Chapter 16:  GABA and Glycine

GABA is formed in vivo by a metabolic pathway referred to as the GABA shunt

The GABA shunt is a closed-loop process with the dual purpose of producing and conserving the supply of GABA. GABA is present in high concentrations (millimolar) in many brain regions. These concentrations are about 1,000 times higher than concentrations of the classical monoamine neurotransmitters in the same regions. This is in accord with the powerful and specific actions of GABAergic neurons in these regions. Glucose is the principal precursor for GABA production in vivo, although pyruvate and other amino acids also can act as precursors. The first step in the GABA shunt is the transamination of α-ketoglutarate, formed from glucose metabolism in the Krebs cycle by GABA α-oxoglutarate transaminase (GABA-T) into l-glutamic acid [4] (Fig. 16-1). Glutamic acid decarboxylase (GAD) catalyzes the decarboxylation of glutamic acid to form GABA. GAD appears to be expressed only in cells that use GABA as a neurotransmitter. GAD, localized with antibodies or mRNA hybridization probes, serves as an excellent marker for GABAergic neurons in the CNS. Two related but different genes for GAD have been cloned, suggesting independent regulation and properties for the two forms of GAD: GAD₆₅ and GAD₆₇. Furthermore, expression of GAD and some GABA receptor subunits has been demonstrated in some non-neural tissues, indicating the likely function of GABA outside of the CNS [5]. GABA is metabolized by GABA-T to form succinic semialdehyde. To conserve the available supply of GABA, this transamination generally occurs when the initial parent compound, α-ketoglutarate, is present to accept the amino group removed from GABA, reforming glutamic acid. Therefore, a molecule of GABA can be metabolized only if a molecule of precursor is formed. Succinic semialdehyde can be oxidized by succinic semialdehyde dehydrogenase (SSADH) into succinic acid and can then reenter the Krebs cycle, completing the loop.

GABA release into the synaptic cleft is stimulated by depolarization of presynaptic neurons. GABA diffuses across the cleft to the target receptors on the postsynaptic surface. The action of GABA at the synapse is terminated by reuptake into both presynaptic nerve terminals and surrounding glial cells. The membrane transport systems mediating reuptake of GABA are both temperature- and ion-dependent processes. These transporters are capable of bidirectional neurotransmitter transport. They have an absolute requirement for extracellular Na⁺ ions with an additional dependence on Cl⁻ ions. The ability of the reuptake system to transport GABA against a concentration gradient has been demonstrated using synaptosomes. Under normal physiological conditions, the ratio of internal to external GABA is about 200. The driving force for this reuptake process is supplied by the movement of Na⁺ down its concentration gradient [6] (see Chap. 5). GABA taken back up into nerve terminals is available for reutilization, but GABA in glia is metabolized to succinic semialdehyde by GABA-T and cannot be resynthesized in this compartment since glia lack GAD. Ultimately, GABA can be recovered from this source by a circuitous route involving the Krebs cycle [4]; GABA in glia is converted to glutamine, which is transferred back to the neuron, where glutamine is converted by glutaminase to glutamate, which re-enters the GABA shunt (see Chap. 15).

The family of GABA transporters is a set of 80-kDa glycoproteins with multiple transmembrane regions; they have no sequence homology with GABA receptors. Pharmacological and kinetic studies have suggested a variety of subtypes, and at least six separate but related entities have been demonstrated by molecular cloning [6,7]. This has led to rapid developments in understanding the localization, pharmacological specificity, structure—function and mechanism of GABA transport.

GABA receptors have been identified electrophysiologically and pharmacologically in all regions of the brain

Because GABA is widely distributed and utilized throughout the CNS, early GABAergic drugs had very generalized effects on CNS function. The development of more selective agents has led to the identification of at least two distinct classes of GABA receptor, GABA_A and GABA_B. They differ in their pharmacological, electrophysiological and biochemical properties. Electrophysiological studies of the GABA_A-receptor complex indicate that it mediates an increase in membrane conductance with an equilibrium potential near the resting level of −70 mV. This conductance increase often is accompanied by a membrane hyperpolarization, resulting in an increase in the firing threshold and, consequently, a reduction in the probability of action potential initiation, causing neuronal inhibition. This reduction in membrane resistance is accomplished by the GABA-dependent facilitation of Cl⁻ ion influx through a receptor-associated channel. On the other hand, increased Cl⁻ permeability can depolarize the target cell under some conditions of high intracellular Cl⁻. This in turn potentially can excite the cell to fire or to activate Ca²⁺ entry via voltage-gated channels and has been proposed as a physiologically relevant event, especially in embryonic neurons.

Electrophysiological data [8] suggest that there are two GABA-recognition sites per GABA_A-receptor complex. An increase in the concentration of GABA results in an increase in the mean channel open time due to opening of doubly liganded receptor forms, which exhibit open states of long duration. It has been demonstrated, using a membrane preparation from rat brain, that the increase in the ionic permeability of the GABA_A receptor complex is transient in the continuing presence of agonist [9]. This phenomenon is known as desensitization and is rapidly reversible. The molecular mechanism of desensitization is not understood, and various hypotheses remain under investigation. The existence of GABA-binding sites specific for the initiation of desensitization and distinct from sites mediating opening of the Cl⁻ channel has been proposed [9].

GABA_B receptors, which are always inhibitory, are coupled to G proteins

Less is known about the GABA_B receptor, primarily due to the limited number of pharmacological agents selective for this site. Originally, GABA_B receptors were identified by their insensitivity to the GABA_A antagonist bicuculline and certain GABA_A-specific agonists [1,10]. The GABA analog ( − )baclofen (β-(4-chloro-phenyl)-γ-aminobutyric acid) was found to be a potent and selective GABA_B agonist.

GABA_B receptors are coupled indirectly to K⁺ channels. When activated, these receptors can decrease Ca²⁺ conductance and inhibit cAMP production via intracellular mechanisms mediated by G proteins. GABA_B receptors can mediate both postsynaptic and presynaptic inhibition. Presynaptic inhibition may occur as a result of GABA_B receptors on nerve terminals causing a decrease in the influx of Ca²⁺, thereby reducing the release of neurotransmitters. The cloning of the GABA_B receptor [11] and its structural similarity to the metabotropic glutamate receptors should allow rapid progress in the pharmacological characterization of receptor subtypes and the development of new drugs of improved selectivity. Pharmacological responses to GABA that are insensitive to both bicuculline and baclofen have been termed GABA_C receptors. Some, but not all, of these responses can be explained by a structural analog of GABA_A receptors, the ρ subunit [10].

The GABA_A receptor is part of a larger GABA/drug receptor—Cl⁻ ion channel macromolecular complex

The complex includes five major binding domains (Fig. 16-2). These include binding sites localized in or near the Cl⁻ channel for GABA, benzodiazepines, barbiturates and picrotoxin as well as binding sites for the anesthetic steroids. These binding domains modulate receptor response to GABA stimulation. In addition, other drugs, including volatile anesthetics, ethanol and penicillin, have been reported to have an effect on this receptor [3,8]. An integral part of this complex is the Cl⁻ channel. The GABA-binding site is directly responsible for opening the Cl⁻ channel. A variety of agonists bind to this site and elicit GABA-like responses. One of the most useful agonists is the compound muscimol, a naturally occurring GABA analog isolated from the psychoactive mushroom Amanita muscaria. It is a potent and specific agonist at GABA_A receptors and has been a valuable tool for pharmacological and radioligand-binding studies [10,12]. Other GABA agonists include isoguvacine, 4,5,6,7-tetrahydroisoxazolo-[5,4-c]py-ridin-3-ol (THIP), 3-aminopropane-sulfonate and imidazoleacetic acid [12]. The classical GABA_A-receptor antagonist is the convulsant bicuculline, which reduces current by decreasing the opening frequency and mean open time of the channel [8,10]. It is likely that bicuculline produces its antagonistic effects on GABA_Areceptor currents by competing with GABA for binding to one or both sites on the GABA_A receptor.

The GABA_A receptor is the major molecular target for the action of many drugs in the brain

Among these are benzodiazepines, intravenous and volatile anesthetics and possibly ethanol. Benzodiazepine receptor-binding sites copurify with the GABA-binding sites [13]. In addition, benzodiazepine receptors are immunoprecipitated with antibodies that were developed to recognize the protein containing the GABA-binding site [14]. This indicates that the benzodiazepine receptor is an integral part of the GABA_A receptor—Cl⁻ channel complex.

Benzodiazepine agonists represent the newest group of agents in the general class of depressant drugs, which also includes barbiturates, that show anticonvulsant, anxiolytic and sedative—hypnotic activity. Well-known examples include diazepam and chlordiazepoxide, which often are prescribed for their anti-anxiety effects [1]. The mechanism of action of benzodiazepine agonists is to enhance GABAergic transmission. From electrophysiological studies, it is known that these benzodiazepines increase the frequency of channel opening in response to GABA, thus accounting for their pharmacological and therapeutic actions [8]. In addition, the benzodiazepine site is coupled allosterically to the barbiturate and picrotoxin sites [2]. Benzodiazepine receptors are heterogeneous with respect to affinity for certain ligands. A wide variety of nonbenzodiazepines, such as the β-carbolines, cyclopyrrolones and imidazopyridines, also bind to the benzodiazepine site.

Barbiturates comprise another class of drugs commonly used therapeutically for anesthesia and control of epilepsy. Phenobarbital and pentobarbital are two of the most commonly used barbiturates. Phenobarbital has been used to treat patients with epilepsy since 1912. Pentobarbital is also an anticonvulsant, but it has sedative side effects. Barbiturates at pharmacological concentrations allosterically increase binding of benzodiazepines and GABA to their respective binding sites [2]. Measurements of mean channel open times show that barbiturates act by increasing the proportion of channels opening to the longest open state (9 msec) while reducing the proportion opening to the shorter open states (1 and 3 msec), resulting in an overall increase in mean channel open time and Cl⁻ flux [8].

Channel blockers, such as the convulsant compound picrotoxin, cause a decrease in mean channel open time. Picrotoxin works by preferentially shifting opening channels to the briefest open state (1 msec). Thus, both picrotoxin and barbiturates appear to act on the gating process of the GABA_A receptor channel, but their effects on the open states are opposite to each other. Experimental convulsants like pentylenetetrazol and the cage convulsant t-butyl bicyclophosphorothionate (TBPS) act in a manner similar to picrotoxin, preventing Cl⁻ channel permeability. The antibiotic penicillin is a channel blocker with a net negative charge. It blocks the channel by interacting with the positively charged amino acid residues within the channel pore, consequently occluding Cl⁻ passage through the channel [8].

There have been numerous studies on the role of GABA_A receptors in anesthesia. A considerable amount of evidence has been compiled to suggest that general anesthetics, including barbiturates, volatile gases, steroids and alcohols, enhance GABA-mediated Cl⁻ conductance. A proper assessment of this phenomenon requires not only a behavioral assay of anesthesia but also in vitro models for the study of receptor function. In this regard, not only electrophysiological methods but also neurochemical measurements of Cl⁻ flux and ligand binding have been useful. For example, a strong positive correlation exists between anesthetic potencies and the stimulation of GABA-mediated Cl⁻ uptake. This is seen with barbiturates and anesthetics in other chemical classes [3].

Comparison of ligand-gated ion channels that vary in sensitivity to anesthetic modulation, using the chimera and site-directed mutagenesis approach, has identified two amino acids in the membrane-spanning domains that are critical for anesthetic sensitivity [14a]. Direct evidence of ethanol augmentation of GABA_A receptor function, measured either by electrophysiological techniques or agonist-mediated Cl⁻ flux, has been reported [3,15]. The similarity between the actions of ethanol and sedative drugs such as benzodiazepines and barbiturates that enhance GABA action suggests that ethanol may exert some of its effects by enhancing the function of GABA_A receptors. Ethanol potentiation of GABA_A receptor function appears to be dependent upon the cell type tested and the method of assay. This suggests that the ethanol interaction may be specific for certain receptor subtypes and/or that it may be an indirect action [3].

Neurosteroids, which may be physiological modulators of brain activity, enhance GABA_A receptor function

This enhancement by steroids involves direct action on the membrane receptor protein rather than through the classical genomic mechanism mediated by soluble high-affinity cytoplasmic steroid hormone receptors (see Chap. 49). Chemically reduced analogs of the hormones progesterone and corticosterone derivatives administered to animals and humans exert sedative—hypnotic and anti-anxiety effects. This led to the development of a synthetic steroid anesthetic, alphaxalone. These neuroactive steroids are potent modulators of GABA_A-receptor function in vitro [2,3,16]. The neuroactive steroids can be produced in the brain endogenously and may influence CNS function under certain physiological or pathological conditions. Some observations suggesting that neurosteroids physiologically affect the CNS include the rapid behavioral effects of administered steroids; diurnal and estrous cycle effects on behavior; gender-specific pharmacology, especially of GABAergic drugs; and the development of withdrawal symptoms following cessation of chronically administered steroids. Neuroactive steroids have effects similar to those of barbiturates in that they enhance agonist binding to the GABA site and allosterically modulate benzodiazepine and TBPS binding [2,3]. Also, like barbiturates, high concentrations of neurosteroids directly activate the GABA_A receptor Cl⁻ channel. These observations led to the hypothesis that the neurosteroid-binding site may be similar to the barbiturate site, but the sites of action for the two classes of drugs are clearly not identical [2].

A family of pentameric GABA_A-receptor protein subtypes exists

The GABA_A receptor was first cloned using partial protein sequence, and verification of these cDNAs as GABA-receptor subunits was made by expression in Xenopus oocytes of GABA-activated channels [17]. Our current understanding of the molecular structure of the GABA_A receptor—ionophore complex is that it is a heteropentameric glycoprotein of about 275 kDa composed of combinations of at least 17 different but closely related polypeptides. The subunits are 50 to 60 kDa and have about 20 to 30% sequence identity between classes α, β, γ, δ, ϵ and ρ [8,14,18–20] (Fig. 16-3A). Variants of most of these subunits have been reported in vertebrate brain, including α_1–6, β_1–4, γ_1–4, δ, ϵ and ρ_1–3. β₄ and γ₄ so far have been identified only in birds. About 70% sequence identity is shared between the isoforms of each subunit class (Fig. 16-3A). This suggests that the genes probably evolved from a common ancestral sequence; the evolutionary relationships are shown in the dendrogram of Figure 16-3B.

Sequencing revealed that the GABA_A receptor is a member of a superfamily of ligand-gated ion channel receptors

This family includes the nicotinic acetylcholine receptor, the strychnine-sensitive glycine receptor and the serotonergic 5-HT₃ receptor. About 10 to 20% homology exists among GABA_A subunits and those of other members of the ligand-gated ion channel gene superfamily. In addition, splice variants exist for several of the subunits. In several cases, these variants provide phosphorylation substrates on the intracellular domain [20].

Differential distribution of GABA_A-receptor-subunit mRNAs and polypeptides in brain is consistent with data indicating variation in physiological function, pharmacology and biochemistry of different brain regions. It is likely that different combinations with differing pharmacologies and conductances are expressed in different neuronal populations. The subunit composition of native isoforms has been deduced by a combination of determining which polypeptides are present in a given cell, which ones can be isolated together as an oligomer using subunit-specific antibodies and what pharmacological properties can be reconstituted from recombinant subunits of known combinations. Table 16-1 summarizes the most abundant isoforms identified, their localization and their unique pharmacological properties [14,18–20]. The ρ subunits are expressed primarily, if not exclusively, in the retina, where they appear to form Cl⁻ channels, possibly homomers, with novel pharmacology, notably insensitivity to baclofen, bicuculline and GABA_A-positive modulators like benzodiazepines and anesthetics. This has led some to designate these ρ receptors as “GABA_C,” but structurally they are part of the GABA_A class [10]. The ϵ class produces receptors as heteromers with α and β subunits, resulting in another sort of novel pharmacology including low sensitivity to anesthetics [19].

Each GABA_A subunit contains four putative α-helical membrane-spanning domains (M1–M4) with a predominantly hydrophobic character. One or more membrane-spanning regions from each subunit, probably M2, form the walls of the channel pore (see also Chap. 11). The sequences of these transmembrane segments are highly conserved between the subunits of the GABA_A receptor as well as between members of the gene superfamily. The region between M3 and M4 contains a long, variable putative intracellular domain. This contributes to the subtype specificity and may participate in intracellular regulatory mechanisms such as phosphorylation and interaction with other cellular constituents [8,17,20].

Photoaffinity-labeling and site-directed mutagenesis of the GABA_A receptors suggest that the binding sites for benzodiazepines are localized at the interface of the α and γ subunits and that those for GABA ligands are located at the interface between the α and β subunits [21,22]. The binding pockets for each class of ligand appear to be formed from three loops of amino acids (Fig. 16-4). These models are consistent with studies on recombinant GABA_A receptors expressed in heterologous cells. Such studies show that the nature of the α and β subunits determines the pharmacological specificity at the GABA and benzodiazepine sites and that the γ subunits are necessary for sensitivity to benzodiazepines and insensitivity to Zn²⁺ inhibition [18]. Combinations including the α₁ subunit have a high affinity for certain “type 1-selective” benzodiazepine site ligands, while those with the α₂, α₃ and α₅ subunits have moderate affinity. The α₅ subunit has a unique specificity to bind most benzodiazepine ligands but not the sedative drug zolpidem [18]. Some GABA_A receptors apparently lack benzodiazepine-binding sites altogether or have a novel pharmacological profile at this site. Subunit combinations containing the α₄ or α₆ subunit with a γ subunit bind benzodiazepine inverse agonists but not agonists and are moderately sensitive to Zn²⁺ and neurosteroids. Combinations containing α₄ or α₆ with a δ subunit instead of a γ subunit do not bind benzodiazepine-site ligands, are highly sensitive to Zn²⁺ and are relatively insensitive to neurosteroids [8,18,20].

Cloning of the GABA_A receptor subunits and deduction of the corresponding amino acid sequences have led to the finding that phosphorylation sites for one or more kinases are present on virtually all of the subunits. Phosphorylation of the β subunits by cAMP-dependent protein kinase (PKA) and phosphorylation of β and γ subunits by protein kinase C and tyrosine kinase have been reported [8,23]. Current studies are directed toward an understanding of the functional consequences of phosphorylation of GABA_A receptors for both acute and more prolonged time frames.

Glycine is synthesized from glucose and other substrates in the brain

The immediate precursor of glycine is serine, which is converted to glycine by the activity of the enzyme serine hydroxymethyltransferase (SHMT). As for GABA, Ca²⁺-dependent release of glycine and specific postsynaptic receptors have been demonstrated. The action of glycine is terminated by its reuptake by a high-affinity transporter system. Synaptosomal uptake of radioactive glycine has been demonstrated in spinal cord and lower brainstem. In supraspinatal regions, glycine can be taken up by transport systems that have lower affinity and specificity for it. It is likely that a family of transporters will be identified for glycine, as for GABA and other neurotransmitters. The metabolic disposal is unclear, but it can be converted to a variety of other substances. However, none of these mechanisms has been associated with glycine neurons.

A number of amino acids can activate, to varying degrees, the inhibitory glycine receptor

The amino acids that can activate the glycine receptor include β-alanine, taurine, l-alanine, l-serine and proline. GABA is inactive at this receptor. There are only few known antagonists of glycine receptors. They include the plant alkaloid strychnine, which is highly selective for the glycine receptor, and the amidine steroid RU 5135, which is less selective. Both compounds bind the glycine receptor with nanomolar affinities. The binding site for glycine is thought to be related closely to the site for antagonist binding, but it may not be identical. Current findings suggest that at least three molecules of glycine are required to activate the glycine receptor. The physiological significance of multiple binding sites for glycine is unclear [24].

Glycine is inhibitory on ligand-gated, strychnine-sensitive Cl⁻ channel receptors but excitatory on N-methyl-d-aspartate receptors

Although acting as a classical neurotransmitter at inhibitory ion channel receptors, glycine is also an activating ligand at a class of excitatory ion channel receptors, the N-methyl-d-aspartate (NMDA) receptors (see Chap. 15). Currently, it is thought that glycine is a necessary cofactor for activation of the NMDA receptor by its neurotransmitter l-glutamate and that glycine is normally present in the extracellular space at suitable concentrations. Regulation of NMDA receptor activity via control of glycine concentrations can be considered more of a neuromodulatory role than a neurotransmitter role. It is curious that common amino acids like glycine and glutamate, which have other roles in metabolism, should be employed as signaling molecules as well and that glycine should be utilized as both an inhibitory and an excitatory signal in the nervous system. It is likely that receptors for signaling molecules evolved from prokaryotic proteins utilized for recognition of nutrients in the environment, including glycine, glutamate and GABA. GABA, a carbon- and nitrogen-storage molecule in plants and algae, appears to act as a colony-stimulating factor for abalone, an invertebrate mollusk.

Glycine receptors belong to the same gene superfamily as the GABA_A receptor

The native receptor is a macromolecular complex of about 250 kDa composed of a combination of two homologous polypeptides identified as α (48 kDa) and β (58 kDa). It has been proposed that the glycine receptor, like the other members of the family, consists of a quasisymmetrical pentameric arrangement forming a central ion pore [24,25] There is approximately 50% amino acid sequence identity between the α and β subunits; significant homology with GABA_A receptors is also evident (Fig. 16-3B). In addition, a 93-kDa polypeptide, gephyrin, copurifies with the glycine receptor. Photoaffinity labeling of the glycine receptor shows that the binding sites for glycine and strychnine are found on the 48-kDa polypeptide α subunit [24]. The α and β subunits span the postsynaptic membrane and are believed to be glycosylated. Like the GABA_A receptor and nicotinic acetylcholine receptor subunits, glycine receptor subunits have four hydrophobic segments, M1–M4, which probably span the lipid bilayer as α helices.

Unlike the α and β subunits, the 93-kDa polypeptide gephyrin is a highly hydrophilic protein, cytoplasmically localized at postsynaptic membranes. The 93-kDa polypeptide is a peripheral component anchoring the glycine receptor in the postsynaptic membrane by attaching it to cytoskeletal elements [25] (Fig. 16-5). Such a role appears analogous to that of the ankyrin family of proteins, which restrict lateral mobility of many membrane proteins, including transporters and channels. The 43-kDa protein rapsyn appears to play this cytoskeleton-linking role with acetylcholine receptors (see Chap. 11); gephyrin or related proteins also may be associated with GABA_A receptors.

Site-directed mutagenesis studies of glycine receptor subunits have led to a greater understanding of domains within the ligand-gated ion channel polypeptides that participate in pentamer assembly and interaction with the cytoskeleton, as well as the agonist-binding pocket and ion channel domains [25]. For example, α subunits can produce abnormal homopentamers in recombinant expression systems, while β subunits cannot. The α homomers are abnormal in channel properties and pharmacology in that they are sensitive to the GABA_A antagonist picrotoxin; coexpression of α and β subunits produces a more native receptor that is insensitive to picrotoxin, as are glycine receptors in real cells. Mutagenesis of certain amino acid residues in the extracellular domain of β subunits, making them similar to α subunits, confers the ability to assemble homomeric channels. Mutagenesis of other amino acid residues that differ between α and β subunits has identified domains involved in binding glycine and Zn²⁺ [25].

Currently, four isoforms of the α subunit, but only one form of β, have been cloned. The α₂ polypeptide represents a subunit primarily expressed neonatally, whereas α₃ is found primarily postnatally. Functional expression of the α₁, α₂ and α₃ transcripts in frog oocytes generates glycine-gated C1⁻ channels that are blocked by nanomolar concentrations of strychnine. Mutations of the α₁ subunit, namely leucine or glutamine for arginine at position 271, have been associated with hyperekplexia, a rare neurological disease characterized by an exaggerated startle response [26]. A variant of the α₂ subunit produces a channel that has a much lower affinity for strychnine and that is thought to correspond with the strychnine-insensitive glycine receptor found neonatally in rat spinal cord. Mutations in both α and β subunits have been implicated in two single-gene mutations in mice that show neurological phenotypes, as well as in one type of human genetic seizure disorder [25].

Immunocytochemical mapping with monoclonal antibodies raised against α-subunit antigen gives results similar to those obtained in autoradiographic studies of [³H]strychnine-binding sites. Although the majority of sites are found in spinal cord and brainstem, a small but significant population is found in more rostral brain regions. Interestingly, β-subunit mRNA is abundant in many brain regions, and even non-neural tissues, where neither [³H]strychnine binding nor known α-subunit mRNAs are found. The implication of this finding is presently unclear.