Role of Cl 2 in Electrogenic Na 1 -coupled Cotransporters GAT1 and SGLT1*

We have investigated the functional role of Cl 2 in the human Na 1 /Cl 2 / g -aminobutyric acid (GABA) and Na 1 / glucose cotransporters (GAT1 and SGLT1, respectively) expressed in Xenopus laevis oocytes. Substrate-evoked steady-state inward currents were examined in the presence and absence of external Cl 2 . Replacement of Cl 2 by gluconate or 2-( N -morpholino)ethanesulfonic acid decreased the apparent affinity of GAT1 and SGLT1 for Na 1 and the organic substrate. In the absence of substrate, GAT1 and SGLT1 exhibited charge movements that manifested as pre-steady-state current transients. Removal of Cl 2 shifted the voltage dependence of charge movements to more negative potentials, with apparent affinity constants ( K 0.5 ) for Cl 2 of 21 and 115 m M for SGLT1 and GAT1, respectively. The maximum charge moved and the apparent valence were not altered. GAT1 stoichiometry was determined by measuring GABA-evoked currents and the unidirectional influx of 36 Cl 2 , 22 Na 1 , or [ 3 H]GABA. Uptake of each GABA molecule was accompanied by inward movement of 2 positive charges, which was entirely accounted for by the influx of Na 1 in the presence or absence of Cl 2 . Thus, the GAT1 stoichiometry was 2Na 1 :1GABA.


Expression in Xenopus Oocytes and Experimental Solutions-Stage
V-VI Xenopus laevis oocytes were injected with 25-50 nl (1 g/l) of cRNA encoding human GAT1 (19) or human SGLT1 (20) or with 50 nl of water (control oocytes) and were maintained in Barth's medium (88 mM NaCl, 1 mM KCl, 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 0.82 mM MgSO 4 , 2.4 mM NaHCO 3 , and 10 mM HEPES, pH 7.4) for 3-14 days (11,16). GAT1 used in this study was polymerase chain reaction-amplified from a human brain library, and sequencing confirmed that it is identical to that previously reported (19). In the recording chamber, oocytes were normally bathed in Na ϩ buffer (100 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM HEPES, pH 7.5). Na ϩ -and Cl Ϫ -free solutions were prepared by replacing Na ϩ with choline and Cl Ϫ with gluconate (i.e. gluconate salts of Na ϩ , K ϩ , Mg 2ϩ , and Ca 2ϩ ). Solutions that were both Na ϩ -and Cl Ϫ -free contained choline/gluconate and the gluconate salts of K ϩ , Mg 2ϩ , and Ca 2ϩ . Chloride replacement was also achieved with MES, with no observed difference in transporter function between gluconate and MES replacement. All experiments were performed at 21 Ϯ 1°C.
Electrophysiological Measurements-Transporter kinetics were studied using the 2-electrode voltage-clamp technique (11). In experiments in which the Cl Ϫ concentration was varied, the reference electrode was connected to the experimental oocyte chamber via an agar bridge (3% agar in 3 M KCl). To obtain current-voltage relations, the pulse protocol (pCLAMP, Axon Instruments, Inc., Foster City, CA) consisted of 100-ms (SGLT1) or 400-ms (GAT1) voltage steps from a holding potential of Ϫ50 mV to a series of test voltages (V m ) from ϩ50 to Ϫ150 mV in 20-mV steps. Currents were low pass-filtered at 500 Hz and sampled at 10 kHz (SGLT1) or 2.5 kHz (GAT1). Substrate-induced steady-state cotransporter currents were obtained from the difference between steady-state currents in the absence and presence of substrates (GABA or ␣-methyl- To obtain the transporter pre-steady-state currents, at every V m , the total current (I(t)) was fitted to Equation 2, I͑t͒ ϭ I 1 e Ϫt/1 ϩ I 2 e Ϫt/2 ϩ I ss (Eq. 2) where I 1 is the initial capacitive current with time constant 1 , t is time, I 2 is the initial transporter (GAT1 or SGLT1) transient current with time constant 2 , and I ss is the steady-state current (11). At each V m , the total transporter-mediated charge (Q) was obtained by integrating the transporter transient currents. The charge-voltage (Q-V) relations obtained were then fitted to a single Boltzmann equation (Equation 3), where Q max ϭ Q dep Ϫ Q hyp (Q dep and Q hyp are Q at depolarizing and hyperpolarizing limits, respectively), z␦ is the effective (or apparent) valence of the movable charge, V 0.5 is the membrane potential at 50% charge transfer, F is Faraday's constant, R is the gas constant, and T is the absolute temperature.
Coupling Stoichiometry-The ion/substrate stoichiometry of GAT1 was determined by simultaneous measurement of GABA-induced inward currents and influx of 36 Cl Ϫ , 22 Na ϩ , or [ 3 H]GABA under voltageclamp conditions. The membrane potential was clamped, and after an initial base-line period in Na ϩ buffer, oocytes that were placed in a chamber (volume of 30 l) were superfused (160 l/min) for 0.5-10 min with a buffer containing GABA and 36 Cl Ϫ , 22 Na ϩ , or [ 3 H]GABA. The experimental conditions were selected to optimize tracer counts, taking into account the specific activity of the reagents and the kinetics of the transporter (see Fig. 2). At the end of the period, GABA and the isotope were removed from the external medium, and the current was monitored until it returned to the base line. The oocyte was removed from the chamber, washed in ice-cold choline buffer, and solubilized for liquid scintillation counting (21)(22)(23). The total substrate-evoked inward charge (see Fig. 5) was obtained from the time integral of the substrateevoked current and was correlated with tracer fluxes in the same cells.
Chloride Efflux Assays-Unidirectional chloride efflux was measured either in voltage-clamped oocytes to quantitate the ratio of chloride efflux to inward charge or in groups of 10 -20 oocytes to examine the time course of efflux (see Fig. 7). The procedure for Cl Ϫ efflux assays was modified from that described in the literature (24). Oocytes were incubated in a modified Barth's medium in which the Cl Ϫ specific activity was 40 Ci/ml, i.e. 40 mM of the total NaCl was provided from the stock Na 36 Cl isotope solution (Amersham Pharmacia Biotech). Oocytes were preincubated in a modified Barth's medium for 24 -72 h to maximize the cytoplasmic 36 Cl Ϫ specific activity. Before each experiment, oocytes were removed from the modified Barth's medium, rinsed, and mounted in the experimental chamber. The chamber was perfused with the NaCl buffer, and GABA (500 M) with or without SKF 89976A (500 M) was added to the buffer. For the non-voltage-clamped oocytes (see Fig. 7, A and B), 10 -20 GAT1-expressing oocytes were placed in a linear perfusion chamber. The chamber was perfused at 1 ml/min, and the superfusate was collected at 1-or 2-min intervals. After an initial control period of 36 Cl Ϫ efflux, GABA (500 M) was added to the superfusate, followed by, in some experiments, SKF 89976A (500 M). Cl Ϫ efflux was assessed by scintillation counting of the superfusate fractions. For the voltage-clamped oocytes (Fig. 7C), after establishment of base-line current, GABA was added to the bathing medium for 10 min, and the GABA-induced inward current was recorded. During this period, the entire superfusate was collected and counted as a single sample. At the end of the experiment, oocytes were rinsed and solubilized to determine the total oocyte tracer content. Cl Ϫ efflux is reported as a percentage of total 36 Cl Ϫ count present in the oocyte at the beginning of the experiment. Total counts at the start of the experiment were determined by adding the efflux counts to the oocyte count at the end of the experiment. The magnitude of Cl Ϫ efflux was estimated by assuming an oocyte cytoplasmic volume of ϳ500 nl and a cytoplasmic [Cl Ϫ ] of 33 mM (25). Control oocytes were obtained from the same batch and were treated in exactly the same manner as the GAT1-expressing oocytes.
All experiments were repeated with a minimum of three oocytes from three different donor frogs. Unless otherwise noted, all statistics are expressed as S.E. of the fit.

RESULTS
Steady-state Kinetics of Human GAT1-We first characterized the kinetics of human GAT1, as they have not been previously reported. The current records from an oocyte expressing human GAT1 in response to step jumps in membrane voltage (V m ) from a holding potential of Ϫ50 mV to a series of test potentials are shown in Fig. 1. In the absence of GABA, after the initial capacitive transient (with time constant Ϸ 1 ms), there was a slower decay to the steady state ( Ϸ 40 -150 ms) (Fig. 1A). These pre-steady-state, or capacitive, transients were not observed in control oocytes and were blocked by the GAT1-specific inhibitor SKF 899776A (data not shown) (10). Addition of GABA (500 M) to the bathing medium induced an inward steady-state current and abolished the pre-steady-state current (Fig. 1B). The GABA-evoked current, the difference in steady-state current in the presence and absence of GABA (Fig.  1C), increased with hyperpolarizing membrane voltages and did not saturate at the most negative V m tested (Ϫ150 mV). In the depolarizing direction, the GABA-evoked current tended to asymptote toward zero, but current reversal was not observed even at the most positive V m tested (ϩ50 mV).
The GABA-evoked inward current depended on the external Na ϩ , Cl Ϫ , and GABA concentrations and on the membrane voltage. The dependences of the half-maximal concentrations (K 0.5 ) for GABA (K 0.5 GABA ), Na ϩ (K 0.5 Na ϩ ), and Cl Ϫ (K 0.5 Cl Ϫ ) are shown in Fig. 2 (A-F, white bars). The maximal currents (I max ) for GABA (I max GABA ), Na ϩ (I max Na ϩ ), and Cl Ϫ (I max Cl Ϫ ) are also shown (Fig.  2, A-F, black bars). The data were obtained at V m ϭ Ϫ110 mV, and the reported values are from the fits of the data to Equation 1. To compare data from oocytes exhibiting different levels of GAT1 expression, the maximal currents at Ϫ110 mV have been normalized to those obtained at 500 M GABA in the NaCl buffer ([Na ϩ ] o ϭ 100 mM and [Cl Ϫ ] o ϭ 106 mM), and at V m ϭ Ϫ150 mV (see Fig. 1C). In the NaCl buffer, K 0.5 GABA was 14 Ϯ 1 M. Decreasing either the Na ϩ or Cl Ϫ concentration increased K 0.5 GABA ; K 0.5 GABA increased to 809 Ϯ 54 M at 0 Cl Ϫ o ( Fig. 2A) and to 27 Ϯ 6 M at 5 mM Na ϩ o (Fig. 2B). The maximal transport rate for GABA (I max GABA ) was independent of the Cl Ϫ concentration ( Fig. 2A); however, it decreased with decreasing [Na ϩ ] o (Fig. 2B). At large negative membrane potentials, I max GABA tended to asymptote toward the same maximal value (data not shown), indicating that the absolute I max GABA was inde- The half-maximal concentration for Cl Ϫ (K 0.5 Cl Ϫ ) increased from 0.6 Ϯ 0.2 to 3.8 Ϯ 1.6 mM as external GABA was increased from 5 to 500 M (Fig. 2C). I max Cl Ϫ also increased (Fig. 2C), and this was the case at more negative potentials (data not shown). At 500 M GABA o , external Na ϩ (20 -100 mM) had little effect on K 0. 5 Cl Ϫ (Ϸ3 mM) (Fig. 2D). I max Cl Ϫ increased from 0.45 Ϯ 0.1 to 0.76 Ϯ 0.1 as [Na ϩ ] o was varied from 20 to 100 mM. However, as the membrane potential approached Ϫ150 mV, the relative I max values approached the same value (data not shown), suggesting that the absolute I max Cl Ϫ was independent of [Na ϩ ] o . GABA exerted a small effect on the apparent affinity for Na ϩ ; K 0.5 Na ϩ increased marginally from 14 Ϯ 1 to 20 Ϯ 5 mM as [GABA] o was increased 100-fold from 5 to 500 M (Fig. 2E). Cl Ϫ increased the apparent affinity of GAT1 for Na ϩ ; K 0.5 Na ϩ decreased from 18 Ϯ 2 to 14 Ϯ 1 mM when [Cl Ϫ ] o was increased from 20 to 106 mM (Fig. 2F).
In summary, Na ϩ and Cl Ϫ increased the apparent affinity of GAT1 for GABA. GABA decreased the apparent affinity for Cl Ϫ , whereas Na ϩ had relatively little effect on the apparent affinity of GAT1 for Cl Ϫ . GABA had no effect on the apparent affinity for Na ϩ , but Cl Ϫ increased the apparent affinity for Na ϩ . The absolute I max values (i.e. I max at hyperpolarizing potentials) were apparently independent of the Na ϩ and Cl Ϫ concentrations and depended on the GABA concentration. All in all, the steady-state and pre-steady-state (see below) kinetics of human GAT1 reported here are similar to those of rat GAT1 reported previously (10,13,15,26,27). The data are consistent with ordered reaction schemes in which Na ϩ and Cl Ϫ bind to the transporter before GABA (2,18), as is the case for Na ϩ and glucose in SGLT1 (17,28).
Effect of Cl Ϫ on Steady-state Kinetics of GAT1 and SGLT1-The dependence of steady-state GABA and glucose transport kinetics on external Cl Ϫ is shown in Fig. 3 (A-F). In 106 mM Cl Ϫ o , the steady-state current-voltage curve of the current induced by GABA (500 M) (Fig. 3A) did not show saturation, whereas the current induced by ␣-MDG (500 M) (Fig. 3B) tended toward saturation. Both currents were reduced upon removal of external Cl Ϫ (gluconate or MES substitution). In GAT1, in the absence of Cl Ϫ , there was virtually no GABAcoupled current at V m more positive than Ϸ0 mV, but as V m was made more negative, the GABA-coupled current increased. At Ϫ150 mV, the current was Ϸ50% of that in the presence of 106 mM Cl Ϫ o . In contrast, in SGLT1, the reduction in ␣-MDGinduced current was small and relatively constant at every voltage tested. Cl Ϫ had no effect on the maximal transport rates (I max ) for GAT1 and SGLT1; I max evaluated at the most negative membrane potentials was the same in the presence and absence of external Cl Ϫ (see Fig. 2A for GAT1; data not shown for SGLT1).
Kinetic analyses of GABA-or glucose-evoked currents indicated that there was an increase in the half-maximal concentration (K 0.5 ) for Na ϩ and/or the organic substrate with decreasing [Cl Ϫ ] o (Figs. 2F and 3, C-F). For example, decreasing [Cl Ϫ ] o from 106 to 20 mM increased K 0.5 Na ϩ for GAT1 from 14 Ϯ 2 to 20 Ϯ 2 mM (at Ϫ110 mV) (Fig. 3C) and that for SGLT1 from 41 Ϯ 1 to 62 Ϯ 2 mM (at Ϫ50 mV) (Fig. 3D). There was a large effect of Cl Ϫ on the affinity of GAT1 for GABA, whereas the effect on the apparent affinity of SGLT1 for ␣-MDG was small. At Ϫ50 mV, K 0.5 GABA increased from 11 Ϯ 2 to 104 Ϯ 13 M, whereas in SGLT1, K 0.5 ␣-MDG increased from 760 Ϯ 20 to 860 Ϯ 30 M (Fig. 3, E and F). For both transporters, there was no change in the observed Hill coefficient for Na ϩ (Ϸ2) or the organic substrate (Ϸ1) in the presence or absence of Cl Ϫ (data not shown). In GAT1, the Hill coefficient for Cl Ϫ was Ϸ1.
Effect of Cl Ϫ on Pre-steady-state Kinetics of GAT1 and SGLT1-For both GAT1 (as illustrated in Fig. 1A) and SGLT1, in the absence of the organic substrate, step jumps in mem-FIG. 1. GAT1 pre-steady-state and steady-state currents. A, in the absence of GABA, oocytes expressing GAT1 exhibited pre-steadystate currents in response to step changes in the membrane voltage. This is illustrated for an oocyte expressing human GAT1 with the membrane held at Ϫ50 mV and stepped to a series of test values (ϩ50, ϩ10, Ϫ10, Ϫ50, Ϫ70, Ϫ110, and Ϫ150 mV). The time constant of the relaxation of the pre-steady-state currents ranged from 40 to 150 ms for the onset of the pulse (on response) and was Ϸ150 ms for the offresponse as the membrane potential was returned to the holding value. B, in the presence of 500 M GABA, pre-steady-state currents disappeared, and in addition, there was a GABA-evoked steady-state inward current. The broken lines to the left of the current traces in A and B denote the zero current level. A and B share the same current and time scales. C, at steady-state, the difference between the current traces in the absence and presence of GABA yielded the net GABA-evoked inward current.
brane voltage elicited pre-steady-state capacitive transients or charge movements (10,11). Fig. 4 (A and B) shows the chargevoltage (Q-V) relations in the presence and absence of external Cl Ϫ . For both transporters, removal of Cl Ϫ from the external medium shifted the V 0.5 (voltage at 50% Q max ) toward more negative potentials (Fig. 4, A and B). The shift in V 0.5 (⌬V 0.5 ) was Ϫ18 Ϯ 3 mV for SGLT1 and Ϫ76 Ϯ 4 mV for GAT1. The shift in V 0.5 can be used to estimate an apparent affinity of the transporter for external Cl Ϫ (Fig. 4, C and D). The apparent affinity for Cl Ϫ was 5-fold greater in SGLT1 than in GAT1 (K 0.5 for Cl Ϫ was 21 Ϯ 1 mM versus 115 Ϯ 18 mM). Chloride removal had no effect on Q max (Fig. 4, E and F) or the apparent valence of the transporter (z Ϸ 1 for both transporters; data not shown).
GAT1 Stoichiometry-To determine the ion/substrate stoichiometry of GAT1, influxes of 22 Na ϩ , 36 Cl Ϫ , and [ 3 H]GABA were measured under voltage-clamp conditions (V m ϭ Ϫ100 mV) (Fig. 5A). At the end of the recording period, the inward charge (Q) and the amount of [ 3 H]GABA, 22 Na ϩ , and 36 Cl Ϫ influx were determined. The slope of the inward charge versus 36 Cl Ϫ influx was 2.0 Ϯ 0.1 charges/Cl Ϫ ion (Fig. 5B). Thus, Cl Ϫ is transported by GAT1, with the net inward movement of 2 positive charges coupled to the influx of each Cl Ϫ ion. The inward charge per [ 3 H]GABA molecule was 2:1 (slope was 1.9 Ϯ 0.1 charges/GABA molecule) and was the same in the presence and absence of Cl Ϫ in the external medium (Fig. 5C). The one-to-one correlation between the inward charge and 22 Na ϩ uptake indicated that each inward charge was accompanied by the influx of one 22 Na ϩ ion (slope was 0.9 Ϯ 0.1 charge/Na ϩ ion) and independent of the presence or absence of external Cl Ϫ (Fig. 5D). Experiments performed with the membrane potential held at Ϫ10, Ϫ30, Ϫ50, and Ϫ70 mV yielded similar results (data not shown).
To determine whether Cl Ϫ was transported by SGLT1, Na ϩ / ␣-MDG currents were evoked for 10 min in the presence of 36 Cl Ϫ (Fig. 6A). There was no difference in 36 Cl Ϫ uptake (4.3 pmol/min) between control oocytes and SGLT1-expressing oocytes (p Ͼ 0.05) (Fig. 6B), indicating that Cl Ϫ is not transported by SGLT1. The inward current in these experiments (Fig. 6A) ranged from 540 to 950 nA, and if one Cl Ϫ ion were transported per inward charge, the Cl Ϫ influx would be expected to be 340 -590 pmol/min, well within the resolution of our experiments.  GAT1 Chloride Efflux-Since the net inward charge is totally accounted for by the influx of Na ϩ , Cl Ϫ influx appears not to contribute to the GABA-evoked inward current. One explanation is that there is an exchange of an internal anion for Cl Ϫ that is transported into the cell, and the most likely candidate is intracellular Cl Ϫ . To investigate this possibility, we monitored the efflux of 36 Cl Ϫ in oocytes preloaded with 36 Cl Ϫ . In the absence of GABA, Cl Ϫ efflux from GAT1-expressing oocytes was similar to that from control cells. In GAT1-expressing oocytes, activation of forward Na ϩ /Cl Ϫ /GABA transport (addition of 500 M GABA to the NaCl buffer) increased the rate of 36 Cl Ϫ efflux from 0.2 to 2%/min (Fig. 7A). The GABA-stimulated Cl Ϫ efflux was blocked by the GAT1 inhibitor SKF 89976A (500 M), as addition of this compound to the external solution reduced the rate of 36 Cl Ϫ efflux to control values (Fig. 7B).
In an effort to gain a more quantitative measure of Cl Ϫ efflux, we monitored Cl Ϫ efflux in individual voltage-clamped oocytes. In these experiments, oocytes that had been preloaded with 36 Cl Ϫ were voltage-clamped; forward Na ϩ /Cl Ϫ /GABA transport was activated for 10 min as was done in the experiments of Fig. 5A; and the inward current was monitored. As The maximal substrate-evoked current was obtained in the presence of both Na ϩ (100 mM) and Cl Ϫ (106 mM). In both GAT1 and SGLT1, the steady-state substrate-evoked current was reduced in the absence of Cl Ϫ (at 100 mM Na ϩ ). C and D, for both GAT1 and SGLT1, respectively, a reduction in [Cl Ϫ ] o (E) led to a lower apparent affinity for Na ϩ (K 0.5 Na ϩ ) than seen in the presence of 106 mM Cl Ϫ (q). E and F, although in the absence of Cl Ϫ (E), there was a dramatic reduction in the apparent affinity of GAT1 for GABA (K 0. was the case in the non-voltage-clamped GAT1-expressing oocytes (Fig. 7, A and B), there was a significant increase in Cl Ϫ efflux as forward Na ϩ /Cl Ϫ /GABA transport was activated (p Ͻ 0.05) (Fig. 7C). Assuming that the internal Cl Ϫ concentration is 33 mM and the cytoplasmic volume of the oocyte is 500 nl (25), the ratio of GABA-evoked inward charge to Cl Ϫ efflux was 2.2 Ϯ 1.0 charges/exported Cl Ϫ ion. DISCUSSION Extracellular Cl Ϫ alters the activity of the brain GABA transporter (GAT1) and the intestinal glucose transporter (SGLT1). A reduction in [Cl Ϫ ] o decreases the steady-state substrate (GABA and glucose)-evoked inward currents of these transporters by reducing their affinity for external Na ϩ and the organic substrate. However, Cl Ϫ is not absolutely required for Na ϩ -dependent substrate transport by either transporter. For Na ϩ -dependent glucose transport (SGLT1), it has been generally believed that transport is insensitive to Cl Ϫ . For GAT1, the sensitivity to external Cl Ϫ is widely accepted (2,29). Our findings that Cl Ϫ is not absolutely required and that negative membrane potentials could substitute for Cl Ϫ agree with uptake studies in membrane vesicles (30) and electrophysiological experiments on GAT1 expressed in Xenopus oocytes (10,15). Here, we add the novel observation that in GAT1, the transport stoichiometry of 2Na ϩ :1GABA is independent of the extracellular Cl Ϫ concentration. Surprisingly, whether external Cl Ϫ is present or absent, 2 net positive charges enter the cell during the GAT1 transport cycle. This is in contrast to the widely held view that the GAT transport cycle introduces only 1 net positive charge into the cell. In addition, we show that forward Na ϩ /Cl Ϫ /GABA transport is accompanied by simultaneous GAT1-mediated Cl Ϫ efflux.
Effect of Cl Ϫ on Steady-state and Pre-steady-state Kinetics-The effect of Cl Ϫ on SGLT1 and GAT1 is to increase the apparent affinity of the transporters for Na ϩ and the substrate FIG . 5. Stoichiometry of GAT1. A, time course of the inward current evoked by GABA at a holding potential of Ϫ100 mV. 22 Na ϩ , 36 Cl Ϫ , and [ 3 H]GABA were superfused simultaneously to correlate ligand influx and transporter currents in the same cell. B, relationship between the GABA-evoked GAT1 inward charge (expressed in nanomoles) and Cl Ϫ influx. The bathing medium contained 100 mM Na ϩ , 3.84 mM 36 Cl Ϫ , and 5 mM GABA. The coupling ratio was 2.0 Ϯ 0.1 charges/Cl Ϫ ion. C, relationship between the GABA-evoked GAT1 inward charge and GABA influx. The coupling ratio was 1.9 Ϯ 0.1 charges/GABA molecule in the absence (E) and presence (q) of external Cl Ϫ . The bathing medium contained 500 M GABA, 100 mM Na ϩ , and 106 mM Cl Ϫ (q) or zero Cl Ϫ (E). D, relationship between the GABA-evoked GAT1 inward charge and Na ϩ influx. The coupling ratio was 0.9 Ϯ 0.1 charges/Na ϩ in the absence (E) and presence (q) of external Cl Ϫ . The bathing medium contained 10 mM 22 Na ϩ , 106 mM Cl Ϫ , and 1 mM GABA (q) or 20 mM 22 Na ϩ , zero Cl Ϫ , and 10 mM GABA (E). Experiments performed with membrane potentials held at Ϫ10, Ϫ30, Ϫ50, and Ϫ70 mV yielded similar results.
FIG. 6. Cl ؊ is not transported by SGLT1. A, shown is the time course of the SGLT1 inward current evoked by ␣-MDG at a holding potential of Ϫ50 mV. 36 Cl Ϫ was superfused simultaneously to correlate Cl Ϫ influx and transporter currents in the same cell. B, despite very high Na ϩ /sugar transport rates, there was no Cl Ϫ influx through SGLT1. The experiment was performed on five control and seven SGLT1-expressing oocytes. In control oocytes, addition of GABA or ␣-MDG did not lead to a detectable inward current (data not shown).
(glucose and GABA), whereas the maximum transport velocity is relatively unaffected by Cl Ϫ . Because the affinities for the organic substrates are dependent on the affinity for Na ϩ , the reduction in the glucose and GABA affinities is, at least in part, secondary to the decrease in Na ϩ affinity.
We examined GAT1 and SGLT1 pre-steady-state currents to gain a better understanding of the nature of Cl Ϫ interaction with these proteins in the absence of GABA and glucose. Presteady-state currents have been investigated in a variety of Na ϩ -coupled transporters, including the Na ϩ /glucose (11,12), Na ϩ /Cl Ϫ /GABA (10,13), Na ϩ /iodide (21), Na ϩ /phosphate (31,32), and Na ϩ /K ϩ /glutamate (33) cotransporters. They are thought to represent a redistribution of protein conformations in response to rapid changes in membrane voltage and/or ligand concentration (9 -11, 28, 34, 35). When voltage perturbations are used, transporter pre-steady-state currents can be represented in the form of charge-voltage (Q-V) relations (10,11).
Q-V curves are characterized by three Boltzmann factors: Q max (maximal charge movement), z␦ (product of the effective valence of the moveable charge and the fraction of the electric field through which the charge moves), and V 0.5 (membrane voltage at 50% Q max ). These curves represent the distribution of a charged species (with valence z), which translocates across a given fraction (␦) of the membrane electric field, thereby distributing between two predominant conformational states depending on the membrane voltage (11,12). Q max is directly related to the total number of functional transporter molecules in the plasma membrane (36,37). V 0.5 represents the voltage at which approximately half of the moveable charge is in either conformational state. In SGLT1, the Q-V relation has been shown to provide a measure of the accessibility of the ligandbinding sites, as there is a direct correspondence between the Q-V curve and the accessibility of alkylating reagents to a cysteine residue (Q457C) placed in the putative sugar-binding/ translocation domain of SGLT1: increasing [Na ϩ ] o shifted V 0.5 to more negative membrane potentials and increased the accessibility of the sugar-binding site from the external medium (11,28). In both SGLT1 and GAT1, a 10-fold reduction in external Na ϩ concentration results in a similar change in occupancy probability: the shift in V 0.5 was Ϫ100 mV for both transporters (10 -12). 2 Since varying [Cl Ϫ ] o had no effect on Q max or z␦ in either GAT1 or SGLT1, the reduction in the rate of substrate transport upon Cl Ϫ removal is not due to a decrease in the number of transporters at the cell surface or to a decrease in the effective valence of each transporter. The shift of V 0.5 toward more negative membrane potentials as [Cl Ϫ ] o was reduced is similar to that seen with a reduction in [Na ϩ ] o . Thus, in GAT1 and SGLT1, Cl Ϫ appears to alter the occupancy probability in a similar way to Na ϩ . From Fig. 4 (A and B), we found that the probability of being in the substrate-binding conformation is decreased from 0.6 to 0.4 for SGLT1 and from 0.8 to 0.3 for GAT1 upon removal of Cl Ϫ from the external medium (at Ϫ50 mV). The larger shift in V 0.5 seen for GAT1 compared with that seen for SGLT1 is responsible for the greater effect of Cl Ϫ on the steady-state transport rates of GAT1 (Fig. 3, A and B). Although the effect of Cl Ϫ on the rate of steady-state substrate transport by GAT1 was much larger than that seen for SGLT1, it should be noted that weaker effects of chloride have been noted for other members of the GABA transporter family even though they exhibit 40 -50% identity to GAT1 (39).
GAT1 Stoichiometry-Previous studies on the stoichiometry of Na ϩ /Cl Ϫ /GABA cotransport have been based on three methods: 1) kinetic analysis of the dependence of cotransport rates on the substrates in oocytes expressing GAT1 (10,26), 2) thermodynamic analysis of the dependence of the reversal potentials of cotransport in membrane vesicles (40,41) and in cells expressing GAT1 (15,27), and 3) comparison of the ratios of substrate fluxes in membrane vesicles (42) and comparison of the inward charge and substrate fluxes in oocytes expressing GAT1 (26). With the exception of the study of Pastuszko et al. (40), which determined the transport of two Na ϩ ions with one zwitterionic GABA molecule (independent of Cl Ϫ ), all other studies determined a 2Na ϩ :1Cl Ϫ :1GABA stoichiometry. The stoichiometry for GABA influx obtained in this study is consistent with this ratio (2Na ϩ :1Cl Ϫ :1GABA). However, our surprising finding is that in the presence or absence of external Cl Ϫ , there were 2 inward charges transported with each GABA FIG. 7. Forward Na ؉ /Cl ؊ /GABA transport induces simultaneous Cl ؊ efflux through GAT1. A, shown is the time course of 36 Cl Ϫ efflux through GAT1. The ordinate is the Cl Ϫ efflux rate, expressed as a percentage of the total 36 Cl Ϫ counts present in the oocyte in the beginning of the experiment. Control (E) and GAT1-expressing (q) oocytes were preloaded with 36 Cl Ϫ by incubation for 24 -72 h in Barth's medium containing 36 Cl Ϫ . Oocytes (20 cells) were rinsed, and 36 Cl Ϫ efflux was monitored at 1-min intervals. At the time indicated by the bar, 500 M GABA was added to the external medium to activate forward Na ϩ /Cl Ϫ /GABA transport. In the absence of GABA, Cl Ϫ efflux rate was similar in both control and GAT1-expressing oocytes (0.2%/ min). GABA stimulated a large increase in Cl Ϫ efflux in the GAT1expressing oocytes (2%/min). B, the GABA-evoked 36 Cl Ϫ efflux in GAT1expressing cells was blocked by the GAT1-specific inhibitor SKF 89976A (500 M). Note that in both A and B, 36 Cl Ϫ efflux was instantaneous with the addition of GABA. C, a similar experiment was performed on control (n ϭ 4) and GAT1-expressing (n ϭ 7) oocytes under voltage-clamp conditions (V m ϭ Ϫ50 mV). The GABA-evoked inward current was monitored and correlated with Cl Ϫ efflux in the same cells. Control oocytes showed no GABA-evoked inward current (data not shown). 36 Cl Ϫ efflux values in control oocytes and GAT1-expressing cells in the absence of GABA were similar. In GAT1-expressing oocytes, activation of Na ϩ /Cl Ϫ /GABA transport led to a significant increase in 36 Cl Ϫ efflux. The coupling ratio was estimated to be 2 Ϯ 1 inward charges for every exported Cl Ϫ ion (similar to the coupling ratio obtained for Cl Ϫ influx). Thus, there appears to be no net Cl Ϫ flux during the transport cycle of GAT1. This is indicative of Cl Ϫ /Cl Ϫ exchange during the transport cycle of GAT1. molecule, and the inward charge was completely accounted for by Na ϩ influx. This is contrary to previous conclusions of 1 inward charge/GABA molecule. In addition, since we found that the Na ϩ /GABA stoichiometry was not altered by Cl Ϫ in the extracellular solution, this implies that Cl Ϫ may not play a role in the thermodynamics of the cotransport process (see below).
GAT1 Cl Ϫ /Cl Ϫ Exchange-GABA stimulated an efflux of Cl Ϫ through GAT1 (Fig. 7). We estimate that the Cl Ϫ efflux is comparable to the Cl Ϫ influx, leading to stoichiometric Cl Ϫ /Cl Ϫ exchange during the transport cycle. Therefore, there is no net Cl Ϫ flux through GAT1. Based on this observation and the finding that the Na ϩ /GABA coupling ratio is independent of Cl Ϫ , it may be inferred that Cl Ϫ plays no role in the thermodynamics of the cotransport process. However, it is difficult to reconcile this direct result with several other reports suggesting a thermodynamic component to Cl Ϫ transport (15,27,41,42). GAT1 has been shown to be able to operate in the reverse mode (15,27,43). This raises the possibility that the observed GABA-stimulated Cl Ϫ efflux is the result of the cotransporter operating in the reverse mode due to GABA accumulated in the cell. Several observations suggest that this is an unlikely scenario. (i) The GABA-stimulated Cl Ϫ efflux was concomitant (1-min sampling resolution) with activation of inward Na ϩ /Cl Ϫ / GABA transport (Fig. 7, A and B), and in view of the unfavorable Na ϩ and voltage gradients, transport in the reverse mode seems unlikely. (ii) An accumulation of GABA should alter the direction of Na ϩ /Cl Ϫ /GABA cotransport. However, after exposing GAT1-expressing oocytes to 500 M GABA for up to 10 min, we did not observe outward currents generated by GAT1. (iii) Assuming that the efflux is caused by accumulation of GABA in the cell, it is expected that it should continue (or even be stimulated) after removal of external GABA. This is not the case, however, as Cl Ϫ efflux ceases immediately after removal of external GABA or addition of the inhibitor SKF 89976A (Fig.  7C). Thus, the observed Cl Ϫ efflux is unlikely to be caused by accumulation of intracellular GABA leading to GAT1 operating in the reverse mode.
Despite overall similarities in external Cl Ϫ sensitivity, there was a major difference between GAT1 and SGLT1. In SGLT1, despite strong interaction with the transporter (as revealed by the dependence of the Q-V relation on Cl Ϫ ), there is no Cl Ϫ flux through the protein. However, in GAT1, for every Cl Ϫ ion transported into the cell, there appears to be an efflux of one Cl Ϫ ion from the cytoplasm mediated by the transporter (Fig. 8).
Conclusion-In summary, our studies show that Cl Ϫ modulates the function of both the Na ϩ -dependent cotransporter SGLT1 and the Na ϩ /Cl Ϫ -dependent cotransporter GAT1. Cl Ϫ dependence is not a good criterion to classify transporter families. This is also emphasized by the recent report on the cloning and characterization of the high affinity choline transporter from rat spinal chord (38). This Na ϩ -and Cl Ϫ -dependent neurotransmitter transporter is a member of the SGLT1 gene family. In both SGLT1 and GAT1, extracellular Cl Ϫ appears to alter the conformational state of the transporter such that it poises the transporter to bind external Na ϩ . The effect is much more pronounced for GAT1. GAT1 appears to mediate Cl Ϫ /Cl Ϫ exchange during the transport cycle, but SGLT1 does not translocate Cl Ϫ across the membrane. All in all, these results point to an allosteric role for Cl Ϫ in Na ϩ -coupled transporters. These experimental observations provide a novel insight into the mechanistic similarity of these transport molecules despite their phylogenetic distance.
FIG. 8. Proposed model for the functional role of Cl ؊ in GAT1 and SGLT1. During every transport cycle of GAT1, two Na ϩ ions, one Cl Ϫ ion, and one GABA molecule are transported across the plasma membrane. Completion of the cycle brings about the efflux of one cytosolic Cl Ϫ ion. Thus, there is no net flux of Cl Ϫ . According to this model, Cl Ϫ is not involved in thermodynamic coupling of Na ϩ and GABA cotransport. The transport cycle of SGLT1 involves the influx of two Na ϩ ions and one glucose molecule. Cl Ϫ binds to SGLT1 and alters transport kinetics, but it is not transported across the plasma membrane. In both proteins, Cl Ϫ facilitates cotransport by favoring a conformation that binds external Na ϩ .