Two-dimensional 1H-NMR study of the spatial structure of neurotoxin II from Naja naja oxiana

Fifty Years of Animal Toxin Research at the Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry RAS

This review covers briefly the work carried out at our institute (IBCh), in many cases in collaboration with other Russian and foreign laboratories, for the last 50 years. It discusses the discoveries and studies of various animal toxins, including protein and peptide neurotoxins acting on the nicotinic acetylcholine receptors (nAChRs) and on other ion channels. Among the achievements are the determination of the primary structures of the α-bungarotoxin-like three-finger toxins (TFTs), covalently bound dimeric TFTs, glycosylated cytotoxin, inhibitory cystine knot toxins (ICK), modular ICKs, and such giant molecules as latrotoxins and peptide neurotoxins from the snake, as well as from other animal venoms. For a number of toxins, spatial structures were determined, mostly by 1H-NMR spectroscopy. Using this method in combination with molecular modeling, the molecular mechanisms of the interactions of several toxins with lipid membranes were established. In more detail are presented the results of recent years, among which are the discovery of α-bungarotoxin analogs distinguishing the two binding sites in the muscle-type nAChR, long-chain α-neurotoxins interacting with α9α10 nAChRs and with GABA-A receptors, and the strong antiviral effects of dimeric phospholipases A2. A summary of the toxins obtained from arthropod venoms includes only highly cited works describing the molecules' success story, which is associated with IBCh. In marine animals, versatile toxins in terms of structure and molecular targets were discovered, and careful work on α-conotoxins differing in specificity for individual nAChR subtypes gave information about their binding sites.

Toxins' classification through Raman spectroscopy with principal component analysis

The method based on the combination of Raman spectroscopy and principal component analysis (PCA) was applied to the set of peptide and protein toxins from animal venoms and to synthetic analogues of peptides. The study demonstrated the possibility of toxin classification according to their primary and secondary structures based on Raman spectroscopy. The method described here allows discrimination of snake venom three-finger toxins from predatory marine mollusks α-conotoxins. Moreover, PCA of the Raman spectra of toxins revealed differences within the group of three-finger toxins and also within the group of conotoxins, related to their spatial structure. In particular, on the basis of the developed technique it is possible to distinguish the disulfide isomers of the same peptide toxin. The results obtained have been confirmed by bioinformatic methods. So, we have proposed a method for the rapid analysis of newly discovered venom-derived protein or peptide toxins by establishing their similarity with other already studied toxins by referring to a particular class. Taking into account a low specimen consumption by Raman spectroscopy, the proposed method could represent a first step in the study of toxins from rare and/or endangered venomous animals. The ability to distinguish configuration of disulfide bonds allows to synthesize the correct isomer of the toxin.

Neurotoxin II bound to acetylcholine receptors in native membranes studied by dynamic nuclear polarization NMR

Methods enabling structural studies of membrane-integrated receptor systems without the necessity of purification provide an attractive perspective in membrane protein structural and molecular biology. This has become feasible in principle since the advent of dynamic nuclear polarization (DNP) magic-angle-spinning NMR spectroscopy, which delivers the required sensitivity. In this pilot study, we observed well-resolved solid-state NMR spectra of extensively (13)C-labeled neurotoxin II bound to the nicotinic acetylcholine receptor (nAChR) in native membranes. We show that TOTAPOL, a biradical required for DNP, is localized at membrane and protein surfaces. The concentration of active, membrane-attached biradical decreases with time, probably because of reactive components of the membrane preparation. An optimal distribution of active biradical has strong effects on the NMR data. The presence of inactive TOTAPOL in membrane-proximal situations but active biradical in the surrounding water/glycerol "glass" leads to well-resolved spectra, yet a considerable enhancement (ε = 12) is observed. The resulting spectra of a protein ligand bound to its receptor are paving the way for further DNP investigations of proteins embedded in native membrane patches.

Solid-state magic-angle spinning NMR of membrane proteins and protein-ligand interactions

Structural biology is developing into a universal tool for visualizing biological processes in space and time at atomic resolution. The field has been built by established methodology like X-ray crystallography, electron microscopy and solution NMR and is now incorporating new techniques, such as small-angle X-ray scattering, electron tomography, magic-angle-spinning solid-state NMR and femtosecond X-ray protein nanocrystallography. These new techniques all seek to investigate non-crystalline, native-like biological material. Solid-state NMR is a relatively young technique that has just proven its capabilities for de novo structure determination of model proteins. Further developments promise great potential for investigations on functional biological systems such as membrane-integrated receptors and channels, and macromolecular complexes attached to cytoskeletal proteins. Here, we review the development and applications of solid-state NMR from the first proof-of-principle investigations to mature structure determination projects, including membrane proteins. We describe the development of the methodology by looking at examples in detail and provide an outlook towards future 'big' projects.

Quantification of casein phosphorylation with conformational interpretation using Raman spectroscopy

Raman spectroscopy is emerging as a powerful method for obtaining both quantitative and qualitative information from biological samples. One very interesting area of research, for which the technique has rarely been used, is the detection, quantification and structural analysis of post-translational modifications (PTMs) on proteins. Since Raman spectra can be used to address both of these questions simultaneously, we have developed near infrared Raman spectroscopy with appropriate chemometric approaches (partial least squares regression) to quantify low concentration (4 microM) mixtures of phosphorylated and dephosphorylated bovine alpha(s)-casein. In addition, we have used these data in conjunction with Raman optical activity (ROA) spectra and NMR to assess the structural changes that occur upon phosphorylation.

A model for short α-neurotoxin bound to nicotinic acetylcholine receptor from Torpedo californica: Comparison with long-chain α-neurotoxins and α-conotoxins

Short-chain alpha-neurotoxins from snakes are highly selective antagonists of the muscle-type nicotinic acetylcholine receptors (nAChR). Although their spatial structures are known and abundant information on topology of binding to nAChR is obtained by labeling and mutagenesis studies, the accurate structure of the complex is not yet known. Here, we present a model for a short alpha-neurotoxin, neurotoxin II from Naja oxiana (NTII), bound to Torpedo californica nAChR. It was built by comparative modeling, docking and molecular dynamics using 1H NMR structure of NTII, cross-linking and mutagenesis data, cryoelectron microscopy structure of Torpedo marmorata nAChR [Unwin, N., 2005. Refined structure of the nicotinic acetylcholine receptor at 4A resolution. J. Mol. Biol. 346, 967-989] and X-ray structures of acetylcholine-binding protein (AChBP) with agonists [Celie, P.H., van Rossum-Fikkert, S.E., van Dijk, W.J., Brejc, K., Smit, A.B., Sixma, T.K., 2004. Nicotine and carbamylcholine binding to nicotinic acetylcholine receptors as studied in AChBP crystal structures. Neuron 41 (6), 907-914] and antagonists: alpha-cobratoxin, a long-chain alpha-neurotoxin [Bourne, Y., Talley, T.T., Hansen, S.B., Taylor, P., Marchot, P., 2005. Crystal structure of Cbtx-AChBP complex reveals essential interactions between snake alpha-neurotoxins and nicotinic receptors. EMBO J. 24 (8), 1512-1522] and alpha-conotoxin [Celie, P.H., Kasheverov, I.E., Mordvintsev, D.Y., Hogg, R.C., van Nierop, P., van Elk, R., van Rossum-Fikkert, S.E., Zhmak, M.N., Bertrand, D., Tsetlin, V., Sixma, T.K., Smit, A.B., 2005. Crystal structure of nicotinic acetylcholine receptor homolog AChBP in complex with an alpha-conotoxin PnIA variant. Nat. Struct. Mol. Biol. 12 (7), 582-588]. In complex with the receptor, NTII was located at about 30 A from the membrane surface, the tip of its loop II plunges into the ligand-binding pocket between the alpha/gamma or alpha/delta nAChR subunits, while the loops I and III contact nAChR by their tips only in a 'surface-touch' manner. The toxin structure undergoes some changes during the final complex formation (for 1.45 rmsd in 15-25 ps according to AMBER'99 molecular dynamics simulation), which correlates with NMR data. The data on the mobility and accessibility of spin- and fluorescence labels in free and bound NTII were used in MD simulations. The binding process is dependent on spontaneous outward movement of the C-loop earlier found in the AChBP complexes with alpha-cobratoxin and alpha-conotoxin. Among common features in binding of short- and long alpha-neurotoxins is the rearrangement of aromatic residues in the binding pocket not observed for alpha-conotoxin binding. Being in general very similar, the binding modes of short- and long alpha-neurotoxins differ in the ways of loop II entry into nAChR.

Towards structure determination of neurotoxin II bound to nicotinic acetylcholine receptor: a solid-state NMR approach

Solid-state magic-angle spinning nuclear magnetic resonance (NMR) has sufficient resolving power for full assignment of resonances and structure determination of immobilised biological samples as was recently shown for a small microcrystalline protein. In this work, we show that highly resolved spectra may be obtained from a system composed of a receptor-toxin complex. The NMR sample used for our studies consists of a membrane preparation of the nicotinic acetylcholine receptor from the electric organ of Torpedo californica which was incubated with uniformly 13C-,15N-labelled neurotoxin II. Despite the large size of the ligand-receptor complex ( > 290 kDa) and the high lipid content of the sample, we were able to detect and identify residues from the ligand. The comparison with solution NMR data of the free toxin indicates that its overall structure is very similar when bound to the receptor, but significant changes were observed for one isoleucine.

Snake and snail toxins acting on nicotinic acetylcholine receptors: fundamental aspects and medical applications

This review covers recent data on interactions of nicotinic acetylcholine receptors (AChR) with snake venom proteins (alpha- and kappa-neurotoxins, 'weak' toxins recently shown to act on AChRs), as well as with peptide alpha-conotoxins from Conus snails. Mutations of AChRs and toxins, X-ray/nuclear magnetic resonance structures of alpha-neurotoxin bound to AChR fragments, and the X-ray structure of the acetylcholine-binding protein were used by several groups to build models for the alpha-neurotoxin-AChR complexes. Application of snake toxins and alpha-conotoxins for pharmacological distinction of muscle, neuronal and neuronal-like AChR subtypes and for other medical purposes is briefly discussed.

A novel short neurotoxin, cobrotoxin c, from monocellate cobra (Naja kaouthia) venom: isolation and purification, primary and secondary structure determination, and tertiary structure modeling

A novel short neurotoxin, cobrotoxin c (CBT C) was isolated from the venom of monocellate cobra (Naja kaouthia) using a combination of ion-exchange chromatography and FPLC. Its primary structure was determined by Edman degradation. CBT C is composed of 61 amino acid residues. It differs from cobrotoxin b (CBT B) by only two amino acid substitutions, Thr/Ala11 and Arg/Thr56, which are not located on the functionally important regions by sequence similarity. However, the LD50 is 0.08 mg/g to mice, i.e. approximately five-fold higher than for CBT B. Strikingly, a structure-function relationship analysis suggests the existence of a functionally important domain on the outside of Loop III of CBT C. The functionally important basic residues on the outside of Loop III might have a pairwise interaction with alpha subunit, instead of gamma or delta subunits of the nicotinic acetylcholine receptor (nAChR).

Structure-function relationship of three neurotoxins from the venom of Naja kaouthia: a comparison between the NMR-derived structure of NT2 with its homologues, NT1 and NT3

Three homologous short-chain neurotoxins, named NT1, NT2 and NT3, were purified from the venom of Naja kaouthia. NT1 has an identical amino acid sequence to cobrotoxin from Naja naja atra [Biochemistry 32 (1993) 2131]. NT3 shares the same sequence with cobrotoxin b [J. Biochem. (Tokyo) 122 (1997) 1252], whereas NT2 is a novel 61-residue neurotoxin. Tests of their physiological functions indicate that NT1 shows a greater inhibition of muscle contraction induced by electrical stimulation of the nerve than do NT2 and NT3. Homonuclear proton two-dimensional NMR methods were utilized to study the solution tertiary structure of NT2. A homology model-building method was employed to predict the structure of NT3. Comparison of the structures of these three toxins shows that the surface conformation of NT1 facilitates the substituted base residues, Arg28, Arg30, and Arg36, to occupy the favorable spatial location in the central region of loop II, and the cation groups of all three arginines face out of the molecular surface of NT1. This may contribute greatly to the higher binding of NT1 with AchR compared to NT2 and NT3.

Modeling the third loop of short-chain snake venom neurotoxins: roles of the short-range and long-range interactions

The influence of long-range interactions on local structures is an important issue in understanding protein folding process and protein structure stability. Using short-chain snake venom neurotoxin as a model system, we have studied the conformational properties of eight different loop III sequences either in the environment of one of the short-chain neurotoxin, erabutoxin b (PDB ID 1nxb), or in free state by Monte Carlo simulated annealing method. The surrounding protein structure was found to be crucial in stabilizing the loop conformation. Although all the eight peptides prefer type V beta turn in solution, three of them (KPGI, KPGV, KSGI) turn to type II beta turn and the other five (KKGI, KKGV, KNGI, KQGI, and KRGV) are confined to more rigid type V beta turn conformation in the protein structure. Using flexible tetra-glycine-peptide to screen the backbone conformational space in the protein environment also validates the results. This study shows that long-range interactions do contribute to the stability and the types of conformation for a surface loop in protein, while short-range interactions may only provide candidate conformations, which then have to be filtered by the long-range interactions further.

High resolution x-ray analysis of two mutants of a curaremimetic snake toxin

A previous mutational analysis of erabutoxin a (Ea), a curaremimetic toxin from sea snake venom, showed that the substitutions S8G and S8T caused, respectively, 176-fold and 780-fold affinity decreases for the nicotinic acetylcholine receptor (AchR). In view of the fact that the side-chain of Ser8 is buried in the wild-type toxin, we wondered whether these affinity changes reflect a direct binding contribution of S8 to the receptor and/or conformational changes that could have occurred in Ea as a result of the introduced mutations. To approach this question, we solved X-ray structures of the two mutants S8G and S8T at high resolution (0.18 nm and 0.17 nm, with R factors of 18.0% and 17.9%, respectively). The data show that none of the mutations significantly modified the toxin structure. Even within the site where the toxin binds to the receptor the backbone conformation remained unchanged. Therefore, the low affinities of the mutants S8T and S8G cannot be explained by a large conformational change of the toxin structure. Although we cannot exclude the possibility that undetectable structural changes have occurred in the toxin mutants, our data support the view that, although buried between loop I and II, S8 is part of the functional epitope of the toxin.

Snake venom alpha-neurotoxins and other 'three-finger' proteins

The review is mainly devoted to snake venom alpha-neurotoxins which target different muscle-type and neuronal nicotinic acetylcholine receptors. The primary and spatial structures of other snake venom proteins as well as mammalian proteins of the Ly-6 family, which structurally resemble the 'three-finger' snake proteins, are also briefly discussed. The main emphasis is placed on recent data characterizing the alpha-neurotoxin interactions with nicotinic acetylcholine receptors.

Snake venom cardiotoxins-structure, dynamics, function and folding

Snake cardiotoxins are highly basic (pI > 10) small molecular weight (approximately 6.5 kDa), all beta-sheet proteins. They exhibit a broad spectrum of interesting biological activities. The secondary structural elements in these toxins include antiparallel double and triple stranded beta-sheets. The three dimensional structures of these toxins reveal an unique asymmetric distribution of the hydrophobic and hydrophilic amino acids. The 3D structures of closely related snake venom toxins such as neurotoxins and cardiotoxin-like basic proteins (CLBP) fail to show similar pattern(s) in the distribution of polar and nonpolar residues. Recently, many novel biological activities have been reported for cardiotoxins. However, to-date, there is no clear structure-function correlation(s) available for snake venom cardiotoxins. The aim of this comprehensive review is to summarize and critically evaluate the progress in research on the structure, dynamics, function and folding aspects of snake venom cardiotoxins.

Substance P derivatives with photoactivatable labels in the N-terminal part of the molecule

Photoactivatable substance P (SP) derivatives containing the p-benzoylbenzoic moiety at the N-terminal alpha-amino group of Arg 1 or at the epsilon-amino group of Lys 3 were prepared. Both derivatives also had a p-hydroxyphenylpropionyl group for radioiodination. To obtain the analogue with the photolabel at Arg 1, SP was first reacted with N-hydroxysuccinimide p-hydroxyphenylpropionate, the Lys 3-modified derivative was isolated by reversed-phase high-performance liquid chromatography (HPLC), reacted with N-hydroxysuccinimide p-benzoylbenzoate and purified by HPLC. To place the photolabel at Lys 3, the order of the reactions was reversed. The structure of the derivatives obtained was confirmed by mass spectrometry. The interaction of the derivatives obtained and of their 125I-labeled forms with the NK-1 neurokinin receptor from the rat brain, as well as with the nicotinic acetylcholine receptor from Torpedo electrocytes was analyzed. The results obtained supported by the data from the literature indicate that benzoylbenzoic acid derivatives should not be considered as universal photolabels, which ensure in all cases a high level of photo-cross-linking.

The emerging three-dimensional structure of a receptor. The nicotinic acetylcholine receptor

The nicotinic acetylcholine receptor is the neurotransmitter receptor with the most-characterized protein structure. The amino acid sequences of its five subunits have been elucidated by cDNA cloning and sequencing. Its shape and dimensions (approximately 12.5 nm x 8 nm) were deduced from electron-microscopy studies. Its subunits are arranged around a five-fold axis of pseudosymmetry in the order (clockwise) alpha H gamma alpha L delta beta. Its two agonist/competitive-antagonist-binding sites have been localized by photolabelling studies to a deep gorge between the subunits near the membrane surface. Its ion channel is formed by five membrane-spanning (M2) helices that are contributed by the five subunits. This finding has been generalized as the Helix M2 model for the superfamily of ligand-gated ion channels. The binding site for regulatory non-competitive antagonists has been localized by photolabelling and site-directed-mutagenesis studies within this ion channel. Therefore a three-dimensional image of the nicotinic acetylcholine receptor is emerging, the most prominent feature of which is an active site that combines the agonist/ competitive-antagonist-binding sites, the regulatory site and the ion channel within a relatively narrow space close to and within the bilayer membrane.

The handedness of the subunit arrangement of the nicotinic acetylcholine receptor from Torpedo californica

Cross-linking an alpha-neurotoxin with a known three-dimensional structure and with photoactivatable groups in known positions to native membrane-bound acetylcholine receptor reveals its quaternary structure, including the handedness of its circular subunit arrangement. Photolabelling with alpha-neurotoxin carrying the photoactivatable group at position Lys46 is inhibited by the competitive antagonist (+)-tubocurarine in a biphasic manner, indicating that it reacts with both alpha-subunits that were shown to have different affinities for this antagonist [Neubig, R. R. & Cohen, J. B. (1979) Biochemistry 18, 5464-5475]. Lys46 is located on loop III of the neurotoxin. The other information necessary for the elucidation of the handedness was provided by the recent finding that the central loop of the toxin (loop II) is oriented towards the central pore of the receptor, securing the overall orientation of the bound toxin [Machold, J., Utkin, Y. N., Kirsch, D., Kaufmann, R., Tsetlin, V. & Hucho, F. (1995b) Proc. Natl Acad. Sci. USA 92, 7282-7286]. Looking at the receptor from the synaptic side of the postsynaptic membrane, it was concluded that the clockwise subunit arrangement is alpha H-gamma-alpha L-delta-beta (alpha H and alpha L are the alpha-subunits binding (+)-tubocurarine with high and low affinity, respectively). Its mirror image alpha alpha L-gamma-alpha H-beta-delta could thus be excluded.

Amino acid residue: is it structural or functional?

A new approach is suggested for delineating the structural and functional amino acid residues in proteins with known three-dimensional structure, basing on the involvement of residues in intramolecular hydrophobic and hydrophilic interactions and additional information about the conservativity of the residues. The approach is applied to the families of homologous neurotoxins and cardiotoxins. The results obtained concerning the role of amino acid residues in both families of toxins accord well with the similarity of their fold, but different mechanisms of action. Current approach can be used for detailed characterization of protein spatial structures, as well as for rational protein engineering.

Photolabeling reveals the proximity of the alpha-neurotoxin binding site to the M2 helix of the ion channel in the nicotinic acetylcholine receptor

A photoactivatable derivative of neurotoxin II from Naja naja oxiana containing a 125I-labeled p-azidosalicylamidoethyl-1,3'-dithiopropyl label at Lys-25 forms a photo-induced cross-link with the delta subunit of the membrane-bound Torpedo californica nicotinic acetylcholine receptor (AChR). The cross-linked radioactive receptor peptide was isolated by reverse-phase HPLC after tryptic digestion of the labeled delta subunit. The sequence of this peptide, delta-(260-277), and the position of the label at Ala-268 were established by matrix-assisted laser-desorption-ionization mass spectrometry based on the molecular mass and on post-source decay fragment analysis. With the known dimensions of the AChR molecule, of the photolabel, and of alpha-neurotoxin, finding the cross-link at delta Ala-268 (located in the upper part of the channel-forming transmembrane helix M2) means that the center of the alpha-neurotoxin binding site is situated at least approximately 40 A from the extracellular surface of the AChR, proximal to the channel axis.

Synthesis of nitrodiazirinyl derivatives of neurotoxin II from Naja naja oxiana and their interaction with the Torpedo californica nicotinic acetylcholine receptor

Five singly modified nitrodiazirine derivatives of neurotoxin II (NT-II) from Naja naja oxiana were obtained after NT-II reaction with N-hydroxysuccinimide ester of (2-nitro-4-[3-(trifluoromethyl)-3H-diazirin-3yl]phenoxy)acet ic acid followed by chromatographic separation of the products. To localize the label positions, each derivative was first UV-irradiated and then subjected to reduction, carboxymethylation, and trypsinolysis. Tryptic digests were separated by reversed phase-HPLC, the labeled peptides being identified by mass spectrometry. The derivatives containing the photolabel at the position Lys 25, Lys 26, Lys 44, or Lys 46 were [125I]iodinated by the chloramine T procedure. Each iodinated derivative was found to form photoinduced cross-links with the membrane-bound nicotinic acetylcholine receptor (AChR) from Torpedo californica. The pattern of labeling the receptor's alpha, beta, gamma, or delta subunits was dependent on the photolabel position in the NT-II molecule and differed from that obtained earlier with an analogous series of p-azidobenzoyl derivatives of NT-II. The results obtained indicate that (i) different sides of the neurotoxin molecule are involved in the AChR binding, and (ii) fragments of the different AChR subunits are located close together at the neurotoxin-binding sites.

Detailed assessment of spatial hydrophobic and electrostatic properties of 2D NMR-derived models of neurotoxin II

2D NMR-derived spatial structures of neurotoxin II (NtII) and several homologous toxins in solution were assessed by comparison with their own amino acid sequences using a three-dimensional (3D) profile method. 3D profiles of all the toxin models match the sequences well and, therefore, the method of 3D profile was demonstrated to work correctly for these well-resolved NMR structures in aqueous solution. At the same time, the profile window plots reveal low scores in the bottom tip of loop II (residues 22-34 in NtII) and in beta-strand of loop III (residues 49-52). Some residues in the first poor-scoring region are of functional importance being involved in binding with nicotinic acetylcholine receptor (AChR). Furthermore, the second segment participates in intermolecular hydrogen bonding upon dimerization of postsynaptic neurotoxins in solution resulting in increasing of the 3D-1D score for residues at the interface between monomers. Therefore, the 3D profile method can be useful for detection functionally-important regions in well-resolved protein structures.

A new class of photoactivatable and cleavable derivatives of neurotoxin II from Naja naja oxiana. Synthesis, characterisation, and application for affinity labelling of the nicotinic acetylcholine receptor from Torpedo californica

A new series of photoactivatable and cleavable derivatives of neurotoxin II from the cobra Naja naja oxiana is investigated which can be used for mapping the surface topology of the nicotinic acetylcholine receptor from Torpedo electric tissue. The preparation and characterisation of five toxin derivatives, each with a radioactive 125I-azidosalicylamidoethyl-1,3'-dithiopropyl group in a defined position within the primary structure, are described. The photoinduced cross-linking reaction of the toxin derivatives with membrane-bound receptor is investigated. The photoactivatable group located at position K25 reacts almost exclusively with the delta subunit of the receptor, whereas the K15 derivative reacts with the alpha and beta subunits. The other derivatives did not react with the receptor to any significant extent. It is shown that, with respect to the receptor subunits, the cross-linking pattern depends on the length and chemical nature of the cross-linking group.

1H-NMR assignments and secondary structure of dendroaspin, an RGD-containing glycoprotein IIb-IIIa (alpha IIb-beta 3) antagonist with a neurotoxin fold

Dendroaspin, also referred to as mambin, was originally isolated from the venom of the Elapidae snake Dendroaspis jamesoni kaimose. It shares a high level of sequence similarity with the short-chain neurotoxins found in other Elapidae but displays approximately 1000-fold lower neurotoxin activity than the closely related protein erabutoxin b. However, unlike neurotoxins, it contains an RGD (Arg-Gly-Asp) motif and functions as an antagonist of platelet aggregation and cell-cell adhesion of comparable potency to the disintegrins from the venoms of Viperidae. We have determined the secondary structure of dendroaspin using 1H-NMR spectroscopy. Its structure resembles that of the short-chain neurotoxins, with three loops extending from a disulphide-bridged core; however, the strands of the triple-stranded beta-sheet are shorter and the loop containing the RGD sequence is moved away from this sheet. The structure bears little resemblance to that of the disintegrins, except in the RGD-containing loop, suggesting that this loop may be of prime importance in its inhibitory function. Comparison of this preliminary structure with that of the neurotoxins and disintegrins furthers our understanding of the mechanism of integrin antagonists and shows how the neurotoxin fold can be manipulated to give a variety of inhibitors.