GABAA Receptor Subtypes: Structural Variety Raises Hope for New Therapy Concepts

Home , Divaplon

Review article e-Neuroforum 2015 · 6:97–103 Margot Ernst1 · Werner Sieghart2 DOI 10.1007/s13295-015-0016-9 1 Department of Molecular Neurosciences, Center for Brain Research, Vienna, Austria Published online: 6 October 2015 2 Center for Brain Research, Vienna, Austria © Springer-Verlag Berlin Heidelberg 2015

GABAA receptor subtypes: structural variety raises hope for new therapy concepts

Introduction chloride channel (GluCl) of Caenorhabdi- the gamma + beta − interface no ligands tis elegans [1]. All cys-loop receptor sub- have been described so far. The binding of Gamma-amino-butyric acid (GABA) is units possess a common domain organiza- GABA to its two sites leads to a change in a major neurotransmitter in the central tion with an N-terminal domain that con- conformation that can open the chloride nervous system (CNS) of adult mam- tains the cys-loop and ligand binding sites channel of the receptor. Binding of ben- mals, but also a messenger molecule in for the agonist or allosteric modulators, a zodiazepines to their site, or of pyrazolo- many nonneuronal tissues. The GABA transmembrane (TM) domain composed quinolinones to the site at the alpha + beta effects are mediated via ionotropic GA- of four helical segments, and an intracel- − interface, does not lead to opening of the BAA receptors and metabotropic GAB- lular domain that is interspaced between channel, but the conformational changes AB receptors. GABAA receptors are GA- TM3 and TM4, see . Fig. 1. The trans- induced by these agents leads to a modu- BA activated chloride channels that con- mitter binding sites of this family are at lation of the GABA effect at the receptor. sist of five subunits (see . Fig. 1) and are extracellular interfaces between two sub- These two binding sites are thus funda- targets of many clinically relevant med- units, whereby one is known as “principal mentally different from those for barbitu- ications [1]. Barbiturates, neuro-active subunit” and the other as “complemen- rates, steroids, or other sedative anesthet- steroids, several general anesthetics, and tary subunit”. The principal subunit, per ics, which are located in the TM domain benzodiazepines are among the class- convention, contributes the “plus” side to of the receptors. Low concentrations of es of drugs acting via GABAA receptors the interface and the complementary sub- these agents lead to an allosteric enhance- (see . Excursion 1). The existence of unit the “minus” side, see . Fig. 1. Each ment of the GABA effect, and high con- multiple ligand binding sites at the recep- subunit contributes several discontinu- centrations open the channel directly in tors as well as of multiple GABAA recep- ous segments to the interface — those in- the absence of GABA. Due to this GA- tor subtypes, results in a highly complex volved in ligand recognition are tradition- BA-mimetic effect, barbiturates, steroids, pharmacology. ally termed “loops A thru G”, see . Fig. 1. or anesthetics are thus much more tox- GABAA receptors are the most wide- In the transmembrane domain the terms ic in overdose than the benzodiazepines. spread inhibitory receptors in the CNS. “plus side”, which is formed by parts of Depending on the structure of ligands, At higher intracellular chloride concen- TM2 and TM3 of the principal subunit; binding events at individual sites can lead trations, however, these receptors can also and “minus side”, formed by parts of TM2 have excitatory effects mediated by chlo- and TM1 of the complementary subunit, Abbreviations ride outflux from the cells. A total of 19 are also commonly used, see . Fig. 1. nAChR nicotinic acetylcholine receptor genes encode for GABAA receptor sub- In the most common GABAA receptors units (alpha1–6, beta1–3, gamma1–3, del- with the arrangement shown in . Fig. 1, AChBP acetylcholine binding protein ta, epsilon, theta, pi, rho1–3). The recep- the GABA binding site is located at the GluCl glutamate-activated chloride tors themselves comprise five subunits extracellular interface which is formed by channel and belong to the family of the pentameric the principal (plus) side of the beta sub- ligand-gated ion channels (cys-loop recep- unit and the complementary (minus) side 5-HT 5- hydroxytryptamine (serotonin) tors), together with the nicotinic acetyl- of the alpha subunit. Thus these receptors GLIC Gloeobacter violaceus ligand- choline receptors (nAChRs), the serotonin feature two GABA binding sites. The ben- gated ion channel type 3 (5-HT3) receptors, the glycine re- zodiazepine binding site is at the homolo- ceptors, the acetlycholine-binding protein gous interface between the alpha subunits’ ELIC Erwinia chrysantemi ligand- gated ion channel (AChBP), the bacterial proteins Gloeo- plus, and the gamma subunits’ minus side. bacter violaceus ligand-gated ion channel The binding site recently identified by our ICD intracellular domain and Erwinia chrysantemi ligand-gated ion group at the alpha + beta − interface can TM transmembrane channel, as well as the glutamate-activated be used by pyrazoloquinolinones [1]. For

e-Neuroforum 4 · 2015 | 97 Review article

many binding sites null-modulators have Excursion 1 Clinical pharmacology of GABAA receptors not yet been identified. Drugs that act on GABAA receptors were in clinical use long before the receptors were identified. The first barbiturate based drug, 5,5- diethyl barbituricacic, was marketed by Merck as sleeping aid Subtype selective in 1903 already under the trade name Veronal. Barbiturates have dose dependently sedative, then ligands as therapeutic hypnotic and at high doses narcotic effects. In addition, they are anticonvulsants. At low concentra- prospects of the future tions they enhance the effects of GABA on the GABAA receptors. At higher concentrations they can activate the receptors directly. At even higher concentrations barbiturates also act at glutamatergic AMPA receptors and voltage-sensitive sodium channels. Due to their high toxicity at overdose, Currently available drugs acting on GAB- barbiturates legally are allowed to be used in Germany and in Switzerland only for certain clinical AA receptors all interact with multiple re- applications since 1992. ceptor subtypes — they thus must be clas- Due to the broad abundance of anxiety disorders, compounds able to fight anxiety were identified early. Presumably, one of the first drugs to relieve anxiety was ethanol, that is even used today to sified as unselective, or partially selective enhance courage and to fight social anxiety by consuming an initial drink at invitations and recep- for certain receptor pools. Benzodiazepines tions. Ethanol at low doses enhances the action of GABA at GABAA receptors and by that, reduces for example act on receptors that contain anxiety. Ethanol might modulate individual GABAA receptor subtypes with differential potency and an alpha1, alpha2, alpha3, or alpha5 sub- seems to act at GABAA receptors via several binding sites at different locations. The actual locations unit together with a neighboring gamma2 of these sites have not been unequivocally unraveled. At higher concentrations, all GABAA receptors subunit, see . Fig. 1. Today it is clear from will become involved in ethanol action, but also other transmitter systems will be modulated by ethanol, leading to sedation, hypnotic actions, and anesthesia. At even higher concentrations, etha- partially selective compounds and genet- nol interacts with many proteins and transmitter systems and develops a cell toxicity independent ically manipulated animals that recep- from the GABA system. tors containing any one of these four al- In addition to ethanol and barbiturates, methaqualone, meprobamate, and anesthetics, such as pha subunits influence distinct brain func- chloralhydrate, were used as anxiolytics. All these drugs seem to act via GABAA receptors, but due to tions in vivo. The sedating and hypnotic their strong sedative action they were not suitable for treating anxiety disorders. The serendipitous effects are mainly mediated by alpha1- re- identification of the anxiolytic action of the benzodiazepine chlordiazepoxide by the pharmaceutical company Hoffmann La Roche then led to a much more specific treatment of these disorders and ceptors, while anxiolytic and antihyperal- since 1960, to a rapid clinical introduction of Librium, Valium, and other benzodiazepines. Benzo- gesic effects can be elicited via alpha2 con- diazepines, due to their anxiolytic, anticonvulsive, muscle-relaxant, and sedative-hypnotic actions, taining receptors [2]. Nonsedating anxio- became soon the most heavily prescribed drugs in clinical use at that time. But only 15 years after lytics or GABAergic antihyperalgetics that their introduction in clinical use evidence accumulated that these compounds might act via the may relieve pathological pain should thus benzodiazepine site of GABA receptors, that are located at the extracellular interface between an A be primarily and strongly acting at alpha2 alpha and a gamma subunit of these receptors (alpha + gamma −, see . Fig. 1). Classical ben zo- receptors, while having ideally no, or only diazepines cannot, or only weakly, distinguish between different GABAA receptor subtypes that are composed of alpha, beta, and gamma subunits, and they thus all have a relatively similar action in weak effects on alpha1 receptors. Alpha5 vivo. Benzodiazepines that exhibit a relatively high selectivity for certain GABAA receptor subtypes subunit containing receptors allow the bi- (see . Excursion 2) already have been developed but are not yet admitted for clinical use. directional modulation of cognitive (mem- In the meantime, around hundred different compound classes have been identified that act via the ory related) processes. A negative alloste- benzodiazepine binding site of GABAA receptors. Several of these compounds are already clinically ric modulator selective for these receptors used. The best known of these compounds is zolpidem, that is marketed in Austria under the com- mercial name Ivadal as a sleeping pill. While zolpidem exhibits a certain receptor subtype-selectivi- is currently investigated in a clinical trial in ty, this is not the case with the structually unrelated compounds zopiclone, zaleplon, and divaplon, Down syndrome patients [2]. that also act via the benzodiazepine site of GABAA receptors. The comparatively small population of In addition, endogenous neurosteroids, such as allopregnanolone, can modulate GABAA receptors delta containing receptors is also turning with differential potency. This can contribute to the mood-variations during puberty and during the out to form an interesting receptor pool female cycle, as well as during and after pregnancy. The steroids act via several steroid binding sites that due to its high steroid sensitivity con- in the transmembrane region of GABA receptors. At low concentrations they enhance the action of A trols many hormonally influenced func- GABA, at higher concentrations they also can activate GABAA receptors directly. Synthetic steroids are used as anesthetics (alfaxalone) or as antiepileptics (ganaxolone). tions in the limbic system and the cor- In addition, some inhalation anesthetics, such as chloroform, isoflurane, halothane, and some in- tex. Compounds with selectivity for one travenous anesthetics, such as etomidate or propofol, are acting via binding sites in the transmem- or the other subtype within this receptor brane region of GABAA receptors. pool should thus have much more specific effects than the so far used GABAA re- to conformational changes of the receptor mazenil does when it is used as antagonist ceptor medications. Possible implications that enhance (positive allosteric modula- of benzodiazepine overdose. Such bidi- of delta containing receptors in pathologi- tion) or reduce (negative allosteric mod- rectional effects of ligand families do not cal mechanisms and as therapeutic targets ulation) the GABA effect. Ligands that only occur among the benzodiazepines, are under intense investigation [3]. induce no or insufficient conformational where positive, negative, and silent mod- Recent developments focus also on changes are called null modulators (or si- ulators are known, but also presumably GABAA receptors on non-neuronal cell lent modulators). They exert no effect on among ligands of other binding sites on types. Outside the CNS, GABAA recep- the receptor activity, but can effectively the receptor. Clear proof of this hypoth- tors are found in the autonomic nervous displace positive or negative modulators esis is still lacking, however, because for system and in endocrine and reproductive and thus, inhibit their actions — as flu- organs [4], in the beta-cells of the pan-

98 | e-Neuroforum 4 · 2015 Abstract creas [5], in certain cells of the immune e-Neuroforum 2015 · 6:97–103 DOI 10.1007/s13295-015-0016-9 system [6], and in epithelial and smooth © Springer-Verlag Berlin Heidelberg 2015 muscle cells of the respiratory tract [7]. M. Ernst · W. Sieghart Thus, it was shown recently, that experimental benzodiazepines with alpha5 re- GABAA receptor subtypes: structural variety ceptor preferring action relax precon- raises hope for new therapy concepts tracted airway smooth muscle and influ- Abstract ence their intracellular Ca2+ [7]. Com- GABAA receptors are ligand-gated chloride have provided insights into the possible loca- pounds with this profile could thus be de- ion channels composed of five subunits that tion of drug interaction sites. Some of these veloped into asthma medications. GABAA can be opened by GABA, and modulated by sites have been confirmed by experimental receptors also occur in many tumor types multiple drugs, some of utmost clinical im- studies. For many receptor ligands, however, and seem to influence tumor cell growth. portance. GABAA receptors occur in the ner- binding sites are not yet known. Here we will vous system as well as in peripheral tissues briefly review the function of distinct types From all these studies receptor subtypes where their function is largely unknown. The of GABAA receptors and provide structural in- emerge that are involved in highly specif- existence of multiple GABAA receptor sub- sights and experimental evidence on binding ic functions in neuronal and nonneuronal types with distinct subunit composition leads sites for ligands that could be of considerable cells. Selective ligands for these subtypes to multiple homologous binding sites with clinical interest. promise enormous therapeutic potential. different degrees of similarity. Crystal structures of proteins homologous to GABAA re- Keywords ceptors and of a GABAA receptor subtype, GABAA receptor · Subtypes · Structure · Structure of GABAA receptors combined with homology modeling studies, Binding sites · Allosteric modulation and related cys-loop receptors

Many of the pharmacological and global structural differences between subtypes The binding pocket that is formed by structural properties of GABAA receptor thus rest on sequence data and homolo- the transmembrane domain interface dis- subtypes were well known prior to struc- gy models. The so far most investigated plays less variability within subunit class- tural studies providing atomic resolution, binding sites are at the extracellular inter- es, but is sufficiently different between such as the localization of the GABA and faces between the principal subunit and classes to be specific for defined receptor benzodiazepine binding sites at specif- its loops A, B, and C and the complemen- pools. For example, the binding site form- ic subunit interfaces, see . Fig. 1. Struc- tary subunit and its loops D, E, F, and G; ing sequence on the plus side of theta, ep- tural details as well as the localization of and at the transmembrane interfaces be- silon, and pi subunits is distinctly differ- many other allosteric binding sites were tween the principal segments of TM 2,3 ent from all others. Thus, this site also unclear for a long time though. There- and the complementary segments of TM would be suitable for selective targeting fore, any discovery of subtype-specific li- 1,2 (see . Fig. 1). Additional binding sites by appropriate ligands. gands was mostly the result of broad drug in the extracellular and transmembrane screens. Sometimes serendipity helped, domains were described in the literature Can the structural variability of binding pock- but such discoveries were only to a small for various compounds, but will not be ets be predicted using crystal structures?. For extent hypothesis driven. Since the first discussed here. all crystal structures, the four letter pro- crystal structure of a homologous pro- Regarding the question of subtype spe- tein database (PDB, http://www.rcsb.org) tein was released in 2001 [8], efforts have cific binding motifs, a sequence compar- identifiers will be indicated below. The on- focused on homology modeling of bind- ison among the binding site forming seg- ly crystal structure of a GABAA receptor ing sites at the atomic level. Such models ments of the 19 GABAA receptor sub- available so far, the one of the beta3 homo- were also successfully used by our group units reveals that the extracellular loops pentamer ([10] 4COF) contains no GABA to investigate the extracellular alpha + C and F feature the largest variability. binding site at the extracellular interface, beta − interface, as well as for computa- Thus, in principle each extracellular in- but one for the synthetic agonist benzam- tional docking studies [9]. The first crys- terface between two subunits is unique idine and for histamine [1]. Even though tal structures were from remote homo- and can therefore offer a more or less spe- no structures of other subtypes are avail- logues (< 20 % sequence identity), but the cific pocket for selective ligands if the se- able yet, analysis of crystal structures of development has reached a new milestone quence differences translate into corre- homologous proteins allows to estimate with the first crystal structure of a GABAA sponding structural differences. Examples the structural variability among subtypes. receptor [10] which raises hopes to soon for selective ligands are some experimen- The same holds true for conformational pursue structure-guided drug design for tal benzodiazepines such as the strongly flexibility. specific receptor subtypes. alpha5-preferring compound SH-053-2’F- For a realistic estimate of structur- R-CH3 [2] that is acting at the benzodi- al and conformational variability, crystal Which binding sites feature subtype-specif- azepine binding site, or the exquisitely al- structures of the glutamate-gated chloride ic structures?. Since so far a crystal struc- pha6/beta2,3 selective “compound 6” that channel (GluCl, particularly 3RIF — one ture of only a single GABAA receptor’s was recently described by us [1] and is act- of the avermectin bound structures — and subtype is available, all statements about ing via the alpha + beta − interface. TNV, the apo-GluCl structure), but also

e-Neuroforum 4 · 2015 | 99 Review article

the alpha isoforms must be expected? The sequence identity between beta3 and gam- ma2 is 34 %, between beta3 and alpha1 it is 36 %. This is in the range shown by two crystallized proteins. One of them (2BYS) is the wild type AChBP, the other is the incomplete conversion of AChBP into the extracellular domain (ECD) of the alpha7 nAChR (5AFH), in which the extracellular interface was converted (chimerized). These two proteins can be regarded as “synthetic subtypes”. In these two structures we find very high structural similarity in loops A, G, and D, small differences in loop E, large differences — as expected on the basis of sequence differences — in loops C and F, see . Fig. 2b. Furthermore, surprisingly large differences are observed around loop B which would not have been expected on the basis of the rather similar sequences. Differences between the GAB- Fig. 1 8 a Schematic primary structure of GABAA receptor subunits, indicating the positions of the in- AA receptor beta3 structure and other sub- terface-forming segments A to G of the principal- and complementary subunits, the positions of the disulfide-forming cysteines (yellow bars), the four transmembrane domaines (TM1-4) with the posi- unit classes, such as gamma or alpha sub- tion of their interface-forming plus and minus segments, and the intracellular domaine (ICD). b Pen- units, will be in the same ballpark range as tameric structure of the most abundant GABAA receptor subtype consisting of two alpha1, two beta2, those shown in . Fig. 2b. Thus, homolo- and one gamma2 subunit, view from extracellular, and based on a homology model using the crys- gy models of alpha or gamma subunits will tal structure of Miller and Aricescu as a template. This structure has no intracellular domain. The five be fairly accurate in the extracellular loops “backbone”-helices shown are the five TM2 helices that form the chloride channel. In addition, the lo- cation of the GABA, benzodiazepine (BZ) and alpha + beta − binding sites of the extracellular interfac- A, G, D and E, but less so in the interest- es is indicated by arrows. The + and − side of the subunits is indicated at the outside circumference. ing variable regions C and F, and will also c Side view of a homology model based on the crystal structure of Miller and Aricescu where the in- be uncertain in the loop B region. Further- tracellular domain is based on the crystal structure of the 5-HT3 receptor. Shown is a principal and a more, the bound conformation of larger li- complementary subunit with the extracellular loops in pink, orange, red, green, and cyan, and the plus gands than the benzamidine present in the (blue) and minus (purple) side of the transmembrane-interfaces, as well as the intracellular domains in the same color code as in a beta3 crystal structure can also not be predicted reliably. To be able to estimate the range of inaccuracy of a model on the ba- structures of more remote homologues, lar loops C and F on the one hand, and sis of structure comparisons like this one specifically of the serotonine type 3 re- from subtle differences in structural de- is very helpful for the interpretation of the ceptor (4PIR), and of a chimera between tails on the other hand. Which differences models. the alpha7-nAChR and the acetylcholine between subtypes must be expected, and binding protein (AChBP, 5AFH and 2BYS) how much will subunits from other class- Binding sites at the trans-membrane inter- are very useful [11–15]. es, such as alpha or gamma, differ from face. This binding site is very similar in The high similarity between the GA- the beta3 structure? These questions de- three different receptors, see . Fig. 2a and BAA receptor structure (4COF) with the mand different answers for the different c, which argues for a high degree of struc- avermectin bound GluCl structure (3RIF) domains and binding sites of the receptor. tural conservation in this region/micro- which share 36 % sequence identity, and domain. The specificity of ligands that even with the more remotely homolo- Binding sites at extracellular interfaces. Does bind here often rests on a single sidechain, gous 5-HT3 receptor (4PIR, only 14 % se- the structure of the beta3 homo-pentamer as is the case for an asparagine/serine quence identity) demonstrates impres- enable us to generate accurate models for polymorphism in the TM2 of the GABAA sively that the structural conservation example of the alpha + gamma − benzo- receptor beta subunits [1], which makes in the cys-loop family is extraordinarily diazepine sites in specific subtypes? Both the beta1 subunit insensitive to etomi- high, see . Fig. 2. The enormous diver- subunits (plus and minus side) contribute date. This binding site is thus not affected sity which furnishes each family member highly conserved segments to this binding by (backbone) structural variability, not and every receptor subtype with unique site, but also highly variable ones, such as even in situations of low sequence simi- pharmacological and electrophysiological loops C and F — which additionally are larity, owing to the high structural conser- properties thus stems from a few highly flexible as well. Which differences between vation of the four helix motif — but it is all variable domains, such as the extracellu- beta3 and gamma2, or between beta3 and the more “in motion”. The TM domains

100 | e-Neuroforum 4 · 2015 Excursion 2 From the discovery of the neurotransmitter GABA to the lization cocktails and the orientation and heterogeneity of GABAA receptors contacts of the receptors within the crystal. However, as the GluCl features wide- In 1950, the amino acid GABA was first identified in the brain, after it had been known to exist in ly different conformations under differ- bacteria, yeasts, and plants. In 1958, it became clear that the conductance increase in crustacean muscle fibers caused by inhibitory nerve stimulation resulted from an increased permeability ent conditions, we must assume that the towards chloride ions and that externally applied GABA exactly duplicated the action of inhibitory mobility of this region is comparable al- nerve stimulation. During the following years a possible role of GABA as a transmitter compound so in GABAA receptors and that the size was heavily disputed with many pro’s and con’s. The identification of bicuculline as an antagonist and shape of the binding pockets undergo of GABA action in 1970 then provided a new powerful tool to study the distribution of GABAergic comparable motion as in the GluCl. This synapses in the CNS. Subsequently, it became clear that most but not all of the GABA actions could is well compatible with the observation be blocked by bicuculline (the respective receptors were defined as GABA receptors), whereas A that ligands of the TM interface binding other GABA actions could be blocked by baclofen (GABAB receptors). Whereas GABAA receptors are GABA-activated anion channels, GABAB receptors are G-protein coupled receptors with a different sites often show strong use-dependent po- structure, function and pharmacology. tency differences [16], that can be caused Benzodiazepines, such as chlordiazepoxide (Librium) or diazepam (Valium), had been introduced by allosteric interactions with other bind- into therapeutic use in the 1960’s by the pharmaceutical company Hoffmann La Roche. In 1975, ing sites. For computational methods Willi Haefely, the research director of Hoffmann La Roche, proposed that benzodiazepines might such as in silico docking ideally the struc- act via GABAA receptors. In 1977, a high affinity benzodiazepine binding site was identified in brain membranes by Möhler and Okada, and by Braestrup and Squires, and it soon was demonstrated ture with the correct state should be cho- that binding of the radiolabeled benzodiazepines [3H]diazepam or [3H]flunitrazepam to this “ben- sen as modeling template — and that is zodiazepine receptor” not only could be enhanced by GABA, but also by barbiturates, neuroactive not necessarily the GABAA receptor itself. steroids and other drugs assumed to act via GABAA receptors. Since it is often not clear which function- 3 In 1980, HannsMöhler discovered that [ H]flunitrazepam could be used as a photoaffinity label for al state a crystal structure best matches, it the benzodiazepine receptor protein. Using this discovery, Sieghart and Karobath [17] for the first is recommend to compare several crystal time provided evidence for a molecular heterogeneity of GABAA receptors. Although this evidence was then further supported by a variety of biochemical and pharmacological studies from the same structures and not to overinterpret mod- as well as from other laboratories, a heterogeneity of GABAA receptors was finally believed only els based on a single template structure. after 1987, when the cloning of various GABAA receptor subunits had been reported by the group The present analysis of the homopen- of Peter Seeburg and others. Today it is clear that a total of 19 distinct GABA receptor subunits A tameric beta3 GABAA receptors in com- (alpha1–6, beta1–3, gamma1–3, epsilon, delta, pi, theta and rho1–3) as well as several alternatively parison with structures of homologues spliced isoforms of some of these subunits are expressed in mammalian nervous system as well as demonstrate the impressively conserved in peripheral mammalian tissues. Since GABAA receptors are composed of five subunits that form the central anion channel, many thousands of different GABAA receptor subtypes could theoretical- scaffold, but also the subtle structural ly be formed from random combinations of the 19 subunits. The overall number of distinct receptor variability of the extracellular domain and subtypes is limited, however, by the rules of assembly, leading to defined subunit stoichiometries, the very high mobility of the upper part as well as by preferential subunit partnerships and excluded subunit combinations. of the transmembrane domain. The crys- Thus, the majority of GABA receptors is composed of two alpha, two beta, and one gamma A tal structures that are now available still subunit. In these receptors a total of four alternating alpha and beta subunits are connected by a gamma subunit (. Fig. 1). Whether all receptors composed of alpha, beta, and gamma subunits or preclude specific conclusions on structur- also those composed of alpha, beta, and delta subunits exhibit the same subunit stoichiometry and al differences between GABAA receptor subunit arrangement presently is not known or controversial. The subunit combinations containing subtypes, but they provide excellent guid- epsilon, theta, or pi subunits are at present unclear. The rho subunits form homo-oligomeric or het- ance in estimating local uncertainties of ero-oligomeric receptors with other rho subunits, and may co-assemble with additional subunits homology models. Particularly notewor- as well. GABA receptor subtypes might contain up to five different subunits, but gamma subunits A thy is the structural variability of the loop seem not to co-assemble with other gamma- or delta-subunits. Nevertheless, even then it was estimated that a maximum of the order of 800 subunit combinations B region (See . Fig. 2b) despite the high of the types observed so far might exist [18]. The true number is likely to be far smaller. Recently, sequence conservation of this region. the evidence accumulated for the existence of 11 native receptor subtypes was considered strong In conclusion, it can be stated that the enough for the assumption that they really exist in vivo. An additional 15 receptor subtypes seem progress in understanding the complexi- to exist with high probability [18], but the number for sure will increase. The actual heterogeneity of ty of the GABAA receptor family is accel- GABAA receptors is thus much larger than previously assumed. erated tremendously by the synergy between structural biology, biochemistry, of the GABAA receptor, the 5-HT3 recep- one hand, and 4TNV as well as 4TNW and pharmacology. The identification of tor (14 % sequence identity with the GA- from [11] on the other hand) are very dif- novel and interesting receptor subtypes in BAA R) and one of the GluCl structures ferent, see . Fig. 2. This is due to the large neuronal and non neuronal tissues and the are very similar and the protein backbone conformational flexibility of this region. study of their function makes them clini- overlays very well, see . Fig. 2. All three In this respect it must be noted that cally interesting target molecules. The de- proteins are in a similar conformation structures in crystals not necessarily look termination of their subunit arrangements and the region around the TM-interface like structures in the cell, because they are and of their structure at the atomic level binding site is structurally very conserved. not only determined by the presence or should enable structure-guided drug de- In contrast, the TM domains of different absence of ligands, but also by stabilizing velopment and soon lead to the identifi- GluCl structures (3RIF from [14] on the antibodies, lipids, detergents, the crystal- cation of specific ligands. These not only

e-Neuroforum 4 · 2015 | 101 Review article

Corresponding address

Asst. Prof. Priv. Doz. Dr. M. Ernst Department of Molecular Neurosciences Center for Brain Research, Spitalgasse 4, 1090 Vienna margot.ernst@meduniwien. ac.at

Prof. Dr. W. Sieghart Center for Brain Research Spitalgasse 4, 1090 Vienna

Asst. Prof. Priv. Doz. Dr. Margot Ernst obtained her PhD at the Georgia Institute of Technology (USA) in computational chemistry. She then did a postdoctoral study in theoretical chemistry. After that she shifted her interests to the life sciences and in 2001 started to work on GABAA receptors in the group of Werner Sieghart at the Center for Brain Research, Vienna. The publications of an ever increasing number of crystal stuctures of cys-loop receptors motivated her to concentrate on modeling of the GABAA receptor structure and their binding sites. Combined with electrophysiological studies on recombinant GABAA receptors she succeeded to demonstrate the existence of a novel binding site for pyrazoloquinolinones at the alpha + beta − interface of GABAA receptors. Her current research is focussed on the structure of cys-loop receptors, the identification of allosteric binding sites thereon, and the development of receptor subtype-selective ligands for the GABAA receptor. In 2014 she became Ass. Prof. and Priv. Doz. at the Center for Brain Research, where she currently holds a career track position.

Prof. Dr. Werner Sieghart studied chemistry at the University of Vienna, and obtained his PhD at the Fig. 2 8 a Two subunits of the homo-oligomeric beta3 GABA receptor structure (4COF). The principal Institute for Biochemistry in Vienna with Hans Tuppy. A He then moved to the Department of Biochemical subunit is shown in blue, the complementary subunit in red. This structure is superposed by the struc- Psychiatry at the University Clinic for Psychiatry, ture of the 5-HT3 receptor (4PIR, cyan). The protein backbone of these two receptors is quite similar to Vienna, where he worked with Manfred Karobath each other. b Comparison between the acetylcholine binding protein (AChBP) (2BYS, gray) and the al- on taurine transport in synaptosomal fractions. In pha7-nicotinic acetylcholine receptor (nACHR)-AChBP chimera (5AFH, brown), in which the loops of his two and a half years of postdoctoral studies with the alpha7 nAChR form the pocket of the AChBP. The loops are shown in the same color code as in Paul Greengard at the Department of Pharmacology, . Fig. 1. c Binding site at the trans-membrane (TM)-interface of the glutamate-activated chloride Yale University, he for the first time demonstrated channel (GluCl) (3RIF, bluegreen) in comparison with the homo-oligomeric beta3 GABAA receptor (blue a connection between the phosphorylation of a and red), shown as top view and side view. In the area of the TM-interface binding site an alpha carbon protein and secretion in mast cells, and also for atom is shown in a space filling mode at a homologous position of the TM2 and TM3 of the one as well the first time demonstrated that the same protein (synapsin) could be phosphorylated by a cAMP- and as at the TM2 and TM1 helix of the other subunit to allow orientation at the interface. In the perspec- a Ca2+-dependent protein kinase. After his return tive view from above it can be seen that the atoms are located at both sides of the interface, and are to the University Clinic for Psychiatry in Vienna, he located quite similarly in the two structures. In the perspective side view the TM2 of the minus side is started to direct the Department of Biochemical located exactly behind the TM1 and is thus, partially obscured. d Comparison of the same binding site Psychiatry in 1980, after Manfred Karobath had moved at the apo-GluCl receptor (4TNV, green) and the GABAA receptor (blue and red)from identical perspec- to the company Sandoz in Basle to become director tives as in c. In the top view of d, the green atom in TM1, that is seen exactly from above in a view par- of preclinical research. In 1980, he also for the first allel to the TM1 helix, is obscured by the red atom and is marked by a green circle. It clearly can be seen time demonstrated a molecular heterogeneity of that the TM helices fit quite well to each other in c, but have clearly different conformations in d. Simi- GABAA receptors and since that time he is applying larly, the labeled amino acid atoms differ much more in d than in c biochemical, pharmacological, electrophysiological and immunohistochemical investigations to study the composition, structure, pharmacology and function of GABAA receptors subtypes. Starting in 1980, he can then be developed to novel therapeutic other organs in which GABA acts as sig- also established a specialized clinical laboratory principles for the treatment of neuropsy- naling molecule. for the determination of drugs of abuse, lithium, chiatric disorders, but also for disorders of anticonvulsants, and hormones, and later on also

102 | e-Neuroforum 4 · 2015 established molecular genetic investigations in the 15. Spurny R et al (2015) Molecular blueprint of al- blood of patients for linkage and association studies. losteric binding sites in a homologue of the ag- In 1982, he became assistant professor, in 1988, he onist-binding domain of the alpha7 nicotin- obtained the title Professor for Neurobiochemistry, ic acetylcholine receptor. Proc Natl Acad Sci U S A and in 1999, he and his research group moved to 112(19):E2543–E2552 the newly founded Center for Brain Research in 16. Franks NP (2015) Structural comparisons of ligand- Vienna. In 2002, he was appointed as Professor for gated ion channels in open, closed, and desensi- Biochemical and Molecular Pharmacology of the tized states identify a novel propofol-binding site Nervous System and as Head of the Department for on mammalian gamma-aminobutyric acid type A Biochemisty and Molecular Biology at the Center for receptors. Anesthesiology 122(4):787–794 Brain Research. He directed this Department until 17. Sieghart W, Karobath M (1980) Molecular het- he retired in October 2011. Currently, he is Guest erogeneity of benzodiazepine receptors. Nature Professor at the Center for Brain Research in Vienna. 286(5770):285–287 18. Olsen RW, Sieghart W (2008) International Union Acknowledgments. We gratefully acknowledge of Pharmacology. LXX. Subtypes of gamma-ami- financial support by the Austrian Science Fund nobutyric acid(A) receptors: classification on the (Margot Ernst and Werner Sieghart) over many years basis of subunit composition, pharmacology, and as well as financial support by grants of the European function. Update. Pharmacol Rev 60(3):243–260 Union to Werner Sieghart.

References

1. Sieghart W (2015) Allosteric modulation of GABAA receptors via multiple drug-binding sites. Adv Pharmacol 72:53–96 2. Rudolph U, Mohler H (2014) GABAA receptor subtypes: therapeutic potential in Down syndrome, affective disorders, schizophrenia, and autism. An- nu Rev Pharmacol Toxicol 54:483–507 3. Brickley SG, Mody I (2012) Extrasynaptic GABA(A) receptors: their function in the CNS and implications for disease. Neuron 73(1):23–34 4. Gladkevich A et al (2006) The peripheral GABAer- gic system as a target in endocrine disorders. Au- ton Neurosci 124(1–2):1–8 5. Wan Y, Wang Q, Prud’homme GJ (2015) GABAergic system in the endocrine pancreas: a new target for diabetes treatment. Diabetes Metab Syndr Obes 8:79–87 6. Barragan A et al (2015) GABAergic signalling in the immune system. Acta Physiol (Oxf) 213(4):819– 827 7. Gallos G et al (2015) Selective targeting of the al- pha5-subunit of GABAA receptors relaxes airway smooth muscle and inhibits cellular calci- um handling. Am J Physiol Lung Cell Mol Physiol 308(9):L931–L942 8. Brejc K et al (2001) Crystal structure of an ACh- binding protein reveals the ligand-binding domain of nicotinic receptors. Nature 411(6835):269– 276 9. Richter L et al (2012) Diazepam-bound GABAA receptor models identify new benzodiazepine binding-site ligands. Nat Chem Biol 8(5):455–464 10. Miller PS, Aricescu AR (2014) Crystal structure of a human GABAA receptor. Nature 512(7514):270– 275 11. Althoff T et al (2014) X-ray structures of GluCl in apo states reveal a gating mechanism of Cys-loop receptors. Nature 512(7514):333–337 12. Hansen SB et al (2005) Structures of Aplysia ACh- BP complexes with nicotinic agonists and antago- nists reveal distinctive binding interfaces and conformations. EMBO J 24(20):3635–3646 13. Hassaine G et al (2014) X-ray structure of the mouse serotonin 5-HT3 receptor. Nature 512(7514):276–281 14. Hibbs RE, Gouaux E (2011) Principles of activation and permeation in an anion-selective Cys-loop receptor. Nature 474(7349):54–60

e-Neuroforum 4 · 2015 | 103