In this study, we characterized the morphological and functional profile of CNS glial cells in response to peripheral nerve injury, to local oro-facial infection and to inflammation. All these three insults triggered a specific pattern of mechanical allodynia on the lower lip. Our results demonstrated that 1) Mental nerve injury triggered both microglia and astrocyte activation in the region of the central terminals of the damaged nerve fibres. However these activated glial cells exhibited low capacity to produce proinflammatory cytokines. The spatial and temporal profile of this glial activation was closely correlated with the development of mechanical allodynia following nerve ligation. 2) LPS injection, a model of oro-facial infection, triggered an acute microglial reaction, starting from the CVOs, which diffused progressively into the CNS parenchyma; and was characterized by a robust induction of IκB-α and IL-1β, IL-6 mRNAs in the CVOs. Most likely this is an acute CNS inflammatory response, not directly relevant to the local mechanical hypersensitivity following LPS injection. No significant astrocyte activation was observed. 3) CFA injection, a model of oro-facial sterile inflammation, only triggered minor changes on microglial morphology and a transient induction of IκB-α mRNA, restricted in the CVOs, and a slight increase of IL-1β and IL-6 mRNAs at late time point, which is not likely correlated to the pain development. Therefore we conclude that nerve injury-induced microglial activation is characterized by morphological changes with a low profile of cytokine expression, whereas infection-induced microglial activation is accompanied by high levels of cytokine production. Astrocyte response was significant only following nerve injury.
Three animal models used in this study provided diversified stimuli to examine the plasticity of CNS glial cells and their potential contribution in the development of pain
Lesion on the peripheral nerve is a major cause for neuropathic pain. Many functional and phenotypic changes occur along pain conducting pathways, from peripheral terminals to higher centers in the brain. In addition, as most animal models of neuropathic pain entail lesions of mixed nerves such as the sciatic nerve, which possesses sensory, motor and sympathetic fibers, the model that we used in this study, loose ligation on the mental nerve offered a unique opportunity to study the effects of lesion on sensory fibers, since in rats the mental nerve is almost exclusively sensory in origin [18, 19]. As a component of the cell wall of Gram-negative bacteria, lipopolysaccharide endotoxin (LPS) is a well established agent to mimic the endogenous effect during infection and sepsis. Systemic injection of LPS induces fever and septic shock and is associated with an increase of different cytokines, including TNF-α, IL-1β and IL-6, in the blood circulation [20, 21] and in the brain [22, 23] in a pattern similar to that seen in natural infection. However, the impact of a local, peripheral injection of LPS on the CNS and relationship with local LPS induced pain behavior were not yet clarified. CFA is an immune response enhancer, effective in stimulating cell-mediated immunity and may lead to increase of the production of certain immunoglobulin in the absence of an infection. Due to its painful reaction and its immunopotentiator properties, CFA is frequently used to study sterile inflammatory reaction-associated pain, including arthritis [24, 25]. In this study, we used CFA as a model of a sterile inflammatory condition.
Nerve injury and infection/inflammation signal the CNS glial cells through different pathways
Within the central nervous system, microglia and astrocytes represent two highly reactive intra-parenchymal cell populations. Microglia, which originate from the bone marrow, provide immune surveillance of any exogenous or endogenous perturbation . Activated microglia are characterized by a specific morphology, proliferation, increased expression of cell surface markers or receptors, and changes in functional activities such as migration to areas of damage, phagocytosis, and production/release of proinflammatory substances . Developmentally different from microglia, astrocytes, normally support neurons by maintaining metabolic homeostasis, and also respond to different insults. The prominent morphological feature of astrocyte activation is hypertrophy. Functionally, this activation is manifested by increased production of a variety of trophic factors and by an alternation of astrocyte function in the maintenance of homeostasis. The fact that peripheral nerve injury can induce microglial/astrocyte activation has been demonstrated in several chronic neuropathic pain models [28–30]. Our detailed systematic analysis of mental nerve injury induced microglia and astrocyte activation in the subnucleus of trigeminal complex revealed a similar pattern of glial activation to that previously described in the spinal cord following sciatic nerve injury . CNS glial response following peripheral nerve damage occurs through a neural pathway, at the region where the central terminals of the damaged nerve fibers end. In the current study, we observed that mental nerve ligation evoked microglia and astrocytes activation predominated at Sp5C where nociceptive primary afferents terminate, and at Pr5 area where larger A-beta afferent fibers synapse. Microglial response preceded that of astrocytes and astrocyte activation outlasted transient microglial activation. While dramatic morphological changes were observed in both microglial and astrocyte responses, cell proliferation also occurred, but the increase in cell numbers was higher for the microglial cell population. We did not find evidence of any significant changes in other pain related regions, either through direct projection from the trigeminal brain stem complex, or through multisynaptic pathways using relays in the reticular formation and adjacent brainstem areas, including thalamus, superior colliculus, periaqueductal gray and others (data not shown), although some previous studies observed glial activation in pain modulatory circuitry, such as rostral ventromedial medullar in a similar nerve injury model . In contrast, inflammatory signals triggered in the periphery following either LPS or CFA injection likely reach the CNS via an endocrine-like mechanism. Indeed, in our study, the subcutaneous injection of LPS triggered an acute brain microglial response, including both morphological changes and a robust induction of cytokine expression, which was initiated from the CVOs, membrane lining on the ventricles and choroid plexus. What is common to these regions is that they are devoid of blood-brain barrier, therefore blood borne molecules have easier access to the CNS than in other regions. This is in agreement with previous observations that the direct action of LPS on cells bearing LPS receptor-CD14 within these organs initiates a rapid activation of myeloid derived cells, including microglia and perivascular macrophages [32, 33]. Cytokines and other inflammatory molecules released by these activated microglia/macrophages spread the signals into the adjacent parenchyma in a migratory-like pattern . We show that CFA seems capable of stimulating the glial reaction in the brain, but the pattern was similar to that following LPS.
Characteristics of nerve injury and infection/inflammation-induced microglial activation
Microglia share the same origin with circulating monocytes and peripheral tissue macrophages and indeed represent CNS tissue specific macrophages. Therefore, it is not surprising that there could be a wide range of phenotypic and functional similarities between microglia cells and peripheral macrophages. Mainly based on in vitro observations, macrophage activation has been described as belonging to two main types, classically activated (or inflammatory) M1 and alternatively activated M2 macrophages. Each subpopulation is characterized by a distinct profile of gene expression and, accordingly, each has different functions [9, 34]. M1 macrophages produce high levels of proinflammatory cytokines and mediators which are vital components of host defence . M2 macrophages have rather an anti-inflammatory profile and are involved in tissue remodelling . A recent in vivo study demonstrated that, in injured spinal cords, macrophages/microglia cells exhibit two phenotypes, while the M1 is neurotoxic, M2 promotes a regenerative growth response . In the current in vivo study, we provide evidence that there is a distinctive phenotype of microglial activation in response to nerve lesion, inflammation and infection. Microglial activation following nerve lesion was characterized by morphologic changes and cell proliferation, but low levels of proinflammatory cytokine expression. In this, it resembles the M2 macrophage activation in peripheral tissue. In contrast, LPS-induced microglial activation possessed a typical inflammatory phenotype, compared to the one of peripheral M1 macrophages. Thus, the morphological and functional changes in microglia reflect altered activation states induced by different signals that arise from injured neurons or from circulating immune mediators. In addition to distinct phenotype of macrophage/microglia activation, our results also suggested that, indeed, the microglial cells may be a heterogeneous population, since monocytes in the blood have been characterized as inflammatory (CCR2high, CX3CR1low and GR1+) and resident (CCR2low CX3CR1 high GR1-) subpopulations . Specific monocyte populations might give rise to specific tissue macrophages , including microglia.
Potential contribution of CNS glial cells to pain in the different experimental conditions
In the last decades, considerable evidence was obtained supporting the concept that activated glial cells can contribute, in one way or another, to experimental pain states. More convincing data were accumulated in models of neuropathic pain following nerve lesions. The concept of glial involvement in pain modulation was initially developed following studies in which there was spinal glial activation , and was further supported by evidence of trigeminal glial activation in oro-facial pain conditions. Intracisternal administration of inhibitors for p38 or ERK1/2, MAPKs phosphorylated on activated glial cells, produced an anti-allodynic effect following infraorbital nerve injury . Microglia inhibitor, minocycline, reduced the tactile hypersensitivity following transection of inferior alveolar nerve and mental nerve . Studies applying the astrocyte inhibitor, fluoroacetate, demonstrated the important role of hyperactive trigeminal astrocytes in oro-facial neuropathic pain . How these activated glial cells affect pain processing is always a hot topic and has not been fully answered. Proinflammatory cytokines have been suggested as signalling molecules between activated glial cells and their surrounding neurons to enhance pain transmission. In the absence of injury, central application of IL-1β and TNF-α induced allodynia and/or hyperalgesia [44, 45]. Intrathecal administration of IL-1ra and soluble TNF-α [46, 47] reduced enhanced nociceptive states. Unexpectedly, our results revealed that, in contrast with LPS induced microglial activation, mental nerve injury-induced CNS microglial activation had a specific phenotype with very low cytokine production profile, which is consistent with what we have observed previously in sciatic nerve injury-induced cytokine expression within the spinal cord . Although such a low level of cytokine expression on highly activated microglia leads us to question the importance of glia-derived cytokines as a link in glia-to-neuron interactions and in the pathogenesis of neuropathic pain, we still cannot exclude the involvement of proinflammatory cytokines in this specific condition. Indeed, since cytokines act in autocrine and paracrine manner, they might need only to be produced in very low amounts to be functional. It is also possible that the level of proinflammatory cytokines is not enough to maintain hyperactivity of surrounding neurons, but might contribute to neuropathic pain though an indirect pathway, e.g., triggering functional changes of astrocytes and modulating the integrity of the blood-brain barrier. At the same time, it should be noticed that in the CNS microenvironment, proinflammatory cytokines are not exclusively secreted by activated glial cells. Indeed, damaged central primary afferent terminals have been shown to release these inflammatory molecules  and even blood born circulating cytokines can contribute to increase the local concentration in the microenvironment (unpublished personal data). In addition, growing evidence suggests that many other candidates expressed by activated microglia can contribute to modulate pain processing, such as cell surface receptors (P2X4, P2X7, CCR2, CX3CR1, etc) [50–54], enzymes  and complement components . Evidence of CNS glial involvement in peripheral inflammatory pain is less substantial. Our results did not provide neither spatial nor temporal correlation between LPS- and CFA-induced CNS glial reaction and enhanced pain behaviour. However, some studies using other inflammatory stimuli, such as application of inflammatory irritant mustard oil in the tooth pulp, demonstrated that inhibition of P38 MAPK signaling and inhibition of astrocytes metabolic processes can completely abolish central sensitization in Sp5C nociceptive neurons [57, 58].
In summary, this study identifies the distinctive phenotype of CNS glial cells in response to remote nerve injury and to local infection/inflammation, which all produced enhanced pain behaviour. It also specifies that both microglial and astrocytic activations are multi-dimensional. Functional and morphological changes were not time-locked, as one could be detected in the absence of the other, depending on the stimulus that triggered activation. Further functional studies will help to delineate whether and how CNS glial cells contribute to different pathological pain conditions.