A Conserved Aspartate Residue Located at the Extracellular End of the Binding Pocket Controls Cation Interactions in Brain Glutamate Transporters*

Background: The cation binding sites of glutamate transporters are not solved. Results: Mutation of a conserved aspartate residue alters apparent cation affinity. Conclusion: This aspartate residue may be part of a novel cation binding site. Significance: Knowledge of the cation binding sites is needed to understand the transport mechanism. In the brain, transporters of the major excitatory neurotransmitter glutamate remove their substrate from the synaptic cleft to allow optimal glutamatergic neurotransmission. Their transport cycle consists of two sequential translocation steps, namely cotransport of glutamic acid with three Na+ ions, followed by countertransport of K+. Recent studies, based on several crystal structures of the archeal homologue GltPh, indicate that glutamate translocation occurs by an elevator-like mechanism. The resolution of these structures was not sufficiently high to unambiguously identify the sites of Na+ binding, but functional and computational studies suggest some candidate sites. In the GltPh structure, a conserved aspartate residue (Asp-390) is located adjacent to a conserved tyrosine residue, previously shown to be a molecular determinant of ion selectivity in the brain glutamate transporter GLT-1. In this study, we characterize mutants of Asp-440 of the neuronal transporter EAAC1, which is the counterpart of Asp-390 of GltPh. Except for substitution by glutamate, this residue is functionally irreplaceable. Using biochemical and electrophysiological approaches, we conclude that although D440E is intrinsically capable of net flux, this mutant behaves as an exchanger under physiological conditions, due to increased and decreased apparent affinities for Na+ and K+, respectively. Our present and previous data are compatible with the idea that the conserved tyrosine and aspartate residues, located at the external end of the binding pocket, may serve as a transient or stable cation binding site in the glutamate transporters.

Glutamate is the major excitatory neurotransmitter in the brain. The synaptic actions of this neurotransmitter are terminated by glutamate transporters, which move the transmitter away from the synapse and back into the cells surrounding the synapse and keep its synaptic concentrations below neurotoxic levels. Glutamate transport is an electrogenic process, (1)(2)(3), which consists of two half-cycles (4 -6): (i) cotransport of the neurotransmitter with sodium and hydrogen ions (1,7) and (ii) countertransport of potassium (1,4). The stoichiometry is three sodium ions, one proton, and one potassium ion per transported glutamate molecule (8,9). The sequential translocation mechanism (Fig. 1A) is supported by the fact that mutants impaired in the interaction with potassium are "locked" in an obligatory exchange mode (6,10). Glutamate transporters mediate two types of substrate-induced steadystate current: an inward-rectifying or "coupled" current, reflecting electrogenic ion-coupled glutamate translocation, and an "uncoupled" sodium-dependent current, which is carried by chloride ions and further activated by the substrates of the transporter (11)(12)(13). Nontransportable substrate analogues, expected to "lock" the transporter in an outward-facing conformation when applied from the external side (Fig. 1A, dashed line), are not only competitive inhibitors of the coupled current and of the substrate-induced uncoupled anion current, but also inhibit the basal sodium-dependent anion conductance (14,15). Moreover, such analogues inhibit the sodiumdependent transient currents (3), which are thought to reflect a charge-moving conformational change in response to sodium binding.
The publication of a high-resolution crystal structure of a glutamate transporter homologue, Glt Ph , from the archeon Pyrococcus horikoshii represented a landmark for the field of glutamate transporters (16). The structure shows a trimer with a permeation pathway through each of the monomers, indicating that the monomer is the functional unit. This is also the case for the eukaryotic glutamate transporters (17)(18)(19)(20). The membrane topology of the monomer (16) is quite unusual but is in excellent agreement with the topology inferred from biochemical studies (21)(22)(23). The monomer contains eight transmembrane domains (TM) 2 and two oppositely oriented reentrant loops, one between domains 6 and 7 (HP1) and the other between domains 7 and 8 (HP2) (Fig. 1B). TMs 1-6 form the outer shell of the transporter monomer, whereas TMs 7 and 8 and the two reentrant loops participate in the formation of the binding pocket of Glt Ph (16,24). Importantly, many of the amino acid residues of the transporter, inferred to be important in the interaction with sodium (25,26), potassium (6,10), and glutamate (27,28) are facing toward the binding pocket. Recent studies indicate that glutamate translocation occurs by an "elevator-like" mechanism (29,30) where the transport domain, which includes HP1 and HP2 and TMs 3, 6, 7, and 8, moves relative to the fixed trimerization domain (31).
Because of the limited resolution of the Glt Ph structure, Tl ϩ ions, which exhibit a robust anomalous scattering signal, have been used in an attempt to visualize the sodium sites in this homologue (24), which also uses 3 Na ϩ ions per transported substrate molecule (32). Two Tl ϩ sites were identified. One was found to be buried just under HP2 and seemed to have only four coordinating main chain carbonyl oxygens from TM7 and HP2. The other Tl ϩ site was buried deeply within the protein, coordinated by three main chain carbonyl oxygens from TM7 and TM8 as well as the two carboxyl oxygens of a conserved TM8 aspartate residue (Asp-405) and possibly a hydroxyl oxygen of a HP1 serine residue (24). There is nevertheless, uncertainty on the assumption that Tl ϩ faithfully reports on Na ϩ because in contrast to Na ϩ , Tl ϩ could not support transport (24). However, functional evidence supports the role of one of the Tl ϩ sites as a sodium binding site (33). In the absence of high resolution structural data, suggestions for additional sodium binding sites have been searched by using a combination of computational and functional studies (34 -36). It has been proposed that one of these sites (34) represents a transient site (35). This suggests the possibility that the number of sodium sites may be even higher than three. In the past, we have obtained functional evidence that the conserved tyrosine residue, corresponding to Tyr-317 of Glt Ph , represents a molecular determinant of ion selectivity in the brain glutamate transporter GLT-1 (10). In the Glt Ph structure, a conserved aspartate residue (Asp-390) is located adjacent to the conserved tyrosine residue (16) (Fig.  1B), suggesting the possibility that these two residues could participate together in the formation of a new and as yet unidentified cation binding site. In this study, we characterize mutants of Asp-440 of the neuronal transporter EAAC1, which is the counterpart of Asp-390 of Glt Ph . We found that only one of the substitution mutants, EAAC1-D440E, exhibited transport activity. The functional characteristics of this mutant are consistent with the possibility that the conserved aspartate and tyrosine residues could be involved in the formation of a novel cation binding site.

EXPERIMENTAL PROCEDURES
Generation and Subcloning of Mutants-The C-terminal histidine-tagged versions of rabbit EAAC1 (37,38) in the vector pBluescript SK Ϫ (Stratagene) was used as a parent for site-directed mutagenesis (39,40). This was followed by subcloning the mutations into the His-tagged EAAC1, residing in the oocyte expression vector pOG 1 (38), using the unique restriction enzymes NsiI and StuI. The subcloned DNA fragments were sequenced between these restriction sites.
Cell Growth and Expression-HeLa cells were cultured (41), infected with the recombinant vaccinia/T7 virus vTF 7-3 (42), and transfected with the plasmid DNA harboring the WT or mutant constructs or with the plasmid vectors alone (41). Transport of D-[ 3 H]aspartate or other radiolabeled substrates was done as described (39). Briefly, HeLa cells were plated on 24-well plates and washed with transport medium containing 150 mM NaCl, 5 mM potassium P i , pH 7.4. Each well was then incubated with 200 l of transport medium supplemented with A, transport cycle. The order of binding of the three sodium ions, glutamate, and the proton is not indicated. After glutamate (or other substrates such as D-and L-aspartate) and the co-ions bind to the transporter from the external medium (out) (step 1), they are translocated (step 2) and released into the inside of the cell (in)(step 3). Subsequently potassium binds from the intracellular side (step 4) and, after translocation to the outside (step 5), is released there (step 6). After completion of this second half-cycle, a new translocation cycle can commence. The steps in this scheme are reversible and therefore elevated levels of extracellular potassium can cause release of intracellular glutamate. If the interaction with potassium is abolished by mutation, the second half-cycle is not operative, but the transporters can still exchange labeled glutamate (or aspartate), added to the outside, with internal substrate by reversible translocation via the first half-cycle (steps 1-3). When a bulky nontransportable analogue (blocker) binds from the extracellular medium together with sodium (the stoichiometry (n) is unknown), the transporter remains locked in the outward-facing conformation (dashed line) because translocation cannot proceed. B, position of the conserved Tyr-Asp pair, Asp-390 and Tyr-317, in a model of the main structural elements of the transport domain of Glt Ph . Shown are reentrant loop HP1, TM7, reentrant loop HP2, and TM8 with the colors gradually changing from blue, green, yellow, and brown with the extracellular side at the top and the intracellular side at the bottom. The position of the L-aspartate substrate (Asp) marks the center of the binding pocket. Asp-390 of Glt Ph corresponds to Asp-440 of EAAC1 and Tyr-317 to Tyr-373. The lithium-bound native state of Glt Ph , PDB accession code 2NWL, was used to make the figure, using PyMOL software. 0.4 Ci of the tritium-labeled substrates for 10 min, followed by washing, solubilization of the cells with SDS, and scintillation counting. Solubilization of transporters expressed in the HeLa cells, their reconstitution in proteoliposomes, and transport experiments were done as described (6,43). Briefly, 10 l of proteoliposomes were diluted into 360 l of 150 mM NaCl, supplemented with 1 Ci of D-[ 3 H]aspartate (11.3 Ci/mmol) or L-[ 3 H]aspartate (12.9 Ci/mmol) and 2.5 M valinomycin for each triplicate time point. For the experiments depicted in Figs. 7 and 8, the amount of radioactive substrate was 4 Ci. For the determination of the kinetic parameters for D-and L-aspartate net flux (1 min) the substrate concentration was varied between 0.15 and 6 M using 0.6 or 4.0 Ci of radioactivity for substrate concentrations below or above 0.5 M, respectively. Exchange was measured exactly as described (6) and the data are presented as net exchange after subtracting the values obtained on proteoliposomes containing only 0.12 M sodium P i , pH 7.4, from those containing 0.12 M sodium P i ϩ 10 mM L-aspartate. Protein was determined by the Lowry method (44). Experiments were done at least three times.
Expression in Oocytes, Electrophysiology, Uptake, and Cell Surface Biotinylation-cRNA was transcribed using mMES-SAGE-mMACHINE (Ambion), injected into Xenopus laevis oocytes, which were maintained as described (26). Oocytes were placed in the recording chamber, penetrated with two agarose-cushioned micropipettes (1%/2 M KCl, resistance varied between 0.5 and 3 m⍀), voltage clamped using GeneClamp 500 (Axon Instruments), and digitized using Digidata 1322 (Axon Instruments) both controlled by the pClamp9.0 suite (Axon Instruments). Voltage jumping was performed using a conventional two-electrode voltage clamp as described previously (38). The standard buffer, termed ND96, was composed of 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM Na-HEPES, pH 7.5. The compositions of other perfusion solutions are indicated in the figure legends. Offset voltages in chloride substitution experiments were avoided by use of an agarose bridge (1%/2 M KCl) that connected the recording chamber to the Ag/AgCl ground electrode. For uptake, four to five oocytes of each mutant were incubated for 20 min in ND96 containing D-[ 3 H]aspartate as described previously (26). Cell surface biotinylation was done as described previously (36). Briefly, 5 oocytes expressing wild type or mutant EAAC1 were treated with 1.5 mg/ml of sulfosuccinimidyl-2-(biotinamide)ethyl-1,3dithiopropionate (Pierce) dissolved in ND96 and the streptavidin beads were eluted with a final volume of 70 l of SDS-PAGE sample buffer. For samples of total cell transporter, 10% of the lysate, before the streptavidin treatment, was run on the same gel as samples eluted from the beads. The Western blots were probed with an affinity purified antibody directed against rabbit EAAT3 (generously provided by N. C. Danbolt, University of Oslo; anti-C491 (Ab,371 (45)).
Expression and Laser-pulse Photolysis and Rapid Solution Exchange-Rat EAAC1 cloned from rat retina (15,46) was subcloned into the vector pBK-CMV (Stratagene) and used for transient transfection of subconfluent human embryonic kidney cell (HEK293; ATCC number CGL 1573) cultures with the calcium phosphate-mediated transfection method, as described (47). Electrophysiological recordings were per-formed 24 h after the transfection for 2 days. The rapid solution exchange (time resolution ϳ100 -200 ms) was performed by means of a quartz tube (opening 350 m) positioned at a distance of Ϸ0.5 mm to the cell. The linear flow rate of the solutions emerging from the opening of the tube was ϳ5-10 cm/s. Laser-pulse photolysis experiments were performed as described previously (15). 2 mM 4-methoxy-7-nitroindolinylcaged glutamate (TOCRIS) was applied to the cells and photolysis of the caged glutamate was initiated with a light flash (355 nm Nd:YAG laser, Minilite series, Continuum). The light was coupled into a quartz fiber (diameter 350 m) that was positioned in front of the cell at a distance of 300 m. Laser energies were varied in the 50 -300 mJ/cm 2 range with neutral density filters. With maximum light intensities of 300 mJ/cm 2 , saturating glutamate concentrations could be released, which was tested by comparison of the steady-state current with that generated by rapid perfusion of the same cell with 1 mM glutamate. Data were recorded using the pClamp6 software (Axon Instruments), digitized with a sampling rate of 25 kHz and low-pass filtered at 3-10 kHz.
Data Analysis-All current-voltage relationships represent steady-state substrate-elicited net currents ((I bufferϩsubstrate ) Ϫ (I buffer )) and were analyzed by Clampfit version 8.2 or 9.0 (Axon Instruments), and the data have been normalized as indicated in the figure legends. Charge movements were quantitated by integrating the current-time relationships using nonsubtracted or subtracted current records as indicated in the figure legends. Kinetic parameters were determined by nonlinear fitting to the generalized Hill equation using the built-in functions of Origin 6.1 (Microcal). For determination of the apparent affinity for substrate, I max and K 0.5 were allowed to vary and the value of n H was fixed at 1. Origin software was also used to fit the transport current responses to the photolytic release of glutamate. The time constants were obtained by a two-exponential fit of the decaying phases.

RESULTS
Transport Activity of Asp-440 Mutants-To test the role of the conserved aspartate residue located at position 440 in EAAC1, we first measured D-[ 3 H]aspartate uptake in HeLa cells expressing mutants where this residue was replaced by glutamate, asparagine, glutamine, serine, or cysteine. As shown in Fig. 2A, none of the mutants tested exhibited significant transport activity, except for D440E, which had activity comparable to that of wild type (EAAC1-WT). Similar results were also obtained by measuring D-[ 3 H]aspartate uptake in oocytes expressing the same mutants (Fig. 2B). To explore the possibility that the lack of transport activity in D440N, D440Q, D440S, and D440C is a consequence of impaired expression at the plasma membrane, we performed cell surface biotinylation in oocytes and found that the biotinylation levels of the Asp-440 mutants were similar or higher than EAAC1-WT (Fig. 2C). The values for the intensity of the transporter bands of the mutants relative to EAAC1-WT (n ϭ 4) were: D440E, 1 D440C mutants could be due to a dramatically increased K m and/or lowered V max . To address these possibilities, without the need to use excessive amounts of radioactive substrate, we measured the transport currents in oocytes expressing the Asp-440 mutants in the presence of 96 mM Na ϩ . No measurable currents were induced by any of the three transporter substrates, D-aspartate, L-aspartate, and L-glutamate (tested at concentrations up to 10 mM) in oocytes expressing D440N, D440Q, D440S, or D440C mutants (data not shown). Only the D440E mutant exhibited transport currents induced by each of the three substrates but not by GABA. However, their voltage dependence was remarkably different from those by WT (Figs.  3 and 4). The substrate-induced currents (Fig. 3, right panel, L-Asp-Na ϩ ) are defined as the currents obtained in the presence of sodium and the substrate (L-aspartate in the case of Fig.  3, middle panel, L-Asp) minus those in sodium without the substrate (Fig. 3, left panel, Na ϩ ). When external Na ϩ was substituted by choline, substrate-induced currents were neither observed in D440E nor in EAAC1-WT (data not shown).
In the presence of 96 mM Na ϩ , but in the absence of substrate, transient currents were observed in oocytes expressing EAAC1-WT (Fig. 3, Na ϩ ). These currents are proposed to reflect charge-moving conformational changes induced by Na ϩ binding and unbinding from the empty transporter. Upon substrate application, these transient currents were converted into steady-state currents (Fig. 3, L-Asp). The ratio of the charge moved (derived from the "off"-transient currents; jumping back to the holding potential) in the presence of Na ϩ (area to the right of the vertical line in inset a) to that in the additional presence of L-aspartate (area to the right of the vertical line in the EAAC1-WT counterpart of inset b) was 9.13 Ϯ 1.68 (n ϭ 3). In contrast, little, if any, transient currents were observed in oocytes expressing the D440E mutant using the same conditions as used in the wild-type (Fig. 3, Na ϩ ), but, remarkably, now L-aspartate-induced transient currents (Fig. 3, L-Asp) and the above ratio of the charge moved was 0.25 Ϯ 0.02 for D440E (n ϭ 3). Similar transient currents were also induced by D-aspartate and L-glutamate but not by GABA, which is not a substrate of the glutamate transporters (data not shown). The substrate-induced transient currents by D440E are reminiscent of those observed in previously characterized glutamate transporter mutants, locked in the exchange mode. These transient currents apparently reflect sodium-coupled substrate movement through the membrane electric field (6,27). It is well documented that mutants, which only mediate electroneutral exchange, still exhibit the substrate-dependent anion conductance (6,10,27,33). The reversal potential, E rev , of substrateinduced steady-state currents by D440E was more negative than in the WT, and in fact close to E rev of Cl Ϫ , consistent with the idea that D440E predominantly operates as an exchanger under these conditions.
In agreement with the idea that the outward currents by D440E (observed in the absence of SCN Ϫ , Fig. 4) and EAAC1-WT are largely carried by chloride ions moving into the oocyte, no outward currents were observed when external Cl Ϫ was replaced by the nonpermeable gluconate (supplemental Fig. 1, A and C). Moreover, when external chloride was replaced with SCN Ϫ , which is 70-fold more permeant than Cl Ϫ (48), much larger outward currents were observed (supplemental Fig. 1, B and D). The concentration dependence of the three substrates to activate the currents by D440E at Ϫ100 mV indicated a higher apparent affinity than by EAAC1-WT (Table 1 and supplemental Fig. 2) (49). Changing the membrane potential to Ϫ60 mV had little effect on these values (Table 1). D440E also exhibited a higher apparent affinity for sodium as judged by the sodium concentration dependence of the currents activated by L-aspartate at Ϫ100 mV (Table 1). In oocytes expressing glutamate transporters, elevation of external potassium induces reverse transport of endogeneous acidic amino acids (6) (Fig. 1) and this results in the activation of the transporter-mediated anion conductance, which is not observed in mutants locked in the exchange mode (6). In medium containing 10 mM of the highly permeant SCN Ϫ , K ϩ induced outward currents at positive potentials in oocytes expressing EAAC1-WT (Fig. 5A, K ϩ ), reflecting stimulation of the anion conductance. These K ϩ -induced anion currents were not seen in oocytes expressing the D440E mutant (Fig. 5B), indicating a defective interaction of this mutant with K ϩ and consistent with the idea that D440E operates as an exchanger, at least under physiological conditions. Similar to EAAC1-WT, L-aspartate induced outward currents by D440E recorded at low K ϩ concentrations, which were dependent on the presence of SCN Ϫ (Fig. 5, A and B, L-Asp versus L-Asp/Cl Ϫ ). Remarkably, in EAAC1-WT K ϩ markedly attenuated the L-aspartate-induced outward currents (Fig. 5A, L-Asp versus L-Asp ϩ K ϩ ), but this effect could not be detected in the mutant (Fig. 5B), again pointing to a defective interaction of the D440E transporters with K ϩ . Neither the L-aspartate-nor the K ϩ -dependent outward currents were observed in uninjected oocytes (data not shown). The inward currents induced by K ϩ , observed in EAAC1-WT as well as in the mutant at negative potentials (Fig.  5, K ϩ ), were also seen in uninjected oocytes (data not shown) and presumably represent cation-leak currents.
D440E Function in Reconstituted Proteoliposomes-The data presented until now suggest that in intact cells EAAC1-D440E The membrane voltage was stepped from a holding potential of Ϫ25 mV to voltages between Ϫ100 and ϩ40 mV in increments of ϩ10 mV. Each potential was held clamped for 250 ms, followed by 250 ms of a potential clamped at Ϫ25 mV. Na ϩ refers to currents recorded in the presence of ND96; L-Asp refers to currents recorded in ND96 medium supplemented with saturating concentration (2 mM) of L-aspartate and both represent nonsubtracted recordings; L-Asp-Na ϩ refers to currents recorded in ND96 medium (Na ϩ ) subtracted from those recorded after L-aspartate superfusion (L-Asp). The dashed lines indicate 0 current. Traces shown are from oocytes representative of at least three. L-Aspartate-induced steady-state currents at Ϫ100 mV ranged from Ϫ400 to Ϫ900 nA and from Ϫ60 to Ϫ400 nA in WT and D440E, respectively. Insets a and b are enlargements of the stippled areas and illustrate the transient currents obtained when jumping back from the test potentials to the holding potential of Ϫ25 mV (to the right of the vertical line in the insets).  D440E (B) in a NaCl-based external medium (ND96), as described under "Experimental Procedures," using the same voltage protocol as described in the legend to Fig. 3. Substrate-induced steady-state currents (I substrate Ϫ I ND96 ) from 210 to 245 ms at each potential were averaged and normalized to the current induced by L-aspartate at Ϫ100 mV (I/I max ). These currents were then plotted against the corresponding membrane potential (V m (mV)). Data are mean Ϯ S.E. of at least three repeats. predominantly acts as an exchanger of externally added acidic amino acid substrates with their intercellular endogenous counterparts, whereas EAAC1-WT is also capable of accumulative net flux. To address this idea experimentally, we have compared the properties of EAAC1-D440E and EAAC1-WT in reconstituted proteoliposomes, where the medium composition on both sides of the membrane can be controlled. HeLa cells, transiently expressing either EAAC1-WT or the D440E mutant, were solubilized and the transporters were reconstituted into liposomes, containing either potassium or sodium and L-aspartate in the internal medium to measure net influx or exchange, respectively. Net influx was measured by dilution of potassium containing proteoliposomes, into a sodium-containing transport medium in the presence of the potassium-specific ionophore valinomycin. The mutant exhibited low but significant D-[ 3 H]aspartate uptake, which was around 15-20% of that by EAAC1-WT (Fig. 6, net flux). On the other hand, the level of exchange by the D440E mutant was around 70% of that in the wild-type (Fig. 6, exchange). Cation Interactions by the D440E Mutant-A possible explanation of the low net flux activity of D440E is that the mutant transporter has a low apparent affinity for potassium. To test this, we measured net influx of substrate by wild-type and D440E proteoliposomes, containing various internal potassium concentrations. Half-maximal net D-[ 3 H]aspartate influx by wild-type was found at a K ϩ concentration of 19.6 Ϯ 2.4 mM, whereas the corresponding value for D440E liposomes is nearly 4-fold higher (76.6 Ϯ 11.9 mM) (n ϭ 3) (Fig. 7A, Table 2). Moreover, the maximum velocity of transport by D440E at elevated potassium concentrations was only around 15-30% of that by EAAC1-WT (Fig. 7A). Because it is possible that the potassium binding site overlaps with one of the three sodium binding sites, we also examined the apparent sodium affinity of substrate transport by the mutant. In contrast to potassium, sodium is cotransported with the substrate. Therefore we determined its apparent affinity at concentrations of the acidic amino acid substrates that were close to saturation. The K m values for D-aspar-tate in proteoliposomes containing EAAC1-WT or D440E were very similar ( Table 2). On the other hand the V max of the mutant was markedly reduced and ranged from 18 to 25 and 95 to 135 pmol/min/mg of protein for D440E and EAAC1-WT, respectively. The K m values for L-aspartate for D440E and EAAC1-WT were similar as well ( Table 2) and also with this substrate the V max of the mutant was markedly reduced and ranged from 17 to 20 and 120 to 130 pmol/min/mg of protein for mutant and EAAC1-WT, respectively. To determine the apparent sodium affinity, we used a final concentration of 1 M of the substrates. The mutant exhibited an increased apparent affinity for sodium using either transport of L-[ 3 H]aspartate (Fig. 7B) or D-[ 3 H]aspartate (Fig. 7C) as a read-out. The sodium concentrations needed to give a half-maximal activation of L-aspartate transport were 24.4 Ϯ 1.1 and 7.6 Ϯ 1.3 mM for FIGURE 5. Anion conductance in wild type and D440E mutant. Steady-state currents by oocytes expressing EAAC1-WT (A) or D440E (B) in the indicated perfusion medium are defined as follows: L-Asp/Cl Ϫ (squares), currents recorded in external ND96-based medium where 48 mM Na ϩ was replaced by the same concentration of choline were subtracted from currents recorded in the same medium supplemented with 1 mM L-aspartate; L-Asp (circles), currents recorded in the same medium as in L-Asp/Cl Ϫ , except that 9.6 mM Cl Ϫ was replaced by the same concentration of SCN Ϫ , were subtracted from currents recorded in the same medium supplemented with 1 mM L-aspartate; L-Asp ϩ K ϩ (triangles), currents recorded in the same medium as in L-Asp, where 48 mM choline was replaced by the same concentration of K ϩ , subtracted from those recorded in the same medium supplemented with 1 mM L-aspartate; K ϩ (inverted triangles), currents recorded in the same medium as in L-Asp, subtracted from those recorded in the same medium where 48 mM choline was replaced by the same concentration of K ϩ (both media were L-aspartate-free). The currents were averaged and normalized to the L-Asp current at ϩ40 mV (circles) (I norm ), and plotted against the corresponding membrane potential (V m (mV)). Currents were recorded using the same voltage protocol described in the legend to Fig. 3. Data are mean Ϯ S.E. of at least three different oocytes. The absolute values of the L-Asp currents at ϩ40 mV ranged from 1100 to 1900 nA, and from 400 to 1200 nA for EAAC1-WT and D440E, respectively. EAAC1-WT and D440E, respectively (Fig. 7B, Table 2) and the corresponding values for D-aspartate transport were 34.1 Ϯ 1.5 and 18.9 Ϯ 1.8 mM, respectively (Fig. 7C, Table 2).
To further explore the cation interactions by the D440E mutant, we examined the lithium/sodium selectivity of transport. In EAAC1, lithium can support L-aspartate transport in the complete absence of sodium, suggesting that in EAAC1, all three sodium ions can be replaced by lithium (26). Net influx of L-[ 3 H]aspartate by D440E proteoliposomes in 150 mM external sodium is around 20% of that by those of EAAC1-WT, but the corresponding value in the presence of lithium is around 75% (Fig. 8). When external Na ϩ was substituted by Li ϩ , very little D-[ 3 H]aspartate net flux was observed in either wild-type or D440E liposomes (Fig. 8). Consistently, measurements of the L-aspartate-induced currents in oocytes at varying external lithium concentrations (in the nominal absence of sodium) by oocytes expressing EAAC1 and D440E at Ϫ100 mV reveal a significantly higher apparent affinity for lithium with the D440E mutant: 16.0 Ϯ 1.4 mM by D440E and in the case of EAAC1-WT, no saturation was observed at 96 mM (Fig. 9). The positive cooperativity seen with D440E (n H ϭ 1.64) is expected because it is likely that the number of Li ϩ ions translocated with the substrate is the same as the number of Na ϩ ions (8,9).
Fast Transport Kinetics of D440E-The markedly reduced V max of net substrate influx of D440E indicates a lower steadystate cycling rate of the mutant than that of EAAC1-WT. We decided to compare the fast kinetics of the mutant with that of EAAC1-WT both under net flux and exchange conditions to pinpoint the step that is slowed down in the mutant. The WT as well as the mutant exhibit fast inwardly directed pre-steadystate currents when in the presence of sodium, glutamate is released by laser photolysis of caged glutamate (Fig. 10, A and  B). These currents are caused by the fast inward movement of sodium ion(s) into the binding site and, most likely the electrogenic glutamate translocation reaction (50). In both cases the current rises very rapidly and decays with two components. The ]aspartate (C) mediated by WT (squares) or D440E (triangles) proteoliposomes containing 120 mM potassium P i , pH 7.4, was measured at the indicated external Na ϩ concentrations (choline substitution) for 1 min using 1 M as the final substrate concentration, as described under "Experimental Procedures." The data were corrected for values obtained by proteoliposomes derived from cells expressing the vector alone, normalized to the activity of EAAC1-WT at 150 mM Na ϩ . The maximal activity values for EAAC1-WT ranged from 95 to 135 and from 120 to 130 pmol/min/mg of protein for D-and L-aspartate, respectively. Data are mean Ϯ S.E. of at least three repeats.

TABLE 2 K m values for cations and substrates in reconstituted proteoliposomes
Acidic amino acids are in micromolar and cations in millimolar concentrations.

WT D440E
L-Aspartate  . Lithium concentration dependence of substrate-induced currents. Steady-state currents induced by L-aspartate (2 mM) at Ϫ100 mV, mediated by oocytes expressing EAAC1-WT (A) or EAAC1-D440E (B), were recorded at varying external lithium concentrations (choline substitution) as described under "Experimental Procedures," using the same voltage protocol described in the legend to Fig. 3. Values were normalized to the induced currents in 96 mM lithium (I/I max ) and plotted against the indicated lithium concentrations ([Li ϩ ] mM). The induced current values at 96 mM lithium at Ϫ100 mV ranged from Ϫ160 to Ϫ250 nA, and from Ϫ120 to Ϫ140 nA in the wild type and D440E, respectively. Data are mean Ϯ S.E. of at least three repeats. The L-aspartate-induced currents in 96 mM Li ϩ relative to the induced currents in 96 mM Na ϩ by the same oocytes were 41 and 66% in EAAC1-WT and D440E, respectively.
time constants of the rapidly and slower decaying components are each very similar for mutant and WT (Fig. 10C). Moreover the time constants of the fast and slower process were similar under net influx and under exchange conditions, for both WT and mutant (Fig. 10C). Because of the absence of potassium during the exchange conditions, the two processes are associated with the sodium and glutamate translocation half-cycle and not with the potassium-dependent relocation step, which is apparently slowed down in the mutant. Na ϩ Binding to the Empty Transporter-In contrast to oocytes expressing EAAC1-WT, almost no transient currents were observed at 96 mM Na ϩ in oocytes expressing the EAAC1-D440E mutant (Fig. 3). This result may indicate that either sodium cannot bind to the empty D440E transporter or that the apparent Na ϩ affinity of the D440E transporters is so high that at 96 mM Na ϩ , almost all transporters are in the Na ϩ -bound state not only at the holding potential of Ϫ25 mV, but even at voltages as positive as ϩ40 mV (the most positive potential in the voltage-jump protocol used). Evidence for the latter possibility was provided by the fact that transient currents did in fact emerge upon lowering the Na ϩ concentration (Fig. 11). At 30 mM Na ϩ , the charge moved by the D440E expressing oocytes using our standard voltage jump protocol was 2.5 Ϯ 0.3-fold of that in 96 mM and at 10 mM Na ϩ the charge moved was 1.15 Ϯ 0.01-fold of that at 30 mM (n ϭ 3), whereas the corresponding ratios for EAAC1-WT were 1.16 Ϯ 0.07 and 0.54 Ϯ 0.08, respectively. The amount of charge moved at 30 mM Na ϩ for the oocytes tested ranged from 9.1 to 11.2 and 3.8 to 5.8 pC for EAAC1-WT and D440E, respectively. These transient currents by the D440E expressing oocytes were blocked by DL-threo-␤benzyloxyaspartate, just like in the case of their EAAC1-WT counterparts (Fig. 11). No transient currents were found in the absence of Na ϩ (Fig. 11), indicating that, as in the wild-type, these transient currents by D440E reflect conformational changes induced by Na ϩ binding/unbinding. In wild-type-expressing oocytes, these transient currents appear symmetrical at a Na ϩ concentration of around 96 mM, indicating that at this sodium concentration, around 50% of the transporters have bound Na ϩ at the holding potential of Ϫ25 mV. Symmetrical transient currents were observed with the D440E mutant at a  sodium concentration as low as 10 mM (Fig. 11), indicating also that the empty mutant transporters exhibit an increased apparent affinity for sodium.

DISCUSSION
Except for substitution by glutamate, Asp-440 of the neuronal glutamate transporter is functionally irreplaceable (Fig. 2). However, the functional properties of D440E are quite different from those of EAAC1-WT. Two-electrode voltage clamp experiments show that, as opposed to EAAC1-WT, substrates induce transient currents in oocytes expressing the mutant (Fig. 3). Moreover, external K ϩ is unable to activate the anion conductance in the mutant (Fig. 5). Thus, under physiological conditions in intact cells, D440E behaves as if it is locked in the exchange mode, effectively carrying out the sodium-coupled glutamic acid half-cycle but not the potassium-dependent relocation step (Fig. 1A). However, unlike other mutants with the same phenotype (6,10,33,36), D440E was capable of net flux, albeit with diminished rates, when reconstituted in proteoliposomes containing only K ϩ but no Na ϩ on the inside (Figs. 6 and 7). The behavior of the mutant as an exchanger in intact cells can be explained by the decreased apparent affinity of the mutant for K ϩ and the increased apparent affinity for Na ϩ (Fig.  7). After translocation to the inside of the cell, glutamate is released before the three Na ϩ ions (51) and this is followed by the binding of K ϩ and the relocation step (Fig. 1). Apparently, even though the intracellular concentration of Na ϩ is low, the high internal K ϩ is not able to displace the Na ϩ in the mutant and as a consequence the mutant acts as an obligate exchanger under physiological conditions.
The functional consequences of a highly conservative mutation of Asp-440 are reminiscent of those of the mutations in the conserved Tyr-403 residue of GLT-1 (equivalent to Tyr-373 of EAAC1) (10), In the Glt Ph structure the two equivalent residues, which are fully conserved, are located adjacent to each other (16) (Fig. 1B). Similarly to Y403F and Y403W, the D440E mutant exhibits an increased apparent affinity for Na ϩ (Fig. 7), as well as a cation-selectivity change ( Fig. 8 and see Ref. 10). This selectivity change is seen with L-aspartate, but not with D-aspartate (Fig. 8). In EAAC1-WT, transport of the latter substrate is basically only seen in the presence of Na ϩ , whereas significant L-aspartate transport is seen in the sole presence of Li ϩ (26), entirely consistent with our earlier findings that the substrate specificity of EAAC1 is determined by the nature of its coupling ion (52). As mentioned above, the effect of the D440E mutation on K ϩ interaction (Fig. 7) is more moderate than in Y403F and Y403W, as the latter two mutants appear to be devoid of any functional interaction with K ϩ (10).
D440E has a markedly decreased v max for D-and L-aspartate, yet the fast kinetic steps, thought to be associated with the Na ϩcoupled substrate translocation step (50), are not slowed down by the mutation (Fig. 10). Thus it appears that it is the K ϩ -dependent relocation step that is slowed down in the mutant. This is also consistent with the transport experiments with reconstituted proteoliposomes, showing that exchange by D440E is relatively intact as opposed to net flux (Fig. 6). It is possible that the longer side chain of D440E has an electrostatic effect on cation binding at a separate site. However, the fact that mutations at the adjacent positions, occupied by the conserved tyrosine and aspartate residues from TM7 and TM8, respectively, have such similar effects on the interactions with cations supports the notion that the residues at these positions may together participate in a novel cation binding site.
Remarkably, E404D of GLT-1, corresponding to E374D of EAAC1, is also locked in the exchange mode. In this case the functional interaction with K ϩ is defective, whereas the interaction with Na ϩ appears to be normal (6). This residue is not fully conserved and is a glutamine in Glt Ph , and in its structure this glutamine points away from the binding pocket (16). Although speculative, it is possible that in the brain transporters, this glutamate also points away in the sodium-coupled glutamate half-cycle, where it is possibly protonated (53), but faces the binding pocket in the K ϩ -mediated relocation step.
It is of interest to note that the impact of the D440E mutation on the apparent affinities for Na ϩ and for substrate is much more pronounced when this was probed using the transport currents rather than the radioactive uptake as a read-out. A likely explanation is that the effects observed during measurements of the transport currents are largely due to the substrateactivated uncoupled anion conductance, especially in the D440E mutant. The net influx measurements on the reconstituted proteoliposomes obviously correspond to the coupled transport cycle, which is lacking in the mutant under physiological conditions, when the transport currents in intact cells are measured. The conformation of the transporter during the mediation of the substrate-activated anion conductance may be different from the conformations visited during coupled transport and there is indeed strong experimental support for this idea (54 -56). Such a difference between the two processes may also account for different properties earlier observed on the D440N mutant (57). This mutant was found to be basically defective in coupled flux, in harmony with our observations (Fig. 1) and therefore the D440N mutant was analyzed using the ability of substrates to stimulate (or inhibit) the anion conductance. In particular it was concluded that D440N was specifically defective in the binding of sodium to the substrate-bound transporter, whereas it is clear that D440E is also affected by the ability of sodium to bind to the substrate-free transporter (Fig.  11).
Two Tl ϩ binding sites were identified in Glt Ph (24), but the inability of this cation to support transport (24,58) cast doubts on the idea that these two Tl ϩ sites represent two of the three sites for Na ϩ minimally needed for sodium-coupled substrate translocation by Glt Ph (32). However, experimental evidence supporting a role of one of these sites, involving Asp-455 of TM8, has been provided (33). The other Tl ϩ site is formed mainly by main chain oxygens from HP2 and TM7 and thus cannot be probed by a mutational analysis. Near this site a serine/glycine residue on HP2 was shown to control cation selectivity (25,26,59), but the side chain at this position does not point to the bound Tl ϩ seen in the Glt Ph structure (24). Nevertheless, alternative side chain conformations cannot be excluded at this point. Using an approach combining functional and computational studies, the groups of Larsson et al. (35) suggested another potential Na ϩ site, involving Asn-451 from TM8, the previously suggested Thr-370 (26), and the substrate.
These authors also presented computational evidence for the existence of an additional site, similar to the one proposed by us (34,36), but suggested that this latter site serves as a transient site. If the idea of the existence of transient sites in addition to stable sites of the Na ϩ -and substrate-loaded transporter is correct, the number of Na ϩ sites may be expected to be larger than three and the number of K ϩ sites in the mammalian transporters may be larger than one (8,9).
The novel putative site, proposed by us in this paper, is located at the extracellular end of the binding pocket, could therefore be a transient site, through which one or more of the Na ϩ ions have to pass on their way into their binding sites in the translocation complex. Alternatively, a site formed by Tyr-373 and Asp-440 may be important to stabilize the spatial relationship of the extracellular parts of TM7 and TM8 and this may be crucial for the ability of the transporter to carry out the translocation step. High resolution crystal structures of Glt Ph and mammalian glutamate transporters will be required to further establish the location and function of the proposed cation binding sites, based on functional and/or computational approaches, including the putative site described here.