Synthesis of full-length homodimer αD-VxXXB that targets human α7 nicotinic acetylcholine receptors

Abstract

αD-Conotoxin VxXXB is a pseudo-homodimer that allosterically inhibits nicotinic acetylcholine receptors (nAChRs) with high potency and selectivity. However, challenges in synthesizing αD-conotoxins have hindered further structure–function studies on this novel class of peptides. To address this gap, we synthesized and characterized its C-terminal domain (CTD) and N-terminal domain (NTD). The CTD inhibited α7 nAChRs (IC₅₀ of 23 nM, measured via FLIPR assays) and bound at the acetylcholine binding protein (Ls-AChBP) through an allosteric binding mode determined from radioligand binding assays. The anti-parallel dimeric NTD synthesised via a regioselective strategy also inhibited α7 nAChRs but with reduced potency (IC₅₀ of 30 μM). The α-ketoacid-hydroxylamine (KAHA) method generated CTD linked to the NTD (VxXXB-NC; α7 IC₅₀ of 27 nM) and full-length synthetic VxXXB variant (α7 IC₅₀ of 11 nM), while the three other native chemical ligation approaches proved unsuccessful. This work underpins further characterisation of the structural components contributing to αD-conotoxin affinity, selectivity and allosteric inhibition of nAChR function that may prove useful in the development of new treatments for nAChR-related disorders.

αD-VxXXB is a homodimer of paired 50-residue peptide with ten conserved cysteines. Here we reported the first synthetic approach to full-length VxXXB via α-ketohydroxylamine ligation that enables further characterisation of this novel class of allosteric nAChR inhibitors.

Introduction

Conotoxins are small disulfide-rich peptide toxins extracted from the venom of predatory marine snails of the family Conidae. A key feature of conotoxins is their highly conserved cysteine frameworks, which stabilize their protein-like secondary structures and unique bioactive folds.^1,2 At least 28 structural frameworks have been reported to-date based on the number of cysteine residues, their loop size and distinct disulfide bond connectivity.³ The αA-, ψ-, αB, αD-, αC- and αS-conotoxins specifically inhibit nAChRs,^4,5 with the αA-conotoxins being the best characterized employing a C^IC^IIX_mC^IIIX_nC^IV framework with globular (Cys^I–Cys^III, Cys^II–Cys^IV) connectivity^6,7 that stabilises a short 3₁₀ α-helical backbone.⁸ In contrast, all other families of conotoxins inhibiting nAChRs show more diverse structures and mechanism of action, with the ψ- and αD-conotoxin families reported to non-competitively inhibit α7 and related nAChRs at an undefined binding site.⁹

The homopentameric α7 nAChRs are implicated in a range of neurological diseases due to its abundant expression in the nervous system. α7 nAChRs are sensitive to positive allosteric modulators (PAMs) and negative allosteric modulators (NAMs), with some PAMs showing clinical promise in neurological diseases.¹⁰ While many PAMs and NAMs appear to act within the conserved transmembrane domain,¹⁰ several bind either beneath the top helix of the extracellular domain of nAChRs, at the subunit interface of the extracellular domain, or in the vestibule pocket opposite the agonist binding site.¹¹ Despite binding allosterically, identified small molecule allosteric modulators often have low selectivity for specific nAChR subtypes, potentially hindering their potential for clinical development.

VxXXA, VxXXB and VxXXC from the venom of Conus vexillum were the first D superfamily α-conotoxins to be characterized.⁹ These αD-conotoxins uniquely exist as 11 kDa covalently linked homo-dimeric proteins (Fig. 1A). VxXXB has highest potency at α7 (0.4 nM), followed by α3β2 and α4β2 nAChRs, and acts allosterically to inhibit orthosteric ligand binding to the acetylcholine-binding protein (AChBP).⁹ Recently, the high-resolution X-ray structure of αD-conotoxin GeXXA from the venom of Conus generalis revealed the disulfide bonding architecture of αD-conotoxins, including three disulfide bonds that stabilize the canonical inhibitory cysteine knot (ICK) fold (Fig. 1B) found in many venom peptides.^12,13

Fig. 1. Synthesis and bioactivity of VxXXB CTD(21–50) and VxXXB CTD(19–50). (A) Sequence of synthetic VxXXB with VxXXB CTD(21–50) and VxXXB CTD(19–50) highlighted in red and blue boxes respectively. (B) RP-HPLC chromatograms of the reduced (lower trace) and oxidised (upper trace) (a) VxXXB CTD(21–50) and (b) VxXXB CTD(19–50) yielding single peaks with expected monoisotopic masses on analytical RP-HPLC (Table S1 and Fig. S1†). (C) Concentration–response curves for VxXXB CTD(21–50) and VxXXB CTD(19–50) inhibition of (a) α7 nAChRs responses in SH-SY5Y cells via FLIPR assays and (b) [³H]-epibatidine binding to Ls-AChBP via radioligand binding assay. (D) Saturation binding of [³H]-epibatidine to Ls-AChBP in the absence and presence of (a) 40 μM of VxXXB CTD(21–50) and (b) 20 μM of VxXXB CTD(19–50). Data represent means SEM of triplicate data from three independent experiments.

As VxXXB has high sequence similarity (52%) to GeXXA, we proposed that the VxXXB pseudo-homodimer adopted the same disulfide bond connectivity, with the dimer formed by disulfide bonds between Cys6 (Cys^I) of one chain and Cys18 (Cys^II) of the other, constituting VxXXB N-terminal domain (NTD) (Fig. 1A). Meanwhile, the C-terminal domain (CTD) of VxXXB flanked either sides of the NTD contains three disulfide bonds (Cys24–Cys36, Cys29–Cys46 and Cys34–Cys48), allowing it to adopt an ICK fold (Fig. 1A). The CTD and NTD joined though a disulfide bond between Cys19–Cys28. Here we couple the CTD (C) and truncated and full-length NTD (N) domains of VxXXB using the α-ketoacid-hydroxylamine (KAHA) strategy to produce VxXXB NC and full length (CNC) and compare their activity at human α7 nAChR and Lymnaea stagnalis-AChBP (Ls-AChBP) (Fig. 4A). The successful synthesis of VxXXB provides new opportunities for the development of novel peptidic modulators nAChRs involved in channel-mediated disorders.

Fig. 4. The synthesis and characterization of VxXXB N(6–18)C and VxXXB CNC by KAHA ligation. (A) Sequences of VxXXB N(6–18)C and VxXXB CNC by KAHA ligation. (B) RP-HPLC chromatograms showing the formation of ligated VxXXB N(6–18)C (a) and ligated VxXXB CNC (b) with monoisotopic mass of 6756.90 Da and 11 651 Da respectively (Fig. S6†). (C) Concentration-response curves for VxXXB NC,VxXXB N(6–18)C and VxXXB CNC at α7 nAChRs on SH-SY5Y cells via FLIPR assay. (D) Displacement of [³H]-epibatidine by (a) Ls-AChBP by VxXXB CTD(19–50) and VxXXB CNC in competition radioligand binding assays and (b) saturation radioligand binding assays. Data represent means SEM of triplicate data from three independent experiments.

Experimental method

Synthesis and oxidation of VxXXB CTDs, CTD(19–50) and CTD(21–50)

αD-Conotoxin CTD variants were assembled using Fmoc chemistry solid-phase peptide synthesis (SPPS) methodology on a Liberty PRIME peptide synthesizer (CEM, USA), with Cys residues orthogonally protected with the acid-labile protecting group trityl, Trt. Purified reduced peptide was dissolved in 0.1 M Tris·HCl, 2 mM EDTA at pH 8.7. 1 mM reduced glutathione (GSH)/oxidized glutathione (GSSG) was added fresh to initiate oxidation. The mixture was stirred at room temperature with exposure to air for 6 h to generate a single major product that was isolated by semi-prep reverse phase-high performance liquid chromatograph (RP-HPLC). This method was used for the synthesis of CTD variants for VxXXB ligation.

Synthesis of VxXXB N-terminal domain (NTD), NTD(1–18) and NTD(6–18) dimers

VxXXB NTD(1–18) and NTD(6–18) variants were assembled using Fmoc-SPPS in a CEM Liberty Prime microwave peptide synthesiser on an Fmoc-Rink-ProTide resin (Iris Biotech GmbH) (scale 0.1 mmol). Cys residues were orthogonally protected in pairs, where Cys^II on chain A and Cys^I on chain B were protected with S-acetamidomethyl (Acm), while Cys^I on chain A and Cys^II on chain B were protected with the Trt group.

Peptide cleavage and purification

Cleavage of all synthesized peptides from resin and global side chain deprotection was achieved by treatment with scavenger mixture (trifluoroacetic acid (TFA)/water/triisopropylsilane (TIPS), 95 : 2.5 : 2.5, v/v/v) for 30 min at 40 °C on a Razor system (CEM, USA). The cleaved peptide was precipitated with cold diethyl ether, centrifuged 3-times at 2000 × g, and the resulting pellet was redissolved in 50% aqueous acetonitrile (ACN) and lyophilized.

The crude products were purified on a C18 Vydac column (Vydac 218TP, Grace) by preparative reverse phase-high performance liquid chromotagraphy (RP-HPLC) using a linear gradient of 5–45% solvent B (90% acetonitrile (ACN), 0.05% aqueous trifluoroacetic acid (TFA)) over 40 min at a flow rate of 20 mL min⁻¹. Oxidised peptides were purified by semi-preparative RP-HPLC on a C18 Vydac column (Vydac 218TP, Grace) using a linear gradient of 10–40% solvent B over 30 min at a flow rate of 4 mL min⁻¹. Final product was collected and analyzed by analytical HPLC using a linear gradient of 10–40% solvent B over 30 min at a flow rate of 1 mL min⁻¹, electrospray ionization-mass spectrometry (ESI-MS) and matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) on an ABSciex API5000 LC/MS/MS. Percentages of peptide yield was calculated relative to the starting crude amount.

Oxidation of VxXXB NTD(1–18)/NTD(6–18) variants by 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) with palladium

Disulfide bonds of the NTD were formed selectively by directed two-step oxidation, with the Trt protecting groups of the first Cys pairs removed after peptide cleavage from resin, while the Acm groups on the second Cys pairs remained intact. Activation of the thiol group of Cys^II in NTD chain B (0.6 μg ml⁻¹) was performed using 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) in 150 mM PBS pH 7.4 for 15 min at room temperature in the dark, followed by purification by RP-HPLC to remove excess DTNB. The first disulfide bond was formed by mixing the activated chain B (0.6 μg μl⁻¹) and chain A (0.4 μg μl⁻¹) in 150 mM phosphate-buffer-saline (PBS) at pH 7.4 for 15 min at room temperature, and the resulting NTD chain A-chain B heterodimer with one disulfide bond was purified by semi-preparative HPLC.¹⁴ To form the second disulfide bond, 0.5 mM of the lyophilized dimeric chain A-chain B was dissolved in 6 M Gn·HCl, pH 1.0 and added with 15 eq. of PdCl₂ (stock of 200 mM in 6 M Gn·HCl) for 5 min at 37 °C. Then, 30 eq. sodium diethyldithiocarbamate (DTC) (stock of 150 mM in H₂O), followed by 10 eq. disulfiram (DSF) (stock of 50 mM in acetonitrile) was added to the mixture with pH adjusted to 7 and incubated at 37 °C for 30 min.¹⁵ The mixture was then diluted with solvent A (0.05% TFA) and the fully oxidised chain A-chain B heterodimer isolated by semi-prep RP-HPLC, with identity confirmed by ESI-MS before lyophilisation.

The ligation of CTD(19–50) and VxXXB NTD variants

Enzymatic ligation

For ligation using sortase (SrtA5°) enzyme, N-terminals of VxXXB CTD(19–50) was incorporated with di-Gly (GG), CTD(19–50)-[1], while the C-terminal NTD(6–18) chain A was added with an –LPTAGG– sequence to make NTD(6–18)-[1]. CTD(19–50)-[1] and NTD(6–18)-[1] were dissolved in SrtA5° ligation buffer containing 50 mM Tris, 150 mM NaCl, and 10 mM CaCl₂ at pH 8 at concentrations of 200 μM and 100 μM, respectively. The solution was added with 30 μM SrtA5° and incubated at 37 °C. The reaction mixture was stirred and monitored at various time points by analytical HPLC. The reaction was quenched with the addition of TFA to a final concentration of 1%.

Hydrazone ligation

For hydrazone ligation, N-terminals of VxXXB CTD(19–50) was incorporated with Ser to make CTD(19–50)-[2], while the NTD chain A was synthesized on a freshly prepared 2-chlorotrityl chloride resin [2-Cl-(Trt)-Cl-NHNH₂] (2-CTC) treated with hydrazine hydrate for hydrazination at its C-terminal to make NTD(6–18)-[2]. To oxidise the N-terminal Ser into an N-terminal aldehyde moiety, 0.5 mM of VxXXB CTD(19–50)-[2] was dissolved in 10 mM Na₃PO₄ (pH 7), followed by the addition of 1.5 eq. of NaIO₄ (a freshly prepared stock of 100 mM NaIO₄ in H₂O) for 2 min at room temperature in the dark.¹⁵ Oxidation was quenched by the addition of N-α-Fmoc-l-serine (Iris Biotech GmbH) to a final concentration of 5 mM. Ligating the NTD(6–18)-[2] and the aldehyde moiety of CTD(19–50)-[2] was performed at a 2 : 1 ratio of hydrazide : aldehyde in 100 mM sodium citrate (pH 4.5). The reaction was incubated at −20 °C in the dark and monitored at various time points by analytical HPLC. The reaction was quenched with the addition of TFA to a final concentration of 1%.

Copper(i)-mediated azide-alkyne cycloaddition (CuAAC) ligation

For CuAAC chemistry, N-terminal of VxXXB CTD(19–50) was incorporated with (S)-2-amino-3-azidopropanoic acid (Aza), CTD(19–50)-[3], while the C-terminal of NTD(6–18) chain A was added with l-propargylglycine [Prg] to make NTD(6–18)-[2]. CTD(19–50)–[3] and NTD(6–18)-[3] were dissolved in a mixture of H₂O/tBuOH (70 : 30 v/v) at a final concentration of 0.5 mg ml⁻¹. To form in situ Cu^I catalyst, CuSO₄·(H₂O)₅ (10 eq.) from a 50 mM CuSO₄·(H₂O)₅ stock and ascorbic acid (5 eq.) from 100 mM ascorbic acid stock were then added to the solution. The reaction mixture was stirred and monitored by analytical HPLC, then quenched with TFA (1% final concentration).

α-Ketohydroxylamine (KAHA) ligation

For KAHA ligation, N-terminal of VxXXB CTD(19–50) was incorporated with (S)-5-oxaproline (Hse) to make CTD(19–50)-[4]. To make VxXXB N(1–18)C and VxXXB N(6–18)C, the NTD chain A was synthesized on an amide resin preloaded with Fmoc-leucine-ketoacid (KA) (scale 0.1 mmol) (provided by Dr Jeffrey Bode, Zurich, Sweden) to make NTD(6–18)-[4] and NTD(1–18)-[4], respectively. For the synthesis of full-length VxXXB CN(1–18)C, both chain A and chain B of NTD(1–18) were synthesised on an Fmoc-leucine-KA-preloaded amide resin (scale 0.1 mmol) to incorporate the ligation points at both C-terminals of homodimeric antiparallel, NTD(1–18)-[5]. Ligation was performed in 80% DMSO containing 0.1 M oxalic acid solution. To make VxXXB N(1–18)C/VxXXB N(6–18)C, the CTD(19–50)-[4] and NTD(6–18)-[4]/NTD(1–18)-[4] were mixed at a concentration of 1.5 mM and 2.0 mM, respectively, while VxXXB CN(1–18)C was achieved by mixing CTD(19–50)-[4] and NTD(1–18)-[5] at a concentration of 1.5 mM and 3.0 mM, respectively. The reaction mixture was stirred and monitored at various time points by analytical HPLC, then quenched with the addition of TFA to a final concentration of 1%. Lyophilized peptide ester was dissolved in 0.1 M NH₄HCO₃ at pH 9.5 for 2 h. The O → N shift was monitored by HPLC, then acidified by 1% TFA and purified.¹⁶

Functional assays

FLIPR assays were performed to measure the inhibition of synthesized peptides on choline-activated intracellular increases in calcium at α7 nAChRs endogenously expressed on SH-SY5Y neuroblastoma cells. Radioligand binding assays were performed to quantify the ability of synthesized peptides to displace the binding of [³H]-epibatidine to the recombinantly expressed Ls-AChBP. Details of the assays are described in ESI.†

Results

Synthesis of monomeric VxXXB CTD variants

VxXXB CTD was synthesized as two variants, CTD(21–50) and CTD(19–50). CTD(21–50) comprised residues 21–50, with the unpaired Cys28 replaced by Ser, while full-length CTD(19–50) had all four disulfide bonds (Fig. 1A and Table S1†). All Met residues were replaced with norleucine (Nle) to prevent side reactions upon oxidation. The thiol groups of six Cys residues were protected with acid-labile protecting group trityl (Trt). Oxidation of the synthetic peptide with reduced glutathione (GSH)/oxidized glutathione (GSSG) yielded one major product with expected mass of 3261.43 Da for fully oxidized VxXXB CTD(21–50) and 3535.13 Da for VxXXB CTD(19–50) obtained on analytical RP-HPLC and ESI-MS (Fig. 1B, S1 and Table S1†).

Both CTD(21–50) and CTD(19–50) showed potency at Ls-AChBP and human α7 nAChRs (Fig. 1C and D), in which CTD(19–50) exhibited 2-fold and 8-fold higher potency at Ls-AChBP and human α7 nAChRs, respectively, compared to CTD(21–50) (Fig. 1C and Table 1). Using a pairwise comparison of the IC₅₀ values for binding affinity for Ls-AChBP and potency at human α7 nAChRs, CTD(19–50) was significantly more potent than CTD(21–50) (p < 0.05). CTD(19–50) and CTD(21–50) incompletely inhibited [³H]-epibatidine binding to Ls-AChBP, 29% and 20% inhibition, respectively, at the highest concentration tested (Fig. 1Cb), suggesting a non-competitive mechanism of action, similar to native full-length VxXXB.⁹ Consistent with this interpretation, these CTD variants also showed non-surmountable inhibition in saturation-binding experiments on Ls-AChBP (Fig. 1D).

IC₅₀ values for displacement of [³H]-epibatidine binding on Ls-AChBPs and inhibition of α7 nAChR current in SH-SY5Y cells by VxXXB variants.

	Ls-AChBP	Ratio	α7 nAChRs	Ratio
	IC50 ± SEM (μM)		IC50 ± SEM (μM)
VxXXB CTD(21–50)	20 ± 0.82	1	0.176 ± 0.020	1
VxXXB CTD(19–50)	10 ± 0.23	0.50a	0.023 ± 0.001	0.13a
NTD(6–18)	ND		46 ± 10	261a
NTD(1–18)	ND		35 ± 10	199a
VxXXB N(6–18)C	ND		0.037 ± 0.003	0.21a
VxXXB NC	ND		0.027 ± 0.005	0.15a
VxXXB CN(1–18)C	0.8 ± 0.09	0.04a	0.011 ± 0.0008	0.06a
VxXXB native	0.011 (ref. 9)	0.00055a	0.0004 (ref. 9)

^aIndicates significant difference in IC₅₀ values to CTD(21–50) (p < 0.05).

Synthesis of VxXXB NTD

A regio-protection strategy was employed to synthesize the homodimer VxXXB NTD formed by two anti-parallel interchain disulfide bonds. A truncated version of VxXXB NTD, VxXXB NTD(6–18), was first made with the five N-terminal residues on both chains omitted (Fig. 3A). To ensure the proper pairing of the disulfide bonds, Cys residues were orthogonally protected in pairs with acid-labile Trt on Cys^I of chain A and Cys^II of NTD chain B and acid-stable S-acetamidomethyl (Acm) on Cys^II of chain A and Cys^I of chain B. Although the first disulfide bond should be formed between Cys^I of NTD chain A and Cys^II of NTD chain B, mixing these two peptides generated three possible products: a chain A-chain A homodimer, a chain B-chain B homodimer, and a chain A-chain B heterodimer. To ensure formation of the heterodimer, the deprotected Cys residue of chain B was activated with the thiol-activating reagent DTNB (Fig. 2A and B). The activated chain B was mixed with the unmodified chain A, to specifically form the first inter-chain disulfide bond between Cys^I of activated chain B and Cys^II of chain A. To form the second bond, we added PdCl₂ to this NTD heterodimer, followed by the Pd scavenger DTC and fresh DSF at pH 7. The second disulfide bond was formed as shown by HPLC-MS analysis and the desired product with an expected mass of 3037.65 Da isolated in 6–15% yield (Fig. 2A and B and S2†). This method was then applied to the synthesis of the NTD(1–18 variants, with γ-carboxyglutamic acids at position 3 and 5 replaced with glutamic acid with a mass of 4187.23 Da (Fig. S2†).

Fig. 3. Schemes to ligate VxXXB N(1–18)C and VxXXB N(6–18)C using four different ligation strategies; enzyme ligation, hydrazone ligation, CuAAC ligation and KAHA ligation, with yields indicated.

Fig. 2. The synthesis of VxXXB NTD(6–18) using 5,5′-dithiobis-2-nitrobenzoic acid (DTNB) and PdCl₂ oxidation. (A) Synthesis scheme of anti-parallel heterodimeric VxXXB NTD(6–18) with DTNB and PdCl₂. (B) RP-HPLC chromatograms of subsequent oxidation step (1–3) for the oxidation of folded VxXXB NTD(6–18) chain A-chain B heterodimer with a mass of 3037.65 Da (Fig. S1†). (C) Concentration–response curves for VxXXB NTD(1–18) and NTD(6–18) at α7 nAChRs on SH-SY5Y cells via FLIPR assay. Data represent means SEM of triplicate data from three independent experiments.

The NTD(1–18) and NTD(6–18) showed comparable inhibitory activity at human α7 nAChRs (p > 0.05), although significantly weaker than CTD(21–50) and CTD(19–50) (p < 0.05) (Fig. 2C and Table 1).

Synthesis of VxXXB CTD-NTD (CN) and VxXXB CNC variants

Two VxXXB variants comprising one CTD and either NTD(6–18) or NTD(1–18), VxXXB N(6–18)C and VxXXB N(1–18)C, respectively, were synthesized to obtain further insight into the binding domains contributing to the affinity of full-length homodimer VxXXB. Four ligation strategies were trialed as outlined in Fig. 3, Table S1† and below.

Enzyme ligation

Linear CTD(19–50) was modified by the addition of di-glycine (GG) to its N-terminal, CTD(19–50)-[1], while a SrtA5° recognition sequence (LPATGG) was added to the C-terminal NTD(6–18) chain A (Fig. 3 and S3 and Table S1†). The linear peptides were oxidized as previously described for their unmodified version. Successful folding was confirmed for both CTD(19–50)-[1] and NTD(6–18)-[1] by RP-HPLC and ESI-MS. Oxidized individual peptides were mixed to be enzymatically ligated with SrtA5° and monitored during 24 h incubation (Fig. 3). This ligation reaction failed to generate the desired ligated product (Fig. S3†).

Hydrazone ligation

Linear CTD(19–50) was introduced with an N-terminal Serine, VxXXB CTD(19–50)-[2]. To attach the hydrazide group to the C-terminal, NTD(6–18) chain A was assembled on a hydrazide-treated 2-chlorotrityl (CTC) resin, NTD(6–18)-[2] (Fig. 3 and S4 and Table S1†). The linear peptides were oxidized as previously described for their unmodified version. Successful oxidation folding generated one major product for each reaction, with CTD(19–50)-[2] and NTD(6–18)-[2] identified by RP-HPLC and ESI-MS. CTD(19–50)-[2] was then selectively converted into the aldehyde and mixed with NTD(6–18)-[2] in sodium citrate buffer (100 mM, pH 4.5) (Fig. 3). Monitoring the reaction for 24 h at −20 °C by RP-HPLC failed to identify the desired ligated product (Fig. 3 and S4†).

CuAAC ligation

Linear CTD(19–50) was incorporated with Prg to the N-terminal as the alkyne moiety employed for triazole formation, CTD(19–50)-[3]. The C-terminal of NTD(6–18) chain A was introduced with an Aza as the azide moieties for the CuAAC chemistry (Fig. 3 and S5 and Table S1†). The linear peptides were oxidized as previously described for their unmodified version. Successful oxidative folding was confirmed for both CTD(19–50)–[3] and NTD(6–18)-[3] by RP-HPLC and ESI-MS. NTD(6–18)-[3] was then conjugated to the alkyne moiety of the CTD(19–50)-[3] by CuAAC chemistry (Fig. 3). Again RP-HPLC analysis over 24 h failed to identify the desired ligated product (Fig. S5†).

KAHA ligation

An N-terminal peptide hydroxylamine group was introduced to the N-terminal of linear CTD(19–50) using (S)-5-oxaproline (Hse) to generate CTD(19–50)-[4], while C-terminal of NTD(6–18) chain A was synthesized on a Rink-amide resin preloaded with leucine ketoacid (LeuKA). These modifications allow the C-terminal α-KA of NTD(6–18)–[4] and N-terminal hydroxylamine of CTD(19–50)-[4] to react chemo-selectively to give an amide bond at the ligation site (Fig. 3, 4 and S6†). The linear peptides were oxidized as previously described for their unmodified versions. Successful folding was confirmed for both the CTD(19–50)-[4] and NTD(6–18)-[4] by RP-HPLC and ESI-MS. Individual oxidized peptides were reacted in 80% DMSO with 0.1 M oxalic acid and monitored for 24 h by RP-HPLC (Fig. 3). A peak with a mass corresponding to ligated VxXXB N(1–18)C/VxXXB N(6–18)C, plus Leu and homoserine (^hSer) introduced at the ligation point between the NTD and CTD, was detected at 3 h that remained constant during a 24 h incubation. Amide bond conversion at the ligation point was achieved following dilution of the reaction with ammonium bicarbonate buffer at pH 9.5 The final yield of ligated VxXXB N(6–18)C with a monoisotopic mass of 6756.90 Da was 10% (Fig. 4Ba and S5 and Table S1†). This method was subsequently used for the production of VxXXB N(1–18)C (or VxXXB NC) (monoisotopic mass of 7912.87 Da) with a yield of ∼10–12% (Fig. S5†).

Preparation of synthetic full-length VxXXB

The KAHA ligation strategy was then employed to produce the full-length VxXXB variant named VxXXB CN(1–18)C or VxXXB CNC. The C-terminals of both chains of NTD(1–18) were incorporated with LeuKA, NTD(1–18)-[5] to allow ligation of CTD(19–50)-[4] to both sides of the NTD(1–18) dimer. After 24 h, a peak corresponding to the ligated VxXXB CNC, with Leu and ^hSer introduced at both NTD-CTD ligation sites, was detected by ESI-MS and RP-HPLC (Fig. 4A, Bb and S7 and Table S1†). At this time, CTD(19–50)-[4] was still present while NTD(1–18)-[5] was no longer detectable (Fig. 4B). Amide bond conversion at the ligation point was achieved following dilution of the reaction with ammonium bicarbonate buffer at pH 9.5. The final yield of ligated VxXXB CNC with an observed mass of 11 651 Da was ∼15–20% (Fig. 4Bb and S7†).

VxXXB N(6–18)C and VxXXB NC showed a comparable potency to CTD(19–50) (p > 0.05) and 2600-fold higher potency compared to NTD(1–18) and NTD(6–18) at human α7 nAChRs (p < 0.05). Interestingly, VxXXB CNC showed a 2-fold and 12.5-fold increase in potency compared to VxXXB NC at human α7 nAChRs and Ls-AChBP, respectively (p < 0.05) (Fig. 4C and Da and Table 1), suggesting VxXXB may require two CTDs to achieve full potency at α7 nAChRs and Ls-AChBP. Interestingly, synthetic VxXXB CNC was 30-fold and 70-fold less potent than previously reported for native VxXXB from electrophysiology at human α7 nAChRs and in binding assays at Ls-AChBP, respectively.⁹ Differences in potency between synthetic VxXXB and native VxXXB may arise from the –L^hS– inserted during ligation that might increase the flexibility of the full-length variant to slow on-rate or hinder optimal binding interactions to increase off-rate.

Saturation binding experiment measuring bound [³H]-epibatidine in the presence of 2 μM of VxXXB CNC revealed a non-surmountable (allosteric) binding mode (Fig. 4Db), supporting an allosteric binding mode outside the orthosteric binding site.

Homology model of native VxXXB and synthetic VxXXB

The potency difference (27.5-fold) between synthetic VxXXB CNC and native full-length VxXXB may arise from the additional residues, –L^hS– motif, introduced during ligation. To examine this possibility, models of both native VxXXB and synthetic VxXXB were built from the crystal structure of GeXXA using the online SWISS server and AlphaFold, a new neural network-based deep learning model powered by Google DeepMind. γ-Carboxyglutamic acids at position 3 and 5 of native VxXXB were replaced with glutamic acid, while homoserine at the ligation point of synthetic VxXXB CNC was replaced with serine. Unlike SWISS server, AlphaFold searches for homologous sequences with existing PDB structures to use as structural scaffolds. The CTDs of the synthetic VxXXB and native VxXXB models generated from both approaches were well superimposed with an RMSD of 0.3–0.4 Å. In contrast, when the NTDs between the synthetic and native VxXXB were aligned, those generated by AlphaFold showed a lower alignment (RMSD of 3.67) than models generated from SWISS server (RMSD of 2.94). Remarkably, the alignment between native and synthetic VxXXB generated from AlphaFold had an RMSD of 6.20 compared to only 2.098 from SWISS model (Fig. 5 and S8†). This significant deviation from the native structure likely arises from the inclusion of three additional structural PDBs (6KC5, 4LJO and 2M48) by the AlphaFold algorithm. Given these differences, the models obtained from SWISS server were chosen for further analysis.

Fig. 5. Model of native VxXXB and synthetic VxXXB (CNC) generated from the online SWISS model. The alignment between native VxXXB and VxXXB CNC viewed from two sides.

Discussion

αD-Conotoxins are antiparallel homodimeric structures that present a number of synthetic challenges to characterize their structure–activity relationships. To address these challenges, a ligation strategy for VxXXB was established, where each component of αD-conotoxin VxXXB was first synthesized individually, followed by the synthesis of the variant VxXXB NC and the full-length VxXXB CNC by native chemical ligation using the KAHA strategy.

Similar to GeXXA,¹³ VxXXB CTD(21–50) possesses three disulfide bonds that stabilize an ICK motif and inhibits human α7 nAChRs. Interestingly, full-length CTD(19–50) includes a fourth disulfide bond connecting the CTD with NTD showed 8-fold and 2-fold higher potency at both human α7 nAChRs and Ls-AChBP respectively, revealing that this bond makes an important contribution to VxXXB potency. We confirmed that both these CTD variants of αD-conotoxin produced non-surmountable inhibition of ligand binding at AChBP. These represent a minimal allosteric pharmacophore that likely overlaps the VxXXB binding site. In contrast, full-length NTD(1–18) displayed comparable potency to truncated NTD(6–18) at human α7 nAChRs, indicating that these five additional C-terminal residues contribute little to the binding interface, at least in the absence of the γ-carboxylglutamates, which were replaced by Glu in this synthesis and are not found in GeXXA.¹³ The comparable potency between full-length NTD and truncated NTD was also observed with αD-GeXXA.^13,14 The reduced potency of NTD variants at α7 nAChRs, compared to the CTD suggests that the NTD plays a secondary role in determining the affinity of full-length VxXXB.

A number of synthetic strategies were utilized to join the NTD and CTD using different linkers and ligation chemistries.¹⁷ As each CTD of VxXXB has four disulfide bonds and the NTD is formed by two interchain disulfide bonds, we considered that the minimum requirement for the synthesis of VxXXB was for each domain to be oxidized prior to ligation to ensure correct disulfide bond connectivities are achieved. A requirement of this approach is to avoid reducing conditions that would potentially allow disulfide bond shuffling during the ligation reaction. Unfortunately, most popular chemical ligation strategies, including a variety of native chemical ligation methods, use reducing agents like tris(2-carboxyethyl)phosphine and thiol additives like 4-mercaptophenylacetic acid or sodium 2-mercaptoethanesulfonate, making these unsuitable for this synthetic approach to VxXXB.^18–20 Thus, we investigated enzyme (sortase),^21,22 hydrazone ligation,^23–26 CuAAC ligation^27,28 and KAHA ligation^16,29,30 approaches, which can form native amide bonds in the absence of reducing agents. Nevertheless, among these four methods, only KAHA ligation was successful in yielding the expected ligated VxXXB NC variants comprising one NTD and one CTD, with an overall yield of ∼10%. The KAHA ligation uses a C-terminal α-KA and N-terminal hydroxylamines, which react chemoselectively to form an amide bond at the ligation site (Fig. 3 and 4). This reaction proceeds at slightly elevated temperatures (60 °C) in DMSO without the formation of problematic by-products (Fig. 4B). The mildly acidic conditions (DMSO) might increase peptide solubility and minimise hydrolysis of sensitive functional groups.^16,31 Particularly, KAHA ligation and hydrazone ligation share a common mechanism where both avoid the addition of reagents to initiate ligation. However, the small hydrazide on the C-termini of NTD compared to the additional Leu in KAHA ligation may influence reaction efficiency near the disulfide bond at the C-terminal of NTD and the N-terminal of NTD. On the other hand, enzymatic ligation and CuAAC ligation occur in the presence of added catalysts. Although extra motifs were introduced at the C-terminal of NTD and N-terminal of CTD, –LPATGG– motif and di-Gly for enzyme ligation and Prg and Aza for CuAAc ligation, the disulfide bonds on both NTD and CTD are near the site of ligation and might hinder access to catalyzing reagents, the SrtA5° enzyme and CuSO₄·(H₂O)_5, thus reducing reaction efficiency (Fig. 3). Recent work characterizing enzymatic ligation approaches to chemical ligation by KAHA, revealed the latter efficiently ligated complex disulfide-rich peptides without disulfide bond rearrangement.¹⁷ Remarkably, KAHA ligated peptides had high serum stability, possibly owing to the unnatural amino acid ^hSer resisting the recognition of protease enzyme.¹⁷ Both the full length and truncated NTD variants had no significant effect on potency of the NC variant compared to CTD(19–50). Although the insertion of –L^hS– between the NTD and CTD may affect potency, this finding supports a secondary role of NTD in VxXXB potency.

Using a similar approach, full-length VxXXB was also successfully synthesized using KAHA ligation in a one-pot reaction (Fig. 4B), where N-terminal hydroxalymine of CTDs reacted chemo-selectively to both C-terminals α-KA of the antiparallel NTD. Interestingly, the full-length VxXXB showed a higher potency compared to VxXXB NC at α7 nAChRs and Ls-AChBP, suggesting VxXXB requires two CTDs to achieve full potency. VxXXB may exhibit a co-operative binding mechanism where two CTDs occupy two adjacent binding sites.

Nevertheless, despite also showing an allosteric binding mode similar to native VxXXB, synthetic full-length VxXXB CNC was less active than native VxXXB that was previously characterized via electrophysiology at human α7 nAChRs and in binding assays at Ls-AChBP.⁹ Differences in potency between synthetic VxXXB and native VxXXB (27.5-fold) may arise from the –L^hS– insertion during KAHA ligation. To examine this possibility, models of synthetic VxXXB and native VxXXB were constructed using SWISS server and AlphaFold. However, non-αD-conotoxin structural data was included in AlphaFold (see ESI†), which uses multiple sequence alignment (MSA) between the input amino acid sequences and related protein sequences from other living organisms to inspect the conservation, coevolution and phylogenetic relationships before converting them into constraints that determine how the proteins folds from these sequences.³² Both MSAs of native and synthetic VxXXB only showed the GeXXA sequence ranked close to VxXXB, followed by the sequence from Conus betulinus. In addition, AlphaFold searched a clustered version of the PDBs also searched against the input amino acid sequences, where the top 20 target templates according to E-value were ranked.³³ In contrast, the synthetic VxXXB sequence included three PDB structures from human and drosophila E3 ubiquitin-protein ligase protein (6KC5, 2M48 and 4LJO, respectively) with significantly lower E-values (3.099 × 10⁻³, 2.959 × 10⁻² and 7.795 × 10⁻², respectively) compared to GeXXA (4.093 × 10⁻²⁴). These additional PDB structures appeared to have introduced a distortion in the construction of the synthetic VxXXB sequence, resulting in a high RMSD compared to native VxXXB (Fig. S7†). Thus, only models constructed from the SWISS server were used for further characterisation. Synthetic VxXXB and native VxXXB models overlayed with an RMSD difference of 2.098 Å, with the –L^hS– insertion accommodated without significant clashes. This insert and the swap from γ-carboxyglutamate to glutamate might influence the probability of achieving the bound conformation (slowed on-rate) and/or its ability to optimally interact with the binding pocket (increased off-rate) compared to native VxXXB. Such differences plausibly explain the 27.5-fold drop in potency observed between native and synthetic VxXXB. Still, the nanomolar potency of synthetic VxXXB can underpin new studies into the structure function of homodimer αD-conotoxins.

Conclusions

We have successfully synthesized and characterized the CTD and NTD of VxXXB, an NTD-CTD variant, and a full-length homodimer VxXXB variant containing a –L^hS– insert at the site of ligation between the CTD and NTD. Similar to native and synthetic VxXXB, the CTD and NTD-CTD variants were allosteric inhibitors. The comparable potency of NTD-CTD and CTD(19–50) variants and the higher potency of full length synthetic VxXXB suggests that NTD directly contributes little to the overall potency of VxXXB but may facilitate the positioning of two CTDs for full potency. The successful synthesis of full-length homodimer VxXXB allows further studies on the structure–function relationship of VxXXB and the development and characterization of truncated allosteric inhibitors of the nAChR. Both VxXXB CTD variants displayed non-surmountable (allosteric) binding to AChBP, providing new avenues for the design of novel allosteric peptidic inhibitors that target nAChR receptor with high potency and subtype selectivity.

Author contributions

TH: study design, functional experiments on FLIPR, radioligand binding studies, data analysis and interpretation, peptide synthesis, HPLC, mass spectrometry, circular dichroism, homology modeling, and prepared the first draft of the manuscript. NA: study design, data analysis and interpretation. RL: study design, data analysis, interpretation, manuscript writing, funding and facilities. All authors reviewed the manuscript.

Conflicts of interest

The authors declare no competing financial interests.