Comparative surface accessibility of a pore-lining threonine residue (T6') in the glycine and GABA(A) receptors.

The substituted cysteine accessibility method was used to probe the surface exposure of a pore-lining threonine residue (T6') common to both the glycine receptor (GlyR) and gamma-aminobutyric acid, type A receptor (GABA(A)R) chloride channels. This residue lies close to the channel activation gate, the ionic selectivity filter, and the main pore blocker binding site. Despite their high amino acid sequence homologies and common role in conducting chloride ions, recent studies have suggested that the GlyRs and GABA(A)Rs have divergent open state pore structures at the 6' position. When both the human alpha1(T6'C) homomeric GlyR and the rat alpha1(T6'C)beta1(T6'C) heteromeric GABA(A)R were expressed in human embryonic kidney 293 cells, their 6' residue surface accessibilities differed significantly in the closed state. However, when a soluble cysteine-modifying compound was applied in the presence of saturating agonist concentrations, both receptors were locked into the open state. This action was not induced by oxidizing agents in either receptor. These results provide evidence for a conserved pore opening mechanism in anion-selective members of the ligand-gated ion channel family. The results also indicate that the GABA(A)R pore structure at the 6' level may vary between different expression systems.

The ligand-gated ion channel (LGIC) 1 superfamily includes the nicotinic acetylcholine receptor (nAChR), serotonin type 3 receptor (5HT 3 R), GABA A receptor (GABA A R), and glycine receptor (GlyR), as well as invertebrate glutamate and histidine receptors (1). Functional receptors of this family comprise five homologous subunits arranged in a ring to form a central ion-conducting pore. Each subunit is composed of a large ex-tracellular ligand-binding N-terminal domain, four membranespanning segments (M1-M4), and a large intracellular domain between M3 and M4.
The pore-lining, second transmembrane (M2) domain has an ␣-helical secondary structure that undergoes a conformational change as the channel is opened (2). To investigate this process in detail, state-dependent differences in the surface exposure of M2 domain residues can be assayed using the substituted cysteine accessibility method (3). In this technique, residues are mutated individually to cysteines, and changes in their reactivity rates with soluble cysteine-reactive reagents can identify structural changes between different functional states. As expected for receptors belonging to the same family, this technique has generally yielded a good correlation between the open state M2 domain secondary structures of the nAChR (4 -7), GABA A R (8), and 5HT 3 R (9, 10).
The M2 domain 6Ј residue, which is a threonine in the GlyR ␣1 subunit and the GABA A R ␣1 and ␤1 subunits (see Fig. 1A), lines a critical part of the pore. It is close to the activation gate (6,11,12) and the ionic selectivity filter (13)(14)(15) and forms the main pore blocker binding site (reviewed in Ref. 16). Therefore, structural differences at this level may be expected to have significant functional consequences. In the homomeric ␣1 T6ЈC GlyR expressed in a mammalian HEK293 cell line, Shan et al. (17) concluded that the surface exposure of introduced 6Ј cysteines was increased in the channel open state. In contrast, in the ␣1 T6ЈC ␤1 T6ЈC GABA A R expressed in Xenopus oocytes, the 6Ј cysteines were found to be exposed in the closed state and rotated to face the adjacent subunits in the open state (18). Thus, despite having a high M2 domain amino acid sequence homology (see Fig. 1A) and a common function in conducting chloride ions, the GlyR and GABA A R appear to be structurally divergent at this position.
The aim of this study was to conduct a detailed comparative study into the surface accessibility of the 6Ј cysteines in the GlyR and GABA A R when both are expressed recombinantly in a common (HEK293 cell) expression system. The main findings are that the respective pore structures at the 6Ј positions are significantly different in the closed states but that there appear to be similarities in the mechanisms of channel opening. The results also reveal distinct differences in the structural and functional properties of GABA A Rs depending on whether they are expressed in Xenopus oocytes or HEK293 cells.

Mutagenesis and Expression of GlyR and GABA A R cDNAs-The
human GlyR ␣1 subunit cDNA was subcloned into the pCIS2 plasmid vector, and the rat GABA A R ␣1 and ␤1 subunit cDNAs were subcloned into the pIRES2-EGFP plasmid vector (Clontech, Palo Alto, CA). Sitedirected mutagenesis was performed using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA), and the successful incorporation of mutations was confirmed by sequencing the clones. Adenovirus-transformed HEK293 cells (ATCC CRL 1573) were passaged in a 50:50 mixture of minimal essential medium and Dulbecco's modified Eagle's medium supplemented with 2 mM glutamate, 10% fetal calf serum and the antibiotics, penicillin (at 50 IU/ml), and streptomycin (at 50 g/ml). Cells were transfected using a calcium phosphate precipitation protocol (19). When co-transfecting the GABA A R ␣1 and ␤1 subunits, their respective cDNAs were combined in a ratio of 1:1. After exposure to transfection solution for 24 h, cells were washed twice using the culture medium and used for recording over the following 24 -72 h.
Electrophysiology-The cells were observed using a fluorescent microscope, and currents were measured using the whole cell patch-clamp configuration. Cells were perfused by a control solution that contained the following (in mM): 140 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 HEPES, 10 glucose, with the pH adjusted to 7.4 with NaOH. Patch pipettes were fabricated from borosilicate hematocrit tubing (Vitrex, Modulohm, Denmark) and heat-polished. Pipettes had a tip resistance of 1.5-3 megohms when filled with the standard pipette solution, which contained the following (in mM): 145 CsCl, 2 CaCl 2 , 2 MgCl 2 , 10 HEPES, 10 EGTA, with the pH adjusted to 7.4 with NaOH. After establishment of the whole cell configuration, cells were voltage-clamped at Ϫ40 mV, and membrane currents were recorded using an Axopatch 1D amplifier and pclamp7 software (Axon Instruments, Union City, CA). The cells were perfused by a parallel array of microtubular barrels through which solutions were gravity-induced. All experiments were conducted at room temperature (19 -22°C).
Methanethiosulfonate ethyltrimethylammonium (MTSET) and methanethiosulfonate ethylammonium (MTSEA) were obtained from Toronto Research Chemicals (Toronto, Ontario, Canada), whereas all other reagents were obtained from Sigma. MTSET and MTSEA were dissolved directly into the bath solution at the final concentrations of 1 and 2.5 mM, respectively, unless indicated otherwise. The oxidizing reagent, copper-O-phenanthroline (Cu:phen) was prepared by mixing CuSO 4 (stored as 100 mM stock solution in H 2 0 at Ϫ10°C) and 1,10phenanthroline (stored as 400 mM stock solution in ethanol at Ϫ10°C). The final concentrations of copper and 1,10-phenanthroline in the control bathing solution were 100 and 400 M, respectively. H 2 O 2 , maintained as a 30% stock solution, was diluted to 0.3% in the control bathing solution. MTSET, MTSEA, Cu:phen, and H 2 O 2 were used for no longer than 10 min after being dissolved into the bathing solution at room temperature. The disulfide-reducing reagent, dithiothreitol (DTT), was prepared daily as a 1 or 10 mM solution in control bathing solution.
The effects of all sulfhydryl-specific reagents were tested using the following procedure. After establishment of the recording configuration, two brief applications of agonist at the half-saturating (EC 50 ) concentration were followed by two brief applications at a saturating (10 -20 ϫ EC 50 ) concentration, all at 30-s intervals. Provided current amplitude remained constant, the averaged current amplitudes were used as the control. Following application of sulfhydryl-specific reagents, cells were washed in control solution for at 1-3 min before the EC 50 and EC 100 agonist-activated currents were measured again.
Data Analysis-All data were analyzed using Origin 4.0 (Northampton, MA) or Sigmastat 1.0 (Jandel Scientific). Results are expressed as means Ϯ S.E. of three or more independent experiments. The empirical Hill equation, fitted by a non-linear least squares algorithm, was used to calculate the EC 50 and Hill coefficient (n H ) values for glycine and GABA activation. Statistical significance was determined by either linear regression or by one-way analysis of variance using the Student's-Newmans-Keul post hoc test for unpaired data, with p Ͻ 0.05 representing significance.

RESULTS
Sulfhydryl Modification of the ␣1 T6ЈC GlyR-This study investigated the surface accessibility of the 6Ј residues of the GlyR ␣1 subunit and the GABA A R ␣1 and ␤1 subunits. As shown in Fig. 1, each of the WT receptor subunits contains a threonine at this position. In this study the threonines were mutated to cysteines to enable cysteine-specific reagents to be used as probes of 6Ј surface accessibility (3). The GlyR ␣1 subunit also contained the C41A mutation, which eliminated the only uncross-linked external cysteine. The GABA A R ␣1 and ␤1 subunits contained no uncross-linked external cysteines.
The mean EC 50 , n H , and I max values for glycine-activated currents in the ␣1 WT and ␣1 T6ЈC GlyRs are summarized in Table I. In the absence of glycine, there was no significant difference in the resting conductance of cells expressing ␣1 WT and ␣1 T6ЈC GlyRs, implying that the T6ЈC mutation did not induce a steady-state leak conductance through the channels.
We demonstrated previously that a 1-min application of 1 mM MTSET had no significant effect on the ␣1 WT GlyR regardless of whether it was applied in the closed or channel open states (17,20). Similarly, MTSET had no effect on the ␣1 T6ЈC GlyR when applied in the closed channel state (17). However, when MTSET was applied to the ␣1 T6ЈC GlyR in the presence of a saturating (0.5 mM) concentration of glycine, the channels remained partially activated following the removal of glycine and MTSET (17). Following the removal of glycine, the currents declined to 86 Ϯ 2.4% (n ϭ 6) of the control current magnitude and remained stable at this level until closed by a 1-min application of 10 mM DTT (e.g. Fig. 2A). When 0.5 mM glycine was applied to the MTSET-modified GlyRs, it reversibly activated an additional current component ( Fig. 2A). At any given time after the completion of the MTSET treatment, the total magnitude of the locked-open plus glycine-gated current was larger than that which could be activated in the same cell by a continuous application of 0.5 mM glycine alone. This point is illustrated in Fig 2, A-C. Fig. 2B shows the effect of a long application of 0.5 mM glycine to the same cell as in Fig. 2A, and both traces are shown superimposed in Fig. 2C. This experiment was repeated in five cells, and the relative current magnitudes were quantitated at a common time point 2 min after the initial application of glycine. It was found that an application of 0.5 mM glycine to the MTSET-modified GlyRs resulted in a net current magnitude that was 167 Ϯ 6% (n ϭ 5) larger than that activated in the same cell by a continuous application of 0.5 mM glycine alone. Together, these observations indicate that MTSET locked the channels into the open state but did not lock significant numbers of channels into either the closed or desensitized states. The MTSET-induced increase in net current magnitude at late times was most likely because of a reduced transition rate from the open to the desensitized state.
Because MTSET induced no current change in the presence of a saturating glycine concentration, its reaction rate in the fully activated state could not be measured. However, in the presence of an EC 50 (30 M) concentration of glycine, the reaction proceeded with a time constant of 1.2 Ϯ 0.1 s (n ϭ 4), indicating a reaction rate of around 830 M Ϫ1 s Ϫ1 . This is about 250 times smaller than the rate constant for the reaction of MTSET with 2-mercaptoethanol in free solution, the decrease because of electrostatic repulsion, steric hindrance, or suppressed ionization of the cysteine thiol (3). The possible contributions of these factors to the reactivity of T6ЈC are considered further below.
When applied at a concentration of 10 mM for 60 s, MTSES had no significant effect on either the ␣1 WT or ␣1 T6ЈC GlyRs regardless of whether it was applied in the absence or presence of a saturating concentration of glycine (17). In addition, a prior MTSES application in either the closed or open state did not significantly attenuate the ability of MTSET to lock the ␣1 T6ЈC FIG. 1. Amino acid sequence alignment of the M2 transmembrane segments of human GlyR ␣1 subunit and the rat GABA A R ␣1 and ␤1 subunits. The residues mutated to cysteine in this study are indicated in bold and numbered 6Ј according to the system of Miller (26), which assigns 1Ј to the most intracellular M2 domain residue and 20Ј to the most extracellular residue. Arrows denote those residues in the GABA A R ␣1 subunit that are exposed to the channel lumen (8).
A 60-s application of 2.5 mM MTSEA also had no significant effect on the ␣1 WT GlyR regardless of whether it was applied in the closed or open states (Table II). Similarly, when applied in the closed state to the ␣1 T6ЈC GlyR, 2.5 mM MTSEA had no significant effect on the magnitude of currents activated by an EC 50 (30 M) or a saturating (500 M) concentration of glycine (Table II). In addition, prior exposure of the ␣1 T6ЈC GlyR to MTSEA in the closed state did not significantly affect the ability of a subsequent application of 100 M MTSET plus 500 M glycine to lock the channels open (Fig. 2D). A 60-s application of MTSEA plus 500 M glycine also had no effect on the magnitude of currents activated by either 20 or 500 M glycine (Table II), although it dramatically attenuated the effect of a subsequent application of MTSET (Fig. 2E). As shown in Table  II, MTSET plus 500 M glycine caused 85 Ϯ 3% of channels to be locked into the open state, while simultaneously reducing the magnitude of the glycine-activable current by 88 Ϯ 2% (both n ϭ 3). Following MTSEA exposure, MTSET plus 500 M glycine caused only 16 Ϯ 5% of channels to be locked into the open state while reducing the magnitude of the glycine-activable current by 17 Ϯ 8% (both n ϭ 4). Both of these values are significantly different from those obtained without MTSEA pre-treatment. Taken together, these results provide strong evidence that MTSEA modifies T6ЈC in the channel open state but not in the closed state.
The effects of MTSEA were likely to have been caused by the covalent attachment of an ethylammonium group to the 6Ј cysteine in the open state. The inability of MTSEA modification to lock the channels open may have been because of the smaller size of MTSEA relative to MTSET. On the other hand, the effects of MTSET may have been because of one of two mechanisms. One possibility is that it directly modified the 6Ј cysteines by covalently attaching an ethyltrimethylammonium group. In this case the reaction would have proceeded only in the open state, and the resulting cysteine modification would have maintained the pore in the open state. However, because the methanethiosulfonate (MTS) group contains a disulfide bond that could directly catalyze the formation of other disulfide bonds, it is also possible that MTSET may have behaved as an oxidizing agent; MTSET can add thioethyltrimethylammonium to one cysteine, and a second cysteine can displace this group in a sulfhydryl-disulfide interchange to generate a cystine-cysteine disulfide. MTSET could thereby induce the for-mation of disulfide bonds between subunits, preventing the channels from closing.
To discriminate between these two possibilities, we tested the effects of oxidizing reagents on the GlyR. We examined the effects of 1-min applications of 0.3% H 2 O 2 and 100:400 M Cu:phen on the ␣1 WT and ␣1 T6ЈC GlyRs. As summarized in Table II, neither reagent had any effect on either the halfmaximal or maximal current magnitudes of the ␣1 WT or the ␣1 T6ЈC GlyRs. Furthermore, neither reagent was able to mimic the effect of MTSET in maintaining the ␣1 T6ЈC GlyR in the open state (n ϭ 3 for each reagent). An example of such an experiment on the ␣1 T6ЈC GlyR is shown in Fig. 3. Although Cu:phen induced a weak transient inhibition, it had no irreversible effects (Fig. 3B). The H109A mutation, which eliminates zinc inhibition (21), had no effect on this transient inhibitory action of copper (data not shown).
Cysteine reactivity with thiol-containing compounds is determined by the local electrostatic potential, the sulfhydryl ionization state, and steric accessibility of the MTS reagent to the sulfhydryl group (3). Unfortunately, it was not possible to determine the contribution of electrostatic potential changes as the only available soluble, negatively charged MTS derivative, MTSES, had no measurable effect (17). However, it is unlikely that electrostatic potential changes alone would have been able to account for the infinitely large observed reaction rate difference (see Ref. 5). Thus, the reaction rate was likely to have been dominated by the sulfhydryl ionization state or steric accessibility. Because the MTS reaction rate increases dramatically with thiol ionization (22), and thiol ionization is suppressed in a hydrophobic environment, one possibility is that the 6Ј cysteines exist in a hydrophobic environment in the closed state (perhaps by facing the protein interior) and increase their exposure to the aqueous environment in the open state. An equally plausible alternative is that the 6Ј cysteines remain in an aqueous environment in the closed state but that access of the externally applied MTS reagents in the closed state is precluded by either an electrostatic impediment or pore constriction external to the 6Ј position. In either scenario, the access of MTSET to the 6Ј cysteines is increased in the open state, and MTSET holds the channel open by covalently attaching a positively charged ethyltrimethylammonium group to T6ЈC.
Sulfhydryl Modification of the ␣1 T6ЈC␤ 1 T6ЈC GABA A R-Both of the above models contrast dramatically with results obtained recently on the structurally and functionally homologous GABA A R by Horenstein et al. (18). That study investigated the a Results for the WT and Phe, Ala, and Cys mutants are reproduced from Shan et al. (17). b The data are significantly different from WT GlyR (p Ͻ 0.05). c The glycine n H was not given, because trace glycine in the control solution distorted the current magnitude at lower glycine concentrations. As a correction, data shown in parentheses were recorded from cells that were switched from 1 M strychnine immediately into glycine-containing solutions.
d ND, not determined.
state-dependent reactivity changes of the T6ЈC residues in the rat ␣1 T6ЈC ␤1 T6ЈC GABA A R expressed recombinantly in Xenopus oocytes. They concluded that the T6ЈC residues are exposed to the external aqueous environment in the closed state and rotate to face the adjacent subunit when the channel is opened. Furthermore, when applied in the open state, Cu:phen promotes the formation of an intersubunit disulfide bond between adjacent ␤1 subunits that locks the channel in the open state (18). We examined the effects of cysteine-reactive reagents on the rat ␣1 WT ␤1 WT and ␣1 T6ЈC ␤1 T6ЈC GABA A Rs expressed recombinantly in mammalian HEK293 cells.
The mean EC 50 , n H , and I max values for GABA-activated currents in the WT and mutant GABA A Rs are summarized in Table III. We were surprised to find that incorporation of the T6ЈC mutations into both the ␣1 and ␤1 subunits resulted in a dramatic increase in the rate of desensitization (e.g. Fig. 4A). In the presence of a saturating 20 M (10 ϫ EC 50 ) GABA concentration, the ␣1 WT ␤1 WT GABA A R desensitized with a time constant of 1370 Ϯ 280 ms (n ϭ 4) whereas in the presence of 100 M (20 ϫ EC 50 ) GABA, the ␣1 T6ЈC ␤1 T6ЈC GABA A R desensitized with a time constant of 87 Ϯ 2 ms (n ϭ 4). This rapid desensitization rate made it difficult to apply cysteine-modifying reagents with a high degree of confidence to the channel open state. In the absence of GABA, there was no significant difference in the resting conductance of cells expressing ␣1 WT␤ 1 WT and ␣1 T6ЈC ␤1 T6ЈC GABA A Rs, implying that the mutations did not induce a steady-state leak conductance through the receptors.
When activated by 20 M GABA, the ␣1 WT ␤1 WT GABA A R was weakly but significantly potentiated by a 2-min application of 10 mM DTT (see Fig. 4B and Table II). Upon removal of DTT, currents gradually returned to the control magnitude over the following 3-5 min. This effect is similar to that observed when the same receptors are expressed in Xenopus oocytes (18). In contrast to this relatively modest effect, a 10 mM application of DTT caused a dramatic potentiation of the ␣1 T6ЈC ␤1 T6ЈC GABA A R when activated by 100 M GABA (see Fig. 4C and Table II). It appears that the T6ЈC residues of both the ␣1 and ␤1 subunits contributed to this effect as DTT had a similar effect on the ␣1 WT ␤1 T6ЈC GABA A R and the ␣1 T6ЈC ␤1 WT GABA A R (Table II). The DTT-potentiated currents in the ␣1 T6ЈC ␤1 T6ЈC GABA A R declined progressively when the cell was perfused in DTT-free bathing solution (Fig. 4C). The potentation observed in both the WT and mutant receptors may have been because of either the reduction of endogenous disulfide bonds or a pharmacological effect of DTT at the alcohol or anesthetic binding site (23). To discriminate between these two possible modes of action, we investigated the effect of 200 mM ethanol in the presence of a saturating (100 M) GABA concentration on both the ␣1 WT␤ 1 WT GABA A R and the ␣1 T6ЈC ␤1 T6ЈC GABA A R. As summarized in Table II, ethanol had no significant effect on either receptor, indicating that DTT was acting by reducing endogenous disulfide bonds.
When applied in the closed channel state, Cu:phen had no effect on the ␣1 WT␤ 1 WT GABA A R (Table II, Fig. 5A, left panel). However, in the ␣1 T6ЈC ␤1 T6ЈC GABA A R, the rate of current reduction upon removal of DTT was accelerated dramatically by Cu:phen (Fig. 5B). Following the removal of DTT, the GABA-activated current reduced to 76 Ϯ 3% (n ϭ 3) after 20 s in the standard bathing solution. However, in the presence of Cu:phen, the GABA-activated current magnitude reduced to 3.3 Ϯ 2% (n ϭ 3) of control magnitude after 20 s. When combined with the results obtained using DTT, these results indicate that disulfide bonds form spontaneously, but relatively slowly, in the closed state in the ␣1 T6ЈC ␤1 T6ЈC GABA A R. Because this slow rate of disulfide bond formation complicated investigations into the reactivity of the 6Ј cysteines, all subsequent experiments on ␣1 T6ЈC ␤1 T6ЈC GABA A Rs in the closed state were performed immediately following a 2-min exposure to 10 mM DTT to ensure that all 6Ј cysteines were in the reduced state. Then, the effects of subsequent pharmacological manipulations were compared with the effects of spontaneous disulfide formation in the same cell.
When applied in the presence of 20 M GABA, Cu:phen had no effect on the ␣1 WT ␤1 WT GABA A R (Table II, Fig. 5A, right panel). However, when Cu:phen was applied to the ␣1 T6ЈC ␤1 T6ЈC GABA A R in the presence of 100 M GABA, it had two distinct effects. First, it reversibly reopened the channel from the desensitized state (Fig. 5C). Second, following the removal of Cu:phen, the peak magnitude of GABA-activated currents was decreased dramatically (see Table II and Fig. 5C). This reduction in current was not spontaneously reversible but was reversed by a 30 -60-s application of 10 mM DTT (see Table II and Fig. 5C).
We were surprised by the ability of GABA ϩ Cu:phen to reopen the channels and investigated this phenomenon further. The reopening effect was found to require the simultaneous presence of GABA and Cu:phen. If either reagent was removed, the receptor immediately resumed a non-conducting configuration (n ϭ 5 for each condition). Application of 100 M CuSO 4 in the presence of GABA caused no detectable current activation (n ϭ 3 cells), thus eliminating a putative pharmacological action of copper. Furthermore, in the continuous presence of GABA, a second application of Cu:phen elicited a current of similar magnitude to the first (n ϭ 3 cells). This last observation eliminated the possibility that the formation of disulfide bonds following the first application of Cu:phen may have closed the channels and prevented Cu:phen from subsequently reopening them. Finally, H 2 O 2 also caused a dramatic 87 Ϯ 3% (n ϭ 4) reduction in the magnitude of the GABAactivable current that was reversed by 10 mM DTT (Fig. 5D). However, H 2 O 2 did not activate the receptors convincingly. Although Cu:phen activated a current with a magnitude of 28 Ϯ 3% (n ϭ 3) of the saturating GABA-activated current magnitude, H 2 O 2 activated a current of only 7 Ϯ 2% (n ϭ 4) of the saturating GABA current magnitude. This difference was significant (p Ͻ 0.05) using a one-way analysis of variance.
MTSET was used to further investigate the state-dependent surface accessibility of the 6Ј cysteines. MTSET had no signif-

6Ј Cysteine Accessibility in Glycine and GABA A Receptors
icant effect on the ␣1 WT ␤1 WT GABA A R regardless of whether it was applied in the absence or presence of GABA (see Fig. 6A and Table II). However, when MTSET was applied to the ␣1 T6ЈC ␤1 T6ЈC GABA A R in the closed channel state, its effects closely resembled those of Cu:phen. Following the removal of DTT, the GABA-activated current reduced to 74 Ϯ 6% (n ϭ 4) after 20 s in the standard bathing solution (e.g. Fig. 6B, left  panel). However, in the presence of MTSET, the GABA-activated current reduced to 14 Ϯ 3% (n ϭ 4) of control magnitude after 20 s.
The effect of MTSET on the ␣1 T6ЈC ␤1 T6ЈC GABA A R was also examined in the desensitized state. In this experiment, GABA was applied 2 s before MTSET to ensure that Ͼ90% of receptors were in the desensitized state. MTSET was found to re-open the channels from this state (Fig. 6C). This reaction proceeded with an average time constant of 35 Ϯ 9 s (n ϭ 4), indicating a mean reaction rate of 29 M Ϫ1 s Ϫ1 . Upon removal of both MTSET and GABA, the current magnitude reduced to a steady-state level of 31 Ϯ 5% (n ϭ 4) of the peak MTSET-induced current magnitude (Fig. 6C), indicating that around one-third of the channels were held in the open state. MTSET modification also strongly reduced the magnitude of the current that was available for activation by GABA (Fig. 6C), indicating that the remainder of the channels were returned to the closed desen-sitized state. The MTSET-modified receptors were returned efficiently to the closed state by DTT, and a subsequent application of GABA activated the currents with a peak magnitude similar to the original control (Fig. 6C). Similar results were observed in each of four cells.
These results indicate that the effects of MTSET on the ␣1 T6ЈC ␤1 T6ЈC GABA A R depend on whether it is applied in the closed or desensitized states. Because it is unlikely that both actions could have been mediated by covalent attachment of the same ethyltrimethylammonium group, it is possible that at least one of the actions may have been mediated by MTSET acting as an oxidizing reagent or by reacting with a nonidentical set of subunits.
MTSEA, applied in either the closed and open states, has been shown previously to irreversibly reduce the magnitude of currents in Xenopus oocyte-expressed ␣1 T6ЈC ␤1␥2 GABA A Rs (8). In this study, we investigated the effect of 2.5 mM MTSEA on the ␣1 WT ␤1 WT GABA A R and the ␣1 T6ЈC ␤1 T6ЈC GABA A R ex-  In the right panel, the current reduction rate was greatly accelerated by 100:400 M Cu:phen and reversed by a subsequent application of 10 mM DTT. C, when applied together with 100 M GABA in the desensitized state, Cu:phen reversibly reopens the channels. However, a subsequent GABA application reveals a dramatic current reduction that is reversed by 10 mM DTT, implying the formation of disulfide bonds in the closed or desensitized states. D, results of a similar experiment to C, but using 0.3% H 2 O 2 in place of Cu:phen. pressed in HEK293 cells. As summarized in Table II, MTSEA had no effect on the ␣1 WT ␤1 WT GABA A R in either the absence or presence of a saturating GABA concentration. However, when applied in the closed state to the ␣1 T6ЈC ␤1 T6ЈC GABA A R, it irreversibly activated the channels to 30 Ϯ 5% (n ϭ 3) of the peak current magnitude while simultaneously reducing the magnitude of the current activated by a saturating (20 M) concentration of GABA (see Fig. 6D and Table II). A 10 mM DTT application efficiently closed the channels and restored the original magnitude of the GABA-activated current. When applied together with 20 M GABA in the channel-desensitized state, MTSEA mimicked the effect of MTSET in returning the channels to the open state (see Fig. 6E and Table II).
Effect of 6Ј Mutagenesis on GlyR Function-To further probe the relationship between the physicochemical properties of the 6Ј residue and the function of the receptor, we introduced a series of mutations at the 6Ј position of the GlyR ␣1 subunit. The identity of these mutations and their effects on I max , EC 50 , and n H values of glycine-gated currents in ␣1 homomeric receptors are summarized in Table I. This table also shows the effect of each mutation on the picrotoxin sensitivity of currents activated by the EC 50 glycine concentrations as indicated. Note that GlyRs incorporating serine, glutamine, glutamic acid, and lysine mutations did not yield measurable currents. Interestingly, glutamine, glutamic acid, and lysine were the most polar amino acids tested.
The EC 50 is a measure of the free energy input required to activate the receptor. If channel opening is accompanied by a movement of the 6Ј residue toward an increasingly hydrophilic environment, it might be expected that the ease of activating the receptor should be a function of the hydropathy of the introduced amino acid. This was investigated by plotting the glycine EC 50 values against some properties of the substituted amino acids (Fig. 7). This figure reveals that there was no significant correlation between glycine EC 50 and side-chain volume, hydrophilicity, hydrophobicity, or hydropathy. We conclude that the relationship between the channel gating energy and the physicochemical properties of the introduced residues is complex.

GlyR in the Closed and Open
States-When applied in the absence of glycine, MTSET has no effect, but when applied in the presence of glycine, MTSET locks the ␣1 T6ЈC GlyR in the open state (17). Because this action is not mimicked by oxidizing reagents, MTSET must act by adding a polar quaternary ammonium group to one or more 6Ј cysteines in the open state only. This attached group prevents the channel from closing either by steric hindrance because of its size or by biasing the conformational equilibrium toward the open state because of its affinity with the aqueous pore environment. The smaller hydrophilic cysteine-specific reagent, MTSEA, also modified T6ЈC in the open state only. However, MTSEA-modified GlyRs closed readily upon removal of glycine. Together, these observations indicate that GlyR channel opening is accompanied by an increase in the exposure of the 6Ј cysteines to the external aqueous environment. This may arise because of either 1) an increase in the ionization state of the cysteines because of a transition  (27), hydrophobicity (28), hydrophilicity (29), and hydropathy (30). The p value refers to the probability that the linear coefficient R value was zero.
On the other hand, substituted cysteine accessibility studies reveal that cationic LGIC family members have remarkably similar patterns of M2 domain residue exposure in the channel open state (4 -7, 9, 10). Of particular relevance to the present study, MTSET modification of 6Ј cysteines irreversibly inhibited current in both the nAChR and 5HT 3 R, whereas MTSES had no effect on either receptor (4 -7, 9, 10). The present study could not directly compare 6Ј cysteine accessibility in the open states of the ␣1 T6ЈC GlyR and ␣1 T6ЈC ␤1 T6ЈC GABA A R because of the fast desensitization rate of the ␣1 T6ЈC ␤1 T6ЈC GABA A R. The observation that MTSET locked both receptors into the partially open state provides strong evidence for a common activation mechanism in this part of the pore. However, the pore structures are unlikely to be identical as MTSEA also locked the ␣1 T6ЈC ␤1 T6ЈC GABA A R in the open state but had no such effect on the ␣1 T6ЈC GlyR.
The present study reveals distinct differences in the properties of GABA A Rs expressed in Xenopus oocytes and HEK293 cells. When expressed in HEK293 cells, the 6Ј cysteines can form disulfide bonds in the closed state. However, this does not occur when the same receptors are expressed in Xenopus oocytes (18). Furthermore, when expressed in HEK293 cells, the GABA A R is locked in the desensitized state by Cu:phen, but when expressed in Xenopus oocytes, it is locked in the open state by Cu:phen (18). Together, these results indicate the surface orientation of the GABA A R 6Ј cysteines varies dramatically depending on the expression system. Moreover, ␣1 T6ЈC ␤1 T6ЈC GABA A Rs expressed in HEK293 cells desensitize at a much faster rate than they do when expressed in Xenopus oocytes. These structural and functional differences could be because of expression system-specific differences in subunit folding and assembly, post-translational modifications, or membrane lipid composition. Regardless of their origin, the results indicate that caution should be applied when comparing results obtained using the two expression systems.