Transmembrane Domain I of the γ-Aminobutyric Acid Transporter GAT-1 Plays a Crucial Role in the Transition between Cation Leak and Transport Modes*

The sodium- and chloride-dependent γ-aminobutyric acid (GABA) transporter is essential for synaptic transmission by this neurotransmitter. GAT-1 expressed inXenopus laevis oocytes exhibits sodium-dependent GABA-induced inward currents reflecting electrogenic sodium-coupled transport. In lithium-containing medium, GAT-1 mediates GABA-independent currents, the relationship of which to the physiological transport process is poorly understood. In this study, mutants are described that appear to be locked in this cation leak mode. When Gly63, located in the middle of the highly conserved transmembrane domain I, was mutated to serine or cysteine, sodium-dependent GABA currents were abolished. Strikingly, these mutants exhibited robust inward currents in lithium- as well as potassium-containing media. Membrane-impermeant sulfhydryl reagents inhibited these currents of the cysteine but not of the serine mutant, indicating that this position was accessible to the external aqueous medium. The cation leak currents mediated by wild-type GAT-1 were inhibited by low millimolar sodium concentrations in a noncompetitive manner. Mutations at other positions of transmembrane domain I increased or decreased the apparent sodium affinity, as monitored by the sodium-dependent steady-state GABA currents or transient currents. In parallel, the ability of sodium to inhibit the cation leak currents was increased or decreased, respectively. Thus, transmembrane domain I of GAT-1 contains determinants controlling both sodium-coupled GABA flux and the cation leak pathway as well as the interconversion of these distinct modes. Our observations suggest the possibility that the permeation pathway in both modes shares common structural elements.

The neurotransmitter ␥-aminobutyric acid (GABA) 1 is a major inhibitory neurotransmitter in the central nervous system. Its synaptic action, like that of most other neurotransmitters, is terminated by sodium-coupled transport. Many neurotransmitters are removed from the synaptic cleft by sodium-and chloride-dependent neurotransmitter transporters (for a review, see Refs. 1 and 2). These transporters form a large family and include transporters for biogenic amines and amino acids. Besides GAT-1, the first member of the family to be identified, it contains three other GABA transporters. One of the best examples of the importance of these neurotransmitter transporters comes from studies of dopamine transporter knockout mice; the decay of extracellular dopamine in brain slices of such mice is ϳ100 times longer than normal (3).
GAT-1 was purified to homogeneity in a form active upon reconstitution (4) and subsequently cloned (5). The transporter catalyzes electrogenic sodium/chloride/GABA cotransport with a stoichiometry of 2:1:1 (6 -9). It should be noted, however, that there still is some dispute regarding this issue. In a recent report, it has been proposed that, during sodium-coupled GABA transport, obligatory Cl out /Cl in exchange takes place (10).
GAT-1, as well as the other members of the family, is predicted to have 12 transmembrane domains linked by hydrophilic loops with the amino and carboxyl termini residing inside the cell (5). Studies on the serotonin transporter SERT suggest that the theoretical topological model is correct (11). Mutagenesis studies (in particular on GAT-1 and SERT, but also on other members of the family) suggest the importance of transmembrane domains I and III in the interaction of the transporters with the neurotransmitters as well as with sodium and chloride (12)(13)(14)(15)(16).
Using biochemical approaches, it has been shown that GAT-1 undergoes extensive conformational changes upon interaction with its substrates (17,18). Electrophysiological measurements of GAT-1 have demonstrated sodium-dependent GABA-induced inwardly rectifying currents reflecting the electrogenic sodium-and chloride-coupled translocation of GABA (7)(8)(9). Moreover, in the absence of GABA, but in the presence of sodium, GAT-1 exhibits capacitative transient currents (8,13,14). These transients are thought to reflect a charge-moving conformational change that takes place after the binding of sodium and that enables subsequent GABA binding (8,13,19). Remarkably, in the presence of lithium, inwardly rectifying steady-state currents are observed in the absence of GABA (13,14). Such leak currents have also been observed in many transporters for neurotransmitters (20 -23) as well as in other transporters such as proton-coupled metal transporters (24). In the latter case, it has been suggested that this leak pathway may protect cells from metal ion overload (24). However, the relationship of these leak currents to coupled transport is poorly understood.
In the case of the uncoupled currents observed with the glutamate transporters, it has been shown that the conformation of the transporter mediating this process is distinct from the coupled process (25)(26)(27)(28). Recently, co-workers and I have obtained some evidence that the same may be true for the leak currents mediated by . In this study, I show that the leak mode of GAT-1 represents a unique conformation of the transporter. Mutants at a water-accessible position in the middle of the highly conserved transmembrane domain I that appear to be locked in the leak mode are described. Moreover, other determinants located in this transmembrane domain that control the sodium-dependent transition of the leak to transport mode have been identified here.

EXPERIMENTAL PROCEDURES
Generation and Subcloning of Mutants-Mutations were made by site-directed mutagenesis of wild-type GAT-1 in the vector pBluescript SK(Ϫ) (Stratagene) according to the method of Kunkel et al. (30) as described (31). Briefly, the parent DNA was used to transform Escherichia coli CJ236 (dut Ϫ ,ung Ϫ ). From one of the transformants, singlestranded uracil-containing DNA was isolated upon growth in uridinecontaining medium according to the standard protocol from Stratagene using helper phage R408. This yields the sense strand; and consequently, mutagenic primers were designed to be antisense. Restriction sites ClaI and AvrII were used to subclone the mutations into the construct containing wild-type GAT-1 in pOG 1 (14). The latter is an oocyte expression vector that contains a 5Ј-untranslated Xenopus ␤-globin sequence and a 3Ј-poly(A) signal. The coding and noncoding strands were sequenced between the above two restriction sites.
cRNA Transcription, Injection, and Oocyte Preparation-Capped runoff cRNA transcripts were made from transporter constructs in pOG 1 linearized with SacII using mMessage mMachine (Ambion Inc.). Oocytes were removed from anesthetized Xenopus laevis frogs and treated with collagenase (type 1A, Sigma C-9891) until capillaries were absent and injected with 50 nl of undiluted cRNA the same or the next day. Oocytes were maintained at 18°C in modified Barth's saline (88 mM NaCl, 1 mM KCl, 1 mM MgSO 4 , 2.4 mM NaHCO 3 , 1 mM CaCl 2 , 0.3 mM Ca(NO 3 ) 2 , and 10 mM HEPES, pH 7.5) with freshly added 2 mM sodium pyruvic acid and 0.5 mM theophylline and supplemented with 10,000 units/liter penicillin, 10 mg/liter streptomycin, and 50 mg/liter gentamycin.
Oocyte Electrophysiology-Oocytes were placed in the recording chamber; penetrated with two micropipettes (back-filled with 2 M KCl, with resistance varied between 0.5 and 2 megaohms); and voltageclamped using GeneClamp 500 (Axon Instruments, Inc.) and digitized using Digidata 1200A (Axon Instruments, Inc.), both controlled with the pClamp6 suite (Axon Instruments, Inc.). Currents were acquired with clampex6.03 and low pass-filtered at 10 kHz every 0.5 ms. Oocytes were stepped from Ϫ160 to ϩ40 mV in 25-mV increments, using Ϫ25 mV as the holding potential unless stated otherwise in the figure legends. Each potential was clamped for 500 ms. The membrane potential was measured relative to an extracellular Ag/AgCl electrode in the recording chamber. The recording solution contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , and 5 mM HEPES, pH 7.4.
In substitution experiments, sodium ions were replaced with equimolar choline or lithium. The records shown in Figs. 2, 3, 6, and 7 are typical and representative of results from four to eight oocytes. In the other figures, the currents were normalized as indicated in the legends to plot results of 3-13 oocytes as means Ϯ S.E. Wherever error bars are not visible, the error was smaller than the size of the symbols. Analysis was performed with Clampfit Version 6.05 and Origin Version 6.1 (Microcal).
Cell-surface Biotinylation-Labeling of wild-type and mutant transporters at the cell surface using sulfosuccinimidyl 2-(biotinamido)ethyl-1,3-dithiopropionate, SDS-PAGE, and Western blot analysis was done as described by Stern-Bach et al. (32). For each determination, five oocytes expressing either the wild type or the indicated mutant were used. One-eighth of the sample was retained as the source of total protein, and the remainder was treated with streptavidin beads to recover the biotinylated proteins. The GAT-1 protein was detected with an affinity-purified antibody to an epitope located in the carboxyl terminus of GAT-1 (residues 571-586), horseradish peroxidase-conjugated secondary antibody, and ECL as described (33).

Characterization of Currents Mediated by the G63S and G63C
Mutants-Of the members of the sodium-and chloridedependent neurotransmitter transporter family, the biogenic amine transporters possess a conserved aspartate residue in the middle of transmembrane domain I. All other members of the transporter family, including GAT-1, have a glycine at the corresponding position (Gly 63 in GAT-1). In the study of the structure/function relationships in transmembrane domain I of GAT-1, I mutated Gly 63 to aspartate as well as to serine, cysteine, and alanine. This led to the complete loss of sodiumand chloride-dependent [ 3 H]GABA uptake, as observed upon expression in HeLa cells (Fig. 1). A similar loss of function was observed with the sodium-dependent GABA-induced transport currents in X. laevis oocytes as shown here for the G63C and G63S mutants (Fig. 2, A, C, and E). Remarkably, however, the two mutants exhibited robust GABA-independent inwardly rectifying leak currents in lithium medium (Fig. 2, D and F). The currents of both of these mutants were similar in magnitude and voltage dependence to that mediated by wild-type GAT-1 (Fig. 2B) and were not influenced by GABA (data not shown). The kinetics (but not the steady-state level) of the currents observed in lithium were dependent on the nature of the substitution at position 63. In the G63S (but not G63C) mutant, the onset of these currents was markedly slowed down ( Fig. 2, B, D, and F). Results similar to those obtained with G63C were obtained when Gly 63 was replaced by alanine, but no lithium leak currents were observed when aspartate was substituted for Gly 63 (data not shown).
Even though position 63 is located in the middle of transmembrane domain I, it is possible that this conserved domain lines an aqueous pathway through the transporter, and a cysteine residue introduced at this position is accessible to hydrophilic membrane-impermeant sulfhydryl reagents. Consistent with this idea, the lithium-dependent currents of G63C (but not of G63S) were almost completely abolished by positively charged MTSET as well as by negatively charged (2-sulfonatoethyl) methanethiosulfonate ( Fig. 3) (data not shown). A conserved cysteine (Cys 74 in GAT-1) located in the first extracellular loop is the only endogenous cysteine that is externally accessible in GAT-1 and most members of this family (36 -38).
To exclude the possibility that the sensitivity of the lithiumdependent currents observed with G63C is due to an increased accessibility of Cys 74 , the G63C and G63S mutants were subcloned into a construct in which Cys 74 was replaced by alanine (C74A). The currents of this latter mutant were very similar to those mediated by the wild type (data not shown). The lithiumdependent currents of the G63C/C74A mutant (Fig. 3A) were almost completely abolished by pretreatment with MTSET ( Fig. 3B). On the other hand, no effect of MTSET on the lithiumdependent currents of the G63S/C74A mutant was observed (Fig. 3, C and D). Similarly, 2-sulfonatoethyl methanethiosulfonate also inhibited these currents of G63C/C74A, but not of G63S/C74A (data not shown). It is of interest to note that the difference in the kinetics of the onset of the currents observed with G63C and G63S (Fig. 2, D and F) was also observed in the mutants constructed in the C74A background (Fig. 3, A and C).
Effects of Sodium on the Lithium Leak Currents Mediated by the Wild Type-The absence of [ 3 H]GABA transport and GABA-induced currents in the G63S and G63C mutants indicated that they were confined to the leak current mode. Further evidence that the leak currents represent a distinct mode of the transporter is presented in Fig. 4. The effect of sodium on the lithium currents of the wild type was measured at varying lithium levels. The lithium-dependent current did not saturate FIG. 2. GABA-induced and lithium currents mediated by wild-type, G63S, and G63C transporters. A, C, and E, currents recorded in sodium medium in the absence of GABA during 500-ms voltage jumps from Ϫ160 to ϩ40 mV (steps of ϩ25 mV) have been subtracted from currents in the same medium containing 1 mM GABA. The prepulse potential was Ϫ25 mV. B, D, and F, currents recorded in choline medium have been subtracted from those in lithium medium using the same oocytes as in A, C, and E, respectively, with the same voltage jump protocol. The dotted lines indicate 0 current. Representative oocytes are shown. WT, wild type. with increasing lithium concentrations, at least not up to 90 mM (Fig. 4). Sodium at a concentration as low as 1 mM inhibited the current at all lithium concentrations tested. However, the inhibition by sodium was larger at higher lithium concentrations (Fig. 4). Thus, these effects are not due to a competition between lithium and sodium, but rather sodium appears to act at a distinct site to suppress the leak mode. As sodium is required for coupled transport, an important question is whether the sodium-induced conformational changes of the transport cycle are related to the suppression of the leakage mode. To address this, mutants with decreased and increased apparent sodium affinities were analyzed.
Effect of Sodium on Currents Mediated by Y60T-Tyr 60 of GAT-1, also located in transmembrane domain I, is unique among the GABA transporters. In the three other GABA transporters (GAT-2-4), a glutamate residue occupies the equivalent position of Tyr 60 . Interestingly, when the Y60E mutant was expressed in HeLa cells, no sodium-and chloride-dependent [ 3 H]GABA uptake could be detected, and the same was true for the threonine substitution mutant, but the Y60C mutant exhibited significant activity (Fig. 1). Upon expression of Y60E in X. laevis oocytes, no sodium-dependent GABA-induced transport currents could be detected, and the same was true for the lithium leak currents (data not shown). In contrast to Y60E, the Y60T mutant exhibited robust lithium leak currents, ranging from Ϫ1195 to Ϫ2791 nA (n ϭ 6) at Ϫ160 mV (Fig. 5A). The same was true for oocytes expressing Y60C; but in contrast to G63C, these currents mediated by Y60C could not be blocked by MTSET (data not shown). The size of the lithium-dependent currents of Y60T was even larger than those of wild-type GAT-1 (Fig. 5B): Ϫ485 to Ϫ2081 nA (n ϭ 13). Sodium-and chloride-dependent [ 3 H]GABA uptake was typically measured at very low GABA concentrations. However, at a saturating concentration of 1 mM, Y60T exhibited significant GABA transport currents (Fig. 5A), ranging from Ϫ31.6 to Ϫ90.8 nA at Ϫ160 mV (n ϭ 6). In the corresponding oocytes expressing the wild type, the GABA transport currents ranged from Ϫ342 to Ϫ1544 nA (n ϭ 13). The voltage dependence of the currents of the mutant and wild type was similar (Fig. 5, A and B), but it is obvious that the ratio of the lithium leak current to the GABA current differed dramatically between Y60T and the wild type.
Even though the GABA transport currents of the mutant were of small magnitude, their size was large enough to determine the sodium dependence of this process (Fig. 5C). More than 5-fold higher sodium concentrations were needed to halfmaximally activate the GABA currents of Y60T compared with those of the wild type (Fig. 5C). It was not possible to determine the exact shift of the apparent sodium affinity in the mutant because, even at 96 mM sodium, the GABA-induced current was not yet saturated (Fig. 5C). In parallel with the decreased apparent affinity of Y60T for sodium, 28.8 mM sodium inhibited the lithium leak current by 26 Ϯ 2% (n ϭ 4). In contrast, the IC 50 for sodium in the same batch of oocytes expressing the wild type was 1.1 Ϯ 0.1 mM (n ϭ 3) (Fig. 5D).
Effect of Sodium on Currents Mediated by R69K-Arg 69 , also located in transmembrane domain I of GAT-1, is fully conserved in the entire family of sodium-and chloride-dependent neurotransmitter transporters. Mutation of this residue, even to a conserved one such as lysine (R69K), leads to defective [ 3 H]GABA uptake (12). Indeed, no steady-state GABA currents were observed in this mutant even though transients were present (Fig. 6A). Even at 5 mM GABA, no currents could be observed, but the transients were somewhat larger (data not

FIG. 5. Effects of sodium on GABA-induced and lithium-dependent currents mediated by Y60T mutant and wild-type GAT-1.
Lithium-induced steady-state currents and GABA-induced currents in sodium-containing medium (both obtained by subtracting the response in sodium) in Y60T (A) and the wild type (WT) (B) were averaged over 40 ms at each of the potentials shown (the interval between 430 and 470 ms after the onset of the voltage jump). The results were normalized to the GABA-induced current at Ϫ160 mV and are the averages of six oocytes from four different batches (Y60T) and 13 oocytes from seven different batches (wild type). The S.E. is indicated and, in most cases, is smaller than the size of the symbols. The GABA-induced currents ranged from Ϫ342 to Ϫ1544 nA, and the lithium-induced currents ranged from Ϫ485 to Ϫ2081 nA both at Ϫ160 mV for the wild type. The corresponding values for Y60T were Ϫ31.6 to Ϫ90.8 nA and Ϫ1195 to Ϫ2791 nA for GABA-and lithium-induced currents, respectively. Net GABA-induced currents (C) and lithium-dependent currents (D) at steady state at Ϫ160 mV were recorded at the indicated sodium concentrations. The results were normalized to the values for the wild type and mutant obtained at 96 mM sodium (C) or at 0 sodium (D). The lithium concentration in D was 67.2 mM, and choline was used to maintain iso-osmolarity. The data are averages from four oocytes of the wild type and Y60T in A and three (wild type) and four (Y60T) oocytes in D.
The observation that transient currents could be isolated by GABA (Fig. 6A) indicated that, even though GABA was not transported, it could still bind to the R69K mutant. This was further substantiated by the fact that, in sodium medium, the non-transportable GABA analog SKF100330A could isolate the sodium-dependent transient currents mediated by R69K (Fig.  7A). In contrast to the corresponding currents mediated by the wild type (Fig. 7B), R69K exhibited transients in the "on" phase only when jumping from a holding potential of Ϫ25 mV to more positive potentials of up to ϩ50 mV, and the opposite was true in the "off" phase (Fig. 7A). The sodium-dependent transients were observed only when Arg 69 was changed to lysine, but not when the substitution was with glutamine or cysteine (data not shown). The transients observed with R69K were capacitative; the transient currents were of the same magnitude as those obtained when jumping back to the holding potential (data not shown), and they increased when the potential was jumped to values more positive than ϩ50 mV (data not shown). An attractive explanation is that, at a holding potential of Ϫ25 mV, all the transporters are already in the sodium-bound state, and the sodium is released by jumps to positive potentials, thereby giving rise to the outward transient current seen in Fig. 7A. Indeed, when the external sodium concentration was progressively decreased, an inward transient became apparent (data not shown). In parallel with the increased apparent affinity of R69K for sodium, this cation inhibited the lithium-dependent currents of R69K at lower concentrations (IC 50 ϭ 0.3 Ϯ 0.1 mM, n ϭ 4) compared with the wild type expressed in the same batch of oocytes (IC 50 ϭ 1.0 Ϯ 0.1 mM, n ϭ 3) (Fig. 8).
Characterization of Potassium Leak Currents-It has been found that GAT-1 also mediates leak currents in cesium, even though they are much smaller than those in the presence of lithium (13). Interestingly, GAT-1 also mediated potassium leak currents, which, at Ϫ160 mV, were 62.8 Ϯ 3.8% (n ϭ 6) of those observed in lithium (Fig. 9A). Such currents were not observed in non-injected oocytes (data not shown). The potassium currents had a similar voltage dependence compared with the currents observed in lithium (data not shown); and as with the lithium currents, they were also inhibited by sodium ions (Fig. 9B). The potassium currents mediated by R69K were more sensitive to sodium than those mediated by the wild type (Fig. 9B). In Y60T, the sensitivity of the potassium currents to sodium was markedly reduced (Fig. 9B), as in the case of the lithium currents (Fig. 5D). Importantly, the ratio of the potassium to lithium currents was different in wild type, R69K, and Y60T (Fig. 9A). The same was also true for the G63S and G63C mutants, in which this ratio was 0.95 Ϯ 0.10 and 0.93 Ϯ 0.11, respectively (n ϭ 3). As for the lithium currents, the potassium currents mediated by G63C/C74A (but not by G63S/C74A) were inhibited by treatment with MTSET (data not shown).
Cell-surface Expression-The transients mediated by R69K were smaller than those mediated by the wild type (Fig. 7). To investigate the possibility that this may be due to decreased cell-surface expression, oocytes expressing wild-type and mutant transporters were treated with the impermeant biotinylation reagent sulfosuccinimidyl 2-(biotinamido)ethyl-1,3-dithiopropionate. Upon solubilization of the membrane proteins with detergent, the biotinylated proteins were isolated using streptavidin beads. After SDS-PAGE of total (unfractionated) and biotinylated proteins, followed by Western blot analysis using affinity-purified antibody against GAT-1, the transporter FIG. 8. Effects of sodium on the lithium leak currents of wildtype and R69K mutant GAT-1. Lithium-dependent currents obtained by subtracting the response in sodium at Ϫ160 mV were normalized to the values for the wild type or mutant and recorded at the indicated sodium concentrations. The lithium concentration was 86.4 mM, and choline was used to maintain iso-osmolarity. The data are averages from three (wild type) and four oocytes (R69K). bands were visualized by ECL. In the case of the wild type, major bands of ϳ50 and ϳ55-60 kDa in the total and biotinylated samples were observed, respectively (Fig. 10). The latter apparently represents the fully processed form of the GAT-1 monomer (33). The mobility of the upper band in the biotinylated sample of the wild type was compatible with that of a dimeric aggregate of the transporter. The expression of R69K in the total sample was similar to that of the wild type. On the other hand, much lower levels of the mutant transporter were detected in the biotinylated sample; and besides a small amount of the ϳ55-60-kDa band, bands of lower mobility were detected (Fig. 10). The expression level and pattern of the biotinylated G63C mutant were similar to those of R69K (Fig.  10). Here, two sets of oocytes expressing G63C were processed in parallel to get an impression of the experimental variability. In contrast to R69K and G63C, the expression level and pattern of the Y60T mutant were very similar to those of the wild type (Fig. 10). DISCUSSION Mutation of Gly 63 , located in the conserved transmembrane domain I of GAT-1, to serine or cysteine resulted in defective GABA transport and transport currents (Figs. 1 and 2). The G63S and G63C mutants did not even exhibit sodium-dependent transient currents (data not shown), but exhibited lithium leak currents, which were at least as large as those exhibited by the wild type (Fig. 2). Therefore, these mutants appear to be locked in a conformation that mediates the leak current of GAT-1.
The model depicted in Fig. 11 describes how the leak mode of GAT-1 may relate to its transport cycle. For reasons of simplicity, the role of chloride is not indicated. In step 1, sodium binds to the outward-facing form of the transporter ( o T). This enables subsequent GABA binding (step 2), followed by the translocation step (step 3). Release of sodium and GABA to the inside (step 4) yields the inward-facing form of the unloaded transporter ( i T); and upon reorientation to the outside (step 5), a new transport cycle can commence. The leak mode of the transporter ( ][ T) can be generated from the unloaded outward-and/or inward-facing form of the transporter upon hyperpolarization (Fig. 5). Sodium inhibited the lithium currents mediated by the GAT-1 transporter (Fig. 4), and this is in accordance with a study that was published during the preparation of this manuscript (39). Moreover, sodium also inhibited the potassium leak currents mediated by GAT-1 (Fig. 9). The inhibition of the leak currents by sodium was noncompetitive (Fig. 4). This indicates that sodium binds at a distinct site from the permeating ions, bringing it into a different conformation. Because sodium induces the initial steps of the transport cycle (Fig. 11), a simple explanation is that these critical steps represent the conformational change resulting in the inhibition of the leak cur- FIG. 9. Comparison between lithium and potassium leak currents. The ratio of steady-state potassium-and lithium-induced currents obtained by subtraction of the response in sodium at Ϫ160 mV is plotted for the wild-type (WT), R69K, and Y60T (A). The potassium current in the presence of 2 or 10 mM sodium (white and black bars, respectively) was normalized to that in the absence of sodium (gray bars; choline substitution) at Ϫ160 mV and is depicted for the wild type and the two mutants (B). can be opened in the absence of extracellular sodium by hyperpolarization (⌬V) of the unloaded outward-facing ( o T) and/or inward-facing ( i T) transporter. Binding of extracellular sodium converts the transporter into the sodium-loaded outward-facing form of the transporter (step 1). Subsequent GABA binding (step 2) is followed by translocation (step 3) and release of sodium and GABA to the inside of the cell (step 4). Upon reorientation of the binding sites to the outside and binding of extracellular sodium, the transport cycle is completed (step 5). For simplicity, chloride is omitted from the scheme. rents. Supporting this explanation are observations on mutants at two other positions, also located in transmembrane domain I. A decreased (Fig. 5C) or an increased (Fig. 7A) apparent affinity for sodium was paralleled by the ability of sodium to inhibit the leak currents (Figs. 5D, 8, and 9).
The model depicted in Fig. 11 is also in harmony with the observations that the GABA currents began to saturate at negative potentials, whereas this was not the case for the lithium leak currents (Fig. 5B). The GABA currents reflect electrogenic sodium-coupled GABA translocation involving steps 1-5. Therefore, increasingly negative potentials stimulate the translocation of positive charges that accompany GABA transport. However, at very negative potentials, a voltageindependent step apparently becomes rate-limiting for transport. It has been suggested that this is a voltage-independent interaction between GABA and the transporter (8). The leak currents reflect only the voltage-dependent gating of the unloaded transporter. They do not require steps 1-4, one or perhaps more of which involve voltage-independent interactions.
In choline medium, no inwardly rectifying leak currents were observed. Moreover, the lithium currents did not saturate with increasing concentrations of this cation (Fig. 4). Thus, it appears that only small cations (but not the larger choline) pass through the permeation pathway when this is activated by hyperpolarization.
The role of Gly 63 in GABA transport is as yet unclear. In the related serotonin transporter SERT, an aspartate residue occupies the position equivalent to Gly 63 of GAT-1. This aspartate residue, which is conserved in all biogenic amine transporters of this family, has been implicated in the binding of the amine group of the substrate and also influences the apparent affinity of SERT for the two cosubstrates, sodium and chloride (16). It is tempting to speculate that, in the amino acid transporter members of the family, the carboxyl group of the transported amino acid fulfills a role played by the carboxyl group of the conserved aspartate in the biogenic amine transporters. If, indeed, Gly 63 is part of the binding pocket for the carboxyl group of GABA, it is easy to see that even small changes at this position impair GABA transport. As sodium-dependent transients are also defective in the G63S and G63C mutants, the leak currents mediated by these mutants are their only functional property. The presence of these leak currents indicates that the conformation of the outward-and/or inward-facing unloaded transporter is not affected by the mutations at position 63. Thus, even though the leak currents do not have an obvious physiological function, they can be an important tool for structure/function studies in GAT-1.
The lithium leak currents and the transient currents mediated by R69K are usually smaller than those mediated by the wild type (Figs. 6 and 7). This indicates a lower cellsurface expression of this mutant, and this is, in fact, supported by surface biotinylation experiments (Fig. 10). It is not clear why bands of lower mobility are observed in the biotinylated samples of R69K (and also of G63C) than in those of the wild type. One possibility is that, due to the lower expression levels at the cell surface, during the processing of the samples, these mutant transporters are more prone to proteolysis and subsequent aggregation. Whatever the reason, the important point for this study is that the expression level is high enough to test the hypothesis that an increased apparent affinity of sodium binding to the transporter can be correlated with an increased potency of sodium to inhibit the leak currents (Figs. 7-9). In R69K, GABA has lost the ability to induce a transport current (Fig. 6A) as well as to catalyze [ 3 H]GABA uptake (12). Nevertheless, GABA can bind inefficiently to the R69K transporters, as judged by its ability to partially isolate the transient currents at high concentrations (Fig. 6A). A possible role of Arg 69 could be to interact with the carboxyl group of GABA. This would be reminiscent of the role of Arg 447 of the glutamate transporter EAAC-1 in binding the ␥-carboxyl group of glutamate (40). In the other amino acid transporters, this conserved arginine could play a similar role.
Determinants located in the highly conserved transmembrane domain I influence the affinity of the transporter for sodium (Figs. 5C and 7), the ability of sodium to inhibit the leak currents (Figs. 5D, 8, and 9B), the potassium/lithium selectivity of the leak currents (Fig. 9A), and the interconversion of the leak to transport mode (Figs. 2 and 3). Consistent with the possibility of a direct involvement of at least one amino acid residue of transmembrane domain I in permeation is the fact that the position of Gly 63 located in the middle of this domain appears to be accessible via an aqueous pathway to the extracellular medium (Fig. 3). The simultaneous effects of the mutations on GABA transport and the leak mode suggest that the permeation pathway of these two modes may share common structural elements.