Appropriate NR1-NR1 disulfide-linked homodimer formation is requisite for efficient expression of functional, cell surface N-methyl-D-aspartate NR1/NR2 receptors.

A c-Myc epitope-tagged N-methyl-D-aspartate receptor NR1-2a subunit was generated, NR1-2a(c-Myc), where the tag was inserted after amino acid 81. NR1-2a(c-Myc) /NR2A receptors when expressed in mammalian cells are not trafficked to the cell surface nor do they yield cell cytotoxicity post-transfection. NR1-2a(c-Myc) was, however, shown to assemble with NR2A subunits by immunoprecipitation and [(3)H]MK801 radioligand binding assays. Immunoblots of cells co-transfected with wild-type NR1-2a/NR2A subunits yielded two NR1-2a immunoreactive species with molecular masses of 115 and 226 kDa. Two-dimensional electrophoresis under non-reducing and reducing conditions revealed that the 226-kDa band contained disulfide-linked NR1-2a subunits. Only the 115-kDa NR1-2a species was detected for NR1-2a(c-Myc)/NR2A. The c-Myc epitope is inserted adjacent to cysteine 79 of the NR1-2a subunit; therefore, it is possible that the tag may prevent the formation of NR1 disulfide bridges. A series of cysteine --> alanine NR1-2a mutants was generated, and the NR1-2a mutants were co-expressed with NR2A or NR2B subunits in mammalian cells and characterized with respect to cell surface expression, cell cytotoxicity post-transfection, co-association by immunoprecipitation, and immunoblotting following SDS-PAGE under both reducing and non-reducing conditions. When co-expressed with NR2A in mammalian cells, NR1-2a(C79A)/NR2A displayed similar properties to NR1-2a(c-Myc)/NR2A in that the 226-kDa NR1 immunoreactive species was not detectable, and trafficking to the cell surface was impaired compared with wild-type NR1/NR2 receptors. These results provide the first biochemical evidence for the formation of NR1-NR1 intersubunit disulfide-linked homodimers involving cysteine 79. They suggest that disulfide bridging and structural integrity within the NR1 N-terminal domain is requisite for cell surface N-methyl-D-aspartate receptor expression.

They are unique in that they require the binding of co-agonists, i.e. L-glutamate and glycine, together with the alleviation of a voltage-dependent blockade by magnesium ions for channel activation. NMDA receptors are critical mediators of excitatory neurotransmission in the brain, being pivotal for long term potentiation. They are also important as a therapeutic target post-ischemia. NMDA receptor channels are highly permeable to calcium ions, and thus overactivation leads to excitotoxic neuronal cell death (reviewed in Ref. 1). Seven genes encode NMDA receptor subunits NR1, NR2A-NR2D, and NR3A-NR3B. The NR1 subunit undergoes extensive splicing to yield eight variants NR1-1a,1b to NR1-4a,4b. Functional NMDA receptors are formed from the co-assembly of the obligatory NR1 glycine-binding subunit with NR2 and/or NR3 subunits, although it was recently shown (2) that NR1/NR3 subunits formed a novel, glycine-gated receptor. The quaternary structure of NMDA receptors is still not yet established. Experimental data support either a tetrameric structure comprising two NR1 and two NR2 subunits or a pentamer with reports of either three NR1 subunits co-assembled with two NR2s or two NR1s with three NR2s (1).
All NMDA receptor subunits share the same transmembrane organization. They have an extracellular N-terminal domain of ϳ550 amino acids, 3 transmembrane domains, 1 re-entrant membrane domain M2 that is thought to form the inner lining of the cation channel, and an intracellular C-terminal tail. The N-terminal region can be subdivided into two discrete domains based on the amino acid sequence homology between the NMDA receptor subunits and amino acid-binding proteins of bacteria. Thus, the first ϳ400 amino acids have homology with the bacterial periplasmic leucine-isoleucine-valine-binding protein (LIVBP), whereas amino acids 420 -550 show similarity with the bacterial lysine-arginine-ornithine and glutaminebinding protein (reviewed in Refs. 1 and 3). The glycine-and glutamate-binding sites of the NR1 and NR2 subunits, respectively, have been localized to the N-terminal regions, i.e. amino acids 420 -550 (S1), that are proximal to the membrane and the S2 extracellular loop found between TM3 and TM4 by a combination of site-directed mutagenesis studies (4 -8) and by analogy with the crystal structure of the soluble extracellular domain of the non-NMDA, amino ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) GluR2 glutamate receptor subunit (9). Most interesting, the N-terminal region distal to the membrane has been shown to be important for the assembly of non-NMDA receptors (10,11). This domain is requisite for GluR-GluR dimer formation, an initial step in the assembly pathway of functional non-NMDA receptors (11).
For the assembly of functional heteromeric NR1/NR2 NMDA receptors, it has been shown that NR2A subunits are localized to the plasma membrane only when co-expressed with an NR1 subunit (12) and that NR1-1a-(1-380) is important for NR1-1a/ NR2A subunit association (13). Furthermore, it was recently suggested that NR1-NR1 dimers are an initial step in NMDA receptor assembly (14). Indeed, using blue native-PAGE, Meddowes et al. (13) showed the presence of NR1-NR1 dimers following co-expression of NR1-1a/NR2A constructs in mammalian cells.
We have described previously an epitope-tagged NR2B subunit (15). The FLAG or c-Myc epitopes were introduced between amino acids 53 and 54 of the mature NR2B subunit to yield NR2B 53-54FLAG and NR2B 53-54c-Myc (15). NR2B FLAG/c-Myc behaved as wild-type NR2B co-assembling with NR1 subunits to form functional cell surface NMDA receptors following coexpression in mammalian cells or in neurons (15,16). The c-Myc epitope tag was subsequently engineered into the Nterminal domain of the NR1-2a splice variant at a position (amino acid 81) similar to that used successfully for the tagging of the NR2B subunit. Most interesting, this epitope-tagged form of the NR1-2a did not result in cell cytoxicity post-transfection suggesting that it was unable to form functional cell surface NMDA receptors. In this paper, we describe the biochemical properties of this NR1-2a c-Myc epitope-tagged NMDA receptor subunit. We show that characterization of the mutant NR1-2a subunit reveals insights into the importance of both NR1 disulfide bridging and the integrity of the N-terminal domain for both the assembly and cell surface expression of functional NMDA receptors.
Mammalian Cell Transfections-Human embryonic kidney (HEK) 293 cells were cultured and transfected with either single subunit NR1 or NR2 NMDA receptor clones, co-transfected with both NR1/NR2 subunit combinations using the calcium phosphate method with a total of 10 g of DNA and a ratio of 1:3 for NR1:NR2 transfections, respectively, or alternatively with NR1, NR2, and post-synaptic density 95 (PSD-95) clones using a total of 20 g of DNA and a ratio of NMDA receptor clones:PSD-95 of 1:1 (20,21). Cells were cultured in the presence of 1 mM ketamine post-transfection. They were harvested 24 -36 h post-transfection by centrifugation at 3000 ϫ g, and homogenates were prepared, adjusted to 0.5 mg protein/ml, and analyzed immediately for [ 3 H]MK801 and [ 3 H]MDL105,519 radioligand binding activities or alternatively frozen at Ϫ80°C and later analyzed by quantitative immunoblotting.
Cell Cytotoxicity-HEK 293 cells were co-transfected with various pCISNR1-2a/pCISNR2A constructs with or without pGWIPSD-95 c-Myc . Twenty hours post-transfection, cell cytoxicity was determined using the Promega CytoTox 96 TM cytotoxicity assay according to the manufacturer's instructions and as described previously (20).
Immunoblotting-Immunoblotting was performed as described previously using 25-50 g of protein/sample precipitated using the chloroform:methanol method and SDS-PAGE under both reducing and nonreducing conditions as specified in 7.5% polyacrylamide slab minigels (15). Affinity-purified anti-NR1 (C2 exon) and anti-NR2A-(1381-1394) antibodies were used at final concentrations of 1-5 g/ml; anti-c-Myc 9E10 mouse monoclonal antibodies were used at a dilution of 1:2000. Rabbit and mouse horseradish-linked secondary antibodies (Amersham Biosciences) were used at a final dilution of 1:2000, and immunoreactivities were detected by using the ECL Western blotting system. Immunoreactive bands were quantified by either molecular densitometry using a Personal Densitometer with ImageQuant (Amersham Biosciences) in the linear range of the film (15) or using the GeneGnome Chemiluminescence Capture and Analysis System (Syngene, Cambridge, UK).
Two-dimensional Electrophoresis-Two-dimensional electrophoresis was carried out as described by Cadieux and Kadner (22) except that after the first dimension of SDS-PAGE under non-reducing conditions, the band of interest was excised from the gel, and the protein was eluted into 25 mM Tris, 192 mM glycine, 0.5% (w/v) SDS, pH 8.5 (2 ml), by electrophoresis at 50 V for 2.5 h followed by electrophoresis for 10 min at 50 V with reversed polarity. The eluted protein was subjected to a second dimension of SDS-PAGE with and without reducing agent. After resolution, proteins were electrophoretically transferred to nitrocellulose membranes, and proteins were detected by immunoblotting as above.
Cell Surface NMDA Receptor Subunit Expression Measured by Enzyme-linked Immunosorbent Assay (ELISA)-The measurement of cell surface NR1-2a/NR2B NMDA receptors was carried out by using an ELISA method modified from that described in Rutter et al. (21). HEK 293 cells were sub-cultured overnight prior to transfection in polylysine (50 g/ml)-coated 24-well dishes, and 0.8 g of total plasmid DNA was used per well. Cells were cultured in the presence of 1 mM ketamine for 24 -36 h post-transfection. Cell culture media was aspirated, and each well was washed with 1 ϫ 500 l of phosphate-buffered saline (PBS). Cells were fixed by the addition of 4% (w/v) paraformaldehyde (250 l) for 5 min at room temperature, washed once with 500 l of PBS, and nonspecific sites blocked by incubation with 4% (w/v) milk powder in PBS (500 l) for 30 min. Cells were incubated for 1 h with the primary antibody, which was anti-NR2B-(46 -60) (0.5 g/ml; 250 l) because NR2 subunits are found on the cell surface only when co-expressed with an NR1 subunit (12). Cells were washed 4 ϫ for 10 min with 4 ϫ 500 l of 4% (w/v) milk powder in PBS; horseradish peroxidase-linked antirabbit Ig was added for 1 h (Amersham Biosciences; 1:2000 dilution in 4% (w/v) milk powder in PBS, 250 l). Cells were washed 3 times for 10 min with 500 l of 4% (w/v) milk powder in PBS, 1 time with 500 l of PBS for 10 min, and then K-blue substrate (Adgen Ltd., Ayr, UK; 900 l) was added. Cells were then incubated for 25 min, and the OD 650 nm was measured. For the determination of total immunoreactivity in the transfected cells, cells were fixed for 20 min with 4% (w/v) paraformaldehyde (250 l), washed 1 time with 500 l of PBS, and then permeabilized by incubation with 0.25% (v/v) Triton X-100 in PBS for 5 min. All subsequent steps were as above for the measurement of cell surface NMDA receptor subunit expression. For the determination of nonspecific binding of anti-NMDA receptor antibodies to transfected cells, HEK 293 cells were always transfected with the plasmid, pCIS, and the OD 650 nm values obtained were subtracted to give specific NMDA receptor total and cell surface binding.
Radioligand  (24) in the presence of 10 M glycine, 10 M glutamate, and 100 M thienylcyclohexylpiperidine for the measurement of nonspecific binding.
Preparation of Rat Forebrain Membranes-Membranes from adult rat forebrains were prepared as described previously and frozen at Ϫ80°C until used (18).

Expression of NR1-2a c-Myc /NR2A Heteromeric NMDA Receptors Does Not Result in the Cell Surface Expression of Functional NMDA Receptors-
We have reported previously that co-expression of wild-type NR1/NR2 heteromeric receptors results in cell cytotoxicity post-transfection (20). This NMDA receptor-induced cell death was eliminated by the inclusion of NMDA receptor antagonists in the cell culture medium posttransfection. Cytotoxicity was 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 . The cell cytotoxicity assay post-transfection is therefore a useful biochemical means to measure functional cell surface NMDA receptors. Most interesting, in contrast to NR1/ NR2B FLAG and NR1/NR2B c-Myc receptors, NR1-2a c-Myc /NR2A did not yield cell cytotoxicity post-transfection (Fig. 1A). The lack of observed cell death post-transfection was not because of a reduced expression of the NR1-2a c-Myc polypeptide, because immunoblots of transfected cell homogenates showed that when expressed alone, there was no significant difference in molecular weight and expression levels between wild-type and epitope-tagged subunits (Fig. 1, B and C). When NR1-2a c-Myc was co-expressed with NR2A, however, the subunit level was reduced 2.8 Ϯ 0.8-fold compared with wild-type NR1-2a/NR2A combinations (Fig. 1, B and C). Cell surface ELISAs confirmed the cell cytotoxicity observations in that no significant cell surface anti-NR2B antibody reactivity was detected following the co-expression of NR1-2a c-Myc /NR2B receptors in HEK 293 cells (Fig. 1D). This contrasted with ketamine-sensitive cell cytotoxicity post-transfection and cell surface expression for wild-type NR1-2a/NR2A and NR1-2a/NR2B receptors, respectively.
NR1-2a c-Myc Co-assembles with NR2A Subunits, Demonstration by Immunoprecipitation, and Radioligand Binding Studies-From the above results, it was evident that although the c-Myc tagged NR1-2a subunit was expressed, it was not trafficked to the cell surface when co-expressed with an NR2 subunit. Further experiments were carried out to characterize the properties of the mutant. In the first instance, immunoprecipitation studies were used to establish whether NR1-2a c-Myc subunits were able to associate with NR2 subunits. Wild-type and mutant NR1-2a were co-expressed with NR2A in HEK 293 cells; transfected cells were detergent-extracted, and immunoprecipitation assays were carried out with anti-NR2A antibodies. The results are shown in Fig. 2, A and B. It can be seen that both NR1-2a and NR1-2a c-Myc subunits were specifically immunoprecipitated by anti-NR2A antibodies demonstrating that c-Myc-tagged NR1-2a subunits do co-assemble with NR2A. Further proof of their co-association was obtained by radioligand binding studies. [ 3 H]MK801 is a use-dependent, noncompetitive antagonist of NMDA receptors. It binds with high affinity only to assembled NR1/NR2 receptors (25,26).  (Fig. 1, B and C). These results imply that the insertion of the c-Myc epitope does not affect the integrity of the NMDA receptor ion channel.
[ 3 H]MDL105,519 is a competitive glycine site antagonist and binds with high affinity to NR1 subunits expressed alone.  Fig. 3A. For wild-type receptors, both NR1 and NR2 subunit immunoreactivities are spread throughout the top part of the gradient. In contrast, for the epitope-tagged receptors, both NR1 and NR2 subunit immunoreactivities are confined to the middle fractions of the gradient where they co-localize with calnexin, a molecular chaperone and endoplasmic reticulum marker. Retention in the endoplasmic reticulum, as might be predicted from insertion of the epitope tag, suggests improper folding of nascent NR1-2a c-Myc polypeptide chains. Indeed, it was shown that although both wild-type NR1 and NR2 subunit antibodies were associated with calnexin, association of calnexin with NR1 in HEK 293 cells transfected with NR1-2a c-Myc/NR2 subunits was significantly increased (Fig. 3B). Native NMDA receptor NR1 but not NR2 subunits have been reported to co-immunoprecipitate with calnexin (27). Co-association of wild-type NR2 subunits described here may be due to the overexpression in mammalian cells compared with regulated NR2 expression in brain.
NR1-2a c-Myc /NR2A Receptors Associate with PSD-95-PSD-95 is a PDZ-containing protein that is enriched in postsynaptic densities of neurons where it is known to associate with NMDA receptors via the E(S/T)XV motif found at the C terminus of the NR2 subunit. It is thought to play a role in the organization of NMDA receptor clusters at synapses (reviewed in Ref. 28). Because NR1-2a c-Myc /NR2A receptors are not targeted to the cell surface, it was of interest to investigate if they co-associated with PSD-95. NR1-2a/NR2A and NR1-2a c-Myc / NR2A receptors were both co-expressed with PSD-95 c-Myc , cell homogenates harvested, immunoprecipitations carried out with anti-NR2A antibodies, and the immune pellets analyzed by immunoblotting. The results in Fig. 4 show that PSD-95 c-Myc is precipitated by anti-NR2 antibodies. NR1 subunits were also present in the immune precipitates as found by immunoblotting with anti-NR1 antibodies (data not shown) and detection of NR1-2a c-Myc (Fig. 4B). Thus, PSD-95 does associate specifically with NR2A subunit receptors despite the fact that they are apparently trapped within the endoplasmic reticulum. This finding concurs with the recent observations of Sans et al. (29) that implied that NMDA receptor/PDZ-containing proteins associate early in the secretory pathway rather than at the plasma membrane.

NR1-2a/NR2A and NR1-2a c-Myc /NR2A Heteromers, Comparative Analysis by SDS-PAGE Under Reducing and Non-reducing
Conditions-In immunoblots of HEK 293 cell homogenates expressing NR1-2a/NR2A and NR1-2a c-Myc /NR2A receptors probed with anti-NR1-C2 antibodies, in addition to the 115 Ϯ 2-kDa (n ϭ 7) immunoreactive species, i.e. the mature, glycosylated NR1 subunit, a second immunoreactive band with 226 Ϯ 5 kDa (n ϭ 8) was observed for wild-type but not for the mutant NR1-2a (Figs. 1B and 5A). This molecular weight is compatible with an NR1-NR1 dimer. It was also detected for the NR1-1a splice variant (data not shown). The c-Myc epitope tag is inserted at NR1-2a-(81). There is a cysteine residue at NR1-2a -(79); thus it was reasoned that if NR1-NR1 subunits form homodimers that are disulfide-bridged, the insertion of   Fig. 5A. Probing with anti-NR1-C2 antibodies, as observed previously under reducing conditions for transfected cells, two immunoreactive bands with 115 and 226 kDa were observed for wild-type with one 116 Ϯ 2 kDa (n ϭ 4) band seen for NR1-2a c-Myc . Under nonreducing conditions for NR1-2a, specific immunoreactive bands with 105 Ϯ 2 kDa (n ϭ 7), 226 Ϯ 5 kDa (n ϭ 6), and 244 Ϯ 4 kDa (n ϭ 6) were observed as well as significant immunoreactivity that did not enter the gel (Fig. 5A). If NR1-NR1 subunits are disulfide-linked as suggested above, one would expect the disappearance of the 115-kDa monomer. However, it has been shown that in heterologous expression systems and indeed in the brain, there exists a pool of unassembled NR1 subunits (18,30); under non-reducing conditions, the 105-kDa species is presumably unassembled NR1. For native NMDA receptors, only the 115-kDa NR1 immunoreactive species was detected under reducing conditions. Under non-reducing conditions, no immunoreactive species were detected. This may be explained by both the relatively low expression of NR1 homomers in brain compared with transfected cells, and the molecular weight of assembled NMDA receptors, i.e. Ͼ600,000, means that they would not enter the running gel.
For NR1-2a c-Myc under non-reducing conditions, interestingly the major immunoreactive band migrated with a significantly different molecular weight to that found for wild-type NR1-2a. Values are as follows: M r ϭ 113 Ϯ 2 (n ϭ 4; NR1-2a c-Myc) versus M r ϭ 105 Ϯ 2 (n ϭ 7; NR1-2a). The 8-kDa difference in molecular mass between wild-type NR1-2a and NR1-2a c-Myc is attributed to differences in folding perhaps due to intradisulfide as opposed to inter-disulfide bond formation in NR1-NR1 homodimers and hence mobility in SDS-PAGE of the NR1-2a subunits (see below for further discussion). No differences were found in the migration properties of monomeric NR2A subunits under reducing and non-reducing conditions (results not shown).
To substantiate that the 226-kDa species in NR1-2a/NR2A receptors contained NR1 disulfide-linked subunits, two-dimensional SDS-PAGE was carried out. The first dimension SDS-PAGE was carried out under non-reducing conditions to max- imize the amount of the 226-kDa species. This 226-kDa band was excised and reduced, and a second dimension SDS-PAGE under reducing and non-reducing conditions was carried out. The results shown in Fig. 5B reveal that following reduction, the 226-kDa immunoreactive species migrated with a molecular mass ϭ 115 kDa thus providing evidence for disulfide bridging of NR1 subunits. Presumed NR1-NR1 homodimers have been detected in cell homogenates of both NR1 or NR1/NR2A transfections by blue native-PAGE (13). Possible disulfide bridging between subunits, however, was not investigated.
Investigation of the Role of Cysteine Residues in the NR1 LIVBP Domain in the Trafficking of NR1/NR2 NMDA Receptors to the Cell Surface-The signal peptide of NR1 is predicted to be the first 18 amino acids of the immature NR1 polypeptide (31). Following signal peptide cleavage, there are thus three cysteines in the LIVBP domain of the mature NR1 subunit, Cys-22, Cys-79, and Cys-308. Each of these cysteines was mutated individually to alanine; a C79A,C308A double mutation and the C22A,C79A,C308A triple mutation were also generated. Mutant NR1-2a/NR2A or NR1-2a/NR2B subunit combinations were expressed transiently in HEK 293 cells, and the following parameters were determined: the percentage cell cytotoxicity post-transfection, the expression level of NR1-2a mutant subunits in total cell lysates prepared from NR1-2a/NR2 co-transfections, NMDA receptor cell surface expression (NR1-2a/NR2B), and the co-association of NR1-2a and NR2 subunits and NR1-2a/NR2A receptors and calnexin by immunoprecipitation. The results are shown in Figs. 6 and 7. Table I is a summary of the properties of the mutants compared with wildtype NR1-2a/NR2A and NR1-2a c-Myc /NR2A expressed receptors.
In Fig. 6B, it can be seen that all NR1-2a/NR2A combinations result in cytotoxicity post-transfection. There was no significant difference in percentage cell death between wild-type and mutant NR1-2a/NR2A receptors. In contrast, it was found that mutation of either Cys-79 or Cys-308 resulted in impairing the efficiency of trafficking of NR2B-containing receptors to the cell surface. For wild-type and NR1-2a C22A /NR2B receptors, 88% and 82%, respectively, of total immunoreactivity was expressed at the cell surface compared with 45% for NR1-2a C79A / NR2B and 45% for NR1-2a C308A /NR2B receptors (Table I).
There was no significant difference in cell surface expression between single, double, and triple mutants. Thus, although the percentage of cell cytotoxicity post-transfection is a useful in-dex of functional NMDA receptors, there is not an empirical correlation with the number of functional NMDA receptors expressed in the plasma membrane.
In immunoblots following SDS-PAGE under reducing conditions, the 226-kDa NR1-NR1 homodimer was not detectable in NR1-2a/NR2A receptors that contained the C79A point mutation (Fig. 6C), but surprisingly, a strong 226-kDa species was always detected for NR1-2a C308A /NR2A. Furthermore, under non-reducing conditions the molecular weight of all the monomeric, NR1-2a mutant subunits except NR1-2a C22A was similar to that found for the NR1-2a c-Myc subunit, i.e. M r ϭ 113,000 compared with M r ϭ 105,000 for native NR1-2a. Under nonreducing conditions, a strong 244-kDa band was observed for NR1-2a C79A -containing receptors (Fig. 6, C and E). These observations suggest the following: (i) Cys-79 is requisite for NR1-NR1 intersubunit disulfide bridge formation; (ii) both cysteine 79 and cysteine 308 are requisite for the proper folding of the NR1 subunit; (iii) in the absence of Cys-308, inapposite folding leads to uncontrolled intersubunit disulfide bond formation; and (iv) appropriate inter-and intra-NR1 subunit disulfide bridging is requisite for the efficient trafficking of NR1/ NR2 NMDA receptors to the plasma membrane. The 244-kDa band seen for NR1-2a C79A and, interestingly, also for NR1-2a c-Myc under non-reducing conditions only may again be due to NR1-NR1 dimer formation but with a different conformation to NR1 subunits linked via Cys-79 residues and, therefore, different mobility to wild-type NR1-NR1 dimers. Disulfide linkage here may possibly favor disulfide bridging between other cysteines, e.g. those located within the lysine-arginine-ornithine and glutamine-binding protein and/or S2 domains.
Despite impaired trafficking to the cell surface, NR1-2a C79A , NR1-2a C308A , and NR1-2a C79A,C308A were all shown to associate with NR2A subunits by immunoprecipitation (Fig. 7A). (Note again the absence of the 226-kDa species in NR1-2a subunits lacking Cys-79.) Furthermore, calnexin was found to co-immunoprecipitate with NR1-2a C79A , NR1-2a C308A , and NR1-2a C79A,C308A all co-expressed with NR2A (Fig. 7B). The amount of calnexin associating with NR1-2a increased with increased cysteine mutations. Thus overall, mutation of Cys-79 in the LIVBP domain yields NR1-2a subunits that possess similar properties to the epitope-tagged NR1-2a. DISCUSSION In this paper we have described the properties of an NR1-2a NMDA receptor subunit tagged with the c-Myc epitope in the N-terminal domain. This NR1-2a c-Myc subunit when co-expressed with NR2A was not targeted to the plasma membrane. Studies aimed at understanding why cell surface expression was not attained led to an investigation into the cysteine residues located within the LIVBP NR1 domain. It was concluded that cysteine residues Cys-79 and Cys-308 are required for efficient insertion into the plasma membrane. Thus, efficient trafficking of functional NMDA receptors to the plasma membrane requires co-synthesis of NR1 and NR2 subunits in the endoplasmic reticulum (12), the formation of NR1-NR1 disulfide-linked homodimers initiated, mediated, and dictated by the maturation of the NR1 LIVBP domain (13,14), and assembly of NR1-NR1 dimers with NR2 subunits possibly via NR1-NR2 dimer-dimer formation (14), followed by successful entry into the exocytotic pathway and insertion in the post-synaptic density.
NMDA receptors are allosterically modulated by several classes of compounds including sulfhydryl redox agents (reviewed in Ref. 1). Indeed, in extensive studies of the electrophysiological characterization of wild-type NR1/NR2 and mutant NR1/NR2 receptors, where defined cysteines were mutated to either alanine or serine, Lipton and co-workers deduced that in the NR1 subunit two pairs of cysteine residues were important for the redox sensitivity of NMDA receptors. One pair in the S2 extracellular loop, i.e. Cys-744 and Cys-798, is responsible for the slow persistent component of the redox modulation of NR1/NR2B, NR1/NR2C, and NR1/NR2D receptors (32). A second pair, Cys-79 and Cys-308, predicted but not proven to form an intra-disulfide bond within NR1, was reported to be responsible for an intermediate component of redox modulation (33). Although the studies reported here agree that these two cysteines are important, they differ in that we predict that NR1 subunits are linked via disulfide bridging involving Cys-79. No studies involving mutation of Cys-22 were reported, but the maximum current response for NR1-1a (C79S, C308S)/NR2A was ϳ60% of wild-type NR1-1a/NR2A receptors. This correlates with the decrease in cell surface expression of NR1-2a C79A , NR1-2a C308A /NR2A receptors reported here (Fig. 6D). It was notable that despite this significant decrease in cell surface expression of the NR1-2a cysteine mutants, there was no change in the percentage cell cytotoxicity post-transfection (Table I). This may be explained by the fact that the cytotoxicity assay was saturated. Alternatively, it may be possible that introduction of the cysteines 79 and 308 affects ligand binding or channel gating leading to enhanced Ca 2ϩ entry and an apparent increased cytotoxicity post-transfection. Choi et al. (33), however, reported that the NMDA and glycine EC 50 values for activation of NMDAR-1a (C79S,C308S)/ NR2A receptor-mediated currents were either not significantly changed (glycine EC 50 ) or were actually increased but only ϳ1.25-fold (NMDA EC 50 ). Thus, although the point mutation was to serine rather than alanine as described here, changes in ligand binding or channel activation are unlikely. The lack of correlation between cell surface expression and cell cytotoxicity may also be explained by an apparent decrease in cell surface expression due to a reduced reactivity of anti-NR2B antibodies, used to measure cell surface expression of NR1-2a/NR2B receptors, and due to possible changes in conformation introduced by cysteine 3 alanine mutations.
AMPA receptors, ionotropic glutamate receptors, belong to the same superfamily as the NMDA receptors. Most interesting, when analyzed under non-reducing conditions, it was reported that multiple high molecular weight species were observed that were absent in reducing conditions. These subunits could correspond to AMPA receptor subunit disulfide-linked dimers (34). Within the N-terminal domain of non-NMDA, AMPA-type receptors, there are four cysteines, Cys-73, Cys-98, Cys-194, and Cys-308. The crystal structure of the ligand binding domain of the GluR2 AMPA receptor subunit has been determined (9,35). This corresponds, however, only to residues GluR2-(404 -796); thus structural information regarding the N-terminal region of GluR2 and related sequences is currently not available. Similarly, the crystal structure of the NMDA receptor NR1 S1S2 domain was recently reported (36), but again, this excluded the LIVBP domain.
Fidelity of disulfide bond formation is important in that it both stabilizes the native conformation of proteins, and it maintains the integrity of secreted soluble proteins and cell surface receptors. Indeed, intradisulfide bonds have been shown to fulfill these functions in, for example, the acetylcholine-binding protein (37) and the Kir2.1 inwardly rectifying potassium channel (38). More recently, the integrity of disulfide bond formation was shown to be important for the trafficking to the cell surface of human P2X, ATP-gated ion channels (39). The results described here reveal another example whereby appropriate NR1-NR1 intersubunit disulfide bond formation involving cysteine 79 is requisite for the efficient trafficking of NR1/NR2 NMDA receptors to the cell surface. It is as yet unclear whether this appropriate disulfide bridging is a necessary step in the formation of NR1-NR1 dimers, whether it initiates the NR1-NR1 dimer formation, whether it occurs subsequently in order to stabilize NR1-NR1 prior to assembly with NR2-NR2 dimers as proposed by Schorge and Colquhoun (14), or indeed whether appropriate disulfide bridge formation merely serves as a quality control checkpoint in the exit of assembled, functional NMDA receptors from the endoplasmic reticulum. a Percentage cell cytotoxicity is the normalized value Ϯ S.E. for (n) determinations, where 100% ϭ the percentage cytotoxicity measured for NR1-2a/NR2A receptors cultured post-transfection in the absence of ketamine.
b Percentage cell surface expression is the percentage of NR2B subunit expressed at the cell surface relative to total NR2B expression in transfected HEK 293 cell homogenates for each construct.