The Biology of Castrate Resistant Prostate Cancer

Background

Prostate cancer is one of the most commonly diagnosed cancers and the second leading cause of cancer death for males in the United States¹. Although there are recently approved therapeutic options for men with advanced and metastatic disease, the unfortunate reality is that advanced prostate cancer is inevitably fatal. Hormonal androgen deprivation therapy (ADT)² is the standard of care once a patient has recurrent disease following primary surgical or radiation therapy; however, the benefits from ADT are typically short-lived. Recurrent disease that follows ADT treatment is termed castrate-resistant prostate cancer (CRPC)², which is the most aggressive and lethal form of prostate cancer. The current treatment regime for CRPC consists primarily of chemotherapy with docetaxel, but other agents such as the immunotherapy agent sipuleucel-T, the taxane cabazitaxel, the CYP17 inhibitor abiraterone acetate, and the androgen receptor (AR) antagonist enzalutamide are also FDA approved for the treatment of CRPC^3-8. While there are clinical trials testing the potential therapeutic benefits of many other compounds, CRPC remains an incurable disease^4,6,9. Recent advancements in our understanding of the biochemical and genetic pathways critical to CRPC progression have identified many novel potential druggable targets as well as biomarkers of disease progression^10,11. This review will focus on the biological aspects of aggressive prostate cancer, biomarkers, therapeutic targets, and future directions for the treatment of CRPC.

The Androgen Receptor

It is well-documented that prostate cancer cells depend on the androgen receptor (AR) and steroid hormones for continued oncogenic growth. Though ADT targets these pathways, this therapy can also select for tumor cells with several alternative survival mechanisms that allow AR to continue to function in the absence of circulating androgens. One such mechanism that generates androgen insensitivity seen in CRPC involves either the amplification of the AR gene or increased AR mRNA expression, along with a concomitant increase in the abundance of AR protein expression in CRPC tumor cells¹². Another survival mechanism relies on the specific mutations in AR that enable binding and activation of this receptor by other steroidal hormones such as progesterone, hydrocortisone, estradiol¹³, or even AR antagonists such as flutamide¹⁴. One critical pathway for CRPC is the intratumoral synthesis of steroidal hormones via upregulation of the cytochrome P450 gene CYP17A1, steroid-5-alpha-reductase, alpha polypeptide 1 (SRD5A1), aldo-keto reductase family 1, member C3 (AKR1C3), and hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 2 (HSD3B2)¹³. These enzymes contribute to de novo synthesis of dihydrotestosterone (DHT) from endogenous cholesterol, as well as from uptake of exogenous dehydroepiandrosterone sulfate (DHEA-S), providing the rationale for treatment with the CYP17A1 inhibitor abiraterone. Furthermore, post-translational modifications such as aberrant phosphorylation of AR, by the Src tyrosine kinase on Y534 leads to increased sensitivity of the AR to low levels of circulating androgens¹⁵. Wang et al. showed through gene expression profiling that AR in CRPC selectively unregulates cell cycle M-phase genes promoting CRPC growth², suggesting that AR in CRPC may regulate a distinct set of genes independent of the typical AR-dependent transcriptional program.

Genome wide localization analysis via chromatin immunoprecipitation and sequencing (ChIP-seq) of AR in CRPC tissues has identified a set of genes regulated by AR that are distinct from those observed from cultured prostate cancer cell lines¹⁶. Analysis of DNA binding motifs adjacent to AR binding sites determined that unlike cultured cells in which AR interacts with FOXA1 and NF-1, in vivo AR interacts with E2F, MYC, and STAT5¹⁶.

PI3K-AKT-mTOR pathway

Activation of the PI3K-AKT-mTOR pathway is extremely common, if not universal, in CRPC. It is critical for the regulation of cell survival, apoptosis, cell proliferation, autophagy, metabolism, and protein synthesis, and has been extensively studied in prostate cancer¹⁷. PI3K is activated by a wide range of growth factor receptors and signaling pathways, including epidermal growth factor receptor (EGFR), insulin-like growth factor 1 receptor (IGF-1R), fibroblast growth factor receptor (FGFR), and platelet-derived growth factor receptor (PDGFR). Activated PI3K activates PDK1, which in turn activates AKT, while PI3K can be inactivated by the PTEN tumor suppressor^18-20. AKT separately phosphorylates and activates mTOR, which promotes cell cycle progression and growth and provides positive feedback via phosphorylation of AKT. Aberrant constitutive AKT activation is one of the most frequent pathway alterations observed in a number of different cancers. Recently, Chin et al. demonstrated that the AKT2 isoform was specifically required for cellular maintenance in prostate cancer tumors lacking PTEN expression²¹. Genome sequencing analysis of CRPC has shown the PTEN gene is mutated in 48% of CRPC samples, and that 33% of samples have mutations in genes that interact with PTEN, resulting in 81% of samples harboring some mutation in the PTEN interaction network²². A number of PI3K and mTOR inhibitors are currently under investigation in clinical trials for CRPC including the dual inhibitor NVP-BEZ235²³, and the mTOR inhibitor RAD001 or everolimus^24,25. Of these compounds, NVP-BEZ235 likely has the most promise, although everolimus may be effective in subsets of patients.

Receptor tyrosine kinase (RTK) growth factor pathways

Multiple RTK growth factor signaling pathways including EGFR^26,27, IGF-1R^28,29, FGFR³⁰, PDGFR^31,32, and HGFR/c-MET^33-35 pathways have been investigated extensively in prostate cancer progression due to their activation of the PI3K-AKT and RAS-MAPK pathways. Combined loss of PTEN with activation of the RAS-MAPK pathway can cooperate to induce epithelial to mesenchymal transition (EMT) and metastases³⁶. Several clinical trials have been conducted targeting EGFR, including two using lapatinib, one of which was negative³⁷, and another that was more encouraging³⁸ with single agent activity in a subset of patients. A third trial using cetuximab showed significant progression free survival in approximately half of the treated patients³⁹. Antibodies against IGF-1R have shown some efficacy in xenografts²⁹ and in human trials⁴⁰.

Recent epidemiological studies have determined that the generic anti-diabetes drug metformin results in lower prostate cancer specific mortality⁴¹, suggesting that not only insulin and IGF, but also glucose regulation may impact patient survival. RTK growth factor pathways can allow cells to meet bioenergetic demands of rapid proliferation by influencing cell signaling via phosphorylation. The Warburg effect has been recognized for nearly a century as perturbation of cancer cell metabolism, but it has only recently become the focus of intense investigation as an avenue for potential therapy^42,43. In the Warburg effect, cancer cells undergo aerobic glycolysis rather than mitochondrial respiration and oxidative phosphorylation in order to generate the necessary precursors for protein, nucleotide, and lipid synthesis that are essential to rapid cell growth. In the reverse Warburg effect, perturbations of metabolic and catabolic metabolism can lead to a shuffling of nutrients between prostate cancer epithelial cells and surrounding stromal cells, in which pyruvate and lactate are transported to prostate tumor cells by the stroma⁴⁴. Almost all glycolysis genes are overexpressed in prostate cancer, especially in advanced disease⁴⁵. These data have led to calls for clinical trials employing combination therapies of RTK inhibitors or PI3K-AKT inhibitors with metformin⁴⁶.

Analysis of CRPC Genomes

Although primary prostate cancers are characterized by genomic translocations that place androgen-responsive promoters such as TMPRSS2 adjacent to ETS-family transcription factors such as ERG⁴⁷, ETV1⁴⁸, ETV5⁴⁹, and ETV4⁵⁰, metastatic prostate cancers can harbor rare gene rearrangements that generate fusions of ubiquitin conjugating enzymes with KRAS⁵¹. Integrated gene expression and copy number analysis of prostate cancers also indicates that more aggressive prostate cancers have increased copy number alterations relative to less aggressive cancers⁵². Moreover, PI3K-AKT was altered in 100% of metastases, while RAS-RAF signaling was altered in 90% of metastases in this study⁵². Exome sequencing of CRPC has identified several recurrent mutations in transcription factors and epigenetic factors that interact with AR such as FOXA1, MLL2, UTX, and ASXL1²².

Epigenetic Pathways

The hypermethylation of CpG islands in gene promoters can inhibit expression of genes such as GSTP1⁵³ and has been implicated in tumorigenesis and cancer progression. Recent data have suggested that methylation of microRNA genes that target AR such as miR-34a can lead to increased AR expression, and that forced expression of miR-34a can reduce AR levels⁵⁴. Since epigenetic modification of genetic DNA is a potentially reversible process, it is an enticing target in the treatment of CRPC. One such pathway target involves the inhibition of DNA methyltransferases (DNMT) to decrease the effects of methylation-related gene silencing. 5-Aza-2’-deoxycytidine (5-Aza-CdR) has been shown to increase expression of miR-146a in LNCaP cells and delay progression of tumor xenografts⁵⁵, but more data is needed on drug efficacy in the face of docetaxel-treated CRPC.

The enhancer of zeste 2 (EZH2) protein is a histone methyltransferase and the catalytic subunit of the Polycomb repressive complex 2 (PRC2) that tri-methylates histone 3 lysine 27 (H3K27me3) to repress many developmental genes, especially in metastatic prostate cancer⁵⁶. EZH2 is elevated in metastatic prostate cancer versus clinically localized disease or benign prostate^57,58, and high expression of EZH2 in patients correlates with an increased probability of developing skeletal metastases, along with seminal vesicle and lymph node infiltration⁵⁹.

Recent data from exome sequencing of CRPC have shown that there are multiple mutational alterations in the chromatin and histone modifying genes. Grasso et al. found that in pre-treated lethal metastatic CRPC tumors, there are interactions between the MLL complex family and AR, further demonstrating that there is dysregulation in epigenetic activation, commonly seen in CRPC²². Berger et al. showed that chromatin modifying genes CHD1, CHD5, and HDAC9 were mutated in 43% of sequenced Gleason 7 or higher prostate cancer tumors⁶⁰. Specifically, CHD1 sequencing exhibited splice site mutations as well as intragenic breakpoints—all leading to truncated protein expression⁶⁰.

Histone and chromatin-remodeling complexes are potential targets in regards to CRPC therapy, with targeting of histone deacetylases (HDACs) that facilitate AR-mediated transcription activation and repression. However, current HDAC inhibitors, such as vorinostat^61,62 and panobinostat⁶³, have shown high rates of side effects and disappointing efficacy in the treatment of docetaxel-refractory CRPC.

Nevertheless, there have been exciting developments in the development of inhibitors of bromodomain-containing proteins that recognize and bind to acetylated lysines of histone proteins. Recently, a new group of small molecules has emerged as novel inhibitors of Bromodomain Containing Protein 4 (BRD4). BRD4, along with the Mediator complex, binds at super-enhancer sites to facilitate initiation of transcription of target genes⁶⁴. BRD4 inhibitors, such as JQ1⁶⁵, bind to bromodomains of proteins such as BRD4, and inhibit BRD4 from binding to super-enhancers of known proto-oncogenes, including MYC^66,67. A mouse model of aggressive prostate cancer with simultaneous loss of PTEN and p53 tumor suppressor genes, termed RapidCaP, demonstrated highly penetrant metastases and activation of MYC⁶⁸. Moreover, these castrate resistant tumors were sensitive to BRD4 inhibition using JQ1 the inhibitor⁶⁸. Additional studies⁶⁹ using JQ1 and the orally bioavailable BRD4 inhibitor IBET762⁷⁰, found that BRD4 inhibitors disrupt AR signaling, and recruitment to and activation of downstream target genes such as the TMPRSS2-ERG gene fusion⁶⁹.

EMT and SOX family genes

Typically, prostate cancer mortality is often related to metastasis to the bone, adrenal gland, liver and lung⁷¹. The epithelial to mesenchymal transition (EMT) is a major step in the metastatic process. To metastasize cancer cells need to acquire migratory and invasive capabilities, a process that involves EMT⁷². EMT encompasses vast molecular changes including gain of mesenchymal markers such as vimentin and N-cadherin, and loss of epithelial markers such as E-cadherin, mediated by aberrant developmental signaling pathway activation that allows epithelial cells to discard differentiated characteristics and acquire migratory and invasive capabilities typical of mesenchymal cells⁷². These changes include the loss of cell-cell adhesion, planar and apical-basal polarity, increased motility, and resistance to apoptosis and anoikis (cell death due to the detachment from the extracellular matrix)^72,73. Among the developmental signaling pathways that are aberrantly activated during EMT is the TGF-β signaling pathway, a highly studied major inducer of EMT⁷⁴. The canonical TGF-β pathway is stimulated via TGF-β induced receptor complex activation, leading to phosphorylation of SMAD 2/3. Subsequently, these SMADs form a trimer with SMAD4 and translocate to the nucleus and associate with other transcription factors to transcribe EMT-inducing genes⁷⁵.

Recently, it was found that SOX4 is a master regulator of TGF-ß induced EMT via induction of EZH2⁷⁶. Tiwari et al demonstrated that SOX4 directly activates EZH2 expression upon TGF-β treatment and that forced expression of EZH2 can overcome SOX4 knockdown and restore TGF-β induced EMT⁷⁶. Moreover, Wang et al. found that, in prostate cancer cells, SOX4 knockdown inhibited TGF-β induced EMT, while SOX4 over expression promoted adoption of the mesenchymal phenotype⁷⁷. They also demonstrated that TMPRSS2-ERG is critical for TGF-β induction of SOX4 expression⁷⁷. Tiwari et al. and Zhang et al. both demonstrated that ectopic expression of SOX4 could induce EMT by increasing the expression of mesenchymal markers and decreasing the expression of epithelial markers^76,78. In addition, SOX4 knockdown was sufficient to cause a reversion from a mesenchymal to epithelial phenotype after a 15-day TGF-β treatment⁷⁶.

Another SOX family factor, SOX9, has also been implicated in prostate cancer progression. Deletion of SOX9 in two different mouse models (TRAM and Hi-Myc) inhibited prostate cancer initiation⁷⁹. ERG redirects AR to a cryptic enhancer of SOX9 to activate SOX9 expression, and knockdown of SOX9 inhibits invasion and growth of VCaP cells in vitro and in vivo⁸⁰. SOX9 cooperates with PTEN deletion to drive prostate tumorigenesis⁸¹, and it activates expression of Wnt pathway components such as LRP6 and TCF4⁸². Like SOX9, SOX4 also plays an important role in Wnt signaling via direct interaction with β-catenin^83,84. SOX4 can act as an oncogene in prostate cells⁸⁵, and activates expression of additional Wnt pathway components such as FZD3, FZD5, and FZD8^83,86.

WNT signaling

There have been many lines of evidence that suggest that Wnt signaling may be important in CRPC⁸⁷. Wnt pathway genes are frequently mutated in metastatic CRPC²². WNT7B ligand is a direct transcriptional target of AR, and can induce osteoblastic bone lesions⁸⁸. TMPRSS2-ERG directly activates LEF1 transcription and expression of ligands such as WNT1 and WNT3A⁸⁹. Moreover, AR interacts directly with β-catenin in CRPC xenografts, whose gene expression profiles exhibit enhanced Wnt signaling⁹⁰. Interestingly, a novel small compound (iCRT3) that inhibits β-catenin transcriptional activity blocks AR binding to target genes, inhibits growth of tumor xenografts, and interferes with self-renewal of bicalutamide resistant cells⁹¹.

Notch and Hedgehog

The Notch and Hedgehog pathways are also developmental pathways that likely play roles in prostate cancer progression. Notch signaling is important for prostate differentiation and maintenance of prostate stem cells⁹². Hedgehog signaling has also been associated with aggressive prostate cancer, and ADT alone or combined with chemotherapy induces hedgehog pathway activation⁹³. Moreover, combined inhibition of AR and hedgehog signaling synergizes to inhibit growth of CRPC xenografts⁹⁴. Docetaxel resistant CRPC cells survive via activation of Notch and Hedgehog pathways that inhibit apoptosis, and depletion of these cells results in resensitization to chemotherapy⁹⁵. Thus, combination therapies that target developmental pathways along with targeting of AR and/or conventional chemotherapy may prove effective in CRPC.

lncRNA-mediated pathways

There is much data suggesting that long non-coding RNAs (lncRNA) greater than 200 base pairs in length play a critical role in tumor biology^96-100. lncRNAs are typically transcribed by RNA polymerase II and are associated with epigenetic modification of histones. These lncRNAs partner with PRC1 and PRC2, leading to downstream ubiquitination and methylation activity that inhibits gene expression¹⁰¹. Furthermore, many lncRNAs have been linked to prostate cancer^{96,97,102-107}. In one study, two lncRNAs, PRNCR1 and PCGEM1, were shown to bind to AR and increase ligand-dependent and ligand-independent AR-mediated gene expression, which causes downstream cellular growth signaling and cancer proliferation¹⁰⁷. However, subsequent studies have contradicted these findings, and suggest that while PCGEM1 is associated with prostate cancer, neither of these lncRNAs interacts with AR signaling nor are prognostic in prostate cancer¹⁰⁸. Nevertheless, other lncRNAs play important roles in prostate cancer. Pickard et al. demonstrated that transiently increased GAS5 lncRNA levels upregulated basal apoptosis in PC3 prostate cancer cells, and a similar effect was seen in 22Rv1 cells¹⁰². Prensner et al. found that SChLAP1 lncRNA overexpression, which is involved in SWI-SNF activity knockdown, was observed in 25% of prostate cancers and was an independent predictor of aggressive disease¹⁰³. Other lncRNAs that have been linked to CRPC include MALAT-1¹⁰⁴, CTBP1-AS¹⁰⁵, and Linc00963¹⁰⁶. Linc00963, specifically, was shown to enhance the evolution from androgen-dependence to androgen-independence through the EGFR pathway¹⁰⁶. From these studies, it is clear that lncRNAs are intimately involved in prostate cancer biology and that dysregulation of lncRNA expression may lead to tumor suppressor antagonism as well as to castrate resistance. Future goals in lncRNA research include defining specific lncRNAs in the human genome, understanding which lncRNAs are altered in the course of tumorigenesis, and delineating the structure and binding mechanisms of lncRNAs to their targets^96,98-100.

Angiogenic Pathways

Angiogenesis refers to the growth of new blood vessels from existing vasculature¹⁰⁹. This process is an important pathway for CRPC progression, and is correlated with both increased rate of disease progression and decreased survival¹¹⁰. The most common target of angiogenesis is vascular endothelial growth factor (VEGF) through the use of an anti-VEGF antibody, such as bevacizumab¹⁰⁹. VEGF can also drive EMT¹¹¹, and this pathway is currently a target of several therapeutic clinical trials testing new drugs including itraconazole¹¹, bevacizumab¹¹², tasquinimod¹¹³, and ramucirumab¹¹⁴. Alternative strategies include inhibiting the binding of VEGF using VEGF-like receptors, such that VEGF is unable to interact with its normal target VEGFR receptor. VEGF ligand directed drugs have seen success in improving patient survival in other cancers, such as colorectal, breast, and non-small-cell lung cancer. However, in two prospective randomized control trials, VENICE and CALGB 90401, neither docetaxel/aflibercept nor docetaxel/bevacizumab combinations showed any significant improvement in median overall survival in patients with CRPC^115,116. Docetaxel plus bevacizumab did show significant improvement in PSA decrease and progression-free-survival when compared to docetaxel alone, but this was not observed in the VENICE trial. A newer small molecule inhibitor of VEGFR2 and MET, cabozantinib, early on displayed a decrease in growth of breast, lung, and glioma tumor models while increasing apoptosis¹¹⁷. Later, in a prospective randomized control trial, patients with CRPC who took cabozantinib significantly increased progression-free survival when compared to placebo¹¹⁸. Furthermore, other trials demonstrated that cabozantinib decreased narcotic use, significantly improved sleep quality¹¹⁹, and inhibited prostate cancer bone growth when tested on androgen-sensitive and castration-resistant cell lines¹²⁰. Other angiogenic strategies are to target the downstream activation pathways of VEGF, or to bypass the VEGF pathway completely and focus on established CRPC vessels and nutrient delivery^109,110.

Biomarkers

Currently, there is a critical need for biomarkers of prostate cancer that can distinguish between locally indolent and metastatically aggressive disease¹²¹. The only FDA approved biomarker, prostate specific antigen (PSA), has historically been used for screening and detection of prostate cancer, but not for prognosis nor detection of CRPC. A major issue with PSA is that it detects prostate cells, not prostate cancer cells, and thus can result in over diagnosis and over treatment. While results of the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial showed no reduction in prostate cancer specific mortality when comparing systematic annual PSA screening to opportunistic screening¹²², the European Randomized Study of Screening for Prostate Cancer (ERPSC) found a small reduction in mortality (1 death per 1000 men screened) in a PSA-screening naïve population¹²³. However, several potential flaws in the ERPSC study¹²⁴ and the associated harms of over diagnosis and over treatment led the US Preventive Services Task Force (USPSTF) to recommend against PSA screening^125,126. Given the recent controversy in its screening effectiveness, PSA is now recommended primarily for the determination of prostate cancer progression and recurrence.

However, there are potential replacements for PSA for detection screening under development, including PCA3 and the ETS fusion gene TMPRSS2-ERG^127-133. TMPRSS2-ERG has been shown to be a predictive indicator of prostate cancer development in patients who present with high-grade prostatic intraepithelial neoplasia¹³⁰. FISH analyses of CRPC and metastatic prostate cancers found that TMPRSS2-ERG was detected more often in metastatic cancer, and that TMPRSS2-ERG positivity was strongly correlated to both AR and ERG expression¹³¹. TMPRSS2-ERG expression has also been seen in otherwise histologically benign radical prostatectomy as well as in cystoprostatectomy specimens. TMPRSS2-ERG and PCA3 can also be detected in patient urine post-digital rectal exam¹³⁴. However, there are still conflicting data regarding the prognostic value of PCA3 or TMPRSS2-ERG. A 2011 study by Danila et al. found that the TMPRSS2-ERG fusion gene could be accurately assayed in circulating prostate tumor cells present in the blood of CRPC patients, but did not show that the presence of the fusion was a significant factor in abiraterone acetate treatment response¹²⁸. Another prospective study of 322 patients illustrated that urinary PCA3 or TMPRSS2-ERG scores were not reliable in staging advanced prostate cancer.

Recent integrated bioinformatics network analyses that have included data mining and profiling of both animal models and human specimens have generated some interesting candidate biomarker gene sets for distinction of aggressive from indolent disease¹³⁵ and drivers of aggressive prostate cancer¹³⁶. Irshad et al. identified a panel of three genes (FGFR1, PMP22, and CDKN1A) that could predict indolence outcome for low Gleason score (Gleason 6) patients via meta-analysis of multiple studies followed by independent validation on a cohort of 95 mRNA samples and 44 fixed biopsy samples at the protein level¹³⁵. Aytes et al. performed cross-species analysis of mouse and human prostate cancer gene expression patterns to identify FOXM1 and CENPF as synergistic regulators of aggressive disease¹³⁶. They also found that FOXM1 and CENPF could function as prognostic biomarkers of metastasis in two independent prostate datasets^137,138. Another distinct 16-gene signature of AR target genes derived from ChIP-seq of AR in CRCP tissues¹⁶ was able to accurately predict CRPC and prostate cancer recurrence in two distinct patient cohorts^52,137.

Our research group at Emory University recently completed RNAseq analysis of 100 formalin-fixed paraffin embedded prostatectomy specimens, and identified a 24-gene biomarker panel that robustly predicts biochemical recurrence following surgery¹³⁹. Our biomarker panel accurately predicted recurrence in an independent patient cohort⁵², and outperformed previously developed RNA biomarkers developed by Myriad Genetics¹⁴⁰. Whether this set of 24 biomarker genes will also detect aggressive CRPC or discriminate aggressive from indolent disease requires further research on additional patient cohorts.

Conclusion

Recent advancements in the molecular understanding of CRPC have given us a number of potential biological targets for the treatment of CRPC. Many of the pathways and targets covered in this review currently have agents that are undergoing clinical trials, and some are FDA approved for the treatment of CRPC. Unfortunately, most of these pharmaceutical agents only moderately increase survival and CRPC still remains incurable. However, recent discoveries and avenues of research may enable more effective molecularly targeted therapies as well as a better understanding of the mechanisms of CRPC.

Figure 1. Signaling Pathways in Castration Resistant Prostate Cancer.

(A) TGF-β pathway is stimulated via TGF-β induced receptor complex activation, leading to phosphorylation of SMAD2/3 and subsequently form a trimer with SMAD4 and translocate to the nucleus and associate with other transcription factors to transcribe SOX4 which activates EZH2 expression leading to EMT. (B) Prostate cancer cells depend on the androgen receptor (AR) pathway and steroid hormones for continued oncogenic growth. This can occur by way of AR over expression or by AR gene amplification. Mutations in AR allow binding and activation of AR by other steroidal hormones. Wnt pathway genes are frequently mutated in CRPC and are transcriptional targets of AR. (C) Activation of the PI3K-AKT-mTOR pathway is extremely common in CRPC. Activated growth factor receptor tyrosine kinases (e.g. EGFR and IGF-1R) leads to activated PDK1, which in turn activates AKT. AKT separately phosphorylates and activates mTOR which promotes cell cycle progression, protein synthesis and decreased apoptosis. AKT can interact with AR in an androgen independent manner. (D) Activation of Receptor Tyrosine Kinase (RTK) pathway (PDGFR, HGFR/c-MET, etc) leads to proliferation through RAS-MAPK. Combined with loss of PTEN, overactive RAS-MAPK can induce EMT. (E) Intratumoral synthesis of steroidal hormones from cholesterol via upregulation of the cytochrome P450 gene CYP17A1. (F) Loss of the PTEN Tumor Suppressor promotes aberrant PI3K-AKT-mTOR signaling. (G) Aberrant Y534 phosphorylation by Src increases AR sensitivity to androgens. (H) Epigenetic Pathways: Hypermethylation of CpG islands in gene promoters inhibits expression of tumor suppressor genes or miRNAs targeting AR.