Nuclear organization and chromatin dynamics – Sp1, Sp3 and histone deacetylases

https://doi.org/10.1016/j.advenzreg.2007.11.016Get rights and content

Introduction

Regulation of gene expression involves the coordinated activities and interplay between chromatin remodeling factors and transcription factor recruitment. Histone acetyltransferases, histone deacetylases, histone kinases, histone phosphatases, histone methyltransferases, histone demethylases and ATP-dependent chromatin remodeling complexes mediate chromatin remodeling and are components of a complex epigenetic network regulating gene expression during development and differentiation. Transcription factors play key roles in the recruitment of histone modifying enzymes and chromatin remodeling complexes to specific gene promoters. Sp1 and Sp3 are two transcription factors that are expressed in all mammalian cells and are involved in the regulation of genes involved in most cellular processes. Remodeling of chromatin is a necessary event in preparing the gene for transcription.

In this review we will cover the organization and remodeling of chromatin, with a focus on dynamic histone acetylation and the histone deacetylase enzymes. The structure and function of transcription factors Sp1 and Sp3 will be presented. The role of these factors in the regulation of the estrogen responsive trefoil factor 1 gene will be highlighted. In the analyses of the factors involved in the regulation of the expression of a specific gene, the chromatin immunoprecipitation assay in which the protein factor of interest is cross-linked to DNA with formaldehyde is an essential tool. The limitations of this assay in cancer cells in which genomic instability is rampant are discussed.

Nuclear DNA is packaged into nucleosomes, the basic repeating structural units in chromatin. The nucleosome consists of a histone octamer, arranged as a (H3–H4)2 tetramer and two H2A–H2B dimers, around which DNA is wrapped. Histone H1 binds to the linker DNA, which joins nucleosomes together, and to core histones, H2A and H4. The core histones (H2A, H2B, H3, H4) have a similar structure with a basic N-terminal domain, a globular domain organized by the histone fold, and a C-terminal tail (Fig. 1). The crystal structure of the nucleosome shows that the N-terminal tails emanate from the nucleosome in all directions and would be available to interact with linker DNA, nearby nucleosomes or with other proteins (Luger and Richmond, 1998).

Core histones undergo post-translational modifications at many sites, including acetylation, methylation, ubiquitination, and phosphorylation (Fig. 1). Most core histones' modifications were found within the N-terminal and C-terminal tails. However, with the advent of sensitive mass spectrometry methods, we have learned that several modifications reside in the histone fold (Cosgrove, 2007, Cosgrove et al., 2004, Cosgrove and Wolberger, 2005). Some modifications (active marks) are generally associated with transcriptionally active chromatin regions while others (repressive marks) correlate with silent regions. Histone acetylation usually marks active genes, and di- or tri-methylation of K4 of H3 is also an active mark whereas di- or tri-methylation of H3 at K9 constitutes a repressive mark (Peterson and Laniel, 2004, Sims and Reinberg, 2006). The role of a given mark or group of marks may vary with the cellular context and the gene under study. Added to the complexity is the dynamics of histone modifications which can change very rapidly and histone variants (Ausio, 2006, Clayton et al., 2006). H2A, H2B and H3 have variants that are expressed at the time of DNA synthesis (e.g. H3.1, H2A.1; replication dependent) and those that are expressed throughout the cell cycle (H3.3, H2A.Z; replication independent). There are also specialized variants, e.g. macroH2A and CENP-A which have roles in mammalian chromosome X inactivation and the centromere, respectively (Ausio et al., 2001, Ausio and Abbott, 2002). H3.1 and H3.3 are components of different histone assembly complexes that distinctly discriminate their functional roles in replication-coupled incorporation from transcription specific deposition, respectively (Ahmad and Henikoff, 2002, Tagami et al., 2004). H3.3 is located in regulatory regions of genes and closely corresponds to DNAase I hypersensitive sites (Jin and Felsenfeld, 2006, Mito et al., 2007). H3.3 is enriched in active marks (di- and tri-methylated K4 and acetylated K9, 14, 18 and 23) (Loyola et al., 2006). Further, H3.3 is important in transcription and in the prevention of heterochromatin spread (Jin and Felsenfeld, 2006). Recent studies found that epitope-tagged H3.3 is enriched in promoter regions in avian cells regardless of the gene's transcriptional activity (Jin and Felsenfeld, 2006) which is in agreement with our earlier studies showing that endogenous newly synthesized H3.3 was incorporated into transcription active/competent chromatin independent of transcription (Hendzel and Davie, 1990). In contrast yeast H3.3 is incorporated only in promoters of active genes (Jamai et al., 2007). At promoters and DNAase I hypersensitive sites, nucleosomes are dynamically disassembled and reassembled with the incorporation of H3.3 (Mito et al., 2007). Mammalian and Drosophila H3.3 is found enriched in active marks including H3 K4 tri-methylated, H3 K9 acetylated and H3 K14 acetylated (Loyola et al., 2006, McKittrick et al., 2004). Recent evidence suggests that a nucleosome containing H3.3 and H2A.Z is less stable than nucleosomes not containing these histone variants (Jin and Felsenfeld, 2007).

At physiological ionic strength chromatin is folded into higher order structures. H1 and core histone N-terminal tails stabilize the folding of the chromatin fiber. Transcriptionally active and poised genes are found in DNAase I-sensitive, decondensed chromatin domains that are accessible to transcription factors and transcription machinery, while transcriptionally repressed genes are in condensed chromatin regions. Transcribed/poised chromatin domains are associated with acetylated histones (Calestagne-Morelli and Ausio, 2006, Espino et al., 2005). The core histones are reversibly modified by acetylation of lysines located in their basic N-terminal domains. These sites of acetylation are evolutionarily highly conserved. Histone acetylation is not limited to transcriptionally active chromatin, but also has a role in DNA replication, DNA repair, and in spermatogenesis.

Core histone acetylation and deacetylation are catalyzed by histone acetyltransferases (HATs, EC number 2.3.1.48) and histone deacetylases (HDACs, EC number 3.5.1), with the level of acetylation being decided by the net activities of these two enzymes (Spencer and Davie, 1999). In mammalian cells, transcriptionally active chromatin regions have core histones undergoing high rates of acetylation and deacetylation, while in repressed chromatin regions the rate of reversible acetylation is slow. In MCF-7 (estrogen receptor α positive, ER+) and MDA MB 231 (ER−) 11–14% of the histones are engaged in rapid acetylation–deacetylation, while 50–60% of the histones are involved in a slow rate of acetylation (Sun et al., 2001). In ER+ human breast cancer cells, estrogen slightly increased the steady state level of acetylated histones. An analysis of the kinetics of histone acetylation/deacetylation revealed that estrogen decreased the rate of deacetylation, with no effect on acetylation rate (Sun et al., 2001).

Approximately 2% of the adult chicken immature erythrocyte genome has core histones that are dynamically acetylated and deacetylated. Most erythrocyte core histones are frozen in unacetylated and monoacetylated states (Zhang and Nelson, 1986). The dynamically acetylated core histones are acetylated at one rate (e.g. t1/2 = 12 min). One population of dynamically acetylated H2B, H3 and H4, called class 2 (Hendzel et al., 1991), is rapidly acetylated to the mono- and di-acetylated states, and is slowly deacetylated. Another population of dynamically acetylated H2B, H3 and H4, called class 1 (Hendzel et al., 1991), is rapidly acetylated to the tetra-acetylated state, and the tetra-acetylated isoform is rapidly deacetylated (t1/2 = 5 min).

Incubating avian immature erythrocytes in the presence of an HDAC inhibitor (butyrate) for 60–90 min will drive the dynamically acetylated (class 1) histones to their highest acetylated states (Walia et al., 1998). Although our previous work suggested that class 1 dynamically acetylated histones were associated with active but not poised chromatin, a further investigation showed that active (β-globin) and poised (ɛ-globin) chromatin had class I acetylated histones (Spencer and Davie, 2001).

There is still much to learn about the role of histone acetylation in chromatin. Histone acetylation destabilizes higher order chromatin structure, enhances solubility of chromatin at physiological ionic strength, alters histone tail conformation, alters the capacity of H1 to condense the chromatin fiber, and maintains the unfolded structure of the transcribed nucleosome (Spencer and Davie, 1999, Wang et al., 2001). A recent study shows that H4 acetylation at K16 has a pivotal role in the decondensation of chromatin (Shogren-Knaak et al., 2006). Acetylation at specific sites on a core histone may also be “read” by other proteins. For example, the bromodomain recognizes an acetylated lysine, and proteins with this motif may bind to a specific acetylated lysine residue (Lee and Workman, 2007).

We have extensively characterized 0.15 M NaCl salt-soluble polynucleosomes isolated from avian immature erythrocytes which are highly enriched in transcribed/poised DNA sequences and depleted in repressed DNA (Delcuve and Davie, 1989). The 0.15 M NaCl-soluble poly- and oligonucleosome fractions contain only 3% of the total nuclear DNA, but approximately 25% of the total active sequences and approximately 45% of the total competent sequences. In contrast, only 0.5% of the total repressed gene sequences is present in these fractions (Delcuve and Davie, 1989). This chromatin fraction is essentially representative of transcriptionally active/poised chromatin. The salt-soluble poly- and oligonucleosomes contain highly acetylated species of H3, H2B and H4, ubiquitinated (u) and polyubiquitinated species of H2A and more strikingly uH2B (Delcuve and Davie, 1989). Other characteristics of active/competent chromatin fractions are the highly dynamic acetylation/deacetylation of the core histones (Hendzel et al., 1991) and the preferential methylation of acetylated H3 and H4 (Hendzel and Davie, 1991). Moreover, newly synthesized histones H2A and H2B and to a lesser extent H3.3 and H4 preferentially exchange with nucleosomal histones of transcriptionally active/competent chromatin domains, suggesting that nucleosomes of active chromatin may be inherently less stable than bulk nucleosomes in vivo (Hendzel and Davie, 1990). Supporting this idea, structural studies using electron spectroscopic imaging have shown that only 66% of the nucleosomes of the active/competent chromatin present the circular profile seen in 90% of the bulk chromatin, while the remaining nucleosomes appear to be u-shaped or elongated. Some of the nucleosomes with an altered morphology have a lower protein mass and may be devoid of an H2A–H2B dimer (Locklear et al., 1990).

Transcribed sequences are also present in the residual nuclear material. As this fraction contains bulk repressed chromatin, biochemical characterization of transcriptionally active chromatin in this fraction has been challenging (Delcuve and Davie, 1989, Hendzel et al., 1991).

HATs, HDACs, histone kinases, histone phosphatases, histone methyltransferases, histone demethylases and ATP-dependent chromatin remodeling complexes mediate chromatin remodeling and are components of a complex epigenetic network regulating gene expression during development and differentiation. Since the first purification and cloning of an HAT (Gcn5) (Brownell et al., 1996) and HDAC (HDAC1) (Tauton et al., 1996), multiple HATs and HDACs have been identified (reviewed in Davie and Moniwa, 2000, Lee and Workman, 2007, Spencer and Davie, 1999, Yang and Seto, 2007). HATs often have transcriptional coactivator activity and when recruited to a gene promoter by a transcription factor will increase the level of acetylated histones and enhance transcriptional activity of the promoter (Lee and Workman, 2007). There are four classes of HDACs (De Ruijter et al., 2003, Glozak and Seto, 2007). Class I HDACs consist of mammalian HDACs, HDAC1, HDAC2 (the mammalian homolog of yeast RPD3), HDAC3 and HDAC8. Class II HDACs include mammalian HDAC4, HDAC5, HDAC6, HDAC7, HDAC9 and HDAC10. Class III HDACs are members of the sirtuin (SIRT) family of NAD+-dependent HDACs, among which yeast Sir2 is the founding member. Class IV HDAC has HDAC11.

Mammalian HDAC1 and HDAC2 are in large multiprotein complexes, Sin3, NuRD and coREST (De Ruijter et al., 2003). The Sin3 complex contains Sin3A, SAP18, SAP30, and retinoblastoma associated proteins (RbAp) 46 and 48 (Grozinger and Schreiber, 2002). The Sin3 complex is directed to its target chromatin location by sequence-specific DNA binding proteins that interact directly with Sin3A and other components of the Sin3 complex. Transcription factors including Sp1, Sp3, Mad-family proteins, unliganded hormone receptors, and p53 recruit the Sin3 complex to regulate target genes. In addition MeCP2 and DNA methyltransferases recruit HDAC1, resulting in cooperativity between DNA methylation and histone deacetylation in gene silencing (Davie and Moniwa, 2000). NuRD (nucleosome remodeling histone deacetylase complex) is about 2 MDa in size and consists of MTA2 (highly related to metastasis-associated protein MTA1), Mi2, RbAp46/48 and MBD3 (methyl-CpG-binding domain-containing protein) (Bowen et al., 2004). NuRD has both ATP-dependent chromatin remodeling and HDAC activities. The coREST complex has HDAC1 and HDAC2 but unlike the other HDAC complexes lacks RbAp46/48 (Grozinger and Schreiber, 2002).

In MCF-7 human breast cancer cells we found that most, if not all, of HDAC2 is in complex with HDAC1. However, HDAC1 is in excess of HDAC2. HDAC1 and HDAC2 account for at least 50% of the HDAC activity in MCF-7 cells. Breast cancer cell lines such as MCF-7 cells have elevated expression of HDAC1, HDAC2 and HDAC3 but not HDAC8 relative to normal breast epithelial cells (Feng et al., 2007).

Several transcription factors repress gene expression by recruiting HDAC1/HDAC2 corepressor complexes to the promoters that they affect. Further, pending the promoter context, transcription factors recruit HDAC1 and HDAC2 corepressor complexes to mediate dynamic deacetylation of histones and non-histone chromosomal proteins associated with or close to the promoter. We reported that Sp1 and Sp3 transcription factors are preferentially associated with phosphorylated HDAC2 in breast cancer cells, although most of the HDAC2 is not phosphorylated in these cells. Others and we reported that mammalian HDAC1 and HDAC2 are phosphorylated by the protein kinase CK2 (Cai et al., 2001, Pflum et al., 2001, Sun et al., 2002, Tsai and Seto, 2002). HDAC1 is phosphorylated at serines 421 and 423, while HDAC2 is phosphorylated at serines 394, 422 and 424. HDAC2 is primarily, if not solely, phosphorylated by CK2 (Tsai and Seto, 2002). Interestingly although the nonphosphorylated form of HDAC2 is considerably more abundant than the phosphorylated form, it is the phosphorylated form that is preferentially cross-linked to chromatin with formaldehyde or cisplatin (Sun et al., 2002).

Mutagenesis studies have demonstrated that phosphorylation of HDAC1 and/or HDAC2 is required for the enzyme's association with RbAp48, which is a component of both the Sin3A and NuRD HDAC complexes. RbAp48 appears to be tightly associated with HDAC1 as it co-purified with HDAC1 (Tauton et al., 1996). Mutation of the CK2 phosphorylation sites in HDAC1 or HDAC2 results in the mutant HDAC not associating with RbAp48, the Sin3A or NuRD complexes, suggesting an important role of HDAC1/HDAC2 phosphorylation in Sin3A and NuRD complex formation. Analyses of the binding of RbAp48 to HDAC1 revealed that RbAp48 bound to the N-terminal section (first 51 amino acids) of HDAC1 in situ in HeLa cells (Taplick et al., 2001). In GST pull down experiments it was demonstrated that RbAp48 interacted with avian HDAC2 regions encompassing amino acids 82–180 and 245–314 (Ahmad et al., 1999). One concern with this study, however, is that the HDAC2 GST fusion is catalytically inactive, a general problem with HDAC1 and HDAC2 produced in bacteria. Thus the folding of the HDAC2 protein in vitro may not be the same as it is in situ. Nevertheless it is interesting that one of the HDAC2 regions interacting with RbAp48 is next to the region containing the three phosphorylation sites. A recent study demonstrated that the C-terminal region of HDAC1 (HDAC1Δ428) is required for its enzymatic activity (Karwowska-Desaulniers et al., 2007). However, the HDAC1 C-terminal region was dispensable for binding to RbAp48. To reconcile the mutagenesis' studies with the deletion experiments, it has been proposed that the phosphorylation events are required to promote proper protein folding and enzyme activity but is not required for activity after a functional enzyme is produced (Karwowska-Desaulniers et al., 2007). Alternatively, the phosphorylation of HDAC2 may induce a protein conformation which exposes the RbAp48 binding sites to RbAp48. Further studies will be required to clarify the role of HDAC1 and HDAC2 phosphorylation in regulating its interaction with other proteins.

The nuclear matrix consists of both nuclear proteins and RNA. The nuclear matrix proteins may be analyzed by isolating the nuclear matrices typically by DNAase I digestion and extractions with 0.25 M ammonium salt (Samuel et al., 1997). Milder methods for isolating nuclear matrices are also available (Jackson and Cook, 1985). Nuclear matrix proteins (NMPs) represent about 30% of the nuclear protein. This subset of the cellular proteome includes proteins with roles in the organization and function of nuclear DNA. NMPs are involved in establishing chromatin loop domains and in the organization of chromosome territories (Coffey, 2002, Jackson, 2003). The nuclear matrix has a pivotal role in the processing of the genetic information (Stein et al., 2004). DNA replication, transcription and DNA repair occur at defined nuclear matrix sites (Dimitrova and Berezney, 2002, Jackson, 2003). Transcription factors including tumor suppressors (e.g. Rb, p53) and hormone receptors (estrogen receptor) dynamically associate with specific nuclear matrix sites. The cancer cell nuclear matrix proteome has proteins involved in the aberrant processing of the genetic information, the disorganization of the genome, and altered nuclear structure. As such, these NMPs are potential biomarkers of the disease.

Enzymes involved in chromatin remodeling such as SWI/SNF, HAT and HDACs (HDAC1, HDAC2) are associated with the nuclear matrix. Similarly, HDAC1 and HDAC2, but not HDAC4, were associated with the nuclear matrix of breast cancer cells (Sun et al., 2001). How the HDACs are targeted to the nuclear matrix is unknown. Presumably the proteins will have a nuclear matrix targeting domain as identified for the RUNX transcription factor (Harrington et al., 2002). F-actin, a component of the nuclear matrix, is involved in the nuclear matrix binding of the HDACs as well as p53 (Andrin and Hendzel, 2004, McDonald et al., 2006, Okorokov et al., 2002). It will be interesting to determine whether HDAC1 and HDAC2 bind directly to actin or more likely to an actin binding protein.

Sp1 was the first mammalian transcription factor to be purified and characterized. In addition to Sp1, mammalian cells express another Sp1-like protein, Sp3. Sp1 and Sp3 regulate the transcriptional activity of many genes involved in a wide range of biological processes, including differentiation, cell cycle progression, and oncogenesis (Jinawath et al., 2005, Li et al., 2004, Sapetschnig et al., 2004). A recent report demonstrates that compound heterozygous Sp1/Sp3 mice are not viable, providing evidence that the levels of both transcription factors are important in maintaining appropriate gene expression programs (Kruger et al., 2007).

Sp1 and Sp3 are members of the Specificity Protein/Krüppel-like Factor (SP/KLF) transcription factor family (Suske et al., 2005). Within this family the nine Sp members are distinguished from the KLF members by the presence of Buttonhead (BTD) domain on the N-terminal side of the DNA binding domain. Sp1, Sp2, Sp3 and Sp4, which have similar modular structure, are a subgroup of the Sp members. In mammalian cells Sp1 and Sp3 are ubiquitously expressed, while Sp2 and Sp4 have a restricted expression pattern. Sp4 expression is high in the central nervous system and in retinal neurons (Lerner et al., 2005). Sp2 is expressed in several cell lines and at higher levels in cancer cells (Phan et al., 2004).

Sp proteins have several sub-domains with respective functions (Li et al., 2004) (Fig. 2). The transactivation domain in Sp1 and Sp3 consists of two sub-domains (A and B), each of which can stimulate transcription. The Buttonhead (BTD) element within domain C may contribute to the factor's transactivation potential. Carboxyl-terminal to the domain C are three Cys2–His2 zinc “fingers”, which are required for sequence-specific DNA binding to GC-rich promoter elements. A carboxyl-terminal domain, termed D, is required for Sp1's synergistic activation (Pascal and Tjian, 1991). The inhibitory domain of Sp1 is located at the N-terminus and that of Sp3 is immediately in front of the zinc-finger domain. This difference in the positioning of the inhibitory domain in the two proteins is believed to be a major reason for the distinct functions of Sp1 and Sp3 (Suske, 1999). Sp1 and Sp3 undergo a number of post-synthetic modifications which add to their complexity (Sapetschnig et al., 2004).

Sp3 has four isoforms, two long (L1-Sp3, L2-Sp3) and two short isoforms (M1-Sp3, M2-Sp3) that are the products of differential translational initiation (Sapetschnig et al., 2004) (Fig. 2). The factors regulating the translational initiation of Sp3 mRNA are poorly understood. The structure of long Sp3 isoforms is very similar to that of Sp1, except for the position of the repression domain (Suske, 1999). The short and long Sp3 isoforms are expressed in all mammalian cells. The Sp3 short isoforms do not have the activation domain A but retain the activation B and the inhibitory domains.

The consensus Sp1 binding sequence is GGGGCGGGG. Sp1/Sp3 binds to variants of this consensus sequence but with reduced affinity (Letovsky and Dynan, 1989). Sp1 will also bind to an Sp site in a nucleosome (Li et al., 1994, Utley et al., 1997). The affinity of Sp1 for a nucleosomal binding site is 10–20-fold less than that in naked DNA. The Sp1 affinity for a nucleosomal binding site diminishes further as the site is placed closer to the center of the nucleosomal DNA (Li et al., 1994).

For promoters containing multiple Sp-binding sites, Sp1 exerts its transcriptional synergism through direct protein–protein interaction (Mastrangelo et al., 1991, Su et al., 1991). In vitro studies demonstrate that Sp3 cannot synergistically activate transcription of promoters containing multiple Sp-binding sites (Yu et al., 2003). On an individual Sp1 binding site, purified Sp1 bound as a multimer, while Sp3 bound as a monomer. Several studies have reported that Sp3 efficiently represses the Sp1-dependent transcription of promoters containing adjacent multiple Sp-binding sites. For these promoters Sp3 competes with Sp1 binding to the GC boxes and thereby blocks the synergistic transactivation function of Sp1. However, there are promoters for which Sp3 activates rather than represses (for review see Li et al., 2004). These and other studies demonstrate that the repressive action of Sp3 is promoter context dependent.

There are significant differences between Sp1 and Sp3 as illustrated in knock out studies and studies investigating their transcriptional roles (Kruger et al., 2007). Our results add to the list of features that distinguish Sp1 from Sp3. Through application of high resolution fluorescence deconvolution microscopy and concurrent indirect immunolocalization of Sp1 and Sp3, we demonstrated that Sp1 and Sp3 are localized in distinct non-overlapping foci in the nucleus (He et al., 2005). The subnuclear foci containing Sp1 or Sp3 were infrequently associated with sites of transcription. Sp1 and Sp3 were present in distinct foci associated with the nuclear matrix. It is possible that the nuclear matrix binding sites, which transiently retain these transcription factors, regulate the level of Sp1 and Sp3 available to associate with Sp-binding sites in chromatin. Alterations in either Sp1 or Sp3 targeting to nuclear matrix sites or changes in the concentrations of these factors could result in aberrant remodeling of chromatin leading to dysfunction of the genome, including genomic instability.

We found that throughout the mitotic process, while being displaced from the condensed chromosomes and dispersed through the cell, Sp1 and Sp3 maintained their separate punctate distributions. In metaphase, Sp1 and Sp3 foci show a high degree of colocalization with actin filaments, suggesting that F-actin is involved in the organization of Sp1 and Sp3 domains during mitosis. In late telophase, Sp1 and Sp3 were equally segregated between daughter cells, and their subnuclear organization as distinct foci is restored. Thus, the maintenance of Sp1 and Sp3 nuclear levels and their correct functionality are assured after cell division (He and Davie, 2006).

Sp1 and Sp3 associate with many other proteins, but these two transcription factors do not bind to each other (Sun et al., 2002, Yu et al., 2003). Sp1 and Sp3 interact directly or indirectly with transcription factors, transcriptional regulators and chromatin remodeling factors (e.g. estrogen receptor (ER) α, HDAC1, p300/CBP, SWI/SNF, an ATP-dependent chromatin remodeling complex) to activate or repress gene expression (Li et al., 2004). Co-immunoprecipitation and indirect immunofluorescence studies demonstrated that Sp1 and Sp3 associate with HDAC1 and HDAC2 and with ERα, albeit at low frequencies in MCF-7 cells (He et al., 2005). Sp1 and Sp3 may act as repressors by recruiting the Sin3A HDAC1/HDAC2 complex by binding to RbAp48 and/or Sin3A (Clem and Clark, 2006, Zhang and Dufau, 2003a). Alternatively, Sp1 and Sp3 may act as a transcriptional activator by recruiting p300 or CBP, which are coactivators with potent HAT activity (Ammanamanchi et al., 2003).

Estrogen can induce the expression of estrogen responsive genes through forming complexes between the ERα and Sp1 or Sp3 (He et al., 2005, Kim et al., 2003). These factors further recruit coactivators that interact with the transcription initiation complex to start transcription (Kim et al., 2003, Stoner et al., 2004). Transcriptional repression can be achieved by perturbation of communication between Sp1/Sp3 and the basal transcription initiator complex (Tan et al., 2003, Zhang and Dufau, 2003b). Alternatively, proteins (e.g. c-Myc) binding to Sp1 can block its transactivation activity (Gartel et al., 2001, Gartel and Shchors, 2003). Taken together, Sp1 and Sp3 can dynamically recruit and form complexes with many proteins, which can cause region-specific changes in histone acetylation and RNA polymerase II recruitment within promoters and in turn activate or repress gene expression.

Sp1 and Sp3 protein levels are often greater in cancer cells than in normal cells (Lou et al., 2005). For example, Sp1 levels were greater in breast carcinomas compared to benign breast lesions (Zannetti et al., 2000); Sp1 was expressed at higher levels in human hepatocellular carcinomas compared to control livers (Lietard et al., 1997); Sp1 levels were greater in human thyroid tumors than in normal thyroid tissues (Chiefari et al., 2002); and Sp1 levels were greater in human gastric cancer tissue than in normal adjacent gastric mucosal tissue (Jiang et al., 2004, Kitadai et al., 1992). Of note, Sp1 expression is a predictor of survival of gastric cancer (Wang et al., 2003). Acknowledging that Sp1 overexpression had a role in gastric cancer, it was proposed that reducing the level of Sp1 would reduce the metastatic potential of the gastric cancer cells. Indeed, decreasing expression of Sp1 with Sp1 specific siRNA in gastric cancer cells reduced their growth and metastatic potential when injected into the stomach wall of mice (Jiang et al., 2004). Further, knocking down Sp1 and Sp3 levels to those of normal cells reduced the potential of fibrosarcoma cells to form tumors in mice (Lou et al., 2005).

Sp1 and Sp3 are involved in the regulation of thousands of genes involved in diverse biological processes (Jinawath et al., 2005, Liang et al., 2004). A recent ChIP on chip study analyzing the distribution of Sp1 binding along human chromosomes 21 and 22 demonstrated a large number of Sp-binding sites (Cawley et al., 2004). Interestingly only about 20% of the Sp sites were those at the 5′ end of protein coding genes, with about 40% of the sites being located at the 3′ end of noncoding RNA genes. Considering that (1) Sp1/Sp3 binding sites are distributed widely in chromosomes, (2) Sp1/Sp3 overexpression is seen in transformed cells and (3) reduction of Sp1/Sp3 overexpression prevents transformation (invasion), it is possible that overexpression of Sp1/Sp3 leads to changes in gene expression and chromatin structure.

Estrogen (E2) responsive elements (EREs) of genes expressed in human breast cancer cells often have an ERE or half site ERE positioned next to an Sp1 binding site (ERE (1/2) (N)x Sp1) (Abdelrahim et al., 2002). Cathepsin D, retinoic acid receptor α (RARα) and tumor growth factor α (TGFα) have a half site ERE located near one or more Sp sites, and these Sp sites are required for the E2 response (Vyhlidal et al., 2000). The Sp sites are also required in the estrogen response of cyclin D1 promoter, which has Sp sites but neither an ERE nor half site ERE (Castro-Rivera et al., 2001). The E2 responsive c-myc promoter has several Sp sites involved in its promoter activity but lacks an ERE (Dubik and Shiu, 1992, Miller et al., 1996). Chromatin immunoprecipitation (ChIP) assays demonstrated that E2 addition to MCF-7 breast cancer cells resulted in the association of Sp1 and ER with the cathepsin D and cyclin D1 promoters (Castro-Rivera et al., 2001). The loading of ER onto E2 responsive promoters may be direct or indirect. For promoters with a consensus or near consensus ERE, ER binding is likely direct. However, for other estrogen responsive promoters ER may bind indirectly through another transcription factor (e.g. Sp1, FoxA1) (Carroll et al., 2005, Lin et al., 2007). Cyclin D1 and c-myc promoters do not have an ERE or half site ERE and thus ER binds indirectly to DNA via Sp1 (Castro-Rivera et al., 2001).

The trefoil factor 1 (TFF1, formerly pS2) gene is responsive to estrogen and TPA in estrogen receptor α positive (ER+) breast cancer cells. TFF1 is a small secreted protein that acts as a proinvasive and angiogenic agent (Prest et al., 2002, Rodrigues et al., 2003a, Rodrigues et al., 2003b). The TFF1 promoter has an Sp1 site, an imperfect ERE and two AP-1 sites with the upstream site being imperfect (Sun et al., 2005). The TFF1 promoter has two positioned nucleosomes, NucE and NucT, which undergo extensive remodeling during transcription (Mellor, 2006, Metivier et al., 2003, Sewack and Hansen, 1997) (Fig. 3). NucE, which harbors the ERE, is positioned between −70 and −410, with the ERE being positioned at −405 and −393. TFF1's NucE remained in place in MCF-7 cells cultured in the absence and presence of E2. The position of NucE is preferred rather than a fixed one (Metivier et al., 2003). The binding site for the TATA-binding protein is on the edge of NucT. Relative to the preferred position of NucE, the Sp1/Sp3 site is located in the linker DNA region upstream of the nucleosome (Fig. 3).

Transient transfection assays in MCF-7 cells showed that the Sp1 site was involved in the E2 responsiveness of a TFF1 promoter–reporter construct (Sun et al., 2005). It should be noted that a promoter in an episomal promoter–reporter construct may not be assembled as chromatin. In ChIP assays we demonstrated that Sp1 and Sp3 were not associated with the episomal TFF1 promoter reporter when the Sp site was mutated, while the episomal wild type TFF1 promoter preferentially bound Sp1 rather than Sp3 in E2 treated cells. This is in marked contrast with our results showing the loading of Sp3 more so than Sp1 onto the endogenous promoter in E2 treated cells. This result suggests that the chromatin context is an important factor in the differential loading of Sp1 and Sp3. Our analyses of the episomal TFF1 promoter illustrate the limitations of the transient transfection assay in deciphering the function of transcription factor binding sites in promoters.

To study E2 mediated changes in factor association with the promoter in situ, breast cancer cells were cultured under E2 deplete conditions. The ChIP assay followed factor binding on the promoter for 30, 60, and 120 min following E2 addition. In the absence of E2, Sp1, Sp3, and low levels of acetylated H3 and H4 were associated with the native promoter. Following E2 addition, levels of ERα and acetylated H3 and H4 bound to the native promoter increased. There was clearance of Sp1, but not of Sp3, from the promoter. At 90 min post-E2, ERα and Sp3 cleared from the promoter, providing evidence for the cycling of ERα and Sp3 on the TFF1 promoter (Sun et al., 2005).

Under E2 deplete conditions general transcription factors (e.g. TFIID) and RNA polymerase II (RNAP II) remain bound to the TFF1 promoter (Reid et al., 2003). To achieve promoter clearance and synchronize the promoters, Gannon and colleagues incubated cells for 2 h with the transcriptional inhibitor, α-amanitin, resulting in the displacement of ERα, the general transcription factors and RNAP II from the TFF1 promoter (Metivier et al., 2003). Following removal of the inhibitor either ethanol or estradiol was added to the MCF-7 cells, and the association of ERα, coactivators, corepressors and chromatin remodeling complexes was followed with time using the ChIP assay. Cycling of ERα on the TFF1 promoter in the presence and absence of estradiol was observed. However, the frequency of cycling and level of association of ERα on the TFF1 promoter were different when ligand was present (20 min without ligand and 45 min with ligand) (Reid et al., 2003). Immediately following the addition of ligand, a nonproductive association of ERα was found with RNAP II not associating with the promoter until the second cycle of ERα binding, which was longer and at a higher level than the first cycle. We repeated this study and observed ERα cycling on the TFF1 promoter with different frequencies with and without E2. However, we found that the first cycle of ERα loading on the TFF1 promoter was transcriptionally productive as indicated by the presence of RNAP II (Lin Li and James Davie, unpublished observations). A cautionary note, however, with these studies is that the use of transcriptional inhibitors will rapidly result in the disorganization of nuclear components (Eskiw et al., 2004, Mintz and Spector, 2000). It is not known how long it will take for the nuclear organization to be corrected following the removal of transcription inhibitors.

The ligand bound ERα recruits coactivators with HAT activity to the TFF1 promoter (Chen et al., 1999, Shang et al., 2000). Our ChIP assays demonstrated that HDAC1 and HDAC2 are associated with the TFF1 promoter when Sp1 and Sp3, but not ERα, are loaded onto the promoter in MCF-7 cells cultured without estradiol. As the phosphorylated form of HDAC2 is preferentially cross-linked to DNA by formaldehyde, we believe that it is phosphorylated HDAC2 that is bound to the TFF1 promoter. Following the addition of estradiol, the level of HDAC1 and HDAC2 associated with the promoter declined. Thus, our results suggest that Sp1/Sp3 recruits HDAC1 and phosphorylated HDAC2 to the TFF1 promoter.

As a low level of HATs (CBP, p300) are associated with the TFF1 promoter in cells grown under estrogen deplete conditions (Chen et al., 1999, Shang et al., 2000), we reasoned that the promoter would be engaged in dynamic histone acetylation. By inhibiting HDACs with butyrate or trichostatin A (TSA), ChIP assays showed that acetylation of H3 and H4 at the promoter and at exon 2 and 3 was increased. Similar to the results of Thomson et al. (2001) with the uninduced c-fos gene, the TFF1 gene is engaged in dynamic acetylation before induction. Interestingly, we found that the HDAC inhibitors had a differential affect on HDAC1 association with the promoter versus coding region. HDAC inhibitors resulted in the partial dissociation of HDAC1 from the promoter, but at the coding region there was either no change or an increased association. Other investigators have noted that exposure of cells to TSA results in displacement of HDAC1 from a promoter (Ghoshal et al., 2002, He and Margolis, 2002, Mishra et al., 2001).

To address the question whether Sp1 and Sp3 co-bind on a TFF1 promoter, we did a re-ChIP assay. Our results show that a TFF1 promoter binds either Sp1 or Sp3 but not both (He et al., 2005). Also in re-ChIP assays we demonstrated that an Sp1 and ERα or Sp3 and ERα load onto a TFF1 promoter, with co-occupation of Sp3 and ERα on the same TFF1 promoter increasing following addition of estradiol (Fig. 3). In a single cell, it is possible that Sp1 is on the TFF1 promoter of one allele, while Sp3 is on the TFF1 promoter on the other, but this could vary from cell to cell. ChIP assays will present an average of these interactions. Pending the Sp3 isoform and its modification state associated with a TFF1 promoter, the transcriptional activity of an Sp3 charged promoter may be quite different from that of an Sp1 occupied promoter.

In addition to E2, TPA will induce the expression of the TFF1 gene in ER+ human breast cancer cells (Espino et al., 2006). TPA addition to MCF-7 cells cultured under estrogen-free serum starved conditions results in the phosphorylation of ERKs and S10 H3 and TFF1 gene expression. TPA-, but not E2-, induced TFF1 expression was diminished with the MEK inhibitor UO126 and the mitogen and stress activated protein kinase (MSK, EC number 2.7.1.37) inhibitor H89. Our transient transfection studies demonstrated that the downstream AP-1 is required for TPA-stimulation of the TFF1 promoter. In ChIP assays we found that following 30 min of TPA-stimulation, there was an increased association of c-Jun, MSK1, acetylated H3 and H3 phosphorylated at S10 with the TFF1 promoter (Espino et al., 2006). MSK1 is an H3 kinase that phosphorylates H3 at S10 or S28 (Drobic et al., 2004, Dunn and Davie, 2005). These results show that E2 and TPA elicit transcriptional activation of the TFF1 gene via alternative routes: E2-induced expression occurs through recruitment of ERα, whereas TPA-induced expression requires AP-1 recruitment, leading to MSK1 loading and H3 phosphorylation on TFF1 promoter. Similar to our results with TFF1, induction of the Hsp70 gene by heat shock or arsenite operates through different pathways and results in different histone modifications: H4 acetylation in the case of heat shock induction and H4 acetylation and H3 phosphorylation for arsenite (Thomson et al., 2004).

The ChIP assay in which formaldehyde is used to cross-link proteins to DNA (the X-ChIP method) is routinely used to determine which transcription factor, cofactor or chromatin remodeling factor is associated with the gene of interest. Our results demonstrated that the X-ChIP assay using anti-HDAC2 antibodies would immunoprecipitate primarily DNA fragments associated with the phosphorylated form of HDAC2. Although the unmodified form of HDAC2 is more abundant than the modified form, the unmodified form of HDAC2 cross-links poorly to DNA with formaldehyde (Sun et al., 2002). This is not a novel observation. Solomon and Varshavsky (1985) reported that, while the cross-linking of histones and nucleosomal DNA by formaldehyde was successful, the cross-linking of DNA to the DNA binding α protein and lac repressor was a failure, demonstrating that the cross-linking efficiency of formaldehyde was unpredictable. It is advisable to determine which DNA cross-linked protein your antibody is recognizing in the X-ChIP assay. In the analyses of histone modifications it must be kept in mind that the histone modifications are dynamic. Thus the results of the ChIP assay for a specific promoter will depend upon the steady state of the modification at the time of cross-linking with formaldehyde (Clayton et al., 2006).

We have analyzed the chromosomes of cancer cells in culture by spectral karyotyping (SKY). Our studies have revealed that breast cancer cells (MCF-7) are aneuploid (He et al., 2007). We were surprised to learn that the MCF-7 cell's karyotype was quite variable, providing evidence that each cell in culture had a different chromosome composition. Thus the copy number of any particular chromosome may vary from cell to cell. For example, chromosome 21, which harbors the TFF1 gene (21q22.3), varied from 1 to 5 copies in MCF-7 cells. It is unclear whether all the TFF1 genes in the MCF-7 cells with multiple copies of chromosome 21 are estrogen responsive. Realizing that genomic instability is rampant in these cells, this presents issues with procedures such as the ChIP assay and chromosome genomic hybridization method as both will average the events occurring in the cell population. Clearly, the analyses of transcription factor dynamic loading onto estrogen responsive genes, for example, are being done in a background of genomic instability.

Section snippets

Summary

The TFF1 gene provides an example in which there are several routes involving different transcription factors, chromatin remodeling complexes and histone modifications in response to different agents to induce TFF1 gene expression in estrogen receptor α positive breast cancer cells (e.g. MCF-7 cells). A TFF1 promoter binds either Sp1 or Sp3 but not both factors. In MCF-7 cells Sp1 and Sp3 are located in non-overlapping foci which are associated with the nuclear matrix. In response to estrogens,

Acknowledgments

This work was supported by Canadian Institute of Health Research (MOP-9186), CancerCare Manitoba Foundation, Inc., National Cancer Institute of Canada (funds from the Canadian Cancer Society), Canada Research Chair to JRD, a studentship from National Cancer Institute of Canada (funds from the Terry Fox Foundation) to BD, a CIHR Canada Graduate Scholarships Doctoral Award to PSE, a US Army Medical and Material Command Breast Cancer Research Program (W81XWH-05-1-0284) studentship to LL, and a

First page preview

First page preview
Click to open first page preview

References (123)

  • S. Cawley et al.

    Unbiased mapping of transcription factor binding sites along human chromosomes 21 and 22 points to widespread regulation of noncoding RNAs

    Cell

    (2004)
  • H. Chen et al.

    Regulation of hormone-induced histone hyperacetylation and gene activation via acetylation of an acetylase

    Cell

    (1999)
  • A.L. Clayton et al.

    Enhanced histone acetylation and transcription: a dynamic perspective

    Mol Cell

    (2006)
  • C.H. Eskiw et al.

    Chromatin contributes to structural integrity of promyelocytic leukemia bodies through a SUMO-1-independent mechanism

    J Biol Chem

    (2004)
  • A.L. Gartel et al.

    Mechanisms of c-myc-mediated transcriptional repression of growth arrest genes

    Exp Cell Res

    (2003)
  • C.M. Grozinger et al.

    Deacetylase enzymes: biological functions and the use of small-molecule inhibitors

    Chem Biol

    (2002)
  • M.J. Hendzel et al.

    Histone deacetylase is a component of the internal nuclear matrix

    J Biol Chem

    (1991)
  • A. Jamai et al.

    Continuous histone H2B and transcription-dependent histone H3 exchange in yeast cells outside of replication

    Mol Cell

    (2007)
  • P. Karwowska-Desaulniers et al.

    Histone deacetylase 1 phosphorylation at S421 and S423 is constitutive in vivo, but dispensable in vitro

    Biochem Biophys Res Commun

    (2007)
  • Y. Kitadai et al.

    The level of a transcription factor Sp1 is correlated with the expression of EGF receptor in human gastric carcinomas

    Biochem Biophys Res Commun

    (1992)
  • L.E. Lerner et al.

    Sp4 is expressed in retinal neurons, activates transcription of photoreceptor-specific genes, and synergizes with Crx

    J Biol Chem

    (2005)
  • B. Li et al.

    Nucleosome binding by the constitutive transcription factor Sp1

    J Biol Chem

    (1994)
  • A. Loyola et al.

    PTMs on H3 variants before chromatin assembly potentiate their final epigenetic state

    Mol Cell

    (2006)
  • K. Luger et al.

    The histone tails of the nucleosome

    Curr Opin Genet Dev

    (1998)
  • J. Mellor

    Dynamic nucleosomes and gene transcription

    Trends Genet

    (2006)
  • R. Metivier et al.

    Estrogen receptor-alpha directs ordered, cyclical, and combinatorial recruitment of cofactors on a natural target promoter

    Cell

    (2003)
  • P.J. Mintz et al.

    Compartmentalization of RNA processing factors within nuclear speckles

    J Struct Biol

    (2000)
  • S.K. Mishra et al.

    Dynamic chromatin remodeling on the HER2 promoter in human breast cancer cells

    FEBS Lett

    (2001)
  • C.L. Peterson et al.

    Histones and histone modifications

    Curr Biol

    (2004)
  • M.K. Pflum et al.

    Histone deacetylase 1 phosphorylation promotes enzymatic activity and complex formation

    J Biol Chem

    (2001)
  • G. Reid et al.

    Cyclic, proteasome-mediated turnover of unliganded and liganded ERalpha on responsive promoters is an integral feature of estrogen signaling

    Mol Cell

    (2003)
  • A. Sapetschnig et al.

    Complexity of translationally controlled transcription factor Sp3 isoform expression

    J Biol Chem

    (2004)
  • G.F. Sewack et al.

    Nucleosome positioning and transcription-associated chromatin alterations on the human estrogen-responsive pS2 promoter

    J Biol Chem

    (1997)
  • Y. Shang et al.

    Cofactor dynamics and sufficiency in estrogen receptor-regulated transcription

    Cell

    (2000)
  • V.A. Spencer et al.

    Role of covalent modifications of histones in regulating gene expression

    Gene

    (1999)
  • V.A. Spencer et al.

    Dynamically acetylated histones association with transcriptionally active and competent genes in the avian adult b-globin gene domain

    J Biol Chem

    (2001)
  • J. Ausio

    Histone variants – the structure behind the function

    Brief Funct Genomic Proteomic

    (2006)
  • J. Ausio et al.

    The many tales of a tail: carboxyl-terminal tail heterogeneity specializes histone H2A variants for defined chromatin function

    Biochemistry

    (2002)
  • J. Ausio et al.

    Histone variants and histone modifications: a structural perspective

    Biochem Cell Biol

    (2001)
  • A. Calestagne-Morelli et al.

    Long-range histone acetylation: biological significance, structural implications, and mechanisms

    Biochem Cell Biol

    (2006)
  • E. Chiefari et al.

    Increased expression of AP2 and Sp1 transcription factors in human thyroid tumors: a role in NIS expression regulation?

    BMC Cancer

    (2002)
  • B.F. Clem et al.

    Association of the mSin3A–histone deacetylase 1/2 corepressor complex with the mouse steroidogenic acute regulatory protein gene

    Mol Endocrinol

    (2006)
  • D.S. Coffey

    Nuclear matrix proteins as proteomic markers of preneoplastic and cancer lesions

    Clin Cancer Res

    (2002)
  • M.S. Cosgrove

    Histone proteomics and the epigenetic regulation of nucleosome mobility

    Expert Rev Proteomics

    (2007)
  • M.S. Cosgrove et al.

    Regulated nucleosome mobility and the histone code

    Nat Struct Mol Biol

    (2004)
  • M.S. Cosgrove et al.

    How does the histone code work?

    Biochem Cell Biol

    (2005)
  • J.R. Davie et al.

    Control of chromatin remodeling

    Crit Rev Eukaryot Gene Expr

    (2000)
  • G.P. Delcuve et al.

    Chromatin structure of erythroid-specific genes of immature and mature chicken erythrocytes

    Biochem J

    (1989)
  • A.J. De Ruijter et al.

    Histone deacetylases (HDACs): characterization of the classical HDAC family

    Biochem J

    (2003)
  • D.S. Dimitrova et al.

    The spatio-temporal organization of DNA replication sites is identical in primary, immortalized and transformed mammalian cells

    J Cell Sci

    (2002)
  • Cited by (69)

    View all citing articles on Scopus
    View full text