Pharmacological characterisation of the highly NaV1.7 selective spider venom peptide Pn3a

Human genetic studies have implicated the voltage-gated sodium channel NaV1.7 as a therapeutic target for the treatment of pain. A novel peptide, μ-theraphotoxin-Pn3a, isolated from venom of the tarantula Pamphobeteus nigricolor, potently inhibits NaV1.7 (IC50 0.9 nM) with at least 40–1000-fold selectivity over all other NaV subtypes. Despite on-target activity in small-diameter dorsal root ganglia, spinal slices, and in a mouse model of pain induced by NaV1.7 activation, Pn3a alone displayed no analgesic activity in formalin-, carrageenan- or FCA-induced pain in rodents when administered systemically. A broad lack of analgesic activity was also found for the selective NaV1.7 inhibitors PF-04856264 and phlotoxin 1. However, when administered with subtherapeutic doses of opioids or the enkephalinase inhibitor thiorphan, these subtype-selective NaV1.7 inhibitors produced profound analgesia. Our results suggest that in these inflammatory models, acute administration of peripherally restricted NaV1.7 inhibitors can only produce analgesia when administered in combination with an opioid.

(PEPD) [5][6][7] . Both disorders are associated with redness, swelling and burning pain, limited to the extremities in IEM and to the rectal, ocular and mandibular areas in PEPD. Therefore, pharmacological inhibition of Na V 1.7 appears to be a promising therapeutic strategy for the treatment of pain. Developing analgesics with Na V 1.7 selectivity is essential, as activity at the skeletal muscle isoform Na V 1.4, the cardiac isoform Na V 1.5, and the neuronal isoforms Na V 1.1, Na V 1.2 and Na V 1.6 are likely to cause dose-limiting adverse effects [8][9][10] .
However, efforts to develop selective small molecule inhibitors have been hampered due to the high sequence identity (> 50%) between Na V subtypes, particularly in the pore forming segments (S5-S6) where local anaesthetics bind. Although some small molecule inhibitors of Na V 1.7 that target the voltage-sensing domain have been described, the in vivo efficacy of these compounds in inflammatory and neuropathic pain animal models has not been reported 11,12 . We therefore know surprisingly little about the therapeutic potential of highly selective Na V 1.7 inhibitors.
Spiders have evolved pharmacologically complex venoms dominated by disulfide-rich peptides, many of which act at ion channels to rapidly immobilise prey or deter predators 13,14 . Insect and vertebrate Na V channel share 55-60% sequence identity and consequently many spider venom-derived peptides act on mammalian Na V channels [15][16][17] . These venom peptides typically bind to the less conserved voltage-sensing domains, and hence they often achieve much better subtype selectivity than small molecules that bind to the pore region of the channel 14,18 .
Here, we report the isolation and characterization of μ -TRTX-Pn3a, a peptide isolated from venom of the South American tarantula Pamphobeteus nigricolor, and show that it has exquisite selectivity for Na V 1.7. Using Pn3a, as well as the small molecule Na V 1.7 inhibitor PF-04856264 and the spider venom peptide phlotoxin 1 11,19,20 , we assessed the therapeutic potential of selective Na V 1.7 inhibition in multiple in vivo rodent models of pain. Surprisingly, despite on-target activity in vivo, these selective Na V 1.7 inhibitors lacked analgesic activity in formalin-, carrageenan-and Freund's complete adjuvant-induced pain models when administered alone. However, supporting the recently described crucial contribution of opioids to the Na V 1.7 loss-of-function pain-free phenotype 21 , co-administration with a sub-effective dose of the opioids oxycodone or buprenorphine or the enkephalinase inhibitor thiorphan yielded analgesia, while co-administration with gabapentin had no effect. These results highlight the lack of broad analgesic efficacy in rodent models of pain with these peripherally restricted selective Na V 1.7 inhibitors and confirm the proposed analgesic synergy between selective Na V 1.7 inhibitors and opioids suggested by a recent study 21 .

Isolation of the novel spider venom peptide μ-TRTX-Pn3a from Pamphobeteus nigricolor.
Crude venom isolated from Pamphobeteus nigricolor (Fig. 1a) partially inhibited veratridine-induced membrane potential changes in HEK293 cells stably expressing rNa V 1.3, with activity-guided fractionation isolating this activity to a single peak with a retention time of 38 min (Fig. 1b). Matrix assisted laser desorption/ionization time-of-flight mas spectrometry (MALDI-TOF MS) of this fraction revealed two dominant masses corresponding to monoisotopic masses of 4210.5 and 4268.5 Da. N-terminal sequencing identified two novel 35-residue sequences differing only by the N-terminal amino acid (Fig. 1c) that we named μ -TRTX-Pn3a (hereafter Pn3a) and μ -TRTX-Pn3b based upon the revised nomenclature recently proposed for spider-venom peptides 22 . We subsequently chose to investigate Pn3a, as this was the major sequence identified by the N-terminal sequencing. Chemical synthesis of Pn3a produced the correctly folded product, as native and synthetic Pn3a co-eluted when assessed using analytical HPLC (data not shown). Synthetic Pn3a was used for all further experiments.
Pn3a is a potent and selective inhibitor of Na V 1.7. Detailed subtype selectivity characterization using whole cell electrophysiology in HEK293 cells revealed that Pn3a most potently inhibits hNa V 1.7 (pIC 50 9.06 ± 0.08 M), with 40-fold selectivity over hNa V 1.1 (pIC 50 7.43 ± 0.06 M), 100-fold selectivity over hNa V 1.2 (pIC 50 6.91 ± 0.07 M), hNa V 1.3 (pIC 50 6.68 ± 0.08 M), hNa V 1.4 (pIC 50 6.84 ± 0.08 M) and hNa V 1.6 (pIC 50 6.89 ± 0.05 M), and greater than 900-fold selectivity over the tetrodotoxin (TTX)-resistant subtypes hNa V 1.5 (pIC 50 6.10 ± 0.06 M), hNa V 1.8 (pIC 50 4.30 ± 0.09 M), and hNa V 1.9 (pIC 50 5.62 ± 0.11 M), making Pn3a one of the most selective Na V 1.7 inhibitors reported to date (Fig. 2a,b and see Table 1 for IC 50 values). Pn3a inhibited mouse Na V 1.7 (pIC 50 8.36 ± 0.11 M) and rat Na V 1.7 (pIC 50 8.83 ± 0.18 M) with similar potency to human Na V 1.7, which is important for pre-clinical in vivo studies performed in rodents (see Table 1 for IC 50 values). At a concentration that nearly abolished inward current at Na V 1.7, Pn3a (10 nM) had little effect on the current traces for other Na V isoforms, with the exception of enhancing inactivation at Na V 1.1 and slowing the inactivation time constant at Na V 1.8 (Fig. 2c). We also assessed the effects of Pn3a at selected off-targets, including voltage-gated potassium (K V ) channels, voltage gated calcium (Ca V ) channels and nicotinic acetylcholine receptors (nAChRs). Pn3a had no significant effect at rK V 2.1 up to 300 nM (see below) or on hCa V 1.2, hCa V 2.2, α 7 nAChR or α 3 nAChR up to 10 μ M (data not shown).
Pn3a is a gating modifier toxin that alters the voltage-dependence of Na V 1.7 activation. In general, spider venom-derived peptides bind to the voltage-sensing domains of Na V channels to modify gating associated with channel activation and inactivation 14 . We thus examined the effect of Pn3a on the current-voltage (I-V) relationship as well as the voltage-dependence of activation and steady-state fast inactivation on Na V 1.1-1.8 (Fig. 3). With Na V 1.7, Pn3a had a slow onset of effect, with time constants of block of 23 s, 167 s and 780 s for concentrations of 100 nM, 30 nM and 10 nM, respectively (Fig. 4a). Interestingly, this slow onset of effect was only observed for Na V 1.7 and is similar to that reported for the Na V 1.7 selective spider peptide ProTx-II 23 . Long application times required to reach steady-state inhibition precluded testing Pn3a at lower concentrations and we thus examined the effects of 100 nM Pn3a on Na V 1.1-1.8.
To investigate the effects of Pn3a on the kinetics of activation and fast inactivation, we measured time-to-peak and the time constants of fast inactivation at Na V 1.7. Pn3a (100 nM) significantly slowed the time-to-peak between − 10 mV and 15 mV (P < 0.05, two-way ANOVA; Fig. 4b) and voltage-dependently slowed the inactivation time constants between − 10 mV and 0 mV (P < 0.05, two-way ANOVA; Fig. 4c). Pn3a (100 nM) also significantly slowed recovery from fast inactivation at Na V 1.7 (control, τ = 3.98 ms; Pn3a, τ = 8.77 ms; P < 0.05; paired t-test;), confirming reduced availability of toxin-bound Na V 1.7 channels for re-activation suggested by the toxin effects on steady-state fast inactivation (Fig. 4d). In addition to fast inactivation, Na V 1.7 channels can inactivate by slow inactivation, a process that occurs over seconds (as opposed to milliseconds) and results in a different conformational state. Pn3a (100 nM) had no effect on the voltage-dependence of steady-state slow inactivation, causing no significant shift in the V 1/2 of slow inactivation (control, − 50.5 ± 1.4 mV; Pn3a, − 55.9 ± 2.9 mV; P > 0.05; paired t-test; Fig. 4e) and had no preference for the slow-inactivated state (data not shown). This is in contrast to the Na V 1.7 selective small molecule inhibitor PF-04856264 and clinical lead PF-05089771, which both preferentially bind to the slow inactivated state 11,24 .
An important biophysical property of Na V 1.7 is the ability to produce currents in response to small, slow depolarization ramps 25 . Pn3a (100 nM) completely inhibited Na V 1.7 ramp currents elicited by a 50-ms depolarization from − 100 to + 20 mV (2.4 mV/ms) (Fig. 4f), indicating that it should inhibit Na V 1.7-mediated generator currents at sensory nerve endings.
Pn3a binds to DII and DIV of Na V 1.7. To determine the Pn3a binding site on hNa V 1.7, we assessed the effects of Pn3a on hNa V 1.7/K V 2.1 channel chimeras, where the S3-S4 linker from each domain of hNa V 1.7 is inserted into the homotetrameric K V 2.1 channel 26 . Pn3a (300 nM) had no effect on wild-type K V 2.1, but inhibited potassium currents in the DII hNa V 1.7/K V 2.1 and DIV hNa V 1.7/K V 2.1 chimeras, indicating that Pn3a interacts with the S3-S4 linkers in both of these domains (Fig. 4g). This is consistent with the demonstration that interaction of Na V channel toxins with any of the voltage-sensing domains I-III affect channel activation, whereas inactivation is only affected by toxins that bind exclusively to voltage-sensing domain IV 27 . Moreover, despite interactions with the DIV voltage-sensing domain, Pn3a potency was not significantly reduced in the hNa V 1.7 M123 mutant (Y1537S/W1538R/D1586E) expressed in HEK293 cells (data not shown), indicating that Pn3a does not overlap with the binding site of the Na V 1.7 selective small molecule inhibitor PF-04856264 11 .
Solution structure of Pn3a. The three-dimensional solution structure of Pn3a was determined using 2D homonuclear NMR spectroscopy. The NMR data were of excellent quality with sharp and well-dispersed resonances, consistent with a single well-ordered structure in solution. NMR-derived interproton distances, hydrogen bonds and dihedral angles were used for structure calculations with simulated annealing and subsequent water minimization 28 . The ensemble of 20 conformations representing the solution structure of Pn3a is shown in Fig. 5a (Protein Data Bank code 5T4R) and the structural statistics presented in Supplementary Table 1. Pn3a forms an inhibitor cysteine knot (ICK) fold that is common in spider-venom peptides. The ICK motif provides rigidity to the peptide, with the only backbone flexibility observed at the termini (Fig. 5a). An anti-parallel β -sheet is formed between β -strands I (K19-K22) and II (R27-R27) that are connected by four tight turns (Fig. 5b). The β -sheet Pn3a most potently inhibited Na V 1.7, with 40-fold selectivity over hNa V 1.1, 100-fold selectivity over hNa V 1.2, 1.3, 1.4 and 1.6, and 900-fold selectivity over Na V 1.5, Na V 1.8, and Na V 1.9. Data are presented as mean ± SEM, with n = 3-9 cells per data point. (c) Representative hNa V 1.1-1.9 current traces before (black) and after addition of Pn3a (red). Currents were obtained by a 20 ms pulse of − 20 mV for Na V 1.1-1.7, a 20 ms pulse + 10 mV for Na V 1.8, and a 40 ms pulse of − 40 mV for Na V 1.9. Pn3a (10 nM) selectively inhibited peak current at hNa V 1.7 only.
Administration of selective Na V 1.7 inhibitors alone does not produce analgesia. To assess the analgesic potential of selective Na V 1.7 inhibition in vivo, we examined the effects of Pn3a, as well as the Na V 1.7-selective small molecule inhibitor PF-04856264, in multiple rodent models of pain. To assess on-target activity at Na V 1.7 in vivo, we assessed analgesic efficacy of Pn3a in a mouse model of Na V 1.7-mediated pain based on intraplantar injection of the scorpion toxin OD1, which impairs inactivation and enhances persistent current produced by Na V 1.7 19 . Pn3a administered by intraperitoneal injection dose-dependently reduced OD1-induced spontaneous pain behaviours in the absence of adverse effects (pain behaviours/10 min: control, 112 ± 8; Pn3a (0.3 mg/kg), 76 ± 19; Pn3a (1 mg/kg), 37 ± 7; Pn3a (3 mg/kg), 22 ± 9; P < 0.05, one-way ANOVA; Fig. 7a), confirming on-target Na V 1.7 activity in vivo. At the highest dose tested (3 mg/kg), analgesic activity persisted for at least 40 min, confirming Pn3a had a duration of action suitable for assessment in other pain models (Fig. 7b). We did not test PF-04856264, as efficacy in the OD1 model at a dose of 30 mg/kg was previously reported 19 .
We next tested Pn3a and PF-04856264 in mouse models of pain where genetic deletion of Na V 1.7 attenuates or abolishes pain behaviours. Surprisingly, at doses that attenuated OD1-induced spontaneous pain, neither Pn3a or PF-04856264 reduced responses to noxious heat assessed by the hot plate test (time to withdrawal: control, 25.9 ± 1.8 s; Pn3a (3 mg/kg i.p.), 30.0 ± 3.1 s; PF-04856264 (30 mg/kg i.p.), 30.7 ± 3.8 s; P > 0.05, one-way ANOVA; Fig. 7c), despite noxious heat responses being abolished the Na V 1.7 Wnt knockout mouse 31 . In the formalin model, where genetic deletion of Na V 1.7 attenuates formalin-induced spontaneous behaviours in both Phase I and Phase II 31,32 , neither Pn3a (3 mg/kg i.p.) (Fig. 7d) or PF-04856264 (30 mg/kg ip.) (Figure 7e) significantly reduced Phase I or Phase II pain behaviours (P > 0.05, two-way ANOVA). Similar results were found with carrageenan-induced inflammation, where thermal allodynia is abolished in the Na V 1.7 Nav1.8 knockouts 33 , but Pn3a and PF-04856264 had no significant anti-allodynic effect on mechanical thresholds (paw withdrawal force; PWF: control, 1.2 ± 0. As Pn3a and PF-04856264 are unlikely to cross the blood-brain barrier 34 , we also assessed if intrathecal administration of Na V 1.7 inhibitors is required to elicit analgesia in Freund's Complete Adjuvant (FCA)-induced inflammation, where mechanical and thermal allodynia are abolished in Na V 1.7 Nav1.8 knockouts 33 . Intrathecal Pn3a (0.3 nmoles) failed to reverse FCA-induced mechanical allodynia (PWF: control, 0.7 ± 0.2 g; Pn3a, 0.6 ± 0.1 g;  Fig. 7j); however motor impairment was observed at a higher dose of 1 nmole (data not shown). Intrathecal PF-04856264 was not tested due to low potency at the rat Na V 1.7 channel 11 .
Selective Na V 1.7 inhibitors synergize with opioids to produce analgesia. Altered expression of endogenous opioid peptides was recently shown to contribute to the pain-free phenotype observed in humans and mice lacking Na V 1.7 21 . We therefore assessed the analgesic effects of the selective Na V 1.7 inhibitors Pn3a and PF-04856264 in combination with subtherapeutic doses of the opioid oxycodone. In the formalin model, administration of Pn3a with a subtherapeutic dose of oxycodone significantly reduced pain behaviours in Phase II only (pain behaviours: control, 227 ± 14; oxycodone (1 mg/kg i.p.), 231 ± 22; Pn3a (3 mg/kg i.p.), 185 ± 20; Pn3a (3 mg/kg i.p.) + oxycodone (1 mg/kg i.p.), 92 ± 15; P < 0.05, two-way ANOVA; Fig. 8a,b). Similar effects were seen in the carrageenan model, where administration of either Pn3a or Figure 3. Effect of μ-TRTX-Pn3a on the electrophysiological parameters of hNa V 1.1-1.8. I-V curves before (black triangles) and after addition of Pn3a (white triangles). Pn3a (100 nM) inhibited peak current at all Na V subtypes but caused a rightward shift in the I-V curve at Na V 1.7 only. G-V curves before (black circles) and after addition of Pn3a (white circles). Pn3a (100 nM) significantly shifted the V 1/2 of voltage-dependence of activation to a more depolarized potential at Na V 1.7 only (Δ + 21.3 mV). Voltage-dependence of steady-state fast inactivation curves before (black squares) and after addition of Pn3a (white squares). Pn3a (100 nM) caused small but significant shifts in the V 1/2 of steady-state fast inactivation at Na V 1.1, 1. To confirm the lack of efficacy wasn't specific to Pn3a, we assessed another previously characterized spider peptide Nav1.7 inhibitor phlotoxin 1 20 , and extended our assessment of opiate synergy to buprenorphine and Voltage-dependence of time-to-peak as a measure of activation kinetics at Na V 1.7. Pn3a (100 nM) delayed time to peak between −20 mV and 15 mV. (c) Voltage-dependence of fast inactivation time constants at Na V 1.7. Pn3a (100 nM) slowed the inactivation rate between −10 mV and 0 mV. (d) Rate of recovery from inactivation at Na V 1.7. Pn3a (100 nM) slowed the rate of recovery from inactivation. Data are fitted with single-exponential functions and the time constant of recovery (τ ) is indicated. (e) Steady-state slow inactivation at Na V 1.7. Pn3a (100 nM) had no significant effect on the voltage-dependence of steady-state slow inactivation. Data are presented as mean ± SEM, with n = 4-6 cells per data point. Statistical significance was determined using two-way ANOVA, *P < 0.05 compared to control. (f) Representative ramp current elicited by a 50 ms depolarization from −100 to + 20 mV at a rate of 2.4 mV/ms at Na V 1.7. Pn3a (100 nM) inhibited ramp currents. (g) Potassium currents before (black) and in the presence of Pn3a (coloured) elicited by depolarisations to 70 mV in hNa v 1.7/rK v 2.1 chimeras. Pn3a (300 nM) inhibited potassium currents in the DII hNa V 1.7/K V 2.1 and DIV hNa V 1.7/K V 2.1 chimeras. Scale bars: 50 ms (abscissa), 2 μ A (DI, DII, DIV, K V 2.1) and 1 μ A (DIII) (ordinate axis).

Discussion
Since loss-of-function mutations in SCN9A were first identified as the cause of CIP in 2006 4 , there has been wide interest in the development of selective Na V 1.7 inhibitors for treatment of pain 35 . However, due to high sequence homology between Na V 1.1-1.9, development of highly selective Na V 1.7 inhibitors has been challenging, making assessment of analgesic potential of truly selective Na V 1.7 inhibition difficult to ascertain. In the search for novel Na V inhibitors, we screened spider venoms, a known rich source of disulfide-containing peptides with potential for subtype-selective activity at Na V channels, and subsequently identified Pn3a as a novel Na V 1.7 subtype selective inhibitor.
Characterisation of Pn3a revealed it as one of the most potent and selective Na V 1.7 inhibitors reported to date, with 40-fold selectivity over Na V 1.1 and at least 100-fold selectivity over all other Na V subtypes. In comparison, ProTx-II, a spider peptide from the venom of Thrixopelma pruriens is at least 85-fold selective over Na V 1.2-Na V 1.8 23 while the aryl sulfonamide PF-04856264 binds to the slow inactivated state of Na V 1.7 with at least 65-fold selectivity and the clinical lead PF-05089771 is only at least 10-fold selective 19,24 . While the required level of subtype-selectivity is difficult to estimate, assuming equivalent tissue penetrance and on-target activity and Hill slope of inhibition close to unity, 100-fold selectivity should enable complete Na V 1.7 inhibition in the absence of marked effects on the function of other isoforms. High subtype-selectivity is important, since activity at other Na V isoforms, such as Na V 1.6, likely contributes to analgesic activity but also adverse effects [36][37][38] . Na V 1.7 is expressed on sensory nerve endings in the skin, where it is thought to have a major role in regulating the excitability of peripheral sensory neurons 39 . The biophysical properties contributing to this specialized function include the slow closed-state inactivation that prevents Na V 1.7 from inactivating during subthreshold depolarisations originating from transducer channels on sensory nerve endings in response to noxious stimuli. This enables Na V 1.7 to amplify subthreshold depolarizations, or generator potentials, by producing ramp currents, which in turn cause sufficient membrane depolarization to activate Na V 1.8, the major contributor to action potential electrogenesis in nociceptors 40,41 . Indeed, enhanced activation of Na V 1.7 at peripheral nerve terminals, whether through toxins like OD1, or due to altered biophysical properties arising from gain-of-function mutations, undoubtedly leads to enhanced afferent firing and pain [5][6][7]42 . Thus, Na V 1.7 at sensory nerve endings could be a key site of action for Na V 1.7 inhibitors. Indeed, at the highest dose tested, Pn3a was able to significantly reduce spontaneous pain behaviours in mice caused by intraplantar injection of the Na V 1.7 activator OD1, confirming on-target activity at Na V 1.7 on sensory nerve endings in vivo. Although pain behaviours were not completely abolished, they were reduced to a similar level to those seen in Na V 1.7 knockout mice, with residual pain responses likely attributable to Na V 1.6 19 .
In addition, Pn3a showed on-target activity on small-diameter DRG neurons, evidenced by inhibition of ~50% of TTX-sensitive current in both IB 4 + and IB 4 − neurons, while TTX− sensitive currents remained largely unaffected in large DRG neurons. This is consistent with global Na V 1.7 knockout mice, in which 37% of TTX-sensitive current is reduced in DRG neurons 32 . Interestingly, the majority of sodium current in small-diameter IB 4 + neurons was TTX resistant, likely attributable to Na V 1.9 43 . However, despite clear inhibition of Na V 1.7 in sensory neurons both in vitro and in vivo, neither Pn3a nor the aryl sulfonamide PF-04856264 produced analgesia in rodent models of acute nociception or inflammatory pain at doses sufficient to reduce OD1-induced spontaneous pain. While it is clear that enhanced activation of Na V 1.7 at peripheral nerve terminals is sufficient to drive pain 19 , this lack of efficacy in animal models suggests that acute pharmacological inhibition of Na V 1.7 at peripheral sensory nerve endings alone is not sufficient to elicit analgesia in these models. Five Na V subtypes are expressed on adult peripheral sensory neurons (Na V 1.1, 1.6, 1.7, 1.8, 1.9), with C-fibres predominantly expressing Na V 1.7 and the TTX-resistant subtypes Na V 1.8 and Na V 1.9 [44][45][46] . While it is thought that Na V 1.7 plays a major role in setting the threshold of activation, Na V 1.9 may also be involved in the production of ramp currents and amplification of  per group (f) Effect of phlotoxin 1 (50 μ g/kg i.p.) and buprenorphine (50 μ g/kg i.p.) alone or in combination on acute heat responses; n = 6 per group. Effect of Pn3a (3 mg/kg i.p.) and gabapentin (100 mg/kg i.p.) alone or in combination on (g) carrageenan-induced mechanical allodynia and (h) carrageenan-induced thermal allodynia in mice; n = 3-8 per group. Data are presented as mean ± SEM. Statistical significance was determined using one-way or two-way ANOVA with Dunnett's post-test as appropriate, *P < 0.05 compared to control. generator potentials 41 . Indeed, in global Na V 1.7 knockout mice, the receptive fields of C-fibres are still mechanically excitable 32 , suggesting that Na V 1.7 channels on sensory nerve endings are not solely responsible for regulating action potential generation in all nociceptive neurons.
The spider peptide ProTx-II is also reported to lack analgesic activity in a rodent model of inflammatory pain, despite reaching plasma levels sufficient to inhibit Na V 1.7 in vitro 23 . ProTx-II's lack of analgesic activity was attributed to an inability to cross the blood-nerve-barrier, a sheath of connective tissue layers that encompasses the axons of peripheral nerves, and thus an inability to inhibit action potential propagation 47 . However, while Na V 1.7 is expressed on axons, it is unlikely to have a major role in action potential propagation, given it has slow recovery from fast inactivation (repriming) compared to other Na V subtypes, such as Na V 1.6 and Na V 1.8, which can sustain repetitive firing 39,48,49 . Na V 1.7 is also expressed on the central projections of DRG neurons in the superficial laminae I and II of the spinal cord, localised on the pre-synaptic terminals, where it is thought to be important in the regulation of neurotransmitter release 24,31,39 . As the physicochemical properties of spider peptides generally preclude crossing of the blood-brain barrier, we assessed if Na V 1.7 inhibition at the level of the spinal cord is required to produce analgesia by administering Pn3a intrathecally. Despite reducing C-fibre evoked eEPSC amplitudes in lamina I neurons by 36%, intrathecal Pn3a at the highest tolerated dose had no analgesic activity in vivo. This therefore suggests that a greater level of eEPSC amplitude inhibition is required to translate to analgesia in vivo. While it is unclear what level of Na V 1.7 inhibition Pn3a achieved when delivered by the intrathecal route, it is clear it was active in vivo by the occurrence of motor adverse effects at the highest dose tested.
Pn3a inhibits Na V 1.7 by decreasing peak current and shifting the voltage-dependence of activation to more depolarized potentials; however, current is not completely inhibited at all membrane potentials. While most mutations associated with CIP lead to a complete loss of Na V 1.7 function, it is unclear what level of pharmacological inhibition is required to recapitulate a pain-free phenotype. Dynamic clamp experiments in small DRG neurons suggest that total loss of Nav1.7 should increase current threshold for single action potentials by about 40%, together with a marked decrease in ability to produce repetitive impulse activity 50 . Two mutations (W1775R and L1831X) have been identified in CIP patients that retain Na V 1.7 function, but cause a voltage-dependent depolarizing shift in channel activation, similar to Pn3a 51 . This suggests that a high level of, but not complete, inhibition is sufficient to cause analgesia. However, it should be noted that these mutations would be associated with substantial loss of Na V 1.7 current at central and peripheral sites. It is plausible that this is not achievable with peptidic inhibitors or small molecule inhibitors such as the aryl sulfonamides which penetrate the blood-brain barrier poorly. Thus, our data support evidence from genetic studies suggesting that substantial reduction of Na V 1.7 current is required for analgesia.
Although Pn3a was able to almost completely reverse OD1-induced pain behaviours, this may represent a specialised case, with the apparent analgesic effect arising from concentration-dependent normalisation of Na V 1.7 function. Consistent with this notion, intraperitoneal injection of Pn3a produced reversal of OD1-induced pain behaviours in a dose-dependent manner. In contrast, antinociception in naïve animals, or the inflammatory models assessed here, may require complete or near complete inhibition of Na V 1.7 at all peripheral and central sites. It thus remains to be determined whether incomplete Na V 1.7 inhibition will lead to partial analgesia in conditions such as IEM, where exaggerated pain develops from enhanced Na V 1.7 function.
Thus, the most likely explanation for the lack of efficacy of Pn3a remains the lack of sufficient inhibition of Na V 1.7, rather than a requirement for a concomitant opioid activation.
In this study, the selective Na V 1.7 inhibitors Pn3a, PF-04856264 and phlotoxin 1 were administered once as a single dose. It is unclear if repeated administration or chronic administration of Na V 1.7 inhibitors is required to elicit analgesia. While limited material and poor pharmacokinetics precluded testing Pn3a chronically, it is interesting to note that the pain-free phenotype develops over several days in inducible Na V 1.7 knockout mice 52 . While this could reflect slow turn-over of Na V 1.7, an alternative explanation could be that analgesia emerges in parallel with upregulation of the endogenous opioid precursor proenkephalin (Penk), an effect unlikely to be reproduced with single administration of a Na V 1.7 inhibitor with a relatively short duration of action 21 .
In further support of an important link between Na V 1.7 and opioids, we were able to produce analgesia by administering the selective Na V 1.7 inhibitors Pn3a, PF-04856264 and phlotoxin 1 in combination with a sub-effective dose of the clinically used opioids oxycodone or buprenorphine or the enkephalinase inhibitor thiorphan, but not with the calcium channel inhibitor gabapentin. While these data clearly demonstrate that analgesic synergy between Na V 1.7 inhibitors and opioids is not restricted to a single agent or single pharmacological class, the mechanisms underlying synergistic analgesia remain unclear. Interestingly, the opioid antagonist naloxone substantially reverses analgesia in both mice and human lacking functional Na V 1.7 21 . In the transcriptome of sensory neurons from Advillin-Cre Na V 1.7 knockout mice, no changes in endogenous opioid receptor transcripts were reported, only upregulation of Penk mRNA and Met-Enkephalin protein 21 . It therefore appears that Na V 1.7 inhibition in combination with opioid receptor activation, whether through upregulation of endogenous opioids, such as the enkephalins as demonstrated in the knockout experiments, or administration of exogenous opioids, as described here, can elicit profound analgesia 53 . Opioid receptors are expressed on peripheral sensory neurons, both on the peripheral and central projections, as well as along the axon, where they are involved in regulating neuronal excitability, action potential propagation, and neurotransmitter release through inhibition of calcium currents 54,55 . Thus, the synergistic activity of opioid receptor agonists with selective Na V 1.7 inhibitors may involve both central and/or peripheral opioid effects. Interestingly, analgesic synergy between oxycodone and Pn3a was not observed after intraplantar administration, suggesting that inhibition of Na V 1.7 at the receptive fields does not contribute to the synergistic effect observed after systemic dosing.
In summary, based on its exquisite selectivity, Pn3a is a useful pharmacological tool to probe the role of Na V 1.7 inhibition in pain. While efficacy in the OD1 model, as well as emerging positive clinical trial results with small molecule aryl sulfonamides 56 , suggest that Na V 1.7 inhibitors may be efficacious in IEM or PEPD, our results reveal that acute administration of peripherally restricted Na V 1.7 inhibitors alone may not have broad efficacy in inflammatory pain. The efficacy of peripherally restricted Na V 1.7 inhibitors in other pain types, such as neuropathic pain, remains to be determined, and underscores the need for further studies to determine how, or whether, Nav1.7 blockade will reduce pain in humans. Our results do highlight that the combination of Na V 1.7 inhibitors and opioids may provide a novel therapeutic approach for the treatment of pain.
Venom fractions were assessed for activity at rNa V 1.3 stably expressed in HEK293 cells using a FLIPR TETRA (Molecular Devices) membrane potential assay as previously described 58 . Active fractions were further fractionated to near-purity and peptide masses were determined using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS using a Model 4700 Proteomics Analyser (Applied Biosystems, Foster City, CA, USA) by spotting HPLC fractions with α -cyano-4-hydroxycinnamic acid (7 mg/mL in 50% ACN). Peptides were then reduced and alkylated and N-terminal sequence determination was performed by the Australian Proteome Analysis Facility (Macquarie University, NSW, Australia). were cultured in MEM containing 10% v/v FBS and selection antibiotics as recommended by the manufacturer. CHO cells stably expressing human Na V 1.8 in a tetracycline-inducible system (ChanTest, Cleveland, OH) were cultured in Ham's F-12 containing 10% (v/v) FBS and selection antibiotics as recommended by the manufacturer. To induce hNa V 1.8 expression, cells were cultured in the presence of tetracycline (1 μ g/ml) for 24 h at 27 °C. HEK293 cells expressing rNa V 1.3 59 or rNa V 1.7 48 were cultured in DMEM containing 10% v/v FBS and the selection antibiotic G-418 (0.5 mg/mL). SH-SY5Y human neuroblastoma cells were cultured in RPMI containing 15% FBS and supplemented with L-glutamine. Cells were grown in a humidified 5% CO 2 incubator at 37 °C, grown to 70-80% confluence, and passaged every 3-4 days using TrypLE Express (Invitrogen). HEK293 cells heterologously expressing human Na V 1.9, mouse Na V 1.7 or hNa V 1.7 M123 mutant (Y1537S/W1538R/D1586E) 11 were generated and cultured by Icagen Inc (Durham, NC, USA).

Calcium responses in SH-SY5Y cells.
To assess activity at Ca V 1.3, Ca V 2.2, α 3-containing and homopentameric α 7 nAChR endogenously expressed in SH-SY5Y human neuroblastoma cells, the effect of Pn3a on Ca 2+ responses was assessed using a FLIPR Tetra assay as described 60  Whole-cell patch-clamp electrophysiology. For hNa V 1.1-1.8 and rNa V 1.7, whole-cell patch-clamp experiments were performed on a QPatch-16 automated electrophysiology platform (Sophion Bioscience, Ballerup, Denmark) as described 61 . The extracellular solution contained in mM: NaCl 145, KCl 4, CaCl 2 2, MgCl 2 1, HEPES 10 and glucose 10; pH 7.4; osmolarity 305 mOsm. The intracellular solution contained in mM: CsF 140, EGTA/CsOH 1/5, HEPES 10 and NaCl 10; pH 7.3 with CsOH; osmolarity 320 mOsm. Pn3a was diluted in extracellular solution with 0.1% BSA at the concentrations stated. All Pn3a effects were compared to pre-toxin control parameters within the same cell, and incubation times were varied between 5 and 40 min depending on the time required for steady-state inhibition to be achieved.
Concentration-response curves were obtained with a holding potential of −80 mV followed by a pre-pulse of −120 mV for 100 ms and then a 20 ms test pulse of −20 mV for Na V 1.1−1.7 and + 10 mV for Na V 1.8 (repetition interval 20 s). I-V curves were obtained with a holding potential of −80 mV followed by a pre-pulse of −100 mV for 50 ms and a series of 50 ms step pulses that ranged from −80 to + 60 mV in 5-mV increments before returning to a holding potential of −80 mV (repetition interval 5 s). Conductance-voltage curves were obtained by calculating the conductance (G) at each voltage (V) using the equation G = I/(V − V rev ), where V rev is the reversal potential and were fitted with a Boltzmann equation. Voltage dependence of steady-state fast inactivation was measured using a series of 500 ms pre-pulses, ranging from −120 to −10 mV in 10-mV increments, followed by a 20 ms pulse of −20 mV for Na V 1.1-1.7 and + 10 mV for Na V 1.8 to assess the available non-inactivated channels (repetition interval 30 s). Time-to-peak was measured from pulse onset to maximal current and fast inactivation time constants were calculated by fitting current decay traces with a single exponential function using the I-V protocol described above. Recovery from fast inactivation was examined using a two-pulse protocol consisting of a depolarizing pulse to 0 mV for 200 ms to inactivate channels, followed by a step to −90 mV of variable duration (1-90 ms) to promote recovery, and a 20 ms test pulse to 0 mV to assess availability of channels. Voltage dependence of steady-state slow inactivation was measured using a series of 15 s pre-pulses, ranging from −100 to −20 mV in 10-mV increments, followed by a 50 ms step to −100 mV to remove fast inactivation, and a 50 ms test pulse to 0 mV to assess the available non-inactivated channels (repetition interval 30 s). Ramp currents were evoked by a depolarization from a holding potential of −100 to + 20 mV at a rate of 2.4 mV/ms.
hNa v 1.7/rK v 2.1 chimeras and electrophysiology. hNa v 1.7/rK v 2.1 chimeras containing the S3-S4 paddle regions of the four domains of hNa v 1.7 were described previously 26  Following resonance assignments the NOESY spectra were assigned automatically using CYANA, and preliminary structures calculated using torsion angle dynamics 62 . Backbone and side chain dihedral angles restraints were determined from TALOS-N 63 . 1 H-1 H TOCSY spectra were recorded at temperatures of 15-35 °C to calculate amide temperature coefficients. Backbone amide protons that showed both slow solvent exchange in 100% D 2 O and had a chemical shift temperature dependence of more than −0.0046 parts per million (ppm)/K were considered to be definitively hydrogen bonded 64 , and included as restraints. Final structures were calculated using water refinement within the CNS program 28  DRG preparation and electrophysiology. Spinal level L3-L5 DRGs were removed from male Sprague-Dawley rats (aged 3-6 weeks) and placed in ice-cold HEPES-buffered saline (HBS) composed of (in mM): 154 NaCl, 2.5 KCl, 1.8 CaCl 2 , 1.5 MgCl 2 , 10 HEPES, and 10 glucose (pH 7.4, 330 ± 5 mOsm). DRG cells were then prepared as described previously 67 . To visualize IB4 binding, isolated DRG cells were pre-treated with 1 μ g/mL Alexa Fluor 488-conjugated Bandeiraea simplicifolia IB4 (Invitrogen) for 5 min at room temperature and washed with HBS for 5 min before fluorescence was examined on the inverted microscope (Olympus, IX50) used for patch-clamp recordings. Cell diameter was determined using a micrometer graticule: < 25 μ m was defined as small-diameter and > 25 μ m was defined as large-diameter.
Whole-cell patch clamp recordings were performed at room temperature (22-24 °C) using an EPC-9 patch-clamp amplifier and corresponding PULSE software from HEKA Electronik (Lambrecht/Pfalz, Germany) or an Axopatch ID amplifier (Axon Instruments, Foster City, CA, USA) using pCLAMP acquisition software (v. 5.5, Axon Instruments). Currents were sampled at 50 kHz. Patch pipettes were pulled from borosilicate glass (AM Systems, Everett, WA, USA) and pipette input resistance was 2-3 MΩ. The capacitance of individual cells ranged between 5 and 60 pF with a series resistance between 1 and 9 MΩ. Series resistance compensation of at least 80% was used in all experiments. Capacitance transients were compensated automatically using a built-in procedure of the HEKA amplifier. Leak current was subtracted online using a P/6 protocol. The liquid junction potential was 9 mV.
Carrageenan. Carrageenan (1% w/v) was administered by i.pl injection in mice as previously described 42 .
Behavioural assessment was performed 3 h post-injection. Pn3a (i.p. 3 mg/kg), PF-04856264 (i.p. 30 mg/kg), oxycodone (i.p. 0.67 mg/kg), gabapentin (i.p. 100 mg/kg) or the combination thereof, were administered 10 min prior to behavioural assessment. Mechanical thresholds were determined using electronic von Frey (MouseMet Electronic von Frey, TopCat Metrology, UK) as previously described 42 . Thermal thresholds were determined using the thermal probe test (MouseMet Thermal, TopCat Metrology) as described 42 . Freund's Complete Adjuvant. FCA was administered undiluted in a volume of 150 μ L by i.pl injection in rats. Behavioural assessment was performed 3 days post-injection of FCA. For intrathecal administration, a polyethylene lumbar catheter was inserted into the intrathecal space between vertebrae L4-L5. Pn3a (0.3 nmoles) was diluted in saline and administered by intrathecal injection in a volume of 10 μ L 30 min prior to behavioural assessment. Mechanical paw withdrawal thresholds were determined using manual von Frey with a series of calibrated filaments 0.4-15.1 g (Stoelting Co., Wood Dale, USA) using the up-down method 69 . Paw withdrawal latency to a thermal stimulus was determined using the Hargreaves test (Ugo Basile, Camerio VA, Italy). Motor co-ordination was assessed using an accelerating rotarod (Ugo Basile). Latency to fall was recorded in seconds, with a maximal cut off of 300 s. At the completion of experiments, catheter placement was checked by injection of lignocaine (2%, 10 μ L) and observation of rapid bilateral hind limb paralysis.
Parallel Rod Floor Test. Motor performance in mice was assessed using the Parallel Rod Floor Test and analysed using ANY-Maze software (Stoelting Co) as described 19 . Pn3a (3 mg/kg), alone or in combination with oxycodone (0.67 mg/kg), was administered by the i.p. route as described above 15 min prior to assessment of motor performance. The ataxia index was calculated by dividing the number of foot slips by the distance travelled (m).

Data analysis.
Data were plotted and analyzed using GraphPad Prism, version 6.0. For concentration-response curves, a four-parameter Hill equation with variable Hill coefficient was fitted to the data. IC 50 values are mean ± SEM of the negative logIC 50 (pIC 50 ). Statistical significance was defined as P < 0.05 and was determined by an paired t-test assuming equal variance, or one-way ANOVA with Dunnett's post-test, or two-way ANOVA with Dunnett's or Sidak's post-test, as appropriate. Data are presented as mean ± SEM.