Excitatory Amino Acid Neurotransmission

Carl W. Cotman, Jennifer S. Kahle, Stephan E. Miller, Jolanta Ulas, and Richard J. Bridges

The amino acid L-glutamate is now recognized as the major excitatory neurotransmitter in the central nervous system (CNS). Accumulating evidence suggests that the glutamate system is involved not just in fast synaptic transmission, but also in plasticity and higher cognitive functions. It is becoming clear that the glutamate transmitter system is organized at a high level of sophistication. This is evident in the organization of the glutamate receptor subtypes, which include both ionotropic and metabotropic receptor families. Ionotropic glutamate receptors (see Introduction to Preclinical Neuropsychopharmacology and Introduction to Preclinical Neuropsychopharmacology) can be distinguished pharmacologically by specific binding of the agonists N-methyl-D-aspartate (NMDA), kainic acid (KA), and a-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) () and include receptors that gate both voltage-dependent and voltage-independent currents carried by Na+, K+, and, in some cases, Ca2+. Because Ca2+ can act as a second messenger, it can initiate a wide range of intracellular responses. Metabotropic glutamate receptors (mGluRs) provide for another level of response complexity through their links with the phosphoinositide (PI) and cyclic nucleotide (cAMP) second messenger systems. Thus, different combinations of receptor subtypes can influence the specific functional capability of individual synapses and neurons. In contrast to a relatively simpler system, such as the neuromuscular junction, this complexity provides the infrastructure necessary of a system involved in higher cognitive function (see Schizophrenia and Glutamate: An Update, Biological Markers in Alzheimer’s Disease, Amyotrophic Lateral Sclerosis, Glutamate, and Oxidative Stress, and Potential Mechanisms of Neurologic Disease in HIV Infection).

This organization, however, appears to come at a cost. The processes of information integration, association, and storage utilize cellular mechanisms for which there is a narrow range of normal function. An imbalance at any one of many key points of the processes can lead to the destruction of the neurons involved and the disruption of circuitries and associated functions. This is most clearly illustrated by the process of excitotoxic-mediated neuronal injury, which may contribute to CNS pathology in a variety of neurodegenerative conditions.

The concept of the pervasiveness and homogeneity of glutamate as the principal transmitter is yielding to an understanding of the highly sophisticated specialized combinations of receptors discretely organized to carry out specific functions. The goal of this chapter is to provide an updated overview of the excitatory amino acid (EAA) transmitter system, in particular, focusing on the glutamate receptors and transport proteins. We have chosen to emphasize two areas that are rapidly developing: (i) the pharmacology of the ligand binding sites on the receptors and transporters () and (ii) the molecular biology of the receptors , , ). Progress in molecular techniques has led to the cloning and expression of many of the components of the glutamate receptor system. This provides a valuable strategy to examine the independent function of the subunits in defined combinations from reconstitution experiments. In parallel, advances in pharmacology have been directed toward the characterization of binding-site pharmacophores and the design of conformationally restrained glutamate analogues. As an acyclic molecule, glutamate can assume a staggered conformer as a result of rotation about its ab and bg sp3 carbon bonds. Conformationally restricted analogues, which mimic stable conformations of glutamate, allow the exploration of the specific stereochemistry of the subunit pharmacology. The integration of molecular biology and pharmacology and the organization of the specific receptors in the CNS holds great promise for the characterization of the complex structure–function relationships within this transmitter system in health and disease.

NMDA receptors were initially identified and separated pharmacologically from other EAA receptors by selective activation by the agonist NMDA. The recent explosion of research on this receptor subtype has revealed that its activity is highly regulated via several allosteric regulatory binding sites on the receptor/channel complex. Activation of the receptor and concurrent depolarization results in the development of a relatively slow-rising, long-lasting current mediated primarily by the influx of Ca2+. Calcium entering the cell can also mediate longer-lasting cellular responses. NMDA receptors appear to have a pivotal role in long-term potentiation (LTP), long-term depression (LTD), and developmental plasticity. Overactivation of NMDA receptors, however, appears to cause damage via excitotoxicity.

At a single CNS synapse, NMDA receptors usually coexist with either AMPA or KA receptors and are thought to be involved in amplification of the glutamate signal, although examples of primarily NMDA-mediated synaptic responses have been reported (70). At resting potentials, NMDA channels are normally blocked by Mg2+ and there must be sufficient concurrent depolarization of the postsynaptic neuronal membrane (to about -30 mV) before the Mg2+ block is relieved and the NMDA channel can contribute to the electrical response of the cell. The level of concurrent depolarization depends on AMPA/KA activation and/or other modulatory postsynaptic signals controlling depolarization.

Precise modulation of NMDA channel activity is required for normal neuronal function, and, as expected, there are several regulatory sites on the NMDA receptor/channel complex which control NMDA-mediated activity further. The binding of glycine to the receptor/channel complex increases the frequency of agonist-induced channel opening. Glycine binding appears to be an absolute requirement for NMDA channel activation. Studies of the NMDA receptor/channel complex at the molecular level indicate that binding of two glycines and two glutamate molecules is required for channel activation (23). The receptor/channel complex also includes a polyamine binding site which, when activated, potentiates the NMDA current by increasing channel open probability. This is due, in part, to an increase in glycine binding. At higher concentrations, the polyamine spermine binds to a second site which results in a reduction of the NMDA current potentiation and a decrease in the channel conductance and/or mean open time. The NMDA receptor/channel complex can also be inhibited via a Zn2+ binding site. In addition, there is a distinct site within the channel that binds MK-801 and PCP inhibiting channel opening (for review see ref. 34). Thus, there is capacity for several different mechanisms for both positive and negative control of NMDA channel function.

The voltage-dependent properties of NMDA-mediated Ca2+ current provides the capacity for Hebbian-type plasticity at synapses where NMDA receptors are located (see ref. 13 for discussion). Repetitive or concurrent activation can depolarize the postsynaptic cell to the level at which Mg2+ block of NMDA-mediated current is relieved and where these channels begin to contribute additional currents to the postsynaptic response. Furthermore, the influx of calcium through this channel initiates long-term synaptic and cellular modification. Thus, NMDA receptor/channels have a key role in plastic events at the synaptic, cellular, and behavioral level.

On the other hand, increasing evidence indicates that the complex mechanisms controlling synaptic plasticity in the brain also introduce points of vulnerability to pathology. In this case, too much intracellular free calcium can be toxic to neurons, and overstimulation can result in excitotoxic cell death. A breakdown of any one of the points of modulation of NMDA channel activity can lead to excitotoxicity. This type of cell death appears to contribute to brain pathology in several conditions, including epilepsy and ischemia, and, possibly, Alzheimer's disease and Huntington's disease (12).

The NMDA receptor subtype has recently been cloned and appears to include two families of subunits (). NMDAR1 (also called NR1 or z1 in mouse) exhibits the major electrophysiological and pharmacological characteristics of the NMDA receptor/channel complex when expressed in isolation in Xenopus oocytes. However, the current carried by the channel is much smaller than that observed in native receptor/channel complexes (36). Presently, seven splice variants have been described and called NMDAR1A–NMDAR1G (NR1a–NR1g). The basic pharmacological and electrophysiological characteristics of channels formed from these splice variants are reminiscent of native NMDA receptors, with slight shifts in specific characteristics (e.g., binding affinities) (56). However, there are several marked differences that have been described. For example, it appears that the NMDAinduced current through NMDAR1B channels are not potentiated by exposure to the polyamine spermine (15).

NMDAR2A–NMDAR2D (also called NR2A–NR2D or e1–e4 in mouse), the second family of subunits to be cloned, are considered to be modulatory subunits and do not form channels by themselves (homomeric channels), but form heteromeric channels (channels consisting of more than one type of subunit). Thus, when NMDAR2 subunits are expressed with NMDAR1A, heteromeric channels are formed that exhibit much greater currents than those of homomeric NMDAR1 channels. Different combinations of specific NMDAR2 subunits and NMDAR1A subunits appear to produce NMDA receptor/channel complexes with subtly different electrophysiological properties. For example, NMDAR1 and NMDAR2C heteromers exhibit an increased current relative to those of NMDAR1 and NMDAR2A (20).

The distribution of different NMDA receptor subunit expression is distinct throughout the brain. In general, the patterns of mRNA expression appear to be consistent with heteromeric receptors made up of a common subunit (NMDAR1) with various NMDAR2 subunit combinations. Thus NMDAR1 is expressed at high levels in most neurons (36), while NMDAR2 mRNAs show distinct temporal and spatial patterns in the developing and adult brain. Previous autoradiographic studies have suggested that NMDA receptors are a heterogeneous population including agonist-preferring and antagonist-preferring receptors (35). Agonist-preferring sites are observed, for example, in the medial striatum, while antagonistpreferring sites are found in the lateral thalamus. These two populations of NMDA receptors may be related to different combinations of NMDA subunits from the two families. The NMDAR2A subunit, for example, has a distribution very similar to that of the antagonistpreferring NMDA subtype (labeled by the antagonist 3H-CPP). Furthermore, when reconstituted in oocytes, NMDAR1/NMDAR2A receptors have a higher affinity for antagonists whereas NMDAR1/NMDAR2B receptors have a higher affinity for agonists (9). Such studies are providing a direct link between subunit composition and receptor physiology.

Besides NMDA, several other dicarboxylic amino acids are also agonists at the NMDA binding site. These include L-glutamate, S-sulfo-L-cysteine, L-homocysteate, L-aspartate, homoquinolinate, L-homo-cysteinesulfinate, L-cysteinesulfinate, L-serine-O-sulfate, L-cysteate, and quinolinate, in order of potency from greatest to least (31). Several conformationally restricted agonists have been identified, some of which are more potent than NMDA. For example, one of the most potent rigid agonists is trans-1-aminocyclobutane 1,3 dicarboxylate (trans-ACBD), a natural product of a seed (genus Atelia). The fact that this analogue is selective and potent is consistent with the suggestion that decreasing conformational flexibility increases selectivity. Modeling studies based on a variety of conformers have led to a proposed stereochemical arrangement of functional groups for the NMDA receptor (10, 42).

The pharmacology of the NMDA receptor/channel complex indicates three major possible mechanisms of antagonism. Several well-characterized competitive antagonists include: D-AP5, D-AP7, D-a-amino adipate, CPP, and CPPene. Antagonists in which the intervening carbons have been rigidified (such as in a cyclic conformation) have led to the synthesis of compounds more potent than AP5 or AP7 (e.g., CGP37849). Several noncompetitive antagonists have been identified that bind to a site within the channel itself (e.g., MK-801 and PCP) (34). There are also non-competitive antagonists that bind to the glycine binding site: kynurenate, 5,7-dichlorokynurenate, MNQX, and L689,560 (29). The study of cyclic derivatives which are conformationally constrained will allow for a rigorous analysis of the structure–activity relationship of the distinct subunits.

Both KA and AMPA receptors mediate fast excitatory synaptic transmission and are associated primarily with voltage-independent channels that gate a depolarizing current primarily carried by an influx of Na+ ions (see ref. 34 for review). The natural plant product KA (isolated from Digenea simplex) and a synthetic analogue of quisqualic acid, AMPA, have had a pivotal role in the characterization of the non-NMDA glutamate receptor subclass. Early electrophysiological studies with these potent excitants (often in combination with EAA antagonists) demonstrated the presence of EAA receptor subtypes that were easily distinguishable from NMDA receptors. Thus, the classification of these receptors as "non-NMDA" evolved as a result of being able to use NMDA antagonists (e.g., D-AP5, MK-801) to easily differentiate KA and AMPA receptor-mediated responses from NMDA receptormediated responses, but not from each other. Glutamate activation of AMPA receptors initiates a current comprised of a fast desensitizing component and a steady-state component. Importantly, specific combinations of subunits regulate the differing desensitization kinetics. This desensitization, in turn, appears to be controlled by an allosteric site that binds compounds in the aniracetam family (68). In the presence of these modulators, desensitization of the receptor is delayed and the synaptic current is correspondingly enhanced. The site is important because aniracetam analogues appear to enhance synaptic activity and improve learning in animal models (63). Receptors relatively specific for KA have been identified with ligand binding (34); however, KA and AMPA can show overlapping patterns and ligand cross-reactivity.

Molecular biology studies on the non-NMDA receptors have not only confirmed the existence of the KA and AMPA classes, but indicate that the potential heterogeneity within these receptor families reveals a remarkable degree of complexity. AMPA receptor channels can be formed by reconstituting one or any two of four subunits (GluR1–GluR4, also called GluRA–GluRD), while the KA subclass of receptors includes GluR5–GluR7 and KA-1–KA-2 ().

AMPA receptors are approximately 900 amino acids in length and occur in two forms distinguished by the presence or absence of an alternatively spliced exonic sequence of 38 residues preceding the last transmembrane domain (TM IV) (62) (for explanation see Basic Concepts and Techniques of Molecular Genetics). This alternative splicing of GluR1–GluR4 results in the so-called "flip" or "flop" variants. The "flip" forms give rise to a larger sustained current (slower to desensitize) than do the "flop" forms. Any one or two of the four subunits assemble into ion channels and are activated by AMPA and, to a lesser degree, KA. The ionic specificity of the assembled channels (e.g., Na+ or K+ versus Ca2+) has been shown to be dependent upon the combination of subunits expressed. GluR1, GluR3, and GluR4 are Ca2+-permeable, while GluR2 appears to possess an optional Ca2+-permeability that is dependent upon the amino acid composition at a single residue. The amino acid composition of this site is controlled by RNA editing. That is, a CAG codon (glutamine) in transmembrane domain II is edited to CIG (arginine) via the enzyme adenosine deaminase. Thus, brain cells can regulate the extent of Ca2+-permeability by controlling the level of GluR2 expression and by RNA editing.

Calcium influx through glutamate receptor channels is thought to play a causal role in glutamate-mediated excitotoxicity. Following transient global ischemia, expression of GluR2, which limits calcium permeability in heteromeric channels, is suppressed in CA1, a region susceptible to ischemic injury, but not in CA3 or the dentate gyrus, regions which are resistant to ischemic injury (45). Such differential regulation suggests that the ratios of expression of the individual receptor subunits is an important determinant in maintaining or disrupting cell viability.

The KA subclass of receptors includes GluR5–GluR7, which correspond to the previously described low-affinity site, and KA1–KA2, which correspond to the highaffinity site. Homomeric expression of GluR5 or GluR6 (but not GluR7) yields a binding pharmacology consistent with a high-affinity site and the formation of ion channels that are activated by KA but not AMPA. In contrast, homomeric expression of KA1 or KA2 subunits does not generate functional ion channels, though binding studies are consistent with a high-affinity KA site. Functional channels are produced, however, if KA1 or KA2 is heteromerically expressed with GluR5 or GluR6. Both GluR5 and GluR6 (not GluR7 or KA1-2) occur in two forms with respect to the Q/R site controlling Ca2+ permeability in the transmembrane II domain that can be modified by RNA editing. These two forms occur at different frequencies in the CNS.

Autoradiographic studies with radioligands clearly demonstrate that while both KA and AMPA sites are principally localized in telencephalic regions, each exhibits a distinctive distribution. The combined patterns of the five-subunit mRNAs of KA approximate the patterns observed for high-affinity 3H-KA sites in the rat brain. KA binding sites are relatively enriched in the hippocampal CA3 stratum lucidum, striatum, deep cortical layers, reticular nucleus of the thalamus, and granule cell layer of the thalamus (14). At present, in situ hybridization studies of the AMPA subunits are difficult to interpret and match with autoradiographic studies. Distribution studies with 3H-AMPA demonstrate that the binding sites are concentrated in CA1 stratum radiatum, outer cortical layers, lateral septum, and molecular layer of the cerebellum (14). This distribution is similar to that exhibited by the NMDA receptors and is consistent with their common action as a functional pair.

The respective pharmacologies of KA and AMPA receptors are similar enough that they are more often distinguished by the relative rank order of potencies of a series of agonists rather than by the action of a single selective compound. Electrophysiological studies demonstrate that the relative potencies of non-NMDA agonists (e.g., KA, domoate, QA, AMPA) vary according to the brain area examined. These studies, however, have been carried out in physiological preparations where the precise receptor subunit compositions of the recorded neurons has yet to be determined. It is clear that the development of more selective agonists and antagonists will play an important role in correlating the receptor structure and function. Considerable progress, however, has been made in identifying agonists that exhibit preferential activity at KA or AMPA receptors ().

KA, a di-substituted proline derivative, contains an embedded L-glutamate moiety that is conformationally restricted about the a,b bond which is probably responsible for the reduced affinity of KA for NMDA receptors, metabotropic receptors, and the high-affinity Na+-dependent glutamate transporter (10). KA can, however, assume several envelope conformers, mimic a number of glutamate conformations, and therefore interact with several types of GluRs (10). The KA derivatives, acromelic acid and domoic acid, have been shown to be more potent than KA as specific KA receptor agonists. Domoic acid is of particular interest because consumption of this neurotoxin by humans has been shown to produce pathological damage to the hippocampus as well as dementia (64). AMPA shows good specificity for AMPA receptors and itself has limited conformational flexibility. Several conformationally constrained analogues of AMPA have been shown to be more potent and selective agonists at the AMPA receptor (27). Other competitive-site AMPA agonists include ATPA, 5-fluorowillardiine, Br-HIBO, and b-L-ODAP, the EAA agonist identified as the causative agent in neurolathyrisms (8). A separate set of agonists exhibit non-competitive modulatory effects on AMPA receptors via action at an allosteric site(s) involved in attenuating receptor desensitization, thereby enhancing synaptic current. Agonists with such allosteric action include aniracetam (68), related analogues [e.g., 1-BCP (63)], and a series of benzothiadiazides [e.g., cyclothiazide (72)].

At present, the most potent and selective of the non-NMDA antagonists are a series of dihydroxyquinoxaline derivatives, CNQX, DNQX, and NBQX. Although these antagonists will competitively block both of the non-NMDA receptors, NBQX may provide insight into the design of more specific antagonists, because it appears to exhibit the greatest selectivity for AMPA receptors. Few KA-selective compounds have been developed, but recently NS-102 has been reported to competitively antagonize low-affinity kainate binding with some selectivity (21). A promising new class of 2,3-benzodiazepines (particularly GYKI 52466) has been demonstrated to non-competitively block both AMPA- and KA-induced responses (50). Importantly, many of the newly developed competitive and non-competitive antagonists have been shown to potently attenuate ischemic neuronal injury in animal models, highlighting the potential role of non-NMDA receptors in CNS pathology. Structural comparisons and molecular modeling with these antagonists and agonists is beginning to shed some light on the respective pharmacophores of the KA and AMPA receptors (10, 27). Progress will no doubt accelerate as analogues are examined in detail on homomeric and heteromeric subunits.

Unlike the ionotropic glutamate receptors which are directly coupled to cation-specific ion channels and mediate fast excitatory synaptic responses, the more recently characterized mGluRs are coupled to a variety of signal transduction pathways via guanine-nucleotide-binding proteins (G proteins) (see Introduction to Preclinical Neuropsychopharmacology and Signal Transduction Pathways for Catecholamine Receptors), producing alterations in intracellular second messengers and generating slower synaptic responses. The prevalence of glutamate as a neurotransmitter in combination with the widespread distribution of metabotropic receptors points to this system as a major modulator of second messengers in the mammalian CNS.

Different subtypes of mGluRs are linked to at least two major signal transduction cascades: PI hydrolysis and the adenylate cyclase/cyclic AMP system. The ability of glutamate to stimulate PI hydrolysis has been extensively characterized in a variety of preparations, including tissue slices, cultured neurons, cultured astrocytes, and transfected cell lines (see ref. 54 for review). Stimulation of PI hydrolysis results from activation of a receptor coupled to a G protein that activates phospholipase C, initiating a signaling cascade by cleaving phosphatidylinositol-4,5-bisphosphate into two second messengers: diacylglycerol which activates protein kinase C and inositol 1,4,5-trisphosphate (IP3) which elicits the release of Ca2+ from intracellular stores.

Metabotropic glutamate receptors can also regulate cAMP levels. Synthesis of cAMP by adenylate cyclase can be altered by activation of either stimulatory Gs proteins or inhibitory Gi proteins which produce opposing effects on adenylate cyclase. In contrast to the excitatory actions of glutamate on ionotropic receptors and PI hydrolysis, glutamate produces an inhibitory effect on cAMP accumulation, an action that has now been characterized in a number of preparations. Some studies have also demonstrated a stimulation of cAMP accumulation with mGluR activation, and evidence exists for coupling of mGluRs to other transduction pathways, including phospholipase D, arachidonic acid, and direct G-protein coupling to cation channels (54).

In addition to biochemical studies, electrophysiological consequences of mGluR activation have also been extensively studied. These studies have been greatly facilitated by the identification of a selective agonist 1-aminocyclopentane-trans-1,3-dicarboxylic acid (trans-ACPD), a conformationally restricted analogue of glutamate which activates metabotropic, but not ionotropic, glutamate receptors (44). Electrophysiologically, activation of mGluRs results in both excitatory and inhibitory actions (54). For example in the hippocampus, trans-ACPD can produce depolarization and reduction of the after-hyperpolarization via a blockade of a Ca2+-activated potassium current. It has also been shown to block accommodation of cell firing, increase the amplitude of population spikes, induce generation of multiple spikes, decrease paired-pulse inhibition, and decrease evoked inhibitory postsynaptic potentials. Inhibitory actions of trans-ACPD in the hippocampus include reduction of the field excitatory postsynaptic potential via a presynaptic action.

To date, six different cDNA clones of mGluRs (mGluR1–mGluR6), all with highly conserved amino acid sequences, have been isolated from rat cDNA libraries (38) (). It has been shown that the family of mGluRs does not share overall sequence similarities with other G-protein-coupled receptors (66). In contrast to ionotropic glutamate receptors, mGluRs contain three structural domains: a large extracellular hydrophilic NH2terminal sequence, seven membrane-spanning domains characteristic of G-protein-linked receptors, and a hydrophilic intracellular COOH-terminal sequence. Members of the mGluR family can be divided into three subgroups according to their sequence similarities, signal transduction properties, and pharmacological profile (i.e., relative potencies of QA, glutamate, ACPD, and AP4) when expressed in cell lines (67).

The members of the first group, mGluR1 (2) and mGluR5 (1), are coupled to PI hydrolysis when expressed in Chinese hamster ovary (CHO) cells. In addition, mGluR1 stimulates cAMP formation and arachidonic acid release with the same order of agonist selectivity (2). Two splice variants of mGluR1 with truncated carboxy terminals have also been isolated (46, 66). The second subgroup, comprised of mGluR2 and mGluR3, are negatively coupled to adenylate cyclase in CHO cells (66, 67). The members of the third group, mGluR4 and mGluR6, also inhibit cAMP accumulation but have a different agonist selectivity (37, 67, 69).

As a group, the mGluRs are widely expressed throughout the brain. The individual subtypes are differentially distributed although sometimes overlap. Considering first the receptors coupled to PI hydrolysis, in situ hybridization (46, 58) and immunocytochemistry (30) have shown a wide distribution of mGluR1 in rat brain, with particularly prominent expression in hippocampal dentate gyrus granule cells, CA4 cells, CA2–CA3 pyramidal cells, cerebellar Purkinje cells, mitral and tufted cells of the olfactory bulb, and neurons of the thalamus and lateral septum. The receptor appears to be localized postsynaptically, and no glial staining has been demonstrated (30). Expression of mGluR5, the other PI-coupled subtype, is prominent in the cerebral cortex, dentate gyrus granule cells, pyramidal cells in CA1–CA4, subiculum, lateral septum, internal granule layer of the olfactory bulb, anterior olfactory nucleus, striatum, and nucleus accumbens (1).

Of the adenylate-cyclase-inhibiting receptors, the distribution of mGluR2 is more restricted than that of the PI-coupled receptors. In situ hybridization showed the most prominent expression in Golgi cells of the cerebellum, mitral cells of the accessory olfactory bulbs, and pyramidal neurons of the entorhinal cortex and parasubicular cortex, as well as in the dentate gyrus (40). It has been suggested that mGluR2 may serve as a presynaptic receptor in the cortico-striatal glutamate projection (40). mGluR3, the subtype with the greatest homology to mGluR2, is more widely distributed. Prominent expression of mGluR3 mRNA was found in neurons of the cerebral cortex, thalamic reticular nucleus, caudateputamen, supraoptic nucleus, and granule cells in the dentate gyrus (67). In addition to the neuronal localization, mGluR3 mRNA was found in glial cells in the corpus callosum, anterior commissure, and throughout the brain (67).

The final subgroup of receptors is comprised of mGluR4 and mGluR6 which also have distinct patterns of expression. mGluR4 is most prominently expressed in cerebellar granule cells, neurons of the internal granule layer of the main olfactory bulb, thalamus, lateral septum, pontine nucleus, and entorhinal cortex, with weaker expression in dentate gyrus and CA3 of the hippocampus (26, 67). The selectivity of mGluR4 for L-AP4 and its expression in the entorhinal cortex suggest that mGluR4 corresponds to the AP4 receptor, a presynaptic autoreceptor initially described in physiological studies (24, 69). The distribution of the final receptor subtype, mGluR6, is the most restricted of all the mGluRs. Appreciable expression was observed only in the inner nuclear layer of the retina (37), the area of the ON-bipolar cells, known to be hyperpolarized by glutamate or L-AP4.

It is interesting to note the degree of differential expression not only between regions but also within a region. In the cerebellum, for example, there is prominent expression of three separate subtypes: mGluR1 in the Purkinje cells, mGluR2 in the Golgi cells, and mGluR4 in granule cells. Such precise segregation of receptors implies an important role for the subtypes in functional specialization.

Glutamate, QA, and IBO were extensively used in the initial studies of mGluRs. These agonists, however, do not differentiate between ionotropic and metabotropic receptors: Glutamate, of course, activates all receptor subtypes, QA also activates AMPA receptors and IBO cross-reacts with NMDA receptors. To further examine the functional roles of mGluR activation required the development of pharmacological tools specific for the mGluRs such as trans-ACPD (44) and (2S,3S,4S)-a(carboxycyclopropyl) glycine [L-CCG-I (59); see )]. However, recent discoveries of the molecular heterogeneity of mGluRs and the coupling of mGluRs to multiple signal transduction pathways have necessitated identification of even more selective agonists to pharmacologically differentiate between mGluR subtypes. The original pharmacological studies of metabotropic responses were conducted primarily with brain slices or primary cultures, which are complex systems with multiple receptor subtypes present. More recently, individual subtypes have been expressed in CHO or baby hamster kidney (BHK) cells allowing examination of the agonist profiles of individual receptors as described above. Among other findings, these studies have demonstrated that L-AP4 is the most potent agonist of mGluR4 and mGluR6 but has little effect on the other receptor subtypes (26, 37, 67). Similarly, L-CCG-I was shown to activate mGluR2 at concentrations that have little effect on mGluR1 and mGluR4 (19). Two promising agonists which have recently been developed and may have some selectivity for specific receptor subtypes are (2S,1¢R,2¢R,3¢R)-2-(2,3dicarboxy cylopropyl)glycine [DCG-IV (59)] and (-)trans-azetidine-2,4-dicarboxylic acid [t-ADA (16)].

Studies of the physiological roles of metabotropic receptors have long been hampered by a lack of effective antagonists. Unlike the overlap with ionotropic receptors in agonist specificity, the mGluRs do not appear to share antagonists with any other glutamate receptor. L-2-Amino-3-phosphonopropionic acid (L-AP3), which has been used most extensively, typically displays a pharmacological action characteristic of a weak partial agonist, but its efficacy varies widely between preparations (54). Recently, a promising series of phenylglycine derivatives, e.g., a-methyl-4-carboxyphenylglycine (MCPG) have been developed which have been shown to competitively inhibit ACPD-stimulated PI hydrolysis and antagonize ACPD-induced depolarization (4) (). Although they have yet to be fully characterized, it appears that some of these analogues may differentially antagonize individual receptor subtypes.

A substantial body of evidence now exists indicating that mGluRs have important roles in development and plasticity. For example, the developmental peak of EAA-stimulated PI hydrolysis occurs between 6 and 12 days of age in neonatal rats and exhibits a high correlation with periods of intense synaptogenesis (39). More direct evidence for a role of mGluRs in plasticity is provided by studies of LTP. Bath application of trans-ACPD can produce LTP in the absence of concomitant tetanic stimulation (5) or can potentiate LTP when applied in conjunction with tetanic stimulation (33). Most recently it has been shown that the newly characterized metabotropic antagonist MCPG can block the induction of LTP without affecting baseline synaptic transmission or previously established LTP (3).

Evidence indicates that activation of mGluRs can have both neuroprotective and neurotoxic effects. Koh et al. (25) demonstrated that trans-ACPD can attenuate NMDA-induced excitotoxicity using a murine cortical culture system. Although other neuroprotective actions of trans-ACPD have been subsequently demonstrated both in vitro and in vivo (11, 41, 48, 61), some recent studies have indicated neuropathological effects of mGluR agonists (28, 32, 53). Whether the net effect of an mGluR agonist is neuroprotective or neuropathological may depend upon the relative expression of specific metabotropic receptor subtypes coupled to either excitatory or inhibitory transduction mechanisms. Expression of the subtypes is differentially regulated according to region, cell type, and developmental state, and the ratios of receptors expressed may represent homeostatic mechanisms that have roles in determining the balance between plasticity and pathology.

While most of the attention and effort aimed at the study of the EAA system has focused specifically on the neurotransmitter receptors, an integrated model of the synapse should include the other steps in synaptic transmission, such as transmitter inactivation. The rapid removal of glutamate, and related transmitters, from the synaptic cleft by high-affinity uptake is thought to contribute to (a) the termination of the excitatory signal, (b) the maintenance of extracellular glutamate levels below those that could induce excitotoxic damage, and (c) the recycling of the transmitter via the glutamine cycle (18, 51, 55, 57).

The uptake process, as well as the potential consequences of decreases in transport capacity, is particularly significant in view of the excitotoxic properties of Lglutamate. This inverse relationship between transport capacity and excitotoxic sensitivity is supported by both in vivo and in vitro studies in which a reduction of transport capacity is associated with a concomitant increase in the damage produced by EAA agonists (51). Interestingly, decreases in the binding density of the transport substrate 3H-D-aspartate have been reported in Alzheimer's disease (43), while the maximum uptake velocity (Vmax) of glutamate is reduced in synaptosomes prepared from the spinal cords of patients with amyotrophic lateral sclerosis (ALS) (52). Such deficits may ultimately contribute to the excitotoxic vulnerability of neurons in these diseases.

Analogous to the heterogeneity found within the receptor systems, accumulating evidence indicates that there are a number of distinct EAA transporters. Currently, these uptake systems can be distinguished on the basis of: (a) their ionic dependency, Na+ versus Cl- versus Ca2+; (b) cell type, neuronal versus glial; (c) anatomical location, forebrain versus cerebellar; (d) substrate, glutamate versus homocysteate; and (e) cell organelle, synaptosome versus synaptic vesicle. Among these various systems, the Na+-dependent high-affinity transporter appears to be the dominant species and is the most thoroughly characterized in regard to pharmacology, distribution, mechanism, and molecular biology (10, 22, 47, 49, 65).

Three glutamate transport gene (22, 47, 65) have been identified, each of which share the common features of: (a) being expressed in brain, (b) lacking apparent signal sequences, (c) containing carbohydrate moieties, and (d) having molecular weights of about 57–64 kD. Importantly, when the proteins are pharmacologically characterized in expression systems, they demonstrate a strong Na+-dependency, are appropriately enantioselective (i.e., D-aspartate, L-aspartate and L-glutamate are substrates, whereas D-glutamate is not), and are inhibited by well-characterized uptake blockers, such as dihydrokainate, b-threo-hydroxy-aspartate, and L-trans-2,4-pyrrolidine dicarboxylate (L-trans-2,4-PDC). Two of these proteins appear to be of glial origin, while the third is reported to be found in neurons, although each exhibits a specific distribution in the CNS.

Sequence analysis demonstrates that while the three transporters share about 50% homology among themselves, they exhibit little homology with any other eukaryotic protein, including the superfamily of neurotransmitter transporters that mediate the uptake of GABA, noradrenaline, serotonin, dopamine, glycine, and choline (for review see ref. 71). Examination of the sequences of each of the three proteins has led to the identification of both glycosylation sites and potential phosphorylation sites, as well as the prediction that each contains a distinct number of membrane-spanning regions (i.e., six, eight, and ten).

Early studies of the pharmacology of the Na+dependent high-affinity glutamate uptake system identified the basic specificity of the system by quantifying the ability of a large number of glutamate analogues to inhibit the uptake of radiolabeled substrates (e.g., 3H-L-glutamate and 3H-D-aspartate) in a variety of preparations (e.g., synaptosomes, cultured astrocytes, tissue slices). In this manner, it was demonstrated that uptake inhibitors generally share the common features of being a-amino acids with a second acidic group separated from the a-COOH by 2–4 methylene groups. Structure–activity studies indicate that the distal COOH group can be derivatized to a hydroxamate (e.g., L-aspartate-b-hydroxamate) or replaced by a sulfonate group (e.g., cysteic acid). Some modification of the carbon backbone is also tolerated, because b-THA and DHK are well-known competitive inhibitors. The system also exhibits enantioselectivity, because D-glutamate is a very weak inhibitor, although L- and D-asparate are excellent substrates.

A major advance in the pharmacological characterization of this system came with the identification of transport inhibitors that are conformationally restricted analogues of L-glutamate, such as L-trans-2,4-PDC (7), L-CCG-III (60), and cis-1-aminocyclobutane-1,3-dicarboxylate (CACB) (17). In contrast to the conformational flexibility of the acyclic molecules, these more rigid analogues are considerably more useful in defining the pharmacophore of the transporter binding site and, in the long term, developing selective transport inhibitors that do not cross-react with the other EAA receptors. Toward this goal, L-trans-2,4-PDC has been shown to be one of the most potent uptake blockers identified, yet it exhibits little or no activity at the NMDA, KA, or AMPA receptors (7).

With the aid of molecular modeling, comparisons have been made between energetically stable envelope conformations of L-trans-2,4-PDC and preferred staggered conformers of L-glutamate. Using this approach, a folded conformation of L-glutamate has been identified that exhibits a high degree of functional group overlap with L-trans-2,4-PDC (6, 10). The specific L-trans-2,4-PDC conformer is one in which the distal carboxyl group occupies an axial-like position at the flap of the pyrrolidine envelope. The L-glutamate conformer, in turn, corresponds to the relatively abundant folded conformation found in solution. Additional modeling studies with CCG-III and CACB further supported this arrangement of functional groups, because significant overlap was also observed with these conformationally restricted uptake blockers (6). These results support the hypothesis that a partially folded conformer of glutamate represents the active conformation at the Na+-dependent transporter and provides a firm basis for further modeling of the binding site pharmacophore and the design of new inhibitors.

As discussed in this chapter, the EAA receptors are a complex class with heterogeneous ligand specificity and subtle differences in the time course and magnitude of current flow even within each subtype. This specificity is created by various combinations of subunits, alternative splicing, and RNA editing. In the case of the ionotropic receptors there are at least 16 cDNAs identified, whereas in the metabotropic class there are over six subtypes, with the number still growing. Their varied distributions suggest that different combinations of EAA receptor and transporter subtypes are essential for functional specialization of individual cell types and at individual synapses. Recent advances in approaches to pharmacological examination of the EAA transmitter system will continue to allow a progressively more detailed understanding of these functions and how they can be modified during dysfunction.

published 2000