Transcriptional network of androgen receptor in prostate cancer progression

Authors

  • Ken-ichi Takayama,

    1. Department of Anti-Aging Medicine, The University of Tokyo, Tokyo, Japan
    2. Department of Geriatric Medicine, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Satoshi Inoue

    Corresponding author
    1. Department of Geriatric Medicine, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
    2. Division of Gene Regulation and Signal Transduction, Research Center for Genomic Medicine, Saitama Medical University, Saitama, Japan
    • Department of Anti-Aging Medicine, The University of Tokyo, Tokyo, Japan
    Search for more papers by this author

Correspondence: Satoshi Inoue M.D., Ph.D., Department of Anti-Aging Medicine, Graduate School of Medicine, The University of Tokyo, Bunkyo-ku, Tokyo 113-8655, Japan. Email: inoue-ger@h.u-tokyo.ac.jp

Abstract

The androgen receptor belongs to the nuclear receptor superfamily and functions as a ligand-dependent transcription factor. It binds to the androgen responsive element and recruits coregulatory factors to modulate gene transcription. In addition, the androgen receptor interacts with other transcription factors, such as forkhead box A1, and other oncogenic signaling pathway molecules that bind deoxyribonucleic acid and regulate transcription. Androgen receptor signaling plays an important role in the development of prostate cancer. Prostate cancer cells proliferate in an androgen-dependent manner, and androgen receptor blockade is effective in prostate cancer therapy. However, patients often progress to castration-resistant prostate cancer with elevated androgen receptor expression and hypersensitivity to androgen. Recently, comprehensive analysis tools, such as complementary DNA microarray, chromatin immunoprecipitation-on-chip and chromatin immunoprecipitation-sequence, have described the androgen-mediated diverse transcriptional program and gene networks in prostate cancer. Furthermore, functional and clinical studies have shown that some of the androgen receptor-regulated genes could be prognostic markers and potential therapeutic targets for the treatment of prostate cancer, particularly castration-resistant prostate cancer. Thus, identifying androgen receptor downstream signaling events and investigating the regulation of androgen receptor activity is critical for understanding the mechanism of carcinogenesis and progression to castration-resistant prostate cancer.

Abbreviations & Acronyms
AF1

activation function 1

AMPK

adenosine monophosphate-activated protein kinase

APC/C

anaphase-promoting complex/cyclosome

APP

amyloid precursor protein

AR

androgen receptor

ARA

androgen receptor-associated

ARBS

androgen receptor binding sites

ARE

androgen responsive element

ARFGAP3

adenosine diphosphate ribosylation factor guanosine triphosphatase-activating protein 3

CAGE

cap analysis of gene expression

CaMKK2

calcium/calmodulin-dependent protein kinase kinase 2

CDH2

cadherin-2

ChIP

chromatin immunoprecipitation

COPA

cancer outlier profile analysis

CRPC

castration-resistant prostate cancer

DHT

dihydrotestosterone

EGF

epidermal growth factor

EGFR

epidermal growth factor 1 receptor

ER

estrogen receptor

ERG

E-twenty six regulated gene

ETS

E-twenty six

ETV1

E-twenty six transcription variant 1

FOXA1

forkhead box A1

H3-K9

histone H3 at lysine 9

HAT

histone acetyltransferase

IL-6

interleukin-6

JMJD2C

Jumonji C domain-containing protein

LBD

ligand binding domain

LNCaP

lymph node carcimona of the prostate

LSD1

lysine-specific demethylase 1

MAPK

mitogen-activated protein kinase

MED1

mediator complex subunit 1

miRNA

micro ribonucleic acid

mRNA

messenger ribonucleic acid

ncRNA

non-coding ribonucleic acid

NTD

N-terminal domain

Oct1

octamer transcription factor 1

PSA

prostate-specific antigen

PTEN

phosphatase and tensin homolog

RefSeq

reference sequence

STAT3

activator of transcription 3

TACC2

transforming acidic coiled-coil protein 2

TMPRSS2

transmembrane protease, serine 2

TSS

transcription start sites

UBE2C

ubiquitin-conjugating enzyme E2 C

Introduction

Prostate cancer is the second leading cause of cancer death in men in the USA.[1] AR signaling plays an important role in the development of the prostate.[2-4] AR belongs to the nuclear receptor superfamily and functions as a ligand dependent transcription factor for gene regulation. When AR binds with a ligand, it translocates to the nucleus, binds to ARE and recruits coregulatory factors.[5] Furthermore, AR also induces prostate cancer development and progression. Despite a favorable response to initial hormone therapy, most patients progress to CRPC with elevation of AR expression, hypersensitivity to androgen, intratumoral steroidogenesis and AR variants.[6-16] Thus, identification of AR downstream signaling events is critical for understanding the progression to CRPC. PSA is the most representative androgen-responsive gene in prostate cancer, and is widely used as a clinical marker for the detection of the disease. Classic analyses of AR functions were carried out using this gene as a model.[17] However, the regulation of other AR target genes and their clinical relevance are not well understood. The present review provides a summary of AR function and downstream signaling in prostate cancer development and progression. We particularly focus on recently developed methodologies for the comprehensive analysis of gene regulatory networks. In addition, functional analyses of AR targets, including protein coding genes, fusion genes, and ncRNA, such as miRNA, are outlined as future directions for clinical use as biomarkers and therapeutic targets.

Androgen receptor signaling pathway in prostate cancer

AR is a member of the nuclear receptor superfamily.[2] AR is the key molecule for androgen signaling in its target organ, the prostate. The AR gene is located on the X chromosome (q11-12) and consists of eight exons. AR has a modular structure in common with other nuclear receptors. It consists of an NTD/transactivation domain, DNA binding domain and C-terminal LBD.[18-23] The NTD is considered to be constitutively active and is important for transcriptional activation independent of ligand binding.[24-27] The NTD contains the transcriptional AF1 domain and mediates aberrant AR activity in CRPC cells. The LBD facilitates binding of androgen to AR. Within the LBD, AF2 interacts with LxxLL-containing coregulators.[28-30] Point mutations in prostate cancer have been mapped to the LBD and associated with resistance to anti-androgens, such as bicalutamide.[31-35] Testosterone is produced in the testes and is the most abundant androgen (∼90%) in the circulation. Androgens produced in the adrenal cortex, such as dehydroepiandrosterone and androstenedione, make up the remaining 10%. Testosterone is able to diffuse into prostate cells and is converted to DHT by the enzyme 5α-reductase.[2, 36] DHT directly binds to and activates AR. DHT binds the receptor more tightly than testosterone.[37] Before ligand binding, AR exists in the cytoplasm in a complex that includes molecular chaperones and co-chaperones from the heat shock protein family. After binding to androgen, a change in the conformation of the complex leads to AR nuclear translocation. In the nucleus, AR binds as a dimer to specific genomic sequences called ARE in the promoter and enhancer regions of its target genes.[38]

Histone modification by AR coregulators

In its genomic action mechanism, AR binds to and recruits various coactivators to activate target genes (Fig. 1a).[39] The representative coactivators are the p160 family (SRC-1, SRC2/NCOA2/GRIP1/TIF2, SRC3/RAC3/AIB1/TRAM1), pCAF, CREB-binding protein and p300.[40-42] The p160 coactivators interact with the AR NTD and also the LBD to enhance ligand-dependent transcriptional activation of the target genes.[43] These coactivators possess intrinsic HAT activity and promote histone acetylation to activate histones. Post-transcriptional modification of histones regulates gene expression. p300 is recruited to the AR complex as a secondary coactivator possessing chromatin remodeling and HAT activities.[44] AR also specifically recruits the ARA coactivators, such as ARA70.[45] The AR-induced complexes finely regulate the level of target gene transcription. The importance of these AR coregulators is shown by evidence that certain AR coregulators are overexpressed in prostate cancer and contribute to disease progression.[46, 47]

Figure 1.

Summary of AR functions and modulation of AR signaling. (a) Coactivator recruitments for histone modification. On binding to DHT, AR translocates to the nucleus, binds to specific sequences called ARE and regulates gene expression. Coactivators for histone acetylation (SRC family or ARA70) or histone demethylation (JMJ2D, LSD1) are shown. (b) Collaborating factors for recruiting AR. For AR-mediated transactivation, multiple interactions with other transcriptional factors are important. The most representative AR interacting partners are the FOX box protein family, such as FOXA1 or FOXP1. FOX binding motifs are widely enriched around AR-binding sites. FOXA1 functions as “pioneer factors” for AR binding. Several other factors, OCT1, NKX3.1 and ERG, have been reported to interact with AR around AR-binding sites. (c) Cross-talk with other signal cascades. AR transactivation is also effected with several signaling pathways arising from extracellular factors, such as epidermal growth factors. Several negative feedback loops by PI3K pathway and PTEN for AR activity have been shown.

In addition to the histone acetylation status, the histone methylation status specifically regulates gene expression. LSD1 interacts with AR and stimulates androgen-receptor-dependent transcription.[48-50] LSD1 relieves repressive histone marks, the mono- or dimethylation of H3-K9, through its demethylase activity. In addition, pargyline, an inhibitor of LSD1, blocks the demethylation of H3-K9 by LSD1 and consequently androgen-receptor-dependent transcription.[48] In another study, LSD1 expression correlated with prostate cancer risk, and overexpression of LSD1 promoted tumor growth.[49] Thus, modulation of LSD1 activity offers a new strategy to regulate androgen receptor functions. Furthermore, JMJD2C is the first histone tridemethylase known to regulate AR activity.[50] JMJD2C interacts with AR and mediates the demethylation of trimethyl H3-K9 for AR-dependent transcriptional activity. Knockdown of JMJD2C inhibits the androgen-induced removal of trimethyl H3-K9, transcriptional activation and tumor cell proliferation. Importantly, JMJD2C colocalizes with androgen receptor, and LSD1 in normal prostate and in prostate carcinomas. AR, JMJD2C and LSD1 assemble on chromatin to remove methyl groups from mono-, di- and trimethylated H3-K9.[50]

Functions of AR-interacting transcription factors

In addition to coregulators of AR, transcription factors interact with AR on the human genome to cooperatively regulate gene expression (Fig. 1b), as shown by recent genome-wide studies. These transcription factors are called “collaborating factors.” FOXA1 is the main AR-interacting partner.[51] FOXA1 modulates ER and AR functions in breast and prostate cancer cells, supporting the postulate that FOXA1 is involved in ER and AR signaling under normal conditions.[52] FOXA1 is considered as a pioneer factor that is essential for AR recruitment and AR-mediated gene expression. According to recent reports that investigated FOXA1 function by analyzing FOXA1 and AR-binding sites genome-wide, FOXA1 depletion elicited extensive redistribution of AR-binding sites on LNCaP cell chromatin.[53] The report identified three distinct classes of AR-binding sites and androgen-responsive genes: (i) independent of FOXA1; (ii) pioneered by FOXA1; and (iii) masked by FOXA1, but functional on FOXA1 depletion.

Interestingly, recent reports about sequencing the prostate cancer exome identified novel mutations in FOXA1 among several patients.[54] FOXA1 is mutated in five of 147 (3.4%) prostate cancers; mutated FOXA1 represses androgen signaling and increases tumor growth. FOXA1 regulates G1-S cell cycle progression in CRPC.[55] In another study, the researchers investigated the correlation of FOXA1 expression with clinical parameters, PSA relapse-free survival and hormone receptor expression in a large cohort of prostate cancer patients at different disease stages.[56] FOXA1 was overexpressed in metastases and particularly in castration-resistant cases, but was expressed at lower levels in both normal and neoplastic transitional zone tissues. FOXA1 levels correlated with higher pT stages and Gleason scores, as well as with AR expression. Importantly, FOXA1 overexpression was associated with poorer prognosis. Furthermore, knockdown of FOXA1 decreased cell proliferation and migration.[56]

In addition to FOXA1, GATA2 and Oct1 were identified as AR-interacting transcription factors that positively regulate AR-binding and androgen-mediated gene induction.[51] We showed that knockdown of Oct1 reduced prostate cancer cell proliferation.[57] We also examined the contribution of Oct1 to prostate cancer development by immunohistochemistry, and showed a significant correlation between high Oct1 immunoreactivity and poor cancer-specific survival. Oct1 can be a prognostic factor in prostate cancer as a coregulator of AR, and might lead to the development of a new therapeutic intervention for prostate cancer.[57]

Furthermore, several studies of AR interacting partners identified p53,[58] nuclear factor 1 (NF1),[59] ETS1[60] and ERG.[61] These studies showed that AR forms complexes with other transcription factors to define AR-binding sites. This is necessary for androgen-mediated transcriptional regulation. In addition, modulation of AR by these collaborating factors promotes prostate cancer progression. Thus, these collaborating factors could be therapeutic target molecules in advanced prostate cancer, particularly CRPC.

Modulation of AR activity by other oncogenic signaling pathways

Interaction of AR with other key signaling events, such as phosphorylation, is also important for AR-dependent gene regulation and subsequent events, such as the proliferation and apoptosis of cancer cells (Fig. 1c).[62] EGF and its membrane receptor, EGFR, have been implicated in the pathogenesis of several cancers, including prostate cancer. Increased expression of EGF and EGFR has been observed in metastatic cancer.[62] EGF treatment enhances the gene transcriptional activity of AR in a manner dependent on the expression level of AR. In LNCaP cells, EGF induces the phosphorylation of AR at tyrosine 267 and 534 by Src and Ack1 kinases in the absence of androgen. Increased levels of STAT3 lead to STAT3-AR complex formation in response to EGF and IL-6. Furthermore, STAT3 increases the EGF-induced transcriptional activation of AR, whereas androgen pretreatment increases STAT3 levels in an IL-6 autocrine-/paracrine-dependent manner, suggesting an intracellular feedback loop.[62] However, EGF negatively regulates the endogenous AR expression level in LNCaP cells.[63] In another report, a signal dependent on a heterodimer of the EGFR dimerization partner HER2/ERBB2 (avian erythroblastosis oncogene B) and another ERBB member, ERBB3, is responsible for the AR signal in conjunction with the PI3K pathway.[64]

Src is a powerful oncogene that is activated in a variety of cancers. In LNCaP cells, Src phosphorylates AR at tyrosine 534, which induces AR's transcriptional activity by increasing its nuclear translocation and DNA binding. Src is activated by various stimuli, such as EGF. Treatment of C4-2 prostate cancer cells with protein phosphatase 2, a Src inhibitor, in androgen-depleted conditions decreases AR activity.[65] P42/44 (ERK1 and ERK2) is another key factor in the MAPK pathway. MAPK inhibitor substantially reduces AR stability and protein expression.[66]

Inactivation of the PTEN tumor suppressor is one of the most common abnormalities found in prostate cancer.[67, 68] PTEN loss activates PI3K and Akt signaling. However, the relationship between PI3K and AR is complex. Recent studies using a PTEN conditional-null mouse model of prostate cancer showed that PTEN/Akt-mediated AR suppression forms a feedback loop. Increased Akt signaling inhibits HER2 kinases and consequently reduces AR activity. In addition, active AR signaling suppressed the Akt pathway through FKBP5.[67] In another study, PTEN negatively regulated the expression and activity of a number of proteins that modulate AR activity, such as EZH2, JUN and EGR1. Thus, PTEN loss combined with AKT activation is sufficient for prostate cancer growth in the absence of AR signaling.[68]

Global analysis of AR-mediated transcriptional action

Since the development of microarray technology, the androgen-mediated transcriptional program has been investigated in normal and cancerous prostate cells.[69] The majority of large-scale gene expression studies have been carried out in LNCaP cells, the most widely used model cell line for prostate cancer research. LNCaP cells are epithelial in origin, express AR, and exhibit androgen-sensitive growth and survival.[5] Expression studies in LNCaP cells characterized the temporal program of transcription that reflects the cellular response to androgens and identified specific androgen-regulated genes or gene networks that participate in these responses. However, in cDNA microarray studies, the locations of specific AR-binding sites and AR dependency were not fully analyzed, although homology to the androgen response-element consensus-binding motif was explored using a computer program.[69]

The rapid development of technology to detect transcription factors binding sites has revolutionized research on steroid hormone receptors (Fig. 2). These technologies are based on ChIP analysis.[58-61, 70-76] After fixing the association between protein and chromatin by formaldehyde treatment, cells are lysed, and chromatin segmentation by ultrasound is carried out. ChIP is carried out using an antibody specific to the transcription factors of interest. Because ChIP-ed DNA contains regions enriched for transcription factor binding sites, there have been several attempts to analyze them.

Figure 2.

Summary of methods to investigate AR signaling. Methods for integrative analysis of AR transcriptional networks are summarized. We used ChIP-on-chip or ChIP-sequence for mapping AR-binding sites in the human genome. To analyze genome-wide androgen-regulated transcripts, we used CAGE (mapping transcriptional start sites of all transcribed mRNA) and microarray (expression profiles of all annotated genes) for this analysis. As another strategy, directional RNA-sequence using next generation sequencer is applied for mapping total sequences of mRNA to the human genome to provide information about transcriptional regulation.

We first used a ChIP-cloning strategy to identify AR-binding sequences.[70] To clone the AR-binding sequences in the presence of the ligand, we subtracted AR ChIP-ed DNA in vehicle control cells from AR ChIP-ed DNA in androgen-treated LNCaP cells using polymerase chain reaction. AR-binding specific DNA was subcloned and prepared for sequencing. We sequenced 100 clones and identified 64 ARBS.[70] For a more comprehensive analysis of transcription factor binding sites, ChIP-on-chip analyses that combine ChIP with genome tiling array technology (chip) have been used. We first carried out ChIP-on-chip using the Encyclopedia of DNA Elements,[71] and a chromosome 21 and 22 genome tiling array[72] in LNCaP cells. The Encyclopedia of DNA Elements includes 1% of the human genome, selected and representative regions of the whole genome. We validated 10 AR-binding sites in the study and identified novel AR-target genes, such as pepsinogen C, UGT1A1 and CDH2. CDH2 is known to be involved in prostate cancer progression. A strong ARBS was identified in intron 1 of the CDH2 gene by this study and validated by conventional ChIP analysis.[71] Our study suggested that unbiased ARBS are not located in the promoter regions and are far from the TSS of RefSeq genes. Several other ChIP analyses were also reported.[59, 60] Wang et al. mapped the AR-binding sites on chromosomes 21 and 22[61] in LNCaP cells. They expanded this to a genome-wide study for comparison of LNCaP cells and LNCaP-derived castration resistant LNCaP-abl cells to identify direct AR-dependent target genes in the disease progression to CRPC.[73] They showed that the role of AR in CRPC is to activate cell cycle progression, mainly by inducing mitotic phase-related genes through a distinct AR program.

More recently, high-throughput analysis of transcription factor binding sites using highly-developed sequencers, called ChIP-sequence analyses, have been developed.[74-76] Several studies using ChIP-sequence technology were recently carried out. Although ChIP-on-chip could not detect ARBS in the regions in which probes were not prepared, ChIP-seq could detect more binding regions. ChIP-sequence has shown a genome-wide cell-based AR transcriptional program. For instance, two sublines of LNCaP prostate cancer cell lines, one overexpressing AR two to threefold and the other four to fivefold compared with control cells, were used for ChIP-sequence.[75] Interestingly, the number of ARBS and the AR-binding strength were positively associated with the level of AR when cells were stimulated with low concentrations of androgens. These data showed that the overexpression of AR sensitizes the receptor binding to chromatin, thus explaining how the AR signaling pathway is reactivated in CRPC cells. In another study, NKX3-1 and AR-binding sites across the prostate cancer genome were analyzed. NKX3-1 is a homeobox gene required for prostate tumor progression, but how it functions is unclear. Two distinct mechanisms by which NKX3-1 controls the AR transcriptional network in prostate cancer were uncovered. NKX3-1 collaborates with AR and FOXA1 to mediate gene expression in advanced and recurrent prostate carcinoma.[76]

Integrative analysis to discover the androgen-regulated transcriptional program

In addition to ChIP-sequence, next-generation sequencers were used to analyze the transcriptome of prostate cancer cells (Fig. 2). Whereas microarray detects the expression levels of transcripts that bind probes, sequence analysis measures the unbiased expression profiles of all transcripts. Therefore, combinational analyses of genome-wide ARBS and the androgen-regulated transcriptome have recently been reported. We have applied these techniques to the analysis of androgen-mediated transcriptional changes. CAGE[77] is a high-throughput method to analyze gene expression and profile TSS, including promoter usage analysis. CAGE is based on the preparation and sequencing of concatemers of DNA tags deriving from the initial 20 nucleotides at the 5′ ends of mRNA.[77] The frequency of CAGE tags correlates well with results from other analyses, such as microarray. The high-throughput nature of this technology paves the way for understanding gene networks through correlation of promoter usage and gene transcription factor expression. We carried out CAGE to determine androgen-regulated TSS and ChIP-on-chip analysis to identify genome-wide ARBS and histone H3 acetylated sites in the human whole genome.[78] CAGE identified 13 110 distinct, androgen-regulated TSS. On the basis of the gene expression database in prostate cancer (Oncomine), the majority of androgen-upregulated genes containing adjacent ARBS and CAGE tag clusters in our study were previously confirmed as upregulated genes in prostate cancer. The integrated high-throughput genome analyses of CAGE and ChIP-on-chip provide useful information for elucidating the AR-mediated transcriptional network that contributes to the development and progression of prostate cancer. Non-coding RNA, including miRNA, were also identified as androgen target transcripts in this study. We found many androgen-dependent TSS widely distributed throughout the genome, including in the antisense direction of RefSeq genes. Several pairs of sense/antisense promoters were newly identified within single RefSeq gene regions, suggesting the involvement of antisense non-coding RNA in transcriptional regulation.[78]

In another study, the whole genome effects of FOXA1 on androgen signaling were investigated. A newly developed technique, global nuclear run-on sequencing, was used to analyze sequential gene expression on androgen treatment.[79] For nuclear run-on reactions, cell nuclei were isolated after treatment with androgen for a specific time. In the run-on step, RNA polymerases are allowed to run on approximately 100 bases in the presence of a ribonucleotide analog (5-bromouridine 5′-triphosphate). 5-Bromouridine-containing RNA was selected through immunopurification with an antibody specific for the nucleotide analog. A cDNA library was then prepared for next-generation sequencing. The authors found the production of enhancer-templated non-coding RNA at a unique class of enhancers that do not require nucleosome remodeling to induce specific enhancer-promoter looping and gene activation. Global nuclear run-on sequencing data also suggest that AR induces both transcription initiation and elongation in a ligand-dependent manner. In combination with the AR-binding and FOXA1 binding data, the authors identified a large repository of active enhancers that can be dynamically tuned to elicit alternative gene expression programs, which might underlie many sequential gene expression events in prostate cancer progression.[79]

Functional analysis of androgen-responsive genes in prostate cancer

Androgen-regulated genes have diverse functions. AR target genes identified by genome-wide research include metabolic enzymes, cell cycle regulators, transcription factors and signal transducers involved in cell survival. We and other groups have reported on functional assays of AR-regulated genes in prostate cancer (Fig. 3).

Figure 3.

Schematic representation of AR responsive genes. AR regulates key regulators through transcriptional activation. The downstream signals include key modulators of diverse cellular functions. Interestingly, AR itself is one of the primary targets and negatively regulated by AR. Furthermore, these signals also include non-coding RNA, such as miRNA. Important genes and pathways are described in detail in the text.

APP

In our integrative analysis of AR gene networks, we identified approximately 500 AR-regulated genes. Among them, we identified APP as a primary androgen target with adjacent functional AR-binding sites.[72] APP is a membrane-bound protein secreted by proteolysis. In the brain, amyloid β peptides produced by specific peptidases form deposits and cause Alzheimer's disease.[80] In contrast, secreted APP function as a growth stimulatory factor to promote cell proliferation in cancer tissues. We showed that APP expression is androgen-inducible in LNCaP cells. Gain-of-function and loss-of-function studies showed that APP promotes tumor growth in prostate cancer. Importantly, APP immunoreactivity correlated with poor prognosis in patients with prostate cancer. Thus, our study showed a novel APP-mediated pathway responsible for the androgen-dependent growth of prostate cancer. APP could be a potential molecular target for the diagnosis and treatment of prostate cancer.[72]

FOXP1 and other forkhead family members

We have reported that the FOXP1 forkhead transcription factor is a novel androgen-regulated gene.[70] FOXP1 can be induced by androgen in hormone-sensitive prostate cancer LNCaP cells at both the mRNA and protein levels. We showed that FOXP1 directly interacts with AR and negatively regulates AR signaling ligand-dependently, as exemplified by transcriptional repression of the PSA gene, which is regulated by androgen-dependent FOXP1 recruitment to its enhancer region. We also showed that several other forkhead transcription factors are androgen-responsive in LNCaP cells.[70] After our report, another group identified three major genomic regions deficient in prostate cancer, p53, PTEN and FOXP1 loci, suggesting that FOXP1 has a tumor suppressive effect in prostate cancer by repressing AR signaling.[81]

ARFGAP3

ARFGAP3 is a guanosine triphosphatase-activating protein that associates with the Golgi apparatus and regulates the vesicular trafficking pathway. We showed that ARFGAP3 expression was induced by DHT at both the mRNA and protein levels in androgen-sensitive LNCaP cells.[82] Small interfering RNA-mediated knockdown of ARFGAP3 markedly reduced LNCaP cell growth. In addition, ARFGAP3 overexpression promoted cell proliferation and migration compared with control cells. Interestingly, ARFGAP3 interacted with paxillin, a focal adhesion adaptor protein that is important for cell mobility and migration. Additionally, paxillin regulates both androgen- and EGF-induced nuclear signaling. A recent report showed that androgen and EGF promoted the MAPK-dependent phosphorylation of paxillin, resulting in nuclear translocation of paxillin and enabling nuclear paxillin to associate with androgen-stimulated AR.[83] The paxillin-AR complex bound AR-sensitive promoters and regulated AR-mediated transcription. Nuclear paxillin also interacts with ERK and ELK. Thus, paxillin is an intermediary between extranuclear MAPK signaling and nuclear transcription in response to androgens and growth factors. AR-dependent transactivation activity on the PSA enhancer was synergistically promoted by exogenous ARFGAP3 and paxillin expression, as shown by a luciferase assay in LNCaP cells. Thus, our results suggest that ARFGAP3 is a novel androgen-regulated gene that can promote prostate cancer cell proliferation and migration in collaboration with paxillin.[82]

14-3-3 family

In humans, seven different 14-3-3 isoforms have been identified. Although 14-3-3 proteins do not function as enzymes, they form homo/heterodimers and bind to phosphorylated serine/threonine motifs on their target proteins.[84, 85] After binding, 14-3-3 proteins can change the conformation of the target protein, thereby affecting protein activity/stability, facilitating protein complex formation or altering protein subcellular localization. 14-3-3 Proteins can interact with hundreds of binding partners to regulate diverse cellular processes, including apoptosis, mitogenic and stress signaling, and cell-cycle progression. In terms of cancer biology, 14-3-3σ is known to be a tumor suppressor, and 14-3-3σ expression is downregulated in prostate cancer.[84] Other 14-3-3 isoforms might have oncogenic roles. We showed the androgen-dependent upregulation of 14-3-3ζ at the mRNA and protein levels. The 14-3-3ζ gene is favorable for cancer-cell survival, as its ectopic expression in LNCaP cells contributes to cell proliferation and the acquired resistance to etoposide-induced apoptosis.[85] 14-3-3ζ Expression was associated with androgen receptor transcriptional activity and PSA mRNA expression. Immunoprecipitation indicated that 14-3-3ζ interacted with AR in the nucleus. Clinicopathological studies further support the relevance of 14-3-3ζ in prostate cancers, as its higher expression is associated with malignancy and lymph node metastasis. Thus, androgen-mediated cell survival is important for prostate cancer development.[85]

CaMKK2

CaMKK2 as a representative AR downstream signal in our integrative analysis.[78] CaMKK2 expression is highly upregulated in prostate cancer in all databases registered in Oncomine. In another study, AR was identified as the core regulator of an anabolic transcriptional network.[86] The investigators highlighted CaMKK2 and showed that this gene regulates cancer cell growth through its unexpected role as a hormone-dependent modulator of anabolic metabolism. In another study, using cellular models of prostate cancer, androgens directly increased the expression of a CaMKK2 splice variant and increased functional CaMKK2 protein levels as determined by the phosphorylation of both CaMKI and AMPK, two of the primary substrates of CaMKK2.[87] Importantly, inhibition of the CaMKK2-AMPK, but not CaMKI, signaling axis in prostate cancer cells by pharmacological inhibitors or small interfering RNA-mediated knockdown blocks androgen-mediated migration and invasion. Conversely, overexpression of CaMKK2 alone leads to increased AMPK phosphorylation and cell migration.[88] Given the key roles of CaMKK2 and AMPK in the biology of prostate cancer cells, these enzymes are potential therapeutic targets in prostate cancer.

TMPRSS2-ERG fusion gene

Chromosomal rearrangements caused by androgen were one of the most important findings in the research field of prostate cancer. A novel gene fusion between the androgen-regulated promoter of TMPRSS2 and the 3′ end of the oncogenic ETS transcription factor family members was identified using a novel approach called COPA.[89] COPA was applied to databases of microarrays to identify overexpressed genes in subsets of malignant versus normal tissues. ERG and ETV1, ETS transcription factors, were overexpressed by chromosomal translocations, which resulted in various fusions between the 5′ end of the TMPRSS2 gene and the 3′ end of either ERG or ETV1. TMPRSS2 was first identified as an androgen-responsive gene in LNCaP cells.[90-92] TMPRSS2 is highly specific to prostatic tissue and localizes to prostate luminal epithelial cells. In microarray datasets, ERG1 or ETV1 was overexpressed in 57% of prostate cancer cases, but not in benign tissues, and the fusion with TMPRSS2 was found in 20 of 22 cases that overexpressed ERG or ETV1. Other cohorts support these findings and suggest that TMPRSS2 fusion with ETS members is the most frequent rearrangement in prostate cancer.[93, 94] In ChIP-sequence analysis of ERG-binding sites in prostate cancer cells, ERG-binding sites overlapped with AR-binding sites, suggesting a role for ERG in modulating AR signaling.[61] Interestingly, TMPRSS2 : ERG fusion has been detected in non-malignant prostate cancer epithelial cells after long-term exposure to DHT.[95, 96] These fusions might be an early event in prostate cancer development induced by androgen.[97-99] Other clinical trials showed that this fusion could be detected in the urine of patients with prostate cancer,[100] and support larger studies on prospective cohorts for the non-invasive detection of prostate cancer. In addition to TMPRSS2, other androgen-regulated genes, such as ACSL3 or FOXP1, were also identified as partners encoding fusion proteins.[101, 102]

Androgen-regulated miRNA

MiRNA are approximately 20 nucleotides-long short RNA that mediate diverse cell signaling pathways by repressing gene expression. Our integration analysis of CAGE and ChIP-on-chip analyses successfully identified a cluster of androgen-inducible miRNA, as exemplified by the miR-125b-2 cluster on chromosome 21. MiR-125b is the first androgen-regulated miRNA to be reported.[103, 104] miR-125b stimulates prostate cancer proliferation. In another study using a next-generation sequencer, we mapped short RNA sequences expressed in LNCaP cells and regulated by androgen. We found that miR-148a and −141 are androgen-regulated miRNA that promote LNCaP cell proliferation.[105] MiR-148a repressed cullin-associated and neddylation-dissociated 1, thereby inducing cell cycle progression by activating SCF complex. In other studies, miR-148a was highly expressed in CRPC.[106] In addition, miR141 is also speculated to be a biomarker of prostate cancer, because miR-141 expression is higher in prostate cancer patient blood.[107, 108] miR-21 is also a representative androgen-regulated onco-miRNA upregulated in prostate cancer.[109] miR-21 overexpression promotes castration-resistant prostate cancer growth. In future studies, investigation of androgen-regulated miRNA and non-coding RNA will identify novel functions of AR.

Role of androgen-regulated genes in CRPC

Although AR overexpression in CRPC has been commonly observed, the mechanism underlying altered AR expression has not been fully understood. However, recent studies showed that a transcriptional change of AR in addition to a genomic change is important for AR upregulation in prostate cancer. Cai et al. analyzed an androgen-dependent prostate cancer cell line and identified AR-binding sites in the introns of the AR gene.[110] Androgen treatment caused AR to bind the enhancer and recruit LSD1, which repressed transcription by inhibiting histone H3-K4 methylation in this situation, thus showing the negative feedback loop that limits endogenous AR expression. However, if cells were incubated in castration levels of androgen, low AR activity induced AR expression and subsequently caused an increase in AR. This group also showed that low levels of AR in CRPC are sufficient to activate AR target genes, but insufficient to recruit AR to suppress the genes, including the AR gene itself.[110]

To analyze the involvement of AR in the progression from hormone-sensitive prostate cancer to CRPC, several cell models have been developed. In particular, AR-positive prostate cancer models have been established by incubating LNCaP cells or VCaP cells in hormone-depleted cells.[110-113] Such model cells were called LNCaP-abl,[114] LNCaP-AI[113] or LTAD[115] cells according to the laboratory that established the cell line. These cell lines express increased AR at both the mRNA and protein levels, suggesting hypersensitivity of AR signaling. In some studies, the functions of AR-regulated genes have been analyzed using these cell models.

UBE2C

A combinational study using AR ChIP-chip and cDNA microarray analysis found that CRPC-unique AR signaling appears in a CRPC cell model derived from LNCaP cells.[73] UBE2C was studied as an androgen target gene. The UBE2C protein is an APC/C-specific E2 ubiquitin-conjugating enzyme that plays a critical role in APC/C-dependent M-phase cell-cycle progression by inactivating the M-phase checkpoint or increasing the pool of active APC/C[114]. UBE2C mRNA and protein are overexpressed in prostate cancer. They increase as the disease progresses. CRPC-specific enhancers drive UBE2C overexpression in both AR-negative and -positive CRPC cells.[73] Recruitment of the co-activator MED1 to UBE2C enhancers is required for long-range UBE2C enhancer/promoter interactions. Importantly, the study found that the molecular mechanism underlying MED1-mediated chromatin looping involves PI3K/AKT-phosphorylated MED1-mediated recruitment of FOXA1, RNA polymerase II and TATA binding protein, and their subsequent interactions at the UBE2C locus. MED1 phosphorylation leads to UBE2C locus looping, UBE2C gene expression and cell growth.[114] These results not only define a causal role for post-translational modification (phosphorylation) of a co-activator (MED1) in forming or sustaining an active chromatin structure, but also suggest that development of specific therapies for CRPC should consider targeting phosphorylated MED1.

TACC2

We discovered that a centrosome- and microtubule-interacting gene, TACC2, is a novel androgen-regulated gene.[115] The members of the TACC family of proteins contain C-terminal coiled-coil domains and are implicated in cancers through their function in the formation of the mitotic spindle during mitosis.[116-119] A recent report showed that TACC2 correlates with poor prognosis in patients with breast cancer.[120] We identified a functional ARBS including two canonical androgen response elements in the vicinity of TACC2 gene, in which activated hallmarks of histone modification were observed. Androgen-dependent TACC2 induction is regulated by AR. Using long-term androgen-deprived cells as cellular models of CRPC, we showed that TACC2 is highly expressed and contributes to hormone-refractory proliferation, as small interfering RNA-mediated knockdown of TACC2 reduced cell growth and cell cycle progression. By contrast, in TACC2-overexpressing cells, an acceleration of the cell cycle was observed. In vivo tumor formation study of prostate cancer in castrated immunocompromised mice showed that TACC2 is a tumor-promoting factor. Notably, the clinical significance of TACC2 was shown by a correlation between high TACC2 expression and poor survival rates. Taken together with the critical roles of TACC2 in the cell cycle and the biology of prostate cancer, we infer that the molecule is a potential therapeutic target in CRPC, as well as hormone-sensitive prostate cancer.[115]

Conclusion and future plans

AR plays an important role in the development of a hormone therapy refractory condition or castration-resistance in prostate cancer. Recent evidence showed the mechanism of activating or renewing AR in CRPC progression. Blocking AR with next-generation AR antagonists has the potential to treat advanced prostate cancers or CRPC. During this decade, highly developed technologies to investigate transcriptional networks have advanced the understanding of androgen signaling in prostate cancer. Microarray analysis and next-generation sequencing of the global transcriptome in prostate cancer cells have identified androgen-regulated gene networks. Various functional studies, including our reports, have cited androgen-regulated genes as preferable candidates for biomarkers and therapeutic targets for therapy resistant prostate cancer. In addition, the global analysis of AR-binding and AR collaborating factors, such as FOXA1, has shown diverse AR genomic functions. Further analysis of signaling, including non-coding RNA, will reveal novel AR functions in prostate cancer cells. Investigating the deeper mechanisms of cancer progression mediated by AR downstream signaling and the regulation of AR activity will be useful for considering new strategies to treat CRPC.

Acknowledgments

The authors acknowledge the generous support by Cell Innovation Program, Grants-in-Aid, and P-DIRECT from the MEXT and JSPS, Grants-in-Aid from the MHLW, and the Program for Promotion of Fundamental Studies in Health Sciences of the NIBIO.

Conflict of interest

None declared.

Ancillary