Differentiation of Glycine Antagonist Sites ofN-Methyl-d-aspartate Receptor Subtypes

The binding site for the co-agonist glycine onN-methyl-d-aspartate (NMDA) receptors has been mapped to the NR1 subunit whereas binding of the principal agonist glutamate is mediated by the NR2 subunits. Using the novel glycine site antagonist and photoaffinity label CGP 61594, distinct contributions of the NR2 subunit variants to the glycine antagonist binding domains of NMDA receptor subtypes are demonstrated. High affinity sites for CGP 61594 were exclusively displayed by NR1/2B receptors, as shown by their co-distribution with the NR2B subunit, by subunit-selective immunoprecipitation and by functional analysis of NR1/2B receptors expressed in Xenopus oocytes (inhibitory potency, IC50 = 45 ± 11 nm). Other NMDA receptor subtypes are clearly distinguished by reduced inhibitory potencies for CGP 61594, being low for NR1/2A and NR1/2D receptors (IC50 = 430 ± 105 nm and 340 ± 61 nm, respectively) and intermediate for NR1/2C receptors (IC50 = 164 ± 27 nm). Glycine antagonist sites with low and intermediate affinity for [3H]CGP 61594 were detected also in situ by radioligand binding in brain areas predominantly expressing the NR2A and NR2C subunits, respectively. Thus, [3H]CGP 61594 is the first antagonist radioligand that reliably distinguishes the glycine site of NMDA receptor subtypes. [3H]CGP 61594 is a promising tool to identify the NR2 subunit domains that contribute to differential glycine antagonist sites of NMDA receptor subtypes.

Inhibition of NMDA receptors by glycine site antagonists is of major interest for the therapeutic intervention in pathological central nervous system states like stroke, head injury, and epilepsy. At doses evoking remarkable anticonvulsant and neuroprotective effects, glycine site antagonists are less prone to psychotomimetic actions and exhibit less severe motoric and cognitive side effects than glutamate site antagonists (22). Further improvements are expected from the targeting of the glycine site of NMDA receptor subtypes.
The NR1 subunit harbors main determinants of the glycine binding site in recombinant receptors as well as in situ. Recombinant homomeric NR1 receptors bind glycine agonists and antagonists (19,23,24), and various amino acid residues involved in high affinity binding have been identified by mutational analysis of the NR1 subunit (25)(26)(27)(28). Similarly, in native receptors, the novel glycine antagonist and photoaffinity ligand [ 3 H]CGP 61594 labels exclusively the NR1 subunit of NMDA receptors in brain membranes. 2 In contrast, the glutamate binding pocket is formed by the NR2 subunits (29). Interestingly, the properties of the glycine site are influenced by the type of NR2 subunit that is assembled with the NR1 subunit. Glycine displays an up to 10-fold lower potency at NR1/2A receptors than at NR1/2B, NR1/2C, or NR1/2D receptors, and a similar difference in affinity has also been observed for the glycine site agonists D-serine, 1-aminocyclobutanecarboxylic acid, D-alanine, HA-966, and L-687,414 (16, 19, 21, 30 -33). However, glycine site antagonists (e.g. L-701,324, L-689,560, L-695,902, CKA) did not clearly differentiate between recombinant NMDA receptor subtypes (23,34). Mutational analysis of the NR1 subunit suggested structural differences in the binding domains of glycine site agonists and antagonists (25)(26)(27)(28). Some of the mutated residues affected the affinity for * This work was supported by the Swiss National Science Foundation (Grant 31-36327.92), by the Deutsche Forschungsgemeinschaft (SFB 169), and by the Fonds der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
For irreversible labeling, membranes containing 60 g of protein were incubated with 25 nM [ 3 H]CGP 61594 in 50 mM Tris/HCl, pH 7.4, for 2 h on ice, followed by irradiation with UV light in a Stratalinker (1 J at 254 nm; distance: 10 cm). Nonspecific binding was determined in the presence of 1 mM glycine. After centrifugation, the membranes were dissolved in SDS-PAGE sample buffer, heated at 60°C for 30 min and subjected to SDS-PAGE using 5% mini gels. After staining for proteins with Coomassie Brilliant Blue R-250, the gel was incubated in Amplify (Amersham Pharmacia Biotech) for 30 min, dried, and exposed to an x-ray film (Fuji RX) for at least 2 weeks.
Receptor Autoradiography-The regional distribution of [ 3 H]CGP 61594 was analyzed on unfixed parasagittal cryostat-cut sections (12 m) of brains of adult male Sprague-Dawley rats (200 -250 g). Slidemounted sections were thawed and washed three times for 10 min at 4°C in 50 mM Tris/HCl, pH 7.5. For [ 3 H]CGP 61594 binding, sections were incubated for 2 h at 4°C with 10 nM [ 3 H]CGP 61594 in 50 mM Tris/HCl, pH 7.5, followed by washing five times for 2 min in buffer. After washing in buffer and a quick dip in distilled water, the sections were dried and exposed to Hyperfilm-3 H (Amersham Pharmacia Biotech) for 15 days. Nonspecific [ 3 H]CGP 61594 binding was assessed in parallel in the presence of 1 mM glycine.
Cell Culture and Transfection-Human embryonic kidney (HEK) 293 cells were grown to 50 -60% confluence and transfected with the NR1 cDNA alone or in combination with the cDNA encoding the NR2A, NR2B, NR2C, or NR2D subunit (25 g total) using the calcium phos-phate precipitation method of Chen and Okayama (41). In co-transfection cultures, either 5 M MK-801 and 10 M DCKA or 500 M ketamine were added. After overnight incubation (35°C, 3% CO 2 ), the medium was renewed and the cells incubated for further 24 -48 h (37°C, 5% CO 2 ). The plates were then rinsed with phosphate-buffered saline (37°C, pH 7.5), and the cells were harvested into the same buffer containing 50 mM KCl. The cell suspension was centrifuged (10,000 ϫ g, 15 min, 4°C), and the resultant pellet homogenized (Polytron) in icecold 0.32 M sucrose, 10 mM EDTA, 1 g/ml phenylmethanesulfonyl fluoride, and centrifuged (1000 ϫ g, 15 min, 4°C). After rehomogenization of the pellet and recentrifugation, the supernatants were combined and centrifuged at 30,000 ϫ g for 30 min at 4°C. The membrane pellet was washed once in 10 mM Tris acetate, pH 7.5, 10 mM EDTA, and frozen to Ϫ70°C at least overnight. Before the binding experiment, the thawed membrane suspension was washed twice with 5 mM Tris acetate, pH 7.0, and once with 50 mM Tris/HCl, pH 7.4.
Expression in Xenopus Oocytes-Linearized plasmid DNAs were used for the in vitro synthesis of cRNA (mCAP mRNA capping kit, Stratagene). cRNAs of NR1 and NR2A-2D subunits were synthesized using T7 or T3 RNA polymerase, respectively. cRNA concentrations were determined by both measuring optical densities at 260 nm and comparing methylene blue staining intensities after gel electrophoresis. For oocyte injection, the concentration of the cRNA samples was adjusted to 200 -500 ng/ml, and diluted aliquots of the NR1 and NR2 cRNAs were mixed at the RNA ratio of 1:1. Microinjection of about 50 nl of cRNA into Xenopus laevis oocytes and voltage-clamp recordings of agonist responses in Mg 2ϩ -free frog Ringer's solution at a holding potential of Ϫ70 mV were performed as described (42).

RESULTS
[ 3 H]CGP 61594 is a novel glycine site antagonist. Upon photoactivation it is selectively cross-linked to the NR1 subunit of NMDA receptors. 2 By analyzing the interaction of [ 3 H]CGP 61594 with various NMDA receptor subtypes, the contribution of the NR2 subunit variants to the glycine antagonist domain was investigated.
Regional and Ontogenetic Distribution of [ 3 H]CGP 61594 Binding Sites-The distribution of [ 3 H]CGP 61594 binding sites was mapped autoradiographically on parasagittal cryostat sections of adult brain. High affinity [ 3 H]CGP 61594 binding was mainly restricted to forebrain regions with highest densities in the hippocampal formation, followed by cerebral cortex, striatum, olfactory bulb, and thalamus (Fig. 1). In midand hindbrain areas only marginal labeling of receptors over background was observed (Fig. 1). In contrast, strychnine-insensitive [ 3 H]glycine binding, which recognizes the entire population of NMDA receptors, is detected in virtually all brain areas, including cerebellum and brain stem (43)(44)(45). Thus, [ 3 H]CGP 61594 appears to interact with high affinity only with a subpopulation of NMDA receptors.
To verify the regional specificity of NMDA receptor labeling by [ 3 H]CGP 61594, high affinity binding sites were identified in crude membranes prepared from various brain regions. No difference in the K D value of high affinity binding of [ 3 H]CGP 61594 was detected among the different brain regions (Table I).
The highest maximum number of binding sites (B max values) was observed in the hippocampal formation, followed by the cerebral cortex and striatum. Lower B max values were apparent in olfactory bulb and thalamus, while very low binding was found in brainstem regions (mesencephalon, medulla/pons). In cerebellar membranes high affinity binding sites were not detected, although specific binding detected at high nanomolar concentrations (150 -200 nM) pointed to the potential presence of low affinity sites (see below). The rank order of regional high affinity labeling closely corresponded to the labeling pattern observed autoradiographically.
As a photoaffinity ligand, [ 3 H]CGP 61594 is known to interact with the NR1 subunit. 2 It was therefore tested whether the molecular target specificity was retained in brain regions containing high affinity [ 3 H]CGP 61594 sites. Following UV irradiation of brain membranes in the presence of 25 nM [ 3 H]CGP 61594, photolabeled proteins were resolved by SDS-PAGE and visualized by fluorography. A single protein with the size of the NR1 subunit (115 kDa) was specifically photolabeled in all brain areas analyzed as shown by its displacement with glycine (Fig. 2). The relative radiolabeling intensity of the 115-kDa protein band varied in different regions with a rank order corresponding to that of the B max values in reversible binding in the respective brain regions. Thus, the molecular specificity of [ 3 H]CGP 61594 photolabeling was retained, and the regional intensity of photolabeling was paralleled by the level of reversibly bound [ 3 H]CGP 61594.
The regional distribution of [ 3 H]CGP 61594 binding was strikingly similar to the distribution of the NR2B subunit previously visualized at the level of the mRNA and protein (38,40,46,47). Furthermore, brain regions that lack expression of the NR2B subunit such as the adult cerebellum (38,40,46,48) also lacked high affinity binding sites for [ 3 H]CGP 61594 (Fig. 1 (36). NMDA receptor subtypes were immunoprecipitated with antisera selective for the subunits NR1, NR2A, NR2B, NR2C, or NR2D, and the radioactivity of the immunoprecipitate was measured by liquid scintillation counting. Nonspecifically precipitated radioactivity was determined by co-incubating the antiserum with an excess of the respective peptide antigen. Photolabeled proteins were specifically immunoprecipitated only with the NR1 or NR2B antisera (Fig. 3A). The amount of radioactivity immunoprecipitated was much higher with the NR1 antiserum than with the NR2B antiserum. This difference may be due to either an excess of free NR1 subunits in brain tissue (50,51) or to differential reactivities of the NR1 and NR2B antisera. The NR2A, NR2C, or NR2D antibodies did not result in specific immunoprecipitation of radioactivity (Fig. 3A), although the respective NMDA receptor subtypes were immunoprecipitated as verified by the detection of the NR1 subunit by Western blotting (Fig. 3B)   lacks expression of NR2B and NR2D subunits (38,46,48). [ 3 H]CGP 61594 was used at a concentration of 180 nM, and nonspecific binding was determined with 1 mM glycine or 10 M CGP 61594. The inhibitory potency of CGP 61594 (IC 50 ) in the cerebellum was found to be of 231 Ϯ 56 nM (n ϭ 4). This value was attributable to NR1/2C or NR1/2A receptors or both. It was therefore tested whether low affinity [ 3 H]CGP 61594 sites were also detected in cerebral cortex and hippocampus, brain areas that express preferentially the NR2A and NR2B subunits and largely lack the NR2C and NR2D subunits (39,40,46,47). In cerebral cortex/hippocampus membranes an IC 50 (Fig. 4A). Thus, the reversible labeling was most likely due to homomeric NR1 receptors present in HEK 293 cells, even when the NR1 subunit is co-expressed with an NR2 subunit. The presence of NR1 homomeric receptors has previously been described in HEK 293 cells cotransfected with the NR2A subunit cDNA (30). Therefore, [ 3 H]CGP 61594 binding was analyzed in a test that is operative only in heteromeric receptors. Glutamate antagonists, such as AP5, interact with heteromeric receptors but virtually lack affinity for homomeric NR1 receptors (K i Ͼ 50 M) (19,52,53). Likewise, glutamate antagonists do not modulate glycine antagonist binding to NR1 homomeric receptors as shown for

. Immunoprecipitation of NMDA receptors irreversibly labeled with [ 3 H]CGP 61594 by NR1 and NR2B subunit-specific antisera.
A, total brain membranes were photolabeled with 25 nM [ 3 H]CGP 61594, solubilized, and immunoprecipitated with antisera directed against the NMDA receptor subunits NR1, NR2A, NR2B, NR2C, or NR2D. The radioactivity in the immunoprecipitates was determined by liquid scintillation counting. Nonspecific precipitation was determined in parallel assays in presence of the respective peptide antigen (10 g/ml). The columns show the mean of two to three independent experiments and the maximum and minimum value. B, the ability of the different antisera to immunoprecipitate the respective receptor subtype was verified by the detection of the NR1 subunits in Western blots of each immunoprecipitate. Nonspecific immunoreactivity (right lane) was determined in presence of the respective peptide antigen (10 g/ml). Electrophysiological Analysis of CGP 61594 on Recombinant Receptors-The differential interaction of CGP 61594 with NMDA receptor subtypes was functionally analyzed in Xenopus oocytes expressing the NR1 subunit in combination with the NR2 subunit variants NR2A, NR2B, NR2C, and NR2D. In this test system, homomeric NR1 receptors were not expected to appreciably interfere with the analysis, since they generate only very small currents compared with heteromeric receptors (32,55,56). Glutamate-evoked responses were measured in the presence of glycine concentrations eliciting a half-maximal response (26,29). The glutamate-evoked currents were inhibited by CGP 61594 with high potency only when the NR1 and NR2B subunits were co-expressed (IC 50 ϭ 45 Ϯ 11 nM) (Fig. 5). The other heteromeric receptors were less sensitive to inhibition by CGP 61594. The IC 50 value determined for the NR1/2C receptor was four times higher (IC 50 ϭ 164 Ϯ 27 nM) as compared with NR1/2B. An even lower inhibitory potency of CGP 61594 was found for the NR1/2A receptor (IC 50 ϭ 430 Ϯ 105 nM) and the NR1/2D receptor (IC 50 ϭ 340 Ϯ 61 nM) (Fig. 5). Thus, the major NR2 variants contribute to distinct glycine antagonist sites as shown by an up to 10-fold difference in their sensitivity to inhibition by CGP 61594. DISCUSSION This report shows that the glycine antagonist [ (19,43,45), [ 3 H]CGP 61594 binding is restricted to the forebrain with a regional distribution that strikingly corresponds to the localization of NR2B mRNA and protein in adult rodent brain (38,40,46,47,57). The presence of low numbers of [ 3 H]CGP 61594 sites in medulla/pons is consistent with the expression of NR2B mRNA in a variety of brain stem nuclei, including superior and inferior colliculus, substantia nigra pars compacta, ventral tegmental area, locus coeruleus, pontine nucleus, and inferior olive (58). Furthermore, the prominent expression of [ 3 H]CGP 61594 binding sites in neonatal brain corroborates their identity with receptors containing the NR2B subunit, since the NR2A and NR2C subunits are virtually absent at birth and the NR2D subunit is expressed only at low levels (37,39,40,46,58).
2) NMDA receptors photolabeled with [ 3 H]CGP 61594 in adult brain could be immunoprecipitated with antisera directed against the NR1 and NR2B subunit only, but not with NR2A, NR2C, or NR2D antiserum. The lack of the NR2A antiserum to immunoprecipitate the photolabeled NMDA receptors is of particular interest in view of the presence of NMDA receptors containing both NR2B and NR2A subunits in situ (36,51,59). Thus, it appears that the NR2A, NR2C, or NR2D subunits are not prominent constituents of the photolabeled receptors.
3) Among the recombinant receptors tested, only the NR1/2B subtype displayed high affinity [ 3 H]CGP 61594 binding sites as shown by its allosteric modulation with the glutamate antagonist AP5. Furthermore, in an electrophysiological analysis, high affinity CGP 61594 antagonism was seen with NR1/2B receptors. The potency of CGP 61594 in inhibiting glutamateinduced currents (IC 50 ϭ 45 Ϯ 11 nM) was similar to the affinity of [ 3 H]CGP 61594 binding determined in forebrain regions (K D ϭ 23 Ϯ 2 nM). Thus, at low ligand concentrations CGP 61594 is a selective ligand for NR1/2B receptors not only in situ but also with recombinant receptors.
In contrast to the NR2B subunit, the NR2A subunit generates a glycine binding domain that binds CGP 61594 with lower potency. This is shown electrophysiologically for recombinant NR1/2A receptors (IC 50 ϭ 430 Ϯ 105 nM) and in situ by radioligand binding in cerebral cortex/hippocampus, where, in addition to the NR1/2B receptors, the NR1/2A receptors are predominant (IC 50 ϭ 371 Ϯ 85 nM). A similar low affinity CGP 61594 binding site is formed upon co-expression of the NR2D subunit (IC 50 ϭ 340 Ϯ 61 nM). Notably, the assembly of the NR1 and the NR2C subunits results in an antagonist domain with intermediate affinity for CGP 61594, most clearly demonstrated by the inhibitory potency of CGP 61594 for NR1/2C receptors (IC 50 ϭ 164 Ϯ 27 nM). A site with comparable affinity is found in situ in the cerebellum (IC 50 ϭ 231 Ϯ 56 nM in radioligand binding), where the NR2C subunit is prevalent (38 -40, 46, 48), although NR2A containing receptors may also contribute to this IC 50 value. Thus, using CGP 61594 as a tool, a comparative analysis of NR2 subunit sequences should help to define the NR2 domains that differentially modulate the glycine site of the major NMDA receptors. New insights into the subunit configurations of NMDA receptor subtypes may be gained.
[ 3 H]CGP 61594 provides the first radioligand that reliably discriminates the glycine site of different NMDA receptor subtypes. This is of particular interest since the glycine site of NR1/2B receptors is of therapeutic relevance as a drug target for the treatment of diseases such as stroke or epilepsy. Similar to CGP 61594, the neuroprotective agents ifenprodil and eliprodil, acting at the polyamine site of NMDA receptors, interact preferentially with NR1/2B receptors (60,61). They are characterized by a favorable side effect profile at neuroprotective doses compared with agents that inhibit NMDA receptors nonselectively (62)(63)(64). Thus, the glycine antagonist CGP 61594 may serve as a valuable tool for developing potential therapeutic agents targeted to the NR1/2B receptor.