Principles of stem cell biology and cancer: future applications and therapeutics. Edited by T. Regad, T. J. Sayers and R. C. Rees. John Wiley & Sons (2015)
Common mutations associated with prostate cancer
Prostate cancer is a common malignancy in men, often showing no symptoms of disease and taking a variable clinical pathological course. Most cases of prostate cancer behave in an indolent manner. However, some cases are characterized by a highly aggressive phenotype, as seen in tumours resistant to standard androgen-deprivation therapy (Kirby et al., 2011). In agreement with the stochastic (clonal) evolution model, the level of prostate cancer malignancy appears to correlate with the prevalence of somatic mutations that affect key pathways involved in prostate physiology (e.g. AR signalling) or signalling pathways, associated with cell survival, apoptosis and DNA repair (Grasso et al., 2012). Several studies have given evidence of different mutation patterns characteristic of different stages of prostate cancer progression, some of which are potentially associated with malignant transformation (Taylor et al., 2010; Kumar et al., 2011; Barbieri et al., 2012; Grasso et al., 2012). The current diagnostic techniques for prostate cancer screening involve digital rectal examination (DRE), transrectal ultrasonography (TRUS) and biopsy histopathological, but these procedures are incapable of providing sufficient indication for the appropriate course of therapy (Madu and Lu, 2010). An insight into prostate cancer mutation patterns could help us better assess the implications of the prostate tumourigenesis ‘mutation signature’ and highlight patterns in aggressiveness and disease progression, providing a molecularly tailored treatment strategy (Kumar et al., 2011). The last 2 decades of cancer research have provided an advancement in genome characterization technologies, bringing most significant prostate cancer-specific aberrations into sharper focus.
Prostate cancer shows a relatively low frequency of point mutations, ranging from one to two alterations per million base pairs, which is much lower than in carcinogen-driven tumours such as lung cancer or melanoma, but a similar rate to that seen in breast, ovarian and renal cancers (Greenman et al., 2007; Pleasance et al., 2010; Berger et al., 2011). Low levels of sequence alterations may explain the lower aggressiveness of prostate cancer compared to other tumours. The prevalence of mutations observed in prostate malignancies is highly dependent on their purity, stage, histological grade (Gleason score) and previous exposure to treatment (Baca and Garraway, 2012). The most common and prostate cancer-specific aberration identified for both primary and advanced tumours is a structural rearrangement in ETS DNA-binding family genes, mostly affecting ERG family members (ETV1, ETV4 and ETV5) (Tomlins et al., 2008). In this mutation, the ETS family-member ERG oncogene is adjoined to a highly active androgen-driven promoter TMPRSS2, leading to its high expression in the prostate epithelium. More recent observations have associated TMPRSS2-ERG fusions with prostate-specific deletion of chromosome 3p14, implicating FOXP1, RYBP and SHQ1 as potential cooperative tumour suppressors (Taylor et al., 2010). Moreover, the overexpression of ERG oncoprotein appears to be involved in a positive regulation of c-Myc and prostaglandin-mediated signalling, potentially contributing to tumour progression (Mohamed et al., 2011; Sreenath et al., 2011). ERG activation has been found in 50 – 70% of PSA-positive prostate cancer; up to 90% of such activations were found to be TMPRSS2-ERG fusions, and despite variations in ethnic distribution, this mutation is to date the most specific described for prostate cancer (Mosquera et al., 2007; Tomlins et al., 2008).
The androgen-regulated pathway has been shown to be crucial, not only for prostate development, but also for prostate oncogenesis (Prins and Putz, 2008). ARs, through their nuclear translocation and interactions with transcriptional factors, regulate the transcription of several prostate genes involved in cellular proliferation and differentiation. Clinical observations show that up to 90% of prostate cancers are dependent on androgens at initial diagnosis. However, on androgen-deprivation treatment, the majority of them develop castration-resistant prostate cancer (CRPC), associated with transactivation of the AR (Kirby et al., 2011). This is consistent with the exome sequencing of prostate cancer, which shows little or no alteration in the AR locus early in prostate cancer and a dramatic AR amplification and point-mutation rate in up to 50% of CRPC and metastatic specimens tested (Visakorpi et al., 1995; Koivisto et al., 1997; Grasso et al., 2012; Linja and Visakorpi, 2004). Kumar et al. (2011) showed a significant novel nonsynonymous variant of AR gain-of-function mutation in CRPC. Malfunctioning AR signalling in primary prostate cancer may be caused by mutations in AR-interacting genes such as NCOR2, NRIPI1, TNK2 and EP300 (Taylor et al., 2010). AR signalling is also regulated by a range of abnormally expressed chromatin/histone remodelers, such as members of the MLL complex (MLL2, MLL, ASH2L), UTX, ASHL1, CHD1 and the direct AR cofactor FOXA1 (Barbieri et al., 2012; Grasso et al., 2012).
Early alterations in prostate cancer oncogenesis predominantly involve signalling that regulates growth, proliferation and normal prostate development. Phosphoinositide-3-kinase (PI3K) is one of the main signalling pathways involved in cellular proliferation that is aberrantly mutated in the early stages of prostate cancer development (Barbieri et al., 2012; Grasso et al., 2012). Overactivation of PI3K signalling is caused by frequent inactivation of the tumour-suppressor gene PTEN, which is unable to counteract PI3K signalling. Loss of heterozygosity at the PTEN locus has been found in up to 70% of primary prostate cancers, whereas ‘inactivation’ mutations have been found in 5 – 10%, although with a marked increase of frequency in advanced tumours (Cairns et al., 1997; Gray et al., 1998; McMenamin et al., 1999; Barbieri et al., 2012). Overstimulation of the PI3K pathway is even more enhanced when PTEN mutations are detected alongside amplification of PI3K itself (PiK3CA amplification in 13 – 29% of primary tumours and 50% of CRPCs). Alterations in INPP4B and PHLPP PAPs were recently implicated in PI3K signalling regulation (Edwards et al., 2003; Gao et al., 2006; Sun et al., 2009 ; Agell et al., 2011). Moreover, loss of PTEN has also been correlated with TP53 loss and overexpression of c-Myc or ERG (King et al., 2009; Wang and Shen, 2011; Kim et al., 2012).
Inactivation of cell-cycle-inhibitory genes appears to disrupt the senescence associated with cancerogenous signalling and possibly the bypass of AR-regulated growth in metastatic and CRPC tumours (Baca and Garraway, 2012). Accordingly, mutations are also commonly found in key tumour-suppressor genes, such a TP53 and RB1, causing the inactivation of p53 protein, which is responsible for positive regulation of p21 cyclin-dependent kinase cell-cycle inhibitor (Holcomb et al., 2009). Mutations within the tumour-suppressor gene TP53 encourage the activation of cell proliferation and suppression of DNA repair (Dong, 2006; Kumar et al., 2011; Barbieri et al., 2012; Grasso et al., 2012). These mutations occur more frequently in prostate cancer that has undergone pharmacological treatment and radiation, resulting in reduced time to development of distant metastasis (Grignon et al., 1997; Dong, 2006). In fact, prostate cancers metastasizing to bone have been found to have the most frequent TP53 mutations (Meyers et al., 1998). It is also likely that mutations in TP53 impair genomic stability, leading to genomic amplification of the AR gene during hormone therapy (Dong, 2006). Exome sequencing of prostate cancer has also revealed frequent aberrations in the tumour-suppressor gene Rb1, reaching up to 20% in both primary and CRPC samples (Kubota et al., 1995; Barbieri et al., 2012; Grasso et al., 2012). Another mutation associated with prostate cancer implicates another key cell-cycle regulator, CDKN1B, which encodes p27, a cyclin-dependent kinase mutated in up to 3% of primary prostate cancer cases, which correlates with poor pathological prognostic outcome (Vis et al., 2000; Dreher et al., 2004; Barbieri et al., 2012). Disruption in CDKN1B promotes prostate cancer growth in coordination with the inactivation of PTEN, which implies an interaction between p27 and the PI3K/Akt pathway (Di Cristofano et al., 2001). These mutations affect the regulation of cell-cycle arrest, contributing to faulty DNA-repair machinery, and therefore potentially being involved in prostate cancer progression.
The downregulation and aberration of Nkx3-1 have been associated with prostatic intraepithelial neoplasia (PIN) lesions and early stages of prostate oncogenesis. Nkx3-1 disruption has been identified in approximately 75% of prostate cancer loss of heterozygosity (Emmert-Buck et al., 1995; Asatiani et al., 2005). Furthermore, in a murine model, the overexpression of Nkx3 results in the suppression of tumour development, strongly implicating Nkx3 as a tumour suppressor gene. Nkx3 has not been found expressed in any other male urogenital system, allowing it to be used as an effective biomarker for carcinogenesis prediction (Bhatia-Gaur et al., 1999). BRCA2 gene mutation is also represented as a hallmark in prostate cancer development. Originally identified as a hallmark of poor prognosis for breast and ovarian cancer, BRCA2 normally mediates DNA repair. Patients with BRCA2 gene mutations show an increasing frequency pattern, with prostate cancer progression into castration-resistant phenotype (Thorne et al., 2011). Finally, it is hypothesized that a high prostate cancer proliferative potential is partly adopted by an embryonic Wnt signalling pathway. Generally silent in differentiated cells, Wnt ligands have been found to be upregulated in prostate cancer, with a marked increase of expression in advanced forms (Kypta and Waxman, 2012). For example, elevated levels of Wnt1, Wnt5a, Wnt7b and Wnt11 expression have been correlated with prostate cancer aggressiveness (Chen et al., 2004; Li et al., 2008; Uysal-Onganer et al., 2010). In addition, DKK1 expression increases during prostate cancer initiation but decreases during metastasis (Hall et al., 2008). The correlation of Wnt activation and skeletal metastasis may be important for therapy.
It is important to note that heterogeneity of a tumour is not entirely attributable to mutations. Epigenetic changes, interactions with the microenvironment and spatial and temporal differences also alter phenotypic expression, posing a challenge when it comes to determining the origin of the cell (Visvader and Lindeman, 2012).
The cancer stem cell (CSC) hypothesis of tumour development was first postulated in the principal study of Park et al. (1971), which showed that a small proportion of cells (0.01 – 1.00%) in tumour isolates are clonogenic and extensively proliferative in vitro and in vivo, indicating that these cells might represent tumour stem cells. The CSC hypothesis postulates the existence of a small subset of cancerous cells, originated from stem cells that accumulated mutations and maintained the inherent ability of self-renewal, which, given their pluripotent nature, are accountable for tumour initiation and maintenance of tumour heterogeneity (Shipitsin and Polyak, 2008; Visvader and Lindeman, 2012). Like normal stem cells, CSCs are able to create a hierarchical organization and to reconstitute the bulk of the tumour (Lawson et al., 2005; Magee et al., 2012; Yu et al., 2012). The CSC hypothesis may also explain the frequent ineffectiveness of standard cancer treatments: this small population of cells are thought to be therapy-resistant, often leading to recurrence and giving rise to metastasis (Wang and Shen, 2011). These observations suggest that thorough CSC characterization could lead to a better understanding of the mechanisms that mediate tumour initiation, progression and metastasis. It is important to distinguish CSCs from tumour-initiating cells (TICs). CSC populations, due to their intrinsic biological stem cell properties, are able to reconstitute a tumour in a recipient animal. This tumour can be identical to the parental one from which it was derived, and can be serially xenotransplanted indefinitely. The definition of TICs is wider and refers to the functional concept of a CSC but does not imply hierarchical organization (Wang and Shen, 2011; Yu et al., 2012).
The origin of CSCs has yet to be securely defined, due to the multiple ways in which this biological phenomenon occurs (Yu et al., 2012). Stem cells differentiate and generate a population of TACs or progenitor cells as a natural order of organogenesis. Maturation-arrests theory states that sporadic mutations may occur within the stem cell/progenitor cells before their full differentiation, activating self-renewal signalling and creating cancerous stem cells (Figure 10.2) (Cozzio et al., 2003; Huntly and Gilliland, 2005; Zhao et al., 2007). The stage of differentiation of the tumour is dependent upon the degree of differentiation at the time of maturation arrest (Sell, 2010). The reverse to the maturation arrest theory, the anaplastic theory, claims that CSCs arise due to mutations of fully differentiated mature cells (Sell, 2010). Once a cell has committed and completed its ‘cell fate’, a mutation may occur, resulting in the alteration of regulatory cell functions. As a consequence, the cell reprogrammes and reacquires its stem-like properties (Yu et al., 2012). Finally, the 19th-century embryonal-rest hypothesis states that residual embryonic or germinal stem cells give rise to CSCs after being left during tissue ontogenesis; this hypothesis become discredited during the 20th century (Sell, 2010).
Figure 10.2. Source of heterogeneity in a tumour, derived from stem cell/TAC mutations, creating a hierarchy. The stages of differentiation of a CSC give rise to the genetic variation of the cell of origin, resulting in tumour variance.
Given their exceptional potential, the identification of CSCs from different types of cancer has become a major challenge for cancer researchers in the last 2 decades. Current efforts to define CSC populations are based on their functional assessment through the use of xenotransplantation assays of cell populations that have been isolated from primary human cancer tissue. Thus, CSCs are identified as separate populations within the total prostate cancer, based on their ability to initiate tumour formation in animal models following transplantation (Wang and Shen, 2011). Their self-renewal ability is determined by analysis of the tumour formation following serial transplantation. In vitro assessments of CSC properties involve the analysis of their ability to form spheres, eject Hoechst dye 33342 and differentiate into nonrenewable heterogeneous cancers (Wicha et al., 2006).
Commonly, CSCs are identified by screening for cancer cell populations expressing stem cell-specific biomarkers or genes and/or by examining the self-renewal and drug resistance properties of a putative population in various in vitro functional assays and in vivo tumorigenicity models (Wang and Shen, 2011). In a majority of studies, stemness or the ability to self-renew has become attributed to the expression of particular protein markers on the cancer cell surface (Yu et al., 2012). Starting from leukaemic (CD34+ /CD38), and breast (CD44+ /CD24) CSCs, CSC populations have been identified in several other cancers, including brain, lung, colon, melanoma, pancreatic, ovarian and prostate (Al-Hajj et al., 2003; Galli et al., 2004; Collins et al., 2005; Kim et al., 2005; Ponti et al., 2005; Ricci-Vitiani et al., 2007; Schatton et al., 2008).
Prostate CSC populations were first identified by fluorescence-activated cell sorting (FACS) from primary human tumours and showed high proliferation potential in clonogenic assays, as well as the ability to give rise to luminal cells. Prostate-cancer CD44+ populations have been shown to have significantly higher proliferation rates, self-renewal abilities and tumorigenic properties than their CD44counterparts (Patrawala et al., 2006). Notably, CD44+ cells show an increased expression of several ‘stemlike’ genes, including Oct-3/4, Bmi, β-catenin and SMO, as well as the ability to differentiate into CD44and AR+ cells in vivo and in vitro (Patrawala et al., 2006). Hurt et al. (2008) have also identified a particular LNCaP CD44+ /CD24population that is able to form prostatospheres in vitro and to reconstitute tumours in nonobese diabetic severe combined-immunodeficient (NOD/SCID) mice from as few as 100 cells. Other populations with a marker signature of CD44+ /CD24/α2β1high have been identified in sphere-forming assays from the DU145 prostate cancer cell line (Rybak et al., 2011). CD133+ cell populations are also capable of self-renewal and increased proliferation (Collins et al., 2005). Gene expression analysis of CD44+ /CD133+ /α2β1high originated from DU145 prostate cancer cell lines reveal high levels of expression of c-myc and β-catenin and a lower expression of Bax (Wei et al., 2007). However, CD133 has also been found to be abundant in nonmalignant tissues, which limits its use as a target of cancer therapies (Visvader and Lindeman, 2012). Rajasekhar et al. (2011) characterized prostate CSCs by investigating the in vitro and in vivo expression of TRA-1-60, CD151, CD166 and NF-kB activity using prostate cancer cells derived from human prostate cancer tissue. The study showed the ability of these triple-positive cells to form spheroids and to recapitulate the original parent tumour heterogeneity in serial xenotransplantations, indicating a tumour cell hierarchy in human prostate cancer development. Also, this sphere-forming CSC population did not express the main prostate markers, such as PSA, AR and Nkx3.1, but showed a marked increase in the expression of stem cell-associated markers like Met-receptor kinase, inhibitor of differentiation 1 (Id1), Musashi-1, phospho-histone 3 and Ki67 (Rajasekhar et al., 2011).
Accumulating data reveal significant resemblances in gene expression between CSCs and embryonic stem cells (ESCs). These similarities have been found through gene expression analysis of CSCs from hepatic, colorectal, nasopharyngeal, brain, breast and ovarian cancers (Zbinden et al., 2010; He et al., 2012; Ling and Jolicoeur, 2012; Zhang et al., 2012; Luo et al., 2013). Genes expressed in CSCs include the embryonic stem cell-specific transcriptional factor NANOG, the sex-determining region Y-box 2 (SOX2) and the octamer-binding transcription factor 4 (OCT4, also called POUF1) heterodimers (Wang et al., 2013). NANOG regulates cell proliferation and cell-cycle arrest via interaction with cyclins D1, D2, D3 and cyclin kinases 1 and 6, as well as through the inhibition of p53 signalling, thereby supporting cancer survival (Choi et al., 2012). Overexpression of NANOG is also involved in metastasis through the inhibition of E-cadherin, FOXO1, FOXJ1 and FOXB1, together with Oct4 induction of the transcription factor Slug, which is involved in epithelial – mesenchymal transition (EMT) (Chiou et al., 2010; He et al., 2012). In fact, knockdown of NANOG significantly reduces the clonogenic and tumorigenic properties of the DU145 prostate cancer cell line (Jeter et al., 2011). On the other hand, SOX2 promotes the apoptosis-resistant phenotype of the DU145 prostate cell line in vitro and in vivo using a NOD/SCID xenograft model (Jia et al., 2011). NANOG and OCT4 also enhance the expression of ABCG2, a member of the ATP binding cassette (ABC) subfamily membrane efflux channels that mediate the exclusion of cytotoxic agents (Chiou et al., 2010; Linn et al., 2010; Jeter et al., 2011). These observations support the existence of a rare, drug-resistant population with a significantly high survival potential, and may also explain the frequent ineffectiveness of conventional cancer treatment strategies. In addition, chemotherapy is an effective means of targeting rapidly dividing cell populations but remains inefficacious in targeting putative CSC populations, due to their quiescent and dormant nature (Klonisch et al., 2008).