The Integrity of the Glycine Co-agonist Binding Site of N-Methyl-d-aspartate Receptors Is a Functional Quality Control Checkpoint for Cell Surface Delivery*

N-Methyl-d-aspartate receptors are a subclass of ligand-gated, heteromeric glutamatergic neurotransmitter receptors whose cell surface expression is regulated by quality control mechanisms. Functional quality control checkpoints are known to contribute to cell surface trafficking of non-N-methyl-d-aspartate glutamate receptors. Here we investigated if similar mechanisms operate for the surface delivery of NMDA receptors. Point mutations in the glycine binding domain of the NR1-1a subunit were generated: D732A, a mutation that results in an ∼3 × 104 decrease in glycine binding affinity; D732E, a conservative change; and D723A, a residue in the same NR1-1a domain that has no effect on glycine binding affinity. Each NR1-1a subunit was co-expressed with NR2A in mammalian cells. Immunoblotting and immunoprecipitations showed that all mutants were expressed to similar levels as wild-type NR1-1a and associated with NR2A. Cell surface expression measured by an enzyme-linked immunosorbent assay found that whereas NR1-1a (D732E)/NR2A and NR1-1a (D723A)/NR2A trafficked as efficiently as NR1-1a/NR2A, there was a 90% decrease in surface expression for NR1-1a (D732A)/NR2A. This was confirmed by confocal microscopy imaging and cell surface biotinylation. Further imaging showed that NR1-1a (D732A) and co-transfected NR2A co-localized with an endoplasmic reticulum marker. Dichlorokynurenic acid, a competitive glycine site antagonist, partially rescued surface expression. Mutation of the NR1-1a ER retention motif showed that the ligand binding checkpoint is an early event preceding endoplasmic reticulum sorting mechanisms. These findings demonstrate that integrity of the glycine co-agonist binding site is a functional checkpoint requisite for efficient cell surface trafficking of assembled NMDA receptors.

ter receptors. They are important due to the pivotal role they play in synaptic transmission, synaptic plasticity, and synaptogenesis during the development of the central nervous system. They are activated by the binding of the co-agonists, L-glutamate and glycine, and the alleviation of voltage-dependent blockade by magnesium ions. Receptor activation results within milliseconds in the opening of an integral ion channel that has a high permeability for calcium ions. The dysfunction of NMDA receptors is implicated in a broad spectrum of neurodegenerative and psychiatric disorders. This is primarily due to the permeability properties of the NMDA receptor channel; overactivation results in inappropriate increases in intracellular calcium ion concentration and excitotoxic neuronal cell death (reviewed in Ref. 1).
NMDA receptors are a family of ligand-gated, heteromeric, integral membrane proteins. There are seven NMDA receptor genes encoding the NR1, NR2A-NR2D, and NR3A-NR3B subunits. Functional NMDA receptors are formed from the coassembly of the obligatory NR1 subunit with NR2 and/or NR3 subunits. The quaternary structure of NMDA receptors is still not definitively proven, although the weight of experimental evidence supports a tetrameric structure comprising two NR1 and two NR2 subunits (reviewed in Ref. 2).
The cell surface expression of NMDA receptors in neurons is tightly regulated. This is important, since, as mentioned earlier, overactivation of receptors or inappropriate increase in receptor number leads to an excessive influx of calcium ions and resultant excitotoxicity and neuronal cell death as found in ischemia (15). Conversely, NMDA receptor hypofunction, exemplified by either a decrease in receptor activity or decreased cell surface receptor number, is associated with schizophrenia (16). With regard to NMDA receptor activity, there is therefore a fine balance between levels that are necessary for normal neuronal functional and those that result in neuropathological conditions. NMDA receptor cell surface expression is regulated by standard quality control mechanisms applicable to all membrane proteins (17). These include primary control checkpoints (i.e. the detection and elimination of misfolded subunits by chaperones in the endoplasmic reticulum (ER) and secondary, more protein-specific checkpoints, such as the masking of ER retention motifs that ensure that only correctly folded and assembled receptors are trafficked to the cell surface (reviewed in Refs. 17 and 18)). For non-NMDA, ionotropic glutamate receptors, additional, functional quality control checkpoints have been shown to contribute to the regulation of cell surface trafficking. It was reported that the integrity of the glutamate binding site is requisite for export of the Caenorhabditis elegans non-NMDA glutamate receptor (GLR-1) (19) and the mammalian non-NMDA GluR6 subunit-containing kainate receptors from the ER (20,21). Further, exit from the ER of GluR2 subunits is controlled by RNA editing at the Q/R site in the poreforming re-entrant membrane domain (22). Usually, GluR2 is retained in the ER, but if the edited site is reversed such that the amino acid is glutamine, as found for the other non-NMDA receptor subunits, rapid ER exit is observed (22). Third, the densitization capability of non-NMDA receptors has also been identified as an ER checkpoint. Priel et al. (23) found that only GluR6 homomers that can desensitize reach the cell surface, whereas Greger et al. (24) reported that GluR2 tetramers containing the point mutation, L483Y that locks the receptor in an active, nondesensitizing conformation state, are retained in the ER.
For NMDA receptors, ER retention motifs in the intracellular C-terminal domain and within the third membrane domain of NR1 and NR2 subunits and ER export motifs in the C termini have been shown to regulate surface trafficking (25)(26)(27)(28)(29). Here we investigated whether determinants within the extracellular N-terminal, agonist binding region contribute additional functional quality control checkpoints, as found for the non-NMDA glutamate receptors. Specifically, we have investigated whether the integrity of the glycine co-agonist binding site is required for cell surface delivery of NMDA receptors. We employed a site-directed mutagenesis strategy for the NR1 glycine binding site in conjunction with a cell surface assay to demonstrate that this is indeed the case; we have investigated the relationship between this checkpoint and ER retention mechanisms and we report partial rescue of cell surface trafficking of the glycine binding site mutant by a pharmacological chaperone.
Mammalian Cell Transfection-Human embryonic kidney (HEK) 293 cells were cultured and transfected with either wildtype or mutant pCISNR1-1a, pCISNR1-4b clones alone or in pCISNR1-1a/pCISNR2A, pCISNR1-4b/pCISNR2A combinations using the calcium phosphate method with a total of 10 g of DNA (30). For transfections containing NR1/NR2A clones only, cells were incubated post-transfection in the presence of 1 mM ketamine to prevent NMDA receptor-mediated cytotoxicity. Cells were harvested 20 h post-transfection, and homogenates were prepared, adjusted to 0.5 mg protein/ml, and analyzed by quantitative immunoblotting. For transfections where cell surface NMDA receptor expression was measured, HEK 293 cells were subcultured overnight prior to transfection in poly-D-lysine (100 g/ml)-coated 24-well dishes, and 0.5 g of total plasmid DNA was used per well. For transfections where cells were imaged by confocal microscopy, a coverslip was added to each well of a 24-well tissue culture plate. Poly-Dlysine (100 g/ml; 1 ml) was added and incubated for 3 h at room temperature, washed two times with 1 ml of H 2 O, and left to air-dry under sterile conditions. HEK 293 or COS-7 cells were added and subcultured overnight prior to transfection with 0.5 g of total plasmid DNA/well. For the pharmacological chaperone experiments, glycine or 5,7-dichlorokynurenic acid (DCKA; Tocris Bioscience, Bristol, UK) was added with ketamine to the cell culture media post-transfection. A stock concentration of DCKA was prepared in 10% (v/v) DMSO; this was diluted to the appropriate DCKA concentrations to yield a final concentration of 0.2% (v/v) DMSO in the cell culture media post-transfection for all test and control transfections.
Cell Cytotoxicity-HEK 293 cells were co-transfected with various pCISNR1/pCISNR2A wild-type and mutant constructs. Twenty hours post-transfection, cell cytotoxicity was determined using the Promega CytoTox 96 TM cytotoxicity assay according to the manufacturer's instructions and as described previously (32).
Immunoblotting-Immunoblotting was performed as previously described using 25-50 g of protein/sample precipitated using the chloroform/methanol method and SDS-PAGE under reducing conditions in 7.5% polyacrylamide slab minigels (34). Rabbit horseradish-linked secondary antibodies (Amersham Biosciences) were used at a final dilution of 1:2000, and immunoreactivities were detected using the ECL Western blotting system.
Determination of NMDA Receptor Cell Surface Expression by Enzyme-linked Immunosorbent Assay-The measurement of cell surface NR1 and NR1/NR2A receptors was carried out using an enzyme-linked immunosorbent assay (ELISA) with affinity-purified antibodies directed against extracellular epitopes of NR1 or NR2A subunits (i.e. anti-NR1-(35-53) (1 g/ml) and anti-NR2A-(44 -58) (0.125 g/ml)), as described by Papadakis et al. (34). For the measurement of total subunit expression in the transfected cells, cells were fixed with 4% (w/v) paraformaldehyde for 15 min at room temperature, and cells were washed with 1ϫ phosphate-buffered saline (PBS), permeabilized for 5 min at room temperature with 0.25% (v/v) Triton X-100 in PBS, washed with 1ϫ PBS, and incubated with 4% (w/v) milk powder in PBS for 30 min at room temperature, and the assay was completed as for the determination of cell surface expression as above.
Biotinylation of Cell Surface Proteins-HEK 293 cells were transfected with either pCISNR1-1a/pCISNR2A or pCISNR1-1a (D732A)/pCISNR2A, and 20 h post-transfection, cell culture medium was aspirated, and cells were incubated with 6 ml of PBS, 0.5 mM EDTA, pH 8.0, for 10 min with shaking. The cell suspension was centrifuged at 200 ϫ g for 5 min at 4°C and washed with 3 ϫ 10 ml of ice-cold PBS, pH 8.0, by centrifuging at 200 ϫ g at 4°C for 5 min. Cells were resuspended in 5 ml of PBS, pH 8.0, and incubated with 1 ml of 10 mM EZ-Link Sulfo-NHS-LC Biotin in double-distilled H 2 O (Pierce) at 4°C for 30 min. Cells were washed with 3 ϫ 10 ml of ice-cold PBS containing 100 mM glycine. Immunoprecipitations were carried out using anti-NR1 C2 and non-immune Ig antibodies as above, except that the solubilization buffer was PBS containing all protease inhibitors, 1% (v/v) Triton X-100, and 240 mM NaCl. Immune pellets were analyzed by immunoblotting using 1 g/ml ImmunoPure streptavidin horseradish peroxidase conjugate (Pierce) to detect biotinylated proteins.
Endoglycosidase H and N-Glycosidase F Digestion of HEK 293 Transfected Cells-HEK 293 cells were harvested 20 h posttransfection as above, cell homogenates were prepared, and samples (30 g of protein) were denatured in Denaturing Buffer (50 l; New England Biolabs (UK) Ltd., Hitchin, UK) at 100°C for 10 min. For endoglycosidase H digestion, 10ϫ G5 reaction buffer (7 l) and 500 units of endoglycosidase H were added to a final volume of 70 l. For N-glycosidase F digestion, 10ϫ G7 reaction buffer (7 l), 10% (v/v) Nonidet P-40, and 500 units of N-glycosidase F were added to a final volume of 70 l. Samples were incubated for 1 h at 37°C. Controls were protein samples incubated for 1 h at 37°C or at 4°C in the absence of enzymes. All samples were precipitated using the chloroform/methanol method and analyzed by immunoblotting.

Confocal Microscopy Imaging of NMDA Receptors Expressed in Mammalian
Cells-For the imaging of surface-expressed NMDA receptors, cells were labeled live at 20 h post-transfection. Each coverslip of transfected cells was washed on ice for 5 min with 1 ml of DMEM containing 25 mM HEPES and 10% (v/v) fetal bovine serum. The primary antibody, (anti-NR2A-(44 -58 Cys); 100 l of 2 g/ml) was added and incubated on ice for 30 min. Coverslips were washed with 3 ϫ 1 ml of DMEM containing 25 mM HEPES and 10% (v/v) fetal bovine serum, incubated with a dilution of 1:5000 anti-rabbit-Ig Alexa Fluor 594 (Invitrogen) for 30 min on ice, and washed with 3 ϫ 1 ml of DMEM containing 25 mM HEPES and 10% (v/v) fetal bovine serum followed by 2 ϫ 1 ml of PBS. Cells were fixed by incubation with 4% (w/v) paraformaldehyde in PBS for 10 min at room temperature, washed with 2 ϫ 1 ml of 25 mM HEPES and 10% (v/v) fetal bovine serum followed by 2 ϫ 1 ml of PBS, and mounted onto a microscope slide using 10 l of mounting solution containing an anti-fading agent (CitiFluor; Citifluor Ltd., Leicester, UK). The coverslips were sealed and kept at 4°C until analysis. For the imaging of intracellular NMDA receptors, the ER, and the Golgi, transfected HEK 293 or COS-7 cells were first fixed for 10 min at room temperature with 4% (w/v) paraformaldehyde in PBS. Cells were washed with 3 ϫ 1 ml of DMEM containing 25 mM HEPES and permeabilized with 0.2% (v/v) Triton X-100 in DMEM containing 25 mM HEPES for 5 min at room temperature. Antibody labeling was then as for immunodetection of surface-expressed NMDA receptors, except that all incubations were at room temperature, and the antibodies used were as follows: for HEK 293 cells, anti-NR2A-(44 -58 Cys) (0.5 g/ml) and for COS-7 cells, anti-NR1 C2 (2 g/ml) and anti-NR2A-(1381-1394) (2 g/ml). The concentrations of mouse anti-calreticulin and mouse anti-58-kDa Golgi protein antibodies were 4 g/ml with 1:400 anti-mouse Alexa Fluor 488 (Invitrogen). Transfected HEK 293 and COS-7 cells were viewed with a Zeiss LSM-510 confocal microscope.

Rationale for the Selection of NMDA Receptor NR1 Subunit
Mutations-As described in the Introduction, several groups have identified key amino acid residues involved in glycine binding to assembled NMDA receptors. We reviewed the published information to determine NR1 mutations that gave the maximum change in terms of decreased apparent affinity for channel activation by glycine and that were also implicated in agonist binding from the x-ray crystallographic studies. The mutation selected was NR1-1a (D732A), which is in the S2 domain. This mutation had a reported decrease in glycine apparent affinity of 35,000-fold for heteromeric NR1 (D732A)/ NR2B receptors expressed in Xenopus oocytes (6). For controls, the conservative mutation NR1-1a (D732E) was generated. Also, a mutation was created in a residue that was within the NR1 S2 domain, but when mutated, it did not result in a change in glycine apparent affinity (6), and the x-ray crystal structure revealed that it did not participate in ligand binding (i.e. NR1-1a (D723A)). A schematic summary of the mutations and the relevant parts of the x-ray crystal structures are shown in Fig. 1.

Characterization of the Mutant NMDA Receptor NR1 and NR2
Subunits-In the first instance, all mutated subunit clones were transfected with the appropriate wild-type subunit clones to determine if each was expressed to the same level. Thus, transfected cells were harvested, and total cell homogenates were analyzed by immunoblotting. The results are shown in Fig. 2, A and B. There was no significant difference in the overall expression levels between wild-type and mutated subunits. All NR1-1a subunits except NR1-1a (D732A) had a molecular mass of 121 Ϯ 1 kDa (n ϭ 18); the molecular mass of NR1-1a (D732A) was 119 Ϯ 1 kDa (n ϭ 17; p Ͻ 0.001). For all subunits, a dimer of NR1-1a immunoreactivity with molecular mass of 217 kDa was clearly visible. This is in agreement with previous reports of an NR1-NR1 disulfide-linked homodimer (34,35); it shows that mutations within the LAOBP domain do not affect dimer formation.
Cell cytotoxicity post-transfection is a useful biochemical means to measure functional cell surface NMDA receptor expression, since the cell death has been attributed to activation of functional cell surface NMDA receptors by L-glutamate and glycine in the cell culture media with a subsequent unregulated influx of Ca 2ϩ (30). To characterize further the NMDA receptor mutated subunits to confirm the decreases in agonist affinity conferred by the selected mutations, each was co-expressed with the wild-type NR2A subunit, and cell death was measured 20 h post-transfection. In the presence of ketamine, the noncompetitive NMDA receptor antagonist, there was no detectable cell death for any of the wild-type and mutant NR1/NR2 combinations (Fig. 2). In the absence of ketamine with wildtype NR1/NR2A transfectants normalized to 100% cell death, for the NR1-1a mutations, there was no significant difference in mortality between wild type and the control NR1-1a (D723A)/NR2A mutation. This was in contrast to the significant decreased cell mortality found for both NR1-1a (D732A)/NR2A and NR1-1a (D732E)/NR2A, where values were 9.9 Ϯ 2.0% (n ϭ 15) and 28.7 Ϯ 9.2% (n ϭ 4) compared with 100% for wild type. These findings are consistent with a decreased affinity for the glycine NR1-1a (D732A) and NR1-1a (D732E) mutations.

Inactivation of the Glycine Binding Site of NMDA Receptors Results in a Deficit in Cell Surface
Trafficking-To investigate if the glycine binding site mutation had an effect on cell surface trafficking of NMDA receptors, each mutant subunit was co-expressed with the appropriate wild-type subunit, and cell surface expression was measured qualitatively by confocal microscopy imaging and cell surface biotinylation and quantitatively by a cell surface ELISA. The results are shown in Fig. 3. The expression of NR1-1a (D732A)/NR2A in mammalian cells resulted in an 89.6 Ϯ 1.9% (n ϭ 17) decrease in cell surface expression compared with wild type; there was no significant difference in cell surface expression between wild type and the NR1-1a (D723A) control mutation, but surprisingly, despite yielding a decrease in cell mortality post-transfection, the conservative mutation NR1-1a (D723E)/NR2A showed no decrease in cell surface expression compared with NR1/NR2A receptors. The same findings were evident when NMDA receptor cell surface expression was assessed by biotinylation and by confocal microscopy imaging. For the biotinylation experiments, an ϳ120,000 kDa band (i.e. NR1 subunit) was detected by streptavidin horseradish peroxidase in anti-NR1 C2 immune pellets from NR1-1a/NR2A transfectants. No signal was detectable for NR1-1a (D732A)/NR2A in immune pellets, and no signal was detectable for either NR1-1a/NR2A or NR1-1a (D732A)/NR2A in control Ig immune pellets.
For the confocal microscopy studies, immunolabeling was carried out on intact and permeabilized transfected cells with anti-NR2A-(44 -58) antibodies. NR2A subunits are only trafficked to the cell surface when co-expressed with NR1 subunits  Endoglycosidase Sensitivity-Since NR1-1a (D732A)/NR2A receptors were not trafficked to the cell surface, their intracellular localization was investigated using confocal imaging in conjunction with antibody markers for the ER and the Golgi apparatus and by their sensitivity to endoglycosidase H digestion. For wild-type NR1/NR2A receptors, some co-distribution of both NR1 and NR2 subunit immunoreactivities was evident with the ER marker antibody (Fig. 4A, a-d and i-l), but this co-distribution was not apparent with anti-58-kDa Golgi protein immunoreactivity (Fig. 4B, a-d and i-l). This is not unexpected, since it is known that the NR1-1a splice form, the splice variant used here, can be retained in the ER due to an ER retention motif in the C-terminal C1 cassette (25)(26)(27). NR2A immunoreactivity may be detectable in the ER due to overexpression. In contrast, for NR1-1a (D732A)/NR2A receptors, a strong codistribution for both NR1 and NR2A immunoreactivities with immunostaining for the ER marker, calreticulin, was observed (Fig. 4A, e-h and m-p). Some co-distribution with the Golgi marker was evident, but this was less marked than observed for the ER. Thus, the majority of the NR1-1a (D732A) and co-expressed NR2A subunit immunoreactivities co-distribute with calreticulin and are thus trapped within the ER.
To substantiate the co-distribution studies, wild-type and mutant NR1-1a/NR2A expressed receptors were subject to deglycosylation using either endoglycosidase H or N-glycosidase F. Endoglycosidase H cleaves high mannose residues of N-linked glycoproteins, whereas N-glycosidase F cleaves N-glycoproteins at the asparagine where N-linked carbohydrate modification occurs. Since high mannose residues are trimmed back during the processing through the Golgi apparatus, N-linked glycoproteins become resistant to endoglycosidase H digestion, and this trait is used as a marker for ER localization. In Fig. 4C, it is shown that for wild-type and NR1-1a (D732A), NR1-1a (D732E), and NR1-1a (D723A) subunits, cleavage with either endoglycosidase H or N-glycosidase F results in a single band at ϳ90 kDa. In contrast, for the co-expressed NR2A subunits a different pattern was observed for the NR2A subunits co-expressed with wildtype or control NR1-1a mutant subunits compared with NR1-1a (D732A). For the control samples, a single band was present following N-glycosidase F digestion with a decrease of ϳ13 kDa in size. After endoglycosidase H treatment, an additional higher molecular weight species (a decrease of ϳ7 kDa) was observed (i.e. an endoglycosidase H-resistant band). This observation is explained by the NR2A subunit having passed through the Golgi to the cell surface. This additional band was not observed in NR1-1a (D732A)/ NR2A transfections, implying that here, in agreement with the confocal imaging studies, that the NR2A subunit is now retained in the ER.
Immunoprecipitations Demonstrate That NR1 Trafficking Mutants Associate with Wild-type NR2A Subunits-Since NR1-1a (D732A) was expressed at the same level as the wildtype subunit (Fig. 2), the deficit in cell surface expression may be a result of an inability to associate with NR2 subunits, since it is known that they are requisite for the trafficking of NR1-1a subunits to the cell surface (36). To investigate this possibility, NR1-1a/NR2A, NR1-1a (D732A)/NR2A, and NR1-1a (D732E)/ NR2A subunits were co-expressed in HEK 293 cells, immunoprecipitations were carried out using anti-NR1 C2 antibodies (or non-immune Ig as a negative control), and immune pellets were probed for reactivity with both anti-NR1 and anti-NR2A antibodies. For all of the NR1-1a/NR2A combinations, anti-NR1 and NR2A immunoreactivities were found in the immune but not in the control pellets (Fig. 5). Quantification of the immunoblots showed that there was no significant difference with respect to the percentage association of NR2A subunits with each NR1-1a. Thus, NR1-1a (D732A) subunits do co-as- semble with NR2A subunits with efficiency comparable with that of wild-type NR1-1a/NR2A association.
Rescue of Cell Surface Trafficking by the Use of a Pharmacological Chaperone-There is a growing literature showing that pharmacological chaperones can rescue misfolded proteins and overcome their aberrant targeting and function. Examples for receptor proteins include the rescue of ER-retained vasopressin V2 receptors by the cell-permeable, nonpeptidic V2 antagonists SR121463A (37), SR49059 (38), and SR121463B (38, 39); of vasopressin V1b/V3 receptors by the nonpeptide antagonist SSR149415 (40); of ER-retained ␦-opioid receptors by the antagonist naltrexone (41,42); of ER-retained dopamine D1 receptors by agonists but not antagonists (43); and the up-regulation of ␣4␤2 nicotinic acetylcholine receptors by nicotine (44,45). To determine if pharmacological chaperones can rescue the cell surface expression of the NMDA receptor NR1-1a (D732A) mutant, we investigated the effects of agonists and antagonists on cell surface trafficking. Thus, HEK 293 cells were co-transfected with either NR1-1a/NR2A or NR1-1a (D732A)/NR2A clones, cultured post-transfection in the pres-ence of 1 mM ketamine and either glycine or the competitive glycine site antagonist, DCKA; cell surface and total expression were measured at 20 h; and the percentage of cell surface trafficking was determined. Glycine in the 1-30 mM concentration range in the cell culture media had no effect on the surface trafficking of either wild-type NR1-1a/ NR2A or NR1-1a (D732A)/NR2A receptors (Fig. 6). DCKA in the 0 -1 mM concentration range had no effect on the surface trafficking of wild-type NR1-1a/NR2A receptors. It did, however, increase NR1-1a (D732A)/NR2A surface trafficking in a concentration-dependent manner (Fig. 6). An approximate 3.4-fold increase in NR1-1a (D732A)/NR2A receptor surface trafficking was found at 1 mM DCKA, the highest concentration permissible due to solubility factors.
The Ligand Binding Quality Control Checkpoint Is an Early Event in Intracellular Processing of NMDA Receptors Preceding ER Protein Sorting Mechanisms-There are eight NR1 splice variants, NR1-1a, NR1-1b; NR1-2a, NR1-2b; NR1-3a, NR1-3b; and NR1-4a, NR1-4b. These splice variants have been shown to contain different motifs governing their cell surface expression. For example, NR1-1a and NR1-1b, when expressed alone in mammalian cells, are not trafficked to the cell surface. This is in contrast to the other six variants, where each is trafficked to the cell surface, but the efficiency at which this occurs varies, with the NR1-4a and NR1-4b splice forms yielding the highest cell surface expression for single NR1 subunits (25). Standley et al. (26) and Scott et al. (27) later demonstrated using tac-NR1 C-terminal reporter constructs that the inability of NR1-1a subunits to be expressed alone at the cell surface was due to an RRR retention motif in the C1 exon; NR1-4b subunits do not have the retention motif, but they do have an ER export motif, TVV, in the C2Ј cassette (27,28). To investigate if these regulatory trafficking mechanisms were affected by the ligand binding site quality control checkpoint, the cell surface expression of NR1-4b (D753A) (the equivalent NR1-1a (D732A) mutation, but due to the insertion encoded by exon 5, the amino acid numbering increases) and NR1-1a (RRR 3 AAA, residues 893-895; D732A; i.e. NR1-1a (D732A/ AAA)) mutant constructs were investigated. When expressed alone, in contrast to NR1-1a, wild-type NR1-4b subunits are efficiently trafficked to the cell surface (Fig. 7A). Introduction of the D753A mutation resulted in an approximately 6-fold HEK 293 cells were transfected in triplicate with wild-type NR1-1a/NR2A or mutated NR1-1a and NR2A subunits, and either the cell surface NMDA receptor expression was determined 20 h post-transfection by an ELISA using antibodies directed against an NR2A extracellular epitope (A) or the biotinylation of cell surface-expressed NMDA receptors (B), or alternatively, surface and total NR2A subunit expression was imaged by confocal microscopy using the same anti-NR2A antibody (C), all as described under "Experimental Procedures." For A, the results are normalized values, where 100% ϭ cell surface expression of wild-type NR1-1a/NR2A NMDA receptors. Values are the means Ϯ S.E. from at least n ϭ 8 independent transfections. In B, HEK 293 cells were transfected with either NR1-1a/NR2A or NR1-1a (D732A)/NR2A, intact cells were biotinylated, detergent extracts were prepared, and immunoprecipitations were carried out with either anti-NR1 C2 or control nonimmune Ig, and the pellets were analyzed by immunoblotting using streptavidin horseradish peroxidase (HRP) to detect biotinylated proteins. Lanes 1 and 4, detergent extract; lanes 2 and 4, control non-immune pellet; lanes 3 and 6, anti-NR1 C2 immune pellet. The immunoblot is representative of n ϭ 2 independent experiments. The positions of molecular mass standards (ϫ10 3 kDa) are shown on the right. In C, the images are a single confocal section of a selected transfected cell representative of 100 cells from n ϭ 3 independent transfections. Scale bar, 10 m. decrease in cell surface trafficking (i.e. a similar level to that determined for the cell surface trafficking of NR1-1a) (D732A). This implies that the ligand binding quality control checkpoint precedes the recognition of the ER export sequence so that the NR1-4b (D753A) subunit, like both NR1-1a and NR1-1a (D732A), is retained in the ER. The same decrease in cell surface trafficking was evident when the mutant NR1-4b (D753A) was coexpressed with NR2A subunits (Fig. 7B).
Mutation of the RRR mutation in NR1-1a resulted in an increase of ϳ1.9-fold for single subunit expression. Expression of NR1-1a (D732/AAA) resulted in a significant decrease in cell surface expression compared with NR1-1a (AAA). In fact, there was now no significant difference in cell surface expression between NR1-1a (D732A) and NR1-1a (D732A/ AAA). Thus, as for NR1-4b, the findings imply that the ligand binding quality control checkpoint is an early determinant of trafficking fate and precedes the mechanisms that permit export from the ER. No differences in surface trafficking were found between NR1-1a/ NR2A and NR1-1a (AAA)/NR2A or between NR1-1a (D732A)/NR2A and NR1-1a (D732A/AAA)/NR2A.

DISCUSSION
The assembly and forward trafficking of NMDA receptors to synaptic membranes, as for other multisubunit integral membrane proteins, is regulated initially by quality control mechanisms in the ER (17). These mechanisms ensure that only correctly folded, assembled, functional receptors can progress along the secretory pathway to the plasma membrane. They include the appropriate folding of the protein to the mature conformational state within the ER lumen by molecular chaperones, such as protein-disulfide isomerase for the formation of disulfide bonds, signal peptide cleavage, and appropriate glycosylation (17). Further, second-  ary ER quality control points, such as the ER retention sequence found in certain NR1 splice forms, ensure that only assembled NMDA receptors exit the ER, since co-association with NR2 subunits and also scaffolding molecules mask the ER retention motif such that forward trafficking can now occur. In this paper, we have described a novel ER primary, functional checkpoint control. Thus, similarly to their non-NMDA glutamate receptor homologues and under the experimental conditions employed here, the integrity of the NR1 glycine co-agonist binding site is requisite for efficient ER exit and cell surface trafficking of assembled NR1/NR2A NMDA receptors. We demonstrate that, for the NR1 glycine binding subunit, this quality control mechanism probably occurs after NR1/NR2 subunit assembly and precedes ER retention mechanisms. The  deficit in trafficking can be partially rescued by a pharmacological chaperone, the competitive glycine site antagonist, DCKA.
The capability of the various NR1 mutants to traffic to the cell surface was correlated with their decrease in glycine agonist apparent affinity. For example, the NR1-1a (D732A) mutation resulted in a reported 35,000 decrease in the EC 50 for the activation of NMDA NR1/NR2B receptors by glycine (6). Here a 90% decrease in cell surface expression was found. The conservative mutation, NR1-1a (D732E), that gave only a 4200 decrease (6) was shown to traffic normally to the cell surface with the same efficiency as wild-type NR1-1a/NR2A. Of course, it could be reasoned that introducing the mutation just results in a misfolded polypeptide that is then targeted for degradation via the ER-associated degradation pathway; however, several points counter this hypothesis. These include the fact that, as above, cell surface NMDA receptor expression is directly correlated with the decrease in agonist apparent affinity; that the nonconservative mutation, NR1-1a (D723A) located in the S2 region in the vicinity of the glycine binding site but not actually implicated in the binding of agonist was trafficked normally; and the observation that the NR1-1a (D732A) mutated subunit co-assembled efficiently with NR2A. If general misfolding were detected, NR1-1a (D732A) might be expected to be targeted for degradation prior to assembly with NR2 subunits (17).
Since inactivation of the glutamate binding site of non-NMDA receptor subunits resulted in their ER retention, Fleck (46) speculated that glutamate itself may function as a molecular chaperone. The concentration of free glutamate in normal brain tissue is 4 -16 mM (47). This is ϳ20,000-fold higher than the concentrations required to activate non-NMDA receptor channels. Further, free glutamate was shown by immunoelectron microscopy to be associated with the ER (48). It was also suggested that glutamate can enter the ER via cystine-glutamate transporters (49). Like glutamate, glycine is a constituent of most proteins, and it has a specialized role as a neurotransmitter. In the lower brain stem and spinal cord, it is the major inhibitory neurotransmitter. In the brain, glycine can also function as an inhibitory neurotransmitter via activation of strychnine-sensitive glycine receptors, but it is also a co-agonist for the activation of NMDA receptors. Glycine in its free form has a concentration similar to that found for glutamate (i.e. in the range 4 -16 mM) (47). This is an ϳ5000 -20,000 higher concentration than that required for activation of NMDA receptors. Free glycine has a widespread distribution in neurons (50 -52). At the ultrastructural level, however, there is little information to suggest that free glycine is associated with the ER (49). An alternative endogenous small molecule chaperone may be D-serine, now a recognized physiological activator at the glycine site of NMDA receptors (reviewed in Ref. 53).
To investigate further the role of glycine as a possible molecular chaperone, its potential as a pharmacological chaperone for the rescue of misfolded proteins was investigated. Glycine had no effect on the surface trafficking of wild-type or mutant NR1/NR2A NMDA receptors. However, the competitive glycine site antagonist, DCKA, enhanced surface trafficking in a dose-dependent manner (Fig. 6). The determination of cell surface NMDA receptors by the ELISA as used here is a steady state measurement. Thus, although an increase in surface traf-ficking was apparent in the presence of DCKA, this may indeed be due to an increased ER export of NR1-1a (D732A)/NR2A receptors. The possibility has to be considered, however, that it may be a result of a decrease in receptor internalization or, simply, stabilization that then results in enhanced NR1-1a (D732A) subunit expression. The latter is unlikely, since no difference was found between the total expression of wild-type NR1-1a versus NR1-1a (D732A) subunits (Fig. 2). It has been reported that in neurons, in the presence of glutamate, glycine primes NMDA receptors for clathrin-dependent endocytosis (54). This was reversed by the glycine site antagonist, L689560 (54). L689560 treatment alone did not permit subsequent glycine-induced internalization, but no data were presented to show whether the antagonist stabilized or increased NMDA receptor surface expression (54).
A second point to consider is that in order to function as a chaperone, glycine and DCKA must permeate the plasma membrane, and they must also enter the ER, since the NR1 glycine binding site is in the lumen of the ER. The octanol-water partition coefficient, logP, a parameter used to predict the membrane pemeabilities of small molecules, was predicted using Molinspiration (available on the World Wide Web) for DCKA and glycine. The values were as follows: DCKA, miLogP ϭ 3.243; glycine, miLogP ϭ Ϫ2.55. These values are in support of DCKA being membrane-permeable. In addition, Asp 732 is not involved in the binding of DCKA to NR1 (13); thus, NR1-1a (D732A) subunits would retain high affinity for the antagonist requiring approximately micromolar DCKA concentrations for full receptor occupancy. In contrast, glycine is unlikely to be membrane-permeable, and this, coupled with the decreased affinity of NR1-1a (D732A) for glycine, would suggest that the intracellular, intra-ER concentrations would not be high enough to bind to the mutated NR1 subunit. Further studies with membrane-permeant glycine agonists and antagonists would show if stabilization of either the glycine co-agonist or antagonist binding sites can facilitate forward ER trafficking.
Since there is a well established ER retention motif in some of the NR1 splice variants, we exploited this to investigate its relationship with the glycine binding quality control checkpoint. In NR1 single subunit expression experiments, we found that mutation of the critical NR1 aspartate residue resulted in the intracellular retention of NR1-4b subunits despite the lack of the NR1 C1 exon ER retention motif and the presence of an ER export motif in the NR1 C2Ј exon. Further, the removal of the ER retention motif in NR1-1a by mutagenesis resulted in an increase in NR1-1a cell surface trafficking. The -fold increase (1.5-fold) in surface trafficking was less than that found by Standley et al. (26), who reported an increase of ϳ5-fold for an NR1-1a (AAA) C-terminal tail fused to a Tac reporter protein rather than the NR1-1a subunit per se. Nevertheless, this ER retention motif was shown to function, but NR1-1a (D732A/ AAA) was still intracellularly retained. These observations permitted the speculation that the ligand binding checkpoint preceded ER retention/ER export regulatory mechanisms. No differences in surface trafficking were found between NR1-1a/ NR2A and NR1-1a (AAA)/NR2A or between NR1-1a (D732A)/ NR2A and NR1-1a (D732A/AAA)/NR2A. This can be ex-plained by the presence of the NR2A subunit; in the heteromers, it is the association with NR2A that masks and overrides the ER retention motif to regulate forward trafficking of assembled NR1/NR2 receptors.
In summary, we have identified a novel, functional checkpoint that is requisite for efficient cell surface trafficking of assembled NR1/NR2 NMDA receptors. This quality control checkpoint control involves the extracellular, ligand binding domain of the NR1 subunit. This mechanism is analogous to that found to operate for the efficient cell surface delivery of non-NMDA glutamate receptors.