<<

The Journal of Neuroscience, September 1994, 74(g): 54716494

The Molecular Basis of NMDA Subtypes: Native Receptor Diversity Is Predicted by Subunit Composition

Amy L. Buller,’ Heidi C. Larson,’ Benjamin E. Schneider,’ Jean A. Beaton,’ Richard A. Morrisett,’ and Daniel T. Monaghan’s’ ‘Department of Pharmacology and 2Division of Neurology, Department of Internal Medicine, University of Nebraska Medical Center, Omaha, Nebiaska 68198-6260

The relationship between four pharmacologically distinct tive NMDA receptor subtypes differ in their NR2 composition. NMDA receptor subtypes, identified in radioligand binding Furthermore, the NR2 subunits significantly contribute to the studies, and the recently identified NMDA receptor subunits anatomical and pharmacological diversity of NMDA receptor (NRl a-g, NRSA-D) has not been determined. In this report, subtypes. we demonstrate that the anatomical distribution of the four [Key words: NMDA, oocytes, in situ hybridization, recom- NMDA receptor subtypes strikingly parallels the distribution binant DNA, receptors, glutamate, autoradiography, voltage of mRNA NRSA-D subunits. The distribution of clamp] NRSA mRNA was very similar to that of “antagonist-prefer- ring” NMDA receptors [defined by high-affinity 3H-2-carbox- NMDA receptors mediate synaptic transmission and neural ypiperazine-4-yl-propyl- 1 -phosphonic PH-CPP) binding sites; plasticity at many sites in the mammalian CNS (Mayer and correlation coefficient = 0.881. -preferring NMDA re- Westbrook, 1987; Collingridge and Lester, 1989; Monaghan et ceptors localized to brain regions expressing both NRPB al., 1989) and also contribute to epileptiform activity and neu- mRNA and NRl- mRNA (NRl splice variant lacking insert ronal cell death in a number of experimental and pathological 1). NRSC mRNA was largely restricted to the cerebellar gran- conditions (Cotman et al., 1989; Greenamyre and Young, 1989; ule cell layer, a region that displays a unique pharmacolog- Meldrum and Garthwaite, 1990). The identification and mo- ical profile. NRPD mRNA localized exclusively to those dien- lecular characterization of functionally distinct NMDA receptor cephalic nuclei that have a fourth, distinct pharmacological subtypeswill contribute to a better understandingof the specific profile (typified by the midline thalamic nuclei). role of these receptors in many normal and pathological con- The pharmacology of native NMDA receptors was com- ditions. pared to that of heteromeric NMDA receptors expressed in Recently, two families of NMDA receptor subunits (NRla- Xenopus oocytes (NRl /NRSA, NRl INR28, NRl INRPC). The g, correspondingto NR 1A-G, Sugiharaet al., 1992; and NR2A- oocyte-expressed NR 1 /NRSA receptor displayed a higher D) have been cloned and sequencedfrom rat (Moriyoshi et al., affinity for antagonists and a slightly lower affinity for ago- 1991; Monyer et al., 1992; Ishii et al., 1993) and mouse(Ikeda nists than the NRl /NRSB receptor. These patterns are anal- et al., 1992; Kutsuwada et al., 1992; Meguro et al., 1992; Ya- ogous to those found for radioligand binding to native re- mazaki et al., 1992). Alternative splicing of a single NRl ceptors in the lateral thalamus and medial , generateseight isoforms with distinct functional properties (An- respectively. NMDA receptors in the lateral thalamus (with antharam et al., 1992; Durand et al., 1992, 1993; Nakanishi et a high density of NRPA subunit mRNA) displayed higher af- al., 1992; Sugihara et al., 1992; Hollmann et al., 1993). Het- finity for antagonists and a lower affinity for than erogeneity within the NR2 subunit family resultsfrom expres- did NMDA receptors of the medial striatum (a region rich in sion of four closely related (Kutsuwada et al., 1992; Mon- NRPB mRNA). Relative to the NRl /NRSA and NRl /NRSB re- yer et al., 1992; Ishii et al., 1993). Native NMDA receptorsare ceptors, oocyte-expressed NRl /NRSC receptors had a lower believed to be heteromeric complexes. While the NR 1 subunit affinity specifically for both D-3-(2-carboxypiperazin-4-yl)-l- can form functional homo-oligomeric receptors, coexpression propenyl-1 -phosphonic acid (o-CPPene) and homoquino- of NRl with NR2A, NR2B, or NR2C in vitro greatly enhances linate (HQ). This pattern was identical to that observed for responsesto NMDA (Kutsuwada et al., 1992; Meguro et al., cerebellar (NRPC-containing) versus forebrain (NRPA- and 1992; Monyer et al., 1992; Ishii et al., 1993). NRSB-containing) NMDA receptors. Taken together, the data Heterogeneity ofNMDA receptorshas been suggestedby pre- in this report suggest that the four previously identified na- vious pharmacologicalstudies. Results from radioligand bind- ing and electrophysiological studieshave identified four phar- macologically distinct populations of NMDA receptors. Two of Received Oct. 29, 1993; revised Mar. 1, 1994; accepted Mar. 15, 1994. thesepopulations are typified by NMDA receptorsfound in the This work was supported by NIH Grant NS 28966. We thank Dr. Shigetada cerebellum (Perkins and Stone, 1983; Sekiguchi et al., 1990; Nakanishi and Dr. Peter Seeburg for their generous gifts of NR 1 and NR2 cDNAs, resoectivelv. We also thank Dr. Jeff Watkins for oroviding several compounds. Ebert et al., 199I; Monaghan and Beaton, 1991; O’Shea et al., in&ding homoquinolinate, and Dr. Paul Herrling for prohding o-CPPene. 1991; Beaton et al., 1992) and the medial thalamic nuclei (Bea- Correspondence should be addressed to Dr. Daniel T. Monaghan, Department of Pharmacology, University of Nebraska Medical Center, 600 South 42 Street, ton et al., 1992) which display distinct pharmacologicalprofiles Omaha, NE 68198-6260. at both the glutamate and binding sites. The Copyright 0 1994 Society for Neuroscience 0270-6474/94/145471-14$05.00/O two remaining groupsof NMDA receptorsdiffer in their relative 5472 Buller et al. * Molecular Basis of NMDA Receptor Diversity

affinities for agonists and antagonists (Monaghan et al., 1988; of insert 1 sequence (Sugihara et al., 1992)]; NRl -, 5’-AAACTGCA- Monaghan, 1991; Sakurai et al., 1993) as well as in specific GCACCTTCTCTGCCTTGGACTCCCGTTCCTCCAG-3’ [comple- biochemical properties (Honor6 et al., 1989; Cunningham and mentary to sequences encoding amino acids 184-197 of NRla which flank the insert 1 insertion site (Sugihara et al., 1992)]. Michaelis, 1990; Ogita et al., 199 1). Examples of these NMDA Oligonucleotides were 3’ end labeled with a-35S-dATP using terminal receptor subtypes are the NMDA receptors in the ventral pos- deoxynucleotidyl transferase (New England Nuclear, Boston, MA). Un- terior nucleus of the lateral thalamus that display a relatively incorporated nucleotide was removed by Nensorb chromatography (New high affinity for antagonists (antagonist-preferring receptor) and England Nuclear, Boston, MA). those of the medial striatum that display a higher affinity for In situ hybridization. Adult male SpragueDawley rats were sacrificed agonists (agonist-preferring receptor). The objective of the pres- and brains were removed and immediately frozen on dry ice. Sections ent study was to determine the relationship between the iden- (12 pm) were thaw mounted, air dried, and fixed for 5 min in 4% paraformaldehyde at 4°C (Wisden et al., 199 1). Slide-mounted sections tified receptor subunits and the native NMDA receptor sub- were washed in phosphate-buffered saline and dehydrated in 70% eth- types. We now report that the pharmacologically distinct NMDA anol prior to storage in 95% . Hybridizations were performed receptor subtypes colocalize with specific NR2 subunit mRNAs. as described by Wisden et al. (199 1). 35S-labeled oligonucleotide probe Furthermore, the native NMDA receptor pharmacology can be was dissolved in hybridization buffer (2000 cpm/pl, -0.2 nM; New England Nuclear, Boston, MA) containing 0.2 M DTT and applied to reproduced with the corresponding combination of recombinant sections. Sections were hybridized overnight under Parafilm-sealed cov- NMDA receptor subunits expressed in Xenopus oocytes. The erslips at 42°C (except NRl - probe, which was incubated at 50°C) and data in this report demonstrate that the pharmacological and then washed for 20 min at a final stringency of 1 x SSC (0.015 M sodium anatomical differences between native NMDA receptor sub- citrate, 0.15 M NaCl) at 60°C (70°C for NRl - probe). Sections were air types depend significantly on the particular type of NR2 subunit dried and exposed to film (@Max; Amersham, Arlington Heights, IL). Exposure time varied for the different probes and is indicated in the present in the NRl/NR2 heteromer. figure legends. Probe specificity was confirmed by incubation of the radiolabeled oligonucleotides with 100 nM unlabeled probe. Only oligonucleotides with the same sequence inhibited probe hybridization. Results of in situ Materials and Methods hybridization using NRZA-D and NRl + probes were consistent with Quantitative receptor autoradiography. Brainswere removed from adult previous reports (Monyer et al., 1992; Watanabe et al., 1992; Ishii et male Sprague-Dawley rats (200-250 gm) and immediately frozen. Sec- al., 1993; Standaert et al., 1993). Cross-hybridization of the NRl- tions (6 pm) were cut in a cryostat, thaw mounted onto gelatin-subbed probe to NR 1+ mRNA was prevented by increasing the incubation slides, and refrozen. L-‘H-Glutamate binding to NMDA receptors was and wash temperatures to 50°C and 7o”C, respectively. This selectively performed as previously described (Monaghan and Beaton, 199 1). Brief- reduced NRl - probe hybridization to brain regions containing high ly, slides were washed for 15-30 min at 0°C in 50 mM Tris acetate, pH densities of NRl + mRNA (e.g., thalamus and layer 4 of the parietal 7.0, and preincubated twice for 10 min at 30°C in 50 mM Tris acetate, cortex). pH 7.0. NMDA receptorswere selectively labeled by incubationin 50 Quantitative analysis of autoradiograms. Anatomical observations nM L-[3H]-glutamate (54.7 Ci/mmol; New England Nuclear, Boston, were confirmed by computer-assisted image analysis of the autoradi- MA) in buffer containing 50 mM Tris acetate, pH 7.0, 10~~ , 5 ograms (MCID, Imaging Research, St. Catharines, Ontario, Canada) that PM amino-3-hydroxy-isoxazole-4-propionate (AMPA), 1 PM kainate, were calibrated by either 3H-Microscales (for ‘H-CPP and L-[3H]-glu- and 100 PM 4-acetamido-4’-isothiocyano-2’,2-disulfonic acid stilbene tamate) or 35S brain paste standards as previously described (Siegel, for 20 min at 0°C to displace all non-NMDA receptor binding. Agonist 1989) for in situ hybridization autoradiograms. To quantify the corre- and antagonist competitors were added as indicated in the Results. lation between NMDA receptor subunits and NMDA receptor binding Nh4DA-specificbinding (defined as binding displaced by 500PM NMDA) sites, the signal intensity in 32 brain regions were determined for in situ represented >95W of the L-3H-glutamate-specific binding. ‘H-(*)-2- hybridization and receptor-radioligand binding autoradiograms. The carboxypiperazine-4-yl-propyl- 1-phosphonic acid (‘H-CPP, 25 Ci/mmol; density values for each animal were normalized to the density found in New England Nuclear, Boston, MA) autoradiography was performed as the parietal cortex and then averaged between animals (prior normal- described (Monaghan et al., 1988). Following two 10 min preincubations ization greatly reduces interexperiment error; Monaghan and Beaton, at 3O”C, slide-mounted sections were incubated in 50 mM Tris-Cl, con- 199 1). Correlation coefficients were determined for the correlations be- taining 100 nM 3H-CPP for 20 min at 0°C. Sections were washed for tween NMDA receptor subunits (NRZA, NR2B, NRl +, and NRl -) 20 set and placed against ‘H-sensitive film (Hyperfilm-3H, Amersham, and NMDA receptor binding sites (‘H-CPP binding sites and NMDA- Arlington Heights, IL) with ‘H standards (Microscales, Amersham, Ar- specific L-3H-glutamate binding sites). lington Heights, IL). Dose-response curves generated for competitors In vitro transcription. pN60 containing the 4.2 kb NRl cDNA (Mo- of L- 3H-glutamate binding in the three brain regions described (lateral riyoshi et al., 1991) was the generous gift of Dr. Shigetada Nakanishi thalamus, medial striatum, and cerebellar layer) were de- (Kyoto University Faculty of Medicine, Kyoto, Japan). This cDNA termined by computer-assisted image analysis of the autoradiograms corresponds to the predominant splice variant found in the CNS, NRlA (MCID, Imaging Research, St. Catharines, Ontario, Canada). Dose-re- (Sugihara et al., 1992) and NRl-SL (Anantharam et al., 1992). cDNAs sponse relationships were analyzed by iterative, nonlinear regression encoding NR2A, NRZB, and NRZC (Monyer et al., 1992), generously analysis using a modification of the method of Marquardt (GRAPHPAD, provided by Dr. Peter Seeburg (University of Heidelberg, Heidelberg, IS1 Software, San Diego, CA) with the goodness of fit assessed by the Germanv)._,, were linearized bv digestion with Not1 (NR1 j. EcoRI (NRZA sum of the squares of the absolute deviation from the curve for un- and NR2C), or EcoRV (NRiB) and transcribed in kro ising T3 &RZA, transformed one- and two-site competitive binding curves (for equa- NRZB, NRZC) or T7 (NRl) RNA polymerase as described (Buller and tions, see Ehlert and Tran, 1990). The Cheng-Prusoff (1973) equation White, 1988). was used to correct for regional variations in L-‘H-glutamate affinity Translation in Xenopus oocytes. Oocytes were removed from mature (Monaghanand Beaton,199 1). Chemicalswere from Sigma(St. Louis, Xenopus laevis (Xenopus One, Ann Arbor, MI) and were dissociated in MO) and Tocris (Essex, UK). All experiments were performed at least 2 mg/ml collagenase (type IA, Sigma Chemical, St. Louis, MO) in Cal+- four times and representative dose-response curves are shown. free OR-2 (82.5 mM NaCl, 2.5 mM KCl, 1 mM MgCl,, 5 mM HEPES, Oligonucleotide probes. Sequences for antisense oligonucleotide probes pH 7.6) and the remaining follicle layer was removed manually. Isolated soecific for NRZA. NR2B. and NR2C were from Monver et al. (1992). stage V and stage VI oocytes were maintained in ND-96 (96 mM NaCl, I& addition, the fdllowing sequences were also used: fiR2D, 5’:CT& 2 mM KCl, 1.8 mM CaCl,, 1 mM MgCl,, 5 mM HEPES, pH 7.6) sup- CGAATCCTCGGAGTCCGAAGGCGAAGGCTCGAGGTCC- plemented with 2.5 mM sodium pyruvate, 1000 U/ml penicillin, 0.1 AGGTA-3’ [complementary to sequences encoding amino acids 1067- mg/ml streptomycin (White et al.,- i985). NMDA receptor RNAs were 108 1 of the c4 subunit (Ikeda et al., 1992) and differs by one nucleotide dissolved in sterile distilled H,O. NRl RNA was mixed in eauimolar from the corresponding rat sequence (Ishii et al., 1993)]; NRI +, 5’- stoichiometry with either NR2& NR2B, or NR2C RNA. The final RNA GGGTCCGCGCTTGTTGTCATAGGACAGTTGGTCGAGGTT- mixture (50 nl) was microinjected (15-30 ng total) into the oocyte cy- TTCATA-3’ [complementary to sequences encoding amino acids 6-20 toplasm as previously described (White et al., 1985). Oocytes were The Journal of Neuroscience, September 1994. 14(9) 5473 incubated in complete ND-96 solution at 19°C for l-4 d prior to elec- Cerebral cortex. High levels of NMDA-sensitive L- 3H-glu- trophysiological assay. tamate binding sites (agonist-preferring) were evident in the Electrophysiology.Electrophysiological responses were measured us- ing a standard two-microelectrode voltage clamp (model OC-725A oo- anterior cingulate, insular, and perirhinal cortices, with lower cyte clamp, Warner Instruments, Hamden, CT). Electrodes were filled levels in the parietal and temporal cortices (Fig. 1B). 3H-CPP with 3 M KC1 and had resistances of0.6-3.0 Ma. The recording chamber binding sites(antagonist-preferring), in contrast, displayeda more was continuously perfused with 116 mM NaCl, 2 mM KCl, 0.5 mM subtle regional variation with a slightly higher density in the CaCl,, 5 mM HEPES, pH 7.4, and all drugs were dissolved in the same parietal and temporal cortical regions than in the anterior cin- solution. Whole cell current elicited by bath application of NMDA receptor in the presence of 10 PM glycine was determined at a gulate, insular, and perirhinal cortices (Figs. lA, 3). Cortical holding potential of -60 mV. Only healthy oocytes with a minimum lamination patterns also differed: while the distribution of ag- resting potential of -30 mV were used. Dose-response curves for ag- onist-preferring NMDA receptors (receptors preferentially la- onist activation of heteromeric NMDA receptor complexes expressed beled by L-3H-glutamate) showed a marked laminar variation in oocytes were fit (GRAPHPAD, IS1 Software, San Diego, CA) according to the equation I = 1,&l + (EC,,/@], where I is the current response, with higher densities in layers I-III and in layer Va of patietal I,,,,, is the current response evoked at maximal agonist concentration, cortex, ‘H-CPP binding sites displayed less laminar variation A is the agonist concentration, n is the Hill coefficient, and EC,, is the with slightly greater binding in layers I-V than in layer VI. Thus, concentration of agonist producing a half-maximal response. Antagonist layer IV displayed the greatest difference between L-3H-gluta- inhibition curves were fit by the equation I = I,,,,. - I,,,,,/[ 1 + (IC,,/ mate and 3H-CPP binding sites with a high density of 3H-CPP A)“], where I is the current response in the presence of antagonist, I,,,,, is the control current response in the absence of antagonist, and A is binding sitesand a low density of L-3H-glutamatesites (compare the antagonist concentration. IC,, values were corrected for agonist Fig. lA,B). In the entorhinal cortex, higher levels of 3H-CPP affinity as described (Durand et al., 1992). binding sites were found in two superficial layers (separated by Microinjection of NRl RNA in combination with NR2A, NR2B, or layer II) while L-3H-glutamate binding sites were of lower and NRZC RNA resulted in expression of NMDA-gated channels in Xenopusoocytes. In all cases, currents elicited by heteromeric NMDA more uniform density. receptors were 10-100 times larger than currents mediated by homo- The cortical distribution of NR2A and NR2B mRNA par- meric NRl receptors (data not shown). Typical current responses for alleled the distribution of antagonist- and agonist-preferring heteromeric NMDA receptors expressed in oocytes assayed in the pres- binding sites, respectively. As found for 3H-CPP binding sites, ence of 100 NM NMDA and 10 NM glycine were approximately 100 nA high levels of mRNA encoding NR2A were detected in the (NRl/NRZA), 20 nA (NRl/NR2B), and 200 nA (NRlLNR2C). The somewhat lower macroscopic currents in this report, relative to other parietal, entorhinal (superficial layers), and temporal cortices reports (e.g., Monyer et al., 1992) were most likely due to the low and lower levels in the anterior cingulate, insular, and perirhinal concentration used in the recording solutions. Cal+ entry through NMDA- cortices (Figs. 2A, 3). In contrast, higher levels of NR2B mRNA gated channels expressed in oocytes has been shown to activate endog- were observed in the anterior cingulate, insular, and perirhinal enous Ca2+ activated Cl- currents. This endogenous current is respon- sible for a large part of both the initial and prolonged phases of the cortices than in the parietal, temporal, and entorhinal cortices NMDA response in oocytes (Leonard and Kelso, 1990). Reduction of (Fig. 2B). NR2A and NR2B mRNA also differed in their lam- the Ca*+ concentration in the perfusion solution reduces both compo- ination pattern in the parietal cortex. Greater densities of NR2A nents of the NMDA response in oocytes (Leonard and Kelso, 1990). subunits were found in layers II-IV than in layers V and VI. NR2B mRNA was present at higher levels in layers II-III than in IV-VI (Figs. 2, 3). Thus, layer IV displayed relatively high Results levels of NR2A mRNA and low levels of NR2B mRNA. Over- Comparison betweenthe distributions of agonist-preferring/ all, both the regional and laminar variations in antagonist-pre- antagonist-preferring NMDA receptorsubtypes and NMDA ferring and agonist-preferring NMDA receptors paralleled the receptorsubunits distribution of NR2A and NR2B mRNA, respectively. The anatomical localizations of agonist-preferring and antago- NRl + mRNA density varied widely acrosscortical regions nist-preferring NMDA receptor populations identified by ra- (Figs. , 3). Patietal, temporal, and superficial entorhinal cor- dioligand binding studies are shown in Figure 1. As described tices displayed approximately threefold higher levels of NRl + previously (Monagban et al., 1988; Monagban, 199 l), the NMDA mRNA than the anterior cingulate, perirhinal, insular, and pyr- antagonist ‘H-2-carboxypiperazine-4-yl-propyl- 1-phosphonic iform cortices. In contrast, higher levels of NR 1 - mRNA were (JH-CPP) and the NMDA agonist L-3H-glutamate label distinct found in the anterior cingulate, perirhinal, and insular cortices. NMDA receptor populations, referred to as antagonist-prefer- Thus, the localization of NRl + and NRl - mRNA between ring and agonist-preferring, respectively. In this report we have cortical regions parallels the distribution of antagonist-prefer- compared, in detail, the distributions of these two binding sites ring and agonist-preferring NMDA receptors, respectively. to the distributions of NMDA receptor subunit mRNAs. Strik- However, lamination patterns within a defined cortical region ing parallels were observed between the localization of NR2A, partially differ between NRl + mRNA and receptor subtypes. NR2B, NRl+ (NRl splice variant with insert I), and NRl- While a high density of 3H-CPP binding sites were found in (NRl splice variant lacking insert 1) mRNAs and the agonist/ layers II-IV, NRl + mRNA was found enriched in layers IV, antagonist-preferring binding sites. In contrast, the distribution Vb, and VI. Thus, in layers II-III, there is a high density of 3H- of antagonist-preferring and agonist-preferring NMDA receptor CPP binding sites (and NR2A mRNA), but a low density of binding sites displayed no similarity to NR2C or NRZD mRNA NRl+. distributions (compare Figs. 1, 6). Agonist-preferring sites (as . The distribution of agonist-preferring NMDA well as the pharmacologically distinct cerebellar and midline receptors displayed a modest medial-to-lateral gradient within thalamic NMDA receptors) were preferentially labeled by LJH- the striatum (Figs. lB, 3). The opposite pattern was observed glutamate both in the presenceand absenceof 6 PM D-2-amino- for 3H-CPP binding sites, with higher levels in the lateral stria- %phosphonopentanoate(D-AP5). The description below for ag- turn than in the medial striatum (Fig. lA, 3). In the nucleus onist-preferring sites was consistent with both conditions and accumbens,a high density of L-‘H-glutamate binding siteswere was confirmed by quantitative image analysis. observed in both the core and shell regions. 3H-CPP binding 5474 Buller et al. l Molecular Basis of NMDA Receptor Diversity

Figure I. Autoradiography of IH-CPP binding sites (antagonist-preferring) (A) and t?H-glutamate binding sites (predominantly agonist-preferring sites) (B) in horizontal sections of rat brain. Note the relatively low levels of )H-CPP binding sites in the medial striatum (MS), lateral septum (L), anteroventral nucleus (4 v), and olfactory bulb glomerular layer (G), and the relatively high levels of ‘H-CPP binding in the ventroposterior nucleus (VP), entorhinalcortex (E), and parietalcortex layer 4 (Pa, 4). 1, CA1 of ;3, CA3 of hippocampus;A, anterior cingulate;CB, cerebellum; EP, external plexiform layer of the olfactory bulb, F-O, frontal-orbital cortices; H, hippocampus; LI, lateral septum, intermediate region; MC. medial geniculate; P, posterior nucleus; PF, parafascicular nucleus; Pr, perirhinal cortex; SH, septohippocampal nucleus; PS, presu- biculum/parasubiculum; T, temporal cortex.

siteswere present at a higher density in the higher levels of 3H-CPP binding were observed in the dorsal shellthan in the core region. As found for 3H-CPP binding sites, lateral septum. The lateral septum had a relatively high density NR2A subunit mRNA was distributed in a distinct lateral to of GH-glutamate binding sites(Figs. 1B, 3). Within the lateral medial gradient within the striatum, with notably higher levels septum, higher levels of 3H-glutamatebinding siteswere found in the shell of the accumbensthan in the core region. NR2B in the septohippocampal nucleus, ventral lateral septum, and mRNA was found at high levels throughout the striatum and distinct subfieldsof the intermediate lateral septum, and lower accumbens,analogous to the agonist-preferring receptor distri- levels were seenin the dorsal lateral septum. bution. Similar to the distribution of ‘H-CPP binding sites in the NRl f mRNA displayed a lateral to medial striatal distri- septum, NR2A mRNA is in relatively low abundance in this bution that was generally similar to NR2A mRNA and antag- region, with higher levels found in the dorsal lateral septum. In onist-preferring NMDA receptors(Figs. 2C, 3). However, unlike contrast to 3H-CPP binding sites and NR2A mRNA, NR2B the distribution of NR2A mRNA and antagonist-preferring mRNA was relatively abundant in the septum with higher den- NMDA binding sites, NRl + mRNA was low throughout the sities found in the septohippocampal nucleus, ventral lateral nucleus accumbens. The density of NR 1 - mRNA was modestly septum, and distinct subfieldsof the intermediate lateral septum higher (15%) in the medial striatum than in the lateral striatum, than in the dorsal lateral septum. NRl + mRNA was present and high levels of this transcript were found in both the core in low levels throughout the septum (Figs. 2C, 3) with even and shell regions of the accumbens.Thus, in the basalganglia, lower levels in the dorsal lateral septum. High levels of NRl- the distribution of antagonist-preferring binding sitesparallels mRNA were found throughout the lateral septum (Figs. 20, 3), that’ of NR2A mRNA but only partially agreeswith NRl + with higher levels in the dorsal lateral septum. Overall, the septal mRNA. The distribution of agonist-preferringbinding sitescor- distribution of antagonist-preferring and agonist-preferring respondsto both NR2B and NRl - mRNAs. NMDA receptors most closely paralleled the distribution of Septal nuclei. Compared to other brain regions,low levels of NR2A and NR2B mRNA, respectively, and only partially par- ‘H-CPP binding sites were present in the lateral septum (Fig. alleled the distribution of NR 1+ and NR 1- mRNA, respec- 1A). Within the lateral septum, the lowest levels of 3H-CPP tively. binding sites were found in the intermediate lateral septum, Thalamic nuclei. Among the various nuclei of the thalamus, ventral lateral septum, and septohippocampal nucleus, while the same distinctive distribution was seenfor both antagonist- The Journal of Neuroscience, September 1994, f4(9) 5475

Figure 2. Localizationof NR2 and NRl subunitmRNAs by in situ hybridization: autoradiograms of in situ hybridization using 35S-labeled antisense oligonucleotide probes specific for NR2A (A), NRZB (B), NRl + (C), and NRl - (0). NRZA and NRl + mRNA expression was low in the striatum (5’) and lateral septum (L), perirhinal cortex (Pr), and anterior cingulate (AC), and high in the ventroposterior nucleus (VP), parietal cortex (Pa), layer 4 (4), and superficial entorhinal cortex (E). NRl + mRNA expression differed from NR2A and antagonist-preferring sites by having high levels in the anteroventral nucleus (A V) and low levels in CA1 hippocampus (I). 3, CA3 of hippocampus; CB, cerebellum; EP, external plexiform layer of the olfactory bulb; F, frontal/orbital cortex; P, posterior nucleus; SH, septohippocampal nucleus. Films were exposed for 21 d (NR2A, NRZB, NRl +) or 13 d (NRl -). preferring NMDA receptors and NR2A mRNA (Figs. lA, 2A). clei and in the dorsal lateral geniculate.Moderate levels of 3H- High-affinity 3H-CPP binding sites and NR2A mRNA were CPP binding sites and NR2A mRNA were observed in the found in highestconcentrations in the ventral posterior nucleus, medial geniculate and paratenial nucleus, lower levels in the with high levels in the laterodorsaland mediodorsallateral nu- posterior and lateroposterior nuclei, and lowest levels in the 5476 Buller et al. * Molecular Basis of NMDA Receptor Diversity

[3H]CPP Binding Sites NRPA mRNA NRl+ mRNA Cortex: Insular Perirhinal r- Anlerior Cingulate Posterior Cingulale FronlaWMolor Parielal 11-111 Parielal IV Parielal V-VI

Slrlatum: Medial Laleral

Septum: Seplohipp. NW Cal. Seplum Venlral Lal. Seplum Intermediate Lat. Septum Dorsal

Thalamus: Anterovenlral Paravenlricular Lalerodorsal Poslerior Laleroposterior Venlrolaleral Mediidorsal Lal. Venlroposlerior

Other:Pre/Parasubiculum lnlerior Colliculus Medial I 0 1 2 0 1 2 3.5 Relative Density Relative Density Relative Density

L-[3H]GIu Binding Sites NRPB mRNA NRl- mRNA

Corlex: Insular Perirhinal Anterior Cingulale Poslerior Cingulale FrontallMolor Parielal 11-111 Parielal IV Parielal V-VI

Strlatum: Medial Laleral

Septum: Seplohipp. Nut. Lat. Septum Ventral Lal. Septum Intermediate Lat. Septum Dorsal

Thalamus: Anlerovenlral Paravenlricular Lalerodorsal Poslerior Laleroposlerior Venlrolaleral Mediodorsal Lal. Venlroposterior

0Iher:PrelParasubiculum lnlerior Colliculus Medial 0 1 2 0 1 2 0 1 2 Relative Density Relative Density Relative Density Figure 3. Quantitative comparison between the density of binding sites for ‘H-CPP and GH-glutamate and NMDA receptor subunit mRNA in representative brain regions. ‘H-CPP binding density displayed a regional variation similar to that of NRZA mRNA and only partially similar to The Journal of Neuroscience. September 1994, 14(Q) 5477

paraventricular and anteroventral nuclei (Figs. IA, 2A). In ad- dition, a high density of both NR2A mRNA and 3H-CPP bind- ing sites was evident within the parafascicular nucleus in a band immediately surrounding the fasciculus retroflexus. NR2B mRNA was present at uniformly high levels throughout most of the thalamus, with lower levels in the paraventricular and intermediodorsal nuclei (Fig. 2B). The distribution of NR2B mRNA was distinct from that of antagonist-preferring receptors and NR2A mRNA, in that NR2B mRNA was found at high levels in the anteroventral nucleus and posterior nucleus and at low levels throughout the parafascicular nucleus. The thalamic distribution of agonist-preferring NMDA re- Subvnit mRNA ceptors was difficult to determine due to the presence of a phar- n NR2A macologically distinct NMDA receptor subtype that is present 0 NR2B in several of the midline thalamic nuclei (Beaton et al., 1992; lid NRl+ see also Fig. 5). Like agonist-preferring NMDA receptor binding q NRI- sites, these binding sites also display higher affinity for L- 3H- glutamate. Thus, the relatively high levels of L- 3H-glutamate binding in the various midline thalamic nuclei (e.g., paraven- L-[3H]GIu [3H]CPP tricular and intermediodorsal nuclei) and in the medial genic- ulate and anteroventral nuclei are largely due to the presence of Figure4. Correlationalanalysis between NMDA receptorbinding sites a distinct NMDA receptor subtype that displays a unique phar- and NMDA receptorsubunit mRNA. Radioligand-receptorbinding and mRNA densitywere determined in 32 brain regions(see Materials macological profile (Beaton et al., 1992). Exclusion of regions and Methods). Thesevalues were usedto determinethe correlation displaying pharmacological properties characteristic of the mid- betweenGH-glutamate or 3H-CPPbinding sites and NMDA receptor line thalamic nuclei (MT-like subtype) revealed that agonist- subunitmRNA. For comparison,the correlationcoefficient for L-)H- preferring NMDA receptors were expressed at a relatively low glutamateand ‘H-CPP bindingsites was 0.09, while the correlation density throughout most of the thalamus (Fig. 1B). Agonist- coefficientfor NRZA mRNA and NR2B mRNA was0.45. preferring NMDA receptors defined as NMDA-specific L-~H- glutamate binding in the presence of 6 MM D-AP5 displayed lower ceptors. Conversely, the presenceof antagonist-preferring sites relative density levels in lateral thalamic nuclei than NMDA- is associatedwith NR2A expression. Quantitative analysis re- specific L-‘H-glutamate binding, consistent with our previous veals that the density of NR2A mRNA and antagonist-prefer- findings (Monaghan et al., 1988). ring binding sitesshow a strong positive correlation (r = 0.88; NR 1 - mRNA was present at low levels throughout the thal- Fig. 4). The weak correlation between NR2B and antagonist- amus (Figs. 20,3), while NR2B mRNA was found at high levels preferring site distributions (r = 0.44) is similar to the correlation throughout most of the thalamus (Fig. 2B). Slightly higher levels between NR2A and NR2B mRNA distributions (r = 0.45). of NR 1 - mRNA were observed in nuclei found along the mid- The distribution of agonist-preferring NMDA receptors dis- line (paraventricular and intermediodorsal nuclei), while high plays many similarities to the distribution of both NRZB and levels of NRl+ mRNA (Fig. 2C) were expressed throughout NRI - mRNAs (each had similar cortical and striatal distri- the thalamus except for the midline structures (paraventricular butions and generally low levels in midbrain and brainstem). and intermediodorsal nuclei). Among the thalamic nuclei, the Overall, regions that display a low density of either NRZB or anteroventral nucleus expressed the highest levels of NRl + NRI - mRNAs, also have a low density of agonist-preferring mRNA (and a low density of 3H-CPP binding sites and NR2A sites. Thus, low levels of agonist-preferring siteswere found in mRNA). 8 the lateral thalamic nuclei (including ventral posterior, poste- In summary, the thalamic organization of antagonist-prefer- rior, and dorsal lateral geniculate) where there is a low density ring NMDA receptors closely paralleled the distribution of NR2A of NR 1- insert 1 mRNA but high NRZB mRNA levels (Figs. 1, mRNA and was distinct from NR2B, NR 1 + , and NR 1 - mRNA 2). Conversely, in the hypothalamus, low levels of agonist-pre- distributions. The low levels of agonist-preferring NMDA re- ferring receptorscorrelates with low levels of NR2B mRNA and ceptors in the thalamus correlates with low levels of NRl- not the higher levels of NRI - mRNA (data not shown). Sim- mRNA and is distinct from the high thalamic levels of NR2A, ilarly, in the lateral septum, the distribution ofagonist-preferring NR2B, and NRl + mRNAs. sitescorresponds better to the distribution of NR2B, rather than Summary and correlational analysis of anatomical observa- NR 1 -, mRNA (Fig. 3). Correlation coefficient analysis(Fig. 4) tions. Overall, there is a marked anatomical correspondence indicates that agonist-preferring NMDA receptors exhibit a between NR2A mRNA and the expression of antagonist-pre- moderate positive correlation to NRl - mRNA (r = 0.46), a ferring. NMDA receptors. Every brain region that expresses better correlation to NR2B mRNA (r = 0.67), and a negative NR2A mRNA also expresses antagonist-preferring NMDA re- correlation to NRl + mRNA (r = -0.5 1). t NR I+ mRNA. GH-glutamatebinding sites displayed a regionalvariation that wasmore similar to NRZB and NR 1- mRNA. The relativedensity of receptorbinding, or in situ hybridization signal,was determined by quantitativeimage analysis using )H or ?S standards,respectively, to correct for nonlinearfilm response.Relative, or normalized,density values were determined by expressingthe observedradioactivity densityas a fraction of the averagedensity obtained for that probein 32 brain regions.The averageSEM was< 101; averagen = 3.8. Septohipp.NW., septohippocampal nucleusof septum;Lat., lateral. 5475 Bullet’ et al. l Molecular Basis of NMDA Receptor Diversity

Figure 5. Distribution of pharmacologically distinct NMDA receptor binding sites. In the absence of inhibitors (A) LJH-glutamate binding sites are found throughout the brain. Addition of 16 PM homoquinolinate (B), 50 PM LY233536 (C), and 10 NM o-CPPene (D) inhibits r?H-glutamate binding in select brain regions. Homoquinolinate potently inhibited binding in all regions except the midline thalamic nuclei (P,P, paratenial and paraventricular-anterior nuclei; PJ, paraventricular-posterior and intermediodorsal nuclei), anteroventral nucleus (A v), the cerebellum (CB), the olfactory bulb glomerular layer (G), periaqueductal gray (PC), and the medial geniculate (MG). In contrast, LY233536 (C) potently inhibited binding in all regions except the cerebellum. At relatively high concentrations (10 NM) D-CPPene (D) potently inhibited L-‘H-glutamate binding in all regions except the cerebellum.

macologically,a variety of agonistsand antagonistswere screened Localization of pharmacologically atypical NMDA receptors for greater subtype specificity. Evaluation of approximately 40 and comparisonto NR2C and NR2D mRNA distributions compounds (data not shown) revealed that two showed greater Receptor-radioligand binding studies. Previous studies have selectivity between brain regions: the agonist homoquinolinate shown that NMDA receptor binding sites in the cerebellum (HQ) and the antagonist D-3-(2-carboxypiperazin-4-yl)- 1-pro- (Ebert et al., 1991; Monaghan and Beaton, 1991; Yoneda et al., penyl- 1-phosphonic acid (D-CPPene). As shown in Figure 5, 1991; Beaton et al., 1992)and in various midline thalamic nuclei D-CPPene and HQ were potent inhibitors of L- ‘H-glutamate (Beatonet al., 1992) display distinct pharmacologicalproperties. binding to NMDA receptorsin the forebrain and poor inhibitors In order to isolate these two NMDA receptor subtypes phar- of binding in the cerebellum and midline thalamus, a pattern The Journal of Neuroscience, September 1994, M(9) 5479

Figure 6. Distributionof NRZC (A) and NRZD (B) mRNA. NRZC mRNA wasfound predominantlyin the cerebellum(CB); low levelswere alsofound in the olfactory bulb glomerularlayer (G); and very low levelswere found in the thalamus.NR2D mRNA wasrestricted to the olfactory bulb glomerularlayer and to various diencephalicnuclei (P,P, paratenialand paraventricular-anteriornuclei; AV, anteroventralnucleus; ML, mediodorsallateral nucleus: P. I. naraventricular-uosteriorand intennediodorsalnuclei; MC, medialgeniculate; PC, periaqueductalgray). Auto- radiogramswere exposed for il d (NR2C)or 4 1 d (NRZD). identical to that previously described for CPP (Beaton et al., and the interpeduncular nucleus. Examination of 10 animals in 1992). At relatively high concentrations (30 times the average several levels of coronal, parasagittal, and horizontal planesdid forebrain K,), D-CPPene (data not shown) and HQ (Fig. 5B) not reveal any NMDA receptors with pharmacological prop- weakly inhibited r/H-glutamate binding in the midline thalam- erties similar to the cerebellum. NMDA receptors of the glo- ic nuclei and cerebellar granule cell layer, although binding in merular layer ofthe olfactory bulb were found have a low affinity most other brain regions was almost completely inhibited. At for HQ and D-CPPene, but specificity for LY233536 was in- even higher concentrations (60 times the averageforebrain K,), conclusive (thus it is uncertain if the glomerular NMDA recep- D-CPPene(Fig. 5D) and HQ (data not shown) still weakly in- tors are cerebellar-like or MT-like). hibited LJH-glutamate binding in the cerebellar granule cell Distribution of NRZC and NRZD sub&it mRNA. In situ hy- layer. While the midline thalamic nuclei and the cerebellum bridization was performed to compare the distribution of NR2 both display a lower affinity for HQ, D-CPPene,and CPP, NMDA subunit mRNAs to that of the pharmacologicallyatypical NMDA receptors in theseregions are differentially sensitive to the an- receptors found in the cerebellum and midline thalamus (Fig. tagonist LY233536 (Beaton et al., 1992; presentedin Fig. 5 for 6). As previously described,NR2C mRNA is almostexclusively comparison). LY233536, at a concentration that was 13 times localized to the cerebellum (Kutsuwada et al., 1992; Monyer et the average forebrain K, , potently inhibited NMDA-sensitive al., 1992; Ishii et al., 1993; Fig. 6A). In addition, NR2C mRNA L-SH-glutamatebinding in the midline thalamic nuclei but not was found at lower concentrations in the pineal gland while in the cerebellum (Fig. 5C). NMDA-sensitive L-3H-glutamate binding siteswere not found The anatomical distributions of NMDA receptors that have in the pineal. the pharmacological properties of the midline thalamus (MT- NR2D mRNA was found throughout the diencephalon and like) and the cerebellum(cerebellar-like) were further evaluated. midbrain (in agreementwith Ishii et al., 1993; Fig. 6B). Regions The distribution of NMDA-sensitive L- 3H-glutamate binding containing the highest levels of NR2D mRNA were identified sites, which were poorly displaced by 10 PM CPP and 10 PM as the paratenial nucleus, paraventricular nucleus, intermedi- D-CPPeneand potently inhibited by 20 I.LMand 50 PM LY233536 odorsal nucleus, mediodorsal lateral nucleus, anteroventral nu- (MT-like pharmacology), was determined. In addition to the cleus, medial geniculate, and rhombiod nucleus (Fig. 6B). Sig- paraventricular and intermediodorsal nuclei, MT-like NMDA nificant levels of NR2D mRNA were also found in the receptors were found in the anteroventral nucleus, the medial periaqueductal gray (especially the dorsal subdivision), inter- geniculate, the mediodorsal lateral nucleus, and the periaque- peduncular nucleus, the glomerular layer of the olfactory bulb, ductal gray (compareFig. 5B, C). In other planesof section (data and the nucleus of the diagonal band. In the adult rat brain, the not shown) NMDA receptors of similar pharmacology were telencephaloncontained very low levels, except the dentate gy- found in the rhomboid nucleus,the nucleusofthe diagonal band, rus, which had low levels. Overall, the distribution pattern of 5480 Buller et al. - Molecular Basis of NMDA Receptor Diversity

NR2D subunit distribution was indistinguishable from that of NMDA receptors, which have a high affinity for LY233536 and a low affinity for HQ, CPP, and D-CPPene (compare Fig. 5, B minus C, to Fig. 6B).

Pharmacology of native and recombinant NMDA receptors Colocalization of specific NR2 subunit mRNAs and distinct NMDA receptor populations suggestedthat the NR2 subunit contributes to the distinct pharmacologies of NMDA receptor subtypes. To test this hypothesis, the pharmacological profiles Log [NMDA]. M of NMDA receptorswere determined in brain regionsthat con- tain relatively pure populations of NMDA receptor subtypes and subunits.Thus, the pharmacologicalproperties ofthe lateral thalamus (enriched in NR2A mRNA and antagonist-preferring receptors), the medial striatum (which displays the purest pop- ulation of both NR2B mRNA and agonist-preferring receptors), !8’:~Q-D and the cerebellar granule cell layer (containing NR2C mRNA) were comparedto the pharmacologyof three heteromericNMDA receptorsexpressed in Xenopus oocytes: NRl/NR2A, NRl/RZB,

0-il and NR 1/NR2C. -5 -4 -3 -5 -4 -3 5B Log [Homoquinohnate]. M The qualitative results shown in Figure indicate that HQ Log [NMDA]. M displays a low affinity for inhibiting r,-3H-glutamate binding to Figure 7. Comparisonof inhibition of GH-glutamatebinding by ag- cerebellar NMDA receptors. Quantitative analysis of HQ in- onists(A, B) in lateral thalamus(solid circles), medialstriatum (open hibition of L- 3H-glutamate binding to NMDA receptors (Fig. circles), and cerebellargranule cell layer (solid triangles) and agonist 7; seealso Table 1) revealed that the weak potency of HQ for activation(plus 10 PM glycine) of receptorcomplexes in Xenopusoocytes (C,D); NRl/NRZA (solid circles), NRl/NRZB (open circles), andNRl/ cerebellar NMDA receptorslabelled by L-3H-glutamatewas due NR2C(solid triangles). Dose-responsecurves were generated for agonist to a low-affinity HQ unique to the cerebellum. In competitors(NMDA and HQ) of L- JH-glutamatebinding (A, B) by each of the three brain regions examined, HQ inhibition curves computer-assistedimage analysis of autoradiograms.Dose-response could be resolved into two components with approximately curvesfor agonistactivation of heteromericNMDA receptorcomplexes equal proportions of high and low affinity binding sites (see (C, D) expressedin oocyteswere fit accordingto the equationI = I,,,/ [l + (EC,,/@], as describedin Materialsand Methods.Each point Table 1). While the high-affinity binding components in the representsthe meanrt SEM of threeto six oocytestaken from the same lateral thalamus, medial striatum, and cerebellum had similar frog. K, values (K, = 0.1,0.2, and 0.3 KM, respectively), the low-affinity

Table 1. Agonist and antagonist potencies against recombinant NMDA receptors expressed in Xerwpus oocytes and GH-glutamate binding to native NMDA receptors in brain

Quantitative autoradiography: Xenopus oocytes: averagek, (PM) -t SEM EC,, or IC,, (PM) & SEM Affinity Affinity ratio ratio LTH: CBG: NRl/ NRl/ NRl/ NRSA: NRZC: LTH MS CBG MS MS NR2A NR2B NR2C NR2B NR2B Agonists NMDA 3.9 + 0.22 2.8 + 0.14 4.10 k 0.21t 1.4 1.4 31.4 + 2.74$# 18.6 + 1.7 18.6+ 1.9 1.7 1.0 Homoquinolinatea 0.1 ? 0.09 0.2 AZ 0.1* 0.3 k O.l8*t 1.5 3.6 14.2-t 3.98 7.4 + 0.855 50.6 + 6.7 1.9 6.8 (54%) (55%) (50%) 5.1 k 3.14 3.3 ZIZ0.67 58 I? 9.61 (46%) (45%) (50%) Antagonists n-AL’5 3.1 + 0.31 8.1 + 0.6* 16 & 1.3*$ 0.38 2.0 0.25 + 0.028 0.44 * 0.126 1.5 k 0.1 0.6 3.4 n-CPPene” 0.12 + 0.02$ 0.19 2 0.03 0.25 + O.O6*t 0.62 14.9 0.09 + 0.025 0.17 k 0.05§ 1.6 -t 0.2 0.6 9.4 (48%) 14.3 + 3.6 (52%) a Statistical analysis and affinity ratio determination conducted on single-site K, values for both homoquinolinate and n-CPPene. * Significantly different from LTH, p < 0.05 (Scheffe F test). t Significantly different from MS, p i 0.05 (Scheffe F test). $ Significantly different from MS, p < 0.05 (Fisher PLSD). 4 Significantly different from NRl/NRZC, p < 0.05 (Scheffe F test). # Significantly different from NRIINRZB, p < 0.05 (Scheffe F test). The Journal of Neuroscience, September 1994, 14(9) 5481 component observed in the cerebellum (K, = 58 PM) differed significantly from the low-affinity component observed in the lateral thalamus (5.1 PM) and medial striatum (3.3 PM; see Table 1). Selective effects of HQ were also observed in Xenopus oocytes expressing heteromeric NR I-NR2 complexes. Relative to NR2A- and NR2B-containing receptor heteromers, NR l/NR2C receptors showed significantly lower affinity for HQ (Fig. 7, Table 1). Interestingly, although HQ was an agonist at all het- eromeric recombinant NMDA receptors expressed in Xenopus oocytes, it was less efficacious than NMDA at the NRl/NR2C Log [D-AP53. M Log [CPPene]. M receptor (data not shown). The Hill coefficients for HQ acti- vation of oocyte expressed NMDA receptors were 1.9 f 0.3 (NRl/NR2A), 1.1 -t 0.1 (NRl/NR2B), and 1.6 + 0.2 (NRl/ NR2C). In contrast to the relatively selective actions of HQ, NMDA did not discriminate between forebrain and cerebellar NMDA receptors labeled by L-3H-glutamate (Fig. 7). Likewise, NMDA potency at the NR l/NR2C complex was not markedly different from NMDA potency at NRl/NR2A or NRl/NR%B receptors. However, HQ and NMDA did display small but consistently higher affinities at medial striatal NMDA receptors than at lat- Log [CPPene]. M eral thalamic NMDA receptors. These differences were also ob- Figure 8. Comparisonof inhibition of L-3H-glutamatebinding by an- served in oocytes expressing heteromeric NMDA receptors; HQ tagonists(A, B) in lateralthalamus (solid circles), medialstriatum (open and NMDA displayed higher potencies at NR 1/NR2B receptors circles), and cerebellargranule cell layer (solid triangles) and antagonist as compared to NR 1/NR2A receptors. The Hill coefficients for inhibition of heteromericreceptor complexes in Xenopuroocytes (C, NMDA activation of oocyte-expressed NMDA receptors were D); NRl-NR2A (solid circles), NRI-NRZB (opencircles), and NRl- 1.3 + 0.1 (NRl/NR2A), 1.3 f 0.1 (NRl/NR2B), and 1.2 f NR2C (solid triangles).Antagonist inhibition curvesfor D-AR (A) and n-CPPene(B) inhibition of L-‘H-glutamatebinding were generated by 0.1 (NRUNR2C). computer-assistedimage analysis of autoradiograms.Inhibition curves Quantitative analysis of antagonist inhibition of L-3H-gluta- for antagonistinhibition of currentselicited by 100PM NMDA, in the mate binding to native NMDA receptor subtypes in lateral thal- presenceof 10 PM glycine,in oocytesexpressing heteromeric NMDA amus, medial striatum, and cerebellum is shown in Figure 8. receptorcomplexes were fit accordingto the equationI = I,,, - I,,,,,/ [ 1 + (IC,,/A)“], and correctedfor agonistaffinity, as describedin Ma- NMDA receptors in the lateral thalamus had the highest affin- terialsand Methods.Each point representsthe mean+ SEM of three ities for all antagonists tested, while receptors in the cerebellum to six oocytestaken from the samefrog. had the lowest antagonist affinities. Compared to D- AP5, n-CPPene displayed a greater selectivity between forebrain and cerebellar NMDA receptors. D-CPPene more potently inhibited with other, unique pharmacological properties (Beaton et al., forebrain NMDA receptors as compared to cerebellar receptors 1992). The resultsofthis report demonstratethat the anatomical (see Table 1). As was observed for HQ, D-CPPene inhibition of distributions of each of the four subtypesshow striking parallels L-3H-glutamate binding in the cerebellum was best fit by a two- to the distributions of individual NR2 subunit mRNAs and that site model with 53% of the binding sites having a low affinity the NR2 subunit is of critical importance in determining recep- for D-CPPene (see Table 1). D- CPPene also displayed greater tor subtype pharmacology. selectivity than D-AP~ for the inhibition of heteromeric NMDA The distribution of the antagonist-preferring subtype, as de- receptor complexes expressed in Xenopus oocytes (Fig. 8C,D). fined by high-affinity 3H-CPP binding, is virtually identical to D-CPPene more potently inhibited NR I-NR2A and NRl-NR2B the distribution of NR2A subunits. In all brain regions exam- heteromers as compared to NR l-NR2C receptors. The oocyte- ined, the presenceof NR2A mRNA predicted the presenceof expressed NR I-NR2A receptor was most sensitive to inhibition 3H-CPP binding sites. Conversely, all brain regions expressing by D-AP~ and D- CPPene, while the NRl-NR2C receptor was antagonist-preferring NMDA receptors also expressedNR2A the least sensitive. mRNA. Furthermore, there is a strong positive correlation (r = 0.88) between the density of 3H-CPP binding sitesand the den- Discussion sity of NR2A subunit mRNA. No other subunit, or combination Relationship between NMDA receptor subunits and agonist- of subunits, displayed such a strong anatomical correspondence. preferring/antagonist-preferring NMDA receptor subtypes The high correlation between high-affinity 3H-CPP binding Four distinct populations of NMDA receptors have been de- sites and NR2A mRNA suggeststhat 3H-CPP poorly labels scribed by radioligand binding and electrophysiological assays: NMDA receptors that do not have NR2A subunits. This hy- (1) agonist-preferring and (2) antagonist-preferring NMDA re- pothesis is consistent with the observation that radiolabeled ceptors (Monaghan et al., 1988; Honor& et al., 1989; Monaghan, antagonistsdo not label NMDA receptorsin brain regions that 199 1; Sakurai et al., 1993) (3) cerebellar NMDA receptors with contain high concentrations of NR2D subunits (e.g., olfactory distinctive pharmacological properties (Perkins and Stone, 1983; bulb glomerular layer, anteroventral nucleus, paraventricular Sekiguchi et al., 1990; Ebert et al., 199 1; Monaghan and Beaton, nucleus, intermediodorsal nucleus). Furthermore, HQ inhibits 1991; O’Shea et al., 1991; Yoneda and Ogita, 1991; Beaton et the binding of the antagonist 3H-CGP39653 in the cerebellum al., 1992) and (4) midline thalamus (MT-like) NMDA receptors (which contains NR2A and NR2C mRNAs) with only a single 5482 Buller et al. l Molecular Basis of NMDA Receptor Diversity

high-affinity component (Monaghan and Beaton, unpublished for agonists. Conversely, receptor complexes containing NR 1 +/ observations), suggesting that 3H-CGP39653 binding in the cer- NRZB subunits would display a reduced affinity for agonists. ebellum may be due to NR2A and not NR2C subunits. In the This hypothesis is consistent with recent findings that NMDA present study, 3H-CPP poorly labeled NMDA receptors in regions receptors composed of NRl + subunits display lower agonist that contain predominantly NR2B mRNA (e.g., striatum and affinity (Nakanishi et al., 1992; Durand et al., 1993; Hollmann lateral septum); however, in other studies (Monaghan, Beaton, et al., 1993) and higher antagonist affinity (Hollmann et al., and Larson, unpublished observations) we have found that the 1993) than NRl - receptors. antagonist 3H-CGP39653 binds with low affinity in regions con- The anatomical distribution of subunit mRNAs suggests that taining NR2B mRNA (e.g., striatum, septum). Taken together, NR2A subunits are associated with antagonist-preferring NMDA these findings suggest that the high-affinity antagonist binding receptors. Thus, for a given NR 1 subunit, NR2A subunits may site (Murphy et al., 1988; Porter et al., 1992; vanAmsterdam confer higher antagonist affinity to the NRl-NR2 complex, et al., 1992; Benke et al., 1993) represents NR2A-containing whereas NR2B subunits may confer higher agonist affinity. Us- NMDA receptors while the low-affinity radiolabeled antagonist ing the Xenopus oocyte expression system, we find that NR2A- binding site represents NMDA receptors that contain NR2B containing receptors consistently display a higher affinity for subunits. antagonists than the NR2B-containing receptor. In contrast, Based upon pharmacological experiments with recombinant NRl/NR2B receptors have a higher affinity for agonists that receptors expressed in Xenopus oocytes, it has been proposed was similar in magnitude to that reported for the mouse ho- that the presence of the NRl + subunit confers the antagonist- molog to the NR2B subunit (t2 compared to tl; Kutsuwada et preferring NMDA receptor subtype properties and the NRl- al., 1992; Yamazaki et al., 1992) and similar in magnitude to subunit is responsible for agonist-preferring subtype (Hollmann that seen in radioligand binding studies for agonist-preferring et al., 1993). The anatomical evidence presented in this report NMDA receptors (compared to antagonist-preferring receptors). suggests that in many brain regions, high-affinity 3H-CPP bind- Thus, the anatomical and physiological evidence is consistent ing sites are coexpressed with NRl+ mRNA (r = 0.59). For with the hypothesis that agonist-preferring and antagonist-pre- example both markers have similar patterns of cortical varia- ferring NMDA receptors are largely determined by NR2B and tion, and both are present in high levels in the thalamus and NR2A subunits. inferior colliculus and low levels in the striatum and septum. Alternative splicing at the second and third NR 1 cassette sites However, there are several differences in the distributions of (deletion 1 and deletion 2, Sugihari et al., 1991) are thought to 3H-CPP binding sites and NRl + mRNA. For example, the have less of an effect upon NMDA receptor pharmacology, since anteroventral nucleus displays high levels of NRl + mRNA and homomeric receptors consisting of subunits alternatively spliced yet displays very low levels of 3H-CPP binding sites (and NR2A at the second or third cassettes do not have significant differences mRNA). Conversely, CA 1 hippocampus has relatively high lev- in NMDA receptor pharmacological properties (Nakanishi et els of 3H-CPP binding sites (and NR2A mRNA) and low levels al., 1992; Durand et al., 1993; Hollman et al., 1993). Consistent of NRl + mRNA. Furthermore, the distinctive patterns of vari- with these findings, we find that mRNAs alternatively spliced ation in 3H-CPP binding sites within the thalamus, septum, and at the second and third cassettes have distribution patterns dis- cortical layers correspond well to NR2A subunit mRNA and similar to that ofapparent NMDA receptor subtypes (Monaghan not to NRl+ mRNA. and Buller, 1994; Monaghan, Schneider, and Larson, unpub- Agonist-preferring NMDA receptors (as defined by L-3H-glu- lished observations). However, it will be important to determine tamate binding that is weakly inhibited by D- AP5 and potently if these alternatively spliced forms can alter the pharmacological inhibited by HQ) are found predominately in the subset of brain properties of heteromeric receptor complexes. regions that contain both NR2B and NRl - mRNA. In many brain regions NR2B and NR 1-insert 1 mRNA are colocalized (r = 0.59). For example, both subunit transcripts show the same Relationship between NRZCINR2D subunit mRNA and detailed pattern of cortical variation and cortical lamination cerebellar/MT-like NMDA receptors and are present in high levels in the striatum, septum, hippo- NR2C subunits display an obvious anatomical correspondence campus, and dorsal medial inferior colliculus, and in low levels to the pharmacologically atypical subtype of NMDA receptor throughout most of the midbrain and brainstem. This pattern found in the cerebellum. Since NR2C mRNA is the only known is similar to agonist-preferring NMDA receptors and dissimilar NMDA receptor subunit that has a distribution that can account to other NMDA receptor subtypes. for the unique pharmacology of the cerebellum, it is significant In regions where there are low densities of either NR2B or that coexpression of the NR2C mRNA with NRl mRNA in NRl - mRNA, there is a low density of agonist-preferring oocytes results in a receptor complex that has pharmacological NMDA receptors as well. Overall, regions that have agonist- similarities to the native cerebellar receptor. Thus, both the preferring NMDA receptors have significant levels of both NR2B anatomical and pharmacological data suggest that the NR2C and NRl - mRNA. Thus, these data suggest that agonist-pre- subunit is responsible for the distinct pharmacological proper- ferring receptors may result from the combination of NR2B and ties of the unique cerebellar NMDA receptor subtype. NRl - subunits. This would account for the low levels of ag- Unexpectedly, the pineal gland expresses NR2C mRNA but onist-preferring NMDA receptors in the thalamus and CA3 not NMDA-sensitive L-3H-glutamate binding sites. The lack of hippocampus (regions where NR 1 + mRNA is expressed at high L- 3H-glutamate binding sites suggests that NR2C mRNA may levels) and would account for the lower density of agonist-pre- not be translated in the pineal. Alternatively, the absence of a ferring binding sites in the dorsal lateral septum, brainstem, and glutamate binding site may be related to our observation that cerebellum (regions where NR2B is expressed at low levels). neither NRl + nor NRl - mRNA (nor NR2A, NR2B, NR2D) Thus, by this hypothesis, when NR2B subunits form a complex was detected in the pineal (Schneider, Larson, and Monaghan, with NR 1 - subunits, the resulting receptor has a high affinity unpublished observations). Thus, the lack of L-‘H-glutamate The Journal of Neuroscience, September 1994, f4(9) 5483

binding may be due to the absence of NRl subunits or the native and recombinant NMDA receptors described in this re- presence of a novel NRl splice variant. port suggest that there are at least four discrete populations of In radioligand binding studies, HQ displays two affinities in NMDA receptors of differing molecular composition that are the cerebellum and it is the low-affinity site that appears to heterogeneous with regard to the glutamate binding site. It has correspond to the NR2C subunit in its affinity for HQ and been suggested that the NR2 subunits are responsible for het- unique cerebellar localization. While it is unclear what the high- erogeneity observed for the binding site on the NMDA affinity HQ site in the cerebellum represents, perhaps NR2A- receptor (Williams et al., 1993). Whether the NR2 subunit also containing NMDA receptors contribute to this component. contributes to NMDA receptor heterogeneity observed at the Likewise, the significance of separate high- and low-affinity HQ glycine binding domain (O’Shea et al., 199 l), the PCP channel binding sites in other brain regions is unknown. blocking site (Yignon et al., 1986; Ebert et al., 1991; Beaton et The anatomical distribution of the MT-like subpopulation of al., 1992), and the site (Reynolds and Palmer, 199 1; NMDA receptors is virtually identical to the distribution of Yoneda and Ogita, 1991) remains to be determined. transcripts encoding the NR2D subunit. Both display a distri- bution that is highly restricted to select diencephalic structures References (intermediodorsal, paraventricular and paratenial, rhombiod, Anand R, Conroy WG, Schoepfer R, Whiting P, Lindstrom J (1991) mediodorsal lateral, and anteroventral thalamic nuclei, the me- Neuronal nicotinic acetylcholine receptors expressed in Xenogus oo- dial geniculate, the nucleus of the diagonal band, the interpe- cytes have a pentameric quaternary structure. J Biol Chem 266:11192- duncular nucleus, and the periaqueductal gray). In contrast, no 11198. other subunit, or combination of subunits, displays a similar Ananthraram V, Panchal RG, Wilson A, Kolchine VV, Treistman SN, Bayley H (1992) Combinatorial RNA splicing alters the surface pattern. These results strongly suggest that the distinctive phar- charge on the NMDA receptor. FEBS Lett 305:27-30. tiacological properties found for this NMDA receptor subtype Beaton JA, Stemsrud K, Monaghan DT (1992) Identification ofa novel (Beaton et al., 1992) are determined by the NR2D subunit. The N-methyl-D-aspartate receptor population in the rat medial thalamus. observation that the NR2D and NR2C subunits show greater J Neurochem 59:754-757. Benke D, Marti T, Heckendom R, Rehm H, Kiinzi R, Allgeier H, Angst homology to each other than to either NR2A or NR2B subunits C, Mohler H (1993) Photoaffinitv labeling of the NMDA recentor. (Ikeda et al., 1992; Ishii et al., 1993) predicts greater structural E& J Pharmacol246: 179-l 80. - - similarities between the cerebellar and MT-like NMDA receptor Buller AL, White MM (1988) Control of Torpedo acetylcholine re- subtypes. Indeed, the observation that the cerebellar and MT- ceptor biosynthesis in Xenopus oocytes. Proc Nat1 Acad Sci USA 85: 8717-8721. like NMDA receptor populations are pharmacologically similar Cheng YC, Prusoff WH (1973) Relationship between the inhibition (both populations have lower affinities for quinolinate, homo- constant (K,) and the concentration of inhibitor which causes 50 per- quinolinate, CPP, and D-CPPene than other NMDA receptors) cent inhibition (IC,,) of an enzymatic reaction. Biochem Pharmacol is consistent with this prediction. 75:3099-3108. -- Collingridge GL, Lester RAJ (1989) Excitatory amino acids receptors General conclusions in the vertebrate central . Pharmacol Rev 4 1:143-2 10. Cooper E, Couturier S, Ballivet M - (1992) Pentameric structure and In this report, we have demonstrated that the distribution of subunit stoichiometry of a neuronal nicotinic . each of the NMDA receptor subtypes parallels the distribution Nature 350:235-238. of a specific NMDA receptor subunit (or combination of sub- Cotman CW, Bridges RJ, Taube JS, Clark AS, Monaghan DT (1989) The role of the NMDA receptor in central nervous system plasticity units). Furthermore, we have demonstrated that recombinant and pathology. J NIH Res 1:65-74. NMDA receptor subunits form heteromeric receptor complexes Cunningham MD, Michaelis EK (1990) Solubilization and partial whose pharmacological behavior can be predicted from the study purification of 3-(( +)-2-carboxypiperizine-4-yl)-[ 1 ,2-3H]propyl-l - of native NMDA receptors. If one assumes a multimeric struc- phosphonic acid recognition from rat brain synaptic mem- ture for the mature NMDA receptor composed of at least one branes. J Biol Chem 265~7768-7778. Durand GM, Gregor P, Zheng X, Bennett ML, Uhl GR, Zukin RS member from each subunit family, as is the case for the neuronal (1992) Cloning of an apparent splice variant of the rat N-methyl-D- nicotinic acetylcholine receptor (Anand et al., 199 1; Cooper et aspartate receptor NMDARl with altered sensitivity to al., 1992), a vast number of heteromeric complexes are possible. and activators kinase C. Proc Nat1 Acad Sci USA 89:9359- Thus, it is intriguing that the simple subunit combinations re- 9363. ported here can mimic the pharmacological properties of iden- Durand GM, Bennett MVL, Zukin RS (1993) Splice variants of the IV- methyl-D-aspartate receptor NRl identify domains involved in tified native NMDA receptor subtypes. That the NR2 subunits the regulation by polyamines and C. Proc Nat1 Acad are predominantly responsible for the pharmacologically dis- Sci USA 90:6731-6735. tinct NMDA receptor subtypes is perhaps unexpected, since Ebert B, Wong EHF, Krogsgaard-Larson P (199 1) Identification of a NRl subunits (Moriyoshi et al., 199 1; Monyer et al., 1992; Ishii novel NMDA receptor in rat cerebellum. Eur J Pharmacol [Mole Pharmacol] 208:49-52. et al., 1993) but not NR2 subunits form functional homo-oligo- Elhert FJ, Tran LLP (1990) Regional distribution of M 1, M2 and non- mers and thus NRl subunits appear to have agonist and an- M 1, non-M2 muscarinic binding sites in rat brain. J Pharmacol Exp tagonist recognition sites (assuming the absence of endogenous Ther 255:1148-1156. NR2 subunits in oocytes). Greenamyre JT, Young AB (1989) Excitatory amino acids and Alz- These data also suggest that different NMDA receptor radi- heimer’s disease. Neurobiol Aging 10:593-602. Hollmann M, Boulter J, Maron C, Beasley L, Sullivan J, Pecht G, oligands should selectively label different NMDA receptor pop- Heinemann S (1993) potentiates agonist-induced currents at ulations. The above hypotheses thus provide specific interpre- certain splice variants of the NMDA receptor. 10:943-954. tations of the various reports where disparate experimental results Honor6 T, Drejer J, Nielsen FO, Watkins JC, Olverman HJ, Nielsen are found when differing NMDA receptor radioligands are used M (1989) Molecular target size of the NMDA-receptor complex in rat cortex. Eur J Pharmacol 172:239-247. (e.g., Honor6 et al., 1989; Ogita et al., 1991; Savage et al., 1991; Ikeda K. Nagasawa M. Mori H. Araki K. Sakimura K. Watanabe M. Kalb et al., 1992; Wieland, 1992). Inoue Y, Gishina M ’ ( 1992) Cloning anclexpression df the 64 subunit The parallel anatomical and pharmacological properties of of the NMDA receptor channel. FEBS L&t 3 13:34-38. 5484 Buller et al. * Molecular Basis of NMDA Receptor Diversity

Ishii T, Moriyoshi K, Sugihara H, Sakurada K, Kadotani H, Yokoi M, Perkins MN, Stone TW (1983) : regional variations in Akazawa C, Shigemoto R, Mizuno N, Masu M, Nakanishi S (1993) neuronal sensitivity. Brain Res 259: 172-l 76. Molecular characterization of the family of the N-methyl-D-aspartate Porter RHP, Cowbum RF, Alasuzoff I, Brings RSJ, Roberts PJ (1992) recentor subunits. J Biol Chem 268:2836-2843. Heterogeneity of N-methyl-D-aspartate (NMDA) receptors labeled Kalb RG, Lidow MS, Halsted MJ, Hockfield S (1992) N-methyl-o- with [3H]-3-((+)2-carboxypiperazin-4-yl)propyl-l-phosphonic acid aspartate receptors are transiently expressed in the developing spinal (CPP): receptor status in Alzheimer’s disease. Eur J Pharmacol225: cord ventral horn. Proc Nat1 Acad Sci USA 89:8502-8506. 195-20 1. Kutsuwada T, Kashiwabuchi N, Mori H, Sakimura K, Kushiya E, Araki Reynolds IJ, Palmer AM (1991) Regional variations in [3H]MK801 K, Meguro H, Masaki H, Kumanishi T, Arakawa M, Mishina M binding to rat brain N-methyl-D-aspartate receptors. J Neurochem (1992) Molecular diversity of the NMDA receptor channel. Nature 56:1731-1740. 358:36-41. Sakurai SY, Penney JB, Young AB (1993) Regionally distinct N- meth- Leonard JP, Kelso SR (1990) Apparent desensitization of NMDA yl-D-aspartate receptors distinguished by quantitative autoradiogra- responses in Xenopus oocytes involves calcium-dependent chloride phy of [3H]MK80 1 binding in rat brain. J Neurochem 60: 1344-l 353. current. Neuron 2:52-60. Savage DD, Ortiz KA, Sanchez CF, Paxton LL (199 1) Prenatal ethanol Mayer ML, Westbrook GL (1987) The of excitatory amino exposure increases hippocampal NMDA binding acids in the central nervous system. Prog Neurobiol 28: 197-276. in 45-day-old rats. Clin Exp Res 15:339. Meeuro H. Mori H. Araki K. Kushiva E. Kutsuwada T. Yamazaki M. Siegel RE (1989) Localization of neuronal mRNAs by hybridization I&ma&hi T, Arakawa M; Sakimura T, Mishina M (1992) Fun& histochemistry. In: Gene probes (Conn, PM, ed), pp 136-150. San tional characterization of a heteromeric NMDA receptor channel ex- Diego: Academic. pressed from cloned cDNAs. Nature 357:70-74. Sekiguchi M, Okamoto K, Sakai Y ( 1990) Glycine-insensitive NMDA- Meldrum B, Garthwaite J (1990) Excitatory amino acid sensitive receptor expressed in Xenopus oocytes by guinea pig mRNA. and neurodegenerative disease. Trends Pharmacol Sci 11:379-387. J Neurosci 10:2148-2155. Monaghan DT (199 1) Differential stimulation of [‘H]MKSO 1 binding Standaert DG, Testa CM, Penney JB, Young AB (1993) Alternatively to subpopulations of NMDA receptors. Neurosci Lett 122:2 l-24. spliced isoforms of the NMDARl subunit: dif- Monaghan DT, Beaton JA (199 1) Quinolinate differentiates between ferential expression in the basal ganglia of the rat. Neurosci Lett 152: forebrain and cerebellar NMDA receptors. Eur J Pharmacol 194: 123- 161-164. 125. Sugihara H, Moriyoshi K, Ishii T, Masu M, Nakanishi S (1992) Struc- Monaghan DT, Buller AL (1994) Anatomical, pharmacological, and ture and properties of seven isoforms of the NMDA receptor gen- molecular diversity of native NMDA receptor subtypes. In: The erated by alternative splicing. Biochem Biophys Res Commun 185:

NMDA recentor. 2d ed (Watkins JC. CollinaridaeI- GL. eds).I New 826-832. York: Oxford UP. ’ vanAmsterdam FTM, Giberti A, Mugnaini M, Ratti E (1992) 3-[( +)- Monaghan DT, Olverman HJ, Nguyen L, Watkins JC, Cotman CW 2-Carboxypiperazin-4-yllpropyl- 1-phosphonic acid recognizes two N- (1988) Two classes of N-methyl-n-aspartate recognition sites: dif- methyl-D-aspartate binding sites in rat cerebral cortex membranes. J ferential distribution and differential regulation by glycine. Proc Nat1 Neurochem 59:1850-1855. Acad Sci USA 85:9836-9840. Vignon J, Privat A, Chaudieu I, Thierry A, Kamenka JM, Chichepor- Monaghan DT, Bridges RJ, Cotman CW (1989) The excitatory amino tiche R (1986) [‘Hlthienyl- ([‘HITCP) binds to two acid receptors: their classes, pharmacology and distinct properties in different sites in rat brain. Localization by autoradiography and bio- the central nervous system. Annu Rev Pharmacol Toxic01 29:365- chemical techniques. Brain Res 378: 133-14 1. 402. Watanabe M. Inoue Y. Sakimura K. Mishina M (1992) Develon- Monyer H, Sprengel R, Schoepfer R, Herb A, Higuchi M, Lomeli H, mental changes in distribution of NMDA receptor channel subunit Bumashev N, Sakmann B, Seeburg PH (1992) Heteromeric NMDA mRNAs. Neuroreport 3: 1138-l 140. receptors: molecular and functional distinction of subtypes. Nature Weiland NG (1992) Estradiol selectively regulates agonist binding sites 256:1217-1221. on the N-methyl-D-aspartate receptor complex in the CA1 region of Moriyoshi K, Masu M, Ishii T, Shigemoto R, Mizuno N, Nakanishi S the hippocampus. Endocrinology 131:662-668. (199 1) Molecular cloning and characterization of the rat NMDA White MM, Mayne KM, Lester HA, Davidson N (1985) Mouse- receptor. Nature 354:31-37. Torpedo hybrid acetylcholine receptors: functional homology does Murphy DE, Hutchinson AJ, Hurt SD, Williams M, Sills MS (1988) not equal . Proc Nat1 Acad Sci USA 82:4852- Characterization ofthe binding of [)H]-CGS 19755: a novel N-methyl- 4856. n-aspartate antagonist with nanomolar affinity in rat brain. Br J Phar- Williams K, Russell SL, Shen YM, Molinoff PB (1993) Developmental macol 95:932-938. switch in the expression of NMDA receptors occurs in vivo and in Nakanishi N, Axe1 R, Schneider NA (1992) Alternative splicing gen- vitro. Neuron 101267-278. erates functional1 distinct N-methyl-n-aspartate receptors. Proc Nat1 Wisden W, Morris BJ, Hunt SP (1991) In situ hybridization with Acad Sci USA 89:8552-8526. synthestic DNA probes. In: Molecular neurobiology: a practical ap- Ogita K, Kouda T, Suzuki T, Ohkawara A, Enomoto R, Nabeshima T, proach (Chad J, Wheal H, eds), pp 205-225. New York: Oxford UP. Yoneda Y (1991) Differential effects of SH-reactive agents on Yamazaki M, Mori H, Araki K, Mori K, Mishina M (1992) Cloning, [3H](-c)3-(2-carboxypiperazin-4-yl)propyl-l-phosphonate and expression and modulation of a mouse NMDA receptor subunit. [-‘HIglutamate binding in brain synaptic membranes treated with Tri- FEBS Lett 300:39-45. ton X- 100. Neurochem Int 18:55-62. Yoneda Y, Ogita K (199 1) Heterogeneity of the N-methyl-o-aspartate O’Shea RD, Manallack DT, Conway EL, Mercer LD, Beart PM (199 1) receptor ionophore complex in rat brain, as revealed by binding Evidence for heterogeneous glycine domains but conserved multiple techniques. J Pharmacol Exp Ther 259:86-96. states of the excitatory amino acid recognition site of the NMDA receptor: regional binding studies with [)H]glycine and [3H]glutamate. Exp Brain Res 86:652-662.