The best evidence for glutamate being a transmitter in cephalopods comes from pharmacological, immunohistochemical and molecular investigations on the giant fibre system in the squid stellate ganglion. These studies confirm there are AMPA/kainate-like receptors on the third-order giant axon. In the (glial) Schwann cells associated with the giant axons both classes of glutamate receptor occur.

Glutamate is an excitatory transmitter in the chromatophores and in certain somatic muscles and its action is mediated primarily via AMPA/kainate-like receptors, but at some chromatophores there are NMDA-like receptors.

In the statocysts the afferent crista fibres are also glutamatergic, acting at non-NMDA receptors.

In the brain (of Sepia) a neuronal NOS is activated by glutamate with subsequent production of nitric oxide and elevation of cGMP levels. This signal transduction pathway is blocked by D-AP-5, a specific antagonist of the NMDA receptor.

Recently immunohistochemical analysis has demonstrated (in Sepia and Octopus) the presence of NMDAR2A /B – like receptors in motor centres, in the visual and olfactory systems and in the learning system. Physiological experiments have shown that glutamatergic transmission is involved in long term potentation (LTP) in the vertical lobe of Octopus, a brain area involved in learning. This effect seems to be mediated by non-NMDA receptors. Finally in the CNS of Sepia NMDA-mediated nitration of tyrosine residues of cytoskeletal protein such as α-tubulin, has been demonstrated.

Glutamate (hereafter L-glu) is now known to be a major excitatory neurotransmitter in the CNS of vertebrates and occurs widely throughout the invertebrates [48]. It has long been accepted as the excitatory transmitter at the nervemuscle junction of insects and crustaceans.

Over the last decade or so it has become clear that L-glu is an important transmitter in both gastropods and cephalopods In such gastropods as Aplysia and Helix there is ample evidence, not only that L-glu is an excitatory transmitter but that it can act at AMPA/kainate or NMDA receptors, and, more importantly in the light of mammalian studies, that the latter receptors in Aplysia are implicated in long-term potentiation (LTP) of the gill-siphon withdrawal reflex [3, 4, 23, 34]. It is not clear, however, whether the NMDA and AMPA receptors are co-expressed at the same site or interact by shaping postsynaptic electrical activity to regulate Ca²⁺ influx differentially [3].

In cephalopods, too, there is good evidence that L-glu is an important excitatory transmitter, usually acting at AMPA/ kainate-like receptors. Experiments with loliginid squids (Loligo Alloteuthis, Lolliguncula), the cuttlefish (Sepia) and Octopus have revealed several glutamatergic systems. These include the squid giant synapse, the neuro-muscular junction of certain somatic muscles, the chromatophores, the statocysts, the ink-gland and the central nervous system. It will be convenient to consider these separately.

Probably the best evidence for L-glu being a transmitter in cephalopods derives from a series of studies on the stellate ganglion of the squids Loligo and Alloteuthis. Here a single second-order giant fibre from the brain branches to make synaptic contact with 10-12 third-order fibres originating in the ganglion (Fig. 1). The most medial third-order fibre is the largest, and the synapse onto this is the so-called “giant synapse”, the largest synapse in the animal kingdom. De Santis & Messenger [19] bath-applied L-glu and a series of glutamate agonist and antagonists to the ganglion while recording intracellularly from the third-order fibre; they found that Lglu and its AMPA/kainate agonists reversibly blocked transmission, presumably by de-sensitising the synapse to the endogenous transmitter. CNQX and DNQX also blocked transmission reversibly, though no NMDA antagonists had an effect. Iontophoretic application of L-glu, quisqualate, kainate and AMPA causes weak depolarisation of the postsynaptic membrane [40] but L-glu released at the synapse from “caged” glutamate (by flash photolysis) depolarises the membrane sufficiently for action potentials to be generated in the third-order fibre (Fig. 2), demonstrating directly that L-glu mimics the effect of the endogenous transmitter strongly [16]. Subsequently Di Cosmo et al. [21] not only showed by HPLC that the second-order fibre contains significantly more L-glu than does the third-order fibre, but also demonstrated by immunocytochemical staining, a crucial difference between the presynaptic and postsynaptic fibres. Application of an antibody raised against L-glu resulted in positive immunofluorescence in the second-order fibre only, whereas an antibody raised against mammalian GluR1 with GluR2/3 led to fluorescence in the third-order fibre only (Fig. 3).

Fig. (1) Diagram of squid stellate ganglion of Loligo vulgaris to show the 2nd-order giant fibre (gf2, black) branching to synapse with (in this specimen) 10 postsynaptic 3rd-order giant fibres (gf3sL, the largest of the 3rd-order fibres; L-1, the next largest, (more ...)

Fig. (2) Postsynaptic action of photolytically released L-glutamate at pH 7.8 at the giant synapse of squid stellate ganglion. Depolarization of postsynaptic fibre by photolysis of 20mM caged glutamate by 1ms 300-350nm light pulse at the time indicated by the (more ...)

Fig. (3) Anti-L-glutamate and anti-glutamate receptor GluR1-GluR2&3 immunofluorescence in 25μm sagittal sections of Loligo vulgaris stellate ganglion. (a) Anti glutamate immunoreactivity is particularly intense (arrows) in the terminal region of (more ...)

Recently Battaglia et al. [5] cloned a putative glutamate receptor subunit, SqGluR, from the stellate ganglion of the squid Loligo opalescens. They showed that its primary structure was homologous to mammalian AMPA-subunits GluR1-GluR4 (44-46% amino-acid identity) and also that intracellular injection of the C-terminal 15- amino-acid peptide of SqGluR into the post-synaptic region of the third-order fibre inhibited synaptic transmission at the giant synapse (Fig. 4). They also showed by in situ hybridisation that SqGluR mRNA is localised in many cells in the stellate ganglion, especially in neurons in the giant fibre lobe that give rise to the third-order fibres. Finally, the effect of L-glu, D-glu, L-asp, and D.asp has been studied on human embryonic kidney cells expressing SqGluR [10]. Only L-glu is able to evoke an inward current, while aspartic acid (both L and D isomers) slows the time course of activation and inactivation of L-glu gated current. This let suggest a role for aspartic acid as neuromodulator during glutamatergic transmission [10].

Fig. (4) Functional effects of the C-terminal peptide of SqGluR on synaptic transmission at the giant synapse. Effect of injected peptide Sqglu15. Postsynaptic responses to presynaptic stimulation: (a) control; (b) 10 min after injection; (c) response to antidromic (more ...)

In a paper that was the first to implicate L-glu as a putative neurotransmitter in cephalopods Bone & Howarth [6] applied drugs to strips of muscle cut from the mantle and fins of two squids (Loligo and Alloteuthis) and the cuttlefish, Sepia. They showed rapid twitches, often superimposed on slower contractures, after bath-application of 50 μM L-glu. Such responses were reversibly blocked by 2APB. They also observed that topical application of L-glu to the skin caused sustained expansion of all colour classes of chromatophore (see below), while its application to such visceral muscles as the branchial hearts, the gut and ink-sac duct caused “immediate repetitive contractions”. These findings have never been followed up physiologically but it is significant that Palumbo et al. [43] have shown the presence of NMDAR-like immunoreactivity in the wall of ink sac duct (see below).

Collins [12, 13] has confirmed and extended Bone & Howarth’s findings on the effect of L-glu on squid somatic muscles. It should be noted that in cephalopods, which lack an internal or external hard skeleton, the muscles fibres in the fins, mantle, arms and tentacles are orthogonally arranged in all three planes, constituting a “muscular hydrostat” [29]. Collins found, in four species of loliginid squids, that all three types of fin muscle (transverse, longitudinal and dorso-ventral) responded to L-glu (25-50μM) though not ACh. In the mantle, however, the longitudinal and radial muscles are cholinergic but the circulars glutamatergic, contracting strongly with 25-75μM L-glu; moreover the responses of the circular muscles are reduced or abolished by CNQX (Fig. 5). In the arms and tentacles Collins [12, 13] found that all muscle types were cholinergic. The use of different excitatory transmitters in the same muscle block is, of course, most curious by vertebrate or arthropod standards [and one can only speculate on its significance]; Bone et al. [7] suggest it may have arisen because the opposing fibres are so closely apposed in the squid mantle muscle that only the use of different transmitters could ensure isolated contraction during successive stages of contraction in the jetting cycle.

Fig. (5) Tension records showing the response of the central zone circular mantle muscular fibres of Loligo vulgaris to 50 μM L-Glu in artificial sea water (ASW). The slice was then washed for 5 min in ASW. After five minutes continuous perfusion of ASW (more ...)

Bone and Howarth’s discovery that the chromatophores expand in the presence of L-glu has been amply confirmed and extended, first by Florey et al. [24] and later by Messenger and his colleagues [14, 30-33, 39, 41]. Cephalopod chromatophores differ from those of vertebrates in that they are complex organs, comprising a sac of pigment granules to which are attached 15-25 radial muscles, innervated directly from the brain. When the muscles contract, the chromatophore expands; when they relax, elastic energy stored in the sac causes the chromatophore to retract. A variety of evidence has accumulated that all points to L-glu being the endogenous excitatory transmitter at the chromatophore nerve-muscle junction: (1) staining of the chromatophore nerves with an antibody to L-glu; (2) topical application of a range of (mammalian) glutamate agonists and antagonists to intact and long-term denervated chromatophores; (3) following the behaviour of intracellular-Ca²⁺ levels during L-glu application to isolated radial muscle fibres (for summary, see [38]). AMPA, kainate quisqualate and domoate all expand the chromatophores, but this effect can be blocked by CNQX and DNQX (50μM). In Sepia Loi & Tublitz [35] have also shown that Joro spider toxin can block the effects of L-glu, so that there is overwhelming evidence that at the majority of chromatophores L-glu acts via AMPA/kainate-like receptors.

This was further supported by the work of Lima et al. [33] who showed that AMPA/kainate-like receptors are present in squid skin by immunoblotting with an antibody raised against a squid AMPA receptor (sGluR). However, more interestingly, these workers also identified a sub-population of chromatophores in the dorsal mantle skin of two species of squid where neurotransmission was partly blocked either by CNQX or DNQX or by the NMDA receptor antagonists AP-5 and MK-801, and completely blocked by the simultaneous application of both classes of antagonist. Using isolated muscle fibres from this same area they iontophoretically applied L-glu to reveal fast inward currents (sensitive to CNQX and DNQX) and slow outward glutamate–sensitive currents (Fig. 6). This slow component was voltage dependent and sensitive to external magnesium and glycine (Fig. 7); moreover L-glu also causes a fast, followed by a slow transient increase in Ca²⁺, similarly sensitive to AP-5, glycine and magnesium. These findings strongly suggest that in the muscles of these chromatophores NMDA-like and AMPA/ kainate-like receptors co-exist, and on the basis of these results Lima et al [33] propose that such an arrangement enables the conspicuous dorsal chromatophores of these squids to maintain expansion as they contribute to a variety of body patterns for concealment or for signalling [15].

Fig. (6) Ionic currents evoked by ionophoretic application of L-glutamate in isolated chromatophore muscle cells from mid-dorsal skin of squid. (A) Glutamate-evoked currents showing a voltage-independent single component. (Aa) Leak-subtracted traces recorded at (more ...)

Fig. (7) Ratio (F340/F380) measurements of Fura-2 AM-loaded squid chromatophore muscle fibres. (a) CNQX (100 μM) blocked L-glutamate (20 mM)-evoked Ca2+ rise, which was obtained after cell was washed in ASW for 5 min. (b) Removal of external Mg2+ enhanced (more ...)

The cephalopod statocysts are complex organs comprising two separate equilibrium receptor systems, hair cells on the macula responding to linear-acceleration and those on the crista responding to angular-acceleration [11]. The sensory epithelia receive massive afferent innervation and in the crista Tu & Budelmann [47] have shown that this is glutamatergic; they also provided good evidence that the receptors were of the AMPA/kainate type.

According to the observations of Bone & Howarth [6] topical application of glutamate to the muscles of the inkgland duct caused “immediate repetitive contractions”. Subsequent immunohistochemical staining, with antibodies to rat NMDAR2/3 subunits, in the regions of the anterior and posterior sphincters of this duct, labelled a few fibres from the visceral nerves that reach the outer circular muscle layer [43]. Somewhat more intense and widely distributed immunoreactivity was detected in fibres reaching the inner longitudinal muscle layer.

All along the wall of the ink sac and below the sphincters, NMDAR2/3-immunopositive fibres reach the trunks of the longitudinal muscle layer but not the circular muscle layer of the ink sac wall. The antibody against subunit 1 of the NMDA receptor did not show any positivity when tested on sections of the ink sac [43].

It is clear that the cephalopod brain contains glutamatergic systems, but the evidence to date is fragmentary and incomplete. The first experiments to implicate L-glu came from Andrews et al., [1, 2]. Using an aortic perfusion technique, by which small “pulses” of drug can be introduced directly into the blood stream of an octopus, these workers obtained dramatic (though transient) motor effects in the arms and mantle with L-glu, kainate and quisqualate. These same substances also expanded the black chromatophores, leading to conspicuous darkening. Control experiments, such as lesioning the pallial nerve to uncouple the mantle muscles on one side (including those of the chromatophores), emphasise that to produce these effects the drugs were acting directly on the CNS.

Subsequently D’Aniello et al. [17] and D’Aniello & Messenger (unpublished data) directly measured L-glu and L-aspartate levels in different regions of the Octopus brain and showed that L-glu was widely distributed in the lower, intermediate and higher motor centres (including those regulating the chromatophores), in the optic lobes and in the superior frontal - vertical lobe system (involved in learning) (see [37]). There are complementary findings for Sepia [43].

Immunohistochemical staining of brain sections with antibodies raised against L-glu has unfortunately produced inconsistent results although L-glu positive cell bodies seem to be widespread. This powerful technique for localising putative transmitter distribution has, however, been more successful with antibodies to glutamate receptors and Di Cosmo et al. [22] have excellent evidence that certain NMDA-like receptors are widely distributed in the brains of Octopus and Sepia. In summary we can note that positive immunoreactivity specific for NMDA receptor subunits 2A and 2B was found in the cell bodies and the neuropil of many (but not all) of the lobes that Boycott [8] defined as Lower and Intermediate Motor Centres: for example the pedal, magnocellular, palliovisceral and (in Sepia) the fin lobes. Similarly widespread specific staining was observed in the Higher Motor Centres, such as the anterior, lateral and median basal lobes and in the “silent areas” - those lobes that give no response to electrical stimulation but which are important in the setting up of memories, visual or tactile: the vertical, subvertical, superior frontal and (in Octopus) the subfrontal lobes [22]. Finally it is interesting to note that immunopositive cells and fibres are also found in several brain areas that influence the reproductive system via the optic gland, a gonadotropic endocrine gland.

These results are supported by SDS-PAGE analysis of brain extracts (from Sepia and Octopus), which revealed a single sharp band at 170 kDa immunoreactive to NMDAR2/3 receptor subunit antibodies. A similar immunoreactive band was detectable in extracts from optic lobes. An exceedingly weak response seems to rule out the presence of the R1 subunit in the cephalopod brain [22].

Complementary results have been obtained by Battaglia et al. [5] on the squid Loligo. Having cloned a glutamate receptor subunit, SqGluR (see above) they also showed by in situ hybridisation that SqGluR mRNA occurs in the second-order “giant” visual cells in the inner and outer granular layers, and in what appears to be the subvertical lobe.

Taking a biochemical approach, Palumbo et al. [43, 44] have shown that exposure of intact brains to either 1.5 mM glutamate or 1.5 mM NMDA caused a rapid increase in the levels of cyclic GMP after the activation of a neuronal nitric oxide synthase (NOS) and the production of nitric oxide (NO), all peaking within 1 min and then rapidly dropping to basal values within 3 min. Significant suppression of the activating effect of NMDA was observed in the presence of the NOS inhibitor L-NA, as well as the glutamate/NMDA antagonist D-AP5 [43].

In recent experiments it has been also demonstrated that glutamate, via the NMDA/NO signal transduction pathway , is involved in post-translational modification of cytoskeletal protein such as 03B1-tubulin [20, 44]. It has been showed that glutamate/NO mediated tyrosine nitration at the C-terminus of α-tubulin is specifically inhibited by D-AP5. The incorporation of nitrotyrosine induces α-tubulin reversible degradation. This temporary process could be associated with physiological mechanism of cytoskeletal protein turnover. Such a mechanism could be involved in microtubule assembly , which is critical for axonal transport, growth, degeneration as well as in synaptic plasticity [20].

Physiological evidence for a glutamatergic system in the cephalopod brain comes also from a very interesting paper by Hochner et al. [25]. Their experiments centred on the superior-frontal/vertical lobe system in Octopus, a region of the brain that has been known for over 50 years to be closely involved in learning and memory [9]. By recording field potentials in brain slices of this area of the brain Hochner and his colleagues obtained good evidence of activity-dependent synaptic plasticity, very similar to the well known, NMDA-mediated. long-term potentiation (LTP) in the mammalian hippocampus [36]. However, LTP in the Octopus vertical lobe apparently does not involve NMDA receptors although the receptors are clearly glutamatergic: CNQX, kynurenate and DNQX block post-synaptic field potentials, and L-glu occludes them (Fig. 8). The same authors, however, do not exclude the presence of currents through NMDA receptors, though D-AP5 insensitive. In fact, for Aplysia, AP5-insensitive NMDA currents have been reported [18, 42]. If so, this could reconcile the LTP results with the immunocytochemical data.

Fig. (8) LTP in the Octopus vulgaris involves non NMDA glutamate receptors. The response during HF [inset (1)] was completely blocked by a mixture of 20 mM kynurenate plus 200 μM CNQX. After rinsing, the test fPSP was highly potentiated with respect to (more ...)

Thus a combination of different experimental strategies clearly implicates L-glu as a neurotransmitter in specific areas of the cephalopod brain: biochemical evidence, linking glutamate and its receptors to nitric oxide, immunocytochemical studies and a classical electrophysiological approach all demonstrate the role of L-glu and show that it exerts its effect via both NMDA and non-NMDA, AMPA/kainate-like receptors.

Cephalopods are well known to neuroscientists primarily because of the squid giant axon. Re-discovered in the 1930s by J.Z.Young [49], this was used in the elegant experiments of Hodgkin and Huxley [26-28] to establish the ionic basis of the action potential in nerves, work that gained them a Nobel Prize and that today still constitutes a cornerstone of modern neurophysiology.

However the sense organs, nervous system and behaviour of these animals have been studied in depth for more than a century. Such studies were given impetus by the realisation that octopuses had learning abilities comparable to those of lower vertebrates and were excellent laboratory animals [9]. As a result there is a substantial literature on the neurobiology of cephalopods including their putative transmitters, notably the “classic” vertebrate transmitters: ACh, dopamine and noradrenaline. This early pharmacology has been admirably summarised by Tansey [46], but shortly after her review was published it began to emerge that L-glu was also an important neuroactive agent in several systems of squids and octopuses (for review see [37]).

In the last ten years this has been confirmed beyond doubt. As this review has shown, cephalopods utilise L-glu as a transmitter at some (but not all) types of nerve-muscle junction; in some sense organs; in the peripheral nervous system; and in the CNS. What is remarkable in the brain is that L-glu occurs in so many systems: in lower, intermediate and higher motor centres; in learning centres; in the visual system; and in the olfactory-endocrine-reproductive regulatory system. In all these its effect appears to be excitatory, as it is in mammals [45].

Evidence is also emerging about the nature of the glutamate receptors in cephalopods. At first it seemed that they had only AMPA/kainate type receptors, but it is now clear that in several systems L-glu acts via receptors similar to the vertebrate NMDA type. The evidence for the presence of NMDAR2A-like and NMDAR2B-like subunits is particularly strong.

At this point it may be worth reminding mammalian neuropharmacologists that molluscs and vertebrates have been evolving independently for some 400Myr so that it is highly unlikely that their receptors will be the same. More important are the facts that they are sufficiently similar (a) for new glutamate agonists and antagonists developed by mammalian pharmacologists to be physiologically active in cephalopods; (b) for antibodies raised against rabbit NMDAR to label cell bodies and fibres in the cephalopod brain; (c) for Western blot analysis of membrane proteins from Octopus and Sepia brains to reveal an immunoactive band at 170 kDa, and (d) for there to be a link between glutamate and NMDAR–mediated NO production in Sepia.