(Received for publication, October 22, 1996, and in revised form, February 26, 1997)
From the Department of Plant Sciences and the
Department of Molecular Microbiology and Biotechnology,
Faculty of Life Sciences, Tel-Aviv University, Ramat-Aviv 69978, Tel-Aviv, Israel, § IBSM-LIDSM-CNRS-UPR 9027, 31 Chemin
Joseph Aiguier, BP 71, 13402, Marseille Cedex 20, France, and the
¶ Laboratoire de Biochimie, CNRS URA 1455, Faculte de
Medicine, Secteur Nord, 13916 Marseille Cedex 20, France
-Neurotoxins from scorpion venoms constitute
the most studied group of modifiers of the voltage-sensitive sodium
channels, and yet, their toxic site has not been characterized. We used an efficient bacterial expression system for modifying specific amino
acid residues of the highly insecticidal
-neurotoxin Lqh
IT from
the scorpion Leiurus quinquestriatus hebraeus. Toxin
variants modified at tight turns, the C-terminal region, and other
structurally related regions were subjected to neuropharmacological
and structural analyses. This approach highlighted both aromatic
(Tyr10 and Phe17) and positively charged
(Lys8, Arg18, Lys62, and
Arg64) residues that (i) may interact directly with
putative recognition points at the receptor site on the sodium channel;
(ii) are important for the spatial arrangement of the toxin
polypeptide; and (iii) contribute to the formation of an electrostatic
potential that may be involved in biorecognition of the receptor site.
The latter was supported by a suppressor mutation (E15A) that restored
a detrimental effect caused by a K8D substitution. The feasibility of
producing anti-insect scorpion neurotoxins with augmented toxicity was
demonstrated by the substitution of the C-terminal arginine with
histidine. Altogether, the present study provides for the first time an
insight into the putative toxic surface of a scorpion neurotoxin
affecting sodium channel gating.
-Neurotoxins, the most abundant group in Buthidae scorpion
venoms, are polypeptides composed of a single chain of 63-65 amino acids, cross-linked by four disulfide bridges, and are responsible for
human envenomation (1, 2). They show variability in their apparent
toxicity to mammals and insects, in their primary structures, and in
their binding features to neuronal membrane preparations (3-6). Among
scorpion neurotoxins, the
-group is the most studied and is useful
in functional mapping of the sodium channel structure (6; reviewed in
Refs. 7-9). Despite the reported structures (10-13), chemical
modifications, and immunochemical studies (14-17) of various scorpion
-toxins, the structural elements dictating molecular recognition of
their binding site have not been characterized. It was postulated that
the molecule surface bearing a cluster of aromatic and hydrophobic
residues, termed the "conserved hydrophobic surface," was
associated with the toxic site (10, 13, 18). However, this postulation
has not gained experimental support.
Despite the differences in their primary structures and phylogenetic selectivity, scorpion neurotoxins affecting sodium channels are closely related in their spatial arrangements and form a compact globular structure that is kept rigid by four disulfide bridges (1, 19). A genetic approach based on comparative analysis of the variable regions may clarify distinct residues involved in toxin-receptor interactions. However, due to difficulties in producing sufficient amounts of recombinant toxins (20, 21), such an approach has been limited thus far.
Recently, we have established an efficient bacterial expression system
of a scorpion -neurotoxin, Lqh
IT, displaying an exceptionally high insecticidal activity (5, 6, 22). Lqh
IT slows the sodium
current inactivation in excitable membranes in a manner characteristic
of other scorpion
-toxins (22). Its binding to the receptor site on
neuronal membranes is competitively inhibited by the sea anemone toxin
ATXII and enhanced by veratridine (6, 23, 24). Lqh
IT binds to a
single class of high affinity sites on insect neuronal membranes
(Kd = 0.2-0.5 nM and 0.03-0.04 nM in locust and cockroach, respectively (6, 23)) and
competes weakly with other
-toxins for binding to rat brain
synaptosomes. AaHII, the most potent anti-mammalian scorpion
-neurotoxin, competes for 125I-Lqh
IT binding sites on
insect sodium channels, suggesting that the two scorpion
-toxins
bind to homologous, nonidentical receptor sites on insect and mammalian
sodium channels (6). Lqh
IT is toxic to mice at relatively high
concentrations as opposed to other scorpion
-neurotoxins such as
AaHII (5, 6), yet it appears to recognize an
-toxin binding site on
both insect and mammalian sodium channels. Using our functional
expression system, we found Lqh
IT very useful for a genetic study
(5) and for two-dimensional 1H NMR studies (25). Here we
report the results of modifications introduced at several regions of
Lqh
IT that highlight a putative molecular surface-mediating
recognition of the receptor site and intoxication.
Escherichia coli DH5a cells were used for plasmid constructions. E. coli BL21 cells lysogen with DE3 phage derivative bearing the T7 RNA polymerase gene under control of the lac promoter (26) were used for expression. A kanamycin-resistant derivative of pET-11c vector (26) was used for expression (5). Sarcophaga falculata blowfly larvae were bred in the laboratory. Albino ICR mice were purchased from the Levenstein farm (Yokneam, Israel).
Site-directed Mutagenesis, Functional Expression, and Purification of ToxinsSubstitutions K8A, K8D, Y10S, Y10W, and
K8D/N9D/Y10V were generated by using back-to-back primers and inverse
polymerase chain reaction (27) with pBluescript bearing LqhIT
cDNA (28). Engineering of the 5
and 3
termini of the Lqh
IT
cDNA for expression was described previously (5). Substitutions
E15A, F17G, R18A, G40S/K41P, and K8D/E15A were performed according to
the method of Deng and Nickoloff (29). Expression of the recombinant
polypeptides in a nonsoluble form, renaturation, and purification of
the active toxin were performed as described previously (5). Sequences of the mutated cDNAs were verified prior to expression with
Sequenase version II (U. S. Biochemical Corp.). Quantification of the
purified recombinant Lqh
IT variants and verification of their
composition was performed by amino acid analysis.
Four-day-old blowfly larvae (S. falculata; 100 ± 20 mg body weight) were injected intersegmentally. A positive result was scored when a characteristic paralysis (immobilization and contraction) was observed 5 min after injection. Nine larvae were injected with five concentrations of each toxin in three independent experiments. ED50 values were calculated according to the sampling and estimation method of Reed and Muench (30). Toxicity to mammals was determined by subcutaneous injection to female mice (20 ± 3 g).
Competition Binding ExperimentsPreparation of
radioiodinated LqhIT was performed according to Gordon and Zlotkin
(23). Cockroach (Periplaneta americana) synaptosomes
(P2L fraction) were prepared from the central nervous system by established methods (24, 31). The binding assays were
performed in the form of equilibrium competition assays using increasing concentrations of the unlabeled toxin in the presence of a
constant low concentration of the labeled toxin (30-50 pM) (6). Analyses of binding assays were carried out by using the iterative
computer program LIGAND (Elsevier Biosoft). Each experiment was
performed at least three times.
The
protein samples used for CD spectrum analyses were in 20 mM
phosphate buffer, pH 7. Spectra were measured at 20 °C in 0.05-mm
path length cuvette from 260 to 178 nm with a JOBIN-YVON (Long-Jumeau,
France) UV CD spectrophotometer (Mark VI). Calibration was performed
with (+)-10-camphorsulfonic acid. A ratio of 2.2 was found between the
positive CD band at 290.5 nm and the negative band at 192.5 nm. Data
were collected at 0.5-nm intervals with a scan rate of 1 nm/min. CD
spectra are reported as per amide. The protein concentration was
in the range of 0.5-1 mg/ml as determined on a Beckman amino acid
analyzer. The secondary structure content was determined according to
the method of Manavalan and Johnson (32).
Structural models were constructed using the Quanta-Charmm program by MSI Ltd. and Insight II-Discover software from MSI Technologies, Inc. (United Kingdom) running on a Silicon Graphics VGX R4000 Crimson Workstation. Models were built by the Homology module of Quanta on the basis of 59.4% identity to AaHII neurotoxin (10). The structures were minimized in vacuo using the Charmm force field. One hundred steps of minimization were performed using the steepest descent algorithm, followed by 5000 steps using the conjugate gradient method until a root mean square deviation of 0.001 was obtained. Electrostatic potentials of the unmodified and mutant toxins were calculated using Delphi software (MSI Technologies, Inc. (UK)).
Selection of sites to be modified was based on comparison between
two homologous -toxins, AaHII and Lqh
IT, displaying two extremes
in their phylogenetic preferences. AaHII, the strongest anti-mammalian
scorpion
-neurotoxin (6, 33), and Lqh
IT, the most insecticidal
toxin among the
-group (5, 6, 22), compete very poorly with each
other for their binding sites on insect or mammalian sodium channels.
This pharmacological difference can be attributed to nonhomologous
residues or structural motifs located on their surfaces. Site-directed
modifications were introduced to Lqh
IT to elucidate the molecular
surface involved in recognition of the receptor site on the sodium
channel. All toxin variants were purified by high performance liquid
chromatography using the toxicity assay on blowfly larvae as a quick
and direct measure of activity. Binding assays to a cockroach neuronal
membrane preparation were used as a measure of direct activity at the
receptor site. Possible structural alterations were assessed by CD
spectroscopy.
Comparison between the
three-dimensional structures and amino acid sequences of AaHII (10) and
LqhIT (5, 25) revealed major differences in the five-residue turn
(residues 8-12) and in the C-terminal region: (i) residues
Asp8-Asp9-Val10 in AaHII are
located on an external region of the toxin surface and are conserved in
a large group of
-toxins, whereas these positions are occupied by
Lys8-Asn9-Tyr10 in Lqh
IT (22,
28) (Fig. 1); (ii) the disposition of the C termini
relative to the five-residue turns shows a clear difference between
Lqh
IT and AaHII (10, 25).
We initiated the mutagenesis program by substituting residues
Lys-Asn-Tyr (positions 8-10) into Asp-Asp-Val (as appear at these
positions in AaHII) and produced the recombinant LqhIT variant using
our expression-reconstitution system (5). This modification affected
the activity of the toxin dramatically: the apparent affinity for the
insect receptor site decreased 21,000-fold, and the toxicity was
practically lost (Fig. 2A, Table
I). To determine whether this detrimental effect was due
to a conformational change of the overall structure, the mutant toxin
was analyzed by CD spectroscopy (Fig. 3A).
Despite some changes in the spectrum, the calculated secondary
structure content (32) was similar to that of the unmodified toxin.
This result suggested that the mutation might have generated a local
effect on the active molecular surface.
|
To identify the specific residue responsible for the detrimental effect, the following separate substitutions were conducted: (i) K8A and K8D to determine the significance of the positive charge at this position; and (ii) Y10S and Y10W to clarify whether a hydroxyl or an aromatic ring is crucial at this site. Neutralization of the positive charge at position 8 (K8A mutant) resulted in a 39.4-fold decrease in the apparent affinity for the receptor site on cockroach sodium channels and a 4-fold decreased toxicity to blowfly larvae (Fig. 2A, Table I) without a significant change in the CD spectrum (Fig. 3B). However, a severe effect was obtained when the charge at position 8 was inverted by replacing lysine with aspartate (K8D). The apparent affinity for the receptor site decreased 1611-fold, and less than 1% of the residual toxicity was determined (Fig. 2A, Table I). CD spectrum analysis of the K8D variant revealed a complete loss of the 190-nm band (Fig. 3A), suggesting changes in its secondary structure (see "Discussion").
In an attempt to visualize the impact of the charge inversion, we
compared the calculated electrostatic potentials of the unmodified, the
K8D, and K8D/N9D/Y10V mutant toxins (Fig. 4). The
unmodified toxin was found to be highly polar. One pole, containing the
N terminus, was negatively charged, and the other pole, containing Lys8, Arg18, Arg58,
Lys62, and the C-terminal Arg64, was positively
charged (Fig. 4, top left). The "bullet-shaped" electrostatic potential was severely disrupted in mutant K8D (Fig. 4,
top right) and only mildly disrupted in mutant K8A (data not shown). We investigated whether this disruption was due to the charge
inversion at position 8, causing a conformational change (Fig.
3A), or to alteration of the overall charge distribution of
the molecule by neutralizing the negative charge of the adjacent Glu15 (E15A mutant). This substitution was based on results
obtained by Delphi calculations with K8D and K8D/E15A computer models
suggesting a detrimental effect on the electrostatic potential caused
by two adjacent negative charges, i.e. Asp8 and
Glu15 (Fig. 4, top right). The E15A mutation
alone revealed no significant change in the binding affinity, a 4-fold
decrease in toxicity (Table I, Fig. 2A), no change in the CD
spectrum (Fig. 3A), and a nonsignificant change in the
electrostatic potential (data not shown). However, an E15A mutation in
addition to the K8D mutation (K8D/E15A double mutant) had no effect on
the CD spectrum, remaining similar to that of mutant K8D (Fig.
3A), but restored the electrostatic charge distribution to a
great extent (Fig. 4, top left). This restoration was
reflected by substantial increases in the apparent affinity for the
receptor site (22-fold) and toxicity (8.3-fold) of the double mutant
compared with the K8D mutant (Fig. 2A, Table I). These
results suggest that the electrostatic potential may be important for
the biological activity of LqhIT.
The possible role of the aromatic Tyr10, belonging to the five-residue turn, was investigated by two substitutions. A Y10W substitution had a minute effect on toxicity (Fig. 2B, Table I). However, replacement by serine caused a 114-fold decrease in the apparent affinity for the receptor site (Fig. 2B, Table I), and only 10% of the toxicity was detected (Table I). In light of the unchanged CD spectra of mutants Y10S (Fig. 3B) and Y10W (not shown), these results suggest that the aromatic side chain at position 10 may interact with the receptor site.
The 40-43 -turn is unique to
-scorpion neurotoxins and thus
could pertain to their characteristic pharmacology (19). In other
-toxins, residues 40 and 41 are serine and proline, respectively, whereas glycine and lysine occupy these positions in Lqh
IT. Still, the G40S/K41P mutation had a mild effect on toxicity. Although the
190-nm band decreased substantially in the CD spectrum of this mutant
(Fig. 3C), the apparent affinity for the receptor site
decreased only 25.7-fold, and 64% of the toxicity was preserved (Table
I). These results suggest that this tight turn does not play a critical
role in the insecticidal activity of Lqh
IT.
From the
solution structure of LqhIT (25) it was apparent that residues
Phe17 and Arg18 in the loop preceding the
-helix (residues 19-28) belong most likely to the molecular surface
common to the aforementioned turns. Phe17 is unique to
Lqh
IT, and Arg18 is found at this position in most
-neurotoxins with high activity against mammals (Fig. 1) (6, 34).
Phe17 was replaced by glycine, found in most
-toxins,
and Arg18 was substituted by alanine to examine the role of
its charge. The IC50 of mutant F17G was affected markedly
(160-fold increase), and 14% of the toxicity was determined (Fig.
2B, Table I). The apparent affinity of mutant R18A for the
receptor site on the cockroach sodium channels decreased 231-fold, and
22% of the toxicity was determined (Fig. 2B, Table I).
However, only minor changes in calculated electrostatic potentials
(data not shown) and CD spectra (Fig. 3B) were observed.
The residue adjacent to Arg18 is the first amino acid in
the -helix. In Lqh
IT, this position is occupied by aspartate
(Asp19), whereas asparagine is found at this position in
most
-neurotoxins (Fig. 1). Mutant D19N revealed no change in the
apparent affinity for the receptor site but some increase in toxicity
to blowfly larvae (Table I). Mutant D19H revealed a slight decrease in
activity. A double mutant, R18K/D19N, exerted some improved binding
affinity but a slight decrease in toxicity (Table I). Apparently, the negative charge at position 19 has no important role in the toxicity of
Lqh
IT.
The region connecting Cys12 (five-residue turn) with Cys63 (C-terminal region) by a disulfide bridge was shown to participate in an immunoreactive epitope whose blocking by specific antibodies inhibited the binding of AaHII to rat brain synaptosomes. Furthermore, Val10, Lys58, His54, and/or His64 were proposed to play a role in the interaction with a monoclonal antibody, which precluded binding of AaHII to the receptor site (35).
The possible involvement of the C terminus in the toxic site of
LqhIT was investigated by modifying the terminal Arg64
and the adjacent Lys62. The positive charge of
Arg64 was either neutralized (R64A, R64N) or inverted
(R64D). Furthermore, the terminal Arg64 was substituted by
histidine, found at the C terminus of several other
-toxins, and by
a histidine-arginine pair, as found in the cDNA sequence of
Lqh
IT (28). Lys62 was converted to leucine to assess the
significance of its positive charge. As shown in Fig. 2C and
Table I, the apparent affinity for the receptor site of mutants R64A
and R64N decreased, whereas their toxicity (Fig. 2C, Table
I) and CD spectra remained practically unchanged (not shown).
Orientation of their electrostatic potentials did not change, albeit
there was a reduction in the overall positive charge (data not shown).
Mutant R64D, however, revealed a marked decrease in the apparent
affinity for the receptor site and 20-fold reduced toxicity (Table I,
Fig. 2C). These effects were in concert with a marked
disruption of the electrostatic potential (Fig. 4) and a change in CD
spectrum (Fig. 3C). Another severe effect occurred with
mutant K62L (220-fold decrease in apparent binding affinity and 5-fold
reduced toxicity) (Table I, Fig. 2C). This effect was
accompanied by an increase in the 190-nm band of the CD spectrum (Fig.
3C) and a marked decrease in the positive component of the
electrostatic potential (not shown).
In contrast to the mutants showing reduced activity, mutant R64H revealed a 3.2-fold increase in toxicity and a 3-fold improved binding affinity for the receptor site (Fig. 2C, Table I). The CD spectrum of this variant remained similar to that of the unmodified toxin (Fig. 3B).
All toxin variants were injected to mice in parallel with the biological tests on insects. Interestingly, the changes in the insecticidal activity of each of the mutant toxins correlated with the changes in the anti-mammalian toxicity (not shown).
The genetic approach in the present study has become possible due
to an efficient expression-reconstitution system established recently
for the highly insecticidal scorpion -neurotoxin Lqh
IT and its
genetic variants (5). Interestingly, the yields of functional
polypeptide variants were similar to that obtained for the unmodified
toxin, varying between 2 and 5 mg of protein per liter of E. coli culture, and no correlation between the yield, bioactivity,
and/or CD spectrum was observed. Point mutagenesis highlighted amino
acid residues having a role in receptor site recognition, structural
arrangement, and formation of a polar electrostatic surface.
Since variations in
animal group specificity among toxins belonging to the same
neuropharmacological class are most likely related to structural
differences at their putative toxic sites, we directed the initial
modifications to such a variable motif, namely the five-residue turn of
LqhIT. This turn (residues 8-12) in Lqh
IT (KNYNC), the most
insecticidal scorpion
-toxin (6, 22), differs greatly from its
corresponding turn in AaHII (DDVNC), the strongest anti-mammalian
scorpion
-neurotoxin (6, 33). The profound decrease in toxicity and
binding affinity to the receptor site caused by the K8D/N9D/Y10V
mutation (Table I) suggested a major role for residues 8-10. At this
point it was important to clarify whether the K8D/N9D/Y10V mutation
affected the overall structure, which could consequently lead to loss
of activity, or whether it was a local effect on residues exposed to
the solvent and, thereby, disrupted direct interactions with the
receptor binding site. CD spectrum (Fig. 3A) and singular
value decomposition-derived data analyses of the mutant toxin revealed
a secondary structure content similar to that of the unmodified toxin,
suggesting preservation of the overall spatial arrangement of the
mutant toxin. The modifications performed with either Lys8
or Tyr10 were made to determine the specific residue
responsible for the detrimental effect obtained with the triple
mutation. The marked decrease in the apparent binding affinity and
toxicity obtained with the K8D mutation (Table I) implied a major role
for this residue. This detrimental effect could be due to either direct disruption of a structural motif or indirect alteration generated by
repulsion between Asp8 and Glu15, elimination
of a direct interaction with a recognition point at the receptor site,
and/or alteration of the polarity of the molecule. Analysis of mutants
K8A, E15A, and K8D/E15A provided data enabling discrimination among
these possibilities. The importance of a positive charge at this
position was shown by the decrease in toxicity of the K8A mutant (Table
I). Since the electrostatic potential and CD spectrum of the K8A mutant
changed only slightly, the decrease in activity did not result from
disruption in secondary structure or from elimination of the putative
interaction with the adjacent Glu15. It is more likely that
Lys8 interacts electrostatically with the receptor site,
and since this interaction is part of a multipoint interacting surface
(31, 36), its effect is rather slim.
However, inversion of the charge at position 8 (K8D mutant) caused a
severe effect on biological activity (Table I) in addition to a
complete disappearance of the 190-nm band in the CD spectrum (Fig.
3A). This band was found to correlate with either -helix and/or turn structures (37). We speculate that the changes observed in
the 190-nm band in the CD spectra of mutants K8D and G40S/K41P are
associated with structural alterations of tight turns alone. That is
because the modifications were introduced to turn structures distant
from the single
-helix of this molecule. On the basis of the
structural difference between AaHII (10) and Lqh
IT (25), we
speculate that this structural change might cause Asp8 to
form hydrogen bonds, as is found in AaHII (10), leading consequently to
a buried position in the mutant K8D. Furthermore, the electrostatic
potential of this variant was disrupted as well (Fig. 4). Restoration
of activity obtained by the E15A suppressor mutation while the CD
spectrum remained similar to that of the K8D mutant (Fig.
3A) suggested that neither the change in secondary structure
nor the negative charge at position 8 were fully responsible for the
detrimental effect caused by mutation K8D. The unchanged CD spectrum of
mutant K8D/E15A compared with that of mutant K8D also contradicted the
possibility that a repulsion between Asp8 and
Glu15 caused the alteration in secondary structure (Fig.
3A). It is possible that introduction of two negative
charges (Asp8 and Glu15) at the positive pole
disrupts the electrostatic potential of the molecule, whereas one
negative charge (Glu15 in the unmodified toxin, provided
that no interaction with Lys8 exists, or Glu15
in the K8A mutant, or Asp8 in the K8D/E15A mutant) is not
detrimental. This conclusion is supported by the results obtained with
the K8D/N9D/Y10V triple mutant. Three negative charges
(Asp8, Asp9, and Glu15) changed the
electrostatic potential of the toxin (Fig. 4); indeed, the apparent
affinity for the receptor site decreased 21,000-fold, and only 0.4% of
the residual toxicity was measured (Table I). These results may suggest
an important role for the electrostatic potential in the biological
activity of the toxin.
Comparison between the structures of the five-residue turns in AaHII
(10) and LqhIT (25) provides further insights regarding the critical
effect obtained by the K8D substitution. In AaHII, hydrogen bonds exist
between Asp8 and each of the residues Cys12,
Val10, and Asn11 (10). This network stabilizes
the five-residue turn and may influence the C-terminal stretch through
the Cys12-Cys63 disulfide bridge. In Lqh
IT,
however, Lys8 is exposed to the solvent, in contrast to the
buried Asp8 in AaHII. Thus, stabilization of the
five-residue turn in Lqh
IT is achieved differently. From the
solution structure of Lqh
IT (25), it is apparent that
Tyr10 interacts most likely with the terminal
Arg64, and Lys8 does not seem to interact with
Glu15 but rather with Tyr14. As mentioned
above, the lack of interaction between Lys8 and
Glu15 reflects the negligible change in the CD spectrum
obtained with mutant E15A (Fig. 3A). Thus, Lys8,
whose substitution by alanine had a moderate effect on toxicity, may
play a dual role by interacting electrostatically with the receptor
site and contributing to the polar electrostatic surface of
Lqh
IT.
The
importance of the aromatic side chain of Tyr10 for toxicity
was apparent from the unchanged toxicity of mutant Y10W compared with
mutant Y10S, which revealed a 110-fold decrease in the apparent binding
affinity for the receptor site (Table I, Fig. 2B). The possible interactions of Tyr10 with the C-terminal
arginine, inferred from the solution structure of LqhIT (25), may
suggest that the precise position of Tyr10 is important and
needs special stabilizing interactions.
Another aromatic residue showing importance for toxicity is Phe17. This is inferred from the substantial decrease in biological activity (Table I) without a significant change in CD spectrum (Fig. 3B) upon its substitution. Its location within the putative molecular surface common to the five-residue turn and the C terminus suggests that Phe17 may interact directly with the receptor site.
Examination of the spatial arrangement of LqhIT suggests that the
side chains of additional aromatic amino acids, such as Tyr14, Tyr21, or Trp38, are located
in the vicinity of the putative toxic surface and may be important for
toxicity. In accordance with this notion is the result obtained by
sulfenylation of Trp38 in AaHII, which decreased the
toxicity and apparent binding affinity for mammalian sodium channels by
more than 1 order of magnitude, whereas the CD spectrum of the modified
toxin remained unchanged (36).
Conversely, modifications of two aromatic residues that belong to the
conserved hydrophobic surface, i.e. sulfenylation of Trp45 in AaHII (the equivalent of Trp47 in
LqhIT) and substitution of Tyr49 by isoleucine in
Lqh
IT (5), had a negligible effect on toxicity to mammals and
insects, respectively. In both instances, the CD spectra of the
modified toxins changed dramatically (Ref. 36 and data not shown).
Thus, it is likely that the conserved hydrophobic surface may not be
involved in the toxic surface.
Our results
pinpoint several charged residues, in addition to Lys8 (see
above), that show significance for activity. One such residue is
Arg18, as shown by the R18A mutation. The CD spectrum of
this mutant (Fig. 3B) was similar to that of the unmodified
toxin, and its electrostatic potential was only moderately affected
(data not shown). Therefore, the decrease in activity of mutant R18A
(Table I) may suggest a direct electrostatic interaction of
Arg18 with the receptor site. Interestingly, substitution
of the adjacent Asp19 by asparagine had no effect on
toxicity, and a D19H mutation affected the activity only moderately.
The double mutant R18K/D19N also had a minor effect (Table I). These
results imply that residue 19, the first residue of the -helix
stretch (residues 19-28), may have no role in the interaction with the
receptor site. This result is in concert with the finding that
antibodies raised against the
-helix region of AaHII and the turn
comprising residues 27-30 were capable of binding to the toxin when
associated with the receptor site (14).
Superposition of the structures of AaHII and LqhIT reveals also that
the C termini of both toxins are positioned between the five-residue
turn and the
-turn formed by residues 40-43. However, their
dispositions relative to the five-residue turns differ in both toxins.
Based on these observations, we modified Lys62 and
Arg64 to investigate their contribution to the putative
toxic surface. Neutralization of the positive charge of the C-terminal
residue (R64A or R64N) decreased the apparent affinity for the receptor site more than 50-fold, although the toxicity remained practically unchanged (Table I). The discrepancy between the effects obtained in vivo, using blowfly larvae, and in vitro,
using cockroach neuronal membranes, may be related to the inherent
differences between fly and cockroach sodium channel structures.
Differences in sodium channel proteins of various insect central
nervous system membranes have been shown at the biochemical level (38,
39). Moreover, the ability of scorpion neurotoxins to distinguish among
sodium channel preparations of various insects has also been reported (6, 24, 39). The similarity between the CD spectra of mutant R64N (not
shown) and that of the unmodified toxin minimized the possibility that
mutation R64N caused a structural change affecting activity. More
likely is the suggestion that Arg64 interacts
electrostatically with the receptor site.
Inversion of the charge at the C terminus, mutation R64D, had a
dramatic effect: the apparent binding affinity decreased 2394-fold, and
the remaining toxicity was 5% (Table I). Such a detrimental effect can
result from a conformational alteration, as inferred from the altered
CD spectrum (Fig. 3C), the change of the electrostatic potential of the toxin (Fig. 4), or the perturbation of the
electrostatic contact with the receptor site. Except for the disruption
of the contacts between Arg64 and Tyr10,
Asp64 may generate a new intramolecular contact with a
nearby residue, leading to some structural change, as was suggested by
the altered CD spectrum of this mutant (Fig. 3C). Yet, this
putative alteration, reflected by the highly sensitive measurement of
the CD spectrum, had only a minute effect on the energy-minimized
-carbon model of the mutant toxin (Fig. 4).
Another positive charge related to the C-terminal domain is Lys62. Its putative interaction with the receptor site is shown by the 220-fold decrease in the apparent binding affinity to the cockroach sodium channel when substituted by leucine (Table I). The moderate change observed in the CD spectrum of mutant K62L (Fig. 3C) may suggest that this residue is involved in an intramolecular interaction as well.
In a previous study, biotinylation of Lys58 of AaHII
resulted in the most devastating effect obtained with chemical agents
applied on scorpion toxins affecting sodium channels (16). Only 1% of the original toxicity toward mammals was determined, and no detectable displaceability of the native toxin from the mammalian receptor site
could be monitored with the modified toxin. From the known structure of
AaHII (10), Lys58 was suggested to interact with
Asn11 and Gly61 by hydrogen bonding. This
network may be important for stabilizing the C terminus relative to the
five-residue turn. By analogy to LqhIT and its putative toxic
surface, we speculate that the prominent effect obtained in the
biotinylated AaHII could be due to a combination of structural
disruption, elimination of a putative electrostatic interaction with
the receptor site, and/or alteration of the electrostatic potential of
the molecule.
Our study pinpoints a putative molecular surface
involved in recognition of the receptor site that differs from the
conserved hydrophobic surface located on a different side of the
molecule (Fig. 5). The plausible suggestion of a
multipoint interaction of scorpion -neurotoxins with their binding
sites (31, 36) is in concert with our results, in which each
substitution of a residue participating in binding has a partial effect
on activity, whereas a structurally devastating mutation may cause a
more severe effect.
Binding to the receptor site may involve electrostatic as well as
hydrophobic interactions with opposing constituents at the receptor
site. This was demonstrated in other toxin receptor site interactions,
e.g. K+ channels and short scorpion neurotoxins
such as charybdotoxin, leiurotoxin, and PO5 (40-43), or the nicotinic
acetylcholine receptor and a sea snake toxin, erabutoxin a
(44). The most recent finding, showing that inversion of a negative
glutamate into a positive residue in the extracellular linker between
sodium channel transmembrane segments S3 and S4 in domain IV disrupts
the binding of an scorpion neurotoxin to rat brain sodium channels
(45), is in concert with our suggestion. The significance of the
electrostatic potential for the binding of a short scorpion toxin, Lq2,
to its receptor site was inferred from the inversion of a negatively
charged residue in a potassium channel, which affected association
rather than the dissociation rate of the toxin (46). From the
correlation between the changes in biological activities and
alterations of the putative electrostatic potentials of the various
modified toxins produced in the present study, the electrostatic
potential seems to play a role in the biological activity of the long
scorpion neurotoxin Lqh
IT. Whether this polarity is related to
association or dissociation kinetics of the toxin is still unknown and
deserves further study.
An alternative approach to the elucidation of the toxic site of neurotoxins affecting the cholinergic binding site was demonstrated by a combination of site-directed mutagenesis of isolated fragments of the receptor (47, 48) and NMR analyses of bound ligands (49).
Three types of animals, i.e. fly larvae, cockroaches, and
mice, were used in the bioassays. Injection to fly larvae provided a
useful and quick measure of the effects caused by the modifications and
throughout the purification process of each of the toxin variants. However, the effects of the mutations on the apparent binding affinity
to the cockroach central nervous system sodium channels were in most
instances more intense than the effects on toxicity to fly larvae. This
phenomenon may result from (i) differences in sensitivity of various
insect sodium channels from different tissues or animals to LqhIT
and to its modified toxin (6) or (ii) variations in susceptibility of
the toxin variants to proteolysis or availability to various tissues in
the whole animal that result from the chemical features of the toxin,
i.e. hydrophobicity or change in charge. Hence,
binding-competition studies using purified neuronal membranes are a
direct measure for toxin receptor site interactions on sodium channels
and thus provide a more sensitive assay than toxicity assays on whole
animals.
The similar changes in toxicity toward fly larvae and mice obtained
with the various mutants suggest that their corresponding receptor
sites are similar. Further modifications of residues that belong to the
putative recognition site may be useful for final determination of the
toxic site and for clarifying the molecular basis of animal group
specificity of scorpion -toxins.
We thank Dr. J. C. Fontecilla-Camps for providing the coordinates for the structure of AaHII; R. Oughideni and N. Zylber (CNRS Marseille) for the amino acid analyses; and Dr. B. Musafia for help with the modeling.