Cell Signaling and the Genesis of Neuropathic Pain

See allHide authors and affiliations

Science's STKE  28 Sep 2004:
Vol. 2004, Issue 252, pp. re14
DOI: 10.1126/stke.2522004re14


Damage to the nervous system can cause neuropathic pain, which is in general poorly treated and involves mechanisms that are incompletely known. Currently available animal models for neuropathic pain mainly involve partial injury of peripheral nerves. Multiple inflammatory mediators released from damaged tissue not only acutely excite primary sensory neurons in the peripheral nervous system, producing ectopic discharge, but also lead to a sustained increase in their excitability. Hyperexcitability also develops in the central nervous system (for instance, in dorsal horn neurons), and both peripheral and spinal elements contribute to neuropathic pain, so that spontaneous pain may occur or normally innocuous stimuli may produce pain. Inflammatory mediators and aberrant neuronal activity activate several signaling pathways [including protein kinases A and C, calcium/calmodulin-dependent protein kinase, and mitogen-activated protein kinases (MAPKs)] in primary sensory and dorsal horn neurons that mediate the induction and maintenance of neuropathic pain through both posttranslational and transcriptional mechanisms. In particular, peripheral nerve lesions result in activation of MAPKs (p38, extracellular signal–regulated kinase, and c-Jun N-terminal kinase) in microglia or astrocytes in the spinal cord, or both, leading to the production of inflammatory mediators that sensitize dorsal horn neurons. Activity of dorsal horn neurons, in turn, enhances activation of spinal glia. This neuron-glia interaction involves positive feedback mechanisms and is likely to enhance and prolong neuropathic pain even in the absence of ongoing peripheral external stimulation or injury. The goal of this review is to present evidence for signaling cascades in these cell types that not only will deepen our understanding of the genesis of neuropathic pain but also may help to identify new targets for pharmacological intervention.


Living organisms need to sense noxious stimuli in their immediate environment to avoid potentially hazardous situations and thus survive. To this end, multicellular creatures have evolved a specialized apparatus (the nociceptor) to differentiate innocuous from noxious stimuli. The recent application of molecular biology to sensory physiology and pain research has led to the discovery of multiple transduction molecules within the nociceptor. These include the transient receptor potential ion channels TRPV1 [also called vanilloid receptor 1 (VR1)], TRPV2, TRPV3, and TRPV4 for sensing heat; TRPM8 [also called cold and menthol receptor 1 (CMR1)] and TRPA1 for sensing cold; the DEG (degenerin), DRASIC (dorsal root acid-sensing ion channel), and TREK-1 (TWIK-related K+ channel–1) ion channels for perceiving noxious mechanical sensations; and chemical sensors, such as the purinoreceptor P2X3 to detect adenosine triphosphate (ATP) and ASIC (acid-activated ion channel), DRASIC, and TRPV1 to detect H+ (1, 2). Activation of sufficient numbers of such transducing molecules in peripheral nociceptors in skin and muscle or in visceral organs initiates action potentials, which are conducted through thin unmyelinated C fibers and myelinated Aδ fibers to the central nervous system (CNS). The signal is first conducted to second-order sensory neurons in the spinal dorsal horn, and then through supraspinal pathways to the cerebral cortex, eliciting the sensation of pain (3).

Although nociceptive pain (physiological pain) requires intense, high-threshold stimulation and is typically transient, pathological pain (clinical pain) associated with inflammation of peripheral tissue that arises from the initial damage (inflammatory pain) or from lesions to the nervous system (neuropathic pain) is often persistent. Such pathological pain manifests as spontaneous pain and as pain from gentle stimuli that are normally innocuous (46). Although neuropathic pain is also associated with inflammation and shares features of inflammatory pain, nerve injury produces distinct neurochemical changes in primary sensory neurons, which are quite different from those produced by inflammation (7, 8). Pathological pain also includes cancer pain (9) and the pain that emerges after sustained opioid treatment, which shares some features with neuropathic pain (10).

Many patients in the pain clinic suffer from neuropathic pain due to injury to the peripheral nervous system (PNS) [including peripheral nerves, dorsal root ganglia (DRG), and dorsal roots] or the CNS (spinal cord and thalamus). These injuries may result from surgery, diabetic neuropathy, amputation, viral infection, trauma, stroke, and so forth. Although neuropathic pain in animal models after specific nerve injury is highly reproducible, at least in certain species (1115), whether similar pain occurs in a particular patient after a similar lesion to the nervous system depends on the site and type of the lesion, medical history, and genetic background of that patient. The most distinctive symptom of neuropathic pain is mechanical allodynia (painful responses to normally innocuous tactile stimuli), although neuropathic pain is also often characterized by hyperalgesia (increased responsiveness to noxious stimuli) to mechanical and thermal stimuli. Mechanical hyperalgesias are further divided into brush-evoked (dynamic), pressure-evoked (static), and punctuate hyperalgesias (6). In addition to stimulus-evoked pain, many patients have spontaneous pain, described as shooting, lancinating, or burning pain. Currently available drugs provide relief of neuropathic pain in only a fraction of such patients, sometimes of enduring quality but often lasting no longer than the drug’s presence at the site (6).

Cases that present clinically are generally treated empirically and are difficult to study, because there are often multiple potentially contributing factors and because the time from the initial insult to its recognition can vary from days to years. Thus, almost all mechanistic studies are based on animal models. Neuropathic pain research has been explored with different animal models in which intentional damage is done to the sciatic nerve and its branches, the spinal nerves, or the spinal cord (6, 1115). How well these different injury models accommodate the clinically presenting pain syndromes remains an important concern for validating common mechanisms and establishing animal models for human drug testing. Pain hypersensitivity can be a manifestation of peripheral sensitization [increased excitability of nociceptors (primary nociceptive neurons) at the distal nerve terminal, at the axon, or at the cell body] or central sensitization involving activity-dependent plasticity of nociceptive neurons in the CNS, particularly in the dorsal horn of the spinal cord (1, 3, 1619). Neuropathic pain can arise from an "injury discharge" originating at the site of nerve injury, as well as from the development of ectopic impulses (increased spontaneous activity) in the injured nerve’s soma within the DRG (20, 21). However, ectopic activity also develops in uninjured fibers (2225), and increased sensitivity to pain is found in the receptive fields of intact nerves (12, 14). The mechanisms for this sensitization are less apparent than for injured nerves; they may involve the transfer of diffusible substances from injured nerves released at the peripheral endings, increased receptive fields of central neurons, or even injury-responsive glia located diffusely in the spinal cord.

Ectopic activity from primary afferents drives central sensitization. Moreover, altered synaptic connectivity in the dorsal horn from synaptic stripping (loss of synaptic input from damaged neurons) or from the formation of novel synapses (6), loss of inhibition in the dorsal horn (6), and increased descending facilitatory influences from the brain (26) all contribute to the development and maintenance of neuropathic pain. Aberrant sympathetic influences (e.g., abnormal sympathetic modulation of pain) have also been implicated in neuropathic pain (2730), but this observation could be species-selective and its relevance for clinical neuropathic pain is debatable (31). All of these scenarios involve the increased excitability and enhanced chemical transmission of peripheral and central neurons—processes that depend on both ion channels and receptors that are targets for acute posttranslational modification and long-term transcriptional regulation. Therefore, this review focuses on signaling pathways in the primary sensory neurons of the DRG and second-order neurons of the spinal cord dorsal horn that likely participate in the hyperexcitability that underlies neuropathic pain. We also focus on neuropathic pain produced by peripheral nerve injury in animal models, with only passing reference to clinical studies.

Peripheral Mechanisms of Pain After Nerve Injury

Signaling Molecules and Signaling Pathways in Sensory Neurons

Peripheral nerve injury is often accompanied by transient local inflammation, which probably contributes to the instigation of neuropathic pain. Multiple inflammatory mediators [IFMs, which include prostaglandin E2 (PGE2), bradykinin, nerve growth factor (NGF), histamine, serotonin, interleukin 1-β (IL-1β), tumor necrosis factor–α (TNF-α), ATP, glutamate, endothelin-1, and various chemokines] can be released from damaged axons or their enclosing Schwann cells, or from satellite cells, mast cells, and infiltrating leukocytes. The soma and axons of primary sensory neurons express receptors for these IFMs, and some receptors are also found on nearby glial cells and invading immune cells. We next consider the role of some of the likely important receptors and of the final physiological targets that they modify.

G Protein–Coupled Receptors

Multiple heterotrimeric GTP-binding protein (G protein)–coupled receptors (GPCRs) are expressed in sensory neurons. These include EP receptors for prostanoids (e.g., EP1-4 for PGE2), bradykinin receptors (B1/B2), serotonin receptors(5-HT1A, 5-HT2A), endothelin receptor (ETA), opioid receptors (μ, δ, κ), cannabinoid receptor (CB1, possibly for anandamide), metabotropic glutamate receptors [both group I (mGluR 1 and 5) and group II (mGluR 2 to 4)], and adrenoceptors (α1 and α2 for norepinephrine) (8, 32, 33). These receptors contribute in part to the dynamic changes in neuronal response. For example, glutamate is probably released from distal and central endings of primary afferents and will depolarize inotropic receptor–containing neurons and adjacent cells, as well as acting on metabotropic receptors. For example, the up-regulation of mGluR5 in A-fiber DRG neurons in response to nerve lesion contributes to neuropathic pain (34). Analogously, whereas the B2 bradykinin receptor is constitutively expressed in primary sensory neurons, expression of the B1 receptor appears to be induced only after nerve injury and then contributes to neuropathic pain (35). Primary sensory neurons express multiple chemokine receptors, such as CXCR4, CX3CR1, CCR4, and CCR5, and chemokines can excite these neurons and produce acute tactile allodynia (36). Neurons and endothelial cells also synthesize endothelin-1 (ET-1), and some neurons express ETA receptors for this peptide (37). ET-1 applied to peripheral nerve or injected into the paw acts through ETA receptors to produce acute, spontaneous nocifensive behavior (behavior that corresponds to the initiation of impulses by ET-1 in peripheral nociceptors) (38, 39). Both cutaneous incision and peripheral nerve injury lead to allodynia and hyperalgesia that is suppressed by antagonists of ETA receptors, implying that endogenous ET-1 plays a critical role in these responses. The complete mechanisms of excitation and pain are not known, but ET-1 applied to isolated DRG neurons expressing tetrodotoxin-resistant (TTX-R) Na+ currents enhances the voltage-dependent activation of these channels, without effect on tetrodotoxin-sensitive (TTX-S) currents in other cells (40).

Opioid receptors (μ, δ, and κ subtypes), which activate pathways that inhibit painful sensations, are normally expressed in small size (C-fiber), primarily nociceptive, DRG neurons (41) and are down-regulated after nerve injury (42). On the other hand, the peptide cholecystokinin (CCK), which antagonizes opioid actions, and its receptor CCKB are up-regulated in DRG neurons after peripheral nerve injury (7). The down-regulation of opioid receptors and up-regulation of the CCK system was thought to cause the decreased efficacy of opioids in treating neuropathic pain (43, 44). Although neuropathic pain has traditionally been considered opioid-resistant, to intrathecal opioids in particular (4547), recent clinical (48) and animal studies (49, 50) [but see (51)] have shown that opioids can be effective in treating neuropathic pain. These conflicting data may result from differences in the type of nerve injury, route of drug administration, and potency and receptor selectivity of the particular opioid agonist (49). In contrast to the findings during neuropathic pain, after hindpaw inflammation the μ opioid receptor is up-regulated in DRG neurons and morphine has been shown to have increased efficacy in reducing this inflammatory pain (41, 52).

Peripheral nerve injury causes sprouting of sympathetic nerves around injured DRG neurons, and sympathetic stimulation is known to release norepinephrine to activate these injured neurons (27). Several growth factors, including NGF, brain-derived neurotrophic factor (BDNF), and leukemia inhibitory factor (LIF), are implicated in this sprouting (30, 53). However, sympathetic axons do not make direct synaptic contacts with DRG somata, even after nerve injury. Thus, functional coupling between sympathetic efferents and DRG neurons may be mediated by diffusion of neurotransmitter molecules in the extracellular space (54). The α1B adrenoceptor is up-regulated in axotomized DRG neurons and plays a role in the development of adrenergic sensitivity in injured sensory neurons, and thus probably contributes to sympathetically maintained pain after spinal nerve ligation (55). However, there is also evidence showing that lumbar sympathectomy or adrenergic antagonists fail to reverse neuropathic pain (31, 56). The effectiveness of sympathetic blockade may depend on the genetic background of the species as well as the nature and locus of the nerve injury.

Activation of Gs- and Gq-coupled receptors, respectively, leads to the activation of protein kinase A (PKA) and phospholipase C (PLC), particularly PLC-β. In contrast, activation of Gi-coupled receptors, such as opioid and cannabinoid receptors and group II mGluRs, inhibits the PKA pathway by suppressing adenylate cyclase (Fig. 1). These and various other kinases act on numerous ion channels and receptors, described below, to effect acute changes in cellular physiology that are necessary for the genesis of pain after nerve injury.

Fig. 1.

Posttranslational mechanisms increase the excitability of primary sensory neurons (terminals, axons, or cell body) through multiple signaling pathways. (A) Peripheral nerve injury leads to the production of various signaling molecules such as bradykinin, PGE2, serotonin, endothelin, NGF, IL-1β, TNF-α, ATP, H+, and glutamate, which are released and stimulate the appropriate receptors on sensory neurons. Stimulation of various G protein–coupled receptors (GPCRs, including the B1 bradykinin receptor, the EP2 PGE2 receptor, the α1B adrenoceptor, group I mGluRs, and the 5HT-1A serotonin receptor) activates adenylate cyclase (AC) through Gs and subsequently stimulates PKA. Enzymes associated with GPCRs can also be suppressed; for instance, AC is inhibited by Gi through stimulation of opioid, cannabinoid, and group II mGluRs. Other GPCRs (such as the B2 bradykinin receptor and the ETA endothelin-1 receptor) activate phospholipase Cβ (PLCβ) through Gq (or G11), thus increasing intracellular Ca2+. Intracellular Ca2+ also increases by means of influx through voltage-gated channels or through stimulation of the ionotropic NMDA, TRPV1, and P2X receptors by glutamate, H+, and ATP, respectively. Increased intracellular Ca2+ stimulates PKC and CaMK. Stimulation of the TrkA receptor by NGF leads to intense activation of the Ras-Raf-MEK-ERK cascade, as well as to activation of the PLC-γ, PI3K, and p38 signaling pathways. PKA, PKC, and PI3K can also activate ERK, and p38 can activate PLA2, eventually increasing PGE2, which acts in autocrine and paracrine modes to enhance excitability. Stimulation of IL-1β and TNF-α receptors by proinflammatory cytokines activates p38 and JNK pathways through apoptosis signaling kinase (ASK). (B) Downstream, protein kinases activated through these signaling pathways can act directly or indirectly through numerous mechanisms to increase or decrease the activity of key ion channels (TTX-S and TTX-R Na+ channels, Ca2+ channels and other Ca2+-permeable channels, and K+ channels), thereby directly modifying the excitability of primary sensory neurons (peripheral sensitization) and generating neuropathic pain.

Ion Channels as Substrates for the Signal Transduction Pathways

Inasmuch as the excitability of primary afferent neurons is one hallmark of experimental neuropathic pain, the ion channels that control excitability are likely targets of those pathways and receptors that respond to nerve injury. Here we consider the evidence for the contributions made by some of the more prominent channels known to be involved in afferent excitability.

Sodium Channels

Sodium (Na+) channels are essential for action potentials in all known mammalian neurons. Several sodium channel isoforms that are inhibited by nanomolar concentrations of TTX, the so-called TTX-S sodium channels (i.e., Nav1.2/1.6/1.7), are expressed in almost all DRG neurons (57, 58). Nerve injury stimulates the expression of TTX-S channel Nav1.3, which is normally present only in embryonic sensory neurons (59). The TTX-R Na+ channels (Nav1.8/1.9) are predominantly expressed in small and in some medium-size sensory neurons with nociceptive properties (59, 60). Although both Nav1.8 and Nav1.9 are down-regulated in the cell bodies of injured DRG neurons after nerve injury (59, 6163), Nav1.8 appears to accumulate in uninjured axons located in hyperexcitable sciatic nerve (63, 64). Knockdown of Nav1.8 with antisense oligodeoxynucleotides (ODNs) can prevent and temporarily reverse neuropathic pain (6365), and systemic application of certain "state-selective" Na+ channel blockers (for instance, lidocaine) produces an acute preferential blockade of impulses arising from membranes containing TTX-R–like Na+ channels and also leads to long-term changes in tactile allodynia in vivo (see below).

Most TTX-S Na+ channels rapidly activate (open) and inactivate (become unavailable to open), providing a transient but sufficiently large inward current that triggers one or a few action potentials in response to transient stimulation. In contrast, the TTX-R Na+ channels Nav1.8 and Nav1.9 activate more slowly and, most important, respectively inactivate slowly or barely at all during a maintained depolarization of the membrane. The Nav1.3 isoform, which is up-regulated in injured neurons, also has a very slowly inactivating component. As a result of this slow inactivation, an inward Na+ current exists immediately after repolarization of the action potential and immediately generates another impulse, thereby establishing a mechanism for sustained, high-frequency repetitive firing after a single stimulus (one of the hallmarks of hyperexcitability). A similar "resurgent" Na+ current that disappears only slowly upon membrane repolarization, which is likely carried by NaV 1.6 and is found selectively in small-diameter neurons, also contributes to repetitive firing (66).

Na+ channel blockers given systemically (such as lidocaine and mexiletine) suppress ectopic discharge and reverse mechanical allodynia and thermal hyperalgesia after nerve injury (6772). Brief intravenous infusions, to plasma concentrations far below those required for blockade of normal nerve impulses, alleviate allodynia transiently and then trigger a slower phase of relief, one that takes about a day to fully develop and can persist for a week or more (71, 72). Although the acute effect of these drugs is consistent with their direct blockade of repetitive impulses arising from membranes containing TTX-R–type slow-inactivating Na+ channels (73, 74), and thus with direct actions on these ion channels, the slow phase may involve biochemical suppression of a signaling pathway.

Local anesthetic-type agents such as lidocaine and mexiletine are not selective for Na+ channels, but also inhibit K+ (75, 76) and Ca2+ channels (7578), vanilloid receptor–1 (VR-1) (79), and various GPCRs (such as B2 bradykinin receptors) (80), and also affect cytoskeletal dynamics and suppress both neurite outgrowth and axonal transport (81, 82). Notably, local anesthetics are unusually potent in inhibiting the priming reaction in neutrophils, which may be an important step in the initiation of an inflammatory response (83). Some of these other effects, or their combination, could account for the late-developing and long-lasting antihyperalgesic actions of these agents.

Calcium Channels

DRG neurons also express multiple calcium channels, including T-, P/Q-, L-, and N-type channels (84). The L- and N-type high voltage–activated channels are found in all sensory neuron cell bodies, P/Q-type channels are represented in a smaller population, and the low voltage–activated T-type isoforms are primarily seen in nociceptive neurons (85, 86).

Calcium currents in DRG are differentially affected by nerve injury. After chronic constriction treatment (CCI) to the sciatic nerve, both L-type and T-type currents are reduced, the former in large- to medium-size cells (87) and the latter in the small-size nociceptors (nociceptive sensory neurons) (88) where they are selectively expressed. Such changes may increase excitability by decreasing the activation of Ca2+-modulated K+ currents that normally limit the duration of repetitive trains of action potentials (89, 90). In contrast, the α2δ subtype of calcium channel is up-regulated in injured DRG neurons, and this channel appears to be a major target for gabapentin (Neurontin), a currently popular clinical choice for the treatment of pain after nerve injury (91, 92). Activation of these calcium channels as well as of N-methyl-𝒹-aspartate (NMDA) receptors, TRPV1, and the inotropic ATP receptor P2X3 all lead to calcium influx. Ca2+ can also be released from intracellular Ca2+ stores after GPCR activation of PLC (Fig. 1). An increase in intracellular Ca2+ activates protein kinase C (PKC) and calcium/calmodulin-dependent protein kinase II (CaMKII) and may also indirectly activate phosphatidylinositol 3-kinase (PI3K), ERK, and p38 mitogen-activated protein kinase (MAPK).

Potassium Channels

Potassium channels are important modulators of cellular excitability. Different types of K+ channels, which can be regulated by voltage or by intracellular Ca2+, by other second messengers such as ATP or cyclic nucleotides, or even directly by G proteins, serve different roles depending on their "gating properties"—in other words, the membrane potentials at which they are activated and the duration of the induced openings. K+ channels function as the major determinant of the resting potential, modulate the duration of single action potentials, and govern the propensity for and duration of repetitive firing.

Several of these channels are associated with nociceptors (93, 94). A class of inwardly rectifying K+ channels that contribute strongly to the resting potential and to firing thresholds as well as to "pacemaker-like" activity is regulated by the direct binding of the βγ subunit of G proteins. These G protein–activated inwardly rectifying potassium channels (GIRKs) appear to be a final common target for different neurotransmitters and neuromodulators that affect pain, including opiates, cannabinoids, and α2-adrenergic analgesics. All of these substances activate GIRK2 and thereby inhibit neuronal excitation, whereas substance P and endothelin-1, which inhibit GIRK, attenuate the resting K+ conductance and promote neuronal excitation. Fast "delayed rectifier" K+ channels are opened by membrane depolarization and modify the duration of single action potentials. In sensory neurons, these channels are modified by protein kinases that are activated by PGE2 and NGF among other signaling molecules. This phosphorylation decreases current through this class of channels, thereby broadening individual action potentials and promoting repetitive firing as well as increasing net Ca2+ influx at terminals and thus potentiating transmitter release. Slow-opening "delayed rectifier" channels also open during depolarization, but their slow gating necessitates the cumulative effect of many brief action potentials for an effective functional recruitment, so their major role is to control the duration of trains of impulses rather than a single spike. Compounds that suppress increased sensory neuron excitability and hyperalgesia in experimental models of neuropathic pain enhance the activity of these so-called KCNQ-type channels, which are classically inhibited by acetylcholine acting through muscarinic receptors (thus the name "M-current"). Some K+ channels are activated by both depolarization and intracellular Ca2+ and have an important role in controlling the duration of trains of action potentials. Repeated transient depolarizations incrementally increase intracellular Ca2+, which acts in a negative feedback mode to activate these K+ channels, hyperpolarize the membrane, and suppress further firing. In addition to regulation by treatments that alter intracellular Ca2+ dynamics, these K+ currents are diminished by the pain-inducing agents bradykinin and substance P and by norepinephrine and the prostaglandins, agents that also promote hyperalgesia. Injury to peripheral nerve leads to reduction in these currents and to the related post-tetanic hyperpolarization, part of which may be due to the decreased Ca2+ current in injured neurons (see above), which changes will in turn heighten membrane excitability.

The examples above do not comprise an exhaustive list of all of the pathways for K+ channel modification—nor of those for the other voltage-gated channels, which would require its own review—but do indicate the potential importance of many of these channels as determinants of membrane excitability and transmitter release. It seems to be more common than not that pronociceptive or prohyperalgesic compounds affect multiple ion channels, apparently by posttranslational mechanisms that use common signaling pathways such as Ca2+-activated PKC, adenylate cyclase–activated PKA, or bioactive lipids such as ceramide (95). The individual changes in these separate ion conductances and gating are usually synergistic for an overall effect of neuronal excitability, including changes that increase stimulated firing, induce spontaneous firing, and enhance neurotransmitter and neuropeptide release at both central and distal terminals. These are the general actions that, by altering the impulse physiology of nociceptive neurons, probably account for the acute development of neuropathic pain symptoms after injury or inflammation.

Nerve Growth Factor, Inflammatory Cytokines, and MAPK Pathways

Although NGF is required for the survival of sensory neurons during fetal development, it is not necessary for the survival of mature sensory neurons. Rather, it maintains the phenotype of sensory neurons in the adult (96) and is thus required for the continued expression of many genes (96). Local and systemic injection of NGF produces hyperalgesia (97, 98), and NGF can rapidly (within minutes) increase the activity of TRPV1 in DRG neurons (99, 100). Nerve injury interrupts the normal supply of target-derived NGF to injured neurons. However, NGF levels increase in the adjacent uninjured DRG (101). NGF typically activates three signaling pathways: p44/42 MAPK (also called ERK, extracellular signal–regulated kinase), PI3K (phosphatidylinositol 3-kinase), and PLC-γ pathways (102).

The MAPKs are a family of evolutionarily conserved molecules that play a critical role in cell signaling by transducing extracellular stimuli into intracellular responses. This family includes three major members: ERK, p38, and JNK (c-Jun N-terminal kinase). Although nonphosphorylated inactive MAPKs are expressed in most DRG neurons, the activated phosphorylated MAPKs (pMAPK) are detected in only a small population of neurons in the unstimulated DRG (103, 104). Peripheral nerve injury increases phosphorylation of ERK, p38, and JNK in DRG neurons (105107). Immunohistochemistry reveals a persistent increase in the percentage of neurons in the DRG that are immunoreactive for p-p38 (the active form of p38), which lasts more than 3 weeks after chronic constriction injury (CCI) treatment or spinal nerve ligation (105, 108). Inhibition of ERK, p38, or JNK alleviates neuropathic pain, indicating an important role for these signaling pathways serving this pathology (105, 107, 109111). Nerve injury can also stimulate MAPK (ERK, p38, JNK) activation in non-neuronal cells (e.g., in satellite cells) in the DRG, causing increased synthesis of inflammatory mediators that then stimulate DRG neurons. We discuss interactions between spinal cord neurons and glia below.

In addition to ERK, NGF also activates MAPK members p38 (104) and JNK (112) in DRG neurons. A recent study showed that PI3K and ERK are required for NGF-induced hyperalgesia (113). It remains to be investigated whether p38 and JNK can also mediate hyperalgesia by NGF. Although most of NGF’s effect is mediated through its high-affinity receptor TrkA, it has also been shown to increase the current of TTX-R Na+ channel through a ceramide-dependent pathway that is mediated through the low-affinity p75 receptor (95). In addition to NGF, GDNF (glial-derived neurotrophic factor) maintains the phenotype of another subset (generally TrkA-negative) of C-fiber DRG neurons (96). However, NGF and GDNF appear to play opposite roles in regulating neuropathic pain (see below), showing that there is no general rule for the pain-related consequences of growth factors in the adult nervous system.

Peripheral axonal injury causes an up-regulation of proinflammatory cytokines (such as IL-1β, TNF-α, and IL-6) produced by non-neuronal cells, including satellite cells and Schwann cells, in the damaged tissue. DRG neurons not only express receptors for these cytokines but also produce some cytokines, especially after nerve injury (30, 114116). TNF-α can evoke neuronal activity in injured and adjacent uninjured DRG neurons (117), and sensory neurons from damaged nerves exhibit enhanced responses to a soup containing multiple proinflammatory substances (118). Although IL-1β has been shown to stimulate release of the pro-algesic neuropeptide substance P, it’s unclear whether this cytokine can directly sensitize sensory neurons (119). Proinflammatory cytokines have been widely implicated in MAPK activation in various cell types, including DRG neurons (18, 116), and MAPK signaling pathways are in turn able to induce the production of cytokines [(116), also see below].

Posttranslational Regulation and Induction of Peripheral Sensitization

Peripheral sensitization refers to the increased sensitivity of primary sensory neurons to stimuli, a phenomenon characterized by lower threshold and increased impulse activity (hyperexcitability) and, sometimes, an adopted responsiveness to stimuli that are ineffective before sensitization (allodynia). Sensitizing changes can occur at the nerve terminals, in axons, and in cell bodies of sensory neurons. Peripheral sensitization uses some well-described mechanisms under conditions of inflammatory pain. For example, PGE2 activates PKA, leading to "sensitization" (increased probability of opening) of TTX-R Na+ channels (Nav1.8 and Nav1.9) and a corresponding reduction of action potential threshold (120, 121). Similarly, bradykinin activates PKC, resulting in phosphorylation of TRPV1 and thereby reducing the minimum temperature (threshold) for activating TRPV1 from 43°C to normal body temperature (37°C) (1). This sensitization occurs in both the damaged axons and surrounding intact axons, as well as in the cell body, under conditions that elicit neuropathic pain, apparently because inflammatory mediators released from non-neuronal cells diffuse to act on nearby neurons. Indeed, TRPV1 and Nav 1.8 are implicated in both inflammatory and neuropathic pain (65). Protein kinases (for instance, PKA, PKC, CaMK, PLC, PI3K, ERK) can increase the activity of Na+, Ca2+, and TRP channels (excitatory channels) through various mechanisms while simultaneously suppressing K+ channels (whose currents antagonize excitation). Kinases may directly phosphorylate Nav1.8, Nav1.9, TRPV1, TTX-S Na+, or Ca2+ channels and thereby increase their activity (1, 100, 122, 123); they may regulate channel activity by promoting various protein-protein interactions, by stimulating channel trafficking and insertion to the membrane, or by promoting the rapid translation of these channels (Fig. 1). It remains to be determined how different protein kinases regulate the activity of other key channels and receptors, in particular the mechanoreceptors mediating mechanical allodynia.

Transcriptional Regulation and Maintenance of Peripheral Sensitization

Peripheral nerve injury changes the expression of neurotransmitters, neuromodulators, their respective receptors, transcription factors, and other molecules implicated in cell signaling (7). Microarray studies have revealed changes in the expression of several hundred genes in DRG in response to peripheral nerve injury (32, 33). For example, in DRG neurons, nerve injury increases expression, at both mRNA and protein or peptide levels, of the neuropeptides galanin, NPY (neuropeptide Y), PACAP (pituitary adenylate cyclase-activating polypeptide), VIP (vasoactive intestinal peptide), and CCK (cholecystokinin) (7, 124); the GPCRs CCKB and adrenoceptor α1B (7, 43, 55); the neurotrophic factors BDNF and bFGF (basic fibroblast growth factor) (125, 126); the calcium channel subtype α2δ; the sodium channel Nav1.3 (59, 91); and the transcription factors ATF-3 and c-Jun (127, 128). Furthermore, nerve injury induces gene expression in uninjured neurons adjacent to injured neurons in the same DRG or, at a larger separation, in uninjured neurons in the DRG adjacent to an injured DRG. For instance, substance P, BDNF, TRPV1, the purinoreceptor P2X3, and mGluR5 are all induced in uninjured DRG neurons after peripheral axonal injury (34, 101, 129131) (Fig. 2) and have been implicated in hypersensitivity to thermal and mechanical stimuli (34, 101, 132, 133).

Fig. 2.

Transcriptional regulation in primary sensory neurons. Inflammatory mediators and elevated impulse activity subsequent to axonal injury (injury discharge) activate MAPKs (p38, ERK, and JNK) and CaMK, leading to the nuclear translocation of these kinases. Phosphorylation of their target transcription factors results in the transcriptional activation of distinct sets of genes in injured and uninjured neurons (nearby intact neurons) of the DRG, which act to maintain neuropathic pain.

The signaling pathways involved in regulating the expression of the above-mentioned genes have begun to be explored. p38 is activated by nerve injury in more than 35% of DRG neurons, including C-fiber and some A-fiber neurons (105), and is essential for NGF-induced TRPV1 gene expression (104). Nerve injury activates ERK in about 10% of DRG neurons, including both C- and A-fiber neurons, and ERK activation is necessary for the nerve injury–induced increase in BDNF (107). JNK activation appears to be very persistent and involves more DRG neurons (106, 111). Translocation of p-p38 and pERK to the nucleus results in the activation of several transcription factors, including CREB (cAMP response element–binding protein), Elk-1, c-Myc, and NF-κB. Because the transcription factor c-Jun is expressed only in injured DRG neurons (61, 128), JNK activation must occur in injured cells but may not be limited to these cells alone. PKA and CaMK can also be translocated to the nucleus to activate CREB. The nuclear form of CaMKIV is expressed in intact DRG neurons but is down-regulated after nerve injury (134). Therefore, this kinase probably does not play an important role in stimulating gene expression in the injured neurons. The transcription factor NF-κB is typically activated by proinflammatory cytokines and growth factors and appears to be up-regulated in DRG neurons by nerve injury (135).

It is noteworthy that peripheral nerve injury also down-regulates many genes in the injured neurons, including the genes encoding the neuropeptides substance P and CGRP, and the ion channels Nav1.8, Nav1.9, and TRPV1 (7, 61, 130). The signaling pathways involved in this down-regulation have yet to be investigated. However, growth factors such as NGF and GDNF can prevent and reverse injury-induced down-regulation in TrkA- and c-Ret–responsive C-fiber neurons of DRG (136, 137). Decreased NGF levels in injured neurons not only lead to down-regulation of substance P, CGRP, Nav1.8, Nav1.9, and TRPV1 but also cause up-regulation of galanin (138, 139). Therefore, continuous infusion of NGF not only can prevent down-regulation of substance P and CGRP but also can suppress the up-regulation of galanin after nerve injury (139). In contrast, NGF levels are increased in adjacent noninjured DRGs, leading to up-regulation of BDNF in uninjured neurons. This latter action appears to be critical for the pathology, because anti-NGF can alleviate neuropathic pain (101). In addition, LIF is induced in non-neuronal cells in the injured DRG and acts on adjacent neurons to induce galanin gene expression (140). GDNF levels in the DRG are decreased after nerve injury (141), and intrathecal infusion of GDNF at high doses(12 to 24 μg per day) normalizes the phenotypic changes (e.g., reverses both up- and down-regulation of gene expression) and correspondingly alleviates pain after nerve injury (136, 142). High doses of artemin, a member of the GDNF-related family with receptors (GFR-α3) specifically expressed in the PNS, can both normalize phenotypic changes and reverse neuropathic pain without eliciting motor and sensory abnormalities (143). The GDNF receptors c-Ret and GFRα1−3 are expressed in DRG neurons, and GFRα1 is up-regulated after nerve injury (136, 144). Although these receptors normally activate PI3K and ERK pathways, the signaling pathways involved in the high-dose action of GDNF and artemin in neuropathic pain conditions are not known.

Central Mechanisms of Pain After Nerve Injury

Increased long-lasting discharge of peripheral nociceptors also modifies neuronal phenotype and function in the CNS. These plastic changes in the spinal cord and in the brain, leading to increased responsiveness, are referred to as central sensitization (Fig. 3). Central sensitization plays a major role in the heightened pain that often follows nerve injury. Central sensitization contributes importantly to the pain produced by normally innocuous low-threshold afferent inputs (e.g., mechanical allodynia) and the spread of hypersensitivity to regions beyond injured tissue (3, 1619, 145).

Fig. 3.

Posttranslational regulation in dorsal horn neurons of the spinal cord and induction of central sensitization. Peripheral nerve injury increases release of the neurotransmitter glutamate and the neuromodulators substance P and BDNF from the central terminals of primary afferents. Activation of NMDA receptors and Ca2+-permeable AMPA receptors in the spinal cord leads to Ca2+ influx. Stimulation of group I mGluRs and NK-1 receptors activates PLC and stimulates Ca2+ release from intracellular stores. The increase in intracellular Ca2+ activates CaMK and PKC. BDNF activates the TrkB receptor and thereby the tyrosine kinase Src. PKC, CaMK, and Src all increase the activity of NMDA receptors. Activation of ERK by TrkB, PKA, PKC, and PI3K can suppress the activity (reduced open probability) of potassium channel Kv4.2 and also increase the trafficking and membrane insertion of AMPA receptors, both of which result in larger depolarizations after glutamate receptor activation. CaMK and PI3K positively regulate the activity, trafficking, or both of AMPA receptors. These posttranslational steps result in an overall increased sensitivity of dorsal horn neurons to afferent input, contributing to the generation of central sensitization.

Pain and Memory: Analogy Through Similar Mechanisms

Investigations of neural plasticity in the spinal cord and in the hippocampus reveal similar mechanisms for central sensitization and long-term potentiation (LTP), which are respectively believed to underlie generation of pain hypersensitivity and memory (145147). LTP can also be recorded in dorsal horn neurons (146). Whereas long-term memory and the late phase of hippocampal LTP require gene transcription, short-term memory and early-phase LTP only require posttranslational modifications. This same dichotomy appears to apply to central sensitization and pain hypersensitivity: The induction of central sensitization and pain hypersensitivity appear to require only posttranslational processes, whereas their maintenance seems to depend on changes in gene transcription (145). Although neural circuits in the spinal dorsal horn and hippocampus are quite different, glutamate is the predominant transmitter in both of these circuits and activation of NMDA receptors is essential for the plasticity of both dorsal horn and hippocampal neurons. The activity of NMDA and AMPA receptors and trafficking of these receptors can be regulated by several protein kinases through posttranslational regulation. Moreover, the ERK-CREB pathway might be critical for the transcriptional regulation of genes in both the dorsal horn and hippocampus that maintain central sensitization and LTP (145). Although direct injury to both peripheral nerve and spinal cord produces a strong immune activation in the spinal cord, which unquestionably contributes to the pathogenesis of neuropathic pain, this phenomenon is not likely to participate in the hippocampal-centered memory processes.

Posttranslational Regulation and Induction of Central Sensitization

Glutamate is the major excitatory neurotransmitter in the spinal cord; it is released from primary afferents and acts on ionotropic AMPA, kainate, and NMDA receptors as well as on mGluRs. Whereas AMPA receptors are important for the rapid excitatory synaptic transmission of physiological nociception [but see (148)], NMDA receptors play a critical role in plasticity in the CNS.

NMDA receptors are more likely to be activated under conditions of neuropathic pain, through several mechanisms. First, glutamate transporter expression in the spinal cord is reduced after nerve injury, and synaptic glutamate concentration is thus likely to be elevated (149). Second, nerve injury produces a loss of inhibition (disinhibition, a reduction of inhibitory GABAergic input onto dorsal horn neurons partially due to a loss of second-order afferent neurons) (2, 150). Removal of GABAergic inhibition facilitates A-fiber–mediated excitatory synaptic transmission in an NMDA-dependent manner (151). Third, nerve lesion reduces the expression of the potassium chloride coexporter KCC2 in the spinal cord, leading to change in Cl concentration and a resulting reversal of the GABAA receptor–conducted current, so that stimulation of GABA and glycine receptors will produce depolarization (152) rather than hyperpolarization. Depolarization, which removes a voltage-dependent Mg2+ block of the NMDA receptors’ channel, is required for activating NMDA receptors.

Strong activation of the NMDA receptor plays a critical role in the induction of neuropathic pain. Thus, NMDA antagonists can prevent and reverse neuropathic pain (153155). Interestingly, cone snail venom–derived antagonists selective for the NR2B (conantokin G) or NR2A and NR2B (conantokin T) subunits of NMDA receptor can reduce neuropathic pain without producing the obvious side effects that are caused by other NMDA receptor antagonists (156). Activation of PKC results in phosphorylation of the NMDA receptor in dorsal horn neurons, which can relieve its voltage-dependent Mg2+ block independent of depolarization and enable glutamate to generate a greater inward current through the NMDA receptor at resting membrane potentials (157). Consistent with these findings, PKC inhibitors reverse neuropathic pain (158, 159), and in transgenic mice lacking PKCγ (an isoform mainly expressed in inner laminae II of the dorsal horn), neuropathic pain is diminished (160).

Postsynaptic density proteins PSD-95 and PSD-93 are important molecules that regulate the trafficking of NMDA receptors and, particularly, their appropriate localization at the cell surface. Therefore, knockdown or knockout of these proteins in the spinal cord leads to decreased NMDA receptor–mediated current and reduced neuropathic pain (161163). Such changes may occur through a reduction in functional receptors on the postsynaptic membrane or by an alteration in their biochemistry. For example, CaMKII’s docking to the NMDA receptor is disrupted by structural manipulation of PSD proteins, which probably compromises this enzyme’s direct influence on activation of the receptor and the subsequent sensitization of pain behavior after nerve injury (161). BDNF released from presynaptic terminals can act on postsynaptic TrkB receptors in the dorsal horn, which is likely to increase the activity of NMDA receptors through tyrosine phosphorylation of the NR2A/B subunit by Src (3, 164) (Fig. 3); consistent with this, blocking of TrkB signaling reduces neuropathic pain (165). The NMDA receptor is also regulated by the redox state of the cell (166), such that shifts in a neuron’s oxidative metabolism due to altered activity, locally active stimuli (such as extracellular ATP), or modified vascular flow after local injury or inflammation (which limits the local tissue oxygen supply) will all modify NMDA receptor activity. Clinical trials have shown that agents that act as NMDA receptor antagonists (ketamine, dextromethorphan, amantadine) can alleviate neuropathic pain (167, 168).

ERK is typically stimulated by growth factors acting through receptor tyrosine kinases. For example, BDNF activates ERK in dorsal horn neurons through the TrkB receptor (169). However, intense noxious stimulation produces a more robust ERK activation in dorsal horn neurons, an activation that appears to involve NMDA and mGluRs (170, 171) and subsequent activation of PKA and PKC (172) (Fig. 3). ERK may also be activated by PI3K after NMDA receptor activation (173). Although inhibition of ERK does not affect basal pain sensitivity, it does block the induction of central sensitization (170, 171). Interestingly, there is enhanced ERK activation in the spinal cord after either Aβ- or C-fiber stimulation under conditions where neuropathic pain already exists, presenting a possible connection between injury-induced changes in one pathway and mechanical allodynia (174). ERK activation appears to stimulate the trafficking and membrane insertion of AMPA receptors, an important step for increasing synaptic strength in both spinal cord and hippocampal neurons (145, 175). Another possible target through which ERK may mediate the induction of central sensitization is the A-type potassium channel Kv4.2, which opens transiently to suppress repetitive firing in many mammalian neurons. ERK can phosphorylate and thereby deactivate this channel, leading to increased excitability (145, 176, 177) (Fig. 3). Inasmuch as ERK can be activated by several protein kinases (e.g., PKA, PKC, Src, and PI3K), it can be inactivated by phosphatases such as MKP-1 (MAP kinase phosphatase–1) (178) and PP2A (protein phosphatase 2A) (179) (Fig. 3). Neuronal activity not only induces ERK activation but also rapidly increases the expression of MKP-1 (178, 180), an immediate early gene protein product that can limit the duration of ERK activation. Intrathecal injection of okadaic acid, a general PP2A inhibitor, enhances central sensitization by prolonging capsaicin-induced mechanical hyperalgesia and allodynia (181). It is likely that ERK activation is involved in this okadaic acid effect.

Transcriptional Regulation and Maintenance of Central Sensitization

Although intensive research has shown that peripheral nerve injury produces robust changes in the expression of numerous genes in primary sensory neurons, gene regulation in postsynaptic dorsal horn neurons has been studied less. Whereas peripheral inflammation and nerve injury often produce different or opposite changes in gene expression in DRG neurons (e.g., up-regulation of substance P and CGRP after inflammation but down-regulation of these two neuropeptides in injured neurons after injury) (7), they appear to induce similar changes in the spinal cord. Thus, nerve injury, like hindpaw inflammation, induces up-regulation of the proteins c-Fos, neurokinin-1 (NK-1), and TrkB and the peptide prodynorphin (182187). Although nerve injury produces a down-regulation of substance P in small-size C-fiber neurons of the DRG, it induces transcription of this peptide in larger-size Aβ-fiber neurons (188). This phenotypic switch (189) appears to underlie mechanical allodynia through up-regulated NK-1 activation in the spinal cord (190). NK-1 antagonists can prevent and reverse nerve-injury pain (hyperalgesia and allodynia) in animal models (191, 192), although clinical trials suggest that they may not be effective in humans (193). Although dynorphin has classically been thought to act on κ opioid receptors to produce antinociception, accumulating evidence indicates that this peptide is overproduced in pathological conditions and becomes pronociceptive through its activation of NMDA receptors (194196). However, in DREAM (downstream regulatory element antagonistic modulator)–deficient mice, where the prodynorphin gene is derepressed, neuropathic pain is diminished (197). It is unclear whether other genes are involved in this phenotype, which could explain the inconsistency between these observations.

ERK is translocated to the nucleus of dorsal horn neurons upon phosphorylation, there activating transcription factors and regulating gene transcription to maintain central sensitization and nerve-injury pain (Fig. 4). Inhibiting ERK activation blocks up-regulation of prodynorphin and NK-1 in response to inflammation (185), and ERK is required for noxious stimulation–induced c-Fos expression after a neuropathy (198). In addition to c-Fos, other transcription factors such as CREB and Elk-1 are also regulated through ERK (116, 172) (Fig. 4). Phosphorylation and activation of CREB is increased in dorsal horn neurons not only after inflammation (199) but also after nerve injury (200); notably, there are CREB binding sites (CREs) in the promoter regions of the genes encoding c-Fos, COX-2, NK-1, prodynorphin, and TrkB (145, 185, 201). Knockdown of CREB results in decreased neuropathic pain (202). ERK also appears to be involved in the up-regulation of the spinal CB1 cannabinoid receptor after nerve injury (203). This up-regulation may explain the preserved antinociceptive effect of exogenous cannabinoid compounds under neuropathic pain conditions. In a spinal cord injury model, the sodium channel Nav1.3 is up-regulated in dorsal horn neurons, and knockdown of this channel can alleviate neuropathic pain (204). Ca2+ channel α1E and α2δ subunits are also up-regulated in the spinal cord by nerve injury (205). The complete story of how signal pathways in peripheral and central neurons are altered by injury and inflammation, and how they regulate different gene expression to maintain pain after nerve injury, will be a far-reaching and complex tale.

Fig. 4.

Transcriptional regulation in second-order sensory neurons of the dorsal horn. Signaling pathways in dorsal horn neurons can be activated not only by glutamate, substance P, and BDNF released in an activity-dependent way from primary afferents but also by diffusible inflammatory mediators (such as IL-1β, PGE2, and NO) released from glial cells. The subsequent activation of ERK, PKA, and CaMK activates the transcription factors CREB, Elk-1, and c-Fos, leading to increased transcription of prodynorphin, NK-1, Cox-2, Nav1.3 Na+ channel, and α2δ Ca2+ channel subunit, among others. The protein or peptide products of these genes, as well as of other unknown genes, act together to maintain heightened excitability and synaptic responsiveness, and thus support central sensitization and neuropathic pain.

Neuro-Glial and Neuro-Immuno Interaction

The CNS is under constitutive immune surveillance by CNS-resident glial cells (e.g., microglia) and CNS-infiltrating immune cells. MAPK cascades play an essential role in the mediation of innate immune responses by signaling the expression of IFMs in different tissues (206). Although pain hypersensitivity was originally thought to result exclusively from altered activity of neurons, there is accumulating evidence for a role of glial cells in the pathogenesis of pain (207209). Several proinflammatory cytokines, such as IL-1β, TNF-α, the prostanoid PGE2, and enzymes involved in the synthesis of IFMs, such as iNOS and COX-2, as well as chemokines, reactive oxygen species (ROS), and cell adhesion molecules (210), are synthesized in microglia and astroglia and are likely involved in pain hypersensitivity (207214). Both microglia and astroglia are activated in the spinal cord after nerve injury, and it is noteworthy that drugs that suppress glial activation can prevent and reverse neuropathic pain (105, 207, 209, 215217). However, the particular roles of microglia versus astrocytes in the creation of pain after nerve injury remain largely unclear because of a lack of specific inhibitors for the different types of glial cells.

Nevertheless, recent evidence suggests a role for spinal microglia in the genesis of neuropathic pain. Minocycline, an antibiotic that can inhibit the activation and proliferation of microglia (218), can prevent the development of neuropathic pain, although it cannot reverse it when administered 5 days after the injury (216). Complementing this pharmacological effect, intrathecal injection of activated microglia can induce pain (219). Expression of the ATP receptor P2X4 per se is induced in the spinal cord after nerve injury but only in the microglia, not in neurons; knockdown of this receptor with antisense oligodeoxynucleotides reduces neuropathic pain (219). The chemokine receptor CCR2 is also expressed in spinal microglia, and neuropathic pain from injury is diminished in mice lacking this receptor (220). Evidence thus supports an active role for spinal microglia in the initiation of neuropathic pain, although the cellular and molecular mechanisms remain obscure and the functional interactions with signaling neurons have not been described.

Neuroinflammation, defined as infiltration of immune cells into the site of CNS injury, plays an important role in various neurodegenerative diseases (221). Peripheral nerve lesion produces infiltration of immune cells not only to the injured nerve but also to the spinal cord in the CNS (210). Disruption of the blood-brain barrier (BBB), a process that involves matrix metalloproteinases, is implicated in neuroinflammation in brain (221, 222). It is unclear whether the BBB in the spinal cord is pathologically disrupted or otherwise altered by peripheral nerve injury and whether the infiltration of leukocytes is substantial enough to contribute to neuropathic pain.

MAPK Activation in Spinal Glia and Pain After Nerve Injury

As a stress-activated protein kinase, p38 can be activated by cellular stress, lipopolysaccharide (LPS), and inflammatory cytokines (18, 116). LPS appears to activate p38 through Toll-like receptors (TLRs). p38 inhibitors can suppress inflammation in different animal models by blocking the synthesis of IFMs such as IL-1β, TNF-α, and the enzyme COX-2, which produces PGE2, in several cell types (18, 116). p-p38 is constitutively present in glial cells in the spinal cord (105, 223). Although peripheral inflammation elicits only a moderate increase of spinal p-p38 levels in glia (104), nerve injury, in particular spinal nerve ligation, elicits a remarkable increase of spinal p-p38 levels that selectively appears in microglia (105, 108, 224) (Fig. 5).

Fig. 5.

p38 activation in spinal microglia in neuropathic pain. Ligation of the L5-spinal nerve, a widely used model of neuropathic pain (A), activates p38 in spinal microglia on the ipsilateral side of the medial dorsal horn (B). Double immunostaining indicates that the active form of p38 (p-p38) is exclusively colocalized with OX-42, a marker for microglia. Scale bar, 100 μm.

Inhibitors of p38 do not affect basal pain perception, but they can reduce inflammatory hyperalgesia as well as pain hypersensitivity induced by substance P, NMDA, and the HIV envelope glycoprotein gp120, and they suppress COX-2 up-regulation and PGE2 release in the spinal cord (104, 223, 225, 226). Furthermore, daily administration of the p38 inhibitor SB203580 can prevent neuropathic pain for up to 2 weeks after spinal nerve injury, and one intrathecal injection of SB203580 will reverse neuropathic pain for several hours (105, 224). In experimental sciatic inflammatory neuropathy, elevated pain can be both prevented and reversed by the p38 inhibitor CNI-1493 (109).

In addition to p38, ERK is also transiently activated in spinal microglia in the first few days after peripheral nerve injury and contributes to the generation of neuropathic pain (110, 227). In a later stage of neuropathic pain(2 to 3 weeks after injury), ERK activation is shifted from microglia to spinal astroglia and ERK activation in astrocytes contributes to the maintenance of neuropathic pain (110). Another MAPK family member, JNK, also appears to be activated in spinal astroglia at later stages of nerve injury and may help to maintain neuropathic pain (111, 228).

Upstream Mechanisms for MAPK Activation in Spinal Glia

Ectopic discharge in injured and adjacent intact peripheral neurons can stimulate release of neurotransmitters and neuromodulators (such as glutamate, ATP, substance P, and the chemokine fractalkine) from primary afferents, or from hyperactivated dorsal horn neurons (nociceptive or pain transmission neurons). These substances are likely to activate the MAPKs (p38, ERK, or both) in microglia (Fig. 6), an activation that appears to be somatotopically organized, as it is restricted (at least initially after injury) to the general area where the injured spinal nerve terminates in the ipsilateral spinal cord (105). Microglia express receptors for glutamate (229), substance P (226), and ATP (219, 230, 231) as well as the chemokine receptors CCR2 (220) and CX3CR1 (209). Some of these receptors (e.g., P2X4 and CX3CR1) appear to be up-regulated in spinal microglia after nerve injury (209, 219). Although normally not expressed in the CNS, cannabinoid receptor CB2 is up-regulated in spinal microglia under conditions of neuropathic pain (232). A CB2 agonist alleviates neuropathic pain, possibly by inhibiting MAPK activation, but the cellular target of the agonist is unknown (209, 233). Microglia also express TLRs for bacterial and viral proteins (234). Peripheral nerve injury induces a rapid increase (within 4 hours) of TLR-4 mRNA expression in the spinal cord (210). The role of bacterial and viral infection on neuropathic pain is not unclear, nor is it known whether these TLRs also have endogenous ligands. However, TLRs and the IL-1 receptor belong to the same superfamily and share similar signaling pathways, both leading to the activation of MAPK cascades (235) (Fig. 7).

Fig. 6.

After peripheral nerve injury, hyperactive primary sensory neurons can release neurotransmitters and neuromodulators such as glutamate, substance P, ATP, and fractalkine from their central axons, leading to activation of signaling pathways and thereby activation of spinal glia (microglia or astroglia). Bioactive substances released from hyperactive dorsal horn neurons can also activate spinal glia. The activated glia synthesize and release multiple inflammatory mediators (IFMs) such as IL-1β, TNF-α, IL-6, NO, and PGE2, which can further act either directly on dorsal horn neurons that transmit pain (nociceptive neurons) or indirectly on primary afferents, both leading to increased sensitivity of the nociceptive neurons (central sensitization). Moreover, these IFMs will diffuse to and act on surrounding glia, causing their further activation. Positive feedback loops between IFMs and glial activation and between glia and neurons lead to the enhancement and maintenance of neuropathic pain.

Fig. 7.

Signal transduction in spinal microglia. Peripheral nerve injury produces selective changes in microglia, such as activation of the MAPK family proteins p38 and ERK. Stimulation of both GPCRs (for instance, chemokine receptors CX3CR1, CXCR4, and CCR2, as well as NK-1) and ionotropic receptors (P2X4, P2X7, and NMDA) elevates intracellular Ca2+, thereby activating p38 and ERK. p38 and ERK can also be activated by IL-1β and TNF-α as well as by viral and bacterial infection through TLRs. Activation of these two MAPKs then results in increased expression, through NF-κB or other transcription factors, of genes encoding membrane receptors CD11b, TLR-4, P2X4, CB2, and CX3CR1; secreted inflammatory mediators, including IL-1β, IL-6, TNF-α, PGE2, NO, and BDNF; and cytoskeletal proteins. Gene transcription can also be stimulated through the JAK/STAT pathway after activation of some cytokine receptors. Whereas secretory molecules will sensitize dorsal horn neurons, the increased production of cytoskeletal proteins participates in morphological changes (such as migration and phagocytosis) necessary for glial function in "injured" tissue.

Initial activation of p38 and ERK in spinal microglia shortly after injury (within the first day) is restricted to the superficial dorsal horn (105) because of restricted diffusion of the signal molecules released from primary afferents and hyperactivated central neurons, although this activation is likely to extend further with continued primary afferent input (105) and can spread to encompass several segments of the spinal cord. This widespread activation is probably triggered by non-neuronal signals. For instance, activated microglia could release multiple IFMs, leading to further activation of surrounding glia (Fig. 6). Infiltration of leukocytes to the spinal cord after nerve injury (210) may also cause widespread activation of MAPK. MAPK appears to be important for glial activation, and p38 appears to be a target of the microglial inhibitor minocycline (229), which is effective in preventing pain after nerve injury (216).

Although both microglia and astrocytes are activated in the spinal cord under conditions of neuropathic pain, the microglial response to injuries precedes those of other glial cell types (236). A recent study of reverse transcription polymerase chain reaction reveals that nerve injury induces increased expression of microglial markers (e.g., integrin αM, CD14) in a few hours, but the increase in the astrocyte-specific marker GFAP (glial fibrillary acidic protein) is not detected until 4 days after the injury (210). Microglial activation is likely to cause astroglial activation (209, 236). ERK and JNK are activated in astroglia at a later stage of nerve-injury pain than ERK activation in microglia.

Downstream Mechanisms for MAPK Action in Spinal Glia

Most studies on MAPKs and injury focus on different mechanisms of gene regulation. First, all MAPKs can induce gene transcription by activating transcription factors such as CREB, ATF-2, Elk-1, c-Jun, and NF-κB. Second, p38 also increases mRNA levels by increasing the stability of mRNA. Third, MAPKs (ERK and p38) can increase protein levels by increasing translation (18, 104, 237). Activation of MAPKs in microglia and astroglia results in production of multiple IFMs, including the proinflammatory cytokines IL-1β, IL-6, and TNF-α, and the biosynthetic enzymes (such as COX-2 and iNOS) for IFMs PGE2 and NO (210). IL-1β induces the synthesis of COX-2 and thus PGE2, leading to the sensitization of dorsal horn neurons (238). In addition, IL-1β and TNF-α might directly stimulate or modulate the activity of primary afferents to increase the release of neurotransmitters and neuromodulators in the spinal cord (18, 104, 208). These cytokines also increase the mechanical sensitivity of the receptive field mapped in recordings from the dorsal root (239). However, TNF-α, COX-2, and iNOS may only be involved in the early development of neuropathic pain (240242). Other mediators such as growth factors (BDNF, NGF) may be involved in the maintenance of neuropathic pain. Activation of JAK (Janus kinase) by cytokines (such as IL-2, IL-3, IL-6, etc.) results in gene transcription through the JAK/STAT (signal transducer and activator of transcription) pathway (Fig. 7). It is noteworthy that the anti-inflammatory cytokine IL-10 may negatively regulate this pathway and suppress the production of inflammatory cytokines and p38 activation (209, 243) (Fig. 7). In particular, activation of cytokine receptors and TLR results in the activation of the transcription factor NF-κB, which is involved in MAPK (p38 and JNK) cascades (235, 244). NF-κB is normally retained in cytoplasm by IκB inhibitor proteins. The phosphorylation of IκB by means of its kinase, IKK, results in IκB degradation, which enables nuclear translocation of NF-κB, thus stimulating the transcription of genes coding many IFMs (245). Because specific inhibition of IKK reduces neuropathic pain, this pathway is also functionally implicated in this pathology (246).

In addition to regulating gene expression in glial cells, MAPKs may also play a role in posttranslational regulation. For example, phospholipase A2 (PLA2) is positively regulated by p38 at what may be a rate (substrate)–limiting step for the synthesis of PGE2, and ERK can activate TNF-α converting enzyme (TACE), resulting in the cleavage of pro-TNF to generate the released and mature TNF-α (247). MAPKs may also phosphorylate cytoskeleton proteins, causing morphological changes of glia (Fig. 7).

Neuro-Glial and Neuro-Immuno Interaction in Peripheral Nerves

Most existing animal models of neuropathic pain involve traumatic injury to the peripheral nerves. Peripheral nerves normally contain not only axons but also Schwann cells and mast cells as well as fibroblasts, endothelial cells, macrophages, and dendritic cells (248). After nerve injury, these cells produce multiple cytokines and chemokines to attract infiltrating leukocytes. The IFMs produced by resident cells and infiltrating leukocytes in the damaged nerve could directly stimulate the axons, producing spontaneous activity and also increasing their response to external stimulation. Mast cells are shown to contribute to the development of hyperalgesia by releasing histamine to recruit neutrophils and monocytes to the injured nerve (249). Nerve biopsies from inflammatory and noninflammatory neuropathies indicate that cytokine levels are directly correlated with the degree of axonal degeneration and with neuropathic pain (250). Schwann cells are the myelinating glia of the PNS. Demyelinating diseases are often associated with tactile allodynia, and focal lysolecithin-induced demyelination of peripheral afferents results in neuropathic pain behavior (251). Degeneration of myelinated efferent fibers induces spontaneous activity in uninjured C-fiber afferents (252). Peripheral demyelination and neuropathic pain behavior also appear in periaxin (Prx)–deficient mice. Mice lacking the Prx gene, which encodes periaxin to stabilize myelin in Schwann cells, show demyelination and neuropathic pain behavior (253). The signaling pathways used by these non-neuronal (e.g., Schwann) cells are not clear but are likely to involve MAPK cascades (254).

Concluding Remarks and Future Directions

Insight into the intracellular signaling pathways in neurons and glia that contribute to neuropathic pain may disclose novel molecular and cellular mechanisms underlying the genesis of this disorder. These mechanisms include the induction of sensitization by posttranslational regulation of molecules in peripheral and central neurons involved in sensation, particularly nociception, and the maintenance of sensitization by transcriptional regulation, all through the action of multiple protein kinases.

This review is limited to nerve injury–induced neuropathic pain in animal models. Clinical neuropathic pain is much more complicated. Whereas neuropathic pain in animal models is usually treated acutely and after only days or a few weeks of existence (i.e., post-injury), clinical neuropathic pain has often been present for months to years before treatment begins and treatment itself may sometimes have a course of months to years, conditions that are difficult and expensive to model in small animals. As a result, both the mechanisms underlying the chronic phase of neuropathic pain and the responses to prolonged dosing of potential therapeutics tend to be underinvestigated by animal testing.

At present, few drugs are effective in treating neuropathic pain. Patients do not respond well to nonsteroidal anti-inflammatory drugs (6), and serious side effects of opioids have limited their usage (255). Gabapentin and clinically available NMDA receptor antagonists are only partially effective in a fraction of patients, and local anesthetic-type Na+ channel blockers, delivered intravenously, also have limited effectiveness and are dose-restricted by side effects. Opioid therapy may be enhanced by adjunct administration of NMDA receptor antagonists or calcium channel blockers (256). The ability of growth factors to alter neuronal phenotypes in the adult, and their specific actions under conditions of experimental neuropathic pain, underline their therapeutic potential. Animal studies of the GDNF family of neurotrophic factors have generated very promising results, producing long-term reversal of experimental neuropathic pain. However, clinical trials of neurotrophic factors for degenerative diseases, including peripheral neuropathy, have often been disappointing (257). Because glial cells are becoming recognized as having an important role in pain regulation, targeting glia might open a new avenue for the management of neuropathic pain (209). A search for membrane or intracellular proteins (258) induced in microglia or astroglia after neuropathy may reveal new targets for pharmacological intervention, targets that are selectively present in these activated glia, providing cellular specificity and, perhaps, a reduction of side effects. The combined use of glial targeting drugs with opioid or other conventional neuropathic pain–alleviating drugs may provide a more effective means to treat this debilitating disease.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
  186. 186.
  187. 187.
  188. 188.
  189. 189.
  190. 190.
  191. 191.
  192. 192.
  193. 193.
  194. 194.
  195. 195.
  196. 196.
  197. 197.
  198. 198.
  199. 199.
  200. 200.
  201. 201.
  202. 202.
  203. 203.
  204. 204.
  205. 205.
  206. 206.
  207. 207.
  208. 208.
  209. 209.
  210. 210.
  211. 211.
  212. 212.
  213. 213.
  214. 214.
  215. 215.
  216. 216.
  217. 217.
  218. 218.
  219. 219.
  220. 220.
  221. 221.
  222. 222.
  223. 223.
  224. 224.
  225. 225.
  226. 226.
  227. 227.
  228. 228.
  229. 229.
  230. 230.
  231. 231.
  232. 232.
  233. 233.
  234. 234.
  235. 235.
  236. 236.
  237. 237.
  238. 238.
  239. 239.
  240. 240.
  241. 241.
  242. 242.
  243. 243.
  244. 244.
  245. 245.
  246. 246.
  247. 247.
  248. 248.
  249. 249.
  250. 250.
  251. 251.
  252. 252.
  253. 253.
  254. 254.
  255. 255.
  256. 256.
  257. 257.
  258. 258.
View Abstract

Navigate This Article