Direct Inhibition of the N-methyl-D-aspartate Receptor Channel by High Concentrations of Opioids 

Subunit-specific mRNAs were synthesized in vitro  with SP6 or T3 RNA polymerase (Ambion MEGAscript) in the presence of cap dinucleotides ⁷mGpppG. The ε1, ε2, ε3, ε4, and ζ1 subunit-specific mRNAs (for the expression of the ε1/ζ1, ε2/ζ1, ε3/ζ1, and ε4/ζ1 channels) were synthesized using pSPGRε1, 13pSPGRε2, 14pSPGRε3, 15pSPGRε4, 16and pSPGRζ1, 15respectively. The ε2, ε2-N589Q, ζ1, and ζ1-N598Q subunit-specific mRNAs (for the expression of the ε2-N589Q/ζ1, ε2/ζ1-N598Q, and ε2-N589Q/ζ1-N598Q channels) were synthesized using pBKSAε2, 17pBKSAε2-N589Q, 11pBKSAζ1, 18and pBKSAζ1-N598Q, 11respectively. The α1 and α2 subunit-specific mRNAs for the α1/α2 AMPA-selective glutamate receptor channel were synthesized using pSPGR1 and pSPGR2, respectively. 19 

Xenopus laevis  oocytes were injected with the wild-type or mutant ε subunit–specific mRNA and the wild-type or mutant ζ subunit–specific mRNA at a molar ratio of 1:1, or with the α1 and α2 subunit–specific mRNAs at a molar ratio of 10:1; the total amount of mRNAs injected per oocyte was approximately 0.6 ng for the ε1/ζ1 and ε2/ζ1 channels; 14 ng for the ε3/ζ1 and ε4/ζ1 channels; 4 ng for the ε2-N589Q/ζ1, ε2/ζ1-N598Q, and ε2-N589Q/ζ1-N598Q channels; and 10 ng for the α1/α2 channel.

After incubation at approximately 19°C for 2 or 3 days, whole-cell currents evoked by bath application of agonists for approximately 15 s were recorded at −70 mV membrane potential with a conventional two-micropipette voltage clamp. 19The current responses of the wild-type and mutant ε/ζ channels to 10 μM L-glutamate plus 10 μM glycine (almost saturating concentrations for all ε/ζ channels) were measured in Ba²⁺-Ringer's solution to minimize the effects of secondarily activated Ca²⁺-dependent Cl^-currents. 20The current responses of the α1/α2 channel to 100 μM kainate were measured in normal frog Ringer's solution. For measurement of the effects of opioids on NMDA receptor channels, opioids were continuously perfused during the experiment. Preapplication of opioids in the absence of agonists did not produce any current response in either wild-type or mutant channels. Agonists were applied three times successively during perfusion of opioids, and the effects on the second and third applications of agonists were averaged. The second and third current responses during perfusion of opioids were of similar magnitude, indicating that the effects of opioids were fully established in this recording system. Ba²⁺-Ringer's solution contained 115 mM NaCl, 2.5 mM KCl, 1.8 mM BaCl², and 10 mM HEPES-NaOH (p  H 7.2). Normal frog Ringer's solution contained 115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl², and 10 mM HEPES-NaOH (p  H 7.2).

Meperidine hydrochloride was purchased from Tanabe Seiyaku (Osaka, Japan). Morphine hydrochloride, codeine phosphate, and fentanyl citrate were from Sankyo (Tokyo, Japan). Naloxone hydrochloride was from Sigma Chemical (St. Louis, MO). Meperidine, morphine, codeine, and naloxone were dissolved in distilled water at a concentration of 100 mM. Fentanyl was dissolved in dimethyl sulfoxide at a concentration of 100 mM. The dimethyl sulfoxide stocks were diluted to appropriate concentrations in Ringer's solution. Perfusion of the highest concentrations of dimethyl sulfoxide used in this investigation (1% for 1 mM fentanyl) inhibited the current responses of the ε2/ζ1 channel by 5 ± 1%(mean ± SEM, n = 4).

The inhibitor concentration for half-control response (IC⁵⁰) and the Hill coefficient values for opioids of the ε/ζ channel were calculated according to the equation Ropi = 1/[1 +(O /IC⁵⁰)ⁿ], where Ropi  represents the relative response, O  the concentration of opioids, and n  the Hill coefficient. The agonist concentration for half-control response (EC⁵⁰) of the ε/ζ channel was calculated according to the equation Rago =Fopi /[1 +(EC⁵⁰/A )ⁿ], where Rago  represents the relative response, Fopi  the residual fraction by opioid inhibition of responses to saturating concentrations of agonists, A  the concentration of agonists, and n  the Hill coefficient. For quantitative estimates of the voltage dependence of block by opioids, data were analyzed using the Woodhull model 21by fitting the data to the equation Ropi = 1/[1 +(O /K  ^d(0)exp(zδ  FE /RT))], where Ropi  represents the relative response, O  the concentration of opioids, K  ^d(0)the equilibrium dissociation constant of opioids at a membrane potential of 0 mV, z  the charge of opioids, δ  the portion of the membrane electric field sensed at the blocking site, E  the membrane potential, F the Faraday constant, R the gas constant, and T the absolute temperature. The results obtained were statistically analyzed by the Student t  test or one-way analysis of variance (ANOVA) followed by Scheffé’s multiple comparison tests. P < 0.05 was considered significant. Data were represented as mean ± SEM.

Four kinds of heteromeric NMDA receptor channels, the ε1/ζ1, ε2/ζ1, ε3/ζ1, and ε4/ζ1 channels, were expressed in Xenopus  oocytes by the injection of respective subunit-specific mRNAs synthesized in vitro  from cloned cDNAs. The sensitivities of these ε/ζ heteromeric channels to opioids were examined by measuring current responses to 10 μM L-glutamate plus 10 μM glycine during continuous perfusion of meperidine at −70 mV membrane potential in Ba²⁺-Ringer's solution. Meperidine inhibited the current responses of the ε/ζ NMDA receptor channels (fig. 2A). After the meperidine was washed out, application of agonists two or three times fully recovered the current responses. The dose-inhibition relationships for meperidine of four kinds of heteromeric channels were examined (fig. 2B). Meperidine inhibited the ε1/ζ1, ε2/ζ1, ε3/ζ1, and ε4/ζ1 channels to a similar extent in a concentration-dependent manner. The IC⁵⁰values (μM) of the ε1/ζ1, ε2/ζ1, ε3/ζ1, and ε4/ζ1 channels for meperidine were 233 ± 14 (n = 6), 206 ± 7 (n = 7), 264 ± 9 (n = 6), and 273 ± 12 (n = 6), respectively. The ε2/ζ1 channel was more sensitive to meperidine than the ε3/ζ1 and ε4/ζ1 channels (log[IC⁵⁰] values were compared by ANOVA followed by Scheffé’s multiple comparison tests;P < 0.01). On the other hand, 1 mM meperidine only inhibited the current responses of the α1/α2 glutamate receptor channel selective for AMPA by 6 ± 3%(n = 5). Thus, the inhibitory effects of meperidine are likely to be selective for NMDA receptor channels out of the glutamate receptor channels. Because meperidine inhibited NMDA receptor channels, the effects of other opioid agonists and antagonists on NMDA receptor channels were examined. Morphine also inhibited the four ε/ζ channels in a dose-dependent manner. The sensitivities to morphine varied between the four ε/ζ channels (log[IC⁵⁰] values were compared by ANOVA;P < 0.0001;fig. 2C). The IC⁵⁰values (μM) of the ε1/ζ1, ε2/ζ1, ε3/ζ1 and ε4/ζ1 channels were 321 ± 48 (n = 7), 187 ± 9 (n = 7), 392 ± 27 (n = 6) and 650 ± 24 (n = 7), respectively. The ε2/ζ1 channel was the most sensitive to morphine among the four ε/ζ channels (Scheffé’s multiple comparison tests, P < 0.05). In addition, fentanyl, codeine, and naloxone also inhibited NMDA receptor channels in a dose-dependent manner (fig. 2D). The IC⁵⁰values (μM) of the ε2/ζ1 channel for fentanyl, codeine, and naloxone were 192 ± 9 (n = 8), 613 ± 25 (n = 7), and 503 ± 34 (n = 8), respectively.

To characterize the inhibitory effects of opioids on NMDA receptor channels, we examined whether opioids affect the apparent affinities of the ε/ζ channel for agonists. The dose–response relationships of the ε2/ζ1 channel for L-glutamate and glycine before and during perfusion of 300 μM meperidine were analyzed (fig. 3A). Meperidine effectively suppressed the maximal current responses to saturating concentrations of both L-glutamate and glycine. The EC⁵⁰values (μM) of the ε2/ζ1 channel for L-glutamate in the presence of 10 μM glycine and those for glycine in the presence of 10 μM L-glutamate during perfusion of 300 μM meperidine (1.04 ± 0.04 [n = 7] and 0.29 ± 0.01 [n = 6], respectively) were not significantly different from those before meperidine perfusion (1.03 ± 0.02 [n = 6] and 0.29 ± 0.01 [n = 6], respectively)(log[EC⁵⁰] values were compared by t  tests;P > 0.76 for L-glutamate and P > 0.81 for glycine). Similarly, morphine (300 μM) inhibited the maximal current responses of the ε2/ζ1 channel without affecting the EC⁵⁰values (fig. 3B). These results suggest the noncompetitive antagonism of NMDA receptor channels by opioids.

To test whether the inhibition by opioids is voltage-dependent, the extent of inhibition was measured at different holding potentials. Figure 4Ashows current–voltage relationships of the ε2/ζ1 channel before and during perfusion of 300 μM meperidine. Meperidine inhibition of the ε2/ζ1 channel exhibited voltage dependence and was quite effective at hyperpolarized potentials. The extent of inhibition was significantly dependent on the membrane potential (ANOVA, P < 0.0001). At a membrane potential of −110 mV, meperidine (300 μM) reduced the current responses of the ε2/ζ1 channel to 17 ± 1%(n = 7) of the control responses, whereas at −10 mV membrane potential, meperidine reduced the current responses to only 85 ± 2%(n = 7). Similarly, morphine (300 μM) inhibited the ε2/ζ1 channel in a voltage-dependent manner (fig. 4B). The degree of voltage dependence of inhibition of the ε2/ζ1 channel by meperidine (300 μM), morphine (300 μM), and naloxone (1 μM) was compared using the Woodhull model 21(fig. 4C). Although the K  ^d(0)values (the affinity of binding) for meperidine, morphine, and naloxone varied (2.6 ± 0.5 [n = 7], 3.8 ± 0.5 [n = 8], and 10.7 ± 1.6 [n = 8] mM, respectively; ANOVA, P < 0.0001), the zδ  values (the degree of voltage dependence of block) for those were not significantly different (0.9 ± 0.05 [n = 7], 1.0 ± 0.04 [n = 8], and 1.0 ± 0.06 [n = 8], respectively; ANOVA, P > 0.07).

To determine whether opioids interact with the Mg²⁺block site of NMDA receptor channels, we examined the effects of meperidine (300 μM) and morphine (300 μM) on the sensitivity of the ε2/ζ1 channel to the Mg²⁺block (fig. 5A). Mg²⁺inhibited the current responses of the ε2/ζ1 channel in a dose-dependent manner with IC⁵⁰values of 15.5 ± 1.0 μM (n = 8). Meperidine and morphine shifted the inhibition curve of Mg²⁺to the right (fig. 5B). The IC⁵⁰values (μM) for Mg²⁺during perfusion of meperidine and morphine were 30.0 ± 1.9 (n = 7) and 37.0 ± 2.5 (n = 7), respectively, which were significantly higher than those for Mg²⁺alone (log[IC⁵⁰] values were compared using ANOVA followed by Scheffé’s multiple comparison tests, P < 0.0001 for control versus  meperidine, and control versus  morphine).

Opioids inhibited the current responses of the NMDA receptor channel in a voltage-dependent manner. The voltage-dependent inhibition is a specific and essential property of the well-characterized NMDA receptor channel blockers, such as Mg²⁺and dissociative anesthetics (phencyclidine [PCP], ketamine, and (+)MK-801). 22,23Furthermore, the Mg²⁺block curve was shifted rightward by opioids. These results suggest that inhibition of NMDA receptor channels by opioids may be a result of channel block mechanisms. We have previously shown that the conserved asparagine residue in the channel-lining segment M2 of the ε2 and ζ1 subunits (the asparagine 589 of the ε2 subunit and the asparagine 598 of the ζ1 subunits) constitutes the Mg²⁺block site of NMDA receptor channels, and that the noncompetitive antagonists, PCP, ketamine, n -allylnormetazocine (SKF-10,047), and (+)MK-801, also act on the Mg²⁺block site. 11,13To reveal whether the same asparagine residue also constitutes the block site of opioids, we examined the effects of replacement by glutamine of the conserved asparagine residue in segment M2 of the ε2 and ζ1 subunits (the mutations ε2-N589Q and ζ1-N598Q, respectively) on the sensitivity to opioids. The mutation ζ1-N598Q reduced the sensitivity to meperidine (fig. 6A). The ε2/ζ1-N598Q and ε2-N589Q/ζ1-N598Q channels were more resistant to meperidine than the ε2/ζ1 and ε2-N589Q/ζ1 channels (log[IC⁵⁰] values were compared by ANOVA followed by Scheffé’s multiple comparison tests, P < 0.0001)(fig. 6B). On the other hand, the sensitivity of the ε2-N589Q/ζ1 channel was not significantly different from that of the ε2/ζ1 channel (Scheffé’s multiple comparison tests, P > 0.99). The IC⁵⁰values (μM) of the ε2/ζ1, ε2-N589Q/ζ1, ε2/ζ1-N598Q, and ε2-N589Q/ζ1-N598Q channels for meperidine were 206 ± 7 (n = 7), 212 ± 16 (n = 6), 1926 ± 83 (n = 6), and 2194 ± 86 (n = 6), respectively. Similarly, the mutation ζ1-N598Q reduced the sensitivity of the ε2/ζ1 channel to morphine and naloxone, whereas the effects of the mutation ε2-N589Q were only slight (figs. 6C and 6D). The involvement of the asparagine residue of the ζ1 subunit in determining the opioid sensitivity was further confirmed by the resistance of the ε1/ζ1-N598Q channel to opioids (data not shown).

The effects of the mutation ζ1-N598Q on the degree of voltage dependence of block by morphine were examined. The inhibitory effects of morphine (300 μM) on the ε2/ζ1-N598Q channel were slight but exhibited voltage dependence (ANOVA, P < 0.0001)(fig. 7A). The degree of voltage dependence of block by morphine (300 μM) was compared between the ε2/ζ1 and ε2/ζ1-N598Q channels using the Woodhull model (fig. 7B). 21Not only were the K  ^d(0)values (the affinity of binding) for morphine different between the ε2/ζ1 and ε2/ζ1-N598Q channels (3.8 ± 0.5 [n = 8] and 6.5 ± 1.0 [n = 6] mM, respectively;t -tests, P < 0.03), but the zδ  values (the degree of voltage dependence of block) of the ε2/ζ1 and ε2/ζ1-N598Q channels were also significantly different (1.0 ± 0.04 [n = 8] and 0.6 ± 0.06 [n = 6], respectively;t  tests, P < 0.0001).

In the present investigation, we have shown that high concentrations of the naturally occurring opioids morphine and codeine, the phenylpiperidine derivatives meperidine and fentanyl, and the opioid antagonist naloxone inhibit NMDA receptor channels. These results are consistent with previous studies that found that high concentrations of opioids and naloxone protect neurons against central nervous system ischemia and injury and NMDA-induced neurotoxicity. 1–3Recent electrophysiologic and receptor binding studies showed that some opioid agonists, such as meperidine, methadone, and ketobemidone, reduce NMDA-induced depolarization in rat-brain slice preparations at concentrations of 1 mM, and inhibit [³H]MK-801 binding in rat cortical and forebrain membranes. 4–6Furthermore, there are some opioid-related compounds that are already known to be noncompetitive NMDA receptor antagonists. The benzomorphans SKF-10,047 and cyclazocine were shown to exhibit NMDA receptor antagonist properties. 24,25The morphinan opioid levorphanol, its dextrorotatory nonopioid enantiomer dextrorphan, and its O -methyl derivative, dextromethorphan, were also able to selectively antagonize the NMDA-induced neuroexcitation. 26,27These findings suggest that the NMDA receptor antagonist property is a common characteristic of various opioids and related compounds.

The NMDA receptor antagonist activity of opioids should be noted because NMDA receptor channels are suggested to be involved in the changing of opioid efficacy in certain clinical situations. In the pain hypersensitivity states in which opioids are not effective, the coadministration of an NMDA receptor antagonist with opioids was shown to restore the antinociceptive effects of opioids. 28,29Furthermore, NMDA receptor antagonists were shown to attenuate or block the development of opioid tolerance and dependence in case of repeated treatment. 30,31Therefore, opioids with NMDA receptor antagonist activities may extend the usefulness of opioids in the clinical management of pain. However, the opioids tested in the present investigation could only block NMDA receptor channels at high micromolar concentrations. Plasma concentrations obtained after systemic administration of meperidine and morphine are at most 1–3 μM, 32,33and those of fentanyl during high-dose fentanyl anesthesia for cardiac surgery are around 0.1 μM. 34,35On the other hand, very high concentrations are obtained in the cerebrospinal fluid (CSF) after epidural or intrathecal administration of opioids in humans. Meperidine concentrations in the CSF after intrathecal injection reach 300–1000 μM, and those after epidural administration come to approximately 100–300 μM because of the rapid absorption across the dural membrane into the CSF. 36,37The initial CSF concentrations of morphine following intrathecal administration are in the high micromolar range, and those after epidural administration are about 10 μM;38,39the CSF concentrations of fentanyl after epidural administration are at most 0.1 μM. 40Thus, the NMDA receptor antagonist property may be clinically significant in the spinal cord following epidural or intrathecal administration of some opioids. Among known opioids with NMDA receptor antagonist properties, methadone and its d - and l -isomers were reported to exhibit relatively high affinities for NMDA receptor channels, approximately similar to those of dextromethorphan. 4,6In the rat formalin test, intrathecal administration of the nonopioid d -methadone was shown to have antinociceptive effects as a result of its NMDA receptor antagonist activity. 41 

We have previously shown that the conserved asparagine residue in channel-lining segment M2 of the ε2 and ζ1 subunits constitutes the Mg²⁺block site of NMDA receptor channels, and that PCP, ketamine, SKF-10,047 and (+)MK-801 also act on the Mg²⁺block site. 11,13The effects of mutations on the sensitivity to ketamine were stronger for the ζ1 subunit than for the ε2 subunit, whereas mutations in both subunits are required for (+)MK-801 resistance. 11,13In the present investigation, the mutation ζ1-N598Q reduced the sensitivity and voltage dependence of opioid inhibition of the ε2/ζ1 channel, whereas the mutation ε2-N589Q barely affected the sensitivity to opioids. These results support the proposition that the block site of opioids may at least partially overlap with those of the established channel blockers. Furthermore, opioids appear to resemble ketamine rather than (+)MK-801 in terms of the contribution of the conserved asparagine residue of the ζ1 subunit to the block site.

The findings that various opioids and related compounds block NMDA receptor channels raise the question as to which chemical structures of these compounds are responsible for the NMDA receptor channel blocking. Studies on the structural requirements for binding at the PCP site of NMDA receptor channels by analyses of PCP derivatives and (+)MK-801–like molecules have proposed two main requirements of molecules, which correspond to a hydrophobic aromatic moiety and a basic nitrogen atom. 42,43Various opioid compounds including both naturally occurring opioids and synthetic compounds such as morphinans, benzomorphans, and phenylpiperidine derivatives, which at first glance seem to be structurally diverse, have a common vital moiety: an aromatic ring and a nitrogen atom that usually originates from a piperidine ring. 44Thus, the aromatic ring and the protonated amine of opioid compounds may interact with structural determinants for the PCP binding site of NMDA receptor channels through the hydrophobic interaction and the hydrogen bond, respectively. This proposition is supported by recent structural analyses of the PCP site, which demonstrated that the morphinan derivative dextromethorphan is able to occupy the binding site in a fashion similar to PCP, with its aromatic ring roughly occupying the same region as the phenyl moiety of PCP. 43 

In the present investigation, the effects of opioids on NMDA receptor channels were fully established by the second application of agonists during continuous perfusion of opioids. After opioids were washed out, application of agonists two or three times fully recovered the current responses. This observation is in contrast to that of the potent channel blocker (+)MK-801, which exhibits the progressive and almost irreversible block by sequential application of agonists. 13The NMDA receptor channel blocking and unblocking kinetics of various noncompetitive antagonists were reported to be highly correlated to their affinities: Lower potency antagonists exhibited faster onset and offset kinetics. 45Thus, the fast onset and recovery of the block by opioids may be caused by their low affinities for NMDA receptor channels.

Electrophysiologic studies showed that 1 mM meperidine reduced NMDA responses in the rat neonatal spinal cord, whereas meperidine was devoid of antagonist activity in the cerebral cortex. 4Because the distribution of the four ε subunits is distinct in the mature and developing brain, 7reported differences in meperidine sensitivities in different central nervous system regions seem to be related to differences in the ε subunits. In the present investigation, however, meperidine inhibited the ε1/ζ1, ε2/ζ1, ε3/ζ1, and ε4/ζ1 channels to a similar extent. Thus, differences in meperidine sensitivities of NMDA receptor channels in different central nervous system regions cannot be explained only by differences in ε subunit species. If the subunit composition of NMDA receptor channels were responsible, differences in a stoichiometry of the ζ1 subunit might be involved.

In conclusion, high concentrations of various opioid compounds inhibited the current responses of heteromeric NMDA receptor channels in a voltage-dependent manner. The conserved asparagine residue in segment M2 of the ζ1 subunit was identified as one of molecular determinants of the opioid binding site at NMDA receptor channels. These results suggest that the low-affinity NMDA receptor antagonist activity is not a property specific for a part of opioids as previously considered, but a common characteristic of various opioid compounds. Furthermore, the inhibition was confirmed to be a result of channel block mechanisms at the site, which partially overlaps with those of Mg²⁺and ketamine. Our results point out the clinical significance of the NMDA receptor antagonist property of some opioids in the spinal cord after local administration and may yield insights into the design of new opioid compounds with higher affinities for NMDA receptor channels.

The authors thank Sankyo Co. Ltd., Tokyo, Japan, for providing fentanyl citrate.