GABA and Glycine

Steven M. Paul

Amino acids are among the most abundant of all neurotransmitters present within the central nervous system (CNS). Studies which have characterized the high-affinity uptake of amino acids, in either brain slices or subcellular fractions, support current dogma that the majority of neurons in the mammalian brain utilize either glutamate or g-aminobutyric acid (GABA) as their primary neurotransmitters. In effect, GABA and glutamate serve to regulate the excitability of virtually all neurons in brain and, not surprisingly, therefore have been implicated as important mediators of many critical physiological as well as pathophysiological events that underlie brain function and/or dysfunction. Pharmacological studies utilizing drugs which selectively block or augment the actions of GABA or glutamate support the notion that these two neurotransmitters, by virtue of their often opposing excitatory and inhibitory actions, control, to a large degree, the overall excitability of the CNS. Thus, drugs which enhance inhibitory synaptic events mediated by GABA often decrease opposing excitatory events mediated by glutamate and vice versa (see Excitatory Amino Acid Neurotransmission, Schizophrenia and Glutamate: An Update, Amyotrophic Lateral Sclerosis, Glutamate, and Oxidative Stress, Potential Mechanisms of Neurologic Disease in HIV Infection, and Abuse and Therapeutic Use of  Benzodiazepines and Benzodiazepine-Like Drugs). The behavioral consequences of such pharmacologically induced changes in the "balance" between inhibition and excitation are often profound (e.g., following administration of convulsant or anesthetic drugs which are known to alter GABAergic or glutamatergic neurotransmission).

GABA and glycine are arguably the most important inhibitory neurotransmitters in the brain and brainstem/spinal cord, respectively. These inhibitory amino acids are of particular interest to the neuropsychopharmacologist, because many commonly studied (and therapeutically useful) drugs work by selectively affecting these two neurotransmitter systems. What follows is a review of these two inhibitory amino acid neurotransmitters, with an emphasis on the important role they play in mediating the actions of a variety of neuropsychopharmacologic agents.

In 1950, GABA was independently identified and reported to be present in the vertebrate brain by Roberts and Frankel (40) and by Awapara et al. (2). These investigators also demonstrated the presence of glutamic acid decarboxylase (GAD) in mouse brain and showed that active enzyme, capable of decarboxylating glutamate to GABA, required pyridoxal 5¢-phosphate (PLP) as cofactor (18). Early electrophysiological work carried out primarily in crustaceans firmly established GABA as an inhibitory neurotransmitter in invertebrates (34). Although GABA was originally shown to be present in very high (up to millimolar) concentrations in the vertebrate CNS, it proved considerably more difficult to unequivocally establish its role as a neurotransmitter in the mammalian brain. By the early 1970s, however, GABA had been shown to satisfy all of the classical criteria of a neurotransmitter (25, 39).

Part of the difficulty in establishing GABA's role as a neurotransmitter stemmed from the very widespread distribution of GABAergic neurons throughout the CNS (in contrast to more discretely localized and less abundant neurotransmitters such as the biogenic amines) and the lack of suitable reagents to positively identify GABAergic neurons. Following the purification of GAD and the generation of GAD antisera, immunohistochemical studies revealed that many (if not most) GABAergic neurons in brain are interneurons and are therefore uniquely able to alter the excitability of local circuits within a given brain region (39). From these (and other) studies it has been estimated that 30–40% of all CNS neurons utilize GABA as their primary neurotransmitter!

GABA is formed in vivo via a metabolic pathway called the GABA shunt. The initial step in this pathway utilizes a-ketoglutarate formed from glucose metabolism via the Krebs cycle. a-Ketoglutarate is then transaminated by a-oxoglutarate transaminase (GABA-T) to form glutamate, the immediate precursor of GABA. Finally, glutamate is decarboxylated to form GABA by the enzyme(s) glutamic acid decarboxylase (GAD) (18, 39). GAD is expressed only in GABAergic neurons and in certain peripheral tissues which are also known to synthesize GABA (see below). Like most neurotransmitters, GABA is stored in synaptic vesicles and is released in a Ca2+-dependent manner upon depolarization of the presynaptic membrane. Following release into synaptic cleft, GABA's actions are terminated principally by reuptake into presynaptic terminals and/or surrounding glia. GABA is also metabolized by GABA-T to form succinic semialdehyde. This transamination will regenerate glutamate when it occurs in the presence of a-ketoglutarate. Succinic semialdehyde is oxidized by succinic semialdehyde dehydrogenase (SSADH) to succinic acid which then reenters the Krebs cycle.

The reuptake of GABA occurs via highly specific transmembrane transporters which have recently been shown to be members of a large family of Na+-dependent neurotransmitter transporters. GABA uptake is temperature- and ion-dependent (both Na+ and Cl- ions are required for optimal uptake). Affinity purification of the GABA transporter protein has recently led to its molecular cloning (20). The principal neuronal GABA transporter appears to be a 70- to 80-kDa glycoprotein which, based on its deduced amino acid sequence, is predicted to contain 12 hydrophobic membrane-spanning domains. To date, at least two other GABA transporter cDNAs have been cloned (9). However, the physiological and pharmacological significance of this heterogeneity is unknown. Nonetheless, specific inhibitors of GABA uptake which directly bind to the transporter itself have been synthesized, and several have been shown to have anticonvulsant and/or antinociceptive properties in laboratory animals.

Unlike the other enzymes involved in GABA synthesis, GAD is expressed only in neurons and certain peripheral tissues which make or utilize GABA for signaling and/or endocrine functions (18). Early work strongly suggested the existence of at least two GAD enzymes which differed in their interaction(s) with PLP as well as in their subcellular distributions (18). Native GAD appears to exist as a dimer—probably a homodimer of two subunits of approximately 60 kDa each. GAD activity is quite high in brain, and it is now clear that approximately 50% of the enzyme(s) exists as apo-GAD (not bound to PLP) whereas the rest is bound to PLP (holo-GAD). Interestingly, there is also evidence that increased neuronal activity (e.g., that induced by depolarizing conditions) results in an increase in local GABA synthesis by promoting the association of PLP with apo-GAD to form active enzyme (18). Although the presence of two GAD isoforms was strongly supported by both biochemical and immunochemical data, their similarities and differences were not fully appreciated until both forms were cloned in an elegant series of studies by Tobin and colleagues (17, 18).

GAD is also expressed outside the CNS. For example, both GAD isozymes are present in b cells of the pancreatic islets where GABA is suspected to play a role in pancreatic endocrine function. (Immunohistochemical and lesion studies with b-cell toxins such as streptozotocin have shown that GAD and insulin coexist in the b cell.) In this regard, Baekkeskov et al. (3) have shown that antibodies to the 64-kDa form of GAD (which appears to be related to GAD65) occur in most, if not all, patients with insulin-dependent diabetes, and their presence appears to precede the clinical onset of disease. Autoantibodies to GAD may therefore underlie the development of insulin-dependent (type I) diabetes as well as that of the relatively rare neurological disorder known as stiff-man syndrome.

Receptors for both inhibitory and excitatory amino acid neurotransmitters are either ionotropic (i.e., their activation results in enhanced membrane ion conductance) or metabotropic (i.e., their activation results in increased intracellular levels of second messenger) in nature. GABAA receptors are ionotropic receptors leading to increased Cl- ion conductance, whereas GABAB receptors are metabotropic receptors which are coupled to G proteins and thereby indirectly alter membrane ion permeability and neuronal excitability (see below). Electrophysiological studies using voltage-clamp and single-channel recording techniques have yielded a rather detailed description and understanding of the operation of the GABAA receptor-gated Cl- ion channel (10, 28). Activation of the GABAA receptor by agonist results in an increase in Cl- ion conductance via the receptor-gated ion channel or pore. This increase in Cl- ion conductance, which requires the binding and cooperative interaction of two molecules of GABA, is actually due to an increase in the mean open time of the Cl- ion channel itself (28). (GABA activates the GABAA receptor at low micromolar concentrations, suggesting that it must be highly compartmentalized within nervous tissue.) The increase in Cl- ion conductance observed following activation of GABAA receptors results in a localized hyperpolarization of the neuronal membrane and therefore leads to an increase in the "threshold" required for excitatory neurotransmitters to depolarize the membrane in order to generate an action potential. This decrease in neuronal membrane "excitability" results in the inhibitory actions of GABA.

GABAA receptors, like most receptors, can be defined by the drugs (and other ligands) which selectively bind to, and either stimulate or block, receptor activity (). A variety of GABA receptor agonists have been discovered and have been shown to selectively activate GABAA receptors. Muscimol, a rigid GABA analogue isolated from the hallucinogenic mushroom Amanita muscaria, is one of the most selective and potent GABA agonists known. Muscimol is also not a substrate for the GABA transporter, which makes it useful for electrophysiological and biochemical studies. Both competitive and noncompetitive GABAA receptor antagonists have also been described (16). Bicuculline is the prototypical competitive antagonist and directly competes with GABA for binding to the receptor complex. Bicuculline reduces both the frequency and mean open time of the GABA-gated Cl- ion channel. Picrotoxin and other extremely potent cage convulsants such as t-butylbicyclophosphorothionate (TBPS) are noncompetitive GABA receptor antagonists which do not compete directly with GABA for its recognition site(s) but, instead, bind to a separate and distinct recognition site(s) associated with the receptor complex (44). Not surprisingly, both classes of GABAA receptor antagonists produce seizures when administered to laboratory animals. The affinity of cage convulsants such as TBPS for GABAA receptors is so high that they have proven to be useful radioligands for measuring GABAA receptors in vitro and for their subsequent biochemical and pharmacological characterization (44). These studies have revealed that GABAA receptors have multiple allosteric binding sites for drugs which, when occupied, modulate (positively or negatively) the inhibitory actions of GABA.

The observation that sedative–hypnotic drugs, which are classified behaviorally as CNS depressants, can augment the inhibitory properties of GABA was first established in 1975 for both benzodiazepines and barbiturates using electrophysiological techniques (23, 32). Benzodiazepines were discovered and developed in large measure to circumvent the potential lethal effects of barbiturates. It is most curious that benzodiazepines and barbiturates, which are structurally dissimilar and which were discovered with no knowledge of their underlying mechanisms of action, actually share the same molecular target(s).

In 1977, specific high-affinity receptors for benzodiazepines were discovered in the brains of many species, including man (30, 43). The excellent correlations between receptor affinity measured in vitro and the in vivo pharmacological potencies of a series of benzodiazepines strongly indicated that these receptors mediate most, if not all, of the pharmacological actions of benzodiazepines (30, 43). In the ensuing 17 years, the pharmacological significance of these receptors has been amply confirmed by many laboratories. It is now clear that all of the major centrally mediated actions of benzodiazepines—that is, their anxiolytic, anticonvulsant, muscle-relaxant, and sedative–anesthetic properties—are mediated by benzodiazepine receptors. Moreover, it has also been shown that the benzodiazepine receptor first demonstrated in 1977 is really a subtype of GABAA receptor (see below) (48, 49).

While both benzodiazepines and barbiturates bind to GABAA receptors to augment GABA-mediated responses, they do so in different ways. Barbiturates have dual actions to enhance GABAA receptor-mediated Cl- ion conductance (42, 45). At low (subanesthetic) concentrations, barbiturates augment the affinity of the GABAA receptor for GABA and increase the mean channel opening time induced by GABA. At higher (anesthetic) concentrations, barbiturates directly increase channel openings, even in the absence of GABA. Benzodiazepines, on the other hand, have no direct effects on channel opening but only increase the affinity of the receptor for GABA as well as the frequency of GABA-activated channel openings (28, 45). This is an important distinction because it means that benzodiazepines will markedly augment GABA's actions at low intrasynaptic GABA concentrations, but will have little to no effect at saturating concentrations of GABA (i.e., benzodiazepines "shift" the concentration–response curve for GABA slightly to the left) (54). These differences undoubtedly contribute to the relatively low toxicity of benzodiazepines compared to barbiturates.

There is now considerable evidence that a number of other sedative-hypnotic-anesthetic drugs also interact with GABAA receptors and at pharmacologically-relevant concentrations. Parenthetically, these studies (cited below) were among the first to suggest that the behavioral effects of alcohols and anesthetics were due, at least in part, to their "specific" actions at critical membrane protein targets, notably ligand-gated ion channels. (It had been widely assumed for over 100 years that the behavioral effects of alcohols and anesthetics were due to their "nonspecific" effects on membrane lipids.)

Ethanol, one of the most commonly used (and abused) sedative-hypnotic agents, has been shown by several investigatiors to augment GABA-activated Cl- ion conductance in a variety of intact and isolated neuronal membrane preparations (46). To date, however, electrophysiological studies of ethanol's actions in augmenting GABA-activated Cl- ion conductance have yielded somewhat mixed results (see ref. 31 for review). More recent studies have generally confirmed that pharmacologically-relevant concentrations of ethanol (10–100 mM) weakly (but significantly) augment GABA-activated Cl- ion conductance (31) as do longer chain-length alcohols and general anesthetics (31, 51). Moreover, several imidazobenzodiazepine inverse agonists of the benzodiazepine receptor (see below) have been reported to ùantagonizeú the sedative/ataxic effects of ethanol (47)—further implicating the GABAA receptor as one of the key central sites mediating at least some of ethanol's neuropharmacological effects. The effects of ethanol on GABAA receptors, coupled with its more recently described actions in inhibiting glutamate (NMDA) receptor-mediated depolarizing events (51), likely contribute to the anxiolytic and sedative effects of alcohols.

Since the discovery of the benzodiazepine receptor (recognition site), a large number of benzodiazepine and nonbenzodiazepine drugs have been found to interact with these receptors—and in unexpected ways. In addition to those receptor ligands which augment GABA responses (now called agonists), two other broad classes of ligands have now been characterized. Selective antagonists such as the imidazobenzodiazepine Ro15-1788 (flumazenil) bind with high affinity to GABAA receptors but are devoid of intrinsic activity of their own (22). However, these antagonists completely block the actions of benzodiazepine receptor agonists in augmenting GABA-mediated responses. Selective antagonists like flumazenil also block (or reverse) the actions of inverse agonists. The latter are benzodiazepine receptor ligands which decrease GABA-activated Cl- ion conductance (by decreasing the frequency of channel openings). The GABAA receptor can thus be positively or negatively modulated by compounds which range in activity from full agonists to full inverse agonists (22). Along this continuum lie compounds with different degrees of intrinsic efficacy—that is, compounds with only partial agonist or inverse agonist actions. Behaviorally, full agonists have sedative– anesthetic properties, whereas full inverse agonists are convulsants (14). Partial benzodiazepine receptor agonists (now in development by several pharmaceutical companies) may prove to be effective anxiolytics, devoid of the sedative effects generally observed with full agonists (22). The fact that benzodiazepine receptor agonists reduce anxiety and that inverse agonists are profoundly anxiogenic, coupled with recent observations that the "sensitivity" of animals to inverse agonists can be altered (in some cases increased) by pharmacological or environmental factors, has prompted considerable speculation that GABAA receptors are involved in at least some forms of human anxiety (see refs. 24 and 54 for reviews).

The large number of drug recognition sites associated with GABAA receptors (which are clearly distinct from those which recognize GABA itself) have led several investigators to propose the existence of endogenous receptor ligands. Several such "candidate" ligands have been identified; however, with the possible exception of two, there is little compelling evidence at present that any interact with GABAA receptors in vivo. One of these ligands is an endogenous peptide called diazepam-binding inhibitor (DBI), which was initially isolated by Guidotti et al. (21) and was shown to interact with GABAA receptors and to have anxiogenic properties (similar to inverse agonists). The other postulated endogenous ligand(s) include two natural reduced steroid metabolites of progesterone and deoxycorticosterone (allopregnanalone and allotetrahydro-DOC) (29). These neuroactive steroids bind with high affinity to GABAA receptors and have "barbiturate-like" actions in augmenting GABA-mediated responses (for review see ref. 35). The plasma and brain levels of these neuroactive steroids increase dramatically following exposure of rats to various stressors. Plasma allopregnanolone levels are also quite high during the third trimester of pregnancy, and they decrease dramatically following parturition (35). None of these putative natural ligands, however, have yet been unequivocally demonstrated to subserve any physiological function.

In 1987, Barnard, Seeburg, and colleagues (4, 41), using partial amino acid sequences from purified bovine brain GABAA receptors succeeded in cloning several of the subunits which comprise the GABAA receptor(s). The deduced amino acid sequences of the a- and b-subunit cDNAs isolated by these investigators indicated that each subunit was approximately 50–60 kDa in size and had four a-helical hydrophobic membrane-spanning sequences of approximately 20–30 amino acids. The predicted structure of the receptor was based on strong evidence that the GABAA receptor is a member of a large superfamily of ligand-gated ion channels which includes the nicotinic-cholinergic, ionotropic glutamate, and glycine receptors (there is approximately 10–20% sequence identity between members of this superfamily) (4, 33).

Currently, it is believed that, like the nicotinic-cholinergic receptor, the GABAA receptor is a heteropentameric glycoprotein of approximately 275 kDa (33) (). To date, five distinct classes of polypeptide subunits (a, b, g, d, and r) have been cloned and multiple isoforms of each have been shown to exist (e.g., there have been six a-subunit cDNAs isolated so far!) (15). There is approximately 70% sequence identity between the polypeptide subunits within a given class, but only approximately 30% between classes.

Although the exact subunit composition of most GABAA receptor(s) is unknown, it appears that their composition varies from brain region to region—and even between neurons within a given region. In situ hybridization studies (now complemented by immunocytochemical studies) have revealed, for example, that some a subunits (e.g., a1) are widely expressed throughout the brain whereas others are only expressed in discrete populations of neurons. Remarkably, a recently cloned a-subunit isoform (a6), which also confers unique pharmacology to recombinantly expressed GABAA receptors, is only expressed in a single neuron subtype—the cerebellar granule neuron (26).

What is the pharmacological and physiological significance of the surprising heterogeneity of GABAA receptor subunit isoforms expressed in brain? A few examples serve to illustrate the critical importance of subunit composition with respect to the pharmacological actions of drugs which, as previously discussed, work by interacting with GABAA receptors. Following the initial report describing the cloning and expression of a and b subunits, it was soon realized that coexpression of these subunits in various combinations reproduced many, but not all, of the properties of native GABAA receptors. The notable exception was the lack of a reproducible response to benzodiazepines when only a and/or b subunits were expressed. It is now clear that coexpression of an additional subunit, called g, is necessary to observe the potentiation of GABA responses by benzodiazepines that is characteristic of most native receptors (38). Moreover, coexpression of individual g-subunit variants (g1, g2, g3), which have now been identified (with a and b subunits), results in varying degrees of modulation by benzodiazepine receptor ligands (agonists, antagonists, inverse agonists). Photaffinity labelling studies suggest that the GABA binding site itself resides on the b subunit, while the benzodiazepine binding site resides on the a subunit () (33). Although these experiments have clearly delineated an important role for individual subunits (such as a and g) in determining the ligand-gating and pharmacological properties of GABAA receptors, it is still not entirely clear where each of the ligand binding sites resides on native GABAA receptors.

Although expression of the g subunit is essential for conferring the modulatory actions of benzodiazepines on recombinant GABAA receptors, it appears that a-subunit heterogeneity determines the diversity of physiological and pharmacological responses characteristic of native GABAA receptors (36, 37). When coexpressed with b1 subunits, for example, the a1 subunit yields a receptor with a relatively high affinity for GABA. (Recall that the a1 subunit is the most widely and abundantly expressed a subunit in brain.) By contrast, coexpression of the a2 or a3 subunits (with the b1 subunit) results in GABAA receptors with far lower affinities for GABA. Thus, the subunit composition of a given receptor may determine the local "response" to synaptically released GABA (27). There are also multiple forms of the b subunit expressed in brain (15, 27). Although their exact role in GABAA receptor function has yet to be determined, each contains a consensus sequence for phosphorylation by protein kinase A. There is some evidence that phosphorylation of the b subunit may result in receptor desensitization seen with continuous exposure to GABA.

Subunit heterogeneity seems also to be relevant to the pharmacological differences observed between drugs, such as the benzodiazepines, which interact with GABAA receptors. Receptors which are composed of a3 subunits (together with b1 and a2 subunits) yield much greater responses to benzodiazepines than do receptors which contain a1 or a2 subunits (27). Early work by Lippa and colleagues delineated pharmacologically distinct subtypes of GABAA receptors (type I versus type II) based on their affinity for CL 218,872 and their regional distribution in brain (see refs. 36 and 54) for discussion of those subtypes). Type I receptors had high affinity for CL 218,872 and were predominantly expressed in cerebellum, whereas type II had relatively low affinity for CL 218,872 and were enriched in the hippocampus. A combination of a1, b1, and g2 subunits results in type I receptors with high affinity for CL 218,872 (type I receptors are enriched in the cerebellum where a1 subunit mRNA is highly expressed) (36). Type I receptors seem to have a high affinity for sedative–hypnotic benzodiazepines as well as for the nonbenzodiazepine hypnotic zolpidem. Type II receptor pharmacology can be reproduced with receptors containing a2 or a3 subunits, and the transcripts for these subunits are expressed in the hippocampus (36). In retrospect, it is not surprising, given the structural simplicity of GABA, that the complexity and diversity of its many functions would require the evolution of a large and heterogeneous number of GABA receptors now known to be expressed in both neurons and glia throughout the brain.

While attempting to identify functional GABA receptors on peripheral nerve terminals, Bowery and Hudson (13) (see ref. 12 for review) described a bicucullineinsensitive action of GABA in reducing the release of [3H]norepinephrine. Subsequently, these investigators extended their findings to the CNS, where it became clear that GABA could also potently inhibit the depolarization-induced release of [3H]norepinephrine from brain slices. Early on, it also became apparent that many GABAA receptor (bicuculline-sensitive) agonists were unable to mimic the actions of GABA in inhibiting neurotransmitter release—leading to the proposal to divide GABA receptors into two subtypes, GABAA and GABAB. Moreover, one compound, b-p-chlorophenyl-GABA (baclofen), which was designed to be a centrally active GABA analogue (and which is still marketed as an antispastic agent), was found to be inactive at GABAA receptors but quite active at GABAB receptors (12). Therefore, baclofen proved to be the first selective GABAB receptor agonist and is still used extensively for characterizing GABAB receptors. Phaclofen, the phosphonic derivative of baclofen, is a selective, albeit weak, GABAB receptor antagonist. Several newer, more potent GABAB antagonists have been discovered; however, the published data on these compounds are rather limited (12). Nonetheless, administration of GABAB antagonists to laboratory animals does not result in the profound behavioral sequelae observed following administration of GABAA receptor antagonists (e.g., seizures). This suggests that GABAA receptors are tonically (and continuously) activated, whereas GABAB receptors may only be activated under certain physiological conditions.

Activation of GABAB receptors in many brain regions results in an increase in K+ channel conductance with a resultant hyperpolarization of the neuronal membrane (10, 12). This increase in K+ conductance is often blocked by pretreatment with pertussis toxin, indicating that many postsynaptic GABAB receptors are indirectly coupled to K+ channels through an intervening G protein (1). There is considerable evidence that a large proportion of GABAB receptors are coupled to G proteins, but there is also evidence that some presynaptic GABAB receptors may be directly linked to K+ channels. The fact that GABAB receptors are coupled to G proteins may also explain, in part, the reported effects of GABAB receptor agonists on Ca2+ conductance and secondarily neurotransmitter release (12). Very little data is available on the structure of the GABAB receptor. To date, attempts to clone the GABAB receptor. To date, attempts to clone the GABAB receptor by microsequencing a portion of the purified protein or by expression cloning in oocytes have proven to be unsuccessful.

Glycine is the major inhibitory neurotransmitter in the brainstem and spinal cord, where it participates in a variety of motor and sensory functions. Glycine is also present in the forebrain, where it has recently been shown to function as a coagonist at the N-methyl-D-aspartate (NMDA) subtype of glutamate receptor. In the latter, context glycine promotes the actions of glutamate, the major excitatory neurotransmitter (for a discussion of glycine's role as a coagonist of the NMDA receptor, see Excitatory Amino Acid Neurotransmission). Thus, glycine subserves both inhibitory and excitatory functions within the CNS.

Glycine is formed from serine by the enzyme serine hydroxymethyltransferase (SHMT). Glycine, like GABA, is released from nerve endings in a Ca2+-dependent fashion. The actions of glycine are terminated primarily by reuptake via Na+/Cl--dependent, high-affinity glycine transporters. The specific uptake of glycine has been demonstrated in the brainstem and spinal cord in regions where there are also high densities of inhibitory glycine receptors.

Recently, two glycine transporters have been cloned and shown to be expressed in the CNS as well as in various peripheral tissues (11, 19). These glycine transporters are members of the large family of Na+/Cl--dependent neurotransmitter transporters, and both share approximately 50% sequence identity with the GABA transporters discussed above. The deduced amino acid sequence of both cDNAs predicts the typical 12 transmembrane domains characteristic of these transporters. The two glycine cloned transporters have been named GLYT-1 and GLYT-2 in the order that they were reported (11). These transporter cDNAs are transcribed from the same gene and are quite similar in their 3¢ nucleotide sequences. They differ in their 5¢ noncoding regions as well as in the first 44 nucleotides of their coding sequence. Expression of GLYT-1 and GLYT-2 yield transporters with similar kinetic and pharmacological properties. Interestingly, however, the distribution of GLYT-1 and GLYT-2 transcripts measured by in situ hybridization are different. GLYT-1 mRNA also closely parallels the distribution of the glycine receptor. These data suggest that GLYT-1 is primarily a glial glycine transporter whereas GLYT-2 is primarily a neuronal transporter. The mapping of both glycine transporter mRNAs, as well as the glycine receptor subunit mRNAs, confirm the importance of this neurotransmitter in the brainstem and spinal cord, but support a more widespread distribution in supraspinal brain regions than was previously suspected.

Inhibitory glycine receptors are blocked by the plant alkaloid strychnine, which was also first used to label glycine receptors in spinal cord membranes (52, 53). Strychnine poisoning results in muscular contractions and tetany as a result of glycinergic disinhibition and overexcitation. Electrophysiological studies primarily carried out in rodent spinal cord neurons have demonstrated that glycine activates Cl- ion conductance (8). Like GABA, this increase in Cl- ion conductance results in a hyperpolarization of the neuronal membrane and an antagonism of other depolarizing stimuli. Other a- and b-amino acids, including b-alanine and taurine, also activate glycine receptors, but with lower potency (6, 8).

The glycine receptor was first successfully solubilized and purified by Betz and colleagues using affinity purification over an affinity matrix derivatized with aminostrychnine (8). The affinity-purified glycine receptor was shown to consist of two polypeptide subunits of approximately 48 kD (a) and 58 kD (b), respectively. Reconstitution of these polypeptide subunits into lipid vesicles resulted in functional receptors, and intramolecular cross-linking experiments suggested that the native glycine receptor is a pentameric structure. Photoaffinity labeling of the glycine receptor with [3H]strychnine revealed that both the strychnine and glycine binding sites are located on the 48-kD a subunit. Purification of the a-and b-receptor subunits was followed closely by their molecular cloning (7).

The deduced amino acid sequences of the a- and b-glycine-receptor subunits predict structures quite homologous to the subunits of other ligand-gated ion channels, including the GABAA receptor (7). Each subunit has four hydrophobic membrane-spanning sequences, and each shares considerable sequence identity with the other. Several glycine-receptor a-subunit variants have now been identified (a1–4), and, not surprisingly, they differ in their pharmacological properties and level of expression. As mentioned, both the agonist and antagonist binding sites are located on the a subunit, but at different amino acids (50). Interestingly, glycine receptors comprised of a1 subunits are efficiently gated by taurine and b-alanine, whereas a2-containing receptors are not (8). The a1 and a2 genes are expressed in the adult and neonatal brain, respectively. Interestingly, the b-subunit transcript is expressed at relatively high levels in the cerebral cortex and cerebellum, where no a transcripts or specific [3H]strychnine binding sites have been observed. Coexpression of b subunits with a subunits (as opposed to homo-oligomeric a-subunit glycine receptors) results in glycine receptors with pharmacological properties quite similar to native glycine receptors. Nonetheless, the widespread distribution of b-subunit mRNA in brain suggests that other, perhaps strychnine-insensitive glycine receptor isoforms will be found.

Recently, the expression of a1 and a2 subunits has been shown to be developmentally regulated with a switch from the neonatal a2 subunit (strychnine-insensitive) to the adult a1 form (strychnine-sensitive) at about 2 weeks postnatally in the mouse (8). The timing of this "switch" corresponds with the development of spasticity in the mutant spastic mouse (5), prompting speculation that insufficient expression of the adult isoform may underlie some forms of spasticity.

A convergence of scientific effort—involving molecular pharmacologists, molecular biologists, and medicinal chemists—has revealed a remarkable and, for the most part, unsuspected degree of complexity and heterogeneity in the biosynthetic enzymes, transporters, and receptors for GABA and glycine. For the neuropsychopharmacologist, GABA and glycine-containing and receptive neurons are of particular significance because they are among the best-characterized of all drug targets. Many psychoactive drugs which alter (increase or decrease) CNS excitability do so by effecting GABAergic or glycinergic neurotransmission. Some of these drugs (e.g., benzodiazepine and nonbenzodiazepine anxiolytic–hypnotics) are commonly prescribed for a variety of disorders. It is likely that the wealth of new information on GABA and glycine will result in an even better understanding of their potential role(s) in various neuropsychiatric disorders and in the discovery even more of effective therapeutic agents.

ACKNOWLEDGMENT

Dedicated to the memory of Daniel X. Freedman - friend, colleague, and mentor.

published 2000