ω-Conotoxins GVIA, MVIIA and CVID: SAR and Clinical Potential

Highly selective N-type voltage-gated calcium (CaV) channel inhibitors from cone snail venom (the ω-conotoxins) have emerged as a new class of therapeutics for the treatment of chronic and neuropathic pain. Earlier in 2005, Prialt (Elan) or synthetic ω-conotoxin MVIIA, was the first ω-conotoxin to be approved by Food and Drug Administration for human use. This review compares the action of three ω-conotoxins, GVIA, MVIIA and CVID, describing their structure-activity relationships and potential as leads for the design of improved N-type therapeutics that are more useful in the treatment of chronic pain.


Introduction
There are approximately 500 different species of the genus Conus. Cone snails, found on the Great Barrier Reef in Australia and most other tropical waters around the world, are predatory marine gastropods they prey on fish, worms or molluscs [1]. They capture their prey by injecting lethal or paralysing venom, comprising more than a hundred different polypeptides [1]. These polypeptides known as conopeptides or conotoxins act selectively at a wide variety of ion channels and receptors in their prey, as well as in mammals. Channels targeted include the nicotinic acetylcholine receptor (AChR), the voltage-gated sodium (Na V ) channel, the voltage-gated potassium (K V ) channel, the voltage-gated calcium (Ca V ) channel, and the N-methyl-D-aspartate receptor [2,3]. Many of these conotoxins have been used to gain further information about their target, either at the pharmacological, physiological or structural level. They are typically 8-30 amino acid residues in length and have been divided into both structural and pharmacological classes. The structural classes are based on the nature of the peptide's secondary structure, which is generally defined by a characteristic pattern of Cysresidues [2,3]. Table 1 illustrates the structural and pharmacological diversity within conopeptides. Importantly, several of these pharmacological classes have therapeutic potential in the treatment of pain, including the noradrenaline transporter inhibitor χ-MrIA (Xen2174), the neurotensin receptor agonist (contulakin-G), the AChR antagonist (α-Vc1.1), and the N-type Ca V channel blockers. This review describes the structure-activity relationship (SAR) and therapeutic potential of ω-conotoxin Ntype Ca V channel blockers form cone snails, the first of the conotoxin classes to be approved as a therapeutic.

ω-Conotoxins
All ω-conotoxins identified to date have been found in piscivorous cone snails ranging from the small Conus catus to the large Conus geographus (Figure 1). The ω-conotoxins are basic peptides of 24-29 residues with an amidated C-terminus [5] and with six Cys-residues arranged to give a fourloop Cys framework (C-C-CC-C-C). ω-Conotoxins target the Ca V channel with most of the ωconotoxins discovered to date being selective for the N-type Ca V channel (Table 2) while a few target the P/Q-type Ca V channel. In addition, the recently isolated ω-conotoxin TxVII, from Conus textile was found to target the L-type Ca V channel [6]. The most studied ω-conotoxins are the selective Ntype Ca V channel blockers GVIA and MVIIA [7,8], and the modestly selective P/Q-type Ca V channel blocker, MVIIC [9]. In 2000, four novel ω-conotoxins CVIA-D were identified from Conus catus [10]. The most interesting of the four being CVID, which displayed high potency at the N-type Ca V channel and low potency at the P/Q-type Ca V channel, making it the most selective ω-conotoxin for the N-type Ca V channel found to date [10].
Sequence hypervariability has been observed between functionally homologous ω-conotoxins, with the six Cys-residues and one Gly being the only residues conserved throughout this set of peptides ( Table 2). Comparison of the sequences of two ω-conotoxins, MVIIA and GVIA, reveals that they share less than one-third of the non-Cys residues [11]. The amino acid composition in each of the corresponding four loops is strikingly different, with GVIA containing three hydroxyprolines and MVIIA containing none. Despite the low sequence identity, they both selectively target the N-subtype of Ca V channels and elicit similar biological effects in animals [3]. Conversely, MVIIC has high sequence identity to MVIIA (Table 2), but quite different selectivity.

Voltage-gated calcium channels
Ca V channels are included in the same transmembrane gene superfamily as the Na V and K V channels [36]. The influx of calcium ions through Ca V channels is involved in a wide range of essential cellular responses including activation of calcium-dependent enzymes, gene transcription, muscle contraction and neurotransmitter release [37]. Ca V channels are part of a suite of ion channels the cell uses to couple electrical signalling at the plasma membrane to a physiological response in the cell. Neuronal Ca V channels are composed of pore forming α 1 subunit that co-assembles with different β and α 2 δ subunits ( Figure 2). While the α 1 subunit is largely responsible for determining the electrophysiological characteristics of the channel, these characteristics are modified by associated β and α 2 δ subunits [38]. To date, six different α 1 channel types named L-, T-, R-, P/Q-and N-type have been identified (Table 3). Ca V Channel as a therapeutic target Because Ca V channels are involved in a multitude of cellular responses, including muscle contraction and neurotransmitter release [39,40], different Ca V channel subtypes have been found to be interesting therapeutical targets. Of the six pharmacologically distinct Ca V channel subtypes (Table  2), the N-type, T-type, and P/Q-type calcium channels are the best validated targets for the treatment of pain [41, 42] although dose-limiting side-effects may limit their clinical application. The N-type Ca V channel has been found to be concentrated in the spinal dorsal horn region [43-47] and helps convey the nociceptive message from the peripheral nervous system (PNS) into the central nervous system (CNS) [48]. Being able to inhibit the nociceptive message has been shown to lead to the interception of the pain signal in animals and will hopefully result in pain relief when used in a clinical setting [49][50][51][52]. The therapeutic potential of MVIIA in pain management has now been identified confirming the role of N-type Ca V channel in pain transmission [53][54][55]. Table 3. Mammalian Ca V channels and selected peptide inhibitors.

ω-Conotoxins as therapeutics
Pain management is currently one of the most actively researched areas in medicine and there has never been a more critical need for novel therapeutics with a number of Cox-2 inhibitors withdrawn from the market [56]. However, despite advances in a number of areas, opioids still play a key part in often inadequate pain management strategies. Since opioids only target certain types of pain and develop tolerance, novel non-opioid analgesic therapies for more severe chronic pain states are desirable. The ω-conotoxins are of fundamental interest as they represent a group of structurally related peptides with a wide range of Ca V channel subtype specificities [57] that are believed to act at or near the outer vestibule of the Ca V 2.2 ion conducting pore [58]. They are now emerging as a new therapeutic class for intrathecal (IT) pain management [26,[28][29][30][31]33, 35] as well as having the potential to treat ischaemic brain injury [24,25,59]. However, all ω-conotoxins evaluated produce dose-limiting side-effects in animal models which may be associated with effects on inhibitory or supraspinal pathways. Peptides are readily targeted by proteases and tend not to display satisfactory bio-availability unless administered directly to the target, hence the need for IT delivery. ω-Conotoxins such as GVIA, MVIIA and CVID, selective towards N-type Ca V channels have been shown to possess neuroprotective and analgesic properties [5,10,60,61] and are therefore interesting in drug development, whereas ω-conotoxins selective to the P/Q-subtypes (i.e., MVIIC and SVIB) are not considered useful leads as they are likely to be lethal to mammals and there have been several reports raising concerns about side effects if these peptides are used as drugs in a clinical setting [62]. It is therefore important to define those structural and functional factors that contribute to N-type as opposed to P/Q-type selectivity so that side-effects associated with P/Q-type current inhibition might be avoided. Several groups have provided insight into the domains and specific residues of the ωconotoxins that are responsible for this selectivity [63][64][65][66][67][68][69][70].

GVIA
GVIA is a selective inhibitor of the N-type Ca V channel [5,[71][72][73] with greater potency in vivo than MVIIA [74] and CVID [32]. Intrathecal administration of morphine, GVIA, MVIIA and CVID is effective in attenuating neuropathic pain in rats [32], with GVIA being about three to four times more potent than MVIIA and CVID and approximately 40-fold more potent than morphine [32]. Due to its slow onset and recovery kinetics [75][76][77][78], GVIA is an almost irreversible inhibitor of the N-type Ca V channel [79] complicating dose control in a clinical setting.

MVIIC and SVIB
Ca V channel blockers, such as MVIIC and SVIB target the P/Q-type Ca V channel [9,15]. Since the P/Q-type current plays an important physiological role including the regulation of transmitter release at neuromuscular junctions [80], the ω-conotoxins are less interesting drug leads for pain management and can be lethal even at low doses [62]. In addition, MVIIC did not provide neuroprotection when tested on a rat model of global ischaemia [81].

MVIIA (Ziconotide or Prialt)
MVIIA is the first ω-conotoxin to enter into clinical trials. Intrathecal MVIIA has recently been approved by the FDA for the management of chronic pain in the USA and Europe [82,83]. MVIIA is a selective, reversible and potent blocker of the N-type Ca V channel [76,84] which shows analgesic and neuroprotective effects in humans [85]. MVIIA has been found to be effective in preventing neuronal cell death following cerebral ischaemia. A single bolus injection provided protection even when administered 24 h after an ischaemic injury [81,86,87]. MVIIA has been reported to inhibit both neuronal excitability and neurotransmission [84,88]. In animal models, IT administered MVIIA reverses acute [26,48,89], persistent [26,48,89] and neuropathic pain [26,90,91], and in humans IT administration of MVIIA provides relief from chronic pain [83,92]. However, it has some undesired side effects such as dizziness, blurred vision, nystagmus, sedation [93] and orthostatic hypotension in humans [78,94,95]. Adverse side effects such as intractable delirium after being treated with MVIIA were reversed with electroconvulsive therapy [96]. Intrathecal MVIIA causes a variety of neurological side effects of unknown origin [97] despite N-type Ca V channels being predominant at synapses carrying nociceptive information to the spinal cord [46]. Importantly, inhibition of N-type Ca V channels with MVIIA produces substantial pain relief in otherwise treatment-refractory patients, and unlike opioid pain management, MVIIA does not develop tolerance or produce addiction [32, 85,98].

CVID (AM336)
Another ω-conotoxin isolated from the fish-hunting C. catus, CVID [10] has also entered clinical trials with AMRAD [99]. CVID is the most selective inhibitor of N-type over P/Q-type Ca V channels showing 6-orders of magnitude selectivity in binding studies [10]. Amongst the ω-conotoxins CVID, GVIA and MVIIA, CVID had the largest ratio of ED 50 (dose causing 50% effect) to TD 50 (dose causing toxicity in 50% of animals) when administered IT in a rat model of neuropathic pain [32]. Due to greater therapeutic margin for CVID seen in animals [34], it is anticipated that AM336 will produce less side effects than MVIIA in the clinic. CVID has undergone a Phase I/IIa assessment for treatment of severe morphine-resistant pain after it was given approval for human trials by the Australian Therapeutic Goods Administration [34, 99].

Opioids and ω-conotoxins
Morphine is a spinal analgesic agent [100] that exerts its effect by binding mainly to µ-opioid receptors. Activating the µ-opioid receptor has been shown to inhibit the N-type Ca V channel current by interacting with the pore-forming α 1 subunit via the G-protein (Figure 3) [101][102][103][104][105]. This inhibition of the N-type Ca V channel contributes to morphine analgesia and uncoupling of the opioid receptor and G-protein [106][107][108][109]   N-type Ca V channel antagonists show analgesic effect alone [97,110] and in combination with µopioids [111,112]. Interestingly, several studies have reported synergistic effects between ωconotoxins (Zicontide or AM336) and morphine when administered IT [32, 98,112]. However, simultaneous administration of morphine and ω-conotoxin did not prevent the development of morphine tolerance [98]. Importantly, this tolerance development did not result in cross-tolerance to MVIIA [98] and instead there was an upregulation of ω-conotoxin binding sites (N-type Ca V channel) in the brain after chronic morphine exposure [113]. Further investigation into the extent of this synergistic effect in the clinic would enable the use of lower doses of N-type antagonists and potentially limit toxic side effects associated with IT administration of ω-conotoxins [32].

Structural studies of ω-conotoxins
To date, all structural information available on the ω-conotoxins has been obtained from 1 H Nuclear Magnetic Resonance (NMR) spectroscopic studies. Although the three dimensional (3D) structures of GVIA [114][115][116][117][118], MVIIA [119][120][121] and MVIIC [57,122,123] have been determined several times, there is scope for improvement as developments in NMR technology and structure calculation methods overcome current limitations. Overall, ω-conotoxins have remarkably similar structures, despite considerable variance in primary sequences within the intercysteine loops [57]. It is possible to identify structural similarities amongst the ω-conotoxins to gain a generalised picture of the features that contribute to their potencies in blocking Ca V channels. However, the identical disulfide arrangement and a conserved Gly in loop 1 are insufficient to define Ca V channel selectivity, but instead provide a structural framework that allows the four hypervariable loops to display key functional groups required for receptor interaction [57]. Therefore, knowledge of the 3D structure of the ω-conotoxins together with an understanding of the critical differences amongst them, are important steps towards understanding the relationship between structure and activity for these peptides.
Structurally the ω-conotoxins are characterised by the common cysteine scaffold that stabilizes the 4-loop framework. This configuration defines the canonical ω-conotoxin fold [57], which comprises a triple-stranded β-sheet/cysteine knot motif [124]. This configuration has been observed in aqueous solution for N-type Ca V channel blockers GVIA [114][115][116][117][118], MVIIA [119][120][121] and CVID [10]. Recent studies have revealed a β-bridge formed by residues 1-2 and 14-16 [10,65]. It is possible that a saltbridge between the side chains of Lys2 and Asp14 in MVIIA, MVIIC and CVID stabilises the βbridge [10,65]. This may perhaps be replaced by a hydrogen bond between Lys2 and Asn14 in GVIA, although this has not been explicitly defined. The network of β-sheet and β-bridge provides a stable base from which critical side chains are anchored. The intervening regions are composed of β-turns. These structural regions are shown schematically for MVIIA [125] in Figure 4 A and B. The three disulfide bridges (yellow ball and stick), the β-bridge region (green arrows), β-sheet region (red arrows) and the orientation of Tyr13 (orange) can be seen. Adapted from Nielsen et al. [126].
Structures of ω-conotoxins targeting different types of calcium channels including TxVII (L-type), MVIIC (P/Q-type), and GVIA, MVIIA and CVID (N-type) are compared in Figure 5, highlighting the common fold adopted by these five peptides. There are structural similarities amongst the N-and the P/Q-type ω-conotoxins, despite their selectivity differences, supporting the notion that the ω-conotoxin macrosites on the N-and P/Q-type Ca V channels are related. The L-type Ca V channel blocker TxVII ( Figure 5A) shares the overall canonical fold but has some significant differences, most notably in loop 4. TxVII has the shortest loop 4, containing only three residues compared to MVIIA ( Figure 5D) with four residues, to MVIIC ( Figure 5B) with five residues, and to CVID ( Figure 5E) and GVIA ( Figure  5C) both with six residues in loop 4. The most notable structural difference between TxVII, MVIIA and MVIIC compared to GVIA is that the loop is oriented outwards in GVIA and downwards in the other peptides. In CVID, loop 4 is significantly different in that it curves towards loop 2, thus presenting a more globular surface. The presence of a H-bond between Gly22 in loop 4 and Lys10 in loop 2 in CVID is likely to brace loop 4 in this different orientation and is also likely to account for the enhanced stabilisation of loop 2 in CVID [10]. Such a H-bond between loops 2 and 4 has not been reported for ω-conotoxins GVIA, MVIIA, MVIIC or TxVII. The structural uniqueness of CVID may have a significant effect in determining its improved selectivity for the N-type Ca V channel over MVIIA and GVIA, since loops 2 and 4 have been shown to act in combination to influence ωconotoxin selectivity [65]. Prior to the determination of the 3D structure of CVID [10], the conformation of loop 2 of ωconotoxins was poorly defined. In MVIIA for example, the ensemble of structures contained two backbone conformations for Tyr13, where this residue was found to lie in different regions of the Ramachandran plot. Other residues in this loop had broadened NH resonances suggesting that conformational averaging was indeed occurring. In GVIA, a similar situation was encountered, although it was believed that a single conformation of the Tyr13 sidechain predominated [116]. A similar situation was reported for MVIIC, based on structure calculations, despite the fact that this was not clear from the raw data, and again, linewidths in loop 2 were broadened with respect to the remainder of the peptide [57]. In contrast, examination of the CVID NMR data provided convincing evidence that the Tyr13 sidechain adopted a single conformation, unusual for such a bulky residue, on the Ramachandran plot. In addition, broadening of NH resonances in loop 2 were reduced or not apparent [10].
The role of structure in structure-activity relationships Many ω-conotoxin SAR studies have neglected the structural component, instead assuming that this is unaffected by residue replacement. This assumption may represent a serious drawback to these studies, as it is not possible to distinguish between structural change and removal of a binding interaction for large ∆pIC 50 values. As enthalpic and entropic terms contribute equally to the free energy of binding (i.e., ∆G = ∆H -T∆S), omission of effects of entropy may lead to misinterpretation of binding results. Regrettably, most of the studies on MVIIA have not been accompanied by structural data [64,84,128,129] and many of these analogues may need to be re-examined before any specific sidechain effects can be fully interpreted. The first attempt to introduce structural studies in ωconotoxin SAR was by Kim et al. [130], where circular dichroism (CD) spectroscopy was used to examine the structures of the analogues. A significant improvement on this is the treatment of Lew et al. [131], where NMR spectroscopy has been used to determine the Hα and NH chemical shifts for the series of GVIA analogues. Unfortunately, the structural affects of several Ala-replacements were not further investigated. Due to the accessibility of high-field NMR equipment these days, the structural component of SAR is becoming more and more carefully evaluated. This approach will help improve the accuracy of the analysis of residues important for binding to a target, since residues important for structural stabilisation of a molecule will be omitted from the pharmacophore.

ω-Conotoxin residues important for binding to the N-type Ca V channel
To determine which residues were important for the binding and activity of ω-conotoxins to the Ntype Ca V channel, Ala-replacement studies have been widely used [64,[129][130][131][132]. These studies have unequivocally shown Tyr13 to be the single most important residue, with the hydroxyl moiety being the key binding determinant. In addition, several other residues, not always conserved across the ωconotoxins, also have an effect on potency when replaced by alanine. In MVIIA, these include Lys2, Arg10, Leu11 and Arg21 [64], while in GVIA these residues are Lys2, Arg17, Tyr22 and Lys24 (functional assays only) [133]. Smaller effects are observed for Hyp10, Hyp21 and Arg25 in GVIA [133]. In both peptides, N-acetylation of the N-terminus results in a significant drop in potency [64,133]. Deamidation of the C-terminus in GVIA and MVIIA also reduces potency [133].
In Figure 6A,B, the structure of GVIA is shown with particular reference to the side chains that are reputedly important for function. For comparison, the structure of MVIIA is shown below ( Figure  6C,D), with residues important for binding highlighted. Unfortunately, any structural effects of replacing these residues are not known for MVIIA, and are not fully analysed for GVIA.
For CVID, a specific alanine scan has not been conducted, but instead several residues have systematically been replaced with residues of either similar character, such as a Lys to Arg replacement or with a residue believed to cause a clash, such as a Gly to Tyr replacement (Lewis et al., unpublished results). As reported for GVIA and MVIIA, residues located in loop 2, such as Lys10, Leu11 and most importantly Tyr13 of CVID are crucial for binding. The other loops are more resistant to changes in activity following residue replacement.
Analysis of SAR of ω-conotoxin GVIA, MVIIA and CVID at the N-type Ca V channel has lead to the development of a couple of ω-conotoxin pharmacophores that can help guide the rational development of N-type Ca V channel inhibitors [126,131]. Initial ω-conotoxin pharmacophores included Lys2 as a major binding determinant. Lys2 is located in loop 1 on the opposite side of the peptide compared to the other important residues located in loop 2. However, recent evidence in our laboratory suggest that Lys2 is not directly involved in binding to the receptor but instead is involved in an important stabilising interaction between Lys2 and residue 14 (see above) (Schroeder et al., unpublished results). A novel minimal pharmacophore was subsequently proposed based on CVID including only residues in loop 2; Lys10, Leu11 and Tyr13 [134]. When designing a pharmacophore it is important to pinpoint which residues that are involved in a direct interaction with the receptor and which are structurally important to facilitate the design and maximise the chance of novel inhibitor being active. These pharmacophores have been utilised in the search for the development of small molecule to block the N-type Ca V channel [134,135]. Whilst these small molecules still only display modest activity, they were found to retain selectivity for the N-type Ca V channel over the undesired P/Q-type Ca V channel and are consequently promising candidates for further development in the search for novel therapies for the treatment of pain. There has never been a more critical need for novel therapeutics for pain management with a number of Cox-2 inhibitors withdrawn from the market [56]. While Prialt, the first FDA approved conotoxin, displays some undesired side effects, the next generation ω-conotoxin AM336 with a larger therapeutic window shows that N-type calcium channel inhibitors can be developed with fewer side effects. Furthermore, the behavioural studies and extensive SAR conducted on the ω-conotoxins over the last two decades now places the medicinal chemists in a strong position to design novel small molecules for the treatment of intractable pain.