help button home button Endocrine Society Molecular Endocrinology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

Molecular Endocrinology, doi:10.1210/me.2006-0361
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
21/2/415    most recent
Author Manuscript (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow NURSA Molecule Pages Link
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Dammer, E. B.
Right arrow Articles by Sewer, M. B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Dammer, E. B.
Right arrow Articles by Sewer, M. B.
Molecular Endocrinology 21 (2): 415-438
Copyright © 2007 by The Endocrine Society

Coregulator Exchange and Sphingosine-Sensitive Cooperativity of Steroidogenic Factor-1, General Control Nonderepressed 5, p54, and p160 Coactivators Regulate Cyclic Adenosine 3',5'-Monophosphate-Dependent Cytochrome P450c17 Transcription Rate

Eric B. Dammer, Adam Leon and Marion B. Sewer

School of Biology, Parker H. Petit Institute for Bioengineering & Biosciences, Georgia Institute of Technology, Atlanta, Georgia 30332-0230

Address all correspondence and requests for reprints to: Marion B. Sewer, School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, Georgia 30332-0230. E-mail: marion.sewer{at}biology.gatech.edu.


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 Note Added in Proof
 REFERENCES
 
Transcription of the cytochrome P450 17 (CYP17) gene is regulated by cAMP-dependent binding of steroidogenic factor-1 (SF-1) to its promoter in the adrenal cortex. Using temporal chromatin immunoprecipitation and mammalian two-hybrid experiments, we establish the reciprocal presence of coactivators [general control nonderepressed (GCN5), cAMP response element-binding protein-binding protein, p300, p300/cAMP response element-binding protein-binding protein CBP associated factor, p160s, polypyrimidine tract associated splicing factor, and p54nrb], corepressors (class I histone deacetylases, receptor interacting protein, nuclear receptor corepressor, and Sin3A), and SWI/SNF (human homolog of yeast mating type switching/sucrose nonfermenting) and imitation SWI chromatin remodeling ATPases on the CYP17 promoter during transcription cycles in the H295R adrenocortical cell line. A ternary GCN5/SRC-1/SF-1 complex forms on the CYP17 promoter with cAMP-dependence within 30 min of cAMP stimulation, and corresponds with SWI/SNF chromatin remodeling. This complex is sensitive to the SF-1 antagonist sphingosine and results in decreased transcription of CYP17. GCN5 acetyltransferase activity and carboxy terminus binding proteins alternatively mediate disassembly of the complex. This work establishes the temporal order of cAMP-induced events on the promoter of a key steroidogenic gene during SF-1-mediated transcription.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 Note Added in Proof
 REFERENCES
 
IN THE ADRENAL CORTEX, the peptide hormone ACTH regulates transcription of cytochrome P450 17 (CYP17), one of the genes required for cortisol synthesis, by activating a cAMP-dependent pathway (1, 2). We have previously shown that ACTH/cAMP increases human CYP17 gene expression by promoting the binding of a protein complex containing steroidogenic factor-1 (SF-1), a 54-kDa nuclear RNA and DNA binding protein (p54nrb), and polypyrimidine-tract binding protein-associated splicing factor (PSF) to a 20 bp region between –57 and –37 of the CYP17 promoter (3). The affinity of this SF-1/p54nrb/PSF complex for this region of the promoter is induced by cAMP and is dependent on phosphatase activity (4). PSF and p54nrb bind the C terminus of RNA polymerase II (Pol II) and preferentially mediate 5'-splicing of pre-mRNA (5, 6). Depending on the promoter context, they are also considered as coactivators (7) or corepressors (8) of transcription.

SF-1 (NR5A1, Ad4BP) is a member of the nuclear receptor superfamily that plays many roles in the regulation of genes involved in steroid hormone biosynthesis, endocrine development, and sex differentiation (9, 10, 11, 12, 13, 14, 15, 16). Targeted disruption of SF-1 in mice resulted in adrenal and gonadal agenesis (17), absence of a differentiated ventromedial hypothalamic nucleus (15, 18), and impaired expression of pituitary gonadotropins (19).

Coregulators of transcription are intimately involved in nuclear receptor-mediated transcriptional activation (20, 21, 22, 23, 24) and trans-activation by other factors, with roles in histone modification (25, 26, 27, 28, 29), chromatin remodeling (30), promoter clearance coupled to transcription (31, 32), transcription-coupled pre-mRNA splicing (6, 7, 33, 34), posttranslational modification of trans-activation complex members (35, 36, 37, 38), and ordered, cooperative, recruitment of basal transcription machinery and other coregulators (28, 39, 40). Many coregulators, including steroid receptor coactivators (SRCs, p160s), are recruited to nuclear receptors in response to ligand-induced conformational changes that provide a mechanism for modulation of receptor activation of transcription in clinical contexts (41, 42). In the case of estrogen receptor (ER){alpha}, the coregulatorsome includes ubiquitination and proteasomal degradation machinery that clears the promoter of the receptor for subsequent rounds of transcription (31, 40), as well as chromatin-remodeling complexes that return chromatin to a repressive conformation immediately after the initiation of transcription in some cases (40). Many laboratories have hypothesized that the cycling of nuclear receptors on and off of chromatin (31) and histone repositioning (40, 43) enable a near real-time response to changing afferent signals that are integrated into the rate of gene expression by coregulators interacting with target gene-specific transcription factors (39). Cycles of transcription on ER{alpha} target promoters (cathepsin D and pS2) in MCF-7 cells occur without fluctuations in coregulator or transcription factor protein levels, but are a cascade of protein-DNA and protein-protein interactions, which result from ligand or signal induction of the nuclear receptor (44).

Temporal cycling of SF-1 and p160 coregulators on the Mc2r promoter has been elegantly demonstrated in Y1 cells and adrenocortical cells of ACTH-treated mice (45). Likewise, in vitro and ex vivo chromatin immunoprecipitation (ChIP) studies on the steroidogenic acute regulatory protein proximal promoter in MA-10 Leydig cells or primary mouse granulosa cells showed a pattern of rapid, transient binding for SF-1 and other transcription factors dependent on cAMP or chorionic gonadotropin treatment time (46).

The aim of the current study was to further characterize dynamics of SF-1 and coregulator cycling on the CYP17 promoter in H295R human adrenocortical cells. Temporal ChIP analysis was carried out to determine the kinetics of SF-1 and coregulator binding to the CYP17 promoter. We also determined the effect of cAMP and complex assembly on the acetylation and methylation of histone lysine residues at the CYP17 promoter, and on the retention of intact histone H2 dimers at this site. We show that cAMP-dependent coregulator recruitment proceeds as a distinct sequence of transient, cyclic binding events on the CYP17 promoter, and that the recruitment of specific acetyltransferases and methyltransferases corresponds with specific covalent histone modifications, and with cyclic, reciprocal exchange of corepressors for coactivators. Recruitment of ATP-dependent chromatin remodeling activity coincides with radical modification of histone H2 promoter occupancy before and after transcription. The intrinsic activity of coregulators, such as the acetyltransferase activity of general control nonderepressed (GCN)5, is shown to have a role in the succession of events occurring during transcription cycles, whereas nicotinamide adenine dinucleotide (NADH) binding to carboxy terminus binding proteins (CtBPs) in the nucleus modulates cooperative interactions of coregulators in these events. Cycles of SF-1-mediated transcriptional activation are also sensitive to the previously identified SF-1-specific antagonist Sph (47), via inhibition of cooperative interaction of the earliest coactivator complexes with SF-1 and CYP17 promoter chromatin. These findings are interpreted with regard to general mechanisms underlying regulation of transcription rate.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 Note Added in Proof
 REFERENCES
 
ACTH, cAMP Induce Cycles of SF-1 Binding to the CYP17 Proximal Promoter
We have previously shown that cAMP stimulates SF-1-dependent transcription by promoting the binding of a complex containing SF-1 and the splicing factors p54nrb and PSF to the cAMP-responsive sequence of the CYP17 promoter (3). Moreover, we have also demonstrated that both ligand binding (47) and phosphorylation (48) modulate the activity of SF-1. Various coregulators of transcriptional activation are known to interact with SF-1 and modulate its ability to transactivate target genes, especially in response to cAMP or protein kinase A (PKA) (37, 49, 50, 51, 52, 53, 54). To examine the kinetics of recruitment and combinatorial effects of a panel of coregulators on SF-1-responsive gene promoters, we treated {alpha}-amanitin-synchronized H295R adrenocortical cells with 1 mM dibutyryl-cAMP (Bt2cAMP) for time periods ranging from 15–240 min and carried out ChIP for SF-1. Primers for ChIP of the CYP17 promoter and transcription start site were designed to most likely capture the state of a single nucleosome, with 147 bp of wrapped DNA in a canonical model. In addition, real-time PCR requires an amplicon length of more than 100 bp for maximum sensitivity. Therefore, the region selected, –104/+43, was chosen to match these criteria, with the –57/–37 SF-1 binding site centrally located.

Both ACTH and Bt2cAMP increased the binding of SF-1 to the CYP17 promoter; however, the relative enrichment of SF-1-bound CYP17 promoters mediated by Bt2cAMP was greater than that seen for ACTH (Fig. 1AGo). The ACTH/Bt2cAMP-stimulated SF-1 binding initially occurs in 120-min cycles (Fig. 1AGo), although elevated SF-1 occupancy of the promoter occurs for less than 90 min. Peaks indicate a stochastic time for maximal recruitment to the promoter within the population of synchronized cells (55). To determine whether the periodic binding of SF-1 results in cyclical changes in transcription, we quantified the changes in CYP17 reporter gene expression in synchronized cells over the same 240-min time frame. As shown in Fig. 1CGo, both ACTH and Bt2cAMP treatment resulted in cyclical increases in CYP17 luciferase activity. However, the periodicity of reporter expression did not overlap with SF-1 binding to endogenous promoter at early time points. CYP17 reporter gene activity peaked at the 30-, 90-, and 210-min time points, suggesting that rhythms of ACTH/cAMP-stimulated SF-1 binding to the CYP17 promoter correspond with an overall increase in P450c17 transcription rate, and these transcription rhythms can be brought forward into corresponding rhythmic protein accumulation.


Figure 1
View larger version (22K):
[in this window]
[in a new window]

 
Fig. 1. ACTH and cAMP Induce Cyclic Binding of SF-1 to the CYP17 Promoter and Increased CYP17 Transcription

A, H295R cells were synchronized for 2 h with 2.5 µM {alpha}-amanitin, and then treated with 50 nM ACTH or 1 mM Bt2cAMP for the indicated times and subjected to ChIP using a polyclonal SF-1 antibody. Real-time PCR analysis of ChIP DNA. Output data are normalized to values obtained for 1% input controls, and results are presented as percent of baseline value obtained for untreated cells at each time point. Graphed data represent the mean ± SEM from 11 experiments performed in duplicate. B, Purified DNA from ChIP analysis of Bt2cAMP-treated cells was amplified PCR with primers for the –104/+43 region of the CYP17 gene and the samples resolved on a 2% agarose gel. Results from a representative experiment are shown. C, H295R cells subcultured onto 24-well plates and transfected with 125 ng pGL3-CYP17 2x57 and 2 ng pRL-TK for 48 h, and then treated with 2.5 µM {alpha}-amanitin for 2 h. Synchronized cells were washed twice with PBS, and then treated with either 50 nM ACTH or 1 mM Bt2cAMP for 30–240 min. Lysates were isolated and luciferase activity quantified by luminometry as described in Materials and Methods. Graphed data are expressed as fold of the control group mean for each time point and represents mean ± SEM from two experiments performed in quadruplicate.

 
trans-Activation of CYP17 Follows SF-1 Binding
Basal transcription machinery as well as nuclear receptors are lost (31), and promoter histone modifications are uniformly reset to patterns associated with low levels of transcription (40) on the promoter when cells are exposed to {alpha}-amanitin in the absence of signals that induce transcription; therefore we extended our ChIP assay to examine the dynamics of RNA polymerase II (Pol II) recruitment to the proximal CYP17 promoter during SF-1 transactivation, particularly after peaks of SF-1 recruitment at 60 min (cycle I) and 180 min (cycle II) of ACTH or Bt2cAMP stimulation. The antibody used recognizes the phosphorylated C-terminal domain heptad repeat. Sixty minutes of both ACTH and Bt2cAMP increase Pol II recruitment, which decreases more slowly than Bt2cAMP-stimuated SF-1 binding; we reference the beginning of SF-1 recruitment until the loss of Pol II as transcription cycle I (Fig. 2Go, A and B). In cycle II, Bt2cAMP promotes Pol II recruitment with more complex dynamics culminating 30 min after peak binding of SF-1 on the CYP17 promoter at 180 min. In contrast to the cycle of Pol II enrichment seen in response to Bt2cAMP, peak ACTH-stimulated Pol II binding in cycle I occurred at the 90-min time point. Further, the comparative effect of ACTH on Pol II recruitment in cycle II was an earlier increase that was sustained longer as compared with Bt2cAMP-stimulated Pol II binding (Fig. 2AGo). SF-1 and Pol II binding after acute 5- and 15-min exposure to ACTH or Bt2cAMP revealed no additional early peaks in binding before the first transcription cycle (data not shown). Because kinetics of SF-1 binding to CYP17 promoter is indistinguishable between Bt2cAMP and ACTH stimulation (Fig. 1AGo), there is a better incremental burst of protein expression with Bt2cAMP (Fig. 1CbGo), and because timed separation of transcription cycles is greater with Bt2cAMP than with ACTH (Fig. 2AGo), the cAMP analog was used as a surrogate for ACTH stimulation in subsequent experiments. SF-1 binding to the CYP17 promoter in response to Bt2cAMP is shown in all subsequent ChIP figures for comparison.


Figure 2
View larger version (44K):
[in this window]
[in a new window]

 
Fig. 2. SF-1 Transcription Cycles Correlate with Histone Acetylation and HAT and Pol II Recruitment

A, Graphic analysis of relative promoter binding of Pol II in response to ACTH and Bt2cAMP. Temporal ChIP assays were performed as described in Materials and Methods and DNA was amplified by quantitative PCR using primers targeted at region –104/+43 of the CYP17 promoter. Two transcription cycles (I and II) are indicated. Graphed data represent the mean ± SEM from three experiments performed in duplicate. B, Representative ethidium bromide-stained agarose gel of temporal ChIP for Bt2cAMP-stimulated Pol II binding to the CYP17 promoter. C, Temporal ChIP analysis of Bt2cAMP-stimulated acetylation of histone H3 and histone H4. Sheared chromatin was immunoprecipitated with anti-SF1, antiacetyl histone H3, or antiacetyl histone H4 antibodies. Output data are normalized to {Delta}Ct values obtained for 1% input controls, and results are presented as percent of baseline value obtained for untreated cells at each time point. D, Graphic analysis of temporal ChIP for HAT recruitment to the CYP17 promoter during Bt2cAMP stimulation. {alpha}-Amanitin-synchronized H295R cells were treated for time periods ranging from 30 min to 4 h with 1 mM Bt2cAMP and exposed to 1% formaldehyde, and purified lysates were immunoprecipitated using antibodies directed against SF-1, GCN5, p300, CBP, and P/CAF. Output data are normalized to values obtained for 1% input controls, and results are presented as percent of baseline value obtained for untreated cells at each time point. Graphed data represent the mean from two experiments performed in duplicate. E, Time course of Bt2cAMP-stimulated recruitment of p160 coactivators to the CYP17 promoter. ChIP was performed on lysates purified from Bt2cAMP-treated synchronized cells using antibodies for SF-1, SRC-1, GRIP-1, and ACTR. Outputs are normalized to {Delta}Ct values obtained for 1% input controls, and results are presented as percent of {delta}Ct values for untreated cells at the corresponding time point. Graphed data represent the mean from two experiments performed in duplicate.

 
Histone H3 and H4 Acetylation Precedes Pol II Recruitment
We postulated that histone acetylation is altered during induction of SF-1-dependent transactivation, and that these changes must precede changes in Pol II occupancy of the proximal CYP17 promoter, as has been demonstrated for SF-1-responsive promoters (45, 46) as well as many other genes in their native context of chromatin (56). ChIP with an antibody for acetylated histone H4 correlates with subsequent Pol II recruitment in both cycles of SF-1 mediated transcription in response to both ACTH and Bt2cAMP, whereas histone H3 acetylation at Lys 9 and 14 coincides with the presence of Pol II in cycle II (Fig. 2CGo). Histone H4 acetylation is decreased just before or as Pol II moves from the –104/+43 region of the promoter, whereas acetylation of H3 remains after cycle II until at least the 240-min time point. Thus, histone H4 acetylation occurs by the time that Pol II binds the proximal promoter, whereas H3 acetylation is not strictly required for Pol II recruitment but does occur at high levels throughout transcription cycle II.

Histone Acetyltransferase (HAT) Recruitment Correlates with Histone Acetylation
We next performed temporal ChIP experiments to determine the kinetics of binding of coactivators with HAT activity (Fig. 2DGo). GCN5 binding peaks at 30 min and again at 180 min of stimulation. GCN5 recruitment coincides with histone H4 acetylation in both transcription cycles (Fig. 2BGo). Early in cycle II, p300 rapidly binds between 120 and 150 min, and this binding coincides with histone H3 acetylation specific to the second transcription cycle (Fig. 2CGo).

HAT binding events that occur after peak SF-1 binding vary between the first and second transcription cycles. CREB-binding protein (CBP) binds preferentially in cycle I, in phase with SF-1 (Fig. 2DGo), and this binding does not correlate with H3 or H4 acetylation. The GCN5 paralog P/CAF (p300/CBP-associated factor) binds in phase with GCN5 but with lower abundance in both cycles; yet, during cycle II at 210 min of stimulation, there is pronounced P/CAF binding independent of GCN5 recruitment 30 min after peak SF-1 binding, and this corresponds with a further increase in H3 acetylation after SF-1 begins to vacate the promoter during the second transcription cycle (Fig. 2CGo).

Each Transcription Cycle Has a Unique Profile of p160 Binding
Only two of the three characterized members of the p160 family of coactivators [SRC-1, glucocorticoid receptor-interacting protein 1 (GRIP-1), activator of thyroid and retinoic acid receptor (ACTR) (see Table 1Go for alternate names)] have confirmed intrinsic HAT activity (26, 27). All three p160s serve as scaffolding for recruitment of other HATs to nuclear receptors (26, 27, 57, 58); thus, they are good candidates for bridging HATs with weak or no ligand-dependent interaction motifs to the SF-1 activation function-2 (AF-2) motif. In temporal ChIP, p160s show binding that mirrors increases in SF-1 on the promoter in both cycles, but in the first transcription cycle, SRC-1 rapidly binds within the first 30 min, whereas SF-1 is recruited for an additional 30 min (Fig. 2EGo). Notably, only SRC-1 is enriched on the promoter in the first transcription cycle. On the other hand, multiple overlapping peaks of lower intensity in the second transcription cycle indicate that p160s may be interchangeable in the second SF-1 transcription cycle.


View this table:
[in this window]
[in a new window]

 
Table 1. Antibodies for ChIP and Nomenclature Cross-Reference

 
GCN5 Interaction with SF-1 Is Strengthened by p160 Coactivators
Initial GCN5 binding coincides with SRC-1 in intensity and timing (cf. Fig. 2Go, D and E), and GCN5 stimulates expression of a SF-1-responsive reporter only in the presence of SRC-1 (group 2 vs. group 6 in supplemental Fig. 1 published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org) and only with SF-1 that has an intact ligand binding pocket (group 6 vs. group 8). We have previously found that sphingosine (Sph) inhibits the ability of SRC-1 to coactivate SF-1-dependent transcription (47). These data suggest the possibility of a concomitant GCN5, SRC-1, and SF-1 interaction on the promoter. Such a complex has been shown to form in yeast cells expressing human thyroid hormone receptor, GCN5, and SRC-1 or GRIP-1 (59). The ability of SRC-1 to mediate the interaction between GCN5 and SF-1 was tested in mammalian two-hybrid experiments in H295R cells. SRC-1 promotes interaction of GCN5 with SF-1 in a dose-dependent manner, and this interaction is potentiated by Bt2cAMP (Fig. 3AGo). Bt2cAMP-stimulated interaction is decreased by cotreatment with the SF-1 antagonist Sph (Fig. 3BGo), suggesting antagonist dissociation is required for cAMP-dependent SRC-1 recruitment of GCN5. GRIP-1 was also able to potentiate interaction of SF-1 with GCN5 (data not shown); however, GRIP-1 coexpression with SRC-1 modestly increased CYP17 reporter expression only in the absence of Bt2cAMP compared with SRC-1 alone, and GRIP-1 alone did not potentiate GCN5 coactivation of this reporter (supplemental Fig. 1, group 7), consistent with a different interaction mechanism.


Figure 3
View larger version (50K):
[in this window]
[in a new window]

 
Fig. 3. p160 Dose-Dependent Interaction of GCN5 with NR5A Receptors Is Mediated by SRC-1 and Sensitive to SF-1 Antagonist

A, Mammalian two-hybrid experiments were performed by transfecting H295R cells with 50 ng pG5 luciferase reporter, 25 ng pBind-SF-1, 15 ng pAct vector, or pAct-GCN5, and SRC-1 using GeneJuice. Cells were treated with 1 mM Bt2cAMP 24 h after transfection, and lysates were harvested 16 h later for dual luciferase assays. B, Mammalian two-hybrid experiments were carried out as described in Materials and Methods. Transfected cells were treated for 16 h with 1 mM Bt2cAMP in the presence and absence of 1 µM Sph and then lysed for dual luciferase assays. C, Cells were transfected for 30 h with the two-hybrid plasmids pG5, pAct-GCN5, and pBind-LRH-1 and expression plasmids for SRC-1, GRIP-1, or ACTR and harvested for dual luciferase assays. D,) H295R cells were transfected with pG5, pBind-SF-1, pAct-GCN5, pAct-GCN5 E214Q, and pBKCMV-SRC-1, and then incubated for 16 h with 1 mM Bt2cAMP. Lysates were isolated and subjected to dual luciferase assays. Data presented in all panels are normalized to Renilla activity (pAct) and represent the mean ± SEM from two experiments performed in triplicate. Statistically significant difference between cells transfected with the pAct empty vector vs. cells transfected with pAct-GCN5 are denoted as follows: {ddagger}, P < 0.05; {ddagger} {ddagger} {ddagger}, P < 0.001. Asterisk denotes statistically significant effect of SRC-1 overexpression: *, P < 0.05, and ***, P < 0.001. #, P < 0.05; or # # #, P < 0.001 denote statistically significant difference compared with pAct-GCN5/pBind-SF-1 transfected cells with control treatment. Ampersand denotes statistically significant difference between untreated and Bt2cAMP-treated cells: &, P < 0.01, and &&, P < 0.01. Caret (^, P < 0.05) denotes statistically significant difference between cells transfected with wild-type GCN5 vs. cells transfected with the GCN5 E214Q mutant. Dollar sign ($$, P < 0.01; and $$$, P < 0.001) denotes statistically significant effect of Sph.

 
To determine the specificity of the GCN5/SRC-1 interaction with nuclear receptors of the NR5A subfamily, we asked whether a complex forms between GCN5, SRC-1, and the SF-1 ortholog liver receptor homolog-1 (LRH-1). We tested for p160-mediated interaction of GCN5 with LRH-1 in the mammalian two-hybrid system in Jeg3 cells. This complex does indeed form and SRC-1, but neither GRIP-1 nor ACTR, potentiates the LRH-1/GCN5 interaction (Fig. 3CGo). The SF-1-specific antagonist Sph has no effect on the LRH-1/GCN5 interaction (data not shown).

GCN5 Acetyltransferase Activity Limits Interaction with SF-1
We hypothesized that acetylation of SF-1 by GCN5 may be a mechanism that ensures the SF-1/p160/GCN5 complex is transient, as observed in the ChIP time course (Fig. 2Go, D and E). To test this hypothesis, we constructed a GCN5 acetyltransferase catalytic site-defective mutant E214Q (60) and repeated two-hybrid experiments in H295R cells. Interestingly, the formation of the SRC-1/SF-1/GCN5 complex was strengthened in the mutant (Fig. 3DGo). These data indicate that GCN5 acetyltransferase activity may destabilize the transient GCN5/SF-1/SRC-1 interaction and suggest that acetylation of SF-1 may be an underlying mechanism. Collectively, the two-hybrid data suggest that acetylation of an unknown target by GCN5 promotes dissociation of the GCN5/SRC-1/SF-1 complex.

Class I HDACs Have Two Roles in Transcription Cycles Mediated by SF-1
There are two major classes of HDACs: class I HDACs (human HDACs 1–3, and 8), which share sequence homology with the yeast transcriptional repressor Rpd3, and class II HDACs (human HDACs 4–7, 9, and 10), which are generally larger and expressed in a more tissue-specific manner, were identified by their homology to yeast Hda1 (61). To determine which class I HDACs are responsible for the loss of histone acetylation on the CYP17 promoter (Fig. 2BGo), we performed temporal ChIP. HDAC8 recruitment occurs within 30 min of Bt2cAMP stimulation, whereas an increase in HDAC1 recruitment peaks at 90 min (Fig. 4AGo). The former peak coincides with early coactivator cooperativity whereas the latter peak coincides with decreases in Pol II occupancy (Fig. 2BGo) and a loss of histone H4 acetylation at this time (Fig. 2CGo). At 120 min, when SF-1 binding to the promoter is at a minimum, HDAC2 is maximally recruited, and whereas levels of this HDAC decline during SF-1 recruitment in the second transcription cycle, they again wax with loss of SF-1 from the promoter at 210 min (Fig. 4AGo). HDAC3 and HDAC8 modestly increase after 120 min until 210 min in the second transcription cycle.


Figure 4
View larger version (23K):
[in this window]
[in a new window]

 
Fig. 4. HDACs and Corepressors Show Promoter Binding Reciprocal to that of SF-1 but Compatible with an Increase in p54nrb and PSF Splicing Factors

A, Temporal ChIP for coregulator protein binding to the CYP17 promoter was carried out as described in Materials and Methods. Lysates were immunoprecipitated using antibodies against SF-1, HDAC1, HDAC2, HDAC3, and HDAC8. B, Time course of corepressor recruitment to the CYP17 promoter in response to Bt2cAMP stimulation. {alpha}-Amanitin-synchronized H295R cells were treated for 30 min to 4 h with 1 mM Bt2cAMP and cross-linked using 1% formaldehyde. Purified lysates containing sheared chromatin were immunoprecipitated using anti-SF-1, anti-NCoR, anti-Sin3A, anti- RIP140, and anti-SMRT antibodies. C, Temporal ChIP of cAMP-dependent binding of p54nrb and PSF splicing factors, SF-1 and Pol II, to the CYP17 promoter. Temporal ChIP for coregulator protein binding to the CYP17 promoter was carried out as described in Materials and Methods. Lysates were immunoprecipitated with antibodies against SF-1, p54nrb, PSF, and Pol II. Data graphed in all panels represent the mean from at least two experiments, each performed in duplicate. Outputs are normalized to {Delta}Ct values obtained for 1% input controls, and results are presented as percent of {Delta}Ct values for untreated cells at each time point.

 
Corepressor Recruitment Reciprocates SF-1 Loss from the CYP17 Promoter
HDACs are often associated with corepressor complexes brought to the promoter by scaffold proteins with repression domains such as nuclear receptor coprepressor (NcoR) or silencing mediator of retinoid and thyroid hormone receptor (SMRT), which can bind nuclear receptors directly in the absence of agonist (62, 63), with dependence on partial agonist (64) or antagonist (47), or in the case of receptor-interacting protein 140 (RIP140), with agonist dependence (65). Notably, temporal ChIP of these nuclear receptor corepressors shows recruitment that reciprocates SF-1 loss from the promoter. NCoR has modest recruitment at 90 min of stimulation, coinciding with HDAC1 recruitment (Fig. 4BGo). RIP140 and Sin3A may be members of a repression complex that is recruited to the promoter in the relative absence of SF-1 at the 120-min time point, coinciding with initial recruitment of HDAC2. Taken together, these data indicate that HDAC1 and 2 are recruited within NCoR and RIP140/Sin3A corepressor complexes, respectively, and these putative complexes bind at the CYP17 promoter during cAMP stimulation without dependence on SF-1. Moreover, these data confirm our previous findings demonstrating that Sin3A mediates CYP17 repression (3).

Corepressor Clearance Coincides with p54nrb/PSF Recruitment and Precedes SF-1 Recruitment
We have shown that the interaction of p54nrb and PSF splicing factors with SF-1 on the CYP17 promoter is stimulated by cAMP (3). Moreover, PSF has been shown to mediate nuclear receptor interactions with Sin3A and associated HDACs (8). Thus, we postulated that these splicing factors may be involved in the dynamics of corepressor-containing complexes and SF-1. Unexpectedly, PSF and p54nrb binding increase in a SF-1-independent manner at 150 min (Fig. 4CGo). Initial recruitment of splicing factors by 150 min reciprocates a loss of corepressors and precedes SF-1 binding in cycle II and also parallels an initial increase in Pol II promoter occupancy at this time (Figs. 2AGo and 4CGo). PSF and p54nrb remain associated with the promoter as SF-1 and Pol II vacate the promoter for the second time by 240 min of Bt2cAMP stimulation. These findings, in combination with our earlier study (3), establish a time course of cAMP-dependent assembly of the p54nrb/PSF/SF-1 complex on the –57/–38 region of the CYP17 promoter and suggest that PSF and p54nrb promote or, at least, are retained during clearance of RIP140, Sin3A, and HDAC2 while SF-1 is recruited. Independent dynamics of promoter occupancy by p54nrb, PSF, and SF-1 indicate unexpected plasticity in their interactions with each other and with promoter DNA and/or other coregulatory proteins bound to the CYP17 promoter.

p54nrb Also Enables Assembly of a HAT/p160 Coactivator Complex
GCN5 and p300 occupancy of the promoter increase before that of SF-1 in the second transcription cycle (Fig. 2DGo). Therefore, we considered the possibility that these HATs may be recruited to the cAMP-responsive p54nrb/PSF transcription complex and together act to initiate the second cycle of SF-1-mediated transcription. We tested for Bt2cAMP-dependent interaction of GCN5 with p54nrb in the mammalian two-hybrid system in H295R cells. This interaction does occur and is potentiated by SRC-1 but decreased by PSF (Fig. 5AGo).


Figure 5
View larger version (29K):
[in this window]
[in a new window]

 
Fig. 5. GCN5 Interaction with SF-1 Can Also Occur via PSF-Sensitive Complexes Containing p54nrb and p160 Coactivators

A, H295R cells were transfected with pG5, pBind-GCN5, pAct-p54nrb, pBKCMV-SRC-1, and pCR3.1-PSF using Gene Juice. After 24 h, transfected cells were treated with 1 mM Bt2cAMP and then harvested for quantification of reporter gene activity. B, Cells were transfected with pG5, pBind-GCN5, and pAct-GRIP-1 in the presence and absence of pCR3.1-p54nrb, pCR3.1-PSF, and/or pcDNA3.1-SF-1, treated with Bt2cAMP, and then harvested for dual luciferase assays. Data graphed in both panels are from two experiments performed in triplicate and normalized to Renilla expression from the pAct vector ± SEM. Statistically significant difference between cells expressing pAct-p54nrb and cells expressing the pAct empty vector is denoted as follows. {ddagger}, P < 0.05; {ddagger}{ddagger}{ddagger}, P < 0.001. Asterisk denotes statistically significant effect of SRC-1 overexpression (A) and statistically significant effect of SF-1 and p54nrb overexpression (B): ***, P < 0.001. Ampersand denotes statistically significant difference between untreated and Bt2cAMP-treated cells: &, P < 0.05; and &&, P < 0.01. Caret denotes statistically significant effect of PSF: ^, P < 0.05; and ^, P < 0.01, comparison with same treatment and transfection without PSF.

 
We next asked whether the interaction between a p160 coactivator specific to cycle II and GCN5 is fostered by p54nrb using the two-hybrid system to detect p54nrb -dependent assembly of a p160/GCN5 complex. A higher order complex among these factors and SF-1 would coincide with simultaneous occupancy on the promoter in the second SF-1-dependent transcription cycle in ChIP time courses, when p54nrb, GCN5, and all three p160 coactivators are enriched (see Fig. 2Go, D and E, and Fig. 4CGo). Modest Bt2cAMP-dependent interaction between GCN5 and GRIP-1 is strengthened only when both SF-1 and p54nrb are coexpressed (Fig. 5BGo). Like the GCN5/p54nrb interaction, this complex is sensitive to PSF. The kinetics of GCN5 binding coincide additively with both p54nrb and SF-1 binding from 120- to 210-min time points (Figs. 2DGo and 4CGo). Together, these data implicate the assembly of a GCN5/p54nrb complex on the CYP17 promoter in H295R cells during cycle II before recruitment of p160s and SF-1 (see Discussion and Fig. 11AGo). SF-1 recruitment between 150 and 180 min destabilizes PSF interaction with the promoter (Fig. 4CGo) and further stabilizes GCN5 and p160 binding in the complex.


Figure 11
View larger version (54K):
[in this window]
[in a new window]

 
Fig. 11. Model of Coregulator Dynamics on the CYP17 Promoter during cAMP Stimulation in Steroidogenic Cells

cAMP initiates formation of a SF-1/SRC-1/GCN5 trimer that associates with the CYP17 promoter (–57/–37) followed by acetylation of histone H4 and another component of the complex. CtBP (not shown) mediates complex disruption, thereby facilitating recruitment of transcription activating factors (TAFs) and Pol II. Corepressors bind in the absence of SF-1 during cAMP-mediated cycling, followed by exchange of the corepressor complex for a complex containing p54nrb and GCN5. This is followed by acetylation and monomethylation of histone H4, which promotes SWI/SNF-dependent remodeling and enables the cooperative binding of SF-1, p160s, and GCN5. Additional remodeling by BRG1 reenables binding of Pol II and TAFs followed by dismissal of SF-1 and coactivators and the initiation of p54nrb/PSF-mediated splicing. **, cAMP-dependent interactions confirmed by this study. Ac, Acetylation; snRNP, small nuclear riboprotein.

 
CtBP Recruitment Corresponds with Exchange of Transcription Activators for Repressors on the CYP17 Promoter
CtBP corepressors were initially identified and characterized due to their ability to suppress transformation by the E1A viral oncoprotein (66) and may also have a role in cases of myeloid leukemia that involve the acute myeloid leukemia 1/myelodysplastic syndrome 1/ectopic viral integration site 1 (AME) fusion oncoprotein, both to promote oncogenic cellular replication (67) and to enable oncogenic transformation (68) via repression of AME-dysregulated genes. A recent report by Zhang et al. (69) shows that CtBP1 can repress E cadherin via promoter binding, inducing cancer cell migration in response to increases in the nuclear NADH:nicotinamide adenine dinucleotide (NAD)+ ratio, which occurs during acute hypoxia or with chemical manipulation of cellular NADH levels. This mechanism requires a DNA targeting factor with CtBP interaction motifs, and CtBP is thought to bridge such factors to HDACs (70); thus, this repression mechanism is sensitive to deacetylase inhibition.

It is not established if or how intrinsic CtBP activity is involved in mechanisms of trans-repression by CtBPs. CtBP1 and -2 bind NADH and/or NAD+ (71, 72, 73, 74). Upon binding NADH, CtBPs self-associate (73, 75) and/or associate with E1A (72, 73). In the absence of NADH, CtBP1 has intrinsic slow dehydrogenase activity (72, 73) and can promote or inhibit histone targeting of the HAT p300 (75). There have also been reports of NADH-dependent inhibition of CBP HAT activity (76, 77). It is thought that binding of CtBPs to the bromodomains of these and other HATs, resulting in loss of chromatin targeting of HAT activity in an HDAC-independent manner, is a second mechanism of CtBP trans-repression (75).

In the monomeric form, CtBP1 binds a signature PxD(L/I)(S/K) motif within the p300 bromodomain (38, 75). GCN5 and two of the three p160s have potential CtBP binding motifs proximal to a conserved nuclear receptor box (Fig. 6AGo). Other HATs and the CBP-related transcription factor CREB show one or more homologous motifs, whereas in p54nrb and PSF, these motifs are near a conserved aromatic residue (underlined in Fig. 6AGo) that is required for small nuclear riboprotein binding and targeting of splicing function in the closely related factor TAT-SF1 (79). Although CtBP1 lacks a nuclear localization sequence, CtBP1-mediated repression is modulated by factors that facilitate nuclear localization of the corepressor (80), including Pinin, a splicing factor with serine-/arginine-rich tract(s) of the SR family (81), with membership and other characteristics shared by PSF and p54nrb.


Figure 6
View larger version (33K):
[in this window]
[in a new window]

 
Fig. 6. CtBP Dehydrogenases Disrupt Cycles of cAMP-Dependent SF-1-Mediated Transcription of the CYP17 Gene via Multiple Interactions with Coregulators

A, Alignment of binding motifs in coregulatory proteins. Interactions that have been experimentally confirmed are cited. Underlined residues: NR4, the fourth nuclear receptor box interaction motif in these p160 coregulators; BrD, conserved acetyllysine binding bromodomain residues; SplFn, a conserved aromatic residue that is important for the splicing activity of TAT-SF1, a homologous corepressor that bridges the Pol II C terminus to both snRNPs and elongation factors (79 ). B–E, Cells were transfected with pG5 and (B) pBind-CtBP1 and pAct-GRIP-1, (C) pBind-GCN5 and pAct-CtBP1, (D) pBind-GCN5 and pAct-CtBP1 or pAct-CtBP1 G183V, (E) pBind-SF-1 and pAct-CtBP1 or pAct-CtBP1 G183V and then treated with 1 mM Bt2cAMP.(Legend continues on next page.) Reporter gene activity was quantified from cell lysates using dual luciferase assays. Open bars are untreated controls and striped bars are treated with 1 mM Bt2cAMP. Graphed data (panels B–E) represent the mean ± SEM and are from two experiments performed in triplicate. Statistically significant difference between cells transfected with pAct empty vector vs. cells transfected with pAct-GRIP-1 (B) or pAct-CtBP1 (C and E) are denoted by {ddagger}, P < 0.05; or {ddagger}{ddagger}{ddagger}, P < 0.001. Ampersand denotes statistically significant difference between untreated and Bt2cAMP-treated cells: &&, P < 0.01; and &&&, P < 0.001. Carets denote statistically significant difference between cells expressing wild-type CtBP1 vs. cells expressing the CtBP1 G183V mutant: ^, P < 0.01; ^^ , P < 0.001. F, H295R cells were synchronized for 2 h with 2.5 µM {alpha}-amanitin and then treated with 1 mM Bt2cAMP for the indicated times and subjected to ChIP using antibodies against SF-1, CtBP1, or CtBP2. Outputs are normalized to {Delta}Ct values obtained for 1% input controls, and results are presented as percent of {Delta}Ct values for untreated cells at the corresponding time point. Graphed data represent the mean from two experiments performed in duplicate.

 
To determine whether CtBPs interact with coregulators involved in cAMP responsiveness and CYP17 transcription, we performed mammalian two-hybrid experiments. CtBP1 interacts with GRIP-1 (Fig. 6BGo) and GCN5 (Fig. 6CGo). Similar positive interactions were detected in both Jeg3 and H295R cells (data not shown). We next determined the effect of mutating the NADH binding site, as done by Kim et al. (75), on the ability of CtBP1 to interact with GCN5. These experiments showed a greater degree of interaction of GCN5 with the CtBP1 dimerization-deficient, NADH-binding site mutant when compared with wild-type CtBP1 (Fig. 6DGo). This finding suggests a role for GCN5 binding to CtBP1 in the disruption of CtBP oligomerization.

Because CtBP1 interacts with a number of factors involved in regulation of CYP17 transcription, we next asked whether CtBPs interact with SF-1. We found that SF-1 interacts with CtBP1, and the interaction signal was weaker in the G183V NADH-binding mutant (Fig. 6EGo). Bt2cAMP weakened the interaction of SF-1 with G183V CtBP but not wild-type CtBP1. These data suggest that both the NADH-sensitive oligomerization of CtBP1 and Bt2cAMP stimulation can affect recruitment of the repressor to SF-1-containing complexes.

To clarify the role of CtBPs in CYP17 transcriptional activation in the context of other coregulator interactions on this promoter, we repeated temporal ChIP on the CYP17 promoter for CtBP1 and -2. Acute CtBP2 binding between 30 and 60 min (Fig. 6FGo) coincides with loss of GCN5 and SRC-1 (cf. Fig. 2Go, D and E), whereas loss by 90 min is concurrent with loss of SF-1 promoter occupancy. CtBP1 also binds to the promoter rapidly between the 180- and 210-min time points (Fig. 6FGo), corresponding with dissociation of the GCN5/p54nrb/SF-1/p160 complex from the promoter (cf. Fig. 2Go, D and E and Figs. 4CGo and 5Go).

CtBP1 Alters cAMP-Dependent Activation of CYP17 and Basal Interactions among SF-1, GCN5, and SRC-1
To assess the ability of CtBPs to regulate CYP17 transcription, we performed cotransfections of CtBP1 and other functional coregulators with SF-1 or LRH-1 and a luciferase reporter with a promoter containing two copies of the CYP17 SF-1 recognition motif (–57/–37 element) (Fig. 7AGo). Wild-type CtBP1 expression in H295R cells lowered basal (compare sets 1 and 2) and SF-1-dependent CYP17 expression (sets 3 and 4), but not LRH-1-dependent expression (sets 6 and 7) with or without Bt2cAMP treatment, and this repression requires the SF-1 AF-2 hexamer (sets 4 and 5). CtBP1 therefore has coregulator function in CYP17 transcription independent of CtBP2, corresponding with increases in promoter occupancy by CtBP1 after 90 min of Bt2cAMP stimulation (Fig. 6FGo). In light of the above results that GCN5 and GRIP-1 interact with CtBP1, and the findings of others demonstrating that GCN5 coimmunoprecipitates with CtBP1 (77), we asked whether SRC-1 coactivator function is also altered by CtBPs. Therefore, we tested whether wild-type CtBP1 disrupts the GCN5/SRC-1/SF-1 interaction in the two-hybrid system in H295R cells, but found that CtBP1 overexpression had a stabilizing effect on the trimer (Fig. 7BGo). On the other hand, CtBP1 decreased the stability of the acetyltransferase-deficient GCN5 E214Q/SRC-1/SF-1 complex in the absence of Bt2cAMP (Fig. 7BGo).


Figure 7
View larger version (33K):
[in this window]
[in a new window]

 
Fig. 7. cAMP- and NADH-Dependent CtBP1 Modulation of GCN5/SRC-1 Cooperativity

A, Cells were transfected with 250 ng pGL3-CYP17 2x57, pcDNA3.1-SF-1, pcDNA3.1-SF1DAF2 mutant, pCI-LRH1, and/or pRC-CMV-CtBP1, stimulated with Bt2cAMP, and then harvested for dual luciferase assays. B, Mammalian two-hybrid assays were performed as described in Materials and Methods by transfecting cells with pBind-SF-1, pAct-GCN5, pAct-GCN5 E214Q, pBKCMV-SRC-1, and/or pRC-CMV-CtBP1, and then incubated with 1 mM Bt2cAMP for 16 h. Interaction was quantified using dual luciferase assays. In both panels A and B, open bars represent untreated control, and striped bars represent Bt2cAMP treatment. Graphed data represent the mean ± SEM and are from two experiments performed in triplicate. In panel A, daggers denote statistically significant effect of overexpressed vectors: {ddagger}, P < 0.01; {ddagger}{ddagger}{ddagger}, P < 0.001. Statistically significant effects of CtBP1 overexpression are denoted by $$; P < 0.01. (&&&, P < 0.001) indicates statistically significant effect of SF-1 AF-2 deletion. Asterisks denote statistically significant differences between cells transfected with only pBind-SF-1 and pAct-GCN5 compared with groups also overexpressing SRC-1 and CtBP1; *, P < 0.05; or ***, P < 0.001.

 
Chromatin Remodeling Occurs before and after Pol II Interaction with the CYP17 Promoter
Studies characterizing cyclical ER{alpha} binding to the pS2 promoter have linked ATP-dependent chromatin remodeling to 1) time points immediately after receptor binding and 2) temporary repressive chromatin structure between cycles of ligand-bound ER{alpha} interaction with the promoter (40). To determine whether ATP-dependent chromatin remodelers act as gatekeepers of promoter accessibility for SF-1-mediated Pol II recruitment, we examined a panel of three chromatin remodeling complex ATPase subunits in assays of temporal CYP17 promoter binding. We found that the Brahma-related gene 1 (BRG1) ATPase, a member of some SWI/SNF remodeling complexes, is associated with the promoter 30 min before peak SF-1 binding in cycle I, and the sucrose nonfermenting 2 (SNF2) component of the imitation SWI or imitation SWI (ISWI) complex binds early in cycle II (Fig. 8AGo). As SF-1 binding peaks during this cycle, Brahma1 (Brm) ATPase exchanges for the BRG1 ATPase. This Brm binding coincides with increased H4 acetylation (Fig. 2CGo). These data suggest that proximal CYP17 promoter chromatin structure is made permissive to Pol II binding and possibly transcription early in transcription cycle I by a SWI/SNF remodeling complex containing BRG1. Additional remodeling by ISWI occurs early in cycle II, reestablishing permissive conditions for PSF and p54 binding before chromatin is further remodeled by SWI/SNF complexes, concomitant with SF-1 and Pol II binding.


Figure 8
View larger version (25K):
[in this window]
[in a new window]

 
Fig. 8. Chromatin Remodeling ATPase Recruitment Is Part of SF-1-Dependent CYP17 Transcription Cycles and Coincides with Rapid Loss and Gain of Histone H2 from Promoter Nucleosomes

A, Temporal ChIP of chromatin remodeling ATPases on the CYP17 promoter and transcription start site was performed as described in Materials and Methods. Lysates were immunoprecipitated with antibodies against SF-1, BRG1, Brm1, and SNF2 and purified DNA subjected to real-time PCR using primers that amplified the –104/+43 region of the CYP17 promoter. B, Temporal ChIP analysis of Bt2cAMP-stimulated SF-1 and histone H2B binding to the CYP17 promoter in {alpha}-amanitin-synchronized cells. Sheared chromatin isolated from cells treated for 30 min to 4 h was immunoprecipitated with antibodies against SF-1 or histone H2B. Output data are normalized to {Delta}Ct values obtained for 1% input controls, and results are presented as percent of baseline value obtained for untreated cells at each time point.

 
Each nucleosome is composed of a core tetramer and two histone H2A/H2B dimers, which may have a role in stabilizing DNA-nucleosome interaction in inactive chromatin. ATP-dependent chromatin remodeling complexes have been shown to exchange histone H2 dimers between nucleosomes on different short chromatin templates in vitro, and it is thought that removal of these dimers disfavors higher order chromatin packing of heterochromatin (82). We asked whether changes in the composition of histone H2 dimers in nucleosomes coincide with remodeling by DNA-interacting ATPases in vivo. When temporal ChIP for histone H2B occupancy of the same promoter region was repeated, we discovered that BRG1 SWI/SNF ATPase recruitment correlates with strong depletion of H2B within 30 min of Bt2cAMP stimulation, which is evidence that H2A/H2B dimers are disrupted during cAMP-dependent CYP17 activation. On the other hand, SNF2H recruitment corresponds with strong enrichment of H2B at the beginning of cycle II, followed within 30 min by Brm SWI/SNF ATPase binding and a subsequent return to histone H2 depletion. Thus, we postulate that SWI/SNF ATPases transfer H2B in H2A/H2B dimers to chaperones or are otherwise exchanged from the nucleosome(s) at the proximal promoter of CYP17 during cAMP-dependent transcription.

Activating Histone Lysine Methyltransferases are Targeted to the CYP17 Promoter in Response to Bt2cAMP
Another class of histone modifications that regulate transcription is the methylation of histone H3 and histone H4 lysine or arginine residues (for a review, see Ref. 83). Histone monomethylation at K20 on the N-terminal tail of histone H4 correlates with hyperacetylation of this tail in transcriptionally competent chromatin; trimethylation at this site represses transcription (84). On the other hand, H3 K4 trimethylation in yeast is required for ATP-dependent chromatin remodeling by Isw1p ATPase and correlates with recruitment of a factor involved in mRNA maturation (85).

Using temporal ChIP, we assayed for these Bt2cAMP-stimulated changes in histone lysine methylation at the CYP17 promoter. Histone H4 K20 monomethylation is up-regulated as SF-1 occupancy peaks at 60 and 180 min, as well as at 120 min (Fig. 9Go); all three peaks correlate with histone H4 hyperacetylation (cf. Fig. 2CGo). Repression-associated trimethylation at this site is not up-regulated during the first 4 h of Bt2cAMP stimulation. Trimethylation at histone H3 K4 occurs transiently during cycle I and also appears to be up-regulated between 210 and 240 min of treatment (Fig. 9Go); the initial event does, in fact, precede recruitment of splicing factors by at least 60 min (cf. Fig. 4CGo). Of the panel of ATPases examined in temporal ChIP assays (Fig. 8AGo), only SNF2H, a mammalian homolog of yeast Isw1p, interacts with the promoter during the interval between histone H3 K4 trimethylation at 60 min and up-regulation of p54nrb and PSF recruitment between 120 and 180 min (cf. Figs. 4CGo, 8AGo, and 9Go). Thus, our data are consistent with roles for at least two histone lysine methylation events already established to correlate with (1) promoter histone H4 acetylation and transcriptional competence, i.e. 1) monomethylated histone H4 K20, and 2) histone H3 K4 trimethylation, associated with subsequent ISWI chromatin remodeling, and then recruitment of factors with roles in RNA maturation, a pattern seen in both yeast (85) and in human adrenocortical cells.


Figure 9
View larger version (35K):
[in this window]
[in a new window]

 
Fig. 9. Histone Lysine Methyltransferases Are Recruited to the CYP17 Promoter during SF-1-Dependent Cycles of Transcription

{alpha}-Amanitin-synchronized H295R cells were treated for time periods ranging from 30 min to 4 h with 1 mM Bt2cAMP and exposed to 1% formaldehyde, and the purified lysates were immunoprecipitated using anti-SF-1, anti-trimethylated (K4) histone H3, antimonomethylated (K20) histone H4, or antitrimethylated (K20) histone. Output data are normalized to values obtained for 1% input controls, and results are presented as percent of baseline value obtained for untreated cells at each time point. Graphed data represent the mean from two experiments performed in duplicate.

 
Sph Modulates SF-1-Mediated Transcription Cycles
The SF-1 antagonist Sph decreases Bt2cAMP-dependent transcription of CYP17 in H295R cells (Fig. 10AGo), in line with Sph-mediated disruption of SF-1/SRC-1/GCN5 cooperativity (Fig. 3CGo). To examine the effect of Sph on SF-1-mediated transcriptional cycling, we repeated synchronized temporal ChIP experiments with 5 µM Sph treatment. In the absence of Bt2cAMP, SF-1 in Sph-treated cells associated with the CYP17 promoter within 30 min of treatment and again increased after 3 h with a longer delay between increases in binding compared with Bt2cAMP-stimulated cycling, and with muted amplitude or binding rate (Fig. 10BGo). Sph strongly attenuated the ability of Bt2cAMP to initiate the binding of SF-1 to the CYP17 promoter in cycle I (Fig. 10BGo). SF-1 recruitment during cycle II was relatively unaffected by the antagonist, indicating significant cAMP-dependent release of Sph from SF-1 within 3 h, consistent with previous results (47).


Figure 10
View larger version (20K):
[in this window]
[in a new window]

 
Fig. 10. cAMP-Stimulated CYP17 mRNA Expression and Cycles of SF-1-Mediated Promoter Binding to the CYP17 Promoter Are Antagonized by Sph

A, H295R cells treated with 1 mM Bt2cAMP and/or 1 µM Sph for time periods ranging from 1–12 h. Total RNA was isolated and subjected to quantitative RT-PCR. Graphed data are expressed as fold change compared with untreated controls and represent the mean ± SEM of CYP17 mRNA expression normalized to the cellular ß-actin mRNA content from four experiments performed in triplicate. B, {alpha}-Amanitin-synchronized H295R cells were treated for time periods ranging from 30 min to 4 h with 1 mM Bt2cAMP and/or 1 µM Sph and exposed to 1% formaldehyde, and the purified lysates were immunoprecipitated using a polyclonal antibody against SF-1. Graphed data in all panels represent the mean from three experiments, each performed in duplicate. Outputs are normalized to {Delta}Ct values obtained for 1% input controls, and results are presented as percent of untreated control {Delta}Ct values at each time point.

 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 Note Added in Proof
 REFERENCES
 
cAMP Induces Chromatin Modifications and Distinct Coregulatory Binding Events during Cycles of CYP17 Transcription
Synchronization with {alpha}-amanitin returns histones on promoters to a modification state that corresponds with promoter chromatin of untranscribed genes (40) in concert with dissociation of RNA polymerase and nuclear receptor (31). Thus, temporal analysis of coregulatory protein binding and histone modification with Bt2cAMP treatment on the promoter of a Bt2cAMP-responsive gene after release of the {alpha}-amanitin blockade allows comprehensive analysis of the sequential binding events that are required for optimal CYP17 transcription in response to cAMP. We have previously shown that a complex containing SF-1, p54nrb, and PSF is required for cAMP-dependent CYP17 transcription (82); however, the precise kinetic details of the assembly of this complex were unknown. Additionally, the kinetic parameters and protein-protein interaction network that define the recruitment of coregulatory proteins, chromatin remodelers, or modifiers, and RNA Pol II to the CYP17 promoter were not defined or incomplete.

An examination of complex assembly revealed cyclic SF-1 binding to the CYP17 promoter in H295R cells in response to Bt2cAMP treatment and highlights the prevalence of coregulatory proteins that act cooperatively and sequentially on chromatin during these cycles. Cycle I is delineated by early binding of GCN5/SRC-1/SF-1 complex, Brg1, and HDAC8 and late CBP and CtBP2 binding events. Cycle II begins with cooperative recruitment of GCN5, PSF, and p54nrb, and then proceeds with recruitment of p300 and SNF2H. At the end of this cycle, SF-1 recruits SWI/SNF remodeling complex containing BRG1, which exchanges for Brm as SF-1 dissociates from the CYP17 promoter. Transient coactivator arginine methyltransferase recruitment to the promoter was also detected at 60 min and, to a lesser extent, at 180 min (our unpublished findings). RNA polymerase is recruited in both cycles, and occupancy decreases after SF-1 dissociates from the promoter. Occupancy of the promoter by corepressors and class I HDACs, including HDACs 1 and 2, is enriched between the two observed transcription cycles when SF-1 is not bound to the promoter.

Our ChIP time courses show that the SNF2H ATPase component of ISWI chromatin remodeling complex and p300 are both on the promoter at 150 min of Bt2cAMP stimulation, which is specifically before the second cycle of SF-1-mediated transcription, and overlaps with a high rate of increase in PSF and p54nrb binding (cf. Figs. 4CGo and 8AGo). Ito et al. (86) have shown that ISWI chromatin remodeling complexes and p300 are sequentially targeted to regions of chromatin bound by sequence-specific DNA-binding domains linked to acidic activation domains. The binding of p54nrb and PSF to the CYP17 promoter at this time agrees with their model. In vitro, chromatin remodeling followed by acetylation enables transfer of a histone H2A/H2B dimer to the nucleosomal chaperone Nap-1, which in turn may enable nucleosome sliding or subsequent stepwise disassembly of nucleosomes on the promoter (cf. Refs. 87 and 88). Additionally, purified chromatin remodeling complexes have been shown to exchange H2A/H2B dimers between chromatin templates in vitro (82). Our data show a period of core histone acetylation and recruitment of remodeling ATPases between 150–180 min of Bt2cAMP stimulation, which are conducive to a rapid increase, and then a decrease in histone H2B (Fig. 8BGo), which likely reflects a gain, followed by a loss, of histone H2 dimers from promoter nucleosomes. Histone H2 dimer loss may further aid nucleosome disassembly or movement and disturb higher order chromatin structure intrinsic to heterochromatin. Continuing increases in H3 and H4 acetylation during this period (Fig. 2CGo) are consistent with core nucleosome tetramer movement, but not prolonged dissociation from the promoter.

We predict that when nucleosome positions are established, BRG1 (at 30 min Bt2cAMP stimulation in this study; see model, Fig. 11AGo, step 1) and Brm (at 180 min, step 6) will mediate nucleosome repositioning. Accordingly, it is likely that this movement facilitates both the stable interaction of SF-1 with the –57/–37 promoter element and SNF2H-dependent changes in the chromatin environment that favor binding of PSF/p54nrb. Additional remodeling events, such as BRG1 recruitment at 210 min, may specifically facilitate access to the TATA box by the basal transcription machinery and strengthen Pol II association (step 7). All SWI/SNF ATPase binding events either precede or coincide with an increase in RNA polymerase on the CYP17 promoter (cf. Figs. 2BGo and 8AGo).

On the pS2 promoter, H3 and H4 acetylation kinetics are not very divergent, and this modification remains prevalent between some cycles of ER{alpha} binding (40). On the CYP17 promoter, H4 hyperacetylation is transient, occurring in phase with recruitment of SF-1 at 60 and 180 min (Fig. 2CGo), or a DNA binding factor that recruits CBP at 120 min of Bt2cAMP stimulation. H3 acetylation between 210 and 240 min is not affected during transient Brg1 binding. In contrast with a strict release of H3 and H4 acetylation just before SWI/SNF binding on the pS2 promoter (40), H3 hyperacetylation remains prevalent at the end of the second transcription cycle.

In temporal ChIP analysis of the proximal promoter of the SF-1-responsive Mc2r gene in Y1 cells recently performed by Winnay and Hammer (45), ACTH stimulation of {alpha}-amanitin-synchronized cells revealed SF-1 promoter occupancy in phase with peaks of histone H4 acetylation as they occur on the CYP17 promoter. Moreover, re-ChIP in their study identified GCN5 in a coactivator complex with SF-1 on the promoter during the first cycle of SF-1 binding, and SRC-1 promoter occupancy was increased preferentially in this cycle. However, on Mc2r promoter, peaks in GCN5 and SRC-1, GRIP-1, and ACTR occur 20 min after peaks in SF-1, and phosphorylated Pol II is at maximal promoter occupancy 40 min after SF-1 peaks, out of phase with SF-1 enrichment of the promoter, in contrast to the binding events on the CYP17 promoter. Although some shift in peak placement may correspond with antibody recognition of specific epitopes of p160 coactivators that become accessible at distinct times during the transcription cycle, Pol II binding on the Mc2r promoter is not closely associated with nuclear receptor promoter occupancy, as it is on the CYP17 (present study), pS2 (31, 40), or cathepsin D (44) promoters.

Although there is only weak conservation of a methyl-lysine binding chromodomain in ISWI ATPases such as SNF2H, Santos-Rosa et al. (85) found striking selectivity of this factor in the binding of histone H3 di- or trimethylated at K4 but not the unmethylated histone. However, our temporal ChIP data show that histone H3 K4 trimethylation is not maintained at the putative SNF2H interaction site until the ATPase binds. There remains the distinct possibility that trimethylation reverts to dimethylation at 60–90 min, which is then maintained, linking histone H3 K4 trimethylation in cycle I to SNF2H binding and subsequent events in cycle II. Based on a conserved sequence of events starting with methylation, followed by ISWI, p300, and finally p54nrb/PSF recruitment (a proxy for RNA maturation machinery and factors with extensive acidic tracts) (85, 86), we cannot discount the possibility that the transient trimethylation of H3 K4 at 60 min transmits a signal through a temporal, sequential cascade of events on the CYP17 promoter that proceeds via delayed recruitment of the ISWI ATP-dependent chromatin remodeling complex that includes SNF2H. With more confidence, we can say that SNF2H in cooperation with p300 affects recruitment of p54nrb/PSF splicing factors and coordinates the timing of cycle II initiation (Fig. 11AGo, step 5). Future studies are warranted to clarify the role of histone H2 dimer exchange among discrete nucleosomes or transfer onto chaperones, as well as precise locations of nucleosome occupancy before and after ATP-dependent remodeling.

Early Coactivator Cooperativity with SF-1 during Each CYP17 Transcription Cycle Is cAMP Inducible and Sensitive to Antagonist
We verified that early cyclic coactivator interactions with the CYP17 promoter observed in both cycles of SF-1-dependent transcription are stimulated by cAMP in mammalian two-hybrid experiments. Stabilization of the ternary GCN5/SRC-1/SF-1 complex (Fig. 3AGo) and of GCN5 interaction with p54nrb (Fig. 5AGo) is cAMP dependent. We found a loss of Bt2cAMP dependence of GCN5 interaction with LRH-1 but not SF-1 when competing HATs were overexpressed (our unpublished observations), implying that the cAMP-dependent induction of a SF-1/SRC-1/GCN5 trimer is intrinsic to this complex. GCN5 association with SF-1 early in each transcription cycle corroborates a report by Fan et al. (52), who found a rapid, PKA-dependent association of GCN5 with SF-1 in KGN cells using confocal fluorescent microscopy. It is possible that GCN5/SRC-1 complexes with other nuclear receptors are cAMP inducible.

Sensitivity of the SF-1/SRC-1/GCN5 interaction to Sph corroborates our earlier report that Sph can antagonize SRC-1 coactivation of SF-1 mediated transcription of the CYP17 gene (47). It remains to be determined whether this complex requires agonist binding. We hypothesize that Bt2cAMP initiation of a transcription cycle in the presence of Sph (Fig. 10BGo) corresponds with loss or exchange of Sph in the SF-1 LBD within 3 h of treatment. Because Sph exchanges readily with other lipids (47), studies are ongoing to define the mechanism by which cAMP alters levels of nuclear lipids and the role of intranuclear SF-1 localization in ligand binding.

The cAMP-inducible interaction between GCN5 and p54nrb (Fig. 5Go) is a novel interaction between a splicing factor, which doubles as a SF-1 coactivator, and a histone acetyltransferase. Although p54nrb already has a sizable acidic tract, PKA may phosphorylate p54nrb, increasing electrostatic interaction with positively charged regions of GCN5. PSF antagonism of this interaction suggests that a binding site on p54nrb is shared by the two factors, or that p54nrb and PSF together have a higher affinity for another factor, possibly p300, which has a larger interaction surface. PSF, p54nrb, and p300 are sharply enriched at 150 min of Bt2cAMP stimulation (cf. Figs. 2DGo and 4CGo); their interaction should be investigated further on promoters where the above factors share roles and synergy of p54nrb and GCN5 coactivator functions is expected.

GCN5 Is the Initial Acetyltransferase in SF-1 Activation of the CYP17 Gene
CBP and P/CAF are only recruited late in cycles I and II after GCN5 or GCN5 and p300, respectively. We have shown that stabilization of GCN5 interaction with SF-1 occurs when GCN5 acetyltransferase activity is disabled (Fig. 3DGo) and CBP or P/CAF compete with GCN5 for interaction with SRC-1-associated LRH-1 (our unpublished observations). It is an intriguing possibility that acetylation-mediated coactivator exchange is occurring on the CYP17 promoter and that this is a paradigm for sequential recruitment of HAT-containing complexes to nuclear receptors during cycles of transcription. Experiments are underway to determine whether SF-1 activation of CYP17 transcription via the recruitment of coactivators depends first on specific factor and/or histone lysine acetylation by GCN5, and second on the specificity of bromodomains in coactivators recruited later in each transcription cycle.

cAMP-dependent acetylation of the placental transcription factor GCMa increases the stability of this transcription factor and promotes GCMa-dependent transcription (89). Thus, cAMP-induced acetylation of a transcription factor can result in targeted modulation of transcription. Moreover, temporally distinct reversal of acetylation may provide another layer of regulation and potentially define a window of time during which induction of transcription can occur in response to an acute signal.

Acetylation can destabilize interactions among protein complexes in a reversible manner, leading to profound changes in function. It has been shown that heat shock protein 90 association with the essential cochaperone p23 is interrupted by acetylation and that this effect is reversed by HDAC6-dependent deacetylation (90). CBP/p300 acetylation of the HIV trans-activating protein (Tat) dissociates it from TAR, the transactivation response RNA element, via competition with P/CAF, which is recruited directly to Tat when the P/CAF bromodomain binds specifically to acetylated Tat lysine 50 (91).

Specific histone acetylation events may precede recruitment of secondary HAT(s) via site-specific bromodomains and transcription factor acetylation (92), although bromodomains are not necessarily specific for the context of residues that surround an acetylated lysine moiety (93). Thus, maintenance of specific histone acetylation marks allows HATs to be recruited at times distinct from when there is transcription factor occupancy of the promoter, as we have seen for CBP and p300 after SF-1 dismissal (Fig. 2DGo), and others have seen for CBP (44) or CBP and p300 (40) after ER{alpha} dismissal from the cathepsin D or pS2 promoters, respectively.

H4 K8 acetylation, carried out by GCN5 or P/CAF, is thought to promote association of nucleosomes with BRG1 via its bromodomain (94). Recruitment of BRG1 follows or matches GCN5 recruitment and H4 acetylation in both cycles of CYP17 transcription (cf. Figs. 2Go, B and C, and 8AGo). For the above reasons, we believe that GCN5 is the primary HAT recruited to the CYP17 promoter by cAMP-dependent interaction with SF-1 and p54nrb in adrenocortical cells, and the subsequent binding of other HATs and chromatin remodeling complexes is stimulated by existing acetylation initiated by GCN5. The increase in GCN5/SRC-1/SF-1 stability with mutation of a residue involved in GCN5 acetyltransferase activity (Fig. 3DGo) suggests that competition by subsequent coactivators and concomitant disassembly of the ternary complex during progress through the transcription cycle depends, at least in part, on acetylation by GCN5.

The role played by GCN5 on the CYP17 promoter does not appear to be interchangeable with other HATs because kinetics for each HAT is unique (Fig. 2DGo). Early GCN5 interaction with SF-1 is mediated by p160 coactivators in both transcription cycles. Therefore, p160 coactivators share a role in initiating ligand or signal-dependent acetylation of histones, and possibly other trans factors through their scaffolding function, although a role for their intrinsic acetyltransferase activity cannot be ruled out.

Acetylation Promotes Sequential Coactivator Recruitment
The bromodomain on acetyltransferases allows recognition of previous acetylation events by earlier acetyltransferases, creating the possibility of an acetylation cascade. We propose a three-step model describing the procession of HAT binding on the CYP17 promoter bound by SF-1: 1) cAMP-dependent association of SRC-1 and GCN5, 2) acetylation of specific residues on histone H4, and 3) exchange of GCN5 with CBP (cycle I) or P/CAF and CBP (cycle II). BRG1 is also recruited after histone H4 lysine 8 acetylation (94), enabling continuation of the promoter-bound cascade, which results in transcription. Thus, specific acetylation states of histones, but also of SF-1 and coregulators, are bookmarks for pr