0.015 0.020 0.025 0.030 a third molecule. Various statistics describing the final multi-
sin 0 /A2 ple-isomorphous-replacement (MIR) phasing are summarized
FIG. 2. Statistics for the Ir derivative (A) and the Pt derivative in Fig. 2. The figure-of-merit is 0.79 for the complete data
(B) as a function of resolution. 0, FH is the calculated heavy-atom set and 0.89 for data from the centric zones. The reliability
contribution to the structure factors; *, E is lack-of-closure error; and
0, AF/Fp is equal to IFpH Fp I/Fp where FpH and FP are the
-
index based on data from the centric zones [ IFH(obs) -
structure factors for the heavy-atom derivative and the native protein, IFH(calc)I |I/2FH(ObS), in which FH(ObS) = IFPH + Fp I ] is 0.52
respectively. for the Pt derivative and 0.51 for the Ir derivative.
 6498 Biochemistry: Fontecilla-Camps et al.
An electron density map was calculated including the 1361
native structure factors with d > 3 A and using centroid phase
angles with figure-of-merit weights. The map was contoured
on acetate sheets at a scale of 0.716 cm/A.
RESULTS
The electron-density map clearly revealed the molecular
boundaries, and most of the polypeptide chain could be traced
in this map. Many of the individual amino acids were identi-
fiable, particularly the proline residues and those with aromatic
side chains. By use of the published primary structure and di-
sulfide bridging patterns, we were able to follow much of the
sequence by inspection of the acetate sheets. Two of the disul-
fide bridges were located in very strong density, but the other
two were in much weaker density. A serious problem was en-
Proc. Natl. Acad. Sci. USA 77 (1980)
FIG. 3. a-Carbon backbone of
scorpion toxin. The numbered
positions coincide with a-carbon
atoms. The sulfur atoms are
shown, along with connections to
the a-carbon atoms of the cysteine
residues. This drawing and the one
in Fig. 5 were prepared by using
the computer program ORTEP
(22).
countered in efforts to trace the chain between residues 25 and
29 (Fig. 1). The electron-density map indicated a disulfide
bridge between residues 25 and 46, in contrast to the published
sequence which assigned a disulfide bridge between residues
27 and 46.
In order to help with the interpretation of this portion of the
map and to assist with assignments in other regions that dis-
played weak electron density, we constructed a 1.25 cm/A
Kendrew wire model by using an optical comparator (19). We
were unable to build a suitable model that corresponded to a
disulfide bridge between residues 27 and 46, but we obtained
an excellent fit by assuming that residue 25 is a cysteine which
forms a disulfide bridge to cysteine-46. In addition, the wire
model indicated that the COOH-terminal dipeptide is Ser-Cys,
rather than Cys-Ser as originally reported (13). The COOH-
FIG. 4. Stereo drawing of the a-carbon
backbone plus those side chains of the variant-3
toxin that correspond to conserved residues in
scorpion neurotoxins. Tyrosine-58 of the vari-
ant-3 toxin, marked by an asterisk, is not a con-
served residue but modification of the lysine
residue at this position in toxin I from Androc-
tonus australis Hector venom inactivates the
toxin.
 Biochemistry: Fontecilla-Camps et al.
terminal cysteine appeared to form a reasonable disulfide
crosslink to cysteine-12. Corroborating evidence for our as-
signments of the disulfide bridges was obtained by calculating
a Fourier map using native anomalous difference coefficients
and multiple-isomorphous-replacement protein phases, as
described by Strahs and Kraut (20). Of the five strongest peaks
in this map, four corresponded to the postulated disulfide
bridges.
At this point, our interpretation of the electron-density map
was heavily dependent upon postulated errors in the published
primary structure of the toxin, and we thought that the amino
acid sequence should be redetermined. Reexamination of the
primary structure for the first 60 residues of the toxin revealed
(unpublished data) that the sequence for residues 25-29 is
Cys-Asp-Thr-Glu-Cys rather than Asn-Thr-Cys-Glu-Cys as
originally reported; the remainder of the first 60 residues agreed
with the earlier report. This new sequence is consistent with the
electron-density map. We do not yet have chemical evidence
to support or refute our current assumption that residues 64 and
65 are Ser-Cys.
Atomic coordinates were measured from the Kendrew
model. These coordinates were refined by several cycles in
which parameter shifts were calculated by the gradient-cur-
vature technique from the 3 A Fourier map, alternated with
imposition of geometrical constraints to the model (21). During
this refinement the reliability index for the data with 3 A < d
< 7 A decreased from 0.50 to 0.40, and the geometry of the
molecule improved considerably. The model was then in-
spected and further improved at the University of North Car-
olina Department of Computer Science by use of the GRIP-75
computer graphics system. This improved model was further
refined by several additional cycles of gradient-curvature shifts
and geometrical constraints. The reliability index for the final
model based on the 3 A data is 0.36, and the average deviations
from ideal bond lengths are 0.07 A.
The a-carbon structure is shown in Fig. 3. The most promi-
nent secondary-structural features are two and a half turns of
a-helix involving residues 23-32 and a short three-strand stretch
of antiparallel 3-sheet, including residues 1-4, 37-41, and
46-50. The 13-sheet runs roughly parallel to the a-helix. The
a-helix and 1-sheet are connected by the disulfide bridges from
cysteine-25 to cysteine-46 and from cysteine-29 to cysteine-48.
A third disulfide bridge, between cysteine-16 and cysteine-41,
is also located near this region. The molecule contains bends that
involve residues 8-11, 18-21, 33-35, and 42-45, and a highly
contorted proline-rich stretch between residues 56 and 60.
There is also a sharp change in the course of the polypeptide
chain at position 50, which is a glycine residue.
DISCUSSION
The overall shape of the scorpion toxin molecule resembles a
right-hand fist: the a-helical segment runs along the knuckles
of the fist; the 1-sheet region lies along the second and third
joints of the fingers; the thumb is coincident with the loops
defined by residues 16-21 and 42-44; the wrist corresponds to
the COOH-terminal stretch of residues 52-65 and the loop
composed of residues 5-14; and the second joint of the little
finger is near the NH2 terminus of the molecule. The a-carbon
backbone displays a large flat surface that roughly coincides
with the plane defined by the COOH-terminus, the NH2 ter-
minus, and residue 43; in terms of the fist representation, this
flat surface lies along the second joints of the fingers and the heel
of the palm.
The complete amino acid sequences of 20 scorpion toxins
have been determined, along with partial sequences for several
others. If the sequences are aligned with the eight cysteine
Proc. Natl. Acad. Sci. USA 77 (1980) 6499
residues in register, and appropriate insertions and deletions
are permitted, it is apparent that there are several sections of
the scorpion toxins that contain invariant or semi-invariant
amino acids (1). The locations of these conserved residues in the
crystal structure of the variant-3 toxin are depicted in Fig. 4.
The entire stretch of six amino acids at the NH2 terminus of the
variant-3 toxin is conserved in most of the other scorpion toxins.
The residues in register with lysine-1 of the variant-3 toxin all
are lysine or arginine in the other scorpion toxins; those corre-
sponding to glutamic acid-2 are either aspartic acid or glutamic
acid in nearly all of the other toxins; the glycine-3 position is
occupied by glycine or alanine in all but one of the other se-
quences; the tyrosine-4 position is occupied by tyrosine in all
of the other scorpion toxins; the leucine-5 position corresponds
to leucine or isoleucine in all scorpion toxins except one which
contains alanine at this position; the valine-6 position is nearly
always valine or alanine, except for two examples in which it
is methionine or isoleucine.
Another conserved region lies near residues 45 and 47: the
alanine-45 position of the variant-3 toxin is occupied by alanine
in the other scorpion toxins, except for one toxin which contains
glycine at this position; the tryptophan-47 position is occupied
by either tryptophan or tyrosine in all of the other scorpion
toxins. A third stretch of conserved residues is found at positions
51-53 of the variant-3 sequence: leucine-51 is invariant; the
proline-52 position is occupied by proline in nearly all of the
other toxins; the glutamic acid-53 position is filled by an acidic
residue in all of the other sequences.
Conserved residues 5, 6, 45, and 51 are embedded within the
scorpion toxin molecule, and it seems likely that they must be
conserved in order to maintain the general three-dimensional
structure of the scorpion toxins. Most of the other conserved
residues are charged or aromatic amino acids which lie on the
flat surface of the molecule. At our current resolution, it appears
that the carboxyl group of glutamic acid-53 may be within
hydrogen-bonding distance to the amino group at the NH2
terminus; the carboxyl group of glutamic acid-2 may be hy-
drogen-bonded to the aromatic N-H group of tryptophan-47
which is in van der Waals contact with tyrosine-4. Conserved
residues leucine-51 and proline-52 might be expected to further
stabilize the orientation of glutamic acid-53. Thus, the con-
served amino acids that lie on or near the flat surface of the
molecule seem to form a stable array of exposed aromatic and
charged sidechains.
Chemical modification studies of toxin I from the scorpion
Androctonus australis Hector have shown that the toxin is in-
activated by modification of the lysine residue in register with
tyrosine-58 in the variant-3 toxin (23). This residue, which also
lies on the flat surface of the molecule, is included in Fig. 4.
Chemical modification studies of toxins I and II from A. aus-
tralis have also indicated that amino acid side chains at regions
away from the flat surface of the molecule can be modified
without concomitant loss of activity (24, 25). It seems likely that
this conserved surface of the molecule may be somehow in-
volved in the interactions that mediate the biological effects of
scorpion toxins.
The crystal structures of erabutoxin (26, 27) and a-cobratoxin
(28) have also been determined. These snake toxins have mo-
lecular weights similar to those of scorpion toxins, and they
contain four or five disulfide bridges. However, the snake toxins
act at the postsynaptic level (29), by binding to the acetylcholine
receptor, whereas scorpion toxins seem to act preferentially at
the presynaptic level (30). A major feature of the scorpion toxin
structure is the rigid portion that contains the a-helix and
A-sheet moieties plus three disulfide bridges. Erabutoxin and
a-cobratoxin also contain a dense, compact region in which four
 6500 Biochemistry: Fontecilla-Camps et al.
of their disulfide bridges are in close proximity. Like scorpion
toxin, these snake toxins have several open loops that extend out
of the dense crosslinked core. Except for this rather general
feature, there appears to be little similarity between the
three-dimensional structures of these snake toxins and the
structure of scorpion toxin. The a-helical segment, which is a
prominent feature of the scorpion toxin structure, is absent from
the snake toxins, and the size and orientation of the loops in
scorpion toxin seem to be quite different than in erabutoxin and
a-cobratoxin. Considering the evidence that scorpion toxins and
these snake neurotoxins display different biological mechanisms
of action, it is not surprising that the two types of molecules have
different three-dimensional structures.
We thank Drs. A. Bhown and J. Mole for redetermining the sequence
of the first 60 residues of the toxin. We gratefully acknowledge the use
of GRIP-75 developed by E. G. Britton, F. P. Brooks, Jr., J. Hermans,
J. S. Lipscomb, J. E. McQueen, M. E. Pique, and W. V. Wright. This
research was supported by National Institutes of Health Research
Grants CA-13148, DE-02670, and NS-12365 and by Grant IN-66Q
from the American Cancer Society. R.J.A. was the recipient of a
postdoctoral fellowship under National Institutes of Health Training
Grant CA-09128.
1. Rochat, H., Bernard, P. & Couraud, F. (1979) in Neurotoxins:
Tools in Neurobiology, eds. Ceccarelli, B. & Clementi, F. (Raven,
New York), Vol. 3, pp. 325-334.
2. Moss, J., Thoa, N. B. & Kopin, I. J. (1974) J. Pharmacol Exp.
Therap. 190, 39-48.
3. Gomez, M. V., Diniz, C. R. & Barbosa, T. S. (1975) J. Neurochem.
24,331-336.
4. Einhorn, V. F. & Hamilton, R. C. (1977) Toxicon 15, 403-
412.
5. Zlotkin, E., Miranda, F. & Lissitzky, S. (1972) Toxicon 10,
207-210.
6. Zlotkin, E., Miranda, F. & Lissitzky, S. (1972) Toxicon 10,
211-216.
7. Couraud, F., Rochat, H. & Lissitzky, S. (1980) Biochemistry 19,
457-462.
8. Koppenhofer, E. & Schmidt, H. (1968) Pfluegers Arch. 303,
133-149.
9. Narahashi, T., Shapiro, B. I., Deguchi, T., Scuka, M. & Wang,
C. M. (1972) Am. J. Physiol. 222,850-857.
Proc. Natl. Acad. Sci. USA 77 (1980)
10. Romey, G., Chicheportiche, R., Lazdunski, M., Rochat, H., Mi-
randa, F. & Lissitzky, S. (1975) Biochem. Biophys. Res. Commun.
64, 115-121.
11. Romey, G., Abita, J. P., Schweitz, H., Wunderer, G. & Lazdunski,
M. (1976) Proc. Natl. Acad. Sci. USA 73,4055-4059.
12. Catteral, W. A. & Ray, R. (1976) J. Supramol. Struct. 5,397-
407.
13. Babin, D. R., Watt, D. D., Goos, S. M. & Mlejnek, R. V. (1974)
Arch. Biochem. Biophys. 164,694-706.
14. Fontecilla-Camps, J. C., Suddath, F. L., Bugg, C. E. & Watt, D.
D. (1978) J. Mol. Biol. 123, 703-705.
15. North, A. C. T., Phillips, D. C. & Mathews, F. S. (1968) Acta
Crystallogr. A24, 351-359.
16. Tickle, I. J. (1975) Acta Crystallogr. B31, 329-331.
17. Lehman, M. S. & Larsen, F. K. (1974) Acta Crystallogr. A30,
580-582.
18. Blundell, T. L. & Johnsun, L. N. (1976) Protein Crystallography,
eds. Horecker, B., Kaplan, N. O., Marmur, J. & Scheraga, H. A.
(Academic, New York).
19. Richards, F. M. (1968) J. Mol. Biol. 37,225-230.
20. Strahs, G. & Kraut, J. (1968) J. Mol. Biol. 35,503-512.
21. Chambers, J. L. & Stroud, R. M. (1977) Acta Crystallogr. B33,
1824-1837.
22. Johnson, C. K. (1965) OR TEP, A Fortran Thermal Ellipsoid Plot
Program for Crystal Structure Illustrations, Report ORNL-3794,
revised (Oak Ridge National Laboratory, Oak Ridge, TN).
23. Sampieri, F. & Habersetzer-Rochat, C. (1978) Biochim. Biophys.
Acta 535,100-109.
24. Rochat, C., Sampieri, F., Rochat, H. & Miranda, F. (1972)
Biochimie 54,445-449.
25. Habersetzer-Rochat, C. & Sampieri, F. (1976) Biochemistry 15,
2254-2261.
26. Tsernoglou, D. & Petsko, G. (1976) Proc. Nati. Acad. Sci. USA
74,971-974.
27. Low, B. W., Preston, H. S., Sato, A., Rosen, L. S., Searl, J. E.,
Rukdo, A. D. & Richardson, J. S. (1976) Proc. Natl. Acad. Sci.
USA 73,2991-2994.
28. Walkinson, M. D., Saenger, W. & Maelicke, A. (1980) Proc. Natl.
Acad. Sci. USA 77,2400-2404.
29. Tamiya, H. & Arai, H. (1966) Biochem. J. 99, 624-630.
30. Zlotkin, E., Miranda, F. & Rochat, H. (1978) in Arthropod
Venoms, ed. Bettini, S. (Springer, Berlin), pp. 317-369.