Ion Fluxes Associated with Excitatory Amino Acid Transport

Home , Niflumic acid

Jacques I. Wadiche,* Susan G. Amara,t et al., 1992). Three homologous excitatory amino acid and Michael P. Kavanaugh* transporter (EAAT) subtypes are widely expressed in hu- *Vollum Institute man brain (EAAT1, EAAT2, and EAAT3; Arriza et al., tHoward Hughes Medical Institute 1994). A kinetic study of one of the human transporters Oregon Health Sciences University (EAAT2; Wadiche et al., 1995) demonstrated that the num- Portland, Oregon 97201 ber of charges translocated per transport cycle varies according to the membrane potential, in contrast to the result expected for a simple transport model involving a fixed Summary stoichiometry. Data are presented here that explain the basis for this variable stoichiometry by demonstrating that Flux of substrate and charge mediated by three cloned members of this transporter family mediate thermodynam- excitatory amino acid transporters widely expressed ically uncoupled CI- currents activated by the molecules in human brain were studied in voltage-clamped Xeno- they transport. pus oocytes. Superfusion of L-glutamate or D-aspartate resulted in currents due in part to electrogenic Na ÷ Results cotransport, which contributed 1 net positive charge per transport cycle. A significant additional compo- Steady-State Currents Activated by Excitatory nent of the currents was due to activation of a revers- Amino Acids ible anion flux that was not thermodynamically coupled The voltage dependence of the currents mediated by to amino acid transport. The selectivity sequence EAAT1, EAAT2, and EAAT3 was examined by clamping of this ligand-activated conductance was NO3- > I- > oocytes expressing the transporters at potentials between Br > CI- > F-. The results suggest that these proteins +60 and -30 mV and superfusing them with the transport mediate both transporter- and channel-like modes of substrate D-aspartate (Arriza et al., 1994) at 100 ~M. At permeation, providing a potential mechanism for negative membrane potentials, amino acid superfusion dampening cell excitability, in addition to removal of induced inward currents in oocytes expressing all three transmitter. transporter subtypes (Figures tA, 1C, and 1E). The amino acid-dependent current mediated by EAAT2 did not re- Introduction verse at potentials up to +60 mV (Figures 1C and 1D). Surprisingly, however, amino acid superfusion induced Reuptake of neurotransmitters is mediated by specific currents that reversed at positive membrane potentials in membrane proteins that couple the electrochemical gradi- oocytes expressing EAAT1 (Erev = 9.3 _.+ 0.7 mV; n -- ents of additional cotransported ions to drive the concen- 46; Figures 1A and 1 B) or EAAT3 (Erov - 38.0 __. 2.7 mV; trative influx of transmitter (for review, see Lester et al., n = 28; Figures 1E and 1F). The outward currents were 1994). The molecular mechanisms underlying this cou- most likely not due to reverse transport of accumulated pling process are unknown. Voltage-clamp studies have substrate because they were observed in response to the revealed substrate-independent ion fluxes mediated by first application of amino acid, when the membrane was some cloned neurotransmitter transporters, including those clamped at depolarized potentials (Figure 1). In addition, for 5-hydroxytryptamine (Mager et al., 1994), ~(-aminobutyric varying external concentrations of amino acid affected the acid (Cammack et al., 1994), and glutamate (Vandenberg amplitudes of the currents without changing the reversal et al., 1995), suggesting the presence of channel-like prop- potential (Figure 2). In the absence of Na÷ (choline substi- erties in these molecules. Uptake of the excitatory amino tution), neither inward nor outward currents were induced acid neurotransmitters glutamate and aspartate in brain by amino acid superfusion in any of the three transporter synaptosomes is associated with influx of Na+ and effiux subtypes (n = 4). The application of 1 mM L-glutamate of K ÷ and OH- (Kanner and Sharon, 1978; Erecinska et al., or D-aspartate to water-injected oocytes did not activate 1983). Voltage-clamp recording in retinal photoreceptors a detectable current (n = 6). and glial cells has been used to isolate currents associated with excitatory amino acid uptake that contribute signifi- A Component of the Transporter Current cantly to these cells' electrical properties (Brew and Att- Is Carried by CI- well, 1987; Tachibana and Kaneko, 1988; Schwartz and The amino acid-dependent outward current observed at Tachibana, 1990; Eliasof and Werblin, 1993). A stoichiom- positive potentials in oocytes expressing EAAT1 or EAAT3 etry proposed for glutamate uptake involves cotransport was abolished when external CI- ions (CI-out)were replaced of 2Na+:lGlu - with countertransport of 1 K + and 1 OH-, by gluconate ions, with little effect on the inward currents resulting in translocation of I net positive charge (Bouvier (Figures 3A-3C). This result suggested the possibility that et al., 1992). the outward currents mediated by EAAT1 and EAAT3 were Isolation of cDNA clones from rat and rabbit has re- carried by CI-. This was tested by examination of the rever- vealed a mammalian gene family of glutamate transport- sal potential of the amino acid-dependentcurrents medi- ers (Kanai and Hediger, 1992; Pines et al., 1992; Storck ated by EAAT1 and EAAT3 as a function of the extracellu- Neuron 722

A EAAT 1 C EAAT2 E EAAT 3

+60 mV +60 mV

) tooo, a¢ 0 see

-30 mV B +300 D F +200

+100 +20 +100 membrane potential (mV) membrane potential (mV) -40 -20 _ +20 +40 w+6~O +80 -40 -20 +20 +40 • • +80 P-- I • 9 ---t--- I I I-- I I I-- I -40 -20 +20 +40 +60 +50 membrane potential (mV) -20 -100 • -200 -200 -300 -60 • -300 o= -80 -400

Figure 1. Currents Mediated by Human Excitatory Amino Acid Transporters Currents induced by bath application (indicated by bar) of 100 pM D-aspartate to Xenopus oocytes expressing EAAT1 (A), EAAT2 (C), and EAAT3 (E). The corresponding steady-state current-voltage relations are shown in (B), (D), and (F). Currents are recorded at potentials from -30 to +60 mV and are offset to align holding currents. Note outward currents observed for EAAT1 and EAAT3. lar CI- concentration ([CI-]o°,). The reversal potential of the tion of the transporter-mediated current activated by excit- EAAT1 and EAAT3 currents shifted 54.1 ± 1.8 mV (n = atory amino acids. 5) and 53.7 ± 4.3 mV (n = 5) per decade change in [CI-]out, To examine the properties of the transporter currents respectively (Figure 3D). Although the magnitudes of the further, intracellular CI- (Cl-~o) was depleted by dialysis of reversal potential shifts were close to predictions for a EAATl-expressing oocytes for 16-24 hr at 17°C in CI--free CI--selective electrode, the absolute values of the reversal medium (gluconate substitution); these oocytes were com- potentials were significantly more positive than the value pared with matched oocytes incubated in ND96. Ec~ was of Ec~, the CI- equilibrium potential (see below). This result measured before and after CI-in depletion by measurement suggested that other ions in addition to CI carried a por- of reversals of endogenous Ca2%dependent CI- channels

A EAAT 1 B EAAT 2 C EAAT 3 membrane potential (mV) +300 membrane potential (mV) +50 membrane potential (mV) I +500 -150 -100 -50 -t O -,00 -50 +100 -150 -1SO -50 J ~ +100

o a o o o eo~e o~o~o

-300 o o ~ ~ -soo e oq°° o ~o ~ o A A <>~ o uG -600 " -100 A o~ ¢ o <> Do -1000 o~ Ao~ o oo -900 ~ o Ao y -150 ~" u o -1500 AoU El o o 0 o El -1200 o ~ -200 o L] o -2000 o -1500 -250 -2500

Figure 2. Concentration and Voltage Dependence of Excitatory Amino Acid Currents Steady-state current-voltage relations of representative oocytes between +80 and -120 mV, obtained by subtraction of control currents from the corresponding currents in the presence of varying concentrations of D-aspartate. Xenopus oocytes expressing EAAT1 (A), EAAT2 (B), and EAAT3 (C). The reversal potential (EAAT1 and EAAT3) does not shift as a function of [D-aspartate]. D-aspartate doses were 10 I~M (closed circles), 30 p.M (triangles), 100 ~M (diamonds), 300 p.M (squares), and 1 mM (open circles). Excitatory Amino Acid Transporter Ion Ftuxes 723

A B C EAAT 1 EAAT 2 EAAT 3 +s00 104raM Cl~= .so 104mM Cir.t 104mM Cl~ut membrane potential (mV) ~o membrane potential (mV) mernbrane potential (rnV) ~[ +500 ~o,// +tnn -150 -100 -,50 ot]~o° +100 -150 -100 -50 /.00 -150 -100 -50 I L ~ ~-~o-b~d I I I __ L i 1 ~e o 9 I ~o ,I°B~[3oD~O ~o \ ~o -500 [3o -soo 0mM Cl~ut -so OmM C/-= ea ss' OmM CI~, ~o 0 -1000 -,o00 o= -100 2 o ~: -1500 -1500 ~' g -100 = © = u g -2000 ~'> -2000 E] -200 -2500

D Figure 3, CI- Carries the Transporter-MediatedOutward Current (A-C) Steady-statecurrent-voltage relationsof oocytes expressing 80 EAAT1 (A), EAAT2(B), and EAAT3 (C) in responseto an application 70 of 1 mM D-aspartate.Removal of CI oct (gluconatesubstitution) abol- ishes the outward current but has no effect on the inward current. 60 (D) Reversalpotential of the current inducedby D-aspartateis depen- 5o dent on [CI ]o°t.The reversalsshifted 54.1 _+ 1.8 and 53.7 + 4.3 mV/ ~ 40 decade for EAAT1 and EAAT3, respectively(mean _+ SEM; n = 5), uJ~ 30 20 10 0 10 100 [CI-]out (mi)

(see Experimental Procedures). With 104 mM [CI-]out, Ec~ dent current-voltage relations (n = 5). In addition, the was shifted from -17 - 1 mV (n = 4) to -81 _ 3 mV oocyte CI- channel blocker niflumic acid (100 I~M; White (n = 3), avalue corresponding to a change in intracellular and Aylwin, 1990) and the nonselective blocker SITS (1 concentration from 53 to 4 raM. Comparison of currents mM; Greger, 1990) had no effect on the current induced activated by D-aspartate in the presence of CI-out revealed by 1 mM D-aspartate, nor did removal of extracellular Ca2+ that the inward current was significantly reduced following (n = 4) dialysis of CI-i,, with little effect on the outward current (Figure 4A). The reversal of the current was shifted from Thermodynamics of Transport 12.5 --- 2.9 mV (n = 5) to 7.5 __. 0.9 mV (n = 5). When The above results suggest that the transporters mediate both Clgn and CI-out were substituted with gluconate, the a reversible CI- current in addition to an inward current remaining excitatory amino acid-induced current exhib- presumably associated with electrogenic cotransport (Fig- ited an exponential dependence on membrane potential ure 4). To test the idea that the excitatory amino acid- (e-fold/75 mV) and did not reverse at potentials up to +80 dependent current remaining in the absence of CI- reflects mV (Figure 4B). The CI--dependent component of the ex- electrogenic amino acid flux, its voltage dependence was citatory amino acid-dependent current was resolved by compared with that of the EAATl-mediated [3H]D-aspar- subtraction of current-voltage recordings in the nominal tate uptake. The [3H]D-aspartate flux displayed a voltage absence of CI ~n and CI-out from recordings in the presence dependence similar to that of the D-aspartate-activated cur- of normal [Cl-]in and [CP]out (Figure 4C). The reversal of rent in the absence of CI- (e-fold/75 mV; Figure 4B; Figure this difference current was -10 mV more negative than 5A, inset). Time integration of currents associated with the measured value of Ec~,which is likely due to incomplete flux of radiolabeled amino acid in recording buffer con- dialysis of C1%. taining CI- revealed that the ratio of the flux of charge to Because Xenopus laevis oocytes express high levels that of amino acid varied with membrane potential (Figure of Ca2+-activated Cl- channels (Robinson, 1979; Barish, 5A). The quantity of charge translocated per transport cy- 1983), the possibility was examined that the transporters cle ranged from approximately +3.5 eo at -100 mV to ap- might activate endogenous CI- channels, e.g., via Ca2+ proximately -2.5 eo at +25 mV (Figure 5B). Thus, amino permeating the transporter or by second messenger- acid flux is not directly proportional to the net ionic flux. mediated intracellular Ca2+ release. However, microinjec- To investigate the influence of the CI electrochemical tion of oocytes expressing EAAT1 with 10 nmol of BAPTA, gradient on uptake of amino acid, flux of [3H]D-aspartate a quantity sufficient to block receptor-mediated intracellu- mediated by EAAT1 was measured under isopotential lar Ca2+ release ([BAPTA]~ = 10 raM; Kavanaugh et al., conditions with the CI- electrochemical gradient directed 1991), had no effect on the excitatory amino acid-depen- either outwardly or inwardly by varying [CI-]out (Figure 5C). Neuron 724

A [Cl']out = 104 mM B [Cl]out = 0 mM C +400 membrane potential (mV) I +500 membrane potential (mV) F +500 membrane potential (mV) DDD DD nO -150 -I00 -50 LoD~,oa -150 -100 -50 | +50 -150L_ -1001 [][]+50 +100 I I I _ I / u±-=~ "5ponOOl ] I ~ ~l~l 0 ~ u~ [] [] dialysed oo~° I--500 dialysed DDO ~O~_.500 -200 DD O I Icl- [] =0 n° 0 0 [3 0 ° 0 0 o 0 ¢~ -1000 u ~0 c~ -1000 [] C -400 [] • {{ ~= 4500 {{ ~ -15oo -600

control ~ -2000 control -2000 -800

Figure 4. Depletion of Ch, Reveals Two Currents (A) Currents activated by 300 ~,M D-aspartate in oocytes expressing EAAT1 dialyzed to remove CI in (gluconate substitution; n = 5) compared with undialyzed cells (n = 6; [CI-]out = 104 mM). (B) Currents activated by 300 ~,M D-aspartate in dialyzed cells compared with undialyzed cells ([CI ]o.t = 0 mM). (C) CI--dependent component of the transport current revealed by subtraction of the mean current-voltage curves recorded in the presence (control; A) and absence (dialyzed; B) of CI-.

At 0 mV, with an inwardly directed CI- gradient ([CI ]out = cated per molecule of amino acid was compared in the 104 mM), uptake during a 100 s pulse of 100 I~M [3H]D- presence and absence of CI- by measuring the time inte- aspartate was 0.48 ± 0.06 pmol/s (n = 9). With an out- grals of currents resulting from a 100 s pulse of 100 ~M wardly directed CI- gradient at the same membrane po- [3H]D-aspartate. At -80 mV, the number of fundamental tential ([CI-]out = 0 mM), flux of [3H]D-aspartate was charges per molecule of D-aspartate was 2.4 -+ 0.02 unchanged (0.47 _+ 0.06 pmol/s; n = 8). This result dem- (n = 4), 2.0 _+ 0.07 (n = 6), and 2.7 _+ 0.3 (n = 5)for onstrates that the driving force responsible for amino acid EAAT1, EAAT2, and EAAT3, respectively. After depletion influx is not related to the CI- electrochemical gradient, of C1%, superfusion of the same concentration of label in suggesting that the CI- and amino acid fluxes occur inde- CI--free buffer resulted in a reduction in the quantity of pendently. charge translocated per molecule of o-aspartate to 1.4 ± Assuming that influx of excitatory amino acid is driven 0.1 (n = 6), 1.1 ± 0.04(n = 5),and 1.2 ± 0.04(n = thermodynamically by electrogenic ion cotransport, the 4), respectively. The flux of radiolabel in nominal CI--free above results suggest that the net current reflects the sum conditions was not significantly changed relative to uptake of the inward current from amino acid flux (1~) and the in the presence of CI-. Thus, with CI-in depleted and in the current arising from the reversible and thermodynamically absence of CI%.t, there is a translocation of - 1 net positive uncoupled amino acid-activated CI- conductance (Io). To charge associated with D-aspartate flux. In the presence test this hypothesis further, the quantity of charge translo- of CI-, less than 1 charge was translocated per molecule

A B C 3 0.6 ~o.6i. 4 ¢

X ~0.4 .*... ~. -6 3 0 0.5 2 2 ~- 0.2 "'- - 2 D 0 0.4 ¢ E ~,o~ .... == 1 o. , U) C3 -100 -60 -20 20 "" 0.3 o--1 potential (mV) C @ x ' : • - C~ o.. 0.2 ...... ~ ...,. ~.i ® 0 <¢ ~-2 C3 0.1 .C o -1 ~ r ~ ~ i _~ -3 0 -100 -80 -60 -40 -20 0 20 -100 -80 -60 -40 -20 0 20 104 0 membrane potential (mV) membrane potential (mV) [Cl']ou t (mM)

Figure 5. The Quantity of Charge Translocated per Transport Cycle Varies with Membrane Potential owing to a Thermodynamically Uncoupled CI- Flux (A) Amino acid uptake and charge translocation were simultaneously measured during a 100 s application of 100 I~M [~H]D-aspartateto voltage- clamped oocytes expressing EAATI. (Inset) Voltage dependence of labeled amino acid flux; superimposed exponential (e-fold/75 mV) derived from fit of transport current under nominal CI -free conditions (see Figure 4B). (B) Quantity of charge translocated per transport cycle varies with the membrane potential. (C) Under is•potential conditions (0 mV), amino acid uptake is not affected by changing the CI electrochemical gradient. Excitatory Amino Acid Transporter Ion Fluxes 725

A B Figure 6. ReversalPotential and RelativeAm- plitude of CI- Flux Is Substrate Dependent membrane potential (mV) +300 (A) Voltage dependenceof EAATl-mediated -150 -100 -50__9~r~@@ +100 "°¢~ 2 I .L c currents induced by application of 100 I~M . oO°[ D~ b-aspartate (circles) or 100 pM L-glutamate oOO[113 0 -300 "~ (squares). Reversalpotentials differ by 37 _+ 0 [3 0 D 3.5 mV (n = 8). 0 G .~oo ~=. ~ I (B) Chargetranslocation per cycle of transport 0 [3 measured for D-aspartate(closed bars) and 0 rJ -9oo ~. '~ [3 L-glutamate (hatched bars) while clamping D-Asp o [] -1200 .¢ 0 cells below(-50 mV) or above (0 mV) Ec~. o D-Asp L-GIu D-Asp L-Glu L-Glu -1500 -50 mV 0 mV of amino acid at potentials positive to Ec~, whereas at po- Discussion tentials negative to EcL, more than 1 charge was translocated. At Ec~, -1 charge was translocated (Figure 5B). Electrogenic uptake of excitatory amino acids is mediated The results are consistent with the interpretation that by a family of membrane proteins that cotransport Na÷ translocation of 1 net positive charge is intrinsically cou- (Kanner and Sharon, 1978; Stallcup et al., 1979) and pled to uptake of a molecule of D-aspartate, but additional countertransport OH- (Erecinska et al., 1983; Bouvier et charge transfer arises from the uncoupled flux of CI- al., 1992) and K + (Kanner and Sharon, 1978; Amato et al., through the transporter. 1994). If transport were tightly coupled to translocation of these inorganic ions, application of amino acid to one membrane face would be expected to result in a unidirec- Gating and Selectivity of the Cl- tional current with a polarity determined by the stoichiome- Permeation Pathway try of the cotransported ions. For a cotransport stoichiome- The reversal potential predicted for a theoretical current try of 1AA-:2Na+ with countertransport of 1 OH- and 1 K+ reflecting the sum of a current flowing through a perfect (Bouvier et al., 1992), a net charge of +1 would accompany inward rectifier and a reversible CI- conductance is depen- influx of each molecule of glutamate or aspartate regard- dent on the relative magnitude of each component (see less of membrane potential. Results from a recent study of Discussion and Figures 8A-8C). The reversal potential of EAAT2 kinetics demonstrate, however, that the net charge the EAATl-mediated current activated by L-glutamate was accompanying translocation of glutamate varies according more positive than that activated by D-aspartate, by 37.0 to membrane potential (Wadiche et al., 1995), and the _ 3.5 mV (n = 8; Figure 6A). This result suggests that present study demonstrates that under appropriate condi- the flux of CI- per transport cycle gated by D-aspartate tions the current associated with excitatory amino acid was greater than that gated by L-glutamate, since the influx can indeed reverse polarity. D-aspartate-induced current reverses closer to Ect. To test These results can be explained by a model involving this possibility, the number of charges translocated per activation of a CI- conductance in parallel with the conduc- molecule of [3H]D-aspartate was compared with that of tance associated with amino acid flux. In this model, the [3H]L-glutamate at potentials positive and negative to EcL. net excitatory amino acid-dependent current represents At -50 mV, - 33 mV more negative than Ec~, the ratio of the sum of the CI- current (Ic~) and the electrogenic trans- charge flux to amino acid flux was greater for D-aspartate port current (IAA). Although the CI- current is activated by than for L-glutamate; the converse was true at 0 mV, - 17 transport substrates, the CI- electrochemical gradient is mV more positive than Ec~ (Figure 6B). These results are not thermodynamically coupled to transport, since amino consistent with translocation of either amino acid being acid flux is unaffected by the direction of the CI- driving coupled to movement of 1 charge, with the magnitude of force (see Figure 5C). Furthermore, excitatory amino acid the CI flux induced by D-aspartate being greater. influx occurs in the absence of CI . From measurements The permeation of anions other than CI through the of the quantity of charge translocated with each molecule transporter was examined by recording EAAT1 currents of amino acid in the presence of CI-, it was ascertained activated by 100 pM D-aspartate under bi-ionic conditions, that the net positive charge translocated into the cell per' with substituted test ions outside at 100 mM and physiolog- transport cycle was >1 eo at membrane potentials negative ical concentrations of CI ~, (-53 mM; see Experimental to Ec~, while at potentials positive to Ec~, it was <1 eo, At Procedures). The amplitudes and the reversal potentials the equilibrium potential for CI , -1 positive charge ac- of the currents varied among the ions tested (Figure 7). companied each molecule of amino acid entering through The general selectivity sequence, reflected in both the the transporter (see Figure 5B). In addition, the voltage current reversal potentials and the outward current ampli- dependence of radiolabeled amino acid influx was similar tudes, was NO3 > I > Br- > CI- > F-. A precise Goldman- to that of the amino acid-dependent current in CI -free Hodgkin-Katz analysis of the relative anion permeabilities conditions (e-fold/75 mV; see Figure 4B; Figure 5A). To- based on the reversal potential is not possible because gether, these results suggest that translocation of excit- the inwardly rectified substrate flux leads to a different atory amino acid is coupled to translocation of I net posi- zero current equation at each potential. tive charge. Neuron 726

+3000 ~ NO 3" a significant percentage (50%-73%) of the total current A activated by D-aspartate was carried by CI- at -80 mV, ~" A ~ ~= +2000 A based on measurements of the net quantity of charge translocated per transport cycle, assuming a charge of +1 A? ~ []DD[] D~I" eo per cycle due to coupled cotransport. u A A E3 o [] The selectivity of the ligand-gated anion conductance ~n @ ~o oOO o oar was revealed by substitution experiments that demonstrated the sequence NO3- > I- > Br- > CI- > F->> gluconate-. An absence of anomolous mole fraction be- "l~0~O~U ~l;00[membra~:otentiLV ) havior observed between I- and (~1- (data not shown) is consistent with a single anion binding site in the trans- Figure 7. Anion Selectivityof EAAT1 porter pore, although more rigorous tests will be required Currents induced by 100 I~M D-aspartate in a representative oocyte to rule out a multi-ion permeation pathway. This selectivity expressing EAAT1 with bath solutionscontaining various test ions as sequence is identical to that of a neuronal CI- channel, 100 mM Na÷ salt plus gluconate salts of Ca2+ (1.8 rnM), Mg2÷ (1 mM), and K+ (2 mM). which displays complex conductance properties that ap- pear to result from anion-cation interactions in the channel pore (Franciolini and Nonner, 1987, 1994). As the excit- The above model predicts that the reversal potential of atory amino acid-activated CI- current through the trans- the total current is independent of amino acid concentra- porter required Na ÷, it is possible that Na ÷ and CI- interact tion only if the concentration dependence for activation of in the pore of the transporter, although binding of Na + may I~ and Ic~ is the same (Figures 8B and 8C). Thus, the simply be required prior to excitatory amino acid binding observed concentration independence of the transporter (Wadiche et al., 1995). Unlike currents mediated by the current reversal potentials (see Figure 2) is consistent with neuronal CI- channel, in which the reversal potential is both Icr and I~ arising from excitatory amino acid binding to affected by cation concentrations (Franciolini and Nonner, a single site. Each transporter subtype exhibited intrinsic 1987), reducing [Na+]oot from 101 to 20 mM reduced the (expression level-independent) differences in the reversal excitatory amino acid-activated current amplitude 53% 4- potential of the net current activated by excitatory amino 8% (-80 mV; n = 4), without changing the reversal po- acids, which would be unlikely if the CI- current were medi- tential. ated by a distinct molecular species. Furthermore, classi- Na ÷- and glutamate-dependent currents with properties cal CI- channel blockers did not affect the transporter- similar to those described here have been reported in ver- mediated currents, supporting the hypothesis that the tebrate photoreceptor cells (Sarantis et al., 1988; Ta- transporters directly mediate both currents. The reversal chibana and Kaneko, 1988; Eliasof and Werblin, 1993). potential of the net current is predicted to shift with In addition, fluctuation analyses of currents in photorecep- changes in [CI-]outl[CI-]~,, but the magnitude of this shift, tors of turtle (Tachibana and Kaneko, 1988) and salaman- as well as the absolute value of the reversal potential, will der (Larsson et al., 1994, Biophys. J., abstract) suggest be influenced by the relative magnitude of IAAand Ic~(Figure the presence of a small conductance channel activated 8C). In general, the greater the contribution of IAA to the by glutamate transporter substrates. The activation of a net current at the reversal potential, the less effect chang- CI- conductance concomitant with transport would provide ing the CI- gradient will have on the reversal potential. a potential mechanism to offset the depolarizing action Similarly, the greater the relative magnitude of Ic~, the of transmitter reuptake and dampen cell excitability. The closer will be the net current reversal potential to Ec~. Thus, molecular mechanisms underlying this current remain to the difference in reversal potentials for the L-glutamate- be elucidated. The thermodynamic independence of the and D-aspartate-activated currents in EAAT1 may be ac- CI- and glutamate fluxes suggests that these ions do not counted for by differences in ligand efficacy for activation simultaneously permeate a single-file pore. Therefore, un- of the CI current, leading to a difference in flux of CI- per less the transporter has a"double-barreled" ion pathway to transport cycle (see Figure 6). In addition to substrate- allow independent permeation, a gating mechanism must dependent changes in reversal potentials, the transport- exist for switching between Na÷-coupled amino acid trans- ers displayed subtype-specific differences in the reversal location and CI- permeation. One potential mechanism for potentials (ranging from +9 mV in EAAT1 to +38 mV in such a switch involves the bound glutamate molecule itself EAAT3 to >80 mV in EAAT2 for D-aspartate currents). In constituting a critical part of the selectivity site required for addition to possible intrinsic differences in activation of CI- permeation (e.g., by contributing a positively charged the CI- current, other reasons for differences in reversal m-amino group with which the anion could interact in tran- potentials between subtypes could include differences in sit). In such a model (see Figures 8D and 8E), the mean the voltage dependence of substrate flux or in intrinsic time that glutamate is bound to the transporter before un- rectification of Ic~. Interestingly, a recently cloned human binding or being translocated through the pore would de- cerebellar transport subtype, EAAT4, mediates an excit- termine the mean lifetime of the anion-conducting state. atory amino acid-induced current that is carried predomi- Variations in the microscopic kinetics of this gating pro- nantly by CI- and reverses close to EcJ (Fairman et al., cess could lead to differences in the relative amplitudes 1995). In all three subtypes examined in the present study, of amino acid and anion fluxes for different transporter Excitatory Amino Acid Transporter Ion Fluxes 727

A B C +500 _~ membrane potential (mY) membrane potential (mY) +s0o ...... membrane potential (mV) +500 -150 -100 -50 -150 -100 -50 -150 -100 -50 I k I L ...... iiii

Icl --- o -1000 c~ -1000 i -lOOO ~

-2000 -2000 '~ -2000 ~ g IAA ~ lative -3000 |total -3000 -3000 4 / AA flux

D E @ outside+G~-u ~N~ 2Na+G,Uo /2.a +Gu transpo, mo0e

TNa2GIu ÷ Cl'°~ ~ fCl'' lchanne,mode inside To Tii TNa2GluCI

Figure 8. Excitatory Amino Acid Transporter Modes (A) Model of total transporter current (solid line) as the sum of a reversible CI- current (Ic,) and an electrogenic current from coupled flux (I,~; see Experimental Procedures). (B) The predicted reversal potential of the net current is independent of amino acid concentration when the concentration dependence of I,~ and Ic, is the same. (C) The absolute reversal potential is dependent on the amino acid flux relative to that of CI . The curves shown represent the effect of scaling the intrinsic transporter turnover rate by the indicated factor while holding the CI- conductance constant. (D and E) Cartoon (D) and corresponding kinetic scheme (E) representing modes of transporter operation. The alternating access model requires a state transition for glutamate permeation (To -~ Ti), while CI permeation requires only that glutamate be bound. For simplicity, a partial reaction cycle is shown that omits the countertransport step for K+ and OH (or HCO~-) as proposed by Bouvier et al. (1992). subtypes and substrates such as those observed in the to bath application of substrates or by off-line subtraction of control present study. current records obtained during 200 ms voltage pulses to potentials between -120 and +80 mV from corresponding current records in the presence of substrate. The CI reversal potential was determined in Experimental Procedures oocytes expressing EAAT1 and in uninjected cells by measuring the reversal potential of currents mediated by Ca2+-activated CI channels Expression and Electrophysiological Recording endogenous to Xenopus oocytes following activation with I ILM A23187 Capped mRNAs transcribed from the cDNAs encoding the human (Barish, 1983). brain glutamate transporters EAAT1-EAAT3 (Ardza et al., 1994) were microinjected into stage V-VI Xenopus oocytes (50 ng/oocyte), and 3H-Labeled Amino Acid Flux membrane currents were recorded 3-6 days later. Recording solution Current measurements were made during superfusion of 100 ~M (ND96) contained 96 mM NaCI, 2 mM KCI, 1 mM MgCI2, 1.8 mM CaCI2, [3H]D-aspartate (0.42 Ci/mmol; Amersham) or [3H]L-glutamate (1 Ci/ and 5 mM HEPES (pH 7.4). In Na÷ or CI- substitution experiments, mmol; Amersham) onto oocytes voltage-clamped at various potentials ions were replaced with equimolar choline or gluconate, respectively. for 100 s. Following washout of the bath (<20 s), oocytes were rapidly Two microelectrode voltage-clamp recordings were performed at 22°C transferred into a scintillation tube, lysed, and measured for radioactiv- with a Geneclamp 500 interfaced to an IBM compatible PC using a ity. In control experiments, no significant effiux of radiolabel was de- Digidata 1200 A/D controlled using the pCLAMP 6.0 program suite tected during this time in oocytes injected with 100 pmol of [3H]D- (Axon Instruments), and to a Macintosh using a MacLab A/D (ADInstru- aspartate (final concentration = 100 p.M). Currents induced by ments). The currents were low pass-filtered between 10 Hz and I kHz 3H-labeled amino acids were recorded using Chart software (ADInstru- and digitized between 20 Hz and 5 kHz. Microelectrodes were filled ments) and integrated off-line, followed by correlation of charge trans- with 3 M KCI solution and had resistances of

the sum of a reversible CI current and an inwardly rectifying transport by membrane vesicles isolated from rat brain. Biochemistry 17, 3949- current as a function of voltage: 3953. Kavanaugh, M. P., Christie, M. J., Osborne, P. B., Busch, A. E., Shen, I,o,(E) = [[EAA]/([EAA] + K05)} [(E- Ec,)(gc,) - (tN)(F)exp(-Ep.)l, K. Z., Wu, Y., Seeburg, P. H., Adelman, J. P., and North, R. A. (1991). Transmitter regulation of voltage-dependent K+ channels expressed where I,o,(E) is the total membrane current as a function of voltage; in Xenopus oocytes. Biochem. J. 277, 899-902. [EAA] is the concentration of excitatory amino acid; Ko.sis the concen- Lester, H. A., Mager, S., Quick, M. W., and Corey, J. L. (1994). Perme- tration of amino acid that activates 50% of each current; (gc~) is the ation properties of neurotransmitter transporters. Annu. Rev, Pharma- CI- conductance due to the channel mode of the transporter, assumed col. Toxicol. 342, 219-249. for simplicity to be ohmic; (Ec0 is the equilibrium reversal potential for CI-; F is the Faraday constant; t is the turnover rate of the transporter Mager, S., Min, C., Henry, D. J., Chavkin, C., Hoffman, B. J., Davidson, at 0 mV for a given amino acid; N is the number of moles of transporter N., and Lester, H. A. (1994). Conducting states of a mammalian seroto- expressed; and p. is a Boltzmann factor determining the voltage depen- nin transporter. Neuron 12, 845-859. dence of the transporter turnover rate. Pines, G., Danbolt, N. C., Bjoras, M., Zhang, Y., Bendahan, A., Eide, L., Koepsell, H., Storm-Mathisen, J., Seeberg, E., and Kanner, B. I. Acknowledgments (1992). Cloning and expression of a rat brain L-glutamate transporter. Nature 360, 464-467. We thank J. Arriza for providing EAAT plasmids and E, McCleskey, Robinson, K. R. (1979). Electrical currents through full-grown and ma- S. Eliasof, and B. Bean for discussion and comments. This work was turing Xenopus oocytes. Proc. Natl. Acad. Sci. USA 76, 837-841. supported by grant GM48709 from the National Institutes of Health. Sarantis, M., Everett, K., and Attwell, D. (1988). A presynaptic action The costs of publication of this article were defrayed in part by of glutamate at the cone output synapse. Nature 332, 451-453. the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC Section 1734 Schwartz, E. A., and Tachibana, M. (1990). Electrophysiology of gluta- solely to indicate this fact. mate and sodium co-transport in a glial cell of the salamander retina. J. Physiol. 426, 43-80. Received March 7, 1995; revised May 25, 1995. Stalicup, W. B., Bulloch, K., and Baetge, E. E. (1979). Coupled transport of glutamate and sodium in a cerebellar nerve cell line. J. Neuro- References chem. 32, 57-65. Storck, T., Schulte, S., Hofmann, K., and Stoffel, W. (1992). Structure, Amato, A., Barbour, B., Szatkowski, M., and Attwell, D. (1994). expression and functional analysis of a Na+-dependent glutamate/ Counter-transport of potassium by the glutamate uptake carrier in glial aspartate transporter from rat brain. Proc. Natl. Acad. Sci. USA 89, cells isolated from the tiger salamander retina. J. Physiol. 479, 371- 10955-10959. 380. Tachibana, M., and Kaneko, A. (1988). L-glutamate-induced depolar- Arriza, J. L., Fairman, W. A., Wadiche, J. I., Murdoch, G. H., Kava- ization in solitary photoreceptors: a process that may contribute to the naugh, M. P., and Amara, S. G. (1994). Functional comparisons of interaction between photoreceptors in situ. Proc. Natl. Acad. Sci. USA three glutamate transporter subtypes cloned from human motor cor- 85, 5315-5319 tex. J. Neurosci. 14, 5559-5569. Vandenberg, R. J., Arriza, J. L., Amara, S. G., and Kavanaugh, Brew, H., and Attwell, D. (1987). Electrogenic glutamate uptake is a M. P. (1995). Pore properties and substrata binding domains of human major current carrier in the membrane of axolotl retinal glial cells. glutamate transporters. J. Biol. Chem., in press. Nature 327, 707-709. Erratum: Nature 328, 742. Wadiche, J. i., Arriza, J. L., Amara, S. G., and Kavanaugh, M. P. Barish, M. E. (1983). A transient calcium-dependent chloride current (1995). Kinetics of a human glutamate transporter. Neuron 14, 1019- in the immature Xenopus oocyte. J. Physiol. 342, 309-325. 1027. Bouvier, M., Szatkowski, M., Amato, A., and Attwell, D. (1992). The glial White, M. M, and Aylwin, M. (1990). Niflumic and flufenamic acids are cell glutamate uptake carrier countertransports pH-changing anions. potent reversible blockers of Ca2+-activated CI- channels in Xenopus Nature 360, 471-474. oocytes. Mol. Pharm. 37, 720-724. Cammack, J. N., Rakhilin, S. V., and Schwartz, E. A. (1994). A GABA transporter operates asymmetrically and with variable stoichiometry. Neuron 13, 949-960. Eliasof, S., and Werblin, F. (1993). Characterization of the glutamate transporter in retinal cones of the tiger salamander. J. Neurosci. 13, 402-411. Erecinska, M, Wantorsky, D., and Wilson, D. F. (1983). Aspartate transport in synaptosomes from rat brain. J. Biol. Chem. 258, 9069- 9077. Fairman, W. A., Vandenberg, R. J., Arriza, J. L., Kavanaugh, M. P., and Amara, S. G. (1995). An excitatory amino acid transporter with properties of a ligand-gated chloride channel. Nature 375, 599-603. Franciolini, F., and Nonner, W. (1987). Anion and cation permeability of a chloride channel in rat hippocampal neurons. J. Gen. Physiol. 90, 453-478. Franciolini, F., and Nonner, W. (1994). Anion-cation interactions in the pore of neuronal background chloride channels. J. Gen. Physiol. 104, 711-723. Greger, R. (1990). Chloride channel blockers. Methods Enzymol. 191, 793-810, Kanai, Y., and Hediger, M. A. (1992). Primary structure and functional characterization of a high-affinity glutamate transporter. Nature 360, 467-471. Kanner, B. I., and Sharon, i. (1978). Active transport of L-glutamate