Covalent modification of GABAA receptor isoforms by a diazepam analogue provides evidence for a novel

Description

JOURNAL OF NEUROCHEMISTRY | 2008 | 106 | 2353–2363 doi: 10.1111/j.1471-4159.2008.05574.x,1 ,1 ,1*Institute of Biochemistry and Molecular Medicine, University of Bern, Bern, SwitzerlandLaboratoire de Chimie Bio-Organique, Unite´ Mixte de Recherche 7175 LC1 CNRS, Faculte´ de Pharmacie, Universite´ Louis PasteurStrasbourg, Illkirch Cedex, FranceAbstract followed by a H126Cb c , a H101Cb c and a H101Cb c ,3 2 2 1 2 2 2 2 2Classical benzodiazepines, for example diazepam, interact while a H105Cb c receptors show little interaction. Our5 2 2with a b c GABA receptors, x = 1, 2, 3, 5. Little is known results allow conclusions about the relative apposition ofx 2 2 Aabout effects of a subunits on the structure of the binding a H101andhomologouspositionsina ,a ,a anda withthe1 2 3 5 6pocket. We studied here the interaction of the covalently position occupied by –Cl in diazepam. During this study wereacting diazepam analog 7-Isothiocyanato-5-phenyl-1,3- found evidence for the presence of a novel site for ben-dihydro-2H-1,4-benzodiazepin-2-one (NCS compound) with zodiazepines that prevents modulation of GABA receptorsAa H101Cb c and with receptors containing the homologous via the classical benzodiazepine site. The novel site poten-1 2 2mutation, a H101Cb c , a H126Cb c and a H105Cb c . tially contributes to the high degree of safety to some of these2 2 2 3 2 2 5 2 2This comparison was extended to a R100Cb c receptors as drugs. Our results indicate that this site may be ...

Informations

• Publié par: Usle
• Nombre de lectures: 22
• Langue: English

Extrait

JOURNAL OF NEUROCHEMISTRY

,1

|2008|106|2353–2363

,1

,1

doi: 10.1111/j.14714159.2008.05574.x

*Institute of Biochemistry and Molecular Medicine, University of Bern, Bern, Switzerland rehc7e71etedeRhc,Faculte5LC1CNRSeicainU,Ped´mrahuiLoassPrsvee´itetruaLotriobarmihieCedrg-OioeBU,euqinaxiM´etin Strasbourg, Illkirch Cedex, France

Abstract Classical benzodiazepines, for example diazepam, interact witha x b 2 c 2 GABA A receptors, x = 1, 2, 3, 5. Little is known about effects ofasubunits on the structure of the binding pocket. We studied here the interaction of the covalently reacting diazepam analog 7-Isothiocyanato-5-phenyl-1,3- dihydro-2H-1,4-benzodiazepin-2-one (NCS compound) with a 1 H101Cb 2 c 2 and with receptors containing the homologous mutation,a 2 H101Cb 2 c 2 ,a 3 H126Cb 2 c 2 anda 5 H105Cb 2 c 2 . This comparison was extended toa 6 R100Cb 2 c 2 receptors as this mutation conveys to these receptors high affinity towards classical benzodiazepines. The interaction was studied at the ligand binding level and at the functional level using electro- physiological techniques. Results indicate that the geometry ofa 6 R100Cb 2 c 2 enables best interaction with NCS compound,

GABA A receptors mediate neuronal inhibition. They are composed of five subunits surrounding a central chloride ion selective channel (Macdonald and Olsen 1994; Rabowet al. 1995; Sieghart 1995; Sieghart and Sperk 2002). Some GABA A receptor isoforms have a high affinity binding site for classical benzodiazepines (Sigel and Buhr 1997; Sigel 2002). The classical benzodiazepine diazepam binds with high affinity and positively modulates recombinanta 1 b x c 2 (x = 1, 2, 3),a 2 b x c 2 ,a 3 b x c 2 anda 5 b x c 2 GABA A receptors, while zolpidem binds with high affinity toa 1 b x c 2 , and with lower affinity toa 2 b x c 2 ,a 3 b x c 2 , anda 5 b x c 2 . Zolpidem strongly potentiates currents ata 1 b x c 2 ,a 2 b x c 2 anda 3 b x c 2 and only weakly affectsa 5 b x c 2 receptors (Sannaet al.2002; Baur and Sigel 2007). Both compounds do not affecta 4 b x c 2 anda 6 b x c 2 receptors. The specificity for classical ben- zodiazepines is due to the amino acid residue homologous to

followed bya 3 H126Cb 2 c 2 ,a 1 H101Cb 2 c 2 anda 2 H101Cb 2 c 2 , whilea 5 H105Cb 2 c 2 receptors show little interaction. Our results allow conclusions about the relative apposition of a 1 H101 and homologous positions ina 2 ,a 3 ,a 5 anda 6 with the position occupied by –Cl in diazepam. During this study we found evidence for the presence of a novel site for ben- zodiazepines that prevents modulation of GABA A receptors via the classical benzodiazepine site. The novel site poten- tially contributes to the high degree of safety to some of these drugs. Our results indicate that this site may be located at the a/bsubunit interface pseudo-symmetrically to the site for classical benzodiazepines located at thea/cinterface. Keywords:benzodiazepine, GABA, GABA A receptors, receptor isoforms. J. Neurochem.(2008)106, 2353–2363.

a 1 H101, which is histidine ina 1 ,a 2 ,a 3 anda 5 and arginine ina 4 anda 6 (Wielandet al.1992; Davieset al.1998; Dunn et al.1999). Replacement of this histidine by arginine ina 1 , a 2 ,a 3 anda 5 abolishes modulation by diazepam (Benson

Received April 11, 2008; revised manuscript received July 7, 2008; accepted July 7, 2008. Address correspondence and reprint requests to Erwin Sigel, Institute of Biochemistry and Molecular Medicine, University of Bern, Bu¨ hlst- rasse 28, CH-3012, Switzerland. E-mail: [e-mail] 1 These authors contributed equally to this study. Abbreviations used: DMSO, dimethyl sulfoxide; HEK, human embryonic kidney; LR, receptor ligand complex, ligand bound to the newly described site 2; NCS compound, 7-isothiocyanato-5-phenyl-1,3- dihydro-2H-1,4-benzodiazepin-2-one; RL, receptor ligand complex, ligand bound to the classical benzodiazepine site (site 1); RLR, receptor ligand complex, ligand bound to both, site 1 and site 2.

2008 The Authors Journal Compilation2008 International Society for Neurochemistry,J. Neurochem.(2008)106, 2353–2363

2353

2354

|

R. Bauret al.

et al.1998). Thus, it is well documented that the benzodi- azepine binding site differs in different receptor isoforms. Benzodiazepines bind at the subunit interfacea x /c 2 (x = 1, 2, 3, 5; Sigel and Buhr 1997; Sigel 2002). Exclusively the type ofasubunit adjacent to thec 2 subunit seems to affect the response of a receptor to these compounds (Minier and Sigel 2004). Diazepam is arranged in the binding pocket of a 1 b 2 c 2 receptors such that the –Cl atom is close toa 1 H101C (Berezhnoyet al.2004, 2005; Tanet al.2007a,b) and the 3¢-atom close toa 1 S205C anda 1 T206C (Tanet al.2007c). In addition to this high affinity site a very low affinity site specific for diazepam located in the lipid bilayer has been described (Walterset al.2000). We were interested in further characterizing the benzodi- azepine binding pocket in different receptor isoforms. We extended previous observations on the interaction of the covalently reacting diazepam analog NCS compound with a 1 H101Cb 2 c 2 receptors (Berezhnoyet al.2004, 2005; Tan et al.2007a,b) to receptors carrying the homologous muta- tion ina 2 H101Cb 2 c 2 ,a 3 H126Cb 2 c 2 anda 5 H105Cb 2 c 2 . We also includeda 6 R100Cb 2 c 2 receptors as this mutation conveys to these receptors high affinity towards diazepam. The interaction was studied at the ligand binding level and at the functional level using electrophysiological techniques. Results indicate that the geometry ofa 6 R100Cb 2 c 2 allows best interaction with NCS compound, followed by a 3 H126Cb 2 c 2 ,a 1 H101Cb 2 c 2 anda 2 H101Cb 2 c 2 , while a 5 H105Cb 2 c 2 receptors show little interaction. During this study, we obtained evidence for the presence in a receptor pentamer of a new site for benzodiazepines, negatively interacting with the well-characterized site. We putatively locate this inhibitory site to thea/bsubunit interface.

Materials and Methods

Synthesis The substance we used is a derivative of diazepam. In the NCS compound, the -Cl group is replaced by –NCS (Fig. 1). The synthesis is described elsewhere (Berezhnoyet al.2004; Tanet al. 2007b). The compound was dissolved in DMSO (Dimethyl sulfoxide) at a concentration of 10 mM and kept at)20C. Final dilutions in assay medium were prepared immediately before each experiment.

Fig. 1Chemical structure of diazepam and NCS compound.

Transfection of GABA A receptors in HEK293 cells, membrane preparation and radioactive ligand binding assay cDNAs coding for thea 1 ,b 2 , andc 2 S subunits of the rat GABA A receptor were transfected in human embryonic kidney (HEK) 293 cells (American Type of Culture Collection, MD, USA, CRL 1573). Culturing of cells, membrane preparation and radioactive ligand 3 binding assay using [ H]Ro15-1788 have been described before (Tanet al.2007a).

Detection of a covalent reaction As detailed in previous work (Berezhnoyet al.2004; Tanet al. 2007a) this procedure included three steps: incubation of mem- branes expressing recombinant wild type or mutant receptors with the reactive agent followed by extensive washing of the membranes in order to remove non-reacted compound and a radioactive ligand binding assay to determine residual binding. No covalent reaction would result in 100% residual binding, and 100% covalent reaction would result in 0% residual binding.

Time course of the covalent reaction The membranes were re-suspended in phosphate buffer (100 mM KCl, 10 mM KH 2 PO 4 , 0.1 mM EDTA, pH 7.4) using a Glass/ Teflon homogenizer. 50–800lg/mL of protein were incubated in a total volume of 120lL with several concentrations of NCS compound for 30–600 s on ice. Membranes were collected with rapid filtration on a round 7 mm diameter glass fiber filter (GF/C; Whatman) that was placed on a round 12 mm diameter glass fiber filter (GF/C; Whatman), both pre-washed with phosphate buffer. The reaction of NCS compound with the receptor was stopped by washing of the filters six times with 5 mL phosphate buffer, each. The membranes on the small filter were incubated in 0.12 mL 3 phosphate buffer containing 5 nM [ H]Ro15-1788. After 30 min the 7 mm filter was placed on a 12 mm filter and washed six times with 5 mL phosphate buffer each. Radioactivity was determined by liquid scintillation counting. Non-specific binding was determined in the presence of 100lM Ro 15-1788. In control experiments, washing efficiency was estimated by placing radioactivity on the small filter. More than 99.95% of the radioactivity were removed (not shown).

Expression and functional characterization in Xenopus oocytes Capped cRNAs were synthesized (Ambion, Austin, TX, USA) from the linearized plasmids with a cytomegalovirus promotor (pCMV vectors) containing the different subunits, respectively. A poly-A tail of about 400 residues was added to each transcript using yeast poly- A polymerase (United States Biologicals, Cleveland, OH, USA). The concentration of the cRNA was quantified on a formaldehyde gel using Radiant Red stain (Bio-Rad) for visualization of the RNA. Known concentrations of RNA ladder (Invitrogen) were loaded as standard on the same gel. cRNAs were precipitated in ethanol/ isoamylalcohol 19 : 1, the dried pellet dissolved in water and stored at)80C. cRNA mixtures were prepared from these stock solutions and stored at)80C. Xenopus laevisoocytes were prepared, injected and defolliculated as described previously (Sigel 1987; Sigel and Minier 2005). They were injected with 50 nL of the cRNA solution containing wild type or mutateda 1 ora 2 ,a 3 ,a 5 ,a 6 and wild typeb 2 andc 2 subunits at a concentration of 10 nM : 10 nM : 50 nM (Boileauet al.2002), or

2008 The Authors Journal Compilation2008 International Society for Neurochemistry,J. Neurochem.(2008)106, 2353–2363

concatenated subunitsc 2 -b 2 -a 6 ,b 2 -a 1 ,c 2 -b 2 -a 1 andb 2 -a 6 (Minier and Sigel 2004; nomenclature anti-clockwise, see also Fig. 4e) at a concentration of 10 nM : 10 nM and then incubated in modified Barth’s solution (10 mM HEPES, pH 7.5, 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.82 mM MgSO 4 mM Ca(NO, 0.34 3 ) 2 , 0.41 mM CaCl 2 , 100 units/mL penicillin, 100lg/mL streptomycin) at +18C for at least 24 h before the measurements. Electrophysiological experiments were performed by using the two-electrode voltage clamp method at a holding potential of )mM80 mV. The perfusion medium contained 90 mM NaCl, 1 KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 5 mM Na-HEPES (pH 7.4). Allosteric modulation via the benzodiazepine site and covalent reaction were measured at a GABA concentration eliciting 2–5% of the maximal GABA current amplitude. GABA was applied for 20 s alone or in combination with allosteric compound. Modulation of GABA currents was expressed as a percentage of the respective control current amplitudes determined in the absence of modulator. The perfusion system was cleaned between drug applications by washing with DMSO to avoid contamination. The concentration response curve for flurazepam was fitted with the equation current potentiation = max/(1 + K 1 /L + L/K 2 + K 1 / K 2 ), where max is a constant, L is the concentration of flurazepam, K 1 is the apparent dissociation constant of the classical benzodiaz- epine binding site and K 2 is the apparent dissociation constant of the newly described inhibitory site.

Modification of receptor function by the reactive compound Modification by the NCS compound was measured as follows. After obtaining a reproducible response to the application of GABA, the NCS compound freshly diluted to 20lM in perfusion medium was applied for 1 min. Maximal final solvent concentration was 0.1%. This concentration of DMSO did not affect the response to GABA in control experiments. Treatment was followed by several GABA applications in intervals of 4 min to reach a steady level. Subsequently, 10lM zolpidem was co-applied with GABA to ensure a covalent reaction with the NCS compound. The irreversible effect was then calculated as Potentiation [(I after NCS /I before NCS )) 1]∙100%. Rate of modification by the NCS compound on mutant receptors was determined by repeatedly exposing an oocyte to 1lM of NCS compound for 5 s every 4 min. The current amplitude elicited by GABA was determined 3 min after exposure to the NCS compound. Values were standardized to the one measured upon exposure of the receptor for 1 min to 10lM NCS compound.

Results

Introduction We wanted to understand differences in the benzodiazepine binding pocket in different GABA A receptor isoforms. For this purpose, we characterized the covalent interaction of a benzodiazepine-like compound witha 1 H101C and homolo- gous residues in receptors containinga 2 ,a 3 ,a 5 anda 6 . In order to explain data obtained at the binding level, we postulate a novel site for benzodiazepines that prevents modulation via the classical benzodiazepine site. Finally, we

An inhibitory benzodiazepine site|2355

compare the different receptor isoforms at the functional level in their covalent reaction at the classical benzodiazepine binding site.

Binding properties of the NCS compound The NCS compound in which the -NCS group replaces the -Cl atom of diazepam (Fig. 1) has an affinity of 3170 ± 1081 nM ata 1 b 2 c 2 GABA A receptors (Berezhnoy et al.2004). Thus, the reactive molecule retained affinity for the benzodiazepine-binding site.

Properties of GABA A receptors carrying a cysteine point mutation The binding properties of the mutant receptora 1 H101Cb 2 c 2 have been described before (Berezhnoyet al.2004). The homologous mutant receptors containinga 2 ,a 3 ,a 5 and a 6 ,a 2 H101Cb 2 c 2 ,a 3 H126Cb 2 c 2 ,a 5 H105Cb 2 c 2 and 3 a 6 R100Cb 2 c 2 H]Ro15-1788 with an estimatedbound [ affinity between 0.6 and 5.0 nM (data not shown). The mutation to cysteine ofa 6 R100, thus increases affinity to Ro15-1788. When this mutant receptor was expressed in Xenopus oocytes 10lM diazepam potentiated currents elicited by GABA by 108 ± 25% (n= 13). In contrast wild type receptors are not responsive to diazepam (Wielandet al. 1992; Bensonet al.1998). The mutation also makes the receptor responsive to zolpidem (see below).

Irreversible reaction of the NCS compound with mutated receptors First, we investigated the concentration dependence of the covalent reaction. The receptors were exposed to different concentrations of the NCS compound, and subsequently non- covalently bound reactive ligand was removed by extensive washing (see Methods). In such experiments, the NCS compound is expected to first occupy its binding site reversibly. Upon proper apposition of the –SH group of the cysteine of the mutated receptor with the –C atom of the – NCS group from the NCS compound, this is followed by the formation of a covalent bond between ligand and receptor. Residual binding to non-covalently reacted sites was quan- 3 tified by reversible binding of [ H]Ro15-1788. These sites 3 bound [ H]Ro15-1788 with the same affinity as naive receptors (not shown). Data fora 1 H101Cb 2 c 2 including protection from the covalent reaction with other reversible ligands and preven- tion of covalent interaction by previous exposure of the reactive compound to cysteine (Berezhnoyet al.2004, 2005) and other mutant receptors (Tanet al.2007a,b) have been documented earlier. The concentration dependence of the irreversible reaction was compared in mutanta 1 H101Cb 2 c 2 ,a 2 H101Cb 2 c 2 and a 6 R100Cb 2 c 2 receptors after treatment for 1 h with various concentrations of NCS compound. Residual binding in the

2008 The Authors Journal Compilation2008 International Society for Neurochemistry,J. Neurochem.(2008)106, 2353–2363

2356

|

R. Bauret al.

Fig. 2Concentration dependence of the reaction of the NCS com- pound witha 6 R100Cb 2 c 2 (closed circles) mutant receptors. They were exposed to increasing concentrations of the NCS compound for 1 h on 3 ice and extensively washed. [ H]Ro15-1788 was used as radioactive ligand to determine the residual binding, which was converted to percentage of binding sites covalently reacted.a 6 b 2 c 2 wild type receptors (open circle) were only exposed to a single concentration of NCS compound. Data are shown as mean ± SD for three experiments each (triplicates of each point in each experiment).

corresponding wild type receptorsa 1 b 2 c 2 ,a 2 b 2 c 2 anda 6 b 2 c 2 was unaltered by this treatment indicating the absence of a covalent reaction. Figure 2 documents the covalent reaction ina 6 R100Cb 2 c 2 receptors. For reasons we do not understand a 3 H126Cb 2 c 2 anda 5 H105Cb 2 c 2 could not be expressed in HEK293 cells. Therefore these receptors are excluded from the binding but not the functional (see below) analysis. As 30 to 50% of protein was lost during the washing steps, each experiment included a control without reactive ligand. Residual binding in this sample was standardized to 100% and residual binding in the other samples were expressed relative to this value. Table 1 summarizes the data. The apparent affinity ofa 6 R100Cb 2 c 2 receptors is about five times higher than fora 1 H101Cb 2 c 2 anda 2 H101Cb 2 c 2 receptors. It should be noted that the covalent reaction never reached 100% (Table 1, Fig. 2), when residual binding was probed 3 using [ H]Ro15-1788. Two reasons could account for this phenomenon. First, the rate of covalent reaction could be too slow to reach the maximum within the 60 min of the experiment. Second, this phenomenon could be due to an equilibrium between on and off rate of the covalent reaction, the covalent adduct being slowly hydrolyzed to produce the receptor as it was prior to the reaction (Hinmanet al.2006; Macphersonet al.2007). To test these two hypotheses it was important to establish the time course of the covalent reaction and the off-rate of the covalent reaction.

Table 1Reaction of the NCS compound with different mutant receptor isoforms

Receptor

a 1 H101Cb 2 c 2 a 2 H101Cb 2 c 2 a 3 H126Cb 2 c 2 a 5 H105Cb 2 c 2 a 6 R100Cb 2 c 2

App. K d (lM)

a 2.7 ± 0.5 3.6 ± 1.2 n.e. n.e. 0.62 ± 0.18

Max (%)

84 ± 3 70 ± 23 n.e. n.e. 81 ± 9

k app25 (/s)

0.045 ± 0.021 (3) 0.025 ± 0.008 (3) 0.072 ± 0.015 (3) very slow 0.115 ± 0.041 (4)

Left two columns: concentration dependence of the covalent reaction at wild type and mutant receptors at the binding level. Apparent K d and % of the maximal binding are given as mean ± SD for three to four experiments where each point was determined in triplicates. The receptors were separately exposed to increasing concentrations of NCS compound for 1 h on ice and extensively washed. Residual 3 binding was determined using [ H]Ro15-1788 as radioactive ligand and converted to percentage of binding sites covalently reacted. n.e.: 3 these receptors did not result in detectable binding ([ H]Ro15-1788 or 3 [ H]flunitrazepam). The value to the right, k app25 is derived from single exponential fits to the data obtained in functional time course experi- ments using 1lM NCS compound at room temperature (22–26C) as shown in Fig. 7 and is an apparent time constant. a Data are from Berezhnoyet al.2005.

Time course of the covalent reaction at the binding level: postulation of a novel, inhibitory site for benzodiazepines We next characterized the time course of the covalent reaction at different concentrations of the reactive com- pound. This time course of the reaction is shown for a 1 H101Cb 2 c 2 anda 6 R100Cb 2 c 2 receptors in Fig. 3(a) and (b), respectively, for three different concentrations of NCS compound each. Similar observations as fora 1 H101Cb 2 c 2 were made witha 2 H101Cb 2 c 2 . At all concentrations tested the reaction reached a plateau within less than 2 min, excluding the first hypothesis outlined in the previous section. To our surprise the apparent maximum of the reaction was not 100% but was dependent on the concen- tration of the reactive ligand. Experiments were carried out to test for a possible slow hydrolysis of the covalent adduct. a 1 H101Cb 2 c 2 receptors were covalently reacted with 10lM NCS compound, the NCS compound removed by filtration and the receptors incubated for different times. No evidence for an off reaction was observed during up to 2 h (not shown). A third possibility is lined out in Fig. 4(a). The receptor may react with NCS compound non-covalently either with the benzodiazepine binding site (1) to form RL, or with a novel site (2) to form LR. Equilibrium between R, RL, LR and LRL is thought to be established instanta- neously. A covalent reaction can follow reversible occu- pancy at both sites, RL and LR with the reaction rates k at site (1) and kk at site (2). If no covalent reaction would occur at site (2), site (1) would form a sink and covalent reaction at this site would reach 100% at any ligand

2008 The Authors Journal Compilation2008 International Society for Neurochemistry,J. Neurochem.(2008)106, 2353–2363

Fig. 3Time course of the reaction of the NCS compound with a 1 H101Cb 2 c 2 (a) anda 6 R100Cb 2 c 2 (b) mutant receptors. The recep- tors were exposed during different times to 10lM (circles), 3lM (squares) or 1lM (triangles) of the NCS compound on ice. After incubation the NCS compound was rapidly removed as indicated in the 3 Methods section. [ H]Ro15-1788 was used as radioactive ligand to determine the residual binding, which latter was converted to per- centage of binding sites covalently reacted. Data are shown as mean ± SD for three experiments each (triplicates of each point in each experiment). Results were fitted as indicated in the method section.

concentration. Occupation of site (2) leads to a conforma- tional change preventing irreversible reaction at site (1), but not reversible binding. This model explains most of our experimental observations.

An inhibitory benzodiazepine site|2357

The reaction scheme may be simplified further for the case that no covalent reaction occurs, i.e. for wild type receptors. Figure 4(b) shows the concentration dependence of RL. Occupancy of site (2) is assumed to allosterically prevent potentiation of currents elicited by GABA through site (1). Thus, LRL is assumed to be non-potentiated. In this case the concentration dependence of RL describes the concentration dependence of the potentiation of currents by benzodiaze- pines. The curve is bell-shaped. Walterset al.(2000) investigated the potentiation of GABA A receptors by diaz- epam. Interestingly, they clearly document in Fig. 1(a) of that paper that the initial concentration-dependent increase in potentiation is followed by a decrease. This decrease was however overruled at even higher concentrations by the previously described potentiation by diazepam at a second low affinity site located in the membrane. As observed before (Walterset al.2000), flurazepam fails to act at this second, stimulatory site at high concentrations, thereby uncovering the here postulated third site. Indeed in non-mutateda 1 b 2 c 2 GABA A receptors, flurazepam potentiated currents induced by GABA in a bell-shaped fashion (Fig. 4c and d). A bell shaped curve was also obtained fora 2 b 2 c 2 ,a 3 b 2 c 2 and a 5 b 2 c 2 , but nota 6 b 2 c 2 receptors (Fig. 4f). The curves were fitted as indicated in the methods section. Apparent affinities for site (1) were 0.26 ± 0.02lM, 0.24 ± 0.02l±M, 1.3 0.11l± 0.004M and 0.065 l± 7M, and site (2) 44 lM, 122 ± 44lM, 127 ± 30l± 4M and 24 lM (n= 3, each) fora 1 b 2 c 2 ,a 2 b 2 c 2 ,a 3 b 2 c 2 anda 5 b 2 c 2 receptors, respectively. We were interested to locate this additional inhibitory site. For this purpose we took advantage of concatenated subunits to place benzodiazepine sensitivea 1 and insensi- tivea 6 subunits in defined positions (Fig. 4e). We looked at the concentration dependence of the potentiation by flurazepam atc 2 -b 2 -a 6 /b 2 -a 1 andc 2 -b 2 -a 1 /b 2 -a 6 receptors (Fig. 4d). Inc 2 -b 2 -a 6 /b 2 -a 1 receptors thea 1 subunit contributes to the classical benzodiazepine binding site anda 6 is located at thea/b4e). Theseinterface (Fig. receptors were potentiated through action at the classical benzodiazepine site, but the decrease in potentiation at higher concentrations of flurazepam was absent in the receptors (Fig. 4d). Obviously, presence of aa 6 at thea/b interface prevents this decrease in potentiation. The above data could indicate that the additional inhibitory site is located at thea/binterface, pseudo-symmetrically to the classical benzodiazepine binding site at thea/csubunit interface. Inc 2 -b 2 -a 1 /b 2 -a 6 receptors thea 6 subunit contributes to the formation of the classical benzodiazepine binding site (Fig. 4e). As expected, these receptors were not potentiated by flurazepam. Interestingly, the postulated occupancy of a site for flurazepam at thea/binterface does neither lead to stimulation nor inhibition of the response to GABA in the absence of a classical benzodi- azepine site.

2008 The Authors Journal Compilation2008 International Society for Neurochemistry,J. Neurochem.(2008)106, 2353–2363

2358|R. Bauret al.

Concentration dependence of the irreversible reaction of the NCS compound at the functional level We then proceeded to the characterization of the covalent reaction at the classical benzodiazepine binding site at the functional level. Figure 5(a) shows current traces obtained from an oocyte expressinga 3 H126Cb 2 c 2 receptors recorded before and after exposure of the oocyte to 20lM NCS compound for 1 min at room temperature (22–26C). NCS compound leads to an increase in the current amplitude elicited by 0.7lM GABA. For the following reasons, we interpret this current increase as the consequence of covalent reaction. First, the current increase is largely resistant to

prolonged perfusion (see Fig. 6), second, covalent reaction is very fast (Fig. 3), and third, control experiments with wild type receptors showed that NCS compound could be washed out under these conditions. Figure 5(b) summarizes exper- iments carried out witha 1 H101Cb 2 c 2 ,a 2 H101Cb 2 c 2 , a 3 H126Cb 2 c 2 ,a 5 H105Cb 2 c 2 anda 6 R100Cb 2 c 2 receptors exposing the receptors to 20 nM–20lM NCS compound. For each concentration a new oocyte was used. Wild type a 1 b 2 c 2 ,a 2 b 2 c 2 ,a 3 b 2 c 2 ,a 5 b 2 c 2 anda 6 b 2 c 2 receptors were similarly exposed to 20lM NCS compound (n= 3 each). In none of the cases a significant increase in the current amplitude was detected, that would have provided evidence

2008 The Authors Journal Compilation2008 International Society for Neurochemistry,J. Neurochem.(2008)106, 2353–2363