The non-adrenergic imidazoline-1 receptor protein nischarin is a key regulator of astrocyte glutamate uptake

Summary Astrocytic GLT-1 is the main glutamate transporter involved in glutamate buffering in the brain, pivotal for glutamate removal at excitatory synapses to terminate neurotransmission and for preventing excitotoxicity. We show here that the surface expression and function of GLT-1 can be rapidly modulated through the interaction of its N-terminus with the nonadrenergic imidazoline-1 receptor protein, Nischarin. The phox domain of Nischarin is critical for interaction and internalization of surface GLT-1. Using live super-resolution imaging, we found that glutamate accelerated Nischarin-GLT-1 internalization into endosomal structures. The surface GLT-1 level increased in Nischarin knockout astrocytes, and this correlated with a significant increase in transporter uptake current. In addition, Nischarin knockout in astrocytes is neuroprotective against glutamate excitotoxicity. These data provide new molecular insights into regulation of GLT-1 surface level and function and suggest new drug targets for the treatment of neurological disorders.


INTRODUCTION
Glutamate transport into cells is mediated by excitatory amino acid transporters (EAATs). Of the five EAAT subtypes found in the CNS, EAAT2/GLT-1 is predominantly expressed in astrocytes and is a major means for glutamate clearance from the extracellular space (Danbolt, 2001). Glutamate buffering through transporter binding helps to maintain a low extracellular glutamate concentration that facilitates termination of fast excitatory synaptic transmission (Diamond and Jahr, 1997;Wadiche et al., 1995aWadiche et al., , 1995bTong and Jahr, 1994). Moreover, lateral diffusion of surface GLT-1 can also regulate glutamate clearance and thus shape glutamatergic neurotransmission (Al Awabdh et al., 2016;Murphy-Royal et al., 2015). The transporters have a long transport cycle ($70 ms) (Wadiche et al., 1995a) compared to the timescale of glutamate presence at the synapse ($1 ms) (Barbour and Hausser, 1997). However, despite this long transport cycle, efficient glutamate clearance occurs during synaptic activity because of the high surface density of transporters (Danbolt, 2001). A low extracellular glutamate concentration (below the submicromolar level that tonically activates NMDA receptors) is also crucial to prevent excitotoxic cell death (Choi et al., 1987;Choi, 1987). Thus, control of GLT-1 density on the astrocyte surface via molecular mechanisms modulating its intracellular trafficking is crucial for normal synaptic physiology and prevention of excitotoxicity.
Four isoforms of GLT-1 have been identified, which exhibit similar functional properties and oligomerize to form homomeric and heteromeric GLT-1 pools. However, the GLT-1 isoforms differ in their N-and C-termini, allowing for interaction with different intracellular proteins and offering an opportunity for differential regulation of the isoforms during physiological and pathological conditions (Peacey et al., 2009). So far, scaffolding proteins, such as PSD-95, PICK-1, and MAGI-1 have been shown to interact with the PDZ domain containing C-terminus of the GLT-1b isoform (Underhill et al., 2015;Sogaard et al., 2013;Zou et al., 2011;Gonzalez-Gonzalez et al., 2008). Here, we have identified the non-adrenergic imidazoline-1 receptor protein Nischarin, as a new physiological GLT-1 N-terminus interacting protein in astrocytes. Nischarin is a cytoplasmic protein that shows a diverse set of functions, including regulation of the cytoskeletal network (through Rac1), interaction with endosomes (via its phox domain), and regulation of receptor (mu opioid and integrin a 5 receptors) surface levels (Alahari et al., 2000(Alahari et al., , 2004Keller et al., 2017;Li et al., 2016;Kuijl et al., 2013;Alahari, 2003;Dong et al., 2017). (C) Schematic diagram of GST fusion constructs for GLT-1 N-terminus, 15 amino acid stretches of GLT-1 N-terminus (A-E), GLT-1 C-terminus, and GLAST C-terminus. GFP-Nisch was successfully pulled down with full-length GST-fused GLT-1 N-terminus and to GST fusions D (amino acids 9-23) and E (amino acids 23-37). Nischarin has been shown to alter the surface levels of receptors, including integrins and mu opioid receptors (Li et al., 2016(Li et al., , 2019Lim and Hong, 2004). To determine whether Nischarin regulated GLT-1 surface density, we used an 'antibody-feeding' immunofluorescence internalization assay in HeLa cells to visualize GLT-1 trafficking. HeLa cells were co-transfected with GLT-1 tagged in its extracellular domain with HA (GLT-1a-HA) along with either GFP-Nisch or GFP as a control. Briefly, antibody against the HA tag, present in the extracellular loop of GLT-1, was incubated with the cells for 15 min before placing them in the incubator at 37 C for 60 min to allow for internalization, followed by differential immunostaining of the surface and internal GLT-1 pool. To obtain a reference value for basal surface GLT-1 levels before internalization, cells were fixed immediately after the antibody surface labeling (constituting the T 0min population). At baseline (T 0min ), in control cells, robust surface and low internal GLT-1 labeling was observed. Even after 60 min, a significant change in GLT-1 internalization was not observed in the control (Figures 2A and 2C), suggesting stable turnover of the transporter under basal culture conditions. In GFP-Nisch expressing cells, no significant difference was observed in GLT-1 distribution initially at T 0min ; however, by 60 min, significant increases were observed in the accumulation of GLT-1 within intracellular compartments (Figures 2B and 2C). This suggests that Nischarin regulates constitutive trafficking of GLT-1 under basal conditions.
As trafficking of GLT-1 is dependent on endocytic and recycling pathways, we next assessed the effects of Nischarin on GLT-1 recycling. No significant differences were found between the GLT-1 recycling rates in GFP-Nisch cells and control (Figures S1B-S1D). Together, these data suggest that Nischarin promotes translocation of GLT-1 from the surface to intracellular compartments but (in contrast to its effect on mu opioid receptors: Li et al., 2016) does not affect GLT-1 recycling.
Next, astrocytes were transfected with either GFP-Nisch or GFP-NischDphox or GFP control, and were cocultured with hippocampal neurons. At DIV 14, immunostaining studies revealed significant colocalization between GFP-Nisch-positive endosomal structures and the early endosomal marker EEA1. The GFP-Nisch-positive vesicles showed variability in size and shape and were distributed throughout the astrocyte cell body and processes. However, GFP-Dphox showed a cytosolic expression, corroborating previous reports that Nischarin is targeted to endosomes via its phox domain (Figures S1A-S1D). In addition, endogenous GLT-1 co-localized significantly with intracellular vesicles positive for GFP-Nisch ( Figure 2E). Together these data suggest that intracellular GLT-1 accumulates in Nischarin-positive early endosomal structures within the astrocyte cell body and processes.
Glutamate promotes Nischarin-mediated GLT-1 intracellular trafficking in fixed and live hippocampal astrocyte cultures It has been previously reported that glutamate treatment decreases clustering and surface expression of GLT-1 (Al Awabdh et al., 2016;Underhill et al., 2015). The PDZ-binding domain containing protein DLG1 interacts with the C-terminal PDZ ligand of GLT-1b to regulate its surface density (Underhill et al., 2015). However, the role of molecules interacting with the N terminus of GLT-1 remains under explored. Therefore, we investigated whether Nischarin regulates GLT-1 trafficking under activity driven conditions, which we mimicked by applying glutamate.
Using a surface biotinylation assay, the effect of glutamate on surface GLT-1 levels in Nischarin-overexpressing astrocytes was assessed. Pure cortical astrocyte cultures were co-transfected with GFP and GLT-1a tagged with V5 (control) or GFP-Nisch and GLT-1a-V5. The transfected cultures were either left untreated or exposed to glutamate (100 mM, 1 h). Glutamate treatment significantly decreased surface GLT-1 level compared to the untreated control ( Figure 3A), as expected (Ibanez et al., 2016). Nischarin overexpression significantly decreased the GLT-1 surface levels compared to the untreated control in the absence of glutamate, corroborating our findings in Figure 2. Glutamate treatment in Nischarin-overexpressing astrocytes did not cause any further decrease in surface GLT-1 levels in comparison to Nischarin overexpression or glutamate application alone, suggesting that overexpression of Nischarin alone is sufficient to drive the internalization of surface GLT-1.
Next, using a proximity ligation assay (PLA), a powerful tool that detects a positive protein interaction only if the two proteins are closer than 40 nm, we determined the effect of glutamate on the endogenous GLT- iScience Article Nischarin-GLT-1 direct association (assessed by number of red puncta per DAPI-labeled (blue) nuclei) was observed compared to single antibody (Nisch/GLT-1) controls. Upon glutamate treatment (100 mM, 1 h), the number of puncta observed were significantly increased compared to the control ( Figure 3B). Together with the biotinylation assay, these results suggest that glutamate enhances the endogenous GLT-1-Nischarin interaction, and that this drives internalization of surface GLT-1.
To monitor live trafficking of GLT-1, we took advantage of a high affinity 13 amino acid a-bungarotoxin (BTX)-binding site (BBS) that has been exploited for tracking AMPA receptor movements in and out of the cell membrane (Sekine- Aizawa and Huganir, 2004). We explored a similar strategy to assess the time course of glutamate's action on the GLT-1-Nischarin interaction by engineering the extracellular loop of the GLT-1 transporter ( Figure 3C) to include the BBS tag. This position in GLT-1 transporters is silent in terms of its impact on receptor structure and function (Peacey et al., 2009). The astrocytes were co-transfected with GLT-1a-BBS and GFP-Nisch and cocultured with hippocampal neurons. The hippocampal iScience Article coculture was incubated with Alexa 555-conjugated BTX (BTX555) at 37 C for 20 min to allow live labeling of surface GLT-1. Live time-lapse imaging using structured-illumination microscopy (SIM)-tracked transporter internalization and individual endosomal events in labeled GLT-1a-BBS astrocytes (cocultured with hippocampal neurons) co-expressing GFP-Nisch and exposed to ACSF alone or ACSF containing 100mM glutamate ( Figure 3C). GFP-Nisch-positive intracellular vesicles were observed within the astrocyte processes and cell body. Glutamate application resulted in internalization of BTX555-labeled surface GLT-1 transporters to GFP-Nisch-labeled vesicles, as seen by the diagonal and co-localized lines in the kymographs representing GLT-1 and Nischarin vesicle movements within the astrocyte process ( Figure 3C). GFP-Nisch was found to be associated with the inner plasma membrane, and the internalized surface GLT-1 was trafficked into GFP-Nisch-positive vesicles, pinched off from the plasma membrane (Videos S1 and S2).

OPEN ACCESS
iScience 25, 104127, April 15, 2022 5 iScience Article Figure 4. GLT-1 surface density and transporter uptake current are enhanced in Nisch KO astrocytes (A) Western blot analysis in cortical astrocytes derived from WT, Nisch HET , and Nisch KO E16 embryos, confirmed decrease and loss of Nischarin in the HET and KO cultures. Surface biotinylation assay showed significant increase in GLT-1 surface density in KO culture compared to WT control. One-way ANOVA, post hoc Tukey's test (n = 3 animals).
(B) Representative images for GLT-1 (green) and Map2 (red) immunostaining in astrocytes derived from DIV14 WT and Nisch KO hippocampal culture. A significant increase in GLT-1 mean fluorescence intensity was observed in Nisch KO astrocytes. Unpaired Student's t test (n = 11-13). iScience Article GFP-Nisch-labeled vesicles (in white, Figure S3A) upon glutamate treatment in comparison to control at all three time points ( Figure 3D). Together, these results reinforce the role of Nischarin in regulating GLT-1 internalization during glutamate application, with colocalization observed as early as 5 min. Further investigation in to the subcellular localization of the co-localized GFP-Nisch vesicles containing BTX-labeled GLT-1 transporters revealed even distribution across the astrocyte cell body and processes in control and glutamate-treated (100mM, 60 min) cultures ( Figures S4A-S4D).

GLT-1 transporter density and function are altered in Nisch KO mice
We further characterized the role of Nischarin-mediated regulation of astrocytic GLT-1 by using transgenic Nisch KO mice ( Figures S3B-S3D). Lack of Nischarin expression was confirmed in homozygous Nisch KO mice, whereas the heterozygous (Nisch HET ) mice showed reduced Nischarin protein expression in comparison to the WT control ( Figure 4A). A surface biotinylation assay revealed significant increases in surface GLT-1 levels in astrocytes derived from Nisch KO transgenic mice compared to from WT control mice. The surface GLT-1 level in astrocytes derived from Nisch HET transgenics were not significantly different from that of WT ( Figure 4A). Immunostaining studies also revealed a significant increase in total astrocytic GLT-1 mean intensity in hippocampal cocultures derived from Nisch KO mice in comparison to WT control ( Figure 4B), which may reflect trafficking to a degradation pathway in the absence of Nischarin. These data support Nischarin's role in regulating GLT-1 surface levels.
Given the increased surface GLT-1 level in Nisch KO astrocytes, we undertook functional studies where glutamate uptake was assessed using whole cell patch-clamp of astrocytes in hippocampal neuron-glial cocultures (Brew and Attwell, 1987). For each glutamate anion transported into astrocytes by GLT-1, three Na + and one H + are also transported in, and one K + is exported from the cell (Levy et al., 1998a(Levy et al., , 1998bZerangue and Kavanaugh, 1996). Thus, two net positive charges are imported per glutamate taken up; therefore, uptake can be measured from the current it produces. Astrocytes were whole cell patch clamped and their identity was confirmed by the following: 1) dye filling (Alexa 488 or 594) to show coupling with other astrocytes ( Figure 4C), 2) their low input resistance ( Figures S5A and S3), and 3) their negative resting potential ( Figure S5B). The input resistance and resting potential were not significantly affected by knockout of Nischarin ( Figures S5A and S5B). After applying blockers of action potentials (150 nM TTX), inward-rectifying K + channels i.e., the main conductance of astrocytes (200 mM BaCl 2 ), and glutamate and GABA receptors (blocked with 10 mM NBQX, 50 mM D-AP5, 10 mM 5,7-dichlorokynurenate, 10 mM MK-801, and 10 mM bicuculline), once a steady membrane current was reached, at a voltage near the cell's resting potential (À90 mV), glutamate transporters were activated by applying D-aspartate (200 mM). D-aspartate evoked an inward current in both WT and Nisch KO astrocytes ( Figure 4E). The currents were confirmed to be mediated by glutamate transporters, because they were blocked by the GLT-1 and GLAST blocker TFB-TBOA (10 mM Figures 4D and 4F). Consistent with the surface biotinylation assay, we found that the glutamate uptake current was almost 2-fold higher in the Nisch KO astrocytes in comparison to the WT astrocytes (p = 0.029, Figures 4E and 4F).
Dysfunction of glutamate clearance can cause overstimulation of glutamate receptors and result in neuronal injury, termed excitotoxicity. To further investigate the neuroprotective function of astrocytes, we carried out an excitotoxicity assay using hippocampal neuron -astrocyte cocultures derived from WT and Nisch KO mice (at DIV 14). The cultures were challenged with 10 mM glutamate and 10 mM glycine for 24 h. Neuronal death was analyzed using propidium iodide (PI) and DAPI staining, and the number of PI-positive nuclei (red) were counted. Glutamate treatment evoked significantly less PI-labeling of neurons in Nisch KO cocultures compared with neurons in wild type cocultures ( Figure 4G). These data suggest that when challenged with neurotoxic glutamate levels, lack of Nischarin is protective against cell death, presumably because of the increased glutamate uptake that lack of Nischarin results in.

DISCUSSION
Glutamatergic neurons are responsible for the majority of the excitatory synaptic transmission and plasticity occurring in the brain. The astrocytic glutamate transporters serve the critical role of efficiently  iScience Article clearing glutamate from the extracellular space (Tanaka et al., 1997) to ensure normal glutamate signaling. GLT-1 is one of the highest expressed proteins in the brain (1% of total brain protein (Lehre and Danbolt, 1998). The high number of surface glutamate transporters (GLT-1 at a density of 8,500 transporters/mm 2 as well as GLAST at 2,500 transporters/mm 2 ) compensate for the slow transport cycle (12-70ms) to ensure effective clearance of the $4000 glutamate molecules released from a single synaptic vesicle (Lehre and Danbolt, 1998;Murphy-Royal et al., 2017). GLT-1 undergoes activity-dependent surface diffusion, and the glutamate-bound GLT-1 from 'synapse facing' sites are continuously replaced with GLT-1-lackingbound glutamate to help maintain a high concentration of available surface transporters at the astrocytic plasma membrane (Murphy-Royal et al., 2017;Al Awabdh et al., 2016). Thus, an in-depth understanding of the molecular mechanisms regulating and maintaining the surface GLT-1 density is crucial. Here, we have identified a new interacting protein partner, Nischarin, a non-adrenergic imidazoline-1 receptor (Alahari, 2003) that regulates intracellular trafficking of GLT-1 in response to the neurotransmitter glutamate. Further work is needed to establish whether the blood pressure lowering effects of imidazoline drugs such as clonidine are in any way mediated by effects on glutamate transport.
We found that Nischarin interacts with the N-terminal tail of GLT-1 through its phox domain. Specifically, Nischarin coprecipitated with amino acids 9-37 within the intracellular, unstructured N-terminal tail of GLT-1. Amino acids 9-23 have also been implicated in the interaction with Ajuba, a scaffolding protein that allows GLT-1 to regulate intracellular signaling or interact with the cytoskeleton (Marie et al., 2002). These amino acids are conserved across the four GLT-1 isoforms (Peacey et al., 2009), suggesting Nischarin could regulate all four isoforms, unlike previously identified regulators of GLT-1 trafficking that bind specifically to the PDZ domain found in GLT-1b (Bassan et al., 2008;Underhill et al., 2015). In addition, a coimmunoprecipitation assay using brain lysates confirmed that the Nischarin-GLT-1 interaction occurs in the intact brain. As we have only examined immunoprecipitation assays from the whole brain, we considered whether regional variation in the Nisch-GLT-1 interaction might occur. However, Nischarin protein levels detected using the same antibody as our study are comparable in the cortex and hippocampus (Ding et al., 2013); therefore, differences in GLT-1 binding capacity are unlikely. Glutamate transporters are known to interact with Na,K-ATPases and can exist with them as part of one macromolecular complex (Rose et al., 2009). Whether Nischarin is able to regulate the entire Glt-1/Na,K-ATPase macromolecular complex remains to be determined. The findings reported here do raise the intriguing possibility that, in addition to Nischarin's previously reported role in regulating cytoskeletal signaling, cell migration, Rabdependent endosomal sorting, and regulation of integrins and mu opioid receptors (Keller et al., 2017;Kuijl et al., 2013;Alahari, 2003), it may have an activity-dependent role in modulating glutamate concentration at the synapse.
GLT-1 is known to undergo constitutive and regulated endocytosis, which determines its availability for glutamate clearance from extracellular compartments in the nervous system (Martinez-Villarreal et al., 2012). Our antibody-feeding assays have revealed that overexpression of Nischarin redistributed surface GLT-1 transporters into endosomal structures but did not alter transporter recycling under basal conditions. Taken together, our data not only support an interaction between Nischarin and GLT-1 but also indicate the possibility that Nischarin alters trafficking of GLT-1 by sequestration. The phox domain of Nischarin is a stretch of $110 amino acids that is a phosphatidylinositol 3-phosphate-binding (PI3P) module, and PI3P is enriched in early endosomal membranes (Lim and Hong, 2004). Our data confirmed previous findings that Nischarin is targeted to early endosomes, marked by EEA1 (Kuijl et al., 2013), but also showed multiple Nischarin-positive yet EEA1-negative vesicular structures, revealing a much wider distribution of Nischarin within the endosomal system. In addition, we observed that the GLT-1-Nischarin vesicular structures are spatially distributed along the astrocyte cell body and processes.
Surface biotinylation studies revealed that astrocytic GLT-1 surface levels decreased (40%) in response to glutamate ( Figure 3A), which is consistent with previous findings (Ibanez et al., 2016). Overexpressing Nischarin in astrocytes exposed to glutamate treatment did not significantly further alter surface GLT-1 levels, implying that Nischarin occludes the effect of glutamate on surface GLT-1. Alteration in surface GLT-1 levels by Nischarin could offer a means for modulating glutamatergic activity at the synapse. Given the increased Nisch-GLT-1 interaction following glutamate exposure, a likely explanation for these results is that Nischarin mediates the effects of glutamate-dependent GLT-1 surface density regulation. The recruitment of Nischarin to GLT-1 could have additional consequences as Nischarin is known to act as a scaffolding platform for signaling pathways through its interactions with multiple proteins including, integrin  berger et al., 2020). The removal of PAPs from the synapse can boost glutamate spillover and shape NMDAreceptor-mediated inter-synaptic cross-talk (Henneberger et al., 2020).
Using time-lapse monitoring of the transporter (employing a GLT-1aBBS construct that can bind fluorophore-conjugated BTX, eliminating the use of bulky antibodies which could promote clustering and affect membrane trafficking properties (Sekine-Aizawa and Huganir, 2004)), we showed that glutamate binding and/or transport triggered intracellular trafficking of GLT-1 into Nischarin-labeled intracellular compartments within 5 min of glutamate exposure and is evenly distributed across the astrocyte cell body and processes. This time course suggests that the GLT-1-Nisch trafficking could be of more relevance in pathological conditions such as ischemia or traumatic brain injury, where the extracellular concentration of glutamate remains elevated (in the 100-200 mM range) for hours (Ibanez et al., 2016). Future studies should focus on a wider range of glutamate concentrations, and examine the GLT-1-Nisch interaction in astrocytes following neuronal stimulation to ascertain its functional relevance during synaptic activity. SIM resolution allowed tracking of single GFP-Nisch-labeled vesicles containing GLT-1BBS bound BTX555, and the resultant kymograph confirmed colocalization and inwardly directed (toward the cell body), slow ($minutes) movement as vesicles traversed the astrocytic process.
Astrocytes derived from Nisch KO transgenic animals exhibit increased surface GLT-1 density and a concomitant 2-fold increase in transporter uptake currents. Future experiments should examine how Nisch KO affects the EPSCs generated by neuronal action potentials, especially during high frequency stimulation or activation of many axons when clearance of glutamate from the synaptic cleft becomes more critically dependent on glutamate transporter activity. This enhanced surface GLT-1 density served to reduce cell death after glutamate insult, demonstrating the relevance of this mechanism to pathology. Dysregulation of Nischarin regulation of GLT-1 transporter surface density and function could affect glutamate clearance. Ineffective glutamate clearance is observed in many neurodegenerative diseases, including amyotrophic lateral sclerosis, epilepsy, Alzheimer's, Huntington's, and Parkinson's disease (Hindeya Gebreyesus and Gebrehiwot Gebremichael, 2020; Peterson and Binder, 2019). Therefore, this work not only reveals a distinct mechanism by which GLT-1 intracellular trafficking and function are regulated but also provides possible new avenues of research for treating neurological disorders.

Limitations of the study
We found that the N-terminal domains of Glt-1 (residues 9-27) are sufficient for the interaction with Nischarin. Future studies should further identify the precise amino acid residues on Glt-1, which would help to allow selective targeting of this interaction to control glutamatergic transmission. In addition, previous studies found that Glt-1 regulation is mediated by phosphorylation. It would be interesting to determine the role of phosphorylation in regulating the Glt-1 -Nischarin interaction. As shown and discussed in the text, most of the imaging results of this study were derived from an in vitro primary astrocyte-neuron cell culture model systems. Further microscopy studies will be needed to validate the effect of Nischarin on Glt-1 membrane dynamics in more intact tissue. However, confirmation in more intact tissue systems (acute brain slices) of the impact on Glt-1 function was achieved using electrophysiology.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Materials availability
Plasmids generated as part of this study will be made available upon request. All data reported in this paper will be shared by the lead author upon request.

REAGENT or RESOURCE SOURCE IDENTIFIER
This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Cell culture
Primary cultures of rat cortical astrocytes were prepared from E18 or P0 Sprague-Dawley rats as previously described (Banker, 1998). Hippocampal astrocyte-neuron rat co-cultures were obtained from E18 rat embryos as described previously with some modifications (Arancibia-Carcamo et al., 2009

METHOD DETAILS
Yeast two-hybrid screen This screen was done as described previously (Marie et al., 2002). Briefly, bait cDNA for the GLT-1 N terminus (amino acids 1-44 of the rat protein sequence) was cloned into the yeast expression vector pPC97 in frame with the GAL4 binding domain. It was screened against a random-primed cDNA library from seizure-stimulated adult rat hippocampus cloned in the yeast expression vector pPC86 in frame with the GAL4 activation domain. Interacting proteins were identified by colony selection on plates lacking leucine, tryptophan, and histidine and confirmed by using a b-galactosidase assay and by checking that in the absence of GLT-1 bait the library protein did not activate the reporter genes (His3, allowing growth on histidine-deficient medium, and LacZ, expressing b-galactosidase).

Plasmid constructs
Mouse GLT-1a cDNA with V5 and HA epitope tag inserted into the extracellular loop of the transporter (between Pro 199 and Pro 200 ) and cloned into pcDNA3.

Proximity ligation assay
The in-situ proximity ligation assay (PLA) was used according to the manufacturer's instructions (Olink Bioscience). Neurons were fixed in 4% PFA/30% sucrose, blocked (10% horse serum, 0.5% BSA, and 0.2% Triton X-100, 10 min at room temperature), and incubated with primary antibodies (1:500, anti-GLT (gift from Dr. N.Danbolt) and anti-Nischarin (1:100, Sigma, Cat. No: HPA023189). For control PLA, single primary antibody was applied. Cells were washed in 1 3 PBS and then incubated with secondary antibodies conjugated to oligonucleotides. Ligation and amplification reactions were conducted at 37 C, before mounting and visualization with confocal laser scanning microscope. Images were thresholded and number of puncta and DAPI stained nuclei were manually counted for each image using the Metamorph software. Experiments were performed blinded during image acquisition and analysis.

Antibody feeding
For receptor internalization and recycling assays, HeLa cells were transfected with GFPNisch and GLT-1a-HA (2mg, 2mg) or GFP and GLT-1a-HA (1mg, 2mg). The transporters were live labeled with anti-HA antibody in DMEM +25mM HEPES at 17 C. Labelled transporters were allowed to internalize for 60 min at 37 C. For recycling assay, surface transporters were stripped using acid wash (0.2M acetic acid and 0.5M NaCl). Cells were then returned to the incubator for 30 and 60 min at 37 C to allow internalized transporters to recycle to the surface. For surface staining, cells were fixed with 4% paraformaldehyde (PFA)/4% sucrose/ PBS, pH 7, for 5 min and blocked with block solution (PBS, 10% horse serum, and 0.5% BSA) for 10 min, followed by Alexa Fluor-555-conjugated anti-mouse secondary antibody for 1h (1:400; Invitrogen). For identifying internalized transporters, cells were subsequently permeabilized with block solution containing 0.2% Triton X-100 for 10min, followed by Alexa Fluor-555-conjugated anti-mouse secondary antibody ll OPEN ACCESS

Recording glutamate uptake in astrocytes
In order to record the glutamate uptake current from astrocytes, voltage-clamp recordings were made at the cell's resting potential (typically around À90 mV). D-aspartate (200 mM, Sigma) was used to evoke a transporter current, since it is taken up by glial glutamate transporters (Davies and Johnston, 1976; Barbour et al., 1991;Furness et al., 2008)  iScience Article glutamate via heteroexchange on transporters (Volterra et al., 1996). Any resulting activation of glutamate receptors might cause membrane potential depolarisation, neuronal action potentials and a rise of [K + ] o into the extracellular space, which could evoke an inward current in astrocytes (which have a highly K +permeable membrane: Meeks and Mennerick, 2007). To prevent these effects we therefore supplemented the aCSF with a selection of blockers, which were present throughout the experiment: TTX to block action potentials (150nM, Tocris), a GABA A receptor blocker (bicuculline 10 mM, Sigma), NMDA receptor blockers (D-AP5 50 mM, Tocris; (+)MK-801 10 mM, Sigma; 5,7-DCK 10 mM, Sigma), an AMPA and kainate receptor blocker (NBQX 10 mM, Sigma), and an inwardly rectifying potassium channel blocker (barium chloride 200 mM, Sigma) which does not affect glutamate transport (Barbour et al., 1991). A non-transported glial glutamate transporter blocker (Shimamoto et al., 2004), TFB-TBOA (10 mM, Tocris) was also used in some experiments to block the glutamate transporter current evoked by D-aspartate. The size of the uptake current was calculated as the inward current recorded in D-aspartate minus the average of the baseline currents measured before and after D-aspartate application (using Clampfit 10.4). Experiments were performed with the experimenter blinded to the genotype.

QUANTIFICATION AND STATISTICAL ANALYSIS
All experiments were performed on astrocytes/cell culture/mixed hippocampal astrocyte-neuron co-culture from at least three individual preparations. For all quantified experiments the experimenters were blind to the condition of the sample analyzed. All image analysis was performed blinded. Values are given as mean G standard error of the mean (SEM). Error bars represent SEM. Statistical analysis was performed in GraphPad Prism (version 8; GraphPad Software, CA, USA) or Microsoft Excel. All data was tested for normal distribution with D'Agostino & Pearson test to determine the use of parametric (Student's t test, one-way ANOVA) or non-parametric (Mann-Whitney, Kruskal-Wallis) tests. When p < 0.05, appropriate post hoc tests were carried out in analyses with multiple comparisons and are stated in the figure legends.