Nicotinic acetylcholine receptors (nAChRs) play a key role in the central and peripheral nervous system and are involved in neuronal growth and plasticity, development, learning, memory and pain sensation 12345. α-Conotoxins, found in the venoms of marine molluscs belonging to the genus Conus, act as antagonists at nAChRs and consequently have a range of potential therapeutic applications 678.

Members of the α-conotoxin family inhibit either muscle, neuronal or both types of nAChRs and prevent channel opening 6910. They comprise 12 to 19 amino acids, including four highly conserved cysteine residues that form two disulfide bonds. Native α-conotoxins have a disulfide connectivity linking the first and the third, and the second and the fourth cysteine residues (Cys^I–Cys^III and Cys^II–Cys^IV), which is referred to as the "globular" isomer. The alternative connectivities (Cys^I–Cys^IV and Cys^II–Cys^III) and (Cys^I–Cys^II and Cys^III–Cys^IV) are referred to as the ribbon and beads forms, respectively and are shown in Fig. 1. The beads form has so far not been reported for naturally occurring conotoxins, but the ribbon connectivity is characteristic of the χ-conotoxins 61112. The segments between cysteine residues define two loops in the peptide backbone, also shown in Fig. 1.

The spacing between the cysteine residues in α-conotoxins is variable, leading to their classification into sub-families 6. Table 1 shows the sequences of selected α-conotoxins with different cysteine spacings. The original α-conotoxins discovered in the 1980's, such as the paralytic toxin GI, have a 3/5 spacing, with three amino acids in the first and five in the second loop 13. They compete mainly for nAChRs of the muscle endplate. Members of the "non-classical" α-conotoxins display 4/X spacings (where X is 3, 4, 6, or 7) and target mainly neuronal nAChRs, with the exception of EI, which is muscle specific 1415. In addition to the variations in cysteine spacing there is a significant degree of sequence variation and it is this diversity that results in the exquisite selectivity of the α-conotoxins. The subtype selectivity of selected α-conotoxins is illustrated in Table 1.

BuIA is the only α-conotoxin with the unique 4/4 spacing (see Table 1 and Fig. 1) and is the first conotoxin identified in the piscivorous Conus bullatus species. Azam et al. 16 used conserved features of the α-conotoxin signal sequence and 3'-untranslated sequence region of α-conotoxin prepropeptide genes to clone the gene coding for BuIA. A synthetic version of the corresponding 13-amino acid mature peptide with the globular disulfide connectivity (Cys²–Cys⁸, Cys³–Cys¹³) is active against α3- and α6-containing nAChR subtypes, whereas α2- and α4-containing nAChRs are blocked less effectively. Strikingly, BuIA kinetically discriminates between β2- and β4-containing receptors, as the off-rates are rapid for β2-subunit, but very slow for β4-containing nAChRs. This feature appears to extend across different mammalian species, as similar results for mouse, rat and human nAChRs were obtained 16. Thus the novel α-conotoxin BuIA appears to be a valuable probe to distinguish among nAChRs containing different α and β subunits.

Understanding the structure-activity relationships of α-conotoxins has been a major focus of conotoxin research in recent years and, as a result the structures of many of the known α-conotoxins have been determined. Conotoxins are particularly suitable for structural analysis with NMR spectroscopy because of their small size, solubility and generally well-defined conformations, and the majority of α-conotoxin structures have been determined using this technique 1718. Despite the range of sequences and spacings between the cysteine residues a well-defined consensus structural motif has been found for the α-conotoxins, with the major features being the restraints imposed by the conserved globular disulfide connectivity and a helical region centred around Cys^II 18. In the current study we have examined the structures and activities of the globular isomer and the non-native ribbon isomer of BuIA with a Cys²–Cys¹³, Cys³–Cys⁸ connectivity and show that the globular isomer is structurally heterogeneous while the non-native ribbon isomer has a well-defined conformation.

Orthogonal syntheses of globular and ribbon BuIA were achieved using solid phase peptide synthesis via Boc chemistry. Purification via HPLC yielded a single product for each isomer whose mass was determined using electrospray ionization mass spectrometry. Both isomers have an observed mass of 1311.6 Da (calculated mass 1311.5 Da). HPLC traces for the purified products are shown in Fig. 1B, from which it is clear that the ribbon form elutes earlier than the globular form, indicating that its hydrophobic residues are buried to a greater extent. Interestingly, under the HPLC conditions used the ribbon form has a slightly broader peak than the globular form despite having a more defined structure as described below. When prepared without selective protection of the cysteine residues BuIA folds predominately into the globular isomer in a 4:2:1 ratio (globular:ribbon:beads), consistent with this being the native form in venom, as is the case with all other α-conotoxins. For the oxidation profile of BuIA without selective protection see Additional file 1.

The globular and ribbon isomers purified from oxidation without selective protection were analyzed by NMR spectroscopy and found to be identical to their respective isomers isolated following selective formation of the disulfide bonds. However, the TOCSY and NOESY spectra revealed considerable differences between the two isomers, as illustrated in the upper panels of Fig. 2. The non-native ribbon form with the Cys²–Cys¹³, Cys³–Cys⁸ disulfide connectivity displayed spectra consistent with a well-defined structure with a single conformation. Specifically, it yielded spectra with good dispersion of chemical shifts in the amide region ranging from 7.7 to 9.1 ppm and exactly the expected number of TOCSY spin systems. In contrast, the spectra of globular BuIA displayed more than the expected number of TOCSY cross-peaks, indicating the likely presence of conformational heterogeneity. The possibility that the extra cross-peaks were due to impurities was eliminated because only a single sharp HPLC peak was observed and only a single mass was detected by mass spectrometry.

The sequential assignment of the spectra of the ribbon and globular forms was undertaken using a combination of TOCSY and NOESY spectra 19. For the ribbon isomer the assignment was straightforward. By contrast, the presence of multiple sets of peaks due to three conformers (referred to here as conformers A, B and C), a lack of detection of sufficient sequential NOESY cross-peaks and peak overlap prevented complete sequential assignment of all of the globular BuIA conformers. Nonetheless, with the aid of the 900 MHz spectra conformers A and B were assigned, as indicated in Fig 2A. The first six residues of conformer C were assigned but the proline residue at position 6 has weak spectral peaks and has an αH shift overlapped with several other residues and thus assignment of Pro7 and sequentially following residues was not possible. A range of pH values (3, 4 and 5.5), temperatures 280–310 K and acetonitrile co-solvent concentrations (20% or 40% v/v) were used to determine if solution conditions influenced the presence of multiple conformations but similar ratios of isomers were present in all the conditions used.

Conformers A and C both display sharp peaks, in contrast to conformer B, which has significant broadening. Based on peak integrals the ratio of the isomers is approximately 3:2:1 (A:B:C). Because peaks from conformers A and C in the same solution are sharp, the broadened peaks of conformer B presumably do not reflect aggregation, but are likely to be the result of exchange broadening. Thus, not only does the globular isomer lead to at least three conformers in slow exchange on the NMR time scale, but one set of peaks probably reflects the presence of two or more conformers in intermediate to fast exchange. Overall this isomer of BuIA appears to be much more conformationally heterogeneous than other conotoxins studied to date.

Inspection of the sequential cross-peaks associated with the proline residues indicated that the conformational heterogeneity observed for globular BuIA originates from cis-trans isomerisation of the two proline residues. Conformer A has both proline residues in a trans conformation, consistent with the report of Chi et al. 20 whereas conformer B has a trans Pro⁶ and cis Pro⁷ based on strong NOEs observed between the α- and δ-protons of Thr⁵ and Pro⁶ and a strong αH-αH NOE between Pro⁶ and Pro⁷ respectively. The lack of a complete assignment for conformer C and peak overlap prevented determination of the proline configurations for this conformation. In the ribbon form of BuIA, which adopts an ordered structure, the same proline conformation pattern as conformer B was observed, namely trans Pro⁶ and cis Pro⁷.

Analysis of the secondary αH chemical shifts of ribbon BuIA indicates a lack of regular secondary structure. There is no continuous stretch of residues exhibiting negative deviations that would be indicative of helices or positive deviations indicative of extended structures, such as β-sheets. Panel D of Fig. 2 shows a comparison of the αH secondary shifts of ribbon BuIA with the ribbon isomer of AuIB 21. The trends are similar, but in general the secondary shifts of ribbon BuIA deviate from random coil to a greater extent than that observed for ribbon AuIB consistent with a well-defined structure for ribbon BuIA.

Despite the lack of complete assignment of all the conformations of globular BuIA an analysis of the chemical shifts was very informative. The secondary shifts of conformer A are very similar to the related α-conotoxin AuIB 21 as shown in Fig. 2C, indicating that the native conotoxin fold is present 18. In contrast, the secondary shifts of conformer B differ significantly from those of AuIB (Fig. 2D). While our work was in progress the structure of the native conformation (i.e. conformer A) of BuIA was reported by Chi et al and confirms that the consensus structure of the α-conotoxins is maintained 20. It was not possible to directly compare our findings with those of Chi et al because chemical shifts or spectra were not provided in the study but consistent with our study, Chi et al reported the presence of additional peaks that they presumed arose from a cis conformation in the globular isomer. In the present study, we have shown there is more than one additional conformation present. The ribbon isomer, which is the focus of our study, was not examined in the Chi et al study 20.

Given the single conformation observed for ribbon BuIA it was of interest to determine the three dimensional structure of this isomer. Structure calculations were carried out with 96 NOE distance restraints (comprising 12 long-range, 18 medium-range, 50 sequential and 16 intraresidual distance restraints) and six dihedral angle restraints, using a simulated annealing protocol in CNS. The ϕ-angle restraints, derived from αN coupling constants, were -65 ± 15° for Ala⁹ and Leu¹¹, -120 ± 15° for Thr⁵, and -120° ± 30° for Cys⁸. The ϕ-angle for Cys¹³ was restrained to 100 ± 80° based on the intraresidual Hα-HN NOE being clearly weaker than the NOE between HN and the Hα of the preceding residue. The χ₁-angle of Cys⁸ was restrained to 180 ± 30° based on the αβ coupling constants and NOE intensities. Hydrogen bonds were not included in the calculations but measurement of the exchange rates for the amide protons of ribbon BuIA revealed that 3 of the 11 amides, namely Val¹⁰, Leu¹¹ and Tyr¹², are slowly exchanging. Their amide proton signals were still detected in NMR spectra recorded 30 minutes after dissolution of the sample in ²H₂O, but all amide protons had exchanged after 45 minutes in ²H₂O. This result suggests that Val¹⁰, Leu¹¹ and Tyr¹² are more protected from the solvent than are the other residues.

Fig. 3 shows a stereoview of the superposition of the 20 lowest energy structures of ribbon BuIA. Despite its small size it adopts a well-defined three-dimensional structure. The ensemble of structures aligns well over the whole molecule with a pairwise backbone RMSD of 0.36 ± 0.12 Å and a RMSD of 1.1 ± 0.49 Å for all heavy atoms, and had good structural and energetic statistics, as shown in Table 2. No NOE violations higher than 0.3 Å and no dihedral angle restraint violations greater than 3° were observed within these 20 structures. In the Ramachandran plot, showing the favored ϕ- and ψ-angle combinations for residues in proteins, all residues of ribbon BuIA (except Pro and Gly) lie in the most favored and additionally allowed regions.

Using the program PROMOTIF 22 the 20 lowest energy structures of ribbon BuIA were examined to identify consensus structural elements. In agreement with the prediction from the secondary shifts, ribbon BuIA shows no α-helices or β-sheets. The structure comprises an inverse γ-turn for residues 2–4 but no regular secondary structure is present in the remainder of the molecule. One of the two disulfide bonds stabilizing the three dimensional structure, namely that formed by Cys³ and Cys⁸, overlays very well in the 20 lowest energy structures. In 17 of 20 structures the torsion angles of this disulfide bond are representative of a right-handed hook 23. By contrast, the disulfide bond joining Cys² to Cys¹³ is classified as right-handed hook in 9 out of 20 structures but not recognized as a standard disulfide type in the other structures. Calculation of the hydrogen bonds using MolMol 24 revealed that in 13 of 20 structures the amide proton of Leu¹¹ is bonded to the carbonyl of Ala⁹. This hydrogen bond is consistent with the relatively slow exchange of the amide proton of Leu¹¹, but Val¹⁰ and Tyr¹² were also slowly exchanging and no hydrogen bond acceptors were apparent from the calculated structures.

Using the program InsightII the solvent accessible surface of ribbon BuIA was calculated and is shown in Fig. 3, from which the large hydrophobic surface formed by Ala, Val, Leu, Tyr and Pro residues is clear. The hydrophobic patch extends across a wide stretch of the whole molecule, as it comprises only a few hydrophilic residues, namely Ser and Thr. Strikingly, there are no charged residues in the molecule. Tyr¹² protrudes conspicuously, making the overall shape of ribbon BuIA non-spherical. In contrast to many other α-conotoxins the cysteine residues are rather exposed to the solvent, presumably as a result of the ribbon disulfide connectivity and the fact that BuIA is only 13 residues in length.

The biological activities of the BuIA globular and ribbon analogues, as well as a form where the cysteine residues were alkylated, were examined in an electrophysiological assay by measuring the % inhibition of ACh-evoked current amplitude at the α3β2 and α3β4 nAChR subtypes expressed in Xenopus oocytes (Fig. 4A). The alkylated form of BuIA was analyzed by mass spectrometry and had an observed molecular mass (1571.9 Da) consistent with the calculated mass (1571.64 Da). Membrane currents were evoked with 100 μM ACh and concentration-response curves were fitted to determine IC₅₀ values and Hill coefficients. The concentration-response curves for inhibition of α3β2 and α3β4 nAChR subtypes by globular BuIA gave IC₅₀ values of 4.8 ± 0.4 nM (n_H = 1.3) and 59.1 ± 2.3 nM (n_H = 1.2), respectively, whereas almost no inhibition was observed with 1 μM ribbon BuIA at either nAChR subtype (n = 4–7; Fig. 4B). Furthermore, the globular BuIA isomer is approximately ten-fold more potent at β2- compared to β4-subunit containing nAChR subtypes. These results for synthetic globular BuIA are consistent with that reported previously 16. Alkylated BuIA was inactive at α3β2 and α3β4 subtypes when tested at 10 nM and 100 nM respectively. Increasing the concentration of alkylated BuIA to 10 μM also failed to inhibit nAChRs.

As BuIA is the only α-conotoxin reported so far with a 4/4 spacing its three dimensional structure is of significant interest. In the current study we examined the structural features and determined the biological activity of the globular (Cys^I–Cys^III, Cys^II–Cys^IV) and ribbon (Cys^I–Cys^IV, Cys^II–Cys^III) disulfide isomers of BuIA. These studies show that the general trend observed for α-conotoxins where the native globular form has a well-defined conformation and the ribbon isomer is disordered is reversed for BuIA.

NMR analysis of the globular isomer of BuIA revealed more cross-peaks than expected for a 13 residue peptide. At least three conformations are present in the two-dimensional spectra of globular BuIA and as native α-conotoxin structures determined to date generally display well-defined structures, the significant heterogeneity differs from usual findings for α-conotoxins. Surprisingly, this conformational heterogeneity is not seen in the ribbon isomer, which forms a single well-defined conformation. The observation that the HPLC trace of the ribbon reveals a broad peak, but the structure is well-defined is interesting and unusual. One possibly is that the different conditions under which the HPLC and NMR are carried out are responsible for these differences. The more hydrophobic environment used for the HPLC analysis coupled with the presence of TFA could potentially affect how the ribbon isomer behaves on RP-HPLC. However, recording NMR spectra of globular BuIA in the presence of varying amounts of acetonitrile co-solvent did not influence the conformational heterogeneity, making the sharper peak observed by HPLC analysis for globular BuIA compared to ribbon BuIA a puzzling result.

The conformational heterogeneity of globular BuIA originates form cis/trans isomerisation of the pair of adjacent proline residues in the middle of the sequence. The proline configurations for the globular isomers and ribbon BuIA are highlighted in Fig. 5. Conformer A has both proline residues in a trans conformation, consistent with the trend observed for other native α-conotoxins, whereas conformer B is broadened and displays a trans/cis configuration for Pro⁶ and Pro⁷ respectively. The conformations of the proline residues in conformer C were not determined because of the lack of a complete assignment. Ribbon BuIA displays the same proline geometry as conformer B despite the difference in disulfide connectivity. The BuIA sequence can clearly accommodate a range of geometries of the proline residues in contrast to what has generally been observed in previous structural analyses of α-conotoxins 1718. One exception is α-conotoxin GI that has previously been shown to have two interconverting conformations 25 but this differs from BuIA where there are clearly multiple conformations. Other peptides are known to contain similar proline motifs to BuIA, including SFTI-126, and in that case the Pro-Pro motif has a cis/trans configuration as shown in Fig. 5. This highlights the complexity that can be associated with the geometry of proline residues in small cysteine-rich peptides and how changes in structure and sequence can result in different geometries being favored.

The structurally heterogeneous globular isomer of BuIA is active on α3-containing neuronal nAChRs, with a preference for receptors comprising β2- rather than β4-subunits. This specificity is quite similar to some of the 4/7 α-conotoxins such as MII 20. The structure of conformer A of globular BuIA, determined by Chi et al 20, has been compared to that of MII and it appears that the two-turn helical motif present in both MII and BuIA is important for binding to the α3-subunit of neuronal nAChRs 20. Recent studies on the binding of α-conotoxins to the acetylcholine binding protein, a homolog of the nAChR extracellular domain, have shown that the rigid consensus structure found in solution does not change significantly upon binding 272829. Thus, as ribbon BuIA does not contain the consensus structural motif it consequently exhibits no inhibitory activity at nAChRs up to 1 μM. It appears that the well-defined conformation of ribbon BuIA prevents it from adopting an appropriate conformation for binding to the nAChR.

We have also shown here that an analogue of BuIA with the cysteine residues reduced and alkylated is inactive at the nAChR. This is consistent with studies on ImI where the reduced and alkylated form had a significant decrease in activity at the α7 receptor 30. Given that the reduced and alkylated form of BuIA is likely to be devoid of structure based on the strong influence of disulfide bonds on the structures of small cysteine-rich peptides, the lack of activity of this peptide indicates that it cannot adopt an appropriate conformation for binding to the nAChR. Thus, the linear sequence alone is not sufficient for activity and a propensity to form a helical structure in solution appears to be correlated with activity at the nAChR.

The ribbon isomer of BuIA, despite having a well-defined structure, is not the preferred isomer formed during air oxidation using standard conditions applicable to other conotoxins. The fact that the folding pathway favors the globular disulfide connectivity indicates that it is energetically more favorable, even though multiple conformations are present in solution. Although the ribbon isomer is less favoured during oxidation, as is often the case for non-native isomers, it displays significantly different properties to both non-native and native conotoxin ribbon isomers. The ribbon isomers of both α-AuIB and α-GI exhibit greater conformational flexibility than their globular counterparts 2131, in contrast to ribbon BuIA, indicating that the unique sequence of BuIA stabilizes the ribbon isomer. Structures of globular and ribbon AuIB are shown in Fig. 6. Although a single structure is shown for the ribbon isomer of AuIB it displays considerable conformational heterogeneity relative to the globular isomer. Furthermore, the ribbon isomer of BuIA lacks any regular secondary structure, in contrast to the χ-conotoxins that naturally contain a ribbon connectivity. The χ-conotoxin MrIA consists of a β-hairpin connected by an inverse γ-turn 32, as shown in Fig. 6.

Another similarity between the ribbon isomers of α-conotoxins is related to the geometry of the proline residues 618. The inactive ribbon isomers of BuIA and GI 31 both contain a cis proline (Pro⁷ and Pro⁵ respectively) in contrast to the trans Pro seen in the native disulfide connectivity for all α-conotoxin structures determined to date 18. In one conformation of globular BuIA (conformer B, Fig. 2) Pro⁷ is also in a cis geometry, but the alternative trans conformation may be stabilized upon binding to the receptor. The implications for the presence of the cis proline in activity have not yet been elucidated.

The native globular isomer of BuIA, although being the favored isomer during oxidative refolding, has multiple conformations in solution, unlike the majority of native α-conotoxins. In addition, the structure of the ribbon isomer is stabilized whereas all other non-native isomers display disordered structures. Overall our results highlight the influence of the disulfide connectivity of BuIA on the dynamics of the three-dimensional structure. This information is potentially important for on-going efforts to understand the structure-activity relationships of this valuable class of peptides.

HPLC, high performance liquid chromatography; NMR, nuclear magnetic resonance; ACh, acetylcholine; nAChRs, nicotinic acetylcholine receptors.

A-HJ designed experiments and carried out synthetic studies. HB carried out structural studies and helped draft the manuscript. STN designed and carried out biological assays. CCT carried out synthetic and folding studies. RJC designed and carried out synthetic and folding studies. DJA designed biological experiments and revised the manuscript. PFA conceived study and designed synthetic experiments. DJC designed structural studies and revised the manuscript. NLD designed and carried out structural studies and drafted and revised the manuscript. All authors have read and approved the final manuscript.