Novel analgesic ω-conotoxins from the vermivorous cone snail Conus moncuri provide new insights into the evolution of conopeptides

Cone snails are a diverse group of predatory marine invertebrates that deploy remarkably complex venoms to rapidly paralyse worm, mollusc or fish prey. ω-Conotoxins are neurotoxic peptides from cone snail venoms that inhibit Cav2.2 voltage-gated calcium channel, demonstrating potential for pain management via intrathecal (IT) administration. Here, we isolated and characterized two novel ω-conotoxins, MoVIA and MoVIB from Conus moncuri, the first to be identified in vermivorous (worm-hunting) cone snails. MoVIA and MoVIB potently inhibited human Cav2.2 in fluorimetric assays and rat Cav2.2 in patch clamp studies, and both potently displaced radiolabeled ω-conotoxin GVIA (125I-GVIA) from human SH-SY5Y cells and fish brain membranes (IC50 2–9 pM). Intriguingly, an arginine at position 13 in MoVIA and MoVIB replaced the functionally critical tyrosine found in piscivorous ω-conotoxins. To investigate its role, we synthesized MoVIB-[R13Y] and MVIIA-[Y13R]. Interestingly, MVIIA-[Y13R] completely lost Cav2.2 activity and MoVIB-[R13Y] had reduced activity, indicating that Arg at position 13 was preferred in these vermivorous ω-conotoxins whereas tyrosine 13 is preferred in piscivorous ω-conotoxins. MoVIB reversed pain behavior in a rat neuropathic pain model, confirming that vermivorous cone snails are a new source of analgesic ω-conotoxins. Given vermivorous cone snails are ancestral to piscivorous species, our findings support the repurposing of defensive venom peptides in the evolution of piscivorous Conidae.

can be radiolabeled for binding assays 26 . MoVIA and MoVIB fully displaced 125 I-GVIA from SH-SY5Y cell membranes with similar potency (p < 0.001) to ω-conotoxins from piscivorous species (Fig. 4A,B, Table 3), suggesting they have overlapping binding site. Since MoVIA and MoVIB may have evolved for defense against fish predators, we determined if MoVIA and MoVIB could also displace 125 I-GVIA from fish brain membranes. Consistent with this hypothesis, MoVIA and MoVIB displaced 125 I-GVIA from fish brain with higher affinity than from human SH-SY5Y cell membranes (p < 0.05, two-way Anova) (Fig. 4, Table 3). Remarkably, MoVIA and MoVIB affinity for fish brain was significantly higher (p < 0.001) than ω-conotoxins from fish hunting species (Fig. 4B, Table 3).  Table 4). In contrast, both peptides were inactive (up to 30 µM) at endogenous human Ca v 1.3 and Ca v 3.1 (data not shown), indicating selectivity for hCa v 2.

MoVIA and MoVIB Selectively
In rat dorsal root ganglia (rDRG) neurons, high voltage-activated (HVA) N-and P/Q-type calcium currents (Ca v 2.2 and Ca v 2.1 channels, respectively) contribute a major components of the endogenous calcium channel current 27,28 . In order to gain insight into the analgesic potential of MoVIA and MoVIB, we assessed their function on depolarization-activated Ba 2+ currents in rat DRG neurons. MoVIA and MoVIB inhibited HVA currents in a concentration-dependent manner ( Fig. 5A-C, Table 4). MoVIA inhibited ~34% of the whole-cell Ba 2+ current that was not reversible following a 5-min washout, suggesting that these peptides have a slow off-rate. percentage of P/Q-type current expressed in DRG neurons 27,29 , suggesting MoVIA had little effect at the P/Q-type currents endogenously expressed in rDRG.

Role of Position 13 in ω-Conotoxins.
Tyrosine at position 13 (Y13) has previously been shown to be essential for high affinity interactions of piscivorous ω-conotoxins at Ca v 2.2 22,23,30,31  were not significantly different from the corresponding native peptides (Fig. 6A,B), indicating that these replacements did not perturb the overall structure or fold.

CKPOGSKCSOSMRDCCTT-CISYTKRCRKYY-
To quantify side effects of MoVIB we used a visual scoring method. MoVIB doses of 0.1-1 nmol produced effects on visual side effect scoring method that were significantly different from vehicle (p < 0.05; one way ANOVA) whereas the lower doses (0.01-0.03) produced no significant effect. The development of side effects was dose-dependent (ED 50 0.09 ± 0.05 nmol, Fig. 7B) and irreversible, lasting for up to 4 h (p < 0.0001; two-way ANOVA). The MoVIB therapeutic index, calculated as the ratio of the ED 50 s for the visual side effect score relative to mechanical PWT was 2.2, indicating a narrow safety window (Fig. 7C) though improved over the safety window for MVIIA (ED 50 of 0.04 nmol and a TI of 0.7) 32 .

Discussion
Cone snails have evolved different cabals comprising groups of venom peptides with complementary or synergistic pharmacology to facilitate prey capture [33][34][35] . Piscivorous cone snails use ω-conotoxins to inhibit vertebrate Ca v 2.2 as part of a 'motor cabal' that blocks neuromuscular transmission and immobilizes fish 16,35 . Recently, ω-conotoxins were hypothesized to have evolved originally for defence by ancestral vermivorous species and later repurposed for fish hunting 12 . In the search for potential ancestral ω-conotoxins in worm-hunting species, we isolated two potent vertebrate-active ω-conotoxins from the venom of the vermivorous Conus moncuri 18 , suggesting that ω-conotoxins could have indeed evolved originally for defence in ancestral worm hunting species.
All venom peptides, including ω-conotoxins, are subjected to evolutionary pressures that enhance ancestral activities where they provide a competitive advantage [10][11][12][13][14][15] . Consistent with this view, similar evolutionary pressures likely also account for the highly potent ω-conotoxins found in vermivorous species. The close relatedness of vermivorous to piscivorous ω-conotoxin signal and propeptide sequences suggest they have related evolutionary origins and target related species. Since vermivorous cone snails are ancestral to piscivorous species 36 , this sequence relatedness confirms the potential for vertebrate-active ω-conotoxins in ancestral vermivorous to have evolved originally for defence against fish and to be later repurposed for fish hunting in piscivorous Conidae. However, MoVIA and MoVIB have an Arg13 replacing the functionally critical Tyr13 found in potent Ca v 2.2 inhibitors from fish hunting species. Despite this important sequence difference, MoVIA and MoVIB potently displaced the prototypical ω-conotoxin GVIA from human SH-SY5Y cell membranes, indicating that vermivorous and piscivorous ω-conotoxins share an overlapping binding site. Interestingly, MoVIA and MoVIB displaced GVIA from fish brain membranes with even higher affinity and were more potent at fish calcium channels than GVIA and MVIIA from piscivorous Conidae. These data support MoVIA and MoVIB being positively selected to target Ca v 2 in predatory fish as part of a defensive strategy, reminiscent of the δ-conotoxins found in the lightning-strike cabal of fish hunters that are also used defensively by other vermivorous cone snail species 37 .
In rat DRG neurons, N-and P/Q-type calcium channel currents are the predominant HVA calcium currents 28   Residues with bad angles 0.00 ± 0.00

Violations from experimental restraints
Total NOE violations exceeding 0.3 Å 0 (highest 0.275) Total Dihedral violations exceeding 3.0°0 (highest 1.97) Table 2. Energies and structural statistics for the family 20 structures of MoVIB with the best overall MolProbity score.
is often found in Ca v 2.2-selective peptides 38,39 . The reduced activity of MoVIA and MoVIB observed in Ca v 2.2 functional vs. binding assays are well known for ω-conotoxins and have been attributed to the non-physiological conditions used to enhance ligand affinity in binding assays 31 . Alternatively, the presence of Ca v 2.2 auxiliary subunits in the functional assays vs. their potential absence in binding assays may contribute to differences in potency, given ω-conotoxins have reduced affinity in the presence of the auxiliary α2δ subunit 24,29,40,41 . To better understand the structure-function of these new vermivorous ω-conotoxins, we used NMR spectroscopy to determine the tertiary structure of MoVIB. Overall, the tertiary structure of MoVIB was similar to those of fish hunting ω-conotoxins, comprising six cysteines connected 1-4, 2-5, 3-6 to display four loops to match the VI/VII framework. MoVIB retained a number of residues in and around the ω-conotoxins pharmacophore 3 , including a significant number of positive (K2, K7, R13, K24, R25, R27 and K28) and hydrophobic (M12, I20, Y22,Y29, T30) residues. Comparing the NMR structures of MoVIB with those of other ω-conotoxins (Fig. 3) showed that Arg13 was oriented similarly to Tyr13 in MVIIA, GVIA and MVIIC, suggesting it may interact with a similar part of the pharmacophore perhaps also through hydrogen bonding interactions 22,30,31 .
Interestingly, MoVIA has an elongated C-terminus, with a Tyr30 and Asn31 in the last two positions, whereas the less potent MoVIB sequence finishes at Tyr30, indicating that Asn31 contributes to the higher affinity of MoVIA for Ca v 2.2. Indeed, Asn31 is located near loop 4 that contributes to the ω-conotoxins pharmacophore 42 , with the most potent ω-conotoxins having an amidated C-termini 22,24 and de-amidation reducing GVIA potency 31 . Therefore, it is conceivable that the side chain or backbone amide of Asn31 in MoVIA could mimic the contribution of the amidated C-terminus found in other ω-conotoxins. Despite Arg13 replacing the conserved Tyr13 critical for potent Ca v 2.2 inhibition in piscivorous ω-conotoxins 22,23 , the pharmacology of MoVIA and MoVIB closely resembles that of ω-conotoxins from fish-hunting cone snails, including those from C. consors, C. catus, C. fulmen, C. geographus, C. magus, C. radiatus, C. striatus and C. tulipa 22,23,38,43,44 . Indeed, MoVIA and/ or MoVIB may be ancestral to GVIIA and/or GVIIB, given Tyr13 is also replaced by Arg13 and both have similar elongated and non-amidated C-termini, albeit those from C. geographus are ~100-fold lower affinity than the highly homologous GVIA at mammalian Ca v 2.2 16,45 . Similarly, the MVIIA-[Y13R] analogue also showed ~100-fold decrease in binding affinity over MVIIA and no functional activity at mammalian Ca v 2.2. In contrast, the MoVIB-[R13Y] analogue failed to show enhanced binding affinity and potency to inhibit calcium influx in FLIPR assays and rat DRG neurons. Since NMR data indicates that MoVIB-[R13Y] and MoVIB have similar structures, it appears that Arg13 may be uniquely preferred to Tyr13 in vermivorous ω-conotoxins.
Consistent with Ca v 2.2 being a validated analgesic target, rat behavioural studies using the PNL model of neuropathic pain confirmed the MoVIB analgesic activity. Intrathecal injection of MoVIB accompanied side effects common to ω-conotoxins, including shaking, tail twitching and serpentine tail movement, indicating that MoVIB would likely need to be carefully titrated intrathecally in a clinical setting to manage dose-limiting side effects also seen for ω-conotoxins MVIIA and CVID 46 . Nonetheless, we calculated an apparently improved safety window for MoVIB compared to that published for MVIIA 32 , which is currently marketed. Future studies will investigate the activity and the therapeutic index of MoVIA and MoVIB in other clinically relevant models of pain.
In conclusion, we have discovered and pharmacologically characterized two novel ω-conotoxins from the venom of a vermivorous cone snail, Conus moncuri. ω-MoVIA and ω-MoVIB had highest affinity for fish Ca v s, suggesting they were positively selected for defense against predation in this worm-hunting species. Given vermivorous cone snails are ancestral to piscivorous species, this finding supports the repurposing of defensive venom peptides during the evolution of piscivorous Conidae, as proposed previously 17 . Alternatively, MoVIA and MoVIB may be an example of convergent evolution of distantly related cone snail toxins that target similar pharmacology in different organisms 47,48 , although the striking structural and sequence similarities suggest otherwise. Like two modestly potent piscivorous ω-conotoxins GVIIA and GVIIB, these vermivorous ω-conotoxins possess an Arg13 instead of the otherwise critical Tyr13, providing new insight into the structural features required for high-affinity interactions of ω-conotoxins at Ca v 2.2 channels.

Methods
Drugs and Chemicals. All drugs and chemicals were analytical reagent grade sourced from Sigma Aldrich, NSW, Australia, unless otherwise detailed throughout the text (in parentheses).

ω-Ctx
Human SH-SY5Y Membrane Fish brain Membrane      DNase-free kit (Qiagen, Hilden, Germany) to remove any genomic DNA contamination. RNA concentration was determined by absorbance measurements at 260 nm. RNA purity/integrity was assessed by analyzing the ratio 260/280 nm using a NanoDrop (Thermo Scientific, MA, USA).
The resulting cDNA was used as template in a polymerase chain reaction (PCR). Primers used on RACE PCR were designed based on previously published sequences from members of the O-superfamily of vermivorous Conidae 49 . The 3′ RACE first strand cDNA was synthesized from 1 μg total RNA using FirstChoice RLM-RACE kit (Ambion), following the manufacturer's instructions. Primer sequences were F1 = 5′-CATCGTCAAGATGAAACTGACGTG-3′ and R1 = 5′-CACAGGTATGGATGACTCAGG-3′. PCR reaction with 500 ng of cDNA as template was performed using FastStart Taq DNA polymerase (Roche, Basel, Switzerland), under the following cycling conditions: 95 °C for 4 min, followed by 40 cycles of 95 °C for 30 s, 58 °C for 30 s, and 72 °C for 1 min and a final elongation step at 72 °C for 7 min. PCR products were analysed and purified after separation on a 1% agarose gel, using a QIAquick Gel Extraction kit (Qiagen). Gel extracted PCR products were sequenced at the Australian Genome Research Facility (AGRF), using the forward and reverse primers F1 and R1, respectively. Sequencing data was transferred to Expasy Tools 50 for sequence translation and amino acid sequence prediction. Sequences from other conotoxins belonging to the O-superfamily were retrieved either from GenBank 51 or Conoserver 52 and compared with MoVIA. Sequence alignment was performed using Clustal W 53 and Jalview version 2.8 54 .
Chemical Synthesis of MoVIA and MoVIB. Synthesis of MoVIA, MoVIB and analogues were performed using in situ neutralization Boc-SPPS on a Boc-Asn-PAM or Boc-Tyr-PAM resin employing HBTU/DIEA activation, respectively, as described previously 55 . Assembled peptidyl-resin was cleaved with hydrogen fluoride for 1 h using p-cresol/p-thio-cresol scavenger (10%) and crude peptide precipitated from ether, filtered and lyophilized from acetonitrile/H 2 O. After HPLC clean up, pure reduced peptides (20 mg each/[0.2 mg/mL]) were oxidized at pH 7.8 in a solution of 0.3 M NH 4 OAc/0.3 M guanidine-HCl in the presence of GSH/GSSG (100:10 mol eq). Two major peptide isomers were obtained after RP-HPLC in quantities of 1-2 mg, corresponding to MoVIA and MoVIB.

Mass Spectrometry. LC-ESI-MS/MS was performed on synthetic MoVIA and MoVIB samples separated
on a ZORBAX 300SB-C18 (2.1 × 100 mm × 1.8 µm) column, eluted with a Shimadzu 30 series HPLC system at 400 μl/min, with a linear gradient from 1-80% over 25 min. The eluent was analysed on a tripleTOF 5600 mass spectrometer (ABSCIEX, MA, USA) with a quadruple TOF system and a DuoSpray ionisation system. The ion-spray voltage was set to 5300 V, with full scanning over 250 ms, followed by full scan product ion data obtained in the information dependant acquisition (IDA) mode over 20 × 50 ms. The mass range was set to 300-4000 (m/z) for TOF MS mode and 80-4000 (m/z) for full scan TOF MS/MS mode. Buffer A was 0.1% FA and buffer B was 90% acetonitrile/0.1% FA. All data analysis was performed using Analyst 1.6 (ABSCIEX).
2D NMR Spectroscopy and 3D Structure Calculations. MoVIB and analogue peptides samples were prepared at 2 mg/ml in 90% H 2 O/10% D 2 O or 100% D 2 O (pH 5.0) for Nuclear Magnetic Resonance (NMR) spectroscopy studies. Two dimensional (2D) homonuclear 1 H-1 H total correlation spectroscopy (TOCSY), nuclear Overhauser effect (NOESY) and exclusive correlation spectroscopy (ECOSY) datasets, and a 2D heteronuclear 1 H-13 C HSQC were recorded at 900 MHz on a Bruker Avance II spectrometer, equipped with a cryogenically cooled probe and processed using Topsin 3.0 (Bruker). Homonuclear data were recorded with 2048 data points in the direct dimension and 512 increments in the indirect dimension over a sweep-width of 12 ppm. The HSQC spectrum was recorded with an indirect dimension sweep-width of 106 ppm. Data analysis were performed using the Computer Aided Resonance Assignment (CARA) software 56 . Structural restraints derived from the NMR data included (i) Inter-proton distances derived from NOESY cross-peak intensities in spectra recorded in either H 2 O or D 2 O with a mixing time of 100 ms. (ii) Backbone dihedral angles (Phi and Psi) derived from a TALOS+ 57,58 analysis of Cα, Cβ, Hα and HN chemical shifts. (iii) Side chain dihedral angles (χ1) derived from analysis of 3 J HαHβ coupling constants and intra residual NOE patterns (iv) Hydrogen-bond restraints derived from amide exchange rates and analysis of preliminary structures. Nuclear Overhauser effect (NOE) cross peaks were manually picked and subsequently calibrated and assigned automatically using the automatic assignment and structure calculation module of CYANA 3.0 59 . For the final structures distance restraint lists from CYANA were used as input for simulated annealing and water minimization within CNS 60 , using protocols from the RECOORD database 61 , modified as described previously 62 . In the final round 50 structures were calculated, and the best 20 based on energies and quality of packing and geometry as judged by MOLPROBITY 63  Fluorimetric calcium imaging assays. For assay-guided fractionation and Ca v selectivity characterization of MoVIA, MoVIB and mutant peptides, we used a cell-based calcium-imaging fluorimetric assay and the FLIPR platform. Human neuroblastoma SH-SY5Y cells were exposed to saturating concentrations of the Ca v 1 inhibitor nifedipine (10 µM) to isolate endogenous Ca v 2.2 40 . For selectivity studies, we isolated Ca v 1.3 using saturating concentrations of Ca v 2.2 inhibitor CVID (3 µM). The small response remaining resistant was inhibited with saturating concentrations of the Ca v 3 inhibitor mibefradil (30 µM) 40 . Briefly, we incubated SH-SY5Y cells with the Ca 2+ dye Fluo 4 in the presence of Ca v inhibitors for 30 min. Test toxins or controls were then added to the cells and responses monitored for 10 min. KCl stimulation buffer was then added to stimulate Ca v channel opening and responses were recorded for an additional 5 min. Electrophysiological Recording from rat DRG neurons. Rat DRG neurons were enzymatically dissociated from 10-16 day old Wistar rats, as described previously 65 . We performed whole-cell patch clamp recording using a MultiClamp 700B Amplifier (Molecular Devices). Data was digitalized with a Digidata 1322 A (Molecular Devices), filtered at 10 kHz and sampled at 100 kHz using pClamp 9.2 software. External recording solution contained (in mM): 150 TEA-Cl, 2 BaCl 2 , 10 D-Glucose and 10 HEPES, adjusted to pH 7.4 with TEA-OH. The pipette electrodes had a final resistance of (1-3 MΩ), with an intracellular solution containing (mM): 140 CsCl, 1 MgCl 2 , 5 BAPTA and 10 HEPES adjusted to pH 7.2 with CsOH. High voltage-activated (HVA) calcium channel currents were recorded using Ba 2+ as the charge carrier and by measuring peak inward current amplitude elicited by 75 ms voltage steps to 0 mV from a holding potential of −70 mV. After current achieved steady state, we applied toxins to the physiological solution and plotted the peak inward current amplitude every 10 s. Series resistance were typically compensated at 70-80%, and leak and capacitance currents were subtracted using a −P/4 pulse protocol. We used selective Ca v 2.1 (P/Q-type) channel inhibitor Agatoxin-IVA (Abcam, Cambridge, United Kingdom) and CVIE, a selective Ca v 2.2 (N-type) channel inhibitor, to examine the selectivity profile of MoVIA and MoVIB in rDRG.
Radioligand Binding Assays. The cell membranes from SH-SY5Y and fish brain preparations and the radioligand binding assays using ω-peptide GVIA radiolabelled at Tyr 22 ([ 125 I]-GVIA, Perkin Elmer) were performed as described previously 26,40 . Partial Nerve Ligation-Induced Neuropathy in Rats. Animal  Tight partial nerve ligation of the sciatic nerve was carried out under isoflurane anaesthesia (1-3% in O 2 ), as previously described 32 . To determine the establishment of neuropathy, PWTs were monitored using von Frey hair filaments and the up-down paradigm 66 . Briefly, a blinded experimenter applied six times von Frey hair stimulus of equal intensity (range 0.4-15 g) to each rat hind paw (at intervals of several seconds from time 0-4 h) and recorded the responses to stimuli (flinching and licking of the hind paws) and side effects. The average of these values per animal served as the pain related score, with no response to 15 g von Frey hair indicating the maximum possible effect (MPE). A visual score of 0-3, following previously described method 32,46,67 , was used to determine the degree of side effect.
One week after PNL surgery rats that developed significant neuropathy received intrathecal long-term polyethylene lumbar catheters inserted between vertebrae L5 and L6. von Frey measurements were taken before to set baseline and after intrathecal treatment with vehicle or MoVIB. MoVIB (0.01-1 nmol, n = 3-6 animals/dose) was freshly dissolved in 0.9% saline and injected via the catheter in a volume of 10 µl, followed by 15 µl of 0.9% saline to wash the drug from the catheter dead space. Control animals received 0.9% saline injections of equal volume. Following intrathecal injections, we monitored rat behaviour at every hour over a 4-h period. After each experiment, we checked the correct placement of catheters by injecting lignocaine (2%) and observing rapid bilateral hind limb paralysis. Animals that had no paralysis or lost weight after the catheter surgery were excluded from the analysis. Statistical analysis. Sigmoidal concentration-response curves and IC 50 values were calculated using GraphPad Prism v5.0, following a nonlinear regression analysis with a four parameter (variable Hill slope) equation fitted to the functional data and a three parameter (Hill slope of −1) equation fitted to radioligand binding data. To calculate average peak current values we used Microsoft Excel version 12.2.0. All results were expressed as the mean ± standard error of the mean (SEM) determined from triplicate data from at least 3 independent experiments. Statistical significance was determined using analysis of variance (ANOVA), with statistical significance defined as p < 0.05, unless otherwise stated. We calculated the MPE values, from animal experiments applying the maximum pain threshold and maximum side effect score. ED 50 values were calculated for % MPE data using a four parameter Hill slope with a variable slope, and compared using one-way or two-way ANOVA. When ANOVA tests were significant, we made post-hoc comparisons between drug/treatment groups and vehicle at individual time points using Bonferroni or Turkey adjustment for multiple comparisons.

Data Availability
All data generated or analysed during this study are included in this published article.