Phosphorylation of the Androgen Receptor

Abstract

The androgen receptor (AR) is a ligand-regulated transcription factor that belongs to the family of nuclear receptors. In addition to regulation by steroid, the AR is also regulated by post-translational modifications generated by signal transduction pathways. Thus, the AR functions not only as transcription factor, but also as a node that integrates multiple extracellular signals. The importance of the AR in disease cannot be understated as it plays a role in many diseases ranging from complete androgen insensitivity syndrome to spinal bulbar muscular atrophy to prostate and breast cancer. In the case of prostate cancer, dependence on AR signaling has been exploited for therapeutic intervention for decades. However, the effectiveness of these therapies is limited in advanced disease due to restoration of AR signaling. Fully understanding the molecular mechanisms of AR action will enable the development of future therapeutics for the wide range of AR dependent diseases. The AR is subject to regulation by a number of kinases through post-translational modifications on serine, threonine and tyrosine residues. Here we review the AR phosphorylation sites, the kinases responsible for these phosphorylations, as well as the biological context and the functional consequences of these phosphorylations. Finally, what is known about the state of AR phosphorylation in clinical samples is discussed.

Introduction

The AR consists of an N-terminal transcriptional regulatory domain (AF-1) that can function in the absence of ligand, a DNA binding domain, a hinge region, and a C-terminal ligand-binding domain that also is associated with a second transcriptional regulatory function (AF-2). The N-terminal and C-terminal domains are involved in homotypic dimerization and binding to other transcriptional regulatory proteins. When not bound to ligand the AR is sequestered with chaperones and is not concentrated in the nucleus. Following ligand binding, a nuclear import signal is exposed and the receptor becomes concentrated in the nucleus where it binds DNA, homodimerizes in a reaction that involves interactions between the N- and C-termini, and interacts with a constellation of transcriptional coregulators, transcription factors, and components of the basal transcription machinery (Langley et al. 1995; Bubulya et al. 1996, 2000, 2001; Doesburg et al. 1997; Aarnisalo et al. 1998; Tillman et al. 1998; Wise et al. 1998; Fronsdal et al. 1998; Nessler-Menardi et al. 2000; Reutens et al. 2001; Wang et al. 2001; Gelmann 2002; Heinlein & Chang 2002; Kotaja et al. 2002a, b; McKenna & O’Malley 2002a, b; Comuzzi et al. 2003). AR phosphorylation is commonly believed to alter AR activity by modifying the protein interactions described in general above.

The AR is phosphorylated on serine, threonine, and tyrosine residues and there are phosphorylations in each of the major protein domains (Figure 1). The N-terminal domain which encompasses the AF1 region, contains the majority of phosphosites: S16, S81, S94, S213, Y223, S256, Y267, T282, S293, S308, Y363, S424, S515, and Y534. In the DNA binding domain S578 is phosphorylated, in the hinge region S650 is phosphorylated, and in the ligand-binding domain, which contains the AF2 domain, S791 and T850 are phosphorylated. Furthermore, under select conditions the AR is also phosphorylated on S405 and Y551/552. Below we describe what is generally known about the major AR phosphorylation sites.

Figure 1.

(A) Schematic of the AR phosphorylation sites with the major functional domains is shown. NTD = N-terminal domain, DBD = DNA biding domain, LBD – Ligand binding domain, AF1 – Activation function 1, AF2 – Activation function 2. (B) Overview of the AR phosphorylations, their corresponding kinases and functional effects. Only those phosphorylation sites with an identified function are listed. Specific details are in the text. The amino acid numbering used in this review is based on NCB Accession No. AAA51729 (Lubahn et al. 1988).

S81 phosphorylation

The highest stoichiometric phosphorylation on the AR in response to hormone is S81 with phosphorylation on S81 increasing over 8 hours in androgen treated prostate cancer cells (Gioeli et al. 2002). Multiple kinases, all from the cyclin-dependent kinase family, have been reported to phosphorylate the AR on S81(Chen et al. 2006, 2012; Gordon et al. 2010; Hsu et al. 2011; Gioeli & Paschal 2012). Studies suggest that CDK1, CDK5, and CDK9 can phosphorylate the AR on S81 (Chen et al. 2006; Gordon et al. 2010; Hsu et al. 2011). Overexpression of CDK1 and AR in 293T cells increased both AR S81 phosphorylation and total AR protein levels, whereas treating both LNCaP cells and AR-transfected HeLa cells with the pan-CDK inhibitor roscovitine decreased S81 phosphorylation and total AR protein levels. Two other CDK inhibitors with overlapping selectivity with roscovitine inhibited S81 phosphorylation and total AR levels further suggesting that CDK1 activity regulates AR S81 phosphorylation in cells (Chen et al. 2006). S81 phosphorylation is higher in nocodozole arrested cells, suggesting an increase in S81 phosphorylation in G2/M when CDK1 is active (Chen et al. 2012). The proteasome inhibitor, MG132, could reverse the decrease in AR protein resulting from CDK1 inhibition although the S81A mutant was stabilized by over-expression of activated CDK1 similar to wild-type receptor suggesting that this phospho-site is not necessary for the CDK1-mediated stabilization of the AR. CDK9 is also an AR S81 kinase (Gordon et al. 2010). CDK9 activity does not fluctuate with the cell cycle, rather it associates with cyclin T to form the multiprotein complex P-TEFb, which phosphorylates multiple substrates necessary for transcriptional elongation. CDK9 and cyclin T overexpression increased S81 phosphorylation levels in cells. Conversely, siRNA knockdown of CDK9 in LNCaP prostate cancer cells decreased hormone induced S81 phosphorylation. Treatment of LNCaP cells with two different CDK9 inhibitors 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole and flavopiridol inhibited S81 phosphorylation and AR-mediated transcription of endogenous genes (Gordon et al. 2010). CDK9 inhibition also impaired AR recruitment to PSA and TMPRSS2 enhancers (Chen et al. 2012). Furthermore, androgen stimulated S81 phosphorylation in chemically arrested G1 and S phase cells when CDK1 is inactive, consistent with CDK9 phosphorylating the AR on S81.

Another kinase reported to phosphorylate the AR on S81 is CDK5 (Hsu et al. 2011). CDK5 regulates neuronal development and differentiation and has multiple substrates including proteins regulating the neuronal cell cytoskeleton and exo- and endocytosis (Dhavan & Tsai 2001). Now there are data indicating CDK5 expression in prostate cancer patient tumors correlates with AR expression (Hsu et al. 2011). In this study, CDK5 associates with the AR by co-immunoprecipitation and phosphorylates the AR on S81 in vitro. However, this result is not consistent with other studies that tested CDK5 as an AR S81 kinase (Chen et al. 2006; Gordon et al. 2010). CDK5 overexpression increased AR levels and AR S81 phosphorylation, whereas CDK5 knockdown with siRNA decreased AR levels and S81 phosphorylation. This increase in AR levels was paralleled by a decrease in AR ubiquitination. Interestingly, the AR S81A mutant was heavily ubiquitinated when proteasome-dependent degradation was blocked with MG132 (Hsu et al. 2011). These latter data are consistent with the earlier results demonstrating that proteasome inhibition could block AR degradation resulting from CDK inhibition (Chen et al. 2006). The data linking AR S81 phosphorylation to protein stability is one potential insight into the functional consequence of this phosphorylation event and may explain the correlation with AR S81 phosphorylation and total AR protein levels and CDK1 or CDK5 expression and activity (Chen et al. 2006; Hsu et al. 2011). Phosphorylation on S81 following hormone stimulation may also explain why AR protein levels increase in response to androgen, which is contrary to the hormone induced protein degradation of other steroid receptors (Gelmann 2002).

The functional consequence of S81 phosphorylation on AR transcriptional activity has been examined using a variety of methods. Analysis of S81 phosphosite mutants using reporter assays failed to show significant effects (Chen et al. 2006, 2012; Gioeli & Paschal 2012) although pharmacologic inhibition of CDK activity did decrease AR transcription of PSA and TMPRSS2 reporters (Chen et al. 2006) suggesting that altering CDK activity may alter AR transcription through effects on CDK regulated cell cycle progression. Examination of gene transcription and chromatin binding on endogenous genes did uncover a function for AR S81 phosphorylation in regulating AR transactivation (Gordon et al. 2010; Chen et al. 2012). A survey of eleven known androgen-regulated genes in AR transfected LHS cells revealed gene-specific changes when S81 in AR was substituted with alanine (Gordon et al. 2010). The majority of AR target genes did not change or showed increased expression; however TMPRSS2 and ORM1 showed a decrease in transcription when comparing the S81A mutant to wild-type AR. These data suggest that S81 phosphorylation is required for regulating the endogenous AR transcriptional program. Another study used a “ligand switch” strategy, where the W741C mutation in the LBD is introduced effectively turning bicalutamide into an AR agonist to assess the effect of AR S81 phosphorylation (Chen et al. 2012). Using this system, loss of S81 phosphorylation also led to a decrease in AR transactivation of endogenous PSA and TMPRSS2 in LNCaP cells.

Multiple studies have demonstrated a role of AR S81 phosphorylation in nuclear localization. Using the ligand switch system, immunofluorescence and biochemical analysis using progressive salt extraction of fractionated cells expressing wild-type, S81A and S81D in the W741C background demonstrated that S81 phosphorylation is required for AR nuclear retention and association with chromatin (Chen et al. 2012). The control W741C mutant and the W741C-S81D mutant show nuclear localization in the absence of bicalutamide and tight association with the insoluble fraction that contains histones. In contrast, the W741C-S81A mutant is excluded from the nucleus at steady state in the absence of bicalutamide and is absent from the insoluble fraction even when bicalutamide is present. Similar observations were made when S81A AR was exogenously expressed in PC3 cells (Hsu et al. 2011). Dephosphorylation of S81 can also lead to loss of AR nuclear translocation and transcriptional activity (Yang et al. 2007; Wu et al. 2013). Studies have demonstrated that PP2A can bind and dephosphorylate the AR on multiple residues, including S81 (Yang et al. 2007). DAB2IP, a tumor suppressor with GTPase activity, regulates AR S81 phosphorylation in a PP2 dependent manner (Wu et al. 2013) and loss of DAB2IP in mice leads to PIN and increased AR nuclear localization, which is consistent with S81 phosphorylation regulating AR nuclear localization.

ChIP assays of the PSA and TMPRSS2 genes demonstrate that the non-phosphorylatable S81A mutant fails to induce chromatin binding in response to hormone (Chen et al. 2012). Importantly, ChIP experiments of endogenous AR in LNCaP cells showed that phosphorylated S81 AR binds to chromatin. However, phosphorylation on this site is not required for chromatin binding; sequential rounds of ChIP suggest that the non-phosphorylated S81 AR also binds to AREs.

The effect of the S81 phosphorylation on growth was tested by stable expression of wild-type and S81A AR mutant in LHS and LAPC4 cells (Gordon et al. 2010). LHS and LAPC4 cells with increased wild-type AR expression grew faster than parental cells and cells expressing the S81A AR mutant, indicating that loss of Ser 81 phosphorylation limits cell growth. Similar results were obtained when AR was overexpressed in CWR22Rv1 cells; the S81A mutant decreased growth relative to wild-type AR (Hsu et al. 2011).

All of these studies follow a similar general paradigm using overexpression, RNAi knockdown, S/A mutant, and pharmacologic inhibitors to interrogate the effect of different CDKs on AR S81 phosphorylation and AR function. Confounding variables include but are not limited to differences in cell and AR expression systems used as well as overlapping selectivity of the pharmacologic inhibitors and off target effects of RNAi approaches. We are left to integrate these studies into a cohesive model that accounts for multiple kinases phosphorylating a single site (Figure 2). Since S81 is the highest stoichiometric site and loss of S81 phosphorylation inhibits cell growth, there may be a need to have multiple kinases phosphorylate the AR on this site. During transcription, the proximity of P-TEFb to the AR suggests that CDK9 phosphorylates the AR during AR transactivation. This is supported by finding S81 phosphorylated AR on chromatin within 15 minutes of androgen stimulation (Chen et al. 2012). The role of CDK1 phosphorylating S81 is more elusive since cells silence most transcription in G2/M; however, some genes remain transcriptionally active during mitosis. Thus, it is plausible that CDK1 phosphorylation of S81 may regulate AR activity on a subset of genes as was observed with the S81A mutant in one study (Gordon et al. 2010). One attractive hypothesis is that CDK1 phosphorylation on S81 is critical for the initiation of gene transcription following mitosis. What role CDK5 plays in regulating S81 phosphorylation in the context of CDK1 and CDK9 phosphorylation of this site is unclear. One hypothesis suggested is that CDK5 phosphorylation of AR S81 causes AR nuclear localization and stabilization (Hsu et al. 2011). Clearly further studies are required to determine the biological consequence of S81 by the different CDKs.

Figure 2. AR S81 phosphorylation regulates AR transcriptional activity, nuclear retention, protein levels and growth.

S81 is phosphorylated by CDK1, CDK5, and CDK9 leading to increased AR transcriptional activity, AR retention in the nucleus, and cell growth. The lack of AR phosphorylation on S81 leads to cytoplasmic localization, AR ubiquitination, and AR degradation.

S213, S791, and T850 phosphorylation

Studies suggest that AR S213 is likely the target of several kinases (Figure 3). At least two kinases, Akt and PIM-1, are known to phosphorylate S213 with different functional consequences (Wen et al. 2000; Lin et al. 2001; Lin et al. 2003; Taneja et al. 2005; Palazzolo et al. 2007; Ha et al. 2012; Kasina & Macoska 2012; Linn et al. 2012; Varisli et al. 2012). Moreover, Akt-dependent AR S213 phosphorylation is regulated by several signaling pathways (Wen et al. 2000; Kasina & Macoska 2012; Varisli et al. 2012). Different functional consequences including effects on AR stability, AR transactivation, and prostate cancer cell growth depending on the context of regulation were reported for the AR S213 phosphorylation.

Figure 3. AR S213 phosphorylation results in different functional outcomes including effects on AR transactivation, AR stability, and prostate cancer cell growth and survival.

S213 phosphorylation regulates AR transactivation, stability, and nuclear translocation in a cell context-dependent manner. This schematic does not discriminate between PIM-1 isoforms. PIM-1S, but not PIM-1L, increases AR degradation by promoting AR association with Mdm2 in an S213-dependent manner. PIM-1S up-regulates transcription of proliferative genes and downregulate transcription of the AR-regulated differentiation genes.

Several studies demonstrated that Akt directly binds and phosphorylates the AR on serines 213 and 791 (Wen et al. 2000; Lin et al. 2001, 2003; Taneja et al. 2005). Co-immunoprecipitation experiments revealed that activated Akt specifically associated with endogenous AR in LNCaP cells (Wen et al. 2000; Lin et al. 2001). Furthermore, Akt was able to phosphorylate the AR on serines 213 and 791 in an in vitro kinase assay (Wen et al. 2000).

Studies on the effect of the Akt-mediated AR serines 213 and 791 phosphorylation on AR transcription yielded conflicting results demonstrating either repression (Lin et al. 2001; Taneja et al. 2005; Palazzolo et al. 2007) or activation of AR function (Wen et al. 2000). Akt repressed AR transactivation in a reporter assay in AR negative castration resistant prostate cancer DU145 cells when the AR was exogenously expressed (Lin et al. 2001). In transfected cells, AR phosphomimetic mutants at the Akt consensus sites repressed transactivation of a reporter in response to DHT, decreased hormone-mediated AR protein stabilization, and abrogated nuclear translocation (Palazzolo et al. 2007). However, in the androgen-dependent prostate cancer LNCaP cell line, Akt phosphorylation activated the PSA reporter and promoted cell survival (Wen et al. 2000; Lin et al. 2001) These differential effects of Akt-mediated AR S213 phosphorylation on AR function could be attributed to cell context. Indeed, Akt was shown to regulate AR activity in a cell passage number dependent manner in LNCaP cells (Lin et al. 2003). The data suggest that modifiers of AR S213 function exist in different contexts, which may be related to disease progression.

AR was also phosphorylated on S213 in response to activation of G-protein coupled receptor chemokine receptor 4 (CXCR4) by its ligand CXCL12 in an androgen-independent manner (Kasina & Macoska 2012). At least one mechanism of AR activation by CXCL12/CXCR4 may be through PI3K/Akt-dependent AR S213 phosphorylation. Experiments with pharmacological inhibitors of MAPK and PI3K pathways revealed that phosphorylation of SRC-1 by MAPK on threonine was required for ligand-independent activation of the AR and that AR S213 phosphorylation is PI3K-dependent. Moreover, treatment with Src and PKC inhibitors showed that multiple kinases are involved in CXCL12/CXCR4-mediated activation of the AR. Although CXCL12/CXCR4-mediated AR transactivation was relatively weak compared to activation by androgen, it serves as an example of how AR activity can be maintained in the absence of androgen.

Interestingly, the AR-regulated gene, hematological and neurological expressed 1 (HN1), has been shown to negatively regulate AR signaling and androgen-regulated growth of prostate cancer cells (Varisli et al. 2012). HN1 regulated AR steady state levels by inhibiting PI3K-dependent Akt phosphorylation on S473. HN1 over-expression in LNCaP cells results in AR S213 dephosphorylation and a decrease in AR levels. This decrease in AR expression could be restored by the proteosomal inhibitor MG132. The converse was also observed; HN1 knockdown resulted in increased AR protein expression and AR nuclear translocation. This work suggests a negative feedback loop between HN1 and the AR involving AR S213 phosphorylation.

AR S213 was also shown to be a target for PIM-1 kinase (Linn et al. 2012). PIM-1 has two isoforms, 44 kDa PIM-1L and 33 kDa PIM-1S (Saris et al. 1991), both of which can directly interact with the AR and phosphorylate it at different residues, namely S213 and T850. AR S213 was phosphorylated by both PIM-1 isoforms in in vitro kinase assays and when exogenously expressed in cells. However, only PIM-1L was able to phosphorylate T850 effectively (Linn et al. 2012). Data from another group supported the notion that AR is phosphorylated on S213 by PIM-1 in a ligand-independent manner. The 33 kDa isoform of PIM-1 phosphorylated the AR on S213 in cells and in vitro (Ha et al. 2012).

Studies show that PIM-1-mediated phosphorylation at S213 and T850 leads to the recruitment of different ubiquitin E3 ligases that have different effects on AR stability (Linn et al. 2012). Co-transfection of PIM-1 kinase isoforms and the AR as well as the S213A mutant into 293T cells revealed that AR phosphorylation on S213 by PIM-1S, but not PIM-1L, decreased the AR half-life, and the phospho-mimetic T850D mutant was resistant to this S213 dependent PIM-1S-induced deregulation. In chemically arrested LNCaP cells, expression of both PIM-1 isoforms was increased in the G2 and M phases of the cell cycle, which correlated with AR phosphorylation on S213 and T850 (Lin et al. 2001). Another study, however, reported that S213 did not increase during the time-course of nocodazole arrest (Chen et al. 2012). Interestingly, when endogenous AR was knocked down and wild-type AR and S213A mutant were re-expressed in LNCaP cells, the S213A mutant was more stable than the wild-type AR in cells synchronized in mitosis (Lin et al. 2001). Co-IP experiments were used to determine the differential effects of S213 and T850 phosphorylation on AR stability. PIM-1S phosphorylation on S213 enhanced the interaction of the AR with Mdm2 leading to AR proteolysis, while PIM-1L phosphorylation on T850 enhanced the interaction of the AR with RNF6, which does not cause AR destabilization (Xu et al. 2009). These data suggest that PIM-1 isoforms have differential effects on AR protein stability with the net effect depending on both S213 and T850 AR phosphorylation.

Only PIM-1L, but not the PIM-1S isoform, could enhance AR transcriptional activity in a reporter assay under low serum conditions and mutations of either S213 or T850 abrogated PIM-1L-induced increase in AR transcriptional activity. In LNCaP cells infected with lentiviruses encoding PIM-1 kinases, PIM-1L activated expression of endogenous AR target genes in a promoter context-dependent manner (Linn et al. 2012). Another group, however, showed that AR S213 phosphorylation by PIM-1S repressed AR function (Ha et al. 2012). Although the reporter assay and cell line systems used in these studies are similar, the experimental design, namely androgen concentrations, was different. In the former study, in which PIM-1S had no effect on AR transcriptional activity in a reporter assay, the experiments were performed under low androgen conditions (Linn et al. 2012), whereas in the later study using a high concentration (10 nm) of synthetic androgen R1881, PIM-1S-mediated AR S213 phosphorylation repressed AR-dependent transcription (Ha et al. 2012). Wild-type AR, but not S213A-expressing cells showed decreased transcription of an androgen-responsive reporter gene. Furthermore, in this study induction of endogenously expressed PSA and Nkx3.1 was reduced in response to hormone in LNCaP cells over-expressing PIM-1S. In addition, over-expression of PIM-1 resulted in activation of AR target genes implicated in cell proliferation including cyclin A1, IL6, and caveolin 2.

Overexpression of PIM-1 kinases in LNCaP cells resulted in an increase in cell growth under low androgen conditions (Linn et al. 2012). Furthermore, replacement of endogenous AR with the codon-switched AR S213 or T850 abrogated the increase in PIM-1-promoted cell proliferation. These experiments suggested that phosphorylation of AR on S213 and T850 could sensitize prostate cancer cells to low androgens.

T282 and S293 phosphorylation

In vitro kinase assays showed that the AR can be phosphorylated by the Serine/Threonine kinase Aurora-A (Shu et al. 2010). Sequence analysis predicted that the N-terminal T282 and S293 were two potential Aurora-A phosphorylation sites. These two sites were confirmed as Aurora-A targets in vitro and in vivo. While T282 and S293 are not induced by androgen, they are required for maximal AR transactivation in cells exposed to androgen. Co-immunoprecipitation carried out in castration resistant cells derived from LNCaP demonstrated that the AR directly associated with Aurora-A. In reporter assays using a synthetic ARE promoter, the double T282A/S293A mutant decreased the Aurora-A-induced AR transcriptional activity in the presence or absence of androgen compared to wild-type AR, suggesting that the phosphorylation of T282 and S293 was not only required for the activation of AR induced by Aurora-A but also the regulation of the response to androgen. T282/S293 phosphorylation by Aurora-A enhanced both AR DNA binding and LNCaP cell growth. These data suggest that the phosphorylation of T282 and S293 resulting from the elevated expression of Aurora-A in prostate cancer could contribute to castration resistance (Shu et al. 2010).

S308 phosphorylation

The AR is phosphorylated on S308 by CDK11^P58, the isoform of CDK11 that is specifically expressed in G2/M. Sequential co-IP using whole cell lysates from transfected cells revealed that the AR forms a ternary complex with CDK11^P58/cyclin D3 in mammalian cells. CDK11^P58 autophosphorylation is necessary for its kinase activity and kinase-dependent effect on AR function (Chi et al. 2011). CDK11^P58 was able to phosphorylate the N-terminal portion of the AR in an in vitro kinase assay. Using Ser to Ala N-terminal AR mutants as substrates in kinase assays, only S308A mutant showed reduced phosphorylation (Zong et al. 2007).

Importantly, phosphorylation of the AR on S308 by CDK11^P58 repressed its function and proliferation of prostate cancer cells. Exogenous expression of cyclin D3 and CDK11^P58 inhibited AR-dependent transcription in reporter assays in several cell lines including prostate cancer cell lines LNCaP and PC3. Furthermore, over-expression of either CDK11^P58 or cyclin D3 or the combination of these two proteins repressed PSA mRNA levels in LNCaP cells. In agreement with this data, siRNA against CDK11^P58 or cyclin D3 abrogated PSA mRNA repression. Moreover, over-expression of CDK11^P58 decreased proliferation in LNCaP cells detected by BrdU incorporation. The S308A mutant failed to repress AR transcriptional activity in reporter assays when co-transfected with cyclin D3/CDK11^P58 (Zong et al. 2007). These data suggest that cyclin D3/ CDK11^P58 represses AR transcriptional activity through phosphorylation on S308.

S405 phosphorylation

A novel AR phosphorylation site was discovered by genomic DNA sequencing of a Caucasian male with Androgen Insensitivity Syndrome (AIS), a disorder that results in the incomplete development of the external genitalia of 46,XY genetic males (Lagarde et al. 2012). This phospho-site was created by a missense mutation in exon 1 of the AR, which resulted in the substitution of arginine for serine at amino acid position 405. The R405S mutation reduced AR transcriptional activity in luciferase assays in CV1 cells expressing wild-type AR or the R405S mutant. Western blot analysis of COS-1 cells transfected with wild-type AR, R405S, or select AR subdomains in the presence or absence of phosphatase inhibitors suggested that the loss of function due to the R405S mutation was associated with the creation of the S405 phosphorylation site. Transcriptional activity in cells expressing the R405A mutant was not affected, supporting the negative effects of phospho-S405 on transcription. The R405S mutation is located in the N-terminal activation function-1 (AF1) domain, 27 residues upstream from the WXXLF motif, which is required for transcriptional activation associated with the AR coregulators primate-specific melanoma antigen-A11 (MAGE-11) and p300 (Lagarde et al. 2012). AR transactivation depends on interactions with co-regulators and the interdomain AR NH2- and carboxyl-terminal (N/C) interaction between the N-terminal FXXLF motif and ligand-binding domain AF2, which is stabilized by androgen (He et al. 2004; Heemers & Tindall 2007). MAGE-11 binds the FXXLF motif and functions synergistically with co-regulators and transcription factors to regulate the AR N/C interaction (Bai et al. 2005). Mutation of the FXXLF motif inhibits AR transcriptional activity by disrupting the AR N/C interdomain interaction. Reporter assays using full-length AR and fragments revealed that the WXXLF motif was also required for MAGE-11-induced AR transactivation since MAGE-11 failed to elicit a transcriptional response in cells expressing fragments lacking the WXXLF motif (Lagarde et al. 2012). Previous studies where the histone acetyltransferase p300 was silenced by siRNA determined that p300 was required in AR transactivation associated with the androgen-dependent AR N/C interaction (Askew et al. 2010). p300 directly associated with MAGE-11, which appeared to be a required factor in the AR transcriptional response to p300. Together, these data suggest that the deleterious effects on transcription resulting from the phosphorylation of S405 are related to the proximity of this phosphosite to the WXXLF motif, thereby interfering with the ability of MAGE-11 to recruit the coactivator p300 to enhance transcription associated with the AR N/C interaction.

S515 and S578 phosphorylation

Another AR phosphorylation site with multiple kinases is S515 with both CDK7 and MAPK reported to phosphorylate the AR on this site. There is evidence that CDK7 phosphorylates the AR on Ser 515 (Chymkowitch et al. 2011). CDK7 is an essential component of the general transcription factor TFIIH, which activates RNAPII initiating transcription (Harper & Elledge 1998). Purified recombinant TFIIH phosphorylated the AR in vitro and this phosphorylation was reduced when a S515A mutant was used as a substrate suggesting that CDK7 phosphorylates the AR on S515. The evidence for MAPK phosphorylation of S515 is indirect but, in aggregate, compelling. Following up on data showing that EGF signaling increased AR transactivation and prostate cancer cell growth (Mellinghoff et al. 2002; Gregory et al. 2004), mutational analysis of candidate AR serine and threonine phosphorylation sites led to identifying S515 as the site of AR phosphorylation in response to EGF treatment (Ponguta et al. 2008). Using mobility shifts on SDS-PAGE as a readout of phosphorylation, the S515A mutant and the MEK1/2/5 inhibitor UO126, both prevented the EGF induced mobility shift of an AR fragment consisting of amino acids 507-660 further suggesting that MAPK phosphorylates AR S515. These results were consistent with an earlier study showing that MAPK phosphorylates an N-terminal fragment of the AR containing S515 (Yeh et al. 1999). Interestingly, AR S515 phosphorylation is co-regulated with S578 phosphorylation by EGF signaling (Ponguta et al. 2008). The evidence for AR phosphorylation on S578 is from immunoreactivity with an anti-phospho-S578 antibody and in vitro kinase assays with purified catalytically active PKC subunit. Phosphorylation of the S578A mutant was reduced when compared to wild-type AR in EGF stimulated cells consistent with S578 phosphorylation by PKC. Exogenous expression of the S578A mutant led to an increase in the mobility shift, consistent with an increase in S515 phosphorylation (Ponguta et al. 2008) and indicating that phosphorylation on S515 is maximal when S578 is not phosphorylated. PAK6 is also an AR S578 kinase (Liu et al. 2013). PAK6 phosphorylation on S578 regulates AR interaction with Mdm2 and proteasomal dependent degradation. In vitro kinase assays using purified proteins and S/A mutants suggests that PAK6 can phosphorylate S578. Exogenous expression of AR and PAK in HEK293 cells revealed that PAK6 kinase activity was required for AR S578 phosphorylation and that S578 phosphorylation was required for association with Mdm2 and subsequent AR ubiquitination and degradation (Liu et al. 2013).

Loss of S515 phosphorylation reduces AR transactivation; the S515A mutant had reduced transcriptional activity in assays using a PSA reporter (Ponguta et al. 2008; Chymkowitch et al. 2011). The S578A mutant reduced AR transactivation more than the S515A mutant (Ponguta et al. 2008). The double mutant S515/578A had virtually no transcriptional activity. Since loss of S578 phosphorylation increases S515 phosphorylation, it seems that the loss of phosphorylation on S578 and subsequent effects on transcription is dominant to S515 phosphorylation. AR phosphorylation on S578 may regulate transcription through affecting AR nuclear localization and association with transcriptional coactivators. Loss of S578 (S578A) led to increased nuclear localization and, by co-IP studies, association with the Ku70/80 regulatory subunits of DNA-PK.

AR S515 phosphorylation may also regulate protein turnover (Chymkowitch et al. 2011). Prolonged AR protein turnover was observed in pulse chase experiments using xeroderma pigmentosum cells carrying a mutation in a TFIIH subunit leading to deregulation of CDK7 or cells expressing a S515A mutant AR. The increase in AR protein stability corresponded with a change in the recruitment of E3 ubiquitin ligase proteins at the PSA promoter. Wild-type AR recruits the E3 ligase MDM2 to the PSA promoter. Loss of S515 phosphorylation by mutation to alanine leads to the recruitment of the E3 ligase CHIP, which replaces MDM2. Both MDM2 and CHIP ubiquitinate the AR. However, S515 phosphorylation by CDK7 may specify which E3 ligase is used to modify the AR activity.

S650 phosphorylation

AR phosphorylation on S650 by stress kinases inhibits AR transcriptional activity by promoting its nuclear export (Gioeli et al. 2006). Evidence suggests that stress kinases MKK4/JNK and MKK6/p38 directly phosphorylate AR on S650. First, exogenous expression of MAPK kinase (MKK) 4 and c-Jun N-terminal kinase (JNK) 1 or MKK6 and p38alpha along with the AR in COS cells leads to a robust increase of AR S650 phosphorylation over basal levels. Second, treatment with pharmacological inhibitors either to p38 or JNK1 dramatically diminished endogenous AR S650 phosphorylation in LNCaP cells. Furthermore, JNK1 and p38alpha were able to directly phosphorylate AR on S650 in vitro. Importantly, stress kinases MKK4 and MKK6 negatively regulate AR-dependent transcription. Inhibiting MKK4 and MKK6 by siRNA in LNCaP cells leads to an increase in basal as well as DHT-stimulated PSA mRNA levels. The increase in PSA mRNA expression was abrogated by either the anti-androgen bicalutamide or AR siRNA demonstrating that the stress kinase-mediated effect on the PSA expression is dependent on the AR. S650 is located in the Hinge region in proximity to the DBD NES. Therefore, its role in nuclear-cytoplasmic shuttling was assessed using the heterokaryon shuttling assay. Using this assay, experiments with phospho-S650 mutants and pharmacological inhibition of p38 and JNK revealed that S650 phosphorylation was required for nuclear export and stress kinase signaling regulates export of the AR. Taken together, these data suggest that MKK4/JNK and MKK6/p38 directly phosphorylate the AR on S650 and negatively regulate AR transcription by facilitating AR nuclear export (Gioeli et al. 2006).

AR Tyrosine Phosphorylation

Multiple tyrosine residues on the AR can be phosphorylated (Figure 4) (Guo et al. 2006; Mahajan et al. 2007). Deletion analysis with truncated AR proteins, mass spectrometry, and immunoblot analysis of point mutants of the phosphorylation sites identified by an artificial neural network-based method expressed in 293T cells along with a constitutively active Ack or a kinase-dead form determined that Y267 and Y363 are phosphorylated by Ack (Mahajan et al. 2007). Various ligands can induce specific tyrosine kinases to phosphorylate distinct tyrosine residues on the AR (Liu et al. 2010). Tyr 267 is phosphorylated by Ack in response to EGF, Heregulin, and Gas6 (Mer receptor ligand). Expression of the Y267F and Y363F mutants decreased gene transcription of and DNA binding to the PSA enhancer. Furthermore, mutation of Y267 blocked the castration-resistant growth of prostate cancer xenografts, implicating Y267 as a critical phosphorylation site for castration-resistant AR transactivation and tumorigenesis. Western blotting of primary prostate cancer specimen lysates revealed increased AR tyrosine phosphorylation associated with active Ack in human castration-resistant prostate cancer (CRPC) tumors but not in normal prostate tissue (Mahajan et al. 2007). ChIP-on-chip analysis of endogenous AR in EGF-stimulated LAPC4 cells showed that the AR was specifically recruited to the ATM enhancer when Y267 was phosphorylated (Mahajan et al. 2012). In vitro reporter assays confirmed that the phosphorylation of Y267 promoted the recruitment of the AR to the ATM enhancer, and increased AR transactivation. EGF-treated LAPC4 and LNCaP cells showed a significant increase in phosphorylation of Y267, and this correlated with increased ATM mRNA and protein levels. DHT did not promote any changes in ATM mRNA levels nor DNA binding to the ATM enhancer. These events were blocked in the presence of the Ack1 inhibitor AIM-100 indicating that the phosphorylation of Y267 is required for AR binding to the ATM enhancer and the subsequent increase in ATM gene transcription. Data from clonogenic survival assays and xenograft tumor growth studies suggest that the inhibition of Y267 phosphorylation with AIM-100 resensitizes CRPC cells to radiotherapy. These data identify a novel signaling pathway where the phosphorylation of Y267 plays a critical role in CRPC resistance to radiation therapy.

Figure 4. AR tyrosine phosphorylation regulates gene transcription, growth, and protein stability.

Y223 phosphorylation by Fer results in the formation of Fer-AR complexes in the nucleus and activation of gene transcription. Ack phosphorylates Y267 and Y363 to enhance transcription of AR target genes (i.e. PSA, ATM) and cellular growth. The phosphorylation of Y534 by Src and Etk leads to increased gene transcription, cell growth, and AR stabilization.

AR Y534 was identified by mass spectrometric analysis as a major Src kinase phosphorylation site. Y534 has been shown to be phosphorylated by Src in the presence of EGF, IL-6 and Bombesin (Guo et al. 2006). Mutation of this tyrosine to phenylalanine decreased AR transcriptional activity in reporter assays in response to EGF and low doses of hormone. Expression of the AR Y534F mutant under hormone-depleted conditions in CWR22-Rv1 and LNCaP cell lines and tumor xenografts resulted in diminished growth when compared to wild-type AR expressing cells. Using phospho-specific antisera to Y534, IHC analysis showed a positive correlation between Y534 phosphorylation during prostate cancer progression and Src activity (Guo et al. 2006). Interestingly, the phosphorylation of Y534 was also strongly correlated with the non-receptor tyrosine kinase epithelial and endothelial tyrosine kinase (Etk) expression in mouse and human prostate tumors (Dai et al. 2010), suggesting that Y534 phosphorylation levels may partly be attributed to Etk. Co-transfection experiments in COS-1 cells using wild-type AR and wild-type Etk, kinase-dead Etk, or constitutively active Src as a positive control showed that wild-type Etk induced AR tyrosine phosphorylation but not to the magnitude induced by Src. Under castration conditions, the upregulation of Etk resulted in increased Y534 phosphorylation and stabilization of the AR, thereby linking Y534 phosphorylation to AR stability (Dai et al. 2010). Introducing the Y534F mutant in LNCaP cells led to increased AR degradation (DaSilva et al. 2009). Mutational studies indicate that Etk also phosphorylates Y551 and Y552 (Dai et al. 2010). Data from glutathione S-transferase pull-down experiments suggest that these phosphotyrosine residues may promote the interaction between the AR and the SH2 domain of Etk. Furthermore, the Etk-induced phosphorylation of Y534 is dependent on the phosphorylation of Y551/552. Therefore, the relative contribution that Src and Etk make to the phosphorylation of the AR may be dependent on the cell context and stimuli.

Gain and loss of function studies show that IL-6 promotes the phosphorylation of AR tyrosine residues by the tyrosine kinase Fer in LNCaP and AR-expressing PC3 cells (Rocha et al. 2013). In vitro kinase assays indicate that Fer phosphorylates Y223 in the presence of IL-6. This phosphorylation leads to the formation of Fer-AR complexes in the nucleus and enhanced PSA transcription in LNCaP and C42 cells exposed to IL-6.

In situ phospho-AR staining of human samples

There is a paucity of information on the levels of AR phosphorylation in patient samples. Few studies have been done although this is an outstanding question of significant importance in the field. The majority of AR phospho-specific antibodies have not been rigorously validated for in situ use. The selectivity of phospho-serine antibodies is notoriously problematic for studies in tissues. Phospho-tyrosine antibodies are typically more selective and thus are more likely to be suitable for in situ experiments. The first study examining phospho-AR in tissues evaluated pS213 in human fetal urogenital sinus epithelial cells (Taneja et al. 2005). When elevated androgen levels and activated AKT were both present, phospho-S213 staining was detected. In this study S213 was not phosphorylated in dividing cells, as determined by Ki-67 staining and there was no S213 phosphorylation in stromal cells although stromal cells were strongly AR positive. This latter observation is an important control, such that when considered with the selectivity of the antibody on western blot analysis of cell lysate and peptide blocking experiments, suggests that the phospho-S213 antibody used in this study is highly selective. In the adult human normal prostate, staining for phospho-S213 was heterogeneous, whereas total AR staining was more homogeneous. These results suggested that S213 phosphorylation occurs in a cell type and developmental specific manner (Taneja et al. 2005).

The first study to examine AR phosphorylation in prostate cancer evaluated AR Y534 phosphorylation (Guo et al. 2006). Y534 phosphorylation was significantly increased in castration-resistant prostate tumors as compared to hormone naïve tumors. IHC analysis of tissue micro arrays showed that the mean intensity of immunostaining for phospho-Y534 was higher in castration-resistant disease. The phospho-Y534 signal correlated with both Src and Etk levels in patient samples (Guo et al. 2006; Dai et al. 2010). The selectivity of the antibody used in this study was rigorously evaluated. Western blotting of whole cell lysates of stimulated and unstimulated prostate cancer cells, peptide-blocking as well as IHC and western blots of wild-type and the phospho-null mutant Y534F all demonstrated phospho-Y534 antibody selectivity (Guo et al. 2006).

Phosphorylation of Y267 was examined by IHC staining of a prostate TMA that was also stained for ATM, which associates with Y267 phosphorylated AR (Mahajan et al. 2012). A significant positive correlation was observed between the levels of Y267 phosphorylation and ATM expression and the severity of disease progression suggesting the clinical importance of Y267 phosphorylation (Mahajan et al. 2012). The antibody used in this latter study was validated by western blotting of lysate from agonist-stimulated cells and from cells overexpressing Ack kinase (Mahajan et al. 2010). In this study AR Y267 phosphorylation correlated with Ack Y287 phosphorylation consistent with elevated Ack activity leading to an increase in AR Y267 phosphorylation.

Other studies using an anti-phospho-S213 antibody have implicated S213 phosphorylation in prostate cancer progression. Studies have shown that phospho-S213 staining increases as prostate cancer progresses from hormone naïve to castration resistance (McCall et al. 2008; Ha et al. 2012; Zellweger et al. 2013). One of these studies describes antibody validation, using calf intestinal alkaline phosphatase treatment of parallel slides (McCall et al. 2008) and another shows agonist stimulated or inhibitor regulation of S213 phosphorylation by western blotting (Ha et al. 2012). These studies showed that phosphorylation of S213 was associated with poor overall survival (McCall et al. 2008; Ha et al. 2012; Zellweger et al. 2013) although in one study this association was restricted to cytoplasmic phospho-S213 staining (Ha et al. 2012). Nuclear AR S213 phosphorylation was significantly increased in cases with high Gleason scores even though nuclear S213 phosphorylation was not predictive of disease outcome (Ha et al. 2012). These studies in general parallel observations of total AR levels in prostate cancer progression to castration resistance (Linja et al. 2001). Overall these studies suggest that AR S213 phosphorylation increases as prostate cancer progresses to castration resistance.

AR phosphorylation has also been examined in breast cancer specimens (Ren et al. 2013). Using the previously validated anti-phospho-S213 AR antibody (Taneja et al. 2005), this study shows elevated phosphorylation of S213 in breast cancers relative to benign breast tissue (Ren et al. 2013). S213 phosphorylation was elevated in invasive ductal carcinoma, metastatic breast cancer and ER negative breast cancers. AR S650 phosphorylation was also evaluated in breast cancer using an antibody validated by western blotting of wild-type and S650A AR, PMA and hormone stimulated AR, phosphatase treated AR, and peptide blocking experiments of IHC samples (Ren et al. 2013; S. Logan, personal communication). AR S650 phosphorylation overall was decreased in breast cancers but phosphorylation of S650 was elevated in select circumstances such as ER-negative breast cancers and breast cancers with distant metastasis. This study suggests that AR phosphorylation may play a role in breast cancer development and progression, which is consistent with the emerging role of the AR breast cancer (Hickey et al. 2012).

AR phosphorylation on serines 81, 94, 308, 515, 650 and 791 in prostate cancer patient specimens has been evaluated (McCall et al. 2013; Willder et al. 2013). In these studies correlations with clinical parameters, including patient outcome, were linked to AR phosphorylation at different sites. A confounding variable in these studies is that antibody selectivity was established using peptide-blocking experiments. While this is an important step for validating antibodies for IHC, in isolation it does not demonstrate specificity (Mandell et al. 1998; Al-Dhaheri & Rowan 2006; Brumbaugh et al. 2011). The concern is that pre-absorption with phosphorylated peptide can block all immunoreactivity, including non-specific antibody staining. Nonetheless, these studies generate interesting and exciting hypotheses about the role of AR phosphorylation in prostate cancer progression, both mechanistically and prognostically.

Summary

The AR integrates input from hormone binding with the cell signaling network through post-translational modifications including phosphorylation. There is still much to learn about AR phosphorylation. For many sites the proximal kinase remains unknown and the biological context of AR phosphorylation in both normal physiology and disease remains an important unanswered question. Furthermore, little is known about the potential interplay between different AR phosphorylation sites. In addition to phosphorylation, the AR is acetylated, sumoylated, ubiquitinated, and methylated. How the multitude of AR post-translational modifications interact to regulate AR function will keep scientists busy for decades to come.