We demonstrated that intrathecal administration of milnacipran, but not citalopram or desipramine mediated an inhibition of NMDA-induced thermal hyperalgesia. Moreover, we documented the inhibition of NMDA-mediated currents by milnacipran, but not by citalopram or desipramine in spinal lamina II. We also demonstrated that activation of pERK induced by NMDA was significantly suppressed by milnacipran in dorsal horn neurons. Taken together, these findings indicate that milnacipran has a direct antinociceptive effect in the spinal cord through its modulation of NMDA receptors.
Some investigators have reported interactions between TCAs and NMDA receptors in nociceptive transmission. Eisenach and Gebhart  reported that intrathecal administration of amitriptyline reversed thermal hyperalgesia via NMDA receptor antagonism in a rat model of inflammation. Kawamata et al.  also using the rat model of inflammation, reported that intrathecally injected desipramine produced analgesia unrelated to NA reuptake inhibition. These reports suggest that TCAs exert a direct inhibitory effect on NMDA receptors to produce analgesia in the spinal cord. TCAs may not only inhibit 5-HT and NA reuptake, but could also interact with various receptors. It has been shown that TCAs block sodium channels [25, 26] as well as voltage-dependent calcium channels [27, 28] and that TCAs inhibit adenosine reuptake . Further, most TCAs have affinity for opioid , NA, 5-HT, histamine, and muscarinic acetylcholine receptors . Therefore, these various mechanisms of action of TCAs might contribute to antinociceptive effects in some kinds of chronic pain models.
Desipramine did not suppress NMDA-induced thermal hyperalgesia in the present study. However, Hwang and Wilcox  reported that desipramine was antinociceptive in three nociceptive tests, tail-flick test, intrathecal substance P-induced behavioral test and intradermal hypertonic saline-induced behavioral test. Moreover, they indicated that the analgesic effect by desipramine probably involves blockade of monoamine reuptake. This discrepancy between our result and that of previous study is likely to be due to different underlying mechanisms of desipramine.
Although there is some evidence that TCAs block NMDA receptor-mediated responses, the site of action is controversial. Based on radioligand binding studies, Reynolds and Miller  have suggested that TCAs act at the Zn2+ recognition site on the NMDA receptor. Sills and Loo  have reported that TCAs bind with higher affinity to the phencyclidine binding site on the NMDA receptor. Moreover, Sernagor et al.  have reported that desipramine blocked NMDA-induced currents in hippocampal neurons by acting on the open channel. In contrast to TCAs, milnacipran has no relevant affinity for any other receptors, including α-adrenergic, 5-HT, histamine, muscarinic acetylcholine, opioid, or NMDA receptors . However, Shuto et al.  reported that milnacipran (IC50 = 6.3 μM), is a class of noncompetitive NMDA receptor antagonist, although the binding affinity of milnacipran for the NMDA receptor is not strong. Although there is no evidence that milnacipran binds to the NMDA receptor, it is assumed that it may act at the recognition site for Zn2+ or at the phencyclidine binding site of the NMDA receptor. Further study is required to clarify this point.
Milnacipran inhibits the presynaptic reuptake of monoamines, 5-HT and NA with an IC50 of 100 to 200 nM, respectively in the brain . However, in this study, milnacipran has the antagonistic effect on NMDA-mediated responses in the spinal cord at a concentration of 10–100 μM. Some previous studies [32, 36, 38] indicate that TCAs also inhibits the NMDA receptors in the brain at a concentration of 10–100 μM. Therefore, it is likely that the concentration of inhibiting the NMDA receptors by milnacipran is higher than that of inhibiting reuptake of the monoamines.
In the present study, milnacipran reduced the amplitudes of exogenously applied NMDA-induced currents in lamina II neurons. Moreover, milnacipran inhibited the amplitudes of dorsal root stimulation evoked NMDA-mediated EPSCs. There are no differences in the degree of depression by milnacipran between NMDA induced-current and dorsal root stimulation evoked NMDA-mediated EPSCs. These results suggest that synaptic and extra synaptic NMDA receptors in dorsal horn neurons are similarly modulated by milnacipran. Moreover, milnacipran inhibited the amplitudes of NMDA-mediated EPSCs in the presence of 5-HT antagonist and α2 receptor antagonist. Therefore, it is unlikely that an antagonistic effect of milnacipran on NMDA receptors is mediated by 5-HT or NA receptors.
We observed that the highest dose of intrathecal milnacipran completely reversed thermal hyperalgesia induced by NMDA. In contrast, desipramine and citalopram did not produce any inhibitory effect, although the concentrations of desipramine and citalopram in this study are sufficiently high to inhibit reuptake of 5-HT or NA, respectively. These results suggest that increases in 5-HT or NA alone in the spinal cord have no effect on NMDA-induced thermal hyperalgesia. The antagonistic action of milnacipran for NMDA receptors may produce additional effects for some types of chronic pain. Previous studies demonstrated that intrathecal administration of milnacipran produced antiallodynic effects in rats with peripheral nerve injury [13, 39, 40], and that the effect was not completely reversed by an α2 receptor antagonist or a 5-HT receptor antagonist . Therefore, it is conceivable that an antagonistic action for NMDA receptors contributes to the antiallodynic effect of milnacipran. Further studies are required to clarify the molecular mechanisms underlying the inhibitory effect of milnacipran on NMDA-mediated responses in dorsal horn neurons. In addition, to elucidate whether the observed effects were specific for milnacipran, further investigations using another SNRI are necessary to resolve this question.
ERK activation is detected in the spinal dorsal horn neurons after stimulation of nociceptive primary afferents and contributes to the development of central sensitization . Activation of the NMDA receptor is partly involved in ERK induction following nociceptive stimulation . Analgesic drugs such as local anesthetics , opioids, or cannabinoids  inhibit ERK induction in the spinal cord. In the present study, we demonstrated that milnacipran inhibited ERK induction following application of NMDA in the spinal cord. This result is consistent with our behavioral data showing that milnacipran attenuated thermal hyperalgesia following intrathecal NMDA injection. Our electrophysiological data clearly indicate that the direct inhibitory effect of milnacipran on NMDA-mediated current underlies these phenomena.
We show in the present study that milnacipran has an antagonistic effect of NMDA receptors in the spinal cord when administered intrathecally. However, it is not clear whether milnacipran has the similar effect in other central and peripheral tissue such as brain and skin. Moreover, NMDA receptor antagonists such as ketamine and phencyclidine have psychotomimetic or anti-depressant properties in humans when administered systemically. There have been no studies demonstrating that milnacipran has these similar properties in humans. Therefore, further study is required to clarify this point.
NMDA glutamate receptors are one of the major receptor channel types mediating rapid excitatory neurotransmission in the central nervous system. This receptor is composed of subunits from at least two families, NR1 and NR2. The NR1 subunit is essential for the function of NMDA receptors and is ubiquitously expressed in most neurons. The functional properties of NMDA receptors are determined by the NR2 subunit composition (NR2A–2D). Previous reports have demonstrated that desipramine inhibited NMDA-evoked responses in hippocampal neurons [32, 43], but not in the present study. The respective NR2 subunits show different expression patterns in various regions of the brain and spinal cord. Whereas NR2A and NR2B subunits are prominent in the hippocampus , these subunits are not identified in spinal dorsal horn neurons . This different composition of NMDA receptors may underlie the variability among tissues in the effects of desipramine on NMDA receptor-responses.
NMDA glutamate receptors also play a key role in central sensitization in chronic pain [3, 4]. The pursuit of an NMDA receptor antagonist for the relief of chronic pain dates from the late 1980s when it was shown that NMDA antagonists inhibit the “wind-up” response [46, 47]. The central sensitization that occurs in the spinal dorsal horn is held to be an important event in the pathway leading to neuropathic pain. For this reason, ketamine is currently widely applied in the treatment of neuropathic pain and for some kinds of chronic pain, including fibromyalgia . Although there is little evidence that NMDA receptor antagonism is involved in the antinociceptive effect of milnacipran, milnacipran has been widely used in patients with fibromyalgia and has provided them with pain relief [49, 50]. These effects of milnacipran for chronic pain have been previously discussed from the perspective of one aspect: balanced inhibition of the reuptake of 5-HT and NA. However, in the present study, our data suggest that milnacipran not only inhibits 5-HT and NA reuptake, but also acts as an NMDA receptor antagonist. Milnacipran may have the potential to prevent or suppress chronic pain related to central sensitization, especially when injected into intrathecal or epidural space.