α–RgIA, a Novel Conotoxin that Blocks the α9α10 nAChR: Structure and Identification of Key Receptor Binding Residues

Abstract

α-Conotoxins are small disulfide-constrained peptides from cone snails which act as antagonists at specific subtypes of nicotinic acetylcholine receptors (nAChRs). The 13-residue peptide α-RgIA is a member of the α-4,3 family of α-conotoxins and selectively blocks the α9α10 nAChR subtype, in contrast to another well characterized member of this family, α-ImI, which is a potent inhibitor of the α7 and α3β2 nAChR subtypes. In this study, we have altered side chains in both the 4-residue and 3-residue loops of α-RgIA, and have modified its C-terminus. The effects of these changes on activity against α9α10 and α7 nAChRs were measured, the solution structures of α-RgIA and its Y10W, D5E and P6V analogues were determined from NMR data, and resonance assignments made for α-RgIA[R9A]. The structures for α-RgIA and its three analogues were well-defined except at the chain termini. Comparison of these structures with reported structures of α-ImI reveals a common two-loop backbone architecture within the α-4,3 family, but with variations in side chain solvent accessibility and orientation. Asp5, Pro6 and Arg7 in loop 1 are critical for blockade of both the α9α10 and α7 subtypes. In loop 2, α-RgIA[Y10W] had activity near that of wild-type α-RgIA, with high potency for α9α10 and low potency for α7, and had a similar structure to wild-type. By contrast, Arg9, in loop 2, is critical for specific binding to the α9α10 subtype, probably because it is larger and more solvent accessible than Ala9 in α-ImI. Our findings contribute to a better understanding of the molecular basis for antagonism of the α9α10 nAChR subtype, which is a target for the development of analgesics for treatment of chronic neuropathic pain.

Introduction

Nicotinic acetylcholine receptors (nAChRs) are acetylcholine (ACh) gated ion channels.¹ They are composed of pentamers of transmembrane proteins organized so as to form a central ion-conducting pore. Many different nAChR subunits exist, with δ, γ, ε, α1-α10 and β1-β4 subunits having been described in vertebrates. Different arrangements of these subunits can be incorporated into functional receptor pentamers, giving rise to diverse nAChR subtypes. For example, at the neuromuscular junction, receptors composed of (α1)₂,β1, δ and γ mediate synaptic transmission at the motor end plate. Heteromeric receptors composed of only α and β subunits also occur, e.g. the α3β2 nAChR is formed from α3 and β2 subunits. Certain nAChR subtypes contain only α subunits, of which there are two types; the α7 and α9 subunits form homopentamers, whereas the α9 and α10 subunits combine in the heteropentameric α9α10 subtype.

Acetylcholine, competitive antagonists and other competitive agonists bind nicotinic receptors at a site in the interface between the N-terminal extracellular domains of adjacent subunits. At these interfaces one side of the inter-subunit cleft is the principal (+) face which is characterized by a vicinal disulfide bond in a backbone loop, the C-loop, which projects from it and partially buries bound competitive ligands.^2,3 Only α subunits are thought to form the principal side of the subunit interfaces and only they have a vicinal disulfide bridge in their C-loop. The complementary (−) side of the ACh site is opposite the (+) face on the other side of the inter-subunit cleft; in heteromeric receptors it is provided by non-α subunits whereas in α7, α9 and α9α10 receptors it is provided by α subunits.

Subtype-specific inhibitors of nAChRs are potentially valuable neurochemical tools for dissecting the roles of the different nAChRs and may also be important therapeutic agents. One particularly rich source of subtype-specific nAChR antagonists is the α-conotoxin family.^4,5 α-conotoxins are venom peptides from snail species in the genus Conus.^6,7 They have the invariant disulfide scaffold shown in Figure 1 but are also characterized by high sequence diversity in their inter-cysteine loops. α-conotoxins can be further subdivided based on the number of amino acids in their inter-cysteine loops; one subclass, found almost exclusively in Conus species that hunt marine worms (unpublished observation), are the α-4,3 toxins which have four amino acid residues in the first loop and three in the second. Two peptides in the α-4,3 family are α-ImI (from Conus imperialis) and α-RgIA (from Conus regius). α-ImI is a potent inhibitor of the α7 and α3β2 nAChR subtypes,⁸ whereas α-RgIA selectively blocks the α9α10 subtype.⁹ The ability of α-RgIA to block α9α10 receptors appears to account for its potent analgesic properties in animal models of neuropathic pain.¹⁰

Figure 1.

The conserved disulfide scaffold of the α-conotoxin family and the sequences of α-RgIA (top) and α-ImI (bottom) are shown. The three differences between the two peptides are in bold and underlined. # represents C-terminal amidation.

Here we describe the solution structures of α-RgIA and three analogues, Y10W, D5E and P6V, sequence-specific resonance assignments for the R9A analogue, and structure/function studies carried out using analogues of α-RgIA. This has enabled the identification and visualization of residues critical for α9α10 potency and specificity. Comparison with previous structures of α-ImI supports the model that different peptides in the α-4,3 subfamily have very similar two-loop backbone structures and their diverse functional properties appear to result from the presentation of different side chains by fundamentally the same scaffold. Our mutational studies indicated that Asp5, Pro6 and Arg7 in loop 1 are critical for α9α10 potency and that Arg9 in loop 2 is critical for high potency for α9α10 and low potency for α7, i.e. for the subtype specificity of α-RgIA. Based on our findings we speculate on how α-RgIA might interact with the α9α10 nAChR.

Results

NMR spectroscopy

α-RgIA and its four analogues gave good quality spectra at pH 3 – 6 and 298 K. The spectra collected at lower pH or lower temperature showed greater peak overlap. At higher pH, cross-peaks from the amide of Cys2 disappeared in NOESY spectra because of exchange with solvent water, but other peaks did not shift appreciably, reflecting the fact that the Asp5 and C-terminal carboxyl groups would be ionized above pH 4 (the amide and aromatic regions of 1D spectra of wild-type α-RgIA at different temperatures and different pH are shown in Figure S1A and B, respectively, of the Supplementary Material). Peaks for Asp5, Arg9, Tyr10 and Cys12 of wild type α-RgIA were broadened in NOESY spectra, but no other spectral inhomogeneities were observed under the experimental conditions, indicating the presence of a single predominant conformation for each peptide in solution. One-dimensional spectra of the four analogues, Y10W, D5E, P6V and R9A, are shown in Figure S1C.

Spin system assignments in TOCSY spectra and sequential connectivities in the fingerprint region of corresponding NOESY spectra acquired on a Bruker DRX-600 spectrometer at 298 K for all five peptides are shown in Figure S2A to E. Chemical shift assignments are presented in Tables S1A to E, and have been deposited in the BioMagResBank.¹¹

Self diffusion coefficients of α-RgIA and its analogues are (1.71±0.08) ×10⁻¹⁰ m²/s, (2.00±0.05) ×10⁻¹⁰ m²/s and (3.26±0.14) ×10⁻¹⁰ m²/s at 5, 10 and 25 °C, respectively. The values at 5 °C are 10% faster than J1cc, a 19-residue peptide with a less well-defined structure,¹² and the values at 10 °C are comparable with extrapolated values for similar size peptides with well-defined structures,¹³ indicating that α-RgIA and its analogues are monomeric in solution.

The trans orientation of the peptide bond preceding Pro6 was established by the intense H^α-H^δ NOEs between residues 5 and 6 for wild-type α-RgIA and its Y10W, D5E and R9A analogues. For wild-type α-RgIA, the backbone amide temperature coefficient for Tyr10 was smaller than 3 ppk/K in magnitude, and those for Cys2, Cys3, Arg7 and Cys12 were around 5 ppk/K, indicating that these amides were partially protected from solvent. However, no slowly exchanging amide protons were observed in 1D spectra recorded after dissolution of α-RgIA in 100% ²H₂O at pH 5 and 5 °C.

Deviations of the backbone NH and H^α chemical shifts of α-RgIA and the four analogues from random coil values¹⁴ are shown in Figures S3A to E in the Supplementary Material. The significant deviations across the sequence indicate that α-RgIA is structured (Figure S3A), as expected from the presence of two disulfide bridges and the structures of other α-conotoxins. The similarities in the patterns of these deviations (Figures S3B to E) with α-ImI (Figures S3F and G) confirm that all three analogues adopt similar backbone structures.

The changes in NH shifts for the R9A analogue relative to wild-type (Figure 2A) were mostly < 0.1ppm. Larger shifts were observed for Ser4 and Arg 7 of α-RgIA[D5E], Val6 of α-RgIA[P6V] and the C-terminal residues of α-RgIA[Y10W]. Differences in H^α shifts for the four analogues relative to wild-type (Figure 2B) were mostly < 0.1 ppm, except for Cys2 and Arg11 in α-RgIA[Y10W], Glu5 and Cys8 in α-RgIA[D5E], and Asp5, Val6 and Cys8 in α-RgIA[P6V].

Figure 2.

Deviations of the chemical shifts of (A) NH and (B) H^α for α-RgIA[D5E], α-RgIA[P6V], RgIA[R9A] and α-RgIA[Y10W] from those of wild type α-RgIA.

Solution structures

A summary of experimental constraints and structural statistics for α-RgIA and the three analogues, Y10W, D5E and P6V, is given in Table 1. Values of the angular order parameters S > 0.8 generally indicate that an angle is well defined within the family of structures; for wild type α-RgIA, the angular order parameters for φ and ψ angles in the final ensemble of 20 structures were both > 0.8 over residues 2-12 (Figure S4A). The mean pairwise RMSD over the backbone heavy atoms of residues 2-12 in this family of structures was 0.43 Å. The closest-to-average structure of α-RgIA (Figure 3) is characterized by two compact loops of four and three residues, defined by the two disulfide bonds. The first loop is better defined than the second. Residues 2-5 approximate a type-I β–turn formed, which is supported by the observation of hydrogen bond between the Asp5 NH and Cys2 O in 13 of the 20 structures, as well as the backbone φ and ψ angles of Ser4 and Asp5. Side chains for residues in the first loop are mostly well defined except for Ser4 (Figures 4 and S5). Although the second loop is less well defined, Tyr10-Cys12 form an inverse γ-turn, with most (18) of the final structures showing a Cys12 NH to Tyr10 O hydrogen bond.

Table 1.

Summary of experimental constraints and structural statistics for α-RgIA and three analogues.

	α-RgIA(wt)	α-RgIA[Y10W]	α-RgIA[D5E]	α-RgIA[P6V]
No. of distance restraints	161	160	134	141
intraresidue (i =j)	82	90	80	82
sequential (|I−j| = 1)	46	44	42	46
medium-range (1 < |i−j| < 5)	25	17	5	8
long-range (|i−j| > 4)	8	9	7	5
hydrogen bond restraints	0	0	0	0
No. of dihedral restraints	11	10	7	7
Energiesa
ENOE(kcal mol−1)	1.7±1.1	1.2±0.5	1.6±0.8	1.6±0.8
RMS deviations from experimental data
NOEs (Å)	0.014±0.004	0.012±0.003	0.015±0.004	0.015±0.004
Dihedrals (deg)	0.50±0.22	0.23±0.26	0.25±0.23	0.14±0.19
Deviations from idealb
angles (deg)	0.58±0.02	0.55±0.02	0.57±0.03	0.50±0.02
bonds (Å)	0.0023±0.0002	0.0018±0.0003	0.0020±0.0003	0.0017±0.0002
impropers (deg)	0.37±0.02	0.35±0.01	0.37±0.02	0.35±0.02
RMS deviations (Å)c
Backbone (bb) atoms (global)	0.71	1.24	1.48	1.49
bb atoms [S(φ) and S(ϕ) > 0.8]	(2–12) 0.43	(2–11) 0.64	(2–10) 0.77	(3–10) 0.67
heavy atoms [S(φ) and S(ϕ) > 0.8]	(2–11) 1.38	(2–11) 1.66	(2–10) 2.10	(3–10) 1.90
Ramachandran plotd
Most favoured (%)	69.5	70.5	65.0	62.7
Allowed (%)	30.5	29.5	34.0	36.5
Additionally Allowed (%)	0	0	1.0	0.9
Disallowed (%)	0	0	0	0

^aThe values for E_NOE are calculated from a square well potential with force constants of 50 kcal mol⁻¹ Ų.

^bThe values for the bonds, angles, and impropers show the deviations from ideal values based on perfect stereochemistry.

^cThe average RMSD over the backbone heavy atoms (N, C^α, C′) over the indicated residues calculated in MOLMOL.

^dAs determined by the program PROCHECK-NMR for all residues.

Figure 3.

Translucent surface view of the closest-to-average structure of α-RgIA. The side chains of Asp, Pro and Arg in the first loop point to the same direction and those of Arg and Tyr in the second loop point in a direction that is roughly perpendicular. This figure was prepared using PyMol (Delano, W.L. The PyMOL Molecular Graphics System (2002) Delano Scientific, San Carlos, CA, USA. http://www.pymol.org).

Figure 4.

Ensembles of 20 structures of (A) α-RgIA, (B) α-RgIA[Y10W], (C) α-RgIA[D5E], and (D) α-RgIA[P6V]. The backbones are shown in black and heavy atoms of the Asp-Pro-Arg motif in the first loop and residues 9-10 in the second loop, are shown. The 2–8 and 3–12 disulfide bonds are shown in orange.

In order to ensure that changes in activity associated with side chain substitutions (see below) were not associated with changes in structure, we have also determined the structure of α-RgIA[Y10W] and sequence-specific resonance assignments for α-RgIA[R9A], both of which have substitutions in loop 2, as well as low resolution structures for α-RgIA[D5E] and α-RgIA[P6V], with substitutions in loop 1. The backbone angular order parameters (S) in the final ensembles of structures for the three analogs are plotted as a function of residue number in Figure S4B-D. For α-RgIA[Y10W], φ and ψ angles were well defined over residues 2-11, with a pairwise RMSD of 0.64 Å over the backbone heavy atoms (N, C^α, C′), while for α-RgIA[D5E] and α-RgIA[P6V], φ and ψ angles were well defined over residues 2-10 and 3-10 and corresponding RMSD values were 0.77 and 0.67 Å, respectively. These structures are not as well-defined as those for α-RgIA because the samples were more dilute and some of the spectra were acquired on lower field spectrometers. Nonetheless, they are all characterized by two compact loops of four and three residues, respectively, and the similar secondary shifts of α-RgIA[R9A] to wild-type α-RgIA indicate these two peptides have similar backbone structures. Cys12 NH to Arg9 O hydrogen bonds were present in 8 out of 20 of the α-RgIA[Y10W] structures, as in α-ImI (see below), and support a sharp bend about Trp10. Hydrogen bonds were also observed from HN of Glu5 or Asp5 to O of Cys2 in 11 of the 20 structures for α-RgIA[D5E] and α-RgIA[P6V], and 7 of 20 for α-RgIA[Y10W]. The second loop was less well defined in all of these peptides than the first loop, as indicated by the reduction in pairwise RMSD values from 0.43, 0.64, 0.77 and 0.67 Å when superimposed over all well-defined residues (Table 1) to 0.20, 0.34, 0.53 and 0.52 Å when superimposed over Cys2 to Cys8, for wild-type α-RgIA, α-RgIA[Y10W], α-RgIA[D5E] and α-RgIA[P6V], respectively.

The other 3,4 α-conotoxin for which three-dimensional structures have been determined is α-ImI.^2,3,15–19 Considering wild-type α-RgIA and the three analogues for which solution structures were determined in this study, α-RgIA[Y10W] is the most similar chemically to α-ImI, so we have compared its structure to α-ImI. Superposition of the backbone heavy atoms (N, C^α and C′) of the final ensembles of 20 structures with α-ImI structures over residues 1-12 (Figure 5) gave average RMSD values of 0.77, 0.74, 1.00, and 0.68 Å for α-ImI solution structures solved by (A) Rogers et al. (PDB code 1IM1, 20 structures),¹⁸ (B) Maslennikov et al. (PDB code 1IMI, 20 structures),¹⁷ (C) Gehrmann et al. (PDB code 1CNL, 10 structures),¹⁵ and (D) Lamthanh et al. (PDB code 1G2G, 20 structures),¹⁶ respectively. Although the 1IMI, 1CNL and 1G2G structures are for C-terminally amidated peptides (unlike α-RgIA and 1IM1) and are apparently very well-defined structures, significant differences are evident in the backbone conformation around Cys3 and Cys8 and in the orientation of the Arg7 side chain and Trp10 ring, as shown in Figure 5B–D. An average RMSD value of 0.72 Å was obtained to the mean crystal structures of α-ImI in complex with Aplysia californica AChBP solved by (E) Hansen et al. (PDB code 2BPY, non C-terminally amidated)² and (F) Ulens et al (PDB code 2C9T, C-terminally amidated).¹⁹

Figure 5.

Ensembles of 20 solution structures of α-RgIA[Y10W] (in red) superimposed with α-ImI structures solved by (A) Rogers et al. (PDB code 1IM1, in black),¹⁸ (B) Maslennikov et al. (PDB code 1IMI, in blue),¹⁷ (C) Gehrmann et al. (PDB code 1CNL, in green),¹⁵ (D) Lamthanh et al. (PDB code 1G2G, in navy),¹⁶ (E) Five crystal structures in Aplysia AchBP complexes solved by Hansen et al. (PDB code 2BYP, in royal blue)² and (F) Eight crystal structures in Aplysia AchBP complexes solved by Ulens et al (PDB code 2C9T, in black).¹⁹

α-RgIA binds at the ACh binding site on the α9α10 nAChR

Three conserved aromatic residues on the principal side of the subunit interfaces in nAChR pentamers are highly conserved and interact with competitive agonists and antagonists. Initially these residues were identified due to cross-linking with reactive competitive ligands of nAChRs²⁰ and their presence at the ACh site was confirmed in structures of ACh binding proteins from Lymnaea stagnalis^3,21 and A. californica,^2,3,19 as well as the Torpedo nAChR.²² Subsequent co-crystal structures with the ACh binding protein from Aplysia revealed interactions between the conserved aromatic residues and competitive nicotinic ligands,^2,23 and receptor mutagenesis indicated that these residues in the α7 nAChR were crucial for α-ImI binding.²⁴ To determine whether α-RgIA was a competitive inhibitor of the α9α10 nAChR, Trp176 of the α9 subunit (Figure 6) was mutated to threonine and this mutant subunit was co-expressed with wild type α10 subunits in Xenopus oocytes. The mutant α9[W176T]α10 receptor was less potently gated by ACh and less potently inhibited by α-RgIA (Figure 7 and Table S2), showing that the toxin is a competitive inhibitor of the receptor and binds at the ACh binding site, i.e. at the interface between adjacent subunits.

Figure 6.

Amino acid sequence alignments. (A) Regions of nAChR α subunits and Aplysia californica AChBP that form part of the principal face. Four conserved principal face aromatic residues are in bold and underlined; the residues Y93, W147, Y188 and Y195 in Aplysia californica AChBP align with Y115, W171, Y210 and Y217, respectively, in the rat α7 subunit. (B) Region of nAChR α subunits that constitutes part of the complementary subunit interface. Discrete loops of nAChR subunits which converge to form the ACh binding site in nAChRs have been identified and the sequences above correspond to loop E according to one nomenclature scheme. Mutation of the bolded residues affects the affinity of α-ImI for the α7 nAChR²⁶ and they closely approach Ala9 or Trp10 of α-ImI in a model of α-ImI bound to the α7 nAChR⁸. The bolded Asn and Ser interact with Trp10 of α-ImI in double mutant cycles analysis²⁶. The α7 sequence is very distinct from the α9 and α10 sequences, and even more distinct from the Aplysia sequence. The spatial disposition of these residues in the crystal structure of A. californica AChBP in complex with α-ImI is illustrated in Figure S7.

Figure 7.

Mutation of the principal face residue W176 in the α9 subunit reduces the potency of both ACh gating of α9α10 nAChRs and blockade by α-RgIA. (A) Concentration dependence of ACh gating of α9α10 nAChRs (black circles) and α9[W176T]α10 nAChRs (open squares). (B) Concentration dependence of α-RgIA block of α9α10 nAChRs (black circles) and α9[W176T]α10 nAChRs (open squares). The error bars are the standard error of the mean for at least three repetitions. The IC₅₀ and Hill slopes for all inhibition curves are in Table S2.

The DPR motif in loop 1 is crucial for high potency block of α9α10 nAChRs by α-RgIA

It has been shown that the Asp-Pro-Arg (DPR) motif in the first loop of α-ImI is crucial for its high affinity binding to the α7 nAChR.^25,26 We therefore made mutations in the first loop of α-RgIA that had been made previously in α-ImI (S4A, D5E, P6V, R7K) and evaluated the potency of the analogues on oocyte-expressed α7 and α9α10 nAChRs. As shown in Figure 8, the changes D5E, P6V and R7K, but not S4A, caused large reductions in the potency of α-RgIA on both α9α10 and α7 receptors, although the effects on α9α10 (Figure 8A) are larger than on α7 (Figure 8B). Since D5E, P6V and R7K mutations in α-ImI were found previously to reduce its affinity for the α7 receptor our data imply that the DPR motifs in α-RgIA and α-ImI play similar roles in anchoring these toxins to both the α9α10 and α7 nAChRs.

Figure 8.

Effects of loop 1 mutations in α-RgIA. Single amino acid substitutions were made in the first loop of RgIA and the resulting analogues were tested on α7 and α9α10 nAChRs expressed in oocytes. (A) The left panel shows the concentration dependence of block of α9α10 nAChRs by wild type α-RgIA and analogues. Error bars are the standard error of the mean for at least three repetitions. The right panel shows IC₅₀ values of wild-type α-RgIA and analogues against α9α10 nAChRs. The error bars are the 95% confidence interval. (B) As for (A) except that block of α7 nAChRs is shown. The IC₅₀ and Hill slopes for all inhibition curves are in Table S3.

Residues in the second half of α-RgIA are crucial for the α9α10 subtype selectivity of α-RgIA

α-RgIA is a much more potent inhibitor of the α9α10 nAChR than the α7 nAChR⁹ (Table S3 in Supplementary Material). The only sequence differences between α-RgIA and α-ImI are in the C terminal halves of the peptides at positions 9 (Arg in α-RgIA, Ala in α-ImI) and 10 (Tyr in α-RgIA, Trp in α-ImI) and the C terminal (α-RgIA ends at Arg13, and α-ImI Cys12, with C-terminal amidation). The preference of α-RgIA for the α9α10 subtype must therefore be due to these amino acid differences. To explore the role of these residues we made three point mutations in α-RgIA that replaced native amino acids with the corresponding residues in α-ImI i.e. ΔR13,#C12, (Arg13 removed and Cys12 amidated), R9A and Y10W. Testing on α7 and α9α10 receptors expressed in oocytes revealed that the R9A and ΔR13,#C12 analogues had mildly increased potency on the α7 receptor whereas the R9A analogue had a dramatically reduced potency on the α9α10 receptor (Figure 9). ΔR13,#C12 had affinity very similar to wild-type α-RgIA on the α9α10 receptor. Surprisingly, Y10W had affinity very similar to wild-type α-RgIA on both receptor types.

Figure 9.

Effects of loop 2 mutations in α-RgIA. Single amino acid substitutions were made in the second loop of α-RgIA and the resulting analogues were tested on α7 and α9α10 nAChRs expressed in oocytes. (A) The left panel shows the concentration dependence of block of α9α10 nAChRs by wild type α-RgIA and analogues. The error bars are the standard error of the mean for at least three repetitions. The right panel shows IC₅₀ values of wild-type α-RgIA and analogues against α9α10 nAChRs. The error bars are the 95% confidence interval. ΔR13,#C12 refers to an α-RgIA analogue in which Arg13 was removed and C-terminal amidation was added to Cys12. (B) As for (A) except block of α7 nAChRs is shown. The IC₅₀ and Hill slopes for all inhibition curves are in Table S3.

Discussion

The mutant α9[W176T]α10 receptor was found to be less potently gated by ACh and less potently inhibited by α-RgIA, as shown in Figure 7. This indicates that the toxin is a competitive inhibitor of the receptor and binds at the ACh binding site; i.e. at the interface between adjacent subunits. Although the competitive inhibition of the α9α10 receptor by α-RgIA might be expected in view of its similarity to the competitive α7 nAChR antagonist α-ImI, another closely related α-conotoxin, α-ImII was shown to block the α7 receptor at a site other than the ACh binding site.²⁷ Therefore, competitive antagonism is not an invariant feature of even similar α-4,3 toxins. The demonstration of competitive block of α9α10 receptors by α-RgIA is thus an important finding. Position 6 of α-ImI (Pro) and α-ImII (Arg) was found to be critical for determining competitive versus non-competitive block.²⁷ Both α-RgIA and α-ImI have a Pro at this position, indicating that Pro6 in α-4,3 toxins may be a general determinant of competitive nAChR antagonism.

The mutations D5E, P6V and R7K in the first loop of α-RgIA caused large reductions in the potency of α-RgIA on α9α10 nAChRs (Figure 8). The same mutations were previously shown to reduce the affinity of α-ImI for the α7 nAChR,^24,25 suggesting that the Asp-Pro-Arg triads in α-RgIA and α-ImI play the same roles in anchoring the peptides to nAChRs. This is supported by the similar structures of the first loops of α-RgIA and α-ImI (Figure S5).

Several lines of evidence identify the regions of nAChRs that interact with the Asp-Pro-Arg motif of α-ImI. Mutation of the α7 nAChR indicated that three (+) face aromatic residues, Y115, W171 and Y217 (which align with Y120, W176 and Y224 of α9, Figure 6), contribute to high affinity α-ImI binding, while double mutant cycle analyses that defined pairwise interactions between toxin and receptor indicated a dominant interaction between α-ImI Arg7 and Y217 on the receptor^8,26 (in earlier work^8,24–26 Y115, W171 and Y217 were identified as Y93, W149 and Y195 corresponding to their position in the mature protein in which an N-terminal signal sequence has been removed). In silico docking of a solution structure of α-ImI¹⁷ to a model of the α7 nAChR based on the structure of an AChBP³ indicated that Arg7 of α-ImI interacts with three aromatic residues (Y115, W171 and Y217) on the principal side of the subunit interfaces.^8,26 Also, the crystal structure of α-ImI bound to the A. californica AChBP supported the interaction of the α-ImI Asp-Pro-Arg motif with (+) face aromatic residues in nAChRs. In this AChBP, Y93, W147 and Y195 align with Y115, W171 and Y217, respectively, of the α7 nAChR (Figure 6), which correspond to Y93, W149 and Y195, respectively, of the mature α7 nAChR. Both AChBP Y93 and Y195 are close to Arg7 of bound α-ImI (the centroid of Y93 is 5.1 Å from Arg7 C^ζ, and Y195 is 5.4 and 5.6 Å from Cα and C^ζ, respectively, of Arg7). The distances from α-ImI Pro6 C^β and C^γ to the centroids of Y93 and W147 are < 5 Å, and Asp5 is close to Y195 of the AChBP (the Y195 ring centroid is 4.7 Å from Asp5 C^γ). In addition, the AChBP (+) face aromatic Y188 is close to Asp5 of bound α-ImI (its centroid is 5.0 Å from Asp5 C^β and Asp5 O^δ is only 2.5 Å from Y188 OH (Figure S7)). Clearly, Asp5, Pro6 and Arg7 in the first loop of α-ImI interact strongly with the conserved aromatic residues on the principal interface of the α7nAChR and A. californica AChBP. Because these aromatic residues are conserved throughout the nAChR subunits (Figure 6), Asp5, Pro6 and Arg7 of α-RgIA are expected to bind to a similar aromatic pocket on the α9α10 receptor.

A model of α-ImI bound to the α7 nAChR⁸ indicates that Ala9 and Trp10 of α-ImI project towards the complementary side of the subunit interfaces; in this model, N133, S135 and Q139 closely approach Ala9 or Trp10 of the receptor-bound toxin⁸ (in the published model these residues were N111, S113 and Q117, corresponding to their positions in the mature protein in which an N-terminal signal sequence has been removed). This is supported by mutagenesis of these three residues, which altered the potency of the toxin for the receptor,²⁶ and by double mutant cycles analysis showing coupling between Trp10 of the toxin and N133 and S135 of the complementary face.²⁶ Since Ala9 in α-ImI is close to the α7 complementary face (Figure S7),² the much larger and more solvent exposed Arg9 of α-RgIA (Figure S6) would be difficult to accommodate when α-RgIA binds to this receptor subtype. This might explain the lower potency of α-RgIA, relative to α-ImI, on α7 receptors and the increased potency of α-RgIA[R9A], relative to α-RgIA, on these receptors (Figure 9). By contrast, Arg9 of α-RgIA is necessary for stabilizing the α-RgIA-α9α10 interaction and its replacement by Ala greatly reduces α-RgIA potency for α9α10 receptors (Figure 9). Thus, differences between the complementary faces of the α7 and α9α10 receptors make Arg9 of α-RgIA stabilizing in the case of α9α10 binding but destabilizing in the case of α7 binding. In fact, the sequence of the α7 subunit in the vicinity of the α-ImI interacting residues N133, S135 and Q139 is quite different from those of both the α9 and α10 subunits (Figure 6B). Clearly, there are distinct binding surfaces at the complementary faces of α9α10 and α7 nAChRs.

In conclusion, the Asp-Pro-Arg motif in the first loop of α-RgIA appears to be crucial for interaction of the toxin with α9α10 nAChRs (Figure 8), just as the same motif in α-ImI appears to be crucial for stabilizing its interactions with the α7 nAChR. Since this motif of α-ImI binds to aromatic residues on the (+) face of the α7 nAChR and since these aromatic residues are conserved throughout α nicotinic subunits (Figure 6), the Asp-Pro-Arg motif in α-RgIA most likely interacts with the (+) face of the α9α10 receptor in the same way. In contrast, Ala9 of α-ImI appears to interact with the complementary face of the α7 receptor, suggesting that Arg9 of α-RgIA might interact with this face of the α9α10 receptor. Since there is greater α7-to-α9α10 sequence variability on the complementary face than in the principal face aromatic residues (Figure 6B), Arg9 of α-RgIA presumably has charge and or shape compatibility only with the unique molecular surface at the complementary face of the α9α10 subtype. Thus, while the first loops of α-ImI and α-RgIA anchor the peptides to α7 and α9α10 receptors in similar fashion, the different residues at position 9 of the toxins account for subtype specificity by allowing unique interactions with the complementary surfaces. The nature of the aromatic side chain at position 10 and of the C-terminus appears to have a negligible effect on affinity for the α9α10 subtype.

The peptide backbone of the structure of α-RgIA determined in this study is similar to that of α-ImI determined in numerous earlier studies (Figure S5). The dramatic differences in subtype specificity among α-4,3 toxins (as exemplified by α-RgIA and α-ImI) must thus arise from the presentation of different side chains from the same, disulfide-constrained framework. This indicates that α-4,3 toxins, and analogues thereof, are a valuable set of tools with which to define pharmacophores that block nAChRs in a subtype-selective manner.

Our findings contribute to a better understanding of the molecular basis for antagonism of the α9α10 nAChR subtype. This is of potential value pharmacologically given the recent recognition¹⁰ that blockade of the α9α10 nAChR may be valuable in treatment of neuropathic pain. α9α10 nAChRs are found in discrete tissues, including immune cells that mediate inflammation.²⁸α-RgIA and α-Vc1.1, another conotoxin that also blocks α9α10 nAChRs, have potent antinociceptive activity in rat animal models of nerve injury induced pain,^10,29 and both α-RgIA and Vc1.1 reduce the immune response to peripheral nerve damage.¹⁰ Understanding the key residues of α-RgIA involved in α9α10 block thus helps to define a pharmacophore useful in certain pain and inflammatory conditions.³⁰

Materials and Methods

Peptide synthesis

Linear α-RgIA and α-RgIA analogues were synthesized by standard F-moc [N-(9-fluorenyl) methoxycarbonyl] chemistry at a University of Utah core facility. Linear peptides were cleaved from solid phase synthesis resin and were oxidized to give to give the correct disulfide connectivity using an orthogonal cysteine protection strategy that has been described previously.³¹

NMR spectroscopy

Samples of wild-type α-RgIA (1.0 mg) and the α-RgIA analogues Y10W (1.6 mg), D5E (0.8 mg), P6V (0.8 mg) and R9A (0.5mg) were each dissolved in 380 μl H₂O containing 5% 2H₂O for NMR studies. Two-dimensional homonuclear total correlation (TOCSY) spectra with a spin-lock time of 70 ms, nuclear Overhauser enhancement (NOESY) spectra with a mixing time of 250 ms, and double quantum filtered correlation (DQF-COSY) spectra were acquired at 600 MHz on a Bruker DRX-600 spectrometer for wild-type α-RgIA and its four analogues. For wild-type α-RgIA a NOESY spectrum with a mixing time of 250 ms was also acquired at 800 MHz on a Bruker Avance 800 equipped with a TCI-cryoprobe. All spectra were collected at 25 °C unless otherwise stated and were referenced to the water resonance. Water was suppressed using the WATERGATE pulse sequence.³² A series of 1D spectra over the temperature range 5–25 °C, at 5 °C intervals, was collected for each peptide. Amide exchange rates in wild-type α-RgIA were monitored by dissolving freeze-dried material in ²H₂O then recording a series of 1D spectra, followed by 70 ms TOCSY and 50 ms NOESY spectra. In addition, ¹H-¹³C HSQC spectra for the assignment of ¹³C chemical shifts and a ¹H-¹⁵N HSQC spectrum for the assignment of ¹⁵N chemical shifts^33,34 were collected at 5 and 25 °C for wild-type α-RgIA and α-RgIA[Y10W] on a Bruker Avance 500 spectrometer equipped with a TXI-cryoprobe.

Diffusion measurements were performed using a pulsed field gradient longitudinal eddy-current delay pulse sequence^35,36 as implemented by Yao et al.¹³ at 5, 10 and 25 °C for wild-type α-RgIA and at 25 °C for its four analogues.

Spectra were processed using TOPSPIN (Version 1.3, Bruker Biospin) and analyzed using XEASY (Version 1.3.13).³⁷

Structural constraints

³J_HNHA coupling constants were measured from DQF-COSY spectra at 600 MHz, and then converted to dihedral restraints as follows: ³J_HNHα > 8 Hz, φ = −120 ± 40º; ³J_HNHα < 6 Hz, φ = −60 ± 30°. χ¹ angles for some residues were determined based on analysis of a short mixing time (50ms) NOESY spectrum. In addition, PREDITOR ³⁸ (http://wishart.biology.ualberta.ca/predictor) was used to predict torsion angle (φ, ψ, χ¹ and ω) restraints based on chemical shifts. The predicted φ and χ¹ angles were compared with the values measured from the NMR experiments mentioned above. Residues with consistent φ angles (range ± 20º) were constrained in preliminary structural calculations in CYANA (version 1.0.6);³⁹ this included seven φ angle restraints for wild-type α-RgIA (Cys2, Cys3, Arg7, Arg9, Tyr10, Arg11 and Cys12), four φ angle restraints for α-RgIA[Y10W] (Cys3, Arg9, Trp10 and Arg11) and the same seven φ angle restraints as in wild-type α-RgIA for α-RgIA[D5E] and α-RgIA[P6V]. Residues with consistent χ¹ angles (Cys3 and Asp5 for α-RgIA; Cys8, Asp5 and Trp10 for α-RgIA[Y10W]) were constrained in later energy minimization in Xplor-NIH.⁴⁰ Stereo-specific assignments were not made for other non-degenerate protons in the peptides. Some φ and χ¹ angles with predicted confidence scores > 0.75 could not be confirmed experimentally because of overlapping or weak cross-peaks in spectra; these were constrained in the structure refinement before energy minimization in water provided they did not increase total or NOE energy or cause significant distance restraints violations (>0.2 Å); there were two such χ¹ angle restraints for α-RgIA (Cys2 and Tyr10 both with −60±30) and three loose φ angle restraints for α-RgIA[Y10W] (−118±50 for Cys2, −90±40 for Asp5 and −60±40 for Arg7). Other non-degenerate protons in the four peptides were assigned only by in silico means without further intervention. The final numbers of dihedral angle constraints for these peptides are listed in Table 1 and details have been deposited along with distance constraints in BioMagResBank¹¹ as entries 15435, 15368, 15367, 15436 and 15586 for wild-type α-RgIA, α-RgIA[Y10W], α-RgIA[D5E], α-RgIA[P6V] and α-RgIA[R9A], respectively. Wild-type α-RgIA and its analogues contain two disulfide bonds, Cys2-Cys8 and Cys3-Cys12, which were included as structural restraints. No hydrogen bond restraints were included for any of the peptides.

Structure calculations

Intensities of NOE cross peaks were measured in XEASY and calibrated using the CALIBA macro of the program CYANA (version 1.0.6).³⁹ NOEs providing no restraint or representing fixed distances were removed. The constraint list resulting from the CALIBA macro of CYANA was used in Xplor-NIH to calculate a family of 200 structures using the simulated annealing script.⁴⁰ The 40 lowest energy structures for each peptide were then subjected to energy minimization in water; during this process, a box of water with a periodic boundary of 18.856 Ǻ was built around the peptide structure and the ensemble was energy minimized based on NOE and dihedral restraints and the geometry of the bonds, angles and impropers. From this set of structures, final families of 20 lowest energy structures were chosen for analysis using PROCHECK-NMR⁴¹ and MOLMOL.⁴² In all cases, the final structures had no experimental distance violations greater than 0.2 Å or dihedral angle violations greater than 5°. The structures have been deposited in the Protein Data Bank⁴³ with accession codes 2JUT, 2JUS, 2JUR and 2JUQ for wild-type α-RgIA, α-RgIA[Y10W], α-RgIA[D5E] and α-RgIA[P6V], respectively. Structural figures were prepared using the programs MOLMOL⁴² and PyMOL (Delano, W.L. The PyMOL Molecular Graphics System (2002) Delano Scientific, San Carlos, CA, USA. http://www.pymol.org).

nAChR mutagenesis

An α9 nAChR subunit clone (see below) was mutated such that the codon for Trp176 was replaced by a codon for threonine using the QuickChange system (Stratagene).

Electrophysiology

A plasmid carrying a cDNA clone encoding the rat α7 nAChR subunit was provided by S. Heinemann and D. Johnson (Salk Institute, San Diego, CA, USA). Plasmids carrying cDNA clones encoding the rat α9 and α10 nAChR subunits were provided by Belen Elgoyhen (Universidad de Buenos Aires, Buenos Aires, Argentina). For all nAChR subunits, plasmids were linearized with NheI and used to prepare cRNA by in vitro transcription using the T7 mMESSAGE mMACHINE system (Ambion TX, USA). The RNA was injected into oocytes that had been isolated and maintained as before.⁴⁴ 50 nl of water containing 5 ng of the α7 RNA, 5 ng each of the α9 and α10 RNAs or 5 ng each of the α9[W176T] and α10 RNAs were injected per oocyte as described earlier.⁴⁴

The oocytes were maintained in a 30 μl bath made from Sylgard and were gravity perfused with ND₉₆ (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 5 mM HEPES, pH ~7.3). The perfusion solution could be changed to one composed of ND96 and ACh using a series of three way solenoid valves. The oocytes were held at −70 mV using a two-electrode voltage clamp (model OC-725B, Warner Instrument, Hamden, CT). The fluidics and voltage clamp apparatus have been described in detail previously.⁴⁴ The oocytes were pulsed for 1 s by 200 μM ACh (α7 receptor), 50 μM ACh (α9α10 receptor) or 1 mM (α9[Y195T]α10 receptor) at 1-min intervals. These concentrations of ACh are close to the EC₅₀ values for the corresponding receptors in this system.

The concentration dependence of block by conotoxins was measured using a static bath method; the ACh pulses and ND96 flow were halted and toxin was applied to various final concentrations in the bath. After 5 min the flow of ND96 was resumed at the same time as an ACh pulse was applied. The peak current elicited by the first ACh pulse following toxin exposure was converted to a percentage (percent response) of the peak current elicited in controls where ND96 alone, instead of toxin, was applied. In all cases where block was seen, the recovery from block was complete within 1 min (i.e. by the second ACh pulse after toxin application).

The concentration dependence of channel gating by ACh for the nAChRs used in this study was measured by pulsing oocytes expressing the receptors with different concentrations of ACh. In these experiments the electrophysiology system was the same as described above except that the oocytes were maintained in a stream of ND96 flowing along the bottom of a 300 μl pipette tip from the narrow end to the wide end; ND96 flowing out of the tip was removed under vacuum. Two holes were cut in the top of the pipette tip, the larger downstream one held the oocyte and allowed access of the recording electrodes to the ND96, the smaller upstream hole allowed ACh to be added. To pulse with ACh the flow of ND96 was stopped and a 10 μl bolus of ACh was added upstream of the oocyte via the smaller hole; following resumption of the ND96 flow the ACh washed over the oocyte giving an agonist pulse of about 1 s. ACh pulses were applied no more frequently than once every 2 min and to ensure that the higher ACh concentrations were not causing receptor desensitization the rate of pulsing by these concentrations was increased to once per min and no reduction in peak currents were seen.

Data were analyzed and plotted using Prism Software (GraphPad Software, San Diego, CA). Inhibition data were fit to the equation: % response = 100/[1+([toxin]/IC₅₀)^nH] where nH is the Hill slope. Data for the concentration dependence of channel gating by ACh were fitted to: normalized response = 1/[1+(EC₅₀/[ACh])^nH] where the best fit upper plateau of each curve was defined as 1 and all responses to ACh were normalized to it.

Supplementary Material

01

Supplementary Data

Supplementary data associated with this article can be found, in the online version, at doi:xxx

Abbreviations used

ACh

acetylcholine

AChBP

acetylcholine binding protein

α-ImI

α-conotoxin ImI

α-RgIA

α-conotoxin RgIA

nAChR

nicotinic acetylcholine receptor

NMR

nuclear magnetic resonance

NOE

nuclear Overhauser effect