Two Novel Residues in M2 of the γ-Aminobutyric Acid Type A Receptor Affecting Gating by GABA and Picrotoxin Affinity*

An amino acid residue was found in M2 of γ-aminobutyric acid (GABA) type A receptors that has profound effects on the binding of picrotoxin to the receptor and therefore may form part of its binding pocket. In addition, it strongly affects channel gating. The residue is located N-terminally to residues suggested so far to be important for channel gating. Point mutated α1β3 receptors were expressed in Xenopusoocytes and analyzed using the electrophysiological techniques. Coexpression of the α1 subunit with the mutated β3 subunit β3L253F led to spontaneous picrotoxin-sensitive currents in the absence of GABA. Nanomolar concentrations of GABA further promoted channel opening. Upon washout of picrotoxin, a huge transient inward current was observed. The reversal potential of the inward current was indicative of a chloride ion selectivity. The amplitude of the inward current was strongly dependent on the picrotoxin concentration and on the duration of its application. There was more than a 100-fold decrease in picrotoxin affinity. A kinetic model is presented that mimics the gating behavior of the mutant receptor. The point mutation in the neighboring residue β3A252V resulted in receptors that displayed an about 6-fold increased apparent affinity to GABA and an about 10-fold reduced sensitivity to picrotoxin.

The GABA A 1 receptors are the major inhibitory neuronal ion channels in the mammalian central nervous system. Two subunits termed ␣ and ␤ have initially been purified from bovine brain (1) and the corresponding DNAs have been cloned (2). Many mammalian subunits have been cloned since (3)(4)(5)(6)(7)(8). These subunits show a high degree of homology to subunits of the nicotinic acetylcholine receptors, the glycine receptor and and the serotonin (type 3) receptor. The GABA A receptor is the site of action of many drugs, among them the benzodiazepines (for review see Ref. 9). The binding site for the channel agonist GABA and that for benzodiazepines are thought to be located at subunit interfaces in homologous positions (for reviews see Refs. 10 and 11).
The second transmembrane segment of these subunits lines the ion channel. In GABA A receptor ␣1 subunits, residues Val 256 , Thr 260 , Thr 261 , Leu 263 , Thr 264 , Thr 267 , Ile 270 , Ser 271 , and Asn 274 have been reported to be exposed to the channel lumen (12). Picrotoxin is likely to interact with residue Val 256 (13,14). Mutation in a homologous residue in Drosophila receptors confers cyclodiene resistance (15). A leucine residue is strictly conserved in the middle of the M2 region of all subunit isoforms and is located at position 263 of the ␣ 1 subunit. Substitution to a serine in ␣ 1 , ␤ 2 , or ␥ 2 resulted in an abnormally high apparent GABA affinity for channel opening (16). Some point mutations of this leucine on ␣ 1 , ␤ 1 , ␤ 2 , or 1 subunits resulted in spontaneous open channels (17)(18)(19)(20)(21).
During work aimed at the understanding of the site in M2 involved in the recognition of tert-butylbicyclophosphorothionate (22), we investigated the properties of chimeric ␣ 1 ␤ 3 receptor subunits coexpressed with ␣1 subunits. The results indicated an importance of Ala 252 and Leu 253 for the tertbutylbicyclophosphorothionate binding affinity. Here, we show that substitutions of ␤ 3 A252 and ␤ 3 L253 result in reduced picrotoxin affinity. For ␤ 3 L253F affinity is more than 100-fold reduced, and a huge transient opening of the channel upon removal of picrotoxin was evident. Such a transient channel opening has not been observed for any other mutation before, and our mathematical model suggests that this is a consequence of the strongly decreased affinity for picrotoxin. Residue Val 256 , neighboring Phe 257 , the homologue on the ␣ 1 subunit, has been shown to covalently interact with other noncompetitive blockers acting at the picrotoxin binding site (14). Leucine 253 of the ␤ 3 subunit (␤ 3 L253) may therefore be part of the contact site for picrotoxin together with ␣ 1 V256. In addition, we report that single point mutation ␤ 3 L253F confers abnormal gating properties to ␣ 1 ␤ 3 receptors. These include spontaneous opening of the channels and a very high GABA sensitivity for channel gating, Thus, our work points to the involvement in GABA A receptor channel gating of more Nterminally located amino acid residues than previously suggested.

EXPERIMENTAL PROCEDURES
Amino Acid Residue Numbering-Residues are numbered according to the mature rat sequences.
Construction of Receptor Subunits-The cDNAs coding for the ␣ 1 , ␤ 3 , and chimeric subunits of the rat GABA A receptor channel have been described elsewhere (22,23). Site-directed mutagenesis was done using the QuikChange mutagenesis kit (Stratagene). In vitro synthesized sequences have been verified by DNA sequencing.
Functional Expression and Characterization-Xenopus laevis oocytes were prepared, injected, and defolliculated, and currents were recorded as described (24,25). Briefly, oocytes were injected with 50 nl of capped, polyadenylated cRNA dissolved in 5 mM K-HEPES, pH 6.8. This solution contained the transcripts coding for the different subunits at concentrations of 75 nM. RNA transcripts were synthesized from linearized plasmids encoding the desired protein using the mMessage mMachine kit (Ambion) according to the recommendations of the manufacturer. A poly(A) tail of ϳ300 residues was added to the transcripts by using yeast poly(A) polymerase (Amersham Pharmacia Biotech). The cRNA combinations were coprecipitated in ethanol and stored at Ϫ20°C. Transcripts were quantified on agarose gels after staining with Radiant Red RNA Stain (Bio-Rad) by comparing staining intensities with various amounts of molecular weight markers (RNA Ladder; Life Technologies, Inc.). Electrophysiological experiments were performed by the two-electrode voltage clamp method at a holding potential of Ϫ80 mV. GABA and picrotoxin (Fluka) were applied for 20 s, and a washout period of 3-15 min was allowed to ensure full recovery from desensitization. The perfusion solution (6 ml/min) was applied through a glass capillary with an inner diameter of 1.35 mm, the mouth of which was placed about 0.4 mm from the surface of the oocyte. The rate of solution change under our conditions has been estimated 70% within less than 0.5 s (25). Current responses have been fitted to the Hill equation: I ϭ I max /(1ϩ(EC 50 /[A]) n ) where I is the peak current at a given concentration of GABA (A), I max is the maximum current, EC 50 is the concentration of agonist eliciting half-maximal current, and n is the Hill coefficient. Currents were measured using a modified OC-725 amplifier (Warner Instruments Corp.) in combination with a xy recorder or digitized using a MacLab/200 (AD Instruments).
Kinetic Modeling-We analyzed the gating behavior of the mutated channel using a kinetic model that omits all states that only become marginally populated (26). For arguments described later, the model consists of four states: two different states for closed channels (C, R), one state for open channels (O) and one state for channels blocked by picrotoxin (O.PTX). The triangular three state model in the absence of picrotoxin (O, C, R) requires six microscopic rate constants. They are calculated using the experimentally determined current decay constants from the open to the closed states ( 1 and 2 ), the equilibrium constant for the spontaneously open state, and the law of detailed balancing. The remaining rate constants were used to adjust the simulation to the data. The binding of picrotoxin to the receptors adds four new rate constants to the system. They have to provide the reopening constant measured in the presence of picrotoxin (), and they have to satisfy the law of detailed balancing. The additional two rates were used to fit the data. Therefore, from the 10 rate constants in total there are only four that can be used to reconcile the simulations with the data. The kinetics of application and washout of picrotoxin is not taken into account. They are assumed to happen instantaneously. This assumption is justified because the processes studied here develop at a much slower time scale. The integration of the differential equation system was performed using the subroutines of the Matlab 5.2.0 library (Math-Works Inc.). Fig. 1 shows the structure of wild type and chimeric subunits. The chimera consist of N-terminal sequences of the ␤ 3 subunit fused to C-terminal sequences of the ␣ 1 subunit. They differ from each other in the residues forming the ion channel pore region. Each of the chimeras was coexpressed together with the ␣ 1 subunit in Xenopus oocytes. Mature receptors incorporated in the surface membrane were analyzed by using the twoelectrode voltage clamp technique. Heteromeric ␣1CH7 receptors failed to respond to GABA (Ͻ10 nA), and an apparent outward current could be measured during application of 1 mM picrotoxin alone ( Fig. 2A). Interestingly, a strong transient inward current was detected during washout of picrotoxin in the absence of GABA ( Fig. 2A). Wild type ␣ 1 ␤ 3 receptors were activated by GABA and showed no response to the application of picrotoxin alone (Fig. 2B). Injection of cRNA coding for CH7 alone did not result in ion currents induced by picrotoxin or GABA (data not shown), indicating that the channel with these unusual properties was formed from ␣ 1 CH7. Three additional chimeras were coexpressed together with the ␣ 1 subunit to localize residues important for the unusual gating behavior. Only ␣ 1 CH74 also displayed the picrotoxin washout current found in ␣ 1 CH7 receptors (Table I). An amino acid comparison between the chimera revealed that a threonine-valine-phenyl-alanine motif was common to CH7 and CH74 and absent in the other two constructs. Three mutant subunits were constructed by individually introducing these three residues of ␣ 1 into the homologous positions of ␤ 3 , with the aim to identify the amino acid residue responsible for the abnormal properties.

Expression of Receptors Containing Chimeric Subunits-
Expression of Point Mutated Receptors-Wild type and mutant ␤ 3 subunits were coexpressed together with the ␣ 1 subunit in Xenopus oocytes. All three mutant receptors expressed GABA-activated chloride currents. Receptors with a leucine to phenylalanine substitution in the channel pore-forming region at position 253 (␣ 1 ␤ 3 L253F) could be activated by very low concentrations of GABA (30 nM). Repeated applications elicited increasingly smaller current amplitudes even when a washout period of up to 15 min was used (data not shown). For this reason it was technically not possible to measure the apparent affinity of GABA for channel opening in ␣ 1 ␤ 3 L253F. Dose response curves for wild type ␣ 1 ␤ 3 , ␣ 1 ␤ 3 V251T, and ␣ 1 ␤ 3 A252V revealed 3.5-and 6.0-fold increases in the apparent affinities to GABA for these mutated receptors (Table II), respectively.
Response to Picrotoxin-Application of 1 mM picrotoxin in the absence of GABA did not induce any apparent outward or inward currents in ␣ 1 ␤ 3 V251T and ␣ 1 ␤ 3 A252V receptors. In contrast, in oocytes expressing ␣ 1 ␤ 3 L253F receptors, picrotoxin application resulted in apparent outward currents (Fig. 2C) similarly to oocytes expressing ␣ 1 CH7. The mutated channel ␣ 1 ␤ 3 L253F also showed a huge transient inward current upon washout of picrotoxin ( Fig. 2C). At 1 mM picrotoxin, the amplitude of the transient inward current was 11 times (average of three determinations) the amplitude of the apparent outward current.
To exclude that the current properties upon injection of cRNAs coding for ␣ 1 and ␤ 3 L253F resulted from homomeric ␤ 3 L253F, we also expressed this mutated subunit alone. No detectable signal could be obtained with 300 M picrotoxin (two independent batches of oocytes).
A voltage ramp protocol was used to measure ion currents in mutant ␣ 1 ␤ 3 L253F channels before and during and the application of picrotoxin and about 5 s after its removal (Fig. 3A). The voltage ramp had a duration of 0.13 s, such that the amplitude of the inward current showed little change during the application of the ramp. Picrotoxin application resulted in a reduction of the membrane conductance, indicating that part of the receptors are spontaneously in an open conformation. All three curves obtained had the same intersection at Ϫ38 Ϯ 1 mV (three experiments). Therefore, it can be concluded that the spontaneous current and the transient inward current have the same ion permeability. Replacing the 95 mM Cl Ϫ with 9.5 mM Cl Ϫ and 84.5 mM acetate Ϫ in the outside medium resulted in an about 60 mV shift to the right in the reversal potential of the transient inward current (not shown), in line with a chloride selective conductance. From the reversal potential of Ϫ38 mV determined at an extracellular chloride concentration of 95 mM, an intracellular chloride concentration of ϳ23 mM may be estimated. The same voltage ramp protocol was used before and during the application of a subsaturating concentration of GABA to oocytes expressing the wild type ␣ 1 ␤ 3 receptor (Fig. 3B). The intersection of the two curves was found at Ϫ25 Ϯ 2 mV (three experiments), indicating an intracellular chloride ion concentration of about 37 mM. The difference in the intracellular chloride concentration in oocytes expressing wild type and the point mutated receptor can be explained by a more negative membrane potential of the oocyte than the chloride reversal potential. Under conditions where the permeability for this ion is increased, the membrane potential drives chloride ions out of the cell.
As mentioned above, the channel opens to a certain degree spontaneously producing an inward current. The initial response to the application of picrotoxin is an apparent outward current reflecting channel closure. The picrotoxin concentration dependence of the peak current amplitude is illustrated in Fig. 4A. It increases between 10 and 1000 M picrotoxin and shows no saturation up to 1 mM picrotoxin (Fig. 4C). The maximum apparent outward current may, however, be estimated, assuming an infinite membrane resistance for the case where all channels are in the closed state. Assuming that most channels are in the open state upon removal of picrotoxin, it can be estimated that less than 9% of the channels are spontaneously open prior to the application of picrotoxin. The shape of the late apparent outward current response during perfusion with picrotoxin depends on the concentration of picrotoxin used (Fig. 4A). In the presence of low concentrations, only a transient inhibition of the current could be detected, followed by reopening of the channels. The time course of this reopening seemed picrotoxin concentration independent in the range of 10 -300 M. Because of the small amplitudes of this component, it could only be estimated. Assuming a mono-exponential time course reopening was characterized by a in the range of ϳ6 -13 s (not shown).
A large transient inward current was observed during washout of picrotoxin. The concentration dependence on picrotoxin Ϫ ϩ amplitude not reproducible of this current is illustrated in Fig. 4C. Its amplitude increases using increasing concentrations of picrotoxin and does not saturate up to 1 mM picrotoxin. Half of the amplitude observed at 1 mM picrotoxin was observed at about 300 M (Fig. 4C). Fig. 5A shows that the size of this current increases with the duration of picrotoxin application. For this experiment 300 M picrotoxin was applied during different time intervals between ϳ1 and 60 s. The time dependence of the increase in inward current amplitude is well fitted with a mono-exponential function with ϭ 9.1 Ϯ 2.2 s (n ϭ 3). Reclosure of the channel after transient opening was independent of the picrotoxin concentration and followed a bi-exponential time course with f ϭ 5.5 Ϯ 2.8 s and s ϭ 23.3 Ϯ 12.0 s (means Ϯ S.D., three experiments at five different concentrations each). The picrotoxin sensitivity of GABA-activated currents was also measured for wild type ␣ 1 ␤ 3 receptors and mutant ␣ 1 ␤ 3 V251T and ␣ 1 ␤ 3 A252V receptors. A GABA concentration eliciting 10 -15% of the maximum current was used in these experiments. Because ␣ 1 ␤ 3 V251T and ␣ 1 ␤ 3 A252V receptors displayed an increased apparent affinity to GABA (Table II), a lower concentration of agonist had to be used for these two mutant receptors. ␣ 1 ␤ 3 A252V receptors displayed an about 10-fold reduced sensitivity to picrotoxin compared with wild type and ␣ 1 ␤ 3 V251T receptors (Fig. 6).
Kinetic Modeling-To explain the following two phenomena, namely the reopening of the channels during continuous application of picrotoxin and the transient inward current after its removal, we performed computer simulations based on the model shown in Fig. 7 Fig. 4B shows simulations of the current during perfusion with different concentrations of picrotoxin. The amplitude of the initial apparent outward current is limited at high picrotoxin concentration because of saturation of channel closure. The following relaxation process into the new equilibrium state depends only weakly on the concentration of picrotoxin (largest relaxation constant ϭ 13-16 s), which is in agreement with the experiments. The simulations revealed that the closed state (R) serves as a reservoir for states C and O so that after binding of the channels to picrotoxin the two latter states become partly refilled by channels from state R. Obviously the higher the concentration of picrotoxin, the larger the population of the state (O.PTX), and as a consequence the amplitude of the rebound current augments with increasing picrotoxin concentration. The true k 2 that describes the transition rate from O.PTX to O is obscured by the rate of solution change (see "Experimental Procedures"). The model is constructed in such a way that the decay of the inward current after removal of picrotoxin is double exponential with time constants 1 ϭ 8 s and 2 ϭ 12 s. Both, the picrotoxin concentration dependence (Fig. 4B) and the time dependence (Fig. 5B) of the responses of the mutated channel to picrotoxin exposure and removal predicted by the model should be compared with the experimentally observed behavior (Figs. 4A and 5A). Except for the initial peak of the apparent outward current predicted from the model and almost absent in the experimental traces, the agreement between model and experiment is remarkable. The difference is probably mostly due to the limited rate of solution change (see "Experimental Procedures"), which has not been taken into account in the simulations. DISCUSSION We studied alterations in the channel lining part of the recombinant ␣ 1 ␤ 3 GABA A receptor. A mutant ion channel was found that displayed a transient chloride current upon removal of the channel blocker picrotoxin. A single point mutation is able to confer this unusual property to the receptor, namely, of leucine 253 in the M2 region of the ␤ 3 subunit to a phenylalanine, which is present in the homologous position of all ␣ subunits. The point mutated strongly affects gating of the channel and its interaction with picrotoxin.
It is interesting to compare the newly identified position with other positions within M2 that are important for receptor function and modulation. Based on radioligand binding experiments to chimeric receptors, it has been suggested that ␤ 3 L253 is important for the binding of another channel blocker tertbutylbicyclophosphorothionate (22). This position is 12 amino acid residues N-terminal to the one on the ␤ 3 subunit implicated in the action of loreclezole (27). It is located 6 residues C-terminal to the predicted cytoplasmic entry point into the membrane and is 6 residues N-terminal to the conserved leucine in the center of the M2 region. Point mutations of this leucine also resulted in spontaneous currents (17)(18)(19)(20). The spontaneous currents were in some cases antagonized by GABA (18,19) and were sensitive to picrotoxin (18 -21). Spontaneous channel activity has also been reported with expression of the rat or mouse ␤ 1 subunit alone (28,29), the mouse ␤ 3 subunit alone (30), and a combination of rat ␣ 5 and ␤ subunits (31). Using the mentioned amounts of cRNA, no spontaneous currents were observed for homomeric wild type ␤ 3 and mutant ␤ 3 L253F receptors. In all the mentioned cases an off-current upon picrotoxin washout was never reported. Our model presented below will predict why this is not the case.
The structure of the GABA A receptor in M2 is presumably close to that of other members of the ligand-gated ion channel family. The above cited studies and studies of the homologous nicotinic acetylcholine receptor suggested that the centrally conserved leucine is structurally critical for channel gating and might even be a part of the channel gate (32). Other studies place the gate more N-terminal to the cytoplasmic end of the channel pore (13). Because ␣ 1 ␤ 3 L253F receptors display an altered channel gating our results support this proposal.
Valine 256 on the ␣ 1 subunit has been proposed to be exposed to the channel lumen (13) and to be in direct contact with picrotoxin (13,14). A homologous residue has also been implicated in the action of the cyclodiene insecticide resistance in invertebrate GABA receptor subunit Rdl (15). Because of the 5-fold symmetry of the channel pore, it is likely that the homologous residue on the ␤ 3 subunit, alanine 252, is also facing the channel pore and might be close to the bound picrotoxin entity. Our results indeed show that replacement of alanine by valine in this position (␤ 3 A252V) results in an about 10-fold reduced affinity for picrotoxin. Mutation of the adjacent leucine 253 to phenylalanine even more drastically reduces the picrotoxin affinity. It cannot completely be ruled out that the substitution might have some indirect effects on the picrotoxin binding pocket. However, it is tempting to speculate that picrotoxin is in direct contact with ␤ 3 L253. In apparent contrast to this speculation, Xu and Akabas (12) found that the homologous phenylalanine 257 on the ␣ 1 subunit after cysteine sub-stitution is not accessible to sulfhydryl reagents. However, this residue is on the same side of the ␣-helix as reactive positions and is also adjacent to them. Side chains other than cysteine might be at least partially accessible to the channel lumen and could be in contact with the picrotoxin molecule.
We propose a kinetic model for ␣ 1 ␤ 3 L253F describing its interaction with picrotoxin. It makes a number of predictions that are discussed in the following. Mutant Both effects are observed for other GABA A receptors with substitution of the centrally conserved leucine (18 -21). Therefore, the model suggests that for these mutated GABA A receptors the dissociation of picrotoxin from the blocked channel (k 2 ) is slowed down. The relative amplitude of this fast transient inward current is also limited by the fraction of channels that are in the open state in the absence of picrotoxin. The larger this fraction of spontaneously open channels is, the smaller the predicted inward current. Both factors may contribute to a different degree to the lack of an transient inward current in the initially cited cases of spontaneous currents.
In summary, we report two important observations. The first is that we describe two point mutations in the M2 region of the ␤ 3 subunit close to the putative cytoplasmic end that both result in an increase of apparent GABA affinity. One of these, ␤ 3 L253F results additionally in spontaneous channel opening and is thus structurally critical for channel gating. Our results therefore strongly suggest that the region important for channel gating has to be extended at least four amino acid positions toward the N terminus as compared with previous conclusions (18). The second observation is that these point mutations result in a reduced affinity for picrotoxin. The residue ␣ 1 V256 has been shown to covalently interact with other noncompetitive blockers acting at the picrotoxin binding site (14). Mutation of the homologous residue ␤ 2 A252 shows a 10-fold effect. However, mutation of the neighboring residue ␤ 3 L253 has drastic effects on picrotoxin affinity and may therefore together with ␣ 1 V256 form part of the contact site for picrotoxin.