Curtatoxins NEUROTOXIC INSECTICIDAL POLYPEPTIDES ISOLATED FROM THE FUNNEL-WEB SPIDER

Three polypeptide neurotoxins (curtatoxins) were isolated from the venom of the spider Hololena curta by reverse-phase high performance liquid chromatog- raphy, gel permeation, and ion-exchange chromatog- raphy. The purified toxins induced an immediate pa- ralysis in the cricket Acheta domestica that resulted in desiccation and death of the insect within 24-40 h (LDbO = 4-20 gg/g); this toxic effect is consistent with irreversible presynaptic neuromuscular blockade. carboxyl-terminal polypeptides of 36 38 identical positions among and an region flanked amino-and carboxyl- terminal ends. By analogy to other cysteine-rich ar- thropod venom proteins, the folded structure of the curtatoxins is likely important for their target speci- ficity and mode of action at the neuromuscular junc- tion.

In recent years the toxic components of spider venoms have attracted considerable attention for their use as probes of vertebrate and invertebrate nervous system functions (1,2). Spider venoms paralyze insects by causing a block in neuromuscular transmission which is mediated by glutamic acid receptors (3). Presently, the structure of only a few of these toxins has been determined. Argiotoxin from the orb-weaving spider Argiope lob&u was the first toxin structure to be elucidated (4) and since then other closely related toxins containing arginine and asparagine linked to polyamine chains have been reported (5)(6)(7). These toxins have been shown to cause a reversible block of the cation-selective, quisqualic acid-sensitive glutamate receptor of the locus (8). Venoms of other orb-weaving (Araneid) spiders contain low molecular weight postsynaptic toxins that also inhibit Lglutamate receptors and associated ion channels in the vertebrate central nervous system and at the invertebrate neuromuscular junction (9)(10)(11)(12)(13)(14)(15)(16)(17)(18). In contrast to the reversible paralysis induced by the orb-weavers, venom of other spiders such as the Clubionid spider Chiracanthium inclusum and the funnel-web (Agelenid) spider, Hololena curta, induce a potent irreversible blockade by venom components that act both preand postsynaptically (17,18). Presently, there is limited information on the structure and specificity of the neurotoxic components of Agelenid spiders. In search of potent inhibitors of neuronal functions, we have isolated and characterized three polypeptide isotoxins from the venom of H. curtu that induce irreversible neuromuscular blockade in the cricket A&eta domestica. EXPERIMENTAL PROCEDURES'

RESULTS
The initial purification of the toxic components from whole venom of H. curtu was performed by reverse-phase HPLC.* Fig. 1 shows the areas of the irreversible paralytic activity in the chromatogram. The active principles (curtatoxins I-III) that caused irreversible paralysis were in fractions 27,33, and 37. Anion-exchange chromatography of fraction 33 resulted in the elution of the paralytic activity in the unretained volume (Fig. 2). The toxic component of this fraction was further purified by gel permeation chromatography (Fig. 3). Final purification of this toxin was obtained by reverse-phase HPLC (Fig. 4) which yielded a single, symmetrical peak indicating a homogeneous component (curtatoxin II). The paralytic factor in fraction 37 was purified further by gel permeation chromatography (Fig. 5). Final purification of this toxin was obtained by reverse-phase HPLC ( Fig. 6) which yielded a single peak indicative of a homogeneous component (curtatoxin III). The toxic component of fraction 27 (curtatoxin I) was purified by microbore reverse-phase HPLC (data not shown) and found to represent a pure component as described below.
Curtatoxin Biological Activity-When injected into the thoracic cavity of crickets (A. domestica), curtatoxins produced a rapid and irreversible flaccid paralysis. For example, with 3.0 Kg of curtatoxin II or III the cricket usually desiccated and died after 24 h, whereas 1.0 pg (3.2 pg/g) resulted in paralysis of the insect after 30 min and death by 48 h. The LDsO for curtatoxin I was -20 fig/g, whereas the LD50 for curtatoxins II and III were = 4 pg/g. The LDbO of the whole venom was 0.5 pi/g (24.9 rg of protein/g). Purification and Structure of Curtatorins irreversible paralytic toxins were hybrid-toxins consisting of peptides with arginine or asparagine linked to polyamine chains was investigated by LC-MS (22). However, no such components were found. In addition, no unusual amino acids were found by microchemical sequence analysis or by ophthalaldehyde/g-fluoroenylmethylchloroformate (23) amino acid analysis. The quantitative Edman degradations on the pyridylethylated spider toxins are given in Table I. Unambiguous assignments were obtained for the first 35 residues of curtatoxin I. Further confirmation of its structure was obtained by sequence analysis of overlapping peptides from tryptic digests (data not shown). Cycle 36 of curtatoxin I indicated Asn as the carboxyl-terminal residue. This assignment was verified by FAB-MS analysis on the isolated carboxyl-terminal tryptic tripeptide Asn-Asn-Asn-NHz (see below), thus completing the sequence. As shown in Table III, FAB-MS analysis confirmed that the carboxyl terminus of curtatoxin I was amidated. The observed protonated molecular weight for the tripeptide (360.2) was identical to the theoretical value. The average protonated molecular weight determined for curtatoxin I was 4103.0 consistent with the theoretical value for the carboxyl-amidated neurotoxin. In addition, the observed molecular weight confirmed that all four disulfide bonds were intact.
The first 35 to 37 residues of curtatoxins II and III were determined by microchemical sequence analysis of the whole toxins. These toxins were highly similar, differing only at positions 9 and 14, i.e. Arg-9 and Ala-14 in curtatoxin II and Lys-9 and Phe-14 in curtatoxin III. The presence of Gln plus Glu at cycle 8 assigned Gln to this position in these two toxins. Cysteine was determined by analysis of the phenylthiohydantoin-derivative of pyridylethylated cysteine. However, pyridylethylation of the nonreduced curtatoxins yielded no phenylthiohydantoin-derivative of pyridylethylated cysteine showing that the 8 cysteine residues form 4 intramolecular disulfide bonds in these toxins. Repeated Edman degradations on these two intact toxins indicated serine at cycle 38 as their carboxyl-terminal residue. To verify the carboxylterminal sequence, pyridylethylated spider toxins II and III were degraded with CNBr to cleave at methionine 29 and the carboxyl-terminal fragments were purified by microbore Cl8 reverse-phase HPLC and sequenced. As is shown in Fig 7, three forms of the carboxyl-terminal fragment were obtained from each toxin; these peptides eluted between 17 and 22 min.
Each peptide had an identical sequence (Table I) and amino acid composition (Table II) Fig. 7). Although the presence of the carboxyl-terminal acid in these samples was estimated to be as much as 20%, the origin was presumed to be from hydrolysis of the amide during sample preparation.  (Table III). These values are consistent with carboxyl-terminal amidation and the presence of 4 intact disulfide bonds. In addition, the data also agree with the assignment of Gln over Glu at position 8 in curtatoxins II and III. Table II compares the amino acid composition of curtatoxins II and III determined from peptide hydrolyses and amino acid sequencing. With the exception of slightly low Cys determined by amino acid analysis, the compositions determined by both methods are in excellent agreement. Notable features of curtatoxins are the high content of Cys (8), Tyr (4)(5), Ser (3)(4)(5), acidic (4) and basic residues (3)(4), the presence of the two dipeptide sequences Cys-Cys and Tyr-Tyr and an absence of His, Thr, and Ile.4

DISCUSSION
This study describes the first complete purification and structure of neurotoxic venom components from the funnelweb spider H. curtu. Three curtatoxins were isolated by reverse-phase HPLC, anion-exchange and gel permeation chromatography. Purification was monitored with the cricket A. domestica using paralysis of the insect as the bioassay. These toxins represent the principle components in H. curta venom (see Fig. 1) responsible for irreversible paralysis and death of the insect upon envenomation. At LD,, doses (4 pg/g) of purified toxin, paralysis occurs within 30 min and death within 24-48 h. However, an LDsO dose of crude venom (24.9 pg/g) contains less than 0.1, 0.4, and 0.2 fig of curtatoxins I, II, and III, respectively, i.e. concentrations below their LDso values. That the potency of the whole venom is greater than that of the purified curtatoxins may be explained by synergism with reversible inhibitors possibly related to the argiotoxin family which have postsynaptic effects on the glutamate receptor (5,8), or, presence of other unidentified toxins. The amount of venom introduced by the bite of H. curta is not known and probably depends upon the type and size of the prey. However, the average amount of venom obtained from a single spider by electrophoretic milking, i.e., 0.1-0.2 ~1,~ agrees with the cricket toxicity data (LDsO = 0.16 ~1) for whole venom reported here.
The amino acid sequences of the curtatoxins are shown in Fig. 8. The difference in chromatographic behavior of curtatoxins II and III is attributable to the substitution of 2 amino acids at positions 9 (Lys -+ Arg) and 14 (Phe + Ala). The Lysg and Phe'" substitutions increase the hydrophobic character of curtatoxin III (24) and presumably account for its increased retention time during reverse-phase and gel permeation chromatography. Curtatoxin I has 36 amino acids as 4  compared to 38 for curtatoxins II and III and is less similar with deletions occurring at residues 1 and 17. The hydropathy (24) profiles of the curtatoxins are similar indicating a region about the Cys-Cys dipeptide that is less hydrophilic than the flanking ends of these molecules (not shown). When allowing for gaps in curtatoxin I, cysteines in all three toxins are conserved at identical sequence positions (Cys motifs) indicating highly similar folded structures. High cysteine contents in short arthropod polypeptide neurotoxins appear as a recurring theme with variation in the sequence position of these residues. In this context, we have observed both similarities and differences among various arthropod toxin Cys motifs relative to those of the curtatoxins. As is shown in Figs. 9 and 10, the Cys motif and hydropathy profile of the insectotoxin-I1 of the Middle-Asian scorpion Buthus eupeus (25) is similar to the insecticidal polypeptide neurotoxins of H. curta. Like curtatoxin I, insectotoxin-I, has 36 amino acids with 8 cysteines organized in a similar sequence fashion (Fig. 9). These findings suggest that these insecticidal toxins may have similar disulfide pairings and hence, folded configurations. This example may be contrasted to the different Cys motifs of the polypeptide neurotoxins with 8 cysteine residues such as the Australian funnel-web spiders Atrax robustus, Atrax versutus (26) and, the Ewing variant 3 scorpion neurotoxin Centroides sculpturatus (27) and toxin II of the North African scorpion Androctonus australis Hector (28). Among these arthropod neurotoxins, the curtatoxins, Atrax toxins, and scorpion toxins ( Fig. 9) form three distinct classes of Cys motifs. It is tempting to speculate that the different Cys motifs have a relation to the allowed molecular topologies (29) describing the stable tertiary structure of these toxins determined during the physical process of folding and disulfide formation. Moreover, the variation in folded configurations may be essential for the different receptor recognition specificites and biological modes of actions among these neurotoxins.
Of additional interest is the mechanism of the inhibitory action of the curtatoxins. Curtatoxins cause an irreversible flaccid paralysis consistent with a presynaptic blockade affecting neuromuscular function in the insect. Bowers et al. (30) reported the isolation of an irreversible presynaptic neurotxoin from H. curta venom consisting of two subunits of M, 7,000 and 9,000. Their work with abnormally excitable Drosophila mutants indicated that inhibition of transmitter release in the motor nerve terminal was due to a specific and direct effect on presynaptic calcium channels. Jackson et al. (31) reported an irreversible inhibitor of M, 5,000 to 10,000 from H. curta. In avian cochlear nucleus neurons the apparent site of action was the postsynaptic receptor-channel complex. Entwhistle et al. (32) isolated a pure, almost neutral polypeptide of M, 5,500-5,900 from Phoneutria nigriventer. This toxin was tested on an isolated locust femur preparation and was found to generate action potentials along the length of the axons in the crural nerve, resulting in rapid and uncontrolled twitching of the skeletal muscles. Its reported amino acid composition shows that it is not identical to the curtatoxins. The mechanism of action of these insecticidal neurotoxins from H. curta remains to be determined.