Conotoxin αD-GeXXA utilizes a novel strategy to antagonize nicotinic acetylcholine receptors

Nicotinic acetylcholine receptors (nAChRs) play essential roles in transmitting acetylcholine-mediated neural signals across synapses and neuromuscular junctions, and are also closely linked to various diseases and clinical conditions. Therefore, novel nAChR-specific compounds have great potential for both neuroscience research and clinical applications. Conotoxins, the peptide neurotoxins produced by cone snails, are a rich reservoir of novel ligands that target receptors, ion channels and transporters in the nervous system. From the venom of Conus generalis, we identified a novel dimeric nAChR-inhibiting αD-conotoxin GeXXA. By solving the crystal structure and performing structure-guided dissection of this toxin, we demonstrated that the monomeric C-terminal domain of αD-GeXXA, GeXXA-CTD, retains inhibitory activity against the α9α10 nAChR subtype. Furthermore, we identified that His7 of the rat α10 nAChR subunit determines the species preference of αD-GeXXA, and is probably part of the binding site of this toxin. These results together suggest that αD-GeXXA cooperatively binds to two inter-subunit interfaces on the top surface of nAChR, thus allosterically disturbing the opening of the receptor. The novel antagonistic mechanism of αD-GeXXA via a new binding site on nAChRs provides a valuable basis for the rational design of new nAChR-targeting compounds.

On the other hand, nAChRs are also the targets of various naturally occurring neurotoxins. These nAChR-targeting toxins have not only facilitated structural and functional studies of nAChRs, but also serve as lead compounds in nAChR-targeting drug development 5 . Among those natural toxins, the peptide neurotoxins produced by marine cone snails, generally termed conotoxins, are of particular interest 6 . Several families of conotoxins with different sequences and chemical structures (α -, Ψ -, α B-, α D-, α C-, and α S-conotoxins) can target nAChRs, with α -conotoxins being the most extensively studied ones [7][8][9][10][11] . Whilst Ψ -conotoxins are competitive inhibitors of nAChRs via binding to the ACh-binding site 12,13 , Ψand α D-conotoxins can inhibit nAChRs noncompetitively, at yet unknown binding sites 8,14 . In particular, α D-conotoxins occur naturally as a dimer with complex disulfide connections (10 disulfide bonds per dimer) 8 . This makes the structural study of the α D-conotoxin family more challenging and the molecular mechanism of their function more intriguing.
Here we present the crystal structure and electrophysiological activity profile of a dimeric α D-conotoxin GeXXA. Based on these results, we elucidated the mechanism of action of this dimeric conotoxin. Furthermore, we identified the binding site of this conotoxin on nAChRs, which is clearly different from that of ACh. Together, our results establish a new antagonistic mechanism at nAChRs, providing a valuable basis for the rational design of novel nAChR-targeting compounds.

Results and Discussion
Identification of αD-conotoxin GeXXA. We identified and isolated a novel α D-conotoxin GeXXA from the venom of Conus generalis (Fig 1A). Reduction of this toxin shifted its molecular weight from 11249.0 Da to 5635.0 Da, and alkylation with N-ethylmaleimide (NEM) increased its weight to 6885.0 Da (Fig. S1A). These results indicate this toxin contains 10 Cys residues per peptide and exists as a homodimer with inter-chain disulfide bond(s). N-terminal sequencing and subsequent cDNA cloning of this toxin revealed that each peptide chain of α D-GeXXA comprises 50 amino acid residues sharing high sequence homology with other known α D-conotoxins (Fig. S1).
As α D-conotoxins can act as noncompetitive inhibitors of nAChRs 8 , we first tested the effects of α D-GeXXA on ACh-evoked currents mediated by different nAChR subtypes expressed in Xenopus oocytes. Our electrophysiology data showed that α D-GeXXA has strong inhibitory activity on α 9α 10, α 7 and α 3β 2 subtypes, moderate inhibitory activity on α 3β 4 and α 1β 1δ ε subtypes, and weak activity on α 4β 2 and α 4β 4 subtypes ( Fig. 1C and Table 1). In particular, α D-GeXXA is most potent against human α 9α 10 subtype with an IC 50 of 28 nM.
Crystal structure of αD-GeXXA. To gain insight into its biological function, we determined the crystal structure of native α D-GeXXA using ab initio methods 15 and refined it to 1.5 Å resolution (Table  S1 and Fig. 2). There is one conotoxin homodimer with a pseudo two-fold symmetry per asymmetric unit. Each peptide chain consists of an N-terminal domain (NTD, residues 1-20) and a C-terminal domain (CTD, residues 21-50). The NTD comprises an N-terminal loop and a β -strand, and the CTD assumes several extended loop conformations. The dimerization involves mainly the N-terminal loops and β -strands of the NTDs, which is further stabilized by two inter-chain disulfide bonds between Cys6 of one chain and Cys18 of the other. There are three disulfide bonds (Cys24-Cys36, Cys29-Cys46 and Cys34-Cys48) in the CTD, making the CTD adopt a compact structure. The two CTDs flank the dimeric NTDs, and the relative conformation of the NTD and the CTD is stabilized by a disulfide bond between Cys19 and Cys28 (Fig. 2).
Preparation of monomeric GeXXA-CTD. Interestingly, the CTD of α D-GeXXA adopts a canonical inhibitory cystine knot (ICK) disulfide linkage, as observed in many 6-Cys-residue-containing bioactive peptides, including several families of conotoxins 16 . This observation prompted us to speculate that the CTD of α D-GeXXA alone may exhibit inhibitory activity against nAChRs. To obtain an isolated CTD, we first synthesized a peptide of residues 21-50, with Cys28 replaced by Ser and the thiol groups of both Cys24 and Cys36 protected by the acetamidomethyl groups (Acm) (Fig. S2A). Oxidation of the synthetic peptide with GSH/GSSH yielded two major products (Fig. S2B). Partial reduction and LC-MS/MS analysis showed that the product in peak 1 has the correctly connected Cys29-Cys46 and Cys34-Cys48 disulfide bonds (Fig. S3). Subsequent iodine oxidation of this intermediate product led to formation of the third disulfide bond between Cys24 and Cys36, thus yielding the monomeric CTD (Fig. S2C).
GeXXA-CTD has nAChR-inhibitory activity. Indeed, the monomeric GeXXA-CTD showed inhibitory activity against human α 9α 10 subtype (IC 50 of 2.02 μ M), but had little or no effect on other nAChR subtypes (Table 1). In general, the activity of GeXXA-CTD is weaker than that of the full-length dimeric α D-GeXXA, making GeXXA-CTD apparently specific to the α 9α 10 subtype. While focusing on this subtype, we found that GeXXA-CTD has a 10-fold higher potency on rat α 9α 10 (IC 50 of 198 nM) than on human α 9α 10 nAChR ( Fig. 3A and Table 2). Interestingly, GeXXA-CTD also exhibited a comparably high potency (IC 50 of 224 nM) on a hybrid receptor of human α 9 and rat α 10 subunits (hα 9rα 10) 17 (Fig. 3A). These results suggest that the α 10 subunit determines the preference of GeXXA-CTD for rat over human α 9α 10 nAChR. Since the nAChR-inhibitory activities of α D-GeXXA and GeXXA-CTD were measured after extracellular application (see materials and methods), the species preference might be due to residue differences in the extracellular domain of human and rat α 10 subunits.
Scientific RepoRts | 5:14261 | DOi: 10.1038/srep14261 Identification of the binding site of αD-GeXXA. To investigate the potential binding site of the α 9α 10 nAChR for α D-GeXXA, we compared the sequences of the extracellular domains of human and rat α 10 subunits and identified differences in 12 residues (Fig. S4). Based on the EM structure of Torpedo α 1β 1δ γ nAChR 18 and the crystal structures of AChBP 12 , the corresponding positions of these 12 differing residues are mostly well scattered on the surface of the extracellular domain ( Fig. S5A and B). However, the pseudo two-fold symmetry of the two CTDs in the dimeric α D-GeXXA implies existence of two equivalent binding sites on each nAChR, presumably on two non-adjacent subunits. On the other hand, the length of α D-GeXXA (up to 52.4 Å between the Cα atoms of the two C-terminal Met50 residues) makes it unlikely that α D-GeXXA binds in the nAChR central pore or on the outside-facing surface of the pentameric extracellular domains (Fig. S5C). Thus, the top surface of nAChR seems the most likely binding site for α D-GeXXA. Among the 12 differing residues, only residue 7 is located on the top surface of nAChR. We therefore hypothesized that His7 is the likely candidate conferring specificity of α D-GeXXA to rat α 10 subunit.
To verify the functional role of residue 7 of nAChR α 10 subunit in α D-GeXXA binding, we mutated Leu7 of human α 10 subunit to His, the corresponding residue in rat α 10 subunit. The inhibitory activity of a competitive α -conotoxin Vc1.1 was not affected by this mutation (data not shown), thus excluding the possibility of the mutation introducing significant structural change. Remarkably, GeXXA-CTD showed an IC 50 of 183 nM on this hα 9α 10[L7H] mutant. This is comparable to the activity on rα 9α 10 but is 11-fold lower than that on hα 9α 10 ( Fig. 3A and Table 2), strongly suggesting that His7 of rat α 10 subunit is involved in the interaction with GeXXA-CTD.
To further confirm this potential binding site for α D-conotoxins, the activity of the natural dimeric α D-GeXXA was also measured on these nAChR subtypes. The dimeric α D-GeXXA exhibited a strong preference on rat α 9α 10 subtype rather than human α 9α 10 subtype, the IC 50 on rat α 9α 10 subtype (1.2 nM, Fig. 3B and Table 2) being clearly lower than that on human α 9α 10 subtype. Similar to the situation of GeXXA-CTD, the dimeric α D-GeXXA showed the same potency on the hybrid receptor of human α 9 and rat α 10 subunits (hα 9rα 10), with an IC 50 of 1.2 nM (Fig. 3B and Table 2). Furthermore, the L7H mutation of human α 10 subtype clearly enhanced the potency of α D-GeXXA, its IC 50 on hα 9α 10[L7H] getting close to that on rα 9α 10 and hα 9rα 10 ( Fig. 3B and Table 2). These results further support the notion that His7 of the rat α 10 subunit confers the species preference of α D-GeXXA and may serve as the binding site for α D-GeXXA.
A cooperative two-site binding model of αD-GeXXA. Based on sequence alignment, the critical residue, His7 of rat α 10 subunit, corresponds to Asn9 of the Torpedo δ subunit or Glu8 of the Torpedo γ subunit (Fig. S4), which is located on the complementary (− ) side of each subunit and faces towards the principal (+ ) side of its clockwise adjacent subunit (Fig. S5C). Therefore, the binding site of α D-GeXXA is probably located at the interface between the α 10 subunit and its clockwise adjacent subunit. In the pentameric α 9α 10 nAChR, there are two α 9 and three α 10 subunits 19 . Following the nomenclature for ACh-binding sites on nAChR 4 , the three α 10-involving interfaces would be two "α 9α 10" interfaces and one "α 10α 10" interface, all of which could be potential binding sites for α D-GeXXA. However, the pseudo two-fold symmetry and the length of the dimeric full-length α D-GeXXA suggest that the two CTDs of this toxin most likely bind the two "α 9α 10" interfaces at the top surface of α 9α 10 nAChR (Fig 4). By doing so, α D-GeXXA, and possibly all the α D-conotoxins, can allosterically and cooperatively perturb the conformational changes of the receptor and opening of the channel.  This cooperative, two-site binding model of α D-GeXXA is also supported by our electrophysiological data. Firstly, inhibition of nAChR by both α D-GeXXA and GeXXA-CTD often gave Hill slopes greater than 1 (Tables 1 and 2), suggesting cooperative rather than single-site binding on nAChR. Secondly, the dimeric α D-GeXXA is considerably more potent than the monomeric GeXXA-CTD (Table 1 and Fig. 3). Thirdly, the dimeric α D-GeXXA exhibits much slower dissociation kinetics than the monomeric GeXXA-CTD (Fig. 3C,D). Interestingly, all these properties have been observed in a study of   Table 2. Inhibitory activities of monomeric GeXXA-CTD and dimeric αD-GeXXA on α9α10 nAChR from different species.
polymer-linked ligand dimers 20 . We now show that this dimerization strategy is adopted by a natural toxin to gain higher potency. In summary, by resolving the crystal structure and performing structure-guided dissection of a dimeric conotoxin α D-GeXXA, we demonstrate that α D-conotoxins inhibit nAChR most likely by binding of the two CTDs cooperatively to two inter-subunit interfaces on the top surface of nAChR. This working mechanism is distinct from that of another dimeric conotoxin that targets AMPA receptors 21 . The binding site of α D-conotoxin on nAChR differs from the binding sites of the endogenous ligand ACh, and the competitive α -conotoxins 12,13 and α -bungarotoxin 22 . However, the binding site stoichiometry of α D-GeXXA is coincidently the same as that of ACh on the muscle subtype and neuronal heterogeneous nAChR (that is, 2 of 5 inter-subunit interfaces) 23 (Fig. 4). The cooperative inhibitory mechanism of α D-GeXXA via a novel binding site on nAChRs provides a valuable basis for the rational design of new nAChR-targeting drugs.

Methods
Toxin purification and characterization. Conus generalis specimens were collected from the South China Sea. To extract the crude venom, the venom duct of living snails was dissected into short fragments and venom was extracted successively with 0.1% (v/v) trifluoroacetic acid (TFA), and 0.1% TFA in 20%, 30%, 40% and 50% acetonitrile. Supernatants were pooled and lyophilized.
Electrophysiological recordings from nAChRs exogenously expressed in Xenopus oocytes. RNA preparation, oocyte preparation, and expression of nAChR subunits in Xenopus oocytes were performed as described previously 24 . Briefly, plasmids with cDNAs encoding the rat α 1, α 3, α 4, α 9, α 10, β 1, β 2, β 4, δ , ε and human α 7 subunits subcloned into the oocyte expression vector pNKS2 and human α 9 and α 10 subunits subcloned into the pT7TS vector were used for mRNA preparation using the mMESSAGE mMACHINE Kit (Ambion Inc., USA). All oocytes were injected with 5 ng of cRNA and kept at 18 °C in ND96 buffer (96 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES, pH 7.4), supplemented with 50 mg/L gentamycin and 100 μ g/ml penicillin/streptomycin for 2-5 days before recording. Membrane currents were recorded from Xenopus oocytes using a GeneClamp 500B amplifier (Molecular Devices) in a two-electrode (virtual ground circuit) voltage-clamp setup. Both the voltage-recording and current-injecting electrodes were pulled from borosilicate glass (GC150T -7.5, The two CTDs of α D-GeXXA, which are shown in cyan and pink, respectively, with roughly 180° rotation, bind at the top surface of the two "α 9α 10" interfaces to inhibit the opening of the nAChR. Harvard Apparatus Ltd.) and had resistances of 0.3-1.5 MΩ when filled with 3 M KCl. All recordings were conducted at room temperature (21-23 °C) using a bath solution of ND96 as described above. During recording, the oocytes were perfused continuously at a rate of 1.5 ml/min, with 300 s incubation times for peptides. Acetylcholine (200 μ M for α 7 and 50 μ M for all other nAChR subtypes) was applied for 1 s at 2 ml/min, with 3-4 min washout periods between applications. Cells were voltage-clamped at a holding potential of − 80 mV. Data were filtered at 100 Hz and sampled at 500 Hz. Peak ACh-evoked current amplitude was measured before and after incubation with peptide.
Concentration-response curves for antagonists were fitted by unweighted nonlinear regression to the following logistic equation 1 where E x is the response, X is the antagonist concentration, E max is the maximal response, n H is the slope factor, and IC 50 is the antagonist concentration giving 50% inhibition of maximal response. All electrophysiological data were pooled (n = 4-8 oocytes for each data point) and represent arithmetic means ± standard error of the fit. Computation was done using GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA, USA).
Crystallization and structure determination. The powder of 1.0 mg native α D-GeXXA was dissolved in a buffer containing 20 mM Tris-HCl (pH 8.0) and 100 mM NaCl. Crystallization of α D-GeXXA was performed using the hanging drop vapor diffusion method by mixing 1.5 μ l protein solution (about 10 mg/ml) and 1.5 μ l reservoir solution at 16 °C. Crystals were grown from drops consisting of a reservoir solution of 0.1 M citrate acid (pH 5.0) and 10% PEG 6000 after about 1 week. The diffraction data were processed, integrated, and scaled together as two datasets with HKL2000 25 . Dataset 1 was processed at high resolution for ab initio phasing and dataset 2 with reasonable statistics was used for structure refinement. The structure of α D-GeXXA was determined by ab initio methods using the program Acorn 15 . Phases were determined with dataset 1 by setting optimal Acorn parameters to start from a random atom (no prior knowledge) to determine substructure and then applied to the program Acorn-MR to produce an interpretable electron density map. An initial model of 63 out of 100 residues for two monomers was constructed automatically by warpNtrace mode of ARP/wARP 26 . The remaining residues and additional water molecules were built manually using Coot with dataset 2 27,28 . Structure refinement was performed using Refmac5 and Phenix 27,29 . The stereochemistry of the protein model was analyzed using MolProbity 30 . Structure analysis was carried out using programs in CCP4 31 . Figures were generated using Pymol (http://www.pymol.org). Statistics of the structure refinement and the quality of the final structure model are summarized in Table S1.
Preparation of GeXXA-CTD. The linear peptide of GeXXA-CTD, with Cys24 and Cys36 protected by Acm, was synthesized by the Chinese Peptide Company (Hangzhou, China). The peptide was first oxidized with 1 mM GSSG/GSH (Fig. S2), which produced two products. To examine the disulfide linkage, the first peak (Peak 1) was partially reduced with Tris(2-carboxyethyl)phosphine (TCEP) and alkylated with NEM, and then fully reduced with DTT and alkylated with iodoacetamide (IAA). The resultant product of Peak 1 was digested with trypsin and analyzed with LC-MS/MS on an Orbitrap Elite (ThermoFisher, USA), revealing the Cys29-Cys46 and Cys34-Cys48 linkages (Fig. S3). Peak 1 was then treated with iodine to remove the Acm group and oxidize the disulfide bond between Cys24 and Cys36 (Fig. S2).