Modulation of Steroid Hormone Receptor Activity

Abstract

Classical steroid hormones (SHs) – estrogens, androgens, progestins, glucocorticoids and mineralocorticoids – play critical roles in the regulation of reproduction, metabolism and cancer. SHs act via their cognate steroid hormone receptors (SHRs) in multiple target tissues throughout the body, exerting their physiological effects through nuclear receptor (NR)-mediated gene transcription. Since SHRs are the mediators of steroid hormone signalling in cells, regulation of their expression and function is critical for appropriate physiological responses to SHs. Cells regulate SHRs by determining the cellular concentration of SHR proteins in the cell and by tightly regulating their activity through post-translational modifications and interactions with coactivator protein complexes. In this chapter we will examine each of these regulatory mechanisms and assess their functional impact on the activity of SHRs.

Introduction

Steroid hormones (SHs) are derivatives of cholesterol that are biosynthesized through a number of biochemical reactions; a flow chart describing their biosynthesis is presented in Fig. 1. In essence, synthesis of SH involves the conversion of cholesterol to pregnolone that is subsequently converted to 17-hydroxypregnolone, 17-hydroxyprogesterone and progesterone. 17-Hydroxypregnolone can be converted into testosterone, a precursor for estrogen, while 17-hydroxyprogesterone can be converted into cortisol. Progesterone is itself a precursor for aldosterone. The biosynthesis of SHs primarily takes place in the adrenal cortex and in the gonads. Adrenocortical steroids include glucocorticoids such as cortisol and mineralocorticoids such as aldosterone. In general, glucocorticoids regulate carbohydrate, lipid and protein metabolism and the responses to stress and inflammation. Mineralocorticoids are involved in the regulation of electrolyte homeostasis and extracellular and intracellular fluid volume. Gonadal sex steroids are produced in the testes in males and in ovaries in females. Both testes and ovaries produce androgens (testosterone) and estrogens (17β-estradiol). Progestins (progesterone) are produced in the ovaries and in the placenta. Like adrenal steroids, gonadal SHs affect a broad array of molecular and physiological processes in cells and tissues throughout the body. Gonadal hormones are involved in the regulation of an organism’s development, sexual maturation and differentiation, behavior and the homeostasis of both reproductive and non-reproductive tissues. Finally, the consequences of reduced levels after menopause or andropause or their misregulation can lead to numerous pathological conditions such as osteoporosis and depression (Guyton and Hall, 2006, Voet and Voet, 2004).

As derivatives of cholesterol, SHs are hydrophobic and are transported in the blood via carrier proteins such as transcortin and albumin. They freely diffuse through the plasma membrane and interact with their cognate receptors in target tissues. The general mechanism of SH action was elucidated in 1972 by O’Malley and colleagues (Means et al., 1972, O’Malley and Means, 1974, Tsai et al., 1975), who were studying the function of estrogen and progesterone and their cognate SHRs in the chicken oviduct and in vitro. Their work laid a foundation for further research on SH function, and eventually to the discovery of the coactivators. These pioneering studies culminated in the present model of molecular events that define the action of the steroid hormones at the cellular level (Fig. 2): (1) steroid hormones bind to steroid hormone receptors located in the cytoplasm or nucleus of the cell; (2) binding of hormone to receptor causes receptor dissociation from chaperone proteins and receptor dimerization; (3) hormone-bound receptors translocate to the nucleus and interact with DNA; (4) in the DNA, hormone-bound receptors cause dissociation of inhibitory repressor molecules and the recruitment of steroid receptor coactivators; (5) receptor–coactivator complexes decondense and open the chromatin by histone modification and displacement; (6) receptor–coactivator complexes recruit general transcription factors (GTFs) and RNA polymerase II and (7) receptor–coactivator complexes maintain an open chromatin state and promote transcriptional elongation, mRNA splicing and eventually degradation of the activated transcription factors at the gene regulation site.

The existence of protein receptors for SHs was first elucidated in the early 1960s by the pioneering work of Jensen and Gorski (Gorski and Nicolette, 1963, Jacobson and Jensen, 1962, Jensen, 1962, Jensen and DeSombre, 1973, Toft and Gorski, 1966), who used tritium-labelled estradiol to isolate and characterize the subcellular and molecular constituents that bind estrogen in the rat uterus. The development of recombinant DNA technologies in the 1970s and 1980s led to the cloning of SHRs and determination of their primary amino acid sequences. The first nuclear hormone receptor to be cloned was the glucocorticoid receptor (GR) in 1985 (Evans, 2005, Hollenberg et al., 1985, Miesfeld et al., 1986). This was followed by cloning of estrogen receptor-α (ERα) (Greene et al., 1986, Green et al., 1986, Miesfeld et al., 1984) and other steroid receptors soon followed, leading to the establishment of the nuclear receptor (NR) superfamily of transcription factors and the realization that they share several major structural and functional features (Evans, 1988a, Evans, 1988b, Green and Chambon, 1988, Gustafsson et al., 1986, Miesfeld et al., 1986). Like other members of the nuclear hormone receptor superfamily, SHRs are functionally composed of three critical modular domains: a hormone-independent activation function 1 (AF1) domain, a DNA-binding domain (DBD) and a hormone-dependent activation function 2 (AF2) domain that is activated allosterically upon ligand binding (Mangelsdorf et al., 1995). Here, we will focus on ERα as a prototypical model for SHR structure and discuss the relationship between its structural features and its biological functions (Fig. 3).

ERα consists of 595 amino acid residues with a molecular mass of 66 kDa (Bourguet et al., 2000, Evans, 1988b, Klinge, 2000, Nagy and Schwabe, 2004, Pike, 2006). It is composed of six domains that comprise the AF1, DBD, AF2 and other portions of the receptor (Fig. 3) (Bourguet et al., 2000, Evans, 1988b, Klinge, 2000, Nagy and Schwabe, 2004). The A and B domains are contained within the receptor’s AF1 and are implicated in hormone-independent transcriptional activation of the receptor. The C domain represents the DBD of the molecule and is composed of two zinc-finger motifs that are responsible for DNA sequence-specific binding to estrogen response elements (EREs). The D domain or the hinge region is a 39 amino acid long linker between the DNA and ligand-binding regions (LBDs) of ERα. Functionally, it contains a nuclear localization signal (NLS) and is implicated in interactions with some coregulator molecules. The E domain of ERα is responsible for ligand binding and doubles as the ligand-activated AF2 domain. The LBD is composed of 12 alpha helixes, five of which (helixes 3, 6, 8, 11 and 12) form a hydrophobic ligand-binding cleft (Bourguet et al., 2000, Evans, 1988b, Klinge, 2000, Nagy and Schwabe, 2004, Pike, 2006). Upon binding to E2, this region undergoes a conformational change such that helix 12 is displaced over the opening of the ligand-binding pocket. This change in the position of helix 12 creates a coactivator-binding surface that forms specific interactions with LXXLL helical motifs present in many coactivators (Klinge, 2000, Lonard and O’Malley, 2007, Lonard and O’Malley, 2008a, Lonard and O’Malley, 2008b). Finally, there is the C-terminal F domain whose role in the receptor function is less clear but has been shown to be involved in receptor dimerization (McKenna et al., 1999).

The functional domains contained in ERα also are found with minor variations in all other steroid hormone receptors. The structural and functional conservation between receptors points not only to the similar mechanism of action and shared evolutionary origin, but also to similar modes of receptor regulation.

Given the significance of SHs and SHRs in an organism’s basic physiological and reproductive processes, regulation of their function depends on numerous regulatory systems. It is not surprising then to learn that SHs regulate hundreds of genes in all tissues of the body. SHR activity is regulated on several different levels: through regulation of SHR expression, protein concentration and stability, through SHR post-translational modifications (PTMs) and through interactions of the receptors with coregulators.

Cells regulate the basal expression of SHRs by regulating the transcription of receptor mRNA, its translation and its protein stability. Given that steroid receptors are important and potent transcription factors, the level and the complexity of regulation of their mRNA expression is not surprising. More than seven different promoters have been identified and characterized that are upstream of the ERα gene. The ERα gene (ESR1) is located on chromosome 6, spans 300 kb and is encoded by eight exons (Ponglikitmongkol and Chambon 1988). ERα promoters are designated as A–F, T1 and T2, and span more than 150 kb of genomic DNA (Fig. 4) (Walter et al., 1985). These promoters produce mRNA with different 5′ untranslated regions that are spliced to a common acceptor site located at the +163 position in the transcript. The location and the amount of endocrine hormone receptor expression in the organism are primary determinants of the endocrine hormone tissue specificity and physiological effect of individual SHs. ERα is primarily expressed in the breast, uterus, ovary, prostate, testes, epididymis, bones and brain, and this expression is controlled by the complex structure of the different ERα promoters. For example, the A promoter is utilized primarily in normal breast and in some breast and uterine cancer cell lines. On the other hand, promoter C is active mostly in breast cancer-derived cell lines, but its activity is less pronounced in normal breast tissue (Grandien et al., 1993, 1995). Promoter B function is linked to ERα expression in endometrial cancer-derived cells but not in breast and uterine cancer cell lines (Grandien et al., 1995). Promoter E is liver specific while promoter F regulates ERα gene transcription in osteoblasts, and T1 and T2 promoters are associated with ERα expression in the testes (Flouriot et al., 1998, Kos et al., 2001, Pinzone et al., 2004). In tissues and cell lines that do not express the receptor, these promoters are methylated and silenced by the action of the DNMT1 DNA methylase (Pinzone et al., 2004). Despite the number of ERα promoters present, their individual intrinsic transcriptional potential is low. Containing no classical TATA, CCAAT or GC boxes, they provide robust ERα mRNA expression with spatial-temporal precision through the combined action of binding of transcription factors interacting at the different sites (Kos et al., 2001). Estrogen and selective estrogen receptor modulators (SERMs) are among the major modulators of ERα promoter activity (Castles et al., 1997, Donaghue et al., 1999, Griffin et al., 1998, Treilleux et al., 1997). Other nuclear receptors such as the progesterone receptor (PR), androgen receptor (AR), vitamin D receptor (VDR) and retinoid X receptor (RXR) also can regulate ERα gene promoter activity. Peptide hormones and growth factors regulate ERα promoter either by direct interaction with the DNA or through the activation of different signalling cascades (involving PKA, PKB, PKC and other signalling kinases). The fact that many growth regulatory pathways control the ERα promoter indicates that ERα is a potent mitogen whose expression must be controlled by a variety of signalling growth factors (Pinzone et al., 2004).

From the transcriptional complexity seen at ERα promoter, a regulatory mechanism emerges that is sensitive to the immediate and long-term needs of the organism, a mechanism capable of converting ERα protein levels from a low level of basal constitutive expression to acute or chronic elevations depending upon acute extracellular events and intracellular signalling cascades. This mechanism provides for dynamic control of ERα expression in response to changes in the environment such as stress or the administration of estrogen analogues in anti-cancer therapy and the consumption of phyto-estrogens in the diet.

Estrogen receptor β (ERβ) is located on chromosome 14 and contains eight exons that are transcribed from at least two different promoters termed ON and OK. Similar to that seen for ERα, the two ERβ promoters are responsible for differential tissue expression of the receptor. ERβ expression also is regulated in a developmentally dependent manner. Unlike ERα, the expression of ERβ is regulated by circadian rhythm. Negative circadian regulators, PER and CRY bind an E-box motif in the ERβ promoter region and regulate circadian ERβ mRNA both in synchronized cell culture lines and in mouse tissues (Zhao et al., 2008a). This finding is of physiological relevance considering the tissue distribution of ERβ, which links it to mood disorders and reproductive problems in female shift workers and in the development of some tumours. The ERβ promoter is methylated and aberrant methylation patterns are correlated with a number of disease states including prostate, breast, endometrial and ovarian cancers and atherosclerosis.

Unlike ERα and ERβ that are coded by two different genes on different chromosomes, progesterone receptors A and B isoforms are produced from two distinct promoters of a single PR gene (Kastner et al., 1990) located on chromosome 11q22. PR-A and PR-B are expressed in an estrogen-dependent manner although no identifiable consensus ERE is present in their 5′ upstream region. Furthermore, both isoforms are expressed in a similar temporal and spatial manner and to a similar extent. Methylation of PR-B can lead to an imbalance between the two isoforms, and it is potentially seen as a causative agent in endometriosis and breast cancer.

The androgen receptor is located on the X chromosome at Xq11-1. Its core promoter does not contain classical TATA or CAAT boxes, but it does contain an SP1-binding site at –46 bp and a cAMP response element (CRE) at –508 bp. Consequently, AR is upregulated by cAMP analogues. In addition, NF-κB and TNFα-binding sites are located in a distal part of the AR core promoter. The expression of AR is developmentally regulated and decreases with age. It is also regulated by its ligand; prolonged treatment with dihydroxytestosterone (DHT) decreases the expression level of AR. However, this effect is cell type specific and occurs in LNCaP and T47-D cell lines but not in human hepatocellular carcinoma or osteoblastic cell lines. Finally, methylation of the proximal promoter and first exon has been observed in some instances and it is correlated with the development of androgen-independent prostate cancers (Waltering et al., 2006a).

The human GR gene is located on chromosome 5 q11–q13 and is composed of nine exons. Like AR, the GR promoter does not contain TATA and CCAAT boxes, but does possess an SP1-binding site. The GR promoter region also contains AP1-, YY1-, NF-κB- and GR-binding sites (Breslin et al., 2001, Nobukuni et al., 2002, Yudt and Cidlowski, 2002). Three distinct promoters (A–C) produce mRNAs that differ in their 5′UTR regions. However, these mRNAs are subsequently spliced to a common translational start site in the second exon. It is speculated that the existence of multiple promoters accommodates cell-type specificity of GR expression and the complex regulation of the GR promoter by a number of transcription factors. In addition, a number of GR isoforms have been identified. These isoforms arise from a single gene through different splicing mechanisms and alternate translational initiations; GR isoforms are regulated in a tissue-specific manner and influence cell-specific response to glucocorticoids (Duma et al., 2006, Lu and Cidlowski, 2006).

The mineralocorticoid receptor (MR) gene is located on chromosome 4q31.1-31.2 and is expressed in many tissues, such as kidney, colon, heart, hippocampus, brown adipose tissue and sweat glands. Its 5′ upstream region contains two promoters (α and β), and it has three CAAT and TATA elements distributed from the 5′ untranslated region to the first intron (Listwak et al., 1996). The MR promoter also contains SP1 and CREB sites (Listwak et al., 1996). The level of MR in the brain is controlled at the mRNA level and is regulated by prolonged administration of anti-depressants such as imipramine (Brady et al., 1991) that increase its levels in the hippocampus.

Regulation of steroid receptor mRNA stability and translation occurs primarily through binding of specific mRNA-binding proteins or micro-RNAs (miRNA) to 3′UTR regulatory elements. Two regulatory elements are most prominent, AU-rich elements (AREs) and C-rich elements. AREs containing multiple copies of a 5′-AUUUA-3′ motif destabilize mRNAs, while C-rich elements are recognized and differentially regulated by poly(C)-binding proteins. The miRNAs are a class of regulatory modalities that interact with specific complementary sequences present in the 3′ UTR region of the receptor’s mRNA. All steroid hormone receptors contain an unusually long 3′UTR and a large number of ARE sequences in the 3′UTR, and allow steroid hormones to auto-regulate the mRNA expression levels of their cognate receptors (Ing, 2005).

The ERα mRNA is 4.3 kb long and has a steady-state half-life of approximately five hours (Fig. 4). The mRNA contains an extensive 3′ UTR which is three times as long as its open reading frame. The ERα 3′UTR is known to contain several regulatory elements including 14 putative class I AREs (Green et al., 1986, Keaveney et al., 1993, Kenealy et al., 2000), but its stability can be altered in response to different stimuli (Kenealy et al., 2000). Proteins such as AUFp45 bind ERα mRNA and increase its stability by protecting it from RNases (Ing et al., 2008). Recently, several miRNAs – miR18a, miR22, miR206 and miR221/222 – have been shown to bind and negatively affect ERα mRNA stability and/or translation (Adams et al., 2007, Liu et al., 2009, Pandey and Picard, 2009, Zhao et al., 2008b). Functionally, the expression of miRNAs leads to a decrease in the cellular receptor pool and subsequently to attenuated cellular responses to estrogen stimulation.

Androgen receptor mRNA spans 7 kbs and is extensively regulated post-transcriptionally (Waltering et al., 2006b). Towards the 5′ end of the AR 3′UTR, the RNA-binding protein HuR binds and stabilizes mRNA by interacting with AREs. HuR belongs to the Elav/Hu family of RNA-binding proteins, which in addition to being a regulator of mRNA stability also has a function in shuttling AR mRNA from the nucleus to the cytoplasm. Two proteins bind the C-rich elements in the AR mRNA-CP1 and -CP2; both affect AR mRNA stability and the rate of translation (Wilce et al., 2002). Recently, a new regulator of AR mRNA translation, hnRNP-K, has been identified in prostate cancer cells. The hnRNP-K can bind to both the 5′ and 3′ UTR regions as well as the coding region of the AR mRNA to inhibit its translation (Mukhopadhyay et al., 2009). Overall, androgens and progestins generally increase the stability of AR and PR mRNA in prostate and endometrial cancers, respectively (Ing, 2005).

Human GR mRNA contains numerous ARE elements and is subject to post-transcriptional regulation (Schaaf and Cidlowski, 2002). In addition to its regulation by a protein binding to an ARE element in the GR 3′UTR, GR mRNA is also negatively regulated by two miRNAs – miR18 and miR124a. Interestingly, while miR18 is broadly expressed in many tissues, miR124a is exclusively expressed in the brain, suggesting a probable tissue-specific regulatory requirement for controlling GR mRNA expression in the nervous system (Vreugdenhil et al., 2009). Generally, glucocorticoids tend to decrease the stability of GR mRNA in the kidney, liver and colon cells (Ing, 2005).

The ubiquitin proteasome system (UPS) regulates SHR protein stability. The UPS is a complex biochemical system that regulates the stability of cellular proteins by virtue of a number of enzymes responsible for protein degradation (Hershko, 1996, Hershko and Ciechanover, 1998, Nandi et al., 2006). In the initial reaction of the UPS, the 76 amino acid protein ubiquitin is covalently attached to protein substrates destined for degradation. Ubiquitin attachment is mediated by three classes of enzymes: a sole ubiquitin-activating enzyme (E1) first charges an ubiquitin molecule and transfers it to an E2-ubiquitin conjugating enzyme; ubiquitin is subsequently transferred to the target protein from E2 by the activity of an E3-ubiquitin ligase. The temporal, spatial, contextual and protein specificities of ubiquitin-protein attachments are regulated by several hundred different ubiquitin E3 ligases. Once a protein substrate is polyubiquitinated, it can become a target for degradation by the 26S proteasome complex. After the initial attachment of the first ubiquitin to a target protein, subsequent ubiquitins can be added to this ubiquitin, forming a polyubiquitin chain. Depending on the lysine linkage of branching ubiquitin molecules (K6, K11, K27, K33, K48 and K63), a polyubiquitinated protein can be fated for degradation or other cellular processes. K48-based polyubiquitin linkages lead to proteasome-mediated protein degradation.

The 26S proteasome is composed of the 20S proteasome proteolytic core and the 19S cap that serves to recruit and channel substrate proteins into the 20S core. The reversal of protein ubiquitination can be effected through a class of proteases termed deubiquitinating enzymes (DUBs). Although ubiquitination is primarily thought of as a PTM event associated with protein degradation, monoubiquitin or polyubiquitin linkages other than K48 can affect protein localization, protein trafficking, secretion, nuclear export, ER processing, transcriptional activity and DNA binding (Nandi et al., 2006).

ERα protein turnover is extensively regulated by the UPS. ERα activity and stability are intimately linked so that inhibition of ERα degradation leads to its stabilization, but also to its loss of transcriptional activity. UPS-dependent turnover of the receptor is required for its transcriptional function (Alarid et al., 1999, Khissiin and Leclercq, 1999, Lonard et al., 2000, Metivier et al., 2003, Nawaz et al., 1999a, Reid et al., 2003, Stenoien et al., 2001, Wijayaratne and McDonnell, 2001).

Ubiquitin E3 ligases that regulate unliganded ERα protein stability and activity include CHIP (carboxyl terminus of Hsc70-interacting protein) and breast cancer 1, early onset (BRCA1) (Ballinger et al., 1999, Fan et al., 2005, Tateishi et al., 2004, Zheng et al., 2001). CHIP is part of a molecular chaperone complex that contains Hsp90, Hsp70, Hsp40 and BAG-1 proteins. In accordance with its established chaperone activity, experimental evidence indicates that CHIP functions to recognize and remove misfolded ERα. CHIP interaction with the ERα ligand-binding domain leads to increased ERα ubiquitination and increased ERα turnover. BRCA1 also is shown to interact and mono-ubiquitinate ERα at lysine 302 in the hinge region of the receptor (Zheng et al., 2001). Although this modification has not been conclusively linked to receptor degradation, it is hypothesized to have a functional effect on ERα transcriptional activity.

In its liganded state ERα is modified by E6-associated protein (E6-AP/UBE3A) and MDM2. ERα association with E6-AP is reciprocal to Ca²⁺-dependent calmodulin binding to ERα. E6-AP destabilizes the receptor and leads to increased receptor activity (Li et al., 2006, Nawaz et al., 1999b). MDM2 interaction with ERα also coactivates the receptor (Liu et al., 2000, Saji et al., 2001). Furthermore, the COP9 signalosome (CSN) is involved in the regulation of ERα degradation by regulating the activity of these E3 ligases (Fan et al., 2003).

Recently, deubiquitinating enzymes have been reported to regulate ERα protein stability and activity. The deubiquitinating enzyme OTUB1 has been found to directly bind and deubiquitinate ERα in cells and in vitro. ERα deubiquitination stabilizes the receptor in the chromatin fraction of Ishikawa endometrial cancer cells and represses ERα transcriptional activity (Stanisic et al., 2009). The presence of both ubiquitinating and deubiquitinating enzymes on ERα suggests that there is a dynamic regulation of ERα ubiquitin status during transcription. Stability of the receptor is coupled with the receptor’s cycling on the promoter of ERα target genes and disruption of either receptor’s stability or its cycling leads to the abolishment of transcription (Reid et al., 2003). As ERα binds to the promoter of target genes, it recruits coactivator complexes and components of the UPS system (ubiquitin E3 ligases, proteasome, deubiquitinating enzymes). In this situation the UPS is hypothesized to ensure that the ordered procession of recruited cofactors proceeds uninterrupted by sequentially removing ‘used up’ coactivator factors, and by eliminating the receptor itself after each successful round of transcription. This process forms a basis for a ‘ubiquitin clock’ present both for the receptor and coactivators, which ensures maximum efficacy of transcription while tightly controlling the amount and usage of available transcriptional constituents (Wu et al., 2007). Experimental data show that if this process is interrupted by the addition of proteasome inhibitors or the removal of cellular ubiquitin (Lonard et al., 2000), transcription is irrevocably stopped by the stabilization and immobilization of the receptor in the nuclear matrix.

Unlike ERα, the degradation of ERβ is not coupled with ERβ activation, but is instead required for the down-regulation of ERβ transcriptional activity. E3 ligase CHIP is charged with the clearing of ERβ and shutting down its signalling in MDA-MB231 breast cancer cells (Tateishi et al., 2006).

Similar to ERα, PR protein is preferentially degraded by the UPS in its liganded state. Within six hours of progesterone treatment, 95% of liganded PR is degraded; in contrast, the half-life of unliganded PR is 21 hours (Ismail and Nawaz, 2005, Nardulli and Katzellenbogen, 1988). Proteasomal degradation of PR is preceded by PR phosphorylation at serine-294 by mitogen-activated protein kinase (Lange et al., 2000). Also, the yeast E3-ubiquitin ligase RSP5 and its human orthologue hRPF1 positively affect PR-mediated transcription (Imhof and McDonnell, 1996). UbcH7, an E2-conjugating enzyme, and E6-AP E3 ligase stimulate PR transcription in a coordinated manner (Faus and Haendler, 2006b, Imhof and McDonnell, 1996, Verma et al., 2004). The ubiquitin E3 ligase BRCA-1 also binds PR; however, the BRCA-1 interaction with PR negatively regulates PR transcriptional activity (Eakin et al., 2007, Ma et al., 2006). A CUE domain containing 2 (CUEDC2) binds PR and enhances ubiquitination of PR on lysine 388 and causes hormone-dependent degradation of PR. Consequently, loss of CUEDC2 leads to a decrease in PR transcriptional activity in breast cancer cells and progesterone-dependent proliferation in cancer cells (Zhang et al., 2007).

AR transcriptional activity is negatively regulated by UPS-mediated receptor turnover. The E3 ligase MDM2 ubiquitinates and destabilizes AR following AR phosphorylation by Akt. These events lead to a decrease in AR activity. The CHIP E3 ligase interacts with AR through its N-terminal domain and limits its function (He et al., 2004). Proteins that stabilize AR, such as tumour susceptibility gene product TSG101, enhance AR monoubiquitination and increase AR transcriptional activity (Burgdorf et al., 2004).

Both GR and MR also are subject to proteasomal degradation. In the case of GR, the E3 ligase hRPF1 binds and coactivates GR transcription (Imhof and McDonnell, 1996). However, a mutation in the GR degron region stabilizes the receptor and causes its transcriptional enhancement. In addition, treatment of cells with the proteasomal inhibitor MG132 both stabilizes GR and leads to increased GR-mediated transcription (Wallace and Cidlowski, 2001). Although MR contains degrons in its protein sequence and is degraded by UPS, the relationship between MR and the proteasome has not been established.

ERα is phosphorylated both in the presence and absence of ligand (Fig. 5). In the unliganded state, ERα is phosphorylated in response to secondary messenger pathways. ERα is phosphorylated on both serine and tyrosine residues, but there is only one tyrosine phosphorylation site recognized in ERα and phosphorylation on this site is not hormone regulated (Lannigan, 2003). In general, phosphorylation of ERα by different kinases changes ERα affinity for E2, 4-hydroxytamoxifen (4HT), EREs in the DNA, and for its interactions with coactivators (Lannigan, 2003, Likhite et al., 2006). ERα is phosphorylated in a hormone-dependent manner on serines S104, S106, S118 and 167 (Likhite et al., 2006). Phosphorylation of ERα on serine S118 is the most prominent phospho-modification of ERα both in vitro and in vivo (Chen et al., 2002, Lannigan, 2003, Murphy et al., 2006). Phosphorylation of S118 is effected by two different independent cellular pathways, and in both cases it conveys activity to the receptor. Hormone binding to the ERα induces rapid S118 phosphorylation by cyclin-dependent protein kinase Cdk7. Also, growth factor signals activate the MAPK signalling pathway and induce hormone-independent phosphorylation of S118 by ERK1/2 kinase. This hormone-independent phosphorylation of S118 mechanistically underscores the hormone-independent activation of ERα via its AF1, and it is implicated as the possible mechanism for the agonistic action of 4HT. In addition to S118, two neighbouring serines, S104 and S106 also have been shown to be phosphorylated in vitro by the MAPK pathway in an E2-dependent manner by the cyclin A/CDK2-dependent kinase (Lannigan, 2003, Thomas et al., 2008). S167 is phosphorylated in a ligand-independent manner by p90 ribosomal S6 kinase RSK, and AKT, in response to the activation of the PI3 kinase pathway (Lannigan, 2003). In vitro, S167 can be phosphorylated by casein kinase B (Lannigan, 2003). In a study of ERα positive breast cancer patients, phosphorylation of ERα at Ser167 was identified as a marker of longer disease-free status and overall survival (Jiang et al., 2007).

The ERα hinge region is extensively modified by several PTMs, and is the site of dynamic interplay between acetylation, methylation, SUMOylation and ubiquitination (Fig. 5). This centre of post-translational processing presumably reflects the functional importance of this region for fine-tuning the interactions of ERα with DNA and coactivators. ERα is acetylated by p300 in a hormone-dependent manner on lysines K266 and K268 (Kim et al., 2006). This acetylation increases ERα estrogen-dependent interaction with DNA and coactivates ERα-mediated transcription in transient reporter assays (Kim et al., 2006, Wang et al., 2001a). Lysines K302 and K303 in the hinge region of ERα also have been recognized as potential p300 acetylation sites since experimentally created mutations of these sites reduce ERα transactivation while at the same time lowering the acetylation status of the receptor (Cui et al., 2004, Wang et al., 2001a). It has been reported that phosphorylation of serine S305 by PKA prevents acetylation of lysine K303 and attenuates ERα transcriptional activity (Cui et al., 2004).

Although the receptor recruits methyltransferases for the purpose of chromatin modification, it has been shown that methyltransferases also can methylate the receptor. Lysine 302 is reported to be methylated in vivo and in vitro by the SET7 lysine methyltransferase. SET7 is recruited by receptors to methylate histone H3K4, but it can act to methylate transcription factors as well. Methylation of ERα K302 stabilizes ERα, indicating that K302 is a potential site of ERα ubiquitination. In addition, loss of SET7 attenuates E2-dependent activation of ERα-regulated genes in MCF-7 cells (Subramanian et al., 2008). Therefore, it is possible that the methylation of the hinge region of ERα stabilizes the receptor and increases its transcriptional activity.

Although ERα does not contain a consensus SUMO conjugation site, ERα has been shown to be SUMOylated in cells and in vitro (Sentis et al., 2005). SUMOylation of ERα appears to be strictly hormone dependent and it occurs in the hinge region of the molecule, affecting K299, K302, K303 lysine triad. SUMOylation of the hinge region positively affects estrogen-dependent ERα transcriptional activity in transient reporter assays. Attachment of the SUMO-1 modifier to the ERα hinge region is mediated by protein inhibitor of activated signal transducer and activator of transcription PIAS1 and PIAS3 E3 SUMO ligases. However, the same report shows that PIAS1 and PIAS3, as well as Ubc9, can coactivate ERα transcription independently from their SUMO-1 conjugation activity (Sentis et al., 2005).

AR is phosphorylated on at least nine serine residues throughout the molecule. Serine S213 was shown to be activated by PI3K/Akt in response to treatment with the synthetic androgen R1881. Another phosphorylation site, Ser 650, is phosphorylated by MAPK kinase (MKK), c-Jun N-terminal kinase (JNK) or MKK6–p38 signalling pathways. Phosphorylation of Ser 650, which is located adjacent to the nuclear export signal, prevents the export of AR from the nucleus and thereby regulates AR activity (Gioeli et al., 2006).

AR activity is substantially linked with its acetylation status. A number of studies have shown that acetylation of AR causes coactivator recruitment and receptor coactivation. AR is acetylated in vitro and in vivo by p300 and cAMP-response element-binding protein (CBP) on lysines K632 and K633 in the vicinity of its DBD. This modification is required for the receptor’s coactivation since the mutation in the conserved acetylation site diminished the receptor’s response to hormone (Fu et al., 2000). Acetylation of the same conserved site was also observed in the study of bombesin-induced AR transactivation. Bombesin and DHT act via Src and PKCδ signalling pathways to activate p300 acetyltransferase, leading to the activation of AR activity in prostate cancer cells (Gong et al., 2006). Furthermore, acetylation of AR leads to the recruitment of coactivator complexes and the dismissal of corepressors, leading to increased residency of the receptor on promoters and enhanced proliferation of prostate cancer cells (Fu et al., 2003). In addition, other studies have shown that AR is also acetylated by coactivator protein Tip60 which plays a key role in prostate cancer development, and that loss of acetylation of the receptor is linked with defects in trafficking, misfolding and aggregation similar to those seen in polyglutamine aggregates (Gaughan et al., 2002, Halkidou et al.,2003, Korkmaz et al., 2004, Thomas et al., 2004).

AR SUMOylation is hormone dependent, and it is generally repressive with respect to the transcriptional function of the receptor. AR is SUMOylated on lysines K386 and K520 by SUMO E3 ligases PIAS1 and PIASxα (Nishida and Yasuda, 2002, Poukka et al., 2000). These events cause transcriptional repression of AR. On the other hand PIAS-like SUMO ligase hZimp10 promotes SUMO-dependent activation of the receptor in an androgen-specific manner (Poukka et al., 1999, Sharma et al., 2003).

PR activity is regulated by several protein kinases including MAPKs, CDK2 and casein kinase II. PR phosphorylation on Ser 294 by MAPK is an activating event that is linked to proteasome-mediated degradation of the receptor protein (Lange et al., 2000, Qiu and Lange, 2003). In addition, PR phosphorylation was found to be important for controlling the receptor’s subcellular localization.

Although PR is likely regulated by protein acetylation, no acetyl transferases have been identified to date to interact with and acetylate the receptor. A SUMOylation site has been mapped in the PR molecule on lysine K388 in the N-terminus. Ligand binding is required for the SUMOylation of this site, and it leads to the auto-inhibition and transrepression of PR transcriptional activity (Abdel-Hafiz et al., 2002). Further studies have shown that SUMOylation also leads to decreased ligand affinity for the receptor and slowed receptor down-regulation. However over-expression of SUMO-1 leads to increased PR activity, presumably through SUMO-dependent enhancement of activity of coactivators such as steroid receptor coactivator-1 (SRC-1) (Abdel-Hafiz et al., 2009).

GR is subject to complex combinatorial phosphorylation interplay by MAPKs, CDKs and casein kinase II. Phosphorylation of the serine residues (S113, S141, S203, S211 and S226) in the N-terminal domain of the molecule has been linked to the receptor’s function and disruption of these phosphorylations led to a decrease in transcriptional activity in some experimental systems (Almlof et al., 1995, Krstic et al., 1997). Furthermore, GR subcellular localization has been shown to be dependent on phosphorylation of S203 and S211 (Ismaili and Garabedian, 2004a). Recently, interplay between phosphorylation of S211 and S226 has been implicated in the regulation of GR transcriptional coactivation. It has been determined that GR transcriptional activation is greatest when the relative phosphorylation of S211 exceeds that of S226 (Chen et al., 2008). Also, GSK-3β has been shown to phosphorylate GR on serine 404. This modification caused inhibition of glucocorticoid-dependent NF-κB transrepression and decreased glucocorticoid-dependent cell death of osteoblasts (Galliher-Beckley et al., 2008). In the rat brain, GR is phosphorylated in the nucleus on residue S232 by Cdk5 following chronic stress. In the same system, phosphorylation of serine S246 by JNK is decreased upon phosphorylation of S246 (Adzic et al., 2009). In addition to kinases, a number of phosphatases also have been implicated in the regulation of GR activity. In MCF-7 and T47-D breast cancer cells, estradiol inhibits GR activity by down-regulating active S211 phosphorylated form of the receptor by inducing protein phosphatase 5 (PP5) (Zhang et al., 2009). Other phosphatases such as PP1 and PP2 also play roles in reversing GR phosphorylation and inhibit its function (Faus and Haendler, 2006a, Ismaili and Garabedian, 2004b).

GR is acetylated in a hormone-dependent manner on lysines K494 and K495. Inhibition of this modification as a result of impaired histone deacetylase 2 (HDAC2) activity causes down-regulation of glucocorticoid-dependent NF-κB-mediated gene expression (Ito et al., 2006). GR is also a target for acetylation by the histone acetyl transferase CLOCK and its heterodimer partner BMAL1. These proteins are self-oscillating transcription factors involved in the regulation of circadian rhythms in the central and peripheral nervous system. CLOCK/BMAL1 has been shown to negatively affect GR-mediated transcription (Nader et al., 2009).

GR is SUMOylated on lysines K277, K293 and K703, and this modification is positively correlated with its transcriptional activity (Le Drean et al., 2002). GR has been found to interact with the SUMO conjugating enzyme Ubc9 through its ligand-binding domain (Cho et al., 2005, Gottlicher et al., 1996).

MR is preferentially phosphorylated on its N-terminal domain in the presence of aldosterone (Faus and Haendler, 2006a). In addition, the activation of PKA has been correlated with increased MR affinity for DNA. MR is SUMOylated on lysines K89, K399, K428, K494 and K953, and the elimination of these sites leads to increased MR transcriptional activity, indicating that SUMOylation has a negative effect on the receptor’s transcription activity. Acetylation of MR has not been reported, despite the presence of a putative acetylation site (Faus and Haendler, 2006a).

Although genome-wide ERα–DNA binding studies have identified several thousand widely dispersed ERα–DNA binding sites, only a handful of these sites have been experimentally shown to regulate the transcription of ERα target genes (Green and Carroll, 2007). These findings together with the finding that forkhead protein binding motifs are enriched in a genome-wide screen of ERα-binding sites suggest the existence of pioneer factors that function to define and license gene promoters for subsequent transcription initiation steps (Cirillo et al., 2002, Green and Carroll, 2007). FoxA1 is thought to initially contact compact chromatin and act to disrupt internucleosomal interactions mediated by H3/H4 histone tetramers (Cirillo et al., 2002). FoxA1 presumably achieves this by mimicking histone molecules (Green and Carroll, 2007). Studies of the cyclin D1 (CCDN1) promoter (Eeckhoute et al., 2006) indicate that FoxA1 relaxes and opens chromatin in the enhancer 2 (enh2) region of the CCDN1 promoter; this event leads to the E2-dependent recruitment of Sp1 and ERα. FoxA1 expression is well correlated with the expression pattern of ERα in luminal breast cancer samples (Green and Carroll, 2007, Thorat et al., 2008). In addition to FoxA1, GATA3 has been shown to play a role as a pioneer factor for ERα (Eeckhoute et al., 2007, Kouros-Mehr et al., 2006). GATA3 expression is recognized as a marker for ERα positive tumours, and its disruption in the GATA3 knockout mouse leads to a phenotype that mimics that of the ERα knockout mouse (Eeckhoute et al., 2007, Mallepell et al., 2006). Importantly, ERα regulates the expression of both FoxA1 and GATA3, while at the same time, these pioneer transcription factors also regulate the expression of ERα (Eeckhoute et al., 2007, Green and Carroll, 2007) forming a feedback regulatory loop. Pioneer factor-mediated chromatin relaxation and opening is thought to lead to hormone-dependent recruitment of ERα to cis regulatory elements in promoter regions.

In subsequent steps, chromatin remodelling coregulator complexes are recruited to the site of impending transcription. There are two major groups of chromatin remodelling coactivator complexes: (1) ATP-dependent chromatin remodelling complexes and (2) enzymes that covalently modify histones by acetylation, methylation, phosphorylation and ubiquitination. ATP-dependent chromatin remodelling complexes such as hSWI–SNF use the energy of ATP hydrolysis to displace nucleosomes along the DNA in a process termed nucleosome sliding, thereby increasing nucleosome accessibility to transcription factors, to change rotational phasing of DNA on the nucleosome and to reduce negative supercoiling of circular chromatin templates (Green and Carroll, 2007, Kassabov et al., 2003). ATP-dependent chromatin remodelling complexes are multi-subunit protein complexes, and ERα is shown to interact in an E2-dependent manner with many of its subunits (Chiba et al., 1994, DiRenzo et al., 2000, Garcia-Pedrero et al., 2006, Green and Carroll, 2007, Ichinose et al., 1997, Kassabov et al., 2003). BRG1 is a component of the hSWI–SNF coactivator complex that interacts with ERα AF2 in an ATP- and E2-dependent manner (Chiba et al., 1994, DiRenzo et al., 2000, Ichinose et al., 1997). BAF57 is another hSWI–SNF subunit that is shown to participate in ERα-mediated gene expression. In cell culture, BAF57 coactivates ERα-mediated transcription. It binds to the ERα hinge region and has been shown to help recruit p160 coactivators (Garcia-Pedrero et al., 2006, Green and Carroll, 2007).

A major addition to the basic model of steroid receptor-mediated transcription came with the observation made using in vitro transcription systems that NR and basic transcription machinery alone are insufficient to induce a strong transcriptional output (Fig. 6) (Lonard and O’Malley, 2007, Lonard and O’Malley, 2008a, Lonard and O’Malley, 2008b, Spelsberg et al., 1972). This realization was followed with the identification of the first steroid receptor-associated protein 160 (ERAP160) (Halachmi et al., 1994). Also, the observation that the addition of two different, non-overlapping steroid receptors into a cell leads to transcriptional interference and squelching suggested that there is a common limiting pool of molecules utilized by different transcription factors during transcription. These observations led to the hypothesis that a group of yet undiscovered transcriptional players is required for robust steroid receptor transcriptional activity. Soon after, Oñate et al. (1995) cloned and identified SRC-1 in a yeast two-hybrid screen using the progesterone receptor LBD as a bait protein; SRC-1 was shown to be a potent activator of transcription for most steroid receptors. In subsequent years many more coactivators and corepressors (>350) were identified with diverse structural and enzymatic capabilities that collectively orchestrate and execute a highly complex sequence of molecular events at the gene promoter that leads to productive transcription (Cavailles et al., 1995, Halachmi et al., 1994, Kamei et al., 1996, Le Douarin et al., 1995, Lee et al., 1995, Lonard and O’Malley, 2007, Lonard and O’Malley, 2008a, Lonard and O’Malley, 2008b, Oñate et al., 1995). There are two types of coactivators – primary coactivators and secondary coactivators. Primary coactivators directly bind to SHRs and usually serve as scaffolds for the recruitment and the exchange of the secondary coactivators. Secondary coactivators do not bind directly to the SHRs but interact indirectly through the primary coactivators, but their enzymatic functions are indispensible for the receptor’s activity. Neither primary nor secondary coactivators bind ligands or DNA but are recruited by the receptors to the chromatin, usually in a hormone-dependent manner.

The p160 coactivators or the steroid receptor coactivators SRC-1, SRC-2 and SRC-3 are the most studied of the primary coactivators; they play critical roles in chromatin remodelling and the assembly of transcription initiation complexes upon the recruitment of ligand-activated receptors to the promoter (Fig. 6). Although they were discovered in the contexts of different receptors, the SRC-family coactivators enhance the transcriptional activity of most steroid receptors and many other transcription factors as well. Although SRC-1 and SRC-3 have intrinsic histone acetyltransferase (HAT) activity, their major function (as well as the function of SRC-2) is to provide scaffolding and to direct the recruitment of other enzymatically active molecules. This function of SRCs is mediated by protein–protein interactions with secondary coactivators via distinct domains in the SRC coactivators. The molecular weight of p160/SRC proteins is ∼160 kDa, and they contain several conserved domains. A key motif contained in SRC proteins structure responsible for their interaction with SHRs is the LXXLL NR box motif. This motif interacts with the hydrophobic cleft formed in the LBD of the receptor upon ligand binding. An important feature of the LXXLL motif is that it binds with different affinities to different SHRs depending upon the amino acid context in which the LXXLL is found. In addition, SRCs can interact with the AF-1 domains of the receptors through their C-terminal regions. SRC interactions with secondary coactivators are mediated by three transcriptional activation domains (ADs). These domains contain intrinsic activating capacity as assessed by the transcriptional ability of isolated AD fused to the Gal4 DBD. AD1 interacts with and recruits acetyltransferases such as p300, CBP and PCAF. AD2 recruits histone methyltransferases CARM1 and PRMT1. Unlike AD1 and AD2 that are located in the C-terminal domain of the SRC molecule, AD3 is located on the N-terminus and contains basic helix-loop-helix PAS domain (bHLH-PAS). This domain interacts with a plethora of coactivators such as ANCO1, BAF57, CoCoA, Flil, GAC63 and others.

The secondary coactivators associate with the steroid receptors mainly through interactions with the p160/SRCs; the HATs p300 and CBP serve as prime examples of secondary coactivators (Fig. 6). Although they can directly interact with the receptors, they primarily function as secondary coactivators in partnership with SRC family coactivators. Once recruited to the promoter, p300 and CBP HAT activities enable chromatin decondensation and the opening of the chromatin. This function results in the neutralization of the positively charged histone tails with negatively charged acetyl groups. Neutralized histone tails lose their affinity for negatively charged DNA, resulting in greater access for other factors to DNA. By acetylating histones, p300 and CBP promote the recruitment of RNA polymerase II and the basic transcription factors TBP and TFIIB.

Protein arginine methyltransferases CARM1 and PRMT1 also are secondary coactivators that impact SHR transcription by methylating histone tails. CARM1 methylates histone H3 on arginine residues R2, R17 and R26 in response to hormone binding to the steroid receptor. In addition to its histone-methylating function, CARM1 is also responsible for methylating coactivators such as SRC-3, p300 and CBP. PRMT1 methylates R3 on histone H4 in response to hormone stimulation of steroid receptors. Methylation of R3 is just one step in a string of sequential methylation, acetylation and ubiquitination events that take place on histone tails, ultimately leading to the recruitment of RNA PolII. Hormone-dependent binding of CoCoA to SRC coactivator is mediated by the AD3 domain of the SRC coactivators. CoCoA is a secondary coactivator that exerts its stimulatory function through protein–protein interactions with mediator complex and basal transcription factors such as TBP and TAF9. CoCoA acts synergistically with p300 and CARM1 to promote hormone-dependent, steroid receptor-mediated transcription.

When steroid hormone receptors recruit coactivator complexes to chromatin, they orchestrate a highly complex and ordered sequence of events that enable transcription to progress from the initial chromatin opening to recruitment of RNA polymerase, transcription initiation, transcription elongation and transcription termination (Fig. 6). In the cell, coactivators serve as ‘master regulators’ that act to integrate cellular and extracellular events with the process of gene transcription with the goal of producing a coherent physiologic transcriptional response (Lonard and O’Malley, 2007, Lonard and O’Malley, 2008a, Lonard and O’Malley, 2008b). In this way, coregulators act as integrators and processors of cellular signals and relay those signals to the chromatin to generate the transcriptional responses appropriate to upstream signals. It has been shown recently that coregulators also participate in processes outside of transcription. Their expanded roles include regulation of mRNA processing, export and mRNA translation into protein and other cellular events (Lonard and O’Malley, 2007, Lonard and O’Malley, 2008a, Lonard and O’Malley, 2008b). Therefore, steroid hormone-dependent transcription is absolutely contingent on the recruitment and interaction of these cofactors with SHRs.

For example, dependent upon cell and gene context, ERα recruits all three SRC coactivators (Anzick et al., 1997, Demarest et al., 2002, Green and Carroll, 2007, Hong et al., 1997, Kamei et al., 1996, Lonard and O’Malley, 2007, Lonard and O’Malley, 2008a, Lonard and O’Malley, 2008b, Torchia et al., 1998). Once recruited to chromatin by ERα, SRC coactivators engage in an elaborate process of recruitment and exchange of enzymatically active molecules that maintain the open state of chromatin, and temporally and spatially direct the assembly of the basal transcription machinery. ERα recruitment of SRC coactivators causes their interaction with CBP and p300. CBP and p300 HATs acetylate histone H3 at lysine K14 and histone H4 at lysines K5 and K8. Furthermore, they acetylate histone H2A, and H2B lysines (Chen et al., 1997, Demarest et al., 2002, Kamei et al., 1996, Kim et al., 2001, Kobayashi et al., 2000, Martinez-Balbas et al., 1998, Schiltz et al., 1999, Spencer et al., 1997, Webb et al., 1998). CBP and p300 are required for the initial decondensation of the chromatin and their functions are often thought to be redundant; however, they are both recruited to the promoter and show different cycling dynamics at the same promoter. In addition to its interaction with coactivators, p300 interacts with the ERα AF2 domain where it promotes functional synergism between the AF1 and AF2 domains (Green and Carroll, 2007, Kobayashi et al., 2000). p300/CBP-associated factor (PCAF) is a homologue of the yeast GCN5 and contains intrinsic HAT activity. PCAF is not directly recruited to ERα, but interacts with ERα via SRC-1 (Santos-Rosa et al., 2003). PCAF has been shown to be a transcriptional coactivator by increasing E2-dependent transcription of ERα by acetylating H3K9 and H3K14 (Santos-Rosa et al., 2003, Xu et al., 1998).

Another class of enzymes that are recruited to the site of ERα transcription by the p160 coactivators includes the methyltransferases CARM1 and PRMT1 (Bedford and Richard, 2005, Metivier et al., 2003, Qi et al., 2002, Strahl et al., 2001, Wang et al., 2001b). PRMT1 adds methyl groups to histone H4 arginine R3 (Strahl et al., 2001, Wang et al., 2001b). This methylation is essential for ERα transcription as it is one of the earliest histone modification marks that ensues after hormone treatment. Histone H4R3 methylation is thought to precede histone acetylation and is considered to be a licensing event on the nucleosome (Strahl et al., 2001, Wang et al., 2001b). CARM1 functions both as a methyltransferases and as an adaptor protein. As a methyl transferase CARM1 methylates histone H3. It does not interact directly with ERα but it is recruited by an SRC coactivator simultaneously with p300. As an adaptor protein, CARM1 acts to recruit BRG1, a member of the ATP-dependent chromatin remodelling complex (Bauer et al., 2002, Bedford and Richard, 2005, Ma et al., 2001, Metivier et al., 2003, Qi et al., 2002, Schurter et al., 2001). The action of methylases is reversed by the action of demethylases such as lysine-specific demethylase LSD1 that removes methylation marks from the histone.

The activity of coactivator proteins at the site of impending ERα-mediated transcription creates a pathway for the recruitment of basal transcription factors. RNA polymerase II (RNAPII) binding and initiation of transcription requires the assembly of a complement of basal transcription factors that include TFII A, B, D, E, F and H.