The cAMP Pathway in Combination with BMP2 Regulates Phox2a Transcription via cAMP Response Element Binding Sites*

Combined BMP2 and cAMP signaling induces the catechola-minergic lineage in neural crest (NC) cultures by increasing expression of the proneural transcription factor Phox2a, in a cAMP response element (CRE)-binding protein (CREB)-mediated mechanism. To determine whether CREB acts directly on Phox2a transcription induced by BMP2+cAMP-elevating agent IBMX, transient transfections of hPhox2a-reporter constructs were performed in avian NC cultures and murine, catecholaminergic CAD cells. Although BMP2+IBMX increased endogenous Phox2a expression, the 7.5-kb hPhox2a reporters expressing either luciferase or DsRed1-E5 fluorescent protein were unresponsive to BMP2+IBMX, but active in both cell types. Cell sorting of fluorescence-positive NC cells expressing the 7.5-kb hPhox2a fluorescent timer reporter differentiated to equal numbers of catecholaminergic cells as fluorescence-negative cells, suggesting inappropriate transcription from the transfected hPhox2a promoter. NC or CAD cells treated with histone deacetylase inhibitor trichostatin A and BMP2+IBMX display increased endogenous Phox2a transcription and prolonged CREB phosphorylation, indicating Phox2a chromatin remodeling is linked to CREB activation. Chromatin immunoprecipitations employing CREB, CREB-binding protein, and acetylated H4 antibodies identified two CRE half-sites at -5.5 kb in the murine Phox2a promoter, which is also conserved in the human promoter. Proximal to the CRE half-sites, within a 170-bp region, are E-box and CCAAT binding sites, also conserved in mouse and human genes. This 170-bp promoter region confers cAMP, BMP2, and enhanced BMP2+cAMP regulation to Phox2a-luciferase reporters. We conclude these CREs are functional, with CREB directly activating Phox2a transcription. Because the E-box binds bHLH proteins like ASH1 induced in NC cells by BMP2, we propose this novel 170-bp cis-acting element is a composite site, mediating the synergistic regulation by BMP2+cAMP on Phox2a transcription.

Pluripotent neural crest (NC) 3 cells derived from the neuroepithelium during neural tube closure migrate along defined routes in the developing embryo and differentiate to diverse cell types (1). NC cells from the trunk region of the neural tube differentiate to the catecholaminergic, sympathoadrenal (SA) lineage, including sympathetic neurons and chromaffin cells of the adrenal medulla (2). SA lineage cells synthesize catecholamines (CA) by expressing tyrosine hydroxylase (TH), the rate-limiting enzyme in CA biosynthesis, and dopamine-␤hydroxylase (DBH), which converts dopamine to norepinephrine (2).
Bone morphogenetic proteins (BMP2, -4, and -7) (3)(4)(5) and cAMPelevating agents (6,7) promote SA lineage development in NC cultures. cAMP signaling in synergy with BMP2 induces development of the SA lineage by increasing the expression of the homeodomain transcription factor Phox2a (7) in a CREB-mediated mechanism (8). cAMP signaling also regulates the transactivation potential of Phox2a (8), consistent with studies by Lo et al. (6). However, whether cAMP signaling directly regulates transcription of the Phox2a gene remains to be determined. Although the regulation of the human hPhox2a promoter has been studied in human neuroblastoma cell lines expressing the endogenous Phox2a gene (9), the direct transcriptional regulation of the Phox2a promoter by BMP2ϩcAMP signaling has not been investigated.
The CNS-derived catecholaminergic Cath.a cell line (22) provides a cellular model for comparative studies of NC-derived and CNS-derived catecholaminergic neurons. The Cath.a cell line derived from brain tumors of transgenic mice expressing the SV40 T-antigen under control of the TH promoter (22), like the catecholaminergic SA cells, expresses TH and DBH, synthesizes catecholamines, and also develops neurites. The CAD cell line used herein is a variant of Cath.a cells that undergoes neuronal differentiation by serum withdrawal (23) or by BMP2ϩcAMP costimulation (24).
Herein, Phox2a promoter activity is examined following costimulation with BMP2ϩcAMP, employing two differentiation models of catecholaminergic neurons, namely, NC-derived SA cells and CNS-derived CAD cells. We report the identification of a novel, composite cis-acting element, located 5.5-kb upstream from the transcriptional start site of the murine Phox2a gene. This regulatory element is comprised of two functional CRE half-sites in proximity to putative E-box and CCAAT binding sites. The identification of the functional CRE sites conclusively demonstrates the direct transcriptional involvement of the cAMP pathway in Phox2a gene regulation.

EXPERIMENTAL PROCEDURES
Culture Reagents-Vitrogen 100 was purchased from Collagen Corp.; bovine fibronectin and IBMX from Sigma; trichostatin A (TSA) from A. G. Scientific Inc. Calcium-and magnesium-free phosphate buffer solution (CMF-PBS) and 0.05% trypsin, 0.53 mM EDTA were purchased from Invitrogen. Recombinant human BMP2 (Wyeth Research) was reconstituted to 10 g/ml as recommended and stored at Ϫ80°C.
NC Cultures-Standard NC culture growth medium containing 10% chicken embryo extract, and primary NC cultures of 47.5 h Japanese quail (Coturnix coturnix) embryos were prepared as previously described (7). NC cells were isolated from primary outgrowths of neural tube explants plated for 42 h on Vitrogen 100-coated culture dishes. Adherent NC cells from the primary neural tube outgrowths were harvested by trypsin-EDTA, resuspended in growth medium (7), and plated in fibronectin-coated dishes at a density of 320 cells/mm 2 resulting in secondary NC cultures. NC cells were allowed to attach for 2 h, and the seeding medium was replaced with 2 ml of growth medium, with or without BMP2 (10 ng/ml) and IBMX (100 M). NC cultures were fed by exchange of 1 ml of growth medium on day 3 after subculture and every other day thereafter.
Western Blot Analysis-Total protein was extracted from NC cultures or CAD cells, grown in 24-well dishes, in radioimmune precipitation assay buffer containing 150 mM NaCl, 2 mM EDTA, 1 mM sodium orthovanadate, 10 g/ml leupeptin, 25 g/ml aprotinin, 1.0% Triton X-100, 50 mM Tris (pH 7.6). Extracts were sonicated on ice for 15 s, and protein concentration was determined by the Bio-Rad protein assay. Total protein (20 g) was analyzed by 10% SDS-PAGE and transferred to nitrocellulose. The membranes were probed with 1 g/ml CREB antibody or 1 g/ml of anti-phospho-CREB (Ser 133 ) (Upstate). Detection was with 1:2,000 dilution of horseradish peroxidase-conjugated anti-rabbit IgG (Vector) using the enhanced chemiluminescence (ECL) detection system (Amersham Biosciences).
Immunocytochemistry-NC cells were fixed for 20 min with 4% paraformaldehyde. Nonspecific background was blocked using PBS containing 10% goat serum for 40 min. Antibody for P-CREB (1:200, Upstate) was applied overnight at 4°C followed by three washes in PBS. Undiluted TH antibody or peripherin antibody (1:200, Chemicon) was applied for 40 min followed by three washes in PBT. A 1:400 dilution of Alexa fluor 488 goat anti-rabbit IgG was applied for 45 min followed by three washes in PBS. For immunofluorescence of CAD cells the following modifications were included: nonspecific background was blocked using PBS containing 1% goat serum. Antibody was added in PBS containing 1% goat serum and applied 1 h at room temperature.
Real-time PCR-Total RNA from NC or CAD cells was extracted with TRIzol (Invitrogen). Total RNA (20 g) was treated with RNasefree DNase I (20 units) in the presence of RNase inhibitor (40 units). cDNA was synthesized using 2.0 g of total RNA, 0.2 g of random hexamers, and Superscript II reverse transcriptase as described (Invitrogen). Oligonucleotide primers (Integrated DNA Technologies) were designed using Omega 2.0 software (Accelrys). PCR reactions were performed in identical triplicates using 2 l of cDNA, 1 l of 20 M F-/Rprimer, and SYBR Green PCR Master Mix as described by the manufacturer (Applied Biosystems), employing an Applied Biosystems Prism 7000 Real-Time PCR system and the accompanying Sequence Detection Software version 1.0. Data analysis was carried out as described by Lee et al. (25), employing 18 S rRNA as the internal control.
Plasmids and Transfections-hPhox2a promoter-Luc reporter constructs (9) were transiently transfected by Fugene 6 (Roche Applied Sciences) according to the manufacturer's instructions, employing 1 g of plasmid DNA for a 6-well NC culture or 2 g for a 6-well of CAD cell culture. Luciferase activity was compared with pGL3-control vector (Promega). CMV-nLacZ plasmid was used to optimize transfections. Several hPhox2a promoter constructs were cloned into the pTimer-1 Vector (BD Biosciences Clontech), encoding the DsRed1-E5 protein, a mutant of Discosoma sp. red fluorescent protein, which changes color over time. Shortly after translation, the protein emits green light; its red fluorophore emerges later, hours after translation. Plasmids were purified by CsCl 2 equilibrium gradient centrifugation. Cells were harvested 48 h after transfection, lysed in 0.2 ml of lysis buffer (Promega) and assayed for luciferase activity (Luciferase Assay system, Promega). Luciferase activity was normalized per microgram of protein extract. Cells transfected with the pTimer-1 reporter constructs were assayed by fluorescence microscopy and flow cytometry. Luciferase reporter plasmids were constructed by cloning the mouse Phox2a upstream element (UE and mtUE) spanning nucleotides Ϫ5754 to Ϫ5340 in the hPhox2a Ϫ32Luc and Ϫ515-Luc vectors. The resulting vectors were transiently transfected in NC or CAD cells as described earlier.
Cell Sorting by Flow Cytometry-NC cells were transfected with 1 g of hPhox2a-Timer-1 plasmid as described earlier, at 24 h of primary NC outgrowths. At 42 h of primary NC outgrowths, neural tubes were removed; NC cells were then replated and treated with BMP2ϩIBMX. After 24 h, NC cells were observed every day for 3 days by fluorescence microscopy. Fluorescence-positive cells were quantified by cell sorting using the Epics Altra flow cytometer (Beckman-Coulter).

RESULTS
In avian NC cultures, moderate activation of cAMP signaling acts synergistically with BMP2 to promote development of the SA lineage (7), by increasing Phox2a and Phox2b transcription via activation and involvement of the transcription factor CREB (8), the downstream effector of the cAMP pathway (27,28).
To directly demonstrate the mechanism of the combined BMP2 and cAMP signaling on Phox2a gene transcription, we employed the Phox2a promoter and its deletion constructs in transient luciferase reporter assays in avian NC cells. It is well established from studies of developmentally regulated and differentiation-specific genes, that functionally important cis-acting elements are conserved across species (29 -33). Accordingly, we employed the human Phox2a promoter to investigate its regulation in the developmentally regulated avian primary NC culture model system. Earlier studies by Hong et al. (9) have reported the cloning of the hPhox2a promoter and its characterization, studied in the human neuroblastoma SK-N-BE (2)M17 and SK-N-BE (2)C Phox2aexpressing cell lines. However, Phox2a regulation in response to BMP2 and cAMP signaling has not been examined.
Employing the Fugene method of transfection in primary NC cultures, we initially established that 40% transfection efficiency was achieved with the CMV-␤-gal reporter, without negative growth effects on the primary culture (data not shown). The hPhox2a-Luc reporter constructs shown in Fig. 1A, spanning sequences from 32 bp to 7.5 kb upstream from the transcriptional start site were transiently transfected in avian NC cells, as a function of cotreatment with BMP2ϩIBMX (a cAMP-elevating agent). The activity of the hPhox2a promoter tested in the avian NC system is qualitatively identical to that reported by Hong et al. (9). Specifically, the highest activity is observed with the 1.3-kb promoter fragment. These results suggest the Phox2a promoter activity is likely conserved between mammalian and avian species.
The Activity of hPhox2a Promoter Differs from the Endogenous Phox2a Gene in NC Cultures-Interestingly, the transient transfection results do not demonstrate regulation of the hPhox2a promoter constructs by BMP2 and cAMP in NC cells (Fig. 1A). In our previous studies we have shown that cotreatment of NC cells with BMP2 and cAMP synergistically induces expression of the endogenous Phox2a gene (7,8), as well as of TH mRNA (Fig. 1B). Thus, there appears to be a discrepancy between the in vivo regulation of the Phox2a gene and the transient transfection results of the hPhox2a promoter.
To exclude that the hPhox2a promoter-luciferase reporter constructs might not be suitable for deciphering regulation of the Phox2a promoter (34), the Phox2a promoter was cloned and studied using the fluorescence-timer (pTimer-1) reporter system (35). This reporter monitors both the onset of transcription of a given promoter, by mon-itoring the appearance of green fluorescence, as well as the time of its transcriptional shut off, i.e. when only red fluorescence is observed (35).
Employing the hPhox2a-fluorescence-timer reporter constructs transfected in NC cultures in conjunction with flow cytometry the fluorescence-positive cells were quantified as a function of BMP2ϩIBMX cotreatment ( Fig. 2A). The results demonstrated the promoter is transcriptionally active in the absence of BMP2ϩIBMX costimulation in NC cultures ( Fig. 2A), and that cotreatment with BMP2ϩIBMX does not mediate any further transcriptional induction ( Fig. 2A). Importantly, Phox2a promoter activation starts on day 1 of secondary NC culture (green fluorescence), which is earlier than the known expression pattern of the endogenous Phox2a gene (7), and is turned off by day 2 FIGURE 1. A, hPhox2a promoter activity in J. quail primary NC cells with BMP2 (10 ng/ml) ϩIBMX (100 M) treatment, as indicated. The hPhox2a-Luc reporter constructs (1 g) listed were transfected via the Fugene 6 method into 6-well NC cultures at day 0 of replating of primary NC outgrowths. Cells were harvested 48 h after transfection; luciferase activity normalized per microgram of protein extract is expressed relative to luciferase activity of the control pGL3 plasmid, considered as 100%. The result is from three independent experiments, each performed in triplicate. B, real-time PCR for TH mRNA expression, employing total RNA isolated from NC cultures at days 1-5 following replating and treated with BMP2ϩIBMX, as indicated. Results represent three independent assays, each performed in identical triplicates. FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 (red fluorescence), when the endogenous Phox2a gene is known to be transcribed (8,36).

Functional CRE Sites in Mammalian Phox2a Promoter
Use of the fluorescent reporter allows cell sorting of the transfected, fluorescence-positive NC cells and thus the study of their phenotype and developmental potential. Following live cell sorting, the Phox2a fluorescence-positive and fluorescence-negative NC cells were placed in culture; 2 h after replating 80% of hPhox2a-fluorescence-positive NC cells remained positive. Likewise, no fluorescence was detected with the sorted fluorescence-negative NC cells (Fig. 2B). The sorted NC cells were then grown either under control conditions or with BMP2ϩIBMX cotreatment. After 4 days of culture the NC cells were immunostained with TH antibody to monitor TH protein expression, a marker of SA cell development (7,8). We hypothesized, if the transcriptional activity of the transiently transfected hPhox2a were authentic, these hPhox2a-fluorescence-positive NC cells would be expressing the endogenous Phox2a gene. Because Phox2a expression is required for SA development, we predicted that all of the sorted hPhox2a-fluorescence-positive NC cells would give rise to SA progenitors, i.e. they would be TH immunoreactive. However, following cell sorting and replating, both the hPhox2a fluorescence-positive and -negative NC cells cotreated with BMP2ϩIBMX, resulted in the appearance of similar numbers of THimmunoreactive cells (Fig. 2C). Interestingly, the appearance of melanocytes was also observed in the fluorescence-positive NC cultures (data not shown).
These results suggest the hPhox2a promoter constructs were inappropriately activated in NC cells, other than those that became SA cells, supporting that the expression of the transiently transfected hPhox2a reporters does not reflect the expression of the endogenous Phox2a gene. Taken together with previous results (Fig. 1A), these data demonstrate that the hPhox2a-Luc construct, comprised of 7.5kb of upstream sequence, does not respond similarly to the endogenous promoter, suggesting that its expression is either misregulated or unresponsive to signals known to induce the expression of the endogenous Phox2a mRNA.
Three possibilities account for the observed results. 1) Regulation of the human Phox2a promoter in avian cells differs from regulation in mammalian cells.
2) The chromatin context is an important determinant in the synergistic BMP2ϩIBMX regulation, and 3) cis-acting ele- ments mediating the BMP2 and cAMP regulation reside outside the 7.5-kb hPhox2a promoter fragment.
Regulation of Phox2a Promoter in Murine, Catecholaminergic CAD Cells-To address whether regulation of the Phox2a promoter in avian cells is different from the regulation in mammalian cells, we employed the murine CAD cell line (23), a variant of the CNS-derived catecholaminergic Cath.a cells (22). The Cath.a cell line was established from a brain tumor that arose in a transgenic mouse carrying the wildtype SV40 T antigen (Tag) under the transcriptional control of the TH promoter (22). Neuronal differentiation in CAD cells is initiated by serum deprivation (23) or addition of BMP2ϩIBMX (24).
Employing real-time PCR we determined the expression of Phox2a and TH mRNAs in CAD cells. Phox2a and TH mRNAs display a slight increase 1 day after serum starvation, followed by a gradual decrease (Fig. 3, A and B). Importantly, BMP2ϩIBMX resulted in an obvious increase in Phox2a and TH expression, both in the undifferentiated (day 0 of treatment) and differentiated CAD cells (Fig. 3, A and B) suggesting regulation of the endogenous Phox2a promoter in mammalian cells is similar to avian cells (7,8). Moreover, BMP2ϩIBMX cotreatment induced differentiation of CAD cells in the presence of serum. 4 Accordingly, employing the CAD cellular model system, the activity of the hPhox2a promoter Luc reporter constructs was investigated following transfection, as a function of BMP2ϩIBMX cotreatment. The results demonstrate, BMP2ϩIBMX cotreatment had no effect on the activity of all hPhox2a promoter constructs tested, similar to the results observed in the avian NC cells (Fig. 3C). Thus, we conclude: 1) The hPhox2a promoter responds similarly in the avian NC and murine CAD cells, and 2) transiently transfected hPhox2a-reporter constructs do not respond to the same regulatory mechanism(s) operative on the endogenous Phox2a promoter.
To investigate the second possibility described earlier, namely whether the chromatin context of the Phox2a gene is an important determinant in the synergistic transcription by BMP2ϩIBMX, we examined the effect of the HDAC inhibitor TSA (41), on endogenous Phox2a transcription. NC cells were treated with various doses of TSA for different periods to determine the TSA effect on NC cell growth (data not shown). Treatment with 100 ng/ml TSA for 3-h intervals at different times during day 1 (24 -36 h after re-plating of the primary NC outgrowths) showed no detectable effect on NC cell growth and differentiation. Interestingly, BMP2ϩIBMX cotreatment in the presence of 100 ng/ml TSA increased endogenous Phox2a mRNA expression compared with treatment with only BMP2ϩIBMX, assessed by real-time PCR (Fig. 4A). TSA treatment without BMP2ϩIBMX does not induce Phox2a mRNA expression (Fig. 4A). It is important to note that BMP2ϩIBMX treatment mediates only minimal Phox2a expression at 24 h of NC cultures, assessed by real-time PCR (8).
Similar studies in CAD cells also demonstrated that endogenous Phox2a mRNA expression is further increased by TSA addition in the presence of BMP2ϩIBMX (Fig. 4B). Likewise, addition of TSA in the presence of BMP2ϩIBMX increases CAD cell neuronal differentiation, assessed by peripherin immunostaining and TH immunoreactivity (  RNA isolated from CAD cells treated with BMP2ϩIBMX as indicated, starting at day 0 prior to serum withdrawal (day 1). Quantification is from three independent RNA preparations, each PCR reaction analyzed in identical triplicates, and normalized relative to 18 S rRNA used as the internal control, as described in Lee et al. (25). C, hPhox2a promoter activity in murine CAD cells treated with BMP2 (10 ng/ml) ϩIBMX (100 M). CAD cells grown in serum-containing medium were transfected with indicated hPhox2a-Luc reporter constructs (2 g) and treated with BMP2ϩIBMX, as indicated. Cells were harvested 48 h after transfection; luciferase activity normalized per microgram of protein extract is expressed relative to luciferase activity of pGL3 plasmid, considered as 100%. Results shown are from three independent experiments each performed in triplicates. FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 ( Fig. 4B) or CAD cell neuronal differentiation (Fig. 4C). Accordingly, these data support a mechanism whereby the chromatin structure is an important parameter in the transcriptional regulation of the endogenous Phox2a promoter by BMP2ϩIBMX.

Functional CRE Sites in Mammalian Phox2a Promoter
HDAC Inhibition Prolongs CREB Phosphorylation in NC and CAD Cells-Our recent studies (8) have demonstrated that dominant negative CREB suppresses Phox2a mRNA expression by BMP2ϩIBMX, whereas the constitutively active CREB DIEDML induces Phox2a expression without cAMP signaling. Based on these studies (8) and the effect of TSA on endogenous Phox2a transcription (Fig. 4), a link has been established between Phox2a transcription, CREB activation, and chromatin remodeling.
To further confirm these observations we monitored by Western blot assays the activation of endogenous CREB, following treatment of NC or CAD cells with BMP2ϩIBMX as a function of TSA addition. Earlier studies (42) demonstrated the phosphorylation of CREB is prolonged following inhibition of HDACs. Following BMP2ϩIBMX treatment of NC cells, CREB phosphorylation was detected after 15 min (Fig. 5A). Whereas TSA addition alone had no effect on CREB phosphorylation (data not shown), TSA in combination with BMP2ϩIBMX prolonged the phospho-CREB levels in NC cells for 1-1.5 h, in agreement with the observations by Michael et al. (42). Likewise, nuclear phospho-CREB immunostaining was virtually undetectable after 1.5 h of stimulation with BMP2ϩIBMX treatment only. By comparison, phospho-CREB immunostaining remained elevated in NC cells cotreated with BMP2ϩIBMX in the presence of TSA (Fig. 5C).
Similar studies investigating the effect of TSA on CREB phosphorylation in CAD cells are shown in Fig. 5B. BMP2ϩIBMX treatment of  FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 FIGURE 6. A, diagram illustrates the putative CRE half-sites and their nucleotide position relative to the ϩ1 transcriptional start site, identified by computer analyses of the mouse and human Phox2a genes. B, ChIP assays, employing murine CAD cells treated with BMP2ϩIBMX as a function of TSA addition as indicated, with the CREB antibody (10 g) or IgG (negative control). Quantification of Phox2a DNA immunoprecipitated by CREB antibody or IgG was by real-time PCR employing mouse Phox2a gene-specific primers spanning putative CRE half-sites 1 and 2, site 3, site 4, and site 5. Data are expressed as fold change of CREB binding quantified relative to IgG, and represent the average of at least three independent CAD cells leads to maximal CREB phosphorylation within 1 h, gradually decreasing to the background level by 4 h (Fig. 5B). Importantly, addition of TSA together with BMP2ϩIBMX prolonged the phosphorylation of CREB, sustaining the level of phospho-CREB for 4 h, and returning to the background unphosphorylated level 6 h after treatment. Immunofluorescence studies of phospho-CREB performed in CAD cells, as a function of TSA addition in the presence of BMP2ϩIBMX demonstrate the same effect (data not shown).

Functional CRE Sites in Mammalian Phox2a Promoter
Identification and Characterization of cAMP Response Elements in Murine Phox2a Promoter-Because CREB activation is necessary for BMP2ϩIBMX-induced Phox2a expression (8), and TSA not only enhances Phox2a mRNA expression (Fig. 4), but also prolongs CREB phosphorylation in NC cells (Fig. 5, A and C) and CAD cells (Fig. 5B), we interpret these data to suggest that CREB directly participates in the transcriptional regulation of the Phox2a gene.
To conclusively prove this mechanism, we sought to identify and characterize the cis-acting elements in the Phox2a promoter that respond to cAMP signaling, via CREB. We obtained the Phox2a promoter sequence from the human and mouse genome databanks. Although the chicken genome sequence is also available, the sequence for both the Phox2a gene and promoter are still unavailable. We searched for putative CRE sites in the human and mouse Phox2a promoter sequences. In general, the CRE consists of an 8-bp palindrome (TGACGTCA) typically found within 100 -300 nucleotides upstream of the TATA box. The palindromic CRE can be separated into two CGTCA motifs or half-sites, which may be configured on the same or on opposite strands, to function cooperatively in response to cAMP stimulation (28,43). Computer analyses revealed that completely palindromic CRE sites (TGACGTCA) are not present in the 15 kb of upstream sequence, either in the human or the mouse Phox2a promoters. However, several variant CRE half-sites are found in the Phox2a upstream promoter region as well as within intron 1 (Fig. 6A).
To determine whether CREB interacts with these CRE half-sites, ChIP assays were performed in CAD cells using the CREB antibody; quantification of the associated Phox2a promoter sequence was carried out by real-time PCR, expressed relative to the negative control value obtained in ChIP assays with IgG, performed in parallel. The results demonstrate (Fig. 6B) that CREB binding was detected in all these CRE half-sites in agreement with the studies by Zhang et al. (44). Treatment with BMP2ϩIBMX mediated a statistically significant (p Ͻ 0.05) stimulation of CREB binding at these CRE half-sites, with the exception of CRE site 4. Importantly, the highest level of CREB binding was observed when CAD cells were cotreated with BMP2ϩIBMX and TSA.
To establish that these CRE half-sites mediate Phox2a transcription in response to BMP2 and cAMP signaling, we performed additional ChIP assays using the CBP antibody, investigating whether activated CREB recruits CBP to these CRE sites. Similar to the ChIP results obtained with the CREB antibody (Fig. 6B), CRE half-sites 1, 2, and 5 also display a statistically significant (p Ͻ 0.05) increased association with CBP upon BMP2ϩIBMX addition. This CBP association with the CRE sites of the Phox2a promoter is more pronounced in the presence of BMP2ϩIBMX and TSA cotreatment (Fig. 6C), consistent with the enhanced Phox2a expression (Fig. 4) and prolonged CREB phosphorylation (Fig. 5) detected under these treatment conditions.
To investigate whether the CRE sites of the Phox2a promoter are associated with nucleosomal modifications that reflect the activation state of the promoter, we examined the acetylation status of histone H4 (Fig. 6D). H4 acetylation in regions encompassing the CRE half-sites is increased with BMP2ϩIBMX treatment. Furthermore, as expected, with BMP2ϩIBMX and TSA cotreatment, high H4 acetylation is observed particularly for sites 1, 2, and 3. Because CRE half-sites 1 and 2 demonstrate the most pronounced association with CREB, CBP, and acetylated H4, we conclude that CRE sites 1 and 2 are the major CRE sites participating in Phox2a gene regulation in response to combined BMP2 and cAMP signaling.
To directly demonstrate that CRE half-sites 1 and 2 have the potential to bind CREB, EMSA was performed employing recombinant purified CREB and oligonucleotide probes encompassing the Phox2a CRE halfsite 1 or 2, the wild-type palindromic CRE or a mutant CRE oligonucleotide probe (Fig. 7A). Both CRE half-sites 1 or 2 probes display CREB binding, similar to the wild-type CRE probe. The sequence specificity of the observed CREB-DNA complex was determined by competition assays employing 100-fold molar excess of unlabeled wild type versus the mutant CRE oligonucleotides (Fig. 7A).
To further demonstrate the functional significance of the Phox2a UE encompassing CRE half-sites 1 and 2, in conferring cAMP regulation to the Phox2a promoter, we cloned the fragment spanning nucleotides Ϫ5754 to Ϫ5340 (UE) in the Ϫ32-Luc and Ϫ515-Luc reporters (Fig.  7B). In addition, we cloned in these reporters the mtUE element in which both CRE half-sites have been mutated (Fig. 7B). The Ϫ32-Luc and Ϫ515-Luc reporters lacked responsiveness to BMP2ϩIBMX cotreatment in either NC or CAD cells (Figs. 1A and 3C). Interestingly, transient transfections of these Phox2a reporters containing the upstream Phox2a element (UE) in either NC (Fig. 7C) or CAD (Fig. 7D) cells demonstrate luciferase induction in response to treatment with IBMX or BMP2 alone, and further enhanced expression in response to cotreatment with BMP2ϩIBMX. Importantly, the Phox2a-luciferase reporters containing the mtUE element with mutated CRE half-sites 1 and 2, did not display luciferase expression in response to BMP2ϩIBMX treatment in NC (Fig. 7C) or CAD cells (Fig. 7D), confirming the functional significance of the CRE half-sites in Phox2a transcription. These results conclusively demonstrate the functional significance of the Phox2a UE in mediating the transcriptional regulation of the Phox2a promoter by cAMP signaling.

DISCUSSION
SA lineage development in cultures of primary NC cells is enhanced by the combined signaling of BMP2 and moderate activation of the cAMP pathway (7). This conclusion is also supported by earlier studies (6, 10) demonstrating, activation of cAMP signaling is required for induction by BMP2 of the SA lineage in NC cultures. Our recent studies (8) have demonstrated the combined BMP2 and cAMP signaling induces transcription of the proneural transcription factor Phox2a. This transcriptional induction of Phox2a requires CREB activation by the cAMP pathway, and CREB-mediated transcription. The mechanism by which cAMP signaling via CREB regulates Phox2a gene expression could be via direct regulation of Phox2a gene transcription, or indirectly, by inducing another transcription factor that in turn, regulates Phox2a transcription.
To determine whether cAMP directly regulates Phox2a gene transcription, we investigated Phox2a promoter activity in the developmentally responsive avian NC cell culture system, and the murine, CNSexperiments (p Ͻ 0.05 where shown). C, CREB recruits CBP at CRE half-sites in mouse Phox2a gene. ChIP assays in CAD cells treated as indicated, using 5 g of CBP antibody and IgG as the negative control. Quantification is by real-time PCR, as described in B. Data are from at least three independent assays. The p value is indicated. D, histone H4 acetylation at Phox2a CRE half-sites. ChIP assays performed with CAD cells treated as indicated, using 1 g of antibody specific for acetylated H4. The amount of H4 acetylation at each CRE half-site was quantified by real-time PCR. Data are expressed as fold change relative to untreated cells, derived from three independent experiments. derived, catecholaminergic CAD cell line (23). Herein, we report the identification within the Phox2a promoter of functional CRE sites that are evolutionarily conserved in the human and mouse promoters. We have identified two functional CRE half-sites located ϳ100 or 250 bp from each other in the 5Ј upstream region of the mouse and the human promoters at 5.5 kb and 10.3 kb, respectively (Fig. 8A). Interestingly, these CRE sites are in proximity to several E-box cis-acting elements, putative binding sites for bHLH transcription factors such as MASH1 (Fig. 8A), suggesting that a composite response element in the Phox2a promoter mediates the combined synergistic effect of BMP2 and cAMP signaling (Fig. 8B).
Transfected 7.5-kb hPhox2a Reporter Responds Differently to Endogenous Phox2a Promoter-Earlier studies by Hong et al. (9) reported the cloning and characterization of the human Phox2a promoter employing human neuroblastoma cell lines expressing the endogenous Phox2a gene. We employed the same hPhox2a constructs and studied them in the developmentally regulated avian NC cellular model system, as a function of BMP2ϩIBMX costimulation. Although employing the avian promoter would have been an ideal approach, we were unable to clone the avian Phox2a promoter. However, as shown by other studies (29 -33) and confirmed in our studies herein (Fig. 8A) functionally important, regulatory cis-acting elements are evolutionarily conserved.
The hPhox2a promoter was studied in NC and CAD cell cultures using either the luciferase or fluorescence-timer reporters. The activity of the hPhox2a promoter studied in the context of the two reporters, and in the context of the avian NC cells versus the murine CAD cell line, is qualitatively similar to results reported by Hong et al. (9). Specifically, the 1.3-kb hPhox2a promoter fragment demonstrates the highest activity in all cellular models tested, i.e. human cell lines (9), avian NC cells, and murine CAD cells. These results support that similar regulatory mechanisms are operative between the avian and mammalian cells.
Furthermore, in contrast to the endogenous Phox2a gene (7,8), none of the hPhox2a constructs, even that comprised of 7.5 kb of upstream sequence, were responsive to the costimulation by BMP2 and IBMX. We conclude that the transfected hPhox2a promoter does not parallel the regulation of the endogenous Phox2a gene in response to the combined BMP2ϩcAMP signaling. These conclusions are further supported by the developmental potential of the NC cells that were sorted based on the hPhox2a-driven expression of the fluorescent reporter (pTimer-1 vector). Both the sorted hPox2a-fluorescence-positive as well as the fluorescence-negative NC cells gave rise to similar numbers of TH-immunoreactive cells, in response to BMP2ϩIBMX. If the activity of the transfected hPhox2a had been authentic, the expression of the endogenous Phox2a gene would have resulted in all hPhox2a-fluorescence-positive cells also being TH immunoreactive, following treatment with BMP2ϩIBMX.
Involvement of Chromatin Remodeling in Phox2a Gene Expression-The importance of chromatin structure in gene regulation is well established (37,38,40,45). However, transiently transfected plasmid templates are thought to assemble incompletely into chromatin, unless it is experimentally demonstrated that they do. Accordingly, we investigated whether the transcription of the endogenous Phox2a gene in NC and CAD cells is modulated by treatment with TSA, an inhibitor of HDACs. The rationale for these studies is based on the work of Michael et al. (42) that demonstrated treatment with the HDAC inhibitor TSA, prolonged the phosphorylation of CREB following activation of the cAMP pathway. In primary NC cultures and the CAD cell line, we demonstrate treatment with the HDAC inhibitor TSA enhanced the BMP2ϩcAMP-mediated expression of the endogenous Phox2a gene, and also prolonged the phosphorylation of CREB. We conclude that inhibition of histone deacetylation increases the BMP2ϩIBMXinduced Phox2a transcription by extending the duration of CREB activation. Because CREB is required for Phox2a transcription (8) these observations support a model linking directly CREB activation to a chromatin-dependent mechanism in Phox2a transcription by BMP2ϩcAMP.
Functional CRE Half-sites in the 5Ј-Upstream Region and Also in the Intron of the Phox2a Gene-We employed computer analyses to address the validity of this proposed model by identifying putative CRE sites located in the 5Ј-flanking region of the human and mouse Phox2a genes as well as in the Phox2a intron 1. We analyzed ϳ15 kb of the upstream sequence. Interestingly, perfectly palindromic CRE sites (TGACGTCA) have not been found in these sequences, and only five imperfect CRE half-sites have been located within the examined 15 kb of the upstream Phox2a sequence and intron 1. Furthermore, none of these CRE halfsites reside within 100 -300 nucleotides from the TATA box where, CRE sites are typically located in CREB-responsive genes. Three of these CRE half-sites are in the upstream 5Ј-flanking region (Fig. 8A).
In murine CAD cells, employing ChIP assays, we demonstrate that these CRE sites: 1) are occupied by CREB in the unstimulated, control conditions in agreement with recent findings (44); 2) display higher CREB binding with BMP2ϩIBMX cotreatment, which is further enhanced by TSA addition, supporting the demonstrated synergy of the BMP2 and cAMP signaling pathways (7,8); 3) display enhanced CBP binding, especially evident upon costimulation with BMP2ϩIBMXϩTSA; CBP is the known, functionally associated CREB co-activator (46); and 4) display enhanced association with acetylated histone 4, a marker of transcriptionally active genes (37,45). These observations derived from the in vivo association of CREB, CBP, and acetylated histone 4 with specific regions of the Phox2a promoter demonstrate the functional significance of the CRE half-sites in Phox2a transcription. Importantly, this analysis is consistent with the conclusions by Cha-Molstad et al. (47) namely, that CREB binding and function is regulated in a cell-specific manner by epigenetic mechanisms, determining active chromatin conformation.
CRE Half-sites 1 and 2 Are the Major Sites Mediating cAMP Regulation of Phox2a Promoter-Because CBP interacts only with activated CREB (46,48), the ChIP assays employing the CBP antibody conclusively demonstrate the functional involvement of the CRE half-sites in the BMP2ϩcAMP-mediated regulation of Phox2a gene transcription. The pronounced association of CBP and acetylated H4 with CRE halfsites 1 and 2, following BMP2ϩIBMXϩTSA costimulation, identify these sites as the major cis-acting elements mediating the combined BMP2ϩcAMP transcriptional regulation of the Phox2a gene. In support of these conclusions we demonstrate that CRE half-sites 1 or 2 bind purified CREB in in vitro DNA binding assays and importantly the upstream Phox2a element (UE) encompassing these CRE half-sites confers cAMP regulation to unresponsive Phox2a promoter constructs. Significantly, the Phox2a gene has been identified as a CREB/cAMPregulated gene in PC12 cells using the SACO method, a new approach for characterizing transcription factor regulatory regions in vivo (43). Thus, these independent observations by Impey et al. (43) further validate our conclusions that cAMP signaling via CREB regulates Phox2a gene transcription.
Concerning sites 3, 4, and 5, the ChIP assays do not identify these sites as major regulators of Phox2a gene expression, although we cannot exclude the possibility that all sites may work together in the overall transcriptional regulation of the Phox2a gene.
Interestingly, further computer sequence analysis of the nucleotide sequence near the CRE half-sites 1 and 2 revealed that they are in prox-imity with other transcription factor binding sites. Of special interest are the E-box (CANNTG) and CCAAT binding sites (Fig. 8A). Intriguingly, each CRE half-site 1 or 2 is located 50 bp from an E-box, and likewise the CCAAT site is 50 bp away from an E-box. The E-box is the putative binding site for bHLH transcription factors such as ASH1. In agreement with this observation the upstream Phox2a element confers enhanced BMP2ϩcAMP responsiveness to previously unresponsive Phox2a promoter constructs. Whether concurrent ASH1 and CREB interactions with CBP, mediate Phox2a gene expression in NC cultures remains to be determined. We have assessed by in vitro protein-protein interaction assays, the occurrence of interactions between recombinant ASH1 with the E1A-binding region of CBP. 5 Studies by others have also shown the interaction of ASH1 and CREB with CBP in mediating VGF transcription (49). The CCAAT binding site, located 50 bp from a putative E-box binds the CCAAT enhancer-binding protein (C/EBP), which is proposed to have intrinsic cAMP-inducible activity, capable of mediating cAMP responsiveness (50). Moreover, C/EBP␣ is also able to stim-5 S. Chen and O. M. Andrisani, unpublished results. FIGURE 8. A, nucleotide sequence of the mouse and human Phox2a promoter sequences. The CRE half-sites 1 and 2 identified by ChIP assays as the major functional CRE sites are in proximity to putative bHLH E-box and C/EBP (CCAAT) binding sites, conserved in mouse and human Phox2a promoters. B, model depicting the transcriptional regulation of the Phox2a gene by the combined action of the BMP2 and cAMP signaling pathways. cAMP induces CREB Ser 133 phosphorylation, which recruits the co-activator CBP. ASH1, the downstream, indirect effector of BMP2, binds to the E-box and interacts with the CBP/phospho-CREB complex bound to each CRE half-sites 1 and 2 to initiate Phox2a transcription.
ulate CBP HAT activity (51). Further studies are required to define the functional significance of these sites in Phox2a gene regulation.
In summary, we propose a model (Fig. 8B) depicting the regulation of Phox2a gene transcription by the combined BMP2ϩcAMP signaling. Phox2a transcription in response to the combined BMP2ϩcAMP signaling involves a composite regulatory, cis-acting element comprised of CRE-half-sites 1 and 2 in proximity (50 bp) to E-box sites and the CCAAT binding site. This composite cis-acting element results in the formation of a protein complex containing CREB, CBP, and ASH1 induced by BMP2 signaling. This model offers an explanation for the synergistic effect of BMP2 and cAMP signaling on Phox2a gene transcription (7,8).