Biochemical Evidence for the Co-association of ThreeN-Methyl-d-aspartate (NMDA) R2 Subunits in Recombinant NMDA Receptors*

Functional characterization of wild-type and mutant cloned N-methyl-d-aspartate (NMDA) receptors has been used to deduce their subunit stoichiometry and quaternary structure. However, the results reported from different groups have been at variance and are thus inconclusive. This study has employed a biochemical approach to determine the number of NMDA R2 (NR2) subunits/receptor together with the NMDA R1 (NR1)/NR2 subunit ratio of both cloned and native NMDA receptors. Thus, human embryonic kidney 293 cells were transfected with the NR1-1a and NR2A NMDA receptor subunits in combination with both FLAG- and c-Myc epitope-tagged NR2B subunits. The expressed receptors were detergent-extracted and subjected to double immunoaffinity purification using anti-NR2A and anti-FLAG antibody immunoaffinity columns in series. Immunoblotting of the double immunopurified NR2A/NR2BFLAG-containing material demonstrated the presence of anti-NR1, anti-NR2A, anti-FLAG, and, more important, anti-c-Myc antibody immunoreactivities. The presence of anti-c-Myc antibody immunoreactivity in the double immunoaffinity-purified material showed the co-assembly of three NR2 subunits, i.e.NR2A/NR2BFLAG/NR2Bc-Myc, within the same NMDA receptor complex. Control experiments excluded the possibility that the co-immunopurification of the three NR2 subunits was an artifact of the solubilization procedure. These results, taken together with those previously described that showed two NR1 subunits/oligomer, suggest that the NMDA receptor is at least pentameric.

The N-methyl-D-aspartate (NMDA) 1 subtype of the excitatory glutamate receptor is a hetero-oligomeric protein composed of the two classes of NMDA receptor subunits, namely, NR1 and NR2. The NR1 subunit is encoded by a single gene, which undergoes extensive alternative splicing to generate eight different splice forms. The NR2 subunit class consists of four genes encoding the subunits NR2A-NR2D. The subunits are thought to co-assemble in different combinations to form functionally distinct NMDA receptors (1). Biochemical and functional studies reported to date are ambiguous with regard to both NMDA receptor subunit stoichiometry and quaternary structure. Functional studies indicate that the binding of at least two molecules of both glutamate and glycine is required for NMDA receptor activation (2)(3)(4). The glutamate-and glycine-binding sites have been localized to the NR2 (5) and NR1 (6,7) subunits, respectively, by in vitro mutagenesis, thus suggesting that at least four subunits must co-assemble to allow binding and activation of the receptor by both coagonists. The molecular size of native NMDA receptors, as determined by both gel filtration and native polyacrylamide gel electrophoresis, is in the range of 605-850 kDa (8 -10), which is consistent with the co-assembly of between four and six subunits. Immunopurification studies of native NMDA receptors have identified a population of receptors containing at least two different NR1 splice variants (10,11). Furthermore, several groups have demonstrated the co-immunoprecipitation of anti-NR2A and anti-NR2B antibody immunoreactivities from rat forebrain, indicating the co-association of at least two NR2 subunits within an NMDA receptor complex (12)(13)(14). The determination of the subunit copy number has also been addressed by the coexpression of wild-type and mutant forms of either the NR1 or NR2 subunit. These results have been inconclusive, with reports of two or three NR1 subunits and two or three NR2 subunits/functional NMDA receptor complex (15)(16)(17).
In this study, human embryonic kidney (HEK) 293 cells were transfected with the NR1-1a and NR2A NMDA receptor subunits in combination with both FLAG-and c-Myc epitopetagged forms of the NR2B subunit. The expressed receptors were purified by anti-NR2A and anti-FLAG antibody immunoaffinity chromatography in series. Immunoblots of the double immunoaffinity-purified material demonstrated the presence of anti-c-Myc antibody immunoreactivity. This is presented as evidence for the co-association of three NR2 subunits, i.e. NR2A/NR2B FLAG /NR2B c-Myc , within the same NMDA receptor complex. Additionally, the NR1/NR2 subunit ratio in both native and cloned NMDA receptors was determined, which further substantiated a 2:3 NR1/NR2 subunit ratio/receptor.

Epitope Tagging
The mouse NR2B subunit was epitope-tagged with either the FLAG epitope (DYKDDDDK) or the c-Myc epitope (EQKLISEEDL) using oligonucleotide-directed mutagenesis (Sculptor TM in vitro mutagenesis system, Amersham Pharmacia Biotech). All tags were positioned between amino acids 53 and 54 of the NR2B subunit using the following oligonucleotides: FLAG, GTCATCTTTCTCGTGCTTGTCATCGTCGT-CCTTGTAGTCGGCATCTTTTATGGC; and c-Myc, GTCATCTTTCTC-GTGTAGGTCTTCTTCTGATATTAGCTTTTGTTCGGCATCTTTTAT-GGC. The position of the tag was confirmed by DNA sequencing. The cDNAs encoding the FLAG-and c-Myc-tagged NR2B subunits were subcloned into the mammalian expression vector pCIS for transfection into HEK 293 cells. gated to thyroglobulin by the glutaraldehyde method. The resultant conjugate was used to generate polyclonal antibodies in rabbits. Antibody production and affinity purification were carried out as described previously (18). Immunoblots probed with affinity-purified anti-NR2A-(1381-1394) antibodies detected a single polypeptide species with a molecular mass of 180 kDa in HEK 293 cells transfected with pCISNR2A (see Fig. 1A). The 180-kDa species is coincident with that previously identified as the NR2A subunit (19). This band was abolished by the prior incubation of the antibody with peptide-(1381-1394). Anti-NR2A-(1381-1394) antibodies displayed no immunocross-reactivity with other NMDA receptor subunits when used against HEK 293 cells transfected with the cDNAs encoding the NR1-1a, NR2B, NR2C, or NR2D subunit (see Fig. 1A).

Transfection of HEK 293 Cells
HEK 293 cells were cultured and transfected using the calcium phosphate method as described previously (19). Post-transfection, the cells were grown in the presence of 1 mM ketamine (Research Biochemicals International, Natick, MA) and harvested 24 h later. The NMDA receptor subunit clones pCISNR1-1a, pCISNR2A, and pCISNR2B were transfected at a ratio of 1:3:3, where the DNA encoding the NR2B subunit was represented by both the FLAG-tagged and c-Myc-tagged forms at a 1:1 ratio.

Cell Cytotoxicity
HEK 293 cells were transfected with pCISNR1-1a in combination with pCISNR2B, pCISNR2B FLAG , or pCISNR2B c-Myc . 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 (9). (20) was performed with both well washed cell homogenates prepared from HEK 293 cells transfected with either NR1-1a/NR2B or NR1-1a/NR2B FLAG and immunoaffinity-purified NMDA receptors. The assay was performed in 50 mM Tris citrate, pH 7.1, containing 5 mM EGTA and 5 mM EDTA. Aliquots (100 l) of the cell homogenate or immunoaffinity-purified NMDA receptor preparations were incubated in the presence of 0.1-50 nM [ 3 H]Ro 25-6981 at 4°C for 2 h. Nonspecific binding was defined by 1 mM spermidine. The reaction was terminated by rapid filtration through either GF/B filters (cell homogenates) or GF/B filters pretreated with 1% polyethyleneimine for 2 h immediately prior to assay (purified samples), followed by five washes with phosphate buffer, pH 7.4, at 4°C using a Brandel cell harvester.

Immunoaffinity Purification of Recombinant and Native NMDA Receptors
Recombinant Receptors-An anti-NR2A antibody immunoaffinity column was synthesized by the directional coupling of affinity-purified anti-NR2A-(1381-1394) antibodies to protein A-Sepharose using an immunopure IgG orientation kit (Pierce). The anti-FLAG M2 monoclonal antibody immunoaffinity column was purchased from Sigma-Aldrich. HEK 293 cells cotransfected with NR1-1a, NR2A, NR2B FLAG , and NR2B c-Myc clones were solubilized with 1% Triton X-100 and 0.5 M NaCl in the presence of a mixture of protease inhibitors as described (11). Anti-NR2A antibody immunoaffinity purification was performed as described for native forebrain NMDA receptors (12), with antigen elution of the affinity column with 50 mM diethylamine, pH 11.5, containing 0.05% (w/v) sodium deoxycholate. Fractions of 1 ml were collected and immediately quenched to a neutral pH with 2 M glycine. The eluted NR2A-containing fractions were pooled and assayed for [ 3 H]Ro 25-6981 radioligand binding activity, or alternatively, they were applied to the anti-FLAG antibody affinity column for 2 h, recirculating at 25 ml/h at 4°C. The column was washed in 50 mM Tris-HCl, pH 7.4, containing 0.5 M NaCl, 5 mM EGTA, 5 mM EDTA, and 1% (v/v) Triton X-100 at 25 ml/h for 2 h at 4°C. Elution of the anti-FLAG antibody affinity column was as described above. The NR2A/NR2B FLAG -containing fractions were analyzed by immunoblotting using the different NMDA receptor subunit-specific antibodies available. Additionally, the percentage co-association of NR2A and NR2B subunits in the NR1-1a/ NR2A/NR2B-transfected cells was determined by quantitative immunoblotting. Thus, equal volumes of the detergent-solubilized cell homogenates and the anti-NR2A antibody affinity column filtrates were analyzed by quantitative immunoblotting using anti-NR2A, anti-c-Myc, and anti-FLAG antibodies. The percentage NR2 subunit retention by the anti-NR2A antibody immunoaffinity column was calculated, and the respective immunoreactivities in the anti-NR2A antibody affinity column filtrates were expressed as a percentage of the activities in the original detergent-solubilized extracts.
Native NMDA Receptors-Membranes prepared from adult rat forebrain were solubilized with 1% (w/v) sodium deoxycholate and 0.15 M NaCl, pH 9, for 30 min at 37°C, and NMDA receptors were purified on anti-NR1 C2 and anti-NR2A antibody affinity columns in series as described (12). The double immunoaffinity-purified material was analyzed by quantitative immunoblotting using affinity-purified anti-NR1 C2 and anti-NR2A-(1381-1394) antibodies. Thus, a fixed volume (100 l) of the double purified material was precipitated and analyzed by quantitative immunoblotting using a range of primary antibody concentrations (0 -50 g/ml), and the ratio of NR1 to NR2 subunits was determined at the saturating antibody concentrations, i.e. 50 g/ml for each antibody. Note that ideally, the anti-NR2A/2B-(1435-1445) antibody would have been used for this quantitative analysis; however, when the antibody concentration dependence in immunoblots was studied, saturation was not obtained.
Immunoblotting SDS-polyacrylamide gel electrophoresis was carried out using 7% polyacrylamide slab gels under reducing conditions. Samples were prepared using the chloroform/methanol method of protein precipitation (19). Anti-NR1 C2, anti-NR2A, and anti-NR2B antibodies were used at a final concentration of 1 g/ml. Anti-FLAG and anti-c-Myc antibodies were used at final dilutions of 1:500 and 1:1000, respectively. Rabbit and mouse immunoglobulin horseradish-linked whole antibodies (Amersham Pharmacia Biotech) were used at a final dilution of 1:2000. Immunoreactivities were detected using the ECL Plus Western blotting detection system (Amersham Pharmacia Biotech). Where applicable, immunoblots were quantified by densitometry using ImageQuant (Molecular Dynamics, Inc., Sunnyvale, California) in the linear range of the film (12).

Characterization of the Epitope-tagged NR2B Subunits
Expression of the FLAG-tagged and c-Myc-tagged forms of the NR2B subunit was verified by immunoblotting. The anti-FLAG and anti-c-Myc antibodies both recognized a species with a molecular mass of 180 kDa in HEK 293 cells transfected with the respectively tagged NR2B subunit (Fig. 1B, lanes 2 and 6). The 180-kDa species is consistent with that previously identified as the NR2B subunit (11). In addition, the anti-c-Myc antibody detected multiple bands with molecular masses lower than that described for the NR2B subunit. These bands were also observed in untransfected HEK 293 cells and thus are cellular proteins that are immunoreactive with the anti-c-Myc antibody. The anti-NR2B-(46 -60) antibody recognized a single polypeptide species of 180 kDa in HEK 293 cells transfected with the wild-type NR2B subunit (Fig. 1B, lane 7). However, anti-NR2B-(46 -60) antibody immunoreactivity was not detected in HEK 293 cells transfected with either of the epitopetagged forms of the NR2B subunit (Fig. 1B, lanes 8 and 9). The introduction of the epitope tags between amino acids 53 and 54 of the NR2B subunit was coincident with the peptide sequence used to raise the anti-NR2B-(46 -60) antibody; thus, the epitope recognized by the anti-NR2B-(46 -60) antibody was abolished. However, quantitative immunoblots of HEK 293 cells transfected with either NR2B FLAG or NR2B c-Myc probed with the anti-NR2A/B-(1435-1445) antibody showed that the two 180-kDa epitope-tagged forms of the NR2B subunit were expressed to similar levels (Fig. 1B, lanes 10 -12).

Solubilization and Immunopurification of Recombinant NMDA Receptors
Anti-NR2A Antibody Immunoaffinity Chromatography-In this study, consistent with results previously reported (11), 1% Triton X-100 and 0.5 M NaCl solubilized ϳ50% of the NR1 and NR2 subunit immunoreactivities expressed in HEK 293 cells transfected with NR1-1a/NR2A/NR2B FLAG /NR2B c-Myc clones. The solubilized material was subjected to immunoaffinity chromatography as shown in Fig. 3A, and immunoblotting was used to characterize the purified receptors. Initially, the solubilized preparation was applied to an anti-NR2A antibody immunoaf- finity column, and the material before and after affinity column application was assayed for anti-NMDA receptor subunit antibody immunoreactivities. The anti-NR2A antibody column was not 100% efficient since only 67 Ϯ 18% (n ϭ 4) of the anti-NR2A antibody immunoreactivity was retained (Fig. 3B, lanes 1 and  2). Further quantification of the immunoblots revealed that only 20 Ϯ 4% (n ϭ 2) of the solubilized anti-FLAG antibody (Fig. 3C, lanes 1 and 2) and 26% (n ϭ 1) of the anti-c-Myc antibody immunoreactivities were retained by the anti-NR2A antibody immunoaffinity column. In the eluted fractions, anti-NR2A antibody immunoreactivity was detected (Fig. 3B, lane  3) in addition to anti-FLAG, anti-c-Myc, and anti-NR1 C2 antibody immunoreactivities (Figs. 3C, lane 3; and Fig. 4, lanes 5  and 11). The presence of anti-NR1 C2 antibody immunoreactivity in anti-NR2A antibody immunoaffinity-purified material is consistent with the co-association of NR1 and NR2 subunits within the same NMDA receptor complex.
Anti-FLAG Antibody Immunoaffinity Chromatography-The NR2A-purified material was subjected to further immunopurification using the anti-FLAG antibody affinity column. The results are shown in Fig. 3C. It can be seen that all the anti-FLAG antibody immunoreactivity was retained by the anti-FLAG antibody immunoaffinity column (Fig. 3C, lanes 3 and  4). The affinity column eluate was further analyzed by immunoblotting. Anti-NR2A, anti-FLAG, and anti-NR1 C2 antibody immunoreactivities were all detected in the double immunoaffinity-purified preparation (Fig. 4, lanes 3, 9, and 12). More important, the immunoblots also demonstrated in the eluted fractions the presence of anti-c-Myc antibody immunoreactivity (Fig. 4, lane 6). The detection of anti-c-Myc antibody immunoreactivity in material that has been purified on anti-NR2A and anti-FLAG antibody immunoaffinity columns in series demonstrates the presence of a population of recombinant NMDA receptors that contain the NR2A subunit co-associated with NR1-1a and both the FLAG-and c-Myc-tagged forms of the NR2B subunit, i.e. in this receptor population, three NR2 subunits are co-associated within one receptor complex. Interestingly, when the anti-NR2A, anti-FLAG, and anti-c-Myc antibody signals in the double immunopurified material were quantified, they gave a ratio of 1:1:1 (Fig. 4, lanes 3, 6, and 9). It should be pointed out, however, that these measurements were made using different specificity antibodies at single antibody concentrations. Full concentration-dependent immunoblots will be required to verify this ratio.

Determination of the NR1/NR2 Subunit Ratios in Native and Recombinant NMDA Receptors
Recombinant NMDA Receptors-The above-described results demonstrate that three NR2 subunits can co-assemble within an NMDA receptor molecule. If the NMDA receptor has a quaternary pentameric structure, this would imply two NR1 subunits/receptor. To test this prediction, the ratio of NR1 to NR2 subunits in recombinant NMDA receptors was determined by quantitative immunoblotting as described previously for the determination of ␥-aminobutyric acid type A receptor subunit stoichiometry (21). Thus, cloned NR1-1a/NR2B FLAG receptors were purified by anti-NR1 C2 followed by anti-FLAG antibody immunoaffinity chromatography. Receptor-containing fractions from the double immunoaffinity-purified material were pooled and analyzed by quantitative immunoblotting using a fixed amount of antigen, but varying the appropriate antibody concentrations, i.e. anti-NR1 C2 and anti-FLAG antibodies. A typical antibody concentration dependence curve and corresponding immunoblot are shown in Fig. 5. For all the quantitative immunoblots, the immunoreactive signal at 50 g/ml antibody was a linear function of antigen concentration. The NR1/NR2 subunit ratio for cloned receptors determined at 50 g/ml antibody was 1:1.5 Ϯ 0.5 for three independent double immunoaffinity-purified preparations of cloned NR1-1a/ NR2B FLAG receptors (data not shown).
Native NMDA Receptors-It is necessary to determine the NR1/NR2 subunit ratios in native receptors to show that the value obtained for the cloned receptors was not an artifact of the expression system. Thus, NMDA receptors were purified from adult rat forebrain by anti-NR1 C2 followed by anti-NR2A antibody immunoaffinity chromatography. As described above for cloned receptors, the double immunoaffinity-purified material was analyzed by quantitative immunoblotting using anti-NR1 C2 and anti-NR2A antibodies (Fig. 5). The NR1/NR2 subunit ratio determined at 50 g/ml antibody was 1:1.4 Ϯ 0.2 for two independent double immunoaffinity-purified preparations (three immunoblots/preparation). From our previous work, this receptor population will be a mixture of NR1/NR2A and NR1/NR2A/NR2B receptors, although the latter contributes only ϳ16% to the total anti-NR1/NR2A antibody double purified population (12). Thus, this 1:1.4 ratio is an underestimate since the NR2 subunit concentration had to be determined using an anti-NR2A antibody (see "Immunoaffinity Purification of Recombinant and Native NMDA Receptors" under "Experimental Procedures"), thus not taking into account the NR2B subunits present in NR1/NR2A/NR2B receptors.
A pentameric receptor composed of two NR1 subunits and three NR2 subunits would have an NR1/NR2 subunit ratio, 2:3, i.e. 1:1.5. This value is approached for native receptors. More important, in all double immunoaffinity-purified receptor preparations, NR2 immunoreactivity was always greater than that found for NR1.

Demonstration That the Co-association of Three NR2 Subunits Is Not an Artifact of the Solubilization Procedure
To discount the possibility that the co-immunopurification of three NR2 subunits is an artifact of the solubilization procedure, control experiments were carried out in which the NMDA receptor subunits NR1-1a, NR2A, NR2B FLAG , and NR2B c-Myc were expressed individually in HEK 293 cells. Homogenates prepared from cells transfected with the different individual subunits were pooled, solubilized, and subjected to anti-NR2A antibody immunoaffinity purification as described under "Experimental Procedures." The results of these experiments are shown in Fig. 6. It can be seen that anti-NR2A antibody immunoreactivity was detected in the anti-NR2A antibody immunoaffinity column eluate (Fig. 6, lane 3). In contrast, anti-NR1 C2, anti-FLAG, and anti-c-Myc antibody immunoreactivities, while detected in the solubilized preparation both before and after anti-NR2A antibody immunoaffinity purification, could not be detected in the anti-NR2A antibody immunoaffinity column eluate (Fig. 6, lanes 6, 9, and 12). This result indicates that the co-immunopurification of three NR2 subunits is unlikely to be due to the random aggregation of NMDA receptor subunits during the solubilization procedure.

Evidence Showing That Immunopurified NR1/NR2A/ NR2B Receptors Retain the Radioligand Binding Properties Commensurate with Those of Assembled Functional NMDA Receptors
Despite the fact that NR2 and, more important, NR1-1a subunits are both copurified by the different specificity anti-NR2 antibody immunoaffinity columns in series, it could be argued that the co-association of the NR1 subunit(s) and three NR2 subunits may be an artifact of the expression system. In the transfected HEK 293 cells, the subunits are overexpressed, and there is a potential for random subunit association. The purified receptor preparations therefore would not then be representative of functional, native NMDA receptors. To address this problem, the binding of the NR2B-specific radioligand [ 3 H]Ro 25-6981 to immunopurified receptors was utilized.