The TRP family of cation-selective channel subunits are encoded by 28 distinct genes, many of which exist as multiple splice variants. This class of channel was ﬁrst identiﬁed in Drosophila, where the channel is involved in visual signalling. TRPs comprise seven sub-families that are structurally related, and form mul- timeric complexes that may involve heteromultimerisation both within and between different subsets. The structure of the channels (Fig. 4) includes six transmembrane domains reminiscent of the voltage-gated channels, and they do indeed seem to form tetrameric complexes. For historical reasons various aspect of TRP physiology have been extensively studied for some classes more than others. For example the TRPP subset, broadly expressed in neuronal and non-neuronal tissues, have been studied in tem of kidney function, because they are associated with polycystic kidney disease, TRPVs have been studied in terms of sensory neuron function, despite broad expression, because one of them is gated by chilli powder, whilst the calcium signalling community have focussed on TRPCs, which seem to be involved in calcium homeosta- sis and the replenishing of intracellular calcium stores. In fact, the range of ligands and stimuli, from mechanical and thermal to eicosanoids and envi- ronmental pollutants, is astonishing and makes these channels a fascinating topic of study for anyone interested in cellular signalling. Another curiosity of the TRP ﬁeld is that the number of reviews written about this class of chan- nel threatens to overtake the number of research papers published on their function.
In mammals the ﬁrst TRP channels to be identiﬁed are known as the canon- ical TRPs or the TRPC family and comprise seven members. TRPC1 has been characterised as a mechano-sensor and is very broadly expressed; TRPC2 is exquisitely selectively expressed in the vomeronasal organ of rodents, and is essential for pheromone-induced normal mating behaviour and is absent in man. TRPC3, -6 and -7 are structurally similar, activated by diacylglycerol and implicated in calcium homeostasis, whilst TRPC4 and -5 are close relatives and are involved in ligands induced calcium signalling. In sensory neurons TRPC1,
-3, -4, -5 and -6 are highly expressed (Fig. 4).
The identiﬁcation of the capsaicin receptor as a member of another TRP class of so-called vanilloid receptors led to the classiﬁcation of the TRPV family of receptors. There are six members of the TRPV class, the ﬁrst four of which are expressed in sensory neurons, and which have been shown to have a potential role in thermoreception and mechano-sensation based on studies of heterologously expressed channels. The polymodal nature of re- sponses to noxious stimuli demonstrated by the capsaicin receptor TRPV1, and the usefulness of capsaicin in treating some chronic pain syndromes suggested that members of this receptor class may play a role in chronic pain. There is indeed evidence that TRPV1 is up-regulated in the DRG neu-
Fig. 4 The structural features of different sets of TRP channels. (Reproduced from Montiel
rons adjacent to those that have been damaged by spinal nerve ligation (Fukuoka et al. 2002) whilst being down-regulated in damaged neurons. The up-regulated TRPV1 expression seems to be associated with A-ﬁbre sen- sory neurons (Hudson et al. 2001). These new receptors have been sug- gested to be important for the analgesic effects of capsaicin cream (Rashid et al. 2003). There are also claims that TRPV1 antagonists may be useful in treating neuropathic pain. Capsazepine, a non-competitive antagonist of TRPV1, markedly attenuated mechanical hyperalgesia in a guinea-pig sci- atic nerve injury model (Walker et al. 2003), but this effect was absent in rodents.
Other TRPs also respond to noxious temperatures in an analogous manner to TRPV1. TRPM8 is activated by cold temperatures, whilst TRPV2 and TRPV4
are activated, like TRPV1, by high temperatures.
Genetic screens have also revealed a number of members of the TRP channel family to be candidate mechano-transducers. Fruit ﬂies lacking the NompC channel showed a substantial loss of movement-evoked receptor potential in bristle mechano-receptor neurons, and NompC is also essential for normal hearing in zebraﬁsh. Recently, Nan, a TRP-related protein expressed exclu- sively in chordotonal neurons in Drosophila was found to be required for mechano-transduction by these cells. Another family member, OSM-9, is re- quired for detection of touch, osmolarity and olfactory stimuli in Caenorhab- ditis elegans (Colbert et al. 1997). The closest mammalian homologue of OSM-9, TRPV4, is activated by osmotically induced cell swelling, and a re- cent report (Suzuki et al. 2003) suggests it contributes to DRG mechano- sensation. Animals lacking this channel showed diminished responses to pres- sure in electrophysiological and behavioural assays. Thus, TRP family mem- bers are good candidates for noxious mechano-sensors; however, few TRP channels, even TRPV1, are selectively expressed only in damage sensing neu- rons, raising the potential problem of side-effects with drugs targeting these channels.
Acid-Sensing Ion Channels
ASICs are voltage-independent H+-gated ion channels belonging to the amilo- ride-sensitive DEG/ENac superfamily of receptor channels. Four genes encod- ing six different ASIC subunits have been cloned so far (ASIC1a, 1b, 2a, 2b,
3 and 4). ASIC1 and ASIC2 each have a splice variant, denoted “b”, which differs from the “a” counterpart only by the N-terminus. Each ASIC subunit
contains intracellular N- and C-termini, two transmembrane domains and a large extracellular loop containing cysteine-rich regions. The ASIC subunits can assemble to form functional homo- or heteromultimers that are mainly permeable to Na+.
All ASIC subunits are expressed in DRG sensory neurons and have been implicated in a number of different physiological sensory processes such as nociception associated with tissue acidosis in inﬂammation and ischaemia
(ASIC1 and ASIC3; Voilley et al. 2001; Chen et al. 2002), indirect regulation of cutaneous and visceral mechano-sensation (ASIC 1–3; Price et al. 2001; Page et al. 2004), sour taste (Ugawa et al. 2002), visual transduction (Ettaiche et al.
2004) and cochlear function (Peng et al. 2004). However, some ASIC subunits also show a wide distribution throughout the brain, where it has been suggested that they may be activated by the transient acidiﬁcation that occurs in the synaptic cleft due to acidic vesicle exocytosis. Recently, 6 ASIC subunits cloned
from the zebraﬁsh were shown to be mainly expressed in the central nervous system, suggesting a conserved role across species in neuronal communication (Paukert et al. 2004). Supporting this hypothesis, several studies in knock-out
mice have demonstrated a role for ASICs in long-term potentiation, learning, memory and fear behaviour (ASIC1 and 2; Wemmie et al. 2002, 2003, 2004). ASIC4 is broadly expressed in the nervous system but is not gated by protons
and has no known function.
ASICs have also been implicated in pathophysiological states. ASIC1a is the main mediator of H+ -gated currents in hippocampal neurons (Wemmie et al.
2002; Alvarez de la Rosa et al. 2003). Ca2+ imaging in COS-7 cells transiently
expressing ASIC1a recently demonstrated that homomeric ASIC1a channels are a major non-voltage-gated pathway for Ca2+ entry in cells (Yermolaieva et al. 2004). The same authors demonstrated that Ca2+ overload through ASIC1a channels makes a major contribution to hippocampal neuron damage in stroke.
A role for ASIC1a in Ca2+ overload during brain ischaemia is consistent with the fact that lactic acid, an enhancer of ASIC currents (Immke and McCleskey
2001), and protons are both produced (by anaerobic glycolysis and ATP hy- drolysis respectively) during ischaemia. In fact, it appears that the main source of Ca2+ entry occurring during ischaemia is not through ionotropic glutamate receptors but through ASIC1a channels, as ASIC1a blockade is a more efﬁcient
way to limit ischaemic damage than the use of glutamate antagonists (Xiong et al. 2004). Thus, apart from its acid-sensing function in various tissues,
ASIC1a activation may be a major factor in the biological events that lead to neuronal damage. Therefore, the search for drugs or factors regulating ASIC1a function has become potentially therapeutically important.
The mammalian ASICs are members of a channel superfamily involved in mechano-sensation in nematode worms (MEC-4 and MEC-10 mutants) and are highly expressed in sensory neurons (Waldmann and Lazdunski 1998). There are four identiﬁed genes encoding ASIC subunits, ASIC1–4, with two alterna- tive splice variants of ASIC1 and -2 taking the number of known subunits to six. Although protons are the only conﬁrmed activator of ASICs, the homology be- tween ASICs and MEC channels, coupled to high levels of expression of ASICs in sensory neurons, has led to the hypothesis that these channels function in mechano-transduction (Lewin and Moshourab 2004). ASIC subunits are found at appropriate sites to contribute to mechano-sensation. However, studies show staining for ASIC subunits along the length of the ﬁbres, not a speciﬁc enrich- ment at the terminals. Expression in sensory terminals is necessary for a role in the transduction of either acidic or mechanical stimuli. The ﬁnding that the
majority of Aβ -ﬁbre sensory terminals are immunoreactive for ASICs is at odds with the long-known observation that low-threshold mechano-receptors are not activated by low pH (see Lewin and Moshourab 2004). Thus, Welsh et al. (2001) have proposed that ASICs may exist, like MEC-4 and MEC-10, in
a multiprotein transduction complex that through an unknown mechanism masks the proton sensitivity of these channels (Fig. 5).
Using the neuronal cell body as a model of the sensory terminal, mechani-
cally activated currents in DRG neurons have been characterized (Drew et al.
2004). Neurons from ASIC2- and ASIC3-null mutants were compared with wild-type controls. Neuronal subpopulations generated distinct responses to mechanical stimulation consistent with their predicted in vivo phenotypes. In
Fig. 5a, b Schematic structure of ASIC channels. (Taken from Krishtal 2003)
particular, there was a striking relationship between action potential duration and mechano-sensitivity as has been observed in vivo. Putative low-threshold mechano-receptors exhibited rapidly adapting mechanically activated cur- rents. Conversely, when nociceptors responded they displayed slowly or in- termediately adapting currents that were smaller in amplitude than responses of low-threshold mechano-receptor neurons. No differences in current ampli- tude or kinetics were found between ASIC2 and/or ASIC3-null mutants and controls. These ﬁndings are consistent with another ion channel type being important in DRG mechano-transduction. Lazdunski’s group also investigated the effect of ASIC2 gene knock-out in mice on hearing, cutaneous mechano- sensation and visceral mechano-nociception. Their data also failed to support a role of ASIC2 in mechano-sensation (Roza et al. 2004).
Despite the mounting evidence that ASICs are not important for mechano- sensation, and the discovery of a number of unexpected functions for these channels, it does appear that in some pain states, for example in visceral pain or in ischaemic heart pain, ASICs may play an important role. However, given their multiple functions, they may not be ideal analgesic drug targets.
HERG channels are heart channels related to the Drosophila ether-a-go-go potassium channels. A structurally diverse range of drugs seems to act as blockers of this channel, resulting in prolongation of cardiac muscle repolari- sation which manifests itself with a prolonged Q-T interval. Premature action potential generation can result in ventricular tachyarrhythmias (torsades de pointes) which may cause lethal ventricular ﬁbrillation.
Sanguinetti et al. (2005) used site-directed mutagenesis to identify several residues of the HERG channel that comprise a common drug binding site. A number of residues located in the S6 domain (Tyr652, Phe656) and the base of the pore helix (Thr623, Ser624, Val625) were important sites of interaction for structurally different drugs. The reason why so many drugs seem to block HERG channels is related to the external atrium of the channel, and so early- stage screening to weed out compounds that interact with and block HERG channels is an essential part of the drug development process, and has halted the progress of many compounds that appeared to be promising ion channel blockers.
Target Validation of Ion Channels Using Transgenic Mice
Many ion channels have been implicated in pain pathways because of, for example, their tissue-speciﬁc pattern of expression. In the absence of speciﬁc
antagonists, genetic approaches to target validation may be useful. Dissecting the role of potential targets in regulating pain thresholds has often relied upon the use of antisense oligonucleotides, the speciﬁc down-regulation of mRNA using siRNA and the generation of mice with targeted mutations.
Each of these approaches has advantages and disadvantages. Antisense tech- nology is cheap, but speciﬁcity is a problem, as high concentrations of oligonu-
cleotide may have some cellular toxicity, and may also target structurally re- lated transcripts. SiRNA technology is still being developed, but has already revolutionised C. elegans genetics, where the speciﬁcity of siRNA action and the catalytic nature of RNA degradation mean that very low concentrations of
dsRNA can be employed. SiRNA is effective in vitro and in vivo in primary sensory neurons, where a 21-bp complementary dsRNA can be used to specif- ically degrade cognate RNA sequences through the formation of a complex
with ribonucleases. SiRNA acts transiently and catalytically and may not lead to long-lived RNA degradation (part of its attraction), but for animal models where neuropathic pain is modelled over a period of weeks this is a major
difﬁculty. Null mutants do not share this problem, but the problems of de- velopmental compensatory mechanisms and death during development have often provided obstacles to interpretation of phenotype. It is also desirable to generate mice where tissue-speciﬁc deletions can be carried out, and ideally
postnatal activation of Cre recombinase should be possible.
Thanks to the work of Sauer and collaborators (Le and Sauer 1999), who have exploited the recombinase activity of a bacteriophage enzyme Cre to delete DNA sequences that are ﬂanked by lox-P sites recognized by this enzyme, it has proved possible to generate tissue-speciﬁc null mutants This powerful technology is likely to be applied increasingly over the next few years, and together with siRNA promises to speed up target validation strategies in animal models of neuropathic pain. DRG-speciﬁc Cre-recombinase mice have been made, and an increasing number of ﬂoxed target genes are also now available for tissue-speciﬁc gene deletion studies.
We now have a clear idea of the vast repertoire of signalling molecules that are activated by tissue-damaging stimuli, and the voltage-gated channels that are responsible for electrical signalling and, eventually, the perception of pain. However, splice variants, post-translational modiﬁcations and altered proper- ties due to association with regulatory molecules all may confound the drug development process that usually relies on cell-based high-throughput screen- ing and tests in animal models. There are nevertheless good reasons for op- timism, as the ﬁrst set of new analgesic drugs acting on individual channels deﬁned over the past decade (for example Nav1.8 and P2X3 antagonists) start
to progress through the drug development pipeline, and genetic studies in both mice and man narrow down the potentially signiﬁcant channels to validated and effective targets.
Acknowledgements We thank our colleagues for helpful insights and suggestions. Our work is supported by the MRC, the BBSRC and the Wellcome Trust.