The use of sample rotation for minimizing convection effects in self-diffusion NMR measurements.

Undesirable temperature gradients in a NMR sample tube are usually generated by an inappropriate temperature regulation system. We have shown that such convection effects can greatly distort the measurement of translational self-diffusion coefficients. The use of sample spinning helps to minimize such undesirable effects by disruption of convection fluxes due to resulting Coriolis forces that have a strongly stabilizing effect on the conducting state of the system (J. Lounila et al., J. Magn. Reson. A 118, 50 (1996)). This simple trick allows the accurate measurement of diffusion coefficients for a wide range of temperatures and solvents without the need for a convection-compensated NMR pulse sequences or more sophisticated temperature control units. Experimental data obtained for some target compounds dissolved in several common deuterated solvents at different temperatures are reported and discussed.

Characterization of the folding and unfolding reactions of a small beta-barrel protein of novel topology, the MTCP1 oncogene product P13.

The equilibrium and kinetic folding properties of a small oncogene product, P13(MTCP1), of novel topology have been investigated using perturbation by guanidine hydrochloride and observation by fluorescence, circular dichroism and two-dimensional heteronuclear NMR spectroscopy. The structure of P13(MTCP1) is comprised of a canonical filled beta-barrel, although the topology of the structure is absolutely unique, rendering the folding properties of this protein of great interest. Equilibrium measurements of the intrinsic fluorescence emission spectrum, the fluorescence decay, the circular dichroism spectrum and the (15)N-(1)H heteronuclear single quantum coherence (HSQC) correlation spectrum as a function of increasing concentrations of denaturant showed no evidence for the population of any equilibrium intermediates, although negative amplitudes on the blue edge of the tryptophan emission and loss of intensity of the native HSQC correlation peaks were indicative of increased conformational dynamics at low denaturant concentrations. The free energy and cooperativity of unfolding as observed by fluorescence and circular dichroism were in relatively good agreement, also consistent with a two-state transition. Kinetics measurements of the fluorescence emission as a function of denaturant concentration revealed that P13(MTCP1) is the slowest folding beta-structure protein reported to date. Comparison of the activation cooperativity values (m(f) and m(u)) indicates that the structure of the transition state is quite close to the folded state in terms of exposed surface area. The calculated contact order of P13(MTCP1) is relatively low and does not appear to explain its slow rate of folding. We suggest that the complex topology of this protein, which would require the ordering of the beta-barrel through a long loop joining the two L-shaped components of the barrel, could provide an explanation for this slow folding.

A comprehensive analysis of multifield 15N relaxation parameters in proteins: determination of 15N chemical shift anisotropies.

This study deals with the exploitation of the three classical 15N relaxation parameters (the longitudinal relaxation rate, R1, the transverse relaxation rate, R2, and the 1H-15N cross-relaxation rate, sigmaNH) measured at several magnetic fields in uniformly 15N-labeled proteins. Spectral densities involved in R1, R2 and sigmaNH are analyzed according to the functional form A + B/(1 + omega(2) taus(2)), where taus is the correlation time associated with slow motions sensed by the NH vector at the level of the residue to which it belongs. The coefficient B provides a realistic view of the backbone dynamics, whereas A is associated with fast local motions. According to the "model free approach", B can be identified with 2tausS(2) where S is the generalized order parameter. The correlation time taus is determined from the field dependency of the relaxation parameters while A and B are determined through linear equations. This simple data processing is needed for obtaining realistic error bars based on a statistical approach. This proved to be the key point for validating an extended analysis aiming at the determination of nitrogen chemical shift anisotropy. The protein C12A-p8(MTCP1) has been chosen as a model for this study. It will be shown that all data (obtained at five magnetic field strengths corresponding to proton resonance of 400, 500, 600, 700, and 800 MHz) are very consistently fitted provided that a specific effective correlation time associated with slow motions is defined for each residue. This is assessed by small deviations between experimental and recalculated values, which, in all cases, remain within experimental uncertainty. This strategy makes needless elaborate approaches based on the combination of several slow motions or their possible anisotropy. Within the core of the protein taus fluctuates in a relatively narrow range (with a mean value of 6.15 ns and a root-mean-square deviation of 0.36 ns) while it is considerably reduced at the protein extremities (down to approximately 3 ns). To a certain extent, these fluctuations are correlated with the protein structure. A is not obtained with sufficient accuracy to be valuably discussed. Conversely, order parameters derived from B exhibit a significant correlation with the protein structure. Finally, the multi-field analysis of the evolution of longitudinal and transverse relaxation rates has been refined by allowing the 15N chemical shift anisotropy (csa) to vary residue by residue. Within uncertainties (derived here on a statistical basis) an almost constant value is obtained. This strongly indicates an absence of correlation between the experimental value of this parameter obtained for a given residue in the protein, the nature of this residue, and the possible involvement of this residue in a structured area of the protein.

Extending the excitation sculpting concept for selective excitation.

Nowadays, excitation sculpting is probably the most efficient way to achieve selectivity in an NMR experiment, since it associates very clean frequency selection with "user-friendliness." In the present report, it is shown that the excitation sculpting concept, originally based on a double pulse field gradient echo acting on a selected transverse magnetization, can be extended through new experiments designed to act on longitudinal magnetization. This leads to outstanding performances, especially when the transverse relaxation rate is a limiting factor as, for example, in the case of biological macromolecules. Several new sequences are proposed, aiming at the selection of magnetization aligned either/both on a transverse axis or/and on the z-axis. Their potentialities are illustrated in light of different applications including multiplet-selective excitation, band-selective excitation, and water suppression.

Backbone dynamics and solution structure refinement of the 15N-labeled human oncogenic protein p13MTCP1: comparison with X-ray data.

Two related oncogenes, TCL1 and MTCP1, are overexpressed in certain T-cell prolymphocytic leukemias as a result of chromosomal rearrangements that involve the translocation of one T-cell receptor gene to either chromosome 14q32 or Xq28, respectively. The human oncoprotein p13MTCP1 is coded by the MTCP1 gene and its primary sequence is highly and only homologous to that of p14TCL1, the product of TCL1. These two proteins likely represent the first members of a new family of oncogenic proteins. A previous model of the three-dimensional solution structure of p13MTCP1 was determined recently using exclusively homonuclear proton two-dimensional NMR methods and, almost simultaneously, high-resolution crystal structures of p13MTCP1 and p14TCL1 appeared in the literature. In order to gain more insight into the details of the solution structure, we uniformly labeled p13MTCP1 with nitrogen-15. The refined structure benefits from 520 additional NOEs, extracted from either 15N-edited 3D experiments or homonuclear 2D NOESY recorded at 800 MHz, and from a nearly complete set of phi angular restraints. Measurements of 15N spin relaxation times and heteronuclear 15N[1H]NOEs at two magnetic field strengths provided additional insights into the dynamics of the protein backbone. On the basis of these new results, a putative binding surface for this particular class of oncogenes is discussed.

Synthesis and NMR solution structure of an alpha-helical hairpin stapled with two disulfide bridges.

Helical coiled-coils and bundles are some of the most common structural motifs found in proteins. Design and synthesis of alpha-helical motifs may provide interesting scaffolds that can be useful as host structures to display functional sites, thus allowing the engineering of novel functional miniproteins. We have synthesized a 38-amino acid peptide, alpha2p8, encompassing the alpha-helical hairpin present in the structure of p8MTCP1, as an alpha-helical scaffold particularly promising for its stability and permissiveness of sequence mutations. The three-dimensional structure of this peptide has been solved using homonuclear two-dimensional NMR techniques at 600 MHz. After sequence specific assignment, a total of 285 distance and 29 dihedral restraints were collected. The solution structure of alpha2p8 is presented as a set of 30 DIANA structures, further refined by restrained molecular dynamics, using simulated annealing protocol with the AMBER force field. The RMSD values for the backbone and all heavy atoms are 0.65+/-0.25 and 1.51+/-0.21 A, respectively. Excised from its protein context, the alpha-hairpin keeps its native structure: an alpha-helical coiled-coil, similar to that found in superhelical structures, with two helices spanning residues 4-16 and 25-36, and linked by a short loop. This motif is stabilized by two interhelical disulfide bridges and several hydrophobic interactions at the helix interface, leaving most of its solvent-exposed surface available for mutation. This alpha-helical hairpin, easily amenable to synthetic chemistry and biological expression system, may represent a stable and versatile scaffold to display new functional sites and peptide libraries.

Rational engineering of a miniprotein that reproduces the core of the CD4 site interacting with HIV-1 envelope glycoprotein.

Protein-protein interacting surfaces are usually large and intricate, making the rational design of small mimetics of these interfaces a daunting problem. On the basis of a structural similarity between the CDR2-like loop of CD4 and the beta-hairpin region of a short scorpion toxin, scyllatoxin, we transferred the side chains of nine residues of CD4, central in the binding to HIV-1 envelope glycoprotein (gp120), to a structurally homologous region of the scorpion toxin scaffold. In competition experiments, the resulting 27-amino acid miniprotein inhibited binding of CD4 to gp120 with a 40 microM IC(50). Structural analysis by NMR showed that both the backbone of the chimeric beta-hairpin and the introduced side chains adopted conformations similar to those of the parent CD4. Systematic single mutations suggested that most CD4 residues from the CDR2-like loop were reproduced in the miniprotein, including the critical Phe-43. The structural and functional analysis performed suggested five additional mutations that, once incorporated in the miniprotein, increased its affinity for gp120 by 100-fold to an IC(50) of 0.1-1.0 microM, depending on viral strains. The resulting mini-CD4 inhibited infection of CD4(+) cells by different virus isolates. Thus, core regions of large protein-protein interfaces can be reproduced in miniprotein scaffolds, offering possibilities for the development of inhibitors of protein-protein interactions that may represent useful tools in biology and in drug discovery.

Refined solution structure and backbone dynamics of 15N-labeled C12A-p8MTCP1 studied by NMR relaxation.

MTCP1 (for Mature-T-Cell Proliferation) was the first gene unequivocally identified in the group of uncommon leukemias with a mature phenotype. The three-dimensional solution structure of the human p8MTCP1 protein encoded by the MTCP1 oncogene has been previously determined by homonuclear proton two-dimensional NMR methods at 600 MHz: it consists of an original scaffold comprising three alpha-helices, associated with a new cysteine motif. Two of the helices are covalently paired by two disulfide bridges, forming an alpha-hairpin which resembles an antiparallel coiled-coil. The third helix is orientated roughly parallel to the plane defined by the alpha-antiparallel motif and appears less well defined. In order to gain more insight into the details of this new scaffold, we uniformly labeled with nitrogen-15 a mutant of this protein (C12A-p8MTCP1) in which the unbound cysteine at position 12 has been replaced by an alanine residue, thus allowing reproducibly high yields of recombinant protein. The refined structure benefits from 211 additional NOEs, extracted from 15N-edited 3D experiments, and from a nearly complete set of phi angular restraints allowing the estimation of the helical content of the structured part of the protein. Moreover, measurements of 15N spin relaxation times and heteronuclear 15N¿1H¿NOEs provided additional insights into the dynamics of the protein backbone. The analysis of the linear correlation between J(0) and J(omega) was used to interpret relaxation parameters. It appears that the apparent relative disorder seen in helix III is not simply due to a lack of experimental constraints, but associated with substantial contributions of sub-nanosecond motions in this segment.

Solution structure of the recombinant human oncoprotein p13MTCP1.

The human oncoprotein p13MTCP1 is coded by the MTCP1 gene, a gene involved in chromosomal translocations associated with T-cell prolymphocytic leukemia, a rare form of human leukemia with a mature T-cell phenotype. The primary sequence of p13MTCP1 is highly and only homologous to that of p14TCL1, a product coded by the gene TCL1 which is also involved in T-cell prolymphocytic leukemia. These two proteins probably represent the first members of a new family of oncogenic proteins. We present the three-dimensional solution structure of the recombinant p13MTCP1 determined by homonuclear proton two-dimensional NMR methods at 600 MHz. After proton resonance assignments, a total of 1253 distance restraints and 64 dihedral restraints were collected. The solution structure of p13MTCP1 is presented as a set of 20 DYANA structures. The rmsd values with respect to the mean structure for the backbone and all heavy atoms for the conformer family are 1.07 +/- 0.19 and 1.71 +/- 0.17 A, when the structured core of the protein (residues 11-103) is considered. The solution structure of p13MTCP1 consists of an orthogonal beta-barrel, composed of eight antiparallel beta-strands which present an original arrangement. The two beta-pleated loops which emerge from this barrel might constitute the interaction surface with a potential molecular partner.

Formation of native disulfide bonds in endothelin-1. Structural evidence for the involvement of a highly specific salt bridge between the prosequence and the endothelin-1 sequence.

The [Lys-Arg]-endothelin-1 analogue (KR-ET-1) yields almost selectively the native disulfide pattern (96%), in contrast to endothelin-1 (ET-1) that gives at least 25% of the non-native disulfide pattern. We have previously shown that the carboxylate-state structure of KR-ET-1 is more constrained and stabilized by a salt bridge between Arg(-1) and the Asp8 or Glu10 side chain [Aumelas et al. (1995) Biochemistry 34, 4546-4561]. To identify this salt bridge and its potential involvement in the disulfide bond formation, [E10Q], [D18N], and [D8N] carboxamide analogues were studied, which led to the unambiguous identification of the Arg(-1)-Asp8 salt bridge. Furthermore, while [E10Q] and [D18N] analogues gave a high yield of the native isomer (>/=90%), the [D8N] analogue afforded a ratio of the two isomers close to that observed for ET-1 (68%) [Kubo et al. (1997) Lett. Pept. Sci. 4, 185-192]. Assuming that the formation of disulfide bonds occurs in a thermodynamically controlled step, we have hypothesized that the Arg(-1)-Asp8 salt bridge and concomitant interactions could be responsible for the increase in yield of the native isomer of KR-ET-1. In the present work, we describe the structural studies of the carboxamide analogues and of the minor non-native KR-ET-1 isomer. On the basis of 1H NMR and CD spectra as a function of pH, [E10Q] and [D18N] analogues display a conformational change similar to that of the parent peptide, whereas the structure of the [D8N] analogue is unchanged. For the non-native isomer, we measured a lower helical content than for the native isomer and observed a marked difference in the orientation of the KRCSC backbone. In addition, no salt bridge was experimentally observed. Altogether, these results allow us to hypothesize that the salt bridge between two highly conserved residues, one belonging to the prosequence [Arg(-1)] and the other to the mature sequence [Asp8], is involved in the formation of the native disulfide isomer of ET-1. The involvement of the prosequence in the formation of the native disulfide isomer strongly suggests that, in the maturation pathway of ET-1, cleavage of the Arg52-Cys53 amide bond occurs after native disulfide bond formation.

Oligomerization of protegrin-1 in the presence of DPC micelles. A proton high-resolution NMR study.

Protegrins are members of a family of five Cys-rich naturally occurring cationic antimicrobial peptides. The NMR solution structure of protegrin-1 (PG-1) has been previously determined as a monomeric beta-hairpin both in water and in dimethylsulfoxide solution. Protegrins are bactericidal peptides but their mechanism of action is still unknown. In order to investigate the structural basis of their cytotoxicity, we studied the effect of lipid micelles on the structure of PG-1. The NMR study reported in the present work indicates that PG-1 adopts a dimeric structure when it binds to dodecylphosphocholine micelles. Moreover, the amide proton exchange study suggests the possibility of an association between several dimers.

Isolation and structure elucidation of a highly haemolytic saponin from the Merck saponin extract using high-field gradient-enhanced NMR techniques.

Saponins SAPO50 and SAPO30, of which SAPO50 is highly haemolytic, have been isolated from the commercial Merck Saponin. Their structures have been determined exclusively by high-field gradient-enhanced NMR methods. The 1H and 13C NMR spectra of these saponins in pyridine-deuterium oxide have been assigned by homonuclear and heteronuclear correlation experiments. Anomeric configurations were obtained by combined use of 1JCH, 3JH-1.H-2, and 1D-NOESY data. Sugar residues were identified by use of 3JHH values obtained from their subspectra recorded using an optimized 1D-zeta-TOCSY sequence. Linkage assignments were made using the ge-HMBC and 1D-NOESY spectra. This study shows that SAPO50 represents a hitherto undescribed saponin with the following structure: 3-O-beta-D-xylopyranosyl-(1-->3)-[beta-D-galactopyranosyl- (1-->2)]-beta-D-glucuronopyranosyl gypsogenin 28-O-(6-deoxy-beta-D-glucopyranosyl)-(1-->4)-[beta-D-xylopyranosyl-(1--> 3)- beta-D-xylopyranosyl-(1-->4)]-alpha-L-rhamnopyranosyl-(1-->2)-beta-D- fucopyranoside. SAPO30, however, corresponds to a saponin previously described [D. Frechet, B. Christ, B. Monegier du Sorbier, H. Fischer, and M. Vuilhorgne, Phytochemistry, 30 (1991) 927-931].

On the convergent evolution of animal toxins. Conservation of a diad of functional residues in potassium channel-blocking toxins with unrelated structures.

BgK is a K+ channel-blocking toxin from the sea anemone Bunodosoma granulifera. It is a 37-residue protein that adopts a novel fold, as determined by NMR and modeling. An alanine-scanning-based analysis revealed the functional importance of five residues, which include a critical lysine and an aromatic residue separated by 6.6 +/- 1.0 A. The same diad is found in the three known homologous toxins from sea anemones. More strikingly, a similar functional diad is present in all K+ channel-blocking toxins from scorpions, although these toxins adopt a distinct scaffold. Moreover, the functional diads of potassium channel-blocking toxins from sea anemone and scorpions superimpose in the three-dimensional structures. Therefore, toxins that have unrelated structures but similar functions possess conserved key functional residues, organized in an identical topology, suggesting a convergent functional evolution for these small proteins.

Solution structure of human p8MTCP1, a cysteine-rich protein encoded by the MTCP1 oncogene, reveals a new alpha-helical assembly motif.

MTCP1 (for Mature-T-Cell Proliferation) is the first gene unequivocally identified in the group of uncommon leukemias with a mature phenotype. The three-dimensional solution structure of the human p8(MTCP1) protein encoded by the MTCP1 oncogene was determined by homonuclear proton two-dimensional NMR methods at 600 MHz. After sequence specific assignments, a total of 931 distance restraints and 57 dihedral restraints were collected. The location of the three previously unassigned disulfide bridges was determined from preliminary DIANA structures, using a statistical analysis of intercystinyl distances. The solution structure of p8(MTCP1) is presented as a set of 30 DIANA structures, further refined by restrained molecular dynamics using a simulated annealing protocol with the AMBER force field. The r.m.s.d. values with respect to the mean structure for the backbone and all heavy atoms for a family of 30 structures are 0.73(+/-0.28) and 1.17(+/-0.23) A, when the structured core of the protein (residues 5 to 63) is considered. The solution structure of p8(MTCP1) reveals an original scaffold consisting of three alpha helices, associated with a new cysteine motif. Two of the helices are covalently paired by two disulfide bridges, forming an alpha-hairpin which resembles an antiparallel coiled-coil. The third helix is oriented roughly parallel to the plane defined by the alpha-antiparallel motif and its axis forms an angle of approximately 60 degrees with respect to the main axis of this motif.

Synthesis and solution structure of the antimicrobial peptide protegrin-1.

Protegrins are members of a family of five Cys-rich, cationic antimicrobial peptides recently isolated from porcine cells. We have synthesised an 18-amino-acid peptide that corresponds to protegrin-1. After Cys oxidation, the peptide has bactericidal activity against gram-positive and gram-negative bacteria, similar to that described for the natural peptide. The solution structure of protegrin-1 was investigated by means of 1H-NMR spectroscopy in water and in (CD3)2SO, with distance-geometry and simulated-annealing calculations. The C6-C15 and C8-C13 disulfide pattern was determined on the basis of NMR-derived constraints. These two parallel disulfide bridges stabilised a beta-sheet structure which comprised two antiparallel strands (residues 5-9 and 12-16) linked by a distorted beta-turn (residues 9-12). The N-terminus and C-terminus were essentially disordered. The distribution of hydrophobic and hydrophilic residues at the peptide surface was found to be a structural feature shared with tachyplesin-1, a related peptide which displays cytolytic activity, and, to a lesser extent, with mammalian defensins. These findings led us to assume that the distribution pattern could be required for the cytolytic activity of these peptides.

Change in membrane permeability induced by protegrin 1: implication of disulphide bridges for pore formation.

Protegrin 1 (PG-1) is a naturally occurring cationic antimicrobial peptide that is 18 residues long, has an aminated carboxy terminus and contains two disulphide bridges. Here, we investigated the antimicrobial activity of PG-1 and three linear analogues. Then, the membrane permeabilisation induced by these peptides was studied upon Xenopus laevis oocytes by electrophysiological methods. From the results obtained, we concluded that protegrin is able to form anion channels. Moreover, it seems clear that the presence of disulphide bridges is a prerequisite for the pore formation at the membrane level and not for the antimicrobial activity.

Determination of the three-dimensional solution structure of noxiustoxin: analysis of structural differences with related short-chain scorpion toxins.

The 3D structure of noxiustoxin, the first identified scorpion toxin acting on K+ channels, has been elucidated by NMR and molecular modeling. Thirty-nine solution structures were calculated using 572 distance and 42 dihedral restraints. The average atomic rms deviation between the refined structures and the mean structure is 0.75 A for the backbone atoms. Noxiustoxin adopts a alpha/beta scaffold constituted of a three-stranded beta-sheet (residues 2-3, 25-30, 33-38) linked to a helix (residues 10-20) through two disulfide bridges. A comparison between the 3D structure of noxiustoxin and those of other structurally and functionally related scorpion toxins (charybdotoxin, PO5-NH2, kaliotoxin) revealed a bending capacity of the helix and a variability in the relative orientations between the helix and the beta-sheet. These two features highlight the plasticity of the alpha/beta scaffold and offer a structural explanation for the capacity of the fold to accommodate an additional alanine residue in the Gly-x-Cys pattern of a previously proposed consensus sequence [Bontems et al. (1991) Science 254, 1521-1523]. Our structural data also emphasize the possibility that the beta-sheet of NTX is implicated in the capacity of NTX to recognize voltage-dependent K+ channels.

Scorpion toxins as natural scaffolds for protein engineering.

A compact, well-organized, and natural motif, stabilized by three disulfide bonds, is proposed as a basic scaffold for protein engineering. This motif contains 37 amino acids only and is formed by a short helix on one face and an antiparallel triple-stranded beta-sheet on the opposite face. It has been adopted by scorpions as a unique scaffold to express a wide variety of powerful toxic ligands with tuned specificity for different ion channels. We further tested the potential of this fold by engineering a metal binding site on it, taking the carbonic anhydrase site as a model. By chemical synthesis we introduced nine residues, including three histidines, as compared to the original amino acid sequence of the natural charybdotoxin and found that the new protein maintains the original fold, as revealed by CD and 1H NMR analysis. Cu2+ ions are bound with Kd = 4.2 x 10(-8) M and other metals are bound with affinities in an order mirroring that observed in carbonic anhydrase. The alpha/beta scorpion motif, small in size, easily amenable to chemical synthesis, highly stable, and tolerant for sequence mutations represents, therefore, an appropriate scaffold onto which polypeptide sequences may be introduced in a predetermined conformation, providing an additional means for design and engineering of small proteins.

An NMR study of the interaction of cardiotoxin gamma from Naja nigricollis with perdeuterated dodecylphosphocholine micelles.

We investigated the interaction of toxin gamma, a cardiotoxin from the venom of the elapid Naja nigricollis, with perdeuterated dodecylphosphocholine (DodPCho) micelles using standard two-dimensional proton NMR spectroscopy. The proton spectrum resonances of the micelle-bound toxin gamma were assigned, and the chemical shifts of the backbone and side-chain protons were compared with those determined in the absence of DodPCho. We observed that DodPCho induced large chemical shift changes on residues localized on the hydrophobic face of the toxin. These changes are not associated with conformational changes of the toxin. However, the micellar environment may induce some stabilization of the triple-stranded beta sheet, the major component of the protein structural core. Since the proton NMR spectrum of toxin alpha, a structurally related neurotoxin extracted from the same venom, was unaffected by the presence of the micelles, we came to the conclusion that the observed effects are specific to cardiotoxins. The present results give direct evidence of the contribution of the hydrophobic face of the toxin to the toxic site and further suggest a possible mechanism of action of cardiotoxin on biological bilayers.

Solution structure of a green mamba toxin that activates muscarinic acetylcholine receptors, as studied by nuclear magnetic resonance and molecular modeling.

The three-dimensional solution structure of the MTX2 toxin (65 amino acids and 4 disulfides) from the green mamba venom (Dendroaspis angusticeps), a toxin that activates the pharmacological M1 muscarinic acetylcholine receptors, has been determined by nuclear magnetic resonance and molecular modeling. Seventeen structures were calculated from 810 distance and 68 dihedral angle restraints using DIANA and X-PLOR. The average rms deviation between the 17 refined structures and the energy-minimized average structure is 0.95 A for the backbone atoms. The overall folding of MTX2 consists of three loops stabilized by the four disulfides and forming a two- and a three-stranded beta-sheet. This structure appears to be very similar to that of other snake toxins, such as neurotoxins, fasciculins, and cardiotoxins, that also possess the same three-finger fold. For instance, the RMSd for the backbone atoms between MTX2 and the curaremimetic toxin alpha (from Naja nigricollis), the acetylcholinesterase inhibitor fasciculin 1 (from Dendroaspis angusticeps), and the cardiotoxic toxin gamma (from Naja nigricollis) are 1.86, 1.87, and 2.04 A, respectively. Local differences are observed between this toxin and the other structurally related toxins. Some of these differences could be relevant for the functional specificity of MTX2. In particular, this toxin presents a large twist at the tip of loop II due to a bulge (V31, T32; N35) that accommodates an inserted amino acid in the loop. This spatial arrangement brings the side chain of K34 in the beta-turn of the loop to be aligned with the beta-sheet. Hypotheses about a possible functional role of this lysine are described. Other characteristics in the side-chain distribution that could be related to the MTX2 function are presented.