We previously reported that an infusion of Ad-GLT-1 into specific brain areas efficiently increased GLT-1 expression at 2 and 5 days, and this expression remained stable up to 8 days after the infusion [29
]. In the present study, we confirmed that intraspinal infusion of the recombinant adenovirus successfully transferred the GLT-1 gene into the spinal cord surrounding the infusion site between 2 and 21 days. Our previous finding showed that infusion of the recombinant adenovirus was accompanied by minimal tissue damage and had no effect on the immunoreactivity of GFAP surrounding the infusion site [29
], suggesting that no toxicity or gliosis resulted from the adenoviral infection. Although an adenovirus transferred the LacZ gene efficiently into both neurons and glial cells [31
], the majority of cells with adenovirus-mediated gene transfer in the dorsal horn were astrocytes when the adenoviruses were infused into the spinal cord, consistent with previous reports [32
]. Because GLT-1 is expressed mainly in astrocytes, the adenovirus-mediated expression system seemed likely to be suitable for the present study. In the ventral horn, the transgene expression was observed in astrocytes and motor neurons. Of all the neurons within the spinal cord, motor neurons have been shown to efficiently uptake and express recombinant adenoviruses [32
]. Although we did not determine the influence of GLT-1 gene transfer in motor neurons in this study, at least, it did not induce behavioral abnormalities and abnormal gait, and had no effect on the escape behaviors from the mechanical and thermal nociceptive stimuli.
Using the recombinant adenoviruses, we showed that gene transfer of GLT-1 into the spinal cord had no effect on the acute nociceptive responses to mechanical and thermal stimuli in naive rats. These data are supported by previous studies using glutamate transporter activators such as MS-153 [34
] and riluzole [4
]. In contrast, intrathecal injection of glutamate transporter inhibitors into naive animals elevates the spinal extracellular glutamate level and produces spontaneous nociceptive behaviors and hyperalgesia [25
]. These results suggest that maintaining a low extracellular glutamate level following an increase in spinal GLT-1 protein does not affect acute spinal pain transmission under normal conditions. In addition, the important finding in this experiment is to show no evidence that adenoviral infection in the spinal cord influenced the basal nociceptive threshold.
Here we showed that spinal gene transfer of GLT-1 reduced i.pl. carrageenan/kaolin-induced inflammatory mechanical hyperalgesia and pSNL-induced tactile allodynia, when the spinal infusion of the adenoviruses was performed 7 days before i.pl. injection of carrageenan/kaolin and pSNL surgery, respectively. Furthermore, it recovered the extent of the spinal GLT-1 expression in the membrane fraction that was decreased at 7 days following pSNL. These results suggest that adenovirus-mediated overexpression of GLT-1 in the spinal cord prevents the induction of inflammatory and neuropathic pain. Consistent with the present data, riluzole, which reduces extracellular glutamate by activating glutamate transporters [28
], inhibits the induction of inflammatory and neuropathic pain [4
]. Glutamate release from primary afferent neurons in the spinal dorsal horn is enhanced by peripheral inflammation and nerve injury [1
]. Excessive and prolonged activation of spinal glutamate receptors and subsequent intracellular adaptation in the postsynaptic dorsal horn neurons lead to a prolonged increase in neuronal excitability, called central sensitization, which produces pathological pain [7
]. Indeed, inhibition of glutamate release by anticonvulsant agents attenuates the induction of hyperalgesia following peripheral inflammation or nerve injury [4
]. Our results suggest that the increase of glutamate uptake activity following overexpression of spinal GLT-1 protein decreases the excessive extracellular glutamate level; this in turn inhibits the generation of neuronal plasticity related to central sensitization following peripheral inflammation and nerve injury.
In the present pSNL model, tactile allodynia was produced also in the contralateral to the nerve injury site, called mirror-image pain. It is considered that mirror-image pain arises from altered contralateral spinal processing of incoming sensory information [36
]. The present study showed that contralateral allodynia was also prevented by the unilateral gene transfer of GLT-1 into the ipsilateral spinal cord, although the unilateral gene transfer did not spread to the contralateral spinal cord. Milligan et al. demonstrated that mirror-image pain may be due to spinal glial activation and release of proinflammatory cytokines, which may spread and reach the contralateral spinal cord to generate sensitization of the contralateral dorsal horn neurons [37
]. The spinal glial activation following peripheral nerve injury is dependent on spinal NMDA receptor activation [38
]. Consequently, it is conceivable that the ipsilateral gene transfer of GLT-1 may attenuate the activation of glial cells and the release of proinflammatory cytokines in the ipsilateral spinal cord by inhibiting a glutamate-dependent pathway, which has the effect of preventing the induction of sensitization of contralateral, as well as ipsilateral, dorsal horn neurons.
In contrast to the ability of exogenous GLT-1 to prevent the induction of pathological pain, when the spinal infusion of the adenoviruses was performed 7 or 14 days after pSNL, spinal gene transfer of GLT-1 did not reverse the established allodynia. Many previous studies have shown that the expression of GLT-1 and glutamate uptake activity are decreased 7–14 days after nerve injury [5
]. Similarly, we found a decrease in GLT-1 expression in the membrane fraction of the spinal cord at 7 days following pSNL, although the total amount of GLT-1 protein was not changed in the present pSNL-induced neuropathic pain model in contrast to previous reports using other neuropathic pain models [19
]. The spinal glutamatergic system contributes to not only the induction of pathological pain, but also to its maintenance [6
]. Sung et al. reported that riluzole twice daily for 4 days beginning on postoperative day 5 gradually reversed the maintenance of thermal hyperalgesia and mechanical allodynia after chronic constriction nerve injury [19
]. However, riluzole has multiple actions in addition to activating glutamate transporters, including blockade of sodium channel α-subunits, glutamate receptors, and γ-aminobutyric acid uptake and the stabilization of voltage-gated ion channels [42
]. The inhibitory effect of riluzole on the maintenance of neuropathic pain may therefore be due to actions other than the activation of glutamate transporters. Our present findings suggest that the expression of GLT-1 in the spinal cord plays little role in the maintenance of neuropathic pain, at least in the present conditions. Otherwise, the expression of other EAAT subtypes, GLAST and EAAC1, was also decreased in the spinal cord of neuropathic pain model animals [19
]. Reduction of glutamate uptake activity via GLAST and EAAC1 may contribute to the maintenance of neuropathic pain.
An accumulating amount of evidence suggests that spinal astrocytes, as well as microglia, contribute to pathological pain [43
]. Early studies indicated that spinal astrocytes are activated in diverse models of pathological pain [37
], and blocking the activation and function of spinal astrocytes prevents and reverses hyperalgesia and allodynia [37
]. Because GLT-1 is expressed mainly in astrocytes, the present findings may further support the importance of spinal astrocytes in pathological pain.