Immunotherapy of prostate cancer: should we be targeting stem cells and EMT?

Abstract

Cancer stem cells have been implicated in a number of solid malignancies including prostate cancer. In the case of localised prostate cancer, patients are often treated with surgery (radical prostatectomy) and/or radiotherapy. However, disease recurrence is an issue in about 30% of patients, who will then go on to receive hormone ablation therapy. Hormone ablation therapy is often palliative in a vast proportion of individuals, and for hormone-refractory patients, there are several immunotherapies targeting a number of prostate tumour antigens which are currently in development. However, clinical responses in this setting are inconsistent, and it is believed that the failure to achieve full and permanent tumour eradication is due to a small, resistant population of cells known as ‘cancer stem cells’ (CSCs). The stochastic and clonal evolution models are among several models used to describe cancer development. The general consensus is that cancer may arise in any cell as a result of genetic mutations in oncogenes and tumour suppressor genes, which consequently result in uncontrolled cell growth. The cancer stem cell theory, however, challenges previous opinion and proposes that like normal tissues, tumours are hierarchical and only the rare subpopulation of cells at the top of the hierarchy possess the biological properties required to initiate tumourigenesis. Furthermore, where most cancer models infer that every cell within a tumour is equally malignant, i.e. equally capable of reconstituting new tumours, the cancer stem cell theory suggests that only the rare cancer stem cell component possess tumour-initiating capabilities. Hence, according to this model, cancer stem cells are implicated in both tumour initiation and progression. In recent years, the role of epithelial–mesenchymal transition (EMT) in the advancement of prostate cancer has become apparent. Therefore, CSCs and EMT are both likely to play critical roles in prostate cancer tumourigenesis. This review summarises the current immunotherapeutic strategies targeting prostate tumour antigens taking into account the need to consider treatments that target cancer stem cells and cells involved in epithelial–mesenchymal transition.

Prostate cancer, tumour antigens and their role in immunotherapy

Prostate cancer constitutes a major health problem worldwide since it is a significant cause of morbidity and mortality in men, particularly in the developed world [1]. It is the third most common cancer in men worldwide and is most prevalent in European and North American men [2]. Prostate cancer accounts for a quarter of all new cancer cases diagnosed in men in the UK; hence, it is the most common male malignancy in the UK. Although in the case of localised prostate cancer, surgery and radiation therapy can prove to be curative, this is not true of the majority of cases which present with locally advanced or widespread disease. In the case of advanced disease, androgen ablation hormone therapy is the standard first-line palliative treatment. However, despite hormone therapy achieving castrate levels of testosterone, cancer progression to biochemical or metastatic hormone-refractory disease often leads to death since metastatic hormone-refractory prostate cancer (HRPC) currently remains an incurable disease [3, 4]. There is an ongoing need to develop new, more accurate screening tests for prostate cancer so that the disease can be identified early on where the treatment options are much more effective. For those hormone-refractory patients, new treatment options are required to help eradicate and/or manage their disease, and to this end, immunotherapy is one line of attack for advanced disease.

In the last two decades, advances in tumour immunology have led to the discovery of many peptides from tumour antigens that are recognised by cytotoxic T lymphocytes [5, 6]. There are two main adoptive immunotherapeutic approaches: undefined antigen-based (whole tumour immunome) therapies and the specific antigen -based therapies [7]. The first approach does not allow monitoring the anticancer immune response to therapy because of the lack of information on the recognised antigen(s), whereas the activity of antigen-specific T cells (CD8⁺/CD4⁺) can be assessed when using specific antigen-based immunotherapies [8]. The criteria for a tumour antigen to be a quintessential target for immunotherapy include its specificity to cancer cells, its high expression in cancer cells, its key role in cancer progression, its high frequency in patients, its location on the cell surface or its ability to be efficiently presented on major histocompatibility complex (MHC) molecules and its ability to induce a specific, potent and lasting immune response.

In order to combat the phenomenon of tumour immunoediting, vaccines should include inhibitors of the activity of regulatory cells (Treg) [9, 10], adjuvants such as toll-like receptor (TLR) agonists (e.g. CpG oligodeoxynucleotides that bind to TLR9) to activate the immune cells [11] and T cells with receptors that have been genetically engineered to be specific to the antigen(s) of interest [12, 13]. Also, antibody-based therapy, radiotherapy, hormonal ablation therapy, chemotherapy or anti-angiogenic therapy should be used in conjunction with the vaccine to increase the death or at least enhance the antigen-presenting capability of cancer cells as well as removing suppressor cells [14–16]. Finally, chemotherapeutic agents can help to increase cytotoxic T-cell entry into the site of the tumour by disrupting the tumour stromal cells [17]. Treating patients with immunotherapy as an adjuvant therapy at earlier disease stages will improve clinical benefits when compared with the vaccination of patients already bearing high-grade tumours [18–20].

Prostate cancer is a promising tumour for targeted treatment using vaccine approaches due to the expression of unique prostate-associated antigens [21]. Tumour antigens were first identified in the early 1970s by Abelev et al. (alpha-fetoprotein) [22] and Gold and Freedman (carcinoembryonic antigen) [23]. Over the last 20 years, the family of tumour antigens identified has vastly increased to include antigen categories such as tumour-associated antigens [24, 25], splice-variant antigens [26], antigens encoded by mutated genes [27], cancer-related autoantigens [28], fusion proteins [29] and, the most promising category of all, cancer testis antigens [30]. Table 1 outlines some of the more promising prostate tumour antigens that have been taken forward into clinical trials in prostate cancer.

Table 1.

Clinically relevant studies utilising prostate-associated antigens

Antigen	Function	Clinical studies
Prostate-specific antigen (PSA)	A 34-kDa kallikrein protein with serine protease activity [53]. Routinely used to assess disease progression in prostate cancer patients. Also expressed by cells in the salivary gland, small intestine, smooth muscle and renal microtubules [54]	PROSTVAC—a recombinant vaccinia-PSA [55]. Dendritophage-rPSA—autologous DCs pulsed with recombinant human PSA protein [56]. PSA-TRICORM—a recombinant pox virus vector containing PSA and 3 human T-cell co-stimulatory molecules [57, 58]. DCs loaded with PSA peptides [38, 59] or transfected with PSA mRNA [36]. Co-delivery of PSA and PSMA DNA vaccines using electroporation [60]
Prostate-specific membrane antigen (PSMA)	A 110-kDa membrane-bound glycoprotein expressed on the surface of prostatic epithelial cells. Its expression has been shown to increase after androgen deprivation therapy [61, 62], and higher levels are correlated with disease stage and Gleason score [61, 63]	Prost30—PSMA directed monoclonal antibody [64, 65]. DCvax—autologous cells exposed to 2 PSMA peptides [34]. DCs loaded with PSMA peptides [33, 54, 66, 67]
Prostatic acid phosphatise (PAP)	A 386-amino acid protein that is secreted by the prostate [68]. Increased levels of PAP have been detected in the circulation of patients with advanced stage disease [69]. Decreases in PAP levels have been correlated with response to therapy [70]	PROVENGE—a DNA vaccine consisting of a combination of PAP and GM-CSF introduced ex vivo into autologous DC before reinjection into the patient [31]
Prostate stem cell antigen (PSCA)	A GPI-anchored cell surface antigen related to the LY-6/Th-1 superfamily with 30 % homology to stem cell antigen 2 [71]. PSCA is localised to the basal cell epithelium in normal prostates and in prostate cancer it is highly expressed on the secretory epithelium. The acquisition of androgen independence and metastatic disease leads to elevated expression [72–74]	DCs loaded with PSCA peptides administered to hormone and chemo-resistant patients [75]. DCs loaded with a combination of PAP, PSMA and PSA peptides administered to hormone-resistant patients [38]

Dendritic cells are known to play an essential role in the establishment of both innate and adaptive antitumour immune responses. Dendritic cell-based vaccines have already shown promising results by producing significant immune responses against prostate tumour antigens such as PAP [31, 32], PSMA [33–35] and PSA [36] as well as being essentially free of side effects [37]. For example, a dendritic cell-based multi-epitope vaccine using prostate stem cell antigen (PSCA14–22), prostatic acid phosphatase (PAP299–307), prostate-specific membrane antigen (PSMA4–12) and prostate-specific antigen (PSA154–163) has elicited significant cytotoxic T-cell responses and the generation of memory T cells [38]. Of particular significance, and an important landmark for the treatment of asymptomatic or minimally symptomatic, metastatic HRPC, was the formulation of the first adoptive cell transfer regimen, called sipuleucel-T (PROVENGE), which was approved by the United States Food and Drug Administration in April 2010 [16]. Sipuleucel-T is composed of autologous dendritic cells which are harvested by leukapheresis and processed in vitro in order to express recombinant PAP (PA2024) linked to GM-CSF. The antigen-presenting cells are then intravenously injected to the patient within 2 days and the regimen is given 3 times in total, once every 2 weeks [16]. Patients participating in late-stage randomised trials had a statistically significant extension of their life of at least 4 months, with an overall survival of about 20 months (P = 0.01) [39]. The treatment was deemed safe and the most common side effects were fever, tremor, rigours and hypersensitivity to cold. Antibody and T cell-specific activities to PA2024 were observed in the treated patient population. A large phase III study was then conducted with 512 patients, and the data confirmed the preliminary findings and also showed that treating this patient population with sipuleucel-T significantly reduced the risk of death by 22% when compared with placebo (P = 0.032) [21]. However, PROVENGE is a good example of the current limitations of cancer vaccines as it is very complex and expensive to produce and only reduces the rate of disease progression rather than curing the cancer [39]. It could be argued that earlier intervention with such forms of immunotherapy would provide longer-lasting clinical benefits.

There are also other prostate antigens that are still in early development and have only been tested in mouse models of prostate cancer. These include six-transmembrane epithelial antigen of the prostate (STEAP), which demonstrated the inhibition of prostate cancer progression when TRAMP-C1 mice were immunised with recombinant DNA and Ankara vectors delivering PSCA and STEAP antigens [40]. Monoclonal antibodies against STEAP-1 significantly inhibited tumour growth in mouse models using patient-derived LAPC-9 prostate cancer xenografts [41]. Prostate secretory protein of 94 amino acids (PSP-94) has also demonstrated efficacy in mouse models; syngeneic models of rat prostate cancer were given either subcutaneous or intracardiac injections of MatLyLu-PTHrP cells, and administration of PSP-94 resulted in tumour growth inhibition and prevention of skeletal metastases [42, 43].

Some of the identified prostate tumour antigens have also been investigated as potential disease biomarkers. Alpha-methyl acyl-CoA racemase (AMACR) is routinely used in combination with p63 on ambiguous clinical prostate cancer biopsies, and trace amounts of autoantibodies against AMACR have been detected in the circulation of prostate cancer patients [44, 45]. Other prostate-related proteins have been included in urine-based diagnostic tests, and these include differential display code antigen 3 [46] and glutathione s-transferase P1 (GSTP1) [47]. In addition, several prognostic and diagnostic markers have been identified, and these include EZH2, a biomarker for post-prostatectomy relapse [48, 49], human kallikrein 2 [50] and early prostate cancer antigen which has been incorporated into an ELISA measuring serum levels for prostate cancer diagnosis [51]. Of particular interest to our group is the prostate-associated tumour antigen termed T21 [52]. T21 was originally identified via the serological screening of a testis cDNA library with sera from prostate cancer patients. The technique of identification alone strongly demonstrates that the immune system of patients with prostate cancer is able to illicit an immune response against T21. Further work on T21 has demonstrated its low-level expression in normal tissues at both the mRNA level and protein level. In prostate cancer, we found T21 mRNA to be significantly over-expressed in malignant glands compared with benign glands and stroma. At the protein level, T21 expression was higher in patients with a high Gleason score, and it was significantly over-expressed in patients with stage pT4 disease. These findings suggest that T21 is a promising marker for late-stage disease and may also prove to be a valuable target for prostate cancer immunotherapy.

Cell types of the prostate

The three most abundant epithelial cell types within the normal, adult prostate are the secretory luminal, basal and neuroendocrine cells, respectively. The androgen-dependent secretory luminal cells are terminally differentiated and secrete prostate-specific antigen (PSA) and prostatic acid phosphatase (PAP) into the glandular lumina in response to androgens. Basal cells, which are not dependent on androgens for their survival and are less differentiated than luminal cells, surround the luminal cells and rest on the basement membrane. The neuroendocrine cells, which are relatively rare in comparison with the other cell types, are terminally differentiated and androgen-insensitive. These cells are involved in the production of neuropeptides which include serotonin and calcitonin. All three cell types are thought to derive from prostate epithelial stem cells which reside in the basal cell compartment [53, 54].

Of late, there has been much evidence implicating epithelial prostate stem cells as targets for transformation in prostate cancer, and this has been the topic of several well-written reviews and research articles [53–59]. Indeed, the role of the so-called cancer stem cells in prostate cancer initiation and progression has sparked huge interest, and, as such, prostate cancer stem cells are the subject of intense investigation.

Cancer stem cells

Cancer stem cells and their potential role in tumourigenesis have been the topic of much debate in recent years as they may help to elucidate some of the unexplained phenomena, including cancer relapse and metastasis. Recent advances in stem cell biology have led to the identification of adult tissue stem cells in various organs which, in turn, has enabled research into such populations as potential candidates for malignant transformation. To date, putative cancer stem cells have been identified and isolated from solid and haematological malignancies including prostate, pancreatic, brain, colon, head and neck, gastric, lung, liver, acute myeloid leukaemia, multiple myeloma, melanoma and ovarian cancers [60–73] (Table 2).

Table 2.

Properties and reported markers of cancer stem cells

Properties
Asymmetrical and symmetrical division
Highly resistant to drugs and toxins
Resistant to apoptosis
Highly proliferative
Marked capacity for differentiation
Responsible for tumour initiation

Reported markers	Cancer types	References
CD133+	Hepatic, brain, colon, lung, prostate, pancreas	[64, 65, 68, 72, 74–78]
CD44+	Breast, prostate, ovarian, pancreatic, head and neck squamous cell carcinoma, colon	[76, 79–84]
CD24-/low	Breast	[79, 83]
CD24+	Pancreatic	[81]
CD34+	Acute myeloid leukaemia	[85–87]
CD38−	Acute myeloid leukaemia	[85–87]
CD123+	Acute myeloid leukaemia	[85–87]
CD117+	Ovarian	[84]
CD90+	Liver	[88]
Nanog+	Ovarian, oral, prostate	[84]
Oct-3/4+	Ovarian, breast, oral, lung, prostate	[84]
ALDH1+	Breast	[89]
EpCAMhigh	Colon	[80]
ESA+	Pancreatic	[81]
ABCB5+	Melanoma	[66]
Integrinα2β1high	Prostate	[76]

The cancer stem cell is hypothesised to be the original cell of a tumour which is held solely responsible for tumourigenesis, tumour differentiation, tumour maintenance, tumour spread (metastasis) and tumour relapse following therapy [74]. Cancer stem cells have been shown to demonstrate a marked capacity for proliferation, self-renewal and differentiation and are generally believed to constitute a very rare population of cells among a majority of tumour cells with a more differentiated phenotype. Aside from their self-renewal and differentiation capabilities, cancer stem cells have been shown to share many other properties with adult stem cells. For example, akin to normal stem cells, cancer stem cells are reported to be highly resistant to drugs and toxins, through the expression of drug efflux pumps and are to resist to apoptosis, through a variety of mechanisms including active DNA repair [75–83].

Considering these facts, the proposed properties of cancer stem cells may explain why disease relapses occur in many cancers since it is believed that current cancer therapies only eliminate the bulk of the tumour, leaving the (resistant) cancer stem cells behind to reinitiate tumour growth [84]. Cancer stem cells have been reported to express many genes known to be important in somatic (normal) stem cell maintenance such as BMI-1 and Oct3/4 [85]. Furthermore, many of the markers known to identify putative CSCs in several tumour types are common to the normal adult stem cells from the healthy tissue [60, 86–88]. Such findings support the notion that cancer stem cells are in fact transformed adult stem cells. However, although the cancer stem cell theory draws parallels between normal tissue stem cells and tumour-initiating cells, thus far, not all studies have addressed whether or not cancer arises from the transformation of normal stem cells and have rather suggested that cancers consist of a heterogeneous population of cells which may be organised in a hierarchical manner much like normal tissues [89]. Hence, tumours are thought to contain a stem-like component which drives proliferation although the source of such a component is controversial, with more than one theory for the origin of cancer stem cells having been proposed.

Figure 1 demonstrates that CSCs could potentially arise from mutated adult tissue stem cells, mutated progenitor cell/transit amplifying cells or differentiated cells which acquire mutations conferring de novo self-renewal and differentiation capabilities. Although the origin of the cancer stem cell has not yet been elucidated for every tumour type, there is evidence to support all three mechanisms and perhaps CSCs can arise by any one, or even all, of the three proposed pathways.

Fig. 1.

The cancer stem cell hypothesis (adapted from [99]). Summary of the cancer stem cell hypothesis and some of the proposed origins of the cancer stem cell

There is a large body of evidence to suggest that cancer stem cells are derived from transformed adult tissue stem cells [58, 90–94]. It has been suggested that normal stem cells may be the only cells which live long enough to acquire sufficient genetic mutations to become cancer stem cells [95]. Alternatively, there is also much evidence to suggest that CSCs derive from a more committed progenitor/transit amplifying cell [90, 92, 94] but there is less data to support the notion that cancer stem cells arise from transformed differentiated cells which acquire stem-like properties. However, since it has been proven that differentiated cells can be reprogrammed to induce pluripotency [96, 97], it is plausible that activating mutations in otherwise silenced stem cell genes such as Oct3/4, Sox2, and c-Myc may confer stem-like properties in differentiated cells.

Stem cells in prostate cancer and EMT

The stem cell model for prostate epithelia was established by Isaacs and Coffey (1989) following a series of androgen cycling experiments in a rodent model. In demonstrating that the prostate, an androgen-dependent organ, could completely regenerate following castration once androgen levels were restored, they proposed that androgen-independent stem cells yield a population of androgen-responsive cells which in turn give rise to the more differentiated androgen-dependent secretory luminal cells. This model for the hierarchy of the prostate gland is now widely accepted and prostate stem cells have since been isolated and investigated [87] Fig. 2.

Fig. 2.

Demonstrating the process of an epithelial to mesenchymal transition whereby epithelial markers are down-regulated and mesenchymal markers are up-regulated, conferring the loss of cell–cell adhesion and cell polarity and the acquisition of migratory and invasive properties

In 2005, Collins et al. isolated ‘tumourigenic prostate cancer stem cells’ using the same markers they had previously established for normal prostate tissue stem cells [58, 87]. The marker profile identified for prostate cancer stem cells was integrinα₂β^high₁/CD44+/CD133+. Integrinα₂β₁ (also referred to as VLA-2) is a transmembrane receptor for extracellular matrix proteins such as laminin, collagen and fibronectin and adhesion molecules such as E-cadherin. Among other functions, it is known to play a role in the generation and organisation of extracellular matrix proteins and mediate interactions between adhesion molecules on adjacent cells. CD44, a cell surface glycoprotein involved in cell–cell interactions, cell adhesion, migration and lymphocyte activation, has been implicated as a cancer stem cell marker for several other cancers including breast, head and neck and colorectal [98–100]. CD133 has also been implicated as a marker for putative cancer stem cells in other cancers including brain, pancreatic and lung [72, 101, 102]. To date, the function of this molecule remains to be elucidated.

Collins et al. [58] demonstrated that cells which were positive for all three markers possessed significant capacity for self-renewal, differentiation and invasiveness in vitro, all properties which have been attributed to cancer stem cells. Such cells may be responsible for both the initiation and progression of prostate cancer, i.e. relapse and metastatic spread despite androgen ablation or other methods. However, this may not necessarily be the case.

Recently, there has been much interest in the phenomenon referred to as epithelial to mesenchymal transition (EMT) and its role in cancer progression and metastasis. Epithelial to mesenchymal transition is a crucial transdifferentiation programme which is known to occur during embryogenesis and in adult tissues following wound repair and organ remodelling in response to injury [103]. During the EMT process, epithelial cells undergo multiple biochemical changes involving the down-regulation of epithelial markers, which confer cell–cell and cell–extracellular matrix (ECM) adhesion, and the up-regulation of mesenchymal markers, conferring increased production of ECM components, enhanced migratory capacity, invasiveness and increased resistance to apoptosis [104, 105]. The hallmarks of an EMT are also characteristic of metastatic cancer cells, as such EMT has recently been implicated in cancer metastasis as the enhanced migratory capacity and invasiveness of mesenchymal cells is believed to facilitate the dissemination of cancer [106]. Furthermore, since EMT has been shown to confer increased resistance to apoptotic agents commonly used in chemotherapy, this phenomenon is thought to be a critical step in tumour metastasis [107–109]. In order to establish new tumours at the metastatic sites, it is believed that the cells which transition from an epithelial to a mesenchymal state and migrate must undergo the reverse procedure, mesenchymal to epithelial transition (MET) [110]. Therefore, metastasis is considered to be a dynamic and complex process involving cellular plasticity.

One study conducted in 2005 demonstrated that the expression levels of Twist, a key transcription factor in EMT, were positively correlated with Gleason grading and metastasis [111]. Furthermore, Twist has been shown to inhibit apoptosis in cancer cells and it is thought to play a central role in the resistance to microtubule-disrupting agents [112]. Indeed, the potential of Twist as a target in advanced and/or metastatic prostate cancer has been discussed [105]. In addition, other studies have correlated the presence of critical EMT drivers with disease relapse and poor survival [113, 114]. A recent study, conducted by Tanaka and colleagues, demonstrated a clear link between the expression of N-cadherin, a mesenchymal cadherin associated with EMT, and invasive, metastatic, castration-resistant prostate cancer (CRPC). In this study, N-cadherin-specific antibodies were shown to delay the progression to castration resistance, inhibit the invasion of surrounding tissues, suppress tumour growth and reduce metastasis in castrated mice. Hence, this work provides further support for the critical role of EMT in prostate cancer progression and the potential of immunotherapy as a strategy to combat this disease [115].

In 2008, Mani et al. [116] established a link between EMT and epithelial stem cells in that the products of an EMT, artificially induced in immortalised mammary epithelial cells, demonstrated stem cell properties and that stem-like cells (CD44^hiCD24^lo) isolated from mouse/human mammary glands or carcinomas expressed EMT markers [116]. These findings indicate that epithelial to mesenchymal transitions of differentiated cancer cells may give rise to cancer stem/stem-like cells in breast cancer. However, this study does not address the cell of origin of breast cancer, and in this human cancer at least, the original tumour-initiating cell may be distinct from the cancer stem/stem-like cells involved in tumour progression and metastasis. One study, conducted on circulating tumour cells (CTCs) of metastatic breast cancer patients, demonstrated that a major proportion of these cells were positive for both stem cell and EMT markers, thus providing further evidence for a link between cancer stem cells and epithelial to mesenchymal transition [117].

The cancer stem cell theory holds the cancer stem cell solely responsible for tumour initiation, progression and metastasis. There is much evidence to support the role of EMT in prostate cancer metastasis [115, 118, 119], yet the putative cancer stem cell markers identified by Collins et al. [58] are not entirely consistent with the EMT phenotype therefore, perhaps both have a role to play in prostate cancer. As such, it is plausible that the progeny of transformed stem cells are also tumourigenic and that such progeny may undergo EMT which then confers the ability to invade and metastasise. Therefore, whereas the stem cell may be the initial site of malignant transformation, their transdifferentiating progeny, rather than the cancer(ous) stem cells themselves, may be accountable for invasion and metastasis. As such, the term ‘cancer stem cell’ may be considered to refer to a number of different cell populations including transformed SCs and EMTs, rather than one ‘super-malignant’ cell population, all of which have a role to play in tumourigenesis. Conversely, EMT may be restricted to the cancer stem cell component of a tumour; however, this remains to be demonstrated in prostate cancer. Consequently, in order to achieve full and permanent eradication of the tumour, cancer therapies should aim to target the tumour-initiating cancer stem cells as well as their more differentiated progeny and potentially the populations undergoing EMT.

Therapies to target prostate cancer stem cells

Chemotherapy can increase patient survival by a few months, but the quality of life is often reduced by uncomfortable side effects [120, 121], and patients with metastatic disease often fail to respond to treatment. Furthermore, drug-resistant proteins (e.g. MDR1 and ABC transporters) are more likely to be expressed by cancer stem cells than the more differentiated cancer cells and render the CSCs more resistant to chemotherapeutic drugs [122]. Therefore, although the tumour mass may decrease following therapy, the few surviving cancer stem cells are thought to eventually regenerate the bulk of the tumour, hence the disease will inevitably progress.

There is a need to develop treatment modalities for metastatic prostate cancer which destroys the cancer stem cells, but not normal tissue stem cells [122]. Normal stem cells and cancer stem cells will share many signalling pathways (e.g. pluripotency programmes); however, it is likely that these pathways will be more active/required in the cancer stem cells due to cell-intrinsic (e.g. mitotic activity) and cell-extrinsic (e.g. regenerative activity in the tissue) factors [122]. As the prostate epithelium is removed in patients undergoing radical prostatectomy, an acceptable side effect of therapy targeting the prostate cancer stem cells would be the loss of prostate epithelial stem cells but not other essential stem cells [122]. Chemotherapeutic drugs have already proven to be selective in killing cancer stem cells in testicular germ cell cancers (e.g. cisplatin) and leukaemia (idarubicin with a proteasome inhibitor) [123, 124].

In order to target the cancer stem cell component of a tumour by immunotherapy, it is essential to establish whether these cells express functional MHC molecules. There is very little literature addressing the MHC status of cancer stem cells; however, normal stem cells, including embryonic stem cells (ESCs) and mesenchymal stem cells (MSCs), have been shown to express MHC class I [125, 126]. Drukker et al. demonstrated that embryonic stem cells expressed low levels of MHC class I which was dramatically up-regulated following treatment with interferon gamma. They did not observe the expression of MHC class II or HLA-G, a non-classical HLA class I molecule involved in immunomodulation. Conversely, Selmani et al [127] identified the expression and secretion of HLA-G by human adult bone marrow-derived MSCs. This study demonstrated the modulation of both adaptive and innate immunity by HLA-G through the suppression of allogenic T-cell proliferation, the expansion of CD4-CD25^highFOXP3 regulatory T cells, the inhibition of natural killer (NK) cell-mediated cytolysis and the inhibition of interferon gamma secretion. The expression of HLA-G by cancer stem cells would prove problematic when considering these cells as targets for immunotherapy. Therefore, it is prudent to investigate the immune status of cancer stem cells, with regard to the expression of non-classical molecules as well as classical HLA molecules, in order to develop strategies to target this population using immunotherapy. One study, investigating the immunogenicity of putative astrocytoma and glioblastoma cancer stem cells (CD133+ cells), revealed that the majority of CD133-expressing cells did not express detectable MHC class I or NK cell-activating ligands. Therefore, not only would such cells prove resistant to adaptive and innate immune surveillance, they would appear to be unsuitable targets for immunotherapy. However, up-regulation of MHC class I and NK cell ligands was successfully achieved following treatment with interferon gamma, hence the immunogenicity of the cells was restored [128]. Despite low levels of MHC class I being reported in both embryonic stem cells and putative astrocytoma and glioblastoma cancer stem cells, expression of this molecule was dramatically increased in both studies following treatment with interferon gamma. As such, the implementation of interferon gamma in an immunotherapy regimen targeting cancer stem cells may render these cells more susceptible to immune attack.

The MHC status of cancer cells undergoing EMT and the reverse process MET is also of importance when considering these cells as targets for immunotherapy. Furthermore, it is critical to identify unique markers expressed by these cancerous transdifferentiating cells which could be exploited for immunotherapy. Although transcription factors involved in EMT, such as Twist and Snail, may appear attractive targets for cancer therapy, they may not prove to be ideal targets for immunotherapy as EMT is an essential programme required for tissue repair and organ remodelling throughout life. Hence, targeting critical EMT factors may prove detrimental to the patient. It is therefore necessary to identify unique antigens, mutated forms of essential transcription factors/key EMT drivers, or antigens which are only up-regulated in cancer cells undergoing EMT. Epithelial to mesenchymal transition has been reported to play a critical role in cancer progression and metastasis. Providing that cancer cells undergoing these transdifferentiation programmes retain malignant characteristics which distinguish them from healthy cells in both states (epithelial and mesenchymal), these cells may prove to be ideal targets for immunotherapy.

Comparison of gene-expression profiles of normal tissues/stem cells versus cancer cells/cancer stem cells can identify novel target antigens for therapy which are preferentially expressed in cancer stem cells [122]. Patients who present a macroscopic tumour could be given immunotherapy against the antigens found on the more differentiated cells in order to remove the bulk of the tumour, as a first-line treatment, while patients who present residual disease following a regimen of therapies could be given immunotherapy tailored against the cancer stem cell-specific antigens [129]. Yawata and colleagues have recently demonstrated the transcriptional activation of the cancer testis antigen (CTA) gene locus in glioma stem cells which showed frequent and high expression of CTA genes [130].

Most of the researchers working on the subject of prostate cancer stem cells agree that studies must be carried out on primary cancer cells rather than cancer cell lines, whenever possible. Changes in the environment are inevitable in in vitro cell culture and in vivo animal models/xenografts, and these could be inducing a deregulation of the expression of cancer stem cell markers leading to inaccurate measurements of their activity. It is therefore essential to precisely recreate the tumour stroma/microenvironment to allow observation and report of true phenomena in order to develop effective therapeutics [131, 132].

Concluding remarks

Although the clinical responses observed with the current immunotherapies are encouraging, they are not entirely satisfactory [133, 134]. For instance, an important drop in the level of PSA is indicative of a good response to any therapeutic intervention in HRPC; however, no such phenomenon was observed following the administration of existing immunotherapies [135, 136]. These treatment failures could be explained by the unsuccessful targeting and elimination of the cells responsible for metastasis and recurrence. Since cancer stem cells and EMT have both been implicated in tumourigenesis, it is critical to examine both populations and determine their immune statuses in order to develop strategies to target these populations using immunotherapy. Novel immunological targets are required and should comprise the specific antigens expressed by differentiated cancer cells, EMTs and CSCs.

It is critical to fully characterise the nature and relationships between these populations in order to develop specific vaccines which will provide the efficient and long-term eradication of all cancer cells, thus a cure and not just another palliative treatment. The target cells must not only express high levels of the prostate tumour antigens of interest but also high levels of MHC class I molecules must be expressed or inducible. Research should also focus on developing tailored treatments by identifying patients who are likely to respond to a specific therapeutic regimen. Molecular and biochemical technologies including whole-genome microarrays and mass spectrometry, respectively, could indicate the presence or absence of tumour antigens, thus allowing an informed decision to be made prior to immunotherapy [137, 138].