Functional Modulation of the Glutamate Transporter Variant GLT1b by the PDZ Domain Protein PICK1*

Background: The scaffolding protein PICK1 (protein interacting with C kinase 1) interacts specifically with the glutamate transporter GLT1b. Results: Co-expression of PICK1 and GLT1b increases membrane leak current and alters GLT1b glutamate transport activity. Conclusion: The interaction between PICK1 and GLT1b has functional consequences on transporter kinetics. Significance: Pathology-related neuronal GLT1b expression may lead to PICK1-dependent changes in neuronal excitability. The dominant glutamate transporter isoform in the mammalian brain, GLT1, exists as at least three splice variants, GLT1a, GLT1b, and GLT1c. GLT1b interacts with the scaffold protein PICK1 (protein interacting with kinase C1), which is implicated in glutamatergic neurotransmission via its regulatory effect on trafficking of AMPA-type glutamate receptors. The 11 extreme C-terminal residues specific for the GLT1b variant are essential for its specific interaction with the PICK1 PDZ domain, but a functional consequence of this interaction has remained unresolved. To identify a functional effect of PICK1 on GLT1a or GLT1b separately, we employed the Xenopus laevis expression system. GLT1a and GLT1b displayed similar electrophysiological properties and EC50 for glutamate. Co-expressed PICK1 localized efficiently to the plasma membrane and resulted in a 5-fold enhancement of the leak current in GLT1b-expressing oocytes with only a minor effect on [3H]glutamate uptake. Three different GLT1 substrates all caused a slow TBOA-sensitive decay in the membrane current upon prolonged application, which provides support for the leak current being mediated by GLT1b itself. Leak and glutamate-evoked currents in GLT1a-expressing oocytes were unaffected by PICK1 co-expression. PKC activation down-regulated GLT1a and GLT1b activity to a similar extent, which was not affected by co-expression of PICK1. In conclusion, PICK1 may not only affect glutamatergic neurotransmission by its regulatory effect on glutamate receptors but may also affect neuronal excitability via an increased GLT1b-mediated leak current. This may be particularly relevant in pathological conditions such as amyotrophic lateral sclerosis and cerebral hypoxia, which are associated with neuronal GLT1b up-regulation.

Glutamate is the most abundant excitatory neurotransmitter in the central nervous system, and its concentration in the synaptic cleft is precisely regulated by transport proteins that reabsorb glutamate released from neurons. This reuptake of glutamate is critical for termination of the synaptic transmission and prevents buildup of excitotoxic levels of glutamate. Disturbance of glutamate reabsorption is implicated in pathophysiological conditions such as brain ischemia, epilepsy, and the neurodegenerative diseases amyotrophic lateral sclerosis and Huntington disease (for review, see Ref. 1). The removal of glutamate from the extracellular space is carried out by a family of glutamate transporters termed excitatory amino acid transporters (EAATs), 2 which comprises EAAT1 (GLAST), EAAT2 (GLT1), EAAT3 (EAAC1), EAAT4. and EAAT5 (for review, see Refs. 2 and 3). The glutamate transporters all co-transport glutamate, Na ϩ , and H ϩ in exchange for K ϩ in an electrogenic process that results in a net flux of two positive charges into the cell per translocated glutamate molecule (4 -6). In addition to the coupled transport current carried by glutamate and co-transported ions, glutamate transporters mediate a glutamate-gated uncoupled anion conductance and a leak anion conductance (7)(8)(9) as well as a leak conductance for monovalent cations (10,11). Functional studies suggest that up to 90% of the glutamate uptake in the forebrain is mediated by the EAAT2 transporter, termed GLT1 in rodents (12). This isoform exists in at least three splice variants differing only in the C termini. GLT1a (Slc2a1, transcript variant 1 in GenBank TM ) is the predominant variant, whereas GLT1b and GLT1c are expressed at lower levels (13,14). In rat hippocampus, GLT1a represents ϳ90% of total GLT1 protein, whereas GLT1b and GLT1c represent ϳ6 and ϳ1%, respectively (15). The regional distribution of GLT1a and GLT1b in rat brain is very similar, and both variants are primarily found in astrocytes. Neuronal expression of both variants has been observed (15)(16)(17)(18)(19)(20), although other research groups failed to detect GLT1b expression in neurons (15,21). Notably, induction of neuronal GLT1b expression occurs in association with amyotrophic lateral sclerosis and hypoxia, which may be a compensatory reaction to the loss of glial GLT1a that is associated with these pathologies (18,22). The intrinsic functional properties of GLT1b are not different from those of GLT1a (21), but interestingly, GLT1b con-tains a PDZ domain binding sequence in its C terminus that is absent in GLT1a (13). Recently, a physical interaction between GLT1b and protein interacting with C kinase 1 (PICK1) was discovered in a yeast two-hybrid screening of a rat neuronal cDNA library (19). PICK1 is a PDZ (PSD-95/DLG1/ZO-1) domain-containing scaffolding protein originally described as a protein kinase C␣ (PKC␣)-binding protein (23). Via its PDZ domain, PICK1 interacts with the AMPA-type glutamate receptors and thereby affects their trafficking and thus synaptic plasticity (24). PICK1 interaction with GLT1b is strictly dependent on the presence of the C-terminal class I PDZ domain binding motif in the extreme C terminus of GLT1b, which contains a unique 11-amino acid C-terminal sequence that replaces the 22 C-terminal amino acids in GLT1a. PICK1 and GLT1b can be co-immunoprecipitated from rat forebrain lysates and are co-localized in hippocampal neurons (19). Apart from a proposed role of PICK1 in the prevention of PKC-mediated down-regulation of GLT1 activity (19), modulatory effects of PICK1 on GLT1b function have not been determined, and the physiological impact of their interaction, therefore, remains unresolved.
In the present study we took advantage of the Xenopus laevis expression system, which allows individual expression of distinct glutamate transporter variants as well as selective co-expression with PICK1. We employed two-electrode voltageclamp recordings combined with uptake of radiolabeled glutamate, immunofluorescence microscopy, and Western blot to examine the functional and regulatory consequences of the interaction between the GLT1b splice variant and the PDZ domain protein PICK1.

EXPERIMENTAL PROCEDURES
Molecular Biology-The cDNA encoding the rat glutamate transporter variant GLT1a (a kind gift from Niels Christian Danbolt) and the PDZ-domain protein PICK1 were subcloned into the oocyte expression vector pXOOM (25). The C-terminal splice variant GLT1b was constructed from the GLT1a cDNA by replacing the 22 C-terminal amino acid sequence of GLT1a (TLAANGKSADCSVEEEPWKREK) with the unique C-terminal 11 amino acid sequence of GLT1b (PFPFLDIETCI) by PCR using a 3Ј flanking primer encoding the latter sequence. The constructs were verified by sequencing. Before in vitro transcription, the cDNA constructs were linearized downstream from the poly-A segment, transcribed using T7 RNA polymerase, and capped with 5Ј-7-methylguanosine using the mMessage Machine kit (Ambion Inc., TX). The cRNA was extracted with MEGAclear (Ambion).
Expression of Glutamate Transporters in X. laevis Oocytes and Electrophysiological Recordings-Defolliculated stage V X. laevis oocytes were obtained from EcoCyte BioScience. Capped cRNA encoding GLT1a, GLT1b, or PICK1 (20 ng) was micro-injected into the oocytes in a total volume of 50 nl. Oocytes were kept at 17-19°C in Kulori medium (90 mM NaCl, 1 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, pH 7.4, 182 mosM) for 3-5 days before the experiments. Transport-induced currents in oocytes were measured at 19 -24°C using the twoelectrode voltage clamp technique with a DAGAN CA-1B High Performance Oocyte Clamp and Axon Instruments Digidata 1440A and MiniDigi 1B A/D interface. Data acquisition and analysis were performed using pCLAMP 10.1 software (Axon Instruments). Recording electrodes had resistances of 1-2 megaohms. Oocytes were placed in a 20-l recording chamber, voltage-clamped at Ϫ50 mV, and continuously superfused (ϳ3 ml/min) with control or test solutions containing L-glutamate (ICN Biomedicals Inc.), D-aspartate, L-cysteine (both from Sigma), or DL-threo-␤-benzyloxyaspartate (TBOA, Tocris Biosciences, Bristol, UK), which were exchanged using a mechanical valve. The recording solution, termed 100Na buffer, contained 100 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES, pH 7.4. The current-voltage relationships for substrate-elicited currents were determined by recording steady-state current measurements in the absence of substrate and presence of substrate during 100-ms voltage pulses from Ϫ50 mV to test potentials ranging from ϩ40 to Ϫ120 mV in 20-mV increments (currents sampled at 10 kHz and low passfiltered at 1 kHz). Currents in the absence of substrate were subtracted from those measured in the presence of substrate to obtain the substrate-elicited currents. The leak current was defined as the membrane current obtained in the absence of substrate at a given clamp potential. Current voltage data were plotted as absolute values or normalized to the current generated by L-glutamate at Ϫ120 or Ϫ100 mV.
Immunofluorescence Microscopy-Five oocytes per construct per experiment were fixed for 1 h in 3% paraformaldehyde in 100Na buffer and stored in 0.3% paraformaldehyde in 100Na buffer at 5°C. Oocytes were subsequently frozen in Tissue-Tek (Sakura Finetek) and sectioned on a cryostat in 10-m sections. Sections were then permeabilized, preblocked in 4% BSA in PBS, and stained with antibodies against GLT1 (anti-B12, 1:200) and/or PICK1 (2G10, 1:400) followed by incubation with the secondary antibodies (goat anti-rabbit, Alexa-680 and donkey anti-mouse, Alexa-488) and phalloidin Alexa-555 (all Invitrogen). After a brief wash, the sections were mounted in Prolong Gold mounting medium (Invitrogen) and imaged with a Zeiss LSM 780 confocal microscope using the same settings for all sections. The staining was repeated to verify the results.

L-[
3 H]Glutamate Uptake Experiments-Uptake of radiolabeled glutamate (L-[ 3 H]glutamate, PerkinElmer Life Sciences) was measured in 24-well culture plates as previously described (29). In brief, 8 -10 cRNA-injected oocytes or 4 non-injected oocytes per condition were placed in 100Na buffer containing 60 nM L-[ 3 H]glutamate (specific activity, 48 Ci mmol Ϫ1 ) and 500 M unlabeled glutamate at room temperature (ϳ22°C). The reaction was terminated after 10 min by washing the oocytes 4 times in ice-cold Na ϩ -free buffer containing 100 mM choline chloride, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM HEPES, pH 7.4. Oocytes were dissolved in SDS (10%) and counted in a Packard Tri-Carb scintillation counter. Initial experiments showed that [ 3 H]glutamate uptake into GLT1a-or GLT1b-expressing oocytes increased linearly with incubation time of up to 12 min. In PMA experiments, oocytes were preincubated for 10 min in 100Na containing 0.1% DMSO (control) or 200 nM PMA dissolved in DMSO (final DMSO concentration 0.1%) after which [ 3 H]glutamate uptake was measured for 10 min in the presence of PMA or DMSO. In all experiments, the [ 3 H]glutamate uptake in non-injected oocytes (non-GLT1-mediated) was subtracted from the uptake measured in GLT1-expressing oocytes to obtain the GLT1-mediated uptake component.
Data Analysis-Current-voltage relationships were analyzed using pCLAMP 10.1. Concentration dependence of glutamateelicited currents in GLT1a-expressing or GLT1b-expressing oocytes was analyzed by nonlinear regression fit to a sigmoidal concentration-response curve using GraphPad Prism 5, which was also used for statistical tests. Data are given as the means Ϯ S.E. Statistical significance was determined using Student's t test or one-way analysis of variance with Bonferroni as post hoc test at a significance level of p Ͻ 0.05. Each set of data is based on oocytes from at least two different batches.

RESULTS
The Basal Electrophysiological Properties of GLT1a and GLT1b Are Similar-We verified the functional expression of the two glutamate transporter splice variants GLT1a and GLT1b in X. laevis oocytes with electrophysiological measurements of glutamate-induced currents. The glutamate-induced net currents (obtained by subtraction of the currents measured in the absence of glutamate from those in its presence) were 204 Ϯ 28 nA for GLT1a and 306 Ϯ 37 nA for GLT1b at Ϫ60 mV (n ϭ 20) when injected with equal amounts of cRNA, and the currents reversed direction at 26 Ϯ 3 and 17 Ϯ 3 mV (n ϭ 20, Fig. 1, A and B). The apparent affinities for glutamate were determined for the two transporter variants from the concentration dependence of glutamate-induced currents (Fig. 1C); bottom, are typical current traces obtained by voltage steps applied to oocytes from a holding potential of Ϫ50 mV in the absence (Ϫglu) or presence (ϩglu) of glutamate. C, glutamate-dependent currents were mediated by GLT1a or GLT1b in oocytes clamped at Ϫ50 mV. The transport currents were normalized to the maximal glutamate-induced current and fitted to a non-linear Hill equation (n ϭ 5 for GLT1a and GLT1b). D, shown are current-voltage relationships of net currents induced by glutamate alone (30 M) or by glutamate in the presence of a saturating concentration of the competitive inhibitor TBOA (100 M). The data were normalized to the current elicited by glutamate at Ϫ120 mV (n ϭ 6). All data are expressed as the means ϮS.E. the EC 50 values of glutamate-induced net transport currents were 27 M (95% confidence interval 22-32 M, n ϭ 5) for GLT1a and 24 M (95% confidence interval 20 -27 M, n ϭ 7) for GLT1b. These values were similar to the glutamate EC 50 values reported for the human GLT1 variants expressed in HEK293 cells (30). Furthermore, the net glutamate transport currents mediated by GLT1a and GLT1b were abolished by the non-transportable competitive glutamate transporter antagonist TBOA (Fig. 1D).
Verification of PICK1 Expression in Oocytes by Western Blot and Immunofluorescence Microscopy-The PDZ-domain containing scaffold protein PICK1 interacts with the C terminus of the glutamate transporter variant GLT1b but not with GLT1a (19). To determine whether PICK1 modulates functional properties of GLT1b and/or GLT1a, we co-expressed PICK1 with each of the transporter variants in Xenopus oocytes. We verified the expression of PICK1 in oocytes by Western blotting with an antibody against PICK1 ( Fig. 2A). A single band corresponding to a protein of ϳ50 kDa was recognized by the PICK1 antibody as previously observed in lysates from PICK1-transfected COS7 cells (31). No proteins were recognized by the PICK1 antibody in non-injected oocytes or those injected with GLT1a or GLT1b alone.
To determine if PICK1 localizes to the plasma membrane, we performed immunofluorescence microscopy on oocytes expressing GLT1b alone or GLT1b/PICK1. GLT1b was robustly expressed at the plasma membrane of the oocytes (as identified by F-actin staining), where it co-localized with PICK1 upon co-expression (Fig. 2B). We found no staining for PICK1 in non-injected oocytes or in oocytes expressing GLT1b alone (Fig. 2B), confirming the observations from the Western blots ( Fig. 2A).
PICK1 Increases the Leak Current in Oocytes Expressing GLT1b-To determine whether co-expression of GLT1b with PICK1 had an effect on transporter function as a consequence of the previously established PDZ-domain interaction between the two proteins, we compared the effect of PICK1 on the function of GLT1b versus GLT1a, the latter of which does not contain the PDZ domain recognition motif. The membrane potential of GLT1b-expressing oocytes was Ϫ27 Ϯ 1 mV (n ϭ 15) versus Ϫ18 Ϯ 1 mV (n ϭ 16) for GLT1b/PICK1-expressing oocytes (p Ͻ 0.0001). In voltage-clamp recordings, oocytes expressing GLT1b and PICK1 (GLT1b/PICK1-expressing oocytes) displayed a characteristically increased inward leak current in the absence of substrate (I leak ) compared with oocytes expressing GLT1b alone (Fig. 3A); at Ϫ50 mV, I leak was 81 Ϯ 7 nA in GLT1b-expressing oocytes (n ϭ 30) versus 365 Ϯ 33 nA in GLT1b/PICK1-expressing oocytes (n ϭ 15), i.e. a ϳ5-fold increase (p Ͻ 0.0001). Notably, PICK1 had no such effect on the leak current in GLT1a-expressing oocytes (I leak was 35 Ϯ 4 nA in GLT1a-expressing oocytes (n ϭ 27) and 23 Ϯ 4 nA in GLT1a/PICK1-expressing oocytes (n ϭ 21)), indicating that a PDZ domain interaction between GLT1b and PICK1 is required to induce the observed high leak current. In agreement with this notion, the leak current in oocytes expressing PICK1 alone (4 Ϯ 3 nA, n ϭ 7) did not differ from that in non-injected oocytes (2 Ϯ 1 nA, n ϭ 7, p ϭ 0.395).

PICK1 Reduces the Substrate-induced Current in GLT1b-
To characterize the effect of PICK1 on the glutamate-induced current as well as the leak current in GLT1b-expressing oocytes, we determined the current-voltage relationship in control solution (Ϫglu) and subsequently in test solution containing 500 M glutamate (ϩglu) in oocytes expressing GLT1b (n ϭ 14) or GLT1b/PICK1 (n ϭ 6). Parallel experiments were performed on oocytes expressing GLT1a (n ϭ 13) or GLT1a/ PICK1 (n ϭ 13) to test whether the observed effects were dependent on the specific GLT1b-PICK1 interaction. In GLT1a-expressing oocytes, PICK1 had no effect on the voltage dependence or magnitude of the leak currents or glutamateinduced currents in the tested range of membrane potentials (Fig. 4, A and B). The net glutamate-induced currents in GLT1a-expressing oocytes were not significantly different FIGURE 2. Verification of PICK1 expression in oocytes by Western blotting and localization of GLT1b and PICK1 in oocytes by immunofluorescence microscopy. A, in Xenopus oocytes injected with cRNA encoding PICK1, a single band of ϳ50 kDa was detected by Western blotting of the oocyte membrane fraction with a monoclonal anti-PICK1 antibody (n ϭ 5). The antibody recognized bands of similar size when the oocytes had been co-injected with cRNA encoding the glutamate transporter variants GLT1a or GLT1b. B, localization of GLT1b and PICK1 by immunofluorescence microscopy in oocytes co-injected with cRNA encoding GLT1b and PICK1, respectively. The antibody used for labeling is denoted above the photo panels and the type of cRNA injected is denoted in the left side. Co-localization of PICK1 (green) and GLT1b (red) was apparent at the plasma membrane as shown by the overlapping localization with F-actin (a plasma membrane marker, cyan labeling). Scale bar ϭ 20 m. from those in GLT1a/PICK1-expressing oocytes at any of the tested membrane potentials (Fig. 4C). Finally, the leak current in oocytes expressing PICK1 alone did not differ from that in non-injected oocytes (n ϭ 7, Fig. 4D). Conversely, the leak current in GLT1b/PICK1-expressing oocytes was significantly enhanced compared with GLT1b-expressing oocytes at all membrane potentials tested (Fig. 4, E and F, p Ͻ 0.01, values at Ϫ20 mV excluded because the current reverses at this membrane potential in GLT1b/PICK1-expressing oocytes). PICK1 co-expression resulted in a significantly smaller magnitude of the net glutamate-induced inward currents (Fig. 4G); the glutamate-induced inward current amplitude at Ϫ40 mV in GLT1b/ PICK1-expressing oocytes was ϳ8-fold smaller than the corresponding currents in GLT1b-expressing oocytes (17 Ϯ 9 versus 134 Ϯ 21 nA, p Ͻ 0.0001), indicating that the glutamate translocation capacity might be compromised in GLT1b/PICK1-expressing oocytes. To obtain a more direct measure of the maximal glutamate transport activity, the uptake of [ 3 H]glutamate into oocytes expressing GLT1b or GLT1b/PICK1 was determined. As shown in Fig. 4H, the GLT1b-mediated glutamate uptake into GLT1b/PICK1-expressing oocytes was 67 Ϯ 8% that of the uptake in GLT1b-expressing oocytes (p Ͻ 0.01, n ϭ 6). Thus, PICK1 co-expression may result in a minor reduction of the transport capacity and/or GLT1b expression level, but the ϳ8-fold smaller amplitudes of the glutamate-elicited inward current in GLT1b/PICK1-expressing oocytes at Ϫ40 mV (Fig. 4G) were not reflective of an equally reduced glutamate transport capacity.
PICK1 Alters the Transporter Kinetics of GLT1b upon Prolonged Glutamate Exposure-Prolonged application of glutamate (30 min) in GLT1b-expressing oocytes produced an initial inward current of 159 Ϯ 76 nA (termed I glutamate and labeled a, n ϭ 5, Fig. 5, A and B, top), which decreased to 74 Ϯ 23 nA after 30 min (termed I glutamate, 30 min and labeled b, Fig. 5, A and B,  top). After removal of glutamate, the leak current (I leak ) was reduced by 12 Ϯ 8 nA (labeled c, Fig. 5, A and B, top). In GLT1b/ PICK1-expressing oocytes, the initial inward current induced by glutamate (labeled a, Fig. 5, A and B, bottom) was 51 Ϯ 27 nA (n ϭ 5) and thus smaller than the corresponding current in GLT1b-expressing oocytes. In contrast to the glutamate response in GLT1b-expressing oocytes, continuous glutamate application led to a slow decrease of the current to a level that was nominally smaller than the initial base line. After 30 min in glutamate, the membrane current was reduced by 109 Ϯ 15 nA (labeled b, Fig. 5, A and B, bottom). After removal of glutamate, the leak current (I leak ) was reduced by 179 Ϯ 30 nA compared with the initial leak current level (from 335 Ϯ 50 to 156 Ϯ 32 nA), which is a significantly larger decrease than in GLT1bexpressing oocytes (p Ͻ 0.001; this change is labeled c, Fig. 5, A  and B, bottom). This is a reduction to 46 Ϯ 5% of the initial I leak . The leak current subsequently increased slowly toward presubstrate levels (data not shown). Co-expression of PICK1 had no effect on the kinetics of GLT1a-mediated currents during identical prolonged substrate applications (data not shown). The striking changes in the kinetics of the glutamate-activated current in GLT1b/PICK1-expressing oocytes were also observed with two other GLT1 substrates, D-aspartate and L-cysteine (32), as summarized in Fig. 5, C and D. To compare the effect of the different substrates, the currents were normalized to the initial, inwardly-directed current in GLT1b-and GLT1b/ PICK1-expressing oocytes upon ϳ1 min substrate exposure (termed I substrate , labeled a in Fig. 5A). I substrate was set to Ϫ1 and is shown as black bars ( Values of I substrate, 30 min of GLT1b/PICK1-expressing oocytes were significantly different from the corresponding values in GLT1b-expressing oocytes for all three substrates (Student's t test with Welch's correction, p Ͻ 0.05). Finally, the change in leak current (I leak ) after prolonged substrate exposure was significantly larger in GLT1b/PICK1-expressing oocytes than in those expressing GLT1b alone for all three substrates when normalized to I substrate (p Ͻ 0.05, Fig. 5, C and D, gray  bars). For GLT1/PICK1-expressing oocytes, values of the change in I leak were (normalized to I substrate ) 6.1 Ϯ 1.5, 4.4 Ϯ 0.2, and 4.4 Ϯ 1.2 with glutamate, aspartate and cysteine, respectively. For GLT1b-expressing oocytes, the corresponding values were 0.1 Ϯ 0.1, 0.2 Ϯ 0.1, and 0.2 Ϯ 0.2 (n ϭ 3-5, as above). The observation that three different GLT1b substrates exerted similar effects on transporter kinetics and leak current suggests that the enhanced leak current in GLT1b/PICK1-expressing oocytes is indeed mediated by GLT1b.
We thus hypothesized that the high leak current in GLT1b/ PICK1-expressing oocytes was mediated by the GLT1b transporter upon its interaction with PICK1 and that exposure to GLT1 substrates induced a concurrent activation of the coupled transport current and inhibition of a PICK1-dependent leak current. To further test this hypothesis and obtain insight into the mechanism(s) responsible for the altered currents in

PICK1 Induces a Leak Current in GLT1b
the absence and presence of substrate in GLT1/PICK1-expressing oocytes, we tested the ability of the competitive glutamate transporter inhibitor TBOA to antagonize the glutamate-induced current response and the leak current in oocytes expressing GLT1b or GLT1b/PICK1. To ensure that glutamate induced a suppression of the current in the GLT1b/PICK1expressing oocytes to a level below the initial leak current level (as shown in Fig. 5A), glutamate (30 M) was applied alone for ϳ6 min, and TBOA was subsequently co-applied with glutamate at a concentration of 100 M. As shown in the representative traces in Fig. 5E, TBOA antagonized the glutamate-dependent current suppression observed after the initial inward current in the GLT1b/PICK1-expressing oocyte (n ϭ 3). This result supports the notion that the glutamate-dependent current suppression is caused by the reduction of a leak current mediated by the glutamate transporter (compare Fig. 5, A, lower panel, with E). However, in the absence of glutamate, TBOA did not inhibit the baseline leak current in GLT1b or GLT1b/ PICK1-expressing oocytes; rather, TBOA promoted a slight increase in the leak current (Fig. 5F, n ϭ 3). PICK1 Alters the Reversal Potential of GLT1-expressing Oocytes-In the absence of substrate, the reversal potential of the membrane current is determined by the sum of endogenous and transporter-associated leak currents. The glutamate-induced activation of the transport cycle of GLT1 (EAAT2) generally shifts the reversal potential toward more positive values due to activation of the coupled transport current with its distinct ion selectivity. We observed a high leak current in GLT1b/ PICK1-expressing oocytes that is likely to be mediated by the transporter in association with PICK1. The presence of a pronounced transporter leak current might influence the ion selectivity of the transporter in distinct conformational states and thereby the reversal potential of transporter currents. Because the leak current in GLT1b/PICK1-expressing oocytes was reduced ϳ50% by a 30-min exposure to substrate, we speculated that a leak current intrinsic to GLT1b might be less dominant after exposure to glutamate, and this might be reflected by the reversal potential. To determine whether prolonged glutamate exposure in GLT1b/PICK1-expressing oocytes restored canonical characteristics of glutamate transporter-associated currents, we measured the reversal potential in control buffer (100Na) before glutamate exposure (a), in the presence of glutamate applied for ϳ1 min (b) and for 30 min (c), and finally after glutamate washout (d) in oocytes expressing GLT1b alone (Fig. 6, A-C) or oocytes co-expressing GLT1b and PICK1 (Fig.  6, D-F). The data are summarized in Fig. 6G. Reversal potentials in GLT1b-expressing oocytes were Ϫ32 Ϯ 2 mV (n ϭ 6) before glutamate exposure (100Na) and Ϫ12 Ϯ 4 mV (n ϭ 6) after brief glutamate exposure (ϳ1 min), and the corresponding values in GLT1b/PICK1-expressing oocytes were Ϫ19 Ϯ 1 and Ϫ19 Ϯ 1 mV (n ϭ 6). Thus, glutamate induced a statistically significant 20-mV positive shift in the reversal potential of the current in GLT1b-expressing oocytes after brief exposure to glutamate (p Ͻ 0.01) but had no effect on the reversal potential of the currents in GLT1b/PICK1-expressing oocytes (p ϭ 0.849). At the end of the 30-min glutamate exposure, the reversal potential in GLT1b-expressing oocytes was Ϫ25 Ϯ 1 mV (n ϭ 4), and washout of glutamate shifted the value to Ϫ38 Ϯ 3 mV (n ϭ 4). Thus, the glutamate-dependent shift in reversal potential was still prominent (p Ͻ 0.05). Conversely, in GLT1b/ PICK1-expressing oocytes, the reversal potential was shifted to Ϫ26 Ϯ 2 mV (n ϭ 3) at the end of the 30-min glutamate exposure and to Ϫ28 Ϯ 2 mV (n ϭ 3) upon washout of glutamate. These values are not significantly different (p ϭ 0.685); hence, the GLT1b/PICK1-dependent currents appear to be dominated by the PICK1-associated leak current throughout the experimental protocol.
GLT1a and GLT1b Are Down-regulated by PKC in a PICK1independent Manner-PICK1 has been proposed to protect GLT1b-containing transporter complexes from being downregulated by activation of PKC (19). However, with a complex cellular experimental system in which the GLT1a and GLT1b splice variants are endogenously co-expressed, it is difficult to assess the regulatory effect of PKC and/or PICK1 on the two GLT1 variants separately. Hence, we determined whether PICK1 is able to prevent PKC-mediated down-regulation of GLT1b in our experimental system expressing this variant alone. Exposure to the phorbol ester PMA, a potent membranepermeable activator of PKC, led to a suppression of the glutamate-elicited transport currents in oocytes expressing GLT1a (Fig. 7A, summarized in C). After 5, 10, and 15 min in PMA, the current amplitudes were reduced to 58 Ϯ 3, 31 Ϯ 3, and 15 Ϯ 3% of the current amplitude measured in the same oocyte just before PMA application (n ϭ 19;21;12, p Ͻ 0.0001 compared with control; Fig. 7C, left panel). The effect of PMA on GLT1b transport activity was similar; glutamate transport currents were reduced to 50 Ϯ 3, 27 Ϯ 2, and 13 Ϯ 2% of control after 5, 10, and 15 min in PMA (n ϭ 20, 22, and 14, p Ͻ 0.0001; Fig. 7C, right panel). There was a slight but statistically significant difference in the degree of glutamate current reduction after 5 min of exposure to PMA when comparing GLT1a and GLT1b (58 Ϯ 3 versus 50 Ϯ 3%, p Ͻ 0.05), that was absent after 10 or 15 min of exposure. The glutamate-induced currents mediated by GLT1a and GLT1b exhibited a similar voltage dependence before and after exposure to PMA (10 min), and the currents were suppressed to a similar degree at all tested potentials in the  PICK1 (B, n ϭ 13). All plotted steady-state currents were measured by applying voltage steps to oocytes from a holding potential of Ϫ50 mV. Currents in glutamate were measured after an ϳ1 min exposure to glutamate. C, shown are current-voltage relationships of net glutamate-induced currents (leak currents measured in the absence of glutamate subtracted) in GLT1a-expressing or GLT1a/PICK1-expressing oocytes (n ϭ 13). D, shown are current-voltage relationships of leak currents measured in non-injected or PICK1-expressing oocytes in the absence of glutamate (n ϭ 7). E and F, current-voltage relationships similar to those in A and B were determined in GLT1b-expressing oocytes (E, n ϭ 14) or GLT1b/PICK1-expressing oocytes (F, n ϭ 6). G, shown are current-voltage relationships of net glutamate-induced currents measured in GLT1b-expressing (n ϭ 14) or GLT1b/PICK1-expressing oocytes (n ϭ 6). H, shown is [ 3 H]glutamate uptake in oocytes injected with GLT1b cRNA (control) or co-injected with GLT1b and PICK1 cRNA. The uptake data are presented as the percentage of control values and are from six independent experiments on different batches of oocytes. All data in A-H represent means Ϯ S.E.; **, p Ͻ 0.01. JULY 12, 2013 • VOLUME 288 • NUMBER 28 range Ϫ120 and Ϫ40 mV (Fig. 7D). Taken together, PKC modulates glutamate transport in a similar manner whether transport is mediated by GLT1a or GLT1b. Because the glutamateinduced currents in GLT1b/PICK1-expressing oocytes were strongly affected by the PICK1-dependent leak current and were thus not reproducible in shape and amplitude with repeated applications of glutamate, we did not attempt to use these currents in the assessment of the influence of PICK1 on PMA-induced down-regulation of GLT1b. Instead, we employed the [ 3 H]glutamate uptake assay to determine the uptake of glutamate in the absence and presence of PMA into oocytes expressing either GLT1a or GLT1b with or without PICK1 co-expression (Fig. 7E). For GLT1a and GLT1a/PICK1, the uptake in PMA was 31 Ϯ 2 and 32 Ϯ 8% of that of control, respectively (n ϭ 4 -5), whereas PMA reduced the uptake in GLT1b and GLT1b/PICK1-expressing oocytes to 27 Ϯ 4 and 38 Ϯ 6% of that of control, respectively (n ϭ 6 and 9). For both transporter variants, the PMA-induced reduction in glutamate uptake was not significantly altered by PICK1 co-expression (p ϭ 0.880 for GLT1a versus GLT1a/PICK1; p ϭ 0.172 for GLT1b versus GLT1b/PICK1). Thus, our data fail to support a role of PICK1 as a modulator of the PKC-mediated down-regulation of GLT1b (or GLT1b-containing transporter complexes).

DISCUSSION
In the present study we have described a functional interaction between the scaffolding protein PICK1 and a variant (GLT1b) of the glutamate transporter isoform GLT1. PICK1 plays an established role in glutamatergic neurotransmission in the central nervous system by regulating the trafficking of glutamate receptors of the AMPA type and thereby modulating learning and memory (24,(33)(34)(35). PICK1 was identified as an interaction partner of the GLT1b C terminus in a yeast twohybrid screen of a neuronal library, whereas it failed to interact with that of the alternative splice variant GLT1a (19). GLT1b co-localizes with PICK1 in hippocampal neurons and co-immunoprecipitates with PICK1 in lysates of rat forebrain (19). Taken together, it is a distinct possibility that PICK1 has a modulatory effect on GLT1b activity and that their interaction may play a physiological role in the glutamatergic neurotransmission.
To obtain an experimental design in which we could identify a functional effect of PICK1 on GLT1a or GLT1b separately, we employed the X. laevis expression system. GLT1a and GLT1b displayed similar kinetics of glutamate-induced currents and EC 50 for glutamate upon expression in oocytes (this study) as well as in mammalian cells in culture (21,30). PICK1 expressed robustly in the cRNA-injected oocytes and localized to the plasma membrane. Co-expression of PICK1 resulted in a depolarization of the membrane potential and a 5-fold enhancement of the leak current in GLT1b-expressing oocytes. This leak current as well as that mediated by GLT1b alone was not blocked by the competitive glutamate transporter blocker TBOA. This observation is in accordance with existing reports that observe TBOA (and kainate) inhibition of the EAAT2/GLT1-mediated leak current only when Cl Ϫ is substituted by a more permeant anion (36 -39).
The GLT1b-mediated glutamate-induced current was reduced by ϳ90% upon co-expression with PICK1. This apparent PICK1-induced restraint on glutamate transporter activity was not reflected in [ 3 H]glutamate uptake experiments in which PICK1 reduced the glutamate transporter activity by a mere 30%. None of these characteristic features was observed in oocytes co-expressing PICK1 and the GLT1a variant, suggesting that a PDZ domain interaction between GLT1b and PICK1 is necessary to alter the transporter function and induce a high leak current. The leak current reversed at a membrane potential of ϳϪ20 mV and was inwardly directed at more negative membrane potentials, indicating that it was carried by either chloride or by a combination of sodium and potassium. The lack of TBOA sensitivity precludes isolation of the GLT1b/PICK1-me- diated leak current and thereby determination of its ion selectivity, but it suggests that the leak current is mediated by a transporter conformation distinct from that associated with the TBOA-sensitive leak anion current of other isoforms of the glutamate transporters (40,41). Curiously, during prolonged glutamate application (30 min), oocytes co-expressing PICK1 and GLT1b displayed characteristic current traces in which the membrane current slowly diminished over time to a level much below that of the original leak current level. After prolonged glutamate application, the leak current was reduced to ϳ50% that of the control level. These effects were not observed in oocytes expressing GLT1b alone or in oocytes expressing GLT1a with or without co-expression of PICK1. This selectivity underscores the requirement for a specific interaction between PICK1 and the C terminus of GLT1b. The characteristic glutamate-induced current reduction of the GLT1b/PICK1-expressing oocytes was mimicked by three other GLT1 substrates and was, in addition, prevented by the addition of the competitive glutamate transporter inhibitor TBOA. GLT1b activity thus underlies the characteristic current decay during substrate application in GLT1b/ PICK1-expressing oocytes. Taken together, PICK1 appears to interact specifically with GLT1b and thereby promotes a large leak current inherent to GLT1b.
We speculate that the current observed during prolonged glutamate exposure in GLT1b/PICK1-expressing oocytes reflects two components: 1) an inward current originating from GLT1b-mediated transport activity and 2) a slowly declining PICK1-dependent GLT1b-mediated leak current. The substrate-induced current decay might reflect a relatively slow change in the steady-state conformational equilibrium away from PICK1-bound transporter states with a high leak current toward transporter states with a low leak current, possibly due to a dissociation of PICK1 from the transporters. The substantial [ 3 H]glutamate uptake in PICK1/GLT1b-expressing oocytes (compared with the small glutamate-induced inward currents) provides support for the notion that the decline of the leak current masks the coupled and uncoupled transport currents. Consequently, these experiments did not permit evaluation of the effect of PICK1 co-expression on the coupled versus the uncoupled currents.
Oocytes expressing GLT1b alone displayed distinctive shifts in reversal potential of the membrane current upon application and removal of substrate in this study. This shift reflects the activation of a glutamate-elicited current mediated by GLT1b (predominantly a coupled transport current) with an ion selectivity distinct from that of the leak current. In contrast, the reversal potential of oocytes co-expressing GLT1b and PICK1 was not shifted by brief glutamate application nor was it significantly shifted by glutamate washout at the end of a prolonged glutamate application. Thus, although the leak current in (right panel) in the absence (Control) or presence (ϩPMA) of PMA. E, [ 3 H]glutamate uptake in the presence of PMA (200 nM, oocytes pre-exposed for 10 min before the uptake experiment) in oocytes expressing GLT1a, GLT1b, GLT1a/PICK1, or GLT1b/PICK1. Data are presented as the percentage of the uptake in oocytes that were not exposed to PMA (control) and represent the means Ϯ S.E. from 4 -9 independent experiments on different batches of oocytes. **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant.
GLT1b/PICK1-expressing oocytes was approximately halved upon prolonged glutamate application, this leak current was still predominant at the end of the prolonged exposure. The contribution from the coupled transport current was, comparably, too small to significantly affect the reversal potential of the total membrane current.
Protein kinase C decreases the activity or cell surface expression of GLT1 in many cellular systems, including primary cultures of astrocytes, mixed cultures of neurons and glia, and cell lines such as glioblastoma C6 and MDCK (Madin-Darby canine kidney) expressing GLT1a (42)(43)(44)(45). Reports on PKC-mediated regulation of the individual GLT1 variants expressed endogenously are sparse, and it remains unresolved whether GLT1b is down-regulated similarly to GLT1a. Notably, PKC-dependent down-regulation of glutamate transport activity by GLT1a involves clathrin-dependent internalization of the transporter, which depends on ubiquitination of the lysine residues 517 or 526 located in the C-terminal region of GLT1a (45,46). These residues are also present in the C terminus of the GLT1b splice variant, and it is thus possible that GLT1b is similarly ubiquitinated and internalized upon PKC activation. However, in cultured cerebellar granule cells, stimulation of PKC resulted in a down-regulation of GLT1a at the plasma membrane level, whereas GLT1b was up-regulated (47). The present data, obtained in Xenopus oocytes, indicate similar PKC-dependent regulation of the GLT1a and GLT1b variants. We cannot, however, exclude the possibility that additional proteins or signaling molecules, required for the differential regulation in cerebellar granule cells, are lacking in Xenopus oocytes. PICK1 has previously been shown to regulate protein kinase phosphorylation of several of its PDZ domain interactions partners, including acid sensing ion channel 2a (48) and mGluR7 (49). PICK1 was proposed to prevent PKC-mediated regulation of GLT1b in rat forebrain neurons via its interaction with GLT1b present in heteromeric GLT1a/GLT1b-containing transporter complexes (19). In Xenopus oocytes, in which we are ensured single-isoform expression, we observed no effect of PICK1 on PKC-dependent down-regulation of GLT1a or GLT1b uptake activity. We thus demonstrated a lack of effect of PICK1 as a modulator of PKC-dependent regulation of GLT1 uptake activity, although additional proteins/factors, which are not present in Xenopus oocytes, may be required to obtain full regulatory control by PICK1. Notably, we cannot rule out that PICK1-mediated scaffolding of PKC could exclusively modulate the electrophysiological properties of GLT1b and thus not be observed with [ 3 H]glutamate uptake experiments.
A well established role of PDZ domain-containing scaffold proteins is to tether membrane receptors and transport proteins to regulatory proteins and to localize them in a correct spatial arrangement relative to each other or to distinct cellular compartments (50). Glutamate transporters have previously been shown to be regulated by PDZ domain proteins. GTRAP48 (glutamate transport-associated protein 48), NHERF3 (Na ϩ /H ϩ -exchanger regulatory factor 3), and PSD-95 (postsynaptic density protein 95) increase transport activity of EAAT4, EAAT3, and GLT1b, respectively, by increasing cell surface expression (20,51,52). We show for the first time a modulatory effect of the neuronal scaffold protein PICK1 on the function of the glutamate transporter, GLT1b. This modulation involves a moderate decrease in activity but, more importantly, a ϳ5-fold increase in leak current. The underlying molecular mechanism, however, remains unclear.
Interestingly, PICK1 expression was recently shown to be increased in reactive astrocytes within the spinal cord of a rat model of amyotrophic lateral sclerosis (53), suggesting a role in the pathophysiology of the disease. Although there is disagreement regarding neuronal expression of GLT1b (17)(18)(19)(20) or lack thereof (15,21), it appears that there is robust neuronal expression of the GLT1b variant under pathological conditions associated with amyotrophic lateral sclerosis and hypoxia (18,22). PICK1-induced GLT1b-dependent leak currents of the relative magnitude observed in the present study would most likely translate to a significant effect on the neuronal membrane potential with associated consequences on the excitability of glutamatergic neurons. Indeed, an uncoupled transporter current mediated by the dopamine transporter (DAT) has been shown to affect neuronal excitability of midbrain dopaminergic neurons (54). PICK1 is an interaction partner of dopamine transporter (55), but it remains unresolved whether there is a functional effect of the interaction or a linkage between PICK1 and the dopamine transporter-mediated current in neurons. Direct determination of a physiological impact of the GLT1b/ PICK1-associated leak current is of high importance; however, such studies are hampered by the lack of TBOA inhibition of the leak current.
In conclusion, our data provide evidence that the scaffold protein PICK1 has a direct functional effect on the glutamate transporter variant GLT1b in the form of an enhanced GLT1bdependent leak current. The interaction is dependent on the specific interaction between PICK1 and the C-terminal protein sequence specific to the GLT1b variant. GLT1a function is thus unaffected by the presence of PICK1. PICK1 may thereby not only affect glutamatergic neurotransmission through its regulatory effect on the AMPA type glutamate receptors but also through its direct effect on the particular subset of the glutamate transporters that are up-regulated during pathological conditions such as amyotrophic lateral sclerosis and cerebral hypoxia.