HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 276, Issue 28, 26568-26576, July 13, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/28/26568    most recent M102199200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Tedford, H. W. Articles by King, G. F. Search for Related Content PubMed PubMed Citation Articles by Tedford, H. W. Articles by King, G. F. Social Bookmarking                      What's this?

Functional Significance of the -Hairpin in the Insecticidal Neurotoxin -Atracotoxin-Hv1a^*

Hugo W. Tedford, Jamie I. Fletcher^§, and Glenn F. King^¶

From the  Department of Biochemistry, University of Connecticut Health Center, Farmington, Connecticut 06032 and the ^§ School of Biochemistry and Molecular Biology, University of Melbourne, Melbourne, Victoria 3010, Australia

Received for publication, March 12, 2001, and in revised form, April 17, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

-Atracotoxin-Hv1a is an insect-specific neurotoxin whose phylogenetic specificity derives from its ability to antagonize insect, but not vertebrate, voltage-gated calcium channels. In order to help understand its mechanism of action and to enhance its utility as a lead compound for insecticide development, we used a combination of protein engineering and site-directed mutagenesis to probe the toxin for key functional regions. First, we constructed a Hairpinless mutant in which the C-terminal -hairpin, which is highly conserved in this family of neurotoxins, was excised without affecting the fold of the residual disulfide-rich core of the toxin. The Hairpinless mutant was devoid of insecticidal activity, indicating the functional importance of the hairpin. We subsequently developed a highly efficient system for production of recombinant toxin and then probed the hairpin for key functional residues using alanine-scanning mutagenesis followed by a second round of mutagenesis based on initial "hits" from the alanine scan. This revealed that two spatially proximal residues, Asn²⁷ and Arg³⁵, form a contiguous molecular surface that is essential for toxin activity. We propose that this surface of the -hairpin is a key site for interaction of the toxin with insect calcium channels.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Public disquiet about the environmental and human health risks (1, 2) associated with chemical pesticides has stimulated the search for "environmentally friendly" pest control strategies. One approach is to engineer plants to produce insect-specific toxins, as exemplified by the engineering of genes encoding insecticidal -endotoxins 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 number of target insects without harming vertebrates and nontarget insects (4-6).

Unfortunately, there is a paucity of well characterized peptide toxins that lend themselves to these genomic approaches. Spider venoms can be viewed as preoptimized combinatorial libraries of insecticidal compounds,¹ and therefore we decided to exploit these venoms in the search for insect-specific peptide toxins. By screening the venom of the lethal Blue Mountains funnel-web spider Hadroncyhe versuta (8), we discovered three families of novel insect-specific toxins that appear to be excellent biopesticide candidates (9-11).¹ The first discovered and so far best characterized of these is the -atracotoxin-1 family (9, 10), of which the prototypic member is -atracotoxin-Hv1a (-ACTX-Hv1a).² The phylogenetic specificity of the toxin derives from the fact that it antagonizes insect, but not vertebrate, voltage-gated calcium channels (9).

Intriguingly, the venom of H. versuta contains at least six variants of -ACTX-Hv1a (10). These sequences, in combination with those of homologs from other funnel-web species (Fig. 1A), reveal that one of the best conserved regions of the toxin corresponds to a -hairpin that protrudes from the globular disulfide-rich core of the molecule. In this study, we examine the functional significance of the conserved -hairpin using a combination of protein engineering and site-directed mutagenesis. In contrast to numerous other disulfide-rich toxins being considered for biological pest control (12, 13), we found that -ACTX-Hv1a could be efficiently expressed in Escherichia coli as soluble, correctly folded protein. This expression system enabled us to perform an alanine-scanning mutagenesis of the entire hairpin region, which revealed a small number of strictly conserved and functionally critical residues that form a contiguous patch on one face of the -hairpin. We propose that this surface of the -hairpin is a key site for interaction of the toxin with insect calcium channels.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Insect Toxicity Assays-- Insecticidal activity was tested by injecting peptides dissolved in insect saline (14) into house crickets (Acheta domesticus Linnaeus, sex undetermined, mass 50-100 mg) as described previously (10). Control insects received injections of insect saline. LD₅₀ values (i.e. the dose that is lethal to 50% of animals) were determined by fitting dose-response data with the following equation,

(Eq. 1)

where y represents the percentage of deaths in the sample population at 48 h postinjection, x is the toxin dose in pmol g^-1, n is the variable slope factor, a is maximum response, and b is minimum response.

Purification of Synthetic and Native Toxin-- Synthetic -ACTX-Hv1a was purified and oxidized/folded as described previously (9). Native -ACTX-Hv1a was purified from the venom of H. versuta as described previously (9, 10).

Folding and Purification of Hairpinless Peptide-- Synthetic Hairpin-less peptide (see Fig. 1 for sequence) was purchased from Auspep (Melbourne, Australia). The reduced peptide was purified to >98% homogeneity using reverse-phase HPLC (rpHPLC) and then oxidized/folded by dilution into a glutathione (GSH) redox buffer that promotes disulfide oxidation/shuffling (15). After 72 h, the folding reaction was quenched by titration to low pH, and then the reaction mixture was dialyzed against H₂O using 1-kDa cut-off dialysis tubing (Membrane Filtration Products) to remove folding buffer components. The lyophilized dialysate was dissolved in H₂O, and then fully oxidized Hairpinless was purified using rpHPLC on a Vydac C18 analytical column (4.6 × 250 mm, 5-µm pore size). The yield of correctly folded peptide (i.e. peptide with the native disulfide pattern as determined by comparison of NMR spectra of the peptide with that of the native toxin) was ~50%.

NMR Spectroscopy-- An NMR sample was prepared by dissolving ~4 mg of Hairpinless peptide in 480 µl of water containing 5% (v/v) D₂O, 25 µM chloramphenicol, and 100 µM d₄-TSP (sodium 3-trimethylsilyl propionate). The pH was adjusted to 4.9, and the sample was passed through a Z-Spin microcentrifuge filter (Gelman Sciences) before being transferred to a 5-mm outer diameter NMR tube. NMR spectra were recorded at 298 K using a 5-mm ¹H probe on a Bruker Avance 600 MHz spectrometer. The WATERGATE module was used for water suppression (16). Two-dimensional DQF-COSY, NOESY (_m = 50 ms and 350 ms), and TOCSY (_m = 100 ms) spectra were acquired using the 5% D₂O sample. The sample was subsequently lyophilized and reconstituted in 480 µl of 99.96% D₂O (Sigma), and then the rate of amide-proton hydrogen-deuterium exchange was monitored by immediately acquiring a series of one-dimensional spectra and two-dimensional TOCSY spectra over a period of 22 h. Two-dimensional E.COSY and NOESY (_m = 350 ms) spectra were then acquired using the 100% D₂O sample. Spectra were processed using XWINNMR (Bruker). Chemical shift assignments were made using XEASY (17).

Structure Calculations-- NOESY cross-peak volumes were integrated in XEASY and converted to distance restraints (with pseudoatoms where appropriate) using the CALIBA macro within DYANA (18). Dihedral angle restraints were derived from ³J_HNH coupling constants measured from either high resolution one-dimensional NMR spectra or from inverse Fourier transforms of in-phase NOESY multiplets (19). Two additional  restraints of 100 ± 80° were applied for residues (Cys¹⁹, Asp²⁵) for which the intraresidue H-H_N NOE was clearly weaker than that between H_N and the H of the preceding residue (20).

Six H stereospecific assignments and seven ₁ restraints were obtained using ECOSY-derived ³J coupling constants in combination with H-H and H_N-H NOE intensities measured from the 50-ms NOESY spectrum. All three X-Pro peptide bonds were clearly identified as trans on the basis of characteristic NOEs (21), and all proline H protons were stereospecifically assigned on the basis of NOE intensities (22). The disulfide bonding pattern was determined unequivocally from preliminary structure calculations. Eight slowly exchanging amide protons were unambiguously assigned as hydrogen bond acceptors on the basis of preliminary structure calculations; corresponding hydrogen bond (i-j) restraints of 1.7-2.3 Å and 2.7-3.3 Å were employed for the HN_i-O_j and N_i-O_j distances, respectively.

The torsion angle dynamics program DYANA was used to calculate 5000 structures from random starting conformations. The program was configured to automatically remove duplicate restraints and restraints that could never be violated. The best 100 structures (selected on the basis of final penalty function values) were then refined in X-PLOR (23). The 20 lowest energy X-PLOR structures were used to represent the solution structure of Hairpinless. MOLMOL (24) was used for molecular graphics.

Construction of a Bacterial Overexpression System-- A synthetic gene encoding -ACTX-Hv1a was created by annealing and ligating a series of overlapping oligonucleotides containing codons optimized for maximal expression in E. coli. First, the following pairs of oligonucleotides (Life Technologies, Inc.) were annealed.

Each pair of complementary oligonucleotides was dissolved at a concentration of 10 µM in 50 mM NaCl, 25 mM Tris-Cl, pH 8, and then incubated at 95 °C for 35 min, followed by a 13-h incubation starting at 92 °C and ending at 17 °C, with 3 °C drops in temperature every 30 min. Annealing was verified by 3% agarose gel electrophoresis.

DsDNA from the pair 2 (Sequence 2) annealing reaction (222 µM) was then treated with 0.09 units/ml of T4 polynucleotide kinase (Life Technologies, Inc.) in 100 mM KCl, 10 mM MgCl₂, 1 mM ATP, 70 mM Tris-Cl, pH 7.6. 5'-Phosphorylated dsDNA was isolated from this reaction mixture by phenol chloroform extraction/ethanol precipitation and used in a ligation reaction with nonphosphorylated dsDNA from the pair 1 (Sequence 1) annealing reaction. The ligation reaction employed T4 DNA ligase (Life Technologies, Inc.) according to the manufacturer's recommendations. Ligation of the annealed duplexes (by virtue of the complementary overhangs in the N2 and C1 oligonucleotides) led to production of the full-length -ACTX-Hv1 gene. The synthetic gene was gel-purified from the ligation reaction and amplified with the following primers using standard PCR methods: N1 primer, 5'-ATATAGGATCCAGCCCGACCTGCAT-3'; C2 primer, 3'-TTGCAACGCTAACTCTTAAGGCCGG-5'. The bases in boldface type in the N1 and C2 primers encode 5'-BamHI and 3'-EcoRI sites, respectively, for cloning purposes. The PCR-amplified synthetic gene was digested with BamHI and EcoRI and subcloned using standard methods into BamHI/EcoRI-digested pGEX-2T (Amersham Pharmacia Biotech) to give the plasmid pHWT1, in which -ACTX-Hv1a is encoded as an in-frame fusion to the C terminus of Schistosoma japonicum glutathione S-transferase (with an intervening thrombin cleavage site).

Point mutations were introduced into the -ACTX-Hv1a gene using either (i) mutagenic PCR (with pHWT1 as template) using the N1 primer and a modified C2 primer incorporating the desired mutation or (ii) overlap PCR (with pHWT1 as template) using a pair of complementary, 25-base primers incorporating the desired mutation in addition to the original N1 and C2 primers.

Overproduction and Purification of -ACTX-Hv1a and Mutant Toxins-- E. coli cells with functional (BL21, trx⁺) or defective (AD494, trx^-) thioredoxin reductase were transformed with pHWT1 (either in its original form or with one or more engineered point mutations) for overproduction of glutathione S-transferase (GST)-toxin fusion protein. Cells were grown at 37 °C in LB medium to an A₆₀₀ of 0.6-0.8, and then expression of the fusion protein was induced by the addition of 150 µM IPTG. Cells were harvested by centrifugation 3-4 h after induction and lysed with a French press. The recombinant toxin was purified from the soluble cell fraction using affinity chromatography on GSH-Sepharose (Amersham Pharmacia Biotech) followed by on-column thrombin cleavage as described previously (25) except that phenylmethylsulfonyl fluoride and dithiothreitol were not present in the lysis buffer, and bovine thrombin (Sigma) was used instead of human thrombin. The recombinant toxin liberated by thrombin cleavage was eluted from the GSH-Sepharose column and dialyzed against water using 1-kDa cut-off dialysis tubing before being lyophilized and stored at 20 °C prior to the final purification step.

In order to resolve correctly folded recombinant toxin from nonnative disulfide bond isomers and other contaminants, the affinity-purified toxins were further purified using rpHPLC on a Vydac C₁₈ analytical column (4.6 × 250 mm, 5-µm pore size). Affinity-purified toxins were eluted at a flow rate of 1 ml min¹ using an initial linear gradient of 12.5-20% acetonitrile over 20 min. Properly folded toxins were recovered as the first major peak (retention time of 9-12 min depending on the toxin being purified); the identity of the toxin was confirmed using electrospray mass spectrometry.

CD Spectroscopy-- CD spectra were recorded at 4 °C using a Jasco J-715 spectropolarimeter calibrated with D-(+)-camphor-10-sulfonic acid. Spectra were recorded using toxins (20 µM) dissolved in 1 mM sodium phosphate, pH 6.5, in a 0.1-cm rectangular quartz cell. Spectra consisted of five scans acquired with a scan rate of 20 nm min¹ and a response time of 4 s. A spectrum of the buffer recorded under the same conditions was subtracted from the spectrum of each of the toxins.

Data Base Files-- Coordinates and experimental restraints for the ensemble of 20 Hairpinless structures have been deposited in the Protein Data Bank (PDB accession code 1HVW), and ¹H chemical shifts have been deposited in BioMagResBank (accession number 4937).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Engineering a Hairpinless Mutant-- -ACTX-Hv1a is a member of the inhibitory cystine knot family of toxic polypeptides (9, 26, 27). The three disulfide bonds form a pseudoknot, thereby providing the toxin with enhanced stability and resistance to proteases (27). The cystine knot toxins essentially comprise a cystine framework onto which four loops are tethered, each bounded by half-cystines. -ACTX-Hv1a differs from the well known conotoxin members of this family (28) in that one of the loops forms an unusually long -hairpin (residues 22-37) that is highly conserved among the -atracotoxin-1 family (Fig. 1A). This hairpin is almost as structurally well ordered as the cystine knot "core"; in the ensemble of 20 deposited structures (Protein Data Bank code 1AXH) the -hairpin (residues 22-37) and disulfide-rich core (residues 4-21) have backbone r.m.s. deviation values, relative to the mean structure, of 0.16 ± 0.05 and 0.13 ± 0.05 Å, respectively.

View larger version (40K): [in this window] [in a new window]   Fig. 1.   Structures of -atracotoxin-1 and Hairpinless. A, primary structures of Hairpinless and all known members of the -atracotoxin-1 family (7, 9). Residues identical to those in -ACTX-Hv1a are boxed in yellow, and conservative substitutions are colored red. The previously determined secondary structure of the prototypic family member -ACTX-Hv1a (9) is indicated above the sequences (-strands 1 and 2 are depicted as blue arrows, and -turns 1-5 are shown as green boxes). Note the well conserved -hairpin (residues 22-37). The three strictly conserved disulfide bonds that form the cystine knot motif are shown in red below the sequences. B, schematic design of the Hairpinless mutant. The bulk of the -hairpin (residues 24-34) and the three unstructured N-terminal residues were excised from -ACTX-Hv1a (left panel), and the small residual -strands were linked with a Gly-Gly dipeptide (middle panel) to create a Hairpinless mutant (right panel). C, stereo view of the ensemble of 20 Hairpinless structures superimposed for best fit over the backbone atoms of the mean coordinate structure. Disulfide bonds are shown in red and labeled. D, superposition of the mean Hairpinless structure (blue) on the corresponding region (residues 4-23 and 35-37) of the mean coordinate structure of the full-length toxin (green; Protein Data Bank file 1AXH). The backbones of these structures overlay very closely with a root mean square deviation of 0.67 Å.

In order to examine the overall functional significance of the -hairpin in -ACTX-Hv1a, we attempted to construct a hairpinless mutant by excising the bulk of the -hairpin without affecting the residual fold of the disulfide-rich core. Detailed analysis of the three-dimensional structure of -ACTX-Hv1a (Protein Data Bank code 1AXH; Ref. 9) revealed that the solvent-accessible tip of the hairpin interacted minimally with the disulfide core, making it probable that excision of this region would not affect the folding or stability of the core region. Thus, we reasoned that residues 24-34 of -ACTX-Hv1a, comprising the bulk of the hairpin, could be replaced by a Gly-Gly linker (Fig. 1B). This leaves the base of the hairpin intact, which is essential to maintain the Cys¹¹-Cys²² and Cys¹⁷-Cys³⁶ disulfide bonds that form part of the hydrophobic core of the toxin (9). The Gly-Gly linker was predicted to induce formation of a -turn at the tip of the small residual -hairpin. We previously showed that excision of the three N-terminal residues caused a small (15-fold) but not critical reduction in biological activity (10), and hence these residues were also excluded from the final construct.

The designed "Hairpinless" peptide ([residues 4-23]-Gly-Gly-[residues 35-37]); see Fig. 1A) was chemically synthesized, oxidized/folded in a glutathione redox buffer, and purified using rpHPLC, and its solution structure was determined using standard homonuclear NMR methods (21). The structure (Fig. 1C, Table I) is well defined with a backbone r.m.s. deviation of 0.20 Å. There are no distance or dihedral angle restraint violations greater than 0.12 Å and 2°, respectively. PROCHECK analysis (29) revealed that 67% of non-Gly, non-Pro residues lie in the most favored regions of the Ramachandran plot, with the remainder in "additionally allowed" regions. The coordinates for the family of 20 structures have been deposited with the Protein Data Bank (accession code 1HVW).

PROMOTIF (30) analysis indicated that Hairpinless contains -turns at residues Pro³-Gln⁶ (type II), Tyr¹⁰-Asn¹³ (type IV), and Cys¹⁴-Gln¹⁷ (type I), which are topologically equivalent to -turns Pro⁶-Gln⁹, Tyr¹³-Asn¹⁶, and Cys¹⁷-Gln²⁰ previously identified in the full-length toxin (9). As predicted, the base of the hairpin is intact, with residues Cys¹⁹-Thr²⁰ and Arg²³-Cys²⁴ forming -strands (blue arrows in Fig. 1D) and the Gly²¹-Gly²² linker inducing formation of a classical type I' -turn (residues Thr²⁰-Arg²³) at the tip of the hairpin.

Activity of the Hairpinless Mutant-- The Hairpinless structure overlays very closely on the corresponding region (residues 3-23 and 35-37) of the structure of the full-length toxin (Fig. 1D). This demonstrates that the Hairpinless design was successful, since this mutant faithfully reproduces the structural details of the disulfide-rich globular core of the parent toxin. Hence, the activity of the Hairpinless mutant should provide an indicator of the functional importance of the excised -hairpin. As seen in Table II, while the native toxin has an LD₅₀ of 269 pmol g¹ in house crickets, the Hairpinless mutant was completely inactive at all concentrations tested up to 91,500 pmol g¹, making it at least 340-fold less potent than the native toxin. Thus, we conclude that the -hairpin is critical to the insecticidal activity of -ACTX-Hv1a.

View this table: [in this window] [in a new window]   Table II Insecticidal activity of -ACTX-Hv1a and mutants thereof

Development of a Recombinant Expression System for -ACTX-Hv1a-- We decided to develop a recombinant expression system for -ACTX-Hv1a so that the key functional residues in the -hairpin could be precisely identified using site-directed mutagenesis. Like most insecticidal toxins being considered for genomic approaches to pest management (4, 31), -ACTX-Hv1a is highly disulfide-bridged, and therefore expression of this toxin in a foreign host such as E. coli requires consideration of how proper oxidation/folding might be accomplished.

Disulfide bonds are only formed in the E. coli cytoplasm in enzymes such as ribonucleotide reductase during their catalytic cycles or in the oxidative response transcription factor OxyR during its regulatory cycle (32). The thioredoxin and glutaredoxin disulfide-reducing pathways ensure that all other cytoplasmic cysteine residues are in the reduced state (32, 33). However, it has been demonstrated that E. coli strains with defective thioredoxin reductase can form cytoplasmic disulfide bonds (34), apparently because they accumulate oxidized thioredoxins that promote disulfide bond formation (32). Thus, we initially attempted to express a fusion of -ACTX-Hv1a to the C terminus of GST in the thioredoxin-deficient E. coli strain AD494 (Novagen).

SDS-polyacrylamide gel electrophoresis analysis (Fig. 2A) revealed that the GST-toxin fusion protein was efficiently expressed in AD494 in a predominantly soluble form (compare lanes 2 and 3) and that following cell lysis the fusion protein could be quantitatively bound to a glutathione-agarose affinity column (compare lanes 4 and 5). We cleaved the fusion protein with thrombin on the affinity column (see lane 6) and eluted the liberated toxin with buffer. -ACTX-Hv1a has six cysteine residues that can be paired in 15 different ways to form three disulfide bonds; thus, if the cysteines pair in a completely random manner, the yield of native disulfide isomer will be only ~7%. In order to determine the yield (if any) of correctly oxidized toxin, the eluate from the affinity column was analyzed using rpHPLC. The major component in the affinity column eluate (see upper trace in Fig. 2B) had a rpHPLC retention time very similar to that of synthetic -ACTX-Hv1a (see lower trace in Fig. 2B), and hence we tentatively identified this peptide as correctly oxidized/folded recombinant -ACTX-Hv1a.

View larger version (42K): [in this window] [in a new window]   Fig. 2.   Purification and characterization of recombinant toxins. A, SDS-polyacrylamide gel illustrating the purification of recombinant toxin from E. coli AD494/pHWT1 cells. Lane 1, molecular weight standards (masses in kDa shown on left of gel); lanes 2 and 3, insoluble and soluble fractions, respectively, resulting from centrifugation of cell lysate; lane 4, eluate obtained from application of soluble fraction to GSH-Sepharose affinity column; lane 5, GSH-Sepharose beads after washing showing relatively pure GST-toxin fusion protein bound to the beads (marked with an arrow); lane 6, GSH-Sepharose beads after incubation with thrombin and elution of liberated toxin with buffer. The complete disappearance of the GST-toxin fusion band and the appearance of a GST band (marked with an arrow) indicates that the proteolytic cleavage reaction has gone to completion. B, the upper trace (which is offset vertically by 0.7 absorbance units) is an rpHPLC chromatogram of eluate from the GSH-Sepharose affinity purification. The major peak, which corresponds to correctly folded recombinant -ACTX-Hv1a, had almost identical retention time to that of synthetic -ACTX-Hv1a (lower trace) prepared as described previously (9). Minor peaks from the eluate correspond to misfolded toxin and non-toxin-related contaminants. The horizontal bar below the upper trace indicates the toxin fraction that was subsequently used for structure-function analyses. Both rpHPLC separations employed a Vydac C18 column with a flow rate of 1 ml min1 and an acetonitrile gradient of 12.5-20% over the first 20 min, 20-95% over the next 7 min, and then isocratic separation at 95% acetonitrile for a further 5 min. C, the most downfield portion of the amide proton region of the one-dimensional 1H NMR spectra of native (upper panel) and recombinant (lower panel) toxin. This region contains resonances from residues that span almost the entire length of the toxin (Ile5-Asp37). The chemical shifts for these residues are almost identical in the two peptides, indicating that the recombinant toxin possesses the native fold. D, log dose-response curves resulting from injection of native recombinant toxin () as well as V33A (), R35A (), N27A (), and N27A,R35A () mutants into A. domesticus. Each data point represents the mean of two or three independent experiments. The solid lines represent the fit of Equation 1 to the data, which yielded the LD50 values given in Table II.

In order to confirm that the recombinant toxin had the native disulfide-bond pattern, we acquired one-dimensional ¹H NMR spectra of rpHPLC-purified recombinant and native toxins. The one-dimensional NMR spectrum of any protein can be considered a structural fingerprint that is extremely sensitive to even minor structural and environmental perturbations. The NMR spectra of the native and recombinant toxins were almost identical, with the exception of a few additional peaks from the N-terminal Gly-Ser extension (remnants of the thrombin cleavage site) in the spectra of the recombinant toxin. One of the most characteristic features of the ¹H NMR spectrum of native -ACTX-Hv1a (9) is a series of highly downfield-shifted amide-proton resonances with chemical shifts between 9 and 10 ppm. As shown in Fig. 2C, these shifts are identical in the native and recombinant toxins, confirming that the recombinant -ACTX-Hv1a is correctly folded. Further evidence that the recombinant toxin is correctly folded comes from the fact that its lethality to house crickets (LD₅₀ = 269 ± 13 pmol g¹; Fig. 2D) is, within experimental error, indistinguishable from that of native toxin (LD₅₀ = 290 ± 15 pmol g¹; see Table II).

We used mass spectrometry to identify which peaks in the rpHPLC chromatogram corresponded to recombinant toxin (both correctly and incorrectly folded) and unrelated contaminants (Fig. 2B, upper trace). The yield of correctly folded recombinant -ACTX-Hv1a (as a percentage of total recombinant toxin observed in the rpHPLC trace) was estimated from integration of the relevant HPLC peaks to be 65-70%. The high yield of correctly folded toxin was somewhat unexpected for several reasons. First, even in thioredoxin reductase-deficient strains of E. coli, the yield of correctly oxidized cytoplasmic alkaline phosphatase (which has two disulfide bonds) is only 25-50% (32). Second, in vitro oxidation of a synthetic -ACTX-Hv1a peptide gives a yield of only ~15% of the native disulfide isomer (9). Hence, we would not expect spontaneous and/or thioredoxin-catalyzed oxidation of cytoplasmic -ACTX-Hv1a to give such a high proportion of the native disulfide isomer.

In order to examine the contribution of thioredoxin-catalyzed oxidation to the final yield of correctly folded recombinant -ACTX-Hv1a, we transformed BL21 cells, which contain functional thioredoxin reductase, with the GST fusion plasmid (pHWT1) and repeated the experiments outlined above. We found no difference in the levels of overproduction of the GST-toxin fusion protein and no difference in the yield of correctly folded recombinant -ACTX-Hv1a (data not shown). Hence, it would appear that thioredoxin-catalyzed oxidation makes no significant contribution to the yield of correctly folded recombinant -ACTX-Hv1a in the AD494 strain. We postulate that most oxidation of -ACTX-Hv1a occurs after the cells have been lysed and the fusion protein is exposed to the periplasmic Dsb system, which normally catalyzes disulfide formation in E. coli (35). Active involvement of the Dsb system, in which DsbC can catalyze disulfide isomerization, may explain why the yield of correctly folded protein is so high.

Future experiments with a panel of dsb and trx mutants may help to define the exact mechanism of -ACTX-Hv1a oxidation in the E. coli expression system. However, regardless of the mechanism of toxin oxidation, it is clear that the expression system we have developed provides a very efficient means of producing recombinant toxin and furnishes us with a facile method for producing site-directed mutants.

Alanine-scanning Mutagenesis of the -Hairpin of -ACTX-Hv1a-- The -hairpin of -ACTX-Hv1a comprises residues Cys²²-Asp³⁷. In order to ascertain the relative functional importance of each residue in the hairpin, we used overlap PCR with pHWT1 as the template to produce a panel of 13 mutants in which individual residues were mutated to alanine. Residues Cys²² and Cys³⁶ were excluded from the mutational analysis, since their side chains are involved in disulfide bonds that are critical to the structure of the toxin. Gly³⁰ was also excluded for two reasons. First, it occurs at the tip of the hairpin in the i + 3 position of a type I -turn (where Gly is by far the most favored residue; see Ref. 36), and therefore it is likely to be structurally important. Second, with only a hydrogen atom side chain, Gly³⁰ is unlikely to be functionally important.

While the mutations we made were chosen to avoid structural perturbations, it was important to establish this experimentally. We did this by acquiring CD spectra of the native and mutant recombinant toxins (Fig. 3). Since the -hairpin is the major secondary structure feature of -ACTX-Hv1a, we expected that its ellipticity would dominate the CD spectrum of the native recombinant toxin, thus allowing structural perturbations in the hairpin to be readily discerned in spectra of the mutant toxins. Consistent with this assumption, the CD spectrum of the native recombinant toxin (Fig. 3) exhibited a -sheet signature, with pronounced minima and maxima at 210 and 196 nm, respectively. A predominantly -sheet protein would be expected to yield minima and maxima at 212-216 and 195-198 nm, respectively, whereas a largely -helical protein would give rise to a CD spectrum with minima at 208 and 222 nm and a maximum near 190 nm (37). The slight blue shift of the minima in the CD spectrum of -ACTX-Hv1a compared with that of a predominantly -sheet protein is expected for -sheets with a small number of antiparallel strands (37) (as is the case for -ACTX-Hv1a) and presumably also reflects the CD contribution from the disordered N-terminal residues (9), which would be expected to yield a minimum near 200 nm.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Circular dichroic spectra of native and mutant recombinant toxins. Comparison of the CD spectrum of wild-type recombinant -ACTX-Hv1a (WT) with the spectra of various alanine scan mutants (A), Asn27 and Arg35 mutants (B), and Lys34 mutants (C).

The CD spectra of all of the alanine scan mutants (with the exception of K34A; see below) were essentially superimposable on the CD spectrum of the native recombinant toxin (Fig. 3A), indicating that none of these mutations significantly perturb the structure of the -hairpin. Thus, any reduction in activity in the mutant toxins should largely reflect the functional significance of the mutated residue(s) rather than being a consequence of structural alterations in the toxin.

The insecticidal potency of the each of the mutants was examined by comparison of their LD₅₀ values in house crickets with that of recombinant toxin containing the native sequence (Fig. 2D, Table II). We considered mutants that increased the LD₅₀ by less than 2-fold to have a negligible effect on activity and considered that the mutated residues were therefore not functionally important. It was previously shown that mutations that caused less than 2.4-fold increase in the ED₅₀ for the insecticidal neurotoxin LqhIT rarely impacted on the channel binding activity of the toxin (38). Based on this criterion, residues Thr²³, Phe²⁴, Lys²⁵, Glu²⁶, Glu²⁸, Asn³¹, Thr³², and Val³³ appear not be critical to the function of -ACTX-Hv1a. We considered mutants that increased the LD₅₀ by 2-4-fold to have a significant effect on activity and considered that the mutated residues were therefore of potential functional importance. Residues in this category are Asn²⁹, Lys³⁴, and Asp³⁷. Residues whose mutation to Ala increased the LD₅₀ by more than 4-fold were considered to be crucial to the function of -ACTX-Hv1a. Only two residues, Arg³⁵ and Asn²⁷, fell into this category. The N27A mutation, which reduced the LD₅₀ by more than 14-fold, was by far the most deleterious of the panel of 13 alanine scan mutants.

Additional Hairpin Mutants-- Alanine-scanning mutagenesis of the -hairpin led to the conclusion that residues Asn²⁷, Asn²⁹, Lys³⁴, Arg³⁵, and Asp³⁷ are the most critical for the insecticidal activity of -ACTX-Hv1a. In order to further explore the molecular basis of toxin action, we constructed a panel of additional mutants.

First, we constructed a double mutant (N27A,R35A) that combined the two most deleterious point mutations from the alanine-scanning mutagenesis. Despite the combination of two spatially proximal radical mutations on the same face of the -hairpin (see Fig. 4B), the mutant was correctly folded as determined from CD analysis (see Fig. 3B). However, the double mutant toxin was completely devoid of insecticidal activity at all concentrations tested up to 4740 pmol g¹ (Table II), thus confirming that these two residues are absolutely critical to the function of -ACTX-Hv1a.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Surface features of the three-dimensional structure of -ACTX-Hv1a. A, three-dimensional structure of the -hairpin of -ACTX-Hv1a (Protein Data Bank file 1AXH) illustrating the hydrophobic interaction between the side chains of Phe24 and Lys34. The two -strands of the hairpin are shown as blue arrows (1 and 2), and the heavy atoms of the Phe24 and Lys34 side chains are depicted in pink and dark green, respectively. The side chain protons of the Lys34 side chain are shown in light green and are labeled. Note the close proximity of the Phe24 aromatic ring and the Lys34 - and -methylene protons. B, molecular surface of -ACTX-Hv1a illustrating the proposed interaction between residues in the -hairpin and insect voltage-gated calcium channels. The side chains of the key interacting residues (Asn27 and Arg35) form a contiguous patch (shown in red) on the surface of the hairpin, and this patch is flanked by two residues (Asn29 and Asp37, shown in yellow) that are proposed to be important, but not as critical, for the toxin-channel interaction.

In order to probe the molecular basis of the toxin-channel interactions involving these residues, we constructed and examined the insecticidal activity of N27D and R35E mutants. CD analysis (Fig. 3B) indicated that both mutants folded correctly. If the major interaction between Arg³⁵ and the targeted insect calcium channel is via a hydrogen bond(s) involving the -guanido group of the Arg³⁵ residue, then an R35E mutation would not be expected to be significantly more deleterious than an R35A mutation and may even be less deleterious if compensating hydrogen bonds can be formed using the Glu³⁵ carboxyl group. On the other hand, if Arg³⁵ forms an ion pair with a negatively charged group on the channel, then we might expect the R35E mutant to be even less potent than the R35A mutant, because it will introduce repulsive electrostatic interactions.

We found that the R35E mutant was 2-fold less potent than the R35A mutant (see Table II). While a larger decrease in potency might be expected from introduction of a repulsive electrostatic interaction with a carboxylate group on the channel, it must be borne in mind that we are measuring the effect of the mutation on the LD₅₀ rather than on channel binding. Previous mutagenesis studies with the scorpion insecticidal neurotoxin LqhIT showed that this tends to attenuate the effects of the mutation; the decrease in binding affinity was generally 10-50-fold greater than the increase in ED₅₀ when the measured -fold increase in ED₅₀ was 4 (38). Given this caveat, we tentatively conclude from the data that Arg³⁵ forms an ion pair with a negatively charged residue on the calcium channel. In future work, it will be interesting to examine the effects of an R35K mutation, which we would predict to be less deleterious than either an R35A or R35E mutation if Arg³⁵ forms an ion pair with a channel carboxylate.

If the major interaction between Asn²⁷ and calcium channels is via a hydrogen bond(s) utilizing the side chain carbonyl group of Asn²⁷, then we might expect the N27D mutant to be almost effective as the wild-type toxin, since the Asp residue retains this moiety. However, the N27D mutant was almost as deleterious as the N27A mutant, suggesting that Asn²⁷ interacts with insect calcium channels primarily via a hydrogen bond(s) involving the Asn²⁷ side chain amine group. Consistent with this hypothesis, the structure of -ACTX-Hv1a (9) reveals that the side chain carbonyl group of Asn²⁷ is largely buried and would be unavailable for interaction with the channel without a significant structural rearrangement of the toxin upon binding. In contrast, the -NH₂ group of Asn²⁷ is solvent-exposed and available for interaction with target channels.

The K34A and D37A mutations caused significant but not striking reductions in insecticidal potency (see Table II). In order to further explore the functional significance of these residues, we constructed K34E and D37K mutants, which reversed the charge on the side chains of these residues. The D37K mutant, which was properly folded according to CD analysis (data not shown), was more than 3-fold less active than the D37A mutant, indicating that Asp³⁷ is indeed a key functional residue that most likely engages in a favorable electrostatic interaction with a positively charged residue on the channel.

In marked contrast, the radical K34E charge reversal mutation caused a similar reduction in potency to the K34A mutation (see Table II). This indicates that Lys³⁴ is almost certainly not involved in an ion pair with a residue on the channel. Rather, we postulate that these mutations lead to a small reduction in insecticidal potency not because Lys³⁴ is functionally important but because mutation of this residue perturbs a key structural interaction that may help to stabilize the -hairpin. The three-dimensional structure of -ACTX-Hv1a (9) reveals that the methylene side chain of Lys³⁴ packs alongside the aromatic ring of Phe²⁴ on the opposite strand of the -hairpin (Fig. 4A). The close packing of these side chains causes a significant upfield shift of the methylene proton resonances of Lys³⁴ (see BioMagResBank chemical shift file 4233) due to a ring current shift from the Phe²⁴ aromatic side chain. The -methylene protons, which are closest to the ring (Fig. 4A), incur the largest shift; one of the Lys³⁴ -protons resonates at 0.49 ppm, which is ~1.0 ppm upfield of its random coil position (21).

The hydrophobic interaction between Phe²⁴ and the methylene protons of Lys³⁴ helps to stabilize what would otherwise be a highly solvent-exposed aromatic ring. Clearly, much of this hydrophobic stabilization would be lost in the K34A mutant where the long lysine side chain is substituted by a single methyl group, thereby removing key interactions between the aromatic ring and the Lys³⁴ - and -methylene protons (Fig. 4B). Similarly, much of the hydrophobic interaction would be lost in the K34E mutant, although it might be expected that the two methylene groups in the glutamate side chain would provide more stabilization than a single alanine methyl group, since the -methylene-ring interaction could be maintained. This appears to be the case, since the K34E mutant is slightly more potent than the K34A mutant.

CD analysis revealed that mutation of Lys³⁴ to either Ala or Glu did indeed perturb the structure of the -hairpin; the -sheet minimum is slightly blue-shifted from 210 to 208-209 nm, and the maximum at 196 nm is reduced in intensity such that it peaks near zero (Fig. 3C). The small positive maximum seen at 236 nm in the CD spectrum of the wild-type toxin is absent from spectra of the Lys³⁴ mutants (Fig. 3C). Thus, we conclude that Lys³⁴ is not functionally critical and probably does not interact with the target calcium channel, but it does play an important role in stabilizing the -hairpin.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

It has been estimated that 20-30% of the world's food supply is lost to insect pests (39). These insects have generally been controlled by foliar application of chemical pesticides, but persistent use of these chemicals has led to resistance development in numerous agronomically important insect species (40). Consequently, there is much interest in the elucidation of new insecticidal compounds. From a resistance perspective, the -atracotoxin-1 family of insecticidal neurotoxins is of considerable interest because these compounds act on a nonconventional target, namely insect voltage-gated calcium channels. In contrast, most commonly used insecticides target insect voltage-gated sodium channels (e.g. DDT, pyrethroids), acetylcholinesterase (e.g. organophosphates, carbamates), or GABA receptors (e.g. the arylheterocycles endosulfan and fipronil) (41).

In order to shed light on the molecular mechanism of action of -ACTX-Hv1a and to facilitate its use as a lead compound for insecticide development, we searched for key functional regions using a combination of protein engineering and site-directed mutagenesis. The C-terminal -hairpin of -ACTX-Hv1a is highly conserved, and this prominent structural feature distinguishes it from other calcium-channel blocking toxins such as the -conotoxins in which the loops tethered to the cystine knot scaffold are all considerably smaller (42). We initially examined the functional importance of the -hairpin by designing, constructing, and determining the three-dimensional structure of a Hairpinless mutant of -ACTX-Hv1a. We showed that this mutant was devoid of insecticidal activity, although the structure of the disulfide-rich core of the molecule was unperturbed, thus confirming that the -hairpin is indispensable for -ACTX-Hv1a activity.

We subsequently developed an expression system for production of recombinant toxin and used site-directed mutagenesis to examine the importance of individual hairpin residues. This study revealed that Asn²⁷ and Arg³⁵, which are conserved in all known members of the -atracotoxin-1 family (Fig. 1A), were the most functionally critical residues. An N27A,R35A double mutant toxin folds properly (Fig. 3B) but is completely devoid of insecticidal activity (Fig. 2D, Table II). Analysis of the structure of -ACTX-Hv1a (9) reveals that these two residues form a contiguous patch on the surface of the molecule that we propose represents a key site for interaction of the toxin with insect voltage-gated calcium channels (Fig. 4B). Analysis of N27D and R35E mutants indicated that this interaction most likely comprises a hydrogen bond interaction between the side chain amine group of Asn²⁷ and a hydrogen bond acceptor on the channel and an electrostatic interaction between the side chain guanido group of Arg³⁵ and a negatively charged residue on the channel. Future experiments with N27Q and R35K mutants should shed further light on the molecular details of this interaction.

Asp³⁷ and Asn²⁹ flank the main Asn²⁷/Arg³⁵ interaction site (see Fig. 4B) and potentially extend the channel interaction surface. However, the N29A mutation was only marginally deleterious (2.24-fold decrease in insecticidal potency), and it would be useful in future studies to further probe the functional significance of Asn²⁹ using N29D and N29Q mutations. Asp³⁷ does seem to be functionally important, and the mutation data suggest that it makes an electrostatic interaction with a positively charged residue on the channel (Fig. 4B).

Elucidation of the key functional residues in the -hairpin of -ACTX-Hv1a represents a starting point for modeling the interaction of this family of insecticidal neurotoxins with insect voltage-gated calcium channels. The development of a system for efficient production of recombinant toxin should allow the functional analysis to be extended in future to the disulfide core of the molecule, thus allowing a complete map of the channel binding surface to be obtained. Furthermore, our demonstration of successful heterologous expression of -ACTX-Hv1a in E. coli suggests that the genes for these toxins could be used to engineer toxin-enhanced viral insecticides.

	ACKNOWLEDGEMENTS

Mass spectral data were obtained at the University of Massachusetts Mass Spectrometry facility, which is supported in part by the National Science Foundation. We thank Dr. Zheng-yu Peng for help with CD analyses.

	FOOTNOTES

^* This work was supported by National Science Foundation Grant MCB9983242 (to G. F. K.), a UConn Health Center Biomedical Scholars Student Fellowship (to H. W. T.), and an Australian Grains Research and Development Corporation junior research fellowship (to J. I. F.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The atomic coordinates and the structure factors (code 1HVW) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).

^¶ To whom correspondence should be addressed: Dept. of Biochemistry, MC3305, University of Connecticut Health Center, 263 Farmington Ave. Farmington, CT 06032. Tel.: 860-679-8364; Fax: 860-679-1652; E-mail: glenn@psel.uchc.edu.

Published, JBC Papers in Press, April 19, 2001, DOI 10.1074/jbc.M102199200

¹ X.-H. Wang, M. Connor, D. Wilson, H. Wilson, G. M. Nicholson, R. Smith, D. Shaw, J. P. Mackay, P. F. Alewood, M. J. Christie, and G. F. King, submitted for publication.

	ABBREVIATIONS

The abbreviations used are: ACTX, atracotoxin; HPLC, high pressure liquid chromatography; rpHPLC, reverse-phase HPLC; GST, glutathione S-transferase; PCR, polymerase chain reaction; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; r.m.s., root mean square.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				H. W. Tedford and G. W. Zamponi Direct G Protein Modulation of Cav2 Calcium Channels Pharmacol. Rev., December 1, 2006; 58(4): 837 - 862. [Abstract] [Full Text] [PDF]

H. W. Tedford, A. E. Kisilevsky, J. B. Peloquin, and G. W. Zamponi Scanning Mutagenesis Reveals a Role for Serine 189 of the He