Spider Knottin Pharmacology at Voltage-Gated Sodium Channels and Their Potential to Modulate Pain Pathways

Voltage-gated sodium channels (NaVs) are a key determinant of neuronal signalling. Neurotoxins from diverse taxa that selectively activate or inhibit NaV channels have helped unravel the role of NaV channels in diseases, including chronic pain. Spider venoms contain the most diverse array of inhibitor cystine knot (ICK) toxins (knottins). This review provides an overview on how spider knottins modulate NaV channels and describes the structural features and molecular determinants that influence their affinity and subtype selectivity. Genetic and functional evidence support a major involvement of NaV subtypes in various chronic pain conditions. The exquisite inhibitory properties of spider knottins over key NaV subtypes make them the best venom peptide leads for the development of novel analgesics to treat chronic pain.


Introduction
Spiders are considered the most speciose and successful terrestrial venomous predators [1]. They comprise 119 families, 4141 genera and 48,255 species at the time of writing [2], with over 150,000 species estimated to exist [3,4]. They form the seventh most diverse order Araneae and completely rely on predation [3,5]. Their venoms comprise highly evolved venom peptides that facilitate both predatory behaviour by killing or paralysing the prey and defence against predation [6]. Medically significant cases of spider envenomation are less common and usually associated with intrusion in the spider's natural habitat or threatening encounters [7]. Spider venoms are highly specialized in targeting the molecular receptors, especially the neuronal system, of insects to immobilize or kill their preys. However, their venoms also impart noxious effects to higher organisms such as mammals. The conserved structure and function of targeted receptors from evolutionarily distant prey species, such as insects and threatening species, including mammals, likely explains the ability of spider venom peptides to potently modulate human receptors [8,9].
Recent advances in analytical technologies have made comprehensive biochemical and functional investigations of spider venoms feasible [10]. In addition, high-throughput technologies, such as fluorescence imaging and automated electrophysiology, have sped up the screening and discovery sensors) are drawn into the membrane where their positively charged residues form ion pairs with negatively charged adjacent residues from S1, S2 and/or S3 segments [50]. However, when the negative membrane potential becomes more positive during membrane depolarization, the S4 segments of DI-DIII move outward, resulting in a conformational change that opens the channel pore followed by the outward movement of DIV S4 that inactivates (blocks) the channel pore intracellularly [50,51]. Channel inactivation is the third cardinal feature of Na V channels in addition to voltage sensing across the membrane and the selective Na + filter [52]. Structurally, the cytoplasmic DIII-DIV linker forms a hinge that facilitates the inactivation and is key to the fast inactivation mechanism. A cluster of hydrophobic amino acids (i.e., Ile, Phe, and Met (IFM motif)) function as a hydrophobic latch to stabilize the inactivated state, and mutations in these residues individually and together alter the kinetics of fast inactivation [53]. Besides this motif, residues in the S4-S5 linkers of DIII (e.g., A1329) and DIV (e.g., N1662) are crucial for the fast inactivation as they form the docking site for IFM motif [54]. 1   Four non-identical domains (DI-DIV) feature six neurotoxin receptor sites (Sites 1-6) and key residues contributing to the outer Na + ion selectivity filter (EEDD) and inner selectivity filter (DEKA). The connecting S5-S6 linker is called P-loop (P) which together with S5 and S6 segments from each domain contributes in forming a Na + ion selective channel pore. (B) Three-dimensional NMR structure of the Na V 1.7 channel (PDB 6J8G) [55]. Four voltage sensing domains (VSDs), DI (yellow), DII (blue), DIII (green), and DIV (orange), are shown with their corresponding pore-forming segments (S5 and S6) arranged to form the pore domain (PD) selective to Na + ions. The P-loop that contributes to forming the inner selectivity filter is coloured in red spheres (DEKA) and outer selectivity filter (EEDD) is coloured in purple. The S6 segments of all the four domains contribute to form the intracellular region of the pore. Site 3 (cyan) and Site 4 (pink) are the major binding sites for spider knottins. The β1 and β2 subunits which interact with DIII and DI, respectively, are highlighted in beige colour. (C) Schematic of the gating cycle of Na V channels. At polarized potentials, the DI-DIV S4 segments are drawn towards the intracellular side due to the positive gating charges to render the closed conformation (down state). Upon depolarization, the forces holding the down state are relieved and DI-DIII S4 segments are rapidly released extracellularly to open the S6 channel gate in the open conformation (up state). The DIV S4 moves up slowly compared to DI-DIII S4 and drives the fast inactivation, where the channel is occluded intracellularly by the Ile, Phe, and Met (IFM) motif. After cell repolarization, the channel returns to a closed (resting) state [56][57][58].
Although the Na V channel gating has multiple kinetic states [59,60], it can be simplified into three distinct physiological states, the resting (closed), open, and inactivated, which develop from the voltage-sensitive conformational changes that occur within the α-subunit ( Figure 1C) [57,61]. Toxins and drugs that interact with Na V channels often bind preferentially to one of these conformational states to alter Na + conductance or the gating properties of the channel. Six neurotoxin receptor sites (Sites 1-6) have been identified on the Na V channel. Site 1 neurotoxin physically occludes the channel pore, whereas neurotoxins acting at Sites 2-6 affect the gating mechanisms of the channel. Venom peptides target four neurotoxin receptor sites in human Na V channels, namely, Site 1 (e.g., µ-conotoxins), Site 3 (e.g., scorpion α-toxins, sea anemone toxins and spider toxins), Site 4 (e.g., scorpion β-toxins, spider toxins) and Site 6 (e.g., δ-conotoxins) [16,62]. Given their physico-chemical properties, it is not surprising that Na V modulatory venom peptides preferentially target the extracellular side of VSDII (Site 4) and VSDIV (Site 3) [14,63].

Spider-Venom ICK Peptides
Inhibitor cystine knot (ICK) peptides have a disulphide-rich structural motif that forms a "knot" that confers high structural, thermal and proteolytic stability, making them attractive starting points for structure-function studies and clinical lead development [64]. This structural motif comprises at least three disulphide bonds with connections between C1-C4, C2-C5 and C3-C6, where two disulphide bonds form a ring threaded by the third (C3-C6) disulphide bond to form the knot. These scaffolds were first described as "knottins" in 1980s [65] and later identified as a "cystine knot" in the crystal structure of nerve growth factor [66,67]. Pallaghy et al. [68], in 1994, coined the term "inhibitor cystine knot" to identify the cystine knot motif with a triple-stranded anti-parallel β-sheet topology. Craik et al. [69] further categorized these disulphide-rich structural motifs as (i) growth factor cystine knots (GFCKs), (ii) inhibitor cystine knots (ICKs) and (iii) cyclic cystine knots (CCKs). ICKs and CCKs have the same disulphide connectivity, but the disulphide connectivity in GFCK differs where C1-C2 threads the ring formed by C2-C4 and C3-C6. ICKs and CCKs are referred to as knottins and cyclotides, respectively [70]. Animal toxin cystine knots have an ICK structural motif [69].
Spider venoms are a rich source of disulphide-rich peptides, including knottins [4]. With few exceptions, such as the atracotoxins, they display the ICK features comprising three disulphide bonds ( Figure 2A). However, certain variations are observed within the β-sheet topology ( Figure 2B,C). Unlike the triple-stranded anti-parallel β-sheet topology defined in knottins, spider venom knottins typically comprise two β-strands with a few exceptions displaying a third strand at the N-terminal [71]. Besides the disulphide bridge connectivity, another conserved structural feature of spider venom knottins is the hydrophobic patch surrounded by charged amino acids on the toxin's surface [72,73] that contributes to potency and selectivity [74][75][76][77][78][79][80][81][82][83]. The UniProt database lists 747 entries on search term "spider venom ICK toxin" (23 September 2019) that are further categorized into their ion channel targets ( Figure 3) which shows approximately 43% of the thus far described ion channel-impairing spider knottins targeted Na V channel [84].   [83] demonstrating the cystine knot motif with three disulphide bridges. A ring structure made up of two disulphide bridges, C1-C4 and C2-C5 (yellow), and the intervening peptide backbone (pink) penetrated by a third disulphide bridge, C3-C6 (orange), to form a pseudo-knot. The three-disulphide bonds form four loops (pink, green and cyan). Below: The primary structure of HwTx-IV with three disulphide bridges and four loops.

Figure 3.
Number of spider venom knottins modulating ion channels. The data were collected from the UniProt database on 23 September 2019 using the following search descriptors: "voltage gated sodium channel impairing spider ICK toxin" for Na V channel targeting spider knottins, "voltage gated calcium channel impairing spider ICK toxin" for voltage-gated calcium (Ca V ) channel targeting spider knottins, "voltage gated potassium channel impairing spider ICK toxin" for voltage-gated potassium (K V ) channel targeting spider knottins, "TRP impairing spider ICK toxin" for transient receptor potential (TRP) channel targeting spider knottins, "ionotropic glutamate receptor impairing spider ICK toxin" for ionotropic glutamate receptor (GluR) targeting spider knottins, "calcium-activated potassium channel impairing spider ICK toxin" for calcium-activated potassium (KCa) channel targeting spider knottins, "mechanosensitive ion channel impairing spider ICK toxin" for mechanosensitive ion channel (MSIC) targeting spider knottins and "ASIC impairing spider ICK toxin" for acid-sensing ion channel (ASIC) targeting spider knottins [84].
The first discovery of Na V -modulating spider venom peptides was in the mid-1980s, when Fontana and Vital-Brazil [87] demonstrated that the crude venom of Phoneutria nigriventer is capable of activating Na V channels in muscle and nerve cells. Later in 1991, Rezende et al. [88] isolated three neurotoxic fractions (PhTx1, PhTx2 and PhTx3) from P. nigriventer venom, and the most toxic fraction (PhTx2) was later shown by Araújo et al. [89] in 1992 to inhibit the Na V channel inactivation. In 1989, Adams et al. [90] isolated µ-agatoxins (I-VI) from Agelenopsis aperta which became the first spider venom with disulphide-rich peptides (8 cysteines arranged into 4 disulphide bridges) targeting Na V channels to induce repetitive firing in the neurons. Since then, a journey of three decades of research on spider venom peptides modulating Na V channels resulted in the discovery of a number of venom peptides which have been applied into research on the mechanisms of Na V channel modulation and potential therapeutics. Based upon the level of sequence identity and inter-cysteine spacing, these Na V channel-targeting spider toxins were classified into 12 families (NaSpTx1-12) [15].

Pharmacology of NaSpTx
The spider knottins' interaction with Sites 3 and 4 displayed diverse pharmacological phenotypes (Tables 1 and 2). Broadly, they either (i) prevent channel opening in response to membrane depolarization by trapping the VSD II in the closed state; (ii) facilitate channel opening by trapping VSD II in open state or (iii) prevent channel inactivation by binding DIV S4 in the closed state to impair the movement of the inactivation gate [15,18,63]. Curiously, an integrated pharmacology of Site 3 and Site 4 indicates multiple binding site interactions by the same toxin [91,92]. For example, the toxic fraction PhTx2 from the venom of P. nigriventer not only prolonged the inactivation and deactivation (Site 3 phenotype) of the Na V channel but also shifted the voltage dependence of activation (Site 4 phenotype) and steady-state inactivation towards negative potentials (Site 3 phenotype) [89]. Later, PnTx2-6 alone confirmed this complex pharmacology was achieved by a single toxin [93]. Further evidence of dual pharmacological profile includes versutoxin (VTX) from Hadronyche versuta, which besides the classical Site 3 features, such as delaying the channel inactivation and shifting the voltage dependence of inactivation towards more negative potential, also displayed the classical Site 4 feature of reducing the maximum (peak) sodium current [94].
Another channel state modulated by spider knottins is channel inactivation [91,93,94,98,116]. For example, when tested over tetrodotoxin sensitive (TTX-S) Na V channels on rat dorsal root ganglia (DRG) neurons, HnTx-III and HnTx-IV from Selenocosmia hainana differed in their channel repriming kinetics. More specifically, HnTx-III delayed channel recovery from inactivation, whereas HnTx-IV had no effects [116]. Similarly, steady-state inactivation can also be differently modulated by spider knottins. For example, JzTx-35 (C. jingzhao), which selectively targeted hNa V 1.5 similarly to JzTx-III (C. jingzhao), differed by shifting the steady-state inactivation of the channel towards more hyperpolarized potentials [112]. The hyperpolarizing shift in the voltage dependence of steady-state inactivation stabilizes the channel in the inactivated state [128]. However, potent inhibitor toxins like HwTx-IV (Ornithoctonus huwena), ProTx-III (Thrixopelma pruriens) and HnTx-IV showed no effect on steady-state inactivation [80,114,129]. Likewise, the excitatory JzTx-I (C. jingzhao) delayed the channel inactivation of other Site 3-acting spider knottins including δ-ACTXs, but the peak current amplitude, I-V relationship and steady-state inactivation remained unaltered [101]. Such characteristics among the spider knottins provide novel paths to modulate Na V channels.
Spider knottins display distinct affinities and modes of action for insect and mammalian Na V channels. Magi 5 from Macrothele gigas interacted with Site 3 on the insect Na V channels and Site 4 on mammalian Na V channels to induce Na + influx [133]. These observations support the concept of common binding "hot spots" proposed by Winterfield and Swartz (2000) [134], where binding sites in the voltage-gated ion channels are not independent structural sites. Recently, PnTx4(5-5) also showed a distinct affinity and mode of action on insect and mammalian Na V channels [126]. On BgNa V from the cockroach Blatella germanica, PnTx4(5-5) strongly slowed channel inactivation (EC 50 213 nM) and increased current amplitude, while it inhibited sodium currents of the mammalian Na V 1.2-1.6 channels, with higher potency on Na V 1.3. This unique selectivity and species-dependent mode of action provides new insight into the molecular mechanisms of spider gating modifier toxins [126].
Cell background can also influence pharmacological profiles of spider knottins [116,117,129]. For example, hyperpolarizing shifts (10−11 mV) in voltage-dependence of steady-state inactivation imparted by HnTx-III and HnTx-IV in TTX-S Na + currents in rat DRG were not observed on heterologously expressed hNa V 1.7. Similarly, HnTX-III which delayed the recovery from inactivation of TTX-S Na + channels on rat DRG did not affect the repriming kinetics of heterologously expressed hNa V 1.7 [116,117]. Indeed, the difference in the relative proportion of Na V subtypes in different DRG cell types (small and large diameter) [117], species and age [117,135], β-auxiliary subunit combinations [136] and difference in the membrane lipid composition [82] likely influence the biophysical and pharmacological properties of Na V subtypes. The pharmacological and biophysical properties of Na V channels are also influenced by the expression system and species differences. For example, Df1a was approximately 8.5 fold less potent at hNa V 1.7 expressed in Xenopus oocytes than at hNa V 1.7 expressed in HEK 293 cells [92]. Similarly, Pn3a was approximately 2 fold less potent at rNa V 1.7 and 5 fold less potent at mNa V 1.7 compared to hNa V 1.7 [61].
Interaction with membrane lipid is another important feature of spider knottins [14] that can influence affinity and potency [75,82,137]. The anionic charges in the polar head groups of the lipid surrounding Na V s probably increase the affinity of positively charged spider knottins, while hydrophobic residues in these toxins can make favourable electrostatic and hydrophobic interactions with the hydrophobic core of the lipid bilayer [138]. Although spider knottins like ProTx-I and ProTx-II exploit electrostatic interactions to increase their potency, HwTx-IV and Hd1a (Haplopelma doriae) rely less on membrane binding for potent inhibition [139,140]. Interestingly, a HwTx-IV analogue engineered to increase the affinity for lipid membrane showed improved inhibitory potency at hNa V 1.7 [137]. On the other hand, the most potent and selective analogues of GpTx-1 (Grammostola porteri) and ProTx-II had reduced affinity for lipid bilayer [75]. Thus, a direct correlation between the toxin-lipid interaction and the potency or the selectivity for Na V s could not be established [86] beyond facilitating the positioning of the toxin at the membrane surface proximal to exposed Na V residues [75]. Indeed, the conserved amphipathicity of the spider knottins may be an evolutionary adaptation favouring toxin promiscuity [86].

Structure-Function of NaSpTx
Structure-function studies highlight the role of knottin residues in determining the toxin potency and selectivity [14]. For example, the distribution of Ser4 and Asp5 at N-terminus instead of cationic Lys4 and Arg5 of δ-ACTX-Hv1b (H. versuta) shifted selectivity for mammalian Na V channels [97]. Similarly, CcoTx-2 (D32Y-CcoTx-1) potently inhibited Na V 1.3 (IC 50 88 nM), while CcoTx-1 was inactive at this subtype. Surprisingly, Asp32 or Tyr32 located on the side opposite the hydrophobic patch also influenced potency, suggesting that residues beyond the hydrophobic patch can also influence toxin-channel interactions [73]. In addition, the ICK fold retained by key mutants suggested that specific amino acid interactions between the toxin and channel are prime in Na V channel modulation and subtype selectivity [83,85].
The NaSpTx family 1-3 incorporate most of the Na V -modulating spider knottins with promising therapeutic lead potential, including ProTx-II [103,141], ProTx-I [132], ProTx-III [80], Df1a [92], Pn3a [61], HwTx-IV [83,137,142,143], GpTx-I [76] and CcoTx-I [144]. Sequence alignments of selected Na V -modulating spider knottins targeting Sites 3 and 4 in Na V channels are shown in Figure 4. Besides the shorter N-terminus and longer C-terminus in Site 3 and Site 4 targeting spider knottins, this alignment highlights conserved hydrophobic residues in loop 1 and positively charged residues in loop 4 of Site 4 targeting spider knottins. In contrast, negatively charged residues are distributed in loop 2 of NaSpTx1-3 spider knottins and in the N-terminal of NaSpTx families 1 and 2. The conserved Arg and Lys in loop 4 of NaSpTx1-3 appear to be crucial for inhibitory function of depressant spider toxins belonging to these families [145]. The NaSpTx family 1-3 generally target Site 4 to inhibit channel activity, except Hm1a which targets Site 3 to excite the channel [20]. JzTx-14 [125] from NaSpTx7 targets Site 4 and inhibits the channel, whereas JzTx-I [100] and JzTx-II [98] from NaSpTx7 targets Site 3 to excite the channel. Yellow highlights conserved cysteines, green highlights hydrophobic residues, cyan indicates positively charged residues, red indicates negatively charged residues and bold letter indicates the aromatic residues. The "*" indicates identical residues, ":" indicates strong conservation, "." indicates weak conservation.
The C-terminal WCK motif is also conserved across well-characterized sodium channel-blocking toxins from NaSpTx families 1 and 3, suggesting that these residues are key determinants of activity [83]. In the cryogenic electron microscopy (cryo-EM) structure of ProTx-II-hNav1.7 VSDII-Na V Ab complex, the corresponding Trp24 served as a hydrophobic anchor to stabilize ProTx-II interactions with hNa V 1.7 [58]. More specifically, the hydrophobic patch Trp5, Trp7, Trp24 and Trp30 stabilized binding by allowing deeper penetration into the membrane lipid [58]. In addition, the ProTx-II-hNa V 1.7 cryo-EM structure confirmed a key role of the C-terminal basic residues capped by hydrophobic residues in anchoring the toxin into the membrane [58]. In contrast, JzTx-14 belonging to NaSpTx7 has a loop 4 that lacks positively charged residues and Arg serving as a C-terminal cap, but still inhibits eight out of nine Na V channel subtypes at nanomolar concentrations [125]. It has additional hydrophobic residues in each of the four loops with only one acidic residue that suggest an alternative binding mode. Indeed, structure-function studies have shown that hydrophobic and aromatic residues in loop 1, loop 4 and C-terminus and positively charged residues distributed in loop 4 and C-terminus were critical for toxins' affinity [83,85,117]. Interestingly, the highly Na V 1.7-selective Pn3a, like other NaSpTx2 spider knottins, lacks positively charged residues in the C-terminus and instead contains hydrophobic and negatively charged Asp residues.
The overall net charge of the peptide also affects the toxin activity. For example, decreased net anionic charge in E1G, E4G, Y33W-HwTx-IV enhanced inhibition by 45 fold (IC 50 0.4 nM) compared to native HwTx-IV (IC 50 17 nM) [142], which was mostly driven by enhanced hydrophobic interactions associated to the Y33W mutation (IC 50 1.4 nM). Amidation of the C-terminus also has a direct influence on the potency of HwTx-IV [142]. While characterizing ProTx-III and Df1a interactions with Na V channels, Cardoso et al. [80,92] elucidated the significance of C-terminal amidation in enhancing potency and altering subtype selectivity of spider toxins.
The comparison of the potency of spider knottins over Na V 1.1-Na V 1.8 for Site 4 ( Figure 5A) and Site 3 ( Figure 5B) toxins highlights the limited pharmacological data for Site 3 spider knottins that are available mostly for Na V 1.3 and Na V 1.5. In contrast, Site 4 targeting spider knottins preferentially target hNa V 1.7 with Cd1a and CcoTx-2 having clear preference over hNa V 1.2, making this class an excellent starting point for the design of analgesic spider knottins. Indeed, the hNa V 1.7 selective spider knottin Pn3a confirms this potential [61,146] and shows analgesic effects in acute postsurgical pain [146]. A more complete list of spider knottins showing potency across different Na V channel subtypes and DRG are listed in Appendix A Tables A1 and A2.  50 of Site 4 spider knottins normalized against Na V 1.7 are shown. These data are from assays performed on human Na V subtypes, except JzTx-14 which was performed at mammalian Na V subtypes. Data for GpTx-1 [122], ProTx-III [80], Cd1a [110], and CcoTx-2 [110] were obtained from Fluorescence Imaging Plate Reader (FLIPR) experiments, while the remainder were acquired using electrophysiology. Black dots indicate that the IC 50 values for the corresponding knottins were less potent than the value indicated. (B) The pEC 50 of Site 3 targeting spider knottins are shown. JzTx-I [100] was tested on rat Na V 1.2-1.4 and human Na V 1.5 and Na V 1.7. JzTx-II [98] was tested on rat Na V 1.3 and human Na V 1.5. PnTx2-1 [147] was tested on rat Na V 1.1 and Na V 1.8, and human Na V 1.5. Hm1a [20] was tested on human Na V subtypes.
Mutagenesis and chimera studies of Na V channels are typically used to determine critical residues for toxin binding on hNa V channels [92,129,148]. These studies have revealed why the spider knottins HnTx-III, HnTx-IV, and HwTx-IV show Site 1-like channel inhibition (pore blocker) although they are gating modifiers [113,116,128]. Specifically, the Y326S mutation in Na V 1.7 decreased the channel sensitivity to tetrodotoxin (TTX) but not to HwTx-IV [114], while three residues (i.e., Glu753, Asp816 and especially Glu818) outside the pore were shown to be involved in the interactions of HnTx-IV with hNa V 1.7 ( Figure 6A,B) [129]. hNa V 1.7/rK V 2.1 S3-S4 paddle chimera studies revealed that Df1a primarily interacted with the DII voltage sensor of hNa V 1.7 and had weaker interactions with VSDs of DIII and DIV [92]. Wingerd et al. [111] showed that the spider knottin Pre1a interacted with the DII and DIV S3-S4 loops of Na V 1.7 as well as the S1-S2 loop of DIV with the latter interaction likely conferring subtype selectivity. They also showed the role of the serine residue in the DIV S2 helix of hNa V 1.1 ( Figure 6C) and rNa V 1.3 in inhibiting the fast inactivation process in these channel subtypes [111]. Spider knottin Hm1a also targets DIV S3b-S4 and S1-S2 loops that likely underlie its subtype selectivity for hNa V 1.1 [20].

hNav1.9 L I I L N M I S M M A E S Y N Q P K A M K S I L D H L N W V F V DIV S1 DIV S1−S2 DIV S2
Pre1a JzTx-II

hNav1.1 V V V I L S I V G M F L A E L I E K − − Y F V S P T L F R V I R L A R I hNav1.2 V V V I L S I V G M F L A E L I E K − − Y F V S P T L F R V I R L A R I hNav1.3 V V V I L S I V G M F L A E M I E K − − Y F V S P T L F R V I R L A R I hNav1.4 V V V I L S I V G L A L S D L I Q K − − Y F V S P T L F R V I R L A R I hNav1.5 V V V I L S I V G T V L S D I I Q K − − Y F F S P T L F R V I R L A R I hNav1.6 V V V I L S I V G M F L A D I I E K − − Y F V S P T L F R V I R L A R I hNav1.7 V V V I I S I V G M F L A D L I E T − − Y F V S P T L F R V I R L A R I hNav1.8 V V V − L S I A S L I F S A I L K S L Q S Y F S P T L F R V I R L A R I hNav1.9 V V V L L S I V S T M I S T L E N Q E H I P F P P T L F R I V R L A R I DIV S3 DIV S4 DIV S3−S4
JzTx-I ProTx-II JzTx-II Figure 6. Sequence alignment of (A) DII S1-S2, (B) DII S3-S4, (C) DIV S1-S2 and (D) DIV S3-S4 of hNa V 1.1-1.9. Identical residues among hNa V subtypes are highlighted in red. hNa V 1.9 demonstrates the highest variation compared to other hNa V subtypes. Boxes highlights key residues in the interactions for the spider knottins HwTx-IV [149], HnTx-IV [129], ProTx-II [58,104], JzTx-III [106], JzTx-II [98], JzTx-I [100], JzTx-34 [119] and Pre1a [111].
Certain spider knottins can interact with multiple sites in the channel as demonstrated by ProTx-II's interactions at DII and DIV of hNa V 1.7 ( Figure 6B,D) [55,104]. Furthermore, the DII residues, Glu753, Glu811, Leu814, Asp816 and Glu818, in hNa V 1.7 that are critical for inhibition of activation by HwTx-IV are partially conserved in hNa V 1.7 DIV. However, the partial conservation of DII residues in DIV (EgLDi) in wild-type hNa V 1.7 made HwTx-IV interact specifically with DII [149].
Mutational studies further identified that Asp1609 in hNa V 1.5 was crucial in determining the JzTx-II potency. However, rNa V 1.8 and rNa V 1.9 resistant to JzTx-II have Ala and Arg, respectively, instead of Asp1609 [98]. Similarly, Glu818 in hNa V 1.7 is conserved in HnTx-IV-sensitive Na V channels rNa V 1.2 and rNa V 1.3 but is replaced by a neutral amino acid in HnTx-IV-resistant rNa V 1.4 and hNa V 1.5 [129]. In addition, Schmalhofer et al. [132] showed the crucial role of Phe813 in ProTx-II's selectivity for hNa V 1.7 over other Na V subtypes. Finally, cysteine palmitoylation, which is a common reversible lipid modification process vital for Na V channel biosynthesis, is also associated with the affinity of spider knottins to Na V channels [150]. It regulates the gating and pharmacology of WT-rNa V 1.2a as observed by the hyperpolarizing shift in steady-state inactivation and slowing in the channel recovery from fast inactivation upon depalmitoylation of intracellular cysteines on both WT-rNa V 1.2a and G1097C-rNa V 1.2a [150].

Knottins for Na V s in Pain Pathways
Pain allows direct perception of noxious stimuli to avoid actual or potential tissue damage. Primary sensory neuron (nociceptor) signals are transmitted to the brain through action potentials generated by ion channels and receptors. Genetic and molecular studies in animals and humans identified six Na V channel subtypes (i.e., Na V 1.1, Na V 1.3, Na V 1.6, Na V 1.7, Na V 1.8 and Na V 1.9) critical for the generation and transmission of pain-related signals [20,21,36,[151][152][153][154][155][156][157][158]. The Na V 1.7, Na V 1.8 and Na V 1.9 are preferentially expressed in the peripheral nervous system (PNS), while Na V 1.1 and Na V 1.6 are found in both the central nervous system (CNS) and PNS. The subtype Na V 1.3 is generally expressed in the CNS and absent in the adult PNS but is re-expressed in peripheral pain-signalling pathways upon neuronal injury [21,154,159]. The PNS localization of Na V 1.3, Na V 1.7, Na V 1.8 and Na V 1.9 avoids reaching the CNS and inducing associated off-target side effects. Besides these four major peripheral targets, there is evidence of the involvement of Na V 1.6 [160][161][162] and Na V 1.1 [20] in various peripheral pain pathways, which suggests these might also be drug targets in chronic pain types.

Na V 1.1
Votage-gated sodium channel subtype 1.1 (Na V 1.1) is a TTX-S sodium channel encoded by the SCN1A gene located on human chromosome 2q24.3 [163]. They initiate action potential and repetitive firing in neurons and are expressed in both the CNS and PNS and including the colonic myenteric plexus [17,164]. In the PNS, Na V 1.1 is predominantly expressed in medium-to-large diameter DRG neurons (A-fibres) but less expressed in small diameter unmyelinated neurons (i.e., C-fibres [165,166]) and has a smaller contribution in C-fibre-mediated nociceptive transmission [166,167]. However, in colonic afferents which predominantly comprise C-fibres, nearly 50% of the neurons are expressed Na V 1.1 [168]. The hNa V 1.1-selective activator Hm1a revealed its role in mechanical but not thermal hypersensitivity in the absence of neurogenic inflammation [20]. Recently, Salvatierra et al. [19] demonstrated the upregulation of Na V 1.1 in chronic visceral hypersensitivity (CVH) and its inhibition reducing the mechanical pain in an irritable bowel syndrome (IBS) rodent model. Furthermore, Na V 1.1 contributed to peripheral nerve injury-associated mechanical hypersensitivity [19].
Besides the role in nociception [19,20], Na V 1.1 participates in the familial hemiplegic migraine type 3 [169]. In addition, the anti-epileptic drug rufinamide, a Na V 1.1 inhibitor [170,171], alleviated spared nerve injury-evoked mechanical allodynia which also stabilizes Na V 1.7 in the inactivated state [172]. In addition to the hNa V 1.1-selective activator Hm1a, which identified a key role for this subtype in mechanosensitive pain, the spider knottins CcoTx-1, CcoTx-2, Df1a, ProTx-III, HwTx-IV and Pre1a inhibit Na V 1.1 in the nanomolar range and are thus potential leads to novel analgesics (Table A1).

Na V 1.3
Voltage-gated sodium channel subtype 1.3 (Na V 1.3) is a TTX-S sodium channel encoded by the SCN3A gene located on human chromosome 2q24.3 [163]. It produces fast activating and inactivating sodium currents and ramp currents due to the fact of its slow closed-state inactivation. These contribute to neuronal hyperexcitability by reducing thresholds and enhancing the repetitive and ectopic firing in injured neurons [17,154,173]. This channel is primarily expressed in embryonic DRG neurons and absent in adult DRG, but re-expressed during peripheral nerve injury and painful neuromas [166,174,175]. It is also expressed in enterochromaffin cells in the large and small intestine of humans and mice, where they participate in responses to chemical and mechanical stimuli [164,176,177].
Neuropathic pain models such as sciatic nerve transection [178], spinal nerve ligation (SNL) [179], SNI [180] and chronic constriction injury (CCI) confirmed the upregulation of Na V 1.3 [181]. Interestingly, this channel is upregulated only when peripheral projections are transected. The upregulation of Na V 1.3 is also associated to hyperexcitability of small DRG neurons [178]. Furthermore, infraorbital nerve-chronic constriction injury (ION-CCI) produced significant upregulation of Na V 1.3 and downregulation of Na V 1.7, Na V 1.8 and Na V 1.9 in trigeminal nerves [182].
Intrathecal administration of Na V 1.3-targeting antisense oligonucleotides attenuated Na V 1.3 upregulation and consequently reduced hyper-responsiveness of dorsal horn neurons and mitigated pain behaviours following CCI [181]. Chen et al. [156] also observed increased expression of Na V 1.3 in a CCI rat model in which neuropathic pain was alleviated by the intrathecal administration of MiR-96, a microRNA that inhibits Na V 1.3 expression. Another microRNA miR-30b attenuated SNL-evoked neuropathic pain by targeting SCN3A and downregulating the expression of Na V 1.3 mRNA and protein both in DRG neurons and the spinal cord [157].
The role of Na V 1.3 is also demonstrated in inflammatory pain [183], where its expression was upregulated in DRG neurons of rats with diabetic neuropathy and showing mechanical allodynia and thermal hyperalgesia [184]. In this same model, Na V 1.3 knockdown by adeno-associated virus (AAV)-shRNA-Na V 1.3 vector reduced neuropathic pain [185]. Varicella zoster virus (VZV) infection causing post-herpetic neuralgia (PHN) also showed upregulation of Na V 1.3 [186]. However, although Na V 1.3 was downregulated by antisense oligonucleotides in an SNI model, mechanical and cold allodynia were not attenuated [180]. Similarly, the mechanical allodynia remained unaltered following nerve injury in nociceptor specific and global Na V 1.3 knockout mouse models [187]. The druggability of Na V 1.3 in chronic pain management requires Na V 1.3-selective inhibitors [17]. Spider knottins ProTx-II, ProTx-III, HwTx-IV, HnTx-III, HnTx-IV and CcoTx-2 have nanomolar potency towards Na V 1.3 and rational engineering may generate Na V 1.3-selective leads (Table A1).

Na V 1.6
Voltage-gated sodium channel subtype 1.6 (Na V 1.6) is a TTX-S sodium channel encoded by the SCN8A gene which is located in human chromosome 12q13.13 [163]. They localize at the axon initial segment (AIS) and nodes of Ranvier in the CNS and PNS, including central projections and soma of the C-fibres [188][189][190]. It underlies persistent and resurgent sodium currents and repetitive neuronal excitability, with loss-or gain-of-function mutations reducing or increases the neuronal excitability, respectively [188]. Persson et al. [191] demonstrated the expression of Na V 1.6 in the axons of small nerve bundles beneath the epidermis and in the nerve terminals of nociceptors. Its increased expression is observed in complex regional pain syndrome Type 1 (CRPS1), post-herpetic neuralgia (PHN) patients [162,192] and in diabetic neuropathy in mice [193]. Recently, gain-of-function mutation M136V in Na V 1.6 was reported in trigeminal neuralgia with significantly increased peaks inward, resurgent currents and overall excitability of trigeminal nerves [194,195]. Deuis et al. [160] showed that oxaliplatin-induced cold allodynia is mediated by Na V 1.6 expressed in peripheral pathways, suggesting a key role in cold pain pathways. Finally, the selective inhibitor Cn2-E15R developed from the Na V 1.6 agonist Cn2 can be used to determine the extent of Na V 1.6 contributions to pain behaviours [196].
Local knockdown of Na V 1.6 alleviated spontaneous pain and mechanical allodynia imparted by the scorpion toxin BmK I [161]. Furthermore, local Na V 1.6 knockdown reversed mechanical pain in an SNL rat model by reducing sympathetic sprouting around Na V 1.6-positive neurons [197].
The Na V 1.6 expression was also upregulated in a pain model of DRG inflammation where Na V 1.6 knockdown reduced the pain behaviours and the abnormal bursting of the sensory neurons, including nociceptors [198]. Spider knottins modulating Na V 1.6 at nanomolar concentrations include ProTx-II, HwTx-IV, JZTx-14, Pre1a and Df1a (Table A1).

Na V 1.7
Voltage-gated sodium channel subtype 1.7 (Na V 1.7) is encoded by the SCN9A gene which is located in human chromosome 2q24.3 [163]. It is preferentially expressed in large-and-small diameter DRG neurons, visceral sensory neurons, olfactory sensory neurons, trigeminal ganglia and sympathetic neurons [151,152,199]. In DRG neurons, they are present from the peripheral to central terminals, with higher expression in small-diameter DRG neurons (C-fibres) [200]. This channel produces a rapidly activating and inactivating TTX-S sodium current that slowly recovers from inactivation and limits the frequency of firing. The slower onset of closed-state inactivation also limits inactivation during sub-threshold depolarizations and facilitates robust action potential generation that can amplify sub-threshold inputs [152,153,201,202]. Besides action potential generation and propagation, Na V 1.7 also contributes to neurotransmitter release in peripheral and central projections of sensory neurons [151,153].
Exclusive localization in the PNS and compelling genetic and functional evidences have centre-staged Na V 1.7 in pain research [153,158,203,204]. For example, mutations have been identified for various gain-of-function pain disorders, such as inherited erythromelalgia (IEM) [204], paroxysmal extreme pain disorder (PEPD) [203], small fibre neuropathy (SFN) [158] and painful diabetic peripheral neuropathy [205]. The gain-of-function mutations lead to hyperpolarizing shifts in activation, increased amplitude of ramp current, impaired inactivation, increased persistent currents and enhanced resurgent currents, all contributing to the hyperexcitability of DRG neurons [152]. Similarly, recessively inherited loss-of-function mutations are linked to congenital insensitivity to pain (CIP), with loss of olfaction being the only known side effect [36]. In addition, Na V 1.7 was upregulated in rodent models of visceral hypersensitivity [164]. These observations have supported Na V 1.7 as a promising therapeutic target for pain [153]. Considering the recent opportunities for spider knottins to unravel the mechanisms of Na V modulation, the discovery and engineering of spider knottins selectively inhibiting Na V 1.7 provided exciting new opportunities for the development of novel pain therapeutics [135]. A number of spider knottins (Table A1) show nanomolar potency to inhibit Na V 1.7, with several undergoing optimization through saturation mutagenesis, directed evolution and/or rational engineering to enhance the potency and selectivity [14].

Na V 1.8
Voltage-gated sodium channel subtype 1.8 (Na V 1.8) is a TTX-R sodium channel encoded by the SCN10A gene in human chromosome 3p22.2 [163] and preferentially expressed in nociceptive DRG and trigeminal neurons (>90%) as well as in low-threshold mechanoreceptors [206,207], skin free nerve terminals [162], and corneal neurons [208]. It generates slowly inactivating rapidly repriming TTX-R sodium currents with a depolarized shift in the voltage dependence of activation and inactivation [154,209,210]. The channel also contributes to slow resurgent currents that probably underlie the excitability of nociceptors in DRG [211]. It also contributes significantly to the action potential upstroke and are sometimes referred to as overshoot channels [212].
The role of Na V 1.8 in inflammatory pain has been well documented. It had increased expression in DRG neurons when carrageenan was injected into rat hind paw [213] and in cultured DRG neurons treated with inflammatory mediators [214,215]. Similarly, Beyak et al. [216] showed increased Na V 1.8 currents in an animal model of colitis. Notably, inhibition of inflammatory pain induced by complete Freund's adjuvant (CFA) was observed in antisense-mediated Na V 1.8 knockdown in rats [217]. In addition, the upregulation of Na V 1.8 in a mouse model of bowel obstruction underpins its role in visceral hypersensitivity [218]. However, its role in neuropathic pain is not well defined with studies demonstrating downregulation of Na V 1.8 mRNA, protein and currents in the sciatic nerve after axonal transection [219][220][221]. On the other hand, increased Na V 1.8 levels were reported in spared axons and neuronal cell bodies of uninjured nerves [222,223]. This discrepancy is probably associated to the effects of inflammatory mediators in neuropathic pain models [154]. Finally, gain-of-function mutations of Na V 1.8 identified in painful neuropathy cases suggest its role in peripheral neuropathy [224]. Although a few spider toxins such as ProTx-I, ProTx-II and JxTx-14 targeted Na V 1.8 at nanomolar concentrations, only the spider knottin Hl1a shows selectivity towards Na V 1.8 (Table A1).
4.6. Na V 1.9 Voltage-gated sodium channel subtype 1.9 (Na V 1.9) is encoded by the SCN11A gene located in the human chromosome 3p22.2 [163]. These TTX-R channels are preferentially expressed in nociceptive DRG neurons, trigeminal ganglia and myenteric neurons [225,226], where they are activated only at hyperpolarized potentials near resting membrane potential to produce ultra-slow inactivating and persistent sodium currents [227]. These biophysical properties assist in amplifying subthreshold stimuli, lowering the threshold for single action potentials, and increasing repetitive firing [151,227,228].
The role of Na V 1.9 has been demonstrated in inflammatory pain with increased Na V 1.9 current density and lower thresholds for action potential generation that ultimately enhances neuronal excitability [151,214,229]. Indeed, Na V 1.9 knockout mice showed diminished mechanical hypersensitivity to formalin and CFA and failed to develop thermal hyperalgesia upon CFA or carrageenan injection [151,[230][231][232]. Rare dominant gain-of-function mutations in SCN11A have been reported in a number of human pain disorders, such as familial episodic pain [233], painful small fibre neuropathy [234,235] and insensitivity to pain [155]. Despite the challenges in obtaining heterologous expression of Na V 1.9, chimeras Na V 1.9/K V 2.1 showed ProTx-I interacted with Na V 1.9 S3b-S4 paddle motif and potentiated Na V 1.9 currents in rat DRG [236].

Conclusions and Future Directions
In addition to a pivotal role of Na V 1.7 in pain processing, other Na V subtypes including Na V 1.1, 1.3, 1.6, 1.8, and 1.9 are increasingly showing supportive and/or pivotal roles in various acute and chronic pain conditions. A large range of spider knottins modulate the Na V function and continue to be developed as potential analgesic leads. In addition, spider knottins provide exquisite research tools to further explore the role of Na V channels and how they can be modulated. Given that multiple sodium channels often contribute to chronic pain conditions, therapeutic leads targeting multiple Na V channels could be advantageous; for example, chronic visceral pain may be best treated by a Na V 1.1/Na V 1.7/Na V 1.8 inhibitor, while diabetic neuropathy may be best treated by a Na V 1.3/Na V 1.6/Na V 1.7 inhibitor. However, despite Na V -inhibiting spider knottin potential to target multiple subtypes, most have been optimized for potency and selectivity towards a single channel subtype hNa V 1.7 [14]. This review summarises spider knottin Na V channel pharmacology that might be useful in guiding structure-function studies and the rational design of multi-valent spider knottin leads towards the development of therapeutic leads for chronic pain management.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Table A1. Inhibitory effects of spider knottins on Na V channel subtypes. Values presented were obtained from electrophysiological experiments unless otherwise as stated.