Distinct N-methyl-D-aspartate receptor 2B subunit gene sequences confer neural and developmental specific expression.

Expression of the N-methyl--aspartate (NMDA) receptor 2B (NR2B) subunit is neural-specific and differentially regulated. It is expressed in the forebrain and in cerebellar granule cells at early postnatal stages and selectively repressed in the cerebellum after the second postnatal week, where it is replaced by the NR2C subunit. This switch confers distinct properties to the receptor. In order to understand the molecular mechanisms that differentially regulate the NR2B gene in the forebrain and cerebellum during development, we have isolated and characterized the promoter region of the NR2B gene. Two 5' noncoding exons and multiple transcription start sites were identified. Transcriptional analysis in transgenic mice reveals that an upstream 800-base pair region, which includes the first exon, is sufficient to direct neural-specific transcription. Developmental repression of the gene in the cerebellum requires additional regulatory elements residing in the first intron or second exon. Sequence elements that may participate in the regulation of the NR2B gene were identified by comparison to other neural genes. These studies provide insight into the molecular mechanisms regulating the switch of NMDA receptor subunit expression in the cerebellum, which ultimately account for the physiological changes in receptor function during development.

Expression of the N-methyl-D-aspartate (NMDA) receptor 2B (NR2B) subunit is neural-specific and differentially regulated. It is expressed in the forebrain and in cerebellar granule cells at early postnatal stages and selectively repressed in the cerebellum after the second postnatal week, where it is replaced by the NR2C subunit. This switch confers distinct properties to the receptor. In order to understand the molecular mechanisms that differentially regulate the NR2B gene in the forebrain and cerebellum during development, we have isolated and characterized the promoter region of the NR2B gene. Two 5 noncoding exons and multiple transcription start sites were identified. Transcriptional analysis in transgenic mice reveals that an upstream 800-base pair region, which includes the first exon, is sufficient to direct neural-specific transcription. Developmental repression of the gene in the cerebellum requires additional regulatory elements residing in the first intron or second exon. Sequence elements that may participate in the regulation of the NR2B gene were identified by comparison to other neural genes. These studies provide insight into the molecular mechanisms regulating the switch of NMDA receptor subunit expression in the cerebellum, which ultimately account for the physiological changes in receptor function during development.
Refinement of synaptic connections and patterning of the visual system, synaptic plasticity in the hippocampus (see Ref. 1), and cerebellar granule cell migration and differentiation (2,3) are regulated by activation of N-methyl-D-aspartate (NMDA) 1 receptors. NMDA receptors have the unique property of functioning as coincident detectors for correlated activity, which is an important characteristic necessary for the refinement of connections during development and the activity-dependent potentiation or depression of synaptic inputs.
The cloning of cDNAs coding for NMDA receptor subunits has provided molecular evidence for receptor heterogeneity and differential control of subunit composition in selective neural populations during development (Refs. 4 -8; for review see Ref. 9). In most cases, it is thought that these multimeric receptors are heterooligomers composed of a widely expressed NR1 (or ) subunit combined with at least one of four NR2 subunits designated A-D (also known as ⑀1-4, respectively), whose expression patterns are regionally and developmentally restricted. The regulation of NMDA receptor channel kinetic properties, by controlling either receptor subunit composition or posttranslational modifications, could have important consequences on the signals transduced by these receptors. The pharmacological and electrophysiological properties of NMDA receptors vary with respect to the composition of different NR2 subunits (5, 7, 10 -14) and NR1 splice variants assembled (15). The most prominent change in NR2 expression is seen with the NR2B subunit, which is widely expressed in the forebrain and cerebellum up to 2 weeks after birth but thereafter is selectively repressed in the cerebellum. Using in situ hybridization histochemistry, it has been shown that during the first 2 weeks after birth premigratory cerebellar granule cells in the external granule layer express NR2B transcripts. However, expression of NR2B mRNAs is repressed to undetectable levels after the cells migrate into the internal granule cell layer, where expression of NR2C is activated (7,16). These changes in NR2 subunit expression are probably responsible for the differences in NMDA receptor properties seen in granule cells during cerebellar development (17). The repression of NR2B expression, which occurs after granule cells have migrated into the internal granule cell layer, where they receive afferent inputs, may result from synaptic activity (18,19). Thus, different NMDA receptor subunits may not only function to distinctly modulate synaptic connections in response to activity, but their expression patterns may also be responsive to synaptic activity. Changes in receptor properties are of the utmost importance, since they may serve to differentially modify characteristics of neural function during a developmental period in which synaptic connections are being refined.
In order to begin understanding the complex mechanisms that direct the tissue-and region-specific expression of different NMDA receptor subunits in the brain during neurogenesis, and that modulate their levels in response to epigenetic factors, we have begun to investigate the mechanisms that regulate expression of the NR2B gene. We chose to focus on the NR2B gene because its differential expression during development is the most prominent among NMDA receptor subunits, and there is strong evidence that its expression in the cerebellum is selectively down-regulated by activity (18,19). We have isolated and characterized the 5Ј end of the gene, and found that NR2B transcription is initiated from multiple sites. However, differential usage of transcription initiation sites cannot account for its selective down-regulation in the cerebellum. Analysis of regulatory regions in transgenic mice showed that dis-tinct DNA regulatory sequences are required for the neuralspecific and developmental down-regulation of the gene in cerebellum. The NR1 promoter has been characterized in vitro (20); the present study constitutes the first report analyzing the DNA regulatory sequences directing transcription of glutamate receptor subunit genes in neural cells during development.

EXPERIMENTAL PROCEDURES
Northern Analysis of NR2B Expression-RNA was prepared from 129/J (Jackson Laboratories) mouse tissues using cesium chloride gradients as described (21) and quantitated spectrophotometrically, and the integrity and relative amounts in each sample were checked by ethidium bromide staining of ribosomal RNA on agarose gels. The RNA was fractionated on 1.5% agarose gels containing 2.2 M formaldehyde and transferred to Nytran membranes (Schleicher & Schuell) by electroblotting. Blots were hybridized with a random prime-labeled 1.3-kb AccI-NarI fragment from the 5Ј end of the rat NR2B cDNA (22). Blots were prehybridized and hybridized at 65°C in Church's buffer (0.5 M sodium phosphate (pH 7.2), 7% SDS, 1% bovine serum albumin, 1 mM EDTA) and 100 g/ml salmon sperm DNA and then washed with 0.1 ϫ SSC, 1% SDS at 65°C. Quantitation was performed on a Molecular Dynamics PhosphorImager system (Sunnyvale, CA).
Cloning of the 5Ј End of the Mouse NR2B Transcript-The 5Ј end of the NR2B cDNA was amplified by rapid amplification of cDNA ends (RACE). Poly(A) ϩ RNA was prepared from 1-week-old C57BL/6 mouse brains using a CsCl gradient as described (21). The oligonucleotide 5Ј-TGGGGGCGCTCTTTTGGG-3Ј (this is antisense to ϩ93 to ϩ102 relative to the ATG) was used to prime cDNA synthesis off 1.0 g of poly(A) ϩ RNA using SuperScript reverse transcriptase (Life Technologies, Inc.). The RNA was degraded using RNase H, and the cDNA was purified using GlassMAX spin cartridges and cDNA tailed with terminal transferase (Life Technologies, Inc.). The dC-tailed cDNA was used as template for 35 cycles of polymerase chain reaction with an annealing temperature of 56°C using the oligonucleotide 5Ј-CAUCAUCAU-CAUCCAACACCAACCAGAACCTTGGG-3Ј (this is antisense to ϩ27 to ϩ49 relative to the ATG) and the anchor primer 5Ј-CUACUACUAC-UAGGCCACGCGTCGACTAGTACGGGIIGGGIIGGGIIG-3Ј (Life Technologies, Inc.). Polymerase chain reaction products were treated with 1 unit of uracil DNA glycosylase and annealed to the pAMP-1 vector at 37°C for 30 min. The resulting colonies were size-selected, and the cDNAs were sequenced with dideoxy nucleotides and Sequenase 2.0 (U.S. Biochemical Corp.).
Isolation of Genomic Clones-The largest cDNA fragment isolated above (784 bp), corresponding to the 5Ј-noncoding region of the NR2B transcript, was labeled to ϳ10 9 cpm/g by random priming (Prime-It, Stratagene) and used to screen a 129/SV mouse genomic library (Stratagene). Hybridization was performed in 40% formamide, 6 ϫ SSC, 5 ϫ Denhardt's, 10 mM EDTA, 100 g/ml denatured salmon sperm DNA, and 1% SDS at 42°C overnight. The hybridized membranes were washed at moderate stringency (0.5 ϫ SSC, 0.5% SDS at 42°C). Two overlapping genomic clones were identified from approximately 10 6 clones screened. clones were restriction-mapped using an oligonucleotide mapping to the 5Ј end of the cDNA. An 8-kb SacI-SacI fragment including 3 kb upstream of the cDNA was subcloned into pBluescript KS. Part of this clone was sequenced as above. Geneworks software (Intelligenetics) was used to analyze the DNA sequence.
RNase Protection Assay-RNase protection was performed on forebrain and cerebellar RNA isolated from mice of different ages to map the NR2B gene transcription initiation sites in these 2 brain regions. Because the results of RACE cloning suggested that NR2B gene transcription initiates at multiple sites, two probes were used to identify transcription initiation sites. Antisense RNA probes were made by subcloning genomic (probe 1) or cDNA (probe 2) fragments (see Fig. 3A) into the Bluescript pSK vector (Stratagene). Protection probe 1 was made from genomic DNA spanning bases Ϫ64 to ϩ147 as numbered in Fig. 2. Probe 2 includes cDNA covering bases ϩ74 to ϩ1324. Subcloning was done by amplifying the region of interest by polymerase chain reaction with primers containing restriction sites. Resulting clones were sequenced prior to use as templates for protection assays. Probes were transcribed with T7 RNA polymerase and [␣-32 P]UTP (800 Ci/ mmol, DuPont NEN) and produced products of the expected size with no evidence of premature termination of transcription.
RNA was isolated from 129/SV mouse tissues using a CsCl step gradient as described previously (21). The conditions for RNase protection assays, using the RPA II kit (Ambion), were initially optimized as recommended by the supplier by varying the amount of RNA and the concentration of RNase used. Unlabeled sense riboprobes mixed with RNA from liver were used as a positive control, and liver RNA alone was used as a negative control. For the assays we used 10 g of RNA, 1 unit of RNase A, and 40 units of RNase T1. The multiple bands seen using NR2B-expressing tissue were specific because there was no evidence for self-protection of probe (which results from RNA secondary structure) in the control experiments. Products were run on a 6% polyacrylamide/ urea sequencing gel with DNA sequencing ladders as molecular weight markers. Quantitation was performed on a Molecular Dynamics Phos-phorImager system. Major bands were defined as those containing at least 5% of the total signal after the relative intensity, size, and base composition of the bands were taken into account.
Production of Transgenic Mouse Lines-The construct 0.8-CAT is a 806-bp region extending from the HindIII to the SacII site, which contains the first exon and upstream sequence. 0.8int-CAT harbors a fragment between the same HindIII site and the end of the second exon (see Fig. 2). The region Ϫ552 to ϩ255 (0.8-CAT) and the region Ϫ552 to ϩ1631 (0.8int-CAT) were cloned into the HindIII-XbaI sites upstream of the CAT gene in the pCAT-Basic vector (Promega). For injection, fragments containing the NR2B gene sequence linked to the CAT reporter gene followed by the SV40 t intron and poly(A) ϩ addition site were cut out with HindIII and BamHI. These fragments were purified by TAE gel electrophoresis, electroeluted, purified with Elutip-D columns (Schleicher & Schuell), ethanol-precipitated, and resuspended in injection buffer (23). Founder transgenic mice were produced by DNX, The National Transgenic Development Facility. Founders were (C57BL/6 ϫ SJL)F2; these were mated to strain C57BL/6. Three CATexpressing lines (6184, 6188, and 6190) for the 0.8-CAT construct and two CAT-expressing lines (6944 and 6948) for the 0.8int-CAT construct were characterized. F1 animals were tested for incorporation of the transgene by slot blotting DNA prepared from tail clips. Slot blots were also used to determine transgene copy number.
Reporter Analysis of Transgenic Mice-For reporter gene assays, animals were sacrificed at the age indicated with CO 2 and subsequent decapitation. Tissues were sonicated in 0.25 M Tris (pH 7.5), 1 mM phenylmethylsulfonyl fluoride, centrifuged in a microcentrifuge for 5 min at 4°C, and the supernatants were heated for 10 min at 65°C to denature endogenous acetylases and centrifuged for 2 min. Final supernatants were assayed for protein content (BCA assay, Pierce) and then CAT activity (21). CAT assays on 30 g of tissue were performed in 0.625 M Tris (pH 8), 0.5 mg/ml acetyl coenzyme A, and 7.7 Ci/ml [ 14 C]chloramphenicol in a final volume of 130 l for a total of 4 h at 37°C with the addition of 10 l of acetyl-CoA (3.5 mg/ml) after 2 h of incubation. The reaction was extracted with ethyl acetate and separated by thin layer chromatography in 5% methanol, 95% chloroform. Quantitation was performed on a Molecular Dynamics PhosphorImager system. A time course was done to show that these conditions are in the linear range of the assay. Data are expressed as percentage of CAT conversion (as measured by PhosphorImager) per mg of protein per hour; background levels (average 0.05 Ϯ 0.01%) from parallel samples obtained from nontransgenic littermates were subtracted.

RESULTS
Northern Analysis of NR2B Expression-Changes in the levels of NR2B transcripts during development were analyzed by Northern blotting. For the analysis, RNA was isolated from the forebrain and cerebellum of P7, P14, and adult (approximately 2-month-old) mice. Hybridization with a NR2B probe detected a single band of about 15 kb, as has been previously shown (8). As shown in Fig. 1, the NR2B mRNA levels are higher in the forebrain than cerebellum at all of the ages studied. Quantification of the signals obtained with the forebrain samples demonstrated that NR2B mRNA levels are highest at P7 and drop to approximately 70% of maximum in the adult (Fig. 1B). Levels in the P7 cerebellum are approximately 30% of that found in the forebrain and become undetectable between P14 and the adult. These results are in accordance with previous reports, where the levels of NMDA receptor transcripts were analyzed using in situ hybridization histochemistry (7,16), RNase protection (24), Northern blotting (8), and reverse transcription-polymerase chain reaction (18). Using these four different methods, NR2B mRNA levels in the cerebellum were found to be repressed to undetectable levels during development in the cerebellum, while expression was maintained in the forebrain. This is consistent with levels of protein as measured by immunoblot (25). Thus, we were interested in identifying the molecular mechanism(s) that may account for the developmental regulation of the NR2B gene.
Identification and Cloning of the NR2B Gene Upstream Region-To enable the cloning of the 5Ј-most sequences of the mouse NR2B gene, we began by obtaining cDNAs corresponding to the full-length 5Ј-noncoding region using RACE. Multiple NR2B cDNA fragments of similar sizes were obtained. The longest cDNA extended 398 bp further in the 5Ј direction than had been previously reported (22). This cDNA (total of 784 bp) was used to screen a mouse 129/SVJ genomic library for clones containing the promoter region of the NR2B gene.
Two distinct overlapping clones were obtained from screening the genomic library and were used for characterization of the upstream region of the NR2B gene. A SacI genomic fragment that hybridized to an oligonucleotide probe made to the 5Ј end of the longest RACE clone was subcloned and partially sequenced (Fig. 2). As shown in the diagram in Fig. 2A, alignment of genomic and cDNA sequences revealed that there are two 5Ј-noncoding exons in the mouse NR2B gene; this result was corroborated by RNase protection experiments. The sequences at the splice junctions match to consensus intron donor/acceptor sites (26). The 2200-bp nucleotide sequence of the NR2B fragment and the location of the intron/exon boundaries are presented in Fig. 2B.
Multiple Transcription Initiation Sites Are Used by the NR2B Gene-The sites of transcription initiation from the NR2B gene were mapped by RNase protection. Initially, the assays were performed using RNA isolated from the forebrains of 129/SVJ adult mice and a cRNA probe derived from genomic DNA that encompasses the 5Ј-most RACE terminus (Fig. 3A, probe 1). In four independent experiments we obtained the same pattern of multiple RNase-protected bands as shown in Fig. 3B. The arrows point to the four major protected bands defined by their relative intensity, size, and base composition (see "Experimental Procedures"). A small amount of fully protected probe 1 was also observed (less than 5%), indicating the presence of minor start site(s) that originate further upstream of probe 1. The results of protection assays using probe 2 (Fig. 3A) ruled out the possibility of major start sites more 3Ј than the group already identified. The multiple bands obtained in the RNase protection assays are specific because samples containing liver (not shown) or adult cerebellar (Fig. 3B) RNA did not show any protected bands. In addition, control experiments (see "Experimental Procedures") eliminated the possibility of secondary structure causing artifactual bands. Using a DNA sequencing ladder as a marker, the sizes of the protected products were calculated, and the major transcription initiation sites were mapped onto the NR2B gene sequence. As shown in Fig. 3A, the four major initiation sites are clustered in a region of 45 base pairs (circles) and map to the same location identified by the RACE assays (triangles). Analysis of the sequences upstream of the transcription start sites did not reveal consensus TATA or CAAT boxes; however, two sequences conforming to the initiator element consensus (27) are located 4 and 22 bp downstream of the 5Ј-most transcription initiation site (Fig. 2B). The NR1 (28) and NR2C (29) genes also contain TATA-less promoters. The presence of multiple transcription initiation sites associated with initiator elements in the NR2B gene is not unique; this is also found in numerous genes expressed in the central nervous system (see "Discussion").
The differential use of promoters during development and in distinct brain regions has been previously described (30). Given the differential expression on the NR2B gene during development, we performed RNase protection assays with RNA isolated from the cerebellum and forebrain of 1-week-old and adult mice. As shown in Fig. 3B, the same pattern of protected bands was observed when using RNA derived either from P7 cerebellum or from the forebrain of P7 and adult mice. There were no protected bands observed in assays using adult cerebellar RNA, consistent with the previous findings that expression of the NR2B gene is essentially shut off during postnatal cerebellar development. Thus, we found no evidence for the differential use of promoters in the NR2B gene.
A 0.8-kb Upstream Region Is Sufficient to Confer Tissue Specificity to a Reporter Construct-Transgenic mice were generated to identify sequences that regulate the neural and developmental specific expression of the NR2B gene. Three inde-pendent transgenic mouse lines (6184, 6188, and 6190) harboring the CAT reporter gene driven by a NR2B fragment that extends from 550 bp upstream of the 5Ј-most start site down through the first noncoding exon (ϩ255), denoted 0.8-CAT, were analyzed (Fig. 4A). Analysis of CAT activity in extracts made from multiple tissues of adult transgenic mice showed that the NR2B 0.8-kb upstream region conferred reporter activity specifically in the brain and failed to be expressed in thymus, liver, kidney, heart, and skeletal muscle; the latter two tissues, like the brain, are rich in excitable cells (Fig. 4B). Analysis of the same tissues from P7 mice showed that CAT expression was also restricted specifically to the brain at this age (data not shown).
The levels of CAT activity were quantitated in extracts from P7 and adult (2-month-old) transgenic mice to test if these . The CAT assay was performed on 30 g of tissue for 4 h, products were separated by TLC, and plates were exposed to film for 2 days. Preliminary experiments were done to ensure samples were in the linear range of the assay with these conditions. C, quantitation of CAT activity driven by the 0.8-CAT construct in three independent transgenic lines at two ages. Forebrain (FB) data are shaded bars; cerebellar (CB) data are solid bars. CAT activity was determined in the forebrain and cerebellum of P7 or adult (2-4-monthold) transgenic animals. Data are expressed as percentage of conversion/mg of protein/h. Samples from nontransgenic mice were used as negative controls, and the values for these samples (0.05 Ϯ 0.01%) were subtracted from all transgenic values to account for background. Bars represent the mean Ϯ S.D. (adults, n ϭ 4; P7, n Ն 2). The activity in the cerebellum is significantly greater than in the forebrain in all lines at both ages (t test). sequences are sufficient to confer proper developmental regulation. As shown in Fig. 4C, the relative levels of CAT expression in the forebrain and cerebellum of P7 and adult transgenic mice were significantly different from those observed for the NR2B gene, where expression of the endogenous gene is fully repressed in the adult cerebellum (see Fig. 1; Refs. 7,8,16,18,24). In the three transgenic mouse lines tested, CAT levels were significantly higher in the cerebellum than in the forebrain of P7 and adult mice. We also used mice up to 6 months of age to test that adult cerebellar expression driven by the 0.8-CAT construct was not simply due to the long half-life of the CAT reporter protein. These animals had levels of CAT activity similar to the 2-month-old mice (data not shown). Thus, the 550-bp upstream sequence plus the first exon is sufficient to direct transgene expression specifically in the brain but is insufficient to repress reporter activity in the cerebellum of adult mice.
Downstream Sequence Is Necessary to Confer Developmental Down-regulation of the NR2B Gene in the Cerebellum-Transcriptional regulatory elements reside within the first intron of numerous genes that contain 5Ј-noncoding exons (see Refs. 31 and 32). To test the potential role of the first intron and second exon in regulating NR2B transcription, we generated mice with a second NR2B reporter construct that originates from the same upstream site as 0.8-CAT and extends through the second noncoding exon (ϩ1627; Fig. 5A). Two independent transgenic lines (6944 and 6948) were developed with this construct, denoted 0.8int-CAT. As shown in Fig. 5B, adult and P7 (not shown) 0.8int-CAT mice also expressed the reporter gene in the central nervous system but not in other tissues. Interestingly, the reporter levels in the forebrain of adult 0.8int-CAT mice were 5-40-fold lower than those measured in the 0.8-CAT mice, suggesting that sequences that reduce transcription may reside in the first intron or second exon (see "Discussion").
To test for developmental regulatory sequences, the levels of CAT activity were quantitated in extracts made from the forebrain and cerebellum of P7 and adult transgenic mice. In contrast to the 0.8-CAT mice, the levels of CAT reporter were dramatically reduced in the cerebellum of adult mice but continued to be expressed in the forebrain (Fig. 5C). CAT activity in cerebellar extracts made from P7 mice dropped from 0.5% (line 6944) and 0.4% (line 6948) chloramphenicol conversion/mg of protein/h to levels that were indistinguishable from background (0.05 Ϯ.01%) in adult mice. In both lines transgene expression in the adult forebrain was maintained at levels that ranged between 0.75% (line 6944) and 0.25% (line 6948) conversion/mg of protein/h. The dramatic down-regulation of reporter activity specifically in the cerebellum is qualitatively similar to that observed for expression of the endogenous NR2B gene. The profiles of reporter expression in 0.8int-CAT mice were similar to the expression patterns of NR2B transcripts in brains from P7 and adult mice. An exception to this were the P7 mice from line 6944, where the levels of CAT reporter expression were higher in the cerebellum than forebrain. Differences in reporter expression profiles among different transgenic lines are not unusual, and the variability is normally attributed to effects imparted by the site of DNA integration (23). As is commonly observed with transgenic mice, the number of integrated copies did not correlate with the levels of CAT expression; the transgene copy number varied from 3 to 16 for 0.8-CAT lines and from 2 to 9 for 0.8int-CAT lines. In conclusion, these results demonstrate that sequences in the first intron and/or second exon are necessary to recapitulate the proper pattern of NR2B expression during cerebellar development.

DISCUSSION
The properties of NMDA receptors have been shown to be regulated in the cerebellum (17,33), superior colliculus (34), and visual cortex (35) during developmental periods that coincide with neural differentiation and the activity-dependent refinement of synaptic connections. The diversity in receptor properties and neural responses to glutamate results, in large part, from differences in the heteromeric composition of NMDA receptor subunits expressed by distinct neural cells during development (Refs. 4 -8; for review see Ref. 9); post-transcriptional mechanisms such as differential splicing (15,36), translation (37,38), and phosphorylation (39) also regulate receptor function. The NR2B subunit is expressed in most forebrain structures throughout postnatal development, but its expression is selectively repressed in the cerebellar granule cells after P14. The repression of NR2B expression in these cells, followed by the expression and assembly of the NR2A and NR2C sub- units to form heteromeric receptors with distinct functional properties, may have important consequences for the maturation of granule cells (see below). To begin understanding the molecular basis for the differential regulation of the NR2B gene, we have investigated the mechanisms controlling its tissue-specific and developmental regulation in the forebrain and cerebellum.
Different Transcription-regulatory Sequences Are Required to Direct Neural Specificity and Developmental Regulation of the NR2B Gene-An explanation of the mechanisms that regulate the NR2B gene must include ways to restrict its expression to neural cells and account for both the down-regulation in the cerebellum during development and the continued expression in the forebrain in the adult. We considered several mechanisms to account for the distinct expression of the NR2B gene in forebrain and cerebellum, including the differential use of multiple genes, promoters, and transcription-regulatory sequences. Our results obtained from genomic library screening, Southern blotting (data not shown), and RACE analysis, as well as a recent report by Madarnas et al. (40) using chromosome mapping, indicate that there is a single copy of the NR2B gene present in the mouse genome. There is precedence for differential promoter utilization in tissues in genes containing multiple transcription initiation sites. For example, this has been previously described for the brain-derived neurotrophic factor gene, which uses different initiation sites in various brain regions (30). Using RNase protection and RACE, we mapped multiple transcription initiation sites in the NR2B gene that were associated with initiator elements, as has been found in other neural genes including the NR2C subunit (28). However, we found no evidence for the differential utilization of promoters in the forebrain or cerebellum during development, indicating that this mechanism does not account for the differences in NR2B mRNA expression. Instead, we have shown that different genomic sequences of the NR2B gene, extending from Ϫ550 bp through either the first or second noncoding exon, are important for its regulation. Using transgenic mice, we found that while the 0.8-CAT construct is sufficient to confine NR2B transcription to neural cells, an additional 1.4-kb region encompassing the first intron and second exon is required to down-regulate NR2B transcription in the cerebellum during development.
Comparison of NR2B Regulatory Sequences with cis-Elements Found in Other Genes Expressed in Neural Cells-Transcriptional regulatory elements work in a combinatorial manner to regulate levels of expression in specific cell types. Therefore, a plausible mechanism to explain the expression pattern of the NR2B gene is that factors that activate transcription are expressed in all NR2B-expressing neurons, but specific repressive factors become active in cerebellar granule cells during development. Alternatively, the cerebellum may lack positive regulatory factors that maintain expression of the NR2B gene later in development.
Transcriptional silencing has been shown to constitute an important mechanism for confining expression of genes specifically to neural tissue (see Ref. 41), as was initially shown from the analysis of the neural-specific genes encoding for type II voltage-sensitive sodium channels (42) and SCG10 (43). A sequence known as the neuron-restrictive silencer element (RE1 or NRSE), which represses transcription by association with a repressor protein known as REST or NRSF (44,45), has been shown to down-regulate transcription of the type II sodium channel and SCG10 genes in nonneural cells. This element has also been shown to be functional in the synapsin (46), Na,K-ATPase (47), dopamine ␤-hydroxylase (48), and neuronal nicotinic acetylcholine receptor ␤2-subunit (49) genes and is found in 14 other neuronal genes (44). It acts as a negative element in neural cells in the dopamine ␤-hydroxylase (48) and the neuronal nicotinic acetylcholine receptor ␤2-subunit (49) genes. Analysis of the NR2B sequences for putative transcription factor binding sites revealed a RE1/NRSE-like sequence located at the SacII site in the first exon of the NR2B gene. As shown in Fig. 6A, the NR2B sequence is highly homologous to the RE1/ NRSE element previously described. Mice harboring either the 0.8-CAT construct (which lacks this sequence) or the 0.8int-CAT construct (which includes this sequence) express CAT in a neural-specific fashion, indicating that other cis-acting sequences are involved in conferring tissue specificity (see below). Interestingly, mice harboring the 0.8int-CAT construct selectively repressed reporter activity to background levels (from 38-to 52-fold over background at P7 in the two lines analyzed) in the adult cerebellum and not the forebrain (Fig. 5C), while the 0.8-CAT failed to do so (Fig. 4B). It is possible that the NR2B RE1/NRSE sequence acts as a negative modulator of transcription in a cerebellum-specific way during development; we are currently testing if the RE1/NRSE sequence has a functional role in regulating NR2B transcription. Another sequence found in the 0.8int-CAT that is absent in the shorter construct, and which could account for repression of the NR2B expression in adult cerebellum, is the element known as the N box (CACNAG; found at ϩ758 and ϩ1472). This sequence is bound by members of the basic helix-loop-helix family of transcription factors, which have been shown to be necessary for the differentiation of central nervous system neurons. Some members of this gene family are specifically expressed in the brain and function as negative regulators of transcription (50).
The larger construct also conferred much lower absolute levels of expression in both the forebrain and cerebellum, suggesting that the longer construct includes a negative element that is active in the forebrain. From our data, we cannot speculate on whether the same element is responsible for different levels of negative regulation in all brain regions or if there are two distinct elements active in the cerebellum and forebrain. Alternatively, the reduced reporter levels observed with the 0.8int-CAT construct may result from sequences in the 5Јnoncoding region that regulate translation (37).
The 0.8-CAT construct that is sufficient to confer neuralspecific transcription includes sequence that is identical to that found in other neural-specific promoters (Fig. 6B). The motif CCAGGAG, which was initially described during the characterization of the type II sodium channel gene (42), is located in FIG. 6. Comparison of NR2B sequence elements with motifs found in other neural genes. Nucleotides in the NR2B gene that match previously described cis-acting elements (see "Discussion") are underlined. Nucleotides conserved among all previously described elements are in boldface type. A, the RE1/NRSE element. rSCG10, rat SCG10 gene (52); rNaChII, rat type II sodium channel (42); hSYNAP-SIN, human synapsin I gene (46); m␤2nAchR, mouse ␤2-subunit of neuronal nicotinic acetylcholine receptor (49). B, the CCAGGAGA motif. mNR2C, mouse NR2C gene (28); rNaChII, rat type II sodium channel (42); rPERIPHERIN, rat peripherin (53); rGAP-43, rat GAP-43 (51). the first exon of the NR2B gene (Fig. 2B). Both reporter constructs include this sequence. Although transient transfection experiments have shown that this element is not sufficient to confer neuron-specific gene expression (42,51), it may play a role in combination with other cis-acting elements.
Transcription as a Mechanism to Regulate NMDA Receptor Properties-The changes in kinetic properties of the NMDA receptor channel and its regulation play an important functional role during development and in synaptic plasticity in the adult. Perhaps the best case supporting changes in subunit composition regulating receptor properties can be made for the developmental switch that occurs in cerebellar granule cells after migration into the internal granule layer. The changes in kinetic properties of the NMDA receptor channel in cerebellar granule cells coincide with replacement of the NR2B subunit by the NR2C subunit (10,17). Moreover, the single channel properties of the premigratory and maturing granule cells closely resemble those of in vitro synthesized dimeric receptors NR1/ NR2B and NR1/NR2C, respectively (10,17). Based on experiments performed with cerebellar slice cultures or cultured granule cells (18,19), it was recently proposed that electrical activity down-regulates expression of the NR2B gene in the cerebellum; the mechanisms activating NR2C expression are not yet understood. The constructs reported here will allow a detailed investigation of the molecular mechanisms that control both the tissue specificity and the developmental downregulation of NR2B expression in the cerebellum. This will include analysis of the potential role synaptogenesis and/or neuronal activity may play in the control of subunit composition of the receptor during development.