Epigenetic regulation of androgen receptor signaling in prostate cancer

Abstract

Prostate cancer is most common cancer in men in the United States, and it is the second leading cause of cancer-related death in American men. The Androgen receptor (AR), a nuclear hormone and transcription factor, is the most therapeutically relevant target in this disease. While most efforts in the clinic are still directed at lowering levels of androgens that activate AR, resistance to androgen deprivation eventually develops, and most prostate cancer deaths are attributable to this castration-resistant form of this disease. Recent work has shed light on the importance of epigenetic events including facilitation of AR signaling by histone-modifying enzymes and also on the role that enzymes such as HDAC6 play in stabilizing AR in prostate cancer cells. Herein, we summarize recent findings on the role of epigenetic enzymes in AR signaling and highlight examples on how interdiction of critical epigenetic enzymes may attenuate AR action in prostate cancer.

Introduction

Prostate cancer is the most common cancer in men in the United States. In 2008, 180,000 new cases of prostate cancer and 27,000 deaths due to this disease in the United States were predicted.¹ With rare exception, the central signaling pathway in human prostate cancer is the androgen receptor (AR).²

AR is a nuclear hormone receptor, which is activated by binding of androgen ligands. Upon androgen binding, AR dissociates from the cytoplasmic chaperone protein HSP90, self-dimerizes, and translocates to the nucleus. There, AR binds to consensus sequences in the genome called AREs (androgen response elements) to activate transcription of its target genes, which is essential for prostate development and maintenance.³

Accumulating evidence also indicates that AR signaling may be important for prostate cancer initiation and progression through the upregulation of a gene fusion of the AR-responsive gene TMPRSS2 and the ERG transcription factor, which is a member of the ETS (E-twenty six) family of transcription factors; cancers harboring these fusions co-opt AR signaling to overexpress ERG, which results in abnormal invasion and transformation of prostatic epithelial cells that may be enhanced by loss of the PTEN tumor suppressor gene.^4–6 Indeed, the presence of the TMPRSS2-ERG gene fusion is a common event in human prostate cancer, occurring in approximately 50% of cases, many of which also harbor PTEN loss.^5,6

The widely used therapies in recurrent or metastatic prostate cancer, the lethal form of this disease, involve androgen deprivation through medical or surgical castration or disruption of androgen binding to AR.² Such treatments are temporarily effective, but over time, most prostate cancers evolve into a castrate-resistant state, in which hormonal interventions are less effective.^7,8 Resistance mechanisms include AR gene amplification and mutation, generation of ligand-independent AR transcript variants, upregulation of pathways that activate AR in the absence of androgens, and persistent intratumoral androgens.^9–13 Of note, even in castrate-resistant prostate cancer, AR still plays an essential role in cancer progression.⁹ Recent work indicates that epigenetic enzymes are important co-activators of AR and may represent targets to affect AR function or stability, thus providing new therapeutic opportunities to overcome mechanisms of resistance and to target AR with non-hormonal therapies.

Epigenetics is defined as the heritable change in gene expression without change in DNA sequence.¹⁴ Among the different types of epigenetic changes, the most important and best characterized are DNA methylation and histone modifications, both of which have been shown to be important for cancer progression.¹⁴

A histone octamer, composed of two copies of histone H2A, H2B, H3 and H4, is wrapped by approximately 146 base pairs of DNA to form the core particle of a nucleosome, the fundamental structural unit of eukaryotic chromatin.¹⁵ The flexible N-terminal tails of histones extend from the nucleosome core, providing sites for post-translational modifications such as acetylation, methylation, ubiquitination and phosphorylation. Such modifications affect chromatin structure and gene transcription.¹⁶ Histone lysine acetylation generally activates gene transcription, whereas lysine methylation can have different effects depending on the position and status of methylation. For instance, methylation of Lys4, Lys36 or Lys79 on histone H3 usually results in transcriptional activation, whereas methylation of Lys9 or Lys27 on histone H3 or Lys20 on histone H4 is usually linked to transcriptional silencing.¹⁴

In addition, homologues of histone modifying enzymes, such as HDAC6, may act to deacetylate non-histone substrates such as the chaperone protein HSP90, which leads to enhanced protein stability of client proteins including AR.^17–20 Due to the central role of AR in all phases of prostate cancer, interference with AR protein stability or AR co-factor activity represents a rationale strategy to overcome all the aforementioned mechanisms of resistance and may have therapeutic implications in this disease.

Targeting Chromatin-Modifying AR Co-Activators or AR Stability

Growing evidence has shown that co-regulators (factors recruited by transcription factors to either activate or repress transcription) are indispensable components of transcriptional regulation.²¹ AR, as a nuclear receptor, is no exception. AR co-regulators, such as FHL2 (four and a half LIM domains 2), have been reported previously.²² However recently, several chromatin modifying enzymes, namely histone demethylase proteins, were shown to complex with AR and facilitate its activation of gene targets (Table 1).^23–26

Table 1.

List of histone and protein modifying epigenetic enzymes that function in AR signaling

Enzyme	Known direct substrates	Effect on AR signaling	References
LSD1 (AOF2, KDM1)	1MK4, 2MK4	Leads to demethylation of repressive 1MK9/2MK9 marks to facilitate AR-dependent transcription	23, 24, 27, 34
JMJD1A (JHDM2A, KDM3A)	1MK9, 2MK9	Demethylates repressive 1M/2MK9 marks to facilitate AR-dependent transcription	24
JMJD2C (GASC1, KDM4C)	2MK9, 3MK9	Demethylates repressive 3MK9 mark to facilitate AR-dependent transcription	25, 38
JARID1B (PLU-1, KDM5B)	1MK4, 2MK4, 3MK4	Facilitates AR-dependent transcription; mechanism unknown	26
HDAC1	Histones, p53, Smad7, Stat3 etc.	Facilitates AR-dependent transcription; mechanism not fully clarified	44, 52
HDAC3	Histones, Smad7, Stat3, SRY, NFκB etc.	Facilitates AR-dependent transcription; mechanism not fully clarified	44, 52
HDAC6	Alpha-tubulin, HSP90	Deacetylates HSP90 which enhances chaperoning of AR protein	17–19, 51

1MK4, mono-methyl lysine 4 on histone H3; 2MK4, di-methyl lysine 4 on histone H3; 1MK9, mono-methyl lysine 9 on histone H3; 2MK9, di-methyl lysine 9 on histone H3; 3MK9, tri-methyl lysine 9 on histone H3.

Histone demethylases

Histone methylation was original thought to be a stable, irreversible mark as only histone methyl-transferases had been identified. However, the discovery of the first histone demethylase LSD1 (Lysine-specific demethylase 1), also known as AOF2 (Amine amine oxidase (flavin containing) domain 2) or KDM1 (Lysine (K) demethylase 1) in 2004 proved that histone methylation is reversible.²⁷ Since then, a dozen more histone demethylases have been identified.^28–30 Histone lysine methylation is a dynamic and versatile process, which has been observed on a number of lysine residues on histone proteins.³⁰ The balance between repressive and active histone modifications (also known as the histone code) ultimately determines whether a gene will be actively transcribed or repressed, and AR target genes are no exception.¹⁶

The status of histone lysine methylation has been shown to be important for AR signaling, and several histone demethylase proteins are expressed or upregulated in prostate cancer. These include LSD1 and the Jumonji class of proteins, which include JMJD2C (Jumonji domain containing 2C, also known as GASC1 (Gene amplified in squamous cell carcinoma 1) and KDM4C (Lysine (K) demethylase 4C)), JMJD1A (Jumonji domain containing 1A, also known as JHDM2A (Jumonji domain containing histone demethylase 2A), and KDM3A (Lysine (K) demethylase 3A)), and JARID1B (Jumonji, AT rich interactive domain 1B, also known as PLU-1 and KDM5B (Lysine (K) demethylase 5B).^23–26 While the transcriptional targets of these proteins are largely unknown, all 4 complex with AR and facilitate its activation of downstream signaling pathways. Although the recruitment of JMJD1A to target genes was shown to occur solely in the presence of androgens, LSD1 and JMJD2C are bound to AR target genes even in the absence of androgen ligand or AR binding.^23–25

Interference with these histone demethylase with siRNA or pharmacological inhibitors leads to distinct changes in histone marks at AR target promoters. LSD1 is the best studied histone demethylase, and its activity and target specificity are largely dependent on interacting factors.³¹ In fact, it is a component of several transcriptional complexes.³¹ While LSD1 may demethylate the active di-methyl lysine 4 (2MK4) mark on histone H3, which leads to reduced gene expression, at AR target genes, several reports show that LSD1 binds to AR target AREs and facilitates demethylation of the repressive mono (1MK9) and di-methyl lysine 9 (2MK9) marks on histone H3 upon recruitment of ligand-bound AR, which results in transcriptional de-repression.^{23–25,27,32–34} The demethylase activity of LSD1 is essential for this process, as deletion of the enzymatic domain of LSD1 or siRNA to LSD1 reduces androgen-induced transcriptional activation of AR targets, although direct biochemical evidence for LSD1 demethylation of 1MK9 and 2MK9 in cell-free systems is lacking.²³ LSD1 may also mediate prostate cancer aggressiveness as high LSD1 mRNA levels or reduced levels of the 2MK4 mark on histone H3 that LSD1 demethylates, are significantly associated with prostate cancer recurrence, making these promising prognostic markers.^35,36

Likewise, JMJD1A interference leads to increased levels of the repressive 2MK9 mark at AR target genes, which attenuates AR target gene transcriptional activation by ligand-bound AR.²⁴ Unlike LSD1, 1MK9 and 2MK9 have been shown, in cell-free assays, to be direct substrates of JMJD1A. Like LSD1, however, knockdown of JMJD1A does not block AR recruitment, suggesting these two demethylases function after AR binding to facilitate AR’s transcriptional program.²⁴ Unlike LSD1 though, promoter occupancy of JMJD1A is dependent upon the presence of androgens or AR binding to its target genes, and RNA interference of JMJD1A results in more pronounced effects on blocking activation of AR targets than RNA interference of LSD1.²⁴ Finally, JMJD1A expression is increased in response to androgens, although it is not clear whether JMJD1A is an AR target gene.³⁷

The expression of another Jumonji class protein, JMJD2C, is also induced by androgens.³⁷ JMJD2C may catalyze the removal of the repressive di and tri-methyl lysine 9 mark on histone H3.^25,38 More importantly, JMJD2C interacts with LSD1 and co-localizes with LSD1 and AR at AR target gene AREs.²⁵ Together, these two demethylases lead to the removal of repressive tri-, di-, and mono-methyl marks from lysine 9 on histone H3 to activate transcription of AR target genes.²⁵

In addition, it was recently shown that JARID1B, whose only known substrate is the active methylated-H3K4 marks, interacts with AR.²⁶ In reporter assays, JARID1B enhanced transcriptional activation by AR while a JARID1B mutant that was unable to bind AR did not.²⁶ Finally, JARID1B is upregulated in prostate cancers samples compared to benign prostate samples.²⁶

Given the importance of these enzymes in the activation of AR target genes, inhibition of these enzymes may be a rational, non-hormonal strategy to disrupt AR signaling. Efficient inhibitors of JmjC domain-containing histone demethylases have yet to be discovered, but two classes of LSD1 inhibitors have been identified. The first class is MAOIs (monoamine oxidase inhibitors) such as pargyline, which have been used clinically as antidepressants and which may also inhibit LSD1 activity.²³ In addition, polyamine analogues, which have been previously shown to inhibit cancer cell growth and more recently which have been shown to inhibit LSD1 from demethylating the active 2MK4 mark, represent another class of agents to target LSD1, although the effect on AR target genes remains poorly characterized.^34,39

Histone deacetylases

Similar to histone methylation, the state of histone acetylation is determined by two classes of enzymes with opposing activities-histone acetyltransferases (HATs) and histone deacetylases (HDACs). Abnormal activities of both have been linked to cancer.⁴⁰ However, the focus in cancer therapy has been on targeting HDAC enzymes due to the availability of active inhibitors of these proteins.⁴⁰

To date eighteen human HDACs have been identified, which are grouped into four classes based on function and homology to their yeast counterparts: class I includes HDAC1, 2, 3 and 8; class II includes HDAC4, 5, 6, 7, 9 and 10; class III includes Sirt1-7, and class IV includes HDAC11.⁴¹ Despite their uniform name, some HDACs deacetylate non-histone substrates. More than 50 non-histone HDAC substrates have been identified thus far, including key transcription factors such as p53, TFIIE and E2F, alpha-tubulin, and the chaperone protein HSP90.⁴⁰ Therefore HDACs regulate gene expression not only through chromatin modification but also through altered activity of non-histone proteins. HDACs are upregulated in many cancers, and in prostate cancer, HDAC1 is overexpressed.⁴²

Many HDAC inhibitors have been tested, but only one, vorinostat, is currently FDA-approved.⁴³ In general, HDAC inhibitors cause histone hyperacetylation, activating many genes, which leads to growth inhibition, differentiation or apoptosis, all of which would be expected to retard tumor growth or effect cell death.^40,41

While the clinical experience with HDAC inhibitors in prostate cancer thus far has been limited and disappointing, for several reasons, HDAC inhibitor therapy in prostate cancer appears rational including a recent report showed that class I HDACs are essential co-activators of AR (Table 1).⁴⁴ The authors found that two widely used HDAC inhibitors, vorinostat and LBH589, block transcriptional activation of many AR targets, including TMPRSS2-ERG.⁴⁴ This effect was recapitulated by siRNA to HDAC1, and, to a lesser extent, by siRNA to HDAC3.⁴⁴ Although certain HDAC inhibitors can reduce AR protein levels in the cell (see below), transcriptional suppression of AR targets in this report was independent of effects on AR protein levels. Further, the authors showed that certain HDAC inhibitors do not block AR recruitment to its targets, rather, they suppress AR target genes activation by blocking the recruitment of AR co-activators and RNA polymerase II.⁴⁴ More importantly, to mimic castration-resistant prostate cancer, these investigators utilized a prostate cancer cell line that exhibits high basal expression of AR targets in the absence of androgens, and they showed that HDAC inhibitors suppress AR target expression in this cell line as well.⁴⁴ These finding may have implications for the treatment of castration-resistant prostate cancer, and this underscores the need for more specific and less toxic HDAC inhibitors in the treatment of this disease.

Not only can HDAC inhibitors disrupt AR signaling, they can also reduce AR protein levels in the cell. Multiple reports have shown that HDAC inhibitors suppress AR expression.^19,44–46 Prior work had demonstrated that HDAC inhibition may be achievable through dietary compounds such as sulforaphane, which is derived from cruciferous vegetables, whose high consumption is associated with lower prostate cancer risk, although the mechanisms for this remained unclear.^47–50 Given the centrality of AR in prostate cancer and the role of the cytoplasmic HDAC6 protein on activation of HSP90, which leads to AR protein stabilization, we tested the hypothesis that sulforaphane treatment of prostate cancer cells would interfere with HDAC6 function and consequently lead to reduced levels of AR protein and attenuated AR signaling.^17–19 Indeed, we found that sulforaphane treatment enhances HSP90 acetylation through HDAC6 inactivation, which leads to disruption of AR binding to HSP90, eventual AR degradation, and reduced expression of AR target genes (Fig. 1).⁵¹ As opposed to other compounds with HDAC inhibitory function, sulforaphane treatment led to reduced AR binding to its target gene AREs.⁵¹ While in cell-free assays, we showed that sulforaphane may inhibit HDAC6 deacetylation of its tubulin substrate, much of the observed effect in prostate cancer cells may be due to degradation of HDAC6 protein after sulforaphane treatment.⁵¹ Taken together, HDAC inhibitors, including compounds such as sulforaphane with effects on HDAC6, inhibit prostate cancer cell growth, which is at least partially explained by effects on AR signaling.

Figure 1.

Sulforaphane attenuates AR signaling in prostate cancer cells through HDAC6 inactivation. Model of sulforaphane’s effect on attenuating AR signaling through HDAC6 inactivation (reviewed in ref. 51). The AR chaperone protein HSP90 is a substrate of HDAC6. When deacetylated by HDAC6, HSP90 is functional and chaperones client proteins such as AR. With sulforaphane treatment, HDAC6 is inhibited or targeted for protein degradation, which leads to hyperacetylated, inactive HSP90. The disassociation of HSP90 from AR results in AR protein degradation, reduced AR binding to its AREs, and ultimately diminished expression of AR target genes.

Concluding Remarks

Androgen deprivation has been the major therapy for prostate cancer for almost 70 years. While it may be effective in retarding cancer growth, many prostate cancers will progress despite castrate serum levels of testosterone. We now have a greater understanding of the mechanisms of sustained AR signaling in these castration-resistant cancers that have progressed despite androgen deprivation, including upregulation of co-factors such as histone demethylases or histone deacetylases that may facilitate AR target gene activation. Whether targeting histone demethylases and other AR coactivators or AR protein stability will be safe and effective remains unknown, but the pre-clinical data to date suggest such clinical trials are rational. For the 180,000 men who are predicted to be diagnosed with prostate cancer this year, the 27,000 men who are predicted to die of castration-resistant prostate cancer this year, and those who care for them, these recent insights cannot be translated into the clinic fast enough.

Abbreviations

AR

androgen receptor

AREs

androgen-response elements

HDAC

histone deacetylase