Chapter Seven - Allosteric Ligands and Their Binding Sites Define γ-Aminobutyric Acid (GABA) Type A Receptor Subtypes

https://doi.org/10.1016/bs.apha.2014.11.005Get rights and content

Abstract

GABAA receptors (GABAARs) mediate rapid inhibitory transmission in the brain. GABAARs are ligand-gated chloride ion channel proteins and exist in about a dozen or more heteropentameric subtypes exhibiting variable age and brain regional localization and thus participation in differing brain functions and diseases. GABAARs are also subject to modulation by several chemotypes of allosteric ligands that help define structure and function, including subtype definition. The channel blocker picrotoxin identified a noncompetitive channel blocker site in GABAARs. This ligand site is located in the transmembrane channel pore, whereas the GABA agonist site is in the extracellular domain at subunit interfaces, a site useful for low energy coupled conformational changes of the functional channel domain. Two classes of pharmacologically important allosteric modulatory ligand binding sites reside in the extracellular domain at modified agonist sites at other subunit interfaces: the benzodiazepine site and the high-affinity, relevant to intoxication, ethanol site. The benzodiazepine site is specific for certain GABAAR subtypes, mainly synaptic, while the ethanol site is found at a modified benzodiazepine site on different, extrasynaptic, subtypes. In the transmembrane domain are allosteric modulatory ligand sites for diverse chemotypes of general anesthetics: the volatile and intravenous agents, barbiturates, etomidate, propofol, long-chain alcohols, and neurosteroids. The last are endogenous positive allosteric modulators. X-ray crystal structures of prokaryotic and invertebrate pentameric ligand-gated ion channels, and the mammalian GABAAR protein, allow homology modeling of GABAAR subtypes with the various ligand sites located to suggest the structure and function of these proteins and their pharmacological modulation.

Introduction

γ-Aminobutyric acid (GABA) is the major inhibitory neurotransmitter in the nervous system (Curtis and Johnston, 1974, Roberts et al., 1975). GABA mediates most fast inhibitory neurotransmission, both synaptic and extrasynaptic, via the GABAA receptors (GABAARs) (Barnard et al., 1998, Enna, 1983, Olsen and Sieghart, 2008). GABAARs are ligand-gated chloride channels, members of the ligand-gated ion channel (LGIC) superfamily that includes the pentameric cys-loop receptors such as nicotinic acetylcholine receptors, inhibitory glycine receptors, and 5HT3 receptors (Dutertre et al., 2012, Galzi and Changeux, 1994, Sigel and Steinmann, 2012, Smart and Paoletti, 2012). GABA also mediates slower, modulatory neurotransmission involving GABAB receptors, G protein-coupled receptors (GPCRs) that regulate other receptors and voltage-gated ion channels involved in neurotransmission, some involving protein phosphorylation (Bowery et al., 2002). The two types of GABA receptor are not related to each other in structure or function (except both are mostly, but not entirely, inhibitory) and differ in their brain regional as well as subcellular localization.

Finally, there were “non-A, non-B” responses to GABA, insensitive to baclofen and bicuculline, sometimes called GABAC receptors, although they were likely a mixture of several discrete receptors (Johnston, 1997). Although insensitive to baclofen and not GPCRs, these responses were similar to those of GABAARs in being chloride ion dependent, and they have significantly different pharmacology including a less extended, more bent, conformation of GABA as the receptor-specific conformation, thus showing selectivity for a series of GABA analogues that are not potent on classical GABAR. Further, the GABACR have little or no sensitivity to a number of GABAAR PAMs (benzodiazepines (BZs), barbiturates, and neurosteroids) (Ng et al., 2011). The modern definition of receptor families is based on gene and thus protein sequence: several subunits (named ρ) that produce homopentameric chloride channels with function similar to GABAARs but GABAC pharmacology had homology with, and were thus classified as a subset of, GABAARs by the receptor nomenclature committee of the International Union of Pharmacology (Barnard et al., 1998, Olsen and Sieghart, 2008).

Sites for GABA, BZ, and picrotoxin allow identification and characterization of GABAA receptors (GABAARs). Historical pharmacology/ physiology/neurochemistry on different organs/central nervous system (CNS) regions and circuits demonstrate subtypes. Selective agonists or antagonists, including those for allosteric ligand sites (medicinal chemistry) were developed for tools: radioligand binding, affinity labeling, and affinity columns.

GABAARs, the subject of this chapter, are abundant in the mammalian CNS where they participate in all circuits and thus functions and behaviors, as well as most neuropsychiatric disorders (Rudolph & Möhler, 2014). GABAARs are most known for a role in, at least, epilepsy, anxiety and sleep disorders, substance abuse, and problems in learning and memory and sensorimotor processing (Olsen, 2014, Olsen et al., 1992, Rudolph and Möhler, 2004). Fortunately, there are about 2 dozen subtypes of GABAARs (Olsen & Sieghart, 2008) made from the heteromeric assembly of 19 highly homologous but distinct subunit genes (Lüddens et al., 1995, Rudolph and Möhler, 2004, Schofield et al., 1987, Whiting et al., 1995). Coupled with the fact that GABAARs are arguably the major molecular target for drugs acting on the brain, mostly positive allosteric modulators, PAM (Barnard et al., 1998, Christopoulos et al., 2014, Lüddens et al., 1995, Olsen, 1982, Olsen, 2014, Olsen and Sieghart, 2008, Rudolph and Möhler, 2006, Wick et al., 1998), the multiple subtypes afford the likelihood that these subtypes might be specifically modulated by drugs and thus function- and disease-specific pharmacological agents could be developed, if not already produced (Möhler, 2012, Whiting, 2003).

GABAARs were identified with radioligand binding assays for the three major drug sites on the protein: the GABA sites (agonist/antagonist), the picrotoxin sites (channel blocker), and the BZ sites (PAM). These three ligand sites will be covered as briefly as possible so I can concentrate on two other classes of site that I have worked on extensively: the anesthetic sites discovered biochemically by our development of the radioactive picrotoxin binding assay, and the high-affinity EtOH sites discovered as novel BZ binding sites on the δ subunit-containing GABAAR. GABAAR studies required an assay defining receptor-specific binding using a radioligand that is specific for the receptor being studied, or a nonradioactive (cold) ligand that is specific for the receptor so that whatever binding of a nonspecific ligand that is displaced by the specific ligand can be used as the “receptor-specific” binding. For example, acetylcholine, or GABA, the neurotransmitters themselves, will bind to many (> 7) proteins in brain/muscle, but one can count only the portion of the binding of acetylcholine, or a nonmetabolizable analogue (either one a nonspecific ligand), that is inhibited by the receptor-specific blocker α-bungarotoxin (Galzi & Changeux, 1994), or the fraction of GABA binding that is measured in sodium-free buffer (to eliminate transporter activity) and inhibited by the receptor-specific agonist muscimol or the receptor-specific antagonist bicuculline (Enna and Snyder, 1975, Greenlee et al., 1978, Krogsgaard-Larsen and Johnston, 1978, Zukin et al., 1974). Of course, one can alternatively use radiolabeled α-bungarotoxin or a closely related snake neurotoxin (Meunier et al., 1971) for the nicotinic acetylcholine receptor, or radiolabeled muscimol (specific agonist) (Beaumont, Chilton, Yamamura, & Enna, 1978) or bicuculline (specific antagonist) (Möhler and Okada, 1977b, Olsen et al., 1984) for the GABAAR, if one can obtain such a ligand. The appropriate radioligand was often the tool that allowed receptor studies to proceed, e.g., isosteric agonists/antagonists for neurotransmitter receptors (Pert and Snyder, 1973, Yamamura et al., 1978), allosteric PAMs like BZ (Braestrup and Squires, 1977, Möhler and Okada, 1977a), and noncompetitive antagonists like picrotoxin (Möhler and Okada, 1977a, Ticku et al., 1978).

In addition, the determination of receptor-specific binding required the comparison of potencies for a series of chemically related compounds (analogues) with restricted conformational flexibility/increased rigidity, both naturally occurring and specifically synthesized as tools, especially the isoxazole series of Krogsgaard-Larsen, including muscimol and THIP (gaboxadol) (Johnston et al., 1979, Krogsgaard-Larsen et al., 2004, Krogsgaard-Larsen et al., 1979, Meera et al., 2011, Olsen et al., 1981); on neuronal inhibitory transmission measured by electrophysiology, with their potency on the receptor binding assay (Egebjerg et al., 2002, Johnston, 2000); likewise keeping in mind the species and tissue specificity associated with the neurotransmitter actions. The in vivo pharmacology of GABA and analogues is not very important for a variety of reasons including rapid metabolism and poor blood–brain barrier penetration, but what is known about analogue potencies must be consistent between in vivo and in vitro activities.

BZ drugs, historically some of the most prescribed drugs ever, were shown to be CNS depressants and eventually to act as PAM on GABAAR (Costa and Guidotti, 1979, Haefely, 1982).

BZ receptors were identified by radioligand binding and shown to exhibit receptor-specific pharmacology (Braestrup and Squires, 1977, Möhler and Okada, 1977a). BZ binding was found to be enhanced in the test tube by GABA (Karobath et al., 1979, Tallman et al., 1980), indicating that the BZ sites are present on the GABAR proteins. This was verified by solubilizing and purifying one protein with GABA, BZ, and modulatory sites (Gavish and Snyder, 1980, King et al., 1987, Sigel and Barnard, 1984, Supavilai and Karobath, 1984), as well as cloning of the BZ site-containing GABAR (Schofield et al., 1987).

A spectrum of partial agonists was found for the BZ sites (Braestrup, Nielsen, Krogsgaard-Larsen, & Falch, 1979). Also, pharmacological antagonists were discovered (Ro15-1788, AKA flumazenil) (Hunkeler et al., 1981) and even ligands with the opposite efficacy as classical BZs (CNS depressants). These excitatory agents (anxiogenic, proconvulsant) were named inverse agonists; they included BZ analogues, plus variants like imidazobenzodiazepines (Ro15-4513) and β-carbolines, such as β-CCE and DMCM (Braestrup, Schmiechen, Neef, Nielsen, & Petersen, 1982). Braestrup, Nielsen, and Olsen (1980) discovered a putative endogenous ligand β-CCE in human urine and brain for the brain BZ binding sites, which exhibited nanomolar affinity and inverse agonist efficacy. GABA inhibited BZ inverse agonist ligand binding (Braestrup & Nielsen, 1981) including β-CCE (remember it enhanced traditional BZ agonist binding). However, the molecule β-CCE is a possible metabolite/breakdown product of tryptophan or serotonin, and this ligand is not actually in the brain but produced by chemical modification of brain ingredients during extraction and characterization. Other putative endogenous ligands for the brain BZ sites, both small molecules, termed “endozepines,” and peptides, notably “diazepam binding inhibitor” DBI, have been described (Papadopoulos, Berkovich, Krueger, Costa, & Guidotti, 1991). Some of them clearly arise from the environment. Whether any of them is biologically active remains controversial (Möhler, 2014). One class of natural products with potent efficacy on BZ receptors are the flavonoids, herbal medicine ingredients from kudzu tree, e.g., daidzein, genistein (Lukas et al., 2013), and Hovenia, e.g., dihydromyricetin (Shen et al., 2012). These molecules have low micromolar affinity for BZ sites and variable efficacy and GABAAR subtype selectivity, the most common being a BZ agonist-like activity (Kahnberg et al., 2002). Several flavonoids have been described as neuroprotectants, antioxidants, and promoters of alcohol elimination (Arolfo et al., 2009); it is not certain if these activities, while interesting, involve the BZ sites on GABAARs or have physiological relevance.

Heterogeneity of binding suggesting pharmacological subtypes is confirmed by cloning: The BZ receptor sites were identified immediately as showing heterogeneity in several characteristics including pharmacology but also brain regional localization. Affinities of many ligands, e.g., zolpidem (Lloyd & Zivkovic, 1988) and triazolopyridazines (Squires et al., 1979), varied with brain region; second, the binding densities of BZ, GABA, and picrotoxin site ligands did not agree across brain regions (Lo et al., 1983, Olsen et al., 1990). Ligands for BZ and picrotoxin sites also allosterically modulate each other (Olsen, 1982, Supavilai et al., 1982), consistent with multiple allosterically coupled sites on a single protein complex, the GABAAR-chloride channel. This helped establish that these drugs acted on the brain via PAM action on GABAARs. This was of major interest because it suggested presumably circuitry and behavioral distinctions consistent with behavioral pharmacological subtypes, and the potential development of new drugs with subtype selectivity and hopefully improved clinical profiles. Molecular cloning of a family of 19 homologous subunit genes yields several dozens of likely heteropentameric subtypes in nature, of which maybe one dozen is sufficiently abundant to contribute significantly to function and pharmacology (Barnard et al., 1998, Hevers and Lüddens, 1998, Olsen and Sieghart, 2008, Rudolph and Möhler, 2004, Schofield et al., 1987, Whiting et al., 1995). Years of work and clever techniques such as genetic engineering were required to sort out the functional and pharmacological subtypes of GABAAR, but the work has been well worth it.

Numerous global knockout mice were engineered for several of the GABAAR subunit genes, e.g., γ2 (Günther et al., 1995), β3 (Homanics et al., 1997), δ (Mihalek et al., 1999), α4 (Chandra et al., 2006), and α5 (Collinson et al., 2002). Many such knockout studies generated informative phenotypes with valuable insights into the function of the gene under study, but a disappointing number gave either lethal or uninformative phenotypes due mostly to multiple compensatory changes in expression of other gene products (Boehm et al., 2004, Olsen and Homanics, 2000, Rudolph and Möhler, 2004). This prompted new technology development, such as the regional- and/or age-specific (conditional) knockouts (see, e.g., Engin et al., 2014, McHugh et al., 2007) or cell type-specific gene expression knock-down using virus-transfected siRNA expressed under control of cell-specific promoters (e.g., Dixon et al., 2012, Liu et al., 2011, Nie et al., 2011, Rewal et al., 2009).

A brilliant refinement of this preliminary understanding of α subunits and pharmacological subtypes was provided by the mouse α subunit knockins, first by Möhler and colleagues (Rudolph et al., 1999) and then by the Merck team of Whiting (McKernan et al., 2000). This approach used the α H101 residue (in α1, rat) that was R in the non-BZ binding α4 and α6 (Hevers & Lüddens, 1998). The H was switched to R by genetic engineering in mouse to give a point-mutated α subunit which did not show any change in expression and localization, GABAAR channel function, or pharmacology, except for the BZ sensitivity, which was eliminated. Then the animals could be tested for sensitivity to BZ in any behavioral assay of interest. The point-mutated α1 subunit resulted in a large reduction in the sedative, but not anxiolytic actions of BZ in the mouse knockin (McKernan et al., 2000, Rudolph et al., 1999). However, the α2 H101R knockin lost the BZ anxiolytic effect but not the sedative action (Low et al., 2000). This strategy was able to elegantly dissect the molecular targets of BZ action on various behaviors, but even more importantly, it gave strong clues to the function of the GABAR subtypes in behaviors and neuropsychiatric disorders (Rudolph & Möhler, 2004) and could be correlated to the brain regional expression and importance of the receptor subtypes (Barnard et al., 1998, Benke et al., 2004, Whiting et al., 1995). The approach was used to study other subunits of GABAAR like β, indicating molecular roles in other phenomena such as general anesthesia (Rudolph & Antkowiak, 2004).

Finally, now that the tremendous importance of GABAARs in virtually all brain functions is established, and the ability to sort out which subtypes mediate which circuit-dependent behaviors, we gain greater understanding of normal brain function and the corresponding neuropsychiatric diseases (Rudolph & Möhler, 2006). In addition to baseline functions, it is also evident that many disorders result from environmentally induced plastic changes in GABAR subunit composition in critical regions (e.g., Kumar et al., 2009, Liang et al., 2006; for alcohol-treated animals) and insights into these mechanisms may be necessary for understanding brain function. The BZ sites are discussed in detail in the chapter “From GABAA receptors to CNS drugs: Past, present, and future” by Möhler.

Picrotoxinin is a non-nitrogenous natural product from plants that possesses potent and specific ability to block GABAAR function (Curtis and Johnston, 1974, Olsen, 1982); it does so noncompetitively, by binding at a site within the transmembrane domain (TMD) channel (Chen et al., 2006, Hibbs and Gouaux, 2011, Olsen, 2014) and nowhere near the GABA binding site, which has been identified on the extracellular domain (ECD) of the protein (Fig. 1) (Miller and Aricescu, 2014, Olsen et al., 2014). Picrotoxin, as isolated from plants, is a molecular pair of two isomers: the much more active picrotoxinin and the less active picrotin. The GABAAR channel blocking activity makes picrotoxin a dangerous convulsant. Its pharmacology is shared by some close structural analogues found in related plants (Jarboe et al., 1968, Olsen, 2014).

The Olsen lab synthesized a radiolabeled analogue [3H]dihydropicrotoxinin and demonstrated specific receptor binding sites in membrane homogenates of crayfish muscle and mammalian brain (Olsen, 1982, Ticku et al., 1978). The binding was inhibited by appropriate concentrations of appropriate picrotoxin analogues, and also by several “cage convulsants” (Ticku & Olsen, 1979), previously demonstrated to be GABAAR noncompetitive inhibitors (Bowery, Collins, & Hill, 1976). These cage convulsants (trioxabicyclo-octanes) were synthesized as a systematic attempt to make invertebrate-selective pesticides acting on the nervous system (Casida, 1993, Chen et al., 2006), including the highly toxic t-butyl bicyclophosphorothionate (TBPS). This ligand was radiolabeled and made a superior affinity ligand tool for the picrotoxin site (Squires, Casida, Richardson, & Saederup, 1983); the binding activity was found to be allosterically inhibited by GABA agonists and shown to be located on the GABAAR protein (King et al., 1987, Sigel and Barnard, 1984, Supavilai and Karobath, 1984). There is no conclusive evidence for any significant difference in sensitivity to picrotoxin site ligands for subtypes of GABAAR, except for perhaps the insecticides.

Many insecticides, e.g., lindane and dieldrin, were found to inhibit the picrotoxin/TBPS site in GABAAR, including Casida's synthetic cage convulsants (Casida, 1993), as well as some new-generation potent compounds such as fipronil (Ikeda et al., 2001, Zhao et al., 2014). Other convulsants considered active on these sites included pitrazepin and tetramethylene disulfotetramine (TETS) (Zhao et al., 2014). This identification of toxin/insecticide binding sites on GABAAR chloride channels was supported by strong evidence. First, genetic variants of insects that were resistant to the insecticide dieldrin allowed cloning of the “resistance to dieldrin” (rdl) gene, which turned out to be the insect homolog protein of the mammalian GABAAR β subunit (Ffrench-Constant, Rocheleau, Steichen, & Chalmers, 1993); the allelic residue conferring dieldrin resistance in the rdl gene sequence was located in the TMD-M2 helical membrane-spanning region implicated in pentameric cys-loop LGIC receptor proteins as the channel domain (Galzi and Changeux, 1994, Macdonald and Olsen, 1994). Additionally, mutagenesis of the amino acid implicated in dieldrin resistance in the GABAAR β subunit to a cysteine allowed an insecticide analogue modified with a sulfhydryl reagent moiety to bind covalently (affinity label) the channel at this residue, so-called proximity-accelerated covalent coupling (PACC; Perret et al., 1999); the attachment of the insecticide affinity label produced an irreversibly inhibited channel and block of the picrotoxin/TBPS binding sites. Some ligands, e.g., TBPS, had more potent actions on mammalian GABAAR than on insect, while others had higher potency on insects, such as fipronil. The latter showed micromolar affinity for vertebrate and nanomolar affinity for insect GABAAR (Ikeda et al., 2001). Modern techniques such as homology structural modeling and molecular dynamics simulations show that the different toxin ligands bind at overlapping but nonidentical sites in the protein channel that indeed vary with subunit subtypes (Zhao et al., 2014). Nevertheless, the sum of this work demonstrated conclusively that the picrotoxin site was on the GABAAR protein, located apparently in the channel, at a site distinct from the GABA site, and cocrystallography of the nematode GluCl LGIC protein with picrotoxinin showed picrotoxinin bound within the channel (Hibbs & Gouaux, 2011), consistent with models of the GABAAR (Olsen, 2014, Zhao et al., 2014). One could use the protein sequence to model the structure of the channel and deduce other amino acids critical to channel function, insecticide binding, conductance ion selectivity, and desensitization (Chen et al., 2006). The first X-ray crystal structure of a human GABAAR has just been published (Miller & Aricescu, 2014). This should greatly enhance our understanding of the structure and function of the protein and its modulation by drugs.

  • (a)

    All the binding sites are on the same protein, verified biochemically, and on cloned products.

  • (b)

    The three ligand categories help establish heterogeneity of binding populations consistent with receptor subtypes, verified by cloning and subsequent analysis of the separate protein distributions and pharmacological and physiological properties.

  • (c)

    Cloning of the GABAAR was achieved by protein purification, partial sequencing, and homology screening of a cDNA library from brain (Schofield et al., 1987, Sigel and Steinmann, 2012). This resulted in a functional GABAAR expressed from two subunit genes, indicating high homology with the other members of the pentameric cys-loop LGIC superfamily (Dutertre et al., 2012, Galzi and Changeux, 1994). This then led to identification of 19 homologous subunit genes, grouped into families by degree of sequence identity (α1–6, β1–3, γ1–3, δ, ɛ, θ, π, ρ1–3) (Barnard et al., 1998, Lüddens et al., 1995, Whiting et al., 1995). Genetic engineering approaches helped to understand subtypes and their functions.

  • (d)

    The wealth of ligands developed, especially for the BZ site, also provide tools for other methods useful in receptor analysis, especially biochemistry, namely, affinity labeling and affinity chromatography. These reagents allow purification of proteins and, coupled with site-directed mutagenesis, identification of functional domains including especially binding site locations for both high- and low-affinity ligands, including agonists, chloride channel blockers, and allosteric modulators (see below).

Ticku and Olsen (1978) found that picrotoxin binding was inhibited by barbiturates with sedative/hypnotic/anesthetic efficacy. This was consistent with evidence from electrophysiological recordings showing that pharmacologically active barbiturates and related drugs enhanced GABA currents at inhibitory GABAAR synapses (Bowery & Dray, 1976), an effect that could be distinguished from that of the well-known GABA positive modulators, the BZs (Macdonald and Olsen, 1994, Study and Barker, 1981). With little knowledge of structure–activity relationships for barbiturates and related drugs, the Olsen lab (Ticku et al., 1978, Ticku and Olsen, 1978, Wong, Leeb-Lundberg, Teichberg and Olsen, 1984) compared a series of these drugs active on binding to their ability to enhance GABAAR channels with an in vitro36Cl flux assay. It turned out that all sorts of CNS depressants with in vivo efficacy ranging from sedation to hypnosis to anesthesia were found to be PAMs for GABAAR in studies using electrophysiological recordings in neurons (Johnston, 1997, Macdonald and Olsen, 1994, Olsen, 1982, Olsen, Fischer and Dunwiddie, 1986, Olsen and Li, 2011, Supavilai et al., 1982), and they allosterically inhibited the binding of picrotoxin/TBPS, and allosterically enhanced the binding of GABA agonists (Johnston, 1997, Olsen and Snowman, 1982, Olsen, Yang, et al., 1986) and BZs (Leeb-Lundberg et al., 1980, Leeb-Lundberg et al., 1981, Supavilai et al., 1982), and inhibited GABA antagonist (Wong, Snowman, Leeb-Lundberg, & Olsen, 1984) and BZ inverse agonist binding (Braestrup & Nielsen, 1981). This activity was found for those compounds and only those compounds, including stereoisomers, that were active pharmacologically in vivo and in vitro. The allosteric interactions in vitro were not limited to barbiturates but also seen for a variety of volatile and intravenous anesthetics of varying chemical structure, including long-chain alcohols and neuroactive steroids (Johnston, 1997, Olsen, 1982, Olsen, Yang, et al., 1986, Supavilai et al., 1982). Thus, the GABAAR protein family has a very large number of sites for allosteric ligands (Fig. 1). The “anesthetic” sites, first biochemically demonstrated indirectly by allosteric inhibition of the picrotoxin binding sites, are distinct from the picrotoxinin site. The inhibition of picrotoxin/TBPS binding to the GABAAR channel by barbiturates and related anesthetics appeared to be possibly competitive but was soon found to be allosteric (Olsen, Yang, et al., 1986).

The pharmacological specificity including stereoisomers of differing potency for barbiturate modulation of GABAAR provided an in vitro “receptor-specific” assay for such drugs that could be utilized as a tool to characterize the drug mechanism of action, screen for activity of unknown analogues, and assay the presence of the receptor (Olsen, 1982). In other words, even though all the known ligands had insufficient potency to be used for radioligand binding, they could still be quantitated and studied by their indirect action on high-affinity ligands. In the absence of tissue-like electric organs from rays and eels that expressed nicotinic acetylcholine receptors in amounts dozens of times more than mammalian muscle or brain, tissues containing the actual sites of action of drugs like anesthetics, these sites were technically difficult to study. Furthermore, even though the highest affinity (submicromolar) ligands among these PAM ligands (notably neurosteroids) (see Bianchi and Macdonald, 2003, Lambert et al., 2003, Olsen et al., 2003, Zheleznova et al., 2008 and below) were made radioactive, even the world's experts in binding could not develop a specific binding assay for any of them due to their horrible nonspecific binding to tissues (e.g., Gee et al., 1988, Olsen, 1982, Squires et al., 1983, Supavilai et al., 1982).

One category of PAMs for GABAAR that may be extremely important because they are active as endogenous modulators are the neurosteroids. This topic is reviewed nicely by numerous other authors, and this paragraph will be limited to a brief discussion of the use of the pharmacology as PAMs for GABAAR and for analysis of GABAAR subtypes, and identification of ligand binding sites and their mechanism of allosteric modulation. The female sex steroid hormone progesterone itself has sedative/hypnotic efficacy, and a synthetic analogue alphaxalone (5α-pregnan-3α-ol-11, 20-dione) marketed by Glaxo, was shown to have clinically useful intravenous anesthetic efficacy, along with PAM activity at GABAAR (Harrison and Simmonds, 1984, Smith, 2003, Smith et al., 1987). Obviously, the anesthetic was an analogue of the nonhormone metabolite dihydroprogesterone (5α-pregnan-3α-ol-20-one (alloprenanolone, AKA 3α-5α-THP)) which was shown to be the prototype for activity on GABAAR, an activity likewise demonstrated by 3α-5β-THP (pregnanolone) and 3α-5α-THDOC (3α-21-dihydroxy-5α-pregnan-20-one, AKA 3α-5α-tetrahydrodeoxycorticosterone). Note that 3β isomers are inactive.

Majewska, Harrison, Schwartz, Barker, and Paul (1986) showed that steroid hormone metabolites are endogenous potent PAMs for GABAAR. Although neuron or even brain regional specific biosynthesis of the neurosteroids has not been demonstrated, the brain levels vary with the precursor hormone, e.g., cortisone and progesterone, cycling, and are increased in critical brain areas during a variety of physiological states, with a net effect of increased anxiolytic GABAergic inhibitory tone (Barbaccia et al., 2001, Concas et al., 1998, Maguire et al., 2005, Morrow et al., 1999, Paul and Purdy, 1992, Smith, 2003). Binding and neuron electrophysiological recording studies demonstrated allosteric enhancement of synaptic and extrasynaptic GABAAR currents (Gee et al., 1988, Majewska et al., 1986, Smith, 2003, Turner et al., 1989). Some evidence for more potent/efficacious actions on extrasynaptic high-affinity δ subunit-containing GABAAR than on synaptic GABAAR (Bianchi and Macdonald, 2003, Stell et al., 2003, Wallner et al., 2003, Zheleznova et al., 2008) is consistent with the evidence that only low nanomolar levels of endogenous neurosteroids are produced under any conditions (Barbaccia et al., 2001, Concas et al., 1998, Morrow et al., 1999, Paul and Purdy, 1992, Smith, 2003). Likewise, in the hippocampal formation CA1 region, pyramidal neurons mediate tonic inhibitory GABAAR currents with extrasynaptic α5 subunit-containing GABAAR; general anesthetics as well as neurosteroids enhance the α5-GABAAR tonic inhibition with resulting inhibition of learning and memory and production of amnesia (Wang & Orser, 2011). Also, α5-selective antagonists, like inverse agonist BZ ligands, reverse this learning and memory inhibition and improve cognitive function (Dawson et al., 2006), possibly in Alzheimer's disease.

Möhler, Battersby, and Richards (1980) discovered that [3H]flunitrazepam could photoaffinity label the BZ sites in brain, and had a sufficient affinity (KD ~ 10 nM) to detect binding in crude homogenates or brain sections, and yield a covalent attachment to a polypeptide on SDS–PAGE at ~ 50 kDa. This affinity label binding was “receptor specific” and could be employed to localize BZ binding sites (GABAAR) in brain sections (Möhler et al., 1980). Others immediately showed (Sieghart & Karobath, 1980) labeling of numerous (four or more) polypeptides in brain membranes on more sensitive SDS–PAGE gels. This showed that radioactive BZ ligands bound to multiple polypeptide bands of similar but not identical molecular weights in brain, presumably (and in retrospect, correctly analyzed) as different gene products, i.e., subunits with different protein sequences. The binding site for BZs was analyzed by mutagenesis and by photoaffinity labeling. The classical agonist [3H]flunitrazepam was found to bind to α1 (and also α2, α3, and α5) H102 in ECD Loop A (cow numbering; rat = 101) (Duncalfe et al., 1996, Smith and Olsen, 2000). Flunitrazepam as a classical BZ agonist did not bind to α4 or α6 (βγ2) subtypes (Barnard et al., 1998, Hevers and Lüddens, 1998). The antagonists and inverse agonists bind the α4 and α6 (βγ2) subtypes, and the binding site on all six α subunits was believed to contain the same residues as flunitrazepam in α1, confirmed by photoaffinity labeling with [3H]Ro15-4513, identified (Sawyer et al., 2002) as attached covalently to Y209 (Loop C, ECD). The BZ binding site was shown to be located at the α +/γ- interface, as opposed to the GABA site at the β +/α- interface. The surprising observation was that the BZ binding sites were in the same homologous sequences (Loops A–H) as the GABA binding sites, which were the same utilized for agonist binding in all members of the cys-loop LGIC superfamily (Amin and Weiss, 1993, Duncalfe et al., 1996, Galzi and Changeux, 1994, Harrison and Lummis, 2006, Sawyer et al., 2002). This was immediately also recognized by workers on the GABAAR (Sigel and Buhr, 1997, Smith and Olsen, 1995) and indicated that the BZ sites were modified GABA agonist sites and these exogenous/ synthetic PAM ligands fortuitously found a site that was almost but not quite an agonist site! This remarkable situation may be unprecedented in protein chemistry. The BZ site was compared for the various subunit subtypes by quantitative structure–activity relationships (e.g., Cook et al., 2005), and the mutagenesis plus affinity labeling experiments, i.e., cysteine replacement and covalent binding of a ligand analogue containing a sulfhydryl reagent (Tan et al., 2007). With the X-ray crystal structure of the invertebrate acetylcholine binding protein (Brejc et al., 2001) available as template, homology structural models of the ECDs of nAChRs and GABAARs were constructed (Fig. 1), including the GABA (Sander et al., 2011) and BZ sites (Ernst et al., 2003, Ernst et al., 2005, Middendorp et al., 2014, Richter et al., 2012).

GABAAR have long been the focus for acute alcohol actions. Evidence was reported for behaviorally relevant low millimolar alcohol actions on tonic inhibitory GABA currents (Hanchar et al., 2005, Liang et al., 2006, Wei et al., 2004) mediated by extrasynaptic α4/6, β, δ subunit-containing GABAAR. Low-dose (≤ 30 mM) EtOH enhances recombinantly expressed δ subunit-containing subtypes of GABAAR, as well as tonic inhibitory GABAR-mediated currents in cells that express the extrasynaptic δ-GABAAR (Sundstrom-Poromaa et al., 2002, Wallner et al., 2003, Wallner et al., 2006, Wallner and Olsen, 2008). Korpi, Kleingoor, Kettenmann, and Seeburg (1993) found a point mutation—or allelic variant—in the GABAAR α6 subunit (R100Q) in a rat strain that exhibited in vivo behavioral hypersensitivity to BZs and also EtOH (Korpi et al., 1993). The BZ sensitivity was explained by the α6-R100Q mutation, which gave high-affinity BZ enhancement when expressed with β and γ2 subunits in vitro in oocytes. Neither wild type nor mutant receptors showed enhancement by EtOH (up to 100 mM) when expressed with β and γ2 subunits. Hanchar et al. (2005) revisited this mutation and found that α6-R100Q, when expressed with β3 and δ subunits, gave even higher sensitivity to EtOH enhancement (low mM concentrations) in oocytes than the already sensitive wild-type α6βδ subtype; further, the rats expressing this allelic α6-R100Q showed greater sensitivity (1–20 mM) to EtOH than wild-type α6-R100 (10–30 mM) for tonic inhibitory currents carried by α6βδ−GABAAR in cerebellar granule cells studied in brain slices (Hanchar et al., 2005). The amino-acid α6-R100 is located in the binding site for traditional BZ including the imidazobenzodiazepine EtOH antagonist drug Ro15-4513. Further, many behavioral effects of EtOH, and the in vitro enhancement of GABAAR in neurons, in culture, and in recombinant expression cells, are reversed by Ro15-4513 (Suzdak et al., 1986, Wallner et al., 2006). They demonstrated that Ro15-4513 not only blocks low mM EtOH enhancement of δ-GABAAR currents, but also this action is reversed by the close structural analogue flumazenil (Ro15-1788), the traditional BZ antagonist. A minor subset of BZ ligands, notably imidazobenzodiazepines and β-carboline compounds, are able to mimic or inhibit Ro15-4513 activity on EtOH in vivo and in vitro, and the same ligands, and only those, are able to inhibit the binding of [3H]Ro15-4513 to native and recombinant δ-GABAAR (Hanchar et al., 2006). Whereas it had been believed that no BZ ligands bound to the δ-GABAAR subtypes, this new evidence suggests that δ-GABAR have a novel BZ/EtOH site with unique pharmacology with respect to BZ ligands (Santhakumar et al., 2007, Wallner et al., 2006). Thus, this work defines an EtOH receptor, when it was not clear that there is such a thing in the traditional sense (Wallner & Olsen, 2008). Perkins, Trudell, Crawford, Alkana, and Davies (2010) showed that there is an ECD region (Loop 2) in glycine LGIC receptors that affects EtOH sensitivity, and introduction of the GABAR δ subunit sequence for that region increases EtOH sensitivity, both in glycine and in GABAAR (Perkins et al., 2010). This domain is implicated in coupling agonist binding to channel gating through physical interaction between Loop 2, which is near the postulated ECD EtOH binding site, and the M2–3 linker (Kash, Jenkins, Kelley, Trudell, & Harrison, 2003), which is near the TMD anesthetic binding sites (Bertaccini et al., 2013, Olsen et al., 2014). This story needs further study to evaluate its importance for both agonist and PAM function.

Further mutagenesis of residues in the extracellular BZ binding domain (ECD) of α6βδ-GABAAR revealed preliminary evidence for an EtOH-sensitive Ro15-4513 binding site at the α +/β- interface (Fig. 2, reproduced from Wallner et al., 2014). Starting with sequence loops in the major subunit for this binding site (α), including α6-R100, they predicted sequences for the minor subunit potentially situated abutting this α pocket domain, which could be either on δ or on β. Mutagenesis of residues in the sequence of the δ subunit, e.g., H68A (Meera, Olsen, Otis, & Wallner, 2010) imparts classical BZ site binding to the heteropentamer, but does not disturb the EtOH-sensitive Ro15-4513 sites in δ-GABAR. This receptor appears to carry two distinct BZ sites. These authors used BZ sensitivity as a marker for (mutated) δ subunit expression in HEK cells. On the other hand, the selectivity of β subunits of the EtOH binding (β3 > β2  β1; Wallner et al., 2003) can be identified in chimera and point mutagenesis studies to involve this loop in the β subunit domain, #62–66 in β3, Loop “D” (Fig. 2, from Wallner et al., 2014). Based on involvement of these two amino-acid residues, α6-R100 and β3-Y66, they propose a model in which δ subunit-containing (and possibly other) GABAAR contain a unique EtOH site at the α +/β- ECD subunit interface. There are other δ-selective PAM ligands that might also interact here, or at a δ-interface (Jensen et al., 2013). This same interface in non-δ-GABAR has been demonstrated to show a significant affinity for a subset of BZ ligands including pyrazoloquinolines and possibly other PAM (Ramerstorfer et al., 2011).

Thus, an ECD site for EtOH modulation may resemble the traditional binding and resulting efficacy of ligands like agonists (e.g., GABA) and more well-understood modulators (e.g., BZ).

A recent review (Olsen et al., 2014) covers most of the evidence for the TMD anesthetic site with low affinity for EtOH and the ECD nonanesthetic high-affinity ethanol-sensitive BZ sites. It is our opinion, supported by considerable literature, that subtypes of the GABAAR represent the brain target for EtOH that is closest to that relevant to blood and brain concentrations affecting humans (~ 17 mM) with the well-known intoxicating actions, and that myriad reports of EtOH acting on other neurotransmitter systems and ion channels require much higher concentrations of EtOH (> 50 mM) that may be involved in anesthesia and fatal overdose for EtOH in humans (Wallner et al., 2014, Wallner and Olsen, 2008). Questions remain in the minds of many in the field, both as to whether low-dose EtOH effects do involve GABAARs at all and whether the important pharmacological effects of low-dose EtOH also involve other targets. The issue is thoroughly covered in a special issue of Alcohol edited by Lovinger and Homanics (2007).

Prior to identification of anesthetic binding contact points on proteins such as GABAAR, mutagenesis was employed to identify amino-acid residues critical for PAM action, by comparing two related LGICs of differing sensitivity to modulation. The weakness of this approach is twofold: the residues to be studied are determined by advance guessing on the part of the investigator, and the residues identified as important to drug modulation are not proven to be actual binding pocket sites, but could be due to allosteric coupling of drug action and function. Affinity labeling on the other hand will generally identify binding pocket residues. Depending on the chemistry of the affinity labeling, a couple of things can go wrong: it is possible that the photoreactive intermediate is long-lived and/or slow to react so that it can move from the actual binding pocket in the protein to a nearby but not critical residue; additionally, there may be no residues in the binding pocket that are actually good substrates for covalent binding of the affinity label ligand; thus, some theoretically active affinity labels will turn out to have no activity (Bouchet et al., 1987, Forman and Miller, 2011). Nevertheless, various criteria for specificity generally support the conclusion that affinity labeling is specific and valuable.

Classical and ground-breaking mutagenesis studies (mutagenesis only, not affinity labeling) demonstrated residues in the TMD that are necessary for anesthetic actions (volatile agents and long-chain alcohols). That work concluded that the two residues identified were in a single binding pocket limited to intrasubunit domains (Mihic et al., 1997). These two residues M2–15′ and M3–4′, in α or β subunits, were analyzed by systematic mutagenesis and characterized for sensitivity to numerous examples of various chemical classes of general anesthetics. In some cases, substitution of larger amino acids for the native one at these two critical positions led to reduced modulation by the anesthetic, but also equal or increased opening probability of the receptor in the absence of agonist, which was interpreted as the substituted amino acid replacing the bound anesthetic as positive allosteric modulator (Jenkins et al., 2001) in the putative binding site.

This work was consistent with identification of a binding pocket for anesthetics, e.g., sensitivity to both volatile agents and long-chain fatty acids exhibited a “cutoff point,” or maximum size or volume of ligand that could fit into the binding site (Krasowski et al., 2001, Wick et al., 1998). Bali and Akabas (2004) showed that sulfhydryl reagents inactivated cysteine-substituted βM3–4′ with respect to propofol enhancement of GABAAR channels, and this was protected by excess free propofol in the reaction; however, no protection was afforded to βM2–15′ (Bali & Akabas, 2004).

Further arguing against the role of M2–15′ in the binding pocket was the observation that the action of numerous chemotypes of PAM was sensitive to the nature of the residue at this position, suggesting a possible allosteric coupling site. For example, loreclezole, etomidate, and propofol were all sensitive to this residue in the β subunits (but not α) (Belelli et al., 1997, Rudolph and Antkowiak, 2004). A convincing demonstration that these residues are important for anesthetic action in vivo were reports establishing that genetic knockin of point mutations in the GABAAR β subunits at position M2–15′ could eliminate the anesthetic actions (Jurd et al., 2003: β3 M265N) or sedative/hypnotic actions (Reynolds et al., 2003; β2M265S) of etomidate. This shows that GABAAR are the major molecular target for etomidate and presumably other general anesthetics and that the β3 subunit-containing subtypes of GABAAR are more important for this in vivo action than other β subunit-containing subtypes, suggesting an anatomic correlate of anesthetic action (Jurd et al., 2003). As mentioned above, this does not prove that this residue is in the binding pocket. However, covalent attachment of sulfhydryl analogues of alcohols like propanethiol to cysteine-substituted M2–15′ leads to irreversible enhancement of channel function, as well as occlusion of the site for any additional anesthetic modulation (Mascia, Trudell, & Harris, 2000), strongly suggesting participation of this residue in PAM binding. Crystal structure evidence is needed and may be forthcoming.

Another residue at the beginning of M1 or pre-M1 in the GABAAR β2 subunit was shown by mutagenesis to alter modulation of GABAAR binding and channel function by several chemotypes of general anesthetics, again consistent more with an allosteric coupling site than a ligand binding pocket (Carlson et al., 2000, Chang et al., 2003, Engblom et al., 2002). This area deserves further studies on its participation.

Mutagenesis suggests critical residues as possible steroid binding sites on GABAAR (Hosie et al., 2006): one residue each in αM1 and αM4 required for PAM activity, and two residues in αM1 and βM3 needed for direct gating, proposed to be part of a single intersubunit pocket. However, the possibility of an intersubunit pocket is ruled out because these two residues are not positioned near each other in the helical wheels generated by models of the intersubunit etomidate site (Li et al., 2009, Li et al., 2006) generated by cysteine substitution cross-linking data (Bali et al., 2009, Olsen and Li, 2011, Olsen et al., 2014, Stewart et al., 2013). Mouse knockins were able to identify residues in GABAARs critical for anesthetic modulation, such as β2/3-M2–15′ for etomidate, propofol, isoflurane (partial), but not steroids (Rudolph & Antkowiak, 2004). Li et al. (2009) showed enhancement of etomidate binding by GABA-enhancing steroids, suggesting that steroid and etomidate sites are distinct (Li et al., 2009). Affinity labeling with a GABA-active neuroactive steroid derivative identified one amino-acid residue of contact in the TMD-M3 of homomeric recombinant β3 GABAAR (Chen et al., 2012). Hopefully, the identification of residues involved in functional binding in native GABAAR of the steroid PAMs will be accomplished soon.

Relatively high-affinity ligands for the anesthetic sites were utilized to make analogues with affinity labeling chemical moieties and radiolabel them to identify the binding pocket amino acids in the GABAAR proteins. A team led by Keith Miller synthesized 3H-labeled azietomidate, azibarbiturate, and azipropofol (Chiara et al., 2013, Forman and Miller, 2011, Jayakar et al., 2014, Li et al., 2006, Olsen, 2014, Olsen and Li, 2011). Radiolabeled [3H]azietomidate (Li et al., 2006) was used as a photoaffinity label on purified GABAAR protein from bovine cerebral cortex. Two residues were labeled, corresponding to the TMDs within M3 of the β subunit (M286 at position 4′) and M1 of the α subunit (M236 at position 11′) (Li et al., 2006). Here, we introduce a shorthand nomenclature for residues in the membrane-spanning domain (TMD). Residues are numbered 1′–20′ starting at the extracellular end and proceeding to the intracellular end of the TMD helices. The TMD helices could be aligned relative to the membrane-spanning sequences based on cysteine scanning mutagenesis and sulfhydryl cross-linking (Bali and Akabas, 2004, Bali et al., 2009), and a three-dimensional model of the TMD generated, showing that the two labeled residues could be located near each other, consistent with a single binding pocket at the intersubunit β/α interface (Li et al., 2009, Li et al., 2006). As with GABA binding, there are two copies of the β/α interface etomidate binding site per pentamer (Forman and Miller, 2011, Olsen and Li, 2011).

Mutagenesis of these two residues established that they were indeed critical for anesthetic modulation of channel functions, both for etomidate and for propofol modulation (Bali and Akabas, 2004, Bali et al., 2009, Desai et al., 2009, Krasowski et al., 2001, Stewart et al., 2008, Stewart et al., 2013); further, the etomidate enhancement of function and modulation of agonist binding could be blocked by large but not small sulfhydryl reagents attaching to a cysteine replacement mutation at αM1–11′ or βM3–4′ (Stewart et al., 2013). These results are consistent with the conclusion that the affinity-labeled residues are part of an etomidate binding pocket (Forman and Miller, 2011, Olsen and Li, 2011).

One of the residues, β(M3–4′), described by Mihic et al. (1997) corresponded with one of the residues identified by etomidate affinity labeling (Li et al., 2006). The same protection of βM3–4′ by etomidate was found after that residue was identified as binding etomidate (Stewart et al., 2013). Mutation to Trp of either of the active site Mets (Stewart et al., 2008) resulted in increased spontaneous opening and sensitivity to GABA (suggested to represent endogenous Trp replacing the exogenous anesthetic). Cysteine replacement mutants of βM2–15′ (Desai et al., 2009) reduced etomidate efficacy but not affinity, nor GABA affinity or efficacy. They suggested that this residue is not part of the anesthetic binding pocket, but could not rule out the possibility that this residue makes contact with the ligand during etomidate-enhanced GABA-gated channel opening. These workers (Forman & Miller, 2011) noted that due to the effects of mutations on basal affinity and efficacy for GABA and modulators, one needed to compare a series of agonists and modulators regarding mutant channel properties, using the Monod/Wyman/Changeux (MWC) allosteric model. Rüsch, Zhong, and Forman (2004) described well the action of etomidate modulation as an MWC co-agonist model (Rüsch et al., 2004), in which both enhancement of agonist by etomidate, at low micromolar concentrations, and direct channel gating, at high micromolar concentrations, could be explained by etomidate binding to a single class of binding sites, now identified (Li et al., 2006). On the other hand, etomidate binding to β (M3–4′) and α (M1–11′) was completely and apparently competitively inhibited by volatile anesthetics at pharmacologically appropriate concentrations, but not by alcohols up to very high doses (Li, Chiara, Cohen, & Olsen, 2010).

In fact, the subunit interface binding etomidate is the same interface where the GABA binding site is located in the ECD about 50 Å above the etomidate site in the TMD, and suggests that intersubunit interfaces are the usual site for ligand binding pockets, both allosteric and agonist (Li et al., 2006). The importance of this conclusion cannot be underestimated. Note that the GABA binding site at subunit interfaces (Amin and Weiss, 1993, Venkatachalan and Czajkowski, 2012) involves residues at positions in the subunit sequence that are homologous with nAChR, GlyR, and 5-HT3R, i.e., the “cys-loop, pentameric LGIC” superfamily (Galzi & Changeux, 1994). We mentioned that the BZ binding sites on GABAR are modified agonist sites, located at different subunit interfaces in the ECD not used by GABA sites (Smith & Olsen, 1995).

Additional important information about these TMD modulatory sites was obtained using another aziridine affinity label analogue of a barbiturate PAM on GABAARs (Chiara et al., 2013); interestingly, residues labeled covalently by the barbiturate ligand were positioned at other intersubunit interfaces homologous to the two that bind etomidate (β +/α-), namely, the α +/β- and γ +/β-. The etomidate sites have low affinity for barbiturates and the barbiturate sites have low affinity for etomidate, while both show mutual allosteric inhibition. Just published is a follow-up from that group (Cohen and Miller) showing that an affinity label analogue of propofol binds to all five subunit interfaces (in recombinant α/β GABAAR lacking γ or δ) employing the pocket residues identified for etomidate and barbiturates (Jayakar et al., 2014). Thus, the sensitivity to these modulators is quite dependent on the subunit composition of the GABAAR subtype and particularly important to subtype selectivity for PAMs is the heteropentameric nature of most native cys-loop, pentameric LGIC (Chiara et al., 2013, Forman and Miller, 2011, Jayakar et al., 2014, Li et al., 2006, Olsen and Li, 2011, Olsen et al., 2014). It will be interesting to see if the remaining TMD interface α +/γ- in native α/β/γ and α +/δ- in α/β/δ GABAAR carry any modulatory sites. On the other hand, another group, using a different propofol-based affinity label, identified a single residue at the ECD surface of TMD-M3 (Yip et al., 2013) which is distinct from those identified by Jayakar et al. (2014).

Looking more closely at Fig. 1, I have shown the location of the known ligand binding sites, both ECD and TMD, on the homology model of the heteropentameric α/β/γ native GABAAR using the templates for the homomeric β3 human GABAAR (Miller & Aricescu, 2014), the prokaryotic LGIC (Corringer et al., 2010), and the invertebrate nematode GluCl (Hibbs & Gouaux, 2011). The left part shows the transmembrane portion (TMD) with the binding sites for various PAMs such as general anesthetics of varying chemotype as described above. The right part shows the ECD with the binding sites for the agonist GABA as well as the two important PAMs, the BZ and high-affinity EtOH as described above.

Section snippets

Conclusion

Note that the native GABAR are primarily heteropentamers, and all the templates are (so far) homomers. Thus, the heteromers, unlike homomers, exhibit different sorts of subunit interfaces and ligand binding sites depending on the subunit composition, i.e., GABAAR subtypes. This model (Fig. 1) can be considered a rough first draft with new experimental evidence hopefully smoothing out the roughness in the near future.

Conflict of Interest

R. W. O. reports no conflict of interest.

Acknowledgments

Special thanks to Dr. Martin Wallner for many years of helpful discussions, and assistance with graphics and references. Thanks to Drs. Guo-Dong Li, Jing Liang, and Kerstin Lindemeyer for helpful discussions. Thanks to Professor Jean-Pierre Changeux for mentoring and encouragement. This study was supported by grants from the US National Institutes of Health.

References (197)

  • M. Ernst et al.

    Comparative modeling of GABAA receptors: Limits, insights, future developments

    Neuroscience

    (2003)
  • J.-L. Galzi et al.

    Ligand-gated ion channels as unconventional allosteric proteins

    Current Opinion in Structural Biology

    (1994)
  • N.J. Harrison et al.

    Locating the carboxylate group of GABA in the homomeric rho GABAA receptor ligand-binding pocket

    The Journal of Biological Chemistry

    (2006)
  • N.L. Harrison et al.

    Modulation of the GABA receptor complex by a steroid anaesthetic

    Brain Research

    (1984)
  • S.S. Jayakar et al.

    Multiple propofol binding sites in a GABAA receptor identified using a photoreactive propofol analog

    Journal of Biological Chemistry

    (2014)
  • M.D. Krasowski et al.

    Methionine 286 in transmembrane domain 3 of the GABAA receptor β subunit controls a binding cavity for propofol and other alkylphenol general anesthetics

    Neuropharmacology

    (2001)
  • P. Krogsgaard-Larsen et al.

    GABAA agonists and partial agonists: THIP (Gaboxadol) as a non-opioid analgesic and a novel type of hypnotic

    Biochemical Pharmacology

    (2004)
  • J.J. Lambert et al.

    Neurosteroid modulation of GABAA receptors

    Progress in Neurobiology

    (2003)
  • F. Leeb-Lundberg et al.

    Interaction of anticonvulsants with the barbiturate-benzodiazepine-GABA receptor complex

    European Journal of Pharmacology

    (1981)
  • G.-D. Li et al.

    Neurosteroids allosterically modulate binding of the anesthetic etomidate to GABAA receptors

    The Journal of Biological Chemistry

    (2009)
  • G.-D. Li et al.

    Numerous classes of general anesthetics inhibit etomidate binding to GABAA receptors

    The Journal of Biological Chemistry

    (2010)
  • K.G. Lloyd et al.

    Specificity within the GABAA receptor supramolecular complex: A consideration of the new omega 1-receptor selective imidazopyridine hypnotic zolpidem

    Pharmacology, Biochemistry, and Behavior

    (1988)
  • D.M. Lovinger et al.

    Editorial: Tonic for what ails us: High affinity GABAA receptors and alcohol

    Alcohol

    (2007)
  • H. Lüddens et al.

    GABAA/benzodiazepine receptor heterogeneity: Neurophysiological implications

    Neuropharmacology

    (1995)
  • J. Amin et al.

    GABAA receptor needs two homologous domains of the β-subunit for activation by GABA but not by pentobarbital

    Nature

    (1993)
  • M.P. Arolfo et al.

    Suppression of heavy drinking and alcohol seeking by a selective ALDH-2 inhibitor

    Alcoholism, Clinical and Experimental Research

    (2009)
  • M. Bali et al.

    Defining the propofol binding site location on the GABAA receptor

    Molecular Pharmacology

    (2004)
  • M. Bali et al.

    GABA-induced intersubunit conformational movement in the GABAA receptor α1M1-β2M3 transmembrane subunit interface: Experimental basis for homology modeling of an intravenous anesthetic binding site

    The Journal of Neuroscience

    (2009)
  • E.A. Barnard et al.

    International Union of Pharmacology. XV. Subtypes of GABAA receptors: Classification on the basis of subunit structure and receptor function

    Pharmacological Reviews

    (1998)
  • D. Belelli et al.

    The interaction of the general anesthetic etomidate with the GABAA receptor is influenced by a single amino acid

    Proceedings of the National Academy of Sciences of the United States of America

    (1997)
  • E.J. Bertaccini et al.

    Assessment of homology templates and an anesthetic binding site within the GABAA receptor

    Anesthesiology

    (2013)
  • M.T. Bianchi et al.

    Neurosteroids shift partial agonist activation of GABAA receptor channels from low- to high-efficacy gating patterns

    The Journal of Neuroscience

    (2003)
  • M.J. Bouchet et al.

    Aryl diazo compounds and diazonium salts as potential irreversible probes of the gamma-aminobutyric acid receptor

    Journal of Medicinal Chemistry

    (1987)
  • N.G. Bowery et al.

    International Union of Pharmacology. XXXIII. Mammalian GABAB receptors: Structure and function

    Pharmacological Reviews

    (2002)
  • N.G. Bowery et al.

    Bicyclic phosphorus esters that are potent convulsants and GABA antagonists

    Nature

    (1976)
  • N.G. Bowery et al.

    Barbiturate reversal of amino acid antagonism produced by convulsant agents

    Nature

    (1976)
  • C. Braestrup et al.

    GABA reduces binding of 3H-methyl beta-carboline-3-carboxylate to brain benzodiazepine receptors

    Nature

    (1981)
  • C. Braestrup et al.

    Partial agonists for brain GABA/benzodiazepine receptor complex

    Nature

    (1979)
  • C. Braestrup et al.

    Urinary and brain beta-carboline-3-carboxylates as potent inhibitors of brain benzodiazepine receptors

    Proceedings of the National Academy of Sciences of the United States of America

    (1980)
  • C. Braestrup et al.

    Interaction of convulsive ligands with benzodiazepine receptors

    Science

    (1982)
  • C. Braestrup et al.

    Specific benzodiazepine receptors in rat brain characterized by high-affinity [3H]diazepam binding

    Proceedings of the National Academy of Sciences of the United States of America

    (1977)
  • K. Brejc et al.

    Crystal structure of an ACh-binding protein reveals the ligand-binding domain of nicotinic receptors

    Nature

    (2001)
  • B.X. Carlson et al.

    A single glycine residue at the entrance to the first membrane-spanning domain of the gamma-aminobutyric acid type A receptor beta(2) subunit affects allosteric sensitivity to GABA and anesthetics

    Molecular Pharmacology

    (2000)
  • J.E. Casida

    Insecticide action at the GABA-gated chloride channel: Recognition, progress, and prospects

    Archives of Insect Biochemistry and Physiology

    (1993)
  • D. Chandra et al.

    GABAA receptor α4 subunits mediate extrasynaptic inhibition in thalamus and dentate gyrus and the action of gaboxadol

    Proceedings of the National Academy of Sciences of the United States of America

    (2006)
  • L. Chen et al.

    Structural model for γ-aminobutyric acid receptor noncompetitive antagonist binding: Widely diverse structures fit the same site

    Proceedings of the National Academy of Sciences of the United States of America

    (2006)
  • Z.-W. Chen et al.

    Neurosteroid analog photolabeling of a site in the third transmembrane domain of the β3 subunit of the GABAA receptor

    Molecular Pharmacology

    (2012)
  • A. Christopoulos et al.

    International Union of Basic and Clinical Pharmacology. Multi-site pharmacology: Recommendations for the nomenclature of receptor allosterism and allosteric ligands

    Pharmacological Reviews

    (2014)
  • N. Collinson et al.

    Enhanced learning and memory and altered GABAergic synaptic transmission in mice lacking the α5 subunit of the GABAA receptor

    The Journal of Neuroscience

    (2002)
  • A. Concas et al.

    Role of brain allopregnanolone in the plasticity of GABAA receptor in rat brain during pregnancy and after delivery

    Proceedings of the National Academy of Sciences of the United States of America

    (1998)
  • Cited by (62)

    • Pharmacology of Alcohol Use

      2022, Comprehensive Pharmacology
    • The efficacy of γ-aminobutyric acid type A receptor (GABA <inf>A</inf>R) subtype-selective positive allosteric modulators in blocking tetramethylenedisulfotetramine (TETS)-induced seizure-like behavior in larval zebrafish with minimal sedation

      2021, Toxicology and Applied Pharmacology
      Citation Excerpt :

      Intoxicated individuals are likely to be treated with benzodiazepines (Barrueto et al., 2003; Li et al., 2012), which are the standard of care for organophosphate-induced seizures. Classified as positive allosteric modulators (PAMs), benzodiazepines do not directly open GABAARs, but increase their sensitivity to the endogenous ligand GABA (Olsen, 2015). Benzodiazepines bind to a conserved site between multiple subtypes of GABAAR (Olsen, 2015; Schoch et al., 1985).

    View all citing articles on Scopus
    View full text