Abstract The distribution of ion channels in neurons associated with pain pathways is becoming better understood. In particular, we now have insights into the molecular nature of the channels that are activated by tissue-damaging stimuli, as well as the mechanisms by which voltage-gated channels alter the sensitivity of peripheral neurons to change pain thresholds. This chapter details the evidence that individual channels may be associated with particular pain states, and describes genetic approaches to test the possible utility of targeting individual channels to treat pain.
Keywords Voltage-gated sodium channels · Calcium channels · Potassium channels · HERG channels · TRP channels · Transgenic mice
Because we know the entire sequence of the human genome we can now predict, clone, express and characterise all known and predicted ion channels and accessory subunits using heterologous expression systems. Molecular probes also enable us to deﬁne patterns of expression, altered transcriptional proﬁles in disease states, and splice variants. Armed with this information we can make an educated guess at which channels are likely to play an important role in somatosensation and pain pathways. Again, gene manipulation allows us to test the possible role of candidate target genes by means of small interfering (si)RNA blockade of translation or transgenic mouse studies. Thus, the post- genomic era is providing unprecedented technology and opportunities for pharmacologists to target subsets of cells and individual subtypes of receptors in order to understand more about and treat disease.
In the case of voltage-gated ion channels, an analysis of the relationship between the pore regions of expressed mammalian voltage-gated ion channels has resulted in the following useful analysis of the ‘chanome’ (Fig. 1) by Yu and Catterall (2004). Note the enormous diversity of potassium channels and the relationship between voltage-gated sodium, calcium, potassium and TRP (transient receptor potential) channels—receptors that have been implicated in a variety of sensory modalities and pain conditions. In this review we will con- centrate on voltage-gated sodium and potassium channels, calcium channels, TRPs, acid-sensing ion channels (ASICs) and other channel subtypes, their role in pain transduction, and thus their potential as analgesic drug targets.
Although the global role of voltage-gated channels in electrical signalling might make one imagine that they would be unattractive as analgesic drug targets, recent studies have identiﬁed a selective role for subsets of sodium, calcium and potassium channels in setting pain thresholds. Thus, microarray analyses of altered gene expression in sensory neurons and the spinal cord following nerve injury have shown a number of altered sodium, calcium and potassium channel transcripts that may be associated with the pathophysiology of chronic pain (Xiao et al. 2002; Costigan et al. 2002; Wang et al. 2003). Consistent with this, drugs targeting sodium (mexiletine), calcium channels (ziconotide) and potassium channels (retigabine) have all been found to have analgesic actions either in man or animal models (or both).
Recapitulating inﬂammatory pain and the human peripheral nerve injuries that lead to neuropathic pain in animal models has proved extremely useful
for mechanistic studies. Most models focus on partial nerve injury to sciatic or sural nerves, which allows altered hind-limb pain sensitivity to thermal and mechanical insults to be measured and compared with the uninjured contralateral paw.
A model of rodent neuropathic pain (Bennett model) relies upon the tight ligation of the sciatic nerve using thread soaked in chrome alum, which also
Fig. 1 The amino acid sequence relationship of pore regions in seven voltage-gated ion channel families. (For full details see Yu and Catterall 2004 and Yu et al. 2005)
imparts a low pH insult to the local tissue (Bennett et al. 1992). Kim et al. (1997) ligated tightly both L5 and L6 (or L5 alone) spinal nerves (Chung model), and compared the evoked behaviour with that found in the Bennett model. Both thermal and mechanical thresholds were affected, with a long-lasting hyperalgesia to noxious heat (at least 5 weeks) and mechanical allodynia (at least 10 weeks) of the affected foot. In addition there were behavioural signs of the presence of spontaneous pain. Seltzer (1990) ligated about half of the sciatic nerve close to the spinal cord. Within a few hours the rats developed guarding and licking behaviour of the ipsilateral hind paw, suggesting the presence of spontaneous pain. This continued for many months. There was a decrease in the withdrawal thresholds in response to repetitive Von Frey hair stimulation at the plantar side. Allodynia and mechanical hyperalgesia were apparent in this model, as was thermal hyperalgesia.
The spared nerve injury model (Decosterd and Woolf 2000) involves a lesion of two of the three terminal branches of the sciatic nerve (tibial and common peroneal nerves), leaving the remaining sural nerve intact. The spared nerve injury model is unique in restricting contact between intact and degenerating axons, allowing behavioural testing of the non-injured skin territories next to the denervated areas.
Erichsen and Blackburn-Munro (2002) investigated the pharmacological sensitivity of the sciatic nerve ligation (SNI) model to analgesic drugs, mea- suring reﬂex nociceptive responses to mechanical and cold stimulation after systemic administration of opioids, sodium channel blockers and other drugs. They found that morphine attenuated mechanical hypersensitivity and cold hypersensitivity. The sodium-channel blocker mexiletine relieved both cold allodynia and mechanical hyperalgesia, but the most distinct and prolonged effect was observed on mechanical allodynia. Gabapentin alleviated mechani- cal allodynia but had no effect on mechanical hyperalgesia.
The Chung model has also been characterised in terms of responses to different type of analgesic drugs that have some utility in the clinic (Abdi et al.
1998). Amitriptyline (an anti-depressant which also blocks sodium channels), gabapentin and lidocaine (a local anaesthetic) were effective in increasing the threshold for mechanical allodynia Amitriptyline and lidocaine reduced the
rate of continuing discharges of injured afferent ﬁbres, although gabapentin did not inﬂuence these discharges, presumably acting centrally. Analgesic drugs found serendipitously in the clinic (e.g. gabapentin) are effective in
animal models, thus giving conﬁdence to the conclusions drawn from animal experimentation.
Voltage-gated sodium channels comprise a family of ten structurally related genes that are expressed in spatially and temporally distinct patterns in the mammalian nervous system. Sodium channel blockers which act as anaesthet- ics at high concentrations are also well know for their analgesic actions at lower concentrations (Strichartz et al. 2002). Local anaesthetics seem to bind to and block sodium channels in a state-dependent manner. Thus, activation of the channels results in increased accessibility for these agents via the cytoplasmic vestibule of the channel, so that channels that are ﬁring at high frequency are selectively inhibited (Fozzard et al. 2005). Such mechanisms are likely to oper- ate in pain states where nociceptors are activated and are selectively targeted by low doses of local anaesthetics. Evidence of a role for sodium channels in various chronic pain situations has come from studies of neuronal excitability, analysis of patterns of expression of channel isoforms in animal models of neuropathic pain, and antisense and knock-out studies. Three sodium chan-
nels, Nav1.7, Nav1.8 and Nav1.9, are selectively expressed within the peripheral nervous system, and these particular isoforms have attracted attention as anal- gesic drug targets. In addition, an embryonic channel Nav1.3 and a β-subunit, β-3, have been found to be up-regulated in dorsal root ganglion (DRG) neurons in neuropathic pain states.
All sodium channel α-subunits consist of four homologous domains that form a single, voltage-gated aqueous pore. The α-subunits are greater than 75%
identical over the amino-acid sequences comprising the transmembrane and extracellular domains. The α-subunits show distinct patterns of expression, and are associated with accessory β-subunits which modify channel properties and interact with cytoskeletal and extracellular matrix proteins. Despite the broadly similar properties of voltage-gated sodium channels, there is evidence for a specialised role of the various isoforms, highlighted for example by the
different phenotypes of null mutant mice. For example, Nav1.2 sodium channel knock-outs show severe hypoxia as a result of brain-stem apoptosis, and the Nav1.8 channel plays a specialised role in pain pathways. Much attention has focussed on the regulation of expression of sodium channels in peripheral no- ciceptive neurons and the functional changes that are associated with various pain states (Fig. 2; Table 1).
Voltage-gated sodium channels provide the inward current that generates the upswing of an action potential in response to supra-threshold depolarisa- tions of the membrane potential. At present, Nav1.1 to Nav1.9 have been char- acterised as voltage-gated sodium channels (Goldin 2001). A further sodium
Fig. 2 The α- and β-subunits of Nav channels. Sites at which Nav channels are regulated are highlighted. S554, S573, S610, S623, S687 and S1506 residues within Nav channels are phosphorylated by either protein kinase A (PKA) or PKC (or both ;see main text for further information). The arrows indicate possible interactions of different regions of the β-subunit with the α-subunit. Abbreviations: An II, annexin II; AnkG, ankyrin G; CaM, calmodulin; Gβγ,G protein β–γ complex; Ub, ubiquitin. (From Chahine et al. 2005)
channel α-subunit has been cloned that lacks the amino acid sequence required for voltage gating. Watanabe et al. (2000) identiﬁed NaX in the body-ﬂuid reg- ulating circumventricular organs of the brain. Behavioural data revealed that knock-out animals, when dehydrated, drink 300 mM NaCl solution in contrast to the avoidance behaviour demonstrated by dehydrated wild-type mice, lead- ing Watanabe et al. (2000) to propose a role for NaX in regulating salt intake behaviour. In support of this hypothesis, more recent data from the same group demonstrated that NaX is gated by extracellular sodium concentration, with approximately 160 mM sodium producing a half maximal response (Hiyama et al. 2002).
Although the α-subunit alone is sufﬁcient for the formation of a functional channel, the accessory β-subunits increase the efﬁciency of channel expression
and are required for normal kinetics and voltage dependence of channel gating. In addition, β-subunits have an important role in the localisation of α-subunits (Malhotra et al. 2000).
Sodium channel α-subunits have a membrane topology consisting of a large intracellular N-terminal domain, four repeated homologous domains (DI, DII, DIII and DIV) containing six transmembrane regions (S1 to S6), a large intra-
cellular loop between DII and DIII involved in sodium channel inactivation and a short intracellular C-terminus.
The loops between transmembrane region 5 and 6 (S5–S6 loop) contribute
residues to the formation of the channel pore and are longer in DI and DIII than DII or DIV. The S5-S6 loop of DI provides residues important for the binding of the pore-blocking toxins saxitoxin (STX) and tetrodotoxin (TTX).
The accessory β-subunits are less well-conserved and signiﬁcantly smaller, consisting of a large, immunoglobulin-like extracellular N-terminal domain
required for functional expression and modulation of α-subunits, a single
transmembrane region involved in modulating the voltage dependence of α- subunit steady state inactivation and a short intracellular C-terminal domain. Other accessory subunits that may modulate channel function have been re- viewed (Malik-Hall et al. 2003).
Voltage-gated sodium channels can be pharmacologically classiﬁed into two main groups according to their sensitivity to TTX. Nav1.1, 1.2, 1.3, 1.4, 1.6 and
1.7 are sensitive to TTX and are thus termed TTX-S, whereas Nav1.5, 1.8 and 1.9
are relatively resistant to TTX and are thus termed TTX-R. An example of these
two classes of current comes from whole cell electrophysiological recordings of dissociated adult rat DRG neurons. Here TTX has a Kd for TTX-S currents of
approximately 300 pM compared to 100 μM for TTX-R currents composed of
Nav1.8 and Nav1.9. Nav1.5, although responsible for a TTX-R current, is more sensitive to TTX (IC50 1 uM) than either Nav1.8 or Nav1.9.
TTX-R currents are mainly restricted to embryonic and denervated mus-
cle and heart muscle (Nav1.5) or small-diameter DRG neurons (Nav1.8 and Nav1.9). Nav1.8 and Nav1.9 are both expressed in sensory neurons and seem to play an import role in nociception and the setting of pain thresholds. Nav1.5 has also been identiﬁed in a small number of sensory neurons. In experimental diabetes there is a signiﬁcant up-regulation of messenger (m)RNA and protein for Nav1.3, and Nav1.7 and a down-regulation of Nav1.6 and Nav1.8 mRNA after the onset of allodynia (Hong et al. 2004). There is also an altered pattern of channel phosphorylation—the level of serine/threonine phosphorylation of Nav1.6 and In Nav1.8 increased in response to diabetes. Increased tyrosine phosphorylation of Nav1.6 and Nav1.7 was also observed in DRGs from dia- betic rats. In most neuropathic pain models exempliﬁed by axotomy, however, Nav1.8 and 1.9 are down-regulated, whilst Nav1.3 is induced (Waxman et al. 1994; Dib-Hajj et al. 1996, 1998).
Nav1.3 is widely expressed in the adult CNS but is normally present at low levels in the adult peripheral nervous system. Axotomy or other forms of nerve damage lead to the re-expression of Nav1.3 and the associated β-3 subunit in sensory neurons, but not in primary motor neurons (Waxman et al. 1994; Dib-Hajj et al. 1996; Hains et al. 2002). This event can be reversed in vitro and in vivo by treatment with high levels of exogenous glial-derived neurotrophic
factor (GDNF). Nav1.3 is known to recover (reprime) rapidly from inactivation (Cummins et al. 2001). Axotomy has been shown to induce the expression of rapidly repriming TTX-sensitive sodium channels in damaged neurons, and this event can also be reversed by the combined actions of GDNF and nerve growth factor (NGF) (Lefﬂer et al. 2002). Concomitant with the reversal of Nav1.3 expression by GDNF, ectopic action potential generation is diminished and thermal and mechanical pain-related behaviour in a rat chronic constric-
tive injury (CCI) model is reversed (Boucher et al. 2000). Moreover, Nav1.3 is up-regulated in multi-receptive nociceptive dorsal horn neurons following experimental spinal cord injury. This up-regulation is associated with hyper- excitability of these nociceptive neurons and pain; antisense knock-down of Nav1.3 attenuates the dorsal horn neuron hyperexcitability and the pain be- haviours in spinal-cord injured animals (Hains et al. 2003). It therefore seems plausible that Nav1.3 re-expression may play a signiﬁcant role in increasing neuronal excitability, thus contributing to neuropathic pain after nerve and spinal cord injury.
Nav1.7, a peripheral nervous system-speciﬁc sodium channel isoform was ﬁrst cloned from the pheochromocytoma PC12 cell line. Its presence at high levels in the growth cones of small-diameter neurons suggests that it is likely to play some role in the transmission of nociceptive information. Immunochemical studies of functionally deﬁned sensory neurons in guinea-pigs supports the view that Nav1.7 is associated with nociceptors. Nav1.7 is expressed exclu- sively in peripheral, mainly small-diameter sensory and sympathetic neurons. Nav1.7 global gene deletion leads to death shortly after birth, apparently as a consequence of failure to feed. This may reﬂect autonomic or enteric sensory neuron dysfunction.
Deleting the gene in a subset of sensory neurons that are predominantly nociceptive demonstrates that Nav1.7 plays an important role in pain mecha- nisms, especially in the development of inﬂammatory pain. The speciﬁc deﬁcits in acute mechano-sensation, rather than thermal sensitivity associated with the Nav1.7 deletion, are striking. However, as TTX does not block mechani- cally activated currents in sensory neurons, Nav1.7 must play a downstream role in mechano-transduction. Nav1.8-nulls are also refractory to Randall– Sellito-induced mechanical insults. These data raise the possibility that sodium channels and mechano-sensitive channels are physically apposed at nociceptor terminals.
The roles of Nav1.7 and 1.8 in pain pathways may be related. In the Nav1.8- null mouse, NGF-induced thermal hyperalgesia is diminished, and there are small deﬁcits in other inﬂammatory pain models. However, in the Nav1.8-null, an up-regulation of Nav1.7 mRNA has been noted, as well as an increase in TTX- S sodium current density. Thus, Nav1.7 may compensate for the depletion of Nav1.8. In contrast, deletion of Nav1.7 is not compensated for by increased TTX- R current activity and leads to a dramatic phenotype in terms of inﬂammatory pain.
Peripheral changes in pain thresholds seem to involve several mechanisms including channel phosphorylation. NGF acting through TrkA-mediated ac- tivation of phospholipase C relieves phosphatidylinositol-4,5-bisphosphate
[PtdIns(4,5)P2] channel block of TRPV1; p38 mitogen-activated protein (MAP) kinase also plays a role in NGF-induced hyperalgesia. Carrageenan increases prostanoid levels that, acting through EP receptors, cause sodium channel phosphorylation involving protein kinase A (PKA). Complete Freund’s adju- vant induces longer-term changes that are partially blocked by aspirin-like drugs. Remarkably, all these forms of inﬂammatory hyperalgesia are attenu- ated by deletion of Nav1.7.
A number of possible mechanisms involving Nav1.7 regulation may occur.
Nav1.7 is unlikely to be a target for kinase modulation in inﬂammatory pain;
in vitro studies in Xenopus oocytes demonstrate a diminution in peak current
density in response to PKA or protein kinaseC (PKC) activation. Nav1.7 con- trasts with Nav1.8, which shows increased peak current and an altered current voltage relationship consistent with nociceptor sensitisation and is relatively faster to reprime. The regulation of Nav1.7 channel density, or localisation with respect to primary signal transducers at nociceptor terminals, could play a signiﬁcant role in setting peripheral pain thresholds. NGF has already been shown to increase excitability of PC12 cells through increasing expression of Nav1.7.
Primary erythermalgia, a chronic inﬂammatory disease that is inherited in
a dominant form in man, maps to gain of function mutation in Nav1.7 that lead to a lowered threshold of activation, demonstrating the utility of mouse models in understanding human pain conditions (Drenth et al. 2005).
Nav1.8 is mainly expressed in nociceptive neurons (Akopian et al. 1996; Djouhri et al. 2003). This channel contributes a majority of the sodium current underly- ing the depolarising phase of the action potential in cells in which it is present (Renganathan et al. 2001). Functional expression of the channel is regulated by inﬂammatory mediators, including NGF, and both antisense and knock-out studies support a role for the channel in contributing to inﬂammatory pain (Khasar et al. 1998; Akopian et al. 1999). Antisense studies have also suggested a role for this protein in the development of neuropathic pain (Lai et al. 2002), and a deﬁcit in ectopic action propagation has been described in the Nav1.8- null mutant mouse (Roza et al. 2003). However, neuropathic pain behaviour at early time points seem to be normal in the Nav1.8-null mutant mouse (Kerr et al. 2000), and studies of double knock-outs of Nav1.7 and Nav1.8 also demon- strate a normal neuropathic pain phenotype using the Chung model (Nassar et al. 2005).
Identiﬁcation of annexin II/p11, which binds to Nav1.8 and facilitates the insertion of functional channels in the cell membrane (Okuse et al. 2002), may provide a target that can be used to modulate the expression of Nav1.8 and hence the level of Nav1.8 current in nociceptive neurons.
Nav1.9 is also expressed in nociceptive neurons (Dib-Hajj et al. 1998; Fang et al. 2003) and underlies a persistent sodium current with substantial overlap between activation and steady-state inactivation (Cummins et al. 1999) that has a probable role in setting thresholds of activation (Dib-Hajj et al. 2002; Baker et al. 2003), suggesting that blockade of Nav1.9 might be useful for the treat- ment of pain. Conversely, it has been suggested that Nav1.9 activators might alleviate pain because Nav1.9 is down-regulated after axotomy (Dib-Hajj et al.
1998; Cummins et al. 2000); the resultant loss of the Nav1.9 persistent cur-
rent and its depolarising inﬂuence on resting potential (Cummins et al. 1999)
might remove resting inactivation from other sodium channels (Cummins and Waxman 1997). Nav1.9 null mutants show normal levels of neuropathic pain but have deﬁcits in inﬂammatory pain. Normal level of expression seems to be dependent on the supply of NGF or GDNF (Cummins et al. 2000). Present evidence thus makes sodium channels highly attractive analgesic drug targets, but speciﬁc antagonists for Nav1.3, 1.8 and 1.9 have yet to be tested in the clinic.
What determines the cell type expression of sodium channels in nocicep-
tors? Although we do not have a comprehensive knowledge of any sodium channel promoter’s structure and association with a particular transcription factor(s), we do have a number of insights into some aspects of sodium chan- nel regulatory motifs. A short sequence found upstream of neuronal sodium channel genes named NRSE (neuron restricted silencing element) or RE-1 (repressor element 1) (Schoenherr and Anderson 1995; Kraner et al. 1992) regulates neuronal expression. Transcription factors that bound to the motif were found to act as inhibitors of gene expression in non-neuronal cells. These proteins were named REST (RE-1 silencing transcription factor), or NRSF (neuron-restrictive silencer factor). The inhibitory activity of the complex can be modulated by double stranded (ds)RNA molecules that have the same se- quence as NRSE/RE-1 and are found in developing neuronal precursors. These regulatory RNA molecules are able to switch the repressor function of the complex to an activator role (Kuwabara et al. 2004). In this way the assump- tion of a neuronal phenotype seems to depend in part upon regulatory RNAs driving gene expression downstream of NRSE/RE-1 motifs. Sodium channels are known to be expressed at the very earliest stages of the appearance of a neuronal phenotype in the mouse. These studies highlight the signiﬁcance of sodium channel expression in neuronal function throughout development (Albrieux et al. 2004).
Apart from the tissue-speciﬁc control of sodium channel expression most obviously demonstrated by the presence of neuronal and muscle isoforms, there is evidence that the relative levels of sodium channel transcripts vary in development. Early studies of the developing rat gave us the ﬁrst indication that the type III sodium channel is prevalent in rat embryos and expressed at much lower levels in adult neuronal tissues, whilst the types 1 and 2 are expressed in variable patterns in adult tissues (Beckh et al. 1989). Interestingly, this pattern of expression does not seem to hold true in cynomolgus monkeys where Nav1.3
is broadly expressed, albeit at low levels in both central and peripheral tissues in the adult (Raymond et al. 2004). Thus, the pattern of expression of human sodium channels may vary markedly from that described in detail in rodents. Felts et al. (1997) extended the rat development analysis with probes against Nav1.1, 1.2, 1.3, 1.6 and 1.7 and showed a complex pattern of developmentally regulated channel expression in both peripheral and central neurons.
Accessory subunits that are assumed to be uniquely associated within voltage-gated sodium channels also show developmentally regulated patterns of expression. Shah et al. (2001) showed that in the developing rat the β-3 sub- unit was prevalent, and this subunit remained expressed in adult hippocampus and striatum. β-1 and -2 subunits were expressed after postnatal day 3 in the rat in central and peripheral tissues. The developmental pattern of expression of the β-2-like subunit β-4 (Yu et al. 2003) that is also expressed both centrally and peripherally has yet to be described. The functional signiﬁcance of β- subunit expression for sodium channel kinetics properties and their tethering to extra-cellular signalling molecules has been explored extensively.