N-Methyl-d-aspartate Receptors Regulate a Group of Transiently Expressed Genes in the Developing Brain*

Mammalian brain development requires the transmission of electrical signals between neurons via theN-methyl-d-aspartate (NMDA) class of glutamate receptors. However, little is known about how NMDA receptors carry out this role. Here we report the first genes shown to be regulated by physiological levels of NMDA receptor function in developing neuronsin vivo: NMDA receptor-regulated gene 1(NARG1), NARG2, and NARG3. These genes share several striking regulatory features. All three are expressed at high levels in the neonatal brain in regions of neuronal proliferation and migration, are dramatically down-regulated during early postnatal development, and are down-regulated by NMDA receptor function. NARG2 and NARG3 appear to be novel, while NARG1 is the mammalian homologue of a yeast N-terminal acetyltransferase that regulates entry into the Go phase of the cell cycle. The results suggest that highly specific NMDA receptor-dependent regulation of gene expression plays an important role in the transition from proliferation of neuronal precursors to differentiation of neurons.

Long before synaptic networks are fully established, electrical activity present in developing neurons regulates neuronal differentiation (1)(2)(3)(4). In particular, electrical activity mediated by the NMDA 1 class of glutamate receptors is required for normal neuronal development. Loss of NMDA receptor function during development increases neuronal cell death (5), prevents the formation of precise neural circuits (6, 7), diminishes respiration and feeding (8 -10), and has been implicated in fetal alcohol syndrome (11) and schizophrenia (12). NMDA receptor function can also regulate neuronal proliferation (13) and migration (14). Despite the importance of NMDA receptors for normal development and adult brain function, knowledge of molecular mechanisms regulated by NMDA receptors in developing neurons is rudimentary.
Evidence from development as well as adult models of learning and memory indicate that regulation of gene expression is an important strategy that can be used to mediate changes in neuronal structure and function. Developmental studies sug-gest that mRNA abundance is rate-limiting for the accumulation of functional neurotransmitter receptors on neurons during development in vivo (15). Changes in gene expression also accompany activity regulated synaptic plasticity events in the developing visual system (16), as well as at the neuromuscular junction (17). Transcription factors have recently been shown to play a critical role for the development of synaptic connectivity (18 -20). Similarly, transcription may be required for activity dependent long term changes in synaptic strength in mature nervous systems (21)(22)(23). Finally, the electrical activity of neurons, including NMDA receptor function, has been shown to regulate gene expression in the adult hippocampus (24 -27), a structure critical for the establishment and maintenance of memories (28 -30).
To test the hypothesis that NMDA receptor-dependent regulation of gene expression is required for, and directs, molecular and cellular mechanisms for neuronal development, we have designed a screen based on the neural circuit that connects highly specialized whiskers (mystacial vibrissae) found on the snout of the mouse to their synaptic targets in the brain stem. These whiskers are richly innervated by pseudounipolar sensory neurons of the trigeminal ganglion that project centrally to the brain stem trigeminal complex. Here synaptic inputs from the trigeminal ganglion that correspond to the mystacial vibrissae are organized in a pattern that matches their topographic organization on the face. This pattern of whisker representations, called barrelettes at the level of the brain stem, is relayed and reiterated in the thalamus and finally the cerebral cortex, and requires NMDA receptor function for normal development (8,10,31,32). The whisker representations are comparable to other sensory maps, for example, for touch, hearing, and vision, that are found in humans, and serve as a powerful mammalian model for highly patterned synaptic development and organization in vivo (33).
By using a unique combination of NMDAR1 knockout mice (8), the trigeminal system of whisker representations, and DNA array analysis, we have discovered a group of three NMDA receptor-regulated genes (NARG1, NARG2, and NARG3) that share striking regulatory features. In the brain stem, where barrelettes are present in wild-type but absent in NMDA receptor knockout mice, these NARGs are expressed at approximately 2-fold higher than normal levels in NMDAR1 knockout animals. Comparison of NARG expression among NMDAR1 knockout, wild-type, and heterozygous brain stems indicates dose-dependence in terms of NMDA receptor-dependent regulation of these genes. In the adult, all three NARGs are expressed at low levels, but with similar spatial distribution. During brain development these genes peak in expression around the time of birth, with highest levels in regions where proliferation and migration are still occurring. Expression is dramatically down-regulated during early postnatal development at the same time that NMDAR1 is up-regulated. The results support the hypothesis that NMDA receptor-dependent regulation of gene expression plays a crucial role for neuronal maturation during mammalian brain development.

EXPERIMENTAL PROCEDURES
RNA Isolation-All tissue dissection for RNA preparation was performed under ice-cold saline. For prenatal brains, the dam was anesthetized by metofane and embryos were quickly removed. For P0 mice, animals were anesthetized briefly on ice, decapitated, and the tissue was rapidly dissected. Older animals were anesthetized by metofane prior to decapitation and dissection. Total RNA was isolated by homogenization and extraction with an acid guanidinium thiocyanate/phenol/ choloroform mixture as previously described (34) with modifications (Atlas Pure Total RNA Isolation Kit, CLONTECH). RNA yield, concentrations, and quality were determined by OD 260 and corroborated by ethidium bromide staining in formaldehyde agarose gels. Genotyping was performed by PCR as previously described (35).
DNA Array Analysis-Total RNA was extracted from P0 mouse brain stems and used to prepare wild-type (ϩ/ϩ) and NMDAR1 knockout (Ϫ/Ϫ) first-strand cDNA probes that were hybridized to nylon and glass cDNA arrays. For nylon array analysis (CLONTECH catalog number 7741-1), first-strand 32 P-labeled cDNA probes were prepared and the membranes processed according to the manufacturer's instructions. Specific signal was detected for 297 out of 588 genes represented. Of these, 161 genes yielded signals at least 2-fold above background; expression was compared between wild-type and knockout for these genes. Three independent experiments were performed, and no significant differences were observed in signals obtained using wild-type and knockout probes. For microarray analysis (Mouse GEM 1, Incyte), RNA from brain stems of 10 wild-type and 10 knockout P0 mice was extracted in parallel, and the genotypes of the final wild-type and knockout RNA preparations were reconfirmed by RT-PCR. The RNA samples were processed by Incyte on GEM 1, which contains a total of 8,734 elements, including at least 7,634 unique genes/clusters, controls, and a number of partially characterized elements (ESTs). Taken together, our array experiments analyzed about 8,500 genes. Assuming that the number of the mouse genes is no more than about 60,000, i.e. similar to human estimates (36), our screen included 10 -15% of all expressed mouse genes.
RNase Protection-RNase protection experiments were carried out as previously described (15). Briefly, total RNA (10 -20 g) was annealed to a molar excess of 32 P-labeled single-stranded antisense RNA probe for 5 min at 85°C in 80% formamide, 40 mM PIPES buffer (pH 6.7), 0.4 M NaCl, and 1 mM EDTA, and the hybridization was continued for 12-18 h at 45°C. Single-stranded RNA was digested by incubating for 60 min at 15°C with 4 g/ml DNase-free RNase (Roche Molecular Biochemicals) and 0.4 g/ml RNase T1 (Roche Molecular Biochemicals). The RNases were then inactivated by treatment with proteinase K and SDS, and the RNA was extracted with phenol/chloroform, precipitated with ethanol, dissolved in loading buffer containing 80% formamide, and electrophoresed in 5% acrylamide containing 8 M urea. The dried gels were exposed at Ϫ70°C to Kodak BioMax MS film, and signals were quantified directly from the gels using a Molecular Dynamics Storm 860 PhosphorImager (except for one NARG1 experiment in which a scanned autoradiograph was quantified using NIH Image). Probes used for RNase protection are as follows: NARG1, nucleotides 19 -291 of AA474587; NARG2 nucleotides 102-365 of AA472833; NARG3, nucleotides 5-283 of AA473329.
In Situ Hybridization-In situ hybridization analysis was performed as described (37,38). Briefly, cryostat sections (12 m) were cut, air dried, fixed for 30 min in 4% paraformaldehyde, dehydrated with ethanol, and stored in Ϫ80°C. Sections were thawed, digested by proteinase K, acetylated, and dehydrated with ethanol. The sections were hybridized at 57°C for 16 h with 5 ϫ 10 6 cpm/ml of 35 S-labeled riboprobe, followed by digestion with 50 g/ml RNase A at 37°C for 30 min, and washed in a series of SSC solutions of increasing stringency, ending with 0.1 ϫ SSC, 1 mM dithiothreitol at 60°C for 30 min. The slides were exposed to Kodak BioMax MS film and were subsequently coated with NTB-2 emulsion and developed after 22-25 days. The sequence of the riboprobes are the same as those used in the RNase protection assays. For all samples, corresponding sense riboprobe was used as control and found to yield only background levels of silver grains. For comparisons of knockout and wild-type brains, sections were placed on the same slide and processed identically.
Northern Hybridization-A mouse poly(A) ϩ RNA Northern blot (Ori-Gene) containing RNA from multiple adult tissues was hybridized with 1 ϫ 10 6 cpm/ml of 32 P-labeled cDNA probe in ULTRAhyb buffer (Am-bion) at 42°C for 16 h. The highest stringency wash was 0.1 ϫ SSC, 0.1% SDS, at 42°C for 15 min. The probes used were: NARG1, the 630-base pair SmaI/KpnI fragment from IMAGE clone 805455; NARG2, the 609-base pair HindIII/BseRI fragment from IMAGE clone 804793; NARG3, the 563-base pair DraI fragment from IMAGE clone 805306. NARG1 (mNAT1) cDNA Clones-To obtain cDNA clones that contain sequence upstream of the NARG1 EST (AA474587), an arrayed mouse testis cDNA library (OriGene) was screened by PCR. PCR primers were designed based on the EST sequence: 5Ј primer, 5Ј-CAGTACGTGTGT-TGTGTTTCCA-3Ј; 3Ј primer, 5Ј-TTCAAGCCATGGTAGAGTTTTC-3Ј. The clones obtained from this screening are conventional (not PCR clones), and our detailed analysis of the complete mNAT1 cDNA sequence is forthcoming. Although the longest NARG1 cDNA clone characterized lacked the 5Ј most 54 base pairs of coding sequence, this was obtained from EST sequences (AA163290 and AA561496). A partial mNAT1 cDNA has previously been reported (39).

A Screen for NMDA Receptor-regulated Genes in Developing
Neurons-Brain stems, including nucleus principalis, nucleus interpolaris, and nucleus caudalis, were dissected from neonatal mice produced from NMDAR1 heterozygous crosses. These nuclei receive synaptic input from the whisker pad via the trigeminal ganglion, and contain whisker representations in wild-type mice that are absent in NMDAR1 knockouts. In initial experiments, duplicate nylon membrane arrays containing previously characterized mouse cDNAs were hybridized with wild-type (ϩ/ϩ) and NMDAR1 knockout (Ϫ/Ϫ) cDNA probes (Fig. 1A). None of the 588 genes represented met our follow-up criteria of at least a 2-fold change in signal between wild-type and NMDAR1 knockout.
The number of genes detected on the nylon array represents less than 1.0% of the mouse transcriptome. To increase the scale of the screen to include 10 -15% of the mouse transcriptome, probes from wild-type and NMDAR1 knockout brain stem were hybridized to a mouse gene expression microarray containing a total of 8,734 elements. NMDAR1 wild-type and knockout brain stem probes detected 8,598 elements, i.e. over 98% of the cDNAs present on the array. Approximately 5,300 elements were detected at sufficient levels to allow comparison between NMDAR1 wild-type and knockout gene expression. Of these, five elements were found that demonstrated differences of between 2-and 2.5-fold (Fig. 1B). We initially checked the validity of the microarray results by semi-quantitative RT-PCR (not shown) and then by RNase protection analysis (Fig. 2). Three of the five original elements identified on the microarray were confirmed as NMDA receptor-regulated genes: NARG1, NARG2, and NARG3. In addition, heterozygous RNA yields signals for NARG1 and NARG3 (signals were too weak to address NARG2 in heterozygotes) that are intermediate between those obtained in homozygous knockout and wild-type preparations (Fig. 2), indicating that NMDA receptor-dependent regulation of these genes is dose-dependent.
NARG1, NARG2, and NARG3 Share Similar Spatial Expression Patterns-In situ hybridization confirmed up-regulation of NARG1 and NARG3 in NMDAR1 knockouts, and revealed similar spatial distribution for NARG1 and NARG3 expression in the neonatal brain ( Fig. 3; unfavorable signal to noise ratios have thus far prevented in situ hybridization analysis for NARG2). Highest levels of expression were observed in the ventricular zone, ganglionic eminence, cortex, olfactory bulb, and basal ganglia. At the time of birth these regions contain numerous proliferating and migrating cells.
To determine whether these genes share similar spatial expression patterns in other parts of the body, we performed Northern blot analysis on multiple tissues from the adult mouse. Although levels of expression were found to be low in all adult tissues examined, similar patterns of expression were observed for these three genes (Fig. 4).
NARG1, NARG2, and NARG3 Are Transiently Expressed in the Developing Brain-To begin to understand the physiological significance of NARG1, NARG2, and NARG3 in the developing brain, we performed RNase protection analyses on whole brain RNA extracted on embryonic day 13 (E13), postnatal day 0 (P0), P7, and P35. These three NARGs display virtually identical patterns of expression: relatively low on E13, high on P0, and very low thereafter (Fig. 5). This pattern is in contrast to that observed for NMDAR1, which is expressed at lowest levels at E13, and expression steadily increases as the brain matures (40) (Fig. 5, bottom panel). The three NARGs are down-regulated as NMDAR1 is up-regulated, precisely as would be predicted by results that demonstrate higher NARG mRNA levels in NMDAR1 knockout neural tissue than in wild-type.

NARG1 Is the Mammalian Homologue of a Yeast N-terminal Acetyltransferase That Regulates Entry into G o Phase of the
Cell Cycle-To determine the identity of NARG1, NARG2, and NARG3 we have isolated and sequenced a number of cDNA clones. Because the ESTs on the microarray correspond to 3Ј regions of the transcripts, we screened libraries to obtain cDNA clones that contain upstream coding regions. Sequence analysis of partial cDNA clones indicates that NARG2 and NARG3 represent previously unknown genes. BLAST searches revealed that NARG1 is the mammalian homologue of N-terminal acetyltransferase 1 (NAT1), a gene that was originally discovered in yeast (Fig. 6). Therefore, we will henceforth refer to NARG1 as mNAT1. Null mutants of nat1 in S. cerevisiae display reduced acetyltransferase activity, derepression of a silent mating type locus, and failure to enter G o (41).
FIG. 1. Comparison of gene expression between wild-type (؉/؉) and NMDAR1 knockout (؊/؊) brain stem from neonatal mice. A, nylon cDNA array analysis; 588 mouse cDNAs. None of these genes met our follow-up criteria of at least a 2-fold change in signal between wild-type and knockout (signals were averaged from three independent experiments). These results also serve as a control; none of the 15 stress response genes and 61 cell death genes present on the array were regulated by the absence of NMDAR1. First-strand 32 P-labeled cDNA probes derived from total RNA preparations were used for these experiments. Identities of the genes on the array are given in CLONTECH's product analysis certificate, catalog number 7741-1. B, microarray analysis; 8734 mouse elements. Five genes were found that demonstrated differences of between 2-and 2.5-fold; the sectors of the array that contain these elements are shown with the black outline removed from the elements of interest. Fluorescently-labeled first-strand cDNA probes from wild-type (ϩ/ϩ; Cy3) and NMDAR1 knockout (Ϫ/Ϫ; Cy5) brain stem were prepared and hybridized to a mouse GEM (custom service by Incyte). The identities of the elements present on the GEM are available (Mouse GEM1; Incyte).

DISCUSSION
Despite the role of electrical activity and NMDA receptors in regulating neuronal development including migration, proliferation, survival, and synaptic plasticity, the question of what genes are regulated by NMDA receptor function in developing neurons has not been systematically addressed. To test the hypothesis that NMDA receptor-dependent electrical activity of neurons directs gene expression to facilitate activity dependent neuronal development, we have performed a screen based on the neural circuit that connects the highly specialized whiskers found on the nose of the mouse to their synaptic targets in the brain stem. We have identified a group of three genes, NARG1, NARG2, and NARG3, that share striking regulatory features including NMDA receptor-dependent regulation, a transient surge in expression early in brain development, low levels of expression in the adult, and similar spatial distribution. All three of these genes are expressed at higher than normal levels in NMDAR1 knockout brains, and at least two of them display dose-dependence in terms of regulation by NMDA receptor function. During neonatal development all three NARGs are sharply down-regulated at the same time that NMDAR1 is up-regulated. The results indicate that maturation of NMDA receptor function in developing neurons plays a role in down-regulating this group of genes, and support the hypothesis that NMDA receptor-dependent regulation of gene expression is required for, and directs, molecular and cellular mechanisms for neuronal development.
Genetic compensation versus direct regulation is a consideration in any screen for genes regulated as a consequence of transgenic manipulation. We have addressed compensation on three levels. First, among ϳ8,500 genes examined, only three NMDA receptor-regulated genes have been identified as a result of NMDAR1 knockout. The results indicate a degree of specificity that argues against large scale induction of compensatory mechanisms. Second, we have investigated genes known to be involved in stress response (15 genes; Fig. 1) and cell death (61 genes; Fig. 1) by using nylon membrane array anal-FIG. 2. RNase protection using 32 P-labeled cDNA antisense riboprobes confirms that NARG1, NARG2, and NARG3 are expressed at approximately 2-fold higher than normal levels in the absence of NMDAR1. The predicted sizes (nucleotides) of undigested probes (arrowheads) and protected species (arrows) are indicated. Autoradiographs from three experiments (horizontal bars) for NARG1 and NARG3 are shown, as well as for one NARG2 experiment. Undigested probe and tRNA controls were performed in all experiments, although only one set of these controls per NARG is shown here. Quantification of knockout/wild-type signal ratios obtained by RNase protection are as follows; NARG1: P0 brain stem, 1.7 Ϯ 0.3 (n ϭ 4), P0 brain (not shown), 1.4 Ϯ 0.1 (n ϭ 4); NARG3: P0 brain stem, 2.5 Ϯ 0.8 (n ϭ 4), P0 brain (not shown), 1.4 Ϯ 0.1 (n ϭ 5). RNase protection signals for NARG2 in the brain stem were very low and were not quantified. However, in all experiments that compared NARG2 signal between wild-type and knockout brain stem, higher signal was observed in knockout samples (n ϭ 3; a typical experiment is shown). Quantification of P0 whole brain signals for NARG2 indicates a level of regulation of about 2-fold (knockout/wild-type ratios were 1.9 and 1.6 in two independent determinations), which is comparable to the level of regulation observed on the cDNA microarray (2.2-fold). Up-regulation of NARG expression also occurs in heterozygotes (n ϭ 7: 5 for P0 brain stem, including 2 for NARG1 and 3 for NARG3; 2 for P7 brain, including 1 for NARG1 and 1 for NARG3). Moreover, signals for heterozygous samples were intermediate between wild-type and knockout. This indicates that the observed regulation by NMDA receptors is dose-dependent. GenBank TM accession numbers of the ESTs that correspond to these genes are: NARG1, AA474587; NARG2, AA472833; NARG3, AA473329.
FIG. 3. In situ hybridization analysis using 35 S-labeled riboprobes on sagittal sections from P0 mouse brains demonstrates that NARG1 and NARG3 are expressed at higher levels in NMDAR1 knockouts (؊/؊) than in wild-types (؉/؉). In wild-types, significant expression of NARG1 is observed in the ventricular zone and intermediate zone (arrowheads), basal ganglia and ganglionic eminence (G), cortical plate (arrows), and olfactory bulb (OB). NARG1 is also expressed in the cerebellar cortex and the caudal inferior colliculus, just rostral and superior to the cerebellum (C). In wild-types NARG3 displays a similar expression pattern to NARG1 in the telencephalon, and shows little signal in more caudal regions of the brain. Sense controls yielded background levels of signal (not shown). BS, brain stem. Scale bar, 1 mm. ysis. None of these genes were significantly regulated by the lack of functional NMDA receptors in the mouse neonatal brain stem. Third, to address the possibility that the regulation observed for NARG1, NARG2, and NARG3 could be attributed to the fact that NMDAR1 knockout mice die within 24 h of birth, we have examined the expression of these genes in heterozygous animals, which are viable and grossly indistinguishable from their wild-type littermates. (Knockout mice appear healthy at birth, and are dissected well before the onset of morbidity, which starts after 10 -24 h (42).) The results demonstrate mRNA levels for these genes in heterozygotes that are intermediate between those found in wild-type and knockout brain stems. Previous reports have also demonstrated graded effects of NMDAR1 gene dosage on the survival of both hippocampal neurons and Purkinje cells in cultures prepared from wild-type, heterozygous, and NMDAR1 knockout mice (43,44). Taken together, these findings argue against genetic compensation and suggest that NMDA receptor-dependent regulation of NARG1, NARG2, and NARG3 is specific and dose-dependent.
Previous screens for neuronal genes regulated by electrical activity have focused on overall levels of activity rather than NMDA receptor-specific activity. Three independent studies have used electrical stimulation (25), the ␥-aminobutyric acid-A antagonist metrazol (26), and the glutamate analogue kainate (24), respectively, to induce hippocampal seizures in adult rats. The rationale of these experiments is that the exceedingly high levels of electrical activity associated with seizures should robustly induce synaptic plasticity events and therefore allow large numbers of candidate genes to be identified. The only previously published screen for genes regulated by electrical activity in developing neurons in vivo used a differential cloning paradigm based on the activity dependent segregation of retinal ganglion cell synaptic inputs into eyespecific domains in the cat visual system (38). By blocking spontaneously occurring action potentials in the developing lateral geniculate nucleus and comparing gene expression with controls, class I major histocompatibility antigens (MHC I) were identified. The MHC I invariant light chain, ␤ 2 -microglobulin, as well as a component of a receptor complex for MHC I, CD3, were also found to be expressed by subsets of central nervous system neurons. These data, as well as recent functional analyses, support the novel hypothesis that MHC I proteins facilitate activity dependent synaptic plasticity in developing neurons (38,45). The MHC I results also demonstrate the viability of the strategy of identifying important genes for neuronal development by virtue of regulation by physiological levels of electrical activity in the developing brain.
Taken together, previous screens have identified ϳ400 activity regulated genes that could play a role in synaptic plasticity and other activity regulated events in neurons. It is somewhat surprising, therefore, that our screen of 10 -15% of the mouse transcriptome identified only 3 genes, two of which are novel. If this sample of the transcriptome is typical, it follows that by saturating the present screen we would have identified on the order of 30 -45 NARGs, most of which would be expected to be novel. Although this is far fewer genes than would be predicted based on adult hippocampus seizure models, the strong agreement between microarray and RNase protection analyses (ϳ2fold regulation for NARGs1-3 was measured by both approaches) make it unlikely that the microarray results greatly underestimated the number of genes regulated by the loss of NMDAR1. Moreover, an examination of microarray analysis studies indicates that, under physiologically controlled conditions, on the order of 0.1-2% of genes are regulated by 1.8-fold or more (46). This level of regulation is consistent with the findings reported here.
There are, however, a number of potential explanations for FIG . 5. NARG1, NARG2, and NARG3 are transiently expressed at high levels in the neonatal brain. RNase protection analysis using 32 P-labeled single-strand antisense RNA probes was performed on wild-type brain total RNA extracted at various stages of development. Riboprobes complementary to NARG1, NARG2, and NARG3 were synthesized, isolated, annealed with the indicated RNA sample, digested with RNases, and fractionated by gel electrophoresis. Autoradiographs of gels are shown, with the predicted sizes of the undigested probes (arrowheads) and protected species (arrows) indicated. Probe, undigested probe; E13, embryonic day 13; P0, postnatal day 0; P7, postnatal day 7; P35, postnatal day 35.
FIG. 6. NARG1 is the mouse homologue of yeast NAT1. Alignment of NARG1 (mouse) amino acid sequence with NAT1 from various species. The first 137 out of ϳ850 amino acids are shown. Over the entire amino acid sequence, NARG1 is 46% homologous (28% identical) to Saccharomyces cerevisiae NAT1. Xenopus: GenBank TM accession number AF247679. Drosophila: AE003512. S. cerevisiae: X15135. The portion of NARG1 shown here is contained in AA163290. the apparent discrepancy between our results and those obtained using seizure models. First, it is possible that more genes are regulated by overall action potential activity than those that are regulated by NMDA receptor function alone. Second, it has been shown that regulation of expression of a given gene by electrical activity is a specific property that can change among brain regions as well as during development (47,48). It is plausible, therefore, that a greater number of genes are regulated by electrical activity in the adult hippocampus, where almost all activity regulated neuronal genes have been discovered, than in the developing brain stem. An additional possibility is that the number of genes regulated by normal physiological levels of electrical activity is lower than the number of genes that can be regulated by seizures. Finally, the set of genes examined on the nylon array and the microarray did not contain a number of genes known to be regulated by electrical activity, e.g. neurotrophins, and therefore may not be statistically representative of a more complete sampling. It is clear that additional analyses of NMDA receptor-regulated gene expression in developing neurons are needed to resolve this issue.
Another issue that should be addressed in future experiments is whether or not the electrical function of NMDA receptors is solely responsible for NMDA receptor-dependent regulation of gene expression. While pharmacological and transgenic studies indicate that this is likely to be the case (e.g. Refs. 24, 25, and 49), there is evidence that NMDA receptors contribute to organizing a network of postsynaptic proteins that connects the synapse to the cytoskeleton (50,51). In addition to disrupting NMDA receptor-mediated electrical activity, knockout of the NMDAR1 subunit is also expected to disrupt this structural function. This could potentially cause downstream changes, including the regulation of gene expression.
The unusual regulatory features shared by NARG1, NARG2, and NARG3 may reflect a common function in developing neurons. NARG1, NARG2, and NARG3 are transiently expressed at high levels in P0 brain, expression is highest in neurons that are just exiting or about to exit the cell cycle, and NARG1 (mNAT1) is the mammalian homologue of a yeast N-terminal acetyltransferase that regulates entry into the G o stage of the cell cycle (41). In addition, pharmacological studies that have established a role for NMDA receptors in cell cycle regulation during the final stages of dentate gyrus maturation in the rat hippocampus (13,52). Taken together, the results suggest that these genes may facilitate the transition from proliferation of neuronal precursors to differentiation of neurons, in part via NMDA receptor-dependent regulation.
In the present study we report the first specific genes that are regulated by normal levels of NMDA receptor function in developing neurons in vivo. The results support the hypothesis that NMDA receptor-dependent regulation of gene expression is required for, and directs, molecular and cellular mechanisms for neuronal development. Moreover, the identification of NMDA receptor-regulated genes provides the insight and tools needed to begin to understand the relationship between NMDA receptor function, gene expression, and ultimately neuronal differentiation. Because NARG1, NARG2, and NARG3 share regulatory features, these genes are likely to be regulated by common cis-regulatory elements and transcription factors that can be identified by promoter analysis. Future studies of these and other NMDA receptor-regulated genes should make an important contribution to the insight needed to obtain an integrated understanding of molecular, anatomical, and electrophysiological principles that guide neuronal development.