Interaction of Non-competitive Blockers within the γ-Aminobutyric Acid Type A Chloride Channel Using Chemically Reactive Probes as Chemical Sensors for Cysteine Mutants*

Selected channel-lining cysteine mutants from the M2 segment of rat α1 γ-aminobutyric acid (GABA) type A receptor subunit, at positions 257, 261, 264, and 272 were co-expressed with β1 and γ2 subunits in Xenopus oocytes. They generated functional receptors displaying conductance and response to both GABA and picrotoxinin similar to the wild type α1β1γ2 receptor. Three chemically reactive affinity probes derived from non-competitive blockers were synthesized to react with the engineered cysteines: 1) dithiane bis-sulfone derivative modified by an isothiocyanate function (probe A); 2) fiprole derivatives modified by an α-chloroketone (probe B) and α-bromoketone (probe C) moiety. These probes blocked the GABA-induced currents on all receptors. This blockade could be fully reversed by a washing procedure on the wild type, the α1T261Cβ1γ2 and α1L264Cβ1γ2 mutant receptors. In contrast, an irreversible effect was observed for all three probes on both α1V257Cβ1γ2 and α1S272Cβ1γ2 mutant receptors. This effect was probe concentration-dependent and could be abolished by picrotoxinin and/or t-butyl bicyclophosphorothionate. These data indicate a major interaction of non-competitive blockers at position 257 of the presumed M2 segment of rat α1 subunit but also suggest an interaction at the more extracellular position 272.

␥-Aminobutyric acid type A (GABA A ) 1 receptors exert their inhibitory effect in the central nervous system of vertebrates by regulating a chloride-sensitive channel which is very likely centered within a protein transmembrane heteropentameric subunits complex (1)(2)(3)(4)(5). The existence of 6␣, 4␤, 4␥, 1␦, and 2 subunits in addition to splicing variants, suggests a large diversity in the constitution of heteropentameric isoforms allowing a subtle tuning of the action of this neurotransmitter (6 -8). However, it has been proposed that a restricted number of combinations condition the functioning of this receptor and it is assumed that the ␣1␤2␥2 represents the major adult isoform (9). GABA A receptors serve as the target for several classes of molecules including important neuroactive drugs such as benzodiazepines, barbiturates, and neurosteroids. In contrast, only three receptor subunits have been cloned from insects up to now, RDL (10), ␤ (11), and GRD (12) leading to an apparently less complex situation for their structural assembly. Of particular interest are the action of non-competitive antagonists which are presumed to interact within the GABA receptor chloride channel leading to powerful insecticidal properties when presenting a selectivity for insect GABA receptor (13)(14)(15)(16).
To investigate, at a molecular level, the interaction of noncompetitive GABA antagonists with the chloride channel associated to the GABA A receptor, we defined an approach which uses chemically reactive non-competitive blockers (NCBs) as chemical sensors for cysteine mutants on the rat ␣1 GABA receptor subunit. This strategy was derived from the extensive work of Akabas and co-workers on several ionic channels including the chloride channel associated to the GABA A receptor (17) which identified the receptor channel-lining residues using a cysteine accessibility method (18). Selected channel-lining cysteine mutants from the ␣1 rat subunit, at positions 257, 261, 264, and 272, respectively, when co-expressed with ␤1 and ␥2 subunits in Xenopus oocytes, were probed for their ability to react covalently with several chemically modified NCBs derived from dithiane bis-sulfones (19,20) and the insecticide fipronil (21). The reactive chemical functions were either aromatic isothiocyanates or ␣-chloroand ␣-bromoketone fiproles. The formation of a selective covalent bond between the cysteine mutant and the reactive NCB, protectable by reference NCBs such as PTX or TBPS, allows positioning in a spatial proximity of the cysteine residue with the reactive group of the NCB and most importantly, has made it possible to discriminate effects due to an allosteric interaction induced by the mutation.
In this study we present evidence that the synthesized reactive NCB probes had a fully reversible effect on the wild type receptor as well as on the ␣1T261C␤2␥2 and the ␣1L264C␤2␥2 mutant receptors. In contrast, a selective irreversible effect could be demonstrated for the ␣1V257C␤2␥2 and ␣1S272C ␤2␥2 mutant receptors. Although, the interaction of NCBs such as PTX at position 257 of the M2 helix was already postulated (22), a specific interaction at position 272 of the helix, close to the entrance of the channel, was not described and if it occurs, would allow new insight on the interaction of NCBs within this ionic channel. In addition, the formation of the covalent bonds, especially at position 257, suggests a positioning of the reactive ligands, in an oriented manner within the channel.

EXPERIMENTAL PROCEDURES
Synthesis of Affinity Probes-The dithiane bis-sulfone isothiocyanate A was synthesized starting from the previously described aromatic difluoroazido derivative (20) by catalytic hydrogenation followed by treatment of the amino group by thiophosgene in acetone to the desired aromatic difluoroisothiocyanate with a global yield of 27% for these two steps. Probes B and C were a gift from Rhône-Poulenc Ag Co. (Research Triangle Park, North Carolina).
GABA A Receptor Subunit cDNA Clones-The cDNA encoding rat ␣1, ␣1V257C, ␣1T261C, ␣1L264C, ␣1S272C, ␤1, and ␥2 in pBluescript SK(Ϫ) plasmids were obtained from Prof. M. H. Akabas (Columbia University, New York). Site-directed mutagenesis was performed as described (23) to construct the ␣1V257S cDNA using the following primers: 5Ј-CCAGCAAGAACTTCCTTTGGAGTGACG-3Ј and 5Ј-CGTC-ACTCCAAAGGAAGTTCTTGCTGG-3Ј in the polymerase chain reaction with the ␣1 subunit cDNA inserted in a pBluescript SK(Ϫ) plasmid (Stratagene) as a template. The underlined position indicates the position of the Val to Ser mutation. After purification of the DNA fragments, the mutation was checked by DNA sequencing.
Expression of GABA Receptor Subunits-For in vitro mRNA transcription, the plasmids containing the ␣1 and ␤1 subunits were linearized with HindIII and the ␥2 subunit was linearized with NotI. mRNA was transcribed in vitro by T 3 (␤1) or T 7 (␣1 an ␥2) polymerase using the Ambion (Austin, TX) mMessage mMachine kit. Stage V and VI Xenopus laevis oocytes were harvested and defolliculated as described (24). One day after the oocytes were harvested, they were injected with 20 ng of total mRNA (mixed in the subunit ␣1␤1␥2 ratio 1:1:1) dissolved in 50 nl of nuclease-free water. After injection, the oocytes were incubated 3-6 days at 18°C in OR-3 (1:2 dilution of L-15 Leibovitz media, 1 mM glutamine, gentamycin (100 g⅐ml Ϫ1 ), 15 mM Hepes adjusted to pH 7.6 with NaOH) before recordings.
Electrophysiological Recordings-GABA-induced currents were recorded from individual oocytes under two-electrode voltage clamp at a holding potential of Ϫ80 mV. Microelectrodes were filled with 3 M KCl and had a resistance of 0.5-3 megohm. The ground electrode was connected to the chamber via a 3 M KCl/agar bridge. Data were acquired with an Oocyte clamp OC-725C (Warner Instruments) using a MacLab Digital Interface Module and Chart software (AD Instruments). The oocytes were held in a 0.3-ml chamber and continuously superfused (2 ml⅐min Ϫ1 ) with Ca 2ϩ -free frog Ringer's solution (CFFR) (115 mM NaCl, 2.5 mM KCl, 1.8 mM MgCl 2 , 10 mM Hepes adjusted to pH 7.5 with NaOH) at room temperature via a rapid superfusion system allowing quick exchange of the various solutions. All drugs were applied via the superfusate. GABA, TBPS, and PTX were purchased from Sigma. Probes A, B, C, and PTX were dissolved in dimethyl sulfoxide and diluted into test concentration, the final concentration of dimethyl sulfoxide being kept below 0.1%. The GABA concentration that induced 50% of the maximal current (EC 50 ) was determined by sequential application of the following concentrations: 1, 5, 10, 50, and 100 M.
The concentration that inhibited 50% of the current induced by 100 M GABA (IC 50 ) was determined by sequential co-application of GABA and six concentrations of PTX: 0.1, 1, 5, 10, 50, and 100 M, respectively. The current was measured after a 20-s co-application of GABA and PTX. EC 50 and IC 50 were calculated from data obtained on six individual oocytes.
We tested the susceptibility of wild type (WT) and mutant GABA A receptors to a 1-min application of probes A, B, or C. To determine the reversible inhibition induced by probes A, B, and C, the following sequence of perfusion was used: 100 M GABA 10 s, CFFR 3 min, 100 M GABA 10 s, CFFR 3 min, 100 M A, 10 M B, or 10 M C, 1 min; A, B, or C ϩ 100 M GABA 10 s. The reversible inhibition was taken as [1 Ϫ (I GABAϩProbe, after /I GABA, before )].
The following sequence of perfusion adapted from a previously described procedure (17)  The average of the two peak currents before probe application was used as the control response. The fractional effect at the different times after application was taken as [1 Ϫ (I GABA, after /I GABA, before )]. The irreversible effect was calculated by comparing the average of the two peak currents 3 and 9 min after probe application. Similar results were obtained on 3-6 single oocytes for each probe and each receptor.
The sensitivity to the GABA response and the effect of the reference NCB PTX were tested for the different recombinant receptors. The observed EC 50 and IC 50 values (Table I) further indicated that the binding properties for GABA and PTX were only slightly affected by the various cysteine mutations, while the ␣1V257S mutant showed an increased affinity for the GABA molecule (EC 50 2 M versus 17 M for the WT), in agreement with the previously observed GABA-induced current increase.
Biochemical and Pharmacological Evaluation of the Reactive Probes-The affinities of the reactive probes A, B, and C ( Fig.  1) for the NCB-binding site were estimated by a displacement of [ 3 H]1-(4Ј-Ethynylphenyl)-4-propyl-2,6,7-trioxabicylo[2,2,2] octane (EBOB) on rat brain and housefly membranes. The obtained IC 50 values (not shown), indicate affinities in the submicromolar range for the vertebrate receptor. A marked difference was noticeable on the rat brain membranes for the 4-acyl-pyrazole side chain substituent in the fiprole series, i.e. IC 50 values for B and C were 0.2 and 1 M, respectively.
The effect of the three probes on recombinant ␣1␤1␥2 WT receptors expressed in Xenopus oocytes was analyzed by recording the GABA-induced chloride currents after applying excess of the probes (100 M probe A and 10 M fiprole derivatives B and C) (Table II). Eighty to 90% inhibition of the GABA-induced currents were obtained for these probes and this effect could be fully reversed by successive washings of the oocytes.
Comparative Effect of Probe A on WT and Mutant ␣1V257C␤1␥2 Receptors- Fig. 2 shows the comparative effect of probe A on the WT receptor and the ␣1V257C␤1␥2 mutant receptor, leading to a reversible effect of GABA-induced currents on the WT ( Fig. 2A) and an irreversible effect on the mutant receptor, with only partial recovery of the initial GABA-induced currents after successive washings (Fig. 2B). Fig. 3A shows the time dependence of the washing procedure on the WT and the ␣1V257C␤1␥2 mutant receptor after inhibition by probe A as well as the probe concentration dependence on the extent of the irreversible effect, varying from 18, 45, and 69% irreversible inhibition for 10, 30, and 100 M probe concentration, respectively (Fig. 3B). Finally, the irreversible effectcouldbepreventedinthepresenceofPTXinaconcentrationdependent manner, i.e. a full protection was observed in the presence of 100 M PTX for a 69% inhibition induced by a 100 M probe concentration (Fig. 2C). We checked that the observed effect was independent of the presence of GABA. The reversible inhibition on the WT receptor and the irreversible inhibition on the ␣1V257C␤1␥2 mutant receptor were similar when probe A was co-applied with 100 M GABA for 1 min (not shown). This series of experiments indicates the occurrence of a specific covalent reaction between the cysteine residue incorporated at position 257 of the ␣ subunit of the mutant receptor and the isothiocyanate moiety of the dithiane bis-sulfone probe A. Table II summarizes the effects on the GABA-induced currents of probes A, B, and C on the different recombinant receptors. These probes blocked efficiently (between 80 and 90%) the GABA-induced currents on all recombinant receptors. While the effect, for the three probes was shown to be fully reversible for the WT, the two cysteine mutants at positions 261 and 264 as well as the serine 257 mutant receptors (Table II), the effect could be demonstrated to be partially irreversible for the two mutant cysteine receptors at positions 257 and 272. The irreversible effect was more pronounced at position 257 for the three probes, at a given probe concentration, and consistent with its greater reactivity, the bromoketone derivative C led to a more efficient irreversible reaction when compared with the chloro derivative B. The irreversible effects were proven to be concentration dependent (not shown) and finally, they were shown to be partially protectable by either PTX or TBPS at both positions 257 and 272. The extent of the protection was, however, depending on the test probe, on the mutant position, and also on the protector examined. While a full protection was observed at position 257 with PTX on probe A-induced inactivation (Fig. 2C), a less efficient protection was established at position 257 over the 272 position for probe C (or B) induced inactivation, i.e. the 100% protection (not shown) observed with 100 M PTX or 10 M TBPS induced by 10 M probe C at position 272 (Table II) was reduced to 60% protection by TBPS or 20% protection by PTX at position 257 (not shown).

DISCUSSION
The GABA A receptors have been proposed to be a protein complex composed of (hetero)pentameric transmembrane subunits assembly comprising a central cavity defining an ionic channel (3,6,25). A series of neurotoxic insecticides, including the toxin PTX, or synthetic compounds such as polychlorocycloalkanes and fipronil, inhibit GABA-induced currents by binding to the NCB site(s) of the GABA A receptor complex of vertebrates and invertebrates with variable efficacy (14,15,26). By analogy to the nicotinic acetylcholine receptor, the M2 membrane-spanning segment of the different GABA receptor subunits have been suggested to line the channel pore (27), and numerous site-directed mutagenesis experiments achieved within this segment of vertebrate or invertebrate receptors have demonstrated an affect on the binding of the NCBs. A typical example being the natural point mutation A302S in the Drosophila melanogaster GABA receptor (RDL subunit) conferring high levels of resistance to PTX and dieldrin (28), this residue being homologous to the rat ␣1Val 257 . Using the cysteine accessibility approach, Akabas and co-workers (22) stud-  ied the action of PTX on the protection of the chemical modification by cysteine-specific permeants (cationic and anionic) at positions 257 and 261 of the rat ␣1 subunit in ␣1␤1␥2 recombinant receptors. They showed a protective effect induced by PTX only at the more cytoplasmic position 257 and this effect was complete, only in an open GABA channel configuration. This action was proposed to result from a direct interaction of PTX in the GABA channel at the level of residue 257, this interaction becoming more effective in an open channel receptor conformation, rather than an allosteric effect induced by PTX binding outside the channel. A series of mutants near the center of the presumed M2 membrane spanning segment were studied to analyze the effect on the binding of NCBs and the conductance properties. Recombinant rat ␣1␤2␥2 receptors containing double mutations at homologous positions at ␣1, ␤2, or ␥2 subunits: ␣1(T261F/T267A), ␤2(T246F/T252A), or ␥2(T271F/T277A) were shown to become insensitive to PTX (29) and even the single mutant ␤2(T246F) could induce this insensitivity to PTX. Among the possible explanations, the authors have forwarded a sterical blockade resulting from the Phe side chain and preventing the access of PTX to its binding site. By analogy to the nicotinic acetylcholine receptor (30,31), the mutation of the conserved leucine residue at the center of the presumed M2 helix of the human GABA ␣1␤1 receptor, ␣1L264T or ␤1L259T, when coexpressed with the non-mutated subunit, generates receptors that form spontaneous open chloride channels in cells, without exposure to GABA. These currents are not blocked by bicuculline or PTX (32). PTX and other NCBs have also been reported to inhibit GABA C receptors (33,34). A naturally occurring mutation within the M2 domain of the rat 2 subunit methionine 300 (aligned with Thr 314 of the rat 1 and Thr 261 of the rat ␣1 subunit) was shown to induce resistance to PTX blockade of native GABA C receptors in the rat retina (35). Independently, proline residue 309 of the human 1 subunit (aligning with rat ␣1Val 257 residue) was identified as controlling the sensitivity to PTX inhibition after expression in oocytes (36). Taken together, these results support a mechanism by which PTX blocks the ionic GABA channel, presumably through binding in the channel lumen without, however, rejecting conclusively allosteric mechanism alternatives. Among the amino acids from the presumed membrane spanning segment M2 of the different GABA receptor subunits, the cytoplasmic position at the level of Val 257 of rat ␣1 subunit, corresponding to the natural mutation in Drosophila Rdl A302S subunit and the Pro 309 in human 1 subunit, represents a crucial position in the NCB interaction with various GABA receptors.
The aim of our approach, which combines a site-directed labeling method to noninvasive site-directed mutagenesis experiments, is to induce a specific irreversible reaction between a reactive electrophilic affinity ligand analog and a nucleophilic cysteine mutant. The affinity probes which were synthesized derived from two structurally unrelated NCBs having insecticidal properties, the dithiane bis-sulfone and the fipronil series (13,15,19) which were chemically modified by an isothiocyanate function or ␣-chloro and ␣-bromo moiety leading to probes A, B, and C, respectively. When tested independently for their chemical reactivity toward nucleophilic amino acids, the cysteine residues were found to be the only amino acids reacting efficiently and instantaneously with our probes at neutral pH (not shown). Therefore we generated cysteine mutants within the ␣ subunit of the GABA A receptor to ensure the chemical reactivity. We selected GABA A channel-lining residues, respectively, at positions 257, 261, 264, and 272, which were (i) spread out along the presumed M2 helix of the receptor and proposed to face the lumen of the channel (17); (ii) known to produce cysteines mutants displaying conductance properties similar to the WT when co-expressed with ␤1 and ␥2 subunits in Xenopus oocytes Table I ( 17); and (iii) which showed unaltered binding properties for the agonist GABA and the NCB PTX (Table I). These four mutants, when co-expressed with ␤1 and ␥2 subunits in X. laevis oocytes, generate GABA receptors having biochemical and pharmacological properties very similar to the WT type receptor and are therefore adapted for the present study. For control experiments we also generated the ␣1V257S mutant which showed an increased affinity for GABA in ␣1V257S ␤1␥2 recombinant receptor.
The three probes A, B, and C ( Fig. 1) are efficient blockers of the GABA A -associated chloride channel at the concentration used as demonstrated by the high inhibition observed on GABA-induced currents from recombinant receptors from rat ␣1, ␤1, ␥2 (WT), or mutants ␣1, ␤1, ␥2, subunits co-expressed in Xenopus oocytes (Table II). The washing procedure that was used allowed complete restoration of the conductance properties on the WT receptor as well as on the two cysteine mutant receptors, ␣1T261C␤1␥2 and ␣1L264C␤1␥2, respectively, while this conductance recovery was only partial for the two cysteine mutant receptors ␣1V257C␤1␥2 and ␣1S272C␤1␥2 (Table II). The observed irreversible inactivations showed the expected probe concentration dependence (Fig. 3B) and all  (Table II). The specificity of the inactivation was demonstrated by the protective effect against the inactivation exerted either by PTX (Fig. 2C) and/or TBPS (not shown). Finally, to assess that the covalent reaction occurred between the mutated cysteine and the reactive moiety of the probe, we performed two additional controls. We checked that no irreversible reaction occurred, either on the serine mutant ␣1V257S␤1␥2 after treatment with probes A, B, or C (Table II) or, between a fiprole derivative having a nonreactive 4-acetyl pyrazole side chain with the cysteine mutant ␣1V257C␤1␥2 (not shown). Clearly, the observed irreversible reactions required the vicinity of both cysteine and reactive probe and even more precisely, it required a spatial proximity between the cysteine side chain and the reactive moiety of the probe, allowing a tentative positioning of the probe within the channel cavity. While reaction at position 257 was fully expected, reinforcing previously described results and observations (23,29), reaction at position 272, presumably very close to the extracellular side, in conjunction with the absence of reaction at the intermediate positions 261 and 264, represents a new result. The simplest tentative explanation to account for these results could be that the NCBs are "filtered" by a narrow part at the channel entrance (position 272), than passing by a more widely open intermediate part before reaching the site located deeply in the channel (position 257) and acting like a plug at this position. Such an overall structure could also explain why PTX does not protect the cysteine mutant at position 261 from the alkylation by the ionic permeants which are smaller molecules and used in large excess (22). Due to the strong hydrophobic character of the probes in conjunction with their respective potencies and the used concentrations, it seems very unlikely that the covalent reaction occurring at position 257 could result through an action of the probe coming from the inside of the cell, after crossing the membrane. For instance, such an explanation would be in contradiction with the fact that all three probes had a much stronger irreversible effect at position 257 over 272, also the efficiency of probe C (59% irreversible inhibition at position 257 at 10 M) would be fully unexpected, knowing its high hydrophobic character and its moderate affinity for the NCB site (IC 50/[ 3 H]EBOB ϳ 1 M on rat membranes).
The proposed mode of action of the NCBs within the GABA receptor channel raises, however, different questions. The description of an ionic channel centered in the cleft of five straight helical transmembrane segments cannot explain satisfactorily our labeling results. They do not coincide with a structural description given for the nicotinic acetylcholine receptor, where the narrow part of the channel is proposed to be located at the middle of the transmembrane segments (37). The irreversible reaction observed at position 272 could either reflect a narrowing of the channel entrance, implying the contribution of other subunits at homologous positions, or it could suggest the existence of a recognition site located mainly on the ␣-subunit(s) and involving additional ␣ subunit residues, e.g. residues from the putative membrane-spanning helix M1, as was proposed for the interaction of a NCB with the open state of the nicotinic acetylcholine receptor (38). Clearly, additional experiments are necessary to describe more precisely the NCB site on the GABA A receptor but again these experiments can be undertaken in the light of our new approach combining cysteine mutants with our reactive probes.