A further level of complexity in the expression of sodium channels is created by the existence of splice variants of sodium channel α-subunits that may be regulated during development and by regulators of splicing choice in the adult. Because ﬂy genetics is so advanced, more information is available about
Drosophila splice choice sodium channel variants and their functional roles than the vertebrate equivalents. However, it seems reasonable to suppose that alternative splicing and RNA editing and transport may also have roles in regulating mammalian sodium channel function. In Drosophila, a mutation in a dsRNA helicase led to a lowering of expression of Para-encoded sodium channels. Reenan et al. (2000) showed that this was due to a failure to edit the Para transcript with adenosine to inosine substitutions, which apparently required the helicase for secondary structure modiﬁcation of the mRNA tran- script. At least three positions in the Para transcript are known to be edited (Hanrahan et al. 2000) by adenosine deaminase to give A to I substitutions, and these events are developmentally regulated.
The editing process requires a complementary sequence of intronic RNA
to form secondary structure with the edited sequence in a similar manner to that demonstrated for mammalian glutamate receptors. Other editing events in cockroach sodium channels and Drosophila para have been correlated with dramatic functional changes. Liu et al. (2004) have shown that a U to C edit- ing event resulting in a phenylalanine to serine modiﬁcation can produce a sodium channel with persistent TTX-sensitive properties rather similar to currents identiﬁed in mammalian CNS neurons, raising the possibility that
similar events could occur in mammals. Song et al. (2004) have catalogued fur- ther editing events that have functional consequences in terms of thresholds of activation and are developmentally regulated in the cockroach. Tan et al. (2002) have also found that alternatively spliced transcripts can have distinct pharmacological proﬁles as well as altered gating characteristics. They found alternative exons encoding transmembrane segments in DIII of a cockroach sodium channel, which had conserved splice sites across evolution in ﬁsh, ﬂies, mice and men. The alternatively spliced forms were found in different tissues. One form with a premature stop codon occurred only in the peripheral nervous system whilst the two other functional forms differed in their sensi-
tivity to pyrethroid insecticides such as δ-methrin. Remarkably, fetal mouse brain also contains transcripts of the SCN8A gene (Nav1.6) that contains a stop codon at the same site as the ﬂy genes predicting the production of two domain
truncated sodium channel transcripts (Plummer et al. 1997). The role of these transcripts is unknown.
In mammals, mutually exclusive exon usage also occurs. The type III channel
exists as an embryonic or adult spliced form with different exons that code for the S3 and S4 segments in DI of the rat channel. Despite the fact that the two exons both encode 29 amino acids, only a single amino acid residue is altered in these two alternative forms. Single amino acid changes may also occur through alternative 3 splice site selection. Kerr et al. (2004) have found that both Nav1.8 and Nav1.5—two TTX-resistant sodium channels found in peripheral neurons—both exist as alternative forms containing an additional glutamine residue within the cytoplasmic loop linking DII and DIII of these channels. As well as alternative exon usage or amino acids insertions, transcripts encoding exon repeats have been identiﬁed in DRG neurons. The presence of a transcript with a three-exon repeat encoding Nav1.8 is enhanced by treatment with NGF, suggesting that this neurotrophin may regulate trans-splicing events in these cells (Akopian et al. 1999). Once again the functional consequences of these conserved changes have yet to be established.
The regulation of splice choice in response to external signals is still little understood. Buchner et al. (2003), studying a modiﬁer locus in different mouse lines that determines the lethality of a Nav1.6 splice site mutation, discovered that the efﬁciency of action of a splice factor determined the amount of func- tional channel produced and hence the lethality of the original mutation. Thus, on a C57Bl6 background little correctly spliced mRNA was produced, causing a lethal phenotype, whilst on a wild-type background, 10% of the transcripts were correctly spliced, leading to a viable if dystonic phenotype.
Recent papers have suggested roles for sodium channels in regulating synap- tic efﬁcacy, as well as functions in immune system cells. Macrophages and microglia express Nav1.6, a channel that is broadly expressed in the nervous system and which is functionally compromised in the naturally occurring med mutant that leads to dystonia. Interestingly, macrophage function is also in- hibited in these animals. When microglia or macrophages are activated, Nav1.6
expression is up-regulated, and this event seems to be important in terms of phagocytic activity, as the uptake of latex beads is partially blocked in med macrophages, or in normal macrophages treated with TTX. This suggests that voltage-gated sodium channels play an important functional role in immune system cell function. These kinds of unsuspected roles for voltage-gated chan- nels may give rise to signiﬁcant problems in attempts to use selective blockers as analgesic in chronic pain states.
Voltage-gated calcium channels comprise a single α-subunit and show strong structural homology with sodium channels, but the accessory subunits associ- ated with these channel are much more complex. The functional calcium chan- nel complexes contain four proteins: α1 (170 kDa), α2 (150 kDa), β (52 kDa), δ (17–25 kDa) and γ (32 kDa). Four α2δ subunit genes have now been cloned.
Both the message and protein for the α2δ-1 subunit is highly expressed in
sensory neurons but is also found almost ubiquitously in other tissues (Gong et al. 2001). All α2δ subunits have a predicted N-terminal signal sequence, in- dicating that the N-terminus is extracellular, with an intracellular C-terminus and potential transmembrane region. There are up to 14 conserved cysteines throughout the α2δ-1, 2 and 3 sequences, six of which are within δ, providing
additional evidence that α2 and δ are disulphide-bonded. Following the iden-
tiﬁcation of α2δ subunits as components of skeletal muscle calcium channels, they have also been shown to be associated with neuronal N- and P/Q-type channels. The α2δ-1 subunit has been shown to bind to extracellular regions
including DIII on the Cav1.2 subunit (Felix et al. 1997).
Voltage-Gated Calcium Channels
High voltage-activated and low voltage-activated calcium channels are now known to comprise a number of cloned α-subunits (Peres-Reyes 2003). Cav2.2, Cav2.3 and Cav3.2 are now known to play an important role in pain sensation and are interesting analgesic drug targets.
The evidence of an important specialised role for certain voltage-gated cal- cium channels in the pathogenesis of neuropathic pain is strong. A variety of drugs targeted at calcium channel subtypes are effective analgesics, and mouse null mutants of N-type Cav2.2 calcium channels show dramatic diminution in neuropathic pain behaviour in response to both mechanical and thermal stim- uli. Opioid peptides are known to inhibit the action of N-type calcium channels through post-translational mechanisms. In addition, two highly effective anal-
gesic drugs used in neuropathic pain conditions selectively target calcium channel subtypes. The conotoxin ziconotide blocks Cav2.2 α-subunits, and the widely prescribed drug gabapentin binds with high afﬁnity to α2δ subunits of calcium channels.
Gabapentin binds to high-afﬁnity sites in the brain, and the target bind- ing site has been identiﬁed as the α2δ-1 subunit. Transient transfection of cells with α2δ-1 increased the number of gabapentin binding sites (Gee et al.
1996). Subsequently, gabapentin has been found to bind to two isoforms of α2δ
subunits (the α2δ-1 and α2δ-2 isoforms, but not α2δ-3 or α2δ-4) (Gee et al.
1996; Gong et al. 2001). The effects of gabapentin on native calcium currents
are controversial, with some authors reporting small inhibitions of calcium currents in different cell types. Gabapentin could interfere with α2δ binding to the α1 subunit, thus destabilizing the heteromeric complex. Interestingly, α2δ1 up-regulation in neuropathic pain correlates well with gabapentin sensitivity (Luo et al. 2002), suggesting that the α2δ-1 isoform is the most likely site of action of gabapentin.
The up-regulation of α2-δ subunits does not occur in all animal models of neuropathic pain that result in allodynia. Luo et al. (2002) compared DRG and spinal cord α2δ-1 subunit levels and gabapentin sensitivity in allodynic rats with mechanical nerve injuries (sciatic nerve chronic constriction in-
jury, spinal nerve transection, or ligation), a metabolic disorder (diabetes) or chemical neuropathy (vincristine neurotoxicity). Allodynia occurred in all types of nerve injury investigated, but DRG and/or spinal cord α2δ-1 sub- unit up-regulation and gabapentin sensitivity only co-existed in mechanical
and diabetic neuropathies. This may partially explain why gabapentin is in- effective in some neuropathic pain patients. Recent studies with knock-in mice have demonstrated that gabapentin loses its effectiveness when it can
no longer bind to a mutated form of α2δ-1, and the details of this work can be found in a patent application (Baron et al. 2005). Thus, the site of action of gabapentin/pregabalin seems resolved, but the mechanism of action is still
Further support for calcium channels as useful drug targets in neuropathic pain comes from an analysis of the characteristics of the Cav2.2-null mutant mouse generated by Saegusa et al. (2001). The same authors (Saegusa et al.
2002) have compared the Cav2.2- and 2.3-null mutants in a variety of pain models. Despite the widespread expression of Cav2.2, it has proved possible to demonstrate major deﬁcits in inﬂammatory and in particular neuropathic pain in this transgenic mouse using the Seltzer model. Thermal and mechanical thresholds were dramatically stabilized in this mutant mouse. The relationship between α2δ subunits and Cav2.2 has not been investigated in detail, so it is possible that gabapentin has sites of action on calcium channels other than
Cav2.2—for example, T-type channels such as Cav3.2.
A role for Cav2.2 in chronic pain is consistent with a known analgesic role
for N-type calcium channel blockers. Ziconotide, a toxin derived from marine
snails, blocks Cav2.2 channels with high afﬁnity and has been found to have analgesic actions in animal models and man. In a study of the anti-nociceptive properties of ziconotide, morphine and clonidine in a rat model of post- operative pain, heat hyperalgesia and mechanical allodynia were induced in the hind paw (Wang et al. 2000). Intrathecal ziconotide blocked established heat hyperalgesia in a dose-dependent manner and caused a reversible blockade of established mechanical allodynia. Intrathecal ziconotide was found to be more potent, longer acting, and more speciﬁc in its actions than intrathecal morphine in this model of post-surgical pain.
Brose (1997) found that intrathecal ziconotide provided complete pain relief with elimination of hyperaesthesia and allodynia in a dose dependent manner in a single patient suffering from intractable pain, although side-effects were experienced. This type of study has led to clinical use of ziconotide in the treatment of intractable pain in late stage cancer patients. Evidence that omega conotoxins also target P2X3 receptors (Lalo et al. 2001) suggests that ziconotide may act at a broader range of targets than ﬁrst suspected.
Interestingly, a sensory neuron-speciﬁc splice variant of Cav2.2 has been identiﬁed in DRG neurons. Lipscombe and collaborators (Bell et al. 2004) showed that the DRG-speciﬁc exon, e37a, is preferentially present in Cav2.2 mRNAs expressed in neurons that contain nociceptive markers TRPV1 and Nav1.8. Cell-speciﬁc inclusion of e37a correlated with the signiﬁcantly larger N-type currents in nociceptive neurons, suggesting that splice-variant speciﬁc antagonists could have useful therapeutic effects. The situation in man has yet to be examined, however.
Cav3.2, a T-type calcium channel has also recently been found to play a role in pain. Antisense targeting Cav3.2 induced a knock-down of the Cav3.2 mRNA and protein expression as well as a large reduction of T-type calcium currents in nociceptive DRG neurons. Concomitantly, the antisense treatment resulted in major anti-nociceptive, anti-hyperalgesic and anti-allodynic effects, suggesting that Cav3.2 plays a major pronociceptive role in acute and chronic pain states. These antisense studies have also been conﬁrmed by the generation and analysis of a knock-out mouse (Bourinet and Zamponi 2005).
Potassium channels act effectively as brakes on neurotransmission and ex- citability, allowing the ﬂow of potassium ions out of the cell in response to a range of different stimuli. For example, some analgesic actions of opioids are mediated by the activation of calcium-dependent potassium channels. Potassium channels have been subdivided into four major subsets based on their structure and mode of gating. Voltage-gated channels are extraordinarily diverse—there are 12 different families, Kv1–Kv12, containing different indi-
vidual members. However, all the voltage-gated channels comprise the classical six-transmembrane monomer that forms a tetrameric voltage-gated structure reminiscent of sodium and calcium channels (Fig. 3).
Evidence has been obtained that potassium channel transcripts are differ- entially regulated at the transcriptional level in animal models of neuropathic pain. Using RT-PCR, Ishikawa et al. (2002) found that, in a chronic constriction injury model of neuropathic pain, Kv1.2, 1.4, 2.2, 4.2 and 4.3 mRNA levels in the ipsilateral DRG were reduced to 63%–73% of the contralateral sides of the same animal at 3 days and to 34%–63% at 7 days following CCI. In addition, Kv1.1 mRNA levels declined to about 72% of the contralateral level at 7 days. No signiﬁcant changes in Kv1.5, 1.6, 2.1, 3.1, 3.2, 3.5 and 4.1 mRNA levels were detectable in the ipsilateral DRG at either time. Interestingly, of the Kv channels present in DRG, Kv1.4 seems to be the main channel expressed in small-diameter sensory neurons, and the expression levels of this channel are much reduced in a Chung model of neuropathic pain (Rasband et al. 1999).
The calcium-activated potassium channel family comprises three subsets that have structural similarities with the Kv channels. They have been classi- ﬁed on the basis of their conductance into big, intermediate and small currents (BK, IK, SK). However, although voltage-dependent via some positive charges in the S4 domain, they are essentially opened by increases in cytoplasmic cal- cium concentrations through an interaction with a calmodulin binding domain found at the C-termini of these channels. They also differ from Kv channels in containing an additional N-terminal transmembrane domain that means the N-terminus is extracellular, unlike voltage-gated channels. Fascinatingly, Xie et al. (2005) have found that splice choice for the BK channel STREK exon is modulated through the actions of a CAM kinase response element, suggesting that neuronal excitability could be altered both directly though the actions of calcium on potassium channels, as well as indirectly by chang-
ing the molecular structure of expressed ion channels through alternative splicing.
Two-pore K channels (K2P channels) comprise a dimer of the two trans- membrane regions that form the functional pore with two P domains, and form functional dimers that are gated by a variety of stimuli. This area has re- cently been reviewed by Kim (2005). Sixteen K2P channel genes are expressed in a range of tissues and have the properties of leak K+ channels. They thus play a major role in setting resting membrane potential and regulating cell excitability. The stimuli that gate K2P channels are very diverse and include lipids, volatile anaesthetics, and heat, oxygen, acid and mechanical stimuli. Inward rectiﬁer potassium channels retain just the two pore-forming subunits found in voltage-gated channels together with the potassium selective P loop and also form tetrameric structures that are, necessarily, voltage-independent. There is an enormous variety of these channels that have been classiﬁed into seven families. G protein-regulated potassium channels are gated by interac-
tions with G protein βγ subunits.
Recently a new set of K channels that are responsive to increased intracellular
levels of sodium have also been identiﬁed (Bhattacharjee and Kaczmarek 2005), but their role in setting pain thresholds has not been explored.
Passmore et al. (2003) have provided evidence that KCNQ potassium cur- rents (responsible for the M-current) may also play a role in setting pain thresholds. Retigabine potentiates M-currents and leads to a diminution of nociceptive input into the dorsal horn of the spinal cord in both neuropathic and inﬂammatory pain models in the rat. Thus, despite the enormous complex- ity of potassium channels, there is some possibility of targeting pain thresholds through activating potassium channels expressed on damage-sensing sensory neurons.
Cyclic nucleotide-regulated hyperpolarisation-activated cation channels (HCN channels) play an important role in cardiac function and are expressed within sensory neurons. Chaplan and collaborators (2003) have described a novel role for HCN channels in touch-related pain and spontaneous neuronal discharge originating in the damaged DRG. Nerve injury markedly increased pacemaker currents in large-diameter DRG neurons and resulted in pacemaker-driven spontaneous action potentials in the ligated nerve. Pharmacological blockade of HCN activity using the speciﬁc inhibitor ZD7288 reversed hypersensitivity to light touch and decreased the ﬁring frequency of ectopic discharges originating
in A-δ and A-β ﬁbres by 90% and 40%. Targeting these channels appears problematic, however, given their other roles in the periphery.