Opioids are well-known to inhibit glutamatergic excitatory transmission by activating opioid receptors in the CNS including the central terminals of primary-afferent fibers, resulting in antinociception ([28,29,30]; see [31,32] for reviews). Not only central but also peripheral terminal opioid receptors in primary-afferent neurons are thought to be involved in antinociception ([33,34,35,36]; see [37] for review). Opioids also exhibit a local anesthetic effect in the PNS. Although it has been reported in decerebrated cats that the perineural administration of an opioid morphine has no effect on CAPs in the superficial radial nerve [38], AP conduction in peripheral nerve fibers is generally blocked by opioids. For example, opioids such as fentanyl and sufentanil decreased the peak amplitudes of CAPs recorded from peripheral nerve fibers [39] and inhibited peripheral nerve AP conduction [40].

A morphine-induced CAP inhibition in mammalian peripheral nerve fibers was antagonized by a non-specific opioid-receptor antagonist naloxone, indicating an involvement of opioid receptors [41]. Consistent with this observation, binding and immunohistochemical studies have shown the localization of opioid receptors in mammalian peripheral nerve fibers [42,43,44]. It has been also demonstrated that a frog sciatic nerve CAP inhibition produced by opioids is sensitive to naloxone [45]. On the contrary, there are reports showing that opioids decrease CAP peak amplitudes [39] and suppress nerve AP conduction [40] in a manner resistant to naloxone.

The compound (1RS,2RS)-2-[(dimethylamino)methyl]-1-(3-methoxyphenyl)cyclohexanol hydrochloride (tramadol) is an orally-active opioid which is clinically used as an analgesic in the CNS [46]. Although tramadol is metabolized to various compounds such as mono-O-desmethyltramadol (M1) via N- and O-demethylation in animals and humans [47], M1 is a therapeutically active drug as a central analgesic [46]. One of cellular mechanisms for the antinociceptive effect of tramadol is the activation of μ-opioid receptors [48,49]. In agreement with this idea, M1 has the highest affinity for cloned μ-opioid receptors among the metabolites of tramadol. M1 is reported to inhibit glutamatergic excitatory transmission in spinal lamina II (substantia gelatinosa) neurons which play a crucial role in regulating nociceptive transmission to the spinal dorsal horn from the periphery, resulting in a decrease in the excitability of the neurons [50,51,52]. In addition to such a central activity, tramadol is known to have a local anesthetic effect following its intradermal injection in patients ([53,54,55]; see [56] for review). Consistent with this result, in vivo studies have shown a spinal somatosensory evoked potential inhibition produced by a direct application of tramadol to the rat sciatic nerve [57].

Tramadol reduced the peak amplitude of CAPs recorded from the frog sciatic nerve in a concentration-dependent manner in a range of 0.2 to 5 mM [19]. A similar CAP inhibitory action of tramadol has been reported by other investigators in the frog [58] and rat sciatic nerve [59,60]. Analysis based on the Hill equation demonstrated that half-maximal inhibitory concentration (IC₅₀) value for tramadol to reduce frog sciatic nerve CAP amplitudes is 2.3 mM; this IC₅₀ value was smaller by about three-fold than that (6.6 mM) reported previously for the frog sciatic nerve [58]. Rat sciatic nerve CAPs were inhibited by tramadol (37% peak amplitude reduction at 4 mM) less effectively than frog sciatic nerve’s ones [59]. This inhibitory action of tramadol in the frog sciatic nerve was not affected by the pretreatment of the sciatic nerve with naloxone (0.01 mM); a μ-opioid receptor agonist (D-Ala², N-Me-Phe⁴, Gly⁵-ol)enkephalin (DAMGO; 1 μM) had no effect on frog sciatic nerve CAPs. Furthermore, CAPs were affected by much smaller extents by M1 (see below) that is similar in chemical structure to tramadol while exhibiting a higher affinity for μ-opioid receptors than tramadol [61]. These results indicate that the tramadol-induced CAP inhibition is not mediated by opioid receptors [19]. Consistent with this result, a spinal somatosensory evoked potential inhibition following the application of tramadol on rat sciatic nerves in vivo was resistant to naloxone [57]. Although tramadol inhibits noradrenaline (NA) and serotonin (5-hydroxytryptamine; 5-HT) reuptake at concentrations similar to those that activate μ-opioid receptors [62,63], a combination of inhibitors of the reuptake of NA and 5-HT (desipramine and fluoxetine, respectively; each 10 μM; see Section 3.3) did not affect frog sciatic nerve CAPs, indicating no involvement of NA and 5-HT reuptake inhibition in the CAP inhibition [19].

The CAP inhibition produced by tramadol is possibly due to an inhibition of voltage-gated Na⁺ and K⁺ channels involved in the production of AP. Tramadol concentration-dependently reduced the peak amplitude of TTX-sensitive Na⁺ channel currents recorded from dorsal root ganglion (DRG) neuroblastoma hybridoma cell line ND7/23 cells with an IC₅₀ value of 0.194 mM [64] and from HEK293 cells expressing rat Nav1.2 channels with an IC₅₀ value of 0.103 mM [65]. These values were smaller than that (2.3 mM) for frog sciatic nerve CAP inhibition. It has been demonstrated that tramadol suppresses the current amplitude of delayed rectifier K⁺-channels (Kv3.1a type) expressed in NG 108-15 cells with an IC₅₀ of 0.025 mM, a value much less than 2.3 mM [66]. Such IC₅₀ values for tramadol to inhibit CAPs, Na⁺ and K⁺ channels were higher than its clinically relevant concentration of about 2 μM in serum [52,67].

Unlike tramadol, M1 (1-2 mM) did not affect frog sciatic nerve CAPs. This was confirmed in the frog sciatic nerve whose CAPs were inhibited by tramadol (1 mM; [19]). CAP peak amplitudes were reduced by 9% by M1 at 5 mM. Consistent with such smaller effects of M1, APs conducting on rat primary-afferent fibers were not blocked when the effect of M1 (1 mM) on dorsal root-evoked excitatory postsynaptic currents was examined by applying the patch-clamp technique to lamina II neurons in spinal cord slices [51]. It is interesting to note that tramadol has -OCH₃ bound to the benzene ring while M1 has -OH and thus that the methyl group is present in tramadol but not M1. This result indicates that the difference in chemical structure between tramadol and M1 is responsible for the distinction in CAP inhibition (see Figure 5a in [19] for a comparison of the chemical structures of the two compounds).

In order to reveal whether such a structure-activity relationship is seen for other opioids, the effects of morphine, codeine and ethylmorphine on frog sciatic nerve CAPs were examined. Morphine concentration-dependently reduced CAP peak amplitude; this extent at 5 mM was 15%. Codeine, which has -OCH₃ in place of -OH in morphine, at 5 mM reduced CAP peak amplitude by 30%. Moreover, CAPs were more effectively inhibited by ethylmorphine, where -OH of morphine is replaced by -OCH₂CH₃; amplitude reduction at 5 mM was 61%. IC₅₀ value for ethylmorphine in reducing CAP peak amplitudes was 4.6 mM. CAP inhibitions produced by morphine (10 mM), codeine (5 mM) and ethylmorphine (2 mM) were resistant to naloxone (0.01 mM). Naloxone at a high concentration such as 1 mM by itself reduced by 9% CAP peak amplitudes, but did not affect morphine (10 mM) activity [20]. These results indicate no involvement of opioid receptors in the CAP inhibition produced by opioids, as reported previously in mammalian peripheral nerves [39,40,68]. A sequence of the opioid-induced CAP peak amplitude reduction was ethylmorphine > codeine > morphine. Thus, CAP amplitude reduction increased in extent with an increase in the number of -CH₂ (see Figure 7A in [20] for a comparison of the chemical structures of the three opioids). This result is consistent with the above-mentioned observation that tramadol having -OCH₃ in the benzene ring inhibits CAPs more effectively than M1 which is different from tramadol only in terms of the presence of -OH in the ring. Interestingly, this result was obtained in spite of the fact that the chemical structures of morphine, codeine and ethylmorphine are quite distinct from those of tramadol and M1 (see [69] for review). Since the increase in -CH₂ number is thought to enhance lipophilicity of opioids, lipophilic opioid-channel interaction is suggested to play a pivotal role in nerve AP conduction block, as shown for local anesthetics [70,71]. This idea is supported by the observation that the potency in CAP inhibition in the rat sciatic nerve was in the order of isopropylcocaine (where the methyl ester group of cocaine is replaced with an isopropyl ester group) > cocaethylene (where its methyl ester group is replaced with an ethyl ester group) > cocaine [72]). It is interesting to note that the sequence of the affinity of opioids for μ-opioid receptors is morphine > codeine > ethylmorphine [73], the order of which is reversed to one for CAP inhibition. If the opioid-induced inhibition of CAPs is mediated by μ-opioid receptors, CAP inhibition sequence will be expected to be morphine > codeine > ethylmorphine. However, this sequence is not seen, a result being consistent with the idea that the opioids-induced frog sciatic nerve CAP inhibition is not mediated by opioid receptors.

The same sequence as that in the frog sciatic nerve has been reported in the rat phrenic nerve [68], although there is a quantitative difference between the two studies. When compared at 5 mM, codeine-induced reduction (about 30%) in frog sciatic nerve CAP peak amplitude was much smaller than that (about 70%) in the rat phrenic nerve, while so a large distinction was not seen in morphine action (about 10%). Frog sciatic nerve CAPs were less sensitive to morphine than those in rabbit and guinea-pig vagus nerves in such that vagus nerve CAP peak amplitudes were reduced by 20–32% at 0.5 mM [41]. APs recorded intracellularly from rat DRG neurons having Aα/β myelinated primary-afferent fibers also exhibited the sequence of ethylmorphine > codeine ≥ morphine (IC₅₀ = 0.70, 2.5 and 2.9 mM, respectively) in AP peak amplitude reduction; this inhibition was resistant to naloxone (0.01 mM) [74].

Although many drugs including narcotics, antiepileptics, local anesthetics, alcohols and barbiturates block AP conduction in peripheral nerve fibers, suggesting a nonspecific interaction of the drugs with membrane bilayers [75], the chemical structure-specific CAP inhibition produced by opioids indicates that opioids act on proteins such as voltage-gated Na⁺ and K⁺ channels (see [76] for review). Morphine is reported to suppress peak Na⁺ currents and steady-state K⁺ currents in single myelinated nerve fibers isolated from the frog sciatic nerve, leading to the prolongation of APs [77]. Intracellularly-applied morphine reduced voltage-gated Na⁺ and K⁺ channel current amplitudes in squid giant axons [78]. Bath-applied morphine reduced TTX-sensitive Na⁺ channel current amplitude in DRG neuroblastoma hybridoma cell line ND7/23 cells with an IC₅₀ value of 0.378 mM [64], although Nav1.2 channels expressed in HEK293 cells were unaffected by morphine at 1 mM [65]. In support of such an idea about ion channel inhibition, it has been reported that an opioid meperidine, which is used for AP conduction blockade and thus analgesia, inhibits Na⁺-channels in a manner similar to that of lidocaine [79]. Table 1 summarizes IC₅₀ values for frog sciatic nerve fast-conducting CAP inhibitions produced by opioids together with those for rat sciatic nerve CAPs and voltage-gated Na⁺ channels.

In clinical practice, although administration of opioids into the nerve sheath results in pain relief (for instance, see [80]), many of pain treatments by use of opioids are due to systemic administration of centrally-penetrating opioids, leading to their actions in the PNS and CNS, both of which contribute to analgesia (see [81] for review). It is possible that centrally-administrated opioids act on not only the CNS but also the PNS, because opioids are reported to be transported to the periphery from brain by P-glycoprotein [82]. In support of an important role of opioids in the PNS, subcutaneous administration of N-methyl-morphine, which did not pass through the blood brain barrier, resulted in antinociception in an acetic acid-writhing model in mice [35]. It has been reported that a subcutaneously-administrated opioid loperamide, which cannot penetrate into the brain, exhibited an antinociceptive effect in the formalin test in rats [34]. Such an action of opioids in the PNS appeared to be mediated by opioid receptors in peripheral terminals of primary-afferent fibers ([33,34,35,36,83]; see [37] for review). In addition, the inhibitory effect of opioids on nerve AP conduction also might contribute to local analgesia following the peripheral perineural administration of opioids (for instance, see [84]) that are expected to lead to a direct action of opioids at high doses on peripheral nerves. Since codeine is metabolized to morphine via O-demethylation in humans and animals ([85,86]; see [81] for review), peripherally-administrated codeine might have a similar effect to that of morphine.

Antinociception produced by NSAIDs is mediated by various mechanisms such as (1) inhibition of the synthesis of prostaglandins from arachidonic acid by inhibiting the cyclooxygenase enzyme ([87,88]; see [89,90,91] for reviews), (2) inhibition of acid-sensitive ion channels [92] and transient receptor potential (TRP) channels [93,94], (3) activation of several K⁺ channels ([95,96,97,98,99]; see [100,101] for reviews), (4) substance P depletion [102], (5) an interaction with the adrenergic system [103] and (6) an involvement of opioids [104,105] and endocannabinoids (see [106] for review). The idea about an involvement of mechanisms other than cyclooxygenase inhibition in antinociception is supported by the observation that there is a dissociation between anti-inflammation and antinociception produced by NSAIDs [107].

An acetic acid-based NSAID diclofenac reduced frog sciatic nerve CAP peak amplitudes in a partially reversible manner. Diclofenac activity was concentration-dependent in a range of 0.01–1 mM with an IC₅₀ value of 0.94 mM. Another acetic acid-based NSAID aceclofenac (a carboxymethyl ester of diclofenac) also exhibited a similar CAP inhibitory action. CAP peak amplitudes were concentration-dependently reduced by aceclofenac in a range of 0.01–1 mM with an IC₅₀ of 0.47 mM, a value smaller than that of diclofenac. Other acetic acid-based NSAIDs had an efficacy smaller than those of diclofenac and aceclofenac. Indomethacin at 1 mM reduced CAP peak amplitudes by 38% and acemetacin (where the -OH group of indomethacin is substituted by -OCH₂COOH) at 0.5 mM did so by 38%. Etodolac at 1 mM reduced CAP peak amplitudes by only 15%, and sulindac and felbinac at 1 mM had no effects on CAP peak amplitudes [21].

A similar frog sciatic nerve CAP inhibition was produced by fenamic acid-based NSAIDs (tolfenamic acid, meclofenamic acid, mefenamic acid and flufenamic acid) whose chemical structures are similar to those of diclofenac and aceclofenac. Tolfenamic acid concentration-dependently reduced CAP peak amplitudes in a range of 0.01–0.2 mM with an IC₅₀ value of 0.29 mM. The activity of meclofenamic acid (where the chloro group bound to the benzene ring of tolfenamic acid is changed in number and position) was concentration-dependent in a range of 0.01–0.5 mM with an IC₅₀ value of 0.19 mM. Moreover, mefenamic acid (where the chloro group bound to the benzene ring of tolfenamic acid is replaced by methyl group) concentration-dependently reduced CAP peak amplitudes in a range of 0.01–0.2 mM with the extent of 16% at 0.2 mM. CAP peak amplitudes were concentration-dependently reduced by flufenamic acid (where one out of two methyl groups bound to the benzene ring of mefenamic acid is lacking and another one is replaced by -CF₃) with an IC₅₀ of 0.22 mM, a value comparable to those of tolfenamic acid and meclofenamic acid [21].

2,6-Dichlorodiphenylamine and N-phenylanthranilic acid (which are similar in chemical structure to diclofenac and tolfenamic acid while being not NSAIDs) reduced frog sciatic nerve CAP peak amplitudes; the former compound lacks the -CH₂COOH group of diclofenac and the latter one lacks chloro and methyl groups bound to the benzene ring of tolfenamic acid. 2,6-Dichloro- diphenylamine activity was concentration-dependent in a range of 0.001-0.1 mM with the extent of 45% at 0.1 mM; N-phenylanthranilic acid activity was concentration-dependent in a range of 0.01–2 mM with the extent of 23% at 1 mM [21].

With respect to other types of NSAIDs, salicylic acid-based (aspirin; 1 mM), propionic acid-based (ketoprofen, naproxen, ibuprofen, loxoprofen and flurbiprofen; each 1 mM) and enolic acid-based [meloxicam (0.5 mM) and piroxicam (1 mM)] NSAIDs had no effects on frog sciatic nerve CAP amplitudes [21].

CAP amplitude reductions produced by the NSAIDs would be mediated by an inhibition of TTX-sensitive voltage-gated Na⁺ channels that are involved in frog CAP production. In support of this idea, diclofenac decreased the peak amplitudes of TTX-sensitive Na⁺-channel currents in rat DRG [108] and mouse trigeminal ganglion neurons [109]. A similar diclofenac-induced Na⁺-channel inhibition has been reported in rat myoblasts [110] and ventricular cardiomyocytes [111]. Flufenamic acid as well as diclofenac decreased Na⁺-channel current amplitudes in rat hippocampal CA1 neurons [112,113,114]. Although IC₅₀ value (0.22 mM) for flufenamic acid in frog sciatic nerve CAP inhibition was similar to that of Na⁺ channel inhibition (0.189 mM) in rat hippocampal CA1 neurons [114], IC₅₀ value (0.94 mM) for diclofenac in CAP inhibition was much larger than those (0.00851 and 0.014 mM in rat myoblasts and DRG neurons, respectively) of Na⁺ channel inhibition [108,110]. Regarding rank order among NSAIDs, the order for CAP inhibition at 0.5 mM was flufenamic acid > diclofenac > indomethacin >> aspirin = naproxen = ibuprofen [21]; this was in part similar to those for Na⁺ channel inhibition in rat cardiomyocytes (diclofenac > naproxen ≥ ibuprofen; [111]) and also in rat DRG neurons (diclofenac > flufenamic acid > indomethacin > aspirin; [108]). With respect to TTX-resistant Na⁺ channels, diclofenac at 0.3 mM reduced peak current amplitudes by about 20% in rat trigeminal ganglion neurons [115]; Nav1.8 channel currents were inhibited by flufenamic acid and tolfenamic acid (current amplitude reduction: ca. 30 and 30%, respectively, at 0.1 mM; [116]). TTX-sensitive Nav1.7 channel currents were more sensitive to flufenamic acid and tolfenamic acid (reduction: ca. 60 and 70%, respectively, at 0.1 mM) than Nav1.8 ones [116]. Alternatively, chemical irritation-induced activity increase of cat corneal sensory nerve fibers was suppressed in extent by NSAIDs; this suppression was different in magnitude among distinct types of NSAIDs [109,117]. Na⁺-channel inhibition produced by NSAIDs appeared to be different in extent among preparations. Concentrations required for NSAIDs to have a significant inhibitory effect on frog sciatic nerve CAPs were in general higher than those needed to inhibit Na⁺ channels; this may be attributed to various reasons including the fact that not only Na⁺ channels but also K⁺ channels are involved in determining CAP amplitudes. To my knowledge, the effects of aceclofenac, indomethacin, etodolac, acemetacin, meclofenamic acid and mefenamic acid on voltage-gated Na⁺ channels have not been reported. Table 2 summarizes IC₅₀ values for frog sciatic nerve fast-conducting CAP inhibitions produced by NSAIDs together with those for voltage-gated Na⁺ channels.

NSAIDs (diclofenac, aceclofenac, tolfenamic acid, meclofenamic acid and flufenamic acid), which are more effective in frog sciatic nerve CAP inhibition compared to the other NSAIDs [21], have two benzene rings that bind a hydrophilic substituent group, both of which rings are linked by -NH- (see Figures 1Aa, 1Ba, 3Aa, 3Ba, 3Da in [21] for the chemical structures of the five NSAIDs), as seen in local anesthetics (see Section 3.4). Mefenamic acid (where one of the two benzene rings has a hydrophobic substituent group; see Figure 3Ca in [21]) appeared to be less effective, albeit not examined at a higher concentration due to a less solubility of this drug (see above). CAPs were effectively inhibited by 2,6-dichlorodiphenylamine and N-phenylanthranilic acid that are not NSAIDs while being similar in chemical structure to NSAIDs having two benzene rings (see Figures 4Aa and 4Ba in [21]). CAPs were also depressed by bisphenol A that have two benzene rings that bind a hydrophilic group such as -OH [26].

Much evidence demonstrates that the other actions of NSAIDs depend on their chemical structures. For instance, an involvement of NO-cGMP-K⁺ channels in antinociception mediated by NSAIDs was dependent on their chemical structures [98,118]. Nonselective cation channels in the rat exocrine pancreas were suppressed by flufenamic acid and mefenamic acid but not indomethacin, aspirin and ibuprofen [119]. There was a distinction in depressing TRP melastatin-3 channels between diclofenac and aceclofenac [94]. Although NSAIDs not only suppress but also activate TRP ankyrin-1 channels, this activation also differed in magnitude among NSAIDs [120]. Moreover, there was a distinction among NSAIDs in the activities of mitochondrial oxidative phosphorylation or electron transport system that may be involved in their adverse side effects [121].

Although the concentrations of NSAIDs tested in the frog sciatic nerve are generally much higher than those for voltage-gated Na⁺-channel inhibition, such high concentrations are likely when NSAIDs are used at high concentrations in the direct vicinity of nerve fibers. At least a part of analgesia caused by NSAIDs used as a dermatological drug for antinociception may be due to a nerve conduction inhibition through their inhibitory action on voltage-gated Na⁺ channels [122].