(Received for publication, July 23, 1996)
From the Vollum Institute, Oregon Health Sciences
University, Portland, Oregon 97201 and § Department of
Biochemistry, Hadassah Medical School, The Hebrew University,
Jerusalem 91120, Israel
Glutamate transporters maintain low synaptic concentrations of neurotransmitter by coupling uptake to flux of other ions. After cotransport of glutamic acid with Na+, the cycle is completed by countertransport of K+. We have identified an amino acid residue (glutamate 404) influencing ion coupling in a domain of the transporter implicated previously in kainate binding. Mutation of this residue to aspartate (E404D) prevents both forward and reverse transport induced by K+. Sodium-dependent transmitter exchange and a transporter-mediated chloride conductance are unaffected by the mutation, indicating that this residue selectively influences potassium flux coupling. The results support a kinetic model in which sodium and potassium are translocated in distinct steps and suggest that this highly conserved region of the transporter is intimately associated with the ion permeation pathway.
The role of electrogenic (Na+ + K+)-coupled glutamate transporters, located in the plasma membranes of nerve terminals and glial cells is to keep the extracellular concentration of the neurotransmitter below neurotoxic levels (1, 2). Furthermore, together with diffusion, they may help to terminate its action in synaptic transmission (3, 4).
A glutamate transporter (GLT-1) has been purified to near homogeneity and reconstituted (5, 6). It represents around 0.6% of the protein of crude synaptosomal fractions and is one (7) of four glutamate transporters expressed in brain cloned thus far (8-10). The stoichiometry of the electrogenic transport process (11-13) is three sodium ions per glutamate (14), whereas potassium is transported in the opposite direction (11). In addition, glutamate transport is accompanied by alkalinization of the external medium (15), and transport of glutamate can be driven by a pH gradient (16). It is not clear whether a proton is cotransported with glutamate or an hydroxyl is antiported (17), but recent evidence favors cotransport with a proton (14). Mechanistic studies have shown that the transport process is ordered and that sodium and potassium are translocated in distinct steps. The sodium ions bind first, followed by glutamate. After their translocation and release on the inside, potassium binds on the inside and is translocated outwards to complete the translocation cycle (18, 19).
Because all substrates are charged molecules, conserved charged amino acids located in hydrophobic stretches of the transporter proteins are likely to be important for the binding and translocation of these substrates. Using site-directed mutagenesis, we have identified four amino acid residues (GLT-1 numbering) important for the transport process: histidine 326 (20) and the acidic amino acid residues aspartate 398, glutamate 404, and aspartate 470 (21). In this report, we describe the in-depth characterization of a mutant in which glutamate 404 has been changed into aspartate. This mutant, E404D, appears to be able to catalyze the exchange of glutamate and aspartate but unable to carry out net flux of the acidic amino acids. We show that glutamate 404 is important for carrying out the potassium transporting limb of the cycle, suggesting that this residue may represent part of the potassium binding site.
Capped
mRNAs transcribed from the cDNAs encoding the rat brain
glutamate transporter GLT-1 (7) or its mutant E404D (21) were
microinjected into state V-VI Xenopus oocytes (50 ng/oocyte), and membrane currents were recorded 3-6 days later (13).
Recording solution (Ringer) contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl2, 1.8 mM CaCl2, and 5 mM HEPES, pH 7.4. In Na+ or Cl substitution experiments, ions
were replaced with equimolar choline or gluconate, respectively. Two
microelectrode voltage-clamp recordings were performed at 22 °C with
a Geneclamp 500 interfaced to an IBM-compatible PC using a Digidata
1200 A/D control using the pCLAMP 6.0 program suite (Axon Instruments)
and to a Macintosh using a MacLab A/D (ADInstruments). The currents
were low pass-filtered between 10 Hz and 1 kHz and digitized between 20 Hz and 5 kHz. Microelectrodes were filled with 3 M KCl
solution and had resistances of <1 megaohms. Offset voltages in
Cl
substitution experiments were avoided by the use of a
3 M KCl-agar bridge from the recording chamber to a 3 M KCl reservoir containing an Ag/AgCl electrode.
Current-voltage relations were determined either by measurement of
steady-state currents in response to bath application of substrates or
by off-line subtraction of control current records obtained during
200-ms voltage pulses to potentials between
100 and +40 mV from
corresponding current records in the presence of substrate.
Current measurements were made during superfusion of 100 µM [3H]D-aspartate (0.42 Ci/mmol; Amersham Corp.) onto oocytes voltage-clamped at various potentials for 100 s. Following washout of the bath (<20 s), oocytes were rapidly transferred into a scintillation tube, lysed, and measured for radioactivity. In control experiments, no significant efflux of radiolabel was detected during this time in oocytes injected with 100 pmol of [3H]D-aspartate (final concentration, 100 µM). Currents induced by 3H-labeled amino acids were recorded using Chart software (ADInstruments) and integrated off-line, followed by correlation of charge transfer with radiolabel flux in the same oocytes.
Uptake in Reconstituted SystemsHeLa cells were cultured in
DMEM supplemented with 10% fetal calf serum (heat-inactivated), 100 units/ml penicillin, 100 µg/ml streptomycin, and 2 mM
glutamine. Infection with recombinant vaccinia/T7 virus vTF7-3 and
subsequent transfection with plasmid DNA was done as described (22)
using pT7-GLT-1 (shortened insert; Ref. 23), pT7-GLT-1 E404D (21), or
vector alone (pBluescript SK) at 2.8 µg/well (3-cm
diameter).
Cells from two wells were washed twice with 1 ml PBS and then taken up in 20 µl of PBS. The cell suspension was mixed with liposomes (4 µmol of asolectin and 0.7 µmol of brain phospholipids) and 0.9% cholate in a final volume of 220 µl of a solution of identical composition as the one used to equilibrate the spin columns, used for the subsequent reconstitution (24). The columns were equilibrated with 5 mM Tris-SO4, pH 7.4, 1 mM MgSO4, 0.5 mM EDTA, 1% glycerol, and other ions and amino acids corresponding to the desired "in-medium" as detailed in the figure legends. In the case of exchange experiments, acidic amino acid, usually 10 mM L-aspartate, was included in the equilibration buffer of the spin-column. The external L-aspartate was removed by passing these proteoliposomes over a spin column of identical composition but lacking the amino acid. The control liposomes (without entrapped L-aspartate) were also passed through a second spin column. Uptake of [3H]D-aspartate was measured as described (21) using 20 µl of proteoliposomes into 360 µl of 0.15 M NaCl and 1 µCi of [3H]D-aspartate (11.5 Ci/mmol) unless indicated otherwise in the figure legends. Uptake in intact cells was performed as described (22).
When expressed in
HeLa cells using a vaccinia/T7 recombinant virus, uptake of
[3H]D-aspartate mediated by the rat brain
glutamate transporter GLT-1 is similar to that in cells expressing a
mutant GLT-1 transporter containing the conservative substitution
E404D (21). Although similar levels of radiotracer uptake were also
seen in voltage-clamped Xenopus oocytes injected with RNA
transcribed from the corresponding wild-type and mutant cDNAs,
membrane currents recorded simultaneously during radiolabel uptake were
dramatically different in the two groups of cells (Fig.
1). While uptake of
[3H]D-aspartate in the two groups was similar,
the magnitude of currents associated with flux mediated by the mutant
transporter was less than 5% of that seen in oocytes expressing the
wild-type transporter (Fig. 1).
These data show that the mutant E404D transporter mediates uptake of
radiolabeled D-aspartate, but they suggest that this uptake
is occurring in an essentially electroneutral fashion at 30 mV.
Uptake of excitatory amino acids by all glutamate transporter clones
thus far characterized results in activation of a current reflecting
the sum of the inward current (resulting from cotransport of coupled
ions such as sodium) together with a chloride current flowing through a
thermodynamically uncoupled conductance pathway (10, 25). The voltage
dependence of the currents mediated by wild-type GLT-1 and the E404D
mutant (GLT-1 E404D) has been examined by clamping oocytes, expressing
the transporters, at potentials between +40 and
100 mV in the
presence and absence of the transport substrate D-aspartate
at 100 µM. In the wild-type GLT-1, inward currents are
observed (Fig. 2A) that do not reverse (Fig.
2, A and C), similar to those observed in its
human homolog (25). The behavior of E404D is quite different. In
addition to the steady-state current, which is greatly diminished as
compared with the wild-type, a relatively slow transient current is
observed (Fig. 2B). Neither of these currents are observed
in water-injected oocytes (13). The very small steady-state current of
E404D does, however, reverse at potentials more positive than
20 mV
(Fig. 2, B and C). The
aspartate-dependent transient currents appear to be
capacitative because the charge movements, following hyperpolarizing or
depolarizing pulses, are equal to those following the return to the
original potential (Fig. 2B). Although we will speculate on
the nature of these transient currents in the "Discussion," it is
important at this point to note that their magnitude indicates that the
greatly reduced steady-state currents of E404D are not due to
corresponding reduced expression levels of the mutant transporter. This
is further supported by the fact that the much faster
kainate-sensitive, sodium-dependent transients, which
reflect sodium binding to the transporter or a subsequent
conformational change (13), are similar for wild-type and mutant
transporters (data not shown). The acidic amino
acid-dependent outward current observed in glutamate transporters expressed in oocytes is largely due to a sodium- and
glutamate-dependent anion conductance not thermodynamically coupled to the transport cycle (25). The outward steady-state current
in the E404D transporter (Fig. 2C) is largely carried by
chloride ions moving into the oocyte because it is greatly attenuated
when the external chloride concentration is reduced from 104 to 4.8 mM (Fig. 2D). Further support for this comes
from the observations that much larger outward currents are observed when the external chloride is substituted by the highly permeant NO3
ion (Fig. 3).
The anion conductance activated during uptake shows a selectivity
sequence favoring chaotropic anions (25, 26), and this selectivity
sequence was unchanged in the E404D mutant (Fig. 3). Raising external potassium induces reverse transport (18, 27). This
phenomenon is observed in Xenopus oocytes, which have
approximately 12 mM intracellular substrate concentrations
(28). The data presented suggest that the E404D transporter may be
impaired in potassium binding. Evidence supporting this is presented in
Fig. 4. Oocytes expressing the wild-type and E404D
transporter have been clamped at 0 mV in Cl-free
Ringer's solution (substituted with nitrate). In the wild-type, superfusion with 100 µM D-aspartate induces
the well characterized anion conductance of the transporter as measured
by a large outward current, which is dependent on nitrate (Fig.
4A; this current is 20-fold lower in normal Ringer's
solution; see also Fig. 2C). This anion conductance can also
be induced by exposing the wild-type transporters to external
potassium, which will cause the transporter to catalyze efflux of
internal glutamate and/or aspartate (Fig. 4A). In the E404D
transporters, this anion conductance can be activated by external
aspartate but not by external potassium (Fig. 4B),
indicating that potassium cannot interact with them.
E404D Function in Reconstituted Systems
The data from the
previous sections indicate that uptake of
[3H]D-aspartate by the E404D transporters is
due to its exchange with intracellular acidic amino acids but not to
its net influx. To prove this, we have solubilized HeLa cells
transiently expressing either wild-type or mutant transporters and
reconstituted those transporters into liposomes. This system allows
control of the composition of the external as well as the internal
medium. Fig. 5A shows that potassium-loaded
proteoliposomes inlaid with E404D transporters are virtually unable to
accumulate [3H]D-aspartate. The accumulation
in the wild-type is completely dependent on the presence of external
sodium (data not shown). This accumulation is dependent on internal
potassium and is not observed when it is replaced by Tris (Fig.
5B, closed triangles). If, however, 10 mM
L-aspartate is included in the internal medium together
with Tris, significant accumulation of
[3H]D-aspartate is observed with wild-type
and E404D transporters (Fig. 5B) but not with liposomes
inlaid with extracts of cells transfected with the vector alone (data
not shown). Under these conditions, the accumulation is absolutely
dependent on external sodium; similar results are also obtained when
L-glutamate is present on the inside and also when sodium
is used instead of Tris (data not shown). These experiments prove that
E404D proteoliposomes catalyze exchange but cannot catalyze net influx,
which is dependent on internal potassium. The ability of external
potassium to interact with the mutant proteoliposomes has been
investigated after their preloading with
[3H]D-aspartate by exchange. After a 10-min
accumulation of [3H]D-aspartate into the
proteoliposomes by exchange, they have been diluted 10-fold into media
containing sodium chloride and 100 µM of unlabeled
L-glutamate or D-aspartate (Fig.
6). The two substrates induce a rapid efflux of the
previously accumulated [3H]D-aspartate in
both the wild-type (Fig. 6A) and the E404D transporter (Fig.
6B). Whereas dilution of the proteoliposomes inlaid with the
wild-type transporter into sodium-containing media without anionic
amino acids does not result in efflux, a rapid net efflux is obtained
into a potassium-containing medium (Fig. 6A). However, no
such net efflux is observed in the case of the mutant (Fig. 6B). Dilution into choline-containing media does not lead to
efflux, neither in the wild-type nor in the mutant (data not
shown).
To further emphasize the selectivity of the defect of
potassium-dependent net flux in the mutant, we have
compared another important parameter, i.e. affinity for
external sodium. Initial rates of sodium-dependent
[3H]D-aspartate transport in HeLa cells
expressing wild-type and mutant have a very similar dependence on the
external sodium concentration (Fig. 7). Although in the
experiment D-aspartate has been used at concentrations
below its Km, similar results are obtained when
saturating levels (200 µM) are used (data not shown). An identical sodium dependence of
[3H]D-aspartate exchange has been observed in
proteoliposomes inlaid with wild-type and E404D transporter (data not
shown). Thus, E404D is selectively impaired in one or more of the
interactions of potassium with the transporter.
The transport cycle of glutamate, as it emerges from mechanistic
studies (18, 19), is illustrated in Fig. 8. On the
outside, the three sodium ions have to bind to the transporter first
(step 1). This is followed by the binding of the acidic amino acid
(glutamate in this example; step 2), followed by translocation (step
3). Then the acidic amino acid debinds (step 4), followed by the three sodium ions (step 5). This leaves the binding sites facing inward. To
start a new cycle, the transporter has to reorient these sites, and for
this purpose, potassium is required. After it binds from the inside
(step 6), it is translocated (step 7) and released from the outside
(step 8). In the case of net efflux, this cycle is reversed, going
clockwise, instead of counterclockwise. Exchange represents a partial
reaction of this cycle (steps 1-4 or 5 and back), including the
reversible translocation of the acidic amino acid. Net flux is
dependent on cis-sodium is external sodium is required for
influx and internal sodium for efflux. Also exchange requires sodium,
but it does not matter on which of the two sides it is present.
Although there are some differences in the relative permeabilities of
glutamate and aspartate between E404D and wild-type (21), the dominant
feature is that E404D cannot carry out the whole cycle (Fig.
5A) but can exchange (Figs. 5B and 6). These data
are in harmony with the electrophysiological data, which indicate that
E404D is "locked" in the exchange mode (Fig. 1). Besides the
ability to translocate sodium and the acidic amino acid, E404D exhibits
an unimpaired substrate-dependent anion conductance (Figs.
2, 3, 4) and a similar affinity to external sodium as the wild-type, both
in intact cells (Fig. 7) as well as in proteoliposomes (data not
shown). Thus, we are not dealing here with some structural mutation
altering the conformation of the transporter, but a specific step is
affected. Two possibilities come to mind: 1) potassium binding and/or
translocation is impaired; and 2) the affinity for sodium is increased
such that the rate of unbinding is dramatically decreased (step 5 for
influx, step 1 for efflux). This latter possibility can be excluded
because potassium-dependent efflux in E404D liposomes is
defective (Fig. 6), whereas the affinity for external sodium is the
same as for the wild-type in this preparation. In addition, no marked
differences in affinity for internal sodium between wild-type and
mutant proteoliposomes are observed (data not shown). Furthermore,
whereas in intact cells the affinity of the mutant transporter for
external sodium is identical to that of the wild-type (Fig. 7), the
mutant transporter has lost the ability to interact with external
potassium (Fig. 4). Therefore, our results indicate that potassium
binding and/or translocation is impaired in the E404D mutant from the
inside as well as from the outside, and that perhaps glutamate 404 is
one of the liganding groups for potassium. It is especially striking
that the removal of a single methylene group is sufficient to cause
this defect. In view of this extremely high specificity, it is perhaps
not surprising that other potassium congeners, i.e.
rubidium, ammonium, and cesium, which can replace potassium in the
wild-type to varying extents, cannot enable net flux in the mutant
transporter (data not shown).
Voltage jumps in the presence and absence of external D-aspartate reveal that exchange mediated by the E404D mutant is accompanied by a marked transient current that decays with a time constant between 10 and 20 ms (Fig. 2). This transient current is significantly slower than the previously described sodium-dependent and kainate-sensitive transient current in EAAT2, the human homolog of GLT-1 (t = 3-4 msec; Ref. 13). This latter current is thought to reflect sodium binding and/or state transitions involving charge movements subsequent to sodium binding. Such sodium-dependent conformational changes have recently been monitored experimentally in GLT-1 (29). The slower transient currents associated with the exchange step may reflect reversible movement of sodium and glutamate across the electric field during this partial reaction cycle.
The E404D mutant mediates a sodium- and aspartate-dependent anion conductance (Fig. 2-4). The selectivity of this conductance is unchanged from the wild-type transporter (SCN > NO3 > I > Cl; Fig. 3; see also Refs. 25 and 26). This result indicates that the anion-conducting state is likely to correspond to a sodium- and glutamate-occupied state (25), rather than a state in the potassium-transporting limb of the transport cycle (Fig. 8). A similar conclusion has been reached with independent experiments (17). The current activated by aspartate in E404D reverses close to the reversal potential for Cl, consistent with electroneutral exchange with little or no net electrogenic transport.
The residue E404D is conserved among all cloned mammalian glutamate transporters and is expected to fulfill a similar role in them. Mutation of the corresponding residue in EAAT3 (30), the human homolog of EAAC1 (9), also abolished potassium-dependent efflux.1 Interestingly, this residue is not conserved in the related neutral amino acid transporter ASCT-1 (31, 32), and recent work shows that this transporter also mediates potassium-independent amino acid exchange (33). Glu404 is located in the middle of a structural domain that influences binding of the glutamate analogue kainate (34) and is also near residue Asp398, which is required for transport (21). Amino acid sequences in this region of the glutamate transporters are highly conserved and contain multiple short hydrophobic segments (7-10). Pore-forming domains in both voltage-gated and ligand-gated ion channels tend to be highly conserved in these proteins (35) and often consist of relatively short loop structures (36). A similar role for the domain encompassing Glu404 in glutamate transporters is suggested both by its critical role in potassium permeation as well as its highly conserved structure.