Maintaining the correct balance in neuronal activation is of paramount importance to normal brain function. Imbalances due to changes in excitation or inhibition can lead to a variety of disorders ranging from the clinically extreme (e.g. epilepsy) to the more subtle (e.g. anxiety). In the brain, the most common inhibitory synapses are regulated by GABAA (γ-aminobutyric acid type A) receptors, a role commensurate with their importance as therapeutic targets. Remarkably, we still know relatively little about GABAA receptor biogenesis. Receptors are constructed as pentameric ion channels, with α and β subunits being the minimal requirement, and the incorporation of a γ subunit being necessary for benzodiazepine modulation and synaptic targeting. Insights have been provided by the discovery of several specific assembly signals within different GABAA receptor subunits. Moreover, a number of recent studies on GABAA receptor mutations associated with epilepsy have further enhanced our understanding of GABAA receptor biogenesis, structure and function.
- γ-aminobutyric acid type A receptor (GABAA receptor)
The balance of excitation and inhibition in the brain is a fundamental requirement for the smooth operation of all sensory, cognitive and motor functions. An imbalance in neuronal activity can result in a number of disease states such as anxiety, schizophrenia, autism, Tourette's syndrome and epilepsy. Indeed, a reduction in the expression of synaptic inhibitory receptors can lead to enhanced anxiety in mice . Importantly, this observation raises the possibility that reduced levels of synaptic GABAA (γ-aminobutyric acid type A) receptors may underlie anxiety disorders and natural variation in response to stress and threat cues in humans. At the extreme, gross imbalances may lead to permanent brain damage, as a result of excitotoxicity, a process that is thought to occur following head trauma and ischaemia and to contribute to neurodegenerative pathology.
Glutamate (excitatory) and GABA (inhibitory) are the most common neurotransmitters in the brain that act on ionotropic receptors to initiate rapid changes in membrane potential. The major classes of ionotropic glutamate receptors are the AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid), kainate and NMDA (N-methyl-D-aspartate) receptors. These receptors are discussed in the accompanying paper by Professor F.A. Stephenson ([1a]; see pp. 877–881) and will not be discussed further. Ionotropic GABA receptors belong to the cysteine-loop family of ligand-gated ion channels, whichs include nicotinic acetylcholine, 5-HT3 (5-hydroxytryptamine) and glycine receptors. Only one class of ionotropic GABA receptor (type A) exists in which receptors are constructed as a pentameric ion channel from a choice of 19 distinct subunits (α1−6, β1−3, γ1−3, δ, ϵ, ρ1−3, θ and π) generating an incredible scope (195/5, ∼500000 possibilities) for receptor diversity. However, only 20–30 functionally distinct receptor subtypes are thought to exist in vivo. Although no account has been made for the existence of functionally equivalent receptor subtypes that may exhibit distinct neuronal localization, trafficking itineraries (recycling, re-targeting or degradation following endocytosis) or modulation by second messengers , it is still clear that receptor construction is not random. Temporal and spatial restrictions on gene expression contribute to the limiting of receptor diversity. However, multiple GABAA receptor genes are often transcribed simultaneously. Thus a defined pathway(s) must exist at the level of receptor biogenesis, during its assembly into the pentameric ion channel.
GABAA receptor composition
Using transgenic ‘knock-in’ approaches to disable benzodiazepine modulation of specific GABAA receptor subtypes, it has been possible to dissect the distinct pharmacological (sedative, anxiolytic, anti-convulsant, muscle relaxant and the development of tolerance) contributions of specific α subunits [3,4]. Using similar approaches, specific β subunit contributions can be attributed to the pharmacological (anaesthesia and post-operative ataxia, sedation and hypothermia) responses to the anaesthetic, etomidate . This large body of research has highlighted the importance of receptor composition and opened our eyes to the potential for the development of highly selective therapeutic agents. It is important, therefore, to understand the molecular composition and structure of these distinct GABAA receptor targets.
It is known that the co-expression of α and β subunits is sufficient for the production of GABA-gated chloride currents and the incorporation of a γ subunit is required for modulation by benzodiazepines [2,6]. In support of the existence of defined pathways, GABAA receptor assembly appears to be strictly controlled, producing receptors with a fixed stoichiometry of two α, two β and one γ subunit [2,7]. At present, it is not known how the δ, ϵ, θ or π subunits are incorporated into GABAA receptors [8,9]. GABAA receptor localization to inhibitory synapses is driven by the γ subunit, with δ-containing receptors being found extrasynaptically . Actually, synaptic receptors exist as a dynamic pool that can interchange rapidly with extrasynaptic, γ-containing, receptors , a process that may enable rapid changes in synaptic inhibition.
Information regarding the order of subunits in the pentamer has been provided by studies elucidating the GABA- and benzodiazepine-binding sites in the receptor , by the forced assembly of concatenated subunits  and by homology to the acetylcholine receptor . The GABAA receptor contains two low-affinity GABA-binding sites, located at the interfaces between α and β subunits, and a single benzodiazepine site formed between the α and γ subunits (Figure 1). Thus the final structural organization of subunits is now established as -γ-(Bz)-α-(GABA)-β-α-(GABA)-β-  (Figure 1). Particularly intriguing is the observation that αβγ receptors are formed preferentially, at the expense of αβ receptor formation, when all three subunits are co-expressed . Thus not only are there strict minimal requirements for the production of GABAA receptors, but particular subunit combinations may be favoured and dominate receptor assembly.
GABAA receptor biogenesis
Achieving such a high fidelity requires a mechanism for rigorous quality control that is provided within the lumen of the ER (endoplasmic reticulum) where chaperone molecules such as calnexin, BiP (immunoglobulin heavy-chain binding protein) and protein-disulfide isomerase operate [6,16]. Together, these chaperone proteins monitor the fidelity of protein folding and assembly within the ER. On the other side of the membrane, the masking of ER retention signals during assembly ensures the cytoplasmic fidelity and transport competence of receptors [17,18].
Receptor biogenesis is facilitated by the use of assembly signals that direct specific subunit interactions [19–29]. Unassembled subunits fail to gain transport competence and are retained in the ER where they are degraded. A tantalizing glimpse into the potential for subunit specificity is provided by the α1 assembly signal, RQS (Arg-Gln-Ser), where a single amino acid (Arg66) has been shown to discriminate between assembly of α1 with β2 and β3 subunits. The basis for this discrimination appears to be the existence of a second assembly signal within β3, GKER (Gly-Lys-Glu-Arg) (see below), but not β2, that does not require the α1(Arg66) for the production of functional α1β3 cell-surface receptors . This is supported by a β2 construct containing the GKER signal that no longer requires the Arg66 in α1. Intriguingly, β3 lacking this GKER signal is still capable of assembling with α1(Arg66), suggesting the existence of an alternative signal within β3. Likewise, these observations suggest the possibility of a second assembly signal (other than RQS) within α1. However, the existence of the QS (Gln-Ser) component of this signal as a prerequisite for the assembly of α1 with either β2 or β3 subunits suggests a differential requirement of the same signal. Thus multiple assembly signals may be available for some, or all, subunits.
Speculatively, each signal might express distinct subunit preferences, and some hierarchical selection mechanism to determine which is used preferentially may exist. In this way, the existence of multiple assembly signals may permit plasticity in the construction of GABAA receptors, with the decision being dependent on subunit availability. It may not be possible to hold out for the prince, when surrounded by so many capable frogs that are in demand. If this is true, GABAA receptor subunits may not be committed to an innate assembly line, churning out identical copies, but may be required to make best choice decisions. Thus a neuron may select the ‘squad’ and leave the receptor subunits to select the best ‘team(s)’.
The existence of a fixed stoichiometry [2,7] and order within the pentamer , combined with the observation of hierarchical assembly , supports the hypothesis that this process follows a preferred strategy that is influenced by the availability of subunits. Indeed, this has been exemplified in a ‘knock-out’ mouse lacking most of the coding region of the α6 polypeptide. Interestingly, a part of α6, including the region thought to possess an assembly signal, was still expected to be present in the truncated protein and generated non-productive intermediates with the δ subunit, targeting them for degradation . Significant reductions in β2 (50%) and γ2 (40%), but not α1 (0%) or β3 (20%), were also observed. Thus a strict adherence to the assembly method appears to exist. In contrast, only when γ2 is unavailable, is γ3 able to take its place , highlighting the existence of plasticity.
Consistent with a requirement for these assembly signals to be located at (or near) inter-subunit contact points, the α/γ signals identified to date are located proximal (with respect to the primary structure) to the GABA- and benzodiazepine-binding sites formed at subunit interfaces (in the quaternary structure) between the α–β and α–γ subunits respectively (Figure 1). However, structural predictions of the quaternary structure of GABAA receptors do not predict the α1(RQS) signal to be located at subunit interfaces (Figure 2), suggesting that either a transient interaction occurs during receptor assembly, or that this signal plays a conformational role in the presentation of the subunit interface, or both. In contrast, both the β3(GKER)  and γ2(SYGY)  signals are located at subunit interface structures (Figure 2). In fact, the β3(GKER) signal is opposed to a signal [HDM (His-Asp-Met) in α1 and INM (Ile-Asn-Met) in γ2] (Figure 2) that is found 11 residues upstream from the RQS signal in α1. This is intriguing as this region lies within loop 2 that, along with loop 7 (cysteine-loop), is thought to participate in conformational changes involving their movement with respect to the TMII–TMIII (where TM is transmembrane region) and pre-TMI regions during receptor gating . That these residues, in either α1 or γ2, are required to direct receptor assembly has not been investigated.
GABAA receptors in epilepsy
A reduction in GABAergic inhibition has been implicated in epilepsy, with GABAA receptor antagonists promoting epileptic seizures, whereas agonists exhibit anticonvulsant activity . Furthermore, a number of GABAA receptor mutations associated with epilepsy have been discovered, each reported to cause a reduction in GABAergic inhibition [34–40]. These studies have provided valuable information on the involvement of particular residues in receptor biogenesis, structure and function.
Analyses of the extracellularly located γ2(R43Q) mutation associated with GEFS (generalized epilepsy with febrile seizures) [35–37,40] revealed an increase in receptor desensitization or a decrease in deactivation, number of ligand-binding sites or current amplitude. Recent studies have revealed a decreased surface expression due to subunit retention within the ER [36,44], suggesting that the primary fault lies in receptor biogenesis. In fact, this mutation leads to a failure to form the β2–γ2 interface, a role shared by this residue at all subunit interfaces . In support of this conserved role in receptor assembly, Arg43 is predicted to be located at each of the five subunit interfaces in the GABAA receptor (Figure 2).
The γ2(K289M) GEFS mutation [35,36,39,40] resides in the second extracellular domain, between TMII and TMIII. Importantly, this region, and notably the conserved lysine residue, has been reported to be involved in channel gating . The mutation has been reported to exhibit a reduced current amplitude, faster deactivation, reduced activation rate and a small decrease in channel opening times [35,36,39,40]. Despite the conservation of Lys289 in GABAA receptor subunits, the mutation had different effects when present in α1 (reduced channel opening times) or β2 (inhibiting receptor assembly), illustrating an asymmetric contribution of this residue to receptor structure and function . If this residue plays a role in receptor assembly, then it is restricted to the incorporation of the β2 subunit. Moreover, given the predicted location of this residue to the outside face of the channel (Figure 2), its involvement, like that of α1(RQS), would probably be transient.
The α1(A322D) mutation is associated with juvenile myoclonic epilepsy [34,35,37] and is located within TMIII and has been reported to decrease cell-surface expression and GABA potency. The mutant subunit is retained in the ER, where it is degraded. That α1(A322D) subunit stability is compromised even when expressed alone, combined with its predicted location facing away from the channel (Figure 2), suggests that this residue is critical for normal receptor structure or membrane stability. Again, this mutation exhibits asymmetric effects depending on which of the two α subunits is mutated. If the mutant α1 contributes to the γ2-(Bz)-α1-(GABA)-β2 interface, receptor stability, assembly and transport to the cell surface are significantly perturbed. In contrast, if the mutant α1 contributes to the β2-α1-(GABA)-β2, the mutation is better tolerated, although it causes rapid desensitization .
Mutations in the δ subunit are associated with GEFS (E177A) or juvenile myoclonic epilepsy (R220C) [36,38] and are both located within the first extracellular domain. Analysis of the mutants revealed reduced current amplitudes and cell-surface expression. However, upon co-expression with wild-type receptor subunits, the only significant effect appeared to be a reduction in channel opening times. Given that only one δ subunit is incorporated into pentameric receptors, a mixed receptor population must co-exist. Although Glu177 does not locate to subunit interfaces (Figure 2), it is present within the cysteine-loop and may be involved in channel gating  and/or receptor biogenesis or stability. How the R220C mutation might contribute to reduced GABAergic function remains unknown. It is not located at subunit interfaces (Figure 2), but may play a role in receptor biogenesis or stability.
In addition to a primary role in the genetic predisposition to epilepsy, a reduction in GABAA receptor function is also induced by epileptic activity as a result of receptor endocytosis [41,42]. Indeed, α1β2γ2(R43Q) and α1β2γ2(K289M), but not α1(A322D)β2γ2, receptors have been reported to exhibit impaired trafficking and/or accelerated endocytosis at raised temperatures that may be relevant to the febrile seizures associated with the γ2 GEFS pedigrees, but absent from the α1 pedigree, described above .
In summary, vast leaps in our understanding of GABAA receptor structure and its impact on function have been made, both as a result of direct attempts and the analysis of epilepsy mutations. However, given the asymmetric contributions of distinct subunits [36,39] and their position-dependency [34,45] within this structure, caution must be observed when drawing conclusions based on the structure of the acetylcholine receptor.
We are particularly grateful to Tim G. Hales (Department of Pharmacology and Physiology, George Washington University, Washington, DC, U.S.A.) for constructive comments on this paper, and Tenovus (Scotland) and The Anonymous Trust (Dundee) for their continued support.
Ion Channels: Communication across Membranes: A Focus Topic at BioScience2006, held at SECC Glasgow, U.K., 23–27 July 2006. Edited by G. Banting (Bristol, U.K.), C. Connolly (Dundee, U.K.), A. Dolphin (University College London, U.K.), J. Henley (Bristol, U.K.), A. Lai (Cardiff, U.K.) and C. Taylor (Cambridge, U.K.).
Abbreviations: ER, endoplasmic reticulum; GABA, γ-aminobutyric acid; GABAA, GABA type A; GEFS, generalized epilepsy with febrile seizures; TM, transmembrane region
- © 2006 The Biochemical Society