Glutamate released into the synaptic cleft is ... - Wiley Online Library

JOURNAL OF NEUROCHEMISTRY

| 2009 | 110 | 264–274

doi: 10.1111/j.1471-4159.2009.06125.x

Facultad de Ciencias, Centro de Biologıá Molecular ‘‘Severo Ochoa’’, Universidad Autońoma de Madrid, Consejo Superior de Investigaciones Cientı´ficas, Centro de Investigacioń en Red de Enfermedades Raras, Madrid, Spain

Abstract The glutamate transporter GLT1 is expressed in at least two isoforms, GLT1a and GLT1b, which differ in their C termini. As GLT1 is an oligomeric protein, we have investigated whether GLT1a and GLT1b might associate as hetero-oligomers. Differential tagging (HA-GLT1a and YFP-GLT1b) revealed that these isoforms form complexes that could be immunoprecipitated when co-expressed in heterologous systems. The association of GLT1a and GLT1b was also observed in mixed primary cultures of rat brain and in the adult rat brain, where specific antibodies for GLT1a immunoprecipitated GLT1b and vice versa. Dual immunofluorescence in mixed cultures demonstrated the partial co-localization of both isoforms in neurons and in glial cells. Because GLT1b interacts with an

organizer of post-synaptic densities, PSD-95, we examined the capacity of GLT1a to associate with this protein. GLT1a was immunoprecipitated from the rat brain in protein complexes that contained not only GLT1b but also PSD-95 and NMDAR. The interaction between GLT1a with PSD-95 and NMDAR was reproduced in transfected COS7 cells and it appears to be indirect as it requires the presence of GLT1b. These results indicate that the major isoform of the glutamate transporter, GLT1a, can acquire the capacity to interact with PDZ proteins through its inclusion in hetero-oligomers containing GLT1b. Keywords: glutamate receptors, glutamate transporter, oligomers, post-synaptic density 95, scaffold proteins, synapse. J. Neurochem. (2009) 110, 264–274.

Glutamate released into the synaptic cleft is inactivated by glutamate transporters, thereby ending neurotransmission and preventing excessive neurotoxic stimulation of glutamate receptors (reviewed in Grewer et al. 2008). Although there are five high-affinity glutamate transporters (GLT1, GLAST, EAAC1, EAAT4, and EAAT5), GLT1 is responsible for up to 90% of extracellular glutamate clearance in the forebrain, of which there are multiple splice variants (Rothstein et al. 1996; Tanaka et al. 1997). There are at least three unique GLT1 isoforms encoded by transcripts with 3¢ untranslated regions of different lengths: GLT1a, GLT1b, and GLT1c (Chen et al. 2002; Reye et al. 2002; Schmitt et al. 2002). GLT1a is mainly expressed in glial cells, although it is also present in neurons. Indeed, 20–30% of the axon terminals in the hippocampus or the somatic sensory cortex can be immunohistochemically stained for this isoform, providing evidence that GLT1a is the pre-synaptic glutamate transporter sensitive to dihydrokainate that has remained elusive for decades (Danbolt et al. 1992; Torp et al. 1994; Lehre et al. 1995; Chen et al. 2004; Melone et al. 2009). The cellular distribution of the GLT1b isoform is less clear but it also seems to be present in neurons and glia (Chen et al.

2002, 2004; Schmitt et al. 2002; Maragakis et al. 2004; Sullivan et al. 2004; Berger et al. 2005). In the C-terminal domain of all glutamate transporters, except GLT1a, there is a sequence motif capable of interacting with proteins that contain PDZ domains. Indeed, GLT1b has been shown to interact with two PDZ proteins, PICK1 (Bassan et al. 2008), and post-synaptic density 95 (PSD-95) (Gonzalez-Gonzalez et al. 2008), and it appears to be capable of interacting with other proteins containing the PDZ domain, at least in vitro (Gonzalez-Gonzalez et al. 2008). PDZ proteins are scaffolds that organize macromolecular complexes many

264

Received December 12, 2008; revised manuscript received April 15, 2009; accepted April 20, 2009. Address correspondence and reprint requests to Dr. F. Zafra, Facultad de Ciencias, Centro de Biologıá Molecular ‘‘Severo Ochoa’’, Universidad Autońoma de Madrid, Madrid, 28049, Spain. E-mail: [email protected] Abbreviations used: DSS, disuccinimidyl suberate; ER, endoplasmic reticulum; FCS, fetal calf serum; GFAP, glial fibrillary acidic protein; HA, Hemaglutinin; MAP2, Microtubule-associated protein 2; MDCK, Madin-Darby canine kidney; PSD 95, post-synaptic density 95; PVDF, polyvinylidene difluoride; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis.

2009 The Authors Journal Compilation 2009 International Society for Neurochemistry, J. Neurochem. (2009) 110, 264–274

Hetero-oligomerization of GLT1 | 265

of which associate plasma membrane receptors with other proteins involved in their regulation (Tomita et al. 2001; Kim and Sheng 2004). Glutamate receptors are good examples of this kind of organization as some subunits of the NMDAR bind to the PSD-95 scaffolding protein, as do a-amino-3hydroxy-5-methylisoxazole-4-propionate receptors, although in this case the association is indirect (Tomita et al. 2001). Many intracellular regulatory proteins can bind to the same scaffold, thereby regulating the intracellular pathways activated by these receptors. Indeed, the binding of transporters to the same scaffold might also fulfill a regulatory function by controlling the concentration of neurotransmitter in the receptor environment. There is evidence that glutamate transporters exist as oligomers in the plasma membrane. Although the threedimensional structure of the mammalian GLT1 is not yet available, the glutamate transporter homolog GltPh from Pyrococcus horikoshii has been crystallized (Yernool et al. 2004). These structural data revealed the bacterial transporter to be a bowl-shaped trimer, although each subunit worked independently of the others in terms of glutamate transport despite its multimeric nature (Grewer et al. 2005; Koch and Larsson 2005). Immunoprecipitation and Fluorescence Resonance Energy Transfer (FRET) analysis are compatible with mammalian glutamate transporters forming trimeric structures (Haugeto et al. 1996; Yernool et al. 2003, 2004; Gendreau et al. 2004), and as GLT1a and GLT1b isoforms may be partially co-expressed in glial cells and in some forebrain neurons, we examined the possibility that these isoforms might assemble as hetero-oligomers. Indeed, immunoprecipitation assays revealed this to be the case in both heterologous expression systems and in native brain tissue. Moreover, while GLT1a is unable to interact directly with the scaffold protein, PSD-95, it can do so when incorporated into hetero-oligomers that also contain GLT1b, thereby permitting it to interact with NMDAR. Accordingly, the minor GLT1b isoform might confer on the major, GLT1a, those properties derived from the interaction with PSD-95, including its anchoring in the neighborhood of NMDAR.

Materials and methods Materials The Lipofectamine-PLUS, Lipofectamine 2000, the pCDNA3 plasmid, the rabbit polyclonal anti-green fluorescent protein (also detects the YFP variant), and the Alexa Fluor 488, Alexa Fluor 594 or Alexa Fluor 647 conjugated secondary antibodies (goat antirabbit, goat anti-mouse goat anti-guinea pig, and goat anti-chicken) were purchased from Invitrogen (Carlsbad, CA, USA), whereas phenylmethanesulfonyl fluoride, the Expand High Fidelity PCR system (Taq polymerase) and all restriction enzymes were obtained from Roche (Mannheim, Germany). Vectashield was purchased from Vector (Burlingame, CA, USA). Standard proteins for sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE

Rainbow markers) and the enhanced chemiluminescence western blotting detection reagents were all obtained from Amersham Pharmacia (Buckinghamshire, UK). The Restore Western Blot Striping Buffer was obtained from Thermo Scientific (Waltham, MA, USA). Guinea pig anti-GLT1a raised against the C-terminal sequence of the rat GLT1a, rabbit anti-NMDAR2A, and Immobilon-P sheets [polyvinylidene difluoride (PVDF)] were obtained from Millipore Corporation (Billerica, MA, USA). The mouse anti-PSD-95 antibody was obtained from Oncogene (San Diego, CA, USA) and the mouse anti-NMDAR1 was purchased from BD Biosciences/Pharmingen (San Jose, CA, USA). Rabbit anti-GLT1b was raised against the Cterminal sequence of this protein and has been characterized previously (Gonzalez-Gonzalez et al. 2008). The mouse monoclonal anti-Hemaglutinin (HA) was prepared in the Centro de Biologıá Molecular ‘Severo Ochoa’ from the 12CA5 hybridoma. The disuccinimidyl suberate (DSS) and the EZ-linkTM sulfo-NHS-SSbiotin were obtained from Pierce (Rockford, IL, USA), the pGEM-T easy cloning vector was purchased from Promega (Madison, WI, USA), and the oligonucleotides used were synthesized by Isogen (Utrech, The Netherlands). The papain dissociation system used was manufactured by Worthington Biochemical (Lakewood, NJ, USA) and fetal calf serum (FCS) was supplied by Gibco (Paisley, Scotland). Poly-D-lysine, mouse anti-glial fibrillary acidic protein (GFAP), mouse anti-Microtubule-associated protein 2 (MAP2), rabbit anti-HA, and all other chemicals were obtained from Sigma Chemical Co. (St. Louis, MO, USA). Monoclonal anti-p58 was a gift of Dr. Fernando Martıń from CBMSO (Madrid, Spain). Plasmid constructs The expression vectors for HA-GLT1a, HA-GLT1b, YFP-GLT1a, and YFP-GLT1b in pCDNA3 were obtained by PCR cloning as described previously (Gonzalez-Gonzalez et al. 2008). HA and YFP tags were placed in the N-terminus and did not interfere either with the uptake capability of the transporters, evaluated by [3H]glutamate uptake assays, or with its subcellular distribution, evaluated by immunofluorescence in different cell types (see Fig. S1). C-terminal tagging was avoided as previous reports indicate that this region of the transporter might be involved in specific interactions and intracellular trafficking (Bassan et al. 2008; Gonzalez-Gonzalez et al. 2008). Electrophoresis and blotting Sodium dodecyl sulfate–polyacrylamide gel electrophoresis was performed in the presence of 2-mercaptoethanol and the minigels were run for 2 h at 20 mA. After electrophoresis, the protein samples were transferred to PVDF membranes in a semi-dry electroblotting system as described previously (Cubelos et al. 2005). When reprobing was necessary, PVDF membranes were stripped using the Restore Western Blot Striping Buffer following the manufacturer’s instructions. Immunoprecipitation COS7 cells were plated in p60 dishes and transfected with Lipofectamine PLUS reagent following the manufacturer’s instructions. Two days later cells were solubilized in lysis buffer (1 mL) for 30 min at 4C (50 mM Tris–HCl, pH 7.5), 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, and 1 mM EGTA. Rat brain tissue was obtained from 3-week-old animals and solubilized for 30 min at


266 | I. M. Gonza´lez-Gonza´lez et al.

4C in RIPA buffer (50 mM Tris–HCl, pH 7.5, 1% Nonidet P40, 0.5% sodium deoxycholate, 150 mM NaCl, 1 mM EGTA, 2 mM EDTA, and 0.1 mM dithiothreitol). The solubilized material was centrifuged at 10 000 g for 20 min and the supernatant was incubated overnight at 4C with the desired antibodies (2 lg/mL). Subsequently, 40 lL of a protein A cross-linked to agarose beads (Sigma, St Louis, MO, USA) was added and the mixture was incubated for 1 h at 4C with constant rotation. The beads were washed five times with ice-cold lysis buffer before adding SDS– PAGE sample buffer to each sample (25 lL). The bound proteins were dissociated from the beads by heating at 92C for 3 min before they were resolved by SDS–PAGE on 7.5% gels as described above. Cell surface biotinylation Cell surface proteins were labeled with the non-permeable sulfo-NHS-SS-Biotin reagent (1 mg/mL) as described previously (Cubelos et al. 2005). Biotinylated proteins were separated by SDS–PAGE and immunoblotted. Biotinylated HA-GLT1 was visualized with an anti-HA antiserum. Transport assays [3H]Glutamate transport in transfected COS7 cells was measured as described previously (Cubelos et al. 2005). Protein determination Protein concentration was determined using the BioRad protein determination kit (Hercules, CA, USA) with bovine serum albumin as the standard. Primary rat brain cultures Primary cortical neurons were prepared from embryonic day 17 Wistar rats using the papain dissociation system (Worthington Biochemical). Briefly, after dissection of the cortex, the tissue was rinsed twice in Earle’s balanced salt solution at 4C, and then the cells were dissociated by incubation with a papain solution (20 U/ mL) in Earle’s balanced salt solution. The tissue was triturated with a fire-polished Pasteur pipette, the papain was then inhibited with an albumin-ovomucoid solution, and the cells were collected by centrifugation at 800 g. The cell pellet was resuspended in Neurobasal medium supplemented with B27 (NB-B27; Invitrogen, Gaithersburg, MD, USA) and 5% FCS, and the cells were counted in the presence of trypan blue before plating them at 7.5 · 104 cells/ cm2 in 24-well plates containing glass coverslips pre-coated with poly-D-lysine (25 lg/mL). Four hours later, the FCS was diluted by the addition of one volume of NB-B27 and the cells were maintained at 37C in a humidified atmosphere containing 5% CO2. Half the medium was replaced twice weekly with fresh NBB27 medium containing glutamax. By 14 days in vitro the cultures consisted of a mixed population of neurons (60–70% MAP2positive cells) and glial cells (30–40% GFAP-positive cells). Immunofluorescence in cultured cells and colocalization analysis Primary cultures or Madin-Darby canine kidney (MDCK) cells grown on poly-D-lysine-treated glass coverslips were transfected with the corresponding expression vectors using Lipofectamine 2000 according to the manufacturers’ instructions. Immunofluorescence was performed as described previously (Cubelos et al. 2005). For detection of the endogenous transporter in primary cultures,

triple-labeling experiments with anti-GLT1a, anti-GLT1b, and antiMAP2 (neuronal marker) or anti-GFAP (glial marker) were performed. For colocalization analysis, images were captured in a confocal microscope LSM510 coupled to an inverted microscope (Axiovert200M, Zeiss, Jena, Germany) using a 63·/1.4 oil PlanApochromat objective (Zeiss). Background was corrected using the subtraction function of ImageJ (rsb.info.nih.gov/ij/index.html). Background intensity levels in each channel were calculated using the ROI Manager function of ImageJ as means values plus three times the standard deviation in the selected area. Regions of interest were selected based on patching patterns of both green and red channels. Areas containing staining artifacts, as indicated by dispersed patterns of fluorescence, were excluded from analysis. The colocalization analysis was performed on paired images using the JACoP plug-in (Bolte and Cordelie`res 2006). Pearson’s correlation and overlap coefficients according to Mander’s were calculated from those images. These analyses quantify the correlation or overlap between GLT1a and GLT1b signals (Bolte and Cordelie`res 2006; Zinchuk and Zinchuk 2008).

Results GLT1a and GLT1b form hetero-oligomers in transfected cells and in primary cultures from rat brain As GLT1 has an oligomeric structure, we investigated whether hetero-oligomers could form between the main isoform of this transporter, GLT1a, and the alternatively spliced GLT1b variant. Accordingly, we performed immunoprecipitation assays in extracts from COS7 cells that had been co-transfected with differentially tagged forms of these transporters, HA-GLT1a and YFP-GLT1b. In preliminary experiments, the inclusion of YFP and HA tags did not appear to affect the glutamate transport characteristics of these isoforms (Fig. S1). Protein complexes were immunoprecipitated with antibodies against one of the tags and the proteins recovered were analyzed by probing immunoblots with antibodies raised against the other tag. Indeed, both YFP-GLT1b (Fig. 1, upper panels) and HA-GLT1a (Fig. 1a, lower panels) were evident in complexes isolated with the anti-HA antibody. It is unlikely that this association occurred after solubilization as similar results were obtained when the cross-linker reagent DSS was incubated with the cells prior to the solubilization step in order to stabilize pre-existing complexes (data not shown). Indeed, GLT1 was preferentially found as oligomers in the presence of cross-linker, suggesting that most of the monomers observed in our experiments were produced by dissociation of oligomers during the manipulation of the samples (see technical comment in Fig. S2a). As an additional control to discard the possibility of a spurious association during the immunoprecipitation procedure, extracts of cells singly transfected with HA-GLT1a or with YFP-GLT1b were mixed together and then subjected to immunoprecipitation with anti-HA antibodies. Under these conditions, these proteins did not



co-immunoprecipitate (Fig. S2b), while HA-GLT1a and YFP-GLT1b co-immunoprecipitated when they had been co-transfected (Fig. S2b and Fig. 1).

To investigate the subcellular localization of both isoforms in transfected cells we co-expressed HA-GLT1a and YFPGLT1b in COS7 cells. In these cells, HA-GLT1a and YFP-GLT1b reached the cell surface as they resulted labeled in a biotinylation assay using the non-permeant reagent sulfo-NHS-SS-Biotin (Fig. 2a). Immunofluorescence assays (a)

Fig. 1 Hetero-oligomerization of GLT1a and GLT1b in transfected cells. COS7 cells were transfected with expression vectors for HAGLT1a and YFP-GLT1b. Two days later, the cells were solubilized and the protein extracts were immunoprecipitated with anti-HA antibodies (IP:a-HA). A sample of the lysates (Lysates) and the immunocomplexes recovered were resolved by SDS–PAGE and immunoblotted to membranes that were probed with anti-green fluorescent protein (WB:a-GFP) and anti-HA antibodies (WB:a-HA). The protein bands were visualized by enhanced chemiluminescence. Note that the transporter appears as monomers (mY-GLT1b, mGLT1a) and oligomers (oY-GLT1b, oGLT1). Bands labeled with an asterisk probably correspond to immature, partially glycosylated forms of the protein.

Fig. 2. Delivery of GLT1 isoforms to the cell surface. (a) COS7 cells were transfected with the constructs indicated and the cell surface proteins were labeled with the sulfo-NHS-SS-Biotin reagent for 30 min at 0C. Subsequently, the biotinylated proteins were isolated with streptavidin-agarose beads. Samples of the lysates (Lysates) or the biotinylated proteins (Biotinylated) were analyzed by immunoblotting with anti-green fluorescent protein (GFP) to detect YFP-GLT1b (WB:aGFP, Y-GLT1b) or anti-HA to detect HA-GLT1a (WB:a-HA, GLT1a). The protein bands were visualized by enhanced chemiluminescence. The asterisk labels an unspecific band produced occasionally by the anti-HA antibody. (b,c) Co-localization of YFP-GLT1a and HA-GLT1b in transfected COS7 cells (b) or non-polarized MDCK cells (c). Cells were transfected and 2 days later they were fixed and immunostained with anti-HA antibodies. Fluorescence for YFP-GLT1a [b(i), c(i), green] and HA-GLT1b [b(ii), c(ii), red] was visualized by confocal microscopy. Co-localization at some clusters is denoted by the yellow color in the merged image [b(iii), c(iii), arrows].

(b) (i)

(ii)

(iii)

(c) (i)

(ii)

(iii)



(a)

(b) (i)

(ii)

(iii)

(ii)

(iii)

(c) (i)

revealed that both isoforms had a tendency to appear co-clustered in the cell surface of co-transfected COS7 cells, while in the intracellular compartment the co-localization was partial (Fig. 2b). Similar immunofluorescence images were obtained in non-polarized MDCK transfected with YFP-GLT1b and HA-GLT1a (Fig. 2c) or with YFP-GLT1a and HA-GLT1b (Fig. S3). In this cell line most of the transporter appeared in the plasma membrane, identified by immunoreactivity with the monoclonal antibody p58 (against the b-subunit of the Na+K+-ATPase) (Fu¨llekrug et al. 2006). Evidence for the formation of hetero-oligomers in native tissue was obtained in primary cultures of fetal rat brain. Immunoprecipitation with anti-GLT1a of primary cells cultured for 2 weeks isolated protein complexes that contained GLT1b, while no such immunoreactive complexes were isolated with pre-immune serum (Fig. 3a, left). As a control of the efficiency of the immunoprecipitation, a fraction of the immunoprecipitated material was electroblotted in parallel and probed with anti-GLT1a (Fig. 3a, right). Previous experiments demonstrated the specificity of these antibodies. and as they were raised against the C-terminal isoform-specific sequences, they displayed no cross-reactivity (Gonzalez-Gonzalez et al. 2008). Immunofluorescence assays were also compatible with the existence of oligomers (Fig. 3b), whereby isoform-specific antibodies revealed a punctate distribution of GLT1a and GLT1b on subpopulations of neurites in primary cultures of neurons in accordance with previous observations (Chen et al. 2002; Bassan et al. 2008). Double-labeling experiments showed that GLT1a and GLT1b co-localized in a number of puncta along these neurites (identified by anti-MAP2 staining in the blue Table 1 Colocalization of GLT1a with GLT1b in primary cultures

Fig. 3 Hetero-oligomerization of GLT1a and GLT1b in primary cultures. (a) Primary cultures from rat brain were solubilized, cleared by centrifugation, and immunoprecipitated (IP) with anti-GLT1a (aGLT1a) or with non-preimmune serum (PI). A sample of the lysate (Lys) or the IP material was resolved by SDS–PAGE and transferred to membranes that were probed with anti-GLT1b (WB:a-GLT1b). To test the efficiency of the IP, a sample of the IP material was analyzed in parallel and probed with anti-GLT1a (WB:a-GLT1a). The protein bands were visualized by enhanced chemiluminescence. The position of monomers (mGLT1a/b) and oligomers (oGLT1a/b) is indicated. (b,c) Co-localization of GLT1a and GLT1b in primary cultures of fetal rat brain. Cultures were maintained for 14 days in vitro, fixed, and stained with guinea pig anti-GLT1a and rabbit anti-GLT1b antibodies that were detected with fluorescent secondary antibodies (Alexa 594 anti-guinea pig and Alexa 488 anti-rabbit). Images for GLT1a [b(i), c(i), red] and GLT1b [b(ii), c(ii), green] were collected on a confocal microscope and the yellow puncta in the merged image indicate co-localization [b(iii), c(iii) arrows]. Neurons (b) or astrocytes (c) in the primary cultures were identified by anti-Microtubule-associated protein 2 (MAP2) or anti-GFAP staining, respectively, in the blue channel (not shown).

Cell type

Pearson’s coefficient (mean ± SD)

Manders coefficient (mean ± SD)

Neurons Astrocytes

0.66 ± 0.10 0.63 ± 0.11

0.71 ± 0.09 0.76 ± 0.10

Primary cultures maintained for 14 days in vitro were fixed and processed for immunofluorescence as indicated in Materials and Methods, using the following primary antibodies: anti-GLT1a (secondary in red) plus anti-GLT1b (secondary in green) and either anti-GFAP or anti-Microtubule-associated protein 2 (MAP2) (secondary in blue) to identify astrocytes or neurons, respectively. Images from the three channels were acquired by confocal microscopy. Overlaps between green and red fluorescence from three regions of at least five fields were calculated by determining the Pearson’s and Manders coefficient. The Pearson’s coefficient describes the correlation of the intensity distribution between channels and varies from )1 to +1. Values over +0.5 are considered as indicative of colocalization. The Manders coefficient indicates the actual overlap of the signals and values over 0.6 are considered to represent colocalization (Bolte and Cordelie`res 2006; Zinchuk and Zinchuk 2008).



channel, not shown). Primary cultures also contained glial cells (about 30–40% of the cells were immunoreactive to anti-GFAP). In glial cells immunoreactivity for both proteins also appeared in clusters in the cell surface and in the intracellular compartment (Fig. 3c). To obtain a measure of colocalization of GLT1a and GLT1b in these primary cultures, images of immunoreacted cells were evaluated by determining the Pearson’s correlation coefficient and the overlap coefficient according to Manders (Table 1). Values obtained for both coefficient support the existence of colocalization in both neuronal and glial cells (Bolte and Cordelie`res 2006; Zinchuk and Zinchuk 2008). Altogether, these results support the existence of oligomeric structures containing both GLT1 isoforms, either in transfected cells or in primary cultures of cells derived from the rat brain. In the rat brain, GLT1a is present in protein complexes with GLT1b, PSD-95, and NMDAR To determine whether hetero-oligomers exist in the rat brain, the material immunoprecipitated from a solubilized tissue (a)

(b)

Fig. 4 GLT1a associates with GLT1b, PSD-95, and NMDAR in the rat brain. (a) Cell extracts from the rat forebrain were immunoprecipitated (IP) with the anti-GLT1a antiserum (aGLT1a) or with pre-immune serum (PI). The proteins recovered and a sample of the lysate (Lys) was resolved by SDS–PAGE and immunoblotted. As indicated in the figure, the membranes were probed with antibodies against GLT1b, PSD-95, NMDAR1, or GLT1a (WB), and their binding was visualized by enhanced chemiluminescence. (b) Extracts were processed as in (a) but IP with anti-GLT1b (aGLT1b) and the blots were probed for

extract with the anti-GLT1a antibody was immunoblotted and probed with an antibody raised against GLT1b. The band pattern obtained in immunoblots showed a diffuse band of 60–65 kDa (GLT1b monomers) as well as higher molecular weight bands that probably corresponded to different oligomeric aggregates, whereas no bands were evident when the same material was immunoprecipitated with pre-immune serum (Fig. 4a). The presence of GLT1a in these complexes was confirmed by reprobing these blots with anti-GLT1a. Moreover, when these membranes were also probed with antibodies against PSD-95 and NR1, bands of the mobility expected for these two proteins were detected but not in material immunoprecipitated with pre-immune serum. Indeed, immunocomplexes containing GLT1b, PSD-95, and NMDAR have already been detected in similar conditions (Fig. 4b and Gonzalez-Gonzalez et al. 2008). The reverse experiment, using anti-PSD-95 to immunoprecipitate this protein in rat brain extracts allowed the isolation of immunocomplexes containing both GLT1a, GLT1b and, as expected, PSD-95 (Fig. 4c). Together, these experiments and (c)

GLT1a followed by GLT1b. (c) Extracts were processed as in (a) but IP with anti-PSD-95 (a-PSD95) and the blots were probed for GLT1b followed by GLT1a and PSD95. Note that GLT1a/b produced a number of bands that presumably correspond to monomers (m) and to at least two forms of oligomeric aggregates (o and o¢). Occasionally monomers appeared as doublet bands (asterisks). It is unclear whether this corresponded to partial degradation or to partially glycosylated intermediates. Additional bands of unclear origin were also occasionally observed in IP material (#).



those reported previously supported the existence of GLT1a– GLT1b hetero-oligomer complexes in the rat brain, and these complexes were associated with components of the PSDs like PSD-95 and NMDAR. GLT1a requires GLT1b to interact with PSD-95 and NMDAR Post-synaptic density 95 contains three PDZ domains, a SH3 domain, and a GK domain that are implicated in a large variety of protein–protein interactions. Indeed, the C terminus of the GLT1b isoform can interact with one of the PDZ domains of PSD-95 and this indirectly enables GLT1b to associate with NMDAR (Gonzalez-Gonzalez et al. 2008). However, as GLT1a does not appear to contain any PDZinteracting motifs, we analyzed the possibility that the formation of GLT1a and GLT1b hetero-oligomers might induce an indirect interaction between HA-GLT1a and mycPSD-95 in transfected COS7 cells. Anti-HA antiserum only immunoprecipitated myc-PSD-95 when GLT1b (YFP tagged) was present (Fig. 5a, lanes 4 and 5). As expected, in immunocomplexes precipitated with the anti-HA antiserum neither myc-PSD-95 nor YFP-GLT1b was observed in the absence of HA-GLT1a (Fig. 5a, lane 6). It was previously reported that GLT1b interacted with PSD-95 through the PDZ interacting motif present in its C-terminus (GonzalezGonzalez et al. 2008). This motif was also required for the interaction of the hetero-oligomer with PSD-95 as the substitution of YFP-GLT1b for a C-terminal truncated

(a)

mutant (YFP-GLT1bD) prevented the formation of immunocomplexes with the scaffold protein (Fig. 5b, compare lanes 4 and 5). As expected, in the absence of HA-GLT1a, anti-HA antibodies did not immunoprecipitate any protein (Fig. 5b, lane 6). When we studied the interaction of HA-GLT1a with NMDAR (Fig. 6), immunocomplexes isolated with anti-HA only contained NMDAR in the presence of mycPSD-95 and YFP-GLT1b (Fig. 6, lane 4) but not in the absence of either of these two proteins (Fig. 6, lanes 5 and 6). Thus, the results obtained here confirm that GLT1a requires GLT1b in order to associate with complexes containing PSD-95 and NMDARs. YFP-GLT1a is recruited to clusters where it co-localizes with PSD-95 in transfected neurons As PSD-95 is a neuronal protein located in dendritic spines, we examined the subcellular distribution of GLT1a when expressed in neurons. YFP-GLT1a was transfected in 14 days in vitro primary cortical neurons and its distribution was analyzed by microscopy 5 days later. YFP-GLT1a was found at the cell surface and had a tendency to concentrate in clusters along the dendritic tree (Fig. 7a), a localization that was compatible with it accumulating in dendritic spines and that was similar to that observed for YFP-GLT1b (Fig. 7b and Gonzalez-Gonzalez et al. 2008). When YFP-GLT1a was co-expressed with GLT1b and mycPSD-95, co-localization of the three proteins in clusters was evident along the dendrites (Fig. 7c–f and insets), supporting the idea that a

(b)

Fig. 5 Association of GLT1a, GLT1b, and PSD-95 in transfected COS7 cells. (a,b) COS7 cells were transfected with expression vectors for HA-GLT1a, YFP-GLT1b, YFP-GLT1bD, and myc-PSD-95, as indicated in the figure. When necessary, pCDNA3 was also included to equalize the amount of DNA per assay. After solubilization, HA-GLT1a was immunoprecipitated (IP) with anti-HA antiserum (IP:a-HA) and the precipitated proteins were analyzed by immunoblotting. The mem-

branes were probed with antisera against PSD-95, green fluorescent protein (GFP), and HA (WB). The lysates were also analyzed to control transfection efficiency (lysates). Arrows refer to the position in the blots of PSD-95, monomers of GLT1 (mGLT1a/b), oligomers of GLT1 (oGLT1a/b), and IgGs (IgG). Asterisks in (a) refer to unspecific bands as they also appear in the IP control (lane 6).



Fig. 6 Association of GLT1a, GLT1b, PSD-95, and NMDARs in transfected COS7 cells. COS7 cells were transfected with expression vectors for the proteins indicated. To prevent excitotoxicity mediated by NMDAR expression, 1 mM kynurenic acid plus 100 lM AP5 were added to the cultures. After 2 days, the cells were solubilized and HAGLT1a was immunoprecipitated (IP) with the anti-HA antiserum. Samples of the cell lysates (lysates) and the IP proteins (IP:a-HA) were analyzed by immunoblotting, and the membranes were probed with antibodies against PSD-95, NR2A, green fluorescent protein (GFP), and HA as indicated in the figure (WB:a-HA). The lysates represent 5% of the total extract that was used in each assay. Note the smaller amount of NR2A in lysates where PSD-95 was absent, probably because of a loss in stability of the receptor in the absence of the scaffold protein. Arrows refer to the position in the blots of PSD-95, NR2 subunit of NMDAR, monomers of GLT1 (mGLT1a/b), and oligomers of GLT1 (oGLT1a/b).

fraction of GLT1a (and GLT1b) might be concentrated in the post-synaptic aspect of glutamatergic synapses.

Discussion In this study we found that the two GLT1 isoforms, GLT1a and GLT1b, can form hetero-oligomers in heterologous expression systems, in primary cultures from fetal rat and in the adult rat brain. Moreover, while GLT1a itself is unable to interact with proteins containing PDZ domains, its associa-

tion with GLT1b enables it to form complexes with a prototypic PDZ protein-like PSD-95 and presumably with proteins bound to this scaffold protein. Indeed, in the rat brain and in heterologous systems, GLT1a appears to associate indirectly with subunits of the NMDAR. Oligomerization is a common feature of glutamate transporters, and trimeric arrangements seem to be the predominant organization that is conserved from archaebacteria to mammals. However, this arrangement does not seem to be related to the mechanism of glutamate transport as each subunit contains all the elements necessary to drive the translocation of the substrate, and each protomer functions independently of the others (Grewer et al. 2005; Koch and Larsson 2005). Nevertheless, the evidence that oligomerization might affect the uncoupled movement of anions mediated by these transporters is still not conclusive (Torres-Salazar and Fahlke 2006; Koch et al. 2007; Leary et al. 2007). Thus, in the absence of a clear effect on the transport cycle itself, oligomerization might be necessary for other aspects of the behavior of these proteins such as their intracellular trafficking and stability. It remains unclear from our data whether hetero-oligomerization occurs early in the biosynthetic pathway or later at the plasma membrane. There is evidence that the GLT1a isoform exits the endoplasmic reticulum as an oligomer, and that oligomerization may serve to occlude a diarginine endoplasmic reticulum (ER) retention signal (Kalandadze et al. 2004). In accordance with recent structural data derived from the archeal GltPh transporter, this diarginine motif resides in the hairpin 2 loop (HP2) close to the substrate binding site. However, it is unclear whether oligomerization of GLT1a is necessary for the occlusion of this putative ER retention motif. Alternatively, this motif might act as a sensor for folding and as such, oligomerization may be required to ensure correct folding. This motif is located in the common region of GLT1a and GLT1b, which suggests that its influence on the exit from the ER would be similar for GLT1a homo-oligomers, GLT1b homo-oligomers, or GLT1a-GLT1b hetero-oligomers, assuming that the latter were formed early in the ER. The existence of GLT1a–GLT1b hetero-oligomers might be a means of generating molecular diversity of this transporter, increasing the repertoire of regulatory mechanisms that could control its activity. Indeed, PICK1, a PDZ protein that regulates the traffic of many synaptic proteins, interacts with the C-terminus of GLT1b which favors the trafficking of GLT1b to the cell surface and that may modulate the protein kinase C-dependent regulation of this glutamate transporter (Bassan et al. 2008). Indeed, in accordance with the data presented here, immunocomplexes containing GLT1b and PICK1 also contained GLT1a, supporting the existence of hetero-oligomers of both these isoforms (Bassan et al. 2008). During the revision of this article another study was published reporting the formation



(b)

(a)

(c)

(d)

(e)

(f)

(i)

(i)

(i)

(i)

of hetero-oligomers that might contain other splicing variant of GLT1 (Peacey et al. 2009). These arrangements are reminiscent of a number of neurotransmitter receptors that are hetero-oligomers whereby each subunit contains specific regulatory sites in their cytoplasmic domains that are targeted by distinct intracellular signaling pathways, or that interact with scaffold proteins which in turn bind signaling or cytoskeletal proteins (e.g., kinases, phosphatases). These multiprotein complexes influence several aspects of receptor function, including gating, trafficking, and their stabilization at synapses. Through its association with GLT1b, GLT1a acquires the capacity to interact with the scaffold protein PSD-95, and it is included in macromolecular complexes that also contain NMDAR. This was evident in heterologous expression systems as well as in solubilized rat brain tissue, indicating that this association might occur in vivo. However, this interpretation does not conform with previous immunohistochemical data that initially detected GLT1a in glial cells and more recently in neurons, albeit exclusively in the presynaptic aspect of asymmetric synapses (Danbolt et al. 1992; Torp et al. 1994; Lehre et al. 1995; Chen et al. 2004; Melone et al. 2009). However, neuronal GLT1 has been difficult to visualize as it is very sensitive to fixation, which may occlude the access of antibodies to their respective epitopes. This issue might still preclude the immunohistochemical detection of GLT1 at PSDs, where the tight packing of proteins and aldehyde cross-linking is known to interfere with the detection of several other proteins by conventional histochemical methods (Fukaya and Watanabe 2000; Burette et al. 2002). However, in addition to the biochemical data

Fig. 7 Expression of GLT1a and GLT1b in clusters and their co-localization with PSD95 in transfected neurons. (a,b) Primary cells of the fetal rat cortex were maintained in culture for 15 days and they were transfected with YFP-GLT1a (a) or YFP-GLT1b (b). Three days later, the cells were fixed and visualized by confocal microscopy. Note the presence of both GLT1 isoforms in clusters (arrows) that give the dendrites a spiny aspect. (c–f) Neurons were transfected as in (a) with expression vectors for YFP-GLT1a, GLT1b, and myc-PSD-95 before they were fixed with cold methanol, stained with anti GLT1b and anti-PSD-95, and visualized by confocal microscopy. The inset is magnified in c(i)–f(i). Co-localization at specific puncta produced a white color in the merged image, as indicated (arrows).

reported here, there are two other lines of evidence that support the existence of GLT1 in PSDs. First, GLT1 peptides have been identified in a proteomic analysis of isolated PSDs (Yoshimura et al. 2004), and second, GLT1a transfected into cortical neurons accumulated in dendrites and spines, supporting the existence of a sorting signal in GLT1a that targets this protein to dendritic spines. We also have observed a similar enrichment of GLT1b in dendritic spines (herein and Gonzalez-Gonzalez et al. 2008) and thus, transfected YFP-GLT1a might be targeted to the spines indirectly through its association with endogenous GLT1b in these neurons. However, the PDZ-interacting C-terminus was not required for this dendritic enrichment of GLT1b (GonzalezGonzalez et al. 2008), suggesting that the hypothetical sorting motif may be located in regions common to both isoforms. Likewise, metabotropic glutamate receptor, mGluR1a, accumulates in dendritic spines independently of the PDZ interaction motif they contain (Das and Banker 2006). As both isoforms are also expressed in glial cells, it is tempting to speculate that they form hetero-oligomers in these cells as well, and that they might also be anchored to macromolecular glial complexes organized around scaffold proteins with PDZ domains. Perhaps the most functionally relevant macromolecular complex in astrocytes is organized around the dystrophin scaffold and it is responsible for regulating water and potassium levels in the brain. This complex contains both aquaporin 4 and the potassium channel Kir4.1 which are anchored to dystrophin through the PDZ domains of syntrophin (Fort et al. 2008). Recent evidence also suggests that GLT1a is present in these



complexes (Hinson et al. 2008) but it is unknown whether GLT1b plays any role in this interaction, a possibility that deserves further study. In summary, we show here that the major glutamate transporter isoform in the brain, GLT1a, can interact with the less abundant one, GLT1b, to form hetero-oligomers. These hetero-oligomers are capable of interacting with scaffold proteins containing PDZ domains and thus, they may be involved in regulating the removal of glutamate from glutamatergic synapses.

Acknowledgments We would like to thank E. Nuń˜ez for expert technical help and members of the Confocal Microscope Department for helping with the image quantifications. We also wish to thank Drs. Baruch Kanner, Johannnes Hell, and Juan Lerma for generously providing us with plasmids. This work was supported by grants from the Spanish ‘Direccioń General de Investigacioń Cientı´fica y Tećnica’ (SAF2008-01059), ‘Comunidad Autońoma de Madrid’, CIBERER, and by an institutional grant from the ‘Fundacioń Ramoń Areces’.

Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. Effect of the inclusion of YFP and HA tags on the activity and trafficking of GLT1. Figure S2. Preferential dissociation of oligomers in lysates versus immunoprecipitated samples. Figure S3. Localization of YFP-GLT1a and HA-GLT1b in the plasma membrane of transfected MDCK cells. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

References Bassan M., Liu H., Madsen K. L. et al. (2008) Interaction between the glutamate transporter GLT1b and the synaptic PDZ domain protein PICK1. Eur. J. Neurosci. 27, 66–82. Berger U. V., DeSilva T. M., Chen W. and Rosenberg P. A. (2005) Cellular and subcellular mRNA localization of glutamate transporter isoforms GLT1a and GLT1b in rat brain by in situ hybridization. J. Comp. Neurol. 492, 78–89. Bolte S. and Cordelie`res F. P. (2006) A guided tour into subcellular colocalization analysis in light microscopy. J. Microsc. 224, 213– 232. Burette A., Zabel U., Weinberg R. J., Schmidt H. H. and Valtschanoff J. G. (2002) Synaptic localization of nitric oxide synthase and soluble guanylyl cyclase in the hippocampus. J. Neurosci. 22, 8961–8970. Chen W., Aoki C., Mahadomrongkul V., Gruber C. E., Wang G. J., Blitzblau R., Irwin N. and Rosenberg P. A. (2002) Expression of a variant form of the glutamate transporter GLT1 in neuronal cultures and in neurons and astrocytes in the rat brain. J. Neurosci. 22, 2142–2152.

Chen W., Mahadomrongkul V., Berger U. V., Bassan M., DeSilva T., Tanaka K., Irwin N., Aoki C. and Rosenberg P. A. (2004) The glutamate transporter GLT1a is expressed in excitatory axon terminals of mature hippocampal neurons. J. Neurosci. 24, 1136–1148. Cubelos B., Gonzalez-Gonzalez I. M., Gimenez C. and Zafra F. (2005) The scaffolding protein PSD-95 interacts with the glycine transporter GLYT1 and impairs its internalization. J. Neurochem. 95, 1047–1058. Danbolt N. C., Storm-Mathisen J. and Kanner B. I. (1992) An [Na+ + K+]coupled L-glutamate transporter purified from rat brain is located in glial cell processes. Neuroscience 51, 295–310. Das S. S. and Banker G. A. (2006) The role of protein interaction motifs in regulating the polarity and clustering of the metabotropic glutamate receptor mGluR1a. J. Neurosci. 26, 8115–8125. Fort P. E., Sene A., Pannicke T. et al. (2008) Kir4.1 and AQP4 associate with Dp71- and utrophin-DAPs complexes in specific and defined microdomains of Muller retinal glial cell membrane. Glia 56, 597– 610. Fukaya M. and Watanabe M. (2000) Improved immunohistochemical detection of postsynaptically located PSD-95/SAP90 protein family by protease section pretreatment: a study in the adult mouse brain. J. Comp. Neurol. 426, 572–586. Fu¨llekrug J., Shevchenko A., Shevchenko A. and Simons K. (2006) Identification of glycosylated marker proteins of epithelial polarity in MDCK cells by homology driven proteomics. BMC Biochem. 13, 7–8. Gendreau S., Voswinkel S., Torres-Salazar D., Lang N., Heidtmann H., Detro-Dassen S., Schmalzing G., Hidalgo P. and Fahlke C. (2004) A trimeric quaternary structure is conserved in bacterial and human glutamate transporters. J. Biol. Chem. 279, 39505–39512. Gonzalez-Gonzalez I. M., Garcia-Tardon N., Cubelos B., Gimenez C. and Zafra F. (2008) The glutamate transporter GLT1b interacts with the scaffold protein PSD-95. J. Neurochem. 105, 1834–1848. Grewer C., Balani P., Weidenfeller C., Bartusel T., Tao Z. and Rauen T. (2005) Individual subunits of the glutamate transporter EAAC1 homotrimer function independently of each other. Biochemistry 44, 11913–11923. Grewer C., Gameiro A., Zhang Z., Tao Z., Braams S. and Rauen T. (2008) Glutamate forward and reverse transport: from molecular mechanism to transporter-mediated release after ischemia. IUBMB Life 60, 609–619. Haugeto O., Ullensvang K., Levy L. M., Chaudhry F. A., Honore T., Nielsen M., Lehre K. P. and Danbolt N. C. (1996) Brain glutamate transporter proteins form homomultimers. J. Biol. Chem. 271, 27715–27722. Hinson S. R., Roemer S. F., Lucchinetti C. F., Fryer J. P., Kryzer T. J., Chamberlain J. L., Howe C. L., Pittock S. J. and Lennon V. A. (2008) Aquaporin-4-binding autoantibodies in patients with neuromyelitis optica impair glutamate transport by down-regulating EAAT2. J. Exp. Med. 205, 2473–2481. Kalandadze A., Wu Y., Fournier K. and Robinson M. B. (2004) Identification of motifs involved in endoplasmic reticulum retentionforward trafficking of the GLT-1 subtype of glutamate transporter. J. Neurosci. 24, 5183–5192. Kim E. and Sheng M. (2004) PDZ domain proteins of synapses. Nat. Rev. Neurosci. 5, 771–781. Koch H. P. and Larsson H. P. (2005) Small-scale molecular motions accomplish glutamate uptake in human glutamate transporters. J. Neurosci. 25, 1730–1736. Koch H. P., Brown R. L. and Larsson H. P. (2007) The glutamateactivated anion conductance in excitatory amino acid transporters is gated independently by the individual subunits. J. Neurosci. 27, 2943–2947.



Leary G. P., Stone E. F., Holley D. C. and Kavanaugh M. P. (2007) The glutamate and chloride permeation pathways are colocalized in individual neuronal glutamate transporter subunits. J. Neurosci. 27, 2938–2942. Lehre K. P., Levy L. M., Ottersen O. P., Storm-Mathisen J. and Danbolt N. C. (1995) Differential expression of two glial glutamate transporters in the rat brain: quantitative and immunocytochemical observations. J. Neurosci. 15, 1835–1853. Maragakis N. J., Dykes-Hoberg M. and Rothstein J. D. (2004) Altered expression of the glutamate transporter EAAT2b in neurological disease. Ann. Neurol. 55, 469–477. Melone M., Bellesi M. and Conti F. (2009) Synaptic localization of GLT1a in the rat somatic sensory cortex. Glia 57, 108–117. Peacey E., Miller C. C., Dunlop J. and Rattray M. (2009) The four major N- and C-terminal splice variants of the excitatory amino acid transporter GLT-1 form cell surface homomeric and heteromeric assemblies. Mol. Pharmacol. 75, 1062–1073. Reye P., Sullivan R., Scott H. and Pow D. V. (2002) Distribution of two splice variants of the glutamate transporter GLT-1 in rat brain and pituitary. Glia 38, 246–255. Rothstein J. D., Dykes-Hoberg M., Pardo C. A. et al. (1996) Knockout of glutamate transporters reveals a major role for astroglial transport in excitotoxicity and clearance of glutamate. Neuron 16, 675– 686. Schmitt A., Asan E., Lesch K. P. and Kugler P. (2002) A splice variant of glutamate transporter GLT1/EAAT2 expressed in neurons: cloning and localization in rat nervous system. Neuroscience 109, 45–61. Sullivan R., Rauen T., Fischer F., Wiessner M., Grewer C., Bicho A. and Pow D. V. (2004) Cloning, transport properties, and differential

localization of two splice variants of GLT-1 in the rat CNS: implications for CNS glutamate homeostasis. Glia 45, 155–169. Tanaka K., Watase K., Manabe T. et al. (1997) Epilepsy and exacerbation of brain injury in mice lacking the glutamate transporter GLT1. Science 276, 1699–1702. Tomita S., Nicoll R. A. and Bredt D. S. (2001) PDZ protein interactions regulating glutamate receptor function and plasticity. J. Cell Biol. 153, F19–F24. Torp R., Danbolt N. C., Babaie E., Bjoras M., Seeberg E., StormMathisen J. and Ottersen O. P. (1994) Differential expression of two glial glutamate transporters in the rat brain: an in situ hybridization study. Eur. J. Neurosci. 6, 936–942. Torres-Salazar D. and Fahlke C. (2006) Intersubunit interactions in EAAT4 glutamate transporters. J. Neurosci. 26, 7513–7522. Yernool D., Boudker O., Folta-Stogniew E. and Gouaux E. (2003) Trimeric subunit stoichiometry of the glutamate transporters from Bacillus caldotenax and Bacillus stearothermophilus. Biochemistry 42, 12981–12988. Yernool D., Boudker O., Jin Y. and Gouaux E. (2004) Structure of a glutamate transporter homologue from Pyrococcus horikoshii. Nature 431, 811–818. Yoshimura Y., Yamauchi Y., Shinkawa T., Taoka M., Donai H., Takahashi N., Isobe T. and Yamauchi T. (2004) Molecular constituents of the postsynaptic density fraction revealed by proteomic analysis using multidimensional liquid chromatography-tandem mass spectrometry. J. Neurochem. 88, 759–768. Zinchuk V. and Zinchuk O. (2008) Quantitative colocalization analysis of confocal fluorescence microscopy images. Curr. Protoc Cell Biol. Chapter 4, 4.19.1–4.19.16.
