Estradiol Regulates Corticotropin-Releasing Hormone Gene (crh) Expression in a Rapid and Phasic Manner that Parallels Estrogen Receptor-α and -β Recruitment to a 3′,5′-Cyclic Adenosine 5′-Monophosphate Regulatory Region of the Proximal crh Promoter

Abstract

In the central nervous system, CRH regulates several affective states. Dysregulation of neuronal crh expression in the paraventricular nucleus of the hypothalamus correlates with some forms of depression, and amygdalar crh expression may modulate levels of anxiety. Because estrogens modulate these states, we sought to determine 17β-estradiol (E2) effects on crh expression. CRH mRNA levels were measured in the AR-5 amygdaloid cell line by RT-PCR analysis. They increased by 1 min of E2 treatment, suggesting that crh behaves as an immediate-early gene. After peaking at 3 min, CRH mRNA returned to basal levels and then increased by 60 min. To dissect some of the molecular mechanisms underlying these events, we measured occupancy of the crh promoter by estrogen receptors (ERs) and coactivators, using chromatin immunoprecipitation. Because this promoter does not contain palindromic estrogen response elements, we targeted the region of a cAMP regulatory element (CRE), implicated in crh regulation. The temporal pattern of the mRNA response was mimicked by recruitment of ERα and -β, phospho-CRE-binding protein, coactivators steroid receptor coactivator-1 and CRE-binding protein-binding protein (CBP), and an increase in histone 3 and 4 acetylation. Lastly, ERα and -β loading were temporally dissociated, peaking at 1 and 3 min, respectively. The ER peaks were associated with coactivators and acetylation patterns. ERα associated with phospho-CRE-binding protein, CBP, steroid receptor coactivator-1, and increased acetylated histone 3. ERβ associated with CBP and increased acetylated histone 4. The tight temporal correlation between E2-induced CRH mRNA levels and promoter occupancy by ERs strongly suggest that E2 regulates crh expression through an ERα- and/or ERβ-CRE alternate pathway.

CRH IS BEST KNOWN for its role in activation of the mammalian stress response. crh expression in the paraventricular nucleus of the hypothalamus (PVH) activates the pituitary-adrenal axis (1), which results in increased circulating glucocorticoids. Glucocorticoids, in turn, provide negative feedback to the axis. Part of this feedback is due to down-regulating expression of CRH (2,3) and its mRNA (4,5). CRH is also expressed in extrahypothalamic regions of the brain (6,7,8). In those sites, both the function and regulation of CRH differ from that in the PVH.

One important extrahypothalamic site of crh expression is the central nucleus of the amygdala (CeA). Here, CRH plays a major role in mediating responses to fear and anxiety (9,10). For example, intraventricular infusion of CRH enhances the startle response in rats (11,12), and CeA lesions attenuate this behavior (13). Similar findings have been reported for rhesus monkeys, in which lesions of the CeA decrease cerebrospinal fluid CRH levels and lessen fear and anxiety behaviors (14). Lastly, in humans, overactivation of CRH is associated with increased cerebrospinal fluid CRH levels in certain anxiety disorders (see Ref. 15 for review).

Clearly, the degree of CRH activity in various brain loci is coordinated by numerous neuronal and hormonal inputs. Curiously, in addition to the well known down-regulation of crh expression in PVH medial parvocellular neurons, glucocorticoids exert an opposite effect on CRH in the CeA. In this case, increases in circulating glucocorticoids effected by systemic corticosterone treatment of adrenalectomized rats result in increased CRH mRNA expression in the CeA (16,17). Glucocorticoid-induced increases in CRH mRNA levels in the CeA are associated with reduced exploratory behavior (18) and increased fear conditioning, suggesting increased anxiety (19).

In addition to glucocorticoids, estrogens play a role in regulating CRH. Although the majority of studies have focused attention on hypothalamic CRH (20,21,22,23,24), evidence is emerging for estrogen regulation of extrahypothalamic CRH as well. Two such reports have been published. One showed that estradiol benzoate treatment increases CRH mRNA in the CeA of ovariectomized mice (25). The other reported that estradiol benzoate increases CRH mRNA in the bed nucleus of the stria terminalis of ovariectomized ewes (26). Thus, estrogens appear to modulate crh expression in extrahypothalamic regions where CRH is associated with anxiogenic effects.

Many estrogen actions are mediated by estrogen receptor (ER)α and -β via a classic pathway, one in which ligand-bound ER binds specific palindromic estrogen response elements (EREs) in target promoters. Interestingly, the crh promoter does not contain palindromic EREs, although ERE half-sites are present. These ERE half-sites have been reported to support 2- to 3-fold promoter activation by ERα in the presence of 17β-estradiol (E2) (27).

Promoter activation may also occur via alternate pathways, in which ERs regulate the activity of distinct transcription factors bound to their own response elements. These include ER interactions with the activator protein complex-1 (AP-1) (28), SP1 (29), and the cAMP regulatory element-binding protein (CREB) (30,31). A proximal CRE has been implicated in estradiol regulation of crh expression (32,33). Miller and colleagues (32) showed that the pattern of ERβ-mediated crh promoter activation in HeLa cells was inconsistent with ERβ regulation via an AP-1 site and suggested that a CRE at approximately −200 bp may be necessary for ER regulation. Mutational analysis of this CRE by Ni and colleagues (33) revealed it is required for ERα-mediated repression of crh transcriptional activity in placental cells. Neither study, however, examined whether ERs were localized to the region of this CRE in the crh promoter in the context of its native chromatin. Data from such studies would suggest direct ER regulation of crh expression via a CRE alternate pathway. Also, neither study addressed the role of the CRE in mediating E2 effects in a neuronal cell line.

Here, we used an immortalized amygdalar cell line known to express CRH (AR-5) (34) to characterize the role of an alternate ER pathway in the regulation of crh expression. E2 treatment elicited a sufficiently rapid elevation of CRH mRNA levels to suggest that crh could be classified as an immediate-early gene (IEG). We then showed that both ERα and ERβ occupy the CRE region of the crh promoter selectively and that E2 further increases ER occupancy. Detailed analyses of ER and coactivator recruitment to the CRE region, and acetylation of histones 3 and 4 (H3 and H4), revealed that the pattern of promoter occupancy is similar to the mRNA profile. Furthermore, the data suggest that two temporally distinct complexes exist at the crh promoter, one containing ERα and the other containing ERβ. Taken together, the findings point to a role of an ER/CRE alternate pathway in mediating ER regulation of crh expression in neurons.

Materials and Methods

Media

Phenol red-free DMEM/F12 Coon’s medium (DMEM F12) was obtained from Hyclone (Logan, UT). Newborn calf serum (NCS) was obtained from Gemini Bioproducts (Woodland, CA; no. 100-504), and the same lot was used throughout the course of these experiments (C82903R). Sodium pyruvate, l-glutamine, nonessential amino acids, and penicillin/streptomycin were obtained from Life Technologies, Inc. (Gaithersburg, MD).

Cell treatments

E2, cycloheximide, and α-amanitin were all obtained from Sigma Chemical Co. (St. Louis, MO).

Antibodies

Antibodies were purchased, as indicated: anti-ERα (MC-20), anti-ERβ (H150), anti-ERβ (L-20), anti-CREB-binding protein (anti-CBP; A-22), and anti-steroid receptor coactivator-1 (anti-SRC-1; M-20) from Santa Cruz Biotechnology Inc. (Santa Cruz, CA); anti-phosphorylated CREB (anti-pCREB; no. 06-519), anti-acetylated H3 (anti-Ac-H3; no. 06-599), anti-Ac-H4 (no. 06-866), anti-ERα (C1355), and anti-ERβ (no. 06-629) antibodies from Upstate/Millipore (Lake Placid, NY); and anti-CREB antibodies from Cell Signaling Technology (no. 9197; Beverly, MA). Tubulin-α (DM1A) was obtained from Neomarkers (Fremont, CA), and Immunopure rabbit IgG (no. 31235) was obtained from Pierce Biotechnology, Inc. (Rockford, IL). Secondary antibodies used for immunoperoxidase staining were goat antirabbit antibody (no. 111-065-144) and goat antimouse (no. 165-065-146) from Jackson ImmunoResearch Laboratories (West Grove, PA) and bovine antigoat antibody (sc-2347) from Santa Cruz Biotechnology.

Buffers

1) Immunocytochemistry (ICC) blocking buffer was PBS (pH 7.4) (138 mm NaCl, 2.6 mm KCl, 1.5 mm KH₂PO₄, 6.3 mm Na₂HPO₄) containing 2% goat serum, 0.1% Triton X-100, and 0.1% sodium azide (Sigma). 2) Nonidet P-40 lysis buffer was 150 mm NaCl, 50 mm Tris (pH 7.5), 50 mm sodium fluoride, 5 mm sodium phosphate, 1% octylphenyl-polyethylene glycol (IGEPAL CA-630) (Sigma I-8896), 0.01% phenylmethylsulfonyl fluoride (PMSF; Sigma P7626), 1 mm sodium vanadate, and 1 μg/ml aprotinin/leupeptin/pepstatin (Roche Diagnostics, Indianapolis, IN). 3) Cytoplasmic extract buffer was 10 mm HEPES (pH 7.6), 60 mm KCl, 1 mm EDTA, 1 mm dithiothreitol, 1 mm PMSF, and 2.5 μg/ml aprotinin/leupeptin/pepstatin. 4) Nuclear extract buffer was 20 mm Tris (pH 8.0), 420 mm NaCl, 1.5 mm MgCl₂, 0.2 mm EDTA, 1 mm PMSF, and 2.5 μg/ml each of aprotinin/leupeptin/pepstatin. 5) Tris-buffered saline Tween 20 (TBS-T) was 20 mm Tris, 137 mm NaCl (pH 7.6), and 0.1% Tween 20 (Sigma). 6) SDS sample buffer was 2% SDS, 60 mm Tris/HCl (pH 6.8), 100 mm dithiothreitol, 2.5% glycerol, and 0.01% bromophenol blue. 7) Stripping buffer was 100 mm 2-mercaptoethanol (Sigma), 2% SDS, and 62.5 mm Tris (pH 7.5). 8) PCR mixture was PCR buffer (Sigma composition proprietary), 0.3 U Taq DNA polymerase (Sigma), 3 mm MgCl₂, 0.25 mm dNTPs (Invitrogen, Carlsbad, CA), and 0.4 μm each of forward and reverse primers in a final volume of 25 μl. For real-time PCR, a 1:75,000 dilution of SYBR Green I dye (Molecular Probes, Eugene, OR) was included in the 25-μl final volume reaction mixture. 9) Chromatin immunoprecipitation (ChIP) sonication buffer (CSB) was 1% Triton X-100, 0.1% deoxycholate (Sigma), 50 mm Tris (pH 8.1), 150 mm NaCl, and 5 mm EDTA (pH 8.0) and a 1:1000 dilution of a prepared protease inhibitor cocktail (Sigma P8340) and 0.2 mm PMSF. 10) LiCl wash buffer was 0.25 m LiCl, 0.005% IGEPAL CA-630 (Sigma), 0.005% deoxycholate, 0.01 m Tris (pH 8.1), and 1 mm EDTA (pH 8.0). 11) Tris-EDTA (TE) buffer (pH 7.5) was 10 mm Tris and 1 mm EDTA. 12) Proteinase K buffer (pH 7.5) was 10 mm Tris-HCl, 0.5% SDS, and 5 mm EDTA. 13) Elution buffer was 1% sodium SDS, 0.1 mm NaHCO₃, and 0.01 mg/ml herring sperm DNA.

Cell culture

The AR-5 rat amygdala cell line (34) was the only cell line used in these experiments. Cells were cultured in DMEM F12 media supplemented with 10% NCS, 1% l-glutamine, 1% sodium pyruvate, 1% nonessential amino acids, and 1% penicillin/streptomycin. For CRH mRNA and ChIP experiments, cells were maintained for 2–3 d in media containing stripped serum supplemented with 5% charcoal-stripped NCS before experimentation. Cell culture plates and chamber slides were obtained from Nunc Inc. (Naperville, IL).

ICC

AR-5 cells (1 × 10⁵ cells/0.5 ml·well) were plated and grown overnight. Cells were fixed in 4% paraformaldehyde for 20 min at room temperature. Cells were washed three times with PBS and then blocked for 1 h with ICC blocking buffer. For detection of ERα, a rabbit polyclonal antibody from Santa Cruz was used at 1:100 (ERα, MC20). For detection of ERβ, a goat polyclonal antibody from Santa Cruz was used at 1:100 (ERβ, L20). Cells were incubated with primary antibodies for 48 h at 4 C. They were then washed three times with ice-cold PBS and then incubated with a goat antirabbit secondary antibody for detection of ERα and a bovine antigoat secondary antibody for detection of ERβ. Secondary antibodies were used at 1:200 dilution for 1 h at room temperature. Immunoreactivity (IR) was visualized using the horseradish peroxidase system (Vector Elite ABC Kit) and diaminobenzidine (Dako, Carpinteria, CA) at 1 mg/ml. Images were recorded using a Scion Color Firewire CCD Camera (Scion Corp., Frederick, MD). The final composite figures were generated using Adobe Photoshop 7.0.

Preparation of cell lysates and Western blot analysis

Cells (1 × 10⁷) were lysed by adding 1 ml ice-cold 0.5× Nonidet P-40 lysis buffer. Cell lysates were clarified by centrifugation at 16,060 × g for 20 min at 4 C, and the supernatants were recovered. Nuclear fractions were prepared as described previously (35). Briefly, cells were washed with ice-cold PBS and scraped from the plate. After centrifugation at 380 × g for 5 min, cell pellets were lysed in 5 pellet volumes of cytoplasmic extract buffer, containing 0.075% IGEPAL CA-630 for 3 min. Nuclei were pelleted by centrifugation at 308 × g for 5 min, and supernatants containing the cytoplasmic proteins were removed to fresh tubes. Nuclei were washed in cytoplasmic extract buffer without IGEPAL CA-630 and centrifuged. Nuclei were resuspended in 1 pellet volume of nuclear extract buffer. The final salt concentration was adjusted to 400 mm. Samples were incubated on ice for 10 min and resuspended every 2 min by vortexing. Finally, both cytoplasmic and nuclear fractions were centrifuged at 16,060 × g for 15 min at 4 C, and supernatants were transferred to fresh tubes.

Protein samples (20 μg) were mixed with SDS sample buffer. Cell lysates and cellular (cytoplasmic or nuclear) fractions were electrophoresed using 12% SDS-polyacrylamide gels. Protein was transferred to polyvinylidene difluoride membrane (Millipore, Bedford, MA) and then incubated with blocking buffer (5% nonfat dry milk in TBS-T) overnight. Subsequently, the blots were washed in TBS-T (three times for 5 min each wash) and incubated with rabbit polyclonal antibodies to ERα (Santa Cruz; MC-20) or ERβ (Santa Cruz; H-150) overnight. Blots were then incubated for 1 h at room temperature with goat antirabbit secondary antibodies conjugated to horseradish peroxidase (Pierce). Protein bands were visualized by incubation with SuperSignal West Dura Extended Duration Substrate signal (Pierce) according to the manufacturer’s protocol. When membranes were reprobed, they were first stripped for 30 min at 37 C and then incubated with blocking buffer overnight.

RT-PCR

Total RNA was extracted with 1 ml TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. The iScript cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA) was used to generate ERα and ERβ cDNA. RT was performed using 1 μg total RNA (1 μg) according to the manufacturer’s instructions. The reaction was carried out at 25 C for 5 min, followed by 42 C for 45 min and 85 C for 5 min. To amplify the cDNA, primers that flanked the coding sequence of rat ERα or ERβ were used. The ERα primer sequences were as follows: forward, 5′-GCA GCA GCG AGA AGGGA AAC A-3′, and reverse, 5′-TCA TGC GGA ATC GAC TTG ACG-3′. The ERβ primers amplify a region of the transcript that spans the ERβ DNA-binding domain (DBD) and ligand-binding domain. This approach, described by Price et al. (36), permits detection of five ERβ variants: the originally characterized ERβ (ERβ1) (37), a variant that contains an additional 54 nucleotides in the ligand-binding domain (ERβ2) (38), splice variants that occur in both ERβ1 and ERβ2 and result in the loss of exon 3 (-δ3), which includes part of the DBD (ERβ1δ3 and ERβ2δ3) (38), and a splice variant missing exon 4 (ERβ1δ4) (36). ERβ primer sequences were as follows: forward, 5′-GTT GTG CCA GCC CTG TTA CTA-3′, and reverse, 5′-CGC CAG GAG CAT GTC AAA-3′. Thermal cycling was performed using an iCycler iQ system (Bio-Rad). Cycling parameters for ERα mRNA were as follows: an initial melting step of 95 C for 4 min, amplification at 95 C for 60 sec, 66 C for 60 sec, and 72 C for 60 sec for 40 cycles, and 72 C for 4 min. Cycling parameters for ERβ mRNA were as follows: an initial melting step of 94 C for 2 min, amplification at 94 C for 35 sec, 54 C for 35 sec, and 72 C for 35 sec for 40 cycles, and 72 C for 5 min.

The High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA) was used to generate CRH cDNA. RT was performed on 5 μg total RNA according to the manufacturer’s instructions. This contained 250 U MultiScribe reverse transcriptase, 10 μl RT buffer (composition proprietary), 1 mm dNTP, 5 μm random primers (all reagents supplied in the High Capacity cDNA Archive Kit; Applied Biosystems), and RNase-free water (Sigma). The reaction was carried out at 25 C for 10 min followed by 37 C for 2 h. To amplify the cDNA, primers were used that flanked the coding sequence of CRH mRNA, as previously described (39). Primer sequences were as follows: forward, 5′-GAA GAG AAA GGG GAA AGG CAA AGA-3′, and reverse, 5′-GCG GTG AGG GGC GTG GAG TT-3′. Primers for endogenous control β-actin mRNAs were as follows: forward, 5′-TCC ATC ATG AAG TGT GAC GT-3′, and reverse, 5′-TAC TCC TGC TTG CTG ATC CAC AT-3′. Cycling parameters were an initial melting step of 95 C for 2.5 min and amplification at 94 C for 30 sec, 56 C for 30 sec, and 72 C for 30 sec for 40 cycles.

ChIP followed by real-time PCR

Cells were grown to 70–90% confluency, and 3 × 10⁷ cells were used per treatment, per time point. They were exposed to α-amanitin (2.5 μm) for 2 h to synchronize transcription start times, washed twice with PBS, and then treated with ethanolic vehicle (0.1% of the final volume of media) or E2 (10⁻⁷ m) for 1, 3, 10, 30, and 60 min. ChIPs were performed as described (40,41) with minor modifications. After treatment, cells were washed once with PBS and then cross-linked with 1% formaldehyde (in PBS) at room temperature for 10 min. Cross-linking was stopped by addition of 0.125 m glycine. Cells were washed twice with ice-cold PBS and then collected in 500 μl PBS. Cells were immediately centrifuged at 590 × g for 1 min at 4 C, and pellets were resuspended in 1 ml CSB. Cells were incubated on ice for 10 min before 12–15 sonication pulses, 5 sec each. The sonicator was set at 40% duty and an output of 4 (Sonicator W375; Heat Systems-Ultrasonics, Inc., Farmingdale, NY). Chromatin was sheared to 800- to 1000-bp fragments. Fragment length was monitored by agarose gel electrophoresis of DNA. Samples were centrifuged at 16,060 × g for 15 min at 4 C. Supernatants were collected and protein concentration determined. Before immunoprecipitation, 1 μg control plasmid DNA (here a reporter construct devoid of a promoter, pFoxLuc) was added to each sample of whole-cell extract to monitor DNA recovery. Aliquots (100 μg) of extract were diluted with CSB to a final volume of 250 μl and then incubated with primary antibody overnight at 4 C with rotation. Subsequently, 60 μl protein A/G-agarose bead slurry (Invitrogen) and 2 μg herring sperm DNA solution (Sigma) were added to each sample and mixed by rotation for 2 h at 4 C. Samples were centrifuged at 590 × g at 4 C for 1 min. Beads were washed sequentially with 1 ml each of the following buffers: 1) CSB, 2) CSB containing NaCl (500 mm), 3) LiCl wash buffer, and 4) TE buffer. Chromatin was eluted from the beads by incubating the pellets in 250 μl elution buffer and mixed by rotation for 15 min at room temperature. The beads were pelleted by centrifugation at 16,060 × g for 1 min, and the supernatants were removed to fresh tubes. This elution step was performed twice. The combined chromatin eluates were reverse cross-linked by incubation in a 65 C water bath for a minimum of 4 h. DNA was ethanol precipitated overnight at −20 C. Samples were centrifuged at 16,060 × g for 20 min. Pellets were resuspended in 100 μl TE buffer containing 1:10 dilution of proteinase K buffer and 19 μg proteinase K (Roche). Samples were heated to 55 C for 1 h, extracted once with phenol-chloroform, and then ethanol precipitated overnight at −20 C. Precipitates were pelleted by centrifugation at 16,060 × g for 15 min at 4 C. The pellets were washed with 70% ethanol and resuspended in 100 μl TE buffer.

The rat crh CRE has been previously described (42). PCR primers were designed to target the region of the CRE most proximal to the TATA box, using the Custom Primers OligoPerfect Designer tool at www.invitrogen.com. Of the three sets of primers designed and empirically tested, the one with the highest melt temperature was chosen for all ChIP PCR. The sequences were as follows: forward primer, 5′-GTC ACC AAG GAG GCG ATA AA-3′, and reverse primer, 5′-CGA CCC TCT TCA GAA AGC AC-3′.

The 5′ upstream ERE half-sites found between −5113 and −1867 were also targeted. Primer pairs beginning with the most 5′ ERE half-site are as follows: 1) forward, 5′-CTG CCC CTA GTT GGT TCT GA-3′, and reverse, 5′-TTC TGG TTC CTG ACC TGC T-3′; 2) forward, 5′-AAG GGG ACA GAG AGG AAG GA-3′, and reverse, 5′-TCC ATG CAG GTA GCA CAC TC-3′; 3) forward, 5′-TGG ATT CAA TGG ACC AGA GAG-3′, and reverse, 5′-TGC CCT TTG GCA TAA ATC TC-3′; 4) forward, 5′-TCC CAT GGA AGC AGT AAG GT-3′, and reverse, 5′-CCA GGC AAG CAA ATA CAC TC-3′; and 5) forward, 5′CAA CAG CAA GGG AGA GAC AG-3′, and reverse 5′-GCC AAT GAG TCC TGA AGA GG-3′. PCR primers were also designed to target an upstream region of the rat crh promoter that contained no known enhancer regions as determined by the Search Transcription Factor Binding Sites tool at www.genomatix.de. This primer pair was forward primer 5′-TTC TTG GCA GCT CTG CAC TC-3′ and reverse primer 5′-TGC CTC TGC TCC TGC ATA AA-3′. The c-fos proximal promoter primers targeted a complex estrogen-responsive region of the proximal promoter where an ERE, AP-1 site, and a CRE are localized within 500 bp upstream of the TATA box (NCBI accession nos. V00727 and DQ089699). These primers were as follows: forward, 5′GGC GAG CTG TTC CCG TCA ATC C-3′, and reverse, 5′-GCG GGC GCT CTG TCG TCA ACT CTA-3′. Lastly, PCR primers directed toward the control plasmid (pFoxLuc), used to correct for DNA recovery were as follows: forward, 5′-TCG CCA GAA AGT AGG GGT CG-3′, and reverse, 5′-GCT TCT GCC AACC GAA CGG AC-3′. All primers were purchased from Operon Biotechnologies, Inc. (Huntsville, AL).

Cycling parameters for the crh promoter and pFoxLuc plasmid DNA were as follows: an initial melting step at 95 C for 3 min and amplifications (95 C for 30 sec, 55 C for 30 sec, and 72 C for 30 sec) for 40 cycles. The threshold cycle was set at the point at which fluorescence rises 10 times above the mean sd of background fluorescence. To ensure that DNA recovered from ChIPs fell within the linear range of the assay, a standard curve was generated using 0.1–100% of input. The amount of pFoxLuc DNA recovered served to correct for sample loss throughout the assay. Calculations performed have been previously reported (40,41).

Statistical analysis

Data represent the averages of at least three separate experiments (n = 3–6). Cells from three plates were harvested, pooled, and carried through the ChIP protocol, and the isolated DNA samples were quantified by real-time PCR in triplicate, the average of which was taken as n = 1. Data are displayed as a best-fit graph. Because data did not meet the criteria for parametric tests, statistical analysis consisted of nonparametric Kruskal-Wallis one-way ANOVAs, followed by Bonferroni post hoc tests. A P < 0.05 was considered significant for these studies. Nonparametric two-sample t tests (Mann-Whitney U test) with Bonferroni corrections to control for multiple comparisons were used to determine the level of significance between individual time points. Nonparametric one-sample t tests (Wilcoxon signed-rank test) with Bonferroni corrections were used to determine treatment effects at individual time points. All data analyses were performed using the Number Cruncher Statistical System program (NCSS, Kaysville, UT).

Results

AR-5 cells express ERα and ERβ

To determine whether the AR-5 cell line express ERα and ERβ, ICC was performed (Fig. 1). To facilitate making the distinction between nuclear and cytoplasmic localization, cells were probed with anti-Ac-H3 and anti-tubulin-α antibodies, respectively (Fig. 1, A and B). ICC using antibodies directed against ERα (Fig. 1C) and ERβ (Fig. 1D) revealed that both were predominantly nuclear. In addition, the presence of ERβ-IR was assessed with the ERβ antibody used for chromatin immunoprecipitations; the pattern of IR was similar to that shown in Fig. 1D (data not shown). To assess specificity, ERα and ERβ antibodies were preincubated with their immunogenic peptides. Preincubation with subtype-specific peptides eliminated ERα-IR (Fig. 1E) and ERβ-IR (Fig. 1F). The reciprocal experiments were also performed. As shown, the patterns of ERα-IR (Fig. 1C) and ERβ-IR (Fig. 1D) were unchanged after preincubation with the ERβ and ERα immunogenic peptides, (Fig. 1, G and H), respectively. As a negative control (Fig. 1I), ICC was carried out in the absence of primary antibody. There was no detectable staining. Phase contrast (Fig. 1J) imaging of the same field in Fig. 1I indicated that cells were present. Thus, by ICC, AR-5 cells express both ERα and ERβ.

Figure 1.

In AR-5 cells, ERα- and -β IR are predominantly nuclear. A and B, Ac-H3 IR (1:100; A) and tubulin-α IR (1:1000; B); C and D, ERα IR (1:175; C) and ERβ IR (1:100; D) are primarily nuclear, with some cytoplasmic localization; E and F, preabsorption of ER antibodies with their antigenic peptide eliminates staining of ERα and ERβ, respectively; G and H, preabsorption with the converse subtype’s antigenic peptide has no effect on ERα (G) or ERβ (H) staining patterns; I, in the absence of primary antibody, no IR is apparent; J, phase contrast shows the presence of cells in the field shown in I. Scale bar in A = 20 μm and applies to all panels.

To determine whether the ERα and -β antibodies recognized proteins of the correct molecular weight and to assess the degree of antibody specificity, Western blots were performed (Fig. 2, A, B, and D). The ERα antibody used for ChIP analysis (MC-20; Santa Cruz Biotechnology) recognized an immunoreactive band that migrates at 62–63 kDa (Fig. 2A). Because the mobility was slightly faster than expected, the blot was stripped and reprobed with a different ERα antibody (C1355; Millipore). The bands were superimposable (Fig. 2, A and B). Lastly, as an independent measure of ERα expression, one that did not involve an antibody-dependent technique, we used RT-PCR to determine whether ERα mRNA is present. This analysis revealed ERα mRNA in AR-5 cells and the amygdala, the region from which AR-5 cells were derived; however, the signal was absent in cortex (Fig. 2C). This region-specific expression in the rat central nervous system is consistent with data reported by Shughrue and colleagues (43). In summary, both ERα protein and mRNA were present in AR-5 cells and uterus but absent in spleen (positive and negative controls, respectively). Thus, by ICC, Western blot, and PCR analysis, AR-5 cells express ERα.

Figure 2.

Detection of ERα and ERβ protein and mRNA expression in AR-5 cells. A–C, Western blot analysis of AR-5 lysates and RT-PCR of AR-5 mRNA revealed ERα IR and mRNA, respectively. A, Antibodies against ERα (MC-20, 1:1000; Santa Cruz) recognized a band with a molecular mass of approximately 62–63 kDa in AR-5 cells and uterus (positive control), but not in spleen. B, After stripping the blot, another antibody against ERα (C1355, 1:5000; Millipore) was used and recognized the same band in AR-5 cells and uterus but not in spleen. C, As an independent measure of ERα expression in AR-5 cells, we detected ERα mRNA levels by RT-PCR. Ethidium bromide staining of PCR products revealed a single ERα amplicon (792 nucleotides). D and E, Western blot analysis and RT-PCR also revealed ERβ IR and mRNA in AR-5 lysates. E, against ERβ (H150, 1:1000; Santa Cruz) recognized two bands with molecular masses of approximately 59 and 70 kDa in AR-5 cells and prostate (positive control) but not spleen. D, Antibodies Detection of ERβ mRNA by RT-PCR revealed the presence of four splice variants of predicted lengths: ERβ2 (788 nucleotides), ERβ1 (734 nucleotides), ERβ2δ3 (671 nucleotides), and ERβ1δ3 (617 nucleotides).

To assess the expression of ERβ in the AR-5 cells, Western blot analyses were performed with the ERβ antibody used for ChIP analysis (H150; Santa Cruz Biotechnology). The antibody recognized a protein that migrates at approximately 59 kDa in both AR-5 cells and in prostate, in which it was originally described (Fig. 2D); spleen served as the negative control. An additional band migrating at a higher molecular weight was observed. In the image shown here, it appears to be only in AR-5 cells; however, this additional band is also visible in prostate with longer exposure of the Western blot (data not shown). This pattern of bands has been previously reported (44). In that report, an explanation set forth for the second band was that it could represent a posttranslational modification of ERβ. An alternate explanation is that it is a splice variant of ERβ, with or without posttranslational modifications. To determine whether AR-5 cells express ERβ isoforms, we performed RT-PCR (Fig. 2E). ERβ primers had been designed to permit detection of ERβ1, ERβ2, their δ3 variants ERβ1δ3 and ERβ2δ3, and a δ4 variant ERβ1δ4. In all tissues except for the spleen (negative control), four bands were clearly present. Their lengths were consistent with predicted lengths of four splice variants. From longest to shortest they are ERβ2, ERβ1, ERβ2δ3, and ERβ1δ3 (Fig. 2E). The absence of an ERβ1δ4 mutant is consistent with the Price et al. (36) finding that it is almost exclusively cerebellar; thus, it would not be expected to be present in the AR-5 line, derived from the amygdala. Lastly, the finding of the ERβ2 splice variants is in accord with our suggestion that the upper band detected by Western blot might be the ERβ2 splice variant. Taken together, data presented in Figs. 1 and 2 indicate that AR-5 cells express ERα, ERβ, and ERβ splice variants.

E2 induces rapid and phasic expression of endogenous CRH mRNA in AR-5 cells

To determine whether E2 treatment of AR-5 cells alters expression of CRH mRNA, we analyzed its levels by real-time PCR amplification of cDNA at 1, 3, 10, 30, and 60 min after treatment. There was a significant effect of time on CRH mRNA levels in the presence of E2 (Fig. 3A; one-way ANOVA: H = 20.393; df = 5; P < 0.001). The pattern of CRH mRNA response to E2 treatment was phasic (by Bonferroni post hoc test; P < 0.05). At both 3 and 60 min, CRH mRNA levels were significantly different from levels at 10 min. In addition, levels at 60 min were significantly different from those at 0 min. The data were also analyzed by two-sample t tests. The approximate 4- and 6-fold increases in CRH mRNA levels at 1 and 3 min were significantly higher compared with 0 min (P < 0.005). The approximately 6-fold increase in CRH mRNA levels at 60 min was significantly higher than at 0 min (P < 0.04) as well. Thus, E2 elicits phasic changes in CRH mRNA levels with a peak at 1–3 min and a second rise from 30–60 min.

Figure 3.

E2 treatment leads to both a rapid and a delayed increase in CRH mRNA levels that differ in the requirement for protein synthesis. AR-5 cells were cultured in media supplemented with stripped serum. A, Cells were treated with ethanolic vehicle (Veh) or E2 (10⁻⁷ m) for 1, 3, 10, 30, and 60 min. B, cells were treated with E2 (10⁻⁷ m) in the absence or presence of the protein synthesis inhibitor cycloheximide (10 mg/ml) for 1, 3, and 60 min. mRNA levels were measured using quantitative real-time RT-PCR. Error bars represent ±sem; n = 5. A: τ, Significantly different from 10 min by Bonferroni post hoc test; *, significantly different from 0 min by Bonferroni post hoc test; #, significantly different from 0 min by two-sample t tests with Bonferroni corrections as follows: 1 min (Z = 2.5491; P < 0.006), 3 min (Z = 2.6740; P < 0.004), and 60 min (Z = 2.6740; P < 0.004). B: #, Significant differences between E2 (60 min) and E2 plus cycloheximide (60 min) by two-sample t test (Z = 1.7457; P < 0.041; n = 3).

The rapid increase in mRNA levels at 1 and 3 min suggested that crh could be classified as an IEG. To determine whether this was the case, we performed the same experiment in the presence or absence of the protein synthesis inhibitor cycloheximide. Cycloheximide failed to alter the E2 response at 1 and 3 min, even though it reduced it by 60 min (Fig. 3B; two-sample t test: Z = 1.7457; P < 0.041). Because the rapid increase in CRH mRNA occurs in 1–3 min and is refractory to cycloheximide treatment, crh meets the criteria of an IEG (45,46,47).

ERα and ERβ occupy the region of the proximal crh CRE

ChIP assays were used to identify the effect of E2 on the presence of ERα and -β at the crh promoter. The promoter regions examined were a CRE (at approximately −200 bases) and five upstream response elements that resemble ERE half-sites (Fig. 4A). All of these have been implicated in regulation of crh expression by E2 (27). At 60 min, ERα and -β were found to localize to the crh CRE region but not to any of the ERE half-sites (Fig. 4B). ERα and -β occupancy of the ERE half-sites after E2 treatment for 1, 3, 10, and 30 min were also negative (data not shown). E2 increased occupancy of ERα and -β at the crh CRE region. Similar levels of the control plasmid were found in each sample, indicating equivalent DNA recovery. As a positive control, ERα and -β occupancy of the known c-fos ERE (−309 to −80 bases) in response to E2 was assessed. Both were present (Fig. 4B). As a negative control, a far upstream region of the promoter was also targeted, and no occupancy was detected (data not shown). Thus, at 60 min, ERα and -β selectively occupy the region of the proximal CRE but none of the far upstream ERE half-sites.

Figure 4.

ERα and ERβ selectively occupy the crh CRE region in response to E2. A, Distribution of putative ER regulatory elements within the rat crh promoter; B, E2 increases ERα and ERβ occupancy in the region of the crh CRE and c-fos ERE. PCR amplicons correspond to the promoter regions, as indicated. AR-5 cells were treated for 60 min. V, Vehicle.

ERα, ERβ, and SRC-1 are rapidly and dynamically recruited to the crh promoter

To investigate the molecular events that underlie E2 regulation of CRH mRNA levels in AR-5 cells, we again used ChIP assays to determine the profile of factor occupancy at the endogenous crh promoter, i.e. in its native chromatin setting. We first assessed occupancy of ERα and -β in the region of the CRE. One-way ANOVA revealed a trend for an effect of time on ERα and ERβ occupancy (H = 10.854; df = 5; P < 0.055 and H = 9.369; df = 5; P < 0.096, respectively; Fig. 5). Furthermore, the Bonferroni post hoc test revealed that ERα peaked at 1 min, and ERβ peaked at 3 min, even though the effects of time were only trends. Two-sample t tests were used to determine the level of significance of ER occupancy at individual time points compared with 0 min. By two-sample t tests, ERα occupancy at 1, 3, 30, and 60 min was significantly higher than at 0 min. The levels of significance were as follows: 1 min (Z = 2.8723; P < 0.002), 3 min (Z = 2.9406; P < 0.003), 30 min (Z = 2.5491; P < 0.006), and 60 min (Z = 2.5491; P < 0.006). ERβ occupancy was increased at all time points (see legend for levels of significance). Although these statistical measures indicate that some overlap in ERα and ERβ occupancy exists, there remain clear peaks in ERα and ERβ occupancy at 1 and 3 min, respectively.

Figure 5.

E2 treatment leads to rapid and dynamic recruitment of ERα, ERβ, and SRC-1 to the crh promoter in the region of the proximal CRE. The same conditions and time points were used as in the CRH mRNA experiment shown in Fig. 3. Here, ChIPs were performed with ERα, ERβ, and SRC-1 antibodies. The region of the crh promoter containing the CRE was amplified by quantitative PCR. Cells were treated for the times indicated. Data represent the fold of E2-treated cells compared with vehicle. Error bars represent ±sem; n = 6. *, P < 0.05, Bonferroni post hoc test for significant time effects; #, significant difference from the 0-min time point by two-sample t test with Bonferroni corrections as follows: ERα, 1 min (Z = 2.8723; P < 0.002), 3 min (Z = 2.9406; P < 0.003), 30 min (Z = 2.5491; P < 0.006), and 60 min (Z = 2.5491; P < 0.006); ERβ 1 min (Z = 2.8627; P < 0.003), 3 min (Z = 2.9406; P < 0.003), 10 min (Z =2.5491; P < 0.006), 30 min (Z = 2.5491; P < 0.006), and 60 min (Z = 2.5491; P < 0.006); SRC-1 1 min (Z = 2.6740; P < 0.004) 3 min (Z = 2.5491; P < 0.006), and 60 min (Z = 2.3910; P < 0.009).

There was also an effect of time on SRC-1 occupancy at the CRE (Fig. 5; one-way ANOVA: H = 12.888, df = 5, P < 0.025). The levels of occupancy at 1, 3, and 60 min all differed from the level at 0 min, as determined by the two-sample t test. Specifically, there was a 4-fold increase at 1 min (Z = 2.6740; P < 0.004) and a 2-fold increase in occupancy at 3 min (Z = 2.5491; P < 0.006). Lastly, there was a 5-fold increase in SRC-1 occupancy at 60 min (Z = 2.3910; P < 0.009). Overall, these results reveal rapid and dynamic patterns of ERα, ERβ, and SRC-1 occupancy at the crh promoter in the region of the proximal CRE, which correlate with the biphasic increases in CRH mRNA levels (Fig. 3).

pCREB and CBP dynamically occupy the crh promoter in the region of the proximal CRE

There was a significant effect of time on pCREB occupancy (Fig. 6; one-way ANOVA: H = 13.549; df = 5; P < 0.020). Additionally, the levels of pCREB occupancy at 1, 30, and 60 min all differed from the level at 0 min by the two-sample t test. There was a 7-fold increase at 1 min (Z = 2.5491; P < 0.006) and approximately 2- and 5-fold increases in pCREB occupancy at 30 and 60 were also significant (Z = 2.3910; P < 0.009 and Z = 2.5491; P < 0.006, respectively).

Figure 6.

E2 elicits rapid and dynamic loading of pCREB and CBP at the crh promoter in the region of the CRE. The AR-5 whole cell lysates used in Fig. 3 were probed with total CREB, phospho-CREB, and CBP antibodies, in parallel. The same region of the CRE was amplified as in Fig. 4. Data represent the fold of E2-treated cells compared with vehicle. Error bars represent ±sem, n = 4. (#) Significant difference from the 0 min time point by two-sample t test with Bonferroni corrections as follows: (pCREB) 1 min (Z = 2.5491; P < 0.006), 30 min (Z = 2.3910; P < 0.009), and 60 min (Z = 2.5491; P < 0.006); (Total CREB) 60 min (Z = 2.5491; P < 0.006); (CBP) 3 min (Z = 2.7765, P < 0.004), and 60 min (Z = 2.6740; P < 0.004). Note that the scale is different from the scale used in Figs. 3 and 5 (maximum of 20 vs. 30). Veh, Vehicle.

One-way ANOVA failed to reveal a significant effect of time for either total CREB or CBP (Fig. 6). However, the 8-fold association of CBP occupancy at 3 min was significantly higher than at 0 min by the two-sample t test (Z = 2.7765, P < 0.004), and the 3-fold association at 60 min was significantly higher than at 0 min (Z = 2.6740; P < 0.004). Lastly, total CREB was significantly different from 0 min only at 60 min (Z = 2.5491; P < 0.006). Taken together, the data again indicate a rapid and dynamic association of a transcription factor and a coactivator with the crh promoter, here pCREB and CBP.

H3 and H4 acetylation increase at the crh promoter in response to E2

Lastly, to determine whether E2 could induce histone acetylation at the crh promoter region targeted, changes in the levels of pan-acetylated H3 and H4 were measured (Fig. 7). Although there was no effect of time on either H3 or H4 acetylation by one-way ANOVA, significant differences were found using the two-sample t test. For H3, the degree of acetylation was significantly greater at 1 min (Z = 2.5491; P < 0.006), at 30 min (Z = 2.5491; P < 0.006), and at 60 min (Z = 2.5491; P < 0.006). A trend toward significance for increased H3 acetylation was observed at 3 min (Z = 1.5598; P < 0.060). H4 acetylation was predominantly a late response, being increased 7-fold at 60 min (Z = 2.3910; P < 0.009). In this case, too, however, there may be an early acetylation event given that there was a trend toward significance at 3 min (Z = 1.5371; P < 0.085). To summarize these data, the timing of H3 acetylation closely parallels the changes in mRNA over time, whereas the acetylation of H4 is predominantly a late event. Thus, as with the transcription factors and coactivators described above, the events are temporally distinct; therefore, acetylation of H3 and H4 probably subserve different roles in ER regulation of crh expression.

Figure 7.

E2 increases acetylation of H3 and H4 at the crh promoter in the region of the CRE. Aliquots of the AR-5 whole-cell lysates used for Figs. 4 and 5 were probed with Ac-H3 and Ac-H4 antibodies in parallel. Data represent the fold of E2-treated compared with vehicle (Veh). Error bars represent ±sem; n = 4. #, Significant difference from the 0-min time point by two-sample t test with Bonferroni corrections as follows: Ac-H3 1 min (Z = 2.5491; P < 0.006), 30 min (Z = 2.5491; P < 0.006), and 60 min (Z = 2.5491; P < 0.006); Ac-H4 60 min (Z = 2.3910; P < 0.009).

The effect of treatment at 1 and 3 min

When viewed collectively, it is apparent that patterns of occupancy and acetylation fall into two groups: an early peak and a late rise, which is significant by 60 min. On closer inspection of the early peak, it becomes apparent that it may be subdivided into two peaks, one at 1 min and the other peak at 3 min. The peak at 1 min coincides with the ERα peak, and the peak at 3 min coincides with the ERβ peak. The effect of treatment on members of these temporally distinct groups is presented in Table 1. At 1 min, there was a significant effect of E2 on the presence of ERα and SRC-1, and on ERβ as well. However, by 3 min, the E2 effect on SRC-1 was no longer significant. Given the high degree of significance in the change of SRC-1 from 1–3 min (P < 0.008), the data suggest that ERα and SRC-1 are in a complex at 1 min but not at 3 min. A trend toward significance for decrease in occupancy by pCREB at 3 min suggests that, like SRC-1, it predominantly associates with ERα at 1 min. In distinction to ERα, ERβ increased 9.71 points from 1–3 min, and the increase was highly significant (P < 0.002). Additionally, the effect of E2 on CBP occupancy was significant at 3 min but not at 1 min. In turn, these data suggest that an ERβ/CBP complex is present at the CRE at 3 but not 1 min. Taken together, the data suggest that an ERα/SRC-1 complex predominates at 1 min, whereas an ERβ/CBP complex predominates at 3 min. The sequential occupancy of these two complexes at the crh promoter suggests that the pattern of CRH mRNA levels is mediated by both ERα and ERβ and their associated coactivators.

Table 1.

Effect of E2 treatment on promoter loading of candidate regulatory factors at 1 and 3 min

	1 min		3 min		Δ (3 min − 1 min)	P
	Mean ± sem	P	Mean ± sem	P
ERα	23.07 ± 7.65	0.016a	8.83 ± 2.64	0.008a	−14.24	0.069
SRC-1	3.78 ± 1.72	0.031a	1.90 ± 0.24	0.063	−1.88	0.008b
pCREB	6.85 ± 2.95	0.063	0.79 ± 0.28	0.625	−6.06	0.056
Ac-H3	6.71 ± 2.86	0.063	2.20 ± 0.88	0.094	−4.51	0.206
ERβ	3.86 ± 1.43	0.008a	13.57 ± 2.51	0.006a	9.71	0.002b
CBP	4.70 ± 2.43	0.125	7.54 ± 3.56	0.016a	2.84	0.421
Total CREB	2.63 ± 0.91	0.156	4.64 ± 2.55	0.313	2.01	0.305
Ac-H4	1.24 ± 0.44	0.250	5.10 ± 3.04	0.125	3.86	0.341

At 1 and 3 min, the numbers represent the E2 effect as fold of vehicle and are the same data shown in Figs. 5–7. The last set of numbers is the difference between the data at 1 and 3 min. The associated P value (last column) indicates the degree to which this difference is significant. 

^aSignificant difference from vehicle, P < 0.025 (Wilcoxon signed-rank test, adjusted for multiple comparisons with Bonferroni’s correction). 

^bSignificant difference between 1 and 3 min, P < 0.025 (Mann-Whitney U test, adjusted for multiple comparisons with Bonferroni’s correction). 

Discussion

The goal of this study was to elucidate mechanisms by which ERα and ERβ modulate expression of crh. To this end, we further characterized an amygdalar cell line and used it to determine the relationship between CRH mRNA levels and ER occupancy of the crh promoter. The recruitment of ER coactivators was also assessed, as was the degree of acetylation of H3 and H4. The region targeted for analysis was the proximal crh promoter that contains a CRE involved in crh regulation. Collectively, the data point to a mechanism that involves two ER complexes distinguished by ER subtype composition, temporal occupancy, associated coactivators, and histone acetylation.

The cell line used for these studies, AR-5, has been previously found to express CRH (34). The data presented here indicate that the AR-5 line expresses ERα and four isoforms of ERβ (Figs. 1 and 2), as well. The cells are responsive to E2, as evidenced by the changes in CRH mRNA (Fig. 3), and are thus suitable for studies of E2 regulation of the endogenous crh promoter. They should also prove useful for studies of estrogen effects on a variety of molecular mechanisms in a neuronal context.

The rapid increase in CRH mRNA within minutes of E2 treatment prompted examination of whether crh could be considered an IEG. Standard criteria for IEG classification are that the gene’s mRNA levels increase rapidly, within minutes, and that the increase is independent of protein synthesis, typically assessed by the lack of a response to cycloheximide treatment (47). Indeed, the peak of CRH mRNA at 1–3 min was unaffected by cycloheximide (Fig. 3). In distinction, levels at 60 min were markedly reduced by cycloheximide treatment. Thus, crh meets the criteria for an IEG. The concept of an IEG originated from observations that mRNA levels of certain genes, expressed in cultured cells, increase within minutes of being exposed to a stressor. The concept was thus originally conceived as one that applied to a cellular response. However, crh is a gene that responds to systemic stressors. Thus, stress-responsive genes may be IEGs either in the context of a cell or of a whole organism.

To begin analysis of molecular mechanisms that underlie the rapid expression of crh, we performed ChIP analysis of putative estrogen targets in the rat crh promoter. The gene contains five ERE half-sites within the far upstream region of the promoter (Fig. 4A). Similar ERE half-sites have been implicated in E2 regulation of the human crh promoter (27). Here, ChIP analysis revealed that neither ERα nor ERβ were localized to these half-sites in response to E2 treatment (Fig. 4B). In the previous study, EMSAs were used to show that the ERα DBD could bind ERE half-site motifs. A subsequent study from a different group (48), however, failed to detect ERα binding to the ERE half-site motifs by EMSAs. Taken together, this latter study and our data suggest that ERs regulate the crh promoter through a mechanism other than ER binding to ERE half-sites.

Instead of direct DNA binding, our data point to regulation via an alternate or tethering pathway (28,49), here one that is mediated through a CRE. The importance of the proximal CRE in regulating crh expression has been previously demonstrated. It mediates glucocorticoid down-regulation of crh expression (50) and is the cognate regulatory element targeted by activation of the cAMP-protein kinase A pathway (51). Furthermore, it plays a role in glucocorticoid-mediated crh down-regulation by recruitment of an inhibitory member of the CREB family, inducible cAMP early repressor (52). Thus, one could consider it to be a site of functional signal integration. Such a proposition was put forth for an AP-1 site in the case of the opposing effects of glucocorticoid and estrogen receptors (53). The analogy is particularly germane given that CBP is a coactivator for both AP-1 family members and for CREB.

The importance of understanding alternate pathways in a given cell type is underscored by comparing data presented here with data presented in the ChIP study of ERα regulation of pS2 in breast cell lines (54). pS2 is regulated via a classic ERE pathway and is one of the standard genes used to determine mechanisms of estrogen actions. Although most of the data reported in the Metivier et al. (54) study was collected at 5-min intervals, in one set of experiments, data collected every minute revealed increased ERα occupancy beginning at 5–6 min, as opposed to the increase we observe at 1 min (Fig. 5). Whether such differences in the timing of occupancy are a result of pathway, or cell context, the point remains that one proposed mechanism of ER action is far from the only mechanism by which estrogens and their receptors can regulate gene expression.

The work presented here raises a number of questions related to CREB’s role in E2 activation of crh expression. The ChIP data reveal that E2 treatment increases pCREB occupancy of the crh promoter (Fig. 6). This is not surprising, because E2 has been shown to phosphorylate CREB in several brain regions (55,56,57) and in cultured hippocampal neurons (58). It is possible, then, that E2 stimulates crh gene expression via two parallel events: 1) one in which E2 leads to increased intracellular cAMP levels (59), which in turn lead to activation of protein kinase A and subsequent pCREB regulation of crh, and 2) one in which E2 binds ERα or -β, which leads to a coactivator complex competent to enhance crh expression through the proximal CRE.

With respect to a complex that forms at 1 min, it is clear that ERα, SRC-1, and pCREB are in the region of the CRE at this time (Fig. 8, A and B). It is not clear, however, what the intermediary factor is that could bridge pCREB to a SRC-1/ERα complex. Certainly, one coregulator that needs to be ruled in or out is CBP, which has been shown previously to interact directly with ERα (60). It is not portrayed in the complex at 1 min (Fig. 9) because its occupancy was not statistically significant at this time. However, there was a trend (Fig. 6; Z = 1.3416; P < 0.090), suggesting that in slightly different conditions it may indeed be the bridging molecule, X (Fig. 9). As for H3 acetylation, the most parsimonious explanation is that it is a result of the histone acetyl-transferase activity of SRC-1 (61). However, it could also be the p300/CBP-associated factor (62), with which SRC-1 has been shown to associate (61).

Figure 8.

Compilation of time courses indicates a sequential and ordered assembly of ERα and ERβ complexes and H3 and H4 acetylation states. A, Data shown in Figs. 3 and 5, superimposed; B, data shown in Figs. 5–7, superimposed. pCREB and SRC-1 occupancy peak at 1 min, as does Ac-H3. CBP peaks at 3 min, as does Ac-H4. Total CREB is omitted because the increase was not significant at any of the times shown.

Figure 9.

Putative ER complexes at the crh CRE. ERα interacts with the SRC-1 coactivator present in a pCREB transcriptional complex. An ERα-SRC-1 complex could bind directly to pCREB or via an intermediary factor, as shown. The ERα complex would promote acetylation of H3 in the region of the CRE. In distinction, ERβ interacts with CBP either via direct or indirect interactions. CBP, in turn, would associate with a CREB family member bound to the CRE. In this case, CBP would promote acetylation of H4.

As for the putative complex at 3 min, it is tempting to portray a direct interaction of ERβ with CBP; however, even though CBP has been shown to interact with ERα (60), an interaction with ERβ has been far less studied, and it is not clear that there is in fact a direct ERβ-CBP interaction. Thus, we have included a bridging protein Y (Fig. 9). As for the CREB family member, activating transcription factor-2 (ATF-2) has been shown to interact with CBP (63) and thus is one candidate for interaction with the proximal crh CRE. H4 is a known target of p300/CBP acetylation (see Ref. 64 for review). Thus, in this context, it is likely to participate in H4 acetylation.

Taken together, our results are consistent with the hypothesis that E2 regulates crh by leading to the integration of coactivator activity at a CRE. This in turn regulates CRH mRNA expression in a dynamic fashion. Further corroboration of this hypothesis requires additional promoter analysis, e.g. walking the promoter with overlapping sets of primers. The requirement for CREB also deserves further analysis, such as performing ChIP experiments in the presence of a dominant-negative CREB and/or small interfering RNA directed against CREB. Lastly, it will be important to provide biochemical evidence, such as mass spectroscopy, to determine whether ERα and -β are present in different complexes. Even though much work needs to be done, the postulate that ERα and ERβ associate sequentially with different proteins is novel and intriguing. It will be important to determine whether temporally distinct ERα and ERβ complexes form at the crh promoter in other neuronal cell lines and in vivo, because the timing of their formation and composition may contribute to cell-specific crh regulation.