Role of disulfide bridges in the folding, structure and biological activity of ?-conotoxin GVIA 1 1

Role of disul¢de bridges in the folding, structure and biological activityof g-conotoxin GVIA1

James P. Flinn a;b, Paul K. Pallaghy a, Michael J. Lew b, Roger Murphy b,James A. Angus b, Raymond S. Norton a;*

a Biomolecular Research Institute, 343 Royal Parade, Parkville, Vic. 3052, Australiab Department of Pharmacology, University of Melbourne, Parkville, Vic. 3052, Australia

Received 9 April 1999; received in revised form 13 July 1999; accepted 13 July 1999

Abstract

g-Conotoxin GVIA (GVIA), an N-type calcium channel blocker from the cone shell Conus geographus, is a 27 residuepolypeptide cross-linked by three disulfide bonds. Here, we report the synthesis, structural analysis by 1H NMR and bioassayof analogues of GVIA with disulfide bridge deletions and N- and C-terminal truncations. Two analogues that retain thecrucial Lys-2 and Tyr-13 residues in loops constrained by two native disulfide bridges were synthesised using orthogonalprotection of cysteine residues. In the first analogue, the Cys-15-Cys-26 disulfide bridge was deleted (by replacing theappropriate Cys residues with Ser), while in the second, this disulfide bridge and the eight C-terminal residues were deleted.No activity was detected for either analogue in a rat vas deferens assay, which measures N-type calcium channel activity insympathetic nerve, and NMR studies showed that this was due to a gross loss of secondary and tertiary structure. Fiveinactive analogues that were synthesised without orthogonal protection of Cys residues as part of a previous study (Flinn etal. (1995) J. Pept. Sci. 1, 379^384) were also investigated. Three had single disulfide deletions (via Ser substitutions) and twohad N- or C-terminal deletions in addition to the disulfide deletion. Peptide mapping and NMR analyses demonstrated thatat least four of these analogues had non-native disulfide pairings, which presumably accounts for their lack of activity. TheNMR studies also showed that all five analogues had substantially altered tertiary structures, although the backbonechemical shifts and nuclear Overhauser enhancements (NOEs) implied that native-like turn structures persisted in some ofthese analogues despite the non-native disulfide pairings. This work demonstrates the importance of the disulfides in g-conotoxin folding and shows that the Cys-15-Cys-26 disulfide is essential for activity in GVIA. The NMR analyses alsoemphasise that backbone chemical shifts and short- and medium-range NOEs are dictated largely by local secondarystructure elements and are not necessarily reliable monitors of the tertiary fold. ß 1999 Elsevier Science B.V. All rightsreserved.

Keywords: Conotoxin; Disul¢de bridge; NMR spectroscopy; Peptide synthesis ; Bioassay; Polypeptide minimisation

0167-4838 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 7 - 4 8 3 8 ( 9 9 ) 0 0 1 6 5 - X

Abbreviations: Acm, acetamidomethyl ; CZE, capillary zone electrophoresis ; MALDI TOF-MS, matrix-assisted laser desorption ion-isation time of £ight mass spectrometry; MeCN, acetonitrile ; NOE, nuclear Overhauser enhancement; NOESY, two dimensional nuclearOverhauser enhancement spectroscopy; RP-HPLC, reverse-phase high performance liquid chromatography; TFA, tri£uoroacetic acid;GVIA, g-conotoxin GVIA from Conus geographus ; MVIIA, g-conotoxin MVIIA from Conus magus

* Corresponding author. Fax: +61 (3) 9903 9655; E-mail : [email protected] Five tables of assigned chemical shifts for the analogues described in the text are available from the corresponding author upon

request.

BBAPRO 35981 2-9-99

Biochimica et Biophysica Acta 1434 (1999) 177^190

www.elsevier.com/locate/bba

1. Introduction

g-Conotoxin GVIA (GVIA) is a polypeptide [1]from the ¢sh-hunting cone shell Conus geographusthat blocks N-type calcium channels [2,3]. This activ-ity confers a number of useful therapeutic properties,including neuroprotective and analgesic activities, onGVIA and the closely related g-conotoxin MVIIA(MVIIA) from Conus magus [4^6].

The solution structure of GVIA was solved byfour groups [7^10] in 1993 and a re¢ned structurehas been reported recently [11]. The g-conotoxinfamily adopts a structure known as the inhibitor cys-tine knot motif [12^14], which is shared by otherinhibitors from diverse sources including spidersand plants. The structural motif is characterised bya small triple-stranded L-sheet and a disul¢de knot.Structure-function studies for GVIA [15,16] haveidenti¢ed Lys-2 and Tyr-13 as the two most impor-tant residues for N-type calcium channel inhibition.Recently, we have described an extended pharmaco-phore for GVIA based on functional bioassays thatidentify residues of secondary importance for chan-nel binding [16] and probed the roles of the key func-tional groups in more detail [17].

Identifying the functional groups of GVIA essen-tial for calcium channel blocking activity representsa key step in elucidating the structure-functionrelationships of the molecule, but it is also necessaryto determine the role of the tertiary structure in pre-senting these groups in the biologically active con-formation. This is particularly relevant to e¡orts tomimic the calcium channel blocking activity ofGVIA in a smaller molecule, either peptidic ornon-peptidic in nature [6]. Towards this end, wehave investigated the e¡ects of deleting individualdisul¢de bonds in GVIA and truncating the peptidechain.

It is di¤cult to predict the e¡ect of single disul¢dedeletions on the structure and activity of polypep-tides containing multiple disul¢de bridges. Removalof single disul¢de bridges in charybdotoxin [18],leiurotoxin I [19] and echistatin Q [20] caused rela-tively small changes in structure or activity. Structur-al studies on bovine pancreatic trypsin inhibitor andinsulin-like growth factor I analogues with disul¢debridge deletions also detected only small changes[21,22]. By contrast, loss of either disul¢de bridge

of enterotoxin B abolished activity, although the sec-ondary structure, as measured by circular dichroism(CD) spectra, of one analogue was maintained [23].Deletion of a disul¢de bridge and three N-terminalresidues in epidermal growth factor resulted in a sub-stantially reduced activity and loss of the ¢rst strandof the L-sheet [24].

Of more direct relevance are studies on other poly-peptides from the inhibitor cystine knot family. Inthe protease inhibitor EETI-II, a stable intermediatelacking the Cys-2-Cys-19 disul¢de bridge (analogousto Cys-1-Cys-16 in GVIA) had a similar structure tothe native molecule [25]. In the case of gurmarin, asweet taste suppressing polypeptide from a plant,each of the three native disul¢de bridges was essen-tial for inhibition of receptors speci¢c for a numberof sugars, excluding sucrose [26]. Goldenberg and co-workers have undertaken an extensive study of thefolding of various g-conotoxins and the role thereinof disul¢de formation and the propeptide [27,28].While the current manuscript was in preparation,they also described the roles of individual disul¢debonds in the stability and folding of MVIIA [29],¢nding that deletion of individual disul¢de bridgesin that molecule led to potency losses of between70- and 5200-fold and gross losses of structure asjudged by CD. Previous work probing the role ofthe disul¢de bridges in GVIA [30] suggested thatthe Cys-8-Cys-19 and Cys-15-Cys-26 bridges werecrucial for biological activity but that Cys-1-Cys-16was less so, although this work did not reportdisul¢de pairings or structural data and the ana-logues were assayed using gold¢sh lethality ratherthan speci¢c activity on a de¢ned tissue prepara-tion.

In this paper, we have investigated the e¡ects ofdeletion of individual disul¢de bridges and trunca-tion of the N- and C-terminal regions in GVIA.The results demonstrate the importance of the nativedisul¢de bonds in the correct folding of GVIA andshow that the Cys-15-Cys-26 disul¢de is essential forbiological activity. NMR analyses show that native-like local turns are present even in analogues withnon-native disul¢des, highlighting the importance oflocal interactions in nucleating these turns. TheNMR data also emphasise that backbone chemicalshifts and short- and medium-range nuclear Over-hauser enhancements (NOEs) are dictated largely

BBAPRO 35981 2-9-99

J.P. Flinn et al. / Biochimica et Biophysica Acta 1434 (1999) 177^190178

by local secondary structure elements and are notnecessarily accurate monitors of the tertiary struc-ture.

2. Materials and methods

2.1. Peptide synthesis

Solid-phase Fmoc synthesis, Me2SO oxidation, pu-ri¢cation and characterisation of [Ser-1,16]GVIA,[Ser-8,19]GVIA and [Ser-15,26]GVIA were describedby Flinn et al. [31], as were the solid-phase Bocsynthesis, aerial oxidation and puri¢cation andcharacterisation of the two truncated analogues,[Ser-16]GVIA8ÿ27 and [Ser-15]GVIA1ÿ19 (where thesuperscript denotes the residues present using thenumbering of the native toxin).

In this study, the analogues [Ser-15,26]GVIA and[Ser-15]GVIA1ÿ19 were resynthesised with Cys-1 andCys-16 protected as the acetamidomethyl (Acm) de-rivative and Cys-8 and Cys-19 protected as the tritylderivative. These analogues were assembled semi-au-tomatically using continuous-£ow methodology at a0.1 mmol scale on NovaSyn TGR resin (0.2 mmol/g).The peptide resins were cleaved and all side-chainprotecting groups except for the Acm groups wereremoved by treatment with reagent K. The crudepeptides were isolated by ¢ltration and their identitywas con¢rmed by matrix-assisted laser desorptionionisation time of £ight mass spectrometry (MALDITOF-MS).

2.2. Peptide mapping of analogues synthesised withoutorthogonally protected half-cystines

Peptide mapping of the analogues [Ser-1,16]GVIAand [Ser-16]GVIA8ÿ27 [31] was achieved through di-gestion with trypsin and [Ser-15,26]GVIA and [Ser-15]GVIA1ÿ19 [31] by digestion with chymotrypsin. Ineach case, the conditions used were 100 mM sodiumphosphate bu¡er, pH 6.5, and an enzyme:substrateratio of 1:50 for 22 h at 37³C. The resulting peptidefragments were isolated using a 5 Wm AdsorbosphereC18 column (150U2.1 mm), eluted with a gradient ofacetonitrile (MeCN) (0^30% over 30 min) in 0.1%tri£uoroacetic acid (TFA) and analysed by MALDITOF-MS.

2.3. Oxidation of orthogonally protected analogues

Oxidative closure of the ¢rst disul¢de bridge (i.e.Cys-8-Cys-19) was accomplished by dissolution ofthe crude lyophilised peptides in 10% Me2SO in 50mM NH4HCO3, pH 8, at a peptide concentration of1 mg/ml and incubation overnight at room temper-ature. The monocyclic peptides were puri¢ed by apreparative reverse-phase high performance liquidchromatography (RP-HPLC) column using a gra-dient of MeCN (5^35% over 60 min) in 0.1% TFA.Fractions were collected and analysed by analyticalRP-HPLC using a gradient of MeCN (0^30% or 10^30% over 30 min) in 0.1% TFA. Appropriate frac-tions (s 95% pure) were pooled and lyophilised. Thisprocedure yielded essentially homogeneous [Cys-(Acm)]monocyclic analogues.

Oxidative closure of the second disul¢de bond (i.e.Cys-1-Cys-16) was carried out by a modi¢cation ofthe method of Kamber et al. [32]. The [Cys(Acm)]-monocyclic analogues (1 Wmol) were dissolved in50% aqueous acetic acid (200 Wl). To this mixture,1 M HCl (50 Wl) was added, followed immediately by50 mM I2 in 50% aqueous acetic acid (2 ml). Thesolution was stirred for 30 min and the crude pep-tides were puri¢ed by preparative RP-HPLC. Evi-dence for homogeneity was provided by RP-HPLC(0.1% TFA H2O/MeCN) and capillary zone electro-phoresis (CZE). The identity of the products wascon¢rmed by MALDI TOF-MS. Yields of oxidised¢nal product expressed as a percentage of crude lin-ear peptide were in the range 1.5^7% for the peptidesexamined in this study.

2.4. NMR spectroscopy

1H NMR spectra of native GVIA in 90% H2O/10% 2H2O at pH 5.6 and 298 K show several back-bone amide resonances that appear to be broadenedby exchange with solvent [10]. At pH 3.4, these res-onances were all sharp, so this pH was used forNMR analysis in the present study. As GVIA lackscarboxyl-bearing or histidine residues, no signi¢cantpH-dependent chemical shift or conformationalchanges would be expected between pH 3.4 and 7.4and we have shown previously that in native GVIA,the non-labile protons show the same NOEs overthis pH range, with and without 150 mM NaCl (un-

BBAPRO 35981 2-9-99

J.P. Flinn et al. / Biochimica et Biophysica Acta 1434 (1999) 177^190 179

published results). Comparison of two dimensional(2D) nuclear Overhauser enhancement spectroscopy(NOESY) spectra at 298 and 283 K indicated thatstronger NOEs were observed at the lower temper-ature (due to an increase in the overall correlationtime of the molecule), so spectra for the disul¢de-deleted analogues were acquired routinely at 283 K.Additional spectra were recorded at 278 and 298 Kto con¢rm assignments in the event of peak overlapor coincidence with the water signal.

1H NMR spectra were recorded on Bruker AMX-500 and -600 spectrometers, essentially as describedpreviously [10,11], with 3.7^15 mM samples of poly-peptide. TOCSY spin-lock and NOESY mixing timeswere 70 and 300 ms, respectively. The 1H chemicalshifts were referenced to 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) at 0 ppm, via the chemical shift ofthe H2O resonance [33].

2.5. Bioassay

Testing of the analogues synthesised without or-thogonal protection was reported in Flinn et al.[31]. The peptide analogues synthesised with orthog-onal protection were tested for their e¡ect on theelectrically evoked twitch responses in the rat vasdeferens in the presence and absence of nativeGVIA, as described by Lew et al. [16].

3. Results

3.1. Synthesis and peptide mapping of analogueswithout orthogonal protection

The amino acid sequences of GVIA and theanalogues investigated here are shown in Fig. 1.The Cys residues participating in the disul¢de bridgesto be deleted were replaced with Ser rather thanthe alternative isostere K-aminobutyric acid dueto the lack of any signi¢cant hydrophobic core innative GVIA [11]. The analogues [Ser-1,16]GVIA,[Ser-16]GVIA8ÿ27, [Ser-15,26]GVIA and [Ser-15]-GVIA1ÿ19 [31] after various stages of puri¢cationwere shown to be essentially homogeneous by ana-lytical RP-HPLC and CZE. The analogue [Ser-8,19]GVIA [31] appeared to be homogeneous by an-alytical RP-HPLC and MALDI TOF-MS, but CZE

analysis showed two closely migrating isomeric spe-cies. No attempt was made to purify this analoguefurther.

Peptide mapping was undertaken to determine thedisul¢de pairings of the oxidation products. NativeGVIA contains four possible tryptic cleavage sites,two adjacent to Lys (Lys-2 and Lys-24) and twoadjacent to Arg (Arg-17 and Arg-25), and three pos-sible chymotryptic cleavage sites (Tyr-13, Tyr-22 andTyr-27). The analogues [Ser-1,16]GVIA and [Ser-16]GVIA8ÿ27 were digested with trypsin, while [Ser-15,26]GVIA and [Ser-15]GVIA1ÿ19 were digestedwith chymotrypsin, in order to distinguish the disul-¢de topology with disul¢de pairings between the ¢rstand second half-cystines (1-2) and the third andfourth (3-4) from those with 1-3/2-4 and 1-4/2-3 pair-ings. These digestions were performed in a phosphatebu¡er at pH 6.5, chosen to minimise disul¢de ex-change reactions [34].

The principal products were separated by RP-HPLC and analysed by MALDI TOF-MS and theresults are summarised in Fig. 2 and Table 1, respec-

Fig. 1. The amino acid sequence of native GVIA and analoguesof GVIA where O represents hydroxyproline and the disul¢debridges are represented by lines. The sequence is C-terminallyamidated. (P) indicates that the analogue was folded with or-thogonal protection. Residues in bold indicate substitutions rel-ative to the native sequence. The analogues [Ser-15,26]GVIA,[Ser-15]GVIA1ÿ19, [Ser-1,16]GVIA and [Ser-16]GVIA8ÿ27 weredetermined by peptide mapping studies to have a non-native,bead topology (see text). The disul¢de pairings of [Ser-8,19]GVIA were not determined.

BBAPRO 35981 2-9-99


tively. The proteolytic digestion products demon-strated clearly that [Ser-1,16]GVIA, [Ser-16]-GVIA8ÿ27 and [Ser-15]GVIA1ÿ19 adopted non-native1-2/3-4 disul¢de pairings rather than the desired(native) 1-3/2-4 pairings. The results for [Ser-15,26]GVIA were equivocal, as only one of the ex-pected fragments for the 1-2/3-4 pairings (C1, Fig.2C) was identi¢ed. However, no fragments corre-sponding to other disul¢de pairings were identi¢edand we assume that this analogue also adoptednon-native pairings.

3.2. Synthesis of orthogonally protected analogues

The analogues [Ser-15,26]GVIA and [Ser-15]GVIA1ÿ19 were resynthesised successfully usingan orthogonal protection strategy and the ¢nal prod-ucts are henceforth referred to as [Ser-15,26]GVIA(P)and [Ser-15]GVIA1ÿ19(P), respectively, the P indicat-ing cyclisation of a protected analogue. All protect-ing groups except for Acm were removed by treat-ment with reagent K. Oxidative closure of the ¢rstdisul¢de bridge was accomplished with 10% Me2SOin aqueous NH4HCO3. The analytical RP-HPLCpro¢le and retention time of the crude [Cys(Acm)]-monocyclic material did not di¡er from that of thecrude linear material and so, the Ellman test [35] wasused to con¢rm that oxidation had occurred. Acm

removal and oxidative closure of the second disul¢deusing I2 in 50% aqueous acetic acid proved to beproblematic. At the completion of the reaction,quenching of I2 with sodium thiosulfate resulted ina complex analytical RP-HPLC pro¢le. These com-ponents were shown by MALDI TOF-MS to repre-sent a series of modi¢ed peptides that did not corre-spond to the required analogue of GVIA.

In contrast, if the I2 was not quenched, RP-HPLCanalysis showed a comparatively simple pro¢le, with

Fig. 2. RP-HPLC pro¢les of the disul¢de-deleted analogues fol-lowing proteolytic digestion. (A) [Ser-1,16]GVIA (digested withtrypsin), (B) [Ser-1,16]GVIA8ÿ27 (trypsin), (C) [Ser-15,26]GVIA(chymotrypsin) and (D) [Ser-15]GVIA1ÿ19 (chymotrypsin). Massspectral data for products corresponding to proteolytic frag-ments (as labelled) are shown in Table 1. Other peaks may cor-respond to smaller proteolytic fragments and were not able tobe assigned on the basis of mass spectral analysis because ofcoincidence with molecular ions from K-cyano-4-hydroxy-cin-namic acid.

Table 1Mass spectral data and inferred fragments for the proteolysis ofsynthetic GVIA analogues synthesised without orthogonal pro-tection of Cys residues

Peptide designations are as for Fig. 2.

BBAPRO 35981 2-9-99


the major product corresponding to the desiredGVIA analogue (Fig. 3A). Interestingly, performingthe I2 oxidation on a large scale (30 mg of peptide)resulted in a signi¢cant quantity of late eluting ma-terial having a deep amber colour. The molecularweight of this material was shown by MALDITOF-MS to be identical to that calculated for [Ser-15,26]GVIA. The hydrophobic nature of this materi-al and its amber colour suggest that under these con-ditions, I2 is involved in a non-covalent interactionwith the disul¢de analogue, possibly with the Z elec-trons of the Tyr residues. Performing the reaction ona smaller scale (3 mg of peptide) resulted in less ofthe late eluting material and more of the requiredcompound. The remaining Acm-protected [Ser-15,26]GVIA was deprotected/oxidised in batches of3 mg and puri¢ed by preparative RP-HPLC to givematerial that was essentially pure by analytical RP-HPLC (Fig. 3B) and CZE (Fig. 3C) and gave a singlemolecular ion by MALDI TOF-MS. The analogue[Ser-15]GVIA1ÿ19 was subsequently deprotected andoxidised using a similar protocol to give material thatwas pure by analytical RP-HPLC and CZE.

3.3. NMR of [Ser-8,19]GVIA

Although [Ser-8,19]GVIA appeared to be pure byRP-HPLC and MALDI TOF-MS, analysis by CZEshowed two isomeric species. Broad signals due to

peak overlap were observed in the one dimensional(1D) NMR spectrum (not shown), suggesting thatmultiple species and/or conformations were present.This was con¢rmed by 2D NMR. Because ofthe spectral complexity, no attempt was made to se-quentially assign either of these conformers. Never-theless, it was clear from the substantial loss of spec-tral dispersion compared with that of native GVIAthat the structure of this analogue was signi¢cantlyaltered.

3.4. NMR of non-orthogonally protected analogueslacking the 1-16 bridge

1D and 2D spectra of the full-length analogue[Ser-1,16]GVIA indicated the presence of multiplespecies, although it was possible to obtain an essen-tially complete set of sequential assignments for themajor form. RP-HPLC and CZE gave only a singlepeak for this sample (data not shown), indicatingthat it was chemically homogeneous and thus, thatthe origin of the spectral heterogeneity was not pep-tide impurities but conformational isomerism. Onepossible source of the observed conformational het-erogeneity is cis-trans isomerism of the peptide bondspreceding the three Hyp residues. Another possibil-ity, albeit less likely, is isomerisation about one orboth of the disul¢de bonds, in which the chirality ofthe disul¢de bond may be either right- or left-

Fig. 3. Analysis of the oxidative folding of [Ser-15]GVIA1ÿ19(P). (A) RP-HPLC analysis of crude [Cys(Acm)]monocyclic product, (B)RP-HPLC analysis of puri¢ed bicyclic product, (C) CZE analysis of puri¢ed bicyclic product.

BBAPRO 35981 2-9-99


handed. Isomerisation of this type has been identi¢edin bovine pancreatic trypsin inhibitor [36].

A small number of native and native-like NOEswas identi¢ed in the region between Ser-9 and Asn-14 and between Asn-20 and Arg-25, which may de-¢ne turns II and IV present in the native structure(Fig. 1). There was no evidence for turn I, re£ectingthe absence of the Cys-1-Cys-16 disul¢de bridge,which stabilises this turn in the native molecule. Inaddition, several non-native NOEs were observed,e.g. Ser-12 CKH-Cys-15 CLH, Tyr-13 C(2,6)H/C(3,5)H-Ser-16 CKH/CLH and Tyr-22 NH-Arg-25NH. Both the reduced spectral dispersion and thesmall number of medium- and long-range NOEs(19, compared with 175 for native) imply that remov-al of the Cys-1-Cys-16 disul¢de bond, even withoutremoval of the N-terminal tail, disrupted the struc-ture signi¢cantly. The exact number of native NOEsshould be interpreted with caution, however, becauseof the limited spectral dispersion, which hinders theunambiguous identi¢cation of these (and other)NOEs even if present.

1H NMR spectra of [Ser-16]GVIA8ÿ27, whichlacks the ¢rst seven residues as well as the 1-16bridge, also showed evidence of heterogeneity, re-£ected in the presence of several weaker resonances.As for the full-length analogue, RP-HPLC and CZEshowed that the origin of this heterogeneity was notchemical impurities but conformational isomerism.An essentially complete set of sequential assignmentsfor the major conformer was obtained. A small num-ber of medium- and long-range NOEs (26) was iden-ti¢ed for this analogue, indicative of a loss of tertiarystructure and consistent with the loss of spectral dis-persion. There were several native and native-likeNOEs in the region between Ser-9 and Cys-15 andbetween Asn-20 and Arg-25, which may de¢ne L-

Fig. 5. Deviations of backbone NH chemical shifts from ran-dom coil values for (A) native GVIA, (B) [Ser-15,26]GVIA, (C)[Ser-15]GVIA1ÿ19 and (D) [Ser-15]GVIA1ÿ19(P). The open ar-rows represent the L-strand locations in native GVIA.

Fig. 4. Deviations of backbone NH chemical shifts from ran-dom coil values for (A) native GVIA, (B) [Ser-1,16]GVIA and(C) [Ser-16]GVIA8ÿ27. The open arrows represent the L-strandlocations in native GVIA.

BBAPRO 35981 2-9-99


Fig. 6. (A) Fingerprint region of the TOCSY spectrum of [Ser-15,26]GVIA(P). (B) Region of TOCSY boxed in part A, together witha row demonstrating the presence of two conformers in [Ser-15,26]GVIA(P). (C) Fingerprint region of the TOCSY spectrum and as-signments for [Ser-15]GVIA1ÿ19(P).

BBAPRO 35981 2-9-99


turns II (Ser-9-Ser-12) and IV (Hyp-21-Lys-24) in thenative structure. There were, however, also severalnon-native NOEs in the region between Cys-8 andSer-16.

Further information on the conformations of theseanalogues can be gleaned from a comparison of thedeviations of their backbone NH and CKH chemicalshifts from random coil values. Plots for the NHresonances of native GVIA, [Ser-1,16]GVIA and[Ser-16]GVIA8ÿ27 are shown in Fig. 4. It is clearthat the NH resonances of [Ser-1,16]GVIA (Fig.4B) are closer to random coil values than those ofnative GVIA, consistent with the loss of spectral dis-persion noted above and re£ecting a destabilisationof the global fold and disruption of the small anti-parallel L-sheet. Interestingly, the pattern of NH

chemical shift deviations for this analogue is similarto that of native GVIA (compare Fig. 4A and B) forseveral residues (2,3, 5^7, 11, 12, 23, 25^27), eventhough the magnitudes of the deviations are reducedcompared with native.

The plots for [Ser-1,16]GVIA and [Ser-16]-GVIA8ÿ27 are very similar (allowing for the absenceof data for the deleted residues 1^7 in the latter), asshown in Fig. 4B and C, indicating that the averagestructures of these two analogues in solution areprobably similar. Thus, the loss of the Cys-1-Cys-16 disul¢de bridge appears to be mainly responsiblefor destabilisation of the structure, rather than dele-tion of the N-terminal seven residues, and the pres-ence of these residues in [Ser-1,16]GVIA has a negli-gible in£uence on the rest of the molecule.

Fig. 6 (continued).

BBAPRO 35981 2-9-99


3.5. NMR of non-orthogonally protected analogueslacking the 15-26 bridge

The NMR data for [Ser-15,26]GVIA show that itexists as multiple conformers in solution. A virtuallycomplete set of resonance assignments was obtainedfor the major conformer, but there appeared to bealmost no medium- and long-range NOEs. This canbe ascribed partly to the lower sample concentrationof this analogue (3.7 mM) compared with that of[Ser-15]GVIA1ÿ19 (10.2 mM, see below). Plots ofthe deviations of the backbone NH chemical shiftsfrom random coil values for native GVIA,[Ser-15,26]GVIA and [Ser-15]GVIA1ÿ19 are shownin Fig. 5. Although some signi¢cant shifts in thenative spectrum are partially maintained in [Ser-15,26]GVIA (e.g. Lys-2 and Gly-5), the overallpattern of shifts is not as well preserved as in thecomparison of [Ser-1,16]GVIA and native GVIA(Fig. 4). This implies an even greater loss of struc-ture, consistent with the lack of native and native-like NOEs.

The analogue [Ser-15]GVIA1ÿ19 also showed evi-dence of multiple conformers, although it was possi-ble to obtain an essentially complete set of resonanceassignments for the major conformer. Loss of terti-ary structure was demonstrated by the reduced spec-tral dispersion and small number of medium- andlong-range NOE interactions. There were only afew native NOEs observed, e.g. Hyp-10 CLH-Ser-12NH, Thr-11 CKH-Tyr-13 NH and Tyr-13 CLH-Ser-15 NH, and some native-like NOEs, such as Ser-3NH-Ser-6 CLH and Thr-11 CLH-Tyr-13 NH. Therewere, however, several non-native medium-range (i/i+2, i/i+3) and long-range (Cys-1-Cys-8, Cys-1-Ser-9)NOEs in the region between Ser-1 and Ser-9, as wellas Thr-11 CKH/CLH-Asn-14 NH, Asn-14 CLH-Cys-16 NH and Arg-17 CKH-Cys-19 NH NOEs, all re-£ecting the non-native disul¢de bridge pairing (seeabove).

Comparison of the plots for [Ser-15,26]GVIA and[Ser-15]GVIA1ÿ19 shows that their patterns of devia-tions are similar to one another (Fig. 5B and C).Again, it appears that deletion of the disul¢debond is the major cause of structural change, withsubsequent deletion of part of the polypeptide chainhaving a minor e¡ect.

3.6. NMR of orthogonally protected analogueslacking the 15-26 bridge

[Ser-15,26]GVIA(P) exists as multiple conformersin solution. Broad signals due to peak overlap wereobserved in the 1D spectrum and the 2D TOCSYspectrum showed that several protons in the mole-cule gave two equally intense resonances (Fig. 6Aand B). These additional resonance positions areclearly conformational in origin, as evidenced by asingle peak on RP-HPLC and CZE and a singlemolecular ion detected by mass spectral analysis.As the corresponding truncated analogue ([Ser-15]GVIA1ÿ19(P), see below) exists in a single majorconformation, the multiple conformations observedin [Ser-15,26]GVIA(P) presumably re£ect the uncon-strained tail between residues 20 and 27. The sourceof this conformational isomerism may be cis-transisomerism, with Hyp-21 free to take up either con-formation as it is not constrained within a disul¢deloop. Due to the presence of multiple conformationsin approximately equal proportions, a complete setof resonance assignments was not obtained.

The NMR data for [Ser-15]GVIA1ÿ19(P) showedevidence of minor conformations (Fig. 6C), althougha complete set of resonance assignments was ob-tained for the major conformer. There appeared tobe no medium- or long-range NOEs even though thesample concentration was 15 mM. 1D NMR data forthis analogue show an even greater loss of tertiarystructure than with the isomer with incorrect disul-¢de pairings, as indicated by the reduced spectraldispersion. Fig. 5D shows the deviation of the NHchemical shifts from random coil values for this ana-logue. Although these deviations are signi¢cantlydampened, the sign of the deviation is the same asthat of native GVIA for most residues, which islargely a re£ection of the presence of two native di-sul¢de bridges (Cys-1-Cys-16 and Cys-8-Cys-19).

3.7. Biological characterisation

None of the initial ¢ve analogues tested up to aconcentration of 1 WM had any observable e¡ect onthe twitch responses of the vas deferens [31]. More-over, GVIA retained its ability to fully inhibit thetwitch in the presence of 1 WM of the analoguesand its potency was not reduced, with 50% inhibi-

BBAPRO 35981 2-9-99


tion between 1 and 10 nM. The analogues [Ser-15,26]GVIA(P) and [Ser-15]GVIA1ÿ19(P) also didnot have any noticeable e¡ect on the twitch re-sponses up to 10 WM and did not a¡ect the abil-ity of GVIA to fully inhibit the twitch responses(Fig. 7).

4. Discussion

The results presented here provide information onthe role of the disul¢de bridges in the folding ofGVIA and their importance for its calcium channelblocking activity. They also emphasise the distinctionbetween local structure and the global fold in in£u-encing backbone chemical shifts.

4.1. Role of disul¢des in folding

Native GVIA adopts a 1-4/2-5/3-6 pattern of di-

sul¢de pairings (Fig. 1). When synthesised and oxi-dised without orthogonal protection of Cys residues,[Ser-1,16]GVIA and [Ser-16]GVIA8ÿ27, both lackingthe native 1-4 bridge, formed a 2-3/5-6 pattern (Cys-8-Cys-15 and Cys-19-Cys-26) instead of the desired2-5/3-6. Likewise, [Ser-15,26]GVIA and [Ser-15]-GVIA1ÿ19, lacking the native 3-6 bridge, gave a1-2/4-5 pattern (Cys-1-Cys-8 and Cys-16-Cys-19)rather than 1-4/2-5. The similar folding of the trun-cated analogues, [Ser-16]GVIA8ÿ27 and [Ser-15]GVIA1ÿ19, to that of the corresponding full-lengthanalogues, [Ser-1,16]GVIA and [Ser-15,26]GVIA, re-spectively, shows that the N- and C-terminal residueswere of no bene¢t in directing the folding of theanalogues toward native pairings.

Goldenberg and co-workers [27] have shown thatreduced GVIA and MVIIA refold to the native formwith e¤ciencies of about 50% in the presence of glu-tathione. Flinn and Murphy [37] found that 10%dimethyl sulfoxide in ammonium bicarbonate bu¡erat pH 8 gave a good yield of native GVIA. In thepresent study, the peptides made without orthogonalprotection were oxidised in ammonium acetate bu¡erat pH 7 containing 5^20% dimethyl sulfoxide (full-length analogues) or by aerial oxidation in ammo-nium bicarbonate bu¡er at pH 8 (truncated ana-logues) [31]. The kinetics of refolding and yield ofthe major product were reportedly similar in thesetwo methods. However, it appears that in both re-folding bu¡ers, the dominant two-disul¢de productscontained non-native pairings and we now considerwhy this occurred.

Alternative bridging patterns of four half-cystinemolecules are referred to by Zhang and Snyder [38]as globule (Cys1-Cys3, Cys2-Cys4), bead (Cys1-Cys2,Cys3-Cys4) or ribbon (Cys1-Cys4, Cys2-Cys3) topolo-gies. These authors studied the rates of disul¢debridge formation for a series of synthetic peptidescontaining a Cys-Alam�0ÿ5-Cys sequence, where mis the number of intervening residues between thetwo Cys residues forming the disul¢de loop. Theydemonstrated that there is a distinctive odd-even pat-tern in loop stability, with the rate of disul¢de for-mation for even m being higher than that for odd m(for m6 6). A similar pattern was found from a stat-istical analysis of naturally occurring proteins. Thispattern diminishes at larger values of m and is pre-dicted to degenerate into a general dependence where

Fig. 7. E¡ects of native GVIA, [Ser-15,26]GVIA(P) and [Ser-15]GVIA1ÿ19(P) on the twitch responses of the rat vas deferensto sympathetic nerve stimulation. Cumulative peptide additionsare indicated by the dots, with the concentration (3log M) in-dicated. Open dots indicate the application of native GVIA at1039 M and 1038 M, sequentially. GVIA inhibited the twitchesby about 50% at 1039 M and completely at 1038 M in the ab-sence (upper record) and presence of the other peptides (lowertraces). Each experiment is representative of at least three repli-cates and the calibration bars apply to each record.

BBAPRO 35981 2-9-99


the rate of loop formation is proportional to m33=2

[39]. The latter relationship represents the generaldecrease in probability of random collisions betweentwo Cys residues as they are separated by longerdistances in the linear sequence.

These considerations can explain the bead topol-ogy of [Ser-1,16]GVIA and [Ser-16]GVIA8ÿ27. With[Ser-15,26]GVIA and [Ser-15]GVIA1ÿ19, the beadtopology (as observed) or ribbon topology wouldbe expected. For [Ser-8,19]GVIA, the native andthe bead topologies are probably equally favouredand may correspond to the two closely migratingpeaks observed by CZE.

For [Ser-1,16]GVIA and [Ser-16]GVIA8ÿ27, someof the NOEs de¢ne chain reversals in the vicinityof turns II and IV in the native structure. In general,an analogue with non-native disul¢de topologywould not be expected to have native-like turns,but the formation of bead topology in these ana-logues would induce and stabilise reverse turns insimilar regions to those of turns II and IV in thenative structure. For [Ser-15]GVIA1ÿ19, some of theNOEs de¢ne chain reversals in the vicinity of turns Iand II in the native structure, but once again, a beadtopology for this analogue would induce and stabi-lise reverse turns in similar regions.

[Ser-1,16]GVIA and [Ser-16]GVIA8ÿ27 were not re-synthesised with orthogonal protection as the func-tionally important Lys-2 residue [16] was either notpresent or not constrained in these analogues. As[Ser-15,26]GVIA and [Ser-15]GVIA1ÿ19 containLys-2 and Tyr-13, they were resynthesised using or-thogonal protection. Virtually no medium- or long-range NOEs were identi¢ed for [Ser-15]GVIA1ÿ19(P)with native disul¢de pairings. In comparison, 21 me-dium- and long-range NOEs were identi¢ed for theanalogue [Ser-15]GVIA1ÿ19 with bead topology, re-£ecting the ability of small disul¢de loops to induceand stabilise a turn structure. The analogue [Ser-15,26]GVIA(P) with native disul¢de pairings showedthe presence of multiple conformations, presumablydue to the poorly constrained free tail. In contrast,the analogue [Ser-15,26]GVIA with bead disul¢detopology existed as a single major conformer, despitehaving a free tail. Once again, this probably re£ectsthe ability of small disul¢de loops to induce andstabilise a turn structure, contributing to a more sta-ble, but not necessarily native, tertiary structure.

Our observations with respect to the accumulationof non-native two-disul¢de intermediates concur withthe recently published results of a study of the refold-ing of the GVIA homologue, MVIIA [29]. ThreeMVIIA analogues lacking one of the native disul¢des(by replacement of the respective half-cystine residueswith alanine) were refolded in three di¡erent bu¡ers,one containing I2, which causes rapid and irreversibledisul¢de formation, the second containing gluta-thione, to permit disul¢de shu¥ing, and the thirdcontaining glutathione and 8 M urea to disruptnon-covalent interactions. In all three refolding sol-utions, the species with two native disul¢des accumu-lated to detectable levels, but in no case, were theythe major product. Moreover, the addition of ureahad little e¡ect, implying that formation of the nativepairings is not signi¢cantly favoured by non-covalentinteractions. Thus, it appears that the oxidative fold-ing of MVIIA in vitro is characterised by a broaddistribution of disul¢de-bonded intermediates, withno particular preference for native pairings amongthe intermediates containing only two disul¢des[29]. Once two native disul¢des form, there is ahigh probability of progression to the native (three-disul¢de) fold, which eventually accumulates in pref-erence to non-native three-disul¢de species because itis energetically more stable. However, when forma-tion of the third disul¢de is blocked, as in the presentstudy of GVIA analogues, non-native two-disul¢despecies dominate. Similar e¡ects on folding to non-native disul¢de con¢gurations have been observedrecently in two out of three disubstituted (Cys to K-aminobutyric acid) analogues of charybdotoxin [40].

4.2. Role of disul¢des in biological activity

Goldenberg and co-workers [29] recently reportedlosses of binding a¤nity between 70- and 5200-fold inMVIIA analogues with single disul¢de bridges de-leted. The MVIIA analogue analogous to [Ser-15,26]GVIA(P) had the greatest loss of binding a¤n-ity (5200-fold). Our results show that [Ser-15,26]GVIA(P) had no detectable activity at 10 WM,corresponding to a loss of potency of s 8000-fold.

Data presented in this study show that [Ser-1,16]GVIA, [Ser-16]GVIA8ÿ27, [Ser-15,26]GVIA and[Ser-15]GVIA1ÿ19, synthesised without orthogonalprotection, all have non-native disul¢de pairings. It

BBAPRO 35981 2-9-99


is not possible, therefore, to draw any conclusionsabout the role of the Cys-1-Cys-16 or Cys-15-Cys-26 disul¢des in the calcium channel blocking activityof GVIA from published data [31] on these products.Moreover, the conclusion that removal of any one ofthe three native disul¢des of GVIA destroys activity[31] is not substantiated. Indeed, the data for MVIIAsuggest that GVIA analogues lacking either the 1-16or 8-19 disul¢des should have calcium channel bind-ing activity within two orders of magnitude of native.

A preliminary study by Sabo et al. [30] in whichdisul¢de bridges were deleted from GVIA reportedsome residual biological activity when the Cys-1-Cys-16 disul¢de bridge was absent, but complete loss ofactivity when either of the other bridges was deleted.However, these results were obtained from a gold¢shlethality assay and there are likely to be some di¡er-ences between the gold¢sh and mammalian re-sponses. Moreover, no disul¢de mapping was re-ported for these analogues.

Pennington et al. [41] found that a disul¢de isomerof GVIA with non-native disul¢de pairings (Cys-1-Cys-8, Cys-15-Cys-19 and Cys-16-Cys-26, represent-ing a 1-2/3-5/4-6 pattern, in contrast to the native 1-4/2-5/3-6 pattern) had only a 77-fold lower activitythan GVIA. Although a structure determination ofthis isomer was not practical because of spectralcomplexity, it was clear from the chemical shiftsthat the major conformer of this isomer adopted aless well-ordered structure in solution and lackedmany of the features of the native toxin [10]. Fur-thermore, isomers of the analogue [Ala-5]GVIA withnon-native disul¢de pairings [16] had a between 290-and 760-fold lower activity than GVIA (data notshown). The observation of some activity for theseanalogues, despite their non-native disul¢de pairings,implies that locally de¢ned elements of the GVIApharmacophore can support weak calcium channelbinding, even though the juxtaposition of all ele-ments in the native structure is required for full ac-tivity. It also suggests that the even greater loss ofactivity for the current disul¢de-deleted analogueswith non-native pairings is due to greater conforma-tional £exibility associated with loss of one of thedisul¢des and the concurrent loss of entropy uponbinding to the calcium channel.

Several successful examples of protein minimisa-tion have been reported in the literature [42] where

a polypeptide or protein has been truncated and thenoptimised using phage display. Protein minimisationmay prove to be a valuable approach in generatingnovel therapeutic polypeptides based on the g-con-otoxin family, given that some regions of the mole-cule can be altered substantially without completeloss of activity.

4.3. Backbone chemical shifts

Initial inspection of the patterns of variation fromrandom coil chemical shifts for the backbone NH andCKH of analogues lacking the Cys-1-Cys-16 disul¢de(Fig. 4) and, to a lesser extent, the Cys-15-Cys-26disul¢de (Fig. 5) suggested that elements of the nativestructure had been retained. This was supported insome cases by the observation of NOEs seen in thenative spectrum. It eventuated, however, that bothparameters were re£ecting the presence of chain re-versals at similar locations. In the case of GVIA,these turns were stabilised by the native fold, whilein the analogues with non-native disul¢de pairings,they were stabilised by the disul¢des linking neigh-bouring Cys residues. In fact, the overall structuresof the analogues were quite di¡erent from native.Thus, while the backbone chemical shifts were con-sistent with similar local topologies, albeit more £ex-ible in the case of the analogues, the exact secondarystructures were not identical and the tertiary struc-tures were very di¡erent. To draw inferences aboutsimilar tertiary structures, it is also necessary to con-sider the chemical shifts of numerous side-chain pro-ton resonances, with those shifted furthest from theirrandom coil values being the most diagnostic.

5. Conclusions

It is clear that the three disul¢de bridges of GVIAare major determinants of the native fold. Analogueslacking one of these three disul¢des become trappedduring folding in non-native disul¢de pairings drivenby local interactions and behave as random polypep-tide sequences in terms of their folding in vitro. As aresult, orthogonal protection is required to generatenative pairings.

The Cys-15-Cys-26 disul¢de bond is essential forthe stability of the GVIA fold and for calcium chan-

BBAPRO 35981 2-9-99


nel blocking activity. Protein minimisation e¡orts de-signed to maintain the conformation of the function-ally important Lys-2 and Tyr-13 residues whilst trun-cating the C-terminus will need to compensate forthe loss of this important constraint and probablyalso for the loss of the central strand of the L-sheet(Lys-24-Tyr-27).

Acknowledgements

We would like to thank Peter Coles and MarkRoss-Smith for expert technical assistance. Thiswork was supported in part by Glaxo-WellcomeAustralia. J.F. was supported by a National HeartFoundation of Australia Postgraduate Scholarship.

References

[1] B.M. Olivera, J.M. McIntosh, L.J. Cruz, F.A. Luque, W.R.Gray, Biochemistry 23 (1984) 5087^5090.

[2] E. Sher, F. Clementi, Neuroscience 42 (1991) 301^307.[3] M.E. Williams, P.F. Brust, D.H. Feldman, S. Patthi, S. Si-

merson, A. Marou¢, A.F. McCue, G. Velicelebi, S.B. Ellis,M.M. Harpold, Science 257 (1992) 389^395.

[4] G.P. Miljanich, J. Ramachandran, Ann. Rev. Pharmacol.Toxicol. 35 (1995) 707^734.

[5] S.S. Bowersox, R.R. Luther, Toxicon 36 (1998) 1651^1658.[6] R.S. Norton, P.K. Pallaghy, J.B. Baell, C.E. Wright, M.J.

Lew, J.A. Angus, Drug Dev. Res. 46 (1999) 206^218.[7] P. Sevilla, M. Bruix, J. Santoro, F. Gago, A.G. Garcia, M.

Rico, Biochem. Biophys. Res. Commun. 192 (1993) 1228^1244.

[8] J.H. Davis, E.K. Bradley, G.P. Miljanich, L. Nadasdi, J.Ramachandran, V.J. Basus, Biochemistry 32 (1993) 7396^7405.

[9] J.J. Skalicky, W.J. Metzler, D.J. Ciesla, A. Galdes, A. Pardi,Protein Sci. 2 (1993) 1591^1603.

[10] P.K. Pallaghy, B.M. Duggan, M.W. Pennington, R.S. Nor-ton, J. Mol. Biol. 234 (1993) 405^420.

[11] P.K. Pallaghy, R.S. Norton, J. Pept. Res. 53 (1999) 343^351.[12] P.K. Pallaghy, K.J. Nielson, D.J. Craik, R.S. Norton, Pro-

tein Sci. 3 (1994) 1833^1839.[13] L. Narasimhan, J. Singh, C. Humblet, K. Guruprasad, T.

Blundell, Nat. Struct. Biol. 1 (1994) 850^852.[14] R.S. Norton, P.K. Pallaghy, Toxicon 36 (1998) 1573^1583.[15] J.I. Kim, M. Takahashi, A. Ogura, T. Kohno, Y. Kudo, K.

Sato, J. Biol. Chem. 269 (1994) 23876^23878.[16] M.J. Lew, J.P. Flinn, P.K. Pallaghy, R. Murphy, S.L. Whor-

low, C.E. Wright, R.S. Norton, J.A. Angus, J. Biol. Chem.272 (1997) 12014^12023.

[17] J.P. Flinn, P.K. Pallaghy, M.J. Lew, R. Murphy, J.A. An-gus, R.S. Norton, Eur. J. Biochem. 262 (1999) 447^455.

[18] J. Song, B. Gilquin, N. Jamin, E. Drakopoulou, M. Guen-neugues, M. Dauplais, C. Vita, A. Menez, Biochemistry 36(1997) 3760^3766.

[19] J.M. Sabatier, C. Lecomte, K. Mabrouk, H. Darbon, R.Oughideni, S. Canarelli, H. Rochat, M.F. Martin-Eauclaire,J. van Rietschoten, Biochemistry 35 (1996) 10641^10647.

[20] L.C. Chuang, P.Y. Chen, C. Chen, T.H. Huang, K.T. Wang,S.H. Chiou, S.H. Wu, Biochem. Biophys. Res. Commun.220 (1996) 246^254.

[21] C.P.M. van Mierlo, N.J. Darby, D. Neuhaus, T.E. Creight-on, J. Mol. Biol. 222 (1991) 353^371.

[22] Q.X. Hua, L. Narhi, W. Jia, T. Arakawa, R. Rosenfeld, N.Hawkins, J.A. Miller, M.A. Weiss, J. Mol. Biol. 259 (1996)297^313.

[23] Y.L. Arriaga, B.A. Harville, L.A. Dreyfus, Infect. Immun.63 (1995) 4715^4720.

[24] K.J. Barnham, A.M. Torres, D. Alewood, P.F. Alewood, T.Domagala, E.C. Nice, R.S. Norton, Protein Sci. 7 (1998)1738^1749.

[25] D. Le-Nguyen, A. Heitz, L. Chiche, M. El Hajji, B. Castro,Protein Sci. 2 (1993) 165^174.

[26] M. Ota, Y. Shimizu, K. Tonosaki, Y. Ariyoshi, Biopolymers46 (1998) 65^73.

[27] M. Price-Carter, W.R. Gray, D.P. Goldenberg, Biochemistry35 (1996) 15537^15546.

[28] M. Price-Carter, W.R. Gray, D.P. Goldenberg, Biochemistry35 (1996) 15547^15557.

[29] M. Price-Carter, M.S. Hull, D.P. Goldenberg, Biochemistry37 (1998) 9851^9861.

[30] T. Sabo, C. Gilon, A. Sha¡erman and E. Elhanaty, in: J.A.Smith and J.E. Rivier (Eds.), Peptides: Chemistry and Biol-ogy. Proceedings of the Twelfth American Peptide Sympo-sium. Escom Science Publishers, Leiden, 1992, pp. 159^160.

[31] J.P. Flinn, R. Murphy, J.H. Boublik, M.J. Lew, C.E.Wright, J.A. Angus, J. Pept. Sci. 1 (1995) 379^384.

[32] B. Kamber, A. Hartmann, K. Eisler, B. Riniker, H. Rink, P.Sieber, W. Rittel, Helv. Chim. Acta 63 (1980) 899^915.

[33] D.S. Wishart, C.G. Bigam, J. Yao, F. Abildgaard, J. Dyson,E. Old¢eld, J.L. Markley, B.D. Sykes, J. Biomol. NMR 6(1995) 135^140.

[34] Y. Nishiuchi, K. Kumagaye, Y. Noda, T.X. Watanabe, S.Sakakibara, Biopolymers 25 (1986) S61^S68.

[35] G.L. Ellman, Arch. Biochem. Biophys. 82 (1959) 70^77.[36] G. Otting, E. Liepinsh, K. Wu«thrich, Biochemistry 32 (1993)

3571^3582.[37] J.P. Flinn, R. Murphy, Lett. Pept. Sci. 3 (1996) 113^116.[38] R. Zhang, G.H. Snyder, J. Biol. Chem. 264 (1989) 18472^

18479.[39] M. Mutter, J. Am. Chem. Soc. 99 (1977) 8307^8314.[40] E. Drakopoulou, J. Vizzavona, J. Neyton, V. Aniort, F.

Bouet, H. Virelizier, A. Menez, C. Vita, Biochemistry 37(1998) 1292^1301.

[41] M.W. Pennington, S.M. Festin, M.L. Maccecchini, W.R.Kem, Toxicon 30 (1992) 755^764.

[42] B.C. Cunningham, J.A. Wells, Curr. Opin. Struct. Biol. 7(1997) 457^462.

BBAPRO 35981 2-9-99


Documents

Role of disulfide bridges in the folding, structure and biological activity of ?-conotoxin GVIA 1 1