Using DSP-4-induced selective ablation of the noradrenergic fibers, we demonstrate that the analgesic effect of DLX in the STZ-induced PDN depends crucially on the presence of intact noradrenergic fibers. Because our analyses indicated drastic changes in the amount of DBH- and NET-expressing fibers in the dorsal horn and the spinal content of NA in the STZ-treated animals, it is highly likely that the potent anti-nociceptive effect of DLX in the STZ-treated animals is mediated by the pharmacological improvement of the pathologically aberrant regulation of spinal NA systems. The mechanisms underlying these effects are discussed below.
The analgesic effect of DLX depends on an intact NA system
DLX is an SNRI that shows inhibitory potency to NA transporters and 5-HT transporters . The present results support the finding that the analgesic effect of DLX is mediated by its effect on NA transport because the suppression of the DLX effect by DSP-4 pretreatment was clearly observed when DSP-4 treatment was combined with injection of an SSRI (fluoxetine)  (see Results). This conclusion that the presence of serotoninergic fibers is not sufficient to produce the anti-nociceptive effect of DLX in PDN is also supported by a recent finding in STZ-treated rats, that the anti-nociceptive effect of another SNRI, venlafaxine, was completely abolished by yohimbine pre-treatment but was only partially inhibited by pretreatment with p-choloroamphetamine, an agent that degenerates serotoninergic fibers . However, partial but significant reduction of analgesic effect of DLX and another SNRI, milnacipran, by 5-HT receptor antagonists has been described in STZ-treated PDN  and postoperative pain  models of rats. A possible interpretation for these results is that activation of 5-HT receptors might affect extracellular NA concentration, such as through modulating NA release, and that the final common mediator that regulate spinal nociceptive network is the NA system, to which the 5-HT system lies upstream. This is a possibility that might explain why manipulations of either 5-HT or NA system affect the effect of SNRI and eliminating only the NA fibers could completely abolish its analgesic effect, as evidenced in this study. Such a primary role of the NA system in the anti-nociceptive effect of SNRI is also supported by observations in other types of chronic pain models in mice that genetically lack central serotoninergic neurons . In these mice, DLX exerted marked analgesic effects in carrageenan- and formalin-induced pain models to a similar degree as those observed in the wild-type mice, again indicating a secondary involvement of 5-HT system in the analgesic effect of DLX. Altogether, in the chronic model of PDN as used in this study and in other types of chronic pain models, the analgesic effect of DLX requires intact NA systems that are capable of releasing NA from nerve terminals.
Impaired NA homeostasis would underlie exaggerated nociception in the STZ-diabetic model
This specific modulation of the NA system in the analgesic effect of DLX in STZ-treated rats supports the notion that STZ administration induces long-lasting aberrant modification of the NA systems, which leads to pro-nociception. NA is one of the principal mediators of endogenous analgesic mechanisms in the descending pain modulatory system in the spinal dorsal horn [26, 27]. The elimination of NA alone by genetic ablation of DBH or DSP-4 administration potently decreases the nociceptive threshold in mice  and rats [32, 33], as confirmed in this study. Conversely, intrathecal NA administration increases tail-flick latency in normal mice  and rats . Additionally, DSP-4 administration, which drastically increased nociception sensitivity in non-STZ-treated rats, did not further affect the lowered nociceptive threshold in STZ-treated animals in this study (Figure 2). This result is a reminiscence of the absence of otherwise pro-nociceptive effect of 6-hydroxydopamine, an NA synthesis neurotoxin, in STZ-treated mice with lowered nociception threshold . These findings suggest that specific defects in the regulation of NA homeostasis in the spinal cord may underlie the pro-nociception in PDN. In support of this interpretation, the effect of intrathecal NA administration in elevating the nociceptive threshold was markedly more potent in STZ-treated mice than in non-diabetic mice . An increase in the extracellular NA level with DLX would be anticipated because it has been shown, albeit not in a diabetic model, that intravenous injection of milnacipran, which is an SNRI, increases extracellular NA levels in the spinal dorsal horn as measured by microdialysis in anesthetized mice with spinal nerve ligation-induced neuropathy . It is therefore speculated that STZ treatment decreases the spinal NA level, which leads to exaggerated nociception.
Nevertheless, contrary to this speculation, the NA content in the spinal cord was significantly increased in STZ-treated rats in the present study. This result was, however, not unexpected because such an increase in NA in STZ-treated rats is consistent with previous reports [44, 45]. In addition to this increase in NA level, the quantities of DBH- and NET-expressing fibers in the dorsal horn were significantly increased in STZ-treated rats. Because these immunopositive fibers in the dorsal horn are drastically abolished after DSP-4 treatment, these molecules are indeed expressed on the segmental branches of the descending noradrenergic fibers.
The NET in the central nervous system is primarily located on the presynaptic membrane of noradrenergic neurons and plays an essential role in the re-uptake of extracellular NA from synaptic clefts to terminals [46, 47]. Recently, accumulated lines of evidence point to a clear role of insulin in the regulation of NET expression and membrane localization. The NA uptake in whole brain neuronal culture is inhibited by insulin . The NET mRNA level in the locus coeruleus is reduced by insulin  and elevated by STZ treatment . Surface expression of functional NETs in the hippocampal neurons is increased in STZ-treated mice, and conversely, NETs are internalized by acute insulin administration through phosphorylation of Ser/Thr kinase Akt/PKB (protein kinase B) pathways .
Our immunohistochemical staining does not allow us to distinguish between surface and internalized NET molecules. However, it is expected that together with the observations of these previous reports, the increased expression of NETs on the fibers in the dorsal horn due to sustained hypoinsulinemia would result in an increased amount of NETs localized on the membrane surface. This activity would lead to an exaggerated NA uptake by the terminals, which leads to decreased extrasynaptic or intracleft NA concentration. Consequently, this decrease in extracellular NA would directly lead to aberrant pro-nociception. The genetic ablation of NETs, which decreases NA content in the spinal cord , produces profound hypoalgesia . This insulin-dependent NET expression and the NA dependency of the spinal nociceptive system support the recent view that hypoinsulinemia itself, rather than hyperglycemia, would play a larger role in the establishment of hyperalgesia . Indeed, insulin, at a dose not affecting the hyperglycemia, has been shown to improve neuropathy and relief hyperalgesia [22, 25]. Because the NET is the primary target molecule of DLX  for its primary effect on NA re-uptake inhibition, the potent anti-nociceptive effect of DLX in STZ-treated rats is, for the most part, attributed to the direct inhibition of exaggerated NA transport in the spinal cord.
Another possibility, which is not incompatible with the interpretation described above, is that the release of NA is lowered in STZ-treated rats. Bitar et al. described a significant reduction in the ratio of 3-methoxy-4-hyroxyphenylglycol (MHPG) to NA in the lumber spinal cord of the rat at 30 days after STZ treatment and suggested a decreased release or turnover of NA in this model . This interpretation is also compatible with the present result of increased NA content in the lumber spinal cord. Decreased NA release would result from decreased firing rate of locus coeruleus neurons and release probability at the spinal noradrenergic axon terminals in STZ-treated rats, possibilities being required to be examined in the future studies.
To date, the molecular mechanisms underlying the increase in the expression of DBH in STZ-treated rats have not been established. The involvement of the CREB pathway in the regulation of tyrosine hydroxylase (TH)  and TH expression in STZ-treated diabetic models  has been documented. Though it has been shown that increase in brain-derived neurotrophic factor (BDNF) following spinal nerve injury results in sprouting of DBH-expressing fibers in the spinal cord , this mechanism is unlikely to primarily underlie the increase in DBH-positive fibers observed in the present study, because the BDNF content in the spinal cord is not significantly affected in a similar PDN model with STZ . In addition to these changes in NA synthesis, the changes in the synaptic expression level of adrenoceptors [55, 59] and agonist potency [55, 59, 60] might also underlie the aberrant NA homeostasis in STZ-treated animals. Whatever the mechanism, increased NA synthesis and storage in the spinal cord in STZ models might result in a larger quantity of NA in the tissue, as evaluated using HPLC (Figure 6). Such augmented NA synthesis and storage would provide support for the effective increase in extracellular NA levels after NET blockade by DLX.
Mechanisms of anti-nociceptive effect of DLX
We failed to detect a significant increase in lumbar NA level after DLX injection using HPLC analysis (Figure 6) unlike previous studies that demonstrated a significant increase in extracellular NA level induced by DLX in the rat frontal cortex using microdialysis [61, 62]. As the expression levels of DBH and the NET were increased in the STZ-treated rats, it is speculated that the changes in extracellular concentration of NA induced by DLX are small compared to the large amount of intracellular stored NA, which obscures the measurement using HPLC. The microdialysis measurement in the dorsal horn in the non-anesthetized animal is a challenging procedure that would provide direct insight into the spinal mechanism of DLX in future studies.
The present results do not necessarily rule out involvement of changes in NA levels in supraspinal structures, such as the limbic system, a pivotal target of nociceptive signals in the brain  as well as a site underlying depressive affection. The synaptic transmission of the amygdala neurons, which shows robust synaptic potentiation in chronic neuropathic pain models  including STZ-models , is modulated by NA . Additionally, the changes in the activities in the amygdala neurons by alpha-2 adrenoceptor agonists affect spinal nocifensive behaviors . These observations might imply that the changes in the amygdala activity by DLX might also underlie these nociceptive effects. Further understanding of the specific molecular facets of supraspinal and spinal NA homeostasis will contribute toward the development of medications with more specific pain-relieving effects in patients with DM.
Clinically, DLX improves pain severity both in type 1 and 2 DM . The PDN model used in this study with STZ treatment mimics the type 1 DM with strong hypoinsulinemia. However, the present finding of the exacerbated spinal nociception through impaired insulin-mediated NA homeostasis might also be of importance in the type 2 DM, in which a larger portion of patients suffer the neuropathic pain . In the animal models for type 2 DM (ob/ob and db/db mice), it has been shown that, despite increased insulin levels, the phosphorylation of Akt is significantly reduced . It is thus expected that such impaired insulin/Akt signal-mediated NA homeostasis would occur and exacerbate nociception also in type 2 DM, which would also be an important target of DLX for its analgesic effect.
We conclude that improvement of extracellular NA homeostasis by inhibiting NETs is the primary mechanism of the anti-nociceptive effect of DLX, which becomes highly potent in painful pathological states that accompany the aberrant increase in NA synthesis and re-uptake in PDN.