Design, Synthesis, and Structure–Activity Relationships of Novel Peptide Derivatives of the Severe Acute Respiratory Syndrome-Coronavirus-2 Spike-Protein that Potently Inhibit Nicotinic Acetylcholine Receptors

Abstract

The spike-protein of SARS-CoV-2 has a distinctive amino-acid sequence (⁶⁸²RRARS⁶⁸⁶) that forms a cleavage site for the enzyme furin. Strikingly, the structure of the spike-protein loop containing the furin cleavage site bears substantial similarity to neurotoxin peptides found in the venoms of certain snakes and marine cone snails. Leveraging this relationship, we designed and synthesized disulfide-constrained peptides with amino-acid sequences corresponding to the furin cleavage-sites of wild-type (B.1 variant) SARS-CoV-2 or the Alpha, Delta, and Omicron variants. Remarkably, some of these peptides potently inhibited α7 and α9α10 nicotinic acetylcholine receptors (nAChR) with nM affinity and showed SARS-CoV-2 variant and nAChR subtype-dependent potencies. Nuclear magnetic resonance spectroscopy and molecular dynamics were used to rationalize structure–activity relationships between peptides and their cognate receptors. These findings delineate nAChR subtypes that can serve as high-affinity spike-protein targets in tissues central to COVID-19 pathophysiology and identify ligands and target receptors to inform the development of novel SARS-CoV-2 therapeutics.

Graphical Abstract

INTRODUCTION

The severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) caused the global COVID-19 pandemic. Camels, pangolins, bats, and other mammals are known reservoirs of β-coronaviruses that include middle-east respiratory syndrome-coronavirus (MERS-CoV), SARS-CoV-1, and the bat virus RaTG13. These viruses express a surface glycoprotein called the spike protein, and each is similar in sequence to that of SARS-CoV-2.¹ However, a key difference between SARS-CoV-2 and SARS-CoV-2-like viruses is the presence of an amino-acid sequence in the spike protein of the former that forms a potential recognition site (⁶⁸²RRAR↓S⁶⁸⁶) for the cleavage enzyme furin.² The absence of this sequence in other SARS-CoV-2-like viruses suggests that acquisition of this furin cleavage site (FCS) may have occurred during human-to-human transmission and might be involved in the pathogenicity of SARS-CoV-2.^3,4 Interestingly, the amino-acid sequence of the FCS and flanking residues shows striking similarity to neurotoxins found in the venom of elapid snakes and marine cone snails that inhibit nAChRs.^5,6 SARS-CoV-2 variants including Alpha (B.1.1.7), Delta (B.1.617.2), and Omicron (B.1.1.529) have different mutations in the FCS as well as residues flanking the ⁶⁸²RRARS⁶⁸⁶ sequence.⁷ Since the region containing the FCS does not overlap with the ligand-binding domain of the primary host target, the angiotensin converting enzyme-2 (ACE2),⁸ a distinct spike-protein target might be involved. Nicotinic acetylcholine receptors are one such proposed target.^6,9,10

Mammalian nAChRs are pentamers assembled from α2–α7, α9, α10, and β2–β4 subunits or α1, β1, δ, and ε/γ in the case of the subtype found at the neuromuscular junction.^11,12 α7, α9, and α10 nAChR subunits are expressed in the lining of the respiratory epithelium (the entry point for SARS-CoV-2).^4,13 Importantly, nAChRs composed of α7, α9, and α10 subunits are expressed by immune cells and modulate the release of inflammatory cytokines.^14,15 Recombinant spike-protein has been tested for inhibitory activity on α7 nAChRs with conflicting results.^16–18 One possible reason for this discrepancy is that commercial spike-protein sources generally mutate the FCS to avoid cleavage by furin and other serine proteases. Although this preserves the “receptor binding-domain” that targets ACE2, such mutations may have deleterious effects on potential nAChR potency. Therefore, we sought to develop peptide ligands that mimic the FCS and test them for activity on nAChRs postulated to be involved in COVID-19.

α-Conotoxins from the venom of the carnivorous marine snail genus Conus are multiply disulfide-bonded peptides with medicinal potential, one of which is an FDA-approved drug for chronic pain.¹⁹ α-Conotoxin selectivity for the various nAChR subtypes generally depends on residues presented by the second backbone-loop. Such an arrangement provides an opportunity to directly assess the binding affinity of sequences found in the FCS loop of SARS-CoV-2 spike-protein variants. Here, we used an α-conotoxin template and inserted sequences of the SARS-CoV-2 FCS into the second loop. The activity of these synthetic peptides (herein called conofurins) on nAChRs was assessed using two-electrode voltage-clamp (TEVC) electrophysiology of Xenopus oocytes that heterologously expressed human nAChRs. We found that conofurins with an intact FCS are potent inhibitors of human α7 and α9α10 nAChRs, whereas those that lack the naturally occurring amino-acid sequence of the FCS are not. Nuclear magnetic resonance (NMR) spectroscopy and molecular dynamics (MD) simulation studies were used to identify key ligand–receptor interactions. Our results suggest that during SARS-CoV-2 infection, the spike protein may interact with α7 and α9α10 nAChRs via residues of the FCS. α7, α9, and α10 nAChR subunits are highly expressed in airway epithelium,²⁰ the initial site of infection and viral replication for SARS-CoV-2.¹³ In addition, both nAChR subtypes are involved in immune cell-mediated cytokine release, and severe elevations in cytokine levels, sometimes referred to as the “cytokine storm”, can be a lethal complication of SARS-CoV-2 infection.²¹ Thus, conofurin-like drugs targeting these receptors have the potential to modify infectivity and alterations in immune-cell activity during COVID-19 disease.

RESULTS

FCS Structure of the SARS-CoV-2 Spike-Protein is Similar to that of α-Conotoxins.

The structure of the wild-type (WT) SARS-CoV-2 spike-protein trimer was previously determined by cryo-electron microscopy (cryo-EM) (Figure 1A).²² The FCS of the spike protein contains a group of poly basic amino-acids frequently polymorphic in SARS-CoV-2 variants. These variations affect the efficiency of viral entry. Despite the importance of the FCS, this portion of the spike protein has not been structurally well-defined. However, computational modeling suggests that the FCS and flanking residues form an extended loop that protrudes in a fashion that may facilitate interaction with host-cell receptors (Figure 1B).²³ We noted that the modeled FCS loop bears strong similarity to the structure of nAChR antagonists known as α-conotoxins (Figure 1C). Interestingly, this resemblance was even more striking when the amino-acid sequences of the FCS were substituted into the second backbone-loop of an α-conotoxin (Figure 1D), suggesting that an α-conotoxin template could be used to target spike-protein sequences to relevant nAChR subtypes.

Figure 1.

(A) Cartoon rendition of the SARS-CoV-2 spike-protein with residue Pro⁶⁸¹ of the FCS (circled) mutated to Arg to represent the Delta variant (PDB: 6XR8);^22,23 (B) enlarged image of the FCS. (C) Cartoon rendition of LvIA residues Asn⁹-Ile¹⁵ (PDB: 2MDQ).²⁴ (D) Cartoon rendition of conofurin-Delta residues Ser⁹-Ser¹⁵ generated by mutating residues of LvIA. All images were generated using PyMOL.²⁵

α-Conotoxin LvIA is a Suitable Template for Insertion of FCS Sequences.

Although nAChRs have been proposed as targets for the SARS-CoV-2 spike-protein, many in vitro tests of the spike protein or peptides containing FCS sequences have shown low potency or no activity.^16,18 We note, however, that there are several potential issues with previous studies. The first is that the tested recombinant spike-proteins have alterations in the FCS sequence to facilitate expression in mammalian cells. Such alterations, though beneficial for protein production, may critically alter nAChR binding affinity. Second, there are mutations in the FCS among SARS-CoV-2 variants that may impart substantially different levels of spike-protein activity. Thus, testing a single variant may underestimate or entirely miss potential activity in the family of SARS-CoV-2 variants. Finally, electrostatic charges present at the N- and C-termini of synthetic peptides, but not in uncleaved spike-protein, may alter activity. We utilized the α-conotoxin framework to present relevant FCS sequences as surrogates to address these issues and assess the interaction between FCS amino-acids and nAChRs. We, therefore, sought an α-conotoxin template that itself was devoid of activity on α7 and α9α10 nAChRs.

α-Conotoxin LvIA is a known antagonist of human α3β2 nAChRs but devoid of activity on rat α7.²⁴ However, the activity on human α7, α9, α10, and α9α10 nAChR subtypes was unknown. We used TEVC electrophysiology to assess LvIA for its potential to be used as a template to engineer peptides containing sequences of the SARS-CoV-2 FCS. LvIA was tested on human α7, α9, and α10 homomers as well as α9α10 heteromers and the results compared to those obtained for α3β2 nAChRs. Native LvIA potently inhibited α3β2 nAChRs but was >400-fold less active on all other subtypes tested (Figure 2A). The low potency of LvIA on the α7 subtype and receptors containing α9 and/or α10 subunits confirmed that LvIA could be used as template to assess potential gains in potency following insertion of FCS sequences. Therefore, we initially designed and synthesized analogs of LvIA containing the amino-acid sequences of the FCS found in WT (SPRRARS), Alpha (SHRRARS), Delta (SRRRARS), and Omicron (KSHRRAR) SARS-CoV-2 variants (Figure 2B).

Figure 2.

α-Conotoxin LvIA is a potent inhibitor of human α3β2 nAChRs but not of α7, α9, α10, or α9α10 subtypes. Xenopus oocytes expressing α3β2, α7, α9, α10, or α9α10 nAChRs were subjected to TEVC electrophysiology and the acetylcholine (ACh)-evoked responses assessed for inhibition by LvIA. LvIA inhibited ACh responses mediated by α3β2 nAChRs with an IC₅₀ value of 25.4 (24.5–26.4) nM (n = 4). By contrast, the potency of LvIA for homomeric α7, α9, α10, and α9α10 heteromers was substantially less (IC_50s > 10 μM). The responses in the presence of LvIA (10 μM) were 81 ± 8% (n = 5) for α7, 91 ± 6% (n = 4) for α9, 74 ± 12% (n = 5) for α10, and 103 ± 8% (n = 5) for α9α10 nAChRs. The error bars and (±) values indicate SD, and values in parentheses indicate the 95% CI of the IC₅₀ estimate. The (n) value indicates the number of oocytes assessed. (B) Sequence alignments of α-conotoxin LvIA and the proposed analogs containing amino-acids found in the WT, Alpha, Delta, and Omicron FCS (red).

Conofurin Peptides Containing Amino-Acids Found in the FCS of the SARS-CoV-2 Spike-Protein Inhibit Human α7 and α9α10 nAChRs.

Residues Asn⁹-Ile¹⁵, which comprise loop 2 of LvIA (Figure 2B), were substituted with the amino-acids found in the FCS of WT, Alpha, Delta, or Omicron variants and tested for activity on α7 nAChRs. In addition, we also tested the peptides on the related α9α10 subtype as both α7 and α9α10 nAChRs are present in lung and olfactory epithelial cells. Conofurin-WT inhibited α7 nAChRs with an IC₅₀ of 2.12 (1.72–2.64) μM, whereas conofurin-Alpha was ~5-fold less active (Figure 3A; Table 1). Conofurin-Omicron showed inhibitory activity like that of conofurin-WT. Conofurin-Delta, however, inhibited α7 nAChRs with high potency [IC₅₀ of 177 (144–220) nM]. Similar results were found for conofurin activity on α9α10 nAChRs but with some notable differences. Conofurins WT and Alpha showed similar potencies for α9α10 nAChRs (IC_50s ~ 1 μM) compared to the α7 subtype but had IC₅₀ values that were >10-fold higher than that of conofurin-Delta [IC₅₀ of 98.1 (76.3–126) nM] (Figure 3B). Interestingly, conofurin-Omicron also showed high potency [IC₅₀ of 133 (103–176) nM] for α9α10 nAChRs. Figure 3C compares the responses of α7 and α9α10 nAChRs in the presence of the various conofurins (10 μM). At the concentration used, conofurin-Delta was significantly more potent than the WT, Alpha, and Omicron variants for α7 nAChRs. Conofurins Delta, Alpha, and Omicron produced equal levels of inhibition but more than conofurin-WT.

Figure 3.

Conofurin peptides containing the FCS sequences of WT, Alpha, Delta, and Omicron SARS-CoV-2 variants are antagonists of human α7 and α9α10 nAChRs. Xenopus oocytes expressing α7 or α9α10 nAChRs were subjected to TEVC electrophysiology and the ACh-evoked responses assessed for inhibition by conofurins WT, Alpha, Delta, and Omicron. (A) Conofurin-Delta potently inhibited α7 nAChRs, whereas conofurins WT, Alpha, and Omicron were 10–65-fold less active. (B) Conofurins Delta and Omicron potently inhibited α9α10 nAChRs and, similar to the results found for α7 nAChRs, conofurins WT and Alpha were less potent on α9α10 nAChRs (see Table 1 for IC₅₀ values). (C) Responses obtained in the presence of the various conofurins (10 μM) were compared to those obtained for conofurin-Delta. Conofurin-Delta was significantly more potent at inhibiting α7 nAChRs than conofurins WT, Alpha, and Omicron; the responses in the presence of the conofurins were 1.4 ± 1.8% (n = 9), 16 ± 8% (n = 9), 46 ± 15% (n = 9), and 13 ± 7% (n = 6), respectively. Conofurin-Delta was also significantly more potent at inhibiting α9α10 nAChRs than conofurin-WT; the responses in the presence of the conofurins were 1.0 ± 0.9% (n = 9) and 10 ± 8% (n = 8), respectively. The responses in the presence of conofurin-Omicron were 0.7 ± 0.4% (n = 5) and no different than those in the presence of conofurin-Delta. The responses in the presence of conofurin-Alpha were 6 ± 6% (n = 9) of control values. Conofurin-Omicron was more potent at inhibiting α9α10 nAChRs than were conofurins WT (p < 0.01**) or Alpha (p < 0.05*), but not significantly different when the same comparisons were made for α7 nAChRs (p > 0.05). The error bars and (±) values indicate SD, and values in parentheses indicate the 95% CI of the IC₅₀ estimate. The (n) value indicates the number of oocytes assessed, and significance was determined using a Kruskal–Wallis test.

Table 1.

Sequences and Potencies of Conofurins Containing Amino-Acids Found in the FCSs of the SARS-CoV-2 WT, Alpha, Delta, and Omicron Variants^a

		α7 nAChRs		α9α10 nAChRs
peptide	amino-acid sequence	IC50 (nM)	n	IC50 (nM)	n
conofurin-WT	GCCSHPACSPRRARSC	2123 (1715–2643)	4	1123 (797–1609)	7
conofurin-Alpha	GCCSHPACSHRRARSC	10,100 (7229–15,950)	6	1286 (878–1882)	5
conofurin-Delta	GCCSHPACSRRRARSC	177 (144–220)	5	98.1 (76.3–126)	6
conofurin-Omicron	GCCSHPACKSHRRARC	1892 (1611–2230)	4	133 (103 – 176)	4
conofurin-WT-SPSRASS	GCCSHPACSPSRASSC	>10,000	5	>10,000	6
conofurin-Delta-SRSRASS	GCCSHPACSRSRASSC	5398 (3815–8079)	4	>10,000	5

^aValues in parentheses indicate the 95% confidence interval of the IC₅₀ estimate. The “n” values indicate the number of oocytes assessed. Underlined and bolded residues indicate substitutions that eliminate the FCS.

During these experiments, we observed that recovery of the ACh responses varied when the receptors were exposed to the different conofurins. Following peptide removal, conofurin-Delta reversed more slowly than did the WT, Alpha, or Omicron variants for α7 and even more so for α9α10 nAChRs. Figure 4 shows examples of current traces recorded from an oocyte expressing α7 nAChRs and exposed to conofurin-WT (A) and then conofurin-Delta (B). Panels (C) and (D) illustrate similar experiments performed using α9α10 nAChR expressing oocytes. These results indicate that the different mutations of the FCS may result in spike-protein interactions with α7 and α9α10 nAChRs that differ substantially among SARS-CoV-2 variants.

Figure 4.

Single amino-acid residue difference between the conofurin-WT and conofurin-Delta determines recovery kinetics. Xenopus oocytes expressing α7 or α9α10 nAChRs were subjected to TEVC electrophysiology and the recovery of the ACh-evoked responses from inhibition by conofurin-WT and conofurin-Delta determined. The oocytes were exposed to conofurins (10 μM) in a static bath for 5 min, and the responses were monitored for recovery during washout of the peptide. (A) ACh-evoked responses from α7 nAChRs in the presence of conofurin-WT were 30 ± 9% (n = 8) of control values, which recovered to 97 ± 8% after 15 min. (B) ACh-evoked responses in the presence of conofurin-Delta were 0.5 ± 0.2% (n = 5), which recovered to 99 ± 4% after 15 min of wash. (C) For α9α10 nAChRs, the ACh-evoked responses in the presence of conofurin-WT were 3 ± 2% (n = 5) of control values, which recovered to 99 ± 2% after 15 min of wash. (D) By contrast, the ACh-evoked responses in the presence of conofurin-Delta were 0.5 ± 0.7% (n = 5), which recovered to only 21 ± 14% after 15 min of wash. Recovery from conofurin-Delta was significantly less than recovery from conofurin-WT (p < 0.0001****). The (±) values indicate SD, and the (n) value indicates the number of oocytes assessed. Significance was determined using a Kruskal–Wallis test.

FCS Residues are Critical for Interaction with α7 and α9α10 nAChRs.

Several spike-protein constructs tested in previous studies have substitutions in the FCS sequence that are not found naturally in SARS-CoV-2 or its variants. These substitutions make the FCS resistant to cleavage by furin and other serine proteases, an advantage for recombinant expression in mammalian cells. We reasoned, however, that such substitutions might impair the interaction between the spike protein and nAChRs subtypes α7 and α9α10. To test this prediction, we synthesized two additional conofurin peptides containing the sequences SPSRASS and SRSRASS in loop 2 of LvIA. Conofurin-WT-SPSRASS and conofurin-Delta-SRSRASS are based on conofurin-WT and conofurin-Delta, respectively, and differ by a single Pro to Arg difference; Pro and Arg are the aminoacids found at position 681 of the SARS-CoV-2 WT and Delta FCSs, respectively (Table 1). Indeed, these conofurins without key FCS residues showed substantially reduced activity on α7 and α9α10 nAChRs (Table 1; Figure S5).

NMR Spectroscopy of Conofurin-Delta.

The β-hairpin loop that contains the FCS residues of the spike protein is not well-ordered, and to date, the three-dimensional (3D) structure of this sequence has not been elucidated through cryo-EM or crystallographic means.^22,26 The solution structures of short, linear peptides are also generally disordered. By contrast, the disulfide bonds in α-conotoxins constrain their structure such that the amino-acid residues of a given backbone loop can be presented to the nAChR in an ordered arrangement. α-Conotoxins are structurally well-defined and contain two backbone loops between pairs of nonadjacent Cys residues.²⁷ Due to the rigid conformation the disulfide bonds impart, the 3D structure of α-conotoxins and synthetic analogs may be determined by two-dimensional NMR. Thus, the structural similarity of the SARS-CoV-2 spike protein with that of α-conotoxins provided an opportunity to examine both the potential structure of the spike-protein FCS residues and, importantly, their possible interactions with nAChRs.

To examine the 3D structure of the FCS sequence predicted to interact with nAChRs, we compared the NMR-derived structure of conofurin-Delta with that of native α-conotoxin LvIA. Although the backbones of both peptides showed considerable overlap, several notable changes in the secondary αH shifts were observed when conofurin-Delta was compared to LvIA (Figure 5A). The largest changes (>0.2 ppm) occurred at positions in the second loop (Cys⁸-Cys¹⁶) of the peptide. More negative secondary αH shifts were observed in conofurin-Delta for residues Cys⁸, Ser⁹, and Ala¹³, which may indicate a stronger helix across the midsection of the molecule and a better-defined turn toward the C-terminal. Figure 5B shows the backbone superposition of the 20 lowest-energy structures of conofurin-Delta, and Figure C shows a ribbon diagram of conofurin-Delta (blue) overlaid with LvIA (gray). An α-helix is present from residues Pro⁶ to Arg¹¹ as well as two short helices at the N- and C-termini. A cartoon rendition of conofurin-Delta with the side chains of residues ⁹SRRRARS¹⁵ shown as sticks to illustrate the similarities between the structures of conofurin-Delta (Figure 5D) and the FCS (Figure 1).

Figure 5.

NMR spectroscopy indicated that residues of the SARS-CoV-2 Delta variant FCS form an α-helical structure when inserted into an α-conotoxin backbone. (A) Secondary αH shifts of LvIA and conofurin-Delta in aqueous solution at 290 K. The horizontal axis represents the sequence of LvIA with substitutions of the Delta variant as shown. (B) Backbone superposition of the 20 lowest energy structures of conofurin-Delta. (C) Ribbon diagram of conofurin-Delta (blue) overlaid with LvIA (gray, PDB: 2MDQ). An α-helix is present from residues Pro⁶-Arg¹¹, with two smaller helical structures at both end termini. (D) Cartoon rendition of conofurin-Delta with the side chains of residues Ser⁹-Ser¹⁵ shown as sticks; the partial helices at the N and C termini in D-E were omitted for visual simplicity. All structures in B–D are shown with the N- and C-termini from left to right, respectively. The cartoon in D was generated using PyMOL.²⁵ Statistical analyses for the conofurin-Delta structure are provided in Table S2.

MD Simulation Reveals Interactions between the FCS Amino-Acid Sequence of Conofurin-Delta and Those of α7 and α9α10 nAChRs.

Initial models of conofurins WT, Alpha, Delta, and Omicron were constructed in PyMOL (Figure S1A–D), and the NMR-determined structure of conofurin-Delta was consistent with these computational models. The interactions between conofurins and nAChR subtypes α7 and α9α10 were then examined through MD simulations. We used the cocrystal structure of PelA and the acetylcholine binding-protein (AChBP) from Aplysia californica (PDB: 5JME)²⁸ as the main template to build homology models for binding of conofurins to α7 and α9α10 nAChRs. We chose this structure because the homology of the conofurin binding-interface encompassed by the sequence of the first disulfide loop (²CCSHPAC⁸) and the principal face (+) of human α7 and α10 subunits with the PelA-AChBP cocrystal structure were highly conserved (sequence alignments of α7, α9, and α10 subunits are provided for reference in Figures S3 and S4). This conserved binding motif allowed us to focus on the interactions between the FCS amino-acid sequences in conofurins and specific nAChR residues. To examine the interaction between conofurins and α9α10 nAChRs, MD simulations were performed with the α10 subunit presenting the principal binding-interface (+) and the α9 subunit presenting the complementary interface (−).^29,30 Models were validated by superimposing MD simulations into homology models and confirming the viability of the binding interfaces, i.e., no significant clashes or observations of poor chemical compatibility. Critical pairwise contacts between residues of α7 nAChRs and conofurins WT and Delta are shown in Figure 6A,B. Figure 6C,D shows the pairwise interactions between α9α10 nAChRs and conofurins WT and Delta, respectively. Consistent with the affinity measurements determined by the electrophysiology experiments, conofurin-WT showed fewer contacts with the α7(+)α7(−) and α10(+)α9(−) interfaces than conofurin-Delta. We therefore measured the evolution of the root-mean-square deviation (RMSD) of the distance (Å) between key Arg residues of conofurin-Delta and α7(+)α7(−) as well as the α10(+)α9(−) binding interface (Figure 7). More stable H-bonds were observed between the Arg residues of conofurin-Delta and the α10(+)α9(−) interface than for the α7(+)α7(−) interface, again, consistent with the activity observed in the electrophysiology experiments.

Figure 6.

MD simulation determined intermolecular H-bond networks for residues of conofurins WT or Delta when bound to the α7(+)/α7(−) or α10(+)/α9(−) ligand-binding interfaces. (A) Arginine¹¹ of conofurin-WT (green) interacted with Glu¹⁹⁴ and Asp²⁶ of the α7(+) subunit (orange). Serine⁹ interacted with Gln⁵⁸ of the α7(−) subunit (blue). (B) Similar interactions were observed between conofurin-Delta (green) and α7(+)/α7(−) nAChRs, but additional contacts between Arg¹⁰ (Pro¹⁰ in conofurin-WT) and residues Trp¹⁵⁰ of the α7(+) subunit and Asn¹⁰⁸ of α7(−) subunit were observed. (C) Intermolecular hydrogen bond network for conofurin-WT when bound to the α10(+)/α9(−) ligand-binding interface. Arginine¹¹ of conofurin-WT interacted with Gln¹⁵⁷ and Glu¹⁹⁷ of the α10(+) subunit (orange). Serine⁹ interacted with Asp¹²⁴ of the α9(−) subunit (blue). (D) Arginine¹¹ of conofurin-Delta (green) made contact with Gln¹⁵⁷, Glu¹⁹⁷, and Tyr¹⁹⁹ of the α10(+) subunit. Serine⁹ interacted with Asp¹²⁴ of the α9(−) subunit. Arginine¹⁰ made contact with Trp¹⁵¹ of the α10(+) subunit and Asn¹¹² and Asp¹²⁴ of the α9(−) subunit. Lastly, Arg¹² interacted with the side chain of Glu¹⁹⁷ of the α10(+) subunit. The dotted lines indicate the interactions between donors and acceptors. Pairwise interactions are listed in Tables S3 and S4. All images were generated using PyMOL.²⁵

Figure 7.

H-bond network of conofurin-Delta’s Arg residues is more stable in α9α10 relative to α7 nAChRs during MD simulations. Evolution of the RMSD of the distance in Å (all Y-axes) as a function of time (ns; all X-axes) between the indicated Arg nitrogen (N) atoms of conofurin-Delta and residues of α7 (A) and α9α10 nAChRs (B) during five concatenated, independent MD simulations of 500 ns each where the first 250 ns of each simulation were discarded for equilibration. Only the most stable bonds (<3 Å) are shown for brevity. At the core of the interaction between conofurin-Delta and α7 nAChRs was Arg¹⁰ of the peptide that made two robust H-bonds with the backbone oxygens (O) of α7(+) Trp¹⁵⁰ and α7(−) Asn¹⁰⁸ (rows 1 and 2). For α9α10 nAChRs, the interactions with Arg¹⁰ included Asp¹²⁴ and Asn¹¹² of the α9(−) subunit (rows 1 and 2). A third less stable H-bond between Arg¹⁰ and the side chain Oδ of α7(−) Asn¹⁰⁸ was also observed (row 3), but this interaction (Arg¹⁰ and the Oδ of α9(−) Asn¹¹²) was more stable for α9α10 nAChRs (row 3). Arginine¹¹ made stable H-bonds with the Oε of a conserved Glu in both the α7(+) and α10(+) subunits (row 4). A second, but less stable, H-bond between Arg¹¹ and the Oδ of α7(+) Asp²⁶ was also observed (row 5). Arginine¹² made no productive H-bonds with α7 subunits and remained solvated. By contrast, the N atom of Arg¹² made stable H-bonds with the Oε of Glu¹⁹⁷ in the α10(+) subunit (row 5) (see Figure S3 for a comparison of conserved residues among the three subunits). Numbering of nicotinic subunit residues follows that of Figure S4. The “@” symbol indicates the atom involved in the H-bond interactions between conofurin-Delta and the indicated nAChR subunit.

DISCUSSION

Although a primary target of the SARS-CoV-2 spike-protein is ACE2,⁸ additional receptor targets are likely.^5,6,9,10 In this study, we synthesized novel peptides wherein the amino-acid sequences of the WT FCS or those of three major SARS-CoV-2 variants were inserted into the second loop of α-conotoxin LvIA (Figure 2). The native α-conotoxin template lacked activity on α7 and α9α10 nAChRs, and insertion of FCS sequences resulted in significant gains in potency in all instances. The magnitude of increased activity was dependent on the SARS-CoV-2 variant with a rank order of potency of Delta > Omicron ≈ WT > Alpha for α7 nAChRs and Delta ≈ Omicron > WT ≈ Alpha for α9α10 nAChRs (Table 1). Although these experiments do not definitively demonstrate that conofurins compete with ACh at the orthosteric ligand-binding site, MDs identified critical contacts between amino-acid residues of conofurin-Delta and the ACh-binding site of nAChR subtypes α7 and α9α10. We found that the presence of three Arg residues (Arg¹⁰-Arg¹²) in conofurin-Delta (corresponding to Delta variant residues Arg⁶⁸¹-Arg⁶⁸³) formed a H-bond network that enhanced peptide potency relative to native LvIA (Figures 6 and 7; Tables S3 and S4). The absence of these Arg aminoacids in key positions of conofurin-WT-SPSRASS rendered the peptide essentially inactive on both α7 and α9α10 nAChR subtypes (Figure S5), demonstrating the importance of the native FCS sequence for activity.

Most studies that examined SARS-CoV-2 spike-protein activity on nAChRs used commercially available, full-length spike-proteins and found only low potency or no activity. However, the current study clearly demonstrates that native FCS sequences enable potent inhibition of α7 as well as α9α10 nAChRs. The structure-activity relationship analysis provides a firm rationale for explaining previous low potency findings. Most recombinantly expressed spike-proteins have an altered FCS to prevent cleavage by endogenous furin expressed in cellular expression systems. Studies that failed to demonstrate spike-protein potency for nAChRs utilized constructs that, for artificial production purposes, lacked Arg residues in the FCS that we demonstrate are critical for potency. Thus, the current study indicates that spike proteins containing an SPSRASS sequence, found in commercially available constructs, rather than the native SXRRARS sequence would have substantially reduced or no activity. Therefore, studies showing low nAChR activity for FCS-containing peptides can be readily rationalized by our findings.

The 3D structure of conofurin-Delta, determined by NMR spectroscopy, showed a closely overlaying peptide backbone with α-conotoxin LvIA (Figure 5A–C). Therefore, the large gain in potency of conofurin-Delta, relative to LvIA, is likely due to the interactions between FCS amino-acid side chains (Figure 5D) and nAChR residues and not due to structural perturbation. It is of interest to consider that the increased potency of conofurin-Delta, relative to conofurin-WT, appears to follow the trend of worse disease severity associated with the SARS-CoV-2 Delta variant.^31,32 Furthermore, conofurin-Omicron showed substantial preference for α9α10 nAChRs over the α7 subtype suggesting that the different SARS-CoV-2 variants may interact with immune and respiratory systems differently depending on nAChR subtype expression patterns.

nAChRs are broadly expressed in olfactory epithelium, bronchia, lung, and epithelial layers of the respiratory tract.^20,33 These tissues likely constitute the first physiological systems that encounter respiratory viruses such as SARS-CoV-2. Thus, spike-protein may interact with nAChRs in these tissues and contribute to the pathophysiology of COVID-19. nAChRs are also expressed by a variety of immune cells including mononuclear phagocytes and multiple populations of lymphocytes,^15,34,35 and importantly, expression of the α7 and α9α10 nAChR subtypes by immune cells has been demonstrated in several studies.^36–38 Seminal reports investigating the role of nAChRs in immune-cell function have shown that α7, α9, and α10 nAChR subunits play an important role in modulating the release of numerous cytokines.^39–41 For a comprehensive review on the modulation of inflammatory cytokines in monocytic cells by α7 and α9α10 nAChRs, see Richter and Grau and references therein.¹⁵ Separately, although symptoms from acute infection of SARS-CoV-2 generally resolve in a period of weeks, postacute sequelae of COVID-19, or long-COVID, occurs in a substantial subset of individuals and affects multiple organ systems.⁴² The cause of long-COVID remains the subject of intense investigation, but it is notable that circulating, uncleaved inflammatory spike-protein has been detected in long-COVID patients up to 12 months after diagnosis.⁴³

Some of the current pharmacotherapies of COVID-19 include antivirals such as a combination of nirmatrelvir and ritonavir, molnupiravir, and corticosteroids such as dexamethasone and methylprednisolone. Novel drugs designed to prevent binding of SARS-CoV-2 to nAChRs could further reduce viral entry and/or minimize adverse interaction with immune cells. Indeed, a review of clinical studies indicated that current smokers have a reduced risk of infection.⁴⁴ That smoking might reduce the risks associated with COVID-19 would be surprising given that cigarette smokers are far more likely to develop other respiratory infections such as influenza,⁴⁵ but would be consistent with a receptor-mediated mechanism unique to SARS-CoV-2. Nevertheless, some studies have reported contrary results and suggest that smoking may, in fact, exacerbate COVID-19 disease.⁴⁶ Furthermore, cigarette smoking is a widespread and preventable cause of cardiac and respiratory illness and mortality worldwide, and great care must be taken in messaging potential positive benefits of nicotine. This fact has prompted concern that support for a nicotinic receptor-based treatment of COVID-19 might lead to unwise or even unscrupulous promotion and use of tobacco-related products.⁴⁷ However, nontobacco-based drug leads, such as those developed in this study could be adapted for therapeutic purposes, and such developments need not preclude vigorous antismoking efforts.

In summary, the present study suggests that during COVID-19, the SARS-CoV-2 spike-protein interacts with α7 and α9α10 nAChRs via FCS residues. The potency of this interaction is altered by single amino-acid differences ⁶⁸⁰SXRRARS⁶⁸⁶ corresponding to spike-protein residue 681 of SARS-CoV-2 variants. Lastly, this study indicates that the FCS sequence of the spike protein can be leveraged to develop nAChR targeted drugs.

EXPERIMENTAL SECTION

Materials.

Acetylcholine chloride was obtained from Tocris (Minneapolis, MN, USA). Strychnine hydrochloride, choline chloride, sodium chloride, potassium chloride, calcium chloride dihydrate, magnesium chloride hexahydrate, 4-(2-hydroxyethyl)-1-piperazinee-thanesulfonic acid (HEPES), sodium pyruvate, bovine serum albumin (BSA), and collagenase A were obtained from Sigma-Aldrich (St. Louis, MO, USA).

Peptide Synthesis.

Synthesis of LvIA was performed using Fmoc solid-phase synthesis techniques, and detailed methods are described elsewhere.⁴⁸ For conofurin synthesis, the peptides were constructed on a preloaded Fmoc-Rink Amide MBHA resin (substitution: 0.34–0.50 mmol/g; Peptides International Inc.; Louisville, KY, USA or NovaBiochem, MilliporeSigma, St. Louis, MO, USA) using the automated peptide synthesizer Apex 396 from AAPPTec (Louisville, KY, USA). All standard amino-acids were purchased from AAPPTec. Cysteine residues were orthogonally protected by trityl for Cys² and Cys⁸ and acetamidomethyl for Cys³ and Cys¹⁶. The peptides were synthesized at 50 μmol scale. Coupling activation was achieved with 1 equiv of 0.4 M benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate and 2 equiv of 2 M N,N-diisopropylethylamine in N-methyl-2-pyrrolidone as the solvent. For standard amino-acid coupling reactions, a 10-fold molar excess of aminoacid was used, and the reaction was carried out for 60 min at room temperature. For peptide sequences with three sequential Arg amino-acids, the third was doubly coupled. The peptides were removed from the resin by treatment with reagent K (trifluoracetic acid)/phenol/ethanedithiol/thioanisol/H₂O [9:0.75:0.25:0.5:0.5 (v/v)] for a minimum of 2.5 and up to 4.5 h at room temperature before being filtered and precipitated with cold methyl-tert-butyl ether. The peptides were purified by reverse-phase high performance liquid chromatography (RP-HPLC) using a preparative C18 column. The disulfide bonds were formed stepwise: the first disulfide bridge (Cys²-Cys⁸) was formed in the presence of 100 mM 2,2′-dithiodipyridine in 50 mM Tris–HCl (pH = 8).⁴⁹ The reaction proceeded for 30 min before being purified by RP-HPLC. Closure of the second disulfide-bridge (Cys³-Cys¹⁶) was accomplished by oxidation in acidic iodine (10 mM) solution for 10 min. The masses of the peptides were verified by electrospray mass spectrometry at the University of Utah Department of Chemistry (Salt Lake City, Utah, USA). The purity of the peptides was determined by RP-HPLC using a buffer gradient of 10–50% B60 (60% vol/vol acetonitrile, 40% vol/vol water, and 0.092% vol/vol trifluoracetic acid) at a rate of 1% per minute. All peptides had purity levels ≥95%. The retention time, purity, and masses for all conofurin peptides are provided in Table S1.

NMR Spectroscopy.

Conofurin-Delta (1.0 mg) was dissolved in 550 μL of 10% vol/vol D₂O/90% vol/vol H₂O (~pH 3), and spectra were recorded on a Bruker ADVANCE III 600 MHz spectrometer equipped with a cryoprobe. Data were collected at 290 K using TOCSY, NOESY, H–N HSQC, and H–C HSQC experiments. Spectra were acquired with mixing times of 80 ms (TOCSY) or 200 ms (NOESY) and 4096 data points in F2 and 512 in F1. Chemical shifts were referenced to internal 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) at 0 ppm. Spectra were processed with Topspin 3.5 (Bruker Biospin) and assigned using the program CcpNmr Analysis.⁵⁰ Structure calculations of conofurin-Delta were based upon distance restraints derived from NOESY spectra and on backbone dihedral angle restraints generated using TALOS+.⁵¹ A family of 20 lowest energy structures consistent with the experimental restraints was calculated using CYANA⁵² and assessed using Molprobity.⁵³ Experimental restraints and stereochemical quality assessment outcomes are provided in Table S2.

Oocyte Electrophysiology.

Protocols (no. 17–07020) for obtaining oocytes from Xenopus laevis frogs were approved by the University of Utah’s Institutional Animal Care and Use Committee. Frogs were purchased from Xenopus1 (Dexter, MI, USA) and maintained by university personnel in an AAALAC accredited facility. Oocytes were obtained from frogs anesthetized with 0.4 wt/vol Tricaine-S (Thermo Fisher Scientific, Waltham, MA, USA) and were sacrificed after removal of the ovarian lobes.

Methods for the preparation of cRNA constructs for expression of nAChRs in X. laevis oocytes have been previously described.²⁸ Briefly, clones for human α3, α7, and β2 were provided by J. Garrett (Cognetix Inc., Salt Lake City, UT, USA). The α9 and α10 clones were provided by L.R. Lustig (University of California San Francisco, San Francisco, CA, USA) and subsequently subcloned into an oocyte expression vector containing an alfalfa mosaic virus sequence as previously described.⁵⁴ Oocytes were injected with cRNA and maintained at 16 °C in frog saline composed of 96 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 100 U/ml penicillin, 100 μg/mL streptomycin, 100 μg/mL amikacin sulfate, 160 μg/mL sulfamethoxazole, 32 μg/mL trimethoprim, 2.5 mM sodium pyruvate, 2 mM Glutamax, 1 mg/mL BSA, and 5 mM HEPES; pH 7.4. Oocytes injected with cRNA for α9 alone were incubated in frog saline containing 5 mM choline, and those with cRNA for α10 alone were incubated with 20 μM strychnine to facilitate functional expression of α9 and α10 homomers, respectively, as previously described.⁵⁵

Stage IV–V oocytes were injected with cRNAs encoding cloned human nAChR subunits and subjected to TEVC electrophysiology 1–4 days after injection. For the assessment of peptide activity, the oocytes were continuously perfused by gravity at a rate of ~3 mL/min with control saline (96 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 1 mg/mL BSA, and 5 mM HEPES; pH 7.4). The perfusion system consisted of a series of solenoid valves (Cat. no. 161T031; NResearch Inc., Caldwell, NJ, USA) controlled by LabVIEW software (National Instruments, Austin TX, USA). The oocyte membranes were clamped at a holding potential of −70 mV using a Warner Instruments Oocyte Clamp OC-725C (Warner Instruments, Hamden CT, USA). The currents were filtered at 5 Hz (FIB1; Frequency Devices, Ottawa, IL, USA) and sampled at 50 Hz using a USB-6009 digital acquisition system (National Instruments). The concentrations of ACh used were 100 μM for α7, α9, and α9α10 subtypes and 3 mM for α10 homomers. The ACh pulses were applied at 60 s intervals for one second except for the α7 subtype, which was stimulated with 500 ms pulses. The oocytes were pulsed with ACh until a stable baseline-response was observed. The saline was then switched to one containing peptide, and the ACh were responses monitored for changes in amplitude. The ACh responses in the presence of peptide were normalized to the average of three responses in control saline. Peptides were applied in this manner for concentrations up to 1 μM. For concentrations >1 μM, peptides were applied in a static bath for 5 min and normalized to the ACh response after a 5 min bath application of control solution.

Molecular Modeling and MD Simulation.

Given the exact homology of the conofurin binding interface encompassed by the sequence around the first disulfide ²CCSHPAC⁸ and the human α7/α9/α10 principal binding-subunit (+) with the cocrystal of PelA and the AChBP from A. californica (PDB: 5JME),²⁸ we used this structure as the main template to build homology models of conofurin binding to nAChRs containing α7, α9, or α10 subunits which are highly conserved. Models were validated by known crystal structures including the conservation of contacts between α7, α9, and α10 subunits with the cryo-EM structure of α-bungarotoxin bound to the α7 nAChR.⁵⁶ The only variations occurred in the flexible Cys loop bound to α-bungarotoxin that adopts a different conformation relative to complexes of α-conotoxin RgIA and α9 (+);⁵⁷ these differences were also observed when the conofurins were modeled with the AChBP. We further superimposed MD simulations into homology models and confirmed the viability of the binding interfaces i.e., no significant clashes or poor chemical compatibility were observed.

The initial peptide structures were generated in PyMOL^25,58 with the disulfide bonds constrained. The MD simulations were run with pmemd.cuda.^58–60 from AMBER18 using AMBER ff14SB force field.⁶¹ We used tLeap binary (part of AMBER18) for solvating the structures in an octahedral TIP3P water box with a 12 Å distance from structure surface to the box edges and closeness parameter of 0.75 Å. The system was neutralized and solvated in a solution of 150 mM NaCl. Six independent simulations were carried out for each peptide equilibrating the system for 1 ns at NPT using Berstein barostat⁶² to keep constant pressure at 1 atm at 300 K, followed by 500 ns NPT production at 300 K, with nonbonded interaction cut off at 10 Å. Hydrogen bonds were constrained using the SHAKE algorithm and integration time-step at 2 fs. Analysis of our repeated MD simulations indicated that 150 ns was an adequate time for equilibration, and therefore the first 150 ns of all our runs was discarded from the analysis (250 ns was discarded for MD simulations of conofurin complexes). Clustering RMSD calculations were generated using CPPTRAJ software.⁶³ Five independent MD simulations of bound conofurins with α7(+)α7(−) and α10(+)α9(−) subunits were performed by constraining the backbone atoms (N, Cα, C, and O) of the subunits using a spring constant k_rest = 5 kcal mol⁻¹ Å⁻². No constraints were imposed on side chains nor the structures of the conofurins.

Statistical Analysis.

All statistical analyses were performed using Prism 10.2.1 (GraphPad Software, San Diego, CA, USA). The estimated IC₅₀ values for inhibition of ACh-evoked currents by the peptides were obtained by nonlinear regression using a four-parameter logistic equation and presented with the corresponding 95% CI to evaluate the precision of the IC₅₀ estimate. The error bars indicate the SD of the data points obtained at each concentration and are provided to assess variance of the data. Differences in ACh response values obtained in the presence of the conofurins were determined using a Kruskal–Wallis test; results were considered significant if p < 0.05*, p < 0.01**, p < 0.001***, or p < 0.0001****.

ABBREVIATIONS

ACh

acetylcholine

COVID-19

coronavirus disease-19

FCS

furin cleavage-site

MD

molecular dynamics

nAChR

nicotinic acetylcholine receptor

NMR

nuclear magnetic resonance

RP-HPLC

reverse-phase high performance liquid chromatography

RMSD

root-mean-square deviation

SARS-CoV-2

sudden acute respiratory syndrome-cornonavirus-2

TEVC

two-electrode voltage-clamp

Data Availability Statement

Authors will release the atomic coordinates and experimental data upon article publication.