Na_v1.3 is a TTX-S VGSC subtype normally expressed in the central nervous system (CNS). More Na_v1.3 is usually expressed in embryogenesis than adulthood [32], but expression levels increase in peripheral DRG neurons after nerve injury and inflammation [33]. This suggests Na_v1.3 plays a role in pain sensation [34]. Na_v1.3 is also up-regulated in mammalian DRGs after spinal cord [35] and motor fiber injury [36]. Patients with trigeminal neuralgia have increased Na_v1.3 expression in gingival tissue [37].

Na_v1.3 has faster kinetics and recovers more quickly from inactivation than TTX-R VGSCs [38]. Therefore, it could play an important role in high-frequency firing, as seen in chronic pain states [22,39]. Reducing Na_v1.3 expression in pain-sensing peripheral neurons has been shown to diminish hypersensitivity in DRGs and pain [33]. However, Nassar et al. showed that Na_v1.3 knock-out mice still exhibit normal neuropathic pain behaviour and ectopic discharges from damaged nerves [40]. This suggests there may be a possible compensation mechanism in knock-out mice.

Na_v1.7 is a TTX-S VGSC subtype, predominantly restricted to the peripheral nervous system (PNS), sensory and sympathetic neurons, and Schwann and neuroendocrine cells [41,42,43,44,45]. It displays rapid activation and inactivation kinetics [46].

Recently, Na_v1.7 received attention as a possible therapeutic target for drugs to treat pain. Mutations in the SNC9A gene, which encodes Na_v1.7, have been shown to produce or prevent pain. Gain-of-function mutations, which produce pain, lead to a syndrome called ‘primary erythromelalgia’, a paroxysmal extreme pain disorder [47,48]. This syndrome is characterized by severe burning pain and redness in the extremities. In contrast, loss-of-function mutations, which prevent pain, lead to a syndrome called ‘congenital insensitivity to pain’ or ‘congenital analgesia’ [49,50,51]. This divergent spectrum, from feeling no pain to feeling constant pain, makes Na_v1.7 an interesting potential therapeutic target. Consequently, Na_v1.7 blockers are prospective analgesic compounds. For example, a peripherally acting Na_v1.7-specific blocker, N-[(R)-1-((R)-7-chloro-1-isopropyl-2-oxo-2,3,4,5-tetrahydro-1H-benzo[b]azepin-3-ylcarbamoyl)-2-(2-fluorophenyl)-ethyl]-4-fluoro-2-trifluoromethyl-benzamide (BZP), has been shown to be as effective as the analgesic drugs gabapentin and mexiletine in reversing hyperalgesia and allodynia in rat models of inflammatory and neuropathic pain, but did not impair motor function [52].

Na_v1.7 has also been shown to be necessary for odour perception in rats, mice and humans [53]. Humans with loss-of-function mutations in the SCN9A gene couldn’t detect odours, and although Na_v1.7-null mice neurons still produced odour-evoked action potentials, they couldn’t initiate synaptic signalling from their axon terminals at the first synapse in the olfactory system. As a result, these mice no longer displayed vital, odour-guided behaviours [54]. Na_v1.7 has also been shown to be expressed in rat olfactory sensory axons and is present in vomeronasal axons, indicating a role for Na_v1.7 in transmitting pheromonal cues [55]. In addition, neuroepithelial injury caused transient expression of Na_v1.7 by dendritic cells of monocytic lineage, suggesting an emerging role for this channel in immune function [55].

Although Na_v1.8 is selectively expressed in sensory neurons and C-type nerve fibers involved in nociception, mediating a slow-inactivating, TTX-R Na⁺ current, its precise role in pain is unclear. Peripheral nerve injury has been shown to reduce Na_v1.8 expression in damaged neurons, indicating that Na_v1.8 does not contribute to neuropathic pain [56]. However, Na_v1.8 has also been shown to redistribute to the axons of uninjured sciatic nerves after spinal nerve ligation, indicating it plays a crucial role in pain states [56]. Zimmermann et al. also showed that Na_v1.8 is essential for nociception in the cold and for cold pain [57]. Changes in Na_v1.8 gene expression are partially attributed to changes in growth factor levels and altered auxiliary β-subunit expression levels [58,59].

Na_v1.8 knock-out mice display increased pain behaviour in comparison with wild-type mice, providing further evidence for a major role for this channel in pain [18,60,61]. Furthermore, discovery of the μO-conotoxin MrVIB, a potent and preferential Na_v1.8 inhibitor, has provided evidence that blocking Na_v1.8 can alleviate chronic pain in rats [61,62]. Nevertheless, the role of Na_v1.8 in neuropathic pain remains a matter for debate [37,45,60,63,64,65].

Na_v1.9 is a TTX-R VGSC subtype whose expression is restricted to the PNS. Its amino acid sequence exhibits only around 50% homology to that of other VGSCs. Na_v1.9 elicits extraordinarily slow persistent currents and its activation kinetics are too slow to contribute to the action potential’s upstroke. Accordingly, Na_v1.9 probably depolarizes the resting membrane potential, lowering the threshold for initiating action potentials [66,67,68].

Studies using Na_v1.9-null mice suggest this subtype plays a predominant role in inflammatory, but not neuropathic, pain [69,70]. However, the role of Na_v1.9 in inflammatory pain is not completely clear, because Na_v1.9 knockdown mediated by anti-sense oligodeoxynucleotides doesn’t reduce thermal hypersensitivity associated with complete Freund’s adjuvant-induced inflammatory pain [71]. It has been suggested that Na_v1.9-mediated currents are down-regulated after nerve injury, but other studies contradict this idea [22,64].

μ-Conotoxins are highly basic, small, rigid peptides ranging from 16 to 26 amino acids long, and contain three disulfide bonds (Table 1). μ-Conotoxins bind to the pore region of VGSCs. Their selectivity for VGSC subtypes has been based predominantly on differences in the turret region (S5-P loop linker), because differences close to the pore are minimal [90].

The first μ-conotoxin isolated and characterized was GIIIA from C. geographus. GIIIA was shown to be a potent and selective blocker of muscle VGSCs [91]. It inhibits the skeletal muscle VGSC, Na_v1.4, with an IC₅₀ in the low nanomolar range [92,93]. GIIIA’s ability to selectively target a single VGSC subtype suggests there may be other μ-conotoxins that could selectively target different VGSC subtypes, particularly those involved in pain signalling. Since the discovery of GIIIA, numerous μ-conotoxins with new sequences have been identified through molecular cloning techniques. Many of their structures have been solved using NMR spectroscopy (Figure 1) [94,95,96,97,98].

The first μ-conotoxin shown to inhibit a neuronal VGSC was PIIIA. Despite having modest affinity for the neuronal channel Na_v1.2 (~500 nM), PIIIA primarily blocks VGSCs in mammalian skeletal muscle [99].

Another μ-conotoxin, KIIIA, was recently shown to preferentially bind to the neuronal channel Na_v1.2 over the skeletal muscle channel Na_v1.4. KIIIA also has a nanomolar affinity for Na_v1.7, a VGSC involved in propagating pain signals [100]. Being only 16 amino acids long, KIIIA is considerably shorter than previously examined μ-conotoxins. The short length is due to fewer residues within the N-terminal half of the peptide, suggesting C-terminal residues are more critical for toxin binding than N-terminal residues.

KIIIA and another small μ-conotoxin, SIIIA, have been shown to have analgesic properties in mouse inflammatory pain assays [101,102,103]. These μ-conotoxins have a framework that could be used to develop new analgesic drugs. Their small size, high affinity and unique VGSC channel selectivity make them strong prospective therapeutic agents.

μ-Conotoxins bind to the outer vestibule to inhibit current through the channel. To understand which of the four channel domains are critical to binding, chimeras between rat Na_v1.4 (rNa_v1.4) and human Na_v1.5 (hNa_v1.5) or human Na_v1.8 (hNa_v1.8) have been constructed and examined [104,108]. These studies showed that SIIIA and CnIIIC are potent blockers of the VGSC subtypes Na_v1.2 and Na_v1.4, but inactive at Na_v1.8 and Na_v1.5. The studies further revealed that determinants for the toxin’s inability to block Na_v1.8 were located across all domains of the channel, whereas those for Na_v1.5 were restricted to DI and DII.

There are few differences among the VGSC isoforms in the pore region near the inner and outer ring charge groups (important for ion selectivity and permeation) of the outer vestibule in each of their four channel domains. However, a single amino acid residue change in this region has been shown to be critical for classical pore blockers saxitoxin and TTX [7]. This residue is located in DI, one residue outside the selectivity filter aspartate. It is either a tyrosine or phenylalanine in TTX-S channels, and a serine or cysteine in TTX-R sodium channels.

This variation among sodium channel isoforms also affects μ-conotoxin binding properties, such as affinity, but this change is modest [100,110]. Nevertheless, it modulates the actions of various μ-conotoxins analogues that incompletely block single channel currents [106]. Potential derivatives that do not completely block single channel currents could fine tune hyperexcitability. Derivatives of GIIIA and PIIIA as well as KIIIA and its derivatives, which have been examined at the single channel level, cannot completely block single channel currents [93,101,110] and could be used in this way. A drug that modulates single-channel currents would be an effective treatment to reduce hyperexcitability without eliminating Na⁺ channel currents completely.

An additional variation in the sequence in this VGSC region is attractive to examine because it may identify potential human Na_v1.7 (hNa_v1.7) blockers. Human Na_v1.7 is the only VGSC subtype, from any species, where the DIII outer ring charge (typically an aspartate) has been replaced by a neutral isoleucine. μ-Conotoxin binding affinity is critically influenced by residues in the outer ring, especially those of DI and DII [111,112,113]. Recent studies using KIIIA showed that charges interacting with this outer ring charge, when neutralized, affected the derivative’s affinity for hNa_v1.7 much less than its affinity for Na_v1.2 and Na_v1.4 [100]. The missing outer ring charge of hNa_v1.7 is believed to be the best site to engineer toxin specificity towards hNa_v1.7. For example, in KIIIA, the derivative H12Q, which, based on molecular dynamics docking simulations, should interact with the DIII outer ring charge, increased toxin selectivity for hNa_v1.7 over Na_v1.2 and Na_v1.4. The R14A derivative increased the toxin’s selectivity for hNa_v1.7 to 10-fold more than that for Na_v1.2 and Na_v1.4 [100]. This is the first example of an engineered μ-conotoxin that is selective for hNa_v1.7. This μ-conotoxin interaction site is a potential target for designing hNa_v1.7-specific blockers, which could increase selectivity towards Na_v1.7 and decrease the risk of possible side effects.

Recent studies using new peptide synthesis techniques are shedding light on how to create stable and selective sodium channel modulators, and identifying new drug leads. Peptides as drug leads have a few drawbacks, such as unstable disulfide bonds, protease degradation and inefficient transport of the peptide to its intended site of action. To address these concerns, researchers have been trying to improve the native μ-conotoxin by creating more stable peptide bridges, using truncated peptides and/or peptide mimetics. They have also used μ-conotoxins as scaffolds to enhance the peptide’s potential as a drug.

μ-Conotoxins have three disulfide bonds, which can be oxidized and reduced, and refolded to inactive conformations. This was examined in detail for PIIIA, where three PIIIA isomers were created with different disulfide connections [114]. Despite differences in connectivity, the three derivatives all had sub-micromolar affinity for Na_v1.4. This suggests that, despite the toxin preferring to fold to a single conformation, other conformations could also be active. It will be of interest to examine these different PIIIA isomers to determine if they exhibit different selectivity to that of the native toxin.

To make the peptide structure stable, different non-disulfide bonds can also be introduced. For example, SIIIA with diselenide bridges instead of disulfide bridges was recently synthesized [115]. Diselenide bonds are considerably more stable than disulfide bonds, making the μ-conotoxin itself more stable [116].

With many new μ-conotoxin sequences being discovered, there seems to be greater divergence in the N-terminus half of the toxin sequence. For example, KIIIA seems to be missing many more residues in the N-terminus than other μ-conotoxins. It may be interesting to examine the effects of smaller μ-conopeptides by removing N-terminal residues. Several groups recently examined short peptides based on KIIIA and SIIIA, with varying success. Norton and colleagues used lactam bridges to stabilize the α-helical segment of μ-conotoxins [117,118]. This segment has most of the critical residues for toxin binding and might be used as a template on which to base drug design. KIIIA and BuIIIC were used to design ‘mini peptides’ ranging from 12 to 16 amino acids, to try to understand the pharmacophore of μ-conotoxins [119,120].

In contrast with these two studies, adding residues to the N-terminus of SIIIA, SIIIB and TIIIA has been shown to change their selectivity preference from rat muscle channels to rat neuronal VGSCs [121]. This suggests that N-terminal residues also contribute to the toxins’ affinity and selectivity.

As peptides are degraded or truncated by proteases, non-peptide backbones have also been examined to try to increase drug bioavailability. KIIIA and SIIIA structure has undergone non-peptide modification to help design new drug leads. For example, Bulaj and colleagues examined the effects of removing a single disulfide bond to see if it is feasible to use non-peptide backbone spacers and disulfide bond deletion to increase bioavailability without decreasing affinity or change selectivity. They also used amino-3-oxapentanoic or 6-aminohexanoic acids to replace non-essential amino acids in the toxin to further minimize the protein component of the toxin [103,122]. Resulting analogues still inhibited VGSCs and produced analgesic effects.

μ-Conotoxins as potential analgesics are still in the preclinical stage. We are learning how these unique toxins can block specific VGSC subtypes, and how we can modify them without changing their selectivity and high affinity for VGSCs. As we further examine these toxins and isolate new toxins that can block specific VGSC subtypes, we will be able to design new selective and clinically relevant drugs to block VGSC subtypes involved in pain signalling.

μO-Conotoxins are hydrophobic peptides belonging to the O-superfamily of conotoxins. This group of toxins was first isolated from Conus marmoreus [82]. To date, five μO-conotoxins have been discovered (Table 2). They range from 28 to 32 amino acids in length, with each toxin containing three intramolecular disulfide bonds. Two of these μO-conotoxins, MrVIA and MrVIB, have high sequence homology. Only two residues differ between them. MrVIA and MrVIB are of interest because they block TTX-R Na⁺ currents in mammalian DRG neurons (which are mainly mediated by Na_v1.8) 10-fold more than TTX-S Na⁺ currents [62,123,124]. However, despite the selectivity of MrVIA and MrVIB for the TTX-R subtype Na_v1.8 over TTX-S subtypes, and the resulting analgesic activity in a variety of animal models of pain [62,123,124,125], little is known of the structure–activity relationships that define μO-conotoxin subtype specificity.

MrVIA inhibits VGSCs and voltage-gated calcium channels (VGCC) in Aplysia californica and Lymnaea stagnalis molluscan neurons [82], and amphibian VGSCs [124]. It also inhibits TTX-S Na⁺ currents in rat hippocampal neurons and heterologously expressed rNa_v1.2 in Xenopus oocytes with IC₅₀ values of 200 nM [126]. MrVIA blocks TTX-R VGSCs from rDRG neurons with an IC₅₀ of 82.8 nM [125]. Consistent with μO-conotoxins acting as gating modifiers, MrVIA inhibition of Na⁺ current appears to be voltage-dependent, with a reduced affinity for the channel after depolarizing voltage steps [77,127].

MrVIB more selectively inhibits TTX-R than TTX-S neuronal VGSCs, and is also selective between different TTX-R VGSC subtypes. It is 100-fold more selective for Na_v1.8 than Na_v1.9 in DRGs [62]. Its selectivity for hNa_v1.8 over other VGSC subtypes was confirmed using Xenopus oocytes expressing a range of different VGSC subtypes [62]. This is particularly interesting, since only a few Na_v1.8-selective modulators have been described to date. Zorn et al. also showed that MrVIB is more selective for Na_v1.4 than Na_v1.2 expressed in HEK 293 cells [128].

MrVIB has exhibited analgesic activity in animal pain models and decreased mechanical and thermal pain sensation [62,125]. Intrathecally applied MrVIB was 30 times more effective than lidocaine on allodynia and hyperalgesia and had only small motor system side effects [62]. The 3D structure was determined using NMR data (Figure 1) [125].

There is considerable interest in finding out how μO-conotoxins bind to VGSCs, since they are modestly selective inhibitors of TTX-R VGSC currents in rat DRG neurons [125] and have potential as analgesic drugs [62,123,124]. Analysis of the MrVIA binding site in rNa_v1.4 channels and competition experiments with the scorpion toxin, β-toxin Ts1, identified that the C-terminal pore loop of DIII is necessary for MrVIA to bind to Na_v1.4 [128]. In contrast, a later study using site-directed Na_v1.4 mutagenesis, revealed that DII’s voltage sensor is the main MrVIA binding site. The same study concluded that MrVIA interaction with DIII of Na_v1.4 appeared to play a lesser role [77]. Neither of these two studies examined the binding site and mechanism of action of μO-conotoxins on Na_v1.8.

Na_v1.8 is only 50% identical to Na_v1.2, a major VGSC of the central nervous system, and its affinity for MrVIB is higher than that of Na_v1.2 [62]. To address the functional and pharmacological significance of these differences, Knapp et al. used a domain swapping strategy between rNa_v1.2 and hNa_v1.8 [129]. Heterologous expression of Na_v1.2/Na_v1.8 chimeras in Xenopus oocytes and analysis of their inhibition by MrVIB revealed that the region between segment S6 of DI and the external loop of DII in Na_v1.8 is the main determinant for the μO-conotoxin family’s high affinity for Na_v1.8 (Figure 2) [129]. Comparing this data with those of Leipold et al. supports the likelihood that MrVIB inhibition of Na_v1.8 is mediated via an interaction between the toxin and DII’s voltage sensor [77,129]. A similar approach was used to study the interaction between Na_v1.9 voltage sensors and scorpion or tarantula toxins. The voltage-sensor paddle motifs of Na_v1.9 were transferred into the voltage-gated potassium channel K_v2.1 [130]. This enabled examination of their interactions with tarantula and scorpion toxins [131]. The possibility that MrVIB could also interact with other domains cannot be excluded. Results from recent studies into MfVIA inhibition of Na_v1.2 and other voltage-sensor toxins such as tarantula toxins may support the idea of multiple binding sites [132,133].

MfVIA, a novel μO-conotoxin, was recently identified from Conus magnificus. Similar to MrVIA and MrVIB, it is a hydrophobic peptide containing 32 residues, but has highest sequence homology to MrVIB [133]. MfVIA’s amino acid sequence is only slightly different to those of MrVIA and MrVIB. Surprisingly, this results in MfVIA having different potencies and selectivity towards mammalian VGSC subtypes from those of MrVIA and MrVIB. MfVIA is three-fold more potent towards Na_v1.4 and five-fold less potent towards Na_v1.2 than MrVIA or MrVIB. MfVIA inhibits TTX-R Na⁺ currents expressed in DRG neurons five times more strongly than those heterologously expressed in rNa_v1.8. It also inhibits Na_v1.4 and Na_v1.8 at low nanomolar concentrations, whereas significantly higher toxin concentrations are needed to inhibit all other VGSC subtypes (Table 2) [133].

The established model of μO-conotoxin mediated block of VGSC channels involves a single toxin molecule binding and inhibiting the channel. This is consistent with MfVIA inhibition of all VGSC subtypes except Na_v1.2. The concentration–dependence for Na_v1.2 inhibition was steeper than the expected Hill slope of −1 [133]. This may indicate that MfVIA interacts with more than one binding site. Another possibility might be that μO-conotoxins have slow on-rates that make it difficult to reach equilibrium at low peptide concentrations [133].

DNA sequencing of C. litteratus has revealed the existence of two novel μO-conotoxins, LtVIC and LtVIIA. These recombinant toxins inhibit Na⁺ currents in DRG neurons in a similar way to known μO-conotoxins. However, their subtype selectivity and structure–activity relationships are yet to be investigated [135,136,137].

MrVIA, MrVIB and MfVIA remain the only μO-conotoxins for which mammalian VGSC subtype selectivity data is available, albeit not across all VGSC subtypes. The unique selectivity profile of μO-conotoxins MfVIA, MrVIA and MrVIB [62,82,126,133] makes these peptides attractive leads for analgesic drugs [62,129]. Synthesis of μO-conotoxins does, however, pose significant challenges. This is mainly due to their highly hydrophobic nature and the difficulty of correctly forming their three disulfide bonds to the native isomer. While replacing cysteines with selenocysteines has helped with MrVIB folding [138], synthesis of the native molecule relies on a semi-selective approach [124,134], which produces a low overall yield. Nevertheless, a novel regioselective protocol has been developed for synthesizing the novel μO-conotoxin MfVIA that only uses a single HPLC purification step and increases peptide yields [133].

Another potential problem with μO-conotoxins is that their affinity for Na_v1.4 and Na_v1.8 is almost identical. This may limit their therapeutic effectiveness, unless derivatives of the toxins can make them more selective for Na_v1.8 than Na_v1.4. Stürzebecher et al. introduced a membrane-tethered isoform of MrVIA (t-MrVIA) into nociceptive neurones in mice, where it was successfully expressed on the cell surface [123]. TTX-R VGSC current densities, which were mainly Na_v1.8-mediated, were reduced by 44 ± 7% in t-MrVIA transgenic mice, without up-regulation of TTX-S VGSC, VGCC or transient receptor potential channel expression in nociceptive neurones [123]. t-MrVIA transgenic mice had less inflammatory mechanical hypersensitivity and firing of cutaneous C-fibres sensitive to noxious cold temperatures, and were more insensitive to cold pain than wild-type mice [123]. Besides proving that MrVIA is analgesic, membrane-tethered MrVIAs could be used to study and manipulate VGSCs in specific cell types in the mammalian nervous system [123], further increasing the toxin’s selectivity.