ABSTRACT
Intracellular recording techniques have been used to provide information on the identity of excitatory transmitters released at synapses formed between dorsal root ganglion (DRG) and spinal cord neurones in two in vitro preparations. Explants of embryonic rat DRG were added to dissociated cultures of embryonic dorsal horn neurones and synaptic potentials recorded intracellularly from dorsal horn neurones after DRG explant stimulation. More than 80% of dorsal horn neurones received at least one fast, DRG-evoked, monosynaptic input. In the presence of high divalent cation concentrations (5 mmoll−1 Ca2+, 3mmoll−1 Mg2+) the acidic amino acid receptor agonists, L-glutamate, kainate (KA) and quisqualate (QUIS) excited all dorsal horn neurones which received a monosynaptic DRG neurone input, whereas L-aspartate and.V-methyl-D-aspartate (NMDA) had little or no action. 2-Amino-5-phosphonovalerate (APV), a selective NMDA receptor antagonist, was relatively ineffective at antagonizing DRG-evoked synaptic potentials and L-glutamate-evoked responses. In contrast, kynurenate was found to be a potent antagonist of amino acid-evoked responses and of synaptic transmission at all DRG-dorsal horn synapses examined. The blockade of synaptic transmission by kynurenate appeared to result from a postsynaptic action on dorsal horn neurones.
Intracellular recordings from motoneurones in new-born rat spinal cord were used to study the sensitivity of the la excitatory postsynaptic potential (EPSP) to antagonists of excitatory amino acids. Superfusion of the spinal cord with APV did not inhibit the la EPSP but did suppress later, polysynaptic components of the afferent-evoked response. Kynurenate was a potent and selective inhibitor of the la EPSP, acting via a postsynaptic mechanism. These findings indicate that L-glutamate, or a glutamate-like compound, but not L-aspartate, is likely to be the predominant excitatory transmitter that mediates fast excitatory postsynaptic potentials at primary afferent synapses with both dorsal horn neurones and motoneurones.
INTRODUCTION
The transmission of sensory information at primary afferent synapses in the spinal cord involves the release of sensory transmitters that elicit excitatory postsynaptic potentials (EPSPs) in spinal neurones (Eccles, 1964). The most intensively studied afferent synapse in the spinal cord is that between muscle spindle afferents (group la afferents) and α-motoneurones (Burke & Rudomin, 1977; Redman, 1979). In the mammalian spinal cord, the chemical nature of the monosynaptic la excitatory postsynaptic potential (EPSP) has been well documented. The la EPSP is preceded by a distinct synaptic delay (Brock, Coombs & Eccles, 1952) and reverses in polarity when the membrane potential is depolarized (Coombs, Eccles & Fatt, 1955; Engberg & Marshall, 1979; Finkel & Redman, 1983). Moreover, synaptic transmission is completely and reversibly blocked by removal of extracellular Ca2+ (Shapolvalov, Shiriaev & Tamarova, 1979).
Intracellular recording from neurones in the superficial dorsal horn of rat spinal cord slices maintained in vitro has demonstrated that activation of cutaneous afferent fibres also elicits fast EPSPs in dorsal horn neurones (Urban & Randic, 1984). In addition, an increase in the frequency of dorsal root stimulation evokes, in the same postsynaptic neurones, slow depolarizing potentials that persist for seconds or minutes.
The identity of the excitatory neurotransmitters released at sensory synapses formed by cutaneous or muscle afferents has not been established (Redman, 1979; Fagg & Foster, 1983; Salt & Hill, 1983a,b). L-glutamate has been proposed as a transmitter in primary afferents based on its distribution in the spinal cord and dorsal roots (Graham, Shank, Werman & Aprison, 1967). It is the only amino acid found in the dorsal roots and dorsal columns in higher concentrations than in the ventral roots (Graham et al. 1967; Duggan & Johnston, 1970; Roberts, Keen & Mitchell, 1973). Biochemical studies have demonstrated the release of endogenous stores of L-glutamate from regions of the CNS containing primary afferent terminals (Roberts, 1974; Takeuchi, Onodera & Kawagoe, 1983). lontophoretic and pressure application of L-glutamate depolarizes the majority of mammalian spinal neurones in vivo and in vitro (Ransom, Bullock & Nelson, 1977; Watkins & Evans, 1981; Salt & Hill, 1983a,b) with a reversal potential (Mayer & Westbrook, 1984) similar to that of the EPSPs evoked by stimulation of DRG neurones (Engberg & Marshall, 1979; Finkel & Redman, 1983; MacDonald, Pun, Neale & Nelson, 1983). Confirmation of the role of L-glutamate as an excitatory neurotransmitter at primary afferent terminals has been hindered by the lack of specific L-glutamate receptor antagonists and by difficulties in administering these compounds, in a controlled way, at identified primary afferent synapses.
Some clarification of the role of L-glutamate in central synaptic transmission has derived from pharmacological studies indicating the existence of distinct amino acid receptor subtypes (Watkins & Evans, 1981; Foster & Fagg, 1984). Selective ligands are available for only one receptor subtype; N-methyl-D-aspartate (NMDA) is an agonist and 2-amino-5-phosphonovalerate (APV) an antagonist at the NMDA subclass of L-glutamate receptors (Davies, Francis, Jones & Watkins, 1981). The NMDA receptor can be further distinguished by a voltage-dependent block of the NMDA-activated ion channel by Mg2+ and other divalent cations (Ault et al. 1980; Mayer, Westbrook & Guthrie, 1984; Nowak et al. 1984). Two other L-glutamate receptor subclasses have been proposed on the basis of the selectivity of the rigid structural analogues of L-glutamate, quisqualate (QUIS) and kainate (KA) (Watkins & Evans, 1981; Foster & Fagg, 1984). QUIS and KA receptors are not activated by NMDA and are not sensitive to APV at concentrations that are sufficient to block NMDA receptors. Moreover, the ion channel associated with QUIS and KA receptors does not appear to be subject to voltage-dependent blockade by Mg2+. At present, no structural or functional distinction between QUIS and KA receptors on spinal neurones has been established, since no antagonists exist that distinguish readily between these receptor subclasses. γ-D-Glutamylglycine (DGG) and cis-2,3-piperidine dicarboxylate (PDA) antagonize all L-glutamate-activated receptors while having no antagonistic effect on responses to non-amino acid compounds (Watkins & Evans, 1981; Foster & Fagg, 1984).
At low agonist concentrations, and at membrane potentials near rest, L-glutamate itself has mixed agonist activity with about equal contributions of NMDA and QUIS/KA receptors to L-glutamate-activated currents (Mayer & Westbrook, 1984). The specificity of action of L-aspartate appears to vary (Davies et al. 1982) although in mouse and rat spinal neurones in culture L-aspartate exhibits reasonable selectivity for NMDA receptors (Mayer & Westbrook, 1984; Jahr & Jessell, 1985).
To provide more direct information on the identity of sensory transmitters released by DRG neurones and to characterize the postsynaptic receptors that mediate fast EPSPs at sensory synapses, we have examined the action of excitatory amino acid receptor ligands at synapses formed between dorsal root ganglion and dorsal horn neurones in cell culture. In addition, intracellular recording from motoneurones in in vitro preparations of spinal cord from new-born rats (Otsuka & Konishi, 1974) has been used to examine the possibility that L-glutamate mediates the fast EPSP produced by a defined class of primary afferents, the group la muscle spindle afferents. The use of these in vitro preparations has permitted analysis of the effects of known concentrations of excitatory amino acid antagonists and of the specificities of the synaptic receptors mediating the action of the afferent transmitter.
SYNAPTIC TRANSMISSION BETWEEN DRG AND DORSAL HORN NEURONES IN CELL CULTURE
Several groups have used dissociated cultures of DRG and spinal cord neurones to examine synaptic transmission at DRG-spinal cord, spinal cord-spinal cord and spinal cord—DRG synapses (Ransom et al. 1977; Choi & Fischbach, 1981; MacDonald et al. 1983). We have found that one major difficulty in studying primary afferent transmission between DRG and dorsal horn neurones in dissociated cell co-cultures is the low detectable incidence of synaptically coupled pairs of neurones. To overcome this problem we have grown explants of DRG neurones in co-culture with dissociated dorsal horn neurones. Simultaneous stimulation of many DRG neurones increases dramatically the probability of recording from dorsal horn neurones with monosynaptic sensory input (Jahr & Jessell, 1985).
Intracellular recordings have been obtained from dorsal horn neurones that received monosynaptic DRG input. In low concentrations of divalent cations (3 mmol l−1 Ca2+, 0·9 mmol l−1 Mg2+), spontaneous postsynaptic potentials are detectable in more than 95% of dorsal horn neurones. Since no spontaneous synaptic activity or action potentials are detectable when recording from DRG neurones in the same cultures, the spontaneous EPSPs probably reflect input from other dorsal horn neurones. Electrical stimulation of DRG explants evokes EPSPs in a high proportion of dorsal horn neurones located near the ganglion explant.
Postsynaptic responses in recording medium containing 3 mmol l−1 Ca2+ and 0·9 mmol l−1 Mg2+, are mediated in part by polysynaptic circuits. Superfusion of cultures with medium containing high divalent cations (5 mmol l−1 Ca2+ and 3 mmol l−1 Mg2+) increases the spike threshold of dorsal horn neurones (Frankenhauser & Hodgkin, 1957; Jahr & Jessell, 1985) and decreases or blocks all but the earliest monophasic EPSP.
Several characteristics of the synaptic potentials elicited in dorsal horn neurones by extracellular stimulation of DRG indicate that they are monosynaptic in origin. Under conditions in which the divalent cation concentration in the recording medium is increased to a level sufficient to block spontaneous synaptic input to dorsal horn neurones (MacDonald et al. 1983), monosynaptic EPSPs generated in dorsal horn neurones by DRG stimulation can be studied without contamination from dorsal horn neuronal input. Evoked EPSPs follow repetitive DRG stimulation at 10 Hz and are monophasic with a constant latency, providing additional evidence that they are monosynaptic. The high proportion of dorsal horn neurones that receive DRG input under these recording conditions has therefore made it possible to compare the pharmacology of monosynaptic EPSPs with that of the potentials evoked by various excitatory transmitter candidates on the same dorsal horn neurones.
Examination of the chemosensitivity of cultured dorsal horn neurones has indicated that excitatory amino acids have excitatory actions consistent with their roles as fast sensory transmitters (Jahr & Jessell, 1985). The sensitivity to excitatory amino acids and related compounds was tested only in those dorsal horn neurones that received monosynaptic sensory input. The effects of excitatory amino acids and their analogues on dorsal horn neurones were consistent. L-glutamate (10–20γmoll−1), QUIS (l–10γmoll−1) and KA (10–20 γmol l−1) were all potent exciters (Fig. 1). KA and L-glutamate were approximately equipotent, whereas QUIS was 10–20 times more potent. These agonists caused rapid depolarizations of up to 40 mV. The depolarizations were often suprathreshold. In contrast, NMDA and L-aspartate, at concentrations up to 200γmol l−1, either had no effect or elicited only small depolarizations (1–5 mV) after repeated application (Fig. 1). Dorsal horn neurones that were insensitive to L-aspartate, however, did receive monosynaptic input from DRG neurones (Fig. 2).
To determine whether amino acids, or compounds acting at amino acid receptors on dorsal horn neurones, might be transmitters at sensory synapses, we examined the effect of several antagonists of excitatory amino acid-evoked responses on the DRG-evoked monosynaptic EPSP recorded from dorsal horn neurones. The tryptophan metabolite kynurenate (Elmslie & Yoshikami, 1983; Ganong, Lanthorn & Cotman, 1983; Perkins & Stone, 1982) was found to be the most potent antagonist of both the EPSP and the depolarization evoked by pressure ejection of L-glutamate, QUIS or KA (Fig. 3). Kynurenate, at a concentration of 1 mmol l−1, completely blocked monosynaptic EPSPs and amino acid-evoked responses.’ In contrast, neither 2-amino-4-phosphonobutyrate (APB) nor L-glutamate diethylester at concentrations up to 1 mmol l−1 significantly affected the DRG-evoked EPSP or the responses of dorsal horn neurones to L-glutamate, KA or QUIS. APB is considered to exert presynaptic inhibitory actions at amino acid-mediated synapses, although in these culture experiments, no inhibitory action was observed. At the same concentration, the selective NMDA receptor antagonist, APV, slightly reduced the amplitude of the EPSP and of the L-glutamate-evoked depolarization. PDA and DGG reduced the amplitude of the EPSP and the L-glutamate-evoked depolarization, but with a potency less than half that of kynurenate. Other compounds related to kynurenate, quinolinate, kynurenine, kynurenamine, xanthurenate and picolinate, had neither agonist nor antagonist activity.
Since kynurenate was the most potent antagonist at DRG-dorsal horn synapses identified in these studies, additional experiments were performed to determine its site and mechanism of action. The reduction by kynurenate of EPSP amplitude and of amino acid-evoked depolarization was not associated with a change in the input resistance of dorsal horn neurones or with changes in the threshold, amplitude or duration of the action potential recorded intracellularly from the cell bodies of DRG neurones (Fig. 3).
The blockade of amino acid-evoked depolarization of dorsal horn neurones clearly demonstrates a postsynaptic site of action of kynurenate. It is possible, however, that kynurenate antagonizes EPSPs by blocking action potential propagation in the axons of DRG neurones presynaptic to dorsal horn neurones. Graded stimulation of DRG explants revealed discrete steps in the amplitude of the EPSP, suggesting that several DRG neurones can form functional synapses with a single dorsal horn neurone (Fig. 4). In the presence of kynurenate the same number of discrete steps could be evoked, although the amplitude of each incremental step was reduced to a similar extent (Fig. 4). Moreover, recruitment of each step in the EPSP occurred at a stimulus strength identical to that required before addition of kynurenate. These findings indicate that kynurenate does not block spike propagation in DRG axons and are consistent with a postsynaptic site of inhibition of the DRG-evoked EPSP.
Additional information on the site and mechanisms of action of kynurenate was obtained by examining its effect on synaptic depression evoked by paired presynaptic stimuli (Jahr & Jessell, 1985). At stimulus intervals of 50–400 ms, the amplitude of the second EPSP was depressed to 50–70% of that of the first EPSP in medium containing 5 mmol l−1 Ca2+ and 3 mmol l−1 Mg2+. When the Mg2+/Ca2+ ratio was increased or Ca2+ channel blockers such as Co2+ or Cd2+ were added to the superfusion medium, synaptic depression was either partly or entirely abolished. In experiments in which the first EPSP was substantially decreased, a potentiation of the second EPSP resulted. In contrast, when the first EPSP was decreased to about 20% of its control amplitude by addition of 0·5 mmol l−1 kynurenate, the second response was decreased to a similar extent. In fact, the addition of kynurenate sometimes resulted in an enhancement of the synaptic depression that may be due to a slight use-dependency of kynurenate blockade. Comparison of the steady-state EPSP amplitudes evoked at 0·2 Hz and 10 Hz in the presence and absence of kynurenate revealed that the antagonism was enhanced approximately 1·2-fold at the higher frequency.
Although kynurenate antagonized the response of dorsal horn neurones to L-glutamate, QUIS and KA, it did not inhibit the response of dorsal horn neurones to other excitatory transmitter candidates. We have shown previously that ATP excites a subpopulation of dorsal horn neurones by activating a membrane conductance similar to that evoked by L-glutamate (Jahr & Jessell, 1983). The reversal potential of both L-glutamate-(Mayer & Westbrook, 1984) and ATP-evoked responses on spinal neurones is near OmV. Superfusion of dorsal horn neurones with concentrations of kynurenate that were sufficient to antagonize both the DRG-evoked EPSP and the response to L-glutamate had no effect on the response of the same neurones to ATP (Jahr & Jessell, 1985).
SYNAPTIC TRANSMISSION AT la AFFERENT SYNAPSES IN VITRO
Examination of the chemosensitivity of DRG-dorsal horn neurone synapses in culture has provided strong evidence for the role of QUIS/KA receptors in mediating fast excitatory synaptic potentials. Our studies in culture, however, have not yet attempted to identify the subclasses of DRG and dorsal horn neurones that represent the pre-and postsynaptic elements of the synapses examined physiologically. Moreover, the use of dorsal horn neurones precludes any comparison of transmission at cutaneous and muscle afferent synapses. To assess the diversity of amino acid-like compounds as primary afferent transmitters, we have used in vitro preparations of new-born rat spinal cord (Otsuka & Konishi, 1974) to test whether a specific class of primary afferents, the group la afferents, releases L-glutamate, or a similar compound, as a transmitter mediating the fast la EPSP recorded from motoneurones (Jahr & Yoshioka, 1986). Two procedures were used to isolate the la EPSP from contaminating polysynaptic potentials. The sciatic nerve and its branches were dissected to the muscles which they innervated, permitting the selective stimulation of individual muscle nerves and smaller intramuscular nerve branches and thus dramatically restricting afferent input. This greatly decreased interneuronal activation and resulted in the partial isolation of the la EPSP. In addition, as in the culture experiments, high concentrations of divalent cations were added to the recording medium in order to increase spike threshold (Frankenhauser & Hodgkin, 1957; Jahr & Jessell, 1985) and block polysynaptic transmission.
Intracellular recordings were obtained from motoneurones with stable resting potentials between –52 and –83 mV. Motoneurones were identified by antidromic spike invasion evoked by stimulating the ventral roots, the sciatic nerve or its larger branches (tibial, common peroneal) or individual muscle nerves. In many cases, stimuli that were subthreshold for antidromic activation resulted in depolarizations of 0·5–1 mV which occurred with a latency of onset very close to or identical with that of the antidromic spike. These short-latency depolarizations probably resulted from antidromic activation of electrically coupled motoneurones (Fulton, Miledi & Takahashi, 1980).
The pattern of convergence of la afferent input to identified motoneurones in the new-born rat spinal cord was similar to that observed in the adult cat spinal cord (Eccles, Eccles & Lundberg, 1957). The largest la input was invariably from hornonymous muscle nerves while heteronymous muscle nerve stimulation produced smaller EPSPs or had no effect. Stimulation of antagonist muscle nerves elicited depolarizing postsynaptic potentials which reversed in polarity at membrane potentials only slightly more depolarized than the resting potential. In motoneurones in in vitro preparations of new born rat spinal cord, the Cl− reversal potential is slightly positive to the resting membrane potential and therefore I PSPs are depolarizing (Fulton et al. 1980; Konishi, 1982; Takashashi, 1984). Collectively, these observations indicate that the synaptic circuitry and membrane properties of neurones in the 4–10 day new-born rat spinal cord have developed sufficiently to permit analysis of the mechanisms underlying the la EPSP.
To test whether excitatory amino acids, or agonists of amino acid receptors, might be transmitters at la synapses, the effects of several compounds which antagonize the responses of motoneurones evoked by excitatory amino acids on the monosynaptic la EPSP were examined. In medium containing low levels of divalent cations, polysynaptic pathways remained intact (Figs 5, 6). The addition of 30–100/zmoll−1 APV suppressed later components of the EPSP (Fig. 6). In the presence of high (4mmoll−1 Ca2+, 8mmoll−1 Mg2+) concentrations of divalent cations, which blocked later components of the EPSP, the same concentration of APV had no effect on the residual EPSP (Jahr & Yoshioka, 1986).
Kynurenate, PDA and DGG were effective antagonists of the la EPSP (Fig. 6). Kynurenate was the most potent of these antagonists. Large reductions in the amplitude of the EPSP evoked by stimulation of the dorsal root or muscle nerve were produced by kynurenate at 250–500γmol l−1 and a complete block was produced at millimolar concentrations (Fig. 6). At these concentrations, kynurenate had no effect on motoneurone resting membrane potential or input resistance (Jahr & Yoshioka, 1986).
The site and mechanism of action of kynurenate was investigated by examining its effect on the depression of the la EPSP produced by repetitive stimulation of the hornonymous or heteronymous muscle nerves (Curtis & Eccles, 1960). When stimuli were paired at intervals from 40 ms to 1 s, a marked diminution of the second EPSP was observed. Decreasing the first EPSP to about 30% of its control amplitude by the addition of 0-5 mmol l−1 kynurenate resulted in a decrease in the second EPSP to a similar extent. In contrast, when the Ca2+/Mg2+ ratio was decreased, or Cd2+ or Co2+ was added in order to decrease presynaptic release of transmitter, the use dependent depression was decreased. These results parallel similar studies in culture (Jahr & Jessell, 1985) and provide evidence that kynurenate also antagonizes the la EPSP by a postsynaptic mechanism.
The specificity of action of kynurenate in neonatal rat spinal cord was determined by examining its effect on the responses of motoneurones to excitatory amino acids and non-amino acid excitants. In the presence of tetrodotoxin, kynurenate had no effect on motoneurone responses to carbachol at concentrations which greatly reduced the depolarization produced by L-glutamate (Fig. 7). The specificity of kynurenate was also tested by examining its action on recurrent inhibitory pathways. To enhance disynaptic activation, we superfused the new-born rat cord with 3 mmol l−1 Ca2+ and 0·5 mmol l−1 Mg2+. Under these ionic conditions, ventral root stimulation elicited a depolarization in motoneurones which reversed into a hyperpolarization when the resting membrane potential was decreased below – 60 mV by d.c. injection (Fig. 8). The reversal indicated that this potential was due to a conductance increase of an ion, probably CU, with an equilibrium potential just positive to the resting potential (Fulton et al. 1980; Konishi, 1982; Takahashi,1984).
In the cat spinal cord, ventral root stimulation results in the activation of recurrent axon collaterals of motoneurones which release acetylcholine onto inhibitory interneurones, the Renshaw cells (Eccles, Fatt & Koketsu, 1954; Curtis & Ryall, 1966). Renshaw cells, in turn, release an inhibitory neurotransmitter, probably glycine, which evokes an IPSP in motoneurones (Eccles et al. 1954). To establish that the inhibitory pathway in the new-born rat spinal cord was the analogue of the recurrent collateral circuit in the cat, we tested its sensitivity to the glycine antagonist, strychnine and to the nicotinic antagonist dihydro-β-erythroidine. The addition of strychnine greatly reduced the response to ventral root stimulation while having very little effect on the early depolarizing phase of the response to dorsal root stimulation. Strychnine did, however, suppress the late, inhibitory component of the response to dorsal root stimulation. Dihydro-β-erythroidine similarly attenuated the response to ventral root stimulation but had no effect on the response to dorsal root stimulation. In contrast to the effects of strychnine and dihydro-β-erythroidine, kynurenate had no effect on the recurrent IPSP while greatly reducing the la EPSP evoked by muscle nerve stimulation (Fig. 8) and the response to dorsal root stimulation. These results suggest that the response of motoneurones evoked by ventral root stimulation is analogous to the recurrent IPSP in the cat spinal cord (Renshaw, 1941, 1946; Jankowska & Roberts, 1972; Walmsley & Tracey, 1981) and is kynurenate resistant.
DISCUSSION
Diversity of primary sensory transmitters
Our studies with amino acid agonists and antagonists provide strong evidence that L-glutamate, or a compound with similar affinity for the synaptic receptor, is the transmitter mediating fast EPSPs at the majority of sensory neurone synapses examined in vitro. The experiments performed in cell culture, in particular, permit a discrimination between L-aspartate and L-glutamate as potential sensory transmitter candidates. The failure of NMDA and L-aspartate to excite dorsal horn neurones that received monosynaptic EPSPs and exhibited L-glutamate sensitivity provides evidence that L-glutamate, but not L-aspartate, mediates the fast afferent EPSPs. The inclusion of high Mg2+ concentrations in all studies on synaptic transmission in culture, however, leaves open the possibility that L-aspartate released from DRG neurones could contribute to sensory transmission via an action at synaptic NMDA receptors that were not detected under these recording conditions.
Biochemical studies demonstrating release of L-glutamate from the terminal regions of primary afferent fibres have provided additional evidence that L-glutamate is a transmitter released from sensory terminals (Roberts, 1974; Takeuchi et al. 1983). Few other endogenous compounds with L-glutamate-like excitatory properties have been identified in the CNS (Luini, Tal, Goldberg & Teichberg, 1984). The high density of L-glutamate receptor binding sites observed in the superficial dorsal horn of the spinal cord is also consistent with a physiological role for L-glutamate at sensory synapses (Greenamyre, Young & Penny, 1984).
From studies in culture, the proportion of DRG neurones that release L-glutamate-like compounds as primary afferent transmitters is not clear. It is possible that the maintenance of neurones in culture has in some way restricted analysis to a subpopulation of primary afferent synapses. Studies on new-born rat spinal cord preparations, however, demonstrate the ability of kynurenate to antagonize both la EPSPs (Jahr & Yoshioka, 1986) and cutaneous input (K. Yoshioka, unpublished) to dorsal horn neurones. Primary afferents conveying diverse sensory modalities therefore appear to release L-glutamate-like compounds as primary sensory transmitters in the mammalian spinal cord.
The amino acid antagonist sensitivity of afferent synaptic potentials elicited by activation of defined classes of sensory neurones in vivo may provide additional information about the universality of L-glutamate as a primary afferent transmitter in the spinal cord (Davies & Watkins, 1983). The broad-spectrum amino acid antagonist PDA has been reported to antagonize extracellularly recorded dorsal horn neuronal responses to noxious mechanical, but not to noxious thermal stimuli (Salt & Hill, 1983a,6; Peet, Leah & Curtis, 1983). Intracellular recording from dorsal horn neurones in isolated guinea pig spinal cord preparations has also indicated that some high-threshold, afferent-evoked synaptic potentials may be insensitive to amino acid antagonists (Schneider & Perl, 1985). These observations raise the possibility that some primary afferents may release non-amino acidfast excitatory transmitters.
From our studies on the chemosensitivity of spinal cord neurones to putative sensory transmitters, the only endogenous compounds examined that have postsynaptic actions consistent with fast excitatory transmitter roles are the acidic amino acids and the nucleotide ATP (Jahr & Jessell, 1983). While results in culture and in the new-born rat spinal cord preparation indicate that L-glutamate-like compounds are prevalent sensory transmitters, the possibility that ATP may be a sensory transmitter released from a small proportion of primary afferent fibres cannot be excluded. ATP selectively excites dorsal horn neurones in cat spinal cord in vivo (Salt & Hill, 1983a,b) in particular, those that receive low threshold C fibre input (Fyffe & Perl, 1984).
Several of the neuropeptides present within subsets of cutaneous sensory neurones may also act as transmitters at primary afferent synapses. Of the peptides localized in sensory ganglia, the role of substance P has been studied in greatest detail (Otsuka et al. 1982; Jessell, 1983). Intracellular recording from dorsal horn neurones in rat spinal cord slices has demonstrated that the slow, synaptically mediated depolarization that is elicited by high-frequency dorsal root stimulation can be mimicked by application of substance P and blocked by substance P antagonists (Konishi, Akagi, Yanagisawa & Otsuka, 1983; Urban & Randic, 1984). Substance P and other peptides may therefore be released as synaptic transmitters mediating slow synaptic potentials in the dorsal horn. The release of peptides from specific subclasses of afferent fibres is likely to have important, but as yet undefined, physiological roles in regulating the excitability of subsets of dorsal horn neurones that receive synaptic input mediated by fast sensory transmitters.
Mechanism and specificity of kynurenate antagonism
The use of kynurenate has been important in establishing that L-glutamate, or a compound interacting with L-glutamate receptors, is a transmitter at sensory neurone synapses examined in vitro. The blockade by kynurenate of L-glutamate sensitivity of spinal cord neurones at concentrations that also reduce or abolish the DRG-evoked monosynaptic EPSP is consistent with the idea that L-glutamate is a primary sensory transmitter. The enhanced synaptic depression that was sometimes observed in the presence of kynurenate may indicate that kynurenate has channel-blocking activity in addition to its action as a competitive antagonist at amino acid receptors. The reason for the variability in enhanced synaptic depression is not known. It is possible that distinct populations of dorsal horn neurones express different classes of amino acid receptors/channels and that these are antagonized by kynurenate by multiple mechanisms. Our results in culture and in the neonatal rat spinal cord, however, clearly indicate that blockade of synaptic transmission by kynurenate can occur independently of enhancement of synaptic depression or change in EPSP time course.
Several lines of evidence indicate that kynurenate antagonism is restricted to excitatory amino acid-mediated synaptic potentials. In cultures of dorsal horn neurones derived from embryonic rat spinal cord, kynurenate antagonizes the effect of excitatory amino acids without affecting the response to ATP, which depolarizes dorsal horn neurones by a mechanism similar to that of L-glutamate (Jahr & Jessell, 1983, 1985). The response of motoneurones to L-glutamate is blocked by concentrations of kynurenate that have no effect on the depolarization produced by carbachol. Kynurenate also has no effect on the depolarization of the ventral root evoked by carbachol, serotonin, noradrenaline, y-amino butyrate or substance P at concentrations that blocked responses to L-glutamate, L-aspartate, QUIS, KA and NMDA (C. E. Jahr & K. Yoshioka, unpublished).
The synaptic specificity of kynurenate was established by comparing its action on the la EPSP to that of the disynaptic recurrent inhibitory input to motoneurones. In the new-born rat, responses to glycine and the recurrent IPSP are depolarizing (Konishi, Saito & Otsuka, 1975; Evans, 1978; Fulton et al. 1980). Both strychnine and dihydro-β-erythroidine antagonized the response to ventral root stimulation indicating the existence of a recurrent inhibitory pathway similar to that in the cat. The lack of effect of kynurenate on recurrent I PSPs at concentrations that greatly suppress the la EPSP demonstrates: (i) selectivity of kynurenate antagonism at central excitatory and inhibitory synapses, (ii) selectivity between excitatory synapses in the spinal cord mediated by amino acid and non-amino acid neuro-transmitters and (iii) selectivity of kynurenate for postsynaptic amino acid receptors.
Postsynaptic receptors and transduction mechanisms at afferent synapses
The development of amino acid agonists and antagonists with enhanced specificity has made it possible to distinguish synaptic events that result from the activation of NMDA and KA/QA receptors. In our studies in vitro, the small degree of inhibition of monosynaptic DRG-dorsal horn and la afferent EPSPs produced by the selective NMDA antagonist APV indicates that amino acid receptors involved in the initial transduction of sensory information transmission are predominantly of the KA/QUIS class. The persistence of DRG-evoked EPSPs at Mg2+ concentrations that are likely to block the NMDA receptor/channel complex provides additional evidence for the involvement KA/QUIS receptors in afferent synaptic transmission. Since Mg2+ blockade of the ion channel associated with NMDA receptor activation is highly voltage dependent, analysis of the voltage sensitivity of synaptic currents at afferent synapses might provide an additional means of assessing the contribution of NMDA receptors to synaptic transmission. Finkel & Redman (1983) observed little voltage dependence in the la afferent motoneurone EPSP in cat spinal cord. This result is consistent with an interaction of the afferent transmitter with KA/QUIS rather than NMDA receptors. However, the voltage dependence of synaptic currents may not be a sensitive indicator of low levels of channel blockade by Mg2+ and it is difficult, therefore, to exclude a contribution of NMDA receptors to synaptic currents evoked by la afferents.
Activation of NMDA receptors may contribute to transmission at some interneuronal synapses in spinal cord. While APV has no effect on the la EPSP it specifically antagonizes later, presumably polysynaptic, components of the post synaptic response evoked by dorsal root or sciatic nerve stimulation in the new-born rat spinal cord. Kynurenate, PDA and DGG are antagonists of both the early and late components of the dorsal root-evoked EPSP. Whether the ability of these compounds to antagonize the late component of the EPSP is due to blockade of mono-or polysynaptic input to motoneurones is unclear. APV-sensitive synaptic potentials have been described in Xenopus embryo spinal cord (Dale & Roberts, 1985) although their origin is still not established. Spontaneous synaptic currents recorded at synapses between spinal interneurones and identified chick moto neurones in dissociated cell culture have been reported to exhibit some voltage sensitivity (O’Brien, 1985). Studies on spinal cord-spinal cord EPSPs in dissociated cell culture, however, have not revealed an APV-or voltage-sensitive component (Nelson, Pun & Westbrook, 1986). APV-sensitive EPSPs have also been reported at interneuronal synapses onto cerebral cortical and hippocampal neurones (Thornson, West & Lodge, 1985; Thornson, 1986; Hablitz & Langmoen, 1986). While there is therefore no conclusive evidence for the involvement of NMDA receptors in excitatory transmission at afferent primary synapses in the spinal cord, excitatory amino acid transmission at many interneuronal synapses in the CNS may include an NMDA-receptor-mediated component.
There is preliminary information about the kinetics and unitary conductance of channels associated with NMDA and KA/QUIS receptors on spinal and other CNS neurones. Nowack & Ascher (1984) have reported that NMDA-activated channels on mesencephalic neurones have conductances in the range of 50 pS whereas KA-and QUIS-activated channels have extremely small conductances. Consistent with its mixed agonist activity, in cultured cerebellar neurones L-glutamate itself appears to activate at least two distinct channels with discrete unitary conductances of less than 0·5 pS and 50pS (Cull-Candy & Ogden, 1985). In horizontal cells in goldfish retina, L-glutamate, QUIS and KA each induce a current with an elemental conductance of 2–3 pS (Ishida & Neyton, 1985).
Voltage-clamp and single-channel recordings from cultured spinal neurones have provided information about the changes in ion permeability that are associated with activation of NMDA and KA/QUIS receptors. Channels associated with both classes of receptors are permeable to Na+ and K+, consistent with the reported reversal potential for amino acid agonist-induced currents at or about 0 mV (Mayer & Westbrook, 1984). The divalent cation sensitivity of the NMDA receptor/channel complex has been examined by several groups. Single-channel analysis on excised patches of CNS neurones has revealed that channel blockade in the presence of extracellular Mg24-, Co2+, Ni2+ and Mn2+ is associated with rapid closures of the NMDA channel (Nowack & Ascher, 1985). The time that channels remain in the closed state is increased with hyperpolarization and with increased extracellular divalent ion concentration. The modification of single-channel kinetics by these ions is not associated with any change in the single-channel conductance. In contrast, Ca2+ and Ba2+, at high concentrations, do not affect channel kinetics but decrease single-channel conductance (Nowack & Ascher, 1985). The reversal potential of NMDA-induced currents is sensitive to extracellular Ca2+ concentration (Mayer & Westbrook, 1985) suggesting that the NMDA channel is permeable to Ca2+. Recent experiments with the calcium indicator Arsenazo III have provided more direct evidence that the inward currents activated by NMDA in voltage-clamped spinal neurones are accompanied by calcium influx and an increase in intracellular Ca2+ (MacDermott et al. 1986). KA appears to be considerably less effective than NMDA in inducing intracellular calcium transients in spinal neurones (MacDermott et al.1986).
Amino acids have also been reported to stimulate inositol phosphate turnover in cultured CNS neurones (Sladeczek et al. 1985; Nicoletti, ladarola, Wrobleski & Costa, 1985). The generation of intracellular inositol tris-and other polyphosphates is likely to result in the mobilization of intracellular calcium stores (Berridge & Irvine, 1984). The potency of amino acid agonists in triggering inositol phosphate turnover, QUIS > L-glutamate S> NMDA = KA (Sladeczek et al. 1985), differs from their ability to induce calcium influx (MacDermott et al. 1986). The activation of amino acid receptors on spinal neurones could therefore lead to increases in intracellular calcium by distinct mechanisms: (i) the depolarization-induced activation of voltage-dependent calcium channels; (ii) changing the calcium permeability of NMDA channels and (iii) an inositol trisphosphate-mediated mobilization of intracellular calcium stores.
Substance P and other peptides released by primary sensory neurones are also known to enhance inositol phosphate turnover (Mantyh et al. 1984) and to increase intracellular calcium (Womack, Hanley & Jessell, 1985). Activation of amino acid and peptide receptors on spinal neurones may therefore result in common intracellular signals. A convergence of these signalling systems could represent one mechanism by which interactions between sensory transmitters integrate sensory information at primary afferent synapses in the spinal cord.
ACKNOWLEDGEMENTS
This work was supported by grants from NINCDS, The McKnight Foundation and The Muscular Dystrophy Association. We are grateful to A. MacDermott for critical comments and to C. L. Miller for preparation of the manuscript.