The problem of chronic pain Chronic pain afflicts one in five adults in Europe and many diseases accompanied by pain are on the rise [1]. The diverse etiology of chronic pain encompasses trauma, metabolic or autoimmune disorders, infection, anti-retroviral treatment, and chemotherapy. Affected individuals typically report a combination of incapacitating sensory abnormalities, including spontaneous pain, hypersensitivity to stimulation, dysesthesias, and paresthesias. Despite significant progress, chronic pain remains refractory to treatment, with only one-third to two-thirds of patients reporting adequate (>50%) pain relief [1]. Moreover, our first-line drugs, non-steroidal anti-inflammatory agents (NSAIDS; e.g., aspirin) and opioids (e.g., morphine), are associated with adverse dose-limiting side-effects, dependence, and tolerance [2]. The lack of improved treatment reflects our incomplete understanding of the molecular pathophysiology underlying these pain states. Nociceptive pathways Pain is usually triggered by the activity of specialized damage-sensing neurons innervating the limbs and torso, whose cell somata cluster paraspinally in the dorsal root ganglion (DRG). These pseudo-unipolar cells project axons that bifurcate into peripheral fibers innervating the skin, muscle, or other organs, and central fibers that synapse with second-order spinal cord neurons. A similar architecture is encountered in trigeminal ganglion neurons located on each side of the cranium, which transduce sensory information from the face. Based on anatomical, neurochemical, and functional attributes, sensory neurons are distinguished into small-diameter with unmyelinated C-fibers, medium-diameter with thinly myelinated Aδ-fibers, and large-diameter that principally give rise to heavily myelinated Aβ-fibers. Because of their ability to encode noxious mechanical, thermal, or chemical stimuli, C- and Aδ-fibers are considered the main nociceptive afferents signaling pain. Aβ-fibers innervating the skin or muscles are predominantly low-threshold mechanoreceptive afferents responding to light touch or pressure, although a proportion are also activated by high-threshold stimuli. Signals initiated at sensory endings are relayed to the dorsal horn of the spinal cord and subsequently the brain via spinal projection systems including the spinothalamic tract, where the information is evaluated and an appropriate response generated. Spinal transmission is not a passive process but rather involves regulatory spinal processing, such as facilitatory or inhibitory modulation by interneurons, astroglia, and descending pathways, which can robustly increase or decrease the output [3]. Under normal conditions, generation of action potentials (APs) in sensory nerves typically originates at their peripheral nerve endings in the presence of a suprathreshold stimulus activating specialized receptors. However, following nerve trauma, electrogenesis can occur spontaneously at the site of injury (neuroma), DRG cell body, or even mid-nerve [4]. Furthermore, inflammation and neuropathic lesions are linked to enhanced responsiveness to supra- or even subthreshold stimulation [5–7]. This hyperexcitability is thought to be a major driver of pain and is ascribed to injury-induced reorganization of membrane ion channels, which are the principal determinants of AP generation and propagation. These maladaptive changes also have downstream effects at the spinal level; C-fiber activity can induce central sensitization, a state of heightened responsiveness of spinal cord neurons, such that innocuous input can now result in abnormally painful responses (e.g., tactile allodynia after Aβ-fiber stimulation) [8]. In addition, lesioned Aβ-fibers can acquire de novo nociceptive qualities that may also contribute to central sensitization [9]. Until recently the search for ion channel correlates of pathological excitability primarily focused on sodium and calcium channels. Unfortunately, despite significant discoveries in acute and inflammatory pain, no decisive involvement has been definitely established yet, particularly in neuropathic pain [10]. New evidence however suggests a previously unappreciated contribution of K+ channels in chronic pain processing, which we review here. K+ channels and pain signaling K+ channels are the most populous, widely distributed, and diverse class of ion channels in neurons, governed by some 78 genes in humans [11]. Upon activation, K+ channels facilitate an extremely rapid transmembrane K+ efflux that can influence AP threshold, waveform and frequency. Because K+ channel opening repolarizes (or even hyperpolarizes) the neuronal membrane, this function can limit AP generation and firing rate. Depending on the biophysical profile and precise subcellular localization in sensory neurons, K+ channel conduction is postulated to inhibit peripheral excitability by counteracting AP initiation at peripheral nerve terminals, reducing conduction fidelity across the axon, or limiting neurotransmitter release at central terminals (Figure 1). In addition, although normal sensory transduction does not rely on cell soma spiking, in chronic pain states K+ channels could provide a brake to the spontaneous activity developing in the DRG soma or other ectopic loci (e.g., the neuroma). Indeed, peripheral application of K+ channel openers on the cell body or terminals invariably decreases DRG excitability, whereas K+ channel blockers augment firing [5,11–13]. In the CNS, K+ channel opening could conceptually lead to enhanced nociception, for instance if the affected neuron participates in an inhibitory circuit. Nevertheless, the available data so far indicate that a variety of antinociceptive drugs mediate their action by directly opening spinal K+ channels [11]. Based on structural and physiological characteristics, K+ channels are organized into four distinct groups: voltage-gated, two-pore, calcium-activated, and inward rectifying, which we discuss in turn below. Voltage-gated K+ channels (Kv) The Kv superfamily is the most numerous among K+ channels, comprising of 40 genes in humans [14–16]. They are further classified in 12 families of α subunits that can interact to form functional homo- or hetero-tetrameric channels. Members of Kv1-Kv4, Kv7 and Kv10-Kv12 are pore-forming subunits, whereas Kv5, Kv6, Kv8, and Kv9 members do not form conducting channels unless associated with pore-forming subunits (Box 1). Channel tetramerization leads to tremendous functional diversity, further elevated by association with auxiliary β subunits, splice variants, and post-translational modifications. The largely overlapping pharmacology in neurons suggests a spectrum of Kv currents rather than fixed groups, reflecting the variant heterotetrameric composition, functional redundancy within families, and complex regulation. The majority of Kv channels are delayed rectifiers, because they are activated slowly to counteract (rectify) depolarization. On the basis of biophysical properties and sensitivity to tetraethylammonium (TEA), α-dendrotoxin, 4-aminopyridine, and muscarinic agonists, Kv currents are broadly distinguished into sustained delayed rectifying (IK), transient slowly inactivating (ID), transient fast-inactivating (IA) and non-inactivating (IM) that, as their names suggest, exhibit different kinetics. Although this classification is an oversimplification, it has value as a starting point to examine the different Kv components in physiological systems. These typical currents are also present in dorsal root and trigeminal ganglia neurons, whereas Gold and colleagues described six distinct K+ currents, three of which in small nociceptors [17–20]. Although it has been known for some time that nerve injury results in a dramatic decrease in K+ conductance of peripheral nerves that correlates with the emergence of hyperexcitability and pain behaviors, it was not until recently that specific subunits were linked to these changes [21]. Kv1.1 and Kv1.2 are delayed rectifiers activated by modest membrane depolarizations, and mainly contribute to the ID current. In many CNS neurons, these channels are preferentially localized at the axon initial segment (AIS, the site of AP initiation in CNS neurons) where they regulate AP threshold and firing rates, as well as nerve terminals where they modulate neurotransmitter release by controlling AP invasion in axonal branches [22,23]. The dominant role of Kv1 becomes apparent in type 1 episodic ataxia, where Kv1.1 mutations drive excitability changes in the cerebellum that cause severe seizures and premature death [24]. In the peripheral nervous system (PNS), Kv1.1 and Kv1.2 are predominantly found in the soma and juxtaparanodes of medium-large DRG neurons, often in heterotetramers [25], and are largely decreased after axotomy [26,27]; this may contribute to the hyperexcitable phenotype. Indeed, Kv1.1 loss-of-function results in reduced firing thresholds, attenuated mechanical and heat pain, and increased sensitivity in both phases of the formalin test [28,29]. By contrast, diminished Kv1.2 activity contributes to mechanical and cold neuropathic pain by depolarizing the resting membrane potential (RMP), reducing threshold current, and augmenting firing rates in myelinated neurons [30]. Moreover, Hao et al. recently reported that Kv1.1 tetramers form a bona fide mechanosensor that acts as an excitability brake in Aβ-mechanoreceptors of mouse DRG, with a minor contribution of Kv1.2 [31]. Interestingly, this mechanosensitive current was also detected in some high-threshold C-mechano-nociceptors (C-HTMRs). Although the literature highlights predominant Kv1.1 expression in myelinated neurons, the authors confirmed the presence of Kv1.1 subunits in a subpopulation of capsaicin-insensitive small neurons and C-fiber terminals in the skin using a monoclonal antibody [30]. This pattern may correspond to the occasional expression Rasband et al. documented in small DRG neurons from rat [25]. Although species differences may account for the discrepancy (and multiple species variations are recognized), other studies implementing molecular, immunohistological, and electrophysiological techniques have also indicated presence of Kv1.1 subunits in rat small sensory neurons [26,29,32,33]. Intriguingly, an accumulating body of research indicates that some human neuropathic pain syndromes are caused by production of autoimmune antibodies against Kv1 subunits that disrupt normal A- or C-fiber function (Box 1). Most of our knowledge on Kv2 comes from CNS studies, where Kv2.1 and Kv2.2 conduct the majority of delayed rectified IK current in several neuron subtypes [15,34]. Kv2 channels are activated slowly after significant depolarization, therefore their opening primarily influences membrane repolarization and inter-spike hyperpolarization during AP firing [15]. Importantly, because Kv2 feature characteristically slow activation and inactivation, the progressive channel recruitment during sustained activity can have a cumulative limiting effect on firing rates. The prominent CNS function of Kv2 is substantiated by specific localization in dendrites and AIS where the channel can exert intricate control over somal AP invasion and back-propagation [35]. Other interesting features of Kv2 are the phosphorylation-dependent regulation by neuronal activity, which can fine-tune excitability of CNS neurons by altering the channel membrane distribution and biophysical properties [36], as well as their modulation by several silent Kv subunits [37]. Despite the pivotal Kv2 role in shaping CNS signaling, an involvement in chronic pain was only recently uncovered. Kv2 subunits are present in small nociceptors (where Kv2.1 conducts the majority of IK [34]) but are also abundantly expressed in myelinated DRG neurons [38]. Transcript and protein Kv2 levels are downregulated by traumatic nerve injury, and this could augment firing by limiting the Kv2 inhibitory effect on spike frequency [26,27,38]. Indeed, application of a Kv2 blocker on ex vivo DRG preparations promotes myelinated neuron hyperexcitability by increasing conduction fidelity to the cell soma during repetitive stimulation [38]. It is possible that particular subcellular Kv2 localization forms the basis of an important filtering capacity (for instance by controlling AP traffic through the T-junction [39]), similarly to somatodendritic Kv2 filtering of somatic input in the CNS. Finally, a role in supraspinal pain pathways has also been demonstrated; cortical expression of Kv2.2 is reduced in oxaliplatin-induced neuropathy, and reproducing this in vivo results in marked cold and mechanical hypersensitivity [40]. All Kv3 channels are high-threshold and are typically encountered in fast-spiking neurons where they facilitate AP repolarization and hence dictate AP duration, but without affecting AP threshold or interspike interval [41]. Kv3.1 and Kv3.2 are delayed rectifiers contributing a small fraction (20%) of IK in small nociceptors, with a possible participation of Kv3.3 heterotetramers [42]. The Kv3.4 member almost certainly underlies the TEA-sensitive high-threshold transient current detected in nociceptors by Gold et al. (1996). This rapid Kv3.4 current accelerates nociceptor repolarization, an effect that restricts Ca2+-dependent neurotransmitter release at central nerve endings, where Kv3.4 is localized [41,43]. Hence, the mechanical hypersensitivity reported after Kv3.4 antisense treatment can be explained by AP broadening and therefore increased neurotransmission [44], although a loss of protein kinase C (PKC)-dependent modulation has also been suggested [43]. It is noted that although APs only spend a brief time at voltages capable of activating Kv3 channels, this restriction may be overcome by enhanced Kv3 densities at sites of action [41]. In addition, in native channels the activation threshold of Kv3.4 could be hyperpolarized following association with other proteins. For instance, heterotetramers of Kv3.4 with the delayed rectifiers Kv3.1 or Kv3.2 are activated at –30 mV [45], whereas Kv3.4 association with the auxiliary MinK-related peptide 2 (MiRP2) yields subthreshold currents in skeletal muscle [46]. In addition to Kv3.4, Kv4 members and Kv1.4 also give rise to transient A-currents (IA) that inactivate rapidly [15]. In contrast to Kv3.4, these A-channels are activated by small depolarizations, and their function in DRG can limit AP threshold, duration, and firing frequency [47]. Two low-threshold IA are detected in DRG neurons [17]; although Kv1.4 might contribute to the low-threshold IA in small DRG neurons, the fast voltage-dependent recovery from inactivation suggests the presence of Kv4 channels [25]. Therefore the low-threshold component may be predominantly mediated by Kv4.1, and the somatically confined Kv4.3 [48], because Kv4.2 is either absent or expressed at very low levels [27,48]. Consistent with a role in nociceptive pathways, A-type subunit expression and currents in the DRG are found to be reduced in a variety of pain models [21,25,44,49]. Mimicking Kv4.3 downregulation via intrathecal antisense is sufficient to induce mechanical hypersensitivity in naïve rats, presumably via reducing firing thresholds in a subset of Mrgprd (Mas-related G protein-coupled receptor D) neurons [44]. A-type blockers or short interfering RNA (siRNA) treatment can also diminish the analgesia by diclofenac in bone cancer [49], although K+ channel-related antinociception by this drug may be principally conferred via direct opening of other voltage-gated (e.g., Kv7 [50]), ATP-sensitive, or Ca2+-activated channels [51]. Finally, despite its negligible involvement in DRG excitability, Kv4.2 can strongly modulate pain plasticity in dorsal horn neurons; thus Kv4.2-null mice exhibit quicker mechanical pain resolution following nerve injury, as well as loss of extracellular signal-regulated kinase (ERK)-dependent sensitization in inflammatory models [52]. Although a few Kv4 activators are available (e.g., NS-5806 and KW-7158), the pacemaking activity of Kv4 channels in cardiac tissue is limiting for systemic applications. Kv7 channels open near RMP and underlie the low-threshold, non-inactivating M-current (IM) [20]. IM serves as a native ‘voltage clamp’ that stabilizes RMP and regulates AP threshold and accommodation within AP trains, affording it a central role in modulation of neuronal excitability. Accordingly, mutations in the human Kv7.2/Kv7.3 encoding genes cause benign familial neonatal epilepsy due to excessive excitability in distal motor axons [20]. In the DRG, IM mediated by Kv7.2, Kv7.3 and Kv7.5 oligomers is the dominant subthreshold K+ current in small neurons and a significant component in larger neurons (together with Kv1.1/Kv1.2) [12]. Kv7.2 and Kv7.3 are enriched in nociceptor AIS and terminals (but see [53]) and in nodes of myelinated fibers, in contrast to the majority of Kv channels which occupy paranodes or juxtaparanodes [12,53]. Kv7.2 and associated currents are reduced in DRG following neuropathic lesions, although the delayed onset of downregulation suggests a link to the maintenance rather than initiation of pain [54]; nevertheless, enhancement of residual IM can reverse pain behaviors [55]. Reduced Kv7 function is also involved in inflammatory pain, where IM inhibition occurs via protease-activated receptor 2 (PAR2) activation and phospholipase C (PLC)-induced depletion of phosphatidylinositol-4,5-bisphosphate (PIP2), inositol trisphosphate (IP3)-mediated Ca2+ augmentation, or a combination of both [12]. Consistent with this, PLC activation by bradykinin results in Ca2+ release which inhibits IM, thus allowing Ca2+-activated Cl− channels to amplify depolarizing input and trigger spontaneous firing [56]. The general purpose anti-inflammatory diclofenac has also been shown to directly activate Kv7.2/Kv7.3 channels [50]. The anticonvulsant retigabine, the most advanced Kv modulator, reduces excitability of animal [12,57,58] and human [59] axotomized nociceptive fibers by enhancing IM via a hyperpolarizing shift in Kv7.2/Kv7.3 activation. Accordingly, both retigabine and its structural analogue flupirtine (used as an analgesic in Europe since 1984) are antinociceptive in a variety of inflammatory and neuropathic pain paradigms through both central and peripheral mechanisms [12,60–62]. Although retigabine failed to produce analgesia in a recent clinical trial of post-herpetic neuralgia, flupirtine is currently in Phase II trials for fibromyalgia pain. New activators such as the Kv7.2/Kv7.3-selective ICA-27243 are in the pharmaceutical pipeline because retigabine and flupirtine do not show strong selectivity among Kv7 subunits and can additionally cause side-effects by interacting with other targets such as GABA receptors [63]. Two-pore K+ (K2P) channels K2P have emerged as promising candidates for pain modulation owing to their cell type-specific expression and lower inter-family sequence identity. They are unique among K+ channels in that they contain two pore domains and co-assemble as dimers rather than tetramers. Under physiological conditions K2P generate hyperpolarizing leak currents that stabilize cells below firing threshold, and disrupting this constitutive conductance results in depolarization and increased excitability [64]. Sensory neurons express many of the 15 members of the K2P superfamily, including TWIK1 (two-pore weak inwardly rectifying K+ channel), the TWIK-related (TR) channels TRESK, TREK1, and TRAAK, as well as TASK1 (acid-sensitive K+ channel), and marked reductions have been documented in pain states [65,66]. The importance of K2P in pain is highlighted by the discovery of a human K+ channelopathy; thus, familial migraine with aura is associated with a dominant-negative mutation in TRESK, a subunit strongly expressed in human trigeminal and dorsal root ganglia [67]. This fits well with the fact that migraine is associated with secretion of neuropeptides such as calcitonin gene-related peptide (CGRP) and substance P by meningeal nociceptors of the trigeminal ganglia, which may lead to sensitization. In a traumatic injury context, TRESK expression is decreased by axotomy, whereas pharmacological or siRNA inhibition induces C-fiber hyperexcitability and pain behaviors [68]. Contrarily, adenovirus-mediated spinal delivery of TRESK can reverse nerve injury-induced mechanical allodynia [69]. Another interesting but relatively unexplored member, TWIK1, is selectively expressed in medium-large DRG neurons and undergoes robust and persistent reductions by neuropathic injury [66]. A particularly noteworthy feature of K2P channels is their activation by a wide range of physicochemical factors including volatile anesthetics [70]. For instance, TREK1 is coexpressed with TRPV1 in nociceptors and can be activated by heat, stretching or lipids; a corresponding current is recorded in small neurons from wild type, but not TREK1 knockout (KO) animals [71]. TREK1 KO animals also show increased sensitivity to heat and mechanical stimulation, suggesting that normal K2P function counterbalances inward currents generated by TRPV1 and mechanosensitive Na+-permeable channels, respectively. Interestingly, TREK1 activity is decreased by inflammatory mediators such as prostaglandin E2 (PGE2) and lysophosphatidic acid, and TREK1-null mice develop more modest mechanical and thermal hyperalgesia during inflammation, presumably due to loss of this inhibition [71,72]. These data suggest that TREK1-modulating drugs may be useful in acute and inflammatory pain. Although TREK1 KOs show reduced cold pain after SNL, the precise involvement of this channel in neuropathic pain has not been thoroughly examined [71]. The member TRAAK is also mechano- and heat-sensitive, and simultaneous deletion of TREK1 and TRAAK has additive effects that may explain some of the mechano-hypersensitivity in colitis [73,74]. Furthermore, double TREK1/TRAAK KOs exhibit defects in acute cold pain processing, traced back to menthol-insensitive nociceptors [74]. Interestingly, oxaliplatin reduces TREK1 and TRAAK expression, and double KOs have modified cold pain responses in this model [75]. Ca2+-activated K+ channels (KCA) Opening of KCA during neuronal firing hyperpolarizes the membrane and provides feedback inhibition that limits Ca2+ influx and excitability, making them powerful regulators of synaptic transmission at nerve terminals [11]. Based on their conductance, they are further divided into BKCA (big conductance), IKCA (intermediate conductance), and SKCA (small conductance). All KCA are found in DRG and respond to increases in intracellular calcium, whereas BKCA are also voltage-sensitive [76,77]. Big conductance KCA are thought to influence excitability more prominently; illustrative of their significance in pain transduction is the recent finding of a functional coupling with TRPV1 (transient receptor potential cation channel, subfamily V, member 1) in nociceptors [78]. Blocking these channels with iberiotoxin reduces outward currents, prolongs AP duration, and increases firing rates in small-medium sensory neurons, with no effect on RMP, AP threshold, or AP amplitude [77]. Accordingly, axotomy decreases BKCA expression and Ca2+-dependent post-spike after-hyperpolarization in small-medium DRG neurons [79]. Contrariwise, the specific BKCA opener NS-1619 suppresses DRG neuron firing and can even antagonize the hyperexcitability evoked by A-channel block [77]. Interestingly, PGE2 and other inflammatory mediators reduce BKCA channel activity in nociceptors [17,76,80] and BKCA deletion in these neurons enhances inflammatory pain without affecting acute or neuropathic behaviours [133]. The BKCA opener andolast is currently in Phase III trials as an anti-inflammatory for chronic obstructive pulmonary disease; it would be interesting to evaluate the antinociceptive properties of NS-1619 and andolast in chronic pain models. Smaller conductance KCA are detected in a mixture of human and rodent DRGs, and may also contribute to pain phenotypes [76,81]. In small neurons, SKCA are downstream targets of NMDA receptor (NMDAR)-mediated Ca2+ influx because deleting the NR1 subunit in DRG induces hyperexcitability and pain hypersensitivity that can be reproduced by NMDAR antagonism or pharmacological SKCA inhibition [82]. Although IKCA expression in large neurons is decreased by nerve injury, SKCA and IKCA subunits in small neurons appear unaltered, suggesting that opening these channels may be a viable approach for chronic pain relief [76]. In line with this, the channel opener 1-ethyl-2-benzimidazolone (1-EBIO) reduces excitability in response to mechanical stimulation; however, the analgesic properties of such compounds remain to be robustly tested [11,83]. KCA also participate in central pain processing. Nerve injury leads to enhanced BKCA expression in second-order neurons near the dorsal root entry zone, and activating these channels by intrathecal NS-1619 reverses pain hypersensitivity [79]. Conversely, KCA blockers can antagonize the antinociceptive effects of muscarinic receptor agonists, gabapentin, and perhaps some NSAIDS [11]. Inward rectifiers (Kir) These channels are expressed mainly (but not exclusively) in supporting cells (Box 2) and can conduct atypical inward (rather than outward) K+ currents at depolarized membrane potentials. This buffering activity adds to glial K+ uptake through electrogenic Na+/K+ pumps to offset extracellular K+ accumulation during neuronal firing [84], thus preventing AP ‘short-circuiting’ and uncontrolled excitability changes [85]. They belong to one of seven families (Kir1-Kir7) and have a relatively simple structure with two transmembrane domains flanking the pore region [11]. Three families implicated in nociception are Kir3 (also known as G protein-regulated inward rectifiers K+ channels, GIRK), Kir2, and the ATP-sensitive channels (KATP). Neuronal GIRK channels are important determinants of spinal analgesia. As their name suggests they can interact with G proteins, an association thought to underlie the analgesic effects of opioids, endocannabinoids, and endogenous pain modulators [11]. Interestingly, enhanced GIRK1 phosphorylation in the dorsal horn following neuropathy or inflammation suggests reduced channel activity [86], whereas ‘pain risk’ GIRK2 alleles are associated with intensity of chronic back pain in humans [87]. Although no GIRK openers are currently available, their development could provide a viable alternative to opiates because this interaction may set in motion the same analgesic pathway without the unwanted side-effects of direct opioid activation [88–90]. Furthermore, a recent study suggests that GIRK2 expressed in sensory neurons also contribute to peripheral opioid-mediated antinociception [134]. Finally, although normally expressed in low levels in the periphery, Kir2.1 channels could also be useful for therapeutic interventions; virus-mediated expression of Kir2.1 in DRG neurons can restore excitability following compression injury, and even preclude pain symptoms when applied pre-emptively [91]. KATP members are tetramers of Kir6.1 or Kir6.2 surrounded by four sulfonylurea receptor subunits (SUR1 or SUR2) [92]. These channels are inhibited by ATP but also modulated by ligands such as ADP, adenosine, NO, vasoactive intestinal polypeptide (VIP) and CGRP. KATP currents are generally thought to play a minor role in setting basal excitability of DRG neurons, where Kv7 and K2P conductances dominate [93]. However, a therapeutic potential in pathological conditions has been proposed. Thus, although Kir6.2 activity is reduced in large DRG neurons post-injury, the ability of KATP openers to hyperpolarize RMP is retained, which could be exploited for neuropathic pain treatments [94,95]. Similarly, the inhibition of Kv7 activity in nociceptors during inflammation may also reveal analgesic roles for KATP channels. Indeed, the activators pinacidil and diazoxide reduce the hyperexcitability and pain induced by a range of peripheral inflammatory stimuli [51,93,96]. Finally, KATP opening in the CNS is linked to the antinociception produced by systemic treatment with morphine, NSAIDs, or even gabapentin [11,97]. Unfortunately, the involvement of KATP in modulation of cardiac rhythmicity, pancreatic insulin secretion, and intestinal function necessitates therapeutic strategies that selectively target the tissues of interest [92]. How does nerve injury trigger K+ channel dysfunction? In the preceding paragraphs we reviewed studies describing distinct expression patterns of K+ channels involved in the peripheral and central processing of painful stimuli (Figure 2) as well as their extensive downregulation after nerve lesions (Table 1). The latter finding has implications for treatment because the analgesia produced by pharmacologically enhancing the remaining K+ activity may be of limited scope. In these cases, targeting upstream cascades that orchestrate K+ channel dysfunction could yield more efficacious treatments. For instance, it was recently reported that an injury-induced endogenous non-coding RNA attenuates Kv1.2 expression, and blocking this pathway diminishes neuropathic pain [30]. Whether similar non-coding RNAs modulate the activity of other K+ channels is a question that warrants further investigation. Similarly, expression of Kv7.2, Kv4.3, and other ion channels in DRG is inhibited by the transcription factor REST (RE1-silencing transcription factor), which is induced by injury or inflammation [54,98]. Accordingly, blocking REST with antisense restores transcript levels and reverses some neuropathic pain symptoms [99]. Ion channel expression is typically controlled by carefully balanced neurotrophic support, which may become disrupted in pain pathology [100]. One of the most interesting messengers downstream of REST is brain-derived neurotrophic factor (BDNF), which has an established sensitizing role; in a diabetic neuropathy model, pre-emptive anti-BDNF treatment can reverse the IA reduction in myelinated neurons [101]. The regulatory role of BDNF may be more general among K+ channels, because the injury-induced BKCA downregulation in DRG can also be reversed by anti-BDNF [102]. There is also evidence that KCA activity can be regulated by nerve growth factor (NGF), glial-derived neurotrophic factor (GDNF), and neurotrophin 3 (NT3) [81,103,104]. Early research showed that NGF treatment can normalize axotomy-induced IA and IK reductions in DRG; however, an inhibitory effect on IA and IM may occur during inflammation [105–107]. The exact influence of these growth factors and the responsive K+ channel subunits remains to be systematically tested and clarified. Concluding remarks The exceptional abundance and breadth of function encountered in K+ channels has complicated efforts to untangle explicit roles in pain syndromes. Owing to advances in molecular, biochemical, electrophysiological, and genetic methods, however, we can now appreciate the involvement of specific subunits in maladaptive pain signaling after injury or inflammation. Nevertheless, there are many potential avenues of K+ involvement that have hardly been explored. It seems likely that unknown mutations in K+ channel genes might contribute to inherited pain syndromes. There are many ‘silent’ K+ channel subunits for which we have little idea of whether and how they might affect pain processing (Box 3). Auxiliary subunits can provide alternative substrates for pharmacological modulation; however, our understanding of these interactions in the PNS is also limited. In many chronic pain models an extensive dysregulation of several K+ channels is seen, and it is unknown whether a common epigenetic control exists. Manipulation of K+ channel subunits with dominant contributions in neuron excitability is likely to play a key role in shaping future pain treatments. The development of novel technologies and increasing availability of structural information creates an optimistic outlook for pharmacological design of K+ channel modulators [108–110]. In the next few years these advancements may be complemented by gene therapy strategies to introduce K+ channel copies at lesioned sites of the nervous system. Given the considerable convergence of pain mechanisms, it is plausible that synergistic treatments with K+ channel openers and other drugs (e.g., sodium or calcium channel blockers) can improve analgesic outcomes and/or circumvent side-effects by expanding the therapeutic window of present drugs to lower, more tolerable doses.
