Nurr1 Protein Is Required for N-Methyl-d-aspartic Acid (NMDA) Receptor-mediated Neuronal Survival*

Background: The mechanism involved in activity-dependent survival of neurons in the central nervous system is not fully understood. Results: Nurr1 is involved in excitatory transmission-dependent survival of glutamatergic neurons by acting downstream CREB and upstream of BDNF. Conclusion: Nurr1 activation mediates activity-dependent survival of glutamatergic neurons. Significance: A novel function of Nurr1 in activity-dependent survival of glutamatergic neurons is reported. NMDA receptor (NMDAR) stimulation promotes neuronal survival during brain development. Cerebellar granule cells (CGCs) need NMDAR stimulation to survive and develop. These neurons differentiate and mature during its migration from the external granular layer to the internal granular layer, and lack of excitatory inputs triggers their apoptotic death. It is possible to mimic this process in vitro by culturing CGCs in low KCl concentrations (5 mm) in the presence or absence of NMDA. Using this experimental approach, we have obtained whole genome expression profiles after 3 and 8 h of NMDA addition to identify genes involved in NMDA-mediated survival of CGCs. One of the identified genes was Nurr1, a member of the orphan nuclear receptor subfamily Nr4a. Our results report a direct regulation of Nurr1 by CREB after NMDAR stimulation. ChIP assay confirmed CREB binding to Nurr1 promoter, whereas CREB shRNA blocked NMDA-mediated increase in Nurr1 expression. Moreover, we show that Nurr1 is important for NMDAR survival effect. We show that Nurr1 binds to Bdnf promoter IV and that silencing Nurr1 by shRNA leads to a decrease in brain-derived neurotrophic factor (BDNF) protein levels and a reduction of NMDA neuroprotective effect. Also, we report that Nurr1 and BDNF show a similar expression pattern during postnatal cerebellar development. Thus, we conclude that Nurr1 is a downstream target of CREB and that it is responsible for the NMDA-mediated increase in BDNF, which is necessary for the NMDA-mediated prosurvival effect on neurons.

During brain development, neuronal activity suppresses apoptosis and promotes survival of neurons, adjusting neuronal population to its connection patterns. The role of the NMDA subtype of glutamate receptors (NMDAR) 5 supporting neuronal survival during CNS development is well documented, and it has been reported that basal NMDAR activity suppresses neuronal apoptosis in the somatosensory thalamus, the hypothalamus, and the dentate gyrus and in CA1 in the hippocampus, among other brain areas (1)(2)(3). The trophic effect of NMDAR stimulation is especially necessary for the survival and differentiation of cerebellar granule cells (CGCs) (4).
CGCs are generated during the postnatal development of the cerebellum in the external granule layer and migrate to the internal granule layer to become mature CGCs (5,6). During its migration, they need glutamatergic stimulation from the mossy fibers. The stimulation of NMDAR is necessary for the maintenance of these neurons during cerebellum development (7)(8)(9). Failure to receive mossy fiber inputs will drive CGCs to apoptotic death (10,11). This scenario can be mimicked in vitro in primary cultures of CGCs. When these neurons are grown in low potassium concentration (5 mM KCl; K5), they undergo spontaneous apoptosis. Nevertheless, they will develop and survive in the presence of high potassium concentrations (25 mM KCl; K25) or NMDA (4,12,13).
It has been widely reported that the activation of survival signaling kinases, including Akt and ERK (14,15), and the suppression of proapoptotic kinases, such as glycogen synthase kinase-3␤ (GSK-3␤) or JNK (16,17), are involved in neuronal survival. The regulation of these kinases modulates the activity of some transcription factors that control the expression of proand antiapoptotic genes. In CGCs, NMDA-mediated prosur-vival effect seems to be dependent on the activation of the phosphatidylinositol 3-kinase (PI3K)/Akt pathway (13,18) and the activation of CREB (19). Other transcription factors, for example MEF2, FKHRL1, DeltaNp73, or Sp1/3, have also been reported to be involved in neuronal survival (20 -23). However, whether these or other factors are involved in the signaling related to the NMDAR-mediated prosurvival effect is still poorly understood.
In the present study, we have analyzed the expression changes in response to NMDA treatment in cultured CGCs. By comparative analysis, we have identified Nurr1 as a candidate to mediate the NMDAR prosurvival response. In fact, we report that NMDA induction of CREB activation promotes Nurr1 expression and consequent up-regulation of its protein levels. In addition, silencing of Nurr1 results in a reduction in brainderived neurotrophic factor (BDNF) levels that correlates with the decrease in the NMDA-mediated neuroprotection in these cells.

EXPERIMENTAL PROCEDURES
Cell Culture-Granule cell cultures were prepared from dissociated cerebella of 8-day-old Wistar rats as described previously (4). Cells were plated (3 ϫ 10 3 cells/mm 2 ) in Eagle's basal medium (Sigma-Aldrich) and 5 or 25 mM KCl (K5, K25) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Sigma-Aldrich), 25,000 units of penicillin, and 25 mg of streptomycin (PAN Biotech Inc.). 10 M cytosine-␤-D-arabinofuranoside was added to the cultures 24 h after plating to prevent proliferation of non-neuronal cells. Neurons were plated onto poly-L-lysine coated 48-well plates for measurement of cell viability and immunocytochemistry, on 35-mm culture dishes for RNA extraction and Western blotting, on 24-well plates for luciferase assays, and on 100-mm plates for ChIP assays. At 2 days in vitro (DIV), K5 neurons were treated or not with NMDA (100 M, Sigma-Aldrich) or K25. The procedures followed were in accordance with guidelines of the Comissió d'Ètica en l'Experimentació Animal i Humana of the Universitat Autònoma de Barcelona.
Cell Viability-Neuronal viability was assessed using the 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT) assay. Cells were incubated with MTT (0.2 mg/ml) for 45 min at 37°C. The blue formazan derivative was solubilized in 100 l of dimethyl sulfoxide, and the dual wavelength was measured at 560 and 620 nm in a Labsystems Multiskan plate reader. Data are presented as mean Ϯ S.E. of values obtained from three or four independent experiments performed in triplicates.
Fluorescence Analysis of Apoptotic Nuclei-Apoptosis was assessed by nuclear DNA staining with Hoechst 33258. Cells were washed twice in Tris-buffered saline (TBS) 24 h after the treatment, fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 45 min at 4°C, and stained with 1 g/ml Hoechst 33258 (Molecular Probes) for 5 min after being washed again twice in TBS. Nuclear DNA staining was observed with an inverted microscope (Nikon eclipse TE-2000-E). In every experiment, more than 1000 cells were individually examined for each experimental condition. Condensed and/or fragmented nuclei were considered as apoptotic nuclei. Data are given as mean Ϯ S.E. of values obtained in three or four independent experiments performed in triplicate. Results are expressed as the percentage of apoptotic nuclei versus total nuclei.
Immunoblotting-CGC cultures were washed with PBS, and total protein was extracted by incubating neurons in lysis buffer containing 20 mM Tris, pH 7.5, 1% Nonidet P-40, 150 mM NaCl, 5 mM EDTA, 1 mM PMSF, 10 g/ml aprotinin, 20 g/ml leupeptin, 1 mM activated orthovanadate. Cell lysates were sonicated for 1 min. Proteins (25 g) were resolved in 10% SDS-PAGE gel and transferred to nitrocellulose membranes (Amersham Biosciences). Membranes were washed with PBS/ Tween (PBS with 0.05% Tween 20) and incubated for 1 h in blocking buffer (5% of nonfat dry milk in PBS/Tween except 10% of nonfat dry milk for Nurr1 and supplemented with 5% FBS for BDNF) at room temperature to block nonspecific binding. Blots were washed and incubated overnight at 4°C with primary antibodies against CREB, phospho-Ser 133 -CREB, and Nur77 from Cell Signaling; Nurr1 and NOR-1 from Abcam; BDNF from Santa Cruz Biotechnology; GAPDH from Ambion Inc.; and actin from Sigma-Aldrich, diluted (1:1000 for all, except GAPDH and actin 1:40,000) in blocking buffer or PBS/ Tween with 0.1% BSA and 2% nonfat dry milk for BDNF. Blots were then washed and incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies (BD Biosciences) diluted 1:3000 in blocking buffer. Immunoreactive bands were visualized using the ECL TM Western blotting detection reagent (Amersham Biosciences, Uppsala, Sweden) and quantified by a computer-assisted densitometer. Actin and GAPDH were used as loading controls.
Immunohistochemistry-Cerebella of P9, P14, P26, and adult rats were fixed for 24 h with 4% paraformaldehyde in PBS at 4°C and embedded in paraffin. 5-m sections were processed for Nurr1 labeling. Briefly, sections were rehydrated and washed twice in TBS. Endogenous peroxidase activity was inactivated with 3% H 2 O 2 in TBS. After extensive washes with TBS/ Tween (TBS with 0.5% Tween), the sections were incubated in blocking solution (TBS/Tween with 10% FBS). Subsequently, the sections were exposed to Nurr1 antibody (1:200 in blocking solution; Abcam) for at least 12 h at 4°C. The sections were again rinsed extensively in TBS/Tween and incubated with anti-mouse antibody coupled to biotin (1:200 in blocking solution; Vector Laboratories) for 1 h at room temperature. After rinsing, the sections were incubated with streptavidin-peroxidase in TBS (1:300; Vector laboratories). Peroxidase was developed with 0.05% diaminobenzidine in TBS and 0.035% H 2 O 2 solution. Nuclei were stained by incubation with hematoxylin for 1 min. Sections were dehydrated and covered with DPX medium. Pictures were obtained in a NIKON Eclipse 901 microscope/Nikon digital sight camera, using 2ϫ, 10ϫ, 20ϫ, and 40ϫ objective lens.
Gene Expression Analysis-CGCs were plated in K5 and treated or not at 2 DIV with NMDA (100 M). After 3 and 8 h, RNA extraction was done with the RNeasy Protect kit (Qiagen), and samples were sent to the genomic service of the Vall d'Hebron Hospital Research Institute to perform the Affymetrix Rat Genome 430 2.0 array. Expression data were obtained using the GCOS software and normalized with robust Nurr1 Mediates CGCs Survival by NMDA multichip average. The LIMMA package was used for statistical analysis to identify up-and down-regulated genes using a multiple test-adjusted p value Ͻ0.2. For genes with multiple probes, we selected the most informative probe showing the lowest p value. The up-regulated genes at 8 h with an increase over 2-fold and a p value Ͻ0.05 were submitted to functional annotation analysis with the DAVID program (david.abcc.ncifcrf-.gov). Filters used in the functional annotation cluster were: Similarity Term overlap 6; Similarity threshold 0.70, and Classification initial group membership 5. Hierarchical clustering of the Affymetrix data was performed with the Cluster software (version 3.0) and represented with TreeView software. The clustering algorithm (unweighted pair group method with arithmetic mean) was used in a centroid linkage fashion on the probes showing an at least 1-fold change, after being centered to the median.
Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-Total RNA was extracted from CGCs at the indicated times after treatment with TRIzol (Invitrogen) according to the manufacturer's instructions. 1 g of total RNA was converted to first strand cDNA using the SuperScript TM II reverse transcriptase protocol (Invitrogen) following the manufacturer's instructions. The resulting cDNA was subjected to PCR analysis. PCR cycling parameters were as follows: 94°C for 2 min for one cycle followed by 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s for 35 cycles and 72°C for 2 min. The PCR products were stained with SYBR Safe TM (Invitrogen). A sample without RNA was always negative; neonatal rat brain RNA was used as a positive control for the products (results not shown). The actual sequences of specific primers are as follows: Nr4a1/Nur77 RNA Interference-Constructs for RNA interference experiments were obtained into the pSUPER.retro.puro plasmid (OligoEngine) using specific oligonucleotides of the Nurr1 and Creb sequence and the scrambled sequence of the oligonucleotide of Nurr1_1, which was used as a control and does not recognize any rat coding sequence, indicated by capital letter, as follows: shRNA Nurr1_1 (forward) gatccccCCAT-TAAGGTAGAAGACATttcaagagaATGTCTTCTACCTT-AATGGttttt, (reverse) agctaaaaaCCATTAAGGTAGAAG-ACATtctcttgaaATGTCTTCTACCTTAATGGggg; shRNA Creb_1 (forward) gatccccCTGAAGAAGCAGCACGAAAtt-caagagaTTTCGTGCTGCTTCTTCAGttttt, (reverse) agcta-aaaaCTGAAGAAGCAGCACGAAAtctcttgaaTTTCGTGC-TGCTTCTTCAGggg; shRNA Creb_2 (forward) gatccccGC-AAGAGAATGTCGTAGAAttcaagagaTTCTACGACATTCT-CTTGCttttt, (reverse) agctaaaaaGCAAGAGAATGTCGTAGA-AtctcttgaaTTCTACGACATTCTCTTGCggg; shRNA scNurr1 (forward), gatccccAGCGAAAAACTTTATCGGAttcaagagaTC-CGATAAAGTTTTTCGCTttttt, (reverse) agctaaaaaAGCGAA-AAACTTTATCGGAtctcttgaaTCCGATAAAGTTTTTCGCTggg. Oligonucleotides were obtained from Invitrogen and were cloned between BglII/HindIII sites of pSUPER.retro.puro plasmid. Lentiviral constructs were achieved by digesting EcoRI-ClaI sites from pSUPER-sh to replace H1 promoter with H1-shRNA cassette in pLVTHM.
Lentiviruses were propagated using methods described previously (24) (25). Briefly, human embryonic kidney 293T (HEK293T) cells were seeded at a density of 2.5 ϫ 10 6 cells in 100-mm dishes. The following day, cells were transfected with 20 g of pLVTHM-derived constructs, 15 g of pSPAX2, and 8 g of pMD2G. The transfection was routinely performed by the calcium phosphate transfection method (26). Cells were allowed to produce lentiviruses for 48 h. After 48 h, the medium was centrifuged at 1200 ϫ g for 5 min, and the supernatant was concentrated at 141,000 ϫ g for 120 min and then resuspended in 50 l of PBS containing 1% BSA and stored at Ϫ80°C. Biological titers of the viral preparations expressed as a number of transducing units per milliliter were determined by transducing HEK293T cells in limiting dilutions. After 72 h of incubation, the percentage of green fluorescent protein (GFP)-positive cells was determined by cytometer, and viruses were used in the experiments at 1 ϫ 10 6 transducing units/ml. Lentiviral infection was performed at the day of plating. Cells remained in contact with lentivirus for 5 h, and then medium was replaced with Eagle's basal medium and 5 or 25 mM KCl supplemented with 10% FBS, 25,000 units of penicillin, and 25 mg of streptomycin. More than 75% of neurons were infected. At 2 DIV, treatments were performed as described before, and chromatin condensation was assessed at 7 DIV, respectively. Cell lysates were obtained 30 h after treatment.
Plasmid Constructs-Rat genomic DNA was used as the template for PCR to amplify the Nurr1 promoter fragment, containing the CREB-response element reported in Ref. 27, with the forward primer A (5Ј-AAGAGCTCGAGAGTTACAGTC-ACGGGA-3Ј) and the reverse primer B (5Ј-AACTCGAGGG-CTTCAGCCGAGTGATTGG-3Ј). The fragment was cloned into the SacI and XhoI sites of pGL3-Basic luciferase reporter vector (pGL3-Basic) to generate plasmid pGL3-Nurr1promoter. The plasmids construct was confirmed by sequencing.
Transfection and Luciferase Assay-CGCs were plated and infected with the indicated lentivirus, and at 1 DIV, they were transfected using Lipofectamine 2000 (Invitrogen) with pGL3-Basic or pGL3-Nurr1promoter. For luciferase assays, 0.75 g of each pGL3 vector and 0.25 g of Renilla plasmid were transfected in each 24-well plate well.
At 2 DIV, cells were treated with NMDA (100 M). 6 h after treatment, cells were washed twice with cold PBS and were lysed with 100 l of lysis buffer (Promega). 30 l of cell extract were used for a luciferase reporter assay (Dual-Luciferase reporter assay system, Promega). Data are given as mean Ϯ S.E. of values obtained in three independent experiments performed in duplicate.
Statistical Analysis-Statistical significance was determined by one-way analysis of variance followed by Tukey's multiple comparison test. A value of p Ͻ 0.05 was considered as statistically significant.

NMDA-mediated Changes in Gene Expression-NMDA
receptor stimulation rescues neurons from apoptosis through activation of several pathways that would finally modulate the activity of transcription factors (13,28,29). To analyze changes in gene expression induced by NMDA in CGC cultures, we extracted RNA from K5 cultures treated or not with the glutamatergic agonist for 3 or 8 h at 2 DIV, and a gene expression analysis was performed with the Affymetrix platform rat genome 430 2.0 array. Expression data were obtained using GCOS software and normalized with robust multichip average. Statistical analysis was done with the LIMMA package to identify up-and down-regulated genes versus control conditions (K5). We considered that a gene was up-regulated or down-regulated if its expression was 1.5ϫ larger or smaller than in control condition (false discovery rate Ͻ0.2).
After exposure to NMDA, we detected 44 up-regulated genes and 7 down-regulated genes at 3 h ( Fig. 1A and supplemental Table I). 188 up-regulated genes and 112 down-regulated were detected at 8 h ( Fig. 1A and supplemental Table II). Moreover, we observed that almost 50% of the genes up-regulated at 3 h were still up-regulated at 8 h and that six of the seven downregulated genes at 3 h had also a reduced expression at 8 h (supplemental Tables I and II).
To determine the meaning of the expression changes, we performed a functional annotation analysis of the data obtained at 3 and 8 h using the DAVID program (data not shown, Fig. 1B and supplemental Table III). At 3 h, the most enriched functions and categories were related to transcriptional activity and nuclear localization. By contrast, at 8 h, the expression changes reflected a more complex situation, having an enrichment of functions such as neurogenesis,  Table III). All of them were in agreement with the category of signal transduction and cell communication that got the higher enrichment scores (Fig. 1B and supplemental  Table III).
Zhang et al. (30) performed an exhaustive study of the gene expression changes in response to NMDAR stimulation comparing the profile between the protective and the excitotoxic stimulation in hippocampal neurons. We compared their results with our list of genes up-regulated at 3 h, and we obtained a list of candidates to participate in the NMDA prosurvival effect in our experimental model (Table l). From this list, we chose the Nr4a subfamily members because of their higher increase in expression, their potential role in apoptosis, and their unknown function in NMDA-mediated neuronal survival during cerebellum development.
Nurr1 Is Involved in NMDA-mediated Neuronal Survival-The observed increase in gene expression of the Nr4a subfamily members was further confirmed by RT-PCR. An increase in Nr4a1, Nr4a2, and Nr4a3 was detected from 1 to 24 h after treatment with NMDA or K25 ( Fig. 2A). However, when the protein levels of the different members of Nr4a family were determined, only Nr4a2/Nurr1 protein levels showed a clear and consistent increase in response to NMDA or K25. Nr4a1/Nur77 and Nr4a3/NOR-1 protein levels were not increased. In fact, a clear decrease in Nr4a1 and Nr4a3 was observed between 8 and 24 h after NMDA or K25 treatment (Fig. 2B).
To further confirm the observed induction of Nurr1 after NMDA treatment, we decided to analyze the activation of its promoter in response to NMDA or K25 treatment. After 6 h of being treated, CGCs were subjected to a luciferase assay with a plasmid containing the Nurr1 promoter (pGL3-Nurr1promoter). Granule neurons presented an increase of Nurr1 promoter activity in the presence of NMDA and K25 versus K5 (Fig. 2C).
Next, we analyzed the participation of Nurr1 in the effect of NMDA on neuronal survival by silencing its expression by shRNA. CGCs were plated in the presence of lentivirus vectors containing the shRNA against Nurr1 (Nurr1_1) sequence or shRNA containing a scrambled Nurr1 sequence (scNurr1) as a control. As shown in Fig. 3A, the presence of Nurr1_1 shRNA reduces NMDA-mediated increase in Nurr1 protein levels. Moreover, the prosurvival effect of NMDA was reduced in the presence of Nurr1_1 shRNA when either cell viability or the number of apoptotic nuclei was determined (Fig. 3, B and C). By contrast, the presence of the Nurr1_1 shRNA did not block the prosurvival effect of K25 (Fig. 3C).
CREB Regulates NMDA-mediated Induction of Nurr1-It is widely supported that CREB is an important transcription factor in neuronal survival signaling. Moreover, it is also known that NMDA activates CREB and that its activation is necessary for NMDA-mediated neuronal survival (28,29). Our results confirmed that in our experimental system, NMDA activates CREB. As shown in Fig. 4A, treatment of CGC cultures with NMDA produced a time-dependent increase in the phosphorylation levels of phospho-Ser 133 -CREB. The increase was evident 1 h after NMDA treatment and reached its higher significance at 8 h (Fig. 4A). We analyzed whether this activation was accompanied with a higher activity of CREB on Nurr1 promoter by performing a chromatin immunoprecipitation assay. K5 cultures were treated at 2 DIV with NMDA or K25, and 90 min after, cells were fixed to promote the cross-linking of DNA and proteins. A co-immunoprecipitation of the soluble chromatin was performed with phospho-Ser 133 -CREB antibody. The products of de-cross-linking were subjected to Q-RT-PCR with the primers flanking the Nurr1 promoter that contains the Cre binding site. We observed a clear increase in CREB binding to the Nurr1 promoter in the presence of NMDA. An increase in the binding was observed neither in K5 nor in K25 cultures (Fig. 4B). Moreover, we assessed the promoter activity in the presence of two CREB shRNA (creb_1 and creb_2). As shown in Fig. 4C, both shRNA were able to reduce CREB levels, with the creb_1 shRNA having a greater effect. Accordingly, both shRNA inhibited NMDA-mediated activation of Nurr1 promoter, with the creb_1 shRNA having a greater effect reducing the NMDA-mediated increase in Nurr1 protein levels than the creb_2 shRNA (Fig. 4, D and E). Blockade of Creb expression by CREB shRNA did not modify the activity associated with Nurr1 promoter and Nurr1 protein levels in K5 cultures (data not shown).
Bdnf Is Regulated by Nurr1 in Response to NMDA-Several genes have been reported to be regulated by the transcription factor Nurr1. Most of them are related to the maturation and maintenance of dopaminergic neurons. However, a few, such as BDNF, vasoactive intestinal peptide (VIP), or ␣-synuclein, have been described to be involved in CGC survival (31)(32)(33). Thus, we decided to explore whether one of these Nurr1 targeted genes was related to the prosurvival effect of NMDA on CGCs. First, we analyzed their protein levels in CGCs treated with NMDA or K25 in comparison with K5 conditions. We did not  MARCH 30, 2012 • VOLUME 287 • NUMBER 14 JOURNAL OF BIOLOGICAL CHEMISTRY 11357 find any differences in VIP and ␣-synuclein when their protein levels were compared in the presence or absence of NMDA (data not shown). However, we observed that although BDNF protein levels were barely detectable in K5, NMDA dramatically increased BDNF levels (Fig. 5A). This fact was in accordance with the NMDA-mediated increase in Bdnf expression observed at 3 and 8 h with the Affymetrix array platform (supplemental Tables I and II). Activity-dependent Bdnf up-regula-tion has been attributed to CREB activity (34). To check whether CREB was involved in NMDA-and Nurr1-mediated increase in Bdnf expression, we decided to perform a ChIP assay with the Bdnf promoter IV (35), 90 min and 3 h after NMDA or K25 treatment. We were unable to detect any significant differences at 90 min in K25-treated cells. However, a robust increase in CREB binding to Bdnf promoter IV was observed in K25treated cells after 3 h. However, we did not find differences between NMDA and K5 either at 90 min or at 3 h (supplemental Fig. 1), suggesting that CREB is not the transcription factor directly involved in NMDA-mediated activation of Bdnf expression. Next, we assessed the levels of BDNF in the presence of Nurr1_1 shRNA to know whether NMDA-mediated BDNF induction could be due to Nurr1. As shown in Fig. 5A, reducing Nurr1 levels produced a reduction in BDNF protein levels, suggesting that Nurr1 is directly involved in NMDAmediated induction of Bdnf expression in CGC cultures. This suggestion was confirmed when we assessed Nurr1 binding to Bdnf promoter IV by ChIP assay. Our results show that Nurr1 binding to Bdnf promoter IV is clearly increased in NMDAtreated cells. No changes in Nurr1 binding were observed in K25-treated cells (Fig. 5B). Moreover, NMDA-mediated Nurr1 binding to Bdnf promoter seems to be selective because no binding was observed to other Bdnf promoters such as promoter III (Fig. 5B).

Nurr1 Mediates CGCs Survival by NMDA
Next, we performed a first approximation of the in vivo situation by determining the protein levels of Nurr1 and BDNF by Western blot in cerebellum lysates at different days after birth. We observed that both mature BDNF and Nurr1 protein levels increased in the cerebellum during the second postnatal week in a similar way (Fig. 5C). We went deeper in the expression analysis of Nurr1 by performing an immunohistochemistry analysis to know which cells were expressing Nurr1 and whether its expression was homogeneous in all lobes of the cerebellum. We detected Nurr1 mainly in the internal granule layer. Moreover, the expression of Nurr1 was more evident in the nodulus (lobe X) than in the other lobes of the cerebellum ( Fig. 6 and supplemental Fig. 2).

DISCUSSION
NMDAR stimulation is necessary for neuronal survival and development in different cerebral areas. One example of such NMDAR dependence is CGCs. Survival of these neurons during their postnatal migration from the external granule layer toward the internal granule layer (5, 6) requires excitatory inputs from the mossy fibers acting through the NMDARs of CGCs. Failure to receive mossy fibers inputs triggers apoptotic CGC death (10,11,36). In culture, CGCs are also sensitive to NMDAR stimulation for their survival. When CGCs are cultured in nondepolarized culture medium containing K5, the CGCs die by apoptosis (12). However, if NMDA is added to the medium, the CGCs will survive for several days (13).
Some studies have shown that the prosurvival effect of NMDA is mediated by synaptic NMDAR (30,37). It is believed that calcium entry through synaptic NMDAR triggers CREB activation (38) likely by stimulation of calmodulin kinases and/or MEK/ERK kinase pathways (14,39). In CGCs,
Because the intracellular signaling triggered by synaptic NMDAR to promote neuronal survival is still only partially known, we have used DNA microarrays to have a complete picture of the NMDA-mediated differences in the transcriptome. This methodology has allowed us not only to identify single genes that could mediate the prosurvival effect of NMDAR stimulation, but also to identify which functional genetic modules are activated. Our ontological analysis revealed that functional clusters related to neuronal development as well as neurogenesis and antiapoptotic programs were enriched in NMDA-treated cells. Accordingly, we detected upregulation of the ␣1, ␥2, and ␦ subunits of the GABA A receptor and a reduction of the catalytic subunit of calcineurin, postu-lated to be responsible for the maturation blockade of CGCs (44). 6 Several genes, related to the development and/or survival of CGCs, were also up-regulated in NMDA-treated cells, such as Bdnf, neurocalcin, Vgf, and Pcsk1/3 (31,(45)(46)(47)(48)(49). We decided to cross-check the list of genes that were up-regulated in NMDA-treated CGCs at 3 h with the list of up-regulated genes obtained by Zhang et al. (30) after an action potential-bursting condition. From all the genes up-regulated in both conditions (listed in Table 1), we focused our interest on the members of the Nr4a subfamily because their role in cerebellum development and NMDA-mediated neuroprotection was barely unknown. Only a previous study had suggested that Nur77 protects hippocampal neurons in culture from staurosporine and growth factor removal-induced toxicity in vitro and protects hippocampal neurons from kainate-induced toxicity in vivo (30). RT-PCR analysis confirmed the increase in the expression of all the Nr4a subfamily members. However, only Nurr1 pro-6 B. Barneda-Zahonero and J. Rodríguez-Alvarez, unpublished data. FIGURE 6. Nurr1 expression is mainly restricted to nodulus lobe. Sagittal sections of cerebellum from P9, P14, and P21 rats were obtained and subjected to Nurr1 immunohistochemistry. Nurr1-increased expression during the different postnatal days was clearly evident in the nodulus lobe. Representative pictures of each age are shown (10ϫ, 20ϫ, 40ϫ, and a reconstruction of whole cerebellum of each age performed with 2ϫ photographs). The squares indicate the magnified area (nodulus), and the arrows indicate cells positive to Nurr1 staining.

TABLE 1 Up-regulated genes in action potential bursting and at 3 h after NMDA treatment
A comparative analysis between the up-regulated genes at 3 h after NMDA treatment in CGC culture and action potential-bursting stimulation for 2 h in hippocampal cultures (30) shows a list of CREB-dependent genes. The symbols indicate: 1 Ͼ1-fold change and Ͻ3-fold change; 11, Ͼ3-fold change and Ͻ5-fold change; 111, Ͼ5-fold change. tein levels were consistently increased after NMDA treatment. By contrast, Nur77 was down-regulated by NMDA in CGCs, precluding its effect as a prosurvival factor in these cells. Although Nurr1 is involved in the development and differentiation of the midbrain dopaminergic neurons (50) and regulates genes such as tyrosine hydroxylase and dopamine transporter (51), the fact is that some studies have also shown that it controls the transcription of proteins such as the VIP (52); ␣-synuclein, (53)); or BDNF (54), which have been related to CGC survival (31)(32)(33)45). These evidences suggested to us that Nurr1 could be a putative candidate to mediate NMDA prosurvival action. Accordingly, Nurr1 shRNA reduced NMDA protective effect, showing that endogenous Nurr1 is involved in the prosurvival signaling triggered by NMDAR. By contrast, no reduction was observed in the prosurvival effect of K25. As mentioned above, Nurr1 expression is regulated by CREB in non-neuronal and in neuronal cells (27,55,56). Our data show for the first time that CREB binds to Nurr1 promoter in response to NMDA treatment. In addition, reducing CREB levels by shRNA leads to a reduction in the Nurr1 promoter activity as well as a decrease of Nurr1 protein levels in NMDAtreated neurons.
Next, we became interested in identifying the substrate/ substrates of Nurr1 that could be responsible for the prosurvival effect of NMDA. As mentioned above, Nurr1 has been described to regulate the expression of Bdnf in midbrain neurons (54), and it is known that BDNF is important in migration and maintenance of CGCs during cerebellum development (45,57,58). Moreover, BDNF has been suggested to mediate NMDA prosurvival effect on CGCs (13,40,41). All these studies allowed us to hypothesize that Nurr1 could be responsible for NMDA-mediated regulation of Bdnf. Our data support that possibility because: (a) silencing Nurr1 expression by shRNA produced a decrease in BDNF levels in NMDA-treated neurons and (b) NMDA increases the binding of Nurr1 to Bdnf promoter IV. This binding seems to be specific to certain Bdnf promoters because no binding of Nurr1 to Bdnf promoter III was observed. It has been also reported that CREB could interact with the Bdnf promoter in cortical neurons after KCl stimulation (34). Our results indicate that in CGCs, the interaction of CREB with the Bdnf promoter IV is enhanced by KCl stimulation (K25), whereas it is not enhanced by NMDA treatment (supplemental Fig. 1). Thus, we can conclude that in CGCs, NMDAR stimulation triggers CREB-dependent Nurr1 activation that results in a BDNF up-regulation. In this context, we have observed in vivo that Nurr1 and mature BDNF levels increase from P9 to P21, suggesting Nurr1 as a mediator of survival of CGCs during cerebellum development. Additionally, our immunohistochemistry analyses of Nurr1 localization during cerebellum development report for the first time a lobe-specific expression of Nurr1.
In summary, we have identified by whole genome expression profiling which neuronal genes change their expression level when survival of CGCs is triggered by NMDAR stimulation. We have characterized Nurr1 as a key mediator factor in NMDA-dependent survival of CGCs. Nurr1 is a downstream target of CREB activation and is responsible of the increase in mature BDNF expression linked to NMDA prosurvival effect. Moreover, the histological analysis that revealed that Nurr1 restricted expression in the nodulus during cerebellum development opens a possibility to consider Nurr1 as a regulator of cerebellar lobe functional definition.