Human α2β1 HI CD133+VE Epithelial Prostate Stem Cells Express Low Levels of Active Androgen Receptor

Stem cells are thought to be the cell of origin in malignant transformation in many tissues, but their role in human prostate carcinogenesis continues to be debated. One of the conflicts with this model is that cancer stem cells have been described to lack androgen receptor (AR) expression, which is of established importance in prostate cancer initiation and progression. We re-examined the expression patterns of AR within adult prostate epithelial differentiation using an optimised sensitive and specific approach examining transcript, protein and AR regulated gene expression. Highly enriched populations were isolated consisting of stem (α2β1 HI CD133+VE), transiently amplifying (α2β1 HI CD133–VE) and terminally differentiated (α2β1 LOW CD133–VE) cells. AR transcript and protein expression was confirmed in α2β1 HI CD133+VE and CD133–VE progenitor cells. Flow cytometry confirmed that median (±SD) fraction of cells expressing AR were 77% (±6%) in α2β1 HI CD133+VE stem cells and 68% (±12%) in α2β1 HI CD133–VE transiently amplifying cells. However, 3-fold lower levels of total AR protein expression (peak and median immunofluorescence) were present in α2β1 HI CD133+VE stem cells compared with differentiated cells. This finding was confirmed with dual immunostaining of prostate sections for AR and CD133, which again demonstrated low levels of AR within basal CD133+VE cells. Activity of the AR was confirmed in prostate progenitor cells by the expression of low levels of the AR regulated genes PSA, KLK2 and TMPRSS2. The confirmation of AR expression in prostate progenitor cells allows integration of the cancer stem cell theory with the established models of prostate cancer initiation based on a functional AR. Further study of specific AR functions in prostate stem and differentiated cells may highlight novel mechanisms of prostate homeostasis and insights into tumourigenesis.


Introduction
Androgen signalling has been shown to be integral to prostate cancer development as it can induce and regulate TMPRSS2-ERG gene fusions, which initiate malignant transformation and drive disease progression [1][2][3]. Even without this fusion, AR signalling remains central to prostate carcinogenesis [4][5][6].
There is increasing evidence that stem cells are the targets for tumourigenesis due to their inherent self-renewal capability, antiapoptotic pathways and maintenance throughout the lifetime of an individual granting time for mutations to accumulate. Human studies of tumourigenesis in xenografts have demonstrated the importance of AR signalling in disease initiation in the basal layer of prostate epithelium [6]. In mice, evidence is growing that there are both basal and luminal stem cells and debate remains over where the critical tumourigenic mutations occur, nevertheless both these models of carcinogenesis required an active AR [7][8][9][10][11]. In the human setting, a common clonal origin has been confirmed for basal, luminal and neuroendocrine cells [12,13]. Human prostate stem cells can be enriched by their gene signature of a 2 b 1 HI and glycosylated CD133 expression, transiently amplifying cells are characterised by a 2 b 1 HI CD133 -VE expression and terminally differentiated cells are defined by the marker a 2 b 1 LOW CD133 -VE [14][15][16][17]. Both stem cells and cancer stem cells described by these signatures from primary human prostates have typically lacked AR expression [14,18]. The existence of AR -VE cancer stem cells has been postulated as a mechanism by which tumours relapse by overcoming androgen ablative therapies that target AR +VE cells [18]. However, it is established that the AR remains active and even amplified in castration resistant prostate cancer (CRPC) [19][20][21]. If the prostate stem cell is the cell of origin for transformation, then this model appears to be at odds with the emerging mechanisms of prostate cancer development and progression dependent upon AR signalling. In this work, we focus on reexamining the expression profiles of AR in prostate epithelial differentiation and challenge the dogma that prostate stem cells lack AR.

Tissue Collection and Isolation of Epithelial Cells
Human prostate samples were obtained from 20 patients following transurethral resection of the prostate for benign prostatic hyperplasia or cystoprostatectomy for bladder cancer. Pathologist assessment confirmed benign histology and the samples underwent processing and selection as previously described [14][15][16]: Magnetic activated cell sorting (MACS) was performed for immunomagnetic selection of Epithelial Cell Adhesion Molecule (EpCAM/CD326) (Miltenyi Biotec, Woking, UK). Epithelial a 2 b 1 HI (stem and transiently amplifying cells) and a 2 b 1 LOW (differentiated) cells were selected by rapid adhesion to collagen-1. Epithelial a 2 b 1 HI CD133 +VE cells were separated by either CD133 immunomagnetic selection (CD133/1, Miltenyi Biotec) or FACS (CD133/2, Miltenyi Biotec). In our work, selected primary samples were never cultured prior to experimentation to avoid adaptations of cells in an in vitro environment and subsequent deviation of their phenotypes [22][23][24].

Maintenance of Prostate Cancer Cell Lines
The human prostate cancer cell lines LNCaP (AR +VE ) and PC3 (AR -VE ) (American Type Culture Collection) were maintained in RPMI1640 medium (Sigma, Dorset, UK) containing 10% foetal calf serum and 2 mM L-glutamine.

Quantitative Real Time PCR Analysis
Prostate epithelia was separated into three distinct fractions, a 2 b 1 HI CD133 +VE stem cells, a 2 b 1 HI CD133 -VE transiently amplifying cells and a 2 b 1 LOW CD133 -VE terminally differentiated cells and underwent RNA isolation (micro RNeasy, Qiagen, Crawley, UK). Message BOOSTER TM cDNA synthesis amplification kit (Epicentre Biotechnologies, Madison, WI, USA) was employed and real-time PCR (Applied Biosystems 7900HT) was performed using SYBR green (Invitrogen) using the following specific

Flow Cytometry
Cells were fixed with Fixation/Permeabilisation solution (BD Bioscience, Oxford, UK) before incubation in methanol at 220uC for 16 hours to permeabilise the nucleus. Cells were labelled with anti-AR antibody (PG-21, Millipore) and secondary FITC (Dako, Ely, UK), and CD133/2 antibody directly conjugated to PE (Miltenyi Biotec). Controls included IgG isotype antibody (Dako) and PE conjugated isotype antibody (Miltenyi Biotec). When required, cells were counterstained with the nuclear stain DRAQ5 TM (Biostatus, Shepshed, UK) according to the manufacturer's recommendations. Samples were analysed using either a FACS Calibur flow cytometer (BD Biosciences) or ImageStreamX Mark II cytometer (Amnis, Ipswich, UK).

Sequential Alkaline Phosphatase Immunostaining
CD133/AR dual staining was carried out on prostate sections using methodologies previously described [25]. Briefly, sections were stained with CD133/1 antibody (Miltenyi Biotec) according to manufacturer's recommendations prior to detection using Poly-AP-GAM/R/R Immunoglobulins (Immunologic, Duiven, The Netherlands) and visualised with Alkaline phosphatase substrate kit III (SK-5300, Vector labs, Burligame, Ca, USA). Sections were cleared of antibodies with a second antigen retrieval before staining for AR (SC-816, Santa Cruz), detection with Poly-AP-GAM/R/R Immunoglobulins and visualisation with Alkaline phosphatase substrate kit I (SK-5100, Vector labs) and mounting slides in Vecta Mount (H-5000, Vector labs).

Results
The Purity of Human EpCaM +VE a 2 b 1 HI CD133 +VE Prostate Cell Selections was Confirmed It is established that primary culture results in changes in phenotype from those seen in vivo [23,24,26] and of particular relevance, studies in glioma have demonstrated that CD133 expression is altered as a result of in vitro conditions [27]. Therefore, in our studies no culturing of samples was carried out prior to experimentation. However, this approach can lead to a greater chance of contamination by unwanted cell lineages, such as blood or stroma cell types, and a previously optimised protocol for epithelial extraction was employed ( Figure 1A) [14][15][16]28,29]. Lineage-specific markers for epithelial (CD24), endothelial (CD146) and haematopoietic (CD45) cells were assessed and the purity of epithelial cell enrichment was confirmed by real time PCR, confirming depletion of unwanted cell lineages with this method ( Figure 1B). In order to assess the enrichment of glycosylated CD133, cells were dual-stained for CD133/1 immunomagnetic beads and an anti-CD133/2 antibody that targets an epitope distinct from CD133/1. Flow cytometry data showed that in the prostate CD133/1 and CD133/2 are exclusively co-expressed, confirming .98% purity in glycosylated-CD133 +VE cells ( Figure 1C). Similarly using real time PCR, immunomagnetic selection for glycosylated-CD133 resulted in enrichment for cells expressing high levels of CD133 mRNA and depleted CD133 expression in the negative fractions ( Figure 1D).

The Sensitivity and Specificity of AR Detection by Flow Cytometry was Validated
In order to accurately determine the presence of the AR in rare cell types within the prostate, a flow cytometry approach was developed, allowing both sensitive and specific quantification.
Using a specific AR antibody (PG-21, Millipore), an optimised staining protocol was developed using prostate cancer cell lines LNCaP (AR +VE ) and PC3 (AR -VE ), allowing identification of the highest concentrations of antibody to increase sensitivity whilst controlling for non-specific labelling ( Figure 2A). Specific AR expression was confirmed by comparison to an IgG-specific isotype control in conjunction with the AR -VE cell line PC3 ( Figure 2B). Specificity of this staining was further confirmed in LNCaP following siRNA knockdown of the AR using flow cytometry ( Figure 2C and 2D). Western Blot analysis was also employed to confirm the specificity of the AR knockdown with siRNA, which correlated directly with the flow cytometry result ( Figure 2E).

Androgen Receptor was Detected at Low Levels in Prostate Stem Cell Enriched Cells
AR transcript levels were examined in a 2 b 1 HI CD133 +VE , a 2 b 1 HI CD133 -VE and a 2 b 1 LOW CD133 -VE cells and were detectable at every stage of differentiation from stem cells through  to terminally differentiated cells from each patient sample examined (n = 10) ( Figure 3A). Flow cytometry confirmed that similar proportion of CD133 +VE stem cell enriched cells and CD133 -VE transiently amplifying cells expressed AR protein ( Figure 3B), with the mean (6SD) fraction of cells stained positive for AR being 77% (66%) in a 2 b 1 HI CD133 +VE cells and 68% (612%) in a 2 b 1 HI CD133 -VE cells ( Figure 3C). Primary progenitor enriched a 2 b 1 HI cells (CD133 +VE stem and CD133 -VE transiently amplifying cells) demonstrated a clear population shift in fluorescence intensity on the histogram from flow cytometry compared to the isotype control when stained with AR ( Figure 3D). In particular, the primary a 2 b 1 HI cells showed 2.54 (SD = 60.50) fold increase in the median and 2.45 fold (SD = 60.31) in the peak fluorescence compared to their isotype controls, whereas a 2 b 1 LOW terminally differentiated cells, showed an increase by 8.69 (SD = 61.54) fold for the median and 6.80 (SD = 61.11) fold for the peak fluorescence compared to their isotype controls ( Figure 3E). In particular, the a 2 b 1 HI progenitor cells did not stain as intensely (3-fold lower) as terminally differentiated prostate epithelial cells (p,0.005) ( Figure 3E). To confirm this result, samples were analysed with the ImageStreamX Mark II cytometer, which combines flow cytometery with fluorescent microscopy, allowing visualisation of rare cell events within a sample. AR was once again confirmed to be expressed by a 2 b 1 HI CD133 +VE and a 2 b 1 HI CD133 -VE cells with higher immunofluorescence expression seen within a 2 b 1 LOW CD133 -VE cells ( Figure 4A). By counterstaining cells with DRAQ5 TM , AR localisation could be estimated within the cells in suspension. All three fractions showed AR expression overlaying with the nucleus, suggesting the AR is active throughout prostate differentiation. In order to establish if the AR was functionally active in the rare prostate progenitor cells, expression of androgen regulated genes was examined ( Figure 4B). Expression of PSA, KLK2 and TMPRSS2 was confirmed at all stages of prostate differentiation, including activity in the a 2 b 1 HI CD133 +VE stem cell enriched cells. Corresponding with the expected increased expression of AR in differentiation, a marked increase in the AR

Distribution of the Androgen Receptor and Prostate Stem Cells within the Human Prostate Epithelium
To evaluate the expression of AR and CD133 within normal prostate histology, dual alkaline phosphatase staining was employed. Using an optimised AR antibody, it was possible to detect AR within the prostate epithelium in both the luminal layer in addition to low levels within the basal epithelium ( Figure 5A). Dual staining also allowed the identification of CD133 positive cells within the prostate epithelium in 2% of sections studied (8/ 342 sections), with cells always restricted to the basement membrane of the epithelium with localised staining of CD133 in keeping with its expression limited to membrane protrusions with a characteristic punctate pattern [14,30]. As with the flow cytometry data, the identified CD133 +VE cells also expressed AR but at a lower level than that seen in luminal cells within the same gland ( Figure 5B).

Discussion
In this study, we established that AR expression is detectable at low levels in the a 2 b 1 HI CD133 +VE stem cell enriched cells of the prostate, using a combination of PCR, dual staining immunohistochemistry and highly specific and sensitive flow cytometry methods. Moreover, we confirmed that AR was active in the stem cell enriched populations by their expression of TMPRSS2, KLK2 and low levels of PSA. However, these expression levels were lower in comparison to differentiated cells.
Previously, a 2 b 1 HI CD133 +VE prostate basal stem cells have been shown to lack AR expression [14,16], and the discrepancies between our data and these studies on CD133 +VE and AR expression may be due to the sensitivity and the specificity of the assay used. In particular, we showed that AR levels were 3-fold lower than differentiated cells, which potentially accounts for the difficulties in detecting the AR in extremely rare prostate progenitor cells using less sensitive approaches such as western blot [31]. A recent report identified a highly conserved site in the second intron of the AR gene that regulates its expression in response to androgen stimulation and withdrawal [32]. Specifically, it was shown that AR binding to this response element decreased AR gene expression by functioning as a transcriptional suppressor at this site and this may be a mechanism to explain why, despite 3-fold lower levels of AR protein, there were similar levels of AR transcripts in a 2 b 1 HI CD133 +VE stem and differentiated cells. Additional explanation for our findings was that in contrast to previous studies we did not culture samples prior to analysis as CD133 expression is altered as a result of in vitro conditions [27]. In particular, AR protein undergoes rapid metabolic turnover in prostate cells and ex vivo culture rapidly leads to low or undetectable levels of AR protein expression [28,33]. These issues associated with measuring CD133 and AR may also explain discrepancies between CD133 +VE cancer stem cell studies where both the presence and absence of AR is reported [18,34]. Having validated AR expression in a 2 b 1 HI CD133 +VE cells, there is a possibility that there are differential AR functions in prostate stem and differentiated populations [35], particularly given that progenitor cells are also indirectly responsive to androgens through paracrine signalling of growth factors from adjacent AR +VE stroma [28,36,37]. Further work exploring this possibility would be of interest and may identify new mechanisms of homeostasis and potential insights into tumourigenesis.
A recent characterisation of prostate stem cells has identified TRA-1-60/CD151/CD166 +VE stem cells which were AR -VE [38]. These cells may represent an acquired phenotype following up-regulation of pluripotent markers seen in advanced cancers [39] that drive de-differentiation [40] into a state more in keeping with embryogenesis (e.g. TRA-1-60 expression) and lacking markers of prostate-specific lineage. However, it is accepted that the TRA-1-60/CD151/CD166 +VE stem cell may have arisen from a cell of origin for cancer that lacks AR, as in our study we did identify a very small fraction (19%) of a 2 b 1 HI CD133 +VE cells lacking AR expression (0.0002% of the total epithelium [14]). This may have clinical relevance where individual tumours may have different cancer stem cell origins, each with their own specific pathobiology requiring tailoring therapies. Further comparative studies are needed to resolve differences between the TRA-1-60/ CD151/CD166 +VE and the a 2 b 1 HI CD133 +VE cancer stem cell models.
In summary, studies of human prostate cancer have demonstrated basal cells, within which the a 2 b 1 HI CD133 +VE stem cell resides, are efficient targets of prostate cancer initiation and that AR expression is required [6,7]. Therefore, the stem cell, which remains the most likely target for transformation, would be expected to have AR expression too. As previously human prostate stem cells were considered to lack AR, the characterisation of AR expression within a 2 b 1 HI CD133 +VE cells offers a resolution to a key paradox about the cell of origin in prostate cancer. Further study of specific AR functions in prostate stem and differentiated cells may highlight novel mechanisms of prostate homeostasis and insights into tumourigenesis.