N-Methyl-D-Aspartate (NMDA) and Non-Nmda (Metabotropic) Type Glutamate Receptors Modulate the Membrane Potential of the Schwann Cell of the Squid Giant Nerve Fibre

The amino acid glutamate plays an important role in intercellular signalling in the nervous system. It functions as the major excitatory neurotransmitter in the vertebrate central nervous system (Monaghan et al. 1989) and is the leading candidate for the excitatory neuromuscular transmitter in a range of invertebrates (Duce, 1988). However, evidence is also increasing that glutamate may play a key role in neuronal-glial signalling in both vertebrate and invertebrate nervous systems (see Barres, 1989). It has been known for some time that glutamate is able to depolarize cultured astrocytes (Bowman and Kimelberg, 1984; Kettenmann and Schachner, 1985), but more recently there has been a discussion as to whether this effect is due to the activation of an electrogenic glutamate uptake mechanism or to a direct activation of glutamate receptors (Brew and Attwell, 1987; Barbour et al. 1988; Schwartz and Tachibana, 1990; Sarantis and Attwell, 1990; Wyllie et al. 1991).

It has also been suggested that glutamate receptor activation underlies the effects of glutamate in a number of invertebrate glial preparations, including those of the adaxonal Schwann cells of the squid giant axon (Villegas, 1978a,b,Lieberman et al. 1989; Evans et al. 1991a) and those of the neuropile glia of the leech segmental ganglion (Ballanyi ef al. 1989; Dörner et al. 1990). In the majority of the above studies the glutamate receptors present on the glial cells have been demonstrated to be of the quisqualate/kainate (non-NMDA) subtype. However, in the adaxonal Schwann cells of the giant axon of the tropical squid Sepioteuthis sepioidea evidence has been presented that both NMDA- and non-NMDA-subtype glutamate receptors are present (Evans et al. 1991a).

In this paper we report further details of the pharmacological properties of the three different subtypes of glutamate receptor found on the Schwann cells of the tropical squid and discuss their possible functional roles.

Giant nerve fibres with a diameter of 300 – 400 μm were dissected in sea water from the hindmost stellar nerve of the squid Sepioteuthis sepioidea. Giant axons with their surrounding Schwann cell sheaths were then isolated and cleaned of adhering bundles of small nerve fibres by dissection in artificial sea water (see below).

Electrophysiological techniques were as described previously and involved the successive measurements of electrical potentials of a series of Schwann cells by brief impalements from inside the axon (Villegas, 1972, 1973, 1975; and see Lieberman et al. 1989). All experiments were carried out at room temperature (20 –22 °C). In experiments where the degree of hyperpolarization of the Schwann cell membrane by a given drug has been quantified, it was calculated as the difference between the mean measurement of the control resting potential in the 2 min period immediately before drug application (averaged over 5 –10 penetrations of individual Schwann cells) and the mean measurement of membrane potential in the 2 min period immediately after the end of the drug pulse (averaged over 5 –10 penetrations of individual Schwann cells). The results are given as the mean hyperpolarization induced by a drug (in mV ±S.E.) of the results obtained from at least four axons A blind protocol was used in which the investigator sampling the Schwann cell membrane potentials did not know the identity of the test pulses being applied to the preparation. All the experiments shown were repeated at least three times in the format shown.

Drugs superfused over the surface of the preparation were dissolved in artificial sea water containing 442 mmol l ⁻¹ NaCl, 10 mmol l⁻¹ KCl, 11 mmol l^{−1 1} CaCl₂, 45 mmol l⁻¹ MgCl₂ and 10 mmol l⁻¹ Tris-Cl buffer (pH 8·0). All the superfused solutions were continuously bubbled with a mixture of 95 % O₂ and 5 % CO₂.

All drugs were obtained from the Sigma Chemical Co., except for glutamylglutamate which was obtained from Peninsula Laboratories Inc., and (IS,3R)-l-aminocyclopentane-1,3-dicarboxylic acid (trans-ACPD), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), cis-(dicarboxyl)-2,4-methanoglutamic acid (cis-DMG) and L(+)-2-amino-3-phosphonopropionic acid (L-AP3), which were obtained from Tocris Neuramin.

The membrane potential of the Schwann cell of the giant axon of the tropical squid Sepioteuthis sepioidea can produce three different responses to the application of pulses of L-glutamate (Villegas, 1981), which appear to be mediated by three pharmacologically distinct classes of receptor (Evans et al. 1991a). They include a rapidly induced and long-lasting hyperpolarization of the membrane potential of the Schwann cell in response to a short 1 min pulse of L-glutamate, a biphasic response consisting of a rapid hyperpolarization followed by a slow depolarization to a long, 10 min, pulse and a rapid depolarization in response to a short pulse of glutamate following the induction of the slow depolarization.

The rapidly induced, long-lasting hyperpolarization has previously been shown to be due to the glutamate-induced release of acetylcholine from the Schwann cells themselves; this feeds back onto nicotinic cholinergic receptors on the Schwann cell to promote the hyperpolarization (Villegas, 1972, 1973, 1975; Lieberman et al. 1989). This response was more sensitive to (IS,3R)-1 -aminocyclopentane-1,3-dicarboxylic acid (trans-ACPD), the selective agonist for metabotropic glutamate receptors (Sugiyama et al. 1987, 1989), than to the other agonists tested. It had a threshold between 10⁻¹¹ and 2× 10⁻¹¹ mol l⁻¹ and a half-maximal response in the region of 10⁻¹¹mol l⁻¹ (Fig. 1A,B). In contrast, the amino acid L-glutamate itself had a threshold between 2×10⁻¹⁰ and 5× 10⁻¹⁰mol l⁻¹ and a half-maximal response in the region of 5×10⁻¹⁰ mol l⁻¹ (Fig. 1A). However, the reduced effectiveness of L-glutamate could be due to the activity of an active uptake system for this amino acid in this preparation (see Evans et al. 1992). Dose-response curves revealed that the response was more sensitive to quisqualate and kainate than to N-methyl-D-aspartate (NMDA). This indicated that the response was likely to be mediated by a metabotropic-type glutamate receptor rather than a quisqualate/kainate-type glutamate receptor as previously suggested (Lieberman et al. 1989; Evans et al. 1991a). In contrast to previous observations in another species of squid, Alloteuthis subulata (Lieberman et al. 1989), L-aspartate was not found to be a very potent agonist of this response in Sepioteuthis sepioidea. A 1 min pulse of 10⁻⁵ mol l⁻¹ L-aspartate gave a hyperpolarization of only 5·8±0.4mV (N-3). The response was stereoselective for the L form of glutamate, as the D form gave no response at concentrations up to 10⁻⁷mol l⁻¹ (Fig. 1A), agreeing with previous work (Villegas, 1981).

The conclusion that the rapid hyperpolarizing response to short pulses (1 min) of L-glutamate was mediated via a non-NMDA subtype of glutamate receptor was supported by pharmacological blocking experiments. Fig. 2A shows that the rapid hyperpolarizations of the Schwann cell membrane induced either by applying a 1 min pulse of 10⁻⁸ mol l⁻¹ L-glutamate or by firing the giant axon at 100 Hz for 1 min were both blocked by D,L-2-amino-4-phosphonobutyric acid (APB or AP4) (10⁻⁶ mol l⁻¹). APB is a potent blocking agent of non-NMDA-type glutamate receptors in squid Schwann cells (Lieberman et al. 1989; Evans et al. 1991a) and in several cephalopod muscles (Bone and Howarth, 1980; Florey et al. 1985), whereas it acts as an agonist at a specific subtype of glutamate receptor in vertebrates. The hyperpolarizing responses in the squid Schwann cell were not blocked by D,L-2-amino-5-phosphonovaleric acid (APV or AP5) (10⁻⁶ mol l⁻¹), a selective NMDA receptor antagonist (Fig. 2A). Another antagonist of non-NMDA-type glutamate receptors, L-glutamic acid diethyl ester (GDEE) did not block either hyperpolarizing effect when used at a concentration of 10⁻⁶ mol l⁻¹, but when used at 10⁻⁴ mol l⁻¹ it produced a partial block of the effect of a 1 min pulse of 10⁻⁸ mol⁻¹ L-glutamate and a complete block of the rapid hyperpolarization induced by firing the giant axon at 100 Hz for 1 min (Fig. 2B). The non-NMDA nature of this subtype of glutamate receptor is further supported by experiments in which 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) (10 mol l⁻¹) blocked the hyperpolarizing effects both of a 1 min pulse of 10⁻⁸ mol l⁻¹ L-glutamate and of firing the giant axon at 100 Hz for 1 min (Fig. 2C). The suggestion that the rapid hyperpolarization induced by short pulses of L-glutamate is mediated by the activation of a metabotropic-like glutamate receptor subtype is further confirmed by the finding that the response was blocked by L(+)-2-amino-3-phosphonopropionic acid (L-AP3) (Fig. 2D), a potent and selective antagonist of the metabotropic glutamate receptor in vertebrate tissues (Irving et al. 1990; Schoepp et al. 1990).

The specificity of the blocking agents for the rapid hyperpolarizations induced by glutamate and by firing the giant axon at 100 Hz for 1 min was tested by comparing their capacity to block the actions of a range of neurotransmitter substances previously described as being capable of modulating the membrane potential of the Schwann cell of the squid giant axon (see Evans et al. 1986, 1990; Reale et al. 1986; Villegas, 1981, 1984). One minute pulses of 10⁻⁶mol l⁻¹ APV, 10⁻⁶mol l⁻¹ APB, 10⁻⁴mol l⁻¹ GDEE or 10⁻⁶ mol l⁻¹ CNQX did not reduce the size of the Schwann cell hyperpolarization induced by 1 min pulses of 10⁻⁷mol l⁻¹ carbachol, 10⁻⁶mol l⁻¹ D,L-octopamine, 10⁻⁹mol l⁻¹ vasoactive intestinal peptide (VIP) or 10⁻⁹mol l⁻¹ substance P (data not shown). Thus, the blocking agents used are specific for the receptors mediating the effects of glutamate in this preparation.

The biphasic response of the Schwann cell membrane potential to a prolonged pulse of L-glutamate can be separated into an initial rapid hyperpolarization, which has the same pharmacological characteristics as the rapid hyperpolarization, induced by a 1 min pulse of L-glutamate (see Evans et al. 1991a), and a slower depolarization. The slower depolarization had a threshold between 10⁻⁹ and 10⁻⁸ mol l⁻¹ L-glutamate (Fig. 3A) and after its induction responses to short (1 min) pulses of L-glutamate became depolarizing. Previous pharmacological studies indicate that the slow depolarization is mediated by NMDA-type glutamate receptors (Evans et al. 1991a). The response was blocked by APV (10⁻⁶ mol l⁻¹), but not by APB (10⁻⁶mol l⁻¹), L-AP3 (10⁻¹ mol l⁻¹) (Fig. 2D) and CNQX (10⁻⁶ mol l⁻¹) (Fig. 3B), and it is mimicked by the application of NMDA with a threshold occurring between 5×10⁻¹⁰ and 10⁻⁹ mol l⁻¹ (Evans et al. 1991a). It is also mimicked by the application of cis-(dicarboxyl)-2,4-methanoglutamic acid (cis-DMG) (Fig. 3C) (10⁻⁸ mol l⁻¹), one of the most potent and selective synthetic agonists of NMDA receptors (Lanthorn et al. 1990; Allan et al. 1990), which is itself only a poor agonist (threshold between 10⁻⁸ and 10⁻⁷ mol l⁻¹) of the rapid hyperpolarizations induced by glutamate in the squid Schwann cell preparation. It was not initiated by the application of 10 min pulses of quisqualate, kainate (Evans et al. 1991a) or L-aspartate up to a concentration of 10⁻⁵ mol l⁻¹ or by trans-ACPD up to a concentration of 1 10⁻⁷ mol l⁻¹ (Fig. 3D).

We have previously shown that the fast depolarization induced by a short (1 min) pulse of L-glutamate given after a prolonged 10 min pulse of L-glutamate has a threshold between 10⁻⁹ and 10⁻⁸ mol l⁻¹, and that such fast depolarizations can also be initiated by 1 min pulses of L-glutamate after a slow depolarization induced by NMDA at concentrations of 10⁻⁹mol l⁻¹ and above (Evans et al. 1991a). Specificity studies (Fig. 4A) indicated that this rapid depolarization could also be induced by 1 min pulses of 10⁻⁵mol l⁻¹ NMDA, as well as similar 1 min pulses of L-glutamate but not by 1 min pulses of quisqualate or kainate up to a concentration of 10⁻⁵ mol l⁻¹ or by 1 min pulses of trans-ACPD up to a concentration of 10⁻⁷mol l⁻¹ (Fig. 3D). Interestingly a 20 min pulse of 2×10⁻⁹mol l⁻¹ L-glutamate (Fig. 4B) did not induce a slow depolarization of the Schwann cell membrane above the resting membrane potential. It only initiated a transient hyperpolarization (cf. prolonged hyperpolarization induced by a 20 min pulse of 10⁻⁹mol l⁻¹ L-glutamate in Fig. 3A) and the Schwann cell membrane potential returned to its resting value after 5 min. Nevertheless, short 1 min pulses of NMDA given after this glutamate pulse still induced a rapid depolarizing response with the same threshold (between 10⁻¹⁰ and 10⁻⁹mol l⁻¹), which is almost three orders of magnitude lower than that needed to produce a hyperpolarizing response before the application of the prolonged glutamate pulse. This indicated that a prolonged slow depolarization of the Schwann cells above their resting potential was not a necessary prerequisite for inducing a rapid depolarizing response. cis-DMG is also a potent activator of this class of NMDA receptor on the squid Schwann cell, with a threshold between 10⁻¹⁰ and 10⁻⁹ mol l⁻¹ (Fig. 3C).

The receptors mediating the rapid depolarizing response appear to be pharmacologically distinct from the receptors mediating the other responses to glutamate in this preparation. Although the response could be induced by NMDA, it did not appear to be blocked by 10⁻⁶mol l⁻¹ APV (Fig. 5A), 10⁻⁶mol l⁻¹ L-AP3 (Fig. 2D) or 10⁻⁶mol l⁻¹ CNQX (Fig. 3B) and was only slightly reduced by 10⁻⁶mol l⁻¹ APB, a concentration that completely blocked the hyperpolarizing responses induced by a 1 min pulse of L-glutamate (Fig. 5B). However, if the 1 min pulse of 10⁻⁸mol l⁻¹ L-glutamate was given in the presence of 10⁻⁶mol l⁻¹ GDEE, the rapid depolarizing response was irreversibly blocked and could not be induced by a subsequent 1 min pulse of L-glutamate given alone (Fig. 5C). Further, if the 20 min long pulse of L-glutamate was given in the presence of 10⁻⁶mol l⁻¹ GDEE, although the initial hyperpolarization and slow depolarization appeared to be normal, no rapid depolarizing responses could be initiated by subsequent 1 min pulses of L-glutamate given at the end of the long glutamate pulse (Fig. 5D). Instead, there was initially no response to the short glutamate pulses and after a delay the normal hyperpolarizing responses returned.

In a number of instances it has been suggested that endogenous dipeptides containing the excitatory amino acids L-glutamate and L-aspartate act directly at the same sites as L-glutamate in vertebrate nervous tissue (see Zaczek et al. 1983; Westbrook et al. 1986). However, it has been pointed out that many of the actions of iV-acetylaspartylglutamate may be produced by a rapid extracellular enzymatic cleavage to release neuroactive L-glutamate (see Moffett et al. 1990) or by contamination of synthetic N- acetylaspartylglutamate with potassium ions (Whittemore and Koerner, 1989). In the present study we have examined the ability of a range of dipeptides containing these amino acids to mimic the various effects of L-glutamate on the membrane potential of the Schwann cell of the squid giant axon.

A survey of the hyperpolarizing actions of a range of dipeptides tested as 1 min pulses at a concentration of 10⁻⁸ mol l⁻¹ on the membrane potential of the Schwann cell is shown in Table 1. The most effective dipeptide tested was N-acetylaspartylglutamate, a naturally occurring dipeptide in vertebrate nervous tissue. The response of the Schwann cell to 1 min pulses of N-acetylaspartylglutamate was dose-responsive (Fig. 6). N- Acetylaspartylglutamate was more potent at inducing this response than L-glutamate itself (Figs 6, 7A,B). The other dipeptides tested, including β-aspartylglycine, an endogenous molluscan dipeptide (McCaman and Stetzler, 1985), produced smaller rapid hyperpolarizations of the Schwann cell membrane.

N-Acetylaspartylglutamate was also able to mimic the effects of a long pulse of L-glutamate in that concentrations of 10⁻⁸ mol l⁻¹ and above produced a slow depolarization after the initial rapid hyperpolarization (Fig. 7B). Furthermore, short pulses of N- acetylaspartylglutamate were also capable of initiating a rapid depolarization of the Schwann cell membrane after the induction of a slow depolarization by a prolonged pulse of (V-acetylaspartylglutamate (Fig. 7B) or L-glutamate (10⁻⁸mol l⁻¹) (Fig. 6B). This effect has a threshold between 10⁻¹¹ and 10⁻¹⁰mol l⁻¹N-acetylaspartylglutamate (Fig. 6B). None of the other dipeptides tested was capable of inducing the rapid depolarization up to a concentration of 10⁻⁸ mol l⁻¹ (Fig. 7B).

Pharmacological parallels with glutamate receptor subtypes in many vertebrate preparations (Foster and Fagg, 1984; Watkins and Olverman, 1987; Monaghan et al. 1989) indicate that the three types of glutamate response observed in the adaxonal Schwann cells of the giant axon of the tropical squid are likely to be mediated via the activation of three distinct subtypes of glutamate receptor rather than by the activation of an electrogenic glutamate uptake system. However, the pharmacological properties of the squid Schwann cell glutamate receptors do not correspond exactly with those of any of the different subclasses of vertebrate glutamate receptor currently identified. The glutamate receptors responsible for the release of acetylcholine from the Schwann cells and for the subsequent nicotinic cholinergic receptor-induced long-lasting hyperpolarization (Villegas, 1972, 1973, 1975; Lieberman et al. 1989) appear to be of the non-NMDA variety since they are activated by trans-ACPD, L-glutamate, quisqualate and kainate in preference to NMDA. This is supported by the fact that they are blocked by the non-NMDA antagonist CNQX and by high doses of the general non-NMDA antagonist GDEE (10⁻⁴ mol l⁻¹) (present experiments and Lieberman et al. 1989) but not by the specific NMDA receptor antagonist APV (10⁻⁶ mol l⁻¹). They appear to express some similarities with the metabotropic glutamate receptors (Sugiyama et al. 1987, 1988) since they are preferentially activated by trans-ACPD, blocked by L-AP3 (Irving et al. 1990; Schoepp et al. 1990) and activated by quisqualate but not by a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) (Lieberman et al. 1989). The Schwann cell metabotropic glutamate receptors appear to be closely related to the vertebrate metabotropic glutamate receptor encoded by the mGluR2 cDNA clone since they are activated better by trans-ACPD and glutamate than by quisqualate (see Tanabe et al. 1992). The receptors also express some differences since they are activated by kainate but not by ibotenate (Lieberman et al. 1989) and they are blocked by CNQX. An additional similarity is that, like some of the metabotropic receptors, they are blocked by APB (or AP4), which conventionally acts like an agonist of some vertebrate glutamate receptor subtypes (Monaghan et al. 1989) whilst at other glutamate receptors it has not been determined whether it acts as an agonist or an antagonist (Watkins, 1988). APB also blocks the actions of non-NMDA-type glutamate receptors in several cephalopod muscle preparations (Bone and Howarth, 1980; Florey et al. 1985). Furthermore, it has been postulated that this class of squid Schwann cell glutamate receptor may be involved in increasing intracellular Ca²⁺ levels to mediate the release of acetylcholine (see Evans et al. 1991b), but it is not known whether these receptors activate an inositol-trisphosphate-mediated release of intracellular Ca²⁺ or an increase in Ca²⁺ permeability of the cell membrane. The lack of involvement of an ion-channel-mediated event in the activation of these receptors is suggested in the tropical squid preparation, since no change in membrane potential can be observed in response to the application of short pulses of glutamate after blockade of the cholinergic receptors with a-bungarotoxin (P. D. Evans, V. Reale, R. M. Merzon and J. Villegas, in preparation). However, in another species of squid, Alloteuthis subulata, blockade of the cholinergic receptors with tubocurare reveals the presence of a glutamate-induced depolarization (Lieberman el al. 1989). Another difference between the glutamate-receptor-mediated effects in these two species of squid is that aspartate was a much more powerful agonist of these receptors in Alloteuthis than in Sepioteuthis. Thus, the receptors mediating this effect on the adaxonal glial cells of the squid may exhibit species differences.

We have previously shown that the slow depolarization of squid Schwann cells induced by long pulses of L-glutamate is likely to be mediated via the activation of NMDA-type glutamate receptors (Evans et al. 1991a). This is the first demonstration of NMDA-type glutamate receptors in any glial cell preparation, although Xenopus laevis oocytes injected with mRNA from an astrocytoma cell line (R-lll) have recently been shown to acquire a small number of NMDA receptors (Matute et al. 1992). This glutamate effect in the squid Schwann cells is blocked specifically by APV (10⁻⁶mol l⁻¹) but not by APB (10⁻⁶mol l⁻¹) (Evans et al. 1991a), GDEE (10⁻⁶mol l⁻¹), L-AP3 (10⁻⁶mol l⁻¹) or CNQX (10⁻⁶mol l⁻¹) (present study). The present study shows that the L-glutamate-induced slow depolarization has a threshold between 10⁻⁹ and 10⁻⁸ mol l⁻¹, whilst NMDA has a lower threshold (between 5×10⁻¹⁰ and 10⁻⁹mol l⁻¹) for the induction of this effect (Evans et al. 1991a). The effect is also mimicked by cis-DMG (10⁻⁸mol l⁻¹), a potent agonist of vertebrate NMDA receptors (Lanthorn et al. 1990; Allan et al. 1990). This receptor in the squid is not activated by quisqualate, kainate (Evans et al. 1991a) or L-aspartate up to a concentration of 10⁻⁵ mol l⁻¹ or by trans-ACPD up to a concentration of 10⁻⁷mol l⁻¹. Since Ca²⁺ is needed to initiate and maintain this slow depolarization (Villegas, 1978a,b; Evans et al. 1991a), as with NMDA receptors in neurones (Malenka et al. 1989), the NMDA receptors on the squid Schwann cell also mediate an increase in calcium permeability. It has been postulated that NMDA receptors in vertebrate preparations can only be activated in the presence of low doses of glycine, which acts as a co-agonist (see Thomson, 1989). The squid Schwann cell NMDA receptors may differ from classical vertebrate NMDA receptors in this respect, since glycine is unlikely to accumulate to significant levels in the superfused preparation. Nevertheless, it is not possible at present to rule out an accumulation of glycine in the small spaces between the Schwann cells and the giant axon. However, a vertebrate glycine-insensitive NMDA receptor has recently been found to be expressed in Xenopus oocytes after the injection of guinea pig cerebellar mRNA (Sekiguchi et al. 1990). Attempts to test directly whether the squid NMDA receptor can be modulated by glycine have been complicated because in some preparations glycine application alone can initiate a very variable change in the membrane potential of the Schwann cell (J. Villegas and R. M. Merzon, unpublished obervations). Another apparent difference between the squid and vertebrate NMDA receptors is that the squid receptor can be activated in artificial sea water containing 53 mmol l⁻¹ Mg²⁺ whereas the vertebrate receptor is blocked in a voltage-sensitive way by 1 mmol l⁻¹ Mg²⁺. This may represent an adaptation of the squid NMDA receptor to a marine environment.

The fast depolarizing responses to L-glutamate induced after the properties of the Schwann cells have been modified by the activation of the NMDA receptors appear to be mediated by a second class of NMDA receptor. They are specifically activated by NMDA with a threshold between 10⁻¹⁰ and 10⁻⁹mol l⁻¹ in contrast to L-glutamate, which had a threshold between 10⁻⁹ and 10⁻⁸ mol l⁻¹ (Evans et al. 1991a), and they are not activated by quisqualate or kainate up to 10⁻⁵mol l⁻¹ or by trans-ACPD up to 10⁻⁷mol l⁻¹. cis-DMG is also a potent activator of these receptors with a threshold between 10⁻¹⁰ and 10⁻⁹mol l⁻¹. However, the properties of these receptors differ from those of the NMDA receptors mediating the slow depolarization since they are not blocked by APV or CNQX, but they are blocked after exposure of the preparation to 10⁻⁶ mol l⁻¹ GDEE. They were also partially blocked by APB. A rapid activation of NMDA receptors has also been described at a variety of vertebrate central synapses (Mody and Heinemann, 1987; D’Angelo et al. 1990). At present the mechanism that enables this second set of NMDA receptors to become activated and inhibits the activation of the metabotropic glutamate receptors for a prolonged period after the stimulation of the NMDA receptors mediating the slow depolarization is not known, but it seems likely to be related to a prolonged elevation of Ca²⁺ concentration in the Schwann cells.

A novel finding of the present investigation is the fact that the dipeptide N- acetylaspartylglutamate is a potent agonist of all three glutamate receptor types found on the squid Schwann cells. This peptide has previously been shown to be localised selectively in various regions of the vertebrate central nervous system and to be capable of activating only a portion of the glutamate receptors present (Zaczek et al. 1983; Moffett et al. 1990). Another dipeptide, β-aspartylglycine, also showed a significant ability to activate the glutamate receptors mediating the release of acetylcholine from the Schwann cells but, as with the other dipeptides tested up to 10⁻⁸mol l⁻¹, it did not activate the receptors responsible for the induction of the rapid depolarization after the induction of a slow depolarization by L-glutamate. β-Aspartylglycine was originally isolated as a dipeptide from the nervous system of the gastropod mollusc Aplysia califomica (McCaman and Stetzler, 1984, 1985) but its function remains unknown. However, as mentioned above, many of the effects of N-acetylaspartylglutamate have been suggested to be brought about either by a contamination of the synthetic peptide with high levels of potassium ions (Whittemore and Koerner, 1989) or by a cleavage of the dipeptide by the enzyme N-acetylated a-linked acidic dipeptidase to release L-glutamate as the active factor (see Moffett et al. 1990). In the present experiments, contamination with potassium ions resulting in a depolarization of the Schwann cell membrane could perhaps account for the ability of this peptide to mimic the fast and slow depolarizing effects mediated by the NMDA receptors but seems unlikely to be able to account for its ability also to mimic the activation of the metabotropic glutamate receptors. In addition, although L-glutamate could be enzymatically released from several of the peptides used in the present study, it could not be produced from β-aspartylglycine and aspartylaspartate and any aspartate released from the latter two peptides would seem unlikely to be responsible for their actions owing to the very low sensitivity of this preparation to applied L-aspartate. Further studies on the specificity and activities of any dipeptidases present in the Schwann cell system of the squid are required to resolve the question of whether these dipeptides containing acidic amino acids are acting directly on receptors in this preparation.

The non-NMDA glutamate receptors mediating the release of acetylcholine from the Schwann cells are extremely sensitive to L-glutamate (threshold between 2×10⁻¹⁰ and 5×10⁻¹⁰mol l⁻¹) and are likely to be activated by glutamate released non-synaptically from the squid giant axon by stimulation at high frequencies (Lieberman et al. 1989; Evans et al. 1991a). The activation of these receptors turns on a complex multistep process that amplifies the original signal and leads to an increase in the permeability of the Schwann cell membrane to potassium ions at times when the squid is likely to be using its giant-fibre-mediated escape response pathway (Villegas, 1981; Villegas et al. 1988; Evans et al. 1991 b). One possible function for this system would be to buffer any accumulation of potassium ions in the adaxonal space that might interfere with the conduction of nerve impulses along the giant axons. The NMDA-type glutamate receptors responsible for the induction of the slow depolarization and the subsequent rapid depolarizing responses have higher thresholds for activation by L-glutamate (between 10⁻⁹ and 10⁻⁸ mmol l⁻¹) and are thus only likely to be activated in vivo under extreme conditions of prolonged activity of the giant axon at high frequency. The slow depolarization above the resting membrane potential is not, however, a prerequisite for the induction of the change in responsiveness. Thus, the Schwann cells of the squid giant axon possess NMDA receptors that can lead to long-term changes in responsiveness as do many other neuronal systems (Davies et al. 1989; Kauer et al. 1988a,b; Malenka et al. 1989; Watkins and Collingridge, 1989).

We thank the British Council for Academic Travel Grants, which made possible the visits of P.D.E. and V.R. to Venezuela. We also thank the Royal Society for a grant from the Marshall and Orr Bequest towards the cost of consumables for this project.