N -Methyl- D -aspartate Receptors Expressed in a Nonneuronal Cell Line Mediate Subunit-specific Increases in Free Intracellular Calcium*

, N -methyl- D -aspartate (NMDA) receptors can mediate cell death in neurons and in non-neuronal cells that express recombinant NMDA receptors. In neurons, in- creases in intracellular calcium correlate with NMDA receptor-mediated death, supporting a key role for loss of cellular calcium homeostasis in excitotoxic cell death. In the present study, free intracellular calcium concen- trations were examined in response to activation of recombinant NMDA receptors expressed in human embry- onic kidney 293 cells. Intracellular calcium was measured in transfected cell populations by cotransfec- tion with the calcium-sensitive, bioluminescent protein aequorin and by single cell imaging with the fluorescent calcium indicator fluo-3. Agonist application to NR1/2A or NR1/2B-transfected cells elicited robust rises in intra- cellular calcium. NR1/2A responses were inhibited by the noncompetitive antagonists MK-801 and dextro- methorphan and were dependent on extracellular calcium but not on intracellular calcium stores. In con- trast, no detectable intracellular calcium responses were observed in NR1/2C-transfected cells. These find- ings indicate that NMDA receptors in the absence of other neuron-specific factors can mediate increases in intracellular calcium with subunit specificity and extracellular calcium dependence. The N -methyl- D -aspartate (NMDA) using 100 ng/well initial fusion protein to bind antibody. Purified anti- bodies were then produced by immunoaffinity chromatography with fusion protein coupled to Reactigel 6 (cid:51) . Purified antibody was then eluted and used in further studies. Western Blotting— For Western blotting, purified protein or cell membranes were solubilized in sample buffer and electrophoresed on 6% polyacrylamide gels. Protein was transferred to nitrocellulose using a Hoeffer/Bio-Rad apparatus. Further steps were carried out as described previously (41). Materials—hcp -Coelenterazine, stock calcium calibration buffers, Fluo-3-AM, fluo-3 free acid, and ionomycin were obtained from Molec- ular Probes. [ 3 H]Glutamate, [ 3 H]MK-801, and 45 Calcium were obtained from DuPont NEN. MK-801 and DL -aminophosphonovaleric acid were obtained from Research Biochemicals Inc. All cDNAs were subcloned into vectors with cytomegalovirus promoters. The full-length cDNA encoding the NR1a subunit was subcloned from a Bluescript/ZAP li-brary as described previously (15, 23). NR 2A, NR 2B, and NR 2C, and GFP were the generous gift of Dr. Peter Seeburg (15, 23); NR1 (N598Q) was produced as described previously (23). The cDNA encoding ae- quorin was cloned into the eukaryotic expression vector pDPaqneo3 as described (42). The cDNA encoding 5HT 2A was cloned as described previously (43). HEK 293 cells were obtained from ATCC. All other chemicals were purchased from Sigma.

The N-methyl-D-aspartate (NMDA) 1 receptor, a subtype of excitatory amino acid receptor, plays a crucial role in glutamatergic transmission in the central nervous system (1)(2)(3). This receptor is essential for induction of long term potentiation, a neurophysiological process thought to underlie learning and memory (3,4), and has also been implicated in a variety of neuropathological processes including ischemia, epilepsy, and some neurodegenerative diseases (5)(6)(7).
Recently, multiple cDNAs that encode NMDA receptor subunits have been identified. These include the NMDAR1 (NR1) subunit, which is thought to exist as at least eight distinct mRNA splice variants termed NR1a-h (8 -10), and four NMDAR2 subunits designated NR2A, NR2B, NR2C, and NR2D (11)(12)(13). Expression of various combinations of these subunits in heterologous systems suggests that heteromeric assembly of the NR1 subunit with at least one type of NR2 subunit best produces functional NMDA receptors with electrophysiological and pharmacological characteristics similar to those observed in brain tissue (11)(12)(13)(14)(15). Thus, expression of heterodimeric combinations of NR1 and NR2 subunits produce channels that resemble native NMDA receptors in many ways, but expression of NR1 or NR2 homomers do not.
Both in vivo and in primary neuronal culture, overactivation of NMDA receptors is toxic to a variety of neuron types (7, 16 -20). This toxicity may be triggered in neurons by calcium influx through the NMDA receptor channel, with consequent increases in the concentration of intracellular calcium (5-7, 18 -19, 21-22). Expression of activated heterodimeric NMDA receptors is also toxic to non-neuronal cells (23)(24)(25), with some heterodimeric receptors mediating greater toxicity than others (23,25). The more toxic heterodimeric receptors have been shown to also mediate substantial 45 calcium influx (26). However, the extent to which these heterodimeric NMDA receptor combinations mediate increases in the levels of free intracellular calcium remains unknown.
In this study, we have investigated the role of NMDA receptors in modulation of intracellular calcium levels. Transfection of cDNAs encoding NMDA receptor subunits allowed us to determine whether heteromeric NMDA receptors, in the absence of any other neuronal-specific factors, could mediate rises in levels of intracellular calcium. The data show that heteromeric NMDA receptors functionally mediate rises in intracellular calcium with subunit and source specificity and with close similarity to reported results on heteromeric NMDA receptor toxicity.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-HEK 293 cells were grown on 10-cm tissue culture plates and routinely subcultured in minimum Eagle's medium containing 10% fetal calf serum supplemented with glutamine, penicillin, and streptomycin in an atmosphere of 5% CO 2 at 37°C. This medium contains 177 M glutamate (23). For aequorin luminescence studies, 24 h prior to transfection, cells were plated into 96-well plates at a density of approximately 10,000 cells per well. Transfection of HEK 293 cells to express subunits of NMDA receptors and apoaequorin was carried out using the calcium phosphate coprecipitation method of Chen and Okayama (27). Briefly, 20 g of purified plasmid DNA was mixed with equal volumes of 0.25 M CaCl 2 and BES-buffered saline. The mixture was then incubated at room temperature for 1 min. 2.9 l of this mixture (116 ng of DNA) was then added to individual wells of the 96-well plates. For aequorin, fluo-3, and 45 calcium uptake experiments, 10 M MK-801 was included during the first 24 h after transfection for protection against NMDA receptor-induced cell death (23) but was removed 24 h before the experiments were performed. On the second day, cells were placed in fresh supplemented medium containing 500 M ketamine, a rapidly dissociating NMDA receptor antagonist, for protection against cell death. 24 h later, the cells were assayed. For fluorescence measurements, cells were grown on 60-mm plates and transfected with 6 g of total DNA per plate. For calcium uptake, cells were grown on 6-well plates and transfected with 3 g of total DNA per well. For ligand binding studies, cells were grown on 100-mm plates, transfected with 20 g of total DNA per plate, and protected with MK-801 both days following transfection (23,28). For electrophysiology experiments, cells were grown on glass coverslips to 10% confluence, transfected with 3 g of total DNA per plate, cotransfected with the transfection marker green fluorescent protein (GFP; 29), and protected with 100 M aminophosphonovaleric acid both days following transfection. For transfection using multiple cDNAs, equimolar ratios of plasmid were used, as all cells which are transfected take up each DNA introduced under these conditions (30). All cDNAs were under the control of an identical cytomegalovirus promoter.
Assay for Aequorin Luminescence-48 h after transfection, hcp-coelenterazine was added to the medium of each well of the 96-well plate at a final concentration of 1 M. The hcp-derivative of coelenterazine was used since it confers to the aequorin complex a higher calcium sensitivity than does native coelenterazine with greater detection limits in the nanomolar calcium range and a greater luminescence intensity at an emission wavelength of 444 nm (31)(32)(33). The cells were then incubated for 30 min in 5% CO 2 at 37°C. Afterward, the medium was replaced with 0.1 ml of HEPES-buffered saline (0.9% saline, 20 mM HEPES, pH 7.5, sterile filtered). Increasing the time before making measurements did not increase the signal. The 96-well plate (Wallac; opaque well walls which eliminate scatter of light between wells) was then placed inside of a microplate reading luminometer (ML 3000, Dynatech). 10 l of 10 ϫ drug solutions were injected by computerdriven dispensers into any well as desired. Responses to drug solutions were measured for 30 s on desired wells, after which the same wells' responses to lysis and exposure to saturating calcium were measured for 30 s after injection of 100 l of a 0.03% Triton X-100, 100 mM CaCl 2 solution. These lysis measurements were carried out on each well for the purpose of quantitating the intracellular calcium response and normalizing between conditions. Drug solutions were prepared in HEPES-buffered saline with or without calcium as desired. Antagonists were added approximately 15 min prior to agonists.
In Vitro Calibration of Aequorin-The calcium calibration curve for aequorin was generated as described previously (34,35) with some modifications. Briefly, HEK 293 cells were grown on 10-cm tissue culture plates and transiently transfected with 20 g of plasmid containing apoaequorin cDNA. Approximately 48 h after transfection, the cells were rinsed with phosphate-buffered saline and then scraped with a rubber policeman into 200 l of 150 mM Tris, 0.8 mM phenylmethylsulfonyl fluoride, 0.1 mM EGTA, pH 7.2. The cells were then lysed through three cycles of freeze-thawing, centrifuged for 10 min (Eppendorf microcentrifuge) at room temperature, and the apoaequorin containing supernatants (cell lysates) were saved, and the cell pellets were discarded. The cell lysates were then frozen at Ϫ20°C and used for the calibration experiments as desired.
Cell lysates were thawed, and 1 ml was incubated with 1 M coelenterazine at 37°C for 30 min in the presence of 140 mM ␤-mercaptoethanol. 10 l of this mixture was added to each well of a 96-well plate and then placed into the luminometer. 90 l of isotonic EGTA/CaEGTA buffers of varying dilutions to obtain a range of free calcium concentrations (stock buffers were obtained from Molecular Probes, Inc., calcium calibration buffer kit 1) were then injected into the desired wells. Luminescence responses to each buffer dilution were measured (reaction 1), followed by measurement of the luminescence response to saturating calcium (10 mM; reaction 2). These experiments were performed at 22°C, pH 7.2. Luminescence responses to known calcium concentrations (P) were normalized to the total aequorin luminescence response (P tot ), thus utilizing a dimensionless value which is independent of aequorin concentration. Since aequorin is progressively used up during the light producing reaction, the peak response at saturating calcium levels was corrected to account for the fraction of aequorin spent in reaction 1. In mathematical terms, these points are described in Equation 1: where P 1 is the peak luminescence response to the calcium concentration measured in reaction 1; P 2 is the peak luminescence response to saturating calcium measured in reaction 2; I 1 is the integral of the luminescence signal measured in reaction 1; and I 2 is the integral of the luminescence signal measured in reaction 2.
A theoretical curve was generated through the experimental data points in order to quantitatively compare aequorin responses from well to well (Fig. 1). The equation that was used to describe the theoretical curve is analogous to the equation used for calibrating fluo-3 (36) but accounts for the cooperativity of calcium binding to aequorin (34, 35) as described in Equation 2: where R refers to log(P/P tot ), R min is the minimum log(P/P tot ), R max is the maximum log(P/P tot ), and n is the Hill coefficient. The results of the curve fit were as follows: R max ϭ Ϫ0.379, R min ϭ Ϫ3.23, K D ϭ 518 nM, and n ϭ 1.86. This Hill coefficient closely matches a recent report which demonstrated that aequorin luminescence is triggered by the binding of two calcium ions (37). Thus, this equation was used to calculate calcium concentrations in transfected cells expressing aequorin. Assay for Fluo-3 Fluorescence-48 h after transfection, the cells were washed once with HEPES-buffered saline, 2 mM CaCl 2 , then loaded with 1 M Fluo-3-AM for 30 min at room temperature, and then washed twice with HEPES-buffered saline, 2 mM CaCl 2 . The cells were then placed onto the stage of a confocal microscope (Leica, TCS), and excitation of fluo-3 was carried out using the 488-nm line of a Kr:Ar laser. Excitation light was reflected off of a (510 nm) dichroic mirror and passed through a 40 ϫ water immersion objective (Zeiss acroplan, NA ϭ 0.75). Emitted light was collected through a bandpass filter (515 nm) to a photomultiplier tube. Digital images were collected every 6 s and were the average of four scans. Responses to drug applications followed by responses to 1 M ionomycin were measured. Individual cells within a field were analyzed by defining a region of interest encompassing each cell and averaging the pixel intensities within the region of interest for each time point. Responding cells were selected on the basis of a 20 nM or greater change in calcium concentration upon agonist application. Responding cells typically represented 30% of all the cells in a field.
A calcium calibration curve was generated for fluo-3 fluorescence. EGTA/CaEGTA buffers of varying dilutions as described for the aequorin calibration containing 2 M fluo-3 were prepared. Relative fluorescence was measured for each buffer dilution upon excitation. A theoretical curve was generated through the experimental data points using Equation 3 as described previously (36): where F min is the minimum fluorescence and F max is the maximal fluorescence. The results of the curve fit yielded a K D of 318 nM. The FIG. 1. Calcium calibration of aequorin luminescence. 10 l of lysate obtained from cells expressing the luminescent protein aequorin was equilibrated with coelenterazine and then mixed with 90 l of isotonic EGTA/CaEGTA buffers of varying calcium concentrations followed by addition of saturating calcium. Peak luminescence response and luminescence integral was measured after each application and used to derive log(P/P tot ) as described under "Experimental Procedures." The curve through the experimental data points corresponds to the best fit theoretical curve described under "Experimental Procedures," where the EC 50 ϭ 518 nM and the Hill coefficient ϭ 1.86. Each data point represents the mean of at least 9 wells tested over three experiments.
parameter "F max " was measured experimentally for the cells by measuring fluorescence intensity after 1 M ionomycin application. The parameter "F min " in the cells was calculated based on the results of the calibration curve fit. 45 Calcium Uptake Assay- 45 Calcium uptake was carried out as described previously (38).
[ 3 H]Glutamate Binding-[ 3 H]Glutamate assays were performed as described previously (39). Briefly, thawed membranes from transfected cells were homogenized in 50 mM Tris acetate (pH 7.4 at 4°C) containing 0.08% Triton X-100 and incubated on ice for 10 min with gentle shaking. Homogenates were centrifuged for 30 min at 50,000 ϫ g. The resultant pellet was resuspended in ice-cold 50 mM Tris acetate, homogenized, and centrifuged for another 30 min. Final pellets were resuspended in 50 mM Tris acetate, homogenized, and used in the binding assays. Assays were initiated by addition of 500 l of homogenate (50 -200 g of protein) to tubes containing 50 l of [ 3 H]glutamate and 50 l of either vehicle or drug to a total volume of 600 l. Nonspecific binding was defined in the presence of 100 M NMDA. Incubations were for 25 min at 0°C. Incubation was terminated by filtration on a Whatman number 32 glass fiber filter followed by three rapid 2-ml washes with a Brandell cell harvester. All buffers were prepared the day of the experiment and filtered through a 0.2-micron filter to remove glutamate-producing microorganisms that could be present in the buffers. Radioactivity retained on the filters was determined by scintillation spectroscopy. Antibody Production-Antibodies to NR 2C were produced by cloning of bases 441-796 (coding for amino acids 147-265) as a PstI to PstI fragment into the appropriate pRSET vector (Invitrogen). The plasmid was transformed into Escherichia coli strain BL 21 LYS (Novagen), and the fusion protein was overexpressed by induction with isopropyl-1thio-␤-D-galactopyranoside. The fusion protein was purified by nickel affinity chromatography with elution by low pH. Preparative SDS-gel electrophoresis was used for further purification. Rabbits were then injected with antigen for production of polyclonal sera. Initial injections contain 500 g of protein with 250-g boosts at 2 weeks and as needed later. Initial injections were injected in Freund's complete adjuvant, and subsequent injections were in incomplete Freund's adjuvant. Antisera were initially screened by enzyme-linked immunosorbent assay Afterward, cell lysis and exposure to saturating calcium with 0.03% Triton, 50 mM CaCl 2 was measured in the same wells. Each trace has a unique y axis, so that the shape of each response can be viewed more easily.
using 100 ng/well initial fusion protein to bind antibody. Purified antibodies were then produced by immunoaffinity chromatography with fusion protein coupled to Reactigel 6 ϫ. Purified antibody was then eluted and used in further studies.
Western Blotting-For Western blotting, purified protein or cell membranes were solubilized in sample buffer and electrophoresed on 6% polyacrylamide gels. Protein was transferred to nitrocellulose using a Hoeffer/Bio-Rad apparatus. Further steps were carried out as described previously (41).
Materials-hcp-Coelenterazine, stock calcium calibration buffers, Fluo-3-AM, fluo-3 free acid, and ionomycin were obtained from Molecular Probes. [ 3 H]Glutamate, [ 3 H]MK-801, and 45 Calcium were obtained from DuPont NEN. MK-801 and DL-aminophosphonovaleric acid were obtained from Research Biochemicals Inc. All cDNAs were subcloned into vectors with cytomegalovirus promoters. The full-length cDNA encoding the NR1a subunit was subcloned from a Bluescript/ZAP library as described previously (15,23). NR 2A, NR 2B, and NR 2C, and GFP were the generous gift of Dr. Peter Seeburg (15,23); NR1 (N598Q) was produced as described previously (23). The cDNA encoding aequorin was cloned into the eukaryotic expression vector pDPaqneo3 as described (42). The cDNA encoding 5HT 2A was cloned as described previously (43). HEK 293 cells were obtained from ATCC. All other chemicals were purchased from Sigma.

Pharmacology of Luminescence
Responses-Aequorin luminescence responses to 100 M glutamate, 100 M glycine, and 2 mM calcium were measured in HEK 293 cells grown in 96-well microtiter plates. Robust luminescence responses were observed in cells transfected with aequorin and either the NMDA receptor NR 1a/2A subunit combination (Fig. 2), the NR 1a/2B subunit combination, or the serotonin 5HT 2A receptor (Fig. 2). Remaining aequorin luminescence was observed in the same cells in response to lysis with 0.03% Triton and exposure to saturating calcium (50 mM, Fig. 2). No aequorin luminescence responses to agonists were detected from cells transfected with NR1a and aequorin only, from cells where calcium was excluded from the extracellular buffer, or from cells where calcium in the absence of NMDA receptor agonists was added (data not shown). The time at which peak luminescence was reached in response to agonists varied with receptor type and occurred over a time course of seconds. In order to determine if agonist-stimulated calcium responses were receptor-mediated, the ability of the NMDA receptor antagonists MK-801 and dextromethorphan to inhibit 100 M glutamate, 100 M glycine, and 2 mM calcium-stimulated calcium responses for the NR 1a/2A subunit combination were evaluated (Fig. 3). The average IC 50 for MK-801 was 1.2 Ϯ 0.5 M (n ϭ 3 experiments). This IC 50 value is significantly higher than that seen in ligand binding experiments (Fig. 3) but more closely matches the IC 50 for toxicity in HEK 293 cells (23). Dextromethorphan significantly inhibited the responses at 1 mM (Fig. 3), and the apparent IC 50 was estimated to be Ͼ1 mM, although determination of the exact IC 50 value was limited by the solubility of dextromethorphan. This IC 50 is also significantly higher compared with that seen in ligand binding (Fig.  3) and also parallels the lack of protection from NMDA receptor toxicity that is observed with dextromethorphan in transfected HEK 293 cells. 2 NMDA Receptor Subunit Specificity of Calcium Responses-Due to the low level of cell death observed with transfection of the NR 1/2C subunit combination in HEK 293 cells as compared with NR 1/2A (23,25), it was of interest to compare these two NMDA receptor subunit combinations in their ability to mediate rises in intracellular calcium in response to agonist stimulation. The two NMDA receptor subunit combinations were expressed to similar extents in HEK 293 cells as measured with both [ 3 H]MK-801 and [ 3 H]glutamate binding (Table  I), since B max values for either radioligand were not significantly different. NR 2C protein in NR 1a/2C-transfected cells was readily detected by Western blot (Fig. 4A) receptors when measured electrophysiologically (Fig. 4B). However, the two NMDA receptor types did not equally mediate rises in intracellular calcium (Fig. 4, C and D). A significant increase in peak intracellular calcium levels was observed upon 100 M glutamate, 100 M glycine, and 2 mM calcium stimulation of the NR 1a/2A receptor (and the NR 1a/2B receptor) as compared with vehicle control, whereas no significant increase in peak intracellular calcium was observed upon agonist stimulation of the NR 1a/2C receptor as compared with vehicle control (Fig. 4D). Agonist stimulation of cells expressing NR 1a/2A receptors resulted in a 120% increase over base-line, whereas only a 10% increase over base-line level was observed with the NR 1a/2C combination (Fig. 4D). This represents a greater than 10-fold difference in intracellular calcium changes mediated by these two receptor combinations. Although the two receptors differ somewhat in their affinities for [ 3 H]glutamate (Table I), this is unlikely to be the reason for the differences observed in their abilities to mediate rises in intracellular calcium since a saturating concentration of glutamate for both receptor types was used to measure agonist stimulation of calcium responses. Cotransfection with the heterotrimeric mixture of NR 1a/2A/2C and aequorin yielded a mean agoniststimulated peak intracellular calcium response of 0.88 Ϯ 0.07 M (n ϭ 24 wells over four experiments). This value was 45% lower than the mean peak intracellular calcium response of NR 1a/2A and aequorin-transfected cells tested in parallel (1.6 Ϯ 0.24 M; n ϭ 14 wells over four experiments), consistent with previously reported excitotoxicity results for this trimeric combination (25).
In order to obtain additional evidence supporting the NMDA receptor subunit specificity results at the single cell level, intracellular calcium changes in NR 1a/2A and NR 1a/2C-transfected HEK 293 cells were imaged with the calcium-sensitive fluorescent dye fluo-3. Robust fluorescence responses to 100 M glutamate, 100 M glycine, 2 mM calcium application were consistently observed in NR 1a/2A-transfected cells (   (29). No responses to NMDA receptor agonists were observed in NR 1a/2C and 5HT 2A -transfected cells that responded to serotonin (Fig. 6). Source of the NMDA Receptor-mediated Calcium Response-The dependence on extracellular calcium and intracellular calcium stores was evaluated to determine the source of the NMDA receptor-mediated intracellular calcium rise. Substitution of the divalent cation barium for calcium in the extracellular buffer eliminated the NR 1a/2A intracellular calcium response to 100 M glutamate, 100 M glycine, 2 mM calcium (Fig. 7A). Transfection of the lower calcium-permeable NR1a (N598Q) mutant with the 2A subunit resulted in a significantly decreased peak intracellular calcium response as compared with wild-type NR 1a/2A (Fig. 7B). Incubation with 45 calcium in the extracellular buffer revealed a significant increase in 45 calcium uptake upon agonist application in NR 1a/2A-transfected cells but not in NR 1a/2C-transfected cells (Fig. 7C). Preincubation with the intracellular calcium store depleter, thapsigargin, resulted in a slight decrease in NR 1a/2A-mediated peak intracellular calcium responses to agonists as compared with vehicle control, although this did not reach statistical significance (Fig. 7D). Since HEK 293 cells normally express purinergic receptors which require intact intracellular calcium stores to mediate cytosolic calcium rises, responses to ATP were measured to assess depletion of intracellular stores. No responses to ATP were observed in thapsigargin preincubated cells, whereas ATP responses were observed in virtually all of the vehicle-treated cells (data not shown). Likewise, pre-treatment with thapsigargin eliminated the response of 5-HT 2A receptor-transfected cells to serotonin (data not shown), further ensuring that the thapsigargin pretreatment was successful in depleting the intracellular stores.

DISCUSSION
The present study demonstrates that heteromeric NMDA receptors expressed in HEK 293 cells mediate agonist-induced increases in intracellular calcium levels with sensitivity to open-channel blockers, extracellular calcium dependence, and subunit specificity. Intracellular calcium measurements made in transfected cell populations with the bioluminescent jellyfish protein aequorin (44 -47) and in single transfected cells with the fluorescent calcium indicator fluo-3 (36) yielded similar results. Utilizing both techniques, the NR 1/2A subunit combination mediated glutamate-induced peak intracellular calcium responses to agonists in the low micromolar range, over a time course of seconds. The NR 1/2B subunit combination likewise mediated glutamate-induced peak intracellular calcium responses to agonists in the high nanomolar range, over a time course of seconds. Similar peak magnitudes and time courses of primarily NMDA receptor-mediated intracellular calcium transients have been reported for a variety of neuron types (22, 48 -49). This suggests that the expression of the combined NR1 and NR2 subunits in non-neuronal cells is sufficient to mediate NMDA receptor-activated intracellular calcium increases, even in the absence of other neuron-specific components.
While the bioluminescent protein aequorin has been used as a tool to measure calcium for decades (44 -47), its use has broadened since the cloning of its cDNA (42, 50 -57). The use of aequorin as a calcium indicator in a transient transfection cell expression system is perhaps the most effective means of measuring the calcium response in 100% of a subset of cells that are transfected with the cDNAs of interest (35). In this study, cotransfection of NMDA receptor subunits and aequorin allowed population calcium measurements to be obtained from the transfected cells. When coupled with a 96-well microplate reader, this high throughput, sensitive assay provided comparable results with those of the more established fluorescent calcium indicator, fluo-3. However, the measures of intracellular calcium levels in cells transfected with NMDA receptors differed slightly between methodologies. The resting level of calcium in 293 cells transfected with NMDA receptor subunits appeared higher when measured by aequorin than by fluo-3. This likely results from the fact that aequorin measurements reflect the average calcium in all transfected cells, many of which are dying due to expression of NMDA receptors and thus already have elevated calcium levels. Single cell imaging favors fields of cells that appear healthier and represent a much smaller number of transfected cells. Thus the two methods provide a complementary methodology for the assessment of NMDA receptor-mediated intracellular calcium responses in transfected cells.
The peak response to agonist was slightly later in NMDA receptors compared with serotonin receptors. This may reflect the difference between plasma membrane-limited calcium entry mediated by NMDA receptors and the amplification obtained in intracellular calcium release mediated by G-proteincoupled receptors. However, it could also reflect differences in receptor expression or calcium buffering mechanisms in this model system.
Transfection of various heterodimeric NMDA receptor subunits did not equally mediate rises in intracellular calcium levels. Both the NR 1/2A and NR 1/2B combinations mediated substantial intracellular calcium rises, although the NR 1/2A combination was larger by approximately 1.5-fold. This difference may be explained by slightly lower expression levels of NR 1/2B as suggested by the 1.5-2-fold lower B max values previously observed with 125 I-MK-801 binding (15) and by the small difference in the rate of calcium change that was observed in the present study. Contrastingly, although the NR 1/2A and NR 1/2B combinations led to robust intracellular calcium rises, the NR 1/2C combination did not produce a significant intracellular calcium response, measured with either aequorin or fluo-3. In addition, the NR 1/2A combination also mediated significant 45 calcium influx while the NR 1/2C combination did not. The lack of a detectable intracellular calcium response or 45 calcium influx with NR 1/2C was not likely to be due to differences in expression levels since B max values, obtained with two different radioligands, for the two dimeric subunit combinations were not statistically different. These ligand binding results matched those previously reported by others (25,58). In addition, 2C protein was readily detectable by Western blot. We have also demonstrated functional channels electrophysiologically, consistent with previous reports that have examined glutamate-induced whole cell currents for this subunit combination (12-14, 23, 59 -61). However, the NR 1/2C and NR 1/2A heterodimeric receptors differ in many specific channel properties, some of which may contribute to the difference in intracellular calcium response to agonist activation. These differences include a lower fractional calcium current for NR 1/2C versus NR 1/2A (62) (13,59,65), and differences in sensitivity to open-channel blockers between the two dimeric receptors (13). Of the seven differences in channel properties listed, the first five are most likely to contribute to the observed differences in agonist-stimulated intracellular calcium rises and 45 calcium uptake. Based on the quantitative data provided by the studies cited, the difference between NR 1/2A and NR 1/2C for each of these channel properties ranges from 1.5-to 3-fold. Thus, it is unlikely that any one difference in specific channel properties may explain the observed difference in calcium response between NR 1/2A and NR 1/2C. How- ever, it is possible that the differences in channel properties, when combined, account for part of the observed differences in intracellular calcium changes, although other contributing factors or channel properties that have not yet been quantitated may also be involved. These could include the activation of other calcium stores or a nonlinearity in the calcium buffering ability of HEK 293 cells. Although small fluorescence changes under conditions where calcium buffering in the cell is dominated by very high concentrations of fura-2 have been reported for the NR 1/2C combination (62), the present results obtained under more physiological calcium buffering conditions suggest that the NR 1/2C combination, in the absence of other neuronspecific components, does not appear to mediate macroscopically detectable rises in levels of intracellular calcium. Thus it is noteworthy that in vivo the granule cell layer of the cerebellum and interneurons in the hippocampus, regions of the adult brain that predominantly express the NR 2C subunit (59), are brain areas also more resistant to ischemic insult (66,67). Therefore, a lack of or a lowered intracellular calcium response with an NMDA receptor containing a 2C subunit may have pathophysiological relevance in vivo.
The NR 1/2A-mediated rise in intracellular calcium requires extracellular calcium. This observation parallels previous reports that demonstrate that extracellular calcium is necessary for NMDA receptor-mediated intracellular calcium rises and toxicity in cultured neurons (5, 18 -19, 21-22). Channel permeability to calcium also played a role in the intracellular calcium response, since transfection of the mutant NR1 (N598Q) with NR2A (65, 68) mediated an attenuated rise in intracellular calcium. This observation is not likely to be the result of decreased expression of the NR1 (N598Q)/2A combination since similar levels of 125 I-MK-801 binding have been demonstrated as compared with wild-type NR 1/2A in these cells (23). Although extracellular calcium was required for the NMDA receptor-mediated intracellular calcium rise, thapsigargin-sensitive intracellular calcium stores appeared not to be a major component. Although some attenuation of the NR 1/2A-mediated calcium response was observed in the presence of the intracellular calcium depleter, thapsigargin, the decrease did not reach statistical significance. In cultured neurons however, intracellular calcium stores, including those sensitive to thapsigargin, give rise to a substantial component of the NMDA receptor-mediated intracellular calcium rise in a variety of neuron types (69 -72). Thus, the NR 1/2A combination in the absence of other neuron-specific components does not mobilize thapsigargin-sensitive intracellular calcium stores to the same degree as observed in neuronal culture.
The present data suggest that increases in intracellular calcium may be the triggering mechanism by which NMDA receptor-transfected nonneuronal cells die in response to glutamate (23)(24)(25), similar to the proposed critical role that intracellular calcium plays in the process of neuronal toxicity (  inhibited by channel blockers MK-801 and dextromethorphan. However, the inhibition curves for both of these antagonists were substantially shifted to the right compared with the corresponding ligand binding curves. A similarly high IC 50 for MK-801 inhibition of NMDA receptor-mediated increases in intracellular calcium in cerebellar granule cells has been reported (73). This pharmacological profile matches that seen for inhibition of glutamate-induced cell death in transfected HEK 293 cells as the IC 50 for MK-801 inhibition of cell death in NR 1/2A-transfected cells is similarly high (0.795 M; Ref. 23), and dextromethorphan is relatively ineffective. The subunit specificity profile of NMDA receptor-mediated intracellular calcium changes also matches previously reported results on NMDA receptor-mediated toxicity (23)(24)(25). The relative levels of peak intracellular calcium rises mediated by the three heterodimeric combinations paralleled the relative levels of cell death reported for these same three combinations (23,25) where NR1/ 2A Ͼ NR1/2B Ͼ Ͼ NR1/2C. In addition, an NR 1/2A/2C heterotrimeric mixture mediated lower peak intracellular calcium responses than the NR 1/2A dimeric combination, consistent with previously reported excitotoxicity results for this trimeric combination (25). The link between the increase in intracellular calcium and susceptibility to cell death is further strengthened by the lower levels of intracellular calcium resulting from activation of receptors made from the lower calcium-permeable NR1 (N598Q) mutant (65,68). NR1 (N598Q)/2A peak levels of intracellular calcium were on average 30% less than in wildtype NR 1/2A, and this mutant subunit combination has approximately 20% less cell death (23). This provides further evidence in favor of the critical role of intracellular calcium levels in NMDA receptor-mediated cell death in transfected cells.
NMDA receptors expressed in non-neuronal HEK 293 cells mediate rises in intracellular calcium, a functional property of NMDA receptors that has been described in neurons and brain. However, the rises in intracellular calcium concentration are subunit combination-specific, which gives rise to an added level of complexity by which NMDA receptors may mediate this response in the brain. The source of the calcium rise in the absence of other neuronal components does not completely reproduce what has been observed in neurons, suggestive of a requirement for other neuronal factors. The heteromeric NMDA receptor calcium responses closely parallel previous reports of heteromeric NMDA receptor toxicity, thus supporting the importance of intracellular calcium in NMDA receptormediated cell death. And finally, the use of recombinant aequorin in conjunction with transient transfection of neurotransmitter receptor subunits provides a simple and convenient means of evaluating receptor-mediated intracellular calcium increases.