Excitatory amino acids and synaptic transmission: the evidence for a physiological function.

For 30 years physiological techniques have been used to investigate excitatory amino acids as neurotransmitters. In the last ten years progress on the definition of receptor subtypes and the availability of more selective agonists and antagonists has fuelled physiological, neurochemical and histochemical approaches to elucidating the involvement of excitatory amino acids at synaptic sites throughout the vertebrate CNS. Here Max Headley and Sten Grillner assess the advances made in defining the roles of excitatory amino acids as functional transmitters, taking examples mainly from studies on the spinal cord, and comment on the limitations of the types of approach that are used in such studies.

NMDA receptor antagonists as analgesics: focus on the NR2B subtype.

Ifenprodil and a group of related compounds are selective antagonists of NR2B-containing NMDA receptors. These compounds are antinociceptive in a variety of preclinical pain models and have a much lower side-effect profile compared with other NMDA receptor antagonists. It remains unclear whether the improved safety of these compounds is due to their subtype selectivity or to a unique mode of inhibition of the receptor. Human trials have so far confirmed the good tolerability of these subtype-selective NMDA receptor antagonists; however, whether they are as effective as other NMDA receptor antagonists in pain patients remains to be demonstrated.

NMDA antagonists and neuropathic pain--multiple drug targets and multiple uses.

NMDA (N-methyl-D-aspartate) receptors are one class of ionotropic receptor for the ubiquitous excitatory neurotransmitter L-glutamate. The receptor is made up of four protein subunits combined from a larger library of proteins, which gives this receptor a great deal of variability. This explains the large number of modulatory sites, a variety of sites at which antagonists can interact, and therefore a number of potential drug targets. Sensitivity of the NMDA ion channel to ambient levels of Mg++ gives it a voltage dependence that suits a function of responding to intense synaptic activation; the ability of the channel to admit Ca++ tends to trigger long-term processes. The receptor is thereby involved in long-term physiological processes such as learning and memory as well as in pathological processes such as neuropathic pain. Separating these functions therapeutically with NMDA antagonists has been a major difficulty, and has not yet been achieved with currently-available agents. This review summarises the preclinical rationale, based on animal models, and the clinical evidence on the use of NMDA antagonists in pain states. It also summarises the details of the receptor so as to explain the rationale for targeting either specific sites on the receptor, or exploiting anatomical differences in subtype expression, so as to provide the beneficial effects of NMDA receptor block with an improved side effect profile. In particular, agents that are selective for receptors that include the NR2B subunit preclinically have a substantially better profile for treating neuropathic pain than do current NMDA antagonists; some emerging clinical evidence supports this view.

The role of nitric oxide in spinal nociceptive reflexes in rats with neurogenic and non-neurogenic peripheral inflammation.

This in vivo electrophysiological study concerns the role of nitric oxide (NO) in mechanical and thermal spinal nociceptive reflexes in alpha-chloralose anaesthetized rats. The effects of the NO synthase inhibitor, NG-nitro-L-arginine methyl ester (L-NAME; 5-40 mg/kg i.v.) on reflexes were compared both in normal rats and in those with peripheral inflammation induced neurogenically (mustard oil) and non-neurogenically (carrageenan). Methoxamine (0.1 mg/kg i.v.) was used to mimic the marked hypertension caused by L-NAME. Thermal nociceptive reflexes were equally reduced by methoxamine and L-NAME in both normal and inflamed rats, implying that NO has no role in mediating thermal reflexes. However, L- (but not D-) NAME dose dependently and significantly inhibited mechanical reflexes in both carrageenan inflamed (to 37 +/- 12% control) and mustard oil inflamed rats (to 75 +/- 8% control). Moreover, these reductions were greater than those by methoxamine. In contrast, L-NAME did not reduce mechanical reflexes in rats with no inflammation or in spinalized rats with inflammation. The inhibition of mechanical reflexes with L-NAME in carrageenan inflamed rats was reversed and prevented by pre- or post-treatment with L- (but not D-) arginine (50-200 mg/kg i.v.). These data imply a supraspinal role for NO in mediating mechanical (but not thermal) nociceptive reflexes only in those rats with peripheral inflammation.

Interactions between metabotropic and ionotropic glutamate receptor agonists in the rat spinal cord in vivo.

Microelectrophoretic application of the non-selective metabotropic glutamate receptor (mGluR) agonist (1S,3R)-1-aminocyclopentane-1,3-dicarboxylic acid [(1S,3R)-ACPD] and the group I selective mGluR agonist (RS)-3,5-dihydroxyphenylglycine [(RS)-3,5-DHPG] potentiated the responses of rat spinal neurones to the cyclically-ejected ionotropic excitatory amino acid (EAA) agonists NMDA, AMPA and kainate in vivo. Potentiation was not selective between the three ionotropic responses and was paralleled by an enhancement of background activity in spontaneously active cells. "Correcting" spike count data for this increase in background activity showed that the EAA responses were not potentiated beyond the apparent enhancement of cell excitability. Neither mGluR agonist produced potentiation when NMDA/AMPA cycling was superimposed on background discharge held constant with kainate. It is concluded that potentiation produced by both (1S,3R)-ACPD and (RS)-3,5-DHPG is secondary to an enhancement of cell excitability rather than being due to a specific interaction with ionotropic EAA receptors. The mechanism of excitability enhancement cannot be determined by extracellular recording, but group I mGluRs are most likely to be responsible.

Thyrotropin-releasing hormone (TRH)-induced facilitation of spinal neurotransmission: a role for NMDA receptors.

Thyrotropin-releasing hormone (TRH) is known to enhance spinal reflexes and modulate NMDA receptors in supraspinal areas. We have investigated the relationship between TRH and NMDA receptors in the spinal cord of alpha-chloralose-anaesthetized spinalized rats. TRH was tested (a) on dorsal horn neurone responses to iontophoretic NMDA, AMPA and kainate and (b) on spinal reflexes evoked by noxious pinch and electrical stimulation (2 Hz) shown to involve NMDA receptor activation. TRH given i.v. (0.5 mg/kg) or iontophoretically selectively potentiated neuronal responses to NMDA. TRH (0.5-2 mg/kg) also dose-dependently increased single motor unit reflex responses. The NMDA antagonist ketamine (2 mg/kg i.v.) abolished these TRH effects; ketamine reduced single motor unit (SMU) reflex responses more effectively when administered after TRH than in pre-TRH control tests. These results indicate that TRH-induced facilitation of spinal sensory transmission involves NMDA receptor activation.

NMDA-receptor contribution to spinal nociceptive reflexes: influence of stimulus parameters and of preparatory surgery.

The contribution of NMDA receptors to nociceptive reflexes has been assessed both in awake rats and in electrophysiological tests on alpha-chloralose anaesthetized spinalized rats prepared with different degrees of surgery. Single motor unit activity was recorded in response to alternating noxious mechanical and electrical stimuli applied to one hindpaw, and the results compared with paw pressure withdrawal reflexes in awake rats. There was little contribution by NMDA receptors to nociceptive paw pinch responses either in awake rats or in rats prepared with minimal surgery, but following extensive lumbar surgery the contribution increased significantly to a level similar to that seen in the wind-up component of responses elicited by electrical stimulation. Surgery therefore has effects several segments from the sensory input that it generates. It enhances the NMDA receptor contribution in responses to some but not all types of afferent input.

The use of cyclic AMP efflux studies in attempts to determine the effects of morphine on cyclic AMP formation in striatal slices.

The effect of morphine and naloxone on basal and forskolin-stimulated efflux of cyclic AMP from rat striatal slices was examined. Neither morphine nor naloxone had any consistent effect on the basal efflux of cyclic AMP. Forskolin produced a time and dose-dependent enhancement of cyclic AMP efflux. Neither morphine nor naloxone affected this forskolin-enhanced release. These results suggest that measurements of cyclic AMP released from brain slices do not accurately reflect effects on adenylate cyclase inhibition by opiates.

The in vivo relevance of the varied channel-blocking properties of uncompetitive NMDA antagonists: tests on spinal neurones.

The voltage dependence and channel-blocking kinetics of uncompetitive NMDA receptor antagonists have been well-described using in vitro techniques, but there is little evidence concerning the functional significance of these properties in vivo. We have now compared the effects of NMDA antagonists that display varied profiles of voltage-dependent block in vitro, on responses of spinal neurones in anaesthetised rats. The compounds examined were the uncompetitive channel blockers memantine, ketamine and MK-801 and, for comparison, an antagonist that acts at the strychnine-insensitive glycine binding site (MRZ 2/502). Using frequency of spike discharge as an indicator of somatic depolarisation, we have compared the effects of these antagonists on responses evoked by iontophoretic NMDA application and on synaptic responses evoked by pinch or electrical stimulation (the latter eliciting "wind-up"). The effectiveness of the antagonists was directly but variably related to the discharge frequency of the test response. The rank order of dependence on firing rate matched the rank order of voltage dependence reported in vitro, namely: memantine > ketamine > MK-801> or = MRZ 2/502. Doses that reduced responses to iontophoretic application of NMDA were less effective at reducing responses to pinch, perhaps due to the major non-NMDA component of the synaptic response. Memantine preferentially reduced "wind-up" relative to responses to pinch, whereas ketamine and MK-801 reduced both types of synaptic responses in parallel. This "filtering" by low affinity, voltage-dependent NMDA antagonists such as memantine, of non-physiological activity whilst leaving normal synaptic events relatively untouched, may contribute to their more favourable clinical profile.

Antagonism of mGlu receptors and potentiation of EPSCs at rat spinal motoneurones in vitro.

The patch-clamp technique has been used to record synaptic responses, elicited by electrical stimulation of dorsal roots, in 28 single motoneurones of in vitro spinal cord preparations from neonate (P5 to P8) rats. The effects of (RS)-alpha-methyl-4-phosphonophenylglycine (MPPG) (200 microM), a potent antagonist at L-2-amino-4-phosphonobutanoate (AP4)-sensitive receptors, and (RS)-alpha-methyl-4-carboxyphenylglycine (MCPG) (500 microM), which is a less selective antagonist of mGluRs, were tested on EPSCs alone and as antagonists of AP4-induced depression of EPSCs. The EC50 for depression of EPSCs by AP4 (1.16 +/- 0.12 microM, n = 8) was increased to 18.9 +/- 0.7 microM (n = 6) by MPPG. MCPG (500 microM) had no significant effect on the depressant potency of AP4. Under control conditions, EPSCs had mean peak amplitudes of 983 pA +/- 64 SEM and mean charge transferred of 306 +/- 37 pC (n = 28). These values were increased significantly (p < 0.05) to 1168 +/- 68 pA and 363 +/- 39 pC by MPPG (n = 6), and 1150 +/- 54 pA and 358 +/- 33 pC (n = 6) by MCPG. There was no significant difference between the enhancement of the initial peak of the EPSCs (mean latency from stimulus artifact 5.9 +/- 0.3 ms) and later components, suggesting mGluRs to be present on primary afferent terminals presynaptic to motoneurones as well as in pathways via interneurones. These results are consistent with the presence of at least two types of presynaptic mGluR that modulate release of glutamate in segmental pathways convergent onto motoneurones. These receptors appear to be activated by interstitial glutamate tonically present in the present preparations.

Do kappa opioids mimic sigma agonists as amino acid antagonists?

The kappa opioids tifluadom and U-50,488 H, when tested on spinal neurones of rats in vivo and frogs in vitro, had no selective effect on responses to microelectrophoretically or bath applied N-methyl-D-aspartate, quisqualate or kainate. In the same preparations ketamine selectively reduces responses to N-methyl-D-aspartate. Amino acid antagonism by dissociative anaesthetics and sigma opioids is thus not mediated by that binding site which kappa and sigma opioids have been reported to have in common.

Antagonism of baclofen-induced depression of whole-cell synaptic currents in spinal dorsal horn neurones by the potent GABAB antagonist CGP55845.

The potencies of two GABAB receptor antagonists P-[3-aminopropyl]- P-diethoxymethyl-phosphinic acid (CGP35348) and the novel compound 3-N[1-(S)-(3,4-dichlorophenyl)ethyl]amino-2-(S)-hydroxypropyl-P- benzyl-phosphinic acid (CGP55845) have been compared in an in vitro spinal cord preparation. They have been tested as antagonists of baclofen-induced depression of EPSCs of patch-clamped dorsal horn neurons following electrical stimulation of dorsal roots. Mean EC50 values for the depressant action of baclofen were increased by 50- and 140-fold respectively in the presence of CGP35348 (200 microM) (n = 5) and CGP55845 (100 nM) (n = 4). This potency of CGP55845 is > 1000-fold higher than that reported previously for other GABAB receptor antagonists.

Actions of kainate and AMPA selective glutamate receptor ligands on nociceptive processing in the spinal cord.

Kainate receptors expressing the GluR5 subunit of glutamate receptor are present at high levels on small diameter primary afferent neurones that are considered to mediate nociceptive inputs. This suggests that GluR5 selective ligands could be novel analgesic agents. The role of kainate receptors on C fibre primary afferents has therefore been probed using three compounds that are selective for homomeric GluR5 receptors. The agonist, ATPA, and the antagonists, LY294486 and LY382884, have been tested in four models of nociception: responses evoked by noxious stimulation of the periphery have been recorded electrophysiologically (1) from hemisected spinal cords from neonatal rats in vitro, (2) from single motor units in adult rats in vivo, (3) from dorsal horn neurones in adult rats in vivo, and (4) in hotplate tests with conscious mice. In some protocols comparisons were made with the AMPA selective antagonist GYKI 53655. The agonist ATPA reduced nociceptive reflexes in vitro, but failed to have effects in vivo. In all tests, the GluR5 antagonists reduced nociceptive responses but only at doses that also affected responses to exogenous AMPA. The AMPA antagonist reduced nociceptive responses at doses causing relatively greater reductions of responses to exogenous AMPA. The results indicate that GluR5 selective ligands do reduce spinal nociceptive responses, but they are not strongly analgesic under these conditions of acute nociception.

New anticonvulsants: Schiff bases of gamma-aminobutyric acid and gamma-aminobutyramide.

Schiff bases of gamma-aminobutyric acid (gammaAbu) and gamma-aminobutyramide (gammaAbuNH2) were prepared and tested for anticonvulsant and gammaAbu mimetic activity. 4-[[(4-Chlorophenyl)(5-fluoro-2-hydroxyphenyl)methylene]amino]butanoic acid monosodium salt (4) and 4-[[(4-chlorophenyl)(5-fluoro-2-hydroxyphenyl)methylene]amino]butanamide (5) blocked bicuculline-induced lethality and convulsions and displaced [3H]gammaAbu from its membrane binding sites. In the rat dorsal root sensory ganglion, compound 4 exhibited gammaAbu agonist properties. Compounds 4 and 5 are thus anticonvulsants and directly acting gammaAbu mimetics.

Stimulus intensity, cell excitation and the N-methyl-D-aspartate receptor component of sensory responses in the rat spinal cord in vivo.

The importance of receptors for N-methyl-D-aspartate in synaptic plasticity and in triggering long-term pronociceptive changes is explained by their voltage-dependence. This suggests that their contribution to acute nociceptive responses would be determined both by the magnitude of synaptic input and by the level of background excitation. We have now examined the role of N-methyl-D-aspartate receptors in acute nociceptive transmission in the spinal cord. Drugs selectively affecting activity mediated by these receptors were tested on responses of dorsal horn neurons to noxious stimuli of different intensities and at different levels of ongoing spike discharge. The drugs used were the N-methyl-D-aspartate receptor channel blocker ketamine; the competitive antagonists, 3-((R)-2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid (D-CPP) and D-2-amino-5-phosphonopentanoic acid (D-AP5), and the positive modulator thyrotropin-releasing hormone. The activity of dorsal horn wide dynamic range neurons was recorded extracellularly in alpha-chloralose-anaesthetized spinalized rats. Their responses to noxious stimuli (pinch, heat and electrical) were monitored in parallel with responses to iontophoretic N-methyl-D-aspartate and (RS)-alpha-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA). Drugs were given i.v. or (D-AP5) iontophoretically. At doses that selectively inhibited responses to exogenous N-methyl-D-aspartate, ketamine (4 or 8, mean 5 mg/kg i.v.) reduced the nociceptive responses of the majority of the cells in deep dorsal horn. Ketamine also reduced wind-up of the responses to repetitive electrical stimulation. Ketamine (4 or 8 mg/kg). D-CPP (2 mg/kg), D-AP5 (iontophoretically) and thyrotrophin-releasing hormone (1 mg/kg) were tested on different magnitude nociceptive responses evoked by alternating intensities of noxious heat or pinch. In percentage terms, the less vigorous responses were affected by all four drugs as much as or more than the more vigorous responses. When background activity of neurones was enhanced by continuous activation of C-fibres with cutaneous application of mustard oil, ketamine was less effective against superimposed noxious pinch responses. Ongoing background activity was affected in parallel with evoked responses. When background discharge of the cells was maintained at a stable level with continuous ejection of kainate, neither the N-methyl-D-aspartate antagonists nor thyrotrophin-relasing hormone affected the responses to noxious pinch or heat, although responses to exogenous N-methyl-D-aspartate were still blocked. The wind-up of the electrical responses was, however, reduced by ketamine irrespective of the level of background activity. The results indicate that under these conditions in vivo, N-methyl-D-aspartate receptors mediate ongoing low-frequency background activity rather than phasic high-frequency nociceptive responses. The effects of N-methyl-D-aspartate antagonists and positive modulators on nociceptive responses are evidently indirect, being secondary to changes in background synaptic excitation. These results cannot be explained simply in relation to the voltage-dependence of N-methyl-D-aspartate receptor-mediated activity; other factors, such as modulation by neuropeptides, must be involved.

Metabotropic glutamate receptor antagonists, like GABA(B) antagonists, potentiate dorsal root-evoked excitatory synaptic transmission at neonatal rat spinal motoneurons in vitro.

Recordings of whole-cell synaptic current responses elicited by electrical stimulation of dorsal roots were made from motoneurons, identified by antidromic invasion, in isolated spinal cord preparations from five- to eight-day-old Wistar rats. Supramaximal electrical stimulation of the dorsal root evoked complex excitatory postsynaptic currents with mean latencies (+/- S.E.M.) of 6.1 +/- 0.26 ms, peak amplitude of -650 +/- 47 pA and duration of 4.30 +/- 0.46 s (n=34). All phases of excitatory postsynaptic currents were potentiated to approximately 20% above control levels in the presence of the metabotropic glutamate receptor antagonists S-2-amino-2-methyl-4-phosphonobutanoate (MAP4; 200 microM; n=15) and 2S, 1'S,2'S-2-methyl-2-(carboxycyclopropyl)glycine (MCCG; 200 microM; n=9). A similar level of potentiation was produced by the GABA(B) receptor antagonist 3-N[1-(S)-(3,4-dichlorophenyl)ethyl]amino-2-(S)-hydroxypropyl-P-benzyl-p hosphinic acid (CGP55845; 200 nM; n=5). MAP4 (200 microM) produced a six-fold rightward shift in the concentration-effect plot for the depressant action of the metabotropic glutamate receptor agonist S-2-amino-4-phosphonobutanoate (L-AP4), whereas CGP55845 produced no significant change in the potency of L-AP4. MAP4 did not antagonize the depressant actions of baclofen (n=8), 1S,3S-1-aminocyclopentane-1,3-dicarboxylate (n=4) or 2-S,1'S,2'S-2-(carboxycyclopropyl)glycine (n=4). The metabotropic glutamate receptor antagonists produced no change in the holding current of any of the neurons, indicating that they had no significant postsynaptic excitatory actions. These results are the first to indicate a possible physiological role for metabotropic glutamate receptors in the spinal cord. Like GABA(B) receptors, they control glutamatergic synaptic transmission in the segmental spinal pathway to motoneurons. This is likely to be a presynaptic control mechanism.

Functional evidence for multiple receptor activation by kappa-ligands in the inhibition of spinal nociceptive reflexes in the rat.

1. The evidence for kappa-receptor heterogeneity is equivocal. We have now investigated this question by comparing the effects of five putatively selective kappa-agonists. The parameters examined were: the relative potencies in depressing hindlimb flexor muscle reflexes to noxious pinch stimuli in both spinalized and sham-spinalized rats; the reversibility of these effects by naloxone; and the effects on blood pressure. 2. Two types of drug effect was discriminated. One drug group, represented by U-50,488, U-69,593 and PD-117,302, had a potency ratio between sham and spinalized rats approximately 10 fold lower than the other group, which comprised GR103545 and CI-977. 3. Under sham-spinalized conditions, CI-977 and GR103545 at high doses caused only sub-maximal reductions of spinal reflexes. U-50,488 was still active when superimposed on these high doses of GR103545. 4. Naloxone reversed all effects, but different doses were required between compounds, with GR103545 taking some 20 times higher doses of naloxone to cause reversal than did U-50,488. 5. The effects on mean arterial pressure were opposite between groups. 6. The results imply that more than one type of naloxone-sensitive non-mu opioid receptor must be involved in mediating these complex actions of ligands that have been claimed to be selective for kappa-receptors.

Spinal antinociceptive actions of mu- and kappa-opioids: the importance of stimulus intensity in determining 'selectivity' between reflexes to different modalities of noxious stimulus.

1. In electrophysiological experiments in spinalized rats, mu- and kappa-opioids were tested intravenously on the responses of single motoneurones to electronically controlled, alternating noxious heat and noxious pinch stimuli. The effects of mu- and kappa-opioids were compared with those of the general anaesthetic alpha-chloralose and the dissociative anaesthetic/PCP ligand ketamine. 2. The kappa-opioids U-50,488 (0.5-16 mgkg-1 i.v.) and tifluadom (0.05-1.6 mgkg-1 i.v.) had very similar actions to the mu-opioid fentanyl (0.5-16 micrograms kg-1 i.v.). Thus all three agonists reduced thermal and mechanical nociceptive reflexes in parallel and in a dose-dependent manner, but only so long as neuronal responses to the alternating stimuli elicited similar excitability levels in the neurone under study. Ketamine (0.5-16 mgkg-1 i.v.) had similar actions to the opioids whereas alpha-chloralose (20 mgkg-1 i.v.) had very little effect on neuronal responsiveness. 3. Apparently 'selective' depressions by both mu- and kappa-opioids could be orchestrated by a deliberate mismatch of the intensities of alternating noxious heat and pinch stimuli; as measured by neuronal firing rate, the weaker of the responses to either type of stimulus was invariably reduced to a greater degree. 4. Similar 'selectivity' could be demonstrated for both mu- and kappa-ligands when the weaker and stronger responses were of the same modality, being applied by the same pincher device but with alternating applied force. 5. It is concluded that the 'selective' spinal actions of kappa-opioids seen in non-thermal over thermal behavioural models of nociception is likely to be related to the relative intensities, rather than the modalities, of the noxious stimuli used. The validity of the interpretation of results obtained in such behavioural studies is discussed.

On the selectivity of intravenous mu- and kappa-opioids between nociceptive and non-nociceptive reflexes in the spinalized rat.

1. In electrophysiological experiments in spinalized, alpha-chloralose anaesthetized rats, opioids and anaesthetics were tested intravenously (i.v.) on the responses of individual motoneurones to alternating noxious (heat or pinch) and non-noxious (tap or vibration) stimuli. 2. On cells that were sensitive to low doses of mu-opioids, both fentanyl (0.5-4 micrograms kg-1 i.v.) and morphine (0.5 mg kg-1 i.v.) selectivity reduced reflexes to noxious stimuli to a greater degree than the higher doses required to reduce nociceptive reflexes (fentanyl 8 micrograms kg-1 i.v.; morphine 1-8 mg kg-1 i.v.) depressed non-nociceptive reflexes to a similar degree. 3. A similar spectrum of selectivity was seen with U-50,488 (0.5-16 mg kg-1 i.v.) although statistically significant selective depression of reflexes was only evident at the lowest dose tested (0.5 mg kg-1 i.v.). All effects of U-50,488 were readily reversed by low doses of the opioid antagonist, naloxone (10-100 micrograms kg-1 i.v.). 4. The dissociative anaesthetic/PCP ligand ketamine (0.5-4 mg kg-1 i.v.) was similar in having selective actions at low doses on sensitive cells but non-selective actions when higher doses were required. In contrast, the general anaesthetics methohexitone (4 mg kg-1 i.v.) and alphadolone/alphaxalone (1 mg kg-1 i.v.) were consistently non-selective between reflexes to noxious and non-noxious stimuli. alpha-Chloralose (20-40 mg kg-1 i.v.) had very little effect on reflexes to any of the synaptic inputs tested.

Reversal by naloxone of the spinal antinociceptive actions of a systemically-administered NSAID.

1. Possible interactions between non-steroidal anti-inflammatory drugs (NSAIDs) and endogenous opioids were examined in electrophysiological experiments in alpha-chloralose anaesthetized spinalized rats without or with carrageenan-induced acute inflammation of one hindpaw. Spinal reflex responses, monitored as single motor unit discharges, were elicited by noxious pinch and electrical stimuli. 2. The mu-opioid agonist, fentanyl, was an effective depressant of reflexes under all conditions (ED50 6-14 micrograms kg-1, i.v.). In rats without peripheral inflammation the NSAID, flunixin, a niflumic acid derivative, had only a small effect that was not dose-dependent. However, in animals with unilateral inflammation, flunixin reduced spinal reflexes evoked both by noxious pinch stimuli (that activate peripheral nociceptors; ID50 4 mg kg-1, i.v.) and by electrical stimuli (that bypass nociceptor endings; ID50 6.5- 11 mg kg-1, i.v.), indicating that it has a central site of action at doses comparable to those used clinically. 3. The opioid antagonist, naloxone (1 mg kg-1, i.v.), reversed all actions of fentanyl. It did not reverse the small effects that flunixin had in rats without inflammation, showing that the NSAID is not a direct opioid agonist. In rats with carrageenan-induced inflammation of the hindpaw, however, naloxone fully reversed or prevented the antinociception by flunixin, but not that by the alpha 2-adrenoceptor agonist, medetomidine. 4. We conclude that under conditions of peripheral inflammation and the resultant central changes, the NSAID, flunixin, has antinociceptive actions that are mediated by endogenous opioids acting within the spinal cord.