This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Heemers, H. V. Articles by Tindall, D. J. Search for Related Content PubMed PubMed Citation Articles by Heemers, H. V. Articles by Tindall, D. J.

Androgen Receptor (AR) Coregulators: A Diversity of Functions Converging on and Regulating the AR Transcriptional Complex

Departments of Urology Research, Biochemistry, and Molecular Biology, Mayo Clinic, Rochester, Minnesota 55905

Correspondence: Address all correspondence and requests for reprints to: Dr. Donald J. Tindall, Departments of Urology Research/Biochemistry and Molecular Biology, Mayo Clinic, 200 1st Street SW, Rochester, Minnesota 55905. E-mail: tindall.donald{at}mayo.edu

	Abstract

Top Abstract I. Introduction II. The Androgen Receptor... III. AR-Interacting Proteins IV. Implications and... V. AR Coregulators in... VI. Conclusions and Future... References

 
Androgens, acting through the androgen receptor (AR), are responsible for the development of the male phenotype during embryogenesis, the achievement of sexual maturation at puberty, and the maintenance of male reproductive function and behavior in adulthood. In addition, androgens affect a wide variety of nonreproductive tissues. Moreover, aberrant androgen action plays a critical role in multiple pathologies, including prostate cancer and androgen insensitivity syndromes. The formation of a productive AR transcriptional complex requires the functional and structural interaction of the AR with its coregulators. In the last decade, an overwhelming and ever increasing number of proteins have been proposed to possess AR coactivating or corepressing characteristics. Intriguingly, a vast diversity of functions has been ascribed to these proteins, indicating that a multitude of cellular functions and signals converge on the AR to regulate its function. The current review aims to provide an overview of the AR coregulator proteins identified to date and to propose a classification of these AR coregulator proteins according to the function(s) ascribed to them. Taken together, this approach will increase our understanding of the cellular pathways that converge on the AR to ensure an appropriate transcriptional response to androgens.

I. Introduction

II. The Androgen Receptor (AR)

III. AR-Interacting Proteins

A. General transcription factors

B. AR coregulators

C. Specific transcription factors

IV. Implications and Significance of the Convergence of a Multitude of Diverse Functions on the AR

A. Accurate transcriptional output by the AR requires the concerted action of numerous cellular pathways and processes

B. Assembly of the AR transcriptional complex

C. Accommodation of coregulators by the AR

V. AR Coregulators in (Patho)Physiology

VI. Conclusions and Future Directions

	I. Introduction

Top Abstract I. Introduction II. The Androgen Receptor... III. AR-Interacting Proteins IV. Implications and... V. AR Coregulators in... VI. Conclusions and Future... References

 
ANDROGENS, WHICH ARE the main male sex steroids, are responsible for the development of the male phenotype during embryogenesis and for male sexual maturation at puberty. In adulthood, androgens remain essential for the maintenance of male reproductive function and behavior. In addition to their effects on reproduction, androgens affect a wide variety of nonreproductive tissues including skin, bone, muscle, and adipose tissues. As a consequence, deregulations in the production or action of androgens can affect different organ systems with a variable degree of severity. This is reflected in pathologies ranging from androgen insensitivity syndromes and prostate cancer to an increased risk and susceptibility to gender-related diseases such as hepatocellular carcinomas, to relatively mild conditions such as acne and male pattern alopecia (1, 2, 3, 4, 5, 6).

Testosterone, the principal androgen in the male circulation, is synthesized by the testes. The remaining androgens in the bloodstream [5–10%, including dehydroepiandrosterone (DHEA), androstenediol, and androstenedione] are either produced by the adrenal cortex and can be converted into testosterone in peripheral tissues or are derived from peripheral conversion from testosterone [dihydrotestosterone (DHT)] (7, 8). Synthesis of androgens is tightly regulated by the hypothalamic-pituitary-gonadal axis. Pulsatile secretion of LHRH by the hypothalamus stimulates secretion of LH by the anterior pituitary, which in turn induces production of testosterone by the testicular Leydig cells. Testosterone acts through a negative feedback loop to prevent LHRH release by the hypothalamus and to decrease the sensitivity of the pituitary to LHRH. The majority of circulating testosterone is bound to carrier proteins: SHBG or albumin. Therefore, only 1–2% of testosterone exists in an unbound, free form. The major androgens in women include DHEA sulfate, DHEA, androstenedione, testosterone, and DHT. Androgen biosynthesis occurs in both the adrenal and the ovary, and it is regulated by ACTH (adrenal synthesis) and LH (ovarian synthesis) (9, 10).

Upon transportation by the blood to its target tissues, unbound, lipophilic testosterone diffuses into its target cell where it can be rapidly and irreversibly converted into its more potent metabolite DHT by action of 5-reductase in some but not all target cells (type I or II, depending on the target tissue) (11). Both testosterone and DHT (either locally produced or from the circulation) exert their activities by binding to a cognate receptor, the androgen receptor (AR), a 110-kDa member of the nuclear receptor superfamily of ligand-activated transcription factors. DHT binds the AR with higher affinity, and its biological activity exceeds that of testosterone up to 10 times (12). DHT dissociates from the AR more slowly than testosterone, and AR bound to DHT is more stable, persisting in cells for greater lengths of time (13). In its basal, unliganded state, the AR resides primarily in the cytoplasmic compartment where it exists in a complex with heat shock proteins (Hsps) and immunophilin chaperones such as Hsp70, 90, 56, and p23. Upon ligand binding, alterations occur in the composition of this Hsp complex, and the AR undergoes a conformational change, allowing nuclear translocation of the AR and AR homodimer formation (14). Inside the nucleus, the activated AR binds to specific recognition sequences known as androgen reponse elements (AREs) in the promoter and enhancer regions of target genes. The ARE-bound AR dimer can either interact directly with components of the transcription preinitiation complex or recruit other components that promote such a functional interaction (15, 16, 17). Recruited elements may be other transcription factors binding to recognition elements in the vicinity of AREs and forming more complex androgen response units, or they can be coregulator proteins. As a general definition, AR coregulators are proteins that are recruited by the AR and either enhance (i.e. coactivators) or reduce (i.e. corepressors) its transactivation, but they do not significantly alter the basal transcription rate and do not typically possess DNA binding ability. Instead, coregulators influence AR-mediated transcription by acting at the target gene promoter region to facilitate DNA occupancy, chromatin remodeling, and/or recruitment of general transcription factors associated with RNA polymerase II, or by assuring the competency of the AR to enhance gene expression directly. The latter can be achieved by modulation of the proper folding of the AR, ensuring its stability or correct subcellular localization (16, 17). In the last decade, an overwhelming and ever increasing number of proteins have been identified as AR coregulators. In the current review, we provide an overview of the AR coregulators that have been identified to date. In view of the remarkable functional diversity displayed by these proteins and the vast number of cellular pathways in which they are involved, we propose a classification of AR coregulatory proteins according to their intrinsic primary function. Finally, we discuss the importance of these factors in the regulation of tissue-selective androgen-dependent gene expression under physiological and pathological conditions.

	II. The Androgen Receptor (AR)

Top Abstract I. Introduction II. The Androgen Receptor... III. AR-Interacting Proteins IV. Implications and... V. AR Coregulators in... VI. Conclusions and Future... References

 
Like other members of the nuclear receptor superfamily, the AR is characterized by a modular structure consisting of four functional domains: an N-terminal domain (NTD), a DNA-binding domain (DBD), a hinge region, and a ligand-binding domain (LBD) (18, 19). The AR NTD is relatively long and displays the most sequence variability among nuclear receptors. It is very flexible and displays a high degree of intrinsic disorder, which has hampered studies into its three-dimensional structure (20). The AR NTD contains the major transactivation function of the AR, termed activation function (AF) 1. When separated from the LBD, AF-1 gives rise to a constitutively active AR. Two motifs in the AR NTD, 23-FQNLF-27 and to a lesser extent 433-WHTLF-473, have been shown to interact with the LBD, resulting in an NH₂-COOH terminal intra- and/or intermolecular AR interaction that has been proposed to be important for the transcriptional activation of some, but not all, AR target genes (21). AF-1 is highly modular and consists of two transactivation units (TAUs), TAU 1 and TAU 5, that participate in transcriptional activation (22). The AF-1 domain undergoes induced folding when contacted by basal transcription factors such as TFIIF, resulting in a more compact and active conformation that enables further coregulator recruitment and transcription (20). In addition, the NTD harbors a variable number of homopolymeric repeats, the most important of which is a polyglutamine repeat that ranges from 8 to 31 repeats in normal individuals, with an average length of 20. Expansion of the glutamine repeats up to 40 residues or more results in spinal and bulbar muscular athrophy (SBMA or Kennedy’s disease), an X-linked pathology characterized by neurological features and late onset symptoms of mild androgen insensitivity (1, 23, 24). Shortening of the polyglutamine stretch, on the other hand, gives rise to a more transcriptionally active AR, which has been suggested to be associated with a predisposition to prostatic neoplasia (25).

The centrally located DBD is the most conserved region within the nuclear receptor family. This region harbors nine cysteine residues, of which eight are involved in forming two zinc fingers, and a C-terminal extension. The first zinc finger, most proximal to the NTD, determines the specificity of DNA recognition, whereas residues in the second zinc finger are involved in AR dimerization. Two AR monomers in a head-to-head conformation bind as a homodimer to AREs (26), which are direct or indirect repeats of the core 5'-TGTTCT-3', or more complex response elements harboring diverse arrangements of AREs (27, 28). The C-terminal extension is important for the overall three-dimensional structure of the DBD and plays a role in mediating the AR selectivity of DNA interaction (27).

The hinge region has long been considered to be a flexible linker between the DBD and LBD. More recently, however, this region was shown to be involved in DNA binding as well as AR dimerization and was suggested to attenuate transcriptional activity of the AR (29, 30). Moreover, a ligand-dependent bipartite nuclear localization signal (NLS) is located in the carboxyterminal part of the DBD and the hinge region, implicating the hinge region in AR nuclear translocation (31, 32).

X-ray crystallographic studies indicate that the AR LBD structure is similar to that of the other members of the nuclear receptor superfamily (33, 34, 35, 36). The LBD in nuclear receptors consists of 12 discrete -helices. Insertion of the agonist into the AR ligand-binding pocket has been suggested to change the conformation of the LBD in such a way that helix-12 is stabilized. This leads to the formation of a shallow hydrophobic groove at the top of the ligand binding pocket, generally referred to as AF-2. AF-2 is the major protein-protein interaction surface used by nuclear receptors to recruit LXXLL-motif containing coactivators (37). The AR, however, differs from other nuclear receptors in this respect and interacts with coactivators in a unique manner. The hydrophobic pocket in the AR-LBD binds preferentially to FXXLF motifs, including the 23-FQNLF-27 found in its NTD, and interacts poorly with LXXLL motifs commonly found in coactivators (38, 39, 40, 41, 42). Consequently, the hydrophobic pocket within the AR LBD facilitates intramolecular and intermolecular interaction between the AR NTD and its C terminus and is apparently not readily available for coactivator binding. Recent data suggest that the AR N/C terminal interactions occur predominantly when the AR is not bound to DNA (43). Interestingly, several AR-associated coactivators that contain FXXLF motifs have been isolated (44), suggesting that competition exists between these regulatory proteins and the NTD for binding to the AF-2. The implications of such competition and the association of NTD and LBD are not clear, but suggest that additional surfaces outside this well-defined coactivator pocket enable the AR to interact with its coactivators and that different classes of coactivators may interact with different AR surfaces. These observations explain why the AF-2 in the AR LBD displays relatively weak ligand-dependent transactivating properties when compared with the AF-2 of other nuclear receptors. Nonetheless, mutation or deletion of AF-2 markedly reduces transcriptional activation in response to ligand. Apart from forming the ligand binding pocket, the AR LBD mediates interaction between the AR and Hsps (14).

	III. AR-Interacting Proteins

Top Abstract I. Introduction II. The Androgen Receptor... III. AR-Interacting Proteins IV. Implications and... V. AR Coregulators in... VI. Conclusions and Future... References

 
In general, proteins that interact with the AR can be divided into three general classes: 1) components of the general transcriptional machinery; 2) functionally diverse proteins with AR coactivating or corepressing properties; and 3) specific transcription factors. AR coregulators differ from general and specific transcription factors in that they do not affect the basal rate of transcription and typically do not bind to DNA.

A. General transcription factors
As is the case for other transcription factors, enhanced transcription by the AR depends on the recruitment of RNA polymerase II to promoters of its target genes. This is achieved by the assembly of general transcription factors that make up the preinitiation complex (PIC). A detailed description of transcription initiation is beyond the scope of this manuscript and has been reviewed previously (45). Briefly, formation of the PIC is accomplished by binding of TFIID, which is composed of TATA-binding protein (TBP) and TBP-associated factors, in the proximity of the transcriptional start site. TFIIB then binds TBP and recruits RNA polymerase II and TFIIF, which ensures specific interaction of RNA polymerase II at the promoter. TFIIE and TFIIH are recruited to RNA polymerase II to facilitate strand separation, which allows transcription initiation. Although many AR-associated coregulators facilitate and mediate communication between the AR and the general transcriptional machinery, the AR has also been shown to interact directly with components of the basal transcriptional machinery. For instance, the AR NTD interacts with RAP74, a large subunit of TFIIF. Binding of RAP74 induces -helical structure in AF-1 and facilitates interaction between the AR and the p160 coactivator steroid receptor coactivator (SRC)-1 (20, 46, 47, 48, 49, 50). Modest binding to the RAP30 subunit of TFIIF and TBP has also been described (46). Moreover, the AR has been shown to interact with TFIIH, and overexpression of the cdk-activating kinase subunit of TFIIH markedly stimulates the AR-mediated transcription (50). the AR interaction with TFIIH may enhance phosphorylation of the RNA polymerase COOH terminal domain (CTD), an event necessary to transition from transcription initiation to transcriptional elongation, suggesting that the AR may increase the efficiency of transcriptional elongation of AR target genes. Consistent with this hypothesis, an interaction between the AR and positive transcription elongation factor b (p-TEFb) has been described (51). The small subunit of p-TEFb, PITALRE (also known as CDK9), harbors protein kinase activity that is able to phosphorylate the CTD of the largest subunit of RNA polymerase II, which is necessary to progress from PIC formation on the promoter to transcriptional elongation. Remarkably, both TFIIH and p-TEFb possess CTD kinase activity, but these activities act at different stages of transcription (15). In addition to its contacts with several general transcription factors, the AR also interacts directly with RNA polymerase II through association with its subunit RPB2. Coexpression of RPB2, which is involved in transcriptional elongation, stimulates AR-mediated transcription of target genes (52). Interaction of the AR with and regulation of AR-mediated gene expression by other subunits of RNA polymerase II has not been observed. Taken together, these findings indicate that the AR may regulate transcription of target genes by regulating both transcriptional initiation and elongation events.

B. AR coregulators
More than 200 nuclear receptor coregulators have been identified since the isolation of the first nuclear receptor coactivator, SRC-1, in 1995 (53). By mining peer-reviewed literature accessible through PubMed (www.pubmed.gov) and consulting specialized websites dedicated to AR function [the AR gene mutation database (androgendb.mcgill.ca) and the Nuclear Receptor Signaling Atlas (NURSA) database (www.nursa.org)], we have attempted to provide an up-to-date overview of proteins that have been listed as putative coregulators for the AR. As of May 2007, the list of proteins that have been classified as potential AR coregulators contains 169 members. Remarkably, these coregulators display a diverse array of functions and are involved in multiple cellular pathways. There are many ways one could categorize or group these proteins. We have chosen to arrange these proteins according to their apparent primary function, i.e., the function for which they are best recognized, even if, in some cases, this particular activity may not be critical for their effect on AR action. Overall, we feel that such a classification, rather than classification by their function as a coactivator or a corepressor, will provide a broader picture of the cellular events that converge on and regulate the transactivation properties of the AR. As will be discussed below, this does not preclude the possibility that some multifunctional AR coregulators can be assigned to multiple categories. For reasons of simplicity and to keep our overview comprehensive, we chose not to include information on the specific cell systems or specific target genes used to ascertain the coregulator properties of the AR cofactors in this section. This information will be addressed in Sections IV and V.

1. Components of the chromatin remodeling complex.
Transcription occurs on a chromatin template, in which DNA is wound around a core of four basic histone proteins (H2A, H2B, H3, and H4) to form nucleosomes. DNA-histone interactions limit the accessibility of the nucleosomal DNA to transcription factors and form a major obstacle to transcription. Chromatin remodeling complexes alter and unwrap the histone-DNA contacts in an ATP-dependent manner catalyzed by ATPases, leading to reorganization of the nucleosomal structure and eventually to a chromatin status that is more permissive to transcription (54, 55). Several AR coregulator proteins have been identified as components of the chromatin remodeling complex. One of the first indications that components of the chromatin remodeling complex may play a role in AR-mediated transcription came from the identification of AR-interacting protein (ARIP) 4, a nuclear ATPase that belongs to the SNF2-like family of chromatin remodeling proteins. ARIP4 interacts with the AR zinc-finger region and stimulates AR-dependent transactivation in cotransfection experiments (56). Alhough ARIP4 displays DNA-dependent ATPase activity, its specific activity was subsequently shown to be considerably lower than that of SNF2-family members, suggesting that it may not be a classical chromatin remodeling protein. Nonetheless, ARIP4 mutants that do not possess ATPase activity behave as dominant-negative regulators of AR function (56, 57). Subsequently, the ATPases Brahma-related gene 1 (BRG1) and human homolog of Drosophila brm gene (hBRM), two core components required for nucleosome repositioning by the mating type switching/sucrose nonfermenting (SWI/SNF) chromatin remodeling complex, were shown to stimulate AR activity potently (58). Depending on the gene context, however, different requirements for these ATPases were noted, with some genes relying solely on hBRM and others relying on both hBRM and BRG1 for androgen regulation (58). The BRG1-associated factor (BAF) 57 subunit, an accessory component of this remodeling complex, is also required for AR-dependent transactivation (59). The AR coactivation function of BAF57 depends on SWI/SNF ATPase activity and cooperates with other classes of coactivators. BAF57 directly binds to the AR and is recruited to AR target genes upon ligand stimulation (59). Interestingly, the SWI3-related gene product (SRG3/BAF155), another component of the SWI/SNF complex, is also able to enhance transactivation by the AR. SRG3 interacts with the AR DBD-hinge region and exists in a complex with the AR on promoters of AR targets genes (60). SRG3 appears to initiate at least some of its coactivation properties by enlisting the SRC-1 coactivator. Remarkably, SRG3 function may not be entirely dependent on the presence of BRG1 or hBRM (60). Furthermore, the SNF2-related CREB-binding protein (CBP) activator protein (SRCAP) is able to coactivate transcription by the AR (61). Direct interaction of SRCAP with the AR, however, has not been reported. Similarly, hOsa1 (BAF250) and hOsa2, the largest subunits of the SWI/SNF complex, stimulate transcription by the AR, but association of these proteins with the AR has not been demonstrated (62).

Overall, the recruitment of these chromatin remodeling proteins to the AR transcriptional complex is consistent with the altered DNA topology and the loss of canonical nucleosomal ladder that is observed at AR target genes after exposure to androgens (63).

2. Histone modifiers: acetyltransferases and deacytelases.
In addition to chromatin remodeling, which represents a higher order level of chromatin reorganization and involves repositioning of components of the nucleosome structure, modification of histone residues can affect transcription efficiency and provide a more localized control over transcriptional events in chromatin (54, 55). Modifications such as acetylation, methylation, phosphorylation, ubiquitination, ADP-ribosylation, and glycosylation of histone residues have been described. In most cases, modification of a histone residue changes the net charge of the nucleosome, which results in loosening or tightening of the DNA-histone interactions. For example, acetylation of a histone lysine residue attenuates its posititive charge and abrogates its interaction with the negatively charged DNA. Some of these histone modifications are associated with transcriptional activation (e.g., acetylation), and others are indicative of active or repressed genes (e.g. methylation). Also, the position of the modified histone residue can affect the activation status of a gene. The concept that a combination of such marks affects binding of transcriptionally effector proteins lies at the basis of the histone code (54, 55).

In keeping with this notion, several dynamic changes in the covalent histone modification status have been associated with androgen-stimulated transcription. These include activating modifications at histone 3 such as acetylation of lysine 9 and lysine 14, dimethylation of arginine 17, phosphorylation of serine 10, and dimethylation as well as trimethylation of lysine 4 (64). In addition, removal of repressive marks has been described, such as demethylation of mono-, di-, and trimethyl marks at lysine 9 of histone 3 and decrease in the dimethylation status of lysine 20 at histone 4 (65, 66, 67, 68). The significance of these modifications will be discussed in more detail in sections of the text describing enzymes executing these events. A summary of the histone modifications associated with androgen action can be found in Fig. 1.

View larger version (24K): [in this window] [in a new window]   FIG. 1. Covalent histone modifications associated with androgen action.

For example, two members of the p160 SRC gene family, SRC-1 and SRC-3 [p300/CBP interacting protein (p/CIP), receptor-associated coactivator-3 (RAC3), acetyltransferase (ACTR), amplified in breast cancer-1 (AIB1), or thyroid hormone receptor activator molecule-1 (TRAM1)] have been reported to possess HAT activity. Similar to SRC-2 [also known as glucocorticoid receptor-interacting protein-1 or transcriptional intermediary factor (TIF)-2], the third member of this family that does not possess HAT activity, SRC-1 and SRC-3 interact directly and ligand-dependently with the AR to enhance AR-mediated transcription (16, 53, 69, 70, 71, 72, 73, 74, 75, 76, 77). Although the LXXLL motif-containing p160 family members have been shown to interact with AR AF-2, they interact primarily with the AR N terminus and possibly the DBD. Moreover, all three SRC family members function as scaffold proteins that attract additional coactivator proteins, including factors with histone-modifying potential (78). Indeed, SRCs have been shown to recruit p300, the p300 homolog CBP, as well as p300/CBP-associated factor (P/CAF), all coactivator proteins that possess HAT functions that are intrinsically stronger than those found in SRC-1 and -3. Moreover, in vitro experiments have demonstrated a direct, SRC-independent interaction between p300, CBP, and P/CAF and the AR (79). The potentiation of ligand-induced AR transactivation by these three coactivators (79, 80, 81) relies on the presence of a functional HAT domain. In addition to their effects on histones, CBP, p300, and P/CAF can acetylate proteins such as transcription factors and coregulators. Noteworthy, p300, as well as P/CAF, acetylates the AR at three lysine residues in its DBD-hinge region (79). Point mutations in these AR acetylation sites selectively prevent androgen induction of androgen-responsive genes, hamper coactivation of the AR by SRC-1, p300, Tip 60, and Ubc9, and result in a 10-fold increase in the binding of the corepressor nuclear receptor corepressor (NCoR) (81). It should, however, be noted that the lysine residues that are acetylated by p300 and P/CAF are part of the AR NLS and that mutations of these sites may therefore be expected to disrupt AR activity regardless of acetylation events at these sites. Nonetheless, histone acetylation by p300 and CBP facilitates recruitment of the SWI/SNF and Mediator coactivator complexes (63). Furthermore, CBP and p300 function as a direct bridge between DNA-bound AR and the basal transcriptional machinery (79). They may also serve as a scaffold interacting with and assembling a number of other transcriptional regulators (79).

Another coregulator that harbors HAT activity is Tat interactive protein 60 kDa (Tip60). Tip60 interacts with the AR LBD and enhances AR-mediated transcription by acetylating histones as well as the AR. Acetylation of AR lysine residues in the AR-hinge region by Tip60 is a requisite for Tip60-mediated coactivation of the AR. Remarkably, action of Tip60 on AR transactivation is counteracted by HDAC1 (82, 83, 84).

Human origin recognition complex interacting protein (HBO1) is another HAT protein that ligand dependently associates with the AR DBD-LBD region. Contrary to the HAT proteins listed above, HBO1 acts as an AR corepressor, inhibiting hormone-dependent AR activation (85). The direct involvement of the HAT function of HBO1 in these events is not clear, because HBO1 has been reported to harbor a relatively weak HAT function. Moreover, histone acetylation by HBO1 has not been demonstrated. Therefore, HBO1 has been proposed to acetylate histones as part of a multisubunit complex (86).

The importance of acetylation and deacetylation of histone and nonhistone proteins in AR-mediated transcription is further emphasized by the functional antagonism between the nicotinamide adenine dinucleotide-dependent HDAC Sirtuin1 (SIRT1) (and p300 at sites of AR acetylation). SIRT1 (Sir2, a class III HDAC) represses androgen-induced AR signaling by a mechanism that involves direct binding to the AR hinge and requires both the nicotinamide adenine dinucleotide-dependent catalytic function of SIRT1 and deacetylation of the lysine residues 630/632/633 in the AR-hinge region that are targeted by p300 and P/CAF (87). Moreover, SIRT1 inhibits p300-mediated interaction between the AR N and C termini. On the other hand, ligand-induced repression of AR function by the class II HDAC HDAC7 is independent of these acetylation sites. Moreover, the deacetylase activity of HDAC7 is at least partly dispensable in the repression of AR function (88). In addition to SIRT1, several HDACs of classes I and II negatively affect AR transactivation. Contrary to SIRT1 and HDAC7, which can interact directly with the AR, recruitment of HDACs to the AR transcriptional complex is usually indirect through association with multisubunit corepressor complexes such as NCoR and silencing mediator of retinoid and thyroid receptors (SMRT) or as binding partners for other AR corepressors (17).

Reflecting the importance of the action of HATs and HDAC enzymes at genomic sites mediating AR transcriptional activity, the acetylation status of histone 3 and 4 residues is often assessed as a marker for the transcriptional activation status of a particular AR target gene.

3. Histone modifiers: methyltransferases and demethylases.
Although histone acetylation is generally associated with active gene transcription, histone methylation can be indicative of both the active and repressed transcriptional states of the chromatin (54, 55). The position of the histone residue affected by this modification is important for making this distinction. For example, methylation of lysine 4 on histone 3; arginines 2, 17, and 26 on histone 3; and arginine 3 on histone 4 is associated with active genes, whereas methylation of lysine 9 on histone 4 is predominantly associated with repressed genes. Moreover, lysine residues can be monomethylated, dimethylated, or trimethylated, and the extent of the methylation serves as an important indication of its transcriptional status. Although methylation has long been considered to be an irreversible epigenetic mark, recently, demethylases that mediate active demethylation of repressive histone segments have been identified (54, 55).

AR-dependent transcription relies on both methyltransferase and demethylase activities. Coactivator-associated arginine methyltransferase 1 (CARM-1) or protein arginine methyltransferase (PRMT)-5, a histone methyltransferase acting at H3R17, was identified initially by its ability to interact with SRC coactivators. Due to its indirect recruitment to ligand-bound nuclear receptors including the AR, CARM-1 has been classified as a secondary coactivator (89). Androgen stimulation leads to recruitment of CARM-1 to androgen-responsive enhancers (90). CARM-1 stimulation of AR transactivation depends entirely on the presence of SRC family members. Moreover, the presence of CARM-1 enhances AR transactivation by p160 family members. Loss of CARM-1 reduces transcription of androgen-responsive genes. Mutation of its S-adenosyl methionine binding site abrogates its methyltransferase activity and prevents CARM-1 from exerting its coactivator function. In addition to its effects on histones, CARM-1 methylates proteins in the transcriptional complex including CBP/p300 and several RNA binding proteins (89, 90). Noteworthy, CARM-1 may also be recruited to the AR transcriptional complex through p44 (MEP50), a component of the methylosome. p44 interacts directly with the AR and CARM-1, is found on promoters of AR target genes upon androgen stimulation, and stimulates the transcription of some AR-target genes. In addition, p44 and CARM-1 synergistically cooperate to enhance transactivation by the AR (91). Similar to CARM-1, PRMT1 is recruited to the AR transcriptional complex and stimulates AR-dependent gene expression via SRC proteins (92). PRMT1, however, preferentially methylates H4R3, which facilitates subsequent acetylation of histone 4 tails by p300. Remarkably, acetylation of histone 4 inhibits its methylation by PRMT1. Like CARM-1, PRMT1 depends on an intact S-adenosyl methionine binding site to fulfill its cofactor function.

Methylation of lysine 9 on histone 3 by the methyltransferase G9a is predominantly associated with repression of transcription. However, for some nuclear receptors including the AR, G9a functions as a coactivator, although weakly (93). Nonetheless, G9a cooperates synergistically with TIF-2, CARM-1, and p300 in activating transcription by the AR. This synergy is strongly dependent on the arginine-specific protein methyltransferase activity of CARM-1, suggesting a link between histone arginine and lysine methylation in AR-mediated transcription. Noteworthy, PRMT methyltransferases cannot substitute for CARM-1 in this respect. On the other hand, cooperation between G9a, CARM-1, and SRC-2 does not absolutely require the enzymatic activity of G9a. The dependency of G9a on SRC-2 indicates that SRC-2 may function as a scaffold to recruit G9a. G9a associates with regulatory regions in AR target genes in the presence as well as the absence of androgens, and loss of G9a expression hampers androgen-induced AR-dependent stimulation of target genes (93).

Recently, histone methyltransferase activity with a specificity for H3-K36 and H4-K20 has been attributed to nuclear receptor-binding Su(var)3-9, Enhancer-of-zeste, Trithorax (SET) domain-containing protein-1 (NSD1, also known as AR-associated (ARA) protein 267) (94). Based on its ability to interact with the AR DBD-LBD region, to stimulate AR transactivation in an androgen-dependent manner, and to cooperate with other AR coregulators, ARA267 has been classified as an AR-associated coactivator (95). Although this possibility remains to be proven, it is tempting to speculate that the AR-coactivating properties of ARA267 are mediated by its histone methyltransferase moiety.

The concept that demethylation of histones is involved in transcriptional activation by the AR has emerged only recently. Interest was triggered by the observation that lysine-specific demethylase 1 (LSD1), which specifically demethylates monomethylated and dimethylated H3K9, interacts with the AR (NTD, DBD, as well as LBD), and stimulates AR-dependent transcription (65). Down-regulation of LSD1 expression abrogates androgen-induced transcriptional activation. Chromatin immunoprecipitation (ChIP) analysis demonstrated that AR and LSD1 form chromatin-associated complexes in a ligand-dependent manner. Androgen exposure leads to a robust decrease in mono-, di-, and trimethyl H3K9 marks at the promoter of AR target genes. Small interfering RNA (siRNA)-mediated loss of LSD1 prevents ligand-induced changes in mono- and dimethyl H3K9 but does not affect trimethyl H3K9 (65). Similar to LSD1, JHDM2A, which demethylates mono- and dimethylated H3K9, interacts directly with the AR and coactivates androgen-mediated transcription (66). Contrary to LSD1, which is constitutively present at AR target genes, JHDM2A exhibits hormone-dependent recruitment. Overexpression of JHDM2A greatly reduces the H3K9 methylation level. A knockdown of JHDM2A expression results in increased dimethyl K9 levels at the promoter region of AR target genes concomitant with a decrease in their expression. Loss of LSD1 does not affect the binding of the AR to promoter regions of target genes or the ligand-induced recuitment of JHDM2A, but it does partially impair hormone-induced reduction of dimethyl H3K9 (66). In addition to LSD1 and JHDM2A, a third demethylase, JMJD2C, interacts with and functions as a coactivator for the AR (67). Interestingly, JMJD2C is a histone tridemethylase able to remove mono-, di-, and trimethyl marks from H3K9. Like LSD1, JMJD2C is constitutively present at promoter regions of AR target genes. Both JMJD2C and LSD1 interact with and stimulate AR-dependent gene transcription in a cooperative manner. Upon androgen treatment, AR, LSD1, and JMJD2C assemble on chromatin, resulting in demethylation of mono-, di-, and trimethyl H3K9 and stimulation of AR-dependent transcription. Conversely, knockdown of JMJD2C inhibits androgen-induced removal of trimethyl H3K9 and transcriptional activation by the AR (67). Taken together, these observations indicate that androgen-dependent gene transcription requires the assembly and coordinate action of methyl transferases and demethylases with distinct substrate specificities.

4. Components of the ubiquitination/proteasome pathway.
Ubiquitination is a reversible posttranslational modification of cellular proteins in which a 76-amino acid polypeptide, ubiquitin, is attached to lysines in target proteins. Ubiquitination of a substrate involves the action of an activating E1 enzyme that transfers ubiquitin to a conjugating E2 enzyme, which in turn enlists an E3 ligase to deliver the ubiquitin tag to the intended target protein. Target proteins can be either polyubiquitinated or monoubiquitinated. The former usually serves as a signal for degradation of the substrate protein by the 26S proteasome, whereas the latter tends to function as a signal that regulates protein stability and protein-protein recognition, activity, and intracellular localization. Both modes of ubiquitination play vital roles in transcriptional regulation because they allow proper progression through rounds of transcription and appropriate assembly of the necessary protein complexes, and they modulate the activation status of transcription factors and coregulators (96, 97). Not surprisingly, several AR coregulator proteins function in the ubiquitin-proteasome pathway. Most of these coregulators demonstrate E3 ligase activity; this is the case for instance for E6-AP, Mdm2, PIRH2, SNURF/RNF4 and Chip. The E3 ligase E6-associated protein (E6-AP) interacts with the AR NTD in a hormone-dependent manner, demonstrates hormone-dependent recruitment to the promoter region of AR target genes, and enhances the transactivation function of the AR (98). E6-AP may also modulate the protein level of the AR because E6-AP null mice demonstrate increased AR levels in androgen-responsive prostate tissues, and overexpression of E6-AP markedly reduces AR protein expression in cells in culture (98). In addition, Mdm2 interacts with the AR and catalyzes its ubiquitination and proteolysis. The interaction of Mdm2 with the AR NTD and DBD is at least in part dependent on its E3 ubiquitin ligase activity (99). After androgen stimulation, Mdm2 complexes with the AR and HDAC1 at active AR target gene promoters and attenuates AR activity. Both the AR and HDAC1 are ubiquitinated in response to androgen. HDAC1 and Mdm2 cooperate to reduce AR-mediated transcription, and this functional interaction is attenuated by the HAT activity of the AR coactivator Tip60 (100). This suggests an interplay between acetylation status and receptor ubiquitylation in AR regulation. Supporting this possibility, Tip60 also interacts with PIRH2, another E3 enzyme. Similar to Mdm2, PIRH2 interacts directly with the AR (NTD) and HDAC1 (101). However, PIRH2 enhances AR-mediated transcription by reducing HDAC1 protein levels and inhibition of HDAC1-mediated transcriptional repression. PIRH2 is recruited to AREs in AR target genes and is required for optimal expression of these genes. The E3 ubiquitin ligase small nuclear RING finger protein (SNURF/RNF4) was isolated as an AR coregulator based on its ability to interact with the AR NLS in a hormone-dependent manner (102). SNURF does not influence the binding of the AR to cognate DNA sequences, but it appears to enhance transactivation by the AR by facilitating its import into the cell nucleus (103). SNURF also retards AR nuclear export on hormone withdrawal (104). To date, there is no evidence of SNURF-mediated changes in the ubiquitination pattern of the AR. In addition to its E3 ubiquitin ligase function, C-terminal Hsp-interacting protein (Chip) acts as a molecular chaperone involved in the folding and hormone binding of nuclear receptors. Interaction between Chip and the AR NTD, which occurs in a highly specific and sequence-dependent manner, suggests a functional link between these processes in the regulation of AR activity (105). Chip negatively regulates AR transcriptional activity by promoting AR ubiquitination and degradation (105, 106). Noteworthy, these effects of Chip are not fully reversed by proteasome inhibitors, suggesting that mechanisms different from proteasome-mediated degradation are involved. Indeed, Chip overexpression reduces the rate of AR degradation, which is consistent with an effect on AR folding (107). Thus, its effect on AR action might not be entirely dependent on its ubiquitin ligase activity.

Other proteins that have been identified as AR coregulators and have been suggested to possess E3 ubiquitin ligase function are ARNIP, ARA54 and MKRN1 (108, 109, 110, 111). AR N-terminal-interacting protein (ARNIP) interacts with the AR NTD, although it does not affect AR ligand-binding kinetics or AR transcriptional activation. However, AR N-terminal-C-terminal interaction is reduced in the presence of ARNIP (108). ARA54 has been isolated based on its ability to ligand-dependently associate with the AR LBD and was subsequently shown to enhance AR-mediated transactivation (109). The ubiquitously expressed Makorin RING zinc finger protein 1 (MKRN1) inhibits the transcriptional activity of the AR (111). Interestingly, disruption of the ubiquitin ligase activity of MKRN1 does not affect its inhibitory transcriptional activity. Whether ARNIP, ARA54, and MKRN1 affect the ubiquitination status of the AR or its associated complex has not been assessed.

In addition to these E3 ubiquitin ligases, proteins with other functions in the ubiquitin/proteasome pathway have been shown to modulate AR-mediated transcription. For instance, the ubiquitin-specific protease USP10 interacts directly with the AR and is part of DNA-bound AR complexes (112). USP10, but not an enzymatically inactive mutant USP10, stimulates AR-dependent transcription. Conversely, loss of USP10 expression impairs responsiveness to androgens. Similarly, the E2 ubiquitin-conjugating enzyme UBCH7 stimulates AR-mediated transactivation in a hormone-dependent manner (113). The ubiquitin conjugation activity of UBCH7 is required for this response. Tumor susceptibility gene 101 (TSG101), an E2-like enzyme deprived of ubiquitin conjugase activity, which has been proposed to act as a dominant-negative inhibitor of polyubiquitination, also modulates the AR transactivating potential. TSG101 was originally shown to repress ligand-dependent AR transcriptional activation (114). Interestingly, in an independent study TSG101 was reported to act as a coactivator for AR-mediated transcription (115). TSG101 enhances monoubiquitination of the AR in a ligand-dependent manner, and this correlates with enhanced transactivating capacity. A dominant-negative mutant of ubiquitin preventing polyubiquitination also stimulates AR-mediated transcription, which cannot be enhanced by TSG101 (115).

These reports establish multiple effectors of distinct steps in the ubiquitinylation pathway as important regulators of the AR transactivating function. Interestingly, the role of the proteasome in these events does not appear to be restricted solely to degradation of the AR or its associated proteins in the cytoplasm (Ref.116 , and also discussed below). Instead, the proteasome itself may be actively involved in AR-governed transcriptional events in the cell nucleus. This hypothesis is supported by the observation that overexpression of the proteosomal subunit PSMA7 enhances AR transactivation (117). Moreover, inhibition of the proteasome prevents the nuclear translocation of the AR, blocks interaction between the AR and several of its coregulators, and ultimately abolishes androgen-induced target gene expression (117). In addition, after androgen treatment, the 19S proteasomal subcomplex is recruited to AR target genes, where its occupancy parallels that of the AR (118).

5. Components of the sumoylation pathway.
The small ubiquitin-related modifier SUMO posttranscriptionally modifies several proteins involved in regulation of transcription and chromatin structure. The machinery responsible for the sumoylation of target genes displays remarkable similarity to that mediating protein ubiquitination because it consists of E1-activating enzymes, an E2-conjugating enzyme, and several E3 ligases. SUMO E1, -2, and -3 enzymes are, however, distinct from their counterpart enzymes in the ubiquitin/proteasome pathway. Similar to the non-proteasomal roles of ubiquitin, SUMO modification of a protein primarily regulates localization and activity (119). Multiple proteins involved in several aspects of the sumoylation pathway can modulate the AR transcriptosome. First, SUMO homologs have been shown to affect AR-mediated transcription. SUMO-1 decreases, whereas SUMO-2 and -3 enhance AR transcriptional activity (120, 121). SUMO-3 interacts with the AR NTD. The positive effect of SUMO-3 on AR-transcriptional activity does not depend on either the sumoylation sites of the AR or the sumoylation function of SUMO-3 (121). Sumoylation of the AR NTD by SUMO-1 is stimulated by androgens, and mutation of the sumoylation sites in the AR NTD increases AR transactivation, indicating that sumoylation serves to attenuate AR function (120). Second, the SUMO E2 conjugating enzyme Ubc9 interacts with the NLS in the AR-hinge region and acts as a AR coactivator (122). Interestingly, the effects of Ubc9 on AR activity occur independently of its SUMO-1 conjugating catalytic activity (122, 123). Third, protein inhibitors of activated STAT (PIAS) family members that function as SUMO E3 ligases can either positively or negatively affect transcription by the AR (124, 125, 126, 127, 128, 129, 130, 131, 132). PIAS proteins (PIAS-1, -3, -x, or ARIP3, -xβ, and -y) bind the AR DBD (126). Although homologous, PIAS proteins can differ markedly in their ability to sumoylate the AR and/or AR-associated coactivators such as TIF-2 (131). PIAS-y, which acts as a corepressor for the AR, does not rely on its E3 ligase activity to influence AR-mediated transcription (132). Moreover, the decision for a particular PIAS protein to function as a corepressor or coactivator depends on the cell type and the target gene (127). Furthermore, the PIAS-like SUMO E3 ligases Zimp7 and Zimp10 both function as AR coactivators (133, 134, 135). Zimp10 was shown to interact with the AR NTD (133). AR sumoylation is enhanced in the presence of Zimp10, and mutation of the AR sumoylation sites abrogates the augmentation of AR activity by Zimp10 (133, 135). Fourth, AR-mediated transcription is markedly enhanced by SENP1, a member of the SUMO-specific protease family (136). Although the AR is a target for SENP1, the ability of SUMO1/sentrin-specific protease 1 (SENP1) to enhance AR-dependent transcription is not mediated through desumoylation of the AR, but rather through its ability to deconjugate HDAC1, thereby reducing its deacetylase activity. The repressive effect of HDAC1 on AR-dependent transcription can be reversed by SENP1 and by deletion of its sumoylation sites. In contrast, SENP2 and SENP3 have only modest effects on AR transactivation (136).

In addition to the ubiquitination/proteasome and sumoylation pathways, activity of the related neddylation pathway appears to be involved in AR-mediated transcription. Ubiquitin-activating enzyme 3 (Uba3), the catalytic subunit of the activating enzyme of the ubiquitin-like NEDD8 (neural precursor cell expressed developmentally down-regulated) conjugation pathway, inhibits transactivation by the AR. The neddylation activity of Uba3 is required for its inhibition of steroid receptor transactivation. Direct interaction between Uba3 and the AR has not been reported (137).

Overall, the identification of numerous proteins with functions in the ubiquitination, SUMOylation as well as NEDDylation pathways as AR coregulators points toward the critical importance of tight regulation of the turnover, stability, degradation, and subcellular relocalization of components of the AR transcriptional machinery in androgen-regulated gene transcription.

6. Proteins involved in splicing and RNA metabolism.
Primary transcripts undergo several modifications before a mature mRNA is generated that can serve as a template for translation. Because the steps from transcription to translation are mechanistically and functionally coupled (138, 139), it is not surprising that several proteins, which demonstrate AR coregulator characteristics, belong to or share high functional or structural homology to members of the family of RNA processing enzymes. For instance, the pre-mRNA splicing proteins PSF (polypyrimidine tract-binding protein-associated splicing factor), PSP1 (paraspeckle protein 1) and PSP2 interact with AF-1 of the AR (140). p54nrb (p54 nuclear RNA binding protein, NonO), another component of the U1snRNP prespliceasome, interacts with the AR NTD in a ligand-dependent manner and potentiates AF-1 function (140). Because p54nrb and PSF also directly interact with the RNA polymerase II CTD, a molecular link between the AR transcriptional and splicing machinery is apparent (138). Furthermore, p102 U5snRNP has been isolated as a protein interacting with AR AF-1, termed ANT-1 (AR N-terminal domain transactivating protein-1). ANT-1 enhances the ligand-independent AF-1 function of the AR but does not affect ligand-dependent AF-2 activity (141, 142).

The splicing factor hnRNPA1 is recruited to the AR through association with the AR interacting coactivator ARA54 and selectively suppresses ARA54-enhanced AR transactivation via interruption of AR-ARA54 interaction (143). Finally, the previously discussed methyltransferases CARM-1 and p44 (MEP50) are components of the methylosome complex that methylates snRNP complex proteins, suggesting that they also fulfill roles in splicing events.

7. Proteins involved in DNA repair.
In a targeted approach to discover components that comprise the AR apo- and holoreceptor complex using tandem mass spectroscopy analysis, the trimeric DNA-dependent protein kinase (DNA-PK) complex was isolated (144). The DNA-PK complex is best known for its role in DNA repair and has emerged as a part of the transcriptional machinery. The AR-LBD interacts directly with the Ku70 and Ku80 regulatory subunits of DNA-PK in a DNA-independent manner. Interaction between the AR and the catalytic DNA-PK subunit has not been observed. Ku proteins bind the AR in both the cytoplasm and the nucleus. Ku proteins are recruited in an androgen-dependent manner to the promoter of AR target genes. Ku70 and Ku80 as well as DNA-PK enhance AR activity in transactivation assays. Ku70 and Ku80 have been shown to exert these effects through recycling of transcriptional factors (144). Other proteins with roles in DNA repair and damage control have been demonstrated to modulate AR-mediated transcription. For instance, the checkpoint protein Rad9 acts as a corepressor to suppress AR transactivation (145). The AR interacts with the C terminus of Rad9 via its LBD. The FXXLF motif within the C terminus of Rad9 interrupts the androgen-induced interaction between the N terminus and C terminus of the AR. Moreover, the tumor suppressor genes BRCA1 and BRCA2 are AR coactivators (146, 147, 148). BRCA2 is an integral component of the homologous recombination machinery, whereas BRCA1 possesses both E3 ubiquitin ligase activity and DNA repair activity. BRCA1 interacts with the AR NTD and enhances transcription of AR target genes. BRCA1-enhanced AR transactivation can be further induced synergistically with AR coregulators SRCs, CBP, ARA55, and ARA70 (146, 147). BRCA2, but not a truncated mutant of BRCA2, synergizes with SRC-2 to enhance transcriptional activation by the AR. BRCA2 associates with the AR NTD and LBD, as well as SRC-2, and further cooperates with P/CAF and BRCA1 to enhance AR- and SRC-2-mediated transactivation (148).

In general, components of the DNA repair machinery are recruited when the transcriptional machinery runs into obstacles or DNA lesions that prevent proper transcription of target genes. The identification of several proteins with functions in DNA repair as coregulators for the AR indicates that this mechanism holds true also for AR-mediated transcription.

8. Chaperones and cochaperones.
In the absence of androgens, the molecular chaperone complex is critical to maintain the AR in a stable, inactive, intermediate configuration that has a high affinity for androgenic ligands. Upon binding of androgens and folding of the AR into an active conformation, selective molecular chaperones remain associated with the AR and are important for downstream events such as AR translocation, AR transcriptional activity, disassembly of the AR transcriptional complex, and AR degradation. The sustained importance of these proteins in overall AR activity is reflected in the interactions between the AR and multiple components of chaperone complexes throughout the life cycle of the AR (14). In the early stages of the AR activation process, AR LBD interacts transiently with Hsp40 (Ydj1), Hsp70 (HSc70), Hip, Hps90, Hop, and p23, leading to an equilibrium in which the AR is maintained in an overall high-affinity ligand-binding state. In this respect, Hsp40 is necessary for hormone binding to the AR (149). Mutations in Hsp40 result in a reduction of AR-Hsp70 complex formation and defects in AR folding (150). In contrast, loss of the Hsp70 cochaperone DjA1 in a knockout mouse model leads to increases in AR protein levels and enhanced transcription of several androgen-responsive genes in Sertoli cells, giving rise to severe defects in spermatogenesis (151). In the same study, DjA1 was suggested to function as a negative regulator of transactivation by the AR. Hormone binding causes the AR to undergo a sequential loss of chaperones (14). With the assistance of Hsp90, the AR is transformed into a DNA-binding competent state. Receptor activation leads to unmasking of the NLS, resulting in Hsp90-dependent translocation of the AR to the nucleus (14). Cdc37 (p50) also functions down-stream of hormone-binding as an Hsp90-associated protein involved in AR trafficking. Mutant forms of cdc37 induce defects in AR transactivation while leaving AR protein levels unaltered (152). Hsp70 and Hsp40 are also believed to reassociate with the AR in the presence of ligand and to facilitate transport of the receptor into the nucleus (14). Hsp90 binding cochaperone FKBP52 (FK506 binding protein of 52kDa) interacts with AR complexes (153). In vivo studies using FKBP52-deficient mice indicate that FKBP52 does not affect hormone binding by the AR or AR nuclear translocation but is critical for AR transactivation (153, 154). FKBP52 enhances AR-mediated transcription, and this effect depends on its ability to interact with Hsp90 (153). FKBP52 may also be necessary to maintain AR protein levels (153). The related cochaperone FKBP51 forms complexes with the AR and stimulates AR transactivation, although these observations could not be confirmed by a second, independent study (154, 155). Apart from their effects on AR folding and trafficking, experimental evidence supports a direct involvement of molecular chaperones in the transcriptional activation of AR target genes. Bag-1 Hsp70 cochaperones, in particular the Bag-1L isoform, bind directly to the TAU 5 domain in the AR NTD and function as a coactivator for the AR (156, 157). Bag-1L relies on its association with Hsp70 to interact with the AR, and loss of this interaction domain markedly suppresses its abilitity to stimulate AR-mediated transactivation. Moreover, Bag-1L as well as Hsp70 are recuited with the AR to promoter regions of AR target genes (156). Noteworthy, Bag-1L harbors a ubiquitin-like domain that facilitates association of Bag-1L with the proteasome, enabling Bag-1L to function as a coupling factor between the chaperone and proteolytic complex (158). This is reminescent of the dual function of Chip as both an E3 ubiquitin ligase and a cochaperone. In fact, AR degradation by an Hsp70-Chip governed system has been reported (158). Bag-1 and Chip interact directly and cooperate with each other during the sorting of chaperone substrates to the proteasome. Taken together, these findings suggest an interdependency between AR-mediated transcription, AR degradation, and folding events.

9. Cytoskeletal proteins.
Actin is a major component of the cytoskeleton. Although the cytoplasmic roles of actin and actin organization in the cytoplasm have been well established, the possibility for a role for actin in the nucleus has been controversial. However, it is now generally accepted that actin plays a role in transcriptional events. Actin as well as actin binding proteins have been shown to mediate nuclear translocation of transcriptional regulators. Moreover, actin is found as part of chromatin remodeling complexes and ribonucleoprotein particles and interacts directly with RNA polymerases (159). Actin binding proteins and actin monomers bind to the AR, indicating that they also play an important role in AR-mediated transcription. For example, supervillin, an actin-binding protein, is able to interact with the AR NTD and DBD-LBD (160). This association is enhanced in the presence of androgens. Supervillin increases AR transactivation and cooperates with other AR coregulators, such as ARA55 or ARA70. Moreover, three different actin isoforms cooperate with supervillin to stimulate further AR transactivation in an additive manner (161). Conversely, an actin chelator that reduces the availability of monomer actin attenuates the coactivator properties of supervillin (161). Supervillin has no effect on cytoplasmic-nuclear translocation of the AR, nor does it affect the half-life of the AR (160). Similar to their effect on supervillin, androgens increase the interaction between the AR DBD-LBD and gelsolin, another actin-binding protein, in a dose-dependent manner. Gelsolin interacts with the AR during nuclear translocation and enhances ligand-dependent AR activity (162). In addition, the F-actin cross-linking protein filamin, which was originally identified as a protein that facilitates nuclear transport of the AR, interacts with the AR DBD-LBD in a ligand-independent manner. However, this interaction is enhanced in the presence of androgens. The absence of filamin hampers androgen-induced AR transactivation (163). A second, smaller 90- to 100-kDa (instead of the 280-kDa form) fragment of filamin (termed filamin A) is capable of nuclear translocation and colocalizes with the AR to the nucleus (164, 165). This naturally occurring filamin fragment represses AR transactivation and disrupts AR interdomain interactions as well as hampers SRC-2-activated AR function. Another cytoskeletal protein, -actinin-2, enhances the transactivation activity of SRC-2 and serves as a primary coactivator for the AR, acting in synergy with SRC-2 to increase AR transactivation function (166). -Actinin-4 also binds to the AR and exhibits coregulating properties (167). Finally, the actin binding protein transgelin suppresses AR function via interruption of AR-ARA54 heterodimerization, resulting in the cytoplasmic retention of the AR and ARA54. Transgelin does not directly interact with the AR but exerts its effects through recruitment to ARA54 (168).

Like actin, microtubules constitute a principal component of the cytoskeleton and have been proposed to play an important role in nuclear receptor function. In the case of AR-mediated transcription, ARA67/PAT1/APPBP, which shows homology to kinesin light chain and binds microtubules, interacts with the AR NTD. ARA67/PAT1 functions as a corepressor for the AR. Interruption of AR cytoplasmic-nuclear shuttling may play a major role in ARA67/PAT1-mediated suppression of AR activity (169).

10. Proteins involved in endocytosis.
Endocytosis mediates the selective uptake of specific macromolecules into the cell. The best-characterized form of endocytosis is that mediated by the budding of clathrin-coated vesicles from specialized regions of the plasma membrane. Clathrin-coated vesicles fuse with endosomes, and the content of the vesicle is sorted for intracellular transport or recycled to the plasma membrane (170). Endocytosis can also be achieved in a clathrin-independent manner by uptake of molecules in small invaginations of the plasma membrane termed caveolae, which are coated with caveolin (171). Interestingly, multiple proteins involved in both mechanisms of endocytosis have been shown to interact with the AR and to influence AR-mediated transcription. One of the adaptor proteins involved in clathrin-mediated uptake, huntingtin-interacting protein 1 (HIP1), associates with the AR and reduces the rate of AR protein degradation (172). Moreover, HIP1 is recruited to AREs upon androgen stimulation. Overexpression of HIP1 enhances AR-mediated transcription. Conversely, transcription by the AR is significantly repressed after knockdown of HIP1 expression. Androgen stimulation results in nuclear translocation of HIP1, an event that relies on a nuclear localization signal at the COOH terminus of HIP1 (172). Remarkably, another endocytic protein termed APPL (adapter protein containing PH domain, PTB domain, and leucine zipper motif), that translocates to the nucleus upon growth factor stimulation, down-regulates AR-mediated transcription in a dose-dependent manner. Interaction between the AR and APPL is mediated by Akt (173). In addition, HAP1 (huntingtin-associated protein 1), which functions in endocytosis of membrane receptors and endosomal trafficking, interacts with the AR through its LBD (174). This interaction is dependent on the length of the AR polyglutamine stretch (stronger with increasing length), and the addition of androgens diminishes the strength of this association. Cyclin G-associated kinase (GAK), or auxilin 2, is an essential cofactor for Hsp70-dependent uncoating of clathrin-coated vesicles. GAK interacts with the AR NTD and enhances the AF-1 function of AR activity in a ligand-dependent manner (175). Caveolin is a principal component of caveolae membranes that serve as a scaffold protein of many signal transduction pathways. Caveolin-1 ligand-dependently interacts with the AR NTD and LBD (176). Overexpression of caveolin-1 significantly increases nuclear localization of the AR and potentiates ligand-dependent AR activation (177). In contrast, down-regulation of caveolin-1 expression diminishes androgen-induced AR-mediated transcription (176).

Overall, these findings support the concept that several proteins involved in different aspects of endocytosis exert AR coregulatory characteristics.

11. Signal integrators and transducers, scaffolds, and adaptors.
In line with the AR coregulator properties of endocytic scaffold proteins such as caveolin-1, several proteins involved in cell-cell contacts and cell-substrate adhesion complexes are able to interact with the AR and regulate its activity. This is the case also for a number of proteins that function as scaffolds and adaptors for multiple signal transduction pathways. Among these are LIN-11, ISL-1, and MEC-3 (LIM) domain-containing proteins that can associate with focal adhesions, such as ARA55/Hic, paxillin and four-and-a-half-LIM-only protein 2 (FHL2). ARA55 binds to the AR-LBD in a ligand-dependent manner and relies on three LIM domains in its C-terminal half for this interaction. ARA55 enhances transcriptional activity of the AR (178). Interestingly, the focal adhesion kinase (FAK) Pyk2 is able to repress AR transactivation by interacting with and phosphorylating ARA55 (179). The ARA55-related protein paxillin also localizes within focal adhesions and can participate in a number of signal transduction pathways. Paxillin directly interacts with the AR, and overexpression of paxillin results in increased targeting of the AR to the nuclear matrix (180). Paxillin functions as a coactivator for the AR. Moreover, like ARA55, paxillin relies on its COOH-terminal LIM domain to interact with the AR. The four-and-a-half LIM domain protein FHL2, which also functions at focal adhesions as well as in the nucleus, directly associates with full-length AR and stimulates AR-mediated transcription in an agonist- and AF-2-dependent manner (181). FHL2 has been found to interact with PELP1/MNAR (proline-, glutamic acid-, and leucine-rich protein-1/modulator of nongenomic actions of the estrogen receptor), which serves as a scaffolding protein that couples nuclear receptors with various signaling complexes (182). The PELP/MNAR interactome harbors transcriptional regulators, chromatin regulators, splicing factors, cell cycle proteins, cytoskeletal regulators, and proteins involved in nongenomic signaling. PELP1/MNAR interacts with the AR and enhances FHL2-mediated AR transactivation. Conversely, knockdown of PELP1/MNAR reduces FHL2-induced AR transactivation (182). Thus, PELP1 functions as a molecular adaptor, coupling FHL2 with the AR. Another focal adhesion protein, vinexin , has been identified as a vinculin-binding protein that plays a key role in cell spreading and cytoskeletal organization. The AR binds to vinexin , and the ligand-induced transactivation function of the AR is stimulated by vinexin (183).

In addition to focal adhesion components, several proteins involved in G-coupled receptor signaling affect AR-induced transcription and/or interact with the AR. The Rho GTPase guanine nucleotide exchange factor (GEF) Vav3 activates Rho family GTPases by promoting the exchange of GDP for GTP. Vav3 potentiates AR transcriptional activity (184, 185). In contrast, knockdown of Vav3 results in decreased AR transactivation. The increase in AR activity by Vav3 involves AF-1 of the AR. However, Vav3 does not interact with the AR, nor does it increase AR levels. Whether its GEF function is required for the stimulation of AR transactivation is still under debate (184, 185). Rho guanine nucleotide dissociation inhibitor (Rho GDI) was originally identified as a negative regulator of the Rho family of GTP binding proteins. Overexpression of Rho GDI increases AR transcriptional activation, suggesting an AR coactivator role (186). Physical interaction between Rho GDI and the AR has not been assessed. Interplay between activated Cdc42-associated tyrosine kinase Ack1 and the AR has also been described. Ack1 binds the AR and phosphorylates several tyrosine residues in its NTD (187). Activated Ack1 is recuited to AREs and promotes induction of AR target gene expression in both the presence and absence of androgens. Moreover, knockdown of Ack1 decreases androgen-stimulated recruitment of the AR to AREs in target genes, suggesting that Ack1 is required for optimal androgen-regulated DNA binding of the AR (187). Protein kinase C-related kinase (PRK) 1/protein kinase N is a member of the protein kinase C (PKC) superfamily of serine/threonine kinases and is one of the first identified effectors for RhoA GTPases. The AR interacts with PRK1 through the TAU 5 domain (188). Blocking of endogenous PRK signaling severely impairs agonist-dependent AR transactivation. Similar effects are seen for the related PRK2. Conversely, stimulation of the PRK signaling cascade results in a ligand-dependent superactivation of the AR. Furthermore, PRK1 promotes a functional complex of AR with the coactivator SRC-2 (188). RanBPM (Ran-binding protein in the microtubule-organizing center) was originally identified by its interaction with the small Ras-like GTPase Ran. RanBPM interacts directly with the AR-NTD and DBD in the presence of ligand and enhances androgen-dependent transcription by the AR (189). RanBPM may also play a role in Ran-dependent nuclear transport. It is noteworthy that Ran/ARA24 interacts with the AR NTD and acts as a coactivator for the AR (190), suggesting that RanBPM, Ran, and the AR work in a multiprotein complex.

The p21-activated kinase (PAK6) that contains a putative amino-terminal Cdc42/Rac interactive binding motif and a carboxyl-terminal kinase domain interacts with either the AR-hinge region or LBD (191, 192, 193). In contrast to most PAKs, PAK6 activity is not stimulated by Cdc42 or Rac, but can be stimulated by AR binding. In response to androgens, PAK6 cotranslocates into the nucleus with the AR and represses AR-mediated transcription. This suppression requires its kinase activity but does not depend upon GTPase binding to PAK6 and is not mimicked by the closely related PAK1 and PAK4 isoforms. Active PAK6 inhibits nuclear translocation of the stimulated AR, suggesting a possible mechanism for inhibition of AR responsiveness (191, 192, 193). Interaction with PAK6 could provide a mechanism for the AR to cross-talk with other signal transduction pathways. In keeping with this concept, the adaptor/scaffolding protein receptor for activated C kinase 1 (RACK1) interacts with the AR through its LBD (194). RACK1 facilitates ligand-independent AR nuclear translocation upon PKC activation and suppresses both ligand-dependent and -independent AR transactivation through PKC activation. ChIP assays reveal a decrease in AR recruitment to AR target genes after stimulation of PKC. These observations support a role for RACK1 as a scaffold for the association and modification of the AR by PKC, enabling translocation of the AR to the nucleus but rendering the AR unable to activate transcription of its target genes (194). An independent study confirmed the repressive effect of RACK1 on androgen-dependent gene expression and showed that androgens can enhance the association between RACK1 and the AR (195). Moreover, RACK1 facilitates the interaction between the AR and Src kinase, which results in increased tyrosine phosphorylation of the AR (195). It is noteworthy that tyrosine phosphorylation by Src is important for AR nuclear translocation (196).

In line with the concept that adaptors for multiple signal transduction pathways can modulate AR-driven transcription, several effectors of such cascades are able to interact with the AR and positively or negatively alter the transcriptional events it mediates. For instance, signal transducer and activator of transcription (STAT)-3 can bind the AR and enhance AR transactivation (125, 197, 198, 199). STAT3 has been described to stimulate the transcriptional activity of the AR in a hormone-dependent manner acting synergistically with SRC-1, P/CAF, CBP, and SRC-2 (197). Conversely, DHT-induced AR activity is increased by IL-6, a cytokine activating and signaling through STAT3 (198). In addition, association of the AR with STAT3 enhances the activity of STAT3 (199). AR activation overcomes the inhibitory effect on STAT3-mediated transcription by PIAS3. The AR relieves STAT3 from STAT3-PIAS3 complex formation (199). Similarly, Smad3, a key component in the TGF β signaling cascade modulates AR-mediated transcription (200, 201, 202, 203). Depending on the experimental setting and the target gene studied, Smad3 can act as an AR coactivator or corepressor. Protein-protein interaction between AR and Smad3 involves the AR DBD-LBD region. Moreover, ligand-bound AR inhibits TGF-β transcriptional responses through selectively repressing Smad3 signaling (203).

Ligand-enhanced binding of Ebp1, an ErbB-3 binding protein, to the AR NTD suggests a link also between ErbB receptor and AR signaling (204). Indeed, ectopic expression of Ebp1 inhibits ligand-mediated transcriptional activation of AR target genes (204, 205, 206). Ebp1 participates in the transcriptional regulation by the AR via its interaction with the corepressors HDAC and Sin3 (207).

Supporting the possibility for cross-talk between Notch and androgen-signaling pathways, Hairy/Enhancer of split related with YRPW motif 1 (Hey1), a member of the basic helix-loop-helix-orange family of transcriptional repressors that mediate Notch signaling, interacts with the AR in a ligand-independent manner (208). Androgen-dependent AR transcriptional activity is inhibited by Hey1, and expression of a constitutively active form of Notch represses transactivation by the AR. It is noteworthy that Hey2, another member of the Hey family, is also able to repress AR transcription. The inhibiting effects of Hey1 on AR transactivation are mediated by AF-1 (208). Similarly, the AR (through its LBD) and the interferon-activated RNase L interact in a ligand-dependent manner (209). In addition, overexpression of RNase L in the presence of interferon reduces androgen-mediated AR transcriptional activity. Conversely, androgens are able to interfere with induction of gene expression by interferon, indicating a functional cross-talk between DHT and interferon signaling (209). Finally, the Wnt signaling pathway also modulates androgen signaling at multiple levels. Androgens promote the cytoplasmic-nuclear translocation of beta-catenin, a critical component of this signaling pathway. Moreover, beta-catenin is able to bind to the AR LBD in a ligand-dependent manner, is recruited to AREs in AR target genes, and enhances transcriptional activity by the AR (210, 211, 212, 213, 214). Furthermore, glycogen synthase kinase-3β (GSK-3β), a protein serine/threonine kinase that regulates beta-catenin degradation, phosphorylates and interacts with the AR and suppresses its ability to activate transcription (215, 216). In contrast, some reports mention a GSK-3β-mediated increase in AR transactivation (217). Furthermore, T cell factor 4, one of the targets of Wnt signaling that relies on beta-catenin for transcriptional activation, interacts with the AR DBD and functions as a corepressor for the AR (214, 218).

12. Cell cycle regulators.
Interestingly, several proteins involved in the regulation of cell cycle progression also associate with the AR and modulate its transcriptional activity. For example, cyclin E increases the transactivation activity of the AR in the presence of DHT. Cyclin E binds directly to the NTD of the AR and enhances its AF-1 transactivation function. Interaction with the AR does not require complex formation with CDK2, nor does it involve phosphorylation of the AR (219). Cdc25B is a cell cycle regulator that functions as a dual-specific phosphatase to mediate cell cycle progression by activating the cyclin-dependent kinases. Cdc25B interacts directly with the AR and stimulates AR-dependent transcription independent of its protein phosphatase activity (220, 221, 222). Moreover, P/CAF and CBP interact and synergize with Cdc25B and further enhance its coactivation activity (220). Furthermore, cyclin-dependent kinase 6 (CDK6) binds to and is activated by cyclin D1 and as such enhances the transition of cells through the G₁ phase of the cell cycle. CDK6 also associates with the AR and stimulates its transcriptional activity in the presence of DHT. This effect does not require its kinase activity and is inhibited by cyclin D1 and p16INK4a (223). Moreover, CDK6 is present in association with the AR at the promoter region of AR target genes. Contrary to the coactivator properties exhibited by these cell cycle regulators, cyclin D1 functions as a corepressor for the AR (224, 225, 226, 227, 228, 229). Ligand-mediated transcriptional activation of AR target genes is inhibited by cyclin D1 as well as cyclin D3 (224). This function of cyclin D1 is independent of its role in cell cycle progression. Cyclin D1 directly binds to the AR in a ligand-independent manner. Cyclin D1 targets the AF-1 transactivation function. Corepressor activity of cyclin D1 can be explained by its ability to recruit HDACs and its inhibition of AR N- and C-terminal interactions (224, 225, 226, 227, 228, 229). A second isoform of cyclin D1, termed cyclin D1b, is compromised in its ability to regulate AR activity, although it retains the ability to associate with the AR (230). The retinoblastoma protein (Rb) functions as a tumor suppressor by controlling progression through the cell cycle. Rb has been described to bind to the AR in an androgen-independent manner and enhances AR transcriptional activity in the presence of DHT (231, 232). Rb and ARA70 cooperate with each other to activate transcription by the AR (231). Interestingly, pp32, which interacts with Rb, also functions as an AR coactivator (233). The presence of Rb modulates this activity. The splicing factors p54nrb and PSF are additional components of the pp32-Rb complex (233). Furthermore, Rb-associated Krüppel protein (RbaK) interacts with the AR LBD in a ligand-dependent manner (234). Finally, apoptosis-antagonizing transcription factor (AATF), also termed Che-1, which has been shown to bind Rb and promote cell cycle progression, enhances AR-mediated transactivation in a hormone- and dose-dependent manner and acts as cooperative coactivator for TSG101 (115).

13. Regulators of apoptosis.
In addition to proteins that govern progression through the cell cycle, a few proteins with prominent roles in the execution of apoptosis act as AR coregulators. For example, the proapoptotic caspase-8 represses AR-dependent gene expression (236). It does this by disrupting AR N/C interaction and inhibiting androgen-induced AR nuclear localization. Noteworthy is the fact that caspase-8 does not depend on its apoptotic protease activity to exert these functions. Caspase-8 interacts directly with the AR NTD FXXLF and WXXLF sequences, and mutations of these AR motifs prevent its repressive effect on AR-mediated transcription. In addition, knockdown of caspase-8 by RNA interference specifically affects the androgen-dependent expression of AR-targeting genes (236). Par-4, another proapoptotic protein, on the other hand, acts as an AR coactivator (237). Par-4 physically interacts with the AR DBD, is recruited to the promoter of an AR-driven gene in the presence of androgens, enhances association of the AR with DNA, and increases AR-dependent transcription. Androgen-induction of this gene is counteracted by a dominant-negative form of par-4 (237).

14. Viral oncoproteins.
The human papillomavirus (HPV) E2 oncoprotein has been suggested to act as an AR coactivator by physical and functional interactions with the AR as well as the AR-associated coactivators SRC-2 and Zac1 (zinc-finger protein which regulates apoptosis and cell cycle arrest 1). SRC-2 and Zac1 are both able to act synergistically with HPV E2 proteins on AR-dependent transcriptional activation (238, 239). Similarly, HPV E6 and E7 are able to interact directly with the AR in the absence and presence of androgens (240). Depending on the cell type and the promoter context, these oncoproteins display AR coactivator or corepressor properties. Interestingly, also the hepatitis B virus nonstructural protein x (HBx) can enhance AR activity (241, 242). HBx dose-dependently increases androgen-stimulated AR-mediated transcription. HBx does not physically associate with ligand-bound AR in the nucleus, and it likely augments AR activity by increasing the phosphorylation of the AR through HBx-mediated activation of the c-Src kinase signaling pathway (242).

15. Other functionally diverse proteins
a. Nuclear receptor coregulators.
A number of proteins that have been identified as AR coregulators cannot be readily classified into the categories listed above. Some of these have been reported to either positively or negatively regulate ligand-dependent transcription by nuclear receptors. General nuclear receptor coactivators that have been shown to enhance androgen-dependent transcription by the AR include Asc-1 (activating signal cointegrator-1) (243), Asc-2 (244), components of the Trap/Mediator complex (63, 245), CoCoA (coiled-coiled coactivator) (246), NRIP (nuclear receptor interaction protein) (247), PNRC (proline-rich nuclear receptor coregulatory protein) (248), TIF-1 (249), MRF1 (modulator recognition factor 1) (250), PDIP1 (PPAR-DBD-interacting protein 1) (251), Zac1 (252), GT198 (253), and ARA70 (254). Corepressors that negatively affect transactivation of the AR as well as other nuclear receptors comprise Alien (255), AES (aminoterminal enhancer of split) (256), components of the SMRT and NCoR repressor complexes (17, 257), RIP