| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Pharmacology (A.K.E., M.A., D.C.), University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104; and Division of Reproductive Biology Research (D.C.), Department of Obstetrics and Gynecology, Robert H. Lurie Comprehensive Cancer Center, Feinberg School of Medicine, Northwestern University, Chicago, Illinois 60611
Address all correspondence and requests for reprints to: Debabrata Chakravarti, Division of Reproductive Biology Research, Department of Obstetrics and Gynecology, Robert H. Lurie Comprehensive Cancer Center, Feinberg School of Medicine, Northwestern University, Chicago, Illinois 60611. E-mail: debu{at}northwestern.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
B ligand gene promoters, respectively. Third, RNA interference-mediated reduction of hAcf1 or vitamin D3 treatment differentially affects the histone modification profile and the histone occupancy in these genes. Together, these results establish that hAcf1 has a critical role in the transcriptional repression of specific NR-regulated genes and indicate that hAcf1 release and histone H3 and H4 eviction are novel mechanisms in hormone-induced gene activation. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Chromatin structure can be changed by the posttranslational modification of histone tails, ATP-dependent chromatin remodeling, histone loss, and the incorporation of histone variants into chromatin. Actively transcribed regions of chromatin are generally rich in acetylated histone H3 and H4 and also show methylation of H3 at K4. Heterochromatin is, however, marked by hypoacetylation, trimethylation of lysine 9 of histone H3 (trimeH3K9), and H3K27 monomethylation (3). ATP-dependent chromatin remodeling enzymes can alter chromatin structure by using the energy of ATP-hydrolysis to disrupt DNA-histone contacts and to change the position of a nucleosome within chromatin in a process known as remodeling (4). The disruption of DNA-histone contacts causes the destabilization or the removal of subunits in the histone octamer and can thus allow for transcriptional activation by RNA polymerase II (Pol II) (5). A physical separation of DNA histone contacts also enables the deposition of histone variants, such as H3.3, which leads to transcriptional activation (6).
Class II nuclear receptors (NRs) of the steroid/thyroid/retinoid superfamily [such as the vitamin D receptor (VDR), the thyroid receptor (TR), and the retinoic acid receptor (RAR)] are ligand-induced transcription factors that are critical for development, tissue growth and differentiation, and metabolism (7). The primary role of NRs is to coordinate the ligand-induced exchange of large coregulatory complexes that have specific enzymatic activity. This leads to alterations in chromatin structure and consequently changes in transcriptional output of their target genes (8, 9).
Transcriptional activation of NR-target genes occurs by sequential recruitment of both histone modification activity and ATP-dependent chromatin remodelers by ligand-bound receptors. There are several coactivator complexes containing intrinsic histone acetyltransferase activity, including members of the SRC family, CBP (cAMP response element-binding protein)/p300, ACTR (activator of thyroid and retinoic acid receptor), and the human mediator complex TRAP (thyroid hormone receptor-associated protein)/DRIP (VDR-interacting protein) (8, 10, 11). In addition to histone acetyltransferase activity (believed to be important for localized disruptions in DNA-histone binding), ATP-dependent enzymes activate the transcription of NR-regulated genes by nucleosome remodeling. This process causes large changes to chromatin structure and so allows the basal transcriptional machinery to access DNA (12).
Transcriptional repression is mediated by unliganded NRs that preferentially associate with the corepressors nuclear receptor corepressor (N-CoR) and silencing mediator for retinoid and thyroid receptor (SMRT) (13, 14). N-CoR/SMRT acts as a platform to coordinate the recruitment of histone deacetylases (HDACs), the histone binding protein TBL1 (transducin ß-like protein-1), and factors of the basal transcriptional machinery to actively repress transcription (15). Ligand binding alters NR structure, leading to the release of the corepression complex and the subsequent recruitment of the coactivation complex to hormone-responsive genes.
Histone modification activity has been extensively documented in N-CoR-mediated transcriptional repression. The role of ATP-dependent chromatin remodeling is, however, not well understood. One biochemical purification of N-CoR identified components of the SWI (switch)/SNF (sucrose nonfermenting) family of ATPases [such as BRG1 (Brm-related gene 1)], but their role in N-CoR-mediated repression was not determined (16). The ATPase- and HDAC-containing NuRD (nucleosome remodeling and HDAC) complex contributes to repression at a TR-regulated gene (17, 18). The ATPase SNF2H (sucrose nonfermenting homolog 2), is also required for the repression of a TR-regulated gene (19). It remains unclear, however, exactly how these enzymes are targeted to specific genes, and whether or not other NRs (in addition to TR) require ATP-dependent chromatin remodeling for transcriptional repression.
ATP-utilizing chromatin assembly and remodeling factor 1 (Acf1) is a member of two SNF2H-containing complexes found in both human (h) and Drosophila (d): ACF and CHRAC (chromatin accessibility complex) (20, 21). Acf1 contains WAC (WSTF/Acf1/cbp146), DDT (DNA binding homeobox and different transcription factors), WAKZ (WSTF/Acf1/KIAA0314/ZK783.4), PHD (plant homeodomain) finger, and bromodomain motifs (22, 23). The SNF2H ATPase subunit is a member of the ISWI (imitation switch) family, which is characterized by a SANT (SWI3, ADA2, N-CoR, and TFIIB) domain in the C terminus (24). In these complexes, SNF2H is the "motor" and the accessory proteins are needed both to produce maximal activity as well as to control the direction of nucleosome repositioning (25). A growing body of evidence suggests that remodeling factors are recruited to specific nucleosomes by site-specific DNA-binding proteins (26). It is, therefore, proposed that the nonenzymatic subunits of the SNF2H complex, such as Acf1, may act as guides to direct the enzyme to its target and to contribute a unique functionality to the complex.
Most of our understanding of the chromatin assembly and remodeling function of Acf1 comes from studies performed in Drosophila or in vitro with purified proteins (24, 27). Knockout of Acf1 in Drosophila decreases the periodicity of nucleosomes, which leads to a loss of transcriptional silencing in pericentric heterochromatin and Polycomb-repressed chromatin (27). In mouse cells, the ACF complex is enriched in heterochromatin and is critical for its replication (28). Purified dACF can assemble chromatin containing histone H1, a histone subunit involved in the formation of higher-order chromatin structure (29). Therefore, ACF may be a key regulator in the assembly and maintenance of higher-order chromatin architecture; such functions likely contribute to transcriptional repression of heterochromatic genes.
The endogenous human ACF complex has not been studied extensively. Even less is known about any direct role of Acf1 in transcription in the mammalian system. This study reports a previously unrecognized role of Acf1 in the repression of euchromatic genes in mammalian cells and shows that hAcf1 affects the hormone responsiveness and the transcriptional repression of specific class II NR-regulated genes. This work demonstrates that hAcf1 is critical for the stable occupancy of the N-CoR corepression complex at the target gene promoters. Hormone treatment, which causes a release of hAcf1 from the promoter, relieves the repression of IGF binding protein 3 (IGFBP3), receptor activator of nuclear factor-B ligand (RANKL), and IGF-I genes attributable to a loss of corepressor recruitment to the gene and to alterations in the local chromatin architecture.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
). A yeast two-hybrid assay was performed with a galactosidase-4 (GAL4) activation-domain fusion human cDNA library, using the RD2 of N-CoR as bait. Among the positive colonies analyzed, cDNA encoding the N terminus of hAcf1 (hAcf1N) (amino acids 1–313) was isolated. Reintroduction of this clone into yeast transformed with either GAL4DBD-RD1 or -RD2 revealed that, although hAcf1N interacted with both RD2 and RD1 (Fig. 1A, b), it had a higher affinity for the RD1 domain: the colonies that were cotransformed with this domain were able to grow even under the selective pressure of 20 mM 3-aminotriazole (Fig. 1A, compare b with c). Together, these results suggest that RD1 and RD2 of N-CoR interact with hAcf1 with varying affinities. Because N-CoR RD1 bound to hAcf1 with higher affinity than did RD2, subsequent studies used RD1 to characterize the interaction.
|
, [35S]-labeled N-CoR (as well as its homolog SMRT) was retained by the GST-hAcf1N beads, but not by the GST-bound beads.
The yeast two-hybrid screen isolated hAcf1N as an N-CoR-interacting protein. This region encompasses the WAC motif, which is a DNA-binding and putative heterochromatin association domain (22). To ascertain whether hAcf1N is necessary for its interaction with N-CoR, [35S]-labeled hAcf1 deletion constructs were used in glutathione S-transferase (GST) pull-down assays with GST-N-CoR RD1. The constructs that lacked amino acids 1–313 did not interact with GST-RD1 to a great extent, whereas those that contained this region were pulled down specifically by the GST-RD1 beads (Fig. 1C). These results show that amino acids 1–313 are both necessary and sufficient for hAcf1 to directly interact with N-CoR, and this region was thus named NID, for N-CoR interaction domain.
The endogenous association of hAcf1 and N-CoR was next tested in eukaryotic cells. For that purpose, coimmunoprecipitations were performed in HEK293T cells using the antibody against hAcf1 and followed by an immunoblot for N-CoR. Anti-hAcf1 antibody specifically immunoprecipitated endogenous N-CoR in HEK293T cells (Fig. 1D). Together, these data show that hAcf1 interacts with N-CoR/SMRT both in vitro and in living cells.
hAcf1 Is Required for Transcriptional Repression of Several Class II NR-Regulated Genes
N-CoR is a corepressor for many NRs, including the unliganded class II NRs. Because hAcf1 interacts with N-CoR, it is likely that this protein has a role in NCoR-mediated repression. To test the role of hAcf1 in establishing transcriptional repression of endogenous, NR-regulated genes, it was overexpressed by transient transfection in Saos-2 cells in hormone-free medium. The mRNA levels of several native NR-regulated genes in these cells were then measured by quantitative real-time PCR (qPCR). Saos-2 osteoblastic cells were chosen for this study because these cells have a well-characterized vitamin D3 (vitD3) signaling pathway. This cell line also responds to thyroid hormone and retinoic acid, thereby allowing studies on TR and RAR target genes.
Transient expression of hAcf1 in transfected cells led to a small but statistically significant increase in repression of specific, endogenous, NR-regulated genes: two VDR-regulated genes, RANKL and IGFBP3, and the TR-regulated gene IGF-I. It did not, however, affect the repression of the VDR-regulated 24-hydroxylase gene or of the RAR-regulated IGFBP6 gene (Fig. 2A). Interestingly, transfection of N-CoR did not significantly enhance repression of these genes (Fig. 2A). This suggests that either hAcf1 is a more potent corepressor or that the endogenous level of N-CoR is sufficient for the maximum repression of these NR-regulated genes. Because chromatin is inherently repressive, the overexpression of components of repression pathways generally has a minimal effect on gene repression.
|
, the N terminus of hAcf1 (hAcf1-NID) was found to be necessary and sufficient for its interaction with N-CoR. This region was described previously to be required for the association of hAcf1 with DNA (22). Protein domains in the C terminus are critical for the association of hAcf1 with both SNF2H and histones (22, 25). To understand the contribution of hAcf1-NID to the repression of genes that are repressed by FL hAcf1, it was transiently expressed in transfected Saos-2 cells. As in Fig. 2A, qPCR was used to measure the mRNA levels of endogenous hormone-responsive genes. hAcf1-NID expression caused an increase in mRNA levels in all tested genes (Fig. 2A). This indicates that the hAcf1-NID is not sufficient for the repression by hAcf1 observed in Fig. 2A. Instead, this region may be acting as a dominant-negative effector by competing for binding of endogenous hAcf1 and thus partially relieving the transcriptional repression of hAcf1 target genes. Because expression of hAcf1-NID relieved repression of all tested genes, this region may be functioning in a gene nonspecific manner, by either sequestering N-CoR away from its target gene or interacting with N-CoR at its target gene and inhibiting its repressive properties (i.e. inhibiting the recruitment of other corepressors). In any case, these results suggest that hAcf1 is involved in repression and prompted additional analysis of the transcriptional regulatory role of endogenous hAcf1.
If hAcf1 is a critical transcriptional corepressor for RANKL, IGFBP3, and IGF-I genes, then knocking down the level of the endogenous protein should partially relieve repression of these genes, even in the absence of hormone. To test this idea, RNA interference (RNAi)-based technique was used to knock down the level of endogenous hAcf1 in Saos-2 cells grown in hormone-free medium. Forty-eight hours after transfection, protein levels were determined by Western blot (Fig. 2B). RNAi treatment specifically reduced hAcf1 but not its paralog Williams syndrome transcription factor (WSTF) or N-CoR. qPCR was then used to measure the mRNA levels of the hormone-regulated genes in these cells (Fig. 2C). Consistent with the results of Fig. 2A, hAcf1 knockdown relieved the transcriptional repression of IGFBP3, RANKL, and IGF-I but not the 24-hydroxylase or the IGFBP6 gene compared with control (Fig. 2C). Together, the results from the hAcf1 overexpression and knockdown studies demonstrate that hAcf1 is an important corepressor of specific hormone-regulated genes.
The mRNA levels of the RANKL, IGFBP3, and IGF-I genes increased 1.5- to 2-fold when hAcf1 was decreased by RNAi compared with their levels in control RNAi-treated cells (Fig. 2C). To determine whether the relief of repression is physiologically significant, the increase in mRNA attributable to hormone-activated transcription was measured (Fig. 2D). After 24 h of hormone treatment, the mRNA level of each gene was elevated about 2- to 3-fold. The increase in mRNA attributable to relief of active transcriptional repression of these genes achieved by RNAi depletion of hAcf1 is, therefore, significant because it is similar to the mRNA increase achieved by hormonal activation of transcription. hAcf1 knockdown thus closely mimicked the role of hormone in the induction of these genes. The 24-hydroxylase gene was not derepressed during hAcf1 knockdown, despite the fact that it is highly responsive to vitD3 treatment. In combination, these results suggest that hAcf1 contributes to the regulation of RANKL, IGFBP3, and IGF-I genes by actively contributing to their repression in the absence of hormone. This study also indicates that hAcf1 exerts gene-specific regulation: hence, it does not have a generalized effect on all class II NR-regulated genes.
Hormone Treatment or hAcf1 Knockdown Decreases hAcf1 and N-CoR Association at the Promoter of the IGFBP3 and RANKL Genes
RANKL and IGFBP3 are VDR-regulated genes that contain a vitamin D response element (VDRE) in their promoter (Fig. 3A). In the absence of vitD3, VDR-regulated genes are repressed via the recruitment of corepressors, such as N-CoR and various histone modification enzymes. Knocking down the level of hAcf1 relieved the repression of RANKL and IGFBP3 genes almost to the same extent as 24 h vitD3 treatment (Fig. 2). All subsequent studies were, therefore, performed on these two genes. To confirm that hAcf1 is indeed present at the IGFBP3 and RANKL promoters and that the RNAi treatment of hAcf1 leads to a decrease of hAcf1 association with these regions of DNA, chromatin immunoprecipitation (ChIP) assays were performed. This experiment used the hAcf1 antibody on Saos-2 cells transfected with RNAi against control or hAcf1 for 48 h in hormone-free medium. Control cells treated with 10 nM vitD3 for 1 h were included in the ChIP assay so that changes in promoter occupancy attributable to hAcf1 RNAi (relief of repression) as well as changes caused by hormone treatment (active transcription) could be examined. The 1-h time point was chosen to capture early changes caused by hormone binding to receptor rather than those caused by secondary effects. The quantity of immunoprecipitated DNA measured by real-time PCR was plotted either as a percentage of input (to look at occupancy levels along the promoter/enhancer) or as normalized to control RNAi-treated cells (to determine the fold change in occupancy under the treatment conditions).
|
, top). On the RANKL gene, highest hAcf1 occupancy was observed in and around the transcription start site (TSS) (primer sets C and D), whereas on the IGFBP3 promoter, it was concentrated at the VDRE (primer set A). The association of hAcf1 on IGFBP3 and RANKL genes was decreased compared with control RNAi-treated cells, when hAcf1 was knocked down (Fig. 3B, bottom). Importantly, vitD3 treatment also significantly lowered hAcf1 occupancy in these promoters. These data, when taken together, indicate that hAcf1 associates with specific regions in its target promoter and its binding is influenced by hormone treatment. It is also important to mention that hAcf1 knockdown did not significantly change promoter occupancy of SNF2H, which is a binding partner for hAcf1 (data not shown). This suggests that the recruitment (or association) of SNF2H with these genes is not dependent on the overall levels of hAcf1. It is possible, however, that minimal levels of hAcf1 are sufficient for the association of SNF2H with the gene because RNAi treatment does not completely abolish hAcf1.
N-CoR is necessary for the repression of vitD3-responsive genes and is released from promoters during hormone treatment. Because hAcf1 associates with N-CoR, it may affect N-CoR occupancy on these genes. If hAcf1 targets or stabilizes N-CoR at the promoter, then lowering its levels should lead to a decrease in the amount of N-CoR that binds to these genes, even in the absence of hormone. To test this hypothesis, ChIP assays using the N-CoR antibody were performed in Saos-2 cells under the same conditions. Interestingly, for the RANKL gene, a significant association of N-CoR was observed close to the TSS rather than at the putative VDRE (Fig. 3C, top left). This region also had a higher level of hAcf1 association. For IGFBP3, maximum N-CoR occupancy was observed near the VDRE, as expected (Fig. 3C, top right); this was also the region of highest hAcf1 occupancy. hAcf1 knockdown decreased N-CoR association with the RANKL (Fig. 3C, bottom left) and the IGFBP3 (Fig. 3C, bottom right) genes when compared with control RNAi-treated cells. The observed decrease in N-CoR occupancy is not a result of a decrease in the level of N-CoR protein (Fig. 2A). The finding that these two proteins are enriched at the same regions of each gene further supports the in vitro and in vivo interaction data. Moreover, hAcf1 is required for N-CoR association near the TSS of the RANKL promoter and near the VDRE of the IGFBP3 promoter. Although 1-h vitD3 treatment caused a release of hAcf1 and N-CoR from the regulatory regions of these two genes, the knockdown of hAcf1 caused a similar level of N-CoR dissociation from these genes as did hormone treatment. These results suggest that hAcf1 is a major target of hormone action and that it is critical for both N-CoR association and hormone-induced N-CoR release from these two gene promoters.
N-CoR is thought to stabilize an NR at its response element in the absence of hormone (30). Because knocking down the levels of hAcf1 led to decreased N-CoR association at the IGFBP3-VDRE, it should also lead to the destabilization of VDR at the VDRE. To test this theory, an anti-VDR ChIP assay was performed under the same conditions. There was a remarkable decrease in the amount of VDR at the VDRE of the IGFBP3 promoter when hAcf1 was knocked down (Fig. 4A, right). In agreement with published results, an increase in VDR association with the IGFBP3-VDRE during vitD3 treatment for 1 h was observed (31) (Fig. 4A, right). This increase in VDR binding could be attributable to the stabilizing effect of the ligand on the DNA-bound receptor or vitD3-promoted movement of VDR from the cytoplasm to the nucleus (32). It is important to note that treatments that changed VDR occupancy did not alter the overall level of the VDR protein (Fig. 4A, bottom).
|
Hormone Treatment or hAcf1 Knockdown Promotes Differential Changes in the Histone Modification Pattern and the Histone Occupancy in the RANKL and IGFBP3 Promoters
To better understand the contribution of hAcf1 to repression of hormone-responsive genes, the local chromatin architecture and the histone modification profile at these promoters were studied by ChIP assays. The results demonstrate that knockdown of hAcf1 significantly altered both the histone modification pattern and the occupancy of histones H3 and H4 at the IGFBP3 and RANKL promoters (Figs. 4B and 5).
|
, left). As expected, there was a significant decrease in the level of trimeK9H3 under treatment conditions that increase transcription: RNAi against hAcf1 and 1-h vitD3 treatment (Fig. 4B, right). It should be noted that there was no statistically significant change in the levels of histone H3 itself (Fig. 4C) or the acetylation of histone H3 (data not shown) under these treatment conditions. It seems, therefore, that one of the important contributions to transcriptional derepression at the IGFBP3 promoter is the decrease of trimeK9H3 that occurs when hAcf1 is knocked down or when cells are treated with vitD3.
The histone modification pattern at the RANKL promoter (Fig. 5) was investigated in parallel with the IGFBP3 promoter. Unlike the IGFBP3 gene, when the level of trimeK9H3 was examined by ChIP, there was no significant difference in this modification during hAcf1 RNAi or hormone treatment (data not shown). Levels of acetylation on lysine 9 of histone H3 (acK9H3) decreased, however (Fig. 5A, right), under hAcf1 knockdown condition in the regions of highest acK9H3 occupancy, primer sets C and D (Fig. 5B, left). There was also a significant decrease in acK9H3 levels during 1-h vitD3 treatment. This result was surprising, because this modification usually increases during transcriptional activation and the knockdown of hAcf1 derepressed the gene. The decrease in acK9H3 could be attributable to a loss of a histone subunit from the promoter. Regions of euchromatin have lowered levels of certain histone subunits, such as histone H3, during transcriptional activation and elongation (37, 38).
The level of histone H3 itself was, therefore, examined by ChIP assay to see whether the decrease in acK9H3 correlated with a decrease in overall histone H3 levels at those regions of the RANKL gene. There was indeed a significant decrease in the level of histone H3 near the TSS (primer sets C and D) under the conditions in which acK9H3 decreased (Fig. 5B, top right). ChIP analyses indicated a concomitant decrease in overall histone H4 levels (Fig. 5B, middle right). These results agree with the belief that histones H3 and H4 are deposited or replaced together in chromatin as dimers or tetramers (39). Surprisingly, vitD3 treatment alone caused significant dissociation of these histones from the target promoter. The decrease in the levels of histones H3 and H4 in this gene was not caused by a loss or degradation of the protein because the treatment conditions did not alter their overall protein levels (Fig. 5B, bottom).
The loss of histones H3 and H4 from the derepressed RANKL promoter during vitD3 treatment and hAcf1 knockdown suggests that alterations in repressive chromatin architecture may allow transcription to occur. To further investigate this, a micrococcal nuclease (MNase) DNA accessibility assay was used in Saos-2 cells that were either treated for 5 h with vitD3 or 48 h with RNAi against hAcf1. If histones H3 and H4 are released from the promoter, then MNase accessibility to the underlying DNA at these regions should increase. As a result, the chromatin isolated from these cells will be more susceptible to cleavage and thus less amplified by qPCR compared with control. A statistically significant increase in MNase accessibility during hormone treatment or hAcf1 knockdown in the RANKL promoter gene region was observed (Fig. 6A). MNase accessibility did not significantly increase at any region of the IGFBP3 gene under these same conditions (Fig. 6B). This is consistent with our ChIP assay result: histones H3 and H4 levels remained unchanged on the IGFBP3 gene promoter during hormone or hAcf1 RNAi treatment.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
). A hormone-dependent release of hAcf1 destabilizes both the receptor-repression complex and the nucleosomes that lead to the derepression/activation of hormone-responsive genes. hAcf1 is, therefore, a major mediator of hormone action because its knockdown mimics hormone responsiveness for a number of NR-regulated genes.
|
Only a few studies have indicated that a link exists between members of chromatin remodeling complexes and the process of N-CoR-mediated repression. The SWI/SNF family of ATPase BRG1 was identified in an N-CoR complex, but it role in repression was not determined (16). The ATPase SNF2H is targeted by N-CoR to enhance active repression of a TR-regulated gene (19). Immunoprecipitation experiments failed, however, to show an association of SNF2H and N-CoR, suggesting that its interaction with N-CoR is indirect or transient.
This investigation provides extensive evidence that hAcf1 binds directly to and forms an endogenous complex with N-CoR. The interaction is mediated by the N-terminus of hAcf1 (termed hAcf1-NID), which is both necessary and sufficient for binding to N-CoR. This region of hAcf1 contains a WAC domain, a conserved motif involved in DNA binding (22). We propose that the WAC domain is also critical for the association of hAcf1 with N-CoR because deletion of the region containing this domain prevented the interaction. Consistent with this idea, overexpression of the hAcf1-NID partially relieved repression of NR-regulated genes. Point mutations at critical conserved residues or internal deletion specifically of the WAC domain would determine whether it is the minimal domain needed for the interaction of hAcf1 and N-CoR.
This work also uncovers a previously unrecognized function of Acf1 in transcriptional repression of euchromatic genes in the mammalian system. Before our study, experiments in Drosophila highlighted a role for Acf1 as a nonenzymatic component of the ACF chromatin assembly/remodeling complex. Some studies have suggested that Acf1 is involved in transcriptional repression because it is required for the replication of heterochromatin. Acf1 also assembles chromatin with evenly spaced nucleosomes and that contain histone H1, hallmarks of repressive heterochromatin (28, 29). These results indicated that Acf1 may have a general role in transcriptional repression. The data of this study suggest for the first time that hAcf1 has a direct and specialized role in the repression of euchromatic genes in the mammalian system. This is evidenced by the finding that two VDR-regulated genes were derepressed during knockdown of hAcf1, but other hormone-responsive genes were not significantly affected.
Only one other study provided direct evidence of the repressive properties of Acf1. Acf1 null Drosophila displayed a suppression of position-effect variegation and reduction of Polycomb-mediated transcriptional silencing in heterochromatin (27). Because this study used a genetic knockout approach, the loss of repression in heterochromatin observed may be, at least in part, attributable to the inability of acf1 null Drosophila to assemble heterochromatin with evenly spaced nucleosomes. It remained unclear whether or not Acf1 has a direct role in repression at heterochromatic loci independent of its role in chromatin assembly and replication. Because our investigation uses RNAi-based knockdown in cells that have fully assembled chromatin, it addresses this question. This study shows that hAcf1 directly regulates the transcription of euchromatic genes, independent of its ability to assemble heterochromatin. It is tempting to speculate that, whereas the chromatin assembly and replication function of hAcf1 might establish repressed heterochromatin, the transcriptional repression of euchromatic genes may require hAcf1 to stabilize repressive chromatin architecture at its target promoters. We propose that, in addition to its role in chromatin assembly and replication, hAcf1 is important for repression of both heterochromatic and euchromatic genes and that it is also a potential transcriptional corepressor across species. It remains to be determined whether or not Acf1, like N-CoR/SMRT, harbors any transcriptional RDs. This investigation should encourage future research to determine whether hAcf1 has a direct transcriptional regulatory role in heterochromatic repression in the mammalian system and, conversely, whether the Drosophila homolog of human SMRT/NCoR, termed SMRTER, interacts with dAcf1 and plays any role in heterochromatin silencing (42).
Novel Mechanisms in Hormone-Mediated Transcriptional Regulation
Our work reveals that hAcf1 is critical for the presence of the repressive N-CoR complex at the hormone-inducible RANKL and IGFBP3 genes. RNAi-mediated depletion of hAcf1 results in lowered levels of N-CoR near the TSS of the RANKL gene and at the VDRE of the IGFBP3 gene even in the absence of hormone. hAcf1 knockdown also causes a dissociation of VDR from the IGFBP3 VDRE, suggesting that it helps to stabilize the repressive complex at the promoter. This finding is supported by a recent study in which the paralog to hAcf1, WSTF, was purified with VDR as part of the ATP-dependent chromatin remodeling WINAC (WSTF-inducing nucleosome assembly complex). WSTF bound to VDR in a ligand-independent manner and stabilized the VDR complex at the VDRE (43). Future studies should determine whether hAcf1 and WSTF show gene specificity based on recruitment to targets via N-CoR and/or the receptor itself.
It is important to note that, although we were not able to immunoprecipitate the VDR in the regions of the RANKL gene that were tested in this study, recent evidence suggests that there are at least five VDREs located at significant distances upstream of the TSS as well as one VDR-binding site located very close to the TSS (34, 35). Because the distal VDREs also associate with RNA Pol II, it has been proposed that these VDREs may be brought near the natural promoter of the gene by extensive chromatin looping (44). This may explain why N-CoR (and perhaps also hAcf1) is enriched near the TSS of RANKL. It will be interesting to determine whether hAcf1 (as a member of a chromatin remodeling complex) regulates transcription by maintaining these special chromatin structures.
The results of this study also strongly suggest that hormone functions to release not only N-CoR but also hAcf1 from promoters to cause the derepression of genes, such as RANKL and IGFBP3. Loss of the corepressors hAcf1 and N-CoR from these gene promoters during hAcf1 knockdown leads to an increase in transcription similar to hormone treatment. Derepression caused by the release of corepressors may be, therefore, more critical to the expression of these genes than the recruitment of coactivators. If coactivator recruitment has a major role in transcriptional activation of these genes, one would expect hormone treatment to cause a more robust increase in mRNA levels than hAcf1 knockdown. Interestingly, the IGFBP3 and RANKL VDREs are located several kilobases upstream of the TSS. Because the VDRE is located distal to the TSS, coactivators recruited to this site may not be as effective in activating transcription.
Hormonal activation of transcription of RANKL, IGFBP3, and IGF-I genes may also be weak because the DNA-bound NR does not completely dissociate with N-CoR. In fact, in our studies, 1-h of vitD3 treatment leads to about a 40% reduction of N-CoR at these genes. This is perhaps attributable to the presence of additional corepressors, such as hAcf1, which may enhance N-CoR association. It is, therefore, possible that hAcf1 may establish repression at a specific subset of VDR-regulated genes, those that have a distal VDRE and that are not greatly activated by hormone. Consistent with this suggestion is the fact that the 24-hydroxylase gene, which has a proximal VDRE and is highly up-regulated by vitD3 treatment, is not derepressed by hAcf1 knockdown. This gene probably does not associate with hAcf1 and it may require the addition of hormone and the subsequent recruitment of coactivators to achieve such a robust activation. (This study did not, however, analyze whether hAcf1 is targeted to the 24-hydroxylase gene.) It will be interesting to determine whether this is a common feature of the NR-regulated genes that require hAcf1 for maximum transcriptional repression.
Although hAcf1 stabilizes the repression complex in the promoter of both the IGFBP3 and RANKL genes, hAcf1 knockdown causes different changes to histone modification profile and histone occupancy at these genes. Overall, hAcf1 seems to play a role in maintaining the features of repressive chromatin (e.g., histone methylation and intact nucleosomes). In the IGFBP3 promoter, for example, hAcf1 RNAi and 1-h vitD3 treatment lowered the levels of trimeK9H3, a modification that stabilizes heterochromatic structure by creating a binding site for the repressive heterochromatin protein 1 (45, 46). It is, therefore, possible that hAcf1 recruits a histone methyltransferase to these promoters and/or a histone demethylase is a component of the derepression/activation pathway for these genes.
At the RANKL promoter, reduced levels of histones H3 and H4 were observed under the conditions in which transcription increased: 1-h vitD3 treatment and hAcf1 knockdown. Although the ChIP assay cannot show whether the observed histone loss is attributable to nucleosome repositioning or nucleosome destabilization, either situation creates an open chromatin structure. This open structure allows for histone variant exchange or for the basal transcriptional machinery to access the DNA. The fact that hormone treatment and hAcf1 knockdown increase MNase accessibility to DNA in the regions of the RANKL gene in which the levels of histones H3 and H4 decreased is consistent with this idea. The histone eviction hypothesis, tested in this mammalian system, agrees with studies in yeast that show that there is histone loss and an incorporation of histone variants in actively transcribed regions of the genome (6, 38, 39, 47).
This study uncovers a novel role of hormone (vitD3) in histone H3 and H4 loss from a hormone-responsive gene. Hormone treatment causes release of hAcf1 (and N-CoR) from promoters, which destabilizes nucleosomal structure of euchromatic genes and leads to the eviction of histones H3 and H4. Such an open chromatin structure is poised for transcriptional activation. Because histone eviction/displacement is observed only in the RANKL gene during vitD3 treatment, it will be important to determine which genes are regulated by hormone-induced histone eviction (48). It will also be of interest to analyze whether or not other transcriptional signaling pathways in the mammalian system use signal-dependent histone eviction as a mechanism for gene regulation (49).
Most work on Acf1 investigates its relationship to its enzymatic binding partner, focusing on how Acf1 regulates the function of the ATPase SNF2H. We find that hAcf1 knockdown, which does not lead to a decrease of SNF2H levels at the promoters, correlates with the relief of transcriptional repression, the loss of the corepressor complex, and even the destabilization of the nucleosome itself in these vitD3-regulated genes. This indicates that hAcf1 alone is crucial to the maintenance of repressive chromatin at these genes. Considering that there are at least four SNF2H-containing complexes in humans, these study results should encourage future research to investigate the independent role of other SNF2H binding partners (e.g. WSTF and RSF) in transcription and other chromatin-based processes. This would help to determine what makes these complexes unique.
Our findings also suggest that hAcf1 and its paralog WSTF do not have a redundant function in transcriptional repression of hormone-inducible genes. WSTF does not compensate for hAcf1 in regulating these genes when the level of hAcf1 is reduced. It was shown recently that WSTF has a role in the ligand-induced activation and in transrepression of certain vitD3-target genes (43, 50). Additional characterization of hAcf1 and WSTF will help determine whether or not they play differential roles in the repression of hormone-sensitive genes.
In conclusion, our results establish a unique and gene-specific role for the chromatin remodeling protein hAcf1 in transcriptional repression of nuclear hormone-responsive genes. Because lowering the levels of hAcf1 causes different changes in the histone tail modifications and the histone occupancy of the IGFBP3 and RANKL promoters, we conclude that the local chromatin architecture for each promoter is unique. These results also uncover new mechanisms in NR-regulated transcriptional repression. This includes a novel role for hormone in hAcf1 release and in the eviction of histones H3 and H4 from specific genes. By elucidating another level of regulation of NRs, our data provide a plausible mechanism involving hAcf1 in the differential hormone responsiveness of some VDR-regulated genes as well as highlight the unique nature of individual hormone-responsive genes.
As a result, Acf1 has multiple roles in chromatin biology. It functions as a component of the ACF complex in chromatin assembly and replication. Current work now suggests that hAcf1 is essential not only for the establishment of transcriptionally repressive heterochromatic regions during DNA replication (27) but also for the maintenance of active repression of euchromatic genes in the mammalian system (this work). Results obtained from the current study will allow future investigations to determine the independent and global roles of hAcf1 (as well as of other nonenzymatic subunits of chromatin assembly/remodeling complexes) in the transcription of hormone-responsive and other cellular and viral genes.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Yeast Two-Hybrid Assay
YRG2 yeast strain was cotransformed with GALDBD-N-CoR RD2 and a human leukemia GAL4 activation domain (AD) fusion cDNA library (Clontech, Mountain View, CA). Transformants, 2 x 107, were selected onto SD-Leu-Trp-His plates. Specificity of interaction was confirmed by cotransforming YRG2 with either GALDBD-RD1 or -RD2 and GALAD-hAcf1N and plating onto Leu-Trp-His plates containing 20 mM 3-aminotriazole.
In vitro Interaction Assays
hAcf1N (amino acids 1–313) and N-CoR RD1 (amino acids 1–312) were subcloned into the pGEX-4T2 vector (GE Healthcare, Little Chalfont, UK) and transformed into BL21 cells. GST fusion proteins were isolated by glutathione-Sepharose 4B beads according to the protocol of the manufacturer (GE Healthcare). N-CoR, SMRT, and hAcf1 deletion constructs subcloned into pCDNA3 vectors were synthesized in vitro in the presence of [35S]methionine following the protocol of the manufacturer (Promega, Madison, WI). In vitro interaction assays were performed as described previously (51).
Cell Culture
Saos-2 and HEK293T cells were cultured as monolayers in DMEM and supplemented with antibiotics and 15 or 10% fetal bovine serum (FBS) (Hyclone, Logan, UT), respectively. Saos-2 cells were kept in DMEM supplemented with 10% charcoal-stripped FBS (Gemini Bio-Products, West Sacramento, CA) for 48 h before hormone treatments.
Coimmunoprecipitation
Coimmunoprecipitations were performed as described with minor modifications (51). HEK293T cell lysate was incubated overnight at 4 C with hAcf1 antibody or equal amount of IgG. After SDS-PAGE, proteins were transferred to nitrocellulose membrane and incubated with anti-N-CoR antibody (1:500). Western blots were visualized using enhanced chemiluminescence detection (GE Healthcare).
Transient Protein Expression
Saos-2 cells were plated in 10-cm plates and transfected at 80% confluency by liposome-mediated transfer (Lipofectamine 2000; Invitrogen, Carlsbad, CA) with 20 µg of the following plasmids: control pcDNA3, pcDNA3-hAcf1FL, pcDNA3-hAcf1-NID, and CMX-N-CoR. Four to 5 h after transfection, medium was changed to DMEM containing 10% charcoal-stripped FBS. Incubation continued for 48 h before harvesting. Under these conditions, cells were transfected at greater than 50% efficiency (as measured by green fluorescent protein-tagged protein expression). Transient overexpression of N-CoR was measured by Western blot. Expression levels of hAcf1FL and hAcf1-NID were measured by qPCR because antibody to hAcf1 does not recognize hAcf1-NID.
RNAi Experiments
Saos-2 cells were transfected at approximately 80% confluency using Lipofectamine 2000. RNAi at 10 nM was used as a double-stranded duplex for hAcf1 (5'-AAC ACU GUG AAC CAC AAG AUG UU-3') and siCONTROL nontargeting siRNA #1 (Dharmacon, Lafayette, CO). Four to 5 h after transfection, medium was changed to DMEM containing 10% charcoal-stripped FBS. Incubation continued for 48 h before harvesting or hormone treatment.
Isolation and RT of mRNA
Saos-2 cells were lysed in RNeasy lysis buffer and homogenized over Qiashredder columns. Total RNA was extracted with the RNeasy kit (Qiagen, Valencia, CA) and reverse transcribed with the Superscript II Ribonuclease H RT Kit (Invitrogen) using 2 µg RNA.
Real-Time PCR
Quantification of mRNA and DNA was performed in an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA), using SYBR Green as detector dye. Reaction mixtures contained 5 µl SYBR Green, 200 nM primer, and 20–40 ng template cDNA. For mRNA quantification, primers were designed to cross introns (sequences available on request). Reaction conditions were 95 C for 2 min, followed by 45 cycles of 95 C for 15 sec and 60 C for 45 sec. Data were analyzed with threshold set in the linear range.
ChIP Assay
DNA immunoprecipitation was performed according to the ChIP protocol of Upstate Biotechnology with minor modifications. Control or hAcf1-RNAi-treated Saos-2 cells were maintained in 10% charcoal-stripped FBS for 48 h. Half of the control cells were treated with 10 nM vitD3 for 1 h. Harvested cells were formaldehyde-treated for 5 min at 37 C and sonicated for three cycles of 15 sec at level 3 of a Misonix Sonicator (Misonix, Farmingdale, NY). Equal amounts of lysate were incubated overnight at 4 C with 3–5 µg antibody. Immunoprecipitates were collected and eluted, and crosslinks were reversed. DNA was purified using the Qiagen PCR purification kit and used in qPCR reactions with primers corresponding to five regions in the RANKL and IGFBP3 genes (sequences available on request).
MNase DNA Accessibility Assay
Saos-2 cells were washed twice in PBS, resuspended in lysis buffer [10 mM Tris-HCl (pH 7.5), 10 mM NaCl, 3 mM MgCl2, and 0.4% Nonidet P-40], and incubated for 10 min on ice. Nuclei were pelleted and resuspended in the same buffer plus 1 mM CaCl2. The A260 of nuclei preparation was adjusted to 0.5, and lysate was aliquoted into 300 µl portions. Samples were digested with either 0 or 3 U of MNase (Sigma) for 15 min at 37 C. Reactions were stopped with 60 µl of 60 mM EDTA, 3% SDS. Samples were treated with 10 µg of ribonuclease A at 37 C for 30 min, followed by 100 µg proteinase K at 55 C for 1 h. Samples were phenol-chloroform extracted twice and then ethanol precipitated. DNA was quantitated by real-time PCR with primers that generate amplicons of about 90 bp (sequences available on request).
Statistical Analysis
Real-time PCR results are expressed as mean ± SD, and statistical analysis was done by paired Student’s t test.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
The authors have nothing to disclose.
First Published Online May 22, 2007
Abbreviations: Acf, ATP-utilizing chromatin assembly and remodeling factor; acK9H3, acetylation on lysine 9 of histone H3; ChIP, chromatin immunoprecipitation; d, Drosophila; FBS, fetal bovine serum; FL, full length; GAL4, galactosidase 4; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GST, glutathione S-transferase; h, human; hAcf1N, N-terminus of human ATP-utilizing chromatin assembly and remodeling factor 1; HDAC, histone deacetylase; MNase, micrococcal nuclease; N-CoR, nuclear receptor corepressor; NID, N-CoR interaction domain; NR, nuclear hormone receptor; Pol II, polymerase II; qPCR, quantitative real-time PCR; RANKL, receptor activator of nuclear factor-B ligand; RAR, retinoic acid receptor; RD, repressor domain; RNAi, RNA interference; SMRT, silencing mediator for retinoid and thyroid receptor; TR, thyroid receptor; trimeK9H3, trimethylation of lysine 9 of histone H3; TSS, transcription start site; VDR, vitamin D receptor; VDRE, vitamin D response element; WSTF, Williams syndrome transcription factor.
Received for publication February 19, 2007. Accepted for publication May 14, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
B ligand gene expression by 1,25-dihydroxyvitamin D3 is mediated through multiple long-range enhancers. Mol Cell Biol 26:6469–6486NURSA Molecule Pages Link:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
