The Prostate And Its Diseases Prostate Anatomy

Published Date: 02 Nov 2017

The prostate is an exocrine gland of the male reproductive system located inferior to the bladder, which surrounding the initial part of the urethra and some of the ejaculatory duct. The function of the prostate is to secrete an alkaline fluid into the semen which to help neutralise the acidity of the female vaginal tract, and prostate-specific antigen (PSA) which liquefies the semen (Kirby, 2003).

The prostate gland is a pyramid-shaped organ with its apex adjacent to the urethra and directed downward and its bases adjacent to the bladder and directed upward. It lies below the urinary bladder and is located in front of the rectum and the seminal vesicles are located at its base. The prostate weighs about 20 g by early adulthood. It is a composite organ made up of several glandular and non-glandular components with no obvious capsular morphology (McNeal, 1988). The prostate can be divided into 5 lobes: an anterior, two lateral, a median and a posterior lobe (Kirby, 2003). However, this idea is disputed and alternatively 3 distinct zones of the prostate can be identified: the central zone (CZ), peripheral zone (PZ), and transition zone (TZ) (McNeal, 1981, McNeal, 1988) (Error: Reference source not found). These zones differ both physiologically and biologically and a zonal description of prostate morphology is central to present understanding of prostatic diseases, particularly prostate cancer.

Figure . Cartoon of the anatomy the normal prostate gland, showing peripheral, central and transitional zones (Potter et al., 2005).

The peripheral zone comprises about 70% of the glandular prostate in young men and is predominantly mesodermal in origin (Argani et al., 1998). Its ducts exit from the posteriolateral recess of the urethral wall and extend mainly laterally in the coronal plane, branching both anteriorly and posteriorly (McNeal, 1988). The central zone is a cone shaped region that surrounds the ejaculatory ducts and makes up approximately 25% of the glandular prostate mass in young adults. The transition zone accounts for only 5% of the glandular prostate tissue and surrounds the proximal urethra. The non-glandular prostate tissue comprises of the preprostatic sphincter, striated sphincter, anterior fibromuscular stroma and the prostatic capsule (McNeal, 1988).

There are three diseases of the prostate: prostatitis (inflammation), benign prostatic hyperplasia (BPH) and prostate cancer (PCa), which affect the different zones of the prostate. BPH develops predominantly in the transition zone and PCa in the peripheral zone.

Prostate Histology

The human prostate gland comprises of two distinct epithelial cell types: basal and secretory luminal cells which along with rarer neuroendocrine cells to form a complex branching ductal structure embedded in a muscular stroma containing smooth muscle cells and fibroblasts (Kirby, 2003). Basal cells form a layer along the basement membrane and luminal cells sit above the basal cells and secrete prostatic proteins into the lumen (Figure ). The primary role of luminal epithelial cells is to secrete PSA, a single-chain glycoprotein consisting of 237 acids.

The two layers of epithelial cells are distinguished by their protein marker expression and secretions, as shown in Figure . Luminal cells express prostate specific antigen (PSA), prostatic acid phosphatase (PAP), androgen receptor (AR), CD57, 15-lipoxygenase (15-LOX2) and low molecular weight cytokeratins K8 and K18. Basal cells express high molecular weight cytokeratins K5 and K14, CD44, p63, telomerase and bcl-2 (Hudson, 2004, Bagley R.G. , 2009).

Figure . Structure of the human prostatic gland. Cartoon of the structure and marker expression of the prostatic ducts, which comprise a basal and a luminal epithelial layer with distinct protein expression (Honorio 2009).

Benign Prostatic Hyperplasia

BPH occurs mainly in the transition zone, and is caused by hyperplasia of the prostatic stromal and epithelial cells (McNeal, 2006). Large fibromuscular nodules form in the periurethral region of the prostate, which can compress the urethral canal and cause obstruction of the urethra and interfere with the flow of urine. Symptoms include urinary hesitancy, nocturia, dysuria (painful urination), increased risk of urinary tract infections, and urinary retention. Although age is the major risk factor for BPH, androgens influence the disease etiology as men castrated before puberty do not develop the disease (Wilson and Roehrborn, 1999).

Diagnosis is usually by digital rectal examination (DRE) and biopsy to rule out PCa. Treatments include 5α-reductase inhibitors that reduce conversion of testosterone to DHT and alpha-adrenergic blockers to relax smooth muscle in the prostate and the bladder neck, thus increasing urine flow (Timms and Hofkamp, 2011). Both 5α-reductase and alpha-adrenoceptor blockade are well tolerated and effective treatments (Tammela, 1997). A surgical alternative is transurethral resection of the prostate (TURP), thus removing the urethral blockage. However, TURP can result in considerable side effects (Tammela, 1997). Other alternatives are transurethral needle ablation (TUNA) and transurethral microwave thermotherapy and although these have lower associated risks, are not as effective as TURP (Schatzl et al., 2000).

Prostate Cancer

Prostate cancer is an adenocarcinoma most frequently found in the peripheral zone of the prostate. Initially, small clumps of cancer cells remain confined to otherwise normal prostate glands, a condition known as prostatic intraepithelial neoplasia (PIN), which can develop into invasive cancers. Prostate cancer most commonly metastasizes to the bone via the lymph nodes, and also can invade rectum, bladder and lower ureters after local progression.

PCa diagnosis is confirmed by a combination of PSA screening, DRE, Magnetic Resonance Imaging (MRI) and prostate biopsy. PSA is a protein produced by normal epithelial cells as well as PCa, and can be elevated in prostatic disorders. Biopsy samples of PCa are classified according to their Gleason score based the morphology of the cancer (Epstein et al., 2005). Cytological features may include hyperchromatic, enlarged nuclei with prominent nucleoli and abundant cytoplasm (Kirby and Madhavan, 2010).

Clinically, localized disease is treated by surgery (prostatectomy) or radiation therapy (external beam or brachytherapy), which have high success rates. An alternative is active surveillance which involves monitoring PSA levels (Bannuru et al., 2011). The use of surgery to treat localized disease can be controversial and some studies have shown no survival advantages when compared to active surveillance in men with low grade disease (Wilt et al., 2012).

Treatment of advanced disease is less successful. First line treatment depends on the androgen sensitivity of PCa, using androgen ablation or blockade of androgen action through the androgen receptor (Miyamoto et al., 2004). Although this treatment initially leads to tumour regression in the majority of patients, the cancer recurs in a median period of 18-24 months, leading to castrate resistance (Miyamoto et al., 2004). Recently tow new treatments, Abiraterone and Enzalutamide have been developed to target castrate resistant disease. Abiraterone is an inhibitor of CYP17 activity which significantly decreases testosterone levels (Barrie et al., 1994). In Phase III trials, it extended median survival to 14.8 months versus 10.9 months. Phase III trials of the androgen receptor antagonist Enzalutamide increased survival to 18.4 months compared to 13.6 months in the control group (Scher et al., 2012). Both drugs have now received FDA approval and offer promising new therapy regimes. Development of alternative treatments is required to treat metastatic disease and improve further the long term prognosis of men with advanced disease (Wolff and Mason, 2012).

Adult Stem Cells

Organs are composed of differentiated cells that perform discrete functions (Miller et al., 2005) and comprise the bulk the cells (Ichim and Wells, 2006). The continuous replacement of differentiated, functional cells by more primitive cells is a normal homeostatic process driven by multi-potential stem cells (SCs). These cells are also known as somatic or adult stem cells and have been identified in most tissues. They replenish dying cells and maintain organ health and functionality. Adult stem cells have features in common, including a large nuclear-to-cytoplasmic ratio, few organelles and they are structurally unspecialised or undifferentiated (Miller et al., 2005).

The Adult Stem Cell Hierarchy

An adult stem cell has two properties which allow it to maintain organ function: Potency and self-renewal. The potential the cell has to differentiate to form all the functional cell types within a tissue is called potency. The SC is at the apex of the hierarchy and is the initiating cell in the cell division and differentiation process producing a large family of differentiated descendants known as clonal expansion. To replenish the stem cell compartment lost during differentiation, the SC can undergo symmetrical division to produce two identical stem cell daughters, a process termed self-renewal (Mackillop et al., 1983).

In addition to self-renewing stem cells, the hierarchy model predicts two other types of cell: proliferating, non-self-renewing cells (transit amplifying) (TA) and non-proliferating, differentiated end cells (DC). The hierarchy in Figure illustrates how the potential for division and differentiation changes as a cell moves down the hierarchy. Following division, the stem cell can give rise to a transit amplifying cell that will undergo further proliferation. The cells progressively differentiate and irreversibly commit to one of the lineages specific to the tissue (Miller et al., 2005). As cells move down the hierarchy acquiring the differentiated features associated with tissue function, the proportion of differentiated cells increases.

Somatic SCs reside in confined tissue compartments, referred to as niches (Loeffler and Roeder, 2002), where the microenvironment suppresses SC proliferation, resulting in a quiescent SC population. The SCs may be triggered to proliferate and differentiate in response to injury to repair damaged tissue (Ghotra et al., 2009). In this way the stem cell has the ability to maintain the organ over its lifetime (Miller et al., 2005).

Figure . Adult Stem Cell Hierarchy. The stem cell can self-renew or divide to produce proliferative transitional cells expand clonally. As cells differentiate they lose their proliferative potential and become more abundant.

Stem Cells in the Haematopoietic System

The first evidence for the existence of SCs in adult animals was generated by Till and McCulloch, who showed that mouse bone marrow (BM) cells injected into irradiated recipient mice developed visible spleen colonies derived from grafted cells. (Till and McCulloch, 1961, McCulloch and Till, 1962) The number of donor cells was directly proportional to the number of colonies that developed within the spleen. This result suggested that the transplanted BM cells were capable of self-renewal and it was hypothesised that these cells were stem cells. In addition, the radiation survival curve of cells that form colonies closely resembled the in vitro cell survival curves of clonogenic cells developed by Puck and Marcus (1956). The clonal origin of spleen colonies was confirmed by transplantation of sub lethally irradiated bone marrow into heavily irradiated recipient mice. Some donor bone marrow cells containing genetic abnormalities caused by the irradiation cells, retained the ability to proliferate and produce clones containing the abnormality, demonstrating their clonal origin (Becker et al., 1963).

If the capacity to form colonies is to be considered as a criterion to identify stem cells, the cell must lose this capacity upon undergoing differentiation. The differentiation and subsequent loss of colony forming capacity was confirmed by applying hypoxia as a differentiating pressure which resulted in a reduction in colony formation in the spleens of hypoxic mice (Bruce and McCulloch, 1964). This result was thought to be due erythropoiesis which stimulates erythropoietin stimulated by hypoxia. The data suggested that an increased demand for differentiated cells reduced the number of stem cells, resulting in a reduction of colony forming ability. This confirmed the property of these stem cells as able to undergo both self-renewal and differentiation.

This data inspired further work studying stem cells, particularly within the haematopoietic system, leading to the discovery that BM contains at least 3 different stem cell populations: haematopoetic stem cells (HSCs), mesenchymal stem cells (MSCs) and endothelial progenitor cells (EPCs) (Alison and Islam, 2009). Single unselected BM cells can also form colonies in non-lympho-haematopoietic tissue, such as hepatocytes (Petersen et al., 1999), muscle fibre (Ferrari et al., 1998), microglia and astroglia (Eglitis and Mezey, 1997) and neuronal tissue (Brazelton et al., 2000, Mezey et al., 2000). The morphology and marker expression of these colonies are similar to the native cells.

Adult Stem Cell Identification

Morphology, proliferative/ cycling rates, location and marker expression are all used to identify and characterise stem cells in adult tissues. Advances in the identification and characterisation of adult stem cells have been aided by the development of techniques to label cells based on surface marker expression. Stem cells are identified by staining cell surface markers with monoclonal antibodies (mAb), exclusion of fluorescent dyes such as Hoechst 33342 or long term labelling with tritiated thymidine (Mittal et al., 2009). mAb technology and flow-cytometery based sorting (FACS) and analysis have been the main driving force in recent SC developments to enrich SC populations.

The identification of SCs is confirmed by the ability of the selected cells to recapitulate the organ of origin. The haematopoietic system and tissues with high cell turnover rates, such as the cells of the digestive tracts and skin, have been extensively studied, although SC in almost all organs have been identified and enriched for based on cell surface marker expression (Alison and Islam 2009). Table lists markers for the identification and selection of adult stem cells in both blood and solid tissues. Of note is CD133 (Prominin 1), a marker which is a potential stem cell marker in several tissues, although its function in this context remains unclear.

The putative SC markers are also compared by clonogenicity, growth in non-adherent culture, the ability to reconstitute the organ when transplanted orthotopically in vivo and lineage tracing studies. Potential molecular markers for stem cells have also been identified in embryonic stem cells. These markers include the transcription factors Oct4, Sox2 and Nanog which are known to control self-renewal and differentiation. These markers can identified in situ by immunocytochemistry and are used in combination to identify the location of the stem cell niche. Reverse transcription polymerase chain reaction (RT-PCR) analysis of cells retrieved from the mouse stomach stem cell niche by laser capture micro-dissection revealed a gene expression profile closely matching that of HSCs (Mills et al., 2002).

Table . Prospective adult stem cell markers

Tissue/ stem cells

Markers Used to Enrich for SCs

Brain

CD133 (Lee et al., 2005)

CD184+ CD271− CD44− CD24+ (Yuan et al., 2011)

Bone marrow (HSC)

CD34+ CD133+ CD45+ and c-kit+ (Bhatia, 2001)

Bone marrow/ (MSC)

CD105+/CD73+/CD90+/CD34-/ CD45-, (Dominici et al., 2006)

Breast

α6integrin + CK19+ ESA+ MUC1− CALLA− (Clarke et al., 2005)

Cardiac

Lin− c-kit+ (Beltrami et al., 2003)

Intestinal

Lgr5+ (Barker et al., 2007)

Kidney

CD133+ CD73+ CD29+ and CD44+ (Bussolati et al., 2005)

Liver

CD29+ CD73+ CD44+ and CD90+ (Herrera et al., 2006)

Skin

CK19+ (Michel et al., 1996)

CD34+ and CK15+ (Blanpain et al., 2004)

Lgr6+ (Snippert et al., 2010)

Potential cell surface markers for the identification of adult stem cells in the respective tissues.

Prostate Stem Cells

The earliest evidence for the existence of stem cells in the adult prostate came from androgen-ablation studies, where it was observed that adult rodent prostate can undergo multiple rounds of castration-induced regression and testosterone-induced regeneration (Isaacs and Coffey, 1989). Following castration there is a rapid reduction in prostate volume due to apoptosis. The remaining epithelial cell population can survive long periods and regenerate the prostate following androgen replacement (English et al., 1987, Evans and Chandler, 1987). This suggests that a small population of SCs possess the ability to self-renew and differentiate to regenerate the prostate, whilst the bulk of the prostate cells which are androgen dependent lack this function.

The castration studies suggest that SCs reside within the basal epithelial compartment of the prostate, as the majority of cells that survive castration have a basal rather than a luminal phenotype. (Kirby, 2003). Expression of proliferation and survival markers such as telomerase, p63 and Bcl-2 are localised in the basal compartment, which further supports this theory (Kasper, 2008). During regeneration SCs differentiate to produce androgen-independent TA cells that give rise to androgen-dependent fully differentiated secretory luminal cells providing evidence for a cellular hierarchy within the adult human prostate (English et al., 1987).

Prostate Stem Cell Identification

Several candidate prostate SC populations have been isolated based on marker expression observed only in the basal layer of the epithelium. Potential markers include K5/K18 double positive cells (Hudson et al., 2000) and CD44+ α2β1hi CD133+ cells (Richardson et al., 2004). α2β1 integrin selects for cells which have a higher colony forming efficiency, are positive for the basal markers K5 and K14 and form prostate-like glands in vivo (Collins et al., 2001). Other potential markers for enrichment of prostate SCs include CD44, CD49f, CD117 (c-kit) and CD133, alone and in combination. Cells expressing these markers demonstrate increased colony forming efficiency (CFE) and can generate glandular structures when implanted under the renal capsule when recombined with rat urogenital mesenchyme (rUGM) (Leong et al., 2008). SCs from the prostate also appear to be enriched in side population (SP) cells. SP cells express the ATP-binding cassette membrane transporter ABCG2 transporter, which actively effluxes Hoechst 33342 from the cell which allows enrichment by FACS (Brown et al., 2007).

Cancer Stem Cells

Models of Tumour Heterogeneity

Like normal tissue, cancers are composed of a heterogeneous mixture of cells with range of capacity to differentiate, proliferate and form tumours (Pierce and Speers, 1988). Studies in vivo have demonstrated that, within a cancer population, only a small percentage of cells are able to initiate tumour development (Bonnet and Dick, 1997, Al-Hajj et al., 2003, Singh et al., 2004). Two models have been proposed to explain phenotypic and functional tumour heterogeneity: the stochastic and hierarchical, which are described in Figure .

Stochastic Model of Stem Cells

The stochastic model predicts that all tumour cells have similar proliferative capacity, but their behaviour is influenced by extrinsic factors (e.g. host factors, immune response, and microenvironment) or intrinsic (e.g. signalling pathways, levels of transcription factors). The randomness and unpredictability of these variables results in heterogeneity in marker expression, proliferation and tumour initiation capacity (Dick, 2008).

For the stochastic model to be correct, tumour cells cannot be permanently affected by these factors and all cells must have equal capacity to act as stem cells (Wang and Dick, 2005). The stochastic models predicts a growth fraction of less than 100% due to cell loss and the result of the constraints of the micro-environment (Miller et al., 2005). In this model tumour initiating activity cannot be enriched or selected for, because no markers are available.

CSC Hierarchy Model of Stem Cells

The second model is the cancer stem cells (CSC) hierarchy model which predicts that the tumour is a ‘caricature of normal tissue development (Pierce and Speers, 1988). Like normal tissue, tumours are hierarchically organised, with the CSCs at the apex, driving tumour growth and regeneration. Like normal SCs, CSCs maintain the hierarchy by self-renewal or they generate transit amplifying cells which divide to produce differentiated offspring which form the bulk of the tumour and lack stem cell properties. The CSCs are thought to be a relatively small population of cells essential for tumour initiation (Frank et al., Reya et al., 2001). As in normal tissue only a small percentage of the tumour population maintain the capacity for long term proliferation, while most cells proceed down the differentiation pathway resulting in terminal differentiation (Miller et al., 2005). Due to differences in their characteristics, including proliferative capacity and marker expression, CSCs can be selected for.

Current cancer treatments may eradicate the tumour bulk but spare the populations of stem cells which are able to regenerate the cancer (Wang et al., 2012). This process may explain why tumour regression does not translate to improved patient survival in many clinical trials. It is still not clear whether CSCs are originally somatic SCs which have undergone oncogenic changes, or TA or DCs which have gained genetic changes which result in SC behaviour. The evidence for the hierarchical model which underlies the CSC theory comes from clonogenic and tumourigenic assays, which will be discussed further.

Clonal Evolution

A parallel and popular idea is the clonal evolution model of cancer in which cancer is a stepwise evolutionary process of Darwinian natural selection (Greaves and Maley, 2012). Intrinsic differences can be caused by stochastic genetic (Nowell, 1986) or epigenetic changes (Baylin and Jones, 2011). This model complements both the CSC and the stochastic model (Shackleton et al., 2009). It proposes that most neoplasms arise from a single cell of origin, and tumour progression results from acquired stepwise genetic changes within the original clone allowing sequential selection of more aggressive sub-lines (Nowell, 1976).

Figure Stochastic and Hierarchical Models of Cancer Stem Cells. The hierarchical model suggests that tumours are composed of functionally distinct cells, including cancer stem cells, which have different functional properties. The stochastic model predicts that all cells functionally are equal and that cell heterogeneity is due to intrinsic and extrinsic influences. Adapted from Dick, 2008.

Cancer Stem Cell Definition

The CSC is defined as "A cell within a tumour that possesses the capacity to self-renew and to cause the heterogeneous lineages of cancer cells that comprise the tumour. CSCs can thus only be defined experimentally by their ability to recapitulate the generation of a continuously growing tumour" (Clarke et al., 2006). Therefore CSCs express the stem cell properties of self-renewal and potency. Self-renewal is the ability to undergo cell division while maintaining stem cell capacity. Potency is the ability of cells to differentiate into the cells that comprise the tumour of origin (Mittal et al., 2009). Experimentally, the ‘gold standard’ for CSC identification is the ability to serially xenograft (Clarke et al., 2006).

Identification of Cancer Stem Cells

The first evidence for the existence of CSCs came from cell proliferation studies. Radiolabelling of cells and the use of autoradiography enabled measurements of proliferation, hierarchical relationships and lifespan in normal and neoplastic tissues (Belanger and Leblond, 1946, Dick, 2008). From these studies came the proposal that tumours are caricatures of normal development and contain stem cells (Pierce and Speers, 1988).

Haematopoietic CSCs

Like somatic stem cells, much of the early work on CSCS relied on cancers of the haematopoietic system. In the 1970s, cytokinetic studies of cell lines, murine models of acute leukaemias and in vivo examination of leukaemia blast proliferation kinetics in human AML (Acute Myeloid Leukaemia) and ALL (Acute Lymphoblastic Leukaemia) patients demonstrated functional heterogeneity (Clarkson et al., 1970, Dick, 2008). The majority of leukaemic blasts were post mitotic the population was replenished from a relatively small fraction of proliferative cells. Only a small number of leukemic blast cells were cycling in vivo and two proliferative fractions were observed: a larger, fast cycling subset with a 24 hour cell cycle time and a smaller, slow cycling subset with a dormancy of weeks to months. It was proposed that the slow cycling fraction generates the fast cycling fraction. These kinetic properties were similar to normal haematopoietic stem cells (Cheshier et al., 1999) and this observation indicates that cancers also exhibit functional heterogeneity in terms of proliferative potential.

The inability of conventional therapies to kill slow cycling leukemic stem cells (LSCs) is predicted to be the cause of relapse and failure of chemotherapy (Cronkite, 1970). LSCs respond to the depletion of the leukemic cell mass following chemotherapy by entering the cell cycle to regenerate the cancer (Clarkson et al., 1975). It has been suggested that the way to eliminate dormant LSCs was to find the time frame in which they are cycling, but the eradication of dormant LSCs by chemotherapeutic agents has, so far, not been fully achieved (Dick, 2008). The inability to identify and assay potential LSCs was a major stumbling block to these studies, and characterising LSCs was impossible due to an inability to identify markers. Therefore attention was focused on the clonogenic assay which was adapted by several groups to study AML. By using the clonogenic assay, these groups managed to identify the phenotype of AML cultures in vitro with differing proliferative potential, providing further proof for hierarchy in AML (McCulloch et al., 1981, McCulloch, 1983, Griffin and Lowenberg, 1986).

Identification of CSCs: In Vivo Tumourigenicity

Advances in the types of immunocompromised animal models makes in vivo serial xeno-transplantation assay the gold standard of CSC identification (Clarke et al., 2006). A CSC enriched cell fraction must display significantly increased tumourigenic capacity to validate the cell surface markers upon which it was selected.

The serial xeno-transplantation model is shown diagrammatically in Figure . Cells from the original tumour are dissociated, usually by enzymatic or mechanical means. The CSCs population is enriched for based on molecular marker expression with mAbs. The cells are then injected either subcutaneously or orthotopically into immunocompromised mice and form a xenograft. The xenografted tumour is in turn harvested, digested and transplanted into the same site into further mice. In vivo limiting dilution assays must be performed with both the target and depleted cell fractions to confirm enhanced tumourigenicity in the positive fraction.

The CSC containing fraction must re-establish the phenotypic characteristics of the original tumour. The non CSC fraction is also injected as a control which, if selection is successful, fails to form a tumour (Visvader and Lindeman, 2008). Clonogenicity in vitro can also be used to estimate CSC frequency and results often correlate with tumourigenicity (O'Brien et al., 2010).

Figure . In vivo serial xeno-transplantation. Human CSCs from human tumours are enriched for by marker selection and injected subcutaneously or orthotopically into immunodeficient mice. The resulting tumour is removed, digested and either reselected or transferred directly into secondary mice. Adapted from (O'Brien et al., 2010).

The xenotransplantation assay was first used to show that human leukaemias contain a small population (≤1%) of cells, with a CD34+, CD38- phenotype, which give rise to differentiated leukaemia cells and recapitulate the heterogeneous phenotype of the bulk tumour cancer in NOD/SCID mice (Bonnet and Dick, 1997). This finding was a crucial first step in demonstrating that subsets of cancer cells with enhanced tumourigenicity can be isolated based on molecular markers. The phenotypically more mature cells failed to engraft in mice, suggesting the presence of an identifiable tumour cell hierarchy. This experiment also demonstrated the hierarchical organisation of human AML.

CSCs in Solid Tissues

Fractions of serially tumourigenic cells have been identified in solid tumours, such as breast (CD44+ CD24-/low) and brain (CD133+) (Singh et al., 2004). In these experiments small numbers of selected cells produced tumours in recipient mice. CD44 and CD133, which are markers frequently used to isolate somatic SC populations, are often used to isolate CSCs. A list of potential CSC markers shown to enrich for tumourigenicity in mice are shown in Table . Putative cancer stem cell markers. . Other CSC markers have been suggested, but although they increased CFE, they have so far failed to enhance tumourigenicity of primary human tumour samples in mice. Most of the currently used markers recognise molecules on the cell surface and not functional stem cell activity. Interestingly, there is a noticeable similarity between markers used to enrich normal adult SCs and CSCs, suggesting that they share some phenotypic traits. However, in some cases it has proven difficult to confirm markers that originally appeared to distinguish tumorigenic from non-tumourigenic cells. Following the initial discovery of some markers, conflicting results have been published and the suitability of several markers refuted (Magee et al., 2012).

Table . Putative cancer stem cell markers.

Neoplasia Location

Marker Used to Enrich for CSCs

Haematopoietic (AML)

CD34+, CD38- (Lapidot et al., 1994, Bonnet and Dick, 1997)

Brain

CD133+ (Singh et al., 2004)

Breast

CD44+ CD24-/low (Al-Hajj et al., 2003)

Colon

CD133+ (O'Brien et al., 2007, Ricci-Vitiani et al., 2007)

Pancreas

CD133+ (Hermann et al., 2007)

ESA+CD44+ CD24+ (Li et al., 2007)

Melanoma

ABC5+ (Schatton et al., 2008)

CD20+ CD166+ Nestin+ (Klein et al., 2007)

Phenotypic populations which enrich for tumour initiating cells when serially transplanted immunocompromised mice.

Studying CSCs in solid tumours has proved more challenging than those in the haematopoietic system and the frequency of CSCs in solid tumours identified by current methods is highly variable (Visvader and Lindeman 2008). Evidence for the existence of cancer stem cells in solid tumours has been more difficult to obtain than in the haematopoietic system for several reasons:

1) The cells within the tumour are less accessible and tissue has to undergo mechanical or enzymatic digestion to obtain a single cell suspension which can be analysed;

2) There is a lack of functional assays suitable for detecting and quantifying non-cancerous stem cells from many organs;

3) Only a few cell surface stem cell markers have been identified and characterised. The identified markers are often used in combination as for most tumour types and no single marker can select for a 100% pure stem cell population.

Improvements to the NOD/SCID murine model by engineering mice to be deficient in natural killer and macrophage activity has increased the reliability of these assays. These improvements have also raised questions about the frequency of CSCs observed in tumours. The number of CSCs estimated in cancer tissue may be affected by the xeno-transplantation model used and the extent to which the host immune system is suppressed may account for some of the variation between studies (Brehm et al., 2010). A study into the frequency of cancer stem cells in melanoma has suggested that the number of stem cells within a tumour can be much higher than previously thought. Using NOD/SCID interleukin-2 receptor gamma chain null (Il2rg2/2) mice, 25% of unsorted primary and metastatic melanoma cells were tumourigenic. Enrichment using markers had no effect on the tumourigenicity of the cells (Quintana et al., 2008).

Prostate Cancer Stem Cells

Like other cancers, there is evidence for CSCs in the progression of prostate cancer, much of which comes from clonogenic assays. Small percentages of cells can establish serially passagable clones or spheres, which are enriched in putative stem cell markers (Tang et al., 2007, Li et al., 2008). However, the majority of these studies involve cancer cell lines rather than primary tumour samples. A combination of markers from primary human cancers which can reconstitute the tumour in immunocompromised mice has not yet been identified.

Cellular Origin of Prostate Cancer

The cell of origin of prostate cancer is still unknown, although recent studies suggest that it may originate in the basal epithelial compartment (Goldstein et al., 2010). The majority of cells in prostate cancer have a luminal phenotype, expressing K8, K18, AR and PSA, which has to suggestions that prostate cancer may arise by the malignant transformation of a luminal cell (Okada et al., 1992). Emerging evidence suggests, however, that this theory may be incorrect and that the ‘cell of origin’ lies within the basal epithelial cell layer of the prostate which is more susceptible to malignant transformation (Goldstein et al., 2010). This theory is supported by the fact that most prostate cancers contain a minor fraction of basal-like cells which express markers such as CD44, p63, ABGG2 and CD133. Surprisingly, one study has shown that the CD133 negative population (the non-stem cell population) is more susceptible to malignant transformation in BPH-1 cells (SV-40 transformed normal prostate epithelial cells) to become tumourigenic (Taylor et al., 2012).

Prostate Cancer Stem Cell Identification

Like other cancers, putative populations of prostate CSCs have been sorted and enriched for based on their cell surface marker expression. In both primary human prostate cancers and cell lines, CD44+/α2β1hi/CD133+ populations isolated by FACS demonstrate increased CFE and SFE (Collins et al., 2005, Gu et al., 2007, Patrawala et al., 2007). Increased serial tumourigenicity is also observed in CD44+ PC-3 cells and CD44+CD24- LNCaP prostate cells, where as few as 100 cells form tumours in mice (Patrawala et al., 2006). These CD44+ CSCs express embryonic stem cell genes such as Oct4, Bmi-1, β-catenin and Smo which are more invasive suggesting a role in metastases (Klarmann et al., 2009). TRA-1-60, CD151 and CD166 triple positive cells from the prostate xenograft model CWR22 have enhanced tumourigenicity and also demonstrate enhanced nuclear factor-κB activity (Rajasekhar et al., 2011).

Clonogenicity

The Clonogenic Assay

In 1956 Puck and Marcus published a paper describing a cell culture technique for assessment of colony forming ability of single mammalian cells (Puck, Marcus et al. 1956). Plated in culture dishes with a suitable medium, human cervical carcinoma cells (HeLa) were supplemented with a large number of irradiated feeder cells and the number of colonies formed was counted. This technique is a simple rapid method for growing single mammalian cells into macroscopic colonies with a colony forming efficiency of 80-100%. The assay was developed further to enable quantification of the effects of x-rays on cell populations in vitro, to produce the first in vitro radiation cell survival curves (Puck and Marcus 1955; Cieciura, Marcus et al. 1956; Puck, Marcus et al. 1956).

The colony forming assay demonstrates heterogeneity in vitro. A clone is defined as a group of cells derived from a single ancestor cell and clonogenicity is the ability of a given cell population, when plated as single cells, to produce one or more clones. The clonogenic potential of a cell population can be measured by clonogenic assay, which quantifies the proportion of colony forming cells, as a percentage of the plated cell number, referred to as colony forming efficiency (CFE). It is believed that colony-forming cells are able to both self-renew and differentiate (Bruce and McCulloch, 1964). Therefore, the ability to measure the capacity of cells to form clones is a useful tool and much of the evidence for the hierarchy model of tumour heterogeneity derives from clonogenic assays.

Several adaptations to the original method have been made. Immobilising cells in a top layer of 0.3% agar avoids formation of tumour cell aggregates by random movement, which can be confused with colony growth (Bizzari and Mackillop 1985). Agar can be replaced by agarose, which is easier to handle or methylcellulose which allows better recovery of the colony for replating (Bizzari and Mackillop 1985) Other groups have simplified the culture medium and omitted feeder cells, although this is dependent on cell type (Franken et al., 2006).

The clonogenic assay has been used for a wide variety of studies, with many cell types, using a range of culture conditions, and for the testing of many potential chemotherapeutic agents. It has played a crucial role in the identification and characterisation of CSCs. Secondary cloning has allowed study of self-renewal and longer term proliferation of CSCs and has the advantage of being able to identify cells that undergo a large number of cell divisions, a fundamental property of SCs (Bizzari and Mackillop 1985). This technique involves selecting specific colonies to determine their proliferative potential over a number of passages.

The role of CSCs in multiple myeloma has been studied using an anchorage independent growth clonogenic assay (Hamburger and Salmon, 1977), Anchorage independence growth is thought to be a characteristic of stem cells (Mori et al., 2009). Bone marrow samples from patients with multiple myeloma and normal volunteers cultured in the presence of an agar feeder layer prepared by either human type O+ washed erythrocytes or adherent spleen cells of BALB/c mice, have a linear relationship between colony formation and the number of nucleated bone marrow cells. Multiple myeloma patient samples have a higher CFE compared to normal volunteers and, crucially, the number of colonies is proportional to the number of cells seeded, suggesting a single cell origin (Hamburger and Salmon, 1977).

The study of stem cell capacity using clonogenic assays demonstrated the presence of a cellular hierarchy in many human cancers, lending support to the stem cell model of tumour growth (Mackillop et al., 1983). A few cells in each tumour are able to give rise to colonies in culture. Some colonies contain transit amplifying cells capable of undergoing a limited number of further divisions (Dick, 2008). Studies of serial CFE and colony size of human tumours has demonstrated the proliferative heterogeneity of a wide range of tumour types including neoplastic human urothelium (Mackillop et al., 1985), melanoma (Asano and Riglar, 1981, Meyskens et al., 1985) and squamous carcinoma (Grenman et al., 1989).

The Human Tumour Stem Cell Assay (HTSCA)

The success of Hamburger and Salmon in showing a relationship between multiple myeloma and colony forming efficiency led to the human tumour stem cell assay (HTSCA) as an in vitro method to test sensitivity of individual tumours to anticancer drugs (Friedman and Glaubiger, 1982, Panasci et al., 1985). Semi-solid agar enriched with medium supports colony growth from cell suspensions from a variety of malignant human tumours (Hamburger and Salmon, 1977). The aim of the HTSCA was to tailor chemotherapeutic regimes to the individual patient and test the effectiveness of new cytotoxic agents (Kirkels et al., 1983) including sensitivity of both leukaemias (Santini et al., 1989) and solid tumours (Von Hoff et al., 1983, Kuczek and Axelrod, 1987).

Although the development of the HTSCA looked promising the results were controversial and it was invalidated (Daniels et al., 1997). Part of the failure is attributed to the relatively small proportion of patient tumour samples that produce sufficient colonies for in vitro testing. Also, only a small proportion of tumours exhibited detectable in vitro sensitivity (Selby et al., 1983).

The response of clonogenic cells to drugs in vitro should correlate with the response of the tumour to the same drug in the patient (Dick, 2008). The stem cell model of human cancer suggests that cure or duration of remission after clinical treatment should correlate with killing of CSCs. Assessment of treatment effects on an unselected cell population (e.g. on the basis of morphological criteria) could be misleading since the effects on a small population of stem cells will be masked by those on the large population of stem cells (Selby et al., 1983).

Some studies directly compared the response in vitro with the subsequent clinical response and showed poor correlation. There have been a wide range of predictive value positives reported for the human clonogenic tumour cell assay when applied to a patient population with an expected clinical response rate of 15-49% (Hug et al., 1984). This value could be misleading and in practice may only be workable for cytotoxicity testing for only one third of specimens tested.

Other problems with the use and interpretation of human tumour clonogenic assays include technical issue such as difficulty in preparing single cell suspensions, production of only small quantities of data, and problems defining drug sensitivity and response criteria (Selby et al., 1983, Hug et al., 1984). These problems lead to the failure of the HTSCA to become a routine tool for analysis and treatment of cancers.

Non-adherent Colonies

Culture of single cells in non-adherent conditions is a clonogenic assay which is used to measure the number of self-renewing cells, by the formation of spheres. Prior to the early 1990s, it was believed that the brain was incapable of regeneration due to an absence of SCs. However, the propagation of neurospheres in a non-adherent system has demonstrated the brain does contain SCs capable of self-renewal and differentiation (Reynolds and Weiss, 1992). In a non-adherent culture system containing growth factors which select for primitive cells, more committed progenitors and mature cells die, positively selecting for proliferating neural stem cells (Galli et al., 2003).

In the sphere forming, assay either freshly digested or previously adherent cells are dissociated and cultured as single cells in non-adherent conditions, either on low attachment tissue culture plates or in a basement membrane matrix suspension (Pastrana et al., 2011). The resultant spheres are dissociated and re-plated under identical growth conditions. Differentiated cells within the original sphere die rapidly, while the neural SCs continue to proliferate exponentially to give rise to secondary spheres. This technique has led to the generation of stable neural SC lines (Galli et al., 2003). The selection of self-renewing cells in non-adherent culture is also possible in mammary cells, which have been shown to form floating structures known as mammospheres, (Dontu et al., 2003a, Dontu et al., 2003b, Dontu and Wicha, 2005) and prostate (prostapheres) (Garraway et al., 2010) which are similarly enriched in stem cells. Putative SC markers have been observed and identified in these systems, although formation of spheres alone is not enough to define a SC (Pastrana et al., 2011).

Barrandon and Green: Holoclones, Meroclones and Paraclones

In a seminal paper, Barrandon and Green (1987b) showed that freshly isolated clonogenic human epidermal cells form colonies with distinct morphologies, which are linked to their proliferative potential. Inoculation of single keratinocyte cells into dishes and transfer of the subsequent colonies into indicator dishes for further growth demonstrated that the founding cells were heterogeneous in their capacity for sustained growth. This provided the first evidence of a relationship between stem cells and colony morphology.

The founding single cells were classified as holoclones, meroclones and paraclones, based on the frequency of terminal colonies produced when the clone was transferred to indicator dishes, shown in Table . When 100% of colonies were terminal the cell was classified as paraclone; when more than 5% but less than 100% of the colonies were terminal, the clone was classified as meroclone; when 0-5% of colonies were terminal the clone was classified as a holoclone. A link between proliferative capacity and colony morphology was also observed with holoclones tending to form large colonies with a smooth outline and consisting of small cells (Figure ). Meroclones tended to form smaller colonies with a wrinkled outline and heterogeneity of cell size and morphology within the colony. Paraclones tended to form small colonies with irregular edges and terminally differentiated cells, which were generally incapable of further division.

Table . Barrandon and Green Colonies.

Typically formed by

Colony Morphology

Colony Area

Cell Size

Proliferative Capacity

Holoclone

Large nearly circular with smooth perimeter.

10-30mm2

Small

<5% cells terminal

Meroclone

Wrinkled colony that is in between holoclone and paraclones in size.

5-10mm2

Mixture

5-95% cells terminal

Paraclone

Small, highly irregular perimeter

<5mm2

Large and flattened

>95% terminal

Colony morphology and proliferative heterogeneity of keratinocyte clones in vitro. (Barrandon and Green, 1987b)

C:\Users\Charlotte\Documents\Charlotte 041111\Presentations\Barrandon and Green 2.jpg

Figure . Barrandon and Green Colonies. Morphologies of colonies derived from holoclone (left), meroclone (middle) and paraclone (right) keratinocyte cells (Barrandon and Green, 1987b).

The terms holoclone, meroclone and paraclone are now synonymous with stem cells, early and late transit amplifying cells respectively (Barrandon and Green, 1987b, Barrandon and Green, 1987a, Rochat et al., 1994). Further analysis in murine keratinocytes has shown that only holoclones can form secondary holoclones and be serially passaged long term (Tudor et al., 2004, Tudor et al., 2007). This has led to the holoclone forming assay being adopted as a surrogate assay to identify adult SCs, particularly in the skin (Mavilio et al., 2006, Murayama et al., 2007, Szabo et al., 2013) , follicular (Rochat et al., 1994) and limbal (Pellegrini et al., 2001, Shortt et al., 2007) tissues. Holoclones also demonstrate expression of survival genes such as p63 (Pellegrini et al., 2001), activation of β-catenin and Akt pathways (Murayama et al., 2007) and increased expression of self-renewal genes such as Bmi1(Claudinot et al., 2005).

Cancer Colony Morphology

Cell lines derived from cancer are a useful tool for studying CSCs. The three colony types including holoclone have been observed in cancer cell lines including head and neck, breast (Locke et al., 2005), prostate (Locke et al., 2005, Wei et al., 2007, Li et al., 2008, Pfeiffer and Schalken, 2009a) and pancreas (Li et al., 2007, Wang et al., 2013). These cell lines are heterogeneous in terms of CFE, secondary plating efficiency, tumourigenicity and marker expression. There is further evidence to indicate the presence of stem cells, including dye-exclusion (Hirschmann-Jax et al., 2004, Setoguchi et al., 2004, Locke et al., 2005, Patrawala et al., 2006), and sphere formation in non-adherent culture conditions (Reynolds, 1992). A fraction of cells within the cell lines also demonstrate increased serial tumourigenicity and chemoresistance (Reynolds and Putnam, 1992).

Prostate cancer cell lines form colonies with three different morphologies when cultured at clonal density (Locke et al., 2005, Li et al., 2008, Pfeiffer and Schalken, 2009a, Zhang and Waxman, 2010). Their morphologies are similar to the Barrandon and Green definitions of holoclone, meroclone and paraclone. This suggests the presence of a cellular hierarchy similar to normal epithelial cell populations containing stem cells, transit amplifying cells and differentiating cells (Locke et al., 2005).

Previous studies have shown that holoclones can be passaged long term (Pfeiffer and Schalken, 2009a) are serially transplantable in immune-compromised mice, and show increased expression of stem cell markers such as CD44, α2β1 integrin and β-catenin in PC-3 (Li et al., 2008) and DU145 clones (Locke et al., 2005) and aldehyde dehydrogenase 1 (ALDH1) activity (Doherty et al., 2011). In contrast meroclones and paraclones can only be passaged for a limited period and are non-tumourigenic. Cells sorted based on CD44+, integrin α2β1+, CD133+ expression have a higher CFE and form more holoclones than CD44+ integrin α2β1low CD133low sorted DU145 cells. (Wei et al., 2007). However, the meroclone fraction has been studied in depth only in PC-3 cells, whilst other studies on other cell lines have concentrated on the differences between holoclones are paraclones and ignored meroclones.

The holoclone forming assay has been extensively utilised in cancer stem cell research, particularly within prostate cancer, as a surrogate stem cell assay. As well as confirming SC enrichment by cell surface expression (Patrawala et al., 2006, Gao et al., 2009, Marian et al., 2010), dye exclusion (Marian et al., 2010) and reduced PSA expression (Qin et al., 2012) the holoclone assay has given insight into how CSCs are controlled and how this may affect metastasis and disease progression. In conjunction with tumourigenicity, an increase in the number of holoclones formed by overexpression of Nanog has demonstrated a functional role for the stem cell gene Nanog in prostate cancer, which supports stem cell model (Jeter et al., 2009, Jeter et al., 2011). The holoclones assay has also demonstrates role for miR-34a in the control of prostate cancer stem cells and metastases (Liu et al., 2011). Despite their frequent use, only the PC-3 cell line colonies have been rigorously characterised.

Drug Discovery by Phage Display

Phage display is a technology that can identify a wide range of biological targets, varying from small molecules to organ specific ligands. Developed by George Smith in 1985, phage display is essentially an affinity selection of random peptides displayed on the surface of a bacteriophage that binds strongly to the target (Smith, 1985).

Bacteriophages are viruses that infect bacterial cells, and a key property is that they can incorporate DNA and translate the sequence to be expressed as peptides on their surface (Smith and Petrenko, 1997). This ability is utilised in the phage display technique, where phage are manipulated to display a potentially infinite range of random peptides. The technique then relies on the ability to rapidly identify ligands with the desired target property from a large population of phage clones (called a library library) displaying diverse surface peptides (Vodnik et al., 2011). Phage display systems that display antibody fragments have also been developed (Clackson et al., 1991) which display either scFv or Fab fragments (Carmen and Jermutus, 2002). Both peptide and antibody phage display libraries have been used in in vivo selection and panning experiments, primarily in mice (Rajotte et al., 1998, Li et al., 2006, Newton et al., 2006, Du et al., 2010). The phage library is injected and the phage homes to and binds to organs. Following sacrifice harvested tissues can then analysed to recover clones binding to them.

Bacteriophages

Bacteriophages have a simple structure which consists of a protein capsid enclosing genetic material. A wide range of bacteriophages with differing protein coats and modes of infectivity exist, although most published phage display work uses filamentous phage strains M13, fd, or f1 as the vector. Filamentous phage are shaped like flexible rods approximately 1 µm long and 6 nm in diameter composed mainly of a tube of helically arranged molecules of the 50-residue major coat protein pVIII, shown in . Inside the tube is single stranded viral DNA (ssDNA) consisting of 6407 bases coding for 11 genes, five of which are coat proteins. At one tip of the protein tube are five copies of each of the minor coat proteins pIII and pVI and at the other tip are minor coat proteins pVII and pIX. 2700 copies of the major coat protein pVIII encapsulate the phage encoded by gene 8 (Griffiths et al., 1994).

http://ars.sciencedirect.com/content/image/1-s2.0-S1389034401000879-gr1a.jpg

Figure . Structure of a filamentous bacteriophage showing the protein coats and their locations. pVIII coats the entire phage with pIX and pVII at one end and pIII and pVI at the other. pIII (or P3) coat protein is present in 5 copies and is frequently the site used to display the peptide library (Sidhu, 2001).

Bacteriophage Infection of an E.coli Host

Bacteriophages used in phage display can only infect an E.coli cell that displays the thread like appendage F pilus. They are non-lytic, meaning that they leave the host cell intact. The pIII protein has two N-terminal domains (N1 and N2) and a C-terminal domain connected by glycine-rich linkers G1 and G2 (Figure ). Infection is initiated when the N-terminal domain of pIII attaches to the tip of the host pilus and the particle enters the cell by dissolving coat proteins on the surface of the cell envelope and the uncoated ssDNA enters the cytoplasm. A complementary DNA strand is synthesised by the host, resulting in the double stranded replicative form (RF). The RF replicates to make progeny RFs and acts as the template for transcription of phage genes and synthesis of progeny ssDNAs. The new progeny ssDNAs are extruded through the cell envelope, acquiring new coat proteins from the membrane, emerging as complete virons which are excreted continuously from the host cell without killing it (Smith and Petrenko, 1997).

Figure . The modular structure of phage coat protein pIII which is essential for phage activity has two N-terminal domains (N1 and N2) and a C-terminal domain connected with glycine rich linkers (G1 and G2). It can be modified to display a peptide library between N2 and CT or at the N terminus (Carmen and Jermutus, 2002).

Peptide phage display

For phage display, foreign peptides have been fused to the coat proteins pIII, pVIII and pVI, although the most commonly used is pIII. The foreign random DNA sequence is inserted between the amino-terminal half and the carboxyl-terminal half of pIII, which minimally disrupts its function, particularly its infectivity (Smith, 1985). A phage display library contains phage clones carrying a wide range of different, random gene inserts and when the phage replicates its foreign peptides are also replicated, producing identical progeny when infecting a new bacterial host (Smith and Petrenko, 1997).

A peptide display library contains a large number of clones, typically 109- 1010 , although there can be up to 1012, and each clone expresses multiple copies of the unique peptide sequence on its surface (Lunder et al., 2005). The library is used for affinity selection assays, where the peptide library vectors are incubated with an immobilised target, a process called panning. Phages that do not bind to the target are washed away, whilst bound phage are eluted and then amplified in the E.coli host.

Amplified phage are again incubated with the immobilised target and panning is repeated 3-4 times. Too many rounds of panning can result in phage with replicative advantage being selected for. The final round eluate should contain phage which are enriched for peptides which bind to the target.

Various targets can be panned for in vitro, from simple proteins to whole live cells, either in solution or adherent. Clones displaying peptides with high affinity for the target bind strongly whilst others are washed away. Selected clones are isolated and DNA coding the displayed peptide can be sequenced. Usually, peptide-encoding DNA libraries are based on partially randomised oligonucleotides (Lindner et al.). Commercially available phage display libraries include Filamentous phage M13 Ph.D-7, Ph.D-12 and Ph.D-C7C (New England Biolabs) and Spherical T7 phage called T7 Select (Merck).

Targeting Tumours by Phage Display

Phage display has been used for a wide range of targets. Although initially phage display was used for relatively simple targets, the system is also utilised for both in vitro whole cell phage display and in vivo panning. Peptide phage display has generally yielded peptide sequences which bind to previously known cellular targets. Studies have screened peptide phage libraries against proteins of interest in cancers such as PSA (Ferrieu-Weisbuch et al., 2006, Shanmugam et al., 2011) , prostate-specific membrane antigen (PSMA) (Lupold and Rodriguez, 2004), ErbB-2 (Karasseva et al., 2002) and isolated putative cancer stem cell protein markers such as CD44 (Park et al., 2011). Of particular interest commercially have been peptides targeting growth factor activity, such as the motif CVRAC which binds to epidermal growth factor receptor (EGFR) (Cardo-Vila et al., 2010) and peptides which inhibit Vascular endothelial growth factor (VEGF) mediated angiogenesis in vitro (Yayon et al., 1993).

Peptides have been screened against whole tumour cells, cancer cell lines and putative stem cell populations. In this way peptides influencing cell attachment and invasion (Romanov et al., 2001, Fukuchi et al., 2010) and targeting specific receptors such as urokinase receptor have been identified (Goodson et al., 1994, Fong et al., 2002).

Using cancer cells in vitro for selection has yielded some interesting potential targets for cancer such as peptide VHLGYAT (Zhang et al., 2007), and HEWSYLAPYPWF (Rasmussen et al., 2002) for colon and breast carcinoma and CPLDIDFYC, believed to be a α4β1 integrin receptor for AML (Jager et al., 2007). Potential ligands have been identified panning against prostate cells using either several cancer cell lines and normal prostate epithelial cells as a negative selection step. Panning against PC-3 cells has yielded DTDSHVNL, DTPYDLTG and DVVYALSDD as potential ligands, and these have proved to be useful for in vivo imaging applications (Jayanna et al., 2010). Screening of the National Cancer Institute panel of human cancer cell lines (NCI-60) identified tri-peptide motifs which were recurrently selected across the panel, although much heterogeneity was observed (Kolonin et al., 2006). This study identified motifs that were similar to domains of human tumour ligands. Antibody phage display has identified antibody fragments which bind to prostate cancer cells but not normal prostate epithelium (Popkov et al., 2004) and single chain antibodies (scFv) which target a putative breast cancer stem cell population (Jakobsen et al., 2007, Gur et al., 2009) but not normal breast tissue.

Phage display has been used to target normal cells such as keratinocytes (Jensen et al., 2003), rat pancreatic islet-cells (Ueberberg and Schneider, 2010) and the luminal surface of polarised endothelium of human umbilical veins (Maruta et al., 2003). Some of these phage ligands have been shown to resemble binding proteins of bacterial and viral pathogens (Writer et al., 2004).

Normal Prostate Epithelial Vs Cancer Colonies

Cell Plasticity

The stem cell traits observed in type 2 DU145 colonies may also be due, in part, to the increased cell plasticity of cancer cells compared to normal prostate epithelial cells. In most normal adult tissue, including the prostate, the majority of cells are post-mitotic and unlikely to ‘dedifferentiate’ to take on a more primitive state in normal homeostatic conditions due to highly regulated genetic controls (Katoh et al., 2004). Plasticity usually occurs in response to injury or experimental manipulation (Tang, 2012) and some cell types can trans-differentiate into other mature cell types. For example, pancreatic progenitor cells which generate glucagon expressing α cells can trans-differentiate into β cells by ectopic expression of the transcription factor Pax4 (Collombat et al., 2009). Plasticity can also be induced experimentally by manipulation of transcription factors to turn one differentiate cell type into another. This technique has been used to programme mature fibroblasts into neurons or hepatocytes and in the derivation of induced pluripotent stem cells (iPS) created by the overexpression of pluripotency factors OKSM, Oct4, KLF4, SOX2 and Myc (Yamanaka, 2009). So without genetic manipulation the primary prostate epithelial cells used in this study and unlikely to demonstrate any plasticity, therefore proliferate and form colonies which reflect their inherent properties.

Cancer cells lack the normal genetic and environmental controls that are present in healthy tissue, therefore may have greater plasticity (Tang, 2012). It has been suggested that non-CSCs can dedifferentiate into CSCs which have a more stem-like phenotype (Chaffer et al., 2011, Gupta et al., 2011). This has been demonstrated to some extent by inter-conversion of ABCG2+ and ABGG2- prostate and breast cancer cells (Patrawala et al., 2005) and CD44+ and CD44- prostate cancer cells (Patrawala et al., 2006). In these studies cells the negative cell fractions were able to generate numbers of positive cells in similar to proportions observed in the original cell line. However, this is not conclusive evidence as there may be overlap between marker expression of CSCs and non-CSCs and inadequate enrichment of unselected populations.

The cancer cell micro-environment may also play a role and control cell plasticity (Tang, 2012). Overexpression of oncogenes such as Nanog (Jeter et al., 2011) and hTERT (Paranjape et al., 2012) can also result in an increase of cancer cells with stem like traits, including clonogenicity and tumourigenicity. Inflammation and infiltration of inflammatory cells and cytokines such as interleukin-1 and TGF-β may cause epithelial-mesenchymal transition (EMT). EMT is the transdifferentiation of polarized epithelial cells to mesenchymal cells which is evoked during tumor invasion and metastasis (Gao et al., 2012). In addition to promoting tumor cell invasion and metastasis, EMT has been shown to increase CSC traits of self-renewal and tumor-initiation (Mani et al., 2008) and expression of CSC markers such as CD24- CD44+ in the breast (Santisteban et al., 2009). The basic micro-environment of the tissue culture dish may not provide the normal controls and influences observed within a tumour in situ, leading to altered clonal behaviour.

Use of the Clonogenic Assay

Colony morphology is a useful surrogate assay for the study of stems in both normal and cancer tissue. The ability of both type 1 and 2 DU145 colonies to form serial tumours in mice suggests that cancers contain a larger proportion of CSCs with tumour initiating capacity than previously believed. The different colony forming efficiency of cancer cell lines, both in this study and previous work has shown that there may be a large variation in the number of CSCs in different cancers. Full characterisation of colonies formed by each cell type is essential to avoid inaccurate interpretation and underestimation of the number of cells with stem cell traits. Increased clonogenicity has been linked to more aggressive disease (Bapat et al., 2005), therefore full characterisation of cancer colonies may enable more accurate testing of potential chemotherapeutic agents for personalised medicine.

Characterised cancer colonies which are known to co