Ligand binding is a critical requirement for plasma membrane expression of heteromeric kainate receptors.

Intracellular trafficking of ionotropic glutamate receptors is controlled by multiple discrete determinants in receptor subunits. Most such determinants have been localized to the cytoplasmic carboxyl-terminal domain, but other domains in the subunit proteins can play roles in modulating receptor surface expression. Here we demonstrate that formation of an intact glutamate binding site also acts as an additional quality-control check for surface expression of homomeric and heteromeric kainate receptors. A key ligand-binding residue in the KA2 subunit, threonine 675, was mutated to either alanine or glutamate, which eliminated affinity for the receptor ligands kainate and glutamate. We found that plasma membrane expression of heteromeric GluR6/KA2(T675A) or GluR6/KA2(T675E) kainate receptors was markedly reduced compared with wild-type GluR6/KA2 receptors in transfected HEK 293 and COS-7 cells and in cultured neurons. Surface expression of homomeric KA2 receptors lacking a retention/retrieval determinant (KA2-R/A) was also reduced upon mutation of Thr-675 and elimination of the ligand binding site. KA2 Thr-675 mutant subunits were able to co-assemble with GluR5 and GluR6 subunits and were degraded at the same rate as wild-type KA2 subunit protein. These results suggest that glutamate binding and associated conformational changes are prerequisites for forward trafficking of intracellular kainate receptors following multimeric assembly.

proteins into tetrameric receptor complexes in the endoplasmic reticulum. The nascent receptors are then transported forward through the secretory pathway to their sites of functional activity, which in neurons can be in postsynaptic densities or at extrasynaptic sites. Synaptic AMPA receptor numbers are modulated by both constitutive and activity-dependent endoand exocytosis processes; for example, AMPA receptors are rapidly transferred to or removed from postsynaptic densities in response to distinct patterns of tetanic stimuli or NMDA receptor activation (3)(4)(5). NMDA receptors in synapses were originally thought to be relatively static compared with AMPA receptors (6), but more recent reports have demonstrated dynamic regulation of NMDA receptor trafficking in ways that might affect induction of synaptic plasticity (Refs. 7 and 8; reviewed in Ref. 9). Relatively few examples of activity-dependent changes in neuronal kainate receptors have been reported (10,11), and the cellular mechanisms that underlie those functional changes remain obscure.
Transit of ionotropic glutamate receptors through the secretory and endocytic pathways is primarily controlled by interactions between cellular chaperone proteins and discrete motifs on the receptor subunits, which are located predominantly on the cytoplasmic carboxyl-terminal domain. These domains are particularly important in the control of receptor egress from the endoplasmic reticulum (ER) after receptor assembly. A number of critical ER-trafficking determinants have been identified in glutamate receptor subunits, which include argininerich sequences in NR1, GluR5-2b, GluR5-2c, GluR6, and KA2 subunits that promote ER retention/retrieval (12)(13)(14)(15)(16) or export (16,17), PDZ binding domains in GluR2 AMPA receptor subunits and splice variants of NR1 that control ER export (12,13,18), and juxtamembrane determinants that promote export of NR2B (19) and GluR␦2 (20) but retention of GluR2 subunitcontaining receptors (18). In addition to cytoplasmic determinants, glutamate receptors in the ER are subject to a resident quality control system that verifies that the proteins are properly folded and assembled before export to Golgi compartments (21). Non-cytoplasmic determinants in the transmembrane and extracellular domains within the receptor subunits are participants in this quality control process. RNA editing at the Q/R site of GluR2 AMPA receptor subunits reduces calcium permeability (22) and channel conductance (23) and was recently shown to control trafficking and assembly in the ER by preventing tetramerization of all-edited subunits into a functional receptor (18,24). In addition, it has been proposed that glutamate binding to receptors in the ER acts as a checkpoint for forward trafficking by probing receptor functionality, because mutations that eliminate binding in the Caenorhabditis elegans GLR-1 AMPA receptor subunit and the GluR6 kainate receptor subunit promoted retention of these receptors (25,26).
Here we have tested the hypothesis that ligand binding is a critical checkpoint in kainate receptor trafficking by introducing a mutation into the ligand binding domain of the KA2 subunit, a subunit that does not form functional homomeric kainate receptors (14,27). Assembly of KA2 subunits with GluR5, GluR6, or GluR7 subunits produces functional heteromeric receptors with distinct pharmacological and physiological characteristics (27,28). KA2 homomers are retained in the ER because this subunit contains an arginine-based retrieval/ retention motif in the cytoplasmic domain. Mutation of this trafficking determinant releases homomeric KA2 receptors from the ER, but despite plasma membrane expression the receptors do not gate currents in response to high concentrations of glutamate (14). We found that elimination of the glutamate binding affinity in the KA2 subunit caused retention of this subunit as well as heteromeric kainate receptor complexes that incorporated the mutated subunit in both transfected cell lines and neurons. Retention was not accompanied by an increased degradation rate, suggesting that the mutation did not cause gross misfolding of the receptors. These data support the hypothesis that glutamate binding is a checkpoint in the biosynthesis of kainate receptors. Further, we suggest that binding-related conformational changes in the receptor structure, rather than ion permeation through the channel, are critical for this quality control pathway.
Immunoprecipitation and Western Blots-Transfected COS-7 cells were washed twice with cold PBS and lysed in 0.1 or 0.5 ml of lysis buffer (50 mM Tris, 150 mM NaCl, 0.5% Nonidet P-40, 2 mM 4-(2aminoethyl)benzenesulfonyl fluoride HCl, 1 mM EDTA, 130 M bestatin, 14 M E-64, 1 mM leupeptin, 0.3 M aprotinin). Supernatants were obtained after lysis and centrifugation at 15,000 rpm at 4°C for 20 min. Immunoprecipitations were performed by incubating lysate supernatants with anti-GFP antibody (2.5 g, JL-8, BD Biosciences Clontech, Palo Alto, CA) followed by incubation with 50 l of 50% protein A/G-Sepharose (Amersham Biosciences) slurry overnight. After four washes in lysis buffer, bound proteins were eluted from the beads by boiling in 2ϫ sample buffer and then separated by electrophoresis on 8% SDS-PAGE gels. Proteins were electro-transferred onto nitrocellulose membranes and probed with anti-myc antibody (0.1 g/ml, Upstate Biotechnology Inc., Lake Placid, NY) overnight at 4°C. Immunoreactive bands were visualized using horseradish peroxidase-conjugated donkey antirabbit secondary antibody with the enhanced chemiluminescence (ECL) detection technique (Amersham Biosciences).
Immunolocalization of Receptors-Cell surface receptors were detected in transfected COS-7 cells following incubation with polyclonal anti-GFP antibody (1:500) or polyclonal anti-myc antibody (1:500) for 1 h at 4°C. The cells were washed with cold PBS and fixed with 4% paraformaldehyde/4% sucrose in PBS for 20 min at room temperature. After fixation, the cells were washed with PBS and incubated with fluorescence-conjugated goat secondary antibody (Alexa 594, Molecular Probes, Eugene, OR) for 1 h at room temperature. For co-localization of surface receptors and intracellular receptors, after surface labeling and fixation transfected cells were permeabilized with 0.2% Triton X-100 in PBS for 5 min. After several washes with PBS, the cells were incubated with polyclonal anti-KA2 primary antibody (1:1000, Upstate Biotechnology Inc.) for 2 h at room temperature. To visualize myc-KA2 subunit co-localization with Golgi, transfected cells were fixed and permeabilized before incubation with polyclonal anti-giantin (1:1000, Covance, Richmond, CA) and monoclonal anti-myc antibodies. To co-localize myc-KA2 and ER compartments, polyclonal anti-KA2 and monoclonal anti-PDI (1:200, Affinity Bioreagents, Golden, CO) antibodies were used. After washing with PBS, the cells were incubated with the appropriate fluorescence-conjugated secondary antibodies (Alexa Fluor 594 or 633, Molecular Probes) for 1 h at room temperature. Surface and intracellular immunofluorescence was captured with a Zeiss LSM510 confocal microscope (Carl Zeiss, Thornwood, NY). For clarity in presentation, the long red fluorescence emitted by Alexa 633 was false-colored as cyan in the figures. Representative images in the figures are single optical slices with thickness of 0.38 -0.40 m.
Enzyme-linked Immunosorbent Assays -ELISAs were performed using a modified version of a described previously protocol (8). Briefly, Dulbecco's modified Eagle's medium media was removed from transfected COS-7 cells cultured in 12-well plates, and cells were washed once with ice-cold PBS. Cells were incubated at 4°C with anti-GFP (1:200) or anti-myc (1:100) antibody for 30 min at 4°C, washed extensively with ice-cold PBS, and fixed for 20 min at room temperature with 4% paraformaldehyde in PBS. After washing with PBS, the cells were incubated for 1 h at room temperature with goat anti-mouse antibody conjugated to horseradish peroxidase (1:1000, Amersham Biosciences). Cells were again washed with PBS, the horseradish peroxidase substrate o-phenylenediamine dihydrochloride was added, and the color reaction was developed for 2-3 h. The optical density of 0.3 ml of supernatant was read on a spectrophotometer at 492 nm. Values for cell surface expression of kainate receptors were determined by averaging at least four replicates in each experiment and subtracting the mean absorbance in wells with non-transfected cells. Three separate experiments were performed with each set of receptor cDNAs.
Neuronal Cell Culture, Transfection, and Imaging -Hippocampal neurons from postnatal day 0 rats were cultured using the methods described previously (29). After 3-4 days in culture, cells were transfected using the calcium phosphate method. Transfected cultures were allowed to grow for another week. Immunocytochemistry was performed when cells were 10 -11 days in vitro. For surface labeling of GFP-GluR6, living cells were incubated with chicken anti-GFP polyclonal antibody in cell culture media (1:500, Chemicon) for 20 min at 37°C, rinsed twice with PBS, and fixed with 4% paraformaldehyde/4% sucrose in PBS. For intracellular labeling of myc-KA2, cells were permeabilized by incubation with PBS/0.15% Triton X-100 solution for 10 min in cold room. The cells were then blocked in PBS containing 10% bovine serum albumin for 2-3 h at room temperature, mouse anti-myc antibody (1:2000, Covance) was applied for 1 h at room temperature, and the cells were washed with PBS/10% bovine serum albumin. Alexa Fluor 568 goat anti-chicken IgG (1:500, Molecular Probes) and Alexa Fluor 350 goat anti-mouse IgG (1:500, Molecular Probes) secondary antibody were applied for 1 h at room temperature. The coverslips were rinsed extensively with PBS and mounted. Neurons were observed on an LSM510 Zeiss confocal laser scanning microscope, and images were analyzed by using a Simple PCI Image Analysis System (Compix Inc., Cranberry Township, PA). Regions of interest were drawn around the fluorescent cell bodies, and the average pixel intensity was calculated. To correct the background, equivalent regions of interest were drawn outside the fluorescent neurons in the same image, and the average pixel intensities were subtracted from the neuronal fluorescence. Fluorescence is expressed in intensity units per pixel, and data are presented as mean Ϯ S.E.
Pulse-chase Assays-HEK 293 QA cells were transfected with plasmid DNA using Lipofectamine reagent (Invitrogen). After 36 h, cells were incubated in cysteine-and methionine-free Dulbecco's modified Eagle's medium starvation media for 30 min. Starvation media was removed and replaced with Dulbecco's modified Eagle's medium labeling media containing EXPRE 35 S 35 S protein labeling mix (PerkinElmer Life Sciences). After 30 min, cells were rinsed twice with PBS and returned to normal growth media, including brefeldin A (10 g/ml, Calbiochem) and methionine for the duration of the chase to the specified time points. Cells were washed twice with ice-cold PBS and incu-bated on ice for 10 min in lysis buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 1% Triton X-100) with a mixture of protease inhibitors (Roche Applied Science) after which solubilized extracts were collected for immunoprecipitation. Proteins were immunoprecipitated with polyclonal anti-myc antibody, resolved with SDS-PAGE, and quantified on a PhosphorImager 425S (Amersham Biosciences). The measured densities of at each time point were normalized to the image density at time point zero.
Radioligand Binding Assays-Crude membrane fractions were prepared from transiently transfected COS-7 cells by first homogenizing in buffer (50 mM Tris-Cl, 0.32 M sucrose, pH 7.4), centrifuging at 800 ϫ g for 10 min at 4°C, and re-centrifuging the supernatant at 13,000 ϫ g for 20 min at 4°C. The pellet was resuspended in 2 ml of homogenizing buffer, and the process was repeated twice. The final membrane pellet was resuspended in 50 mM Tris-Cl, pH 7.4, and 25 g of membrane protein was used in radioligand binding assays with 5, 10, and 50 nM [ 3 H]kainate. Nonspecific binding was measured in the presence of 1 mM glutamate. Binding was carried out for 1 h at 4°C before rapid filtration on Whatman GF/C glass filters through a Brandel Cell Harvester (Biomedical Research and Development Laboratories, Gaithersburg, MD), three washes with cold 0.9% saline solution, addition of scintillation fluid and counting on a Beckman LS5000 TD scintillation counter (Beckman Instruments).

RESULTS
To determine how elimination of glutamate binding affinity of KA2 subunits affected the biosynthesis and pharmacological properties of heteromeric kainate receptors, we first altered a critical threonine residue in the putative binding domain of the KA2 subunit to either an alanine (T675A) or a glutamate (T675E) using site mutagenesis of the receptor cDNA. The hydroxyl moiety of the equivalent threonine in GluR2 (T655) forms an essential hydrogen bond with the ␥-carboxyl group of glutamate and kainate (30). Mutation of the equivalent residue in the NR2A subunits increased the EC 50 for glutamate by three orders of magnitude (31) and eliminated specific binding of [ 3 H]kainate to chick kainate-binding protein (32). Consistent with these results, neither KA2(T675A) nor KA2(T675E) receptors bound [ 3 H]kainate in radioligand assays using membrane preparations from transfected COS-7 cells; that is, total binding was not different from nonspecific binding at any concentration tested (5-50 nM, data not shown). In contrast, specific binding of [ 3 H]kainate was robust in membranes from cells expressing wild-type KA2 subunits and was consistent with the K D of ϳ15 nM described previously (27). Thus, mutation of Thr-675 eliminated detectable ligand binding to the KA2 subunit.
We next determined if mutation of the ligand binding site occluded multimeric assembly of receptor subunits, thereby preventing formation of heteromeric receptors. Membranes were prepared from HEK 293 cells co-expressing GFP-tagged GluR6 and myc-tagged KA2 receptor subunits, and receptor proteins were isolated by immunoprecipitation with an anti-GFP antibody. Proteins were separated by denaturing SDS-PAGE and transferred to nitrocellulose membranes for Western analysis with an anti-myc primary antibody (Fig. 1C). Myc-KA2, myc-KA2(T675A), and myc-KA2(T675E) protein were abundant following immunoprecipitation of GFP-GluR6 protein (Fig. 1C, IP lane), indicating that the wild-type and mutant KA2 subunits had formed multimeric receptors with GFP-GluR6 subunits. These results suggested that the absence of AMPA-evoked currents in heteromeric receptors was not due to compromised assembly of mutant KA2 subunits with GluR6.
In our initial physiology recordings we were not able to resolve whether glutamate-evoked currents in a subset of cotransfected cells arose from heteromeric GluR6/KA2(T675A) and GluR6/KA2(T675E) receptors or rather from homomeric GluR6 receptors. To determine if cells expressing both GluR6 and the KA2 mutants contained heteromeric kainate receptors on their plasma membrane, we assessed their localization using immunofluorescent techniques. COS-7 cells were co-transfected with GFP-GluR5-2b or GFP-GluR6 in combination with myc-KA2 or myc-KA2(T675A) cDNAs ( Fig. 2A). Cells were first stained for surface-expressed GFP-GluR5-2b or GFP-GluR6 with an anti-GFP primary antibody under non-permeabilizing conditions, followed by permeabilization and labeling of intracellular myc-KA2 (or KA2 mutants) with an anti-myc antisera. Fig. 2A shows representative examples of the three-color analysis we performed in these experiments: green fluorescence arises from total GFP-GluR5-2b or GFP-GluR6 receptors, red from surface GFP-tagged receptors, and cyan from intra-and extracellular myc-KA2 (originally acquired at long red wavelengths and false-colored for clarity). In cells expressing wildtype GFP-GluR5-2b/myc-KA2 and GFP-GluR6/myc-KA2, GFPtagged receptors were abundant on the plasma membrane of co-transfected cells ( Fig. 2A, center column). In contrast, cells co-transfected with myc-KA2(T675A) or myc-KA2(T675E) had significantly lower levels of detectable surface GFP-GluR5-2b or GFP-GluR6, and indeed most of the co-transfected cells imaged had no apparent receptor subunits on the plasma membrane. These results suggest that the altered KA2 subunit acts as dominant suppressor of forward trafficking in heteromeric kainate receptors and support the interpretation that the residual glutamate currents in patch clamp recordings arose from homomeric GluR6 receptors that did not contain a mutant KA2 subunit.
We also tested if mutation of the ligand binding site prevented surface expression of a homomeric KA2 subunit. As described previously (14), wild-type KA2 receptors are efficiently retained in the endoplasmic reticulum by virtue of an arginine-based retention/retrieval signal in the cytoplasmic tail. Alanine substitution of this motif in KA2-R/A mutants releases KA2 receptors to the plasma membrane (see Fig. 2B), but the receptors still do not gate currents in response to glutamate application (14). Mutation of Thr-675 in myc-KA2-R/A subunit eliminated surface expression of the homomeric receptors, as shown with surface-staining immunofluorescence for anti-myc epitopes in Fig. 2B (top row of images). Introduction of the Thr-675 mutations also prevented surface expression of a chimeric KA2 receptor containing the carboxyl-terminal domain of the GluR6 subunit (myc-KA2⌬R6, bottom row in Fig. 2B), which contains a forward trafficking determinant that efficiently drives plasma membrane localization of homomeric receptors (17). Thus ligand binding is a critical prerequisite for surface expression of both heteromeric and homomeric kainate receptors, even when the latter do not respond to glutamate application with channel gating.
The reduced surface expression was also observed in cell ELISA assays to quantitate surface expression of receptor protein (Fig. 3). Live COS-7 cells expressing GFP-GluR6 with myc-KA2 or myc-KA2 Thr-675 mutants were incubated with anti-GFP antisera to label surface receptors and a second antibody conjugated to horseradish peroxidase that generated a color reaction upon addition of substrate. GFP-GluR6/myc-KA2 cells exhibited robust surface expression of GFP epitopes (Fig.  3A), whereas surface expression of GFP-GluR6 was reduced by Ͼ90% when co-expressed with myc-KA2(T675A) and myc-KA2(T675E) subunits. Homomeric myc-KA2-R/A receptors on the plasma membrane were expressed by ϳ80 -85% when the Thr-675 mutations were introduced (Fig. 3B), again consistent with the immunofluorescent results in supporting a critical role for ligand binding to the intracellular receptor.
We next tested if heteromeric kainate receptors expressed by cultured neurons also require an intact ligand site. There is currently no specific antibody that recognized extracellular epitopes of kainate receptor subunits, so we transfected GFP-GluR6 alone or in combination with myc-KA2, myc-KA2(T675A), or myc-KA2(T675E) cDNAs into cultured hippocampal neurons and localized receptor subunits using immunofluorescence imaging (Fig. 4). Live neurons were exposed to anti-GFP antibody before fixation and permeabilization, followed by anti-myc antibodies to localize myc-KA2-expressing neurons. As shown in the figure, GFP-GluR6 and GFP-GluR6/myc-KA2 receptors localized to the somatic and dendritic membranes (red panels in the top row of images) and internal compartments (green GFP fluorescence). In contrast, co-expression of GFP-GluR6 with myc-KA2(T675A) or myc-KA2(T675E) greatly reduced surface expression of GFP-GluR6containing receptors (Fig. 4A, bottom two sets of images), as was observed in the transfected COS-7 cells. In addition to preventing surface expression of the heteromeric kainate receptors, myc-KA2(T675A) and myc-KA2(T675E) also redistributed intracellular GFP-GluR6 protein into a predominantly somatic compartment, with little staining evident in the dendritic processes, in contrast to GFP-GluR6/myc-KA2 receptors- Recordings were made in whole cell patch clamp mode at a holding potential of Ϫ70 mV. B, residual glutamate currents in GluR6/KA2(T675A)-and GluR6/KA2(T675E)-expressing cells likely arose from homomeric GluR6 receptors, because their desensitization rates were significantly slower than that of wild-type GluR6/KA2 receptors but not significantly different from GluR6 receptors (34). C, co-immunoprecipitations demonstrate that wild-type and mutant myc-KA2 subunits co-assemble with GFP-GluR6 subunits. Following coexpression of GFP-GluR6 and myc-KA2, myc-KA2(T675A), or myc-KA2(T675E), GFP-GluR6 subunit protein was immunoprecipitated with anti-GFP antibody, and Western blots were carried out with anti-myc antibodies. L, lysate; PC, protein from pre-clear beads without anti-GFP antibody; IP, immunoprecipitated protein using anti-GFP antibody. expressing neurons. To quantitate these changes in surface expression of GFP-GluR6 containing receptors, we measured the fluorescent intensities in the red (surface GFP-GluR6) and green (total GFP-GluR6) channels and expressed those values as a ratio (Fig. 4B). GFP-GluR6 and GFP-GluR6/myc-KA2 had relatively high red:green ratios (0.85 Ϯ 0.03 and 0.68 Ϯ 0.06, respectively, n ϭ 7) compared with GFP-GluR6/myc-KA2(T675A) and GFP-GluR6/myc-KA2(T675E) (0.24 Ϯ 0.04 and 0.23 Ϯ 0.04, respectively, n ϭ 8), indicating that the mu-tant KA2 subunits reduced proportional surface expression of GFP-GluR6 containing receptors by ϳ65% in neurons.
To determine if elimination of glutamate binding in the KA2 mutants caused redistribution intracellularly, we co-localized myc-KA2-R/A and myc-KA2-R/A(T675E) with two markers for prominent organelles: protein di-sulfide isomerase (PDI), which is resident in the ER, and giantin, a cis-and medial-Golgi resident protein. As shown in Fig. 5, myc-KA2-R/A colocalized extensively in PDI-stained reticular structures characteristic of ER in COS-7 cells. Little overlap was observed with giantin, despite the surface localization of this homomeric receptor, suggesting that the proportion of total receptor contained within the Golgi was very small. This pattern of expression was similar for myc-KA2-R/A(T675E) (Fig. 5, right panels), which also co-localized with PDI but not with giantin. This analysis and similar immunofluorescent co-localization assays with heteromeric GFP-GluR6 and myc-KA2 receptors (data not shown) did not allow us to identify unequivocally the intracellular site of retention. We also attempted to assess maturation of myc-KA2-R/A and GFP-GluR6/myc-KA2 receptors using biochemical assays for resistance to endoglycosidase H but were unable to detect enzyme-resistant receptor protein (even in heteromeric subunit combinations that produced functional receptors). Additionally, we assessed whether kainate receptors were subject to sulfation, a post-translational modification dependent upon the Golgi apparatus that has been shown to modify the GluR2 AMPA receptor subunit (18). However, neither myc-KA2-R/A receptors nor GFP-GluR6/myc-KA2 receptor subunits were sulfated in detectable quantities (data not shown), precluding use of this assay as a means of analyzing the site of retention of ligand-binding mutants. In summary, we conclude that the gross subcellular distribution of the KA2 receptor was not altered by elimination of ligand binding but FIG. 2. Immunofluorescent localization of heteromeric kainate receptor subunits demonstrates that mutation of the KA2 ligand binding eliminates surface expression. A, GFP-GluR5-2b or GFP-GluR6 were expressed in combination with myc-KA2 or myc-KA2(T675A) subunits on COS-7 cells. The left panels show the GFP fluorescence, indicative of total expression of GFP-GluR5-2b or GFP-GluR6. The center panels show extracellular GFP epitopes arising from plasma membrane receptors probed with anti-GFP primary antibody under non-permeabilized conditions. The right panels show immunoreactivity detected by anti-myc antibodies to verify the presence of myc-KA2 or myc-KA2(T675A) in the cells. Weak surface GFP staining was observed in a subset of GFP-GluR6-expressing cells, consistent with the presence of homomeric GluR6 receptors in physiological recordings. B, surface expression of homomeric myc-KA2-R/A and myc-KA2⌬6 receptors is eliminated by mutation of the binding site reside Thr-675. The left panels are representative images from COS-7 cells expressing the indicated receptors stained with anti-myc antibody under non-permeabilizing conditions. The center and right panels demonstrate that introduction of T675A or T675E mutations into myc-KA2-R/A or myc-KA2⌬6 results in a profound reduction in surface localization of these receptors.

FIG. 3. Surface expression of heteromeric kainate receptors is significantly reduced in cell ELISA assays by binding site mutation of KA2.
A, cell ELISAs were used to quantitate surface expression of GFP-GluR6 when co-assembled with myc-KA2 and Thr-675 mutants in COS-7 cells. Both myc-KA2(T675A) and myc-KA2(T675E) reduced surface expression of GFP-GluR6 by Ͼ10-fold. B, surface expression of homomeric myc-KA2-R/A is reduced by mutation of Thr-675 to either alanine or glutamate in cell ELISA assays. the precise site of retention remains unclear.
An increased rate of receptor degradation in principle could account for the reduction in surface expression of KA2-contain-ing receptors. We tested this possibility in pulse-chase experiments to compare the degradation rate of homomeric myc-KA2 and Thr-675 mutant receptors (Fig. 6). Transfected HEK 293 (data not shown). Quantitative densitometry yielded the relative density data shown in the graph, which were fit best by a bi-exponential decay function. There were no significant differences in the predominant (faster) time constants of decay for myc-KA2 (solid squares, 1.4 h), myc-KA2(T675A) (open circles, 1.4 h), and myc-KA2(T675E) (open triangles, 1.8 h). These data demonstrate that the reduction in surface expression is not accounted for by an increased rate of degradation of the KA2 subunits lacking glutamate binding sites. DISCUSSION It has become clear recently that excitatory synaptic strength can be modulated by regulation of receptor trafficking and biosynthesis (1,9). The carboxyl-terminal domains of ionotropic glutamate receptor subunits, which are cytoplasmic, are the primary site in the subunit proteins that contains critical determinants of subcellular and synaptic receptor localization. Recent studies have elucidated a number of cytoplasmic determinants of kainate receptor trafficking and biosynthesis, revealing that a diverse array of amino acid motifs play critical roles in the intracellular trafficking (14 -17). In addition, it has been shown that AMPA and kainate receptors contain noncytoplasmic determinants of subcellular localization. For AMPA receptors, these include the Q/R site in the pore-forming re-entrant loop (18,24) and a short motif in the amino-terminal domain (35). As well, both AMPA and kainate receptors appear to require an intact ligand-binding domain for forward progression in the secretory pathway from the ER (25,26), although it has been unclear how ligand binding operated as a permissive mechanism for forward trafficking.
Our results support this role for glutamate binding as a quality-control checkpoint in trafficking and surface expression of kainate receptors and clarify potential mechanisms underlying this regulation (25,26). We found that homomeric and heteromeric kainate receptors lacking affinity for their endogenous ligand are sequestered intracellularly and that retention occurs even in receptors only partially composed of bindingdeficient subunits. Dominant ER retention of receptors incorporating binding-deficient mutants occurred in both heterologous cell lines and cultured neurons. Mutant subunits were not compromised in their ability to co-assemble into multimeric receptors with wild-type receptor subunits nor were they degraded at a more rapid rate than wild-type KA2 subunits. These results suggest that each component subunit within a tetrameric glutamate receptor requires an intact binding site for transit forward from the ER.
Dominant retention by the mutant KA2 subunits could be accounted for mechanistically in at least three ways: mutations caused misfolding of the subunit protein sufficient to activate quality-control systems in the ER, binding-related structural changes necessary for forward trafficking were prevented, or channel activation and gating equivalent to agonist-evoked steady-state currents were necessary for forward transit through the secretory pathway. We believe that misfolding is unlikely, because no change was detected in the degradation rate of wild-type and mutant KA2 subunit proteins, suggesting elimination of the ligand binding site did not activate the ER-associated degradation quality control system or other molecular chaperones that would promote more rapid elimination of aberrant receptors. This result is qualitatively similar to that observed for Q/R site-edited GluR2 AMPA receptor subunits, which reside stably in the ER as monomers or dimers and are not subject to accelerated degradation (18,24). However, GluR2(R) subunits are substantially longer-lived in the ER compared with KA2 subunits, which exhibited a fairly rapid turnover rate of a couple of hours in our experiments. This rate of degradation is also significantly faster than that of homo-meric GluR6 receptors, which are stable at least 4 h when trapped in the ER by brefeldin treatment before appreciable degradation occurs (17). This difference in rates suggests that assembly of heteromeric GluR6/KA2 receptors might be ratelimited by the degradation of KA2 subunits, which could in part explain the common observation that mixtures of homomeric GluR6 and heteromeric GluR6/KA2 receptors are formed despite the preponderance of KA2 cDNA in transfection conditions like those used in our study. In summary, these data demonstrate that the homomeric form of KA2 subunits, which do not normally traffic to the plasma membrane, are targeted by quality control systems in the ER but mutation of the glutamate binding site does not alter this turnover rate.
Because altered degradation does not appear to play a role in the retention mechanism, we conclude that glutamate binding to ER-resident receptors provides a read-out of the receptor functional state as a mechanism for biosynthetic quality control. We attempted to test this hypothesis more directly by incubating kainate receptor-expressing COS-7 cells in the antagonist CNQX for 24 h prior to cell ELISA assays, but this assay did not result in reductions in surface expression of the receptors (data not shown). These negative data are difficult to interpret, however, because it is possible that CNQX did not penetrate the critical intracellular compartment containing nascent receptors binding glutamate. Additionally, CNQX is a competitive antagonist and might have not been present in sufficient concentration to completely displace endogenous glutamate binding. The concentration of glutamate in ER is unknown, but relatively low concentrations induce conformational changes in glutamate receptor structure correlated with gating and desensitization; for example, application of 1-5 M glutamate effectively desensitized kainate (36) and AMPA and NMDA receptors (37). Thus, it would not require particularly high levels of ambient ER glutamate to elicit structural changes in fully assembled receptors.
Glutamate binding to receptors in ER is predicted to have two primary consequences: desensitization-related structural changes in the "extracellular" domains of the subunits and equilibrium cation currents through the receptor channels. Our results are most consistent with the former as a mechanism for quality control mediated by ligand binding. Desensitization of ionotropic glutamate receptors has been modeled structurally from crystallographic analysis of the GluR2 subunit ligandbinding domain as a destabilization of the subunit interface domains (38). This rearrangement in principle could expose chaperone binding sites on the receptor that are necessary for further transit in the biosynthetic pathway. Alternatively, partially assembled subunits might contain ER retention/retrieval motifs that are masked upon tetramerization and glutamate binding. We favor structural rearrangements as being relevant to quality control rather than ion permeation, because the T675E and T675A binding site mutations cause retention of homomeric myc-KA2-R/A or myc-KA2⌬R6 receptors. These receptors bind glutamate and traffic to the plasma membrane, unlike wild-type KA2 receptors, but nonetheless do not gate currents in response to glutamate (14), and thus it is unlikely that ion permeation occurs in the ER. This hypothesis is consistent with previous observations that mutations in gating domains also causes retention of homomeric GluR6 subunits (26); such mutations might occlude necessary conformational rearrangements necessary for chaperone recognition.
It is also possible that conformational changes associated with ligand binding are necessary for efficient assembly of tetrameric receptors, and that mutation of ligand binding or gating sites occludes oligomerization. AMPA receptors assemble as dimer of dimers, and this process is regulated by the amino acid residing at the Q/R site in GluR2 subunits (18,24). It is not clear if kainate receptors, which exhibit much greater variability than GluR2 in the degree of their Q/R site RNA editing in situ (39,40), are subject to the same rules of oligomerization, but domain-swapping experiments suggested that this might be the case (41). Our co-immunoprecipitations of GFP-GluR6 with mutant KA2 subunits suggest that, despite retention, at least one subunit of KA2 can assemble with a GluR6 subunit in the ER, although we do not know if tetrameric receptors are formed. Similarly, myc-KA2-R/A receptors traffic forward to the plasma membrane (14), but because these receptors remain insensitive to glutamate we cannot be sure they are tetrameric rather than monomeric or dimeric receptor subunits.
In summary, we have identified an important quality-control mechanism that controls functional expression of ionotropic glutamate receptors through ligand-induced conformational changes in the ER. Future correlation of binding and gating mutations with changes in intracellular trafficking and assembly will facilitate an understanding of the critical elements of this regulatory process.