|Home | About | Journals | Submit | Contact Us | Français|
Prostate cancer remains a leading cause of cancer death, as there are no durable means to treat advanced disease. Treatment of non-organ confined prostate cancer hinges on its androgen dependence. First line therapeutic strategies suppress androgen receptor (AR) activity, via androgen ablation and direct AR antagonists. While initially effective, incurable, "castration-resistant" tumors arise due to resurgent AR activity. Alterations of AR and/or associated regulatory networks are known to restore receptor activity and support resultant therapy-resistant tumor progression. However, recent evidence also reveals an unexpected contribution of AR ligand, wherein alterations in pathways controlling androgen synthesis support castrate-resistant AR activity. Herein, mechanisms underlying the lethal pairing of AR deregulation and aberrant androgen synthesis in prostate cancer progression will be discussed.
To date, prostate cancer remains the second leading cause of cancer death and the most frequently diagnosed malignancy amongst men in the United States . Locally confined prostate cancers can be effectively treated through either surgical resection or radiation therapy . However, non-organ confined tumors represent a significant clinical challenge, accounting for significant morbidity. While the underlying mechanisms are not well understood, prostate cancers respond poorly to standard antimitotics used for chemotherapy. Thus, the first line of clinical intervention for patients with non-organ confined disease capitalizes on the established addiction of prostate cancers to androgen receptor (AR) signaling, and consists of a variety of mechanisms to ablate AR function [3–8]. These regimens are initially effective, resulting in AR activity suppression and tumor regression. However, incurable, “castration-resistant” cancers (CRPC, castration-resistant prostate cancers) develop in patients with disseminated disease within a median time of 2–3 years, wherein AR activity has been reactivated [4, 9]. Based on these clinical observations, there has been an intensive effort in the field to discern the mechanisms by which AR is reactivated in recurrent disease and to develop novel strategies to thwart this process.
As a member of the steroid receptor subclass of nuclear receptors, AR functions as a ligand-dependent transcription factor (Figure 1) [4, 10]. In the absence of ligand (androgen) binding, the receptor is present diffusely throughout the cytoplasm and held in an inactive state in association with chaperones such as heat shock proteins (HSPs) . While testosterone (T) is the most prevalent androgen present in sera of human males, it is converted into 5α-dihydrotestosterone (5α-DHT), a higher affinity ligand for AR, in prostatic epithelia or prostatic adenocarcinoma cells . The Kd of T for AR=10−9 M while that of 5α-DHT for AR=10−11M. Ligand binding releases AR from HSPs, facilitating AR homodimerization, rapid nuclear translocation, post-translational modification, and receptor stabilization . Activated AR dimers subsequently bind DNA at specific sequences deemed androgen responsive elements (AREs), serve as a platform for recruitment of coactivators and basal transcriptional machinery, and initiate a program of gene transcription that results in diverse biological outcomes dependent on cell context [13–15]. To date, the best-characterized AR target gene is KLK3, which encodes the PSA (prostate specific antigen) protein. The ability to monitor PSA as a readout of AR function has had a major impact on diagnoses and management of human disease [16, 17]. As a serine protease secreted by the prostate, detection of serum PSA affords a facile mechanism through which to assess prostate-specific AR activity. As such, PSA screening is widely utilized to monitor disease development, tumor progression, and response to therapeutic intervention. Recent advances revealed that AR requires cooperating “pioneer” factors such as FoxA1 to enhance transcription, and that the predominant sites of AR action lie outside classical promoter regions of androgen responsive genes; in fact, AR function appears to be largely manifest through sites of action that lie in distal enhancers located in intragenic, intronic and intergenic regions [18, 19].
While challenges exist for matching identified sites of AR binding to transcriptional output or disease relevance, several recent advances revealed the divergent function of AR in RPC. First, Wang et al. elegantly showed that in CRPC cells, AR occupies binding sites that overlap with but are distinct from those observed in tumor cells that are responsive to hormone ablation. Remarkably, AR appears to be enriched in castrate-resistant cells to regulate genes whose products control transitions into and through mitosis. One target gene of relevance, UBE2C, is required for castration-resistant cell growth, and is overexpressed in clinical CRPC specimens [18, 19]. Second, resurgent AR activity in VCaP xenograft models following castration promotes recurrent ERG expression from the TMPRSS2:ERG chromosomal translocation present within this model system . Given the high frequency of TMPRSS2:ERG translocations in prostate cancer and the importance of ERG signaling for tumor phenotypes, these findings are of clinical significance. Most recently, two independent groups demonstrated that AR activation induces a close proximity of the TMPRSS2 and ERG chromosomal loci, and in the presence of ionizing radiation, actually promotes formation of the TMPRSS2-ERG chromosomal translocation [21, 22]. These striking findings present a model wherein AR activation binds chromatin to alter gene expression, but therein actively supports chromosomal aberrations. Combined, it is clear that the ability of AR to bind AREs and elicit a resultant gene expression program is critical for both early stage and castration-resistant disease, and these gene-expression profiles may be different.
The biochemical goal of first line intervention for non-confined tumors is to effectively suppress the transactivation potential of AR, regimens collectively referred to as androgen depletion or endocrine therapy [4, 5, 7]. GnRH agonists (e.g. leuprolide) represent the most frequently utilized means to deprive AR of ligand . These drugs suppress the release of LH from the anterior pituitary and prevent Leydig cell testosterone biosynthesis in the non-castrate male. While these agents result in an initial spike of T release , testicular androgen synthesis is subsequently suppressed, resulting in serum T levels similar to those seen in surgically castrated men (<0.2 ng/mL) . These strategies are sometimes used in combination with direct AR antagonists such as bicalutamide, which compete for agonist binding [4, 7]. In addition, the most commonly administered AR antagonists are thought to elicit conformational changes in the receptor that recruit corepressors (rather than coactivators) to sites of AR binding, thus assisting in active transcriptional repression . That regimens are initially effective is incontrovertible; the vast majority of patients with disseminated disease show marked PSA declines (thus providing biochemical evidence of AR inactivation) and tumor remission [4, 7]. At the cellular level, hormone therapy results in both cell cycle arrest and cell death [15, 26]. However, recurrent tumor formation is common, and there remains no effective, durable means to treat this latter stage of disease [4, 7]. For CRPC, docetaxel shows modest efficacy in prolonging life but is not curative . Development of detectable, recurrent tumors is almost invariably heralded by a preceding rise in serum PSA, thus indicating that AR is re-activated for disease progression . Based on these observations and interrogation of AR function in models of human disease, it is widely accepted that AR is the key driver of prostate cancer progression and is required at all stages of disease for tumor maintenance [4, 7]. As such, it is of the utmost importance to fully delineate the mechanism(s) by which AR becomes reactivated in recurrent disease, and to discern the underlying pathways that impinge on this process.
In the last decade, it has become quite clear that there are multiple mechanisms through which AR can be inappropriately reactivated in the presence of GnRH agonists (chemical castration) and direct AR antagonists. Means of AR reactivation in such castrate-resistant tumors can be loosely classified into mechanisms that: i) impinge directly on AR modulation; ii) involve alterations in AR cofactors, or iii) promote intratumor ligand synthesis (Figure 2). Each will be considered herein.
Aberrant AR expression, alterations in upstream regulatory factors, and/or upregulation of required cofactors each significantly contributes to resurgent AR activity in CRPC. First, the great majority of CRPCs show marked induction of AR mRNA and protein expression [28, 29]. A fraction can be accounted for by amplification of the AR locus itself [30–32], thus implicating genomic instability in CRPC. Little is understood about the additional mechanisms that promote enhanced AR expression. Loss of Pur-alpha, which can modulate the AR transcript through the 5’UTR, has been implicated in this process in model systems, but its clinical relevance remains uncertain . Regardless of mechanism, it has been clearly demonstrated that deregulation of AR alone can have a major impact on prostate biology and therapeutic response. Transgenic animals wherein AR was modestly overexpressed in a prostate-specific manner showed evidence of both hyperplasia and carcinoma in situ, providing in vivo evidence for the pro-tumorigenic functions of AR overexpression . Oncogenic transformation and progression to metastatic disease was observed in a transgenic model of prostate specific AR-E231G expression, thus validating the concept that AR gain-of-function mutations are sufficient to drive tumor development and progression . In addition, analyses of xenograft models wherein hormone therapy-sensitive tumors progressed to recurrent tumors post-castration revealed that the major molecular change associated with CRPC was elevation of AR itself, reinforcing the hypothesis that disease progression is reliant on sustained AR signaling . These same studies effectively demonstrated that AR induction alone is sufficient to bypass androgen depletion therapy and weaken the antagonist capabilities of bicalutamide. Collectively, these studies point to AR deregulation as a major mechanism of recurrent AR activity and CRPC formation.
Alterations of AR also occur that can either significantly alter the spectrum of ligands that act as agonists or bypass the need for ligand altogether. A large number of somatic, tumor-derived mutations of AR have been identified, and the majority of these result in “promiscuous” ligand binding, facilitating activation of the receptor by non-androgen steroid hormones (e.g. progesterone, estrogens, cortisol and weak androgens [6, 9, 37]). A subset of these somatic mutations also convert known AR antagonists into agonists. Clinical situations referred to as “the anti-androgen withdrawal effect”, wherein cessation of therapies based on direct AR antagonism resulted in lower PSA levels, suggest that alterations in AR may change the cellular response to these therapeutically used receptor antagonists . Indeed, a somatic mutation of AR identified in human disease, AR-T877A, results in flutamide-mediated receptor activation . More recent studies identified mutations that result in receptor activation by bicalutamide ), and analyses of specimens from CRPC support the current hypothesis that specific AR antagonists select for development of specific AR somatic mutations . Since the overall frequency and impact of somatic AR mutation in prostate cancer remains uncertain (and ranges from 8–25% in tumors analyzed), additional studies using relevant tissue (CRPC) are needed. In addition, it will be important to define the ligands that act as agonists for clinically relevant somatic mutants of AR. At present, it is clear that specific mutations are selected for during disease progression in a subset of tumors, resulting in increased ligand promiscuity and responsiveness, and active promotion of CRPC via non-androgens.
In addition, AR can be alternatively spliced in CRPC, resulting in receptors that decisively bypass the need for ligand [42–44]. These “constitutively active” splice variants were first identified in prostate cancer cell lines, and shown to result in both cryptic exon usage and exon exclusion [42–44]. While the precise number and frequency of the alternatively spliced variants remains to be rigorously determined, those identified to date retain the most critical transactivation domain of the receptor (AF1, located in the N-terminus) and the DNA binding domain, but are devoid of the ligand binding domain (LBD). It has been long appreciated that deletion of the LBD results in constitutively active AR , and revolutionary amongst these findings was the observation that LBD-deficient splice variants are enhanced in CRPC. Since these mutants would not be amenable to inhibition by LBD-directed AR antagonists (such as bicalutamide, TOK-001, or MDV3100), upregulation of such AR splice variants presents a significant clinical challenge. Combined, it is apparent that alterations in the AR coding region, either through somatic mutation or alternative splicing of the resultant transcript, play significant roles in human disease progression.
Recurrent AR activity can be achieved in the presence of hormone therapy through post-translational modification(s) that do not require alterations in the AR locus or mRNA processing. Not surprisingly, AR activity is modulated by disparate mechanisms that include serine/threonine phosphorylation, tyrosine phosphorylation, acetylation, ubiquitylation, and sumoylation [9, 46]. Some uncertainty remains with regard to the overall impact of individual modification events on subsequent modifications and total AR activity, but several key findings point to evidence for aberrant AR modification playing a role in human disease. For example, phosphorylated AR is associated with reduced survival in patients that have failed hormone therapy, thus implicating phosphorylation-derived AR modifications with disease progression . Underlying mechanisms of aberrant phosphorylation events are emerging, and in many cases are attributed to growth factor receptor activation. Deregulated epidermal growth factor (EGF) activity can induce AR phosphorylation at Ser-578, resulting in castration-resistant receptor activity and tumor cell proliferation . Other growth factors including IGF1 (recently reviewed in ) bolster AR activity in a low-ligand environment, supporting the contention that under certain conditions, peptide growth factors support overall AR activity. These findings are of potential disease relevance, as IGF1 is locally induced in human disease . Conversely, tyrosine phosphorylation of AR appears to be predominantly driven by oncogene activation, especially via Src. This phosphorylation event is found with higher frequency in castrate-resistant tumors, and modeling of aberrant tyrosine phosphorylation supports the contention that deregulated tyrosine phosphorylation promotes ligand-independent AR activity and concomitant cellular proliferation . Notably, EGF function is also partially dependent on Src-mediated AR tyrosine phosphorylation, supporting a role for multiple phosphorylation events in mediating growth factor-induced AR activation . Intriguing new studies suggest that the tumor microenvironment may promote both events, as a neuroendocrine cell-derived protein (parathyroid hormone related protein, PTHrP) appears to promote EGF and Src-mediated AR modification and resultant adaptation to a low androgen environment . Neuropeptides released by this cell type appear to serve similar functions . Alternatively, AR phosphorylation can be enhanced through altered phosphatase activity. Recent reports indicate that PP1 associates with AR and regulates both receptor subcellular localization and stability . These collective observations underscore the importance of external signals in modulating nuclear receptor function through phosphorylation cascades.
In addition to phosphorylation, AR is regulated by ubiquitylation, sumoylation, and acetylation events that may influence disease progression. The ubiquitin E3 ligase RNF6 promotes AR activity through selective modulation of cofactor recruitment (such as ARA54), and this function is enhanced in castrate-resistant tumors . While similar observations were reported with TRIM68 , the Mdm2-mediated ubiquitylation of AR results in receptor destabilzation and loss of activity . Thus, different ubiquitylation events appear to result in disparate effects on AR activity, and the underlying events that control these processes are incompletely understood. By contrast, conjugation of SUMO-1 to AR occurs rapidly after androgen binding, and cleavage of this process by SENP1 and SENP2 promotes gene-specific AR activation . It has been suggested that this post-translational modification helps “fine tune” receptor activity. It will be of interest to determine how this process is regulated in human disease.
AR cofactors are cis-acting transcriptional modulatory proteins that substantially influence AR function. Given the prevailing posit that agonists induce recruitment of coactivators and antagonists promote conformational changes that promote recruitment of corepressors, two hypotheses emerge. First, it would be predicted that deregulation of coactivators or loss of corepressors promote unchecked AR activity and disease progression. Second, it is predicted that changes in overall AR levels alter the stoichiometry of assembled complexes. Both predictions appear to be correct and have disease relevance.
To date, several hundred putative AR coactivators have been identified that enhance ligand-dependent AR activity in model systems. These coactivators serve pleiotropic functions at the chromatin level, including recruitment of basal transcriptional machinery, modulation of chromatin remodeling enzymes function or recruitment (e.g. histone acetylase), and/or altered AR conformational changes. A subset of AR coactivators appear to be enhanced in human disease, including SRC1, SRC2, SRC3, or ARA70 [13, 59–61]. The importance of deregulated coactivator expression may be significant, as excessive coactivator expression may not only sensitize cells to a low hormone environment but also convert nuclear receptor antagonists into partial or full agonists. As AR is known to regulate a distinct transcriptional program in hormone sensitive versus castrate-resistant models of disease , an attractive hypothesis is that altered cofactor expression and/or regulation assists in eliciting the CRPC-specific transcriptional program.
Loss of AR corepressor function can convert therapeutic antagonists into agonists or promote agonist sensitization (reviewed in ). Such events can occur through downregulation of the corepressor itself (such as occurs with prohibitin) [63, 64], through dismissal of the corepressor from the AR complex (as seen with NCoR in the presence of macrophage induced TAB2 signaling) , and/or through aberrant corepressor mislocalization (such as observed with Hey1) . In addition to AR modulation, corepressors perturbed in prostate cancer may crosstalk with pathways directly associated with prostate cancer growth. For example, reduction of the AR corepressor Ebp1 is not only associated with resistance to hormone therapy, but also alters the proliferative response to heregulin . Similarly, crosstalk between the AR and cell cycle machinery is mediated by cyclin D1, which acts through cyclin-dependent kinase-independent functions to suppress AR activity [15, 68]; this function of cyclin D1 is abrogated in human disease through downregulation, mislocalization, or alternative splicing events [15, 69]. As a result of such growth factor and cell cycle crosstalk functions embedded within selected AR cofactors, alterations therein may impinge both on AR signaling and connected pathways to yield a powerful pro-tumorigenic signal. Challenges remain with regard to discerning which of the several hundred co-repressors identified to date play critical roles in recurrent AR activity, and prioritizing those which could be developed as viable therapeutic targets.
Most recently, it has become apparent that resurgent AR activity in CRPC can be accounted for in part through intratumoral androgen synthesis mediated by intracrine and paracrine mechanisms. As mentioned, prostate cancer is a disease whose growth is dependent on the male sex hormone T which is converted in the prostate by steroid 5α-reductase type 2 (SRD5A2) to yield the more potent androgen 5α-DHT  (Figure 3). Importantly, prostate cancer is a disease of the aging male and thus grows under the influence of androgens even as testicular output of T wanes. An alternative source of androgens in the aging male is the adrenal, whereby circulating dehydroepiandrosterone (DHEA) is converted in the prostate via the sequential actions of 3β-hydroxysteroid dehydrogenase [3β-HSD/ ketosteroid isomerase type 1 and type 2 (HSD3B1, HSD3B2), type 5 17β-HSD (AKR1C3)]; and 5α-reductase type 2 to yield 5α-DHT [71–73]. In CRPC this intratumoral synthesis of androgens provides a mechanism by which the effects of a GnRH agonist on Leydig cell T synthesis can be surmounted. Indeed, increases in the androgenic synthetic pathway occur in CRPC as part of an adaptive response to androgen ablative therapy [74–76].
The role of intratumoral synthesis in CRPC has been controversial since it did not adequately explain why AR receptor antagonists, e.g. flutamide and biaclutamide, failed and why early chemopreventive trials of finasteride (a selective 5α-reductase type 2 inhibitor) decreased prostate cancer incidence but resulted in the appearance of a more aggressive disease in some patients . Proponents of the intratumoral synthesis of androgens point out that T and 5α-DHT are very potent hormones, and low concentrations might be sufficient to outcompete the effects of low affinity AR antagonists and activate AR . In addition, the more aggressive tumors observed in the original finasteride prostate cancer chemopreventive trial are now widely accepted as being due to a sampling artifact due to the increased sensitivity of biopsies to detect cancer in the drug arm . Studies on the use of finasteride to reduce intraprostatic 5α-DHT show that hormone levels are suppressed by only 68–86%, suggesting that other routes to this hormone exist . Two routes to the synthesis of 5α-DHT that are independent of SRD5A2 are steroid 5α-reductase type 1 (SRD5A1) and “RoDH like 3α-HSD” (HSD17B6) [80, 81] (Figure 3). The latter enzyme catalyzes the back (oxidative) conversion of 3α-androstanediol to 5α-DHT. The recognition that two 5α-reductase isoforms are involved in intraprostatic synthesis of 5α-DHT has led to the development of dutasteride which inhibits both isoforms. Dutasteride is currently in clinical trial for the treatment and prevention of prostate cancer . Earlier studies with dutasteride to treat benign prostatic hyperplasia indicated that it failed to reduce serum DHT levels altogether , and intraprostatic levels of DHT fell from 3.23 ng/g to 0.209 ng/g . Alternatively, 5α-DHT can be formed by the “backdoor pathway” in which 3α-androstanediol is oxidized to 5α-DHT via RoDH-like 3α-HSD. In this pathway, Δ4-androstene-3,17-dione and testosterone are not precursors of 5α-DHT [84, 85]. This pathway starts with the conversion of pregnenolone to progesterone catalyzed by 3β-HSD2 (HSD3B2), formation of 17α-hydroxyprogesterone catalyzed by 17α-hydroxylase (CYP17A1), 5α-reduction to yield 5α-pregnane-17α-ol-3,20-dione (catalyzed by 5α-reductase isoforms), 3-ketone reduction to yield 5α-pregnane-3α,17α-diol-20-one (catalyzed by AKR1C2) followed by CYP-17,20-lyase (CYP17A1) to yield androsterone. Androsterone is then reduced to 3α-androstanediol by the action of AKR1C3. This “backdoor pathway” is thought to be important in the aging male when adrenal output of these steroids contributes more to prostate steroidogenesis. Thus far, “RoDH-like 3α-HSD” (17BHSD6) has not been targeted for androgen ablative therapy in prostate cancer since its role has only been recently elucidated.
The importance of intratumoral androgen synthesis following chemical or surgical castration has gained credence based on several observations. First, critical genes involved in androgen synthesis in the prostate are up-regulated at the transcript level in CRPC; these include HSD3B2 (1.8 fold increase); AKR1C3 (5.2 fold increase), SRD5A1 (2.1 fold increase); AKR1C2 (3.4 fold increase); and AKR1C1 (3.1 fold increase), where the latter two enzymes produce 3α-androstanediol and 3β-androstanediol, respectively from 5α-DHT  (Figure 3). These findings were observed in Affymetrix expression microarrays and validated by qRT-PCR . In a separate study, the relative expression of the following transcripts changed in castrate-resistant metastases versus primary prostate tumors; CYP17A1 (16.9 fold increase) HSDD3B2 (7.5 fold increase) AKR1C3 (8.0-fold increase), SRD5A1 (2.6 fold increase) and SRD5A2 (9.4 fold decrease) . Since these studies show that the ratio of AKR1C3:SRD5A2 transcripts increases, this may result in a decrease in the ratio of 5α-DHT:T within tumor samples. It was found that primary prostate tumors from eugonadal patients had a 5α-DHT:T ratio of 10:1, while this ratio was reversed to 0.25:1 in metastatic tumors . Importantly, T levels measured by liquid chromatography-mass spectrometry in metastatic tumors are well within the range to stimulate AR, These studies suggest that in CRPC, the disease may become more dependent on T than 5α-DHT.
The second piece of evidence showing the importance of intratumoral androgen synthesis following chemical or surgical castration comes from xenograft studies. Using a LNCaP (an AR dependent prostate cancer cell line) mouse xenograft model for CRPC, increases in transcripts for androgen synthesizing enzymes were observed following extended castration and were coincident with increased PSA . These studies suggest that during tumor reoccurrence, not only is there an increase in local androgen synthesis but this is sufficient to cause the induction of androgen sensitive genes. Importantly, there were additional changes in proteins responsible for the build up of free cholesterol and cholesterol synthesis (LDL-r, SR-B1, HMG-CoA reductase, StAR ACAT1, 2 and ABCA1) [88, 89] as well as changes in the expression of side-chain cleavage enzyme (CYP11A1) , suggesting that denovo steroidogenesis from cholesterol may take place in CRPC. Metabolism studies in the castrate resistant tumors provided evidence for denovo synthesis of 5α-DHT from [14C]-acetate . In addition, metabolism studies with [3H]-progesterone provided evidence that intermediates in the backdoor pathway to 5α-DHT accumulate . One caveat with these xenograft experiments is that in mice, CYP17A1 is not expressed in the adrenal, placing additional selection pressure on these tumors to synthesize their own androgens. However, the importance of this work is that even under conditions in which the mice are castrated and the adrenal is not producing DHEA, the tumors adapt to make their own androgens. These data indicate that following chemical or surgical castration, CRPC can be promoted by intratumoral androgen synthesis, and that denovo synthesis from cholesterol may also occur.
Third, abiraterone acetate (a CYP17α-hydroxylase/CYP17,20 lyase inhibitor), has shown important clinical response in individuals with CRPC leading to a reduction in bone metastases . This response suggests that non-localized disease is still dependent on androgens, since this inhibitor blocks the conversion of either pregnenolone to DHEA or progesterone to Δ4-androstene-3,17-dione (Figure 3). This blockade could occur either in the adrenal or the prostate to prevent DHEA formation. Irrespective of where the blockade occurs, subsequent conversion of DHEA to T in the prostate is prevented. An important clinical outcome of abiraterone was a further decline in plasma T levels in CRPC by one log unit. Use of abiraterone acetate to treat CRPC can have the unintended consequence of inhibiting the conversion of pregnenolone to DHEA in the adrenal and lead to the diversion of pregnenolone into desoxycorticosterone which is a potent mineralocorticoid with glucocorticoid activity. To prevent the overproduction of desoxycorticosterone, abiraterone acetate is usually co-administered with dexamethasone to suppress the adrenal-pituitary axes and block ACTH formation . Clearly, the effectiveness of abiraterone acetate has stimulated a reemergence of therapeutic approaches to block adaptive androgen synthesis in CRPC.
Fourth, in a recent small clinical study involving 10 patients, 80% showed slower progression of CRPC when they were given ketoconazole (a less selective CYP17α-hydroxylase/CYP17,20 lyase inhibitor than abiraterone acetate) in combination with dutasteride (Figure 3) . A combination of agents that block androgen synthesis at multiple steps in CRPC could be a useful treatment strategy.
AKR1C3 is a prime therapeutic target downstream of CYP17α-hydroxylase/CYP17,20 lyase. This enzyme catalyzes the penultimate step in T biosynthesis in the prostate. Moreover, in metastatic disease we have seen that the ratio of 5α-DHT: T clearly favors T accumulation, suggesting that as the disease progresses, T may be the more dependent hormone . These findings also raised the question whether AKR1C3 is the only reductive 17β-HSD expressed in prostate cancer. Type 3 17β-HSD (HSD17B3), also known as androgenic 17β-HSD catalyzes the conversion of Δ4-androstene-3,17-dione to testosterone in the Leydig cells, and was thought to be Leydig cell specific [94, 95]. Recently, evidence has emerged that this enzyme is also expressed in prostate cancer, but based on transcript levels AKR1C3 appears to be the dominant player . Interestingly, AKR1C3 is potently and selectively inhibited by indomethacin suggesting that NSAID analogs that do not inhibit COX-1 or COX-2 might be effective agents for CRPC . Taken together, it is apparent that alterations in androgen synthesis and androgen metabolism pathways are frequently observed in CRPC, presenting new opportunities for means to target AR activity and resultant tumor progression.
The studies described herein illustrate that deregulation of the mechanisms that control both AR activity and androgen levels promote disease progression and lethal tumor phenotypes. With this knowledge in hand, critical next steps and questions should be considered. First, how can the information be clinically translated? For example, if it is known that a patient harbors tumors with somatic mutations of AR, AR splice variants, or altered cofactor expression, is this information useful for directing treatment options? Second, are the mechanisms that underlie recurrent AR activity mutually exclusive? It is likely that different tumors display a repertoire of these mechanisms. Thus, molecular profiling of the tumor may ultimately have diagnostic and therapeutic value. In the future, a component of this molecular profiling must include sensitive and specific methods for measuring intratumoral androgen levels so that changes in the expression of androgen synthesizing genes can be validated at the functional level and enzymes targeted with specific inhibitors. Third, can more potent ligands be produced that suppress receptor function given the known mechanisms of androgen and AR deregulation? Recently, the darylthiohydantoins RD162 and MDV3100 were developed as more potent ligands for AR than bicalutamide. These compounds also reduce the efficiency of nuclear accumulation of AR, and impair binding to androgen response elements and coactivator recruitment. Of the first 30 patients with CRPC treated with MDV3100, 13/30 (43%) showed sustained declines in PSA . There is also promise in designing single agents that could block androgen biosynthesis and AR function simultaneously. Analogs of abiraterone acetate e.g. 3-hydroxy-17-(1H-benzimidiazole-1-yl)androsta-5,16-diene, not only block CYP17α-hydroxylase/CYP17,20-lyase but are also potent anti-androgens and cause marked down-regulation of AR protein expression . These agents are now entering clinical trial (TOK-001) and could be potentially important if CRPC is characterized by adaptive androgen synthesis and concurrent AR mutation to make the receptor more promiscuous to other ligands. Fourth, what mechanisms underlie the observed induction of enzymes that drive intratumor androgen synthesis? It will be important to determine the mechanisms by which these genes are induced and how their expression levels become permanently elevated. The promoters of the AKR1C1-AKR1C3 genes contain a number of half-sites for steroid hormone response elements. In addition, they all contain an anti-oxidant response element . Whether inflammatory responses leading to reactive oxygen species generation to activate the antioxidant response element occurs remains an open question. Fifth, can AR cofactors or other critical effectors of ligand-bound receptor be developed as therapeutic targets? Advances in this area could provide new means to suppress AR activity, even in the presence of deregulated ligand synthesis.
In summary, it is clear that current means of therapeutic intervention for disseminated prostate cancer are circumvented in CRPC by both AR deregulation and aberrant androgen synthesis. This lethal pairing represents a major mechanism of tumor progression, and future efforts for development of new means to treat CRPC will need to consider both partners in crime.
This article was written with partial support from grants from the National Institutes of Health 1R01-CA90744 and P30-ES013508 awarded to TMP, CA116777 and CA099996 to KEK. The authors thank Dr. Steven Balk, Dr. Daniel Gioeli, Matthew Schiewer, and Sucharitha Balasubramaniam for critical commentary, and Elizabeth Gosnell for assistance with artwork. We regret omissions of related citations due to space constraints.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.