Discovery and Structure of a Potent and Highly Specific Blocker of Insect Calcium Channels*

From the ‡Department of Biochemistry, University of Connecticut Health Center, Farmington, Connecticut 06032, Departments of §Pharmacology and §§Biochemistry, University of Sydney, Sydney, New South Wales 2006, Australia, ¶Institute for Molecular Bioscience and **Department of Biochemistry, University of Queensland, Brisbane, Queensland 4072, Australia, Department of Health Sciences, University of Technology, Sydney, New South Wales 2007, Australia, and ‡‡John Curtin School of Medical Research, Australian National University, Canberra, Australian Capital Territory 0200, Australia

Unfortunately, there are few well characterized peptide/protein toxins that lend themselves to these genomic approaches. Spider venoms can be viewed as preoptimized combinatorial libraries of insecticidal peptides, and therefore we decided to exploit these venoms in the search for insect-specific toxins suitable for engineering into plants and insect viruses. Here we describe a new family of insecticidal neurotoxins isolated by screening the venom of the lethal Australian funnel-web spider Hadronyche versuta (Fig. 1, inset). These toxins are the most potent blockers of insect voltage-gated calcium channels reported to date, but they are virtually inactive on vertebrate ion channels, making them ideal biopesticide candidates. The structure of one of the toxins reveals a compact, disulfide-rich core and a structurally disordered lipophilic extension that is essential for channel blocking activity.

EXPERIMENTAL PROCEDURES
Purification of Toxins-Funnel-web spiders were collected from the Blue Mountains west of Sydney (H. versuta), from Fraser Island, Queensland (H. infensa), and from the Illawarra region of New South Wales (Atrax sp. Illawarra). Lyophilized crude venom was fractionated using a Vydac C 18 analytical reverse phase high pressure liquid chromatography (rpHPLC) 1 column as described previously (6). Semi-pure -ACTX-Hv2a obtained from this initial fractionation was further purified on the same column using a gradient of 30 -48% acetonitrile over 35 min at a flow rate of 1 ml min Ϫ1 . Once purified to Ͼ98% homogeneity, peptides were lyophilized and stored at Ϫ20°C until further use. Cysteine residues were alkylated before sequencing (7).
Insect and Vertebrate Toxicity Assays-Insecticidal activity was tested by injecting peptides into house crickets (Acheta domesticus Linnaeus) as described previously (7). Vertebrate activity was assayed as described previously (8) using vertebrate smooth (vas deferens) and skeletal (biventer cervicis) nerve-muscle preparations; tissue contractions were recorded in the absence of additives or after injection of peptides directly into the bath buffer. -Atracotoxin-Hv1a (100 nM), a modulator of voltage-gated sodium channels (9), was used as a positive control. Vertebrate toxicity was determined by subcutaneous injection of -ACTX-Hv2a in 0.1 ml of saline into young BALB/c mice (3.1 Ϯ 0.2 g; n ϭ 3). Toxicity was monitored over 72 h.
Preparation of cDNA Libraries and RACE Analysis-Venom glands were dissected from a single subdued specimen of H. infensa (mature female) and Atrax sp. Illawarra (mature male), and mRNA was imme-* This work was supported by National Science Foundation Grant MCB9983242 (to G. F. K.), Australian Research Council grants (to G. M. N., M. J. C., and G. F. K.), and Postgraduate Research Scholarships (to X.-h. W., D. W., and H. I. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The

40306
New methods of insect control are urgently required due to the evolution of insect resistance to classical chemical pesticides (1), growing appreciation of the environmental damage caused by many agrochemicals, and increased public concern about the human health risks associated with prolonged insecticide exposure (2). One promising approach is to engineer plants to produce insect-specific toxins, as exemplified by the engineering of genes encoding insecticidal toxins from the soil bacterium Bacillus thuringiensis into a variety of agricultural cultivars (3). A potentially more selective method is to use insect-specific viruses as vectors to deliver toxins to a restricted This is an open access article under the CC BY license. diately isolated using a QuickPrep Micro mRNA Purification Kit (Amersham Pharmacia Biotech). For the H. infensa mRNA template, firststrand cDNA synthesis employed Superscript II reverse transcriptase (Life Technologies, Inc.) to extend a 3Ј universal poly(dT) anchor primer (NotI-dT 18 ; Amersham Pharmacia Biotech). Second-strand synthesis used DNA polymerase I. Marathon adapters (CLONTECH) were then ligated to the cDNA ends. For the Atrax template, full-length singlestranded cDNAs were obtained by including a 5Ј SMART II oligonucleotide (CLONTECH) in addition to the NotI-dT 18 primer.
5Ј-RACE of the H. infensa cDNA library employed a redundant primer based on the partial N-terminal sequence of mature toxin (5Ј-(AGTC)GT(AG)TT(AGTC)AC(AGTC)AC(AG)CA(AG)TC) and a 5Ј universal adapter primer (CLONTECH). Cloning and sequencing of the derived leader sequence allowed a gene-specific 3Ј-RACE primer (5Ј-gtggacgccATGAAATTTTCAAAGC) to be designed based on the 5Ј-untranslated region, translation start site, and N-terminal signal sequence. This gene-specific primer was used in conjunction with a 3Ј universal primer (Pacific Oligos) to amplify entire coding sequences from both the H. infensa and Atrax cDNA libraries. Final polymerase chain reaction products (450 -470 base pairs) were purified, cloned, and sequenced.
Electrophysiology-Neurons were dissociated from brains of adult European honeybees (Apis mellifera) as described previously (6). Adult C57B16/J mice of either sex were anesthetized with halothane and then killed by cervical dislocation. Trigeminal ganglion neurons were isolated by gentle trituration of the minced ganglia after a 20-min treatment at 34°C with papain (20 units/ml Ϫ1 ) in a HEPES-buffered saline solution of 140 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl 2 , 1.5 mM MgCl 2 , 10 mM HEPES, and 10 mM glucose, pH 7.3 (HBS).
Standard whole cell voltage clamp recordings (10) were made of bee brain calcium channel (I Ca ), sodium channel (I Na ), and potassium channel (I K ) currents and mouse sensory neuron I Ca and I Na at ambient temperature (22°CϪ24°C). For bee neurons, recordings were made with fire-polished borosilicate pipettes of ϳ6 megaohm resistance when filled with an intracellular solution of either of the following composi-tions: (a) 120 mM CsCl, 5 mM NaCl, 5 mM MgATP, 0.3 mM Na 2 GTP, 10 mM EGTA, 2 mM CaCl 2 , and 10 mM HEPES, pH 7.3 (for I Na and I Ca ), or (b) 130 mM KF, 10 mM EGTA, 2 mM CaCl 2 , and 10 mM HEPES, pH 7.3 (for I K ). For recordings of I Ca and I Na , the external solution consisted of 135 mM NaCl, 20 mM tetraethylammonium chloride, 5 mM CsCl, 5 mM BaCl 2 , 10 mM HEPES, 10 mM glucose, and 0.05% bovine serum albumin, pH 7.3. For I K recording, the external solution consisted of 130 mM NaCl, 20 mM KCl, 2.5 mM CaCl 2 , 1.5 mM MgCl 2 , 10 mM HEPES, 10 mM glucose, and 0.05% bovine serum albumin, pH 7.3. The same internal solution was used for recordings of mouse sensory neuron I Ca and I Na ; electrodes had a resistance of 1-2 megaohms. I Ca external solution contained 140 mM tetraethylammonium chloride, 2.5 mM CaCl 2 , 2.5 mM CsCl, 10 mM HEPES, 10 mM glucose, and 0.05% bovine serum albumin, pH 7.3, whereas I Na was recorded in HBS.
Neurons were voltage clamped at Ϫ90 mV, and currents were evoked by stepping the membrane potential from Ϫ60 to ϩ60 mV. Toxin effects on I Ca and I Na were tested at the potential with the largest inward current, usually Ϫ10 or 0 mV. In bee neurons, the peak inward currents were usually abolished by 100 M Cd 2ϩ , suggesting that the current was largely carried by Ca 2ϩ channels. In a few bee neurons, there was a rapidly activating, transient, and Cd 2ϩ -insensitive current that was blocked completely by tetrodotoxin (1 M). In mouse sensory neurons, the peak inward currents evoked in the presence of potassium and sodium channel blockers were abolished by 30 M Cd 2ϩ . The inward currents recorded in HBS consisted of both tetrodotoxin-sensitive and tetrodotoxin-resistant components. Toxin effects on bee brain I K were determined over a range of membrane potentials (from Ϫ40 to ϩ60 mV). Data were collected and analyzed as described previously (11).
Folding and Purification of Truncated -ACTX-Hv2a-A synthetic peptide (90% purity) encompassing residues 1-32 of -ACTX-Hv2a was purchased from Auspep (Melbourne, Australia). The reduced peptide (referred to hereafter as CT-Hv2a) was oxidized/folded at ambient temperature (22°C) in a glutathione redox buffer that promotes disulfide oxidation/shuffling (12). After 48 h, the reaction mixture was quenched with HCl and dialyzed against H 2 O using 1-kDa cutoff cellulose dialysis The inset shows a female specimen of H. versuta in an aggressive/defensive "ready-to-strike" stance with forelegs and palps raised and fangs exposed. b, comparison of primary structures of -atracotoxins from H. versuta (-ACTX-Hv2a), H. infensa (-ACTX-Hi2a and -ACTX-Hi2b), and Atrax sp. Illawarra (-ACTX-As2a and -ACTX-As2b). The -ACTX-Hv2a sequence was derived from protein sequencing, whereas the other primary structures are inferred from cDNA sequences (see the text). Residues boxed in yellow are identical in four or more of the sequences, whereas the strictly conserved cysteine residues are highlighted in red. The signal, propeptide, and mature peptide regions of the prepropeptide are indicated, and the percentage of amino acid identity within each of these regions is given in parentheses. The red arrowhead marks the site of additional C-terminal processing of the H. infensa toxins (see the text). The secondary structure of -ACTX-Hv2a, as determined in the current study, is shown below the sequences (␤-strands and 3 10 -helices are depicted as arrows and cylinders, respectively).
tubing (Membrane Filtration Products) to remove folding buffer components. The lyophilized dialysate was dissolved in H 2 O and then applied to a Vydac C 18 analytical rpHPLC column; fully oxidized CT-Hv2a was eluted with a retention time of 19 min using a gradient of 22-47% acetonitrile over 30 min at a flow rate of 1 ml min Ϫ1 .
Structure Calculations-NOESY cross-peaks were integrated in XE-ASY and converted to distance restraints (with pseudoatoms where appropriate) using CALIBA (14). Dihedral-angle restraints were derived as described previously (6). The intense intraresidue H ␣ ϪH N NOE for Arg-26, combined with a 3 J HNH␣ value of ϳ7 Hz, allowed its angle to be restrained to 50 Ϯ 40°for both peptides (15). H ␤ stereospecific assignments and 1 restraints for -ACTX-Hv2a were obtained using ECOSY-derived 3 J ␣␤ coupling constants in combination with H ␣ ϪH ␤ and H N ϪH ␤ NOE intensities measured from the 60-ms NOESY spectrum. Proline H ␤ protons were stereospecifically assigned as described previously (6). All X-Pro peptide bonds were clearly identified as trans on the basis of characteristic NOEs (16). The disulfide bonding pattern was determined unequivocally from preliminary structure calculations.
The torsion angle dynamics program DYANA was used to calculate 5000 structures from random starting conformations. The best 100 structures (selected on the basis of final penalty-function values) were then refined in X-PLOR (17). The 20 lowest-energy X-PLOR conformers were used to represent the solution structures of -ACTX-Hv2a and CT-Hv2a. MOLMOL (18) was used for molecular graphics. Fig. 1a shows a typical rpHPLC fractionation of crude venom from H. versuta. 50 fractions were individually assayed for insect and vertebrate toxicity. The late-eluting peak marked with an arrow caused immediate and sustained paralysis when injected into crickets (PD 50 ϭ 160 Ϯ 9 pmol g Ϫ1 ; mean duration of paralysis at a dose of 250 -500 pmol g Ϫ1 ϭ 4 -5 h). Injection of crickets with a second dose (250 -500 pmol g Ϫ1 ) of toxin before reversal of paralysis was lethal. The toxin was inactive in vertebrate smooth and skeletal nerve-muscle preparations at a concentration of 1 M (data not shown). These two neuromuscular preparations were chosen for the vertebrate toxicity screen because in combination they contain most of the potential neurophar- FIG. 2. -ACTX-Hv2a is a specific antagonist of insect voltage-gated calcium channels. a, whole cell calcium channel currents (I Ca ) recorded from a bee brain neuron in the absence (control) or presence of -ACTX-Hv2a. b, whole cell calcium channel currents recorded from a bee brain neuron in the absence (control) or presence of CT-Hv2a. c, dose-response curves for inhibition of I Ca in bee brain and rat trigeminal neurons by -ACTX-Hv2a (q and E) and -Aga-IVA (f and Ⅺ). Each data point is the mean Ϯ S.D. of 7-10 recordings. The curves are the result of fitting a simple logistic function to the data. d, time course for inhibition of I Ca recorded from a bee brain neuron after the addition of 1 and 10 nM -ACTX-Hv2a at the indicated times. Inhibition was rapid and was not significantly reversed by prolonged washing (indicated by the horizontal bar). macological targets of spider toxins such as ligand-and voltagegated ion channels. As further evidence of its lack of vertebrate toxicity, the toxin did not cause any adverse effects when injected into newborn mice at doses of up to 800 pmol g Ϫ1 , which is 5-fold higher than the PD 50 in crickets.

Isolation of a Novel Insecticidal Toxin-
Proteolytic digestion combined with N-and C-terminal sequencing revealed the complete amino acid sequence of this 45-residue toxin (Fig. 1b). Consistent with its long rpHPLC retention time, the toxin contains an unusually high proportion (55%) of apolar residues, including a highly hydrophobic Cterminal tail. We named the peptide -ACTX-Hv2a (Swiss-Prot accession number P82852) based on its molecular target (see below) and published nomenclature rules (11). The toxin has no homologs in the protein/DNA sequence data bases.
Elucidation of Precursor Structure-A peptide with a rpHPLC retention time similar to that of -ACTX-Hv2a was also evident in the venom of H. infensa. However, despite several attempts, N-terminal sequencing yielded only 12 residues (GVLDCVVNTLGC), and C-terminal sequencing indicated that the peptide had a blocked C terminus. Hence, we used RACE analysis (19) to extract the complete mRNA sequences corresponding to this toxin using cDNA libraries prepared from the venom glands of H. infensa and Atrax sp. Illawarra (see "Experimental Procedures").
Sequencing of RACE-derived clones revealed two 306-base pair coding sequences from H. infensa (corresponding to two 102-residue translation products, -ACTX-Hi2a and -ACTX-Hi2b) and two 300-base pair coding sequences from the Atrax species (corresponding to two 100-residue translation products, -ACTX-As2a and -ACTX-As2b). The DNA sequences have been deposited in GenBank TM (GenBank TM accession numbers AF329442-329445). The derived amino acid sequences (Fig.  1b) indicate that these peptides are homologs of -ACTX-Hv2a and reveal that the mature toxins are obtained by processing of a much larger prepropeptide precursor. The propeptide cleavage site was readily discerned from the known N-terminal sequence of -ACTX-Hv2a and -ACTX-Hi2a, whereas the signal peptide cleavage site was predicted using SignalP (20).
The prepropeptide architecture is similar to that determined for conotoxins (21,22) and provides the first circumstantial evidence that Australian funnel-web spiders have evolved a strategy similar to that of the cone snails for diversifying their toxin pool. The signal sequence is extremely well conserved (78% identity and 100% similarity if conservative substitutions are included; see Fig. 1b), whereas the mature peptide sequence is more diversified (53% identity). This finding is consistent with accelerated evolution (hypermutation) of the Cterminal region of the precursor to generate a library of functionally diverse toxins with identical cystine framework (21,22). It will be interesting in future studies to directly examine whether the venom contains families of functionally disparate toxins with the same signal sequence.
Mass spectral analysis of -ACTX-Hi2a (predicted oxidized mass ϭ 4408 Da; observed mass ϭ 4009 Da) indicated that it undergoes posttranslational deletion of the C-terminal four residues. C-terminal "trimming" has been noted for several spider (23,24) and scorpion (25) toxins. The scorpion toxin AaH II from Androctonus australis Hector undergoes posttranslational cleavage at a C-terminal Gly-Arg followed by an amidation process that eliminates the C-terminal glycine (25). Similar processing at the C-terminal Gly-Arg sequence in the H. infensa toxins would yield a toxin with the experimentally observed mass and would explain the block encountered during C-terminal sequencing of -ACTX-Hi2a. Mass analysis of the H. versuta and Atrax toxins indicated that their C termini are not trimmed, consistent with the absence of the C-terminal Gly-Arg sequence.
-ACTX-Hv2a Is a Potent and Specific Blocker of Insect Calcium Channels-Application of -ACTX-Hv2a (10 pM to 100 nM) to bee brain neurons inhibited calcium channel currents (I Ca ) in all cells examined (n ϭ 37; Fig. 2a), with maximum inhibition occurring at concentrations of Ͼ10 nM. The EC 50 for -ACTX-Hv2a inhibition of I Ca was ϳ130 pM (Fig. 2c). Inhibition was rapid at high concentrations and was not significantly reversed by prolonged washing (Fig. 2d). Application of -agatoxin (Aga)-IVA, the prototypic antagonist of vertebrate P-type voltage-gated calcium channels (26), also inhibited I Ca in all bee neurons examined (n ϭ 19), but the EC 50 (10 nM) and the concentration required for maximum inhibition (Ͼ100 nM) were both significantly higher than those for -ACTX-Hv2a (Fig. 2c).
In striking contrast to its effect on invertebrate neurons, superfusion of high concentrations of -ACTX-Hv2a (1 M; n ϭ 10) for 5 min had little effect on I Ca in mouse sensory neurons, whereas application of -Aga-IVA inhibited a component of I Ca in all mouse sensory neurons with an EC 50 of about 20 nM (maximum I Ca inhibition ϳ40%; Fig. 2c). -ACTX-Hv2a (100 nM) did not inhibit the tetrodotoxin-sensitive I Na of bee brain neurons (I Na was 98 Ϯ 4% of control; n ϭ 4), nor did it significantly affect I Na in mouse sensory neurons (I Na was 97 Ϯ 3% of control with -ACTX-Hv2a ϭ 1 M; n ϭ 5). -ACTX-Hv2a (100 nM; n ϭ 5) had no effect on bee brain I K at any potential when neurons were stepped from Ϫ90 mV to between Ϫ40 and ϩ60 mV.
We conclude that -ACTX-Hv2a is a potent and extremely specific blocker of insect voltage-gated calcium channels. Based on the data in Fig. 2c, we calculate that -ACTX-Hv2a has at least a 10,000-fold preference for insect versus vertebrate calcium channels.
Three-dimensional Structure of -ACTX-Hv2a-The solution structure of -ACTX-Hv2a purified from H. versuta venom was determined using standard homonuclear NMR methods (16). The ensemble of structures ( Fig. 3; Table I; Protein Data Bank accession code 1G9P) is highly precise with a backbone r.m.s. difference of 0.18 Å for the structured region (residues 3-32). According to PROCHECK (27), 75% of the non-Pro/Gly residues in the structured region lie in most favored sector of the Ramachandran plot, with the remaining 25% located in "additionally allowed" regions.
The three disulfide bridges in -ACTX-Hv2a form an inhibitory cystine knot motif (28) in which the Cys-17-Cys-29 disulfide passes through a 15-residue ring formed by the other two disulfide bridges and the intervening sections of polypeptide backbone (Fig. 3b). Although the N-terminal disulfide bridge of the inhibitory cystine knot motif does not generally contribute to the hydrophobic core of inhibitory cystine knot toxins and is not essential for formation of the basic inhibitory cystine knot fold (6,29), a complete cystine knot motif has been found in all four atracotoxin structures reported to date (this study and Refs. 6, 9, and 11). Presumably, the additional stability and protease resistance conferred by the complete knot (28) are critical for effective delivery of these neurotoxins to their sites of action.
Structural/Functional Homology with -Agatoxin-IVA-A search of the protein structure database using DALI (30) revealed weak but functionally significant structural homology between -ACTX-Hv2a and -Aga-IVA from the unrelated American funnel-web spider Agelenopsis aperta (Fig. 4a). We previously noted close structural/functional homology between the sodium channel modifiers ␦-ACTX from H. versuta and -Aga-I from A. aperta (6,9). Given the large evolutionary distance between these arachnids (Australian funnel-web spiders are primitive mygalomorphs, whereas American funnelweb spiders are modern araneomorphs), these results imply a remarkable case of convergent evolution.
In addition to the significant structural homology between the disulfide-rich domains of -ACTX-Hv2a and -Aga-IVA, both toxins have an unstructured, lipophilic C-terminal extension that was demonstrated to be critical for the activity of -Aga-IVA (31). To examine the functional role of the unstructured C-terminal domain in -ACTX-Hv2a (i.e. residues 33-45), we produced a synthetic peptide comprising only residues 1-32 of the parent toxin and determined its solution structure using NMR spectroscopy (Table I; Protein Data Bank accession code 1HP3). As expected, the truncated toxin has the same fold as the corresponding region of the full-length parent toxin (Fig.  4b). However, we found that the C-terminally truncated toxin did not inhibit insect calcium channels (Fig. 2b), nor did it competitively inhibit the activity of the native toxin (data not shown). Thus, we conclude that the lipophilic C-terminal extension is essential for interaction of -ACTX-Hv2a with insect calcium channels.

DISCUSSION
Insecticide Development-Most commonly used insecticides target voltage-gated sodium channels (e.g. DDT, pyrethroids), GABA receptors (e.g. cyclodienes and fipronil), or acetylcholinesterase (e.g. organophosphorus and carbamate insecticides) (32). This narrow target range has accelerated resistance development (1) and stimulated interest in the elucidation of new insecticidal compounds that act on novel targets. We have shown in this study that -ACTX-Hv2a acts on a nonconven- tional target, namely, insect voltage-gated calcium channels. Furthermore, this toxin appears to be the most potent blocker of these channels reported to date; its EC 50 on bee brain neurons (ϳ130 pM; this study) is significantly lower than that obtained for -Aga-IVA on bee (ϳ10 nM; this study) or cockroach (17 nM; Ref. 33) brain neurons. The unprecedented phylogenetic specificity of -ACTX-Hv2a significantly augments its utility as a lead compound for insecticide development, with our studies indicating that the toxin has at least a 10,000-fold preference for insect over vertebrate calcium channels. Whereas the toxin was inactive in all vertebrates tested in this study (chicken, rat, and mouse), we have found that -ACTX-Hv2a is toxic to a wide range of insect orders, including Leptidoptera, Diptera, and Orthoptera. 2 PLTX-II appears to be the only peptide toxin with comparable potency on insect calcium channels (34), but its ion channel and phylogenetic specificity remains to be determined.
Mode of Action-Surprisingly, despite a complete lack of sequence similarity, we found that -ACTX-Hv2a has weak but functionally significant structural homology with -Aga-IVA/B from the American funnel-web spider A. aperta. Both toxins contain a highly ordered disulfide-rich core and an unstructured lipophilic C-terminal region that protrudes from this globular domain. Both toxins have markedly reduced activity when the lipophilic C-terminal extension is deleted. Furthermore, we demonstrated that C-terminally truncated -ACTX-Hv2a does not competitively inhibit the activity of the fulllength toxin, suggesting that the disulfide-rich core does not bind the channel in the absence of the C-terminal tail. These results lead us to propose a possible model for the mode of action of these toxins.
It seems improbable that the structurally disordered C-terminal tails of -ACTX-Hv2a and -Aga-IVA make specific interactions with residues on the extracellular surface of voltagegated calcium channels. First, it is difficult to envisage how these lipophilic tails could make extensive favorable contacts with the largely polar surface of the channel. Second, the C-terminal apolar tail of -ACTX-Hv2a is a low complexity sequence, comprising a triple (G/P)G(L/I)(L/V) repeat, which seems unlikely to make specific high-affinity contacts with the channel surface. Third, in this model, immobilization of the C-terminal tail upon channel binding would incur a huge loss of conformational entropy, which is difficult to reconcile with EC 50 values in the picomolar range.
Thus, we suggest that -ACTX-Hv2a and -Aga-IVA share a similar mechanism of action in which the lipophilic tail does not make specific high-affinity contacts with the extracellular surface of the targeted calcium channel but rather initiates toxin binding by penetrating the membrane either adjacent to the channel or by intercalation between transmembrane segments of the channel protein (Fig. 4c). A similar model has been proposed for the mode of action of -Aga-IVB (35). Limited motion of the tail region within the membrane might minimize the loss of conformational entropy suffered by the toxin upon channel binding. We propose that anchoring of the C-terminal tail in the membrane somehow facilitates direct interaction of the disulfide-rich core region with the extracellular surface of the channel (i.e. the tail "targets" the structured region to the channel). One possibility is that binding of the C-terminal tail alters the channel conformation sufficiently to reveal a cryptic high-affinity binding site for the disulfide-rich portion of the toxin. In this model, C-terminal truncates would not be expected to bind the channel or act as competitive inhibitors of the wild-type toxin, which is what we observed experimentally.
While further experiments will clearly be required to test this hypothesis, it is salient to note that the insect calcium channel blocker PLTX-II from the spider Plectreurys tristis contains a C-terminal palmitoyl group that is essential for biological activity (36), and therefore it may function similarly. Thus, it will be instructive in future experiments to examine whether the C-terminal tail of -ACTX-Hv2a can be replaced by nonspecific hydrophobic anchors such as a palmitoyl group.  4. a, overlay of -ACTX-Hv2a (gold) on -Aga-IVA (green; Protein Data Bank accession code 1OAW); the core disulfide bridges of each toxin are shown as red and sage tubes, respectively. A similar comparison can be made with -Aga-IVB (Protein Data Bank accession code 1AGG). The folds are homologous, but loops 1 and 3 are more highly elaborated in -Aga-IVA/B. The molecules are rotated ϳ180°around the long axis of the ␤-hairpin relative to the view in Fig. 3. b, the mean 20 CT-Hv2a structure (cyan) superimposed for best fit over the backbone of residues 3-32 of the mean -ACTX-Hv2a structure (gold). The molecular orientation is similar to that in Fig. 3. c, hypothetical models of the mechanism of action of -ACTX-Hv2a and -Aga-IVA in which the lipophilic C-terminal tail (orange) penetrates the lipid bilayer either adjacent to the channel (top panel) or by intercalation between transmembrane segments of the calcium channel (bottom panel). In either case, this positions the disulfide-rich core (shaded sphere) for direct interaction with the extracellular surface of the channel.