|Home | About | Journals | Submit | Contact Us | Français|
The discovery that glial activation plays a critical role in the modulation of neuronal functions and affects the spinal processing of nociceptive signalling has brought new understanding on the mechanisms underlying central sensitization involved in chronic pain facilitation. Spinal glial activation is now considered an important component in the development and maintenance of allodynia and hyperalgesia in various models of chronic pain, including neuropathic pain and pain associated with peripheral inflammation. In addition, spinal glial activation is also involved in some forms of visceral hyperalgesia.
We discuss the signalling pathways engaged in central glial activation, including stress pathways, and the neuron-glia bidirectional relationships involved in the modulation of synaptic activity and pain facilitation. In this expanding field of research, the characterization of the mechanisms by which glia affect spinal neuro-transmission will increase our understanding of central pain facilitation, and has the potential for the development of new therapeutic agents for common chronic pain conditions.
Pain is commonly defined as a multidimensional process in which physical, emotional and perceptual integration serves the primary function of survival in safeguarding the individual from potential sources of a tissue damaging stimulus. In healthy condition, this process is adaptive, transient and has a protective role. The pain signal initiates from the stimulation of peripheral nociceptor nerve terminals, via activation of specific receptors/ion channels. Sensory information is conducted via small diameter C fibers which terminate within distinct regions of the dorsal horn of the spinal cord (laminae I-IV), from where the signal is transmitted to the brainstem, thalamus and higher cortical centers (see review (1)). Small diameter fibers include unmyelinated C fibers (slow conducting and responding to high intensity activation by noxious heat, mechanical and chemical stimuli) and myelinated Aδ fibers (intermediate conduction velocity, responding to high intensity activation). Visceral afferents constitute only 10% of all afferents and are divided into splanchnic and pelvic afferents projecting to the thoracic and sacral spinal cord respectively. Significant viscerovisceral and viscerosomatic convergence occurs at the level of the dorsal horn (2). Visceral afferent terminals are present in the mucosal epithelium, the serosa, the muscles and the myenteric ganglia, and are able to monitor changes in the gut milieu and participate in the transmission of visceral sensory information (3).
Pathological pain refers to conditions characterized by hyperalgesia and allodynia, in which maladaptive neuroplastic changes lead to persistent increased perception and responsiveness to noxious stimuli, or response to normally non-noxious stimuli. Such neuroplastic changes can occur in primary afferent terminals (peripheral sensitization) but also in the spinal cord and in the brain (central sensitization) thereby altering the processing of sensory information (4).
While the concept of sensitization of nociceptive pathways has long been considered as a neuro-centric plastic mechanism, it has become increasingly recognized that spinal glia are dynamic modulators of this neuronal network (5) and accumulating evidence supports a crucial role of glia in central sensitization and pathological pain observed in various experimental animal models of peripheral inflammation, spinal injury or nerve injury (6-8). The aim of this review is to discuss the most recent findings in this rapidly expanding field of glial involvement in pain facilitation, and to address the emerging area of the role of glia in various models of chronic pain, including visceral pain.
Glia in the CNS, including the spinal cord, refer to several distinct non-neuronal cell types including oligodendrocytes, microglia, and astrocytes. Oligodendrocytes play an important role in myelinating axons in the CNS. The study of their full role in the modulation of neuronal function is still in its infancy. Microglia are heterogeneous cell types (perivacular and resident), constitute 5-12% of all cells in the CNS (9) and share a common mesodermal lineage with macrophages. Studies based on in vivo two-photon microscopy in transgenic mice revealed that microglia perform continuous biochemical sensing and interpretation of their environment, providing a close surveillance of homeostasis of their milieu (10, 11). This surveillance function, active in resting state, is processed via highly motile protusions and a variety of surface receptors, which enable prompt reaction to environmental changes. A wide range of conditions (stimuli related to trauma, ischemia, invading pathogens, or stress) which are interpreted as threat to the structural or functional integrity of the CNS, can trigger a shift in microglia activity state, becoming activated or “reactive”. Astrocytes constitute 40-50% of all glial cells and in contrast to microglia, are derived from the neuroectoderm (12). They become activated in response to the same of array of stimuli that activate microglia but also in response to by-products released by activated microglia. The unique intimate contact of astrocytes with neurons, enwrapping the synapses, provides the anatomical support for neuronal trophic function, regulation of extracellular ions (via voltage gated channels), modulation of neurotransmitters release, and the regulation of neuronal survival, differentiation, and formation of synapses (13-15). Astrocytes include different cellular subpopulations depending on the relative expression of glia fibrillary acidic protein (GFAP), Na+ currents /K+ voltage-dependent or independent currents, glutamate transporters, S100 calcium-binding protein beta (S100B) or expression of the proteoglycan NG2 (16) (17) (for review (6)). The astrocytic Na/K+ ATPase-mediated removal is considered as the major mechanism of potassium regulation in the synapse (18). Astrocytes also play a major role in the clearance of synaptic glutamate via high affinity glutamate reuptake transporters expressed on their surface such as GLT1 and GLAST (19). Depending on the nature and intensity of the stimulus, glia activation is characterized by a variety of phenotypic transformations engaging distinct intracellular signaling cascades. Activation of these signaling systems, results in different patterns of morphological changes and secretory activities, making it difficult to establish definitive makers of glial activation (20). Across the literature, the status of glial activity has been assessed by a combination of observations of changes in morphology (retracted processes and hypertrophy), expression of specific cellular proteins and cell surface markers (histocompatibililty complex (MHC-II) or complement receptor 3 (CD11b or OX-42 for microglia and GFAP for astrocytes), kinase activity (phosphorylation of p38 mitogen-activated protein kinase [p38 MAPK] and extra cellular signal-regulated kinases [ERK] or c-Jun N-terminal kinase [JNK] as well as by measuring the release of a variety of mediators including pro-inflammatory cytokines, chemokines, ATP and glutamate. Glia activity can be also assessed by the expression of molecules such as glutamate transporters, and gap junction proteins (connexin 43, CX43) and toll like receptors TLR (21, 22).
The role of spinal glia in pathological pain has been mainly studied in experimental models of neuropathic and inflammatory pain (23). Neuropathic pain models include manipulations such as nerve transection, resection or crush, complete or partial ligation of the sciatic nerve, spinal nerves or lumbar spinal cord roots, but also spinal cord injury or spinal infection with the human immunodeficiency virus (HIV-1) glycoprotein envelope gp120 (24). Persistent hyperalgesia and allodynia are observed after injury and occur in the skin areas innervated by the injured nerve and in other regions (extraterritorial and mirror pain). It is associated with a robust activation of spinal microglia and astrocytes and the release of pro-inflammatory mediators (25-27). Reports that pharmacological agents with relative specificity to glia cells, such as minocycline (an antibiotic targeting mostly microglia), or fluorocitrate (which disrupt microglia and astrocytic metabolism by blocking aromatase), can prevent and/or reverse the increased pain behavior but also reduce the release of pro-inflammatory mediators, support the role of spinal glia activation in the mechanism underlying neuropathic pain (7, 28-30). Inflammatory pain models are generally initiated by peripheral tissue damage or inflammation using subcutaneous injections of complete Freund's adjuvant, phospholipase A2, snake venom, formalin or zymozan. Hypersensitivity is observed at the site of injection and in adjacent non inflamed tissues and does not persist once the inflammation is resolved (23). The role of glia activation in inflammatory pain is supported by studies showing spinal microglia and astrocyte activation following subcutaneous CFA injection or intra-articular CFA injection in a model of arthritis (23). It is further supported by reports showing the inhibitory effect of cytokines, NO, or glia blockers (fluorocitrate and propentofylline) on inflammatory hyperalgesia (31). However, conflicting data have been reported on the temporal profile of microglia and astrocytes activation in inflammatory models (23, 32) and more studies are needed to better assess the role of spinal glia in the acute, sub-acute and chronic phases of CFA-induced inflammatory pain.
Most studies related to glial-neuron interactions underlying chronic pain have been gained from experimental models of inflammation or injury related to somatic pain responses. In view of the growing number of studies showing the role of glia in central sensitization, a process which is thought to contribute to visceral hyperalgesia in various animal models (3), the role of spinal glia in the modulation of visceral pain has only been recently addressed. One study using a model of visceral hypersensitivity induced by neonatal colonic irritation previously shown to exhibit peripheral and central sensitization, (33) showed an increase spinal immuno-reactivity for OX42 (indicating microglia proliferation) in sensitized rats compared to controls (34). Visceral hypersensitivity to colonic distension in these animals was blocked by an acute treatment with minocycline. In another report from Riazi et al. (35), intracolonic administration of 2,4,6-trinitrobenzene sulfonic acid (TNBS) in rats, known to produce visceral hyperalgesia, was found to induce microglia activation associated with increased levels of TNF-α in the hippocampus. These results are consistent with the literature on the contribution of glia to pain facilitation induced by traumatic peripheral stimulation.
Our group evaluated whether spinal glia activation occurs in a rat model of chronic visceral hyperalgesia induced by chronic psychological stress (chronic water avoidance stress, WA) (32). We observed stress-induced increase of p38 phosphorylation in OX42 positive cells, and this effect was blocked by treatment with minocycline. We also demonstrated that stress-induced visceral hyperalgesia can be blocked by intrathecal treatment with minocycline or the p38 inhibitor SB203580, supporting a functional role of spinal microglia activation in the development of visceral hypersensitivity and providing the first demonstration of microglia activation in the spinal cord in the absence of peripheral/nerve inflammation or injury. The modulatory influence exerted by spinal microglia on visceral nociception was further supported by the findings that spinal injection of the microglia activator fractalkine in naïve rats, induces visceral hyperalgesia, also shown by Saab et al. (34) These data corroborate previous work showing that spinal administration of fractalkine produces mechanical allodynia and thermal hyperalgesia, whereas blockage of spinal CX3CR1 attenuates neuropathic pain or prevent its development (36, 37). Together, these data strongly support the concept that transmission of visceral nociceptive signals may be enhanced in various conditions of spinal microglia activation.
In addition to the evidence that nerve injury induces activation of glia within the dorsal horn of the spinal cord, It has recently been shown that glia may also plays a role in supraspinal regions involved in the modulation of pain perception (such as the anterior cingulate cortex) (38). In a rat model of chronic constriction of the infraorbital nerve (CCI-ION) (as a model of orofacial painful neuropathy), Wei et al. found a prolonged astrocytic reaction and increased expression of cytokines including tumor necrosis factor- (TNFα) and interleukin-1β (IL-1β) in astrocytes and their receptors in RVM neurons after nerve injury (39). Intra-RVM microinjection of glial inhibitors and neutralization of endogenous TNF-α and IL-1β significantly attenuated CCI-induced mechanical hyperalgesia and allodynia. Further supporting the activation of glia in other CNS structures, microglia activation was described in the gracile nucleus in a model of peripheral nerve injury (40) and Zhao et al demonstrated up-regulation of the chemokine CCL21 (known to activate microglia) in the ventral posterolateral (VPL) nucleus of the thalamus after spinal cord injury (41). Together these reports suggest that glia activation in other CNS structures involved in the modulation of pain ought to be examined and may provide new insight into the modulating role of CNS glia on nociceptive function.
However, other studies have not confirmed the role of glia in supraspinal pain modulation. For example, Zhang et al., using transgenic mice in which all microglia are labeled by green fluorescence protein (GFP), performed a systemic mapping of microglia in major pain-related brain areas in control mice and mice with nerve injury (42). Results from this study indicated that microglia showed a uniform distribution throughout the CNS, and peripheral nerve injury selectively activated microglia in the spinal cord dorsal horn and related ventral horn. In contrast, microglia were not activated in supraspinal regions of the CNS, including the anterior cingulate cortex (ACC), prefrontal cortex (PFC), primary and secondary somatosensory cortex (S1 and S2), insular cortex (IC), amygdala, hippocampus, periaqueductal gray (PAG) and rostral ventromedial medulla (RVM).
Although the focus of this review is on the role of spinal glia in pain facilitation, recent reports suggest a possible role of satellite glial cells (SGCs) in the modulation of the sensory transmission. SGCs, surrounding the sensory neuron somata in dorsal root ganglion (DRG) neurons, are often thought as the equivalent of CNS astrocytes. Although of different morphologies, they share similar features with astrocytes such as the expression of K+ ion channels and glutamate recycling components (GLAST) (43). Interestingly, GFAP in SGCs, while not detectable in resting state, is increased following nerve injury. There are also reports of increased expression of GFAP and IL-1β in SGCs following inflammation (44). While recent evidence indicates that activation of sensory ganglionic SGCs can modulate the excitability of primary nociceptive neurons (45), further studies are needed to establish which factors activate satellite glia and whether SGCs play an important role in pathological pain.
In conditions of peripheral tissue inflammation or nerve injury, immune cells are recruited at the site of injury and release pro-inflammatory mediators such as ATP, proinflammatory cytokines, chemokines, prostaglandins and nerve growth factors (NGF), which may act as sensitizing agents on peripheral nerve terminals, increasing the sensitivity and excitability of nociceptors. In response to intense stimulation or when injury or inflammation is sustained, ongoing excitation of primary nociceptive neurons leads to the release of neurotransmitters including glutamate, substance P, BDNF, chemokines and ATP from central terminals in the dorsal horn. These mechanisms are well known to contribute to subsequent plasticity of spinal sensory neurons (7). Such changes may occur pre- or post synaptically in the dorsal horn but also in supra-spinal pain processing areas receiving projections from these neurons, leading to a chronic pain facilitatory state. The activation of postsynaptic NMDA receptors, release of NO, enhanced release of substance P and EAA from presynaptic terminals, combined with an increased expression of neuronal NK1R are the best characterized signaling mechanisms engaged in the process of central sensitization. These mechanisms have been reported in both somatic and visceral pain facilitation (7, 46).
Interestingly, similar signaling pathways have been observed in rodent models of stress-induced facilitation of pain, in the absence of peripheral nociceptive stimuli (47), and there is evidence showing glial responsiveness to environmental stressors (48). The fact that glia can be activated via activation of nociceptive pathways, and by environmental stressors may have implications for the role of stress in the development and maintenance of hyperalgesic states. A number of studies have demonstrated increased pro-inflammatory cytokines within selected areas of the brain in response to experimental stressors in rats (49-53) and there is evidence suggesting that microglia are the primary cellular source for these mediators (54, 55). Chronic restraint stress in mice was found to induce brain microglia activation and proliferation (48). In the hippocampus, restraint stress induces microglia activation (56) and chronic unpredictable stress reduces brain GFAP expression (57). Stress can also sensitize microglia immune reactivity, such that microglia can be ‘primed” by stress and exhibit changes in phenotype that may lead to potentiation of CNS immune response to further peripheral stimulation or immune/pro-inflammatory challenge (58).
Although the signals triggering glia activation remain to be fully understood, several candidate molecules involved in glia activation in response to both inflammatory and stress stimuli have been documented.
CNS glia activation may occur in response to circulating factors or signals originating from activated monocytes and lymphocytes in the blood stream. A progressive activation of resident microglial cells has been described following systemic bacterial or viral infections or in experimental models using systemic lipopolysaccharide injection (59). Interestingly CNS microglia can be activated when the blood brain barrier remains intact, via initial glial response in circumventricular organs (60). Microglia and astrocytes in the spinal cord can also be activated by signals initiated by neuromodulators and neurotransmitters released from sensitized spinal terminals of primary afferent neurons or by excited postsynaptic sensory neurons in the spinal dorsal horn (23).The expression of common receptors for a wide range of mediators in both glia and neurons, combined with the release of similar neuromodulators by both cell types has posed a challenge in identifying glia specific molecules. Candidate molecules involved in glia activation signaling include neurotransmitters such as substance P or glutamate, but also purinergic agents, opioids, chemokines and glucocorticoids, some of which are discussed below.
While the involvement of substance P and glutamate in spinal sensitization has been well characterized, their role as signaling mediators in glial activation is incompletely understood (61-63). Although reports have shown NK1 receptor expression on microglia and astrocytes (64, 65), a direct implication of substance P-induced glial activation in central sensitization has yet to be demonstrated. Studies have also suggested CGRP as a potential mediator signaling glia activation (66).
Glutamate receptors (NMDA, AMPA and kainite receptors) and glutamate uptake transporters are expressed at the cell surface in astrocytes (67). Functional NMDARs have been reported on astrocytes and in contrast to NMDAR in neurons, their activity is independent of the presence of magnesium (68, 69). Activation of microglia p38 has been reported in response to NMDAR activation (70) and microglia also have the potential to release glutamate (71). Astrocytes are able, via their glutamate transporter system, to take up glutamate from the extracellular system, but also to release glutamate (72). These combined functions suggest that astrocytes and microglia may respond to stimulation by releasing glutamate, which may feed further glial activation via GluRs.
Among the different ATP receptors expressed on microglia (including P2X7, P2Y2, P2Y6 and P2Y12), P2X4 is thought to play a major role in pain facilitation. While expressed at low level in normal conditions, P2X4 is upregulated specifically on microglia in response to peripheral nerve injury and inflammation (73) and the inhibition of ATP signaling or P2X4 function in mouse microglia was found to delay and reduce neuropathic pain (74, 75). In recent studies, fibronectin was proposed to be involved in microglia P2X4 upregulation (75). In addition, P2X7 is also expressed on microglia and transgenic mice lacking P2X7 fail to develop neuropathic pain (76). The role of P2X7 in pathological pain is supported by recent studies using antagonists developed for the P2X7 receptors, showing an inhibitory effect on tactile allodynia in 3 different models of neuropathic pain in rats (77). While other pharmacological studies have shown a possible role of several other purinergic receptors in neuropathic pain, the specificity of the pharmacological tools targeting purinergic receptors has been argued and further studies are necessary to clearly establish their specific role in microglia activation and nociceptive signaling.
Chemokines and their receptors are expressed on peripheral immune cells, and neurons, but also in glial cells. The monocyte chemoattractant protein-1 (MCP-1 or CCL2) has been proposed as a potential candidate molecule signaling microglia activation. Several reports have shown that the expression of MCP-1 and its receptor CCR2 are markedly increased in DRGs and spinal cord neurons after peripheral nerve injury (78) and coincide with microglial and astrocytes activation. Mice lacking the CCR2 receptors were found to exhibit a reduced phosphorylation level of p38 in spinal microglia and do not develop tactile allodynia after nerve injury (79). Finally, studies reported that spinal injection of a MCP-1 neutralizing antibody can reduce microglia activation and that MCP-1 injected intrathecally produces an increase in microglia number (80). While these data strongly suggest the role of MCP-1 in microglia activation, a direct effect via CCR2 on microglia has yet to be demonstrated.
Another member of the chemokine family, fractalkine (CXCL1), which specifically binds to the receptor CX3CR1 (expressed uniquely on microglia) has been documented to play a crucial role in spinal pain signaling. Fractalkine is highly expressed in naïve DRG and dorsal horn neurons and in astrocytes after nerve injury (81), while it is not expressed in microglia. Spinal fractalkine is bound to the extracellular membrane of neurons in which it is expressed and its cleavage from the membrane may occur following peripheral nerve injury or neuronal excitation and involves cathepsin S, a cysteine protease expressed in microglia. Increased expression of CX3CR1 was found in the spinal cord after nerve injury (82).
The hypothesis that fractalkine released from neurons and astrocytes may induce microglia activation, leading to hyperalgesia, is supported by a number of studies. For example, intrathecal injection of fractalkine causes p38 activation in microglia and nerve injury induced spinal microglia activation is blocked by a CX3CR1 neutralizing antibody (82). Recent work showed that intrathecal injection of fractalkine in naïve rats produces thermal hyperalgesia and this effect is blocked by a neutralizing antibody against CX3CR1 (83). In addition, fractalkine-induced mechanical allodynia, thermal hyperalgesia, or visceral hyperalgesia is blocked by the microglia blocker minocycline (37, 47). Together, these data support the potential role of fractalkine in neuron to glia signaling involved in pain facilitation. It is not known whether fractalkine expression is altered in stress conditions.
Both microglia and astrocytes express toll-like receptors (TLRs) including TLR2, TLR3 and TLR4 (mostly expressed in microglia) (84). In general, the function of TLRs is to recognize exogenous pathogens and alarm signals (bacteria, viruses, and molecular components released by damage cells). TLRs (TLR1, 2, 4) are up-regulated in neuropathic pain conditions and TLRs activation is known to induce the release of proinflammatory cytokines such as IL-1β, IL-6, TNFα (84). Several studies using pharmacological tools or gene depletion have demonstrated the specific role of TLR4 in neuropathic pain (upregulated after nerve injury) and microglia activation (85). While several ligands have been shown to induce TLR4 activation, such as fibronectin, heat shock proteins (HSPs), saturated fatty acids, ATP or opioids, the nature of the mediators involved in TLR4 mediated changes in nociception in neuropathic pain remains unclear (85).
Glucocorticoid receptors (GRs) are expressed on microglia in culture and in vitro treatment of microglia with glucocorticoids was found to decrease the production of TNF-α and IFN-γ in response to LPS (86), consistent with the generally accepted immunosuppressive properties of GCs. In contrast, in a recent study, Sierrra A, et al. (87) demonstrated that isolated microglia from adult mice express glucocorticoid receptors and reported a downregulation of these GRs after a LPS challenge suggesting that this mechanism may be involved in the suppression of the anti-inflammatory actions of endogenous steroid hormones on the immune system, contributing to a sustained activation of microglia. Another report indicates increased glucocorticoid immunoreactivity in both microglia and astrocytes in the gerbil hippocampal C1 region after transient ischemia (88). Together, these data suggest that glia might be a target of GCs in the CNS, and may respond differently depending on the duration and the context of exposure. Further studies are needed to assess whether this signaling pathway may contribute to the modulation of peripheral nociception.
Activation of glial cell surface receptors evokes intracellular signaling cascades. The initial increase in intracellular calcium leads to the activation of calcium sensitive signaling molecules such as p38. Other cascades such as activation of ERK1/2, JNK or NFκB have been reported in glia.
Increased p38 phosphorylation (indicative of p38 activation) has been demonstrated in spinal microglia in several animal models of pathological pain induced by peripheral inflammation, nerve injury, spinal cord injury (89, 90), in visceral hyperalgesia induced by chronic stress (47) and in response to intrathecal injections of various molecules considered candidates for microglia activation, such as fractalkine, TLR agonists or cytokines (74, 82). Whether these effects correspond to a direct or indirect activation remains to be verified. Although contrasting results regarding the effectiveness of treatment at different time points after injury have been reported, pharmacological inhibition of p38 activity was repeatedly shown to attenuate the development of allodynia or hyperalgesia in different models, supporting the role of spinal p38 signaling in microglia in spinal pain processing (91). In general, p38 activation appears to be a transient phenomenon occurring in the early phase of development of hyperalgesia (91, 92). Activation of p38 in microglia has been associated with increased synthesis and/or secretion of several mediators such as COX 2, IL1-β, iNOS, phospholipase A2 (PLA2) and prostaglandins (PGE2) (93) which are known to be involved in the signaling of inflammation and nociception.
ERK was also found to be activated in microglia in the early phase after nerve injury (27). It remains unclear whether phosphorylation of ERK and p38 occur in the same population of microglia and discrepancies exist regarding the timing of ERK activation compared with p38. Interestingly, a sequential activation of ERK1/2 has been observed in a model of spinal nerve damage, with first activation of ERK1/2 in neurons, followed by microglia and astrocytes in the latter phase (27). Astrocyte-microglia signaling may be part of the mechanisms of turn off of some of the signals in microglia, but this remains to be investigated.
Sustained activation of ERK was described in astrocytes in the spinal cord in the later stage of neuropathic pain development (27) suggesting that ERK activation in astrocytes is mostly involved in the maintenance of pain facilitation. This is supported by findings that inhibition of ERK activation using treatment with a MEK inhibitor can reverse nerve injury-induced mechanical allodynia (94). In addition to the transcriptional regulation of several mediators including cytokines, ERK was also found to exhibit post-translational regulatory activity. For example, ERK activates the TNFα converting enzyme (TACE) that cleaves pro-TNF to generate mature TNFα, supporting the role of ERK in the glial signaling leading to the production of proinflammatory and pronociceptive mediators (93).
In contrast to the pattern of activation of pERK and p38, persistent increase of P-JNK has been primarily reported in astrocytes in the spinal cord after spinal nerve ligation (27).The preferential activation of JNK in astrocytes is supported by observations that c-Jun the major transcriptional factor downstream of JNK, is also activated in spinal astrocytes (95) and that factors required for JNK activation, including the transforming growth factor activated kinase 1 (TAK1) are also expressed in spinal astrocytes (96). Intrathecal treatment with a JNK inhibitor was found to attenuate neuropathic pain in the nerve ligation model (97) supporting a role of JNK signaling pathway activation in the maintenance of neuropathic pain. The growth factor FGF2 has been proposed as a potential mediator involved in the up stream mechanisms causing JNK activation in the spinal cord (98). When activated, JNK translocates to the nucleus and phosphorylates a range of substrates, including transcription factors such as c-Jun, leading to gene transcription and to the production of mediators (TNF-α, IL-1β, COX1/2). In addition to transcriptional regulation, non-transcriptional regulation can also be involved. Activation of ERK and JNK in astrocytes, can also regulate the expression of GLT1 and gap junction proteins as well as the release of glutamate or serine D (93).
In summary, there is substantial evidence that MAPKs are activated in spinal glia in models of allodynia and hyperalgesia and play an important role in chronic pain sensitization by signaling engaging inflammatory mediators. Most studies demonstrated that spinal microglial activation precedes astrocyte activation and this pattern corresponds to a sequential activation of p38 and ERK in microglia followed by ERK and JNK in astrocytes in the later phase of neuropathic pain. In addition to the MAPK pathways, the nuclear transcription factor NK-κB has been described as an important mediator involved in the intracellular signaling events occurring in activated astrocytes and microglia and is thought to play an important role in the transcriptional regulation of several genes encoding for pro-inflammatory mediators such as intracellular adhesion molecules ICAM-1, vascular adhesion molecule-1 (VCAM-1), IL8, iNOS, TNF1, IL6, NGF and MMP-9 (99).
Together both the MAPK and NF-kB pathways in microglia and astrocytes respond to a wide variety of stimuli including stress and inflammation. There is strong evidence that cytokines play a crucial role in the activation of these intracellular pathways in glia. This ability of glia to respond to cytokines is a key component of the positive feedback loops engaged in the microglia-astrocyte-neural network leading to the production of more inflammatory mediators, thought to contribute to the mechanisms of central sensitization.
Pro-inflammatory mediators released from glia, including cytokines, chemokines and neuromodulators such as ATP or NO can act on neuronal functions, but also on glial cells themselves. Such reciprocal interactions between glia and neurons are thought to contribute to the activation of a neuro-glial amplification loop leading to the amplification of the pain signals (100). For example, the cytokines TNF-α, IL-1β and IL-6, of which the expression has been found increased in pain facilitation models (101, 102), have been implicated in this positive feedback loop. The fact that glia inhibitors such as minocycline or fluorocitrate can decrease the level of cytokines in models of pain facilitation in rats, support the hypothesis of a glial source for those cytokines (29, 103, 104). In addition, microglial cells can produce IL-1β in vitro (105). In the spinal cord, receptors for cytokines (TNF-α, IL-6, IL-1β) are expressed on neurons and glia and both microglia and astrocytes exhibit activation in response to cytokines (25, 106). For example, activation of IL-1β or TNF-2 receptors in glial cells was found to induce the activation of the NFκB signaling pathways, in turn activating the transcription of further cytokine or iNOS and COX 2 (107), leading to the release of NO and PGE2, which have a well documented role in the modulation of pain processing in the spinal cord. In neurons, activation of cytokine receptors (TNF receptor 1) produce changes in the behavior of sensory neurons at different levels. Some of the long-term changes involve alterations in gene transcription and protein expression, and the subsequent up-regulation of other pro-inflammatory cytokines, and ion channels expression (106). However, TNF-α and other upstream cytokines can produce very rapid changes in neuronal excitability which arise from direct effects of cytokine signaling on the properties of important ion channels, including voltage-dependent sodium channels and transient receptor potential (TRP) channels expressed by sensory nerves (107).
An indirect effect of cytokines on neuronal excitability via the modulation of the glutaminergic functions was also found to play an important component in the neuro-glia matrix. In a series of elegant studies, Guo et al demonstrated a coupling of IL-1β signaling with increased activity of neuronal NMDARs, leading to pain facilitation associated with inflammation (103). In vivo, they showed that the glial inhibitor fluorocitrate and IL-1ra inhibited inflammatory hyperalgesia and inflammation-induced NMDAR phosphorylation. In addition, IL1Rs were found to co localize with neurons and direct application of IL-1β to an in vitro brain stem slice preparation induced an enhanced NMDAR phosphorylation, which was blocked by IL-1ra. The lack of effect of fluorocitrate in this preparation suggests that the effect of IL-1β on NMDAR is downstream to glial activation.
A relationship between glutamatergic transmission and TNF-α was demonstrated in several studies (108). Activation of TNF-α receptors on astrocytes leads to increased extracellular glutamate concentration, which may activate mGluR2 on microglia, which may release more TNF-α in addition to other pro-inflammatory cytokines. Beattie et al showed that glial TNF-α causes an increase in surface expression of neuronal AMPA receptors, thereby increasing synaptic efficacy (109). TNF-α was also found to induce endocytosis of neuronal inhibitory GABA receptors (110). Additionally, the effect of TNF-α on glutaminergic synaptic transmission can be mediated via a modulatory action on the expression of the glutamate transporters (TNF-α-induced downregulation of GLT1) (111). It has been demonstrated that activated astrocytes are less efficient at clearing glutamate under the specific phenotype characterized by low GLT-1 expression, high levels of GLAST, and high levels of pro-inflammatory mediator (112). In various experimental conditions including acute CNS trauma, axonal injury, or ischemia, glutamate excess is observed, and this is generally accompanied by reduced expression of glutamate transporters on astroglia (113).
Pharmacological inhibition of glutamate transporters in the spinal cord produces hypersensitivity to mechanical and thermal stimuli (114), which is blocked by a NMDAR antagonist, supporting the role of astrocytic modulation of glutamate transmission in nociceptive functions. Similarly, an increased concentration of glutamate in the synaptic cleft in the CNS has been reported in experimental stress models in rodents (115-117). In vitro experiments indicate that GLT-1 expression in microglia cultures from rat cerebral cortex is inhibited by corticosterone, although the modulation of GLT1 by glucocorticoids seems to vary depending on the cell type (astrocytes vs microglia) and the experimental set up (in vitro versus in vivo)(118).
ATP released from activated astrocytes has been proposed as another candidate involved in the neuro-glia communication, participating in the facilitation of synaptic transmission. Dorsal horn neurons have also been reported to be excited by ATP (119) and astrocyte-derived ATP may facilitate synaptic transmission by acting on both pre- and post-synaptic neurons. In vitro studies demonstrated that the amount of ATP secreted by astrocytes in response to glutamate stimulation is dose-dependently increased by the presence of SP (120), supporting a facilitatory role of astrocytes in synaptic neurotransmission. The specific role of ATP released from glia in central sensitization and hyperalgesia remains to be determined.
It is now clear that neuron-glia interactions are part of a dynamic and complex system in control of synaptic activity and play a crucial modulatory role in the transmission of nociceptive signals. Although knowledge about the influence of pro-inflammatory cytokines and other mediators on neuronal activity and excitability has greatly increased in the past decade, further studies are necessary to better understand the specific role that glia play in these effects. Similarly, it is important to better understand the pathways leading to CNS glial activation. In the context of stress, sustained elevation of glucocorticoids is a possible signaling mechanism, as evoked in the previous section. One may also hypothesize that alteration of the blood brain barrier permeability may facilitate the passage of circulating mediators originating from the periphery, which may be involved in CNS glial activation. Stress-induced increased expression of cytokine receptors on brain endothelium may also facilitate the passage of pro-inflammatory mediators to the CNS leading to glial activation. In a subset of gastrointestinal disorders, the increased presence of immune mediators in the gut may initiate a cascade of signaling events, via the circulation or via activation of vagal afferent terminals, resulting in subsequent CNS glial activation. Glial activation may in turn be implicated in the symptoms associated with immune hyper-reactivity in the gastrointestinal tract, as observed in chronic inflammatory bowel conditions. Evidence showing such signaling pathways from the gut to CNS glia activation or demonstrating the role of CNS glia in gastrointestinal disorders have yet to be investigated.
To date, evidence of CNS glial activation in humans suffering from chronic pain has not been clearly established. One recent report showed high serum level of IL-8 in the cerebrospinal fluid of patients suffering from fibromyalgia compared with healthy controls, suggesting the presence of an inflammatory context in the CNS in this condition (121). While reports of glial activation in clinical chronic pain conditions remain sparse, increasing amount of data indicate that patients with depression have increased systemic levels of pro-inflammatory cytokines, chemokines and adhesion molecules (reviewed in (122). These observations have led to the development of a new field of investigation on the role of CNS cytokines in the pathogenesis of depression (107) It is interesting to note that the current hypothesis for the inflammatory response in depression involves mechanisms which in part, occur through complex interactions with stress-responsive pathways involving the neuroendocrine and autonomic nervous systems, which are known to be involved in the modulation of the pain pathways. Considering the significant overlap of depression with many other painful conditions such as fibromyalgia, chronic widespread pain or functional gastrointestinal disorders, the newly discovered role of cytokines in depression may have significant implication in the future research in the field of chronic pain. Pro-inflammatory cytokines in the CNS, possibly related to glia activation, may be potential common targets chronic pain conditions, including gastrointestinal conditions, for which effective therapeutic treatment remain a challenge. It is anticipated that the development of pharmacological tools targeting specifically glia or some specific components of the glia-neurons relationship will provides valuable informations that may have important implications in the development of therapeutic tools for the treatment of chronic pain in humans.
In conclusion, the study of neuro-glial interaction has become a dynamic and prolific field of investigation, challenging old established concepts in the neurobiology of chronic pain. It is becoming clear that glia in the CNS are involved in many aspects of neuronal function and play an important role in the modulation of pain signal processing. New mechanisms engaging spinal glia activation have been revealed in the physiopathology underlying chronic neuropathic or inflammatory pain, as well as in stress induced pain models and in depression.
Supported by : RO1 DA026597-01A2