CAAT/Enhancer-binding Protein δ and cAMP-response Element-binding Protein Mediate Inducible Expression of the Nerve Growth Factor Gene in the Central Nervous System*

Nerve growth factor (NGF) synthesis in the rat cerebral cortex is induced by the β2-adrenergic receptor agonist clenbuterol (CLE). Because NGF is a crucial neurotrophic factor for basal forebrain cholinergic neurons, defining the mechanisms that regulate its transcription is important for developing therapeutic strategies to treat pathologies of these neurons. We previously showed that the transcription factor CCAAT/enhancer-binding protein δ (C/EBPδ) contributes to NGF gene regulation. Here we have further defined the function of C/EBPδ and identified a role for cAMP response element-binding protein (CREB) in NGF transcription. Inhibition of protein kinase A in C6-2B glioma cells suppressed CLE induction of an NGF promoter-reporter construct, whereas overexpression of protein kinase A increased NGF promoter activity, particularly in combination with C/EBPδ. A CRE-like site that binds CREB was identified in the proximal NGF promoter, and C/EBPδ and CREB were found to associate with the NGF promoter in vivo. Deletion of the CRE and/or C/EBP sites reduced CLE responsiveness of the promoter. In addition, ectopic expression of C/EBPδ in combination with CLE treatment increased endogenous NGF mRNA levels in C6-2B cells. C/EBPδ null mice showed complete loss of NGF induction in the cerebral cortex following CLE treatment, demonstrating a critical role for C/EBPδ in regulating β2-adrenergic receptor-mediated NGF expression in vivo. Thus, our findings demonstrate a critical role for C/EBPδ in regional expression of NGF in the brain and implicate CREB in CLE-induced NGF gene transcription.

Alzheimer disease is characterized by devastating changes in plasticity of basal forebrain cholinergic neurons and deficits in memory and other cognitive functions (1). Among the most important questions in Alzheimer disease pathobiology is whether it is possible to restore cholinergic neurotransmission and improve cognition in patients. A significant amount of data indicates that nerve growth factor (NGF) 3 is a trophic factor for experimentally injured cholinergic neurons (reviewed in Ref. 2) and in Alzheimer disease patients (3). Thus, pharmacological compounds that increase the availability of NGF in the brain (4) would provide a useful therapeutic approach for the treatment of Alzheimer's disease, as has been demonstrated for focal ischemia (5)(6)(7).
In the adult brain NGF is produced by cholinergic targets within the hippocampus and neocortex (8,9). Various agents increase NGF expression in these areas, including glucocorticoid hormones (10 -12) and the neurotransmitters glutamate (13,14) and acetylcholine (15). Furthermore, activation of ␤2-adrenergic receptors (BARs) by treatment with the lipophilic agonist, clenbuterol (CLE), also leads to increased levels of NGF mRNA and protein in the rat cerebral cortex (16,17). However, little is known about signal transduction and transcriptional events underlying the neuronal specific activation of the NGF gene. NGF gene transcription is induced by increases in cAMP and activation of the protein kinase A (PKA) signaling pathway (reviewed in Ref. 18). Among transcription factors that are cAMP-inducible, CCAAT/enhancer-binding protein ␦ (C/EBP␦) has been implicated in regulating NGF gene expression in the brain in response to BAR activation (19). C/EBP␦ belongs to a family of six structurally and functionally related transcription factors (C/EBP␣, -␤, -␦, -, -␥, and -) that regulate multiple aspects of cell function, including proliferation, differentiation, stress responses (20), and neuronal development (21). Several C/EBPs are expressed in specific brain structures. For example, C/EBP␣ mRNA is found in the murine hippocampus, cerebellum, and cortex (22), whereas C/EBP␤ is widely expressed in the mouse central nervous system and is a downstream target of NGF receptor activation (23). Additionally, C/EBP␤ and C/EBP␦ expression and DNA-binding activities are enhanced by stimulation of cAMP-dependent signaling pathways in rat hippocampal neurons, and both proteins have been implicated in learning and memory formation (24 -26).
Previous studies suggested that C/EBP␦ may function as a regionspecific regulator of NGF gene transcription in the brain (19). Stimuli known to increase NGF synthesis enhanced C/EBP␦ DNA-binding activity in a glioma cell line and in rat brain, and C/EBP␦ was shown to transactivate a NGF promoter-reporter construct in glioma cells. Using DNase I footprinting analysis, C/EBP␦ was found to interact with a sequence spanning nucleotides Ϫ90 to Ϫ59 of the NGF promoter (19). However, C/EBP␦ alone is probably insufficient to mediate the induction of NGF in response to BAR activation. Indeed, inspection of the NGF promoter sequence revealed a putative cAMP-response element (CRE) residing within the C/EBP␦ footprint (19). The presence of a CRE-like site within the NGF promoter suggests that CRE-binding protein (CREB) may function together with C/EBP␦ to regulate NGF promoter activity. CREB is ubiquitously and constitutively expressed throughout the central nervous system, and CREB activation in the brain in response to cAMP-induced signaling has been well documented (reviewed in Ref. 27). In this study we provide evidence that C/EBP␦ is a central component of NGF transcriptional induction following BAR stimulation. Our results suggest that CREB also participates in the induction of NGF gene transcription.
Transient Transactivation Assay-C6-2B cells were transiently transfected with Fugene6 transfection reagent (Roche Applied Science, Indianapolis, IN). Transfections were carried out using 0.1 g of pNGF promoterreporter constructs, 0.1 g of pcDNA or pcDNA-C/EBP␦ expression constructs, 0.5 g of pPKA-wt, and 0.2 g of pRSV-␤-galactosidase reporter plasmid as an internal control. Twenty-four hours following transfection, cells were washed with phosphate-buffered saline and placed in serum-free medium. Cell lysates were analyzed for luciferase activity 48 h after transfection using the Promega Luciferase Assay System (Promega, Madison, WI). ␤-Galactosidase activity was determined using a luminescent ␤-galactosidase detection kit (BD Biosciences-Clontech, Palo Alto, CA). CLE (Sigma, 1 or 10 M) was included where indicated for the final 6 h of transfection. N- ethyl)-5-isoquinolinesulfonamide hydrochloride (H89, Sigma) treatment (10 M, 6 h) was carried out in similar fashion.
Oligonucleotides were labeled with [ 32 P]dCTP and Klenow DNA polymerase or [␥-32 P]ATP and T4 polynucleotide kinase. 5 g of nuclear extract was incubated with 0.5 ng of probe for 20 min at room temperature in a 25-l reaction containing 10 mM HEPES, pH 7.6, 134 mM NaCl, 4% (w/v) Ficoll, 5% (v/v) glycerol, 1 mM EDTA, 10 mM dithiothreitol, 0.25 g of bovine serum albumin, 0.06% bromphenol blue, 1 g of poly(dI-dC). Protein-DNA complexes were separated on 6% PAGE. For supershift assays, extracts were preincubated with 1 l of pre-immune serum or 1 l of antisera against total CREB (a kind gift from David Ginty) or P-CREB (Ser-133, Upstate Biotechnology) at 4°C for 20 min prior to addition of probe. Binding assays using recombinant CREB-bZIP or C/EBP␦ were carried out in 25-l reactions containing 20 mM HEPES, pH 7.2, 100 mM NaCl, 5% Ficoll, 1 mM EDTA, 10 mM dithiothreitol, 0.25 g of bovine serum albumin, 0.5 g of poly(dI-dC), 0.5 ng of probe. Competition assays included the indicated amounts of unlabeled oligonucleotide.
RNase Protection Assay-RNA was prepared from C6-2B cells using TRIzol reagent (Invitrogen) following transient transfection with pcDNA, pcDNA-C/EBP␦, or pcDNA-C/EBP␤. Where indicated, cells underwent 3-h stimulation with 10 M CLE. RNase protection assay was carried out using the rat neurotrophin probe set, rNT-1 (Multi-Probe RNase Protection Assay System, BD Biosciences). Probe labeling and hybridization were carried out per the manufacturer's instructions. Briefly, 40 g of RNA was hybridized overnight at 56°C with 1 ϫ 10 6 cpm of probe/sample. After hybridization, samples were digested with RNase for 45 min at 30°C and precipitated for 30 min in a dry ice/EtOH bath. Pellets were resuspended in 3 l of loading buffer and analyzed on a 6% polyacrylamide sequencing gel. Densitometric scanning was used to normalize protected fragments against glyceraldehyde-3-phosphate dehydrogenase or L32 internal control probes.
RNase protection assay was used to determine the relative levels of c-Fos mRNA in mouse brain samples (see below). The assay was performed with a 404-base 32 P-labeled mouse c-Fos cRNA as described (19). [ 32 P]Cyclophilin cRNA was used as a reference to correct for RNA loading (19).
Analysis of NGF mRNA Expression in WT and C/EBP Knock-out Mice-C/EBP␦ and C/EBP␤ knock-out mice have been described (24,33). Animals received saline or CLE intraperitoneally and were sacrificed by cervical dislocation at various times after the injection. The brain was removed, and brain areas were dissected on ice, frozen on dry ice, and stored at Ϫ80°C until further processing. NGF mRNA levels in mice were determined by Northern blot analysis as previously described (34,35). In brief, total RNA was extracted and size-fractionated by agarose/formaldehyde gel electrophoresis. RNA was transferred to nylon membranes (Hybond-XL, Amersham Biosciences) and hybridized overnight at 65°C in hybridization buffer (50% formamide, 5ϫ SSC (1ϫ SSC is 0.15 M NaCl/15 mm sodium citrate), 0.2% (w/v) polyvinylpyrrolidone, 20 mM EDTA, pH 8, 1% (w/v) SDS), containing 32 P-labeled NGF cRNA (specific activity, ϳ6 -7 ϫ 10 7 cpm/mg of RNA). The probe contained 721 bases of the rat NGF coding region (36); NGF cDNA was generated by digesting the plasmid with EcoRI. Linearized plasmid was used as a template for in vitro transcription using T3 polymerase (Promega). Blots were washed once for 15 min at room temperature in 2ϫ SSC buffer containing 0.1% SDS, washed three times for 15 min each at 68°C in 0.1% SSC buffer containing 0.1% SDS, and then exposed to x-ray film with Hyperscreen (Amersham Biosciences). After development, blots were stripped and re-hybridized with [ 32 P]cyclophilin cRNA as a standard to control for RNA loading.
Relative levels of NGF mRNA are expressed as arbitrary units and were calculated by measuring the optical density of the NGF mRNA band on the autoradiograph analyzed by a densitometer (Bio-Rad GS-710, Bio-Rad) normalized to cyclophilin, as previously described (19,35). Several exposure times were used to keep the signal intensity in a linear range.

Ectopic Expression of C/EBP␦ in Combination with CLE Treatment
Increases Endogenous NGF mRNA Levels-We previously showed that the lipophilic BAR agonist CLE (37) increases C/EBP␦ expression in vitro and in vivo (19). To examine the role of C/EBP␦ in BAR-mediated activation of the endogenous NGF gene, we analyzed NGF mRNA levels in cells overexpressing C/EBP␦. C6-2B glioma cells were transfected with expression constructs for C/EBP␦ or C/EBP␤, a C/EBP family member that does not transactivate the NGF promoter (19). Following transfection, cells were left untreated or stimulated with CLE for 3 h, RNA was prepared, and NGF transcripts were analyzed by ribonuclease protection assays. In the absence of BAR activation, expression of C/EBP␦ or C/EBP␤ had no effect on NGF mRNA levels (Fig. 1A). How-ever, overexpression of C/EBP␦ combined with CLE stimulation resulted in a 4.7-fold increase in NGF mRNA expression. In contrast, CLE treatment of cells expressing either C/EBP␤ or transfected with the empty vector caused only a 2-fold increase in NGF mRNA accumulation. Western blot analysis of lysates from C/EBP␦ or C/EBP␤ transfected cells demonstrated comparable expression levels under both stimulated and unstimulated conditions (Fig. 1B). These data allow several conclusions. First, C/EBP␦ promotes expression of the endogenous NGF gene. Second, C/EBP␦ activation of NGF expression is specific, because C/EBP␤ had no effect on NGF mRNA levels. Third, C/EBP␦mediated activation of the endogenous NGF gene requires an additional event that is induced by BAR stimulation.
C/EBP␦ and CLE Stimulate Transcription from the NGF Promoter-To further investigate the role of C/EBP␦ in BAR-mediated NGF promoter activity, we performed reporter assays in C6-2B cells transfected with a fragment of the NGF promoter fused to luciferase (pNGF-luc Ϫ615/ϩ50). Treatment of the cells with CLE for 6 h resulted in a 3.5fold increase in reporter activity compared with unstimulated cells (Fig.  2A). Cotransfection of a C/EBP␦ expression vector elicited a similar level of induction. Overexpression of C/EBP␦ combined with CLE treatment yielded a further increase in luciferase activity compared with either alone. Thus, both BAR stimulation and C/EBP␦ overexpression activate the NGF promoter in reporter assays, supporting the idea that C/EBP␦ acts as a downstream effector of BAR signaling in regulating NGF gene transcription.
C/EBP␦ and PKA Cooperate to Activate the NGF Promoter-The data of Fig. 1 suggest that a CLE-induced event is required for C/EBP␦ to transactivate the NGF promoter. BAR stimulation activates adenylyl cyclase, which causes elevation of intracellular cAMP levels and subsequent activation of PKA (38). To examine the possible requirement for PKA activation on CLE-and C/EBP␦-induced NGF promoter activity, we performed transient transactivation assays in the presence of H89, a chemical inhibitor of PKA. H89 modestly decreased C/EBP␦-induced transcription (ϳ40%) but had no significant effect on vector-transfected FIGURE 1. CLE, in combination with C/EBP␦ overexpression, increases endogenous NGF mRNA levels. A, C6-2B cells overexpressing C/EBP␦ or C/EBP␤ were transiently transfected with C/EBP␦, C/EBP␤, or control plasmids, serum-starved overnight, and treated for 3 h with serum-free medium with or without CLE. Total RNA was prepared, and NGF mRNA levels were analyzed by ribonuclease protection assay. Relative levels of NGF mRNA were determined by densitometric scanning of protected fragments and were normalized to glyceraldehyde-3-phosphate dehydrogenase or L32 transcripts. Data are the mean Ϯ S.E. of three independent experiments. *, p Յ 0.02 C/EBP␦ versus C/EBP␦ϩCLE as determined by t test analysis. B, levels of overexpressed C/EBP␦ and C/EBP␤ in transfected cells were confirmed by Western blot analysis. The figure represents two separate blots probed with antibodies against C/EBP␦ and C/EBP␤, respectively. cells (Fig. 2B). H89 also diminished NGF promoter activity in CLEstimulated cells transfected with C/EBP␦ (45% decrease) or the empty vector (35% reduction) (Fig. 2B). Thus, PKA signaling may contribute to NGF gene activation by CLE and C/EBP␦.
To address the relationship between C/EBP␦ and PKA in regulating NGF promoter activation, we tested the effect of overexpressing PKA alone or in combination with C/EBP␦. Transfection of the catalytic subunit of PKA in C6-2B cells elicited similar levels of NGF-luc activity compared with overexpression of C/EBP␦ (Fig. 2C). However, coexpression of PKA and C/EBP␦ caused a further 5-to 6-fold increase in NGF-luc activity compared with PKA or C/EBP␦ alone (Fig. 2C, p Ͻ 0.05). This synergistic activation of the NGF promoter by PKA and C/EBP␦ shows that PKA signaling augments the ability of C/EBP␦ to transactivate the NGF promoter.
To determine whether PKA enhances the intrinsic activity of C/EBP␦, we performed transient transactivation studies using an artificial C/EBP-responsive promoter construct (2xC/EBP-luc) that bears two C/EBP binding sites and lacks any known CRE sequences. C/EBP␦ and PKA stimulated 2xC/EBP promoter activity by 9-and 17-fold, respectively (Fig. 2D). C/EBP␦ and PKA together acted cooperatively, eliciting a 244-fold enhancement of reporter activity. These results indicate that the transactivation function of C/EBP␦ is stimulated by PKA signaling, because we have not observed increased DNA-binding activity when C/EBP␦ is coexpressed with PKA (data not shown).
CREB Binds to a CRE-like Sequence in the Proximal NGF Promoter-PKA stimulation of NGF promoter activity could occur solely through C/EBP␦ or, alternatively, PKA may activate one or more other factors that cooperate with C/EBP␦ to activate transcription. One such candidate is CREB. We previously noted a putative cAMP-response element (CRE), TGACGAGC, in the NGF proximal promoter (19). To investigate whether CREB plays a role in regulating inducible transcription from NGF promoter, we first examined CREB binding to the CRElike sequence. Recombinant CREB protein was incubated with oligonucleotide probes bearing a canonical CRE (Fig. 3A, lane 1) or the NGF CRE-like element (NGF-CRE, lane 2). EMSA revealed that recombinant CREB binds to the NGF-CRE probe, which shares a half-site with the consensus CRE motif (Fig. 4A). CREB binding to the NGF-CRE probe was efficiently competed by 100-fold excesses of either the unlabeled NGF-CRE oligonucleotide, the entire NGF C/EBP␦-CRE region (NGF Ϫ90/Ϫ60), or the consensus CRE sequence (cCREB) (lanes 3-5).
To determine whether endogenous CREB binds to NGF-CRE, C6-2B cells were exposed to CLE or dibutyryl cAMP (Bt 2 cAMP, a cell-permeable analogue of cAMP) for 5 and 15 min. Nuclear extracts were prepared and analyzed by EMSA to assess CREB binding to the NGF-CRE probe. Two DNA-protein complexes were observed in extracts from unstimulated cells (Fig. 3B, lane 1), and binding was not appreciably enhanced by Bt 2 cAMP or CLE treatment (lanes 2-5). To examine whether either binding species contains CREB, extracts were preincubated with anti-CREB or anti-phospho-CREB antibodies. The CREB antibody completely blocked formation of the upper DNA-binding complex (Fig. 3B, lanes 6 -10). In addition, the antibody specific for active, phosphorylated CREB shifted most of the upper complex and caused the appearance of a supershift band whose intensity increased moderately after 5 min treatment with Bt 2 cAMP or CLE (Fig. 3B, lanes  11-15). Collectively, these data demonstrate that CREB binds to the NGF-CRE motif and that CREB activation results in enhanced phos-

FIGURE 2. NGF promoter is activated by CLE or overexpression of C/EBP␦ or PKA. A,
C6-2B cells were transfected with pNGF-luc-615/ϩ50 promoter-reporter construct and either pcDNA control or pcDNA-C/EBP␦ vectors. Luciferase activity was then determined in cells exposed to medium alone or medium containing CLE (6 h). Data are expressed as -fold induction relative to the control (pNGF-luc-615/ϩ50 plus pcDNA) after normalization to ␤-galactosidase activity. Data represent the mean Ϯ S.E. from four independent experiments, each performed in duplicate. B, luciferase data from C6-2B cells transfected with pNGFϪ615/ϩ16 and either pcDNA control or pcDNA-C/EBP␦ expression vectors followed by 6-h treatment with CLE, H89, or CLE plus H89. Data are expressed as the mean Ϯ S.E. of three independent experiments, performed in duplicate. C, activation of the NGFϪ615/ϩ50 promoter by overexpression of C/EBP␦, PKA, or C/EBP␦ and PKA. The -fold induction was determined as described above and represents the mean Ϯ S.E. of three independent experiments performed in duplicate. D, activation of the 2ϫ C/EBP promoter in C6-2B cells by C/EBP␦, PKA, or both. The -fold induction was determined as above from three independent experiments performed in duplicate.

FIGURE 3. CREB binds to the NGF-CRE promoter element.
A, DNA binding assay using recombinant CREB-1 bZIP domain (CREB bZIP) and the NGF-CRE probe. 2 ng of CREB bZIP protein was incubated with 0.5 ng of 32 P-labeled consensus CREB or NGF-CRE oligonucleotides, and the complexes were separated on a 6% polyacrylamide gel. The CREB bZIP protein-DNA complex is indicated by the arrow. Competition assays were carried out by adding 100-fold excess (50 ng) of the indicated unlabeled oligonucleotides. B, C6-2B cells were exposed to control medium, Bt 2 cAMP, or CLE for 5 or 15 min, and nuclear extracts were prepared and tested (5 g) for binding to 32  pho-CREB binding to this site. Thus, CREB may contribute to NGF promoter induction by CLE.
Identification of NGF Promoter Sequences Mediating Activation by C/EBP␦ and CLE-We next sought to identify sequences within the NGF promoter that are necessary for induction by C/EBP␦ and CLE. Candidate sequences include the CRE element and a C/EBP␦ binding site in the Ϫ90/Ϫ60 region previously identified by DNase footprinting (19). We performed transient transactivation assays on a series of 5Ј deletion mutants driving luciferase (shown in Fig. 4A). The reporter constructs did not include the AP-1 site located at ϩ35, because C/EBP␦ transactivation of the promoter is independent of this element (19). Truncation of the NGF promoter to Ϫ100 (NGFϪ100/ϩ16) caused a 7-fold increase in basal promoter activity and a 5-fold augmentation of CLE-induced activity compared with the full-length promoter (Fig. 4B). These data demonstrate the existence of a novel repressive element located within the Ϫ615/Ϫ100 interval. Constructs bearing deletions to Ϫ85 (NGFϪ85/ϩ16) or Ϫ72 (NGFϪ72/ϩ16) also showed increased basal and CLE-induced promoter activity relative to the full-length promoter. Deletion to Ϫ60 (NGFϪ60/ϩ16) reduced this increase in basal activity and diminished CLE responsiveness to less than 2-fold over the unstimulated level (Fig. 4B). Because the CRE is located between Ϫ72 and Ϫ60, these findings suggest that the NGF-CRE element plays an important role in CLE-induced transcription.
We also used the deletion constructs to examine sequence requirements for stimulation by C/EBP␦ and/or PKA (Fig. 4C). As was seen with CLE treatment, the NGFϪ100/ϩ16 construct showed 2-to 3-fold increases in C/EBP␦, PKA, or C/EBP␦ ϩ PKA-induced luciferase activity compared with the Ϫ615/ϩ16 construct. In contrast, activation of NGFϪ85/ϩ16 or NGFϪ72/ϩ16 by C/EBP␦ or PKA was comparable to that of the full-length promoter. Because deletion to Ϫ72 removes the putative C/EBP binding site, it was surprising that responsiveness to C/EBP␦ was not compromised. It is possible that promoter activation by C/EBP␦ in the absence of the putative C/EBP binding site occurs through direct C/EBP␦ binding to the NGF-CRE region, consistent with previous observations that C/EBP␦ has some affinity for CRE-like elements (39). Notably, resection to Ϫ60 abrogated or substantially impaired responses to C/EBP␦, PKA, and C/EBP␦ ϩ PKA (Fig. 4C). These data implicate the NGF-CRE motif as an important element regulating inducible transcription from the NGF promoter.
C/EBP␦ and CREB Bind to the NGF Promoter in Vivo-We next used chromatin immunoprecipitation (ChIP) experiments to determine whether C/EBP␦ and CREB are associated with the NGF promoter in cells. ChIP assays using a C/EBP␦ antibody and either of two CREB antisera demonstrated binding of both transcription factors to the NGF promoter region in C6-2B cells (Fig. 5). Binding to the promoter was similar in the absence or presence of CLE stimulation. Specificity was demonstrated by decreased ChIP signals when blocking peptides were included in the immunoprecipitation reactions or when a negative con- . Sequence requirements for NGF promoter activation by CLE or C/EBP␦. A, diagram of the NGF promoter and endpoints of deletion mutants. Positions of the putative C/EBP and CRE motifs and their homology to consensus sites are indicated. B, activity of NGF promoter deletion constructs in C6-2B cells exposed to medium alone or medium containing CLE for 6 h. C, analysis of the same deletion constructs cotransfected with PKA, C/EBP␦, or PKA plus C/EBP␦. The -fold induction of luciferase activity in B and C represents the activity of each deletion construct (normalized to ␤-galactosidase) relative to the fulllength promoter fragment (NGFϪ615/ϩ16 plus pcDNA). Data are the mean Ϯ S.E. for three independent experiments, performed in duplicate. FIGURE 5. ChIP assays of C/EBP␦ and CREB binding to the NGF promoter in C6-2B cells. Cells were exposed to serum-free medium or medium containing 10 M CLE. Chromatin was immunoprecipitated using the indicated antibodies (C/EBP␦, CREB-1 #1 (Santa Cruz Biotechnology), CREB-1 #2 (polyclonal antiserum provided by D. Ginty), and normal rabbit IgG), and the recovered DNA was analyzed by PCR using primers specific for the indicated genes. Specific binding of antibodies was confirmed by preincubating the antibodies with their respective blocking peptides (BP), where possible, prior to the immunoprecipitation reaction. ␤ 2 -microglobulin (␤ 2 -MG) and IL-6 genes were used as negative and positive controls, respectively. Input refers to the chromatin sample before immunoprecipitation.
trol antibody (IgG) was used. Based on these criteria, positive signals were obtained for C/EBP␦ and CREB binding to the NGF promoter and to a positive control, the IL-6 promoter, which has a well characterized C/EBP site (40). Binding was not apparent using a negative control sequence from the ␤2-microglobulin gene (31). This amplicon exhibited higher background PCR signals that were generally not diminished by the blocking peptides, although a weakly positive signal was suggested with the competing peptide for the CREB #1 antiserum. Collectively, the data of Fig. 5 indicate that C/EBP␦ and CREB are associated with the chromatin-embedded NGF promoter even in the absence of an inducing signal.

Directed Mutation of the C/EBP or CRE-like Sites within the Proximal NGF Promoter Impairs Induction by C/EBP␦ and CLE-To
further demonstrate that the C/EBP and CRE-like motifs are required for NGF promoter activation, we generated luciferase reporter constructs using the Ϫ615/ϩ16 promoter fragment in which the two binding sites were mutated individually or in combination. The mutant constructs were analyzed for induction by CLE, C/EBP␦, and/or PKA. Mutation of either the CRE or C/EBP sites resulted in a ϳ40% reduction in CLE-induced promoter activity (Fig.  6A). However, mutation of both binding sites together did not further diminish promoter activity (Fig. 6B). C/EBP␦-and/or PKAinduced activation of the mutant promoter constructs was decreased to a similar extent (ϳ50%) compared with the WT promoter (Fig.   6C). The reduction in promoter activity when the C/EBP or CRE sites are mutated indicates that both motifs are important for maximal induction of NGF gene transcription.
To confirm that these mutations disrupted binding of CREB and C/EBP␦, we performed competition DNA-binding assays using recombinant proteins and consensus CRE or C/EBP site probes. Binding of each protein to its cognate probe was competed by an excess of WT NGF-CRE or NGF-C/EBP oligonucleotides but not by their respective mutant sequences (Fig. 6D, left and middle panels). Also, 32 P-labeled NGF-CRE probe bound to CREB while the mNGF-CRE probe did not (right panel). Thus, the point mutations effectively eliminate binding of CREB and C/EBP␦. C/EBP␦ Knock-out Mice Are Defective for CLE-induced NGF mRNA Expression in the Cortex-Previous studies using rats showed that intraperitoneal injection of CLE results in 2-to 3-fold increase of NGF mRNA and protein specifically in the rat cerebral cortex (16,17). This increase correlates with elevated levels of C/EBP␦ DNA-binding activity in the same brain region (19). To definitively test the role of C/EBP␦ in NGF transcription in vivo, we examined cortical NGF mRNA levels after CLE treatment of wild-type (WT) or C/EBP␦ knock-out (KO) mice (Fig. 7, upper panel). A slight decrease in NGF mRNA levels was observed in saline-treated C/EBP␦ KO animals compared with agematched WT animals, although this effect was very minor when NGF expression was normalized to cyclophilin levels. Thus, C/EBP␦ FIGURE 6. Mutation of the CRE or C/EBP sites impairs NGF promoter activation. Transient transactivation assays of NGF promoter-reporter constructs bearing CRE and/or C/EBP site mutations. A, promoter-reporter activity of CRE or C/EBP mutant constructs in C6-2B cells with or without CLE stimulation (6 h). B, CLE-induced luciferase activity of the NGF CREϩC/EBP double mutant. C, effect of promoter mutations on responses to PKA, C/EBP␦, or PKA plus C/EBP␦. The -fold induction of luciferase activity in A-C was determined as described in Fig. 3 and is calculated relative to the wild-type unstimulated construct (NGFϪ615/ϩ16 (WT) plus pcDNA). Data are the mean Ϯ S.E. of three independent experiments performed in duplicate. D, EMSA using recombinant CREB and C/EBP␦. Binding of CREB-bZIP to a consensus CRE probe (cCRE) was challenged with unlabeled WT or mutant NGF-CRE oligonucleotides or cCRE (left panel). A similar experiment using recombinant full-length C/EBP␦ protein and cC/EBP probe is shown in the middle panel. Comp., competitor. In the right panel, labeled WT and mutant NGF-CRE probes were analyzed for binding to CREB-bZIP.
is not required for basal NGF expression in the cortex. We also measured normalized NGF mRNA levels in WT and mutant animals at various times following CLE treatment (Fig. 7, lower panel). CLE elicited a time-dependent increase in NGF mRNA in WT mice beginning at 5 h, which returned to basal levels after 18 h. In contrast, no induction of NGF mRNA was observed for C/EBP␦ KO animals at any time examined. CLE-induced BAR signaling in C/EBP␦ KO animals was confirmed by RNase protection analysis of c-Fos mRNA, which is also induced by CLE treatment (19). CLE caused a 2-fold increase in cortical c-Fos mRNA expression in both WT and C/EBP␦ KO mice within 1 h after treatment (data not shown). Thus, C/EBP␦ deficiency does not cause a general impairment of BAR activation or downstream signaling.
To examine whether other C/EBP family members might contribute to NGF regulation, we performed a similar experiment using C/EBP␤ null mice. CLE caused a ϳ2.5-fold increase in NGF mRNA levels in both WT and C/EBP␤ KO animals (Fig. 8). This result, together with the complete lack of NGF induction observed for C/EBP␦ null mice, demonstrates that C/EBP␦ selectively regulates NGF gene expression in vivo.

DISCUSSION
Although C/EBP protein function has been studied extensively in a variety of tissues such as liver and adipose, much less is known about the role of these transcription factors in modulating neurotransmitter-induced changes in gene expression in the CNS. In the present study we have further defined the role of C/EBP␦ in regulating neurotrophin expression by demonstrating that C/EBP␦ is a critical component of BAR-induced NGF gene transcription. Our observations in C6-2B glioma cells and, more importantly, in vivo using C/EBP␦ null mice, establish C/EBP␦ as an essential transcription factor in cAMP-mediated regulation of NGF expression. We show that C/EBP␦ is specific, because C/EBP␤, a closely related family member whose expression is also increased by CLE stimulation (19), does not affect NGF mRNA induction in vitro or in vivo.
Our results also support a role for CREB, operating in conjunction with C/EBP␦, to achieve full promoter activation. Transcription from an NGF reporter construct was synergistically induced by PKA and C/EBP␦, suggesting that C/EBP␦ and CREB may function together to induce NGF expression. EMSA studies demonstrate that activated CREB binds to the CRE-like element in vitro, and CREB is associated with the NGF promoter in vivo. Additional experiments have ruled out binding of two other CREB/ATF family members, activating transcription factors 1 and 2 (ATF1 and ATF2), to the NGF-CRE probe (data not shown), indicating that CREB mediates transcriptional activation via this cis-regulatory element.
CREB acts synergistically with other transcription factors to activate the brain-derived neurotrophic factor gene during neuronal survival and adaptive responses (41,42). Interestingly, we have observed increases in endogenous brain-derived neurotrophic factor expression in C6-2B cells following CLE stimulation and C/EBP␦ overexpression (data not shown). Thus, C/EBP␦ and CREB may function in a combinatorial fashion to regulate neurotransmitter-induced transcription of neurotrophin genes through BAR signaling.
Another explanation for enhanced NGF promoter activity in response to PKA is direct activation of C/EBP␦. Phosphorylation and activation of C/EBP␦ by PKA has been proposed previously, although the site of modification was not identified (43). Our observations tend to support PKA regulation of C/EBP␦ activity, because PKA enhanced the ability of C/EBP␦ to transactivate an artificial C/EBP-responsive reporter construct. Thus, we favor a model in which PKA signaling stimulates NGF transcription through modification and activation of both C/EBP␦ and CREB. It remains to be determined whether collaboration of these proteins occurs via direct physical interaction or by some other means. Physical interaction of CREB with another C/EBP family member, C/EBP␤, has been demonstrated for regulation of the c-Fos promoter in C6 glioma cells (44), supporting the possibility of a direct association mechanism. Additionally, recent studies demonstrate interaction between C/EBP␦ and the coactivator, CREB-binding protein, in osteoblasts. This observation suggests a possible mechanism involving CREB-binding protein as a tether that binds C/EBP␦ and CREB and facilitates their interaction (43).
Interestingly, 5Ј deletions that remove the putative C/EBP site in the proximal NGF promoter did not impact promoter activity in transient transactivation assays. In contrast, deletion to Ϫ60, which also elimi- FIGURE 7. CLE fails to induce NGF mRNA in the cortex of C/EBP␦-deficient mice. C/EBP␦ null mice and age-matched WT animals received an intraperitoneal injection of either CLE or saline (sal) and were sacrificed at the indicated times. Cerebral cortex was dissected, and RNA was extracted and analyzed for NGF mRNA by Northern blot. The upper panel shows a representative Northern blot of NGF mRNA in the cerebral cortex 5 h after CLE or saline injection. The blot was first hybridized with 32 P-labeled cRNA NGF probe, washed, and exposed to x-ray film for 72 h. The blot was then stripped and reprobed with 32 P-labeled cyclophilin (Cyc) cRNA probe and re-exposed to x-ray film for 9 h. NGF mRNA levels were calculated by densitometric analysis of the NGF mRNA band normalized to cyclophilin mRNA. Lower panel, time course of NGF mRNA expression after CLE injection. Data are the mean Ϯ S.E. (normalized to cyclophilin) of three independent samples for each time point. FIGURE 8. CLE increases cortical NGF mRNA levels in C/EBP␤ ؊/؊ mice. Adult (3-month-old) WT and C/EBP␤ KO mice received saline or CLE and were sacrificed 5 h after the injection. Cortical RNA was extracted, and NGF mRNA was determined by Northern blot analysis (upper panel). Relative levels of NGF mRNA were determined by densitometric scanning and normalization to cyclophilin mRNA, as described in Fig. 7. Lower panel, data are the mean Ϯ S.D. of three independent samples.
nates the CRE site, abolished promoter activation by CLE, C/EBP␦, PKA, or combined C/EBP␦ and PKA overexpression. Similarly, point mutations that disrupt both sites reduced but did not entirely abolish promoter activity. Several explanations could account for these results. Recombinant C/EBP␦ binding to the NGF promoter protects the region spanning Ϫ90/Ϫ59 (19). However, the putative C/EBP binding site within this region (Fig. 4A) is a non-canonical sequence and does not bind C/EBP␦ with high affinity. Therefore, although this sequence appears to be important for full promoter activity, additional C/EBP binding sites may exist within the Ϫ615/ϩ16 region. Further studies are required to identify these putative sites. Alternatively, C/EBP␦-mediated promoter activation could occur indirectly through binding of C/EBP␦ to CREB, thereby enhancing the transactivation potential of CREB. Finally, the CRE-like element is located within the C/EBP␦ footprint, raising the possibility that C/EBP␦ has some affinity for the CRE sequence and promotes transcription via this cis element. Additional transcription factors may also be involved; however, in vivo studies demonstrate an absolute requirement for C/EBP␦ in CLE-mediated NGF expression (Fig. 7). C/EBP␦ has been shown to regulate expression of inflammatory mediators and pro-inflammatory cytokines (45)(46)(47), and the importance of CREB family members in promoting neuronal survival following oxidative stress, neurotransmitter toxicity, and ischemic tolerance has been well documented (48 -50). Moreover, studies using animals lacking CREB or its related family member, CRE modulator, have established CREB-dependent transcription as a critical feature of neuronal survival pathways (51). We have identified C/EBP␦ and CREB as key regulators of cellular responses to BAR stimulation. Thus, while C/EBP␦ and CREB can participate in controlling inflammation in the brain, they may also be involved in the regulation of neuronal plasticity. Indeed, it has been shown that activation of BAR by norepinephrine affects the cellular mechanisms typically associated with various aspects of learning and memory (52)(53)(54). Thus, we propose that the functions of C/EBP␦ and CREB in activating NGF gene expression represent one means of modulating neuronal plasticity in response to neurotransmitters. It is now important to establish whether noradrenergic regulation of NGF biosynthesis is a viable strategy to prevent cell death at early stages of chronic neurodegenerative diseases.