SEARCH

SEARCH BY CITATION

Keywords:

  • Cancer;
  • Stem cell-microenvironment interactions;
  • Tissue-specific stem cells;
  • Embryonic stem cells;
  • Experimental models

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Normal prostatic epithelium is composed of basal and luminal cells. Prostate cancer can be initiated in both benign basal and luminal stem cells, but because basal cell markers are not expressed in patient tumors, the former result was unexpected. Since the cells of origin of prostate cancer are important therapeutic targets, we sought to provide further proof that basal stem cells have tumorigenic potential. Prostatic basal cells were enriched based on α2β1integrinhi expression and further enriched for stem cells using CD133 in nontumorigenic BPH-1 cells. Human embryonic stem cells (hESCs) were also used as a source of normal stem cells. To test their tumorigenicity, we used two alternate stromal-based approaches; (a) recombination with human cancer-associated fibroblasts (CAFs) or (b) recombination with embryonic stroma (urogenital mesenchyme) and treated host mice with testosterone and 17β-estradiol. Enriched α2β1integrinhi basal cells from BPH-1 cells resulted in malignant tumor formation using both assays of tumorigenicity. Surprisingly, the tumorigenic potential did not reside in the CD133+ stem cells but was consistently observed in the CD133 population. CAFs also failed to induce prostatic tumors from hESCs. These data confirmed that benign human basal cells include cells of origin of prostate cancer and reinforced their importance as therapeutic targets. In addition, our data suggested that the more proliferative CD133 basal cells are more susceptible to tumorigenesis compared to the CD133+-enriched stem cells. These findings challenge the current dogma that normal stem cells and cells of origin of cancer are the same cell type(s). STEM CELLS2012;30:1087–1096


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The glandular epithelium of the normal prostate is divided into basal and luminal compartments. In prostate cancer, this arrangement is disrupted, with an associated loss of cells with a basal cell phenotype. A minor luminal stem cell population was shown to be an efficient target for oncogenic transformation in mice [1], although this was not surprising given that mouse models of prostate cancer are commonly induced by luminal-specific promoters, including probasin and/or prostate-specific antigen (PSA) [2]. More surprisingly, murine and human basal epithelial stem cells expressing high Trop2 and CD49f were also susceptible to malignant transformation [3, 4]. Altogether, these exemplary studies demonstrated that both basal and luminal cell types contain potential cells of origin of prostate cancer [1, 3].

The notion that basal cells can be malignantly transformed is a seminal but controversial finding, because basal cell markers are not generally detected in patient tumor tissues. Rare cells (<1%) that coexpress basal and malignant markers (including CKs 5/14, p63 and AMACR, or TMPRSS2-ERG fusion gene) were identified in a few prostate tumors [5–7], although this low percentage is not within the detection limits of immunodiagnosis for prostate cancer, which is typically classified as p63-negative. Nonetheless, as cells of origin of prostate cancer, basal cells are important therapeutic targets. To provide further substantive proof of concept that human stem cells within the basal population have tumorigenic potential, we used an alternate approach both for selection of basal stem cells themselves and for assessment of tumorigenicity.

CD133 is a cell surface marker commonly used to select for stem cells in many tissues [8], including prostate. In combination with α2β1integrinhi expression, we previously used CD133 to enrich for stem cells from benign tissues that express basal cell markers (CK5+, CK14+, AR, and PSA), have increased proliferative potential, and undergo full prostatic differentiation in vivo [9–11]. Although CD133+ cells also enriched for cancer repopulating stem cells from prostate cancer [5, 12], the oncogenic potential of CD133+-enriched stem cells from benign specimens was not tested, and therefore their potential to act as cells of origin of prostate cancer is unknown. The goal of this study was to test the potential for CD133+ or CD133 basal cells to act as cells of origin of prostate cancer, by assessing the ability of benign epithelial subpopulations to undergo malignant transformation. This approach is in contrast to other studies using prostate cancer cells to test the cancer repopulating or stem cell activity by assessing which malignant subpopulations can regenerate tumors.

Here, we enriched for basal cells from the nonmalignant BPH-1 cell line using selection for α2β1integrinhi cells, and then further enriched for CD133+ stem cells to assess tumorigenicity of benign stem cell populations, by recombining them with human tumor stroma (cancer-associated fibroblasts [CAFs]) or high-dose testosterone and 17β-estradiol (T + E2) treatment. Our results showed that benign basal cells formed tumors by either method, but surprisingly the CD133+-enriched stem cells were resistant to malignant transformation, while the more proliferative CD133 fraction were consistently tumorigenic. Nevertheless, these data confirm that cells from the basal cell layer are indeed tumorigenic but suggest a more complex paradigm, where the potential for oncogenic transformation resides in the more proliferative cells within the basal population, rather than the stem cell subpopulation themselves.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Patient Samples

Prostatic cancer tissues were obtained with patients consent (Human Ethics Research Approvals 34306 at Epworth Hospital, 03-14-04-08 at Cabrini Hospital and RMO 2006/6108-2004000145 at Monash University) from patients undergoing radical prostatectomy. Primary prostate cancer and adjacent nonmalignant tissues (confirmed by a pathologist examining histopathology of adjacent frozen tissue fragments) were obtained from tumors of five patients at radical prostatectomy for isolation of CAFs.

Cell Lines and Stem Cell Enrichment

CAFs were derived from five patients with Gleason 7 prostate cancer, based on protocols published by Olumi et al. [13]. Briefly, prostate tissues were washed with calcium- and magnesium-free phosphate buffered saline (PBS) containing 100 μg/ml gentamycin to remove any debris. Tissue was then cut into 2 mm3 pieces using a scalpel. Tissue was firstly digested in Roswell Park Memorial Institute Medium (RPMI) 1640 containing penicillin/streptomycin, 0.5 μg/ml fungizone, 0.1 mg/ml gentamycin, 25 mM HEPES, 10% fetal calf serum (FCS), 225 U/ml collagenase, and 125 U/ml hyaluronidase for 16–18 hours at 37°C and gentle rotation. Single-cell suspensions were separated from remaining undigested organoids using repeated gravity sedimentation washes. Remaining organoids were further digested in 0.1% trypsin-versene for 30 minutes at 37°C into single cells. Single cells obtained through out protocol were then cultured in RPMI 1640 with 5% FCS and 1 nM testosterone that allows continual growth of only the stromal populations. Each line was confirmed to display properties of CAFs in vitro including growth characteristics, immunomarkers, and RNA expression similar to previous reports [13, 14]. In Supporting Information Figure S1, we show that SFRP-1 gene expression was significantly elevated in CAFs used in this study, compared to patient-matched normal prostatic fibroblasts (NPF)s; stromal lines were used between passages 3-6.

The BPH-1 prostatic epithelial cell line was obtained from American Type Culture Collection (Manassas, VA, http://www.atcc.org/). BPH-1 cells were routinely cultured in a 5% CO2 humidified incubator at 37°C in complete RPMI 1640 with penicillin/streptomycin and 5% FCS and subfractionated using previously published protocols. Briefly, single-cell suspensions were fractionated based on adhesion to type I collagen, with rapidly adherent cells constituting the α2β1integrinhi population [9]. The nonadherent cells (that had not adhered within 10 minutes, to type I collagen) were also recovered during washing. CD133+ cells were selected from the adherent fraction using magnetic-activated cell sorting (MACS) microbeads linked to anti-human CD133, according to the manufacturer's instruction (Miltenyi Biotec Ltd., Victoria, Australia, http://www.miltenyibiotec.com/) [10].

Human embryonic stem cells (hESCs) were supplied by ES Cell International (Melbourne, Victoria, Australia) and Australian Stem Cell Centre (Melbourne, Australia, http://www.stemcellcentre.edu.au/). Experiments were conducted on two cells lines, hES 2 and hES 3, to determine whether genetic sex determination had any impact on the differentiation potential [15]. Briefly, hESCs were cultured on a mitomyocin C-treated (i.e., mitotically inactivated) mouse embryonic fibroblast feeder layer, isolated from day 13.5 postcoitum fetuses of 129/SV mouse strain, in gelatin-coated tissue culture dishes. The culture media consist of Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% fetal bovine serum (FBS), 0.1 mM β-mercaptoethanol, 1% nonessential amino acids, 2 mM glutamine, and penicillin/streptomycin. Seven days after plating, hESC colonies are mechanically cut into transfer pieces using a micropipette and replated on fresh feeder layers or placed in tissue recombinants with prostatic stromal cells.

Flow Cytometry

hESCs were dissociated using TrypLE Select (Invitrogen, Grand Island, NY, http://www.invitrogen.com) and resuspended at 5 × 106 cells per milliliter in PBS-EDTA plus 1% bovine serum albumin (BSA) buffer. Aliquots of the cell suspension (50 μl) were prepared as (a) unstained control, (b) propidium iodide (PI) only, (c) CD133/1-APC (Miltenyi Biotec), and (d) mouse IgG1-APC isotype control. Antibodies were diluted 1:10 according to the manufacture's protocol and incubated for 30 minutes at 4°C. Cells were resuspended in 400 μl of PBS with 1 μg/ml PI, except for the unstained control, which was resuspended in PBS alone. Cells were analyzed at Monash University FlowCore on a LSRII flow cytometer (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com/)), gating for live, single cells. The isotype control was used to determine positive CD133 staining.

Animals

All animal handling techniques and procedures were conducted in accordance with National Health and Medical Research Council guidelines for the care and use of Laboratory Animal Act according to the Animal Experimentation Ethics Committee at Monash University (Approval Numbers: MMCA/2007/04 and MMCA/2008/33). Sprague-Dawley day 0 and 1 rat pups (for seminal vesicle mesenchyme; SVM) or E17 rat fetuses (for urogenital mesenchyme; UGM) were obtained from Monash University Central Animal Services (Clayton, Australia) and culled. SVs or the urogenital sinus (UGS) were dissected out of the pups or E17 fetuses, respectively, and left in DMEM F-12 with fungizone until trypsinization. The mesenchyme was dissociated from the epithelium by digestion in 1% trypsin for 70 (SV) or 100 minutes (UGS) and separated into mesenchyme and epithelial components by dissection. SVMs and UGMs were transferred to fresh DMEM with 20% FCS where they were stored on ice until recombinant tissues were made. Severe combined immune-deficient (SCID) 6–8-week-old mice were obtained from Animal Resources Centre (Canningvale, Australia) for grafting of recombinants.

Tissue Recombination

Tissue recombination involves mixing of stromal and epithelial cell populations that are combined and then grafted into immune-deficient host mice, as previously described [13, 16-18]. For this study, tissue recombinants were generated from CAFs, SVM, or UGM (∼2.5 × 105 cells), mixed with the various epithelial populations both unsorted and sorted CD133+ and CD133 BPH-1 cells, or hESCs. While standard tissue recombinants contained 1 × 105 unsorted epithelial cells, experiments comparing sorted BPH-1 cells were restricted to 500 cells per graft since the yield of CD133+ cells was so small (0.16% of total cell population). For all recombinants, mixtures of stromal and epithelial cells were set in a collagen gel of 10–50 μl of type I rat tail collagen. Each recombinant gel was incubated overnight in a 5% CO2 humidified incubator at 37°C in complete RPMI 1640 with 1 nM testosterone. The following day, recombinant gels were surgically implanted under the renal capsule of SCID mice, typically involving 3-4 grafts on each kidney per mouse. During surgery, SCID hosts were supplemented with a subcutaneous hormone implants; all mice received 5 mm testosterone (T) implants, except for T + E2 experiment hosts that were supplemented with a 10 mm T plus 5 mm 17β-estradiol (E2) implants for the treated hosts and 5 mm T plus empty 10 mm implants for the untreated hosts. Recombinant tissues were grown in host mice for a period of 8–16 weeks. At the time of harvest, wet weights of tissue recombinants were recorded and tissues were fixed in 10% formalin (for tissue containing BPH-1 cells) or Bouin's fixative and embedded in paraffin wax for histological analysis.

Immunohistochemistry

In order to measure tumorigenicity of the CAF and T + E2 recombinant tissues, we used various histological and immunohistochemical techniques. All tissues were stained by hematoxylin and eosin (H&E) for pathological examination. Immunohistochemistry was performed using the Leica BOND-MAX automated system (Leica Microsystems Pty Ltd, Mount Waverly, Australia, http://www.leica-microsystems.com/), according to manufacturer's instructions. Antibodies to α-actin and androgen receptor (AR) were purchased from Sigma Aldrich (St. Louis, MO, http://www.sigmaaldrich.com/; 1A4, polyclonal; 1/2000, 1/500); to CK8/18 from Novocastra (Leica Microsystems; NCL-L-5D3; 1/1600); to E-Cadherin from BD Biosciences (36/E-Cadherin; 1/2000); to p63 and SV40 from Santa Cruz (Santa Cruz, CA, http://www.scbt.com/; P64 4A4, Pab101; 1/800, 1/100); and to PSA from Dako (Carpinteria, CA, http://www.dako.com; A0562; 1/3000). Slides stained for AR had pretreatment with Bond TM Epitope Retrieval 1 and slides stained for CK8/18, E-Cadherin, p63, and SV40 had pretreatment with Bond TM Epitope Retrieval 2. All antibodies were incubated for 15 minutes, except SV40 that was incubated for 60 minutes. Detection was achieved with the Bond Refine Detection Kit except for CK8/18 to which Bond Refine Red Detection Kit was applied.

Assessment of Tumor Incidence

In order to assess whether tissue recombinants contained tumors, we established a three-point criteria based on observation of SV40 immunohistochemistry and H&E staining. Criteria for CAF recombinants included assessment of (a) squamous differentiation, (b) presence of atypical nuclei, and (b) presence of undefined glandular borders. Criteria for the hormonal carcinogenesis model included assessment of (a) presence of undefined glandular borders, (b) presence of atypical nuclei, and (c) loss of α-actin organization. Tumors were defined as having all three criteria and ≤2 features were classified as not having tumor. A forth criteria, being the invasion of epithelial cells into the stroma and/or toward the renal parenchyma, was noted to be present in aggressive tumors but was not a requirement of tumorigenicity. In order to estimate the epithelial mass (measured as percentage epithelium per graft) in CAF tissue recombinants, we performed stereology using the new-CAST component (version 2.14; Visiopharm, Hoersholm, Denmark, http://www.visiopharm.com/) of Visiopharm Integrator System (version 2.16.1.0; Visiopharm) [19]. Briefly, every 20th section of grafted tissues was stained with the SV40 antibody (see above) and underwent uniform systematic random sampling. Sections were mapped at ×10 magnification to define tissue boundaries. At ×40 magnification, a grid of 25 evenly distributed points was overlaid on each counting frame. Random sampling occurred in 25% of each tissue section and for each counting frame, the 25 points on the frame was counted as either being positive or negative for SV40 structures. For each tissue, the percentage of points positive for SV40 was calculated and expressed as percentage epithelium per graft. Statistical analysis was conducted using either a one-way ANOVA and Tukey's post hoc test or a Student's t test where significance of p < .05 is noted as *.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Selection of Epithelial Stem Cells

Although prostate cancer can arise from luminal (stem) cells [1, 2], the concept that basal-enriched stem cells can also be cells of origin of prostate cancer is new and controversial [3, 4]. In order to confirm that benign basal cells can be malignantly transformed, we used cells from the initiated but nontumorigenic BPH-1 epithelial cell line [20] that form tumors in recombination with tumor stroma (CAFs) [13]. Different to Goldstein et al. [3], we used an alternate selection protocol to enrich for basal cells based on rapid adhesion to collagen. This is an established protocol used in the field of prostate cancer research, which has been shown to select for basal-enriched α2β1integrinhi cells [9]. We then further subfractionated α2β1integrinhi cells for CD133+ prostatic stem cells [10, 12]. Of the starting cell population, 77.39% was the α2β1integrinlo luminal cell fraction; the remaining 22.61% was enriched for basal cells. Of this latter cell population, ∼0.16% was CD133+ and ∼22.45% was CD133; the CD133 was used as a control for the CD133+ population as well as the unsorted BPH-1 cells. As a second source of stem cells, we used hESCs; by definition, these are CD133+ and are proven to be pluripotent [21, 22].

Stromal-Mediated Tumor Assays

Establishment of CAF-Induced Tumors from BPH-1 and hESCs

Previously, prostatic CAFs in tissue recombination with parental BPH-1 cells formed tumors with squamous differentiation that were significantly increased in size, displayed abnormal nuclear architecture, and showed invasion of BPH-1 cells into the stromal and renal parenchyma [13]. In our study, we observed that tumor formation did not reliably correlate with graft weight. Therefore, we established an unbiased method to assess the incidence of tumors in CAF + BPH-1 recombinants using a three-point scoring system, including (a) expanded epithelial mass and squamous differentiation, (b) atypical nuclei, and (c) undefined glandular boarders (Table 1). A forth criteria of invasion into the local parenchyma was recorded but was not essential to meet the requirements of tumor formation (Table 1). Using this criteria, we determined that CAFs induced tumors in 100% of grafts (n = 10; Fig. 1A), while normal mesenchyme (rat SVM) was unable to induce tumors (0%; n = 6; Fig. 1B).

thumbnail image

Figure 1. CAFs induced malignancy in α2β1integrinhi/CD133 BPH-1 basal cells. BPH-1 cells recombined with prostatic CAFs derived from a single patient line or seminal vesicle mesenchyme (SVM) and subrenally grafted into severe combined immune-deficient mice for 8–12 weeks. (A--D): Photomicrographs showing immunolocalization of SV40 to identify BPH-1 cells, hematoxylin and eosin staining to determine structural morphology. (A): Grafts of unsorted BPH-1 cells (1 × 105 cells per grafts) combined with CAFs formed tumors in 100% of grafts; invasive tumor cells (black arrow) and abnormal nuclear architecture (white arrows). (B): Grafts of unsorted BPH-1 cells (1 × 105 cells per grafts) combined with mouse SVM formed benign cord structures with defined borders and normal nuclear architecture. (C): Grafts of CD133+ BPH-1 cells (500 cells per graft) predominantly formed benign cords with normal nuclear architecture observed. (D): Grafts of CD133 BPH-1 cells (500 cells per graft) formed tumors with atypical nuclei (white arrow). (E): Weight (mg) and epithelial mass (percentage epithelium per graft; determined using stereological analysis of SV40-positive cells) of CAF + BPH-1 cell grafts. Tumor formation did not correlate with graft weight p = .8, while there was a positive correlation between tumor incidence and epithelial mass. Values represented as mean ± SEM; n = 4–10 per group; **, p < .01 and ***, p > .001 using one-way ANOVA and Tukey's post hoc test; Scale bar = 1,250 μm (A--D left panel); 125 μm (A, B midpanel); and 50 μm (A--D right panel). Abbreviations: CAF, cancer-associated fibroblast; SVM, seminal vesicle mesenchyme.

Download figure to PowerPoint

Table 1. Summary of tumor incidence measured by point criteria in grafts in CAF-induced model
  1. Abbreviation: CAF, cancer-associated fibroblast.

inline image

Assessment of relative tumorigenicity of unsorted and sorted BPH-1 cells shows CD133 cells are most tumorigenic.Using the three-point criteria, we quantified the relative tumorigenic potential of the subpopulations of BPH-1 cells. Using equal cell numbers, we demonstrated that unsorted BPH-1 cells showed the lowest efficiency of tumor formation (0%; n = 10) and the CD133+ cells were more tumorigenic (25%; n = 4); however, CD133 recombinants most consistently formed tumors (100%; n = 7; Table 1). Histologically, CD133+ grafts showed controlled differentiation and defined glandular borders (Fig. 1C), while CD133 tumors showed large squamous mass, atypical nuclei, undefined borders, and invasion of a proportion of grafts (Fig. 1D).

Although tumors were not significantly larger in weight (Fig. 1E), we determined the percentage of epithelial mass using unbiased stereological quantitation throughout the graft. Epithelial mass was smallest in CAF + BPH-1 grafts, and a significant increase in epithelial constituent was observed in CAF + CD133+ recombinants. However, CAF + CD133 grafts consistently formed the largest epithelial mass per graft, consistent with tumor formation (p < .05; Fig. 1E) This pattern was consistent across multiple CAF lines generated from individual patient specimens; regardless of the aggressiveness of each CAF line, the incidence of tumors was always greatest in the CD133 cells (Supporting Information Fig. S2). Overall, these results show that the tumorigenic potential of the initiated basal-enriched BPH-1 cells was greatest in the CD133 subpopulation.

Assessment of CAF-induced tumorigenicity of hESCs. Given that we were surprised that the CD133 BPH-1 cells were more tumorigenic than unsorted or CD133+ cells in recombination with CAFs, we sought to further investigate the potential for normal stem cells to form tumors using hESCs that are without any genetic variations or evidence of initiation. While previous studies showed hESCs are predominantly CD133+ [21, 22], we used fluorescence-activated cell sorting analysis to confirm this and showed that the hES 3 cells used in this study were ∼92.8% CD133+ (Supporting Information Fig. S3). Previously, we demonstrated that normal stroma directed prostatic epithelial differentiation [16], and now we tested whether recombination of CAFs with normal pluripotent hESCs harbored cells of origin of prostate cancer. When grafted alone under the kidney capsule, hESCs underwent spontaneous differentiation into all three germ lineages, forming teratomas, consisting of a wide range of structures, including rare glandular epithelia structures (Fig. 2A). In comparison, grafts of CAF + hESCs showed controlled teratoma formation and consisted of predominantly glandular structures (Fig. 2A). Using cytokeratin 8/18 and p63 expression, we identified human-derived epithelial glands; CAF + hESC grafts contained polarized ducts containing basal and luminal cells, while hESC-only grafts were not differentiated (Fig. 2B). Expression of smooth muscle α-actin showed organization around epithelial ducts in CAF + hESC grafts compared to a disorganized stroma in hESC only grafts (Fig. 2C). While CAFs controlled teratoma formation of hESCs and consistently induced AR-positive epithelial differentiation (Fig. 2D), PSA was not detectable in hESC-derived glands, perhaps reflecting the inability of CAFs to induce differentiation to a functional secretory epithelium (Fig. 2E). Of note, one of the three CAF lines also induced a predominantly mesoderm phenotype, containing bone and cartilage adjacent to epithelial structures (Supporting Information Fig. S4). Importantly, (prostatic) malignancy was not evident in any grafts examined (Fig. 2). Therefore, although CAFs appeared to control differentiation of hESCs to an extent, these data failed to show induction of tumorigenicity of hESCs by CAFs but also failed to show complete prostatic differentiation resulting in secretory epithelial cells.

thumbnail image

Figure 2. hESCs were resistant to CAF-induced malignancy. Representative photomicrographs of hESCs grafted alone or in combination with up to three individual CAF lines under the renal capsule of severe combined immune-deficient mice for 8–12 weeks. Images are of (A) hematoxylin and eosin staining, (B) coimmunolocalization of p63 (basal cell marker; brown) and CK8/18 (luminal cell marker; pink), (C) immunolocalization of smooth muscle α-actin organization, (D) AR and, (E) PSA. hESCs grafted alone formed teratomas containing all three germ lineages; primitive glandular structures (black arrows) in these teratomas failed to express organized basal and luminal cell compartments, AR or PSA with absence of α-actin organization around the glands. In comparison, CAF + hESC grafts contained glandular structures (black arrows) that expressed some basal and luminal organization, AR, α-actin organization around glands but not PSA; no evidence of malignancy was detected in any graft (determined by pathological evaluation). Scale bar = 500 μm (A) and 125 μm (B--E). Abbreviations: AR, androgen receptor; CAF, cancer-associated fibroblast; hESC, human embryonic stem cell; PSA, prostate-specific antigen.

Download figure to PowerPoint

Confirmation of Tumor Formation in CD133 BPH-1 Cells Using Hormonal Carcinogenesis

Although our data using CAF tissue recombinants supported existing evidence that basal-enriched cells act as cells of origin for prostate cancer, the finding that the CD133 fraction was more highly tumorigenic (and normal stem cells were less tumorigenic) was unexpected. To clarify and confirm this finding, we applied an alternative oncogenic stimulus to further investigate the tumorigenic potential of basal-enriched CD133+ stem cells, using hormonal carcinogenesis, by administration of high doses of T + E2 [23]. Initially, we generated tissue recombinants using unsorted or CD133-enriched BPH-1 cells with inductive prostatic UGM (UGM + BPH-1 cells) resulting in both the CD133+ and CD133 fractions forming benign glandular structures (Supporting Information Fig. S5).

Again, we established a three-point scoring criteria to determine the incidence of tumor formation in grafts. UGM + BPH-1 (T-only control) grafts rarely displayed any malignant features (0% of grafts formed tumors; n = 6; Table 2) and consistently formed organized cords with normal nuclear architecture, defined glandular borders surrounded by organized bands of smooth muscle α-actin and no evidence of invasive epithelial cells (Fig. 3A). In contrast, similar to previous reports, UGM + BPH-1(plus T + E2) grafts consistently underwent malignant transformation (100% of grafts formed tumors; n = 7; Table 2), evident by presence of undifferentiated squamous carcinoma, defined by (a) loss of defined borders, (b) atypical nuclei, and (c) α-actin expression that was unorganized or lost in tumor regions (Fig. 3B). We also observed invasion of epithelial cells into the surrounding stroma in aggressive grafts, although this hallmark was not essential to be defined as tumorigenic (Fig. 3B).

thumbnail image

Figure 3. α2β1integrinhi/CD133 BPH-1 basal cells underwent malignant transformation by hormonal carcinogenesis. (A--D): Photomicrographs showing immunolocalization of SV40 to identify BPH-1 cells, hematoxylin and eosin staining to determine structural morphology, and immune localization of smooth muscle α-actin to visualize stromal organization. (A): urogenital mesenchyme (UGM) + BPH-1 (T-only control) grafts formed organized cords, which normal nuclei and organized bands of α-actin surrounding cords. (B): UGM + unsorted BPH-1 (T+E2) grafts formed tumors with undefined borders, invasion (black arrow), abnormal nuclear architecture (white arrow), and loss of α-actin organization. (C): UGM + CD133+ BPH-1 (T+E2) grafts formed benign cords with normal nuclei architecture surrounded by bands of α-actin positive smooth muscle. (D): UGM + CD133 BPH-1 (T+E2) grafts contained malignant foci in 43% of grafts involving abnormal nuclei (white arrows), loss of α-actin organization, and local invasion (black arrow). Scale bar = 500 μm (A--D left panel); 50 μm (A--D mid panel); and 250 μm (A--D right panel).

Download figure to PowerPoint

Table 2. Summary of tumor incidence measured by point criteria in grafts in hormonal carcinogenesis model (high dose T + E2)
inline image

When we applied the T + E2 hormonal carcinogenesis to basal-enriched BPH-1 cells, no tumors were observed in UGM + CD133+ (T + E2) grafts (0% formed tumors; n = 6), whereas UGM + CD133 (T + E2) grafts showed tumors that met all the malignant criteria, in 43% of grafts (n = 7) (Table 2), including presence of undifferentiated squamous carcinoma and atypical nuclei shown morphologically in Figure 3C and 3D. Overall, these data were consistent with the CAF-induced model, where basal-enriched CD133+ cells appear to be resistant to tumorigenesis, while the CD133 fraction has greater potential to form tumors.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The cells of origin of prostate cancer are important because of their potential as therapeutic targets [24]. The basal epithelial cell compartment contains normal prostatic stem cells [9, 25, 26], but the finding that CD49fhi/Trop2hi basal cell populations also include cells of origin of prostate cancer was a seminal and surprising discovery [3], because prostate cancer is itself diagnosed by a loss of basal cell markers. This study provides further independent proof of concept that basal cells are cells of origin of cancer, by exploiting a different stem cell selection approach (α2β1integrin and CD133) and two different stromally mediated methods to induce tumorigenicity in benign cells (recombination with CAF or T + E2 treatment), further substantiating the pivotal findings of Goldstein and coworkers.

However, our data raise additional questions about the specific cell type within the basal cell population that is most susceptible to malignant transformation. CD133+ stem cells in human prostate tissues are growth quiescent, with long-term proliferative capacity and a high efficiency of establishing colonies in vitro and in vivo [10, 12, 27]. Primary human CD133 basal cells are proliferative and also have the ability to form colonies, but they form fewer colonies over a shorter time period [10]. Hence, the prostatic CD133 population is postulated to contain transit-amplifying cells, which are defined as fast cycling and proliferative cells [28, 29] (Fig. 4A). However, in BPH-1 cells, the colony-/sphere-forming capacity resides predominantly within the CD133 fraction [30]. The results of our study show CD133 BPH-1 cells were more susceptible to tumor formation, compared to the CD133+ stem cells, and may implicate transient-amplifying cells as cells of origin of prostate cancer in the basal epithelial compartment (Fig. 4A).

thumbnail image

Figure 4. Schematic overview. (A): Normal stem cells and cells of origin of prostate cancer reside within the basal cell compartment of prostatic epithelium, isolated by expression of CD49fhi/Trop2hi. Herein, our study provides further independent proof of concept that basal cells are cells of origin of cancer, by using an entirely independent stem cell selection approach (α2β1integrin and CD133). While the basal-enriched CD133+ stem cells can form colonies and regenerate prostatic tissues in vitro and in vivo, they were resistant to tumorigenesis by CAF or hormonal induced methods, while the highly proliferative basal-enriched CD133 more readily underwent tumorigenesis using these approaches. (B): Our data suggest a new paradigm where the basal-enriched CD133 fraction, not the CD133+ cells, show greater potential to act as cells of origin for prostate cancer by stromal-based assays. Hence, normal stem cells and cells of origin of cancer, within the basal compartment, are not necessarily the same cells, and the latter may involve rapidly proliferating or transient-amplifying cells within the basal cell fraction. This is despite the fact that in human prostate cancer, it is well established that CD133 also isolates cells with cancer repopulating activity that are therapeutic targets.

Download figure to PowerPoint

The fact that CD133 cells are more susceptible to malignancy does not negate the proof that CD133+ identifies normal prostatic stem cells that are responsible for normal tissue replenishment [10, 27] (Fig. 4B). Nor does it disagree with the fact that CD133+ isolates cancer repopulating cells within the tumor itself [12, 27, 31]. It does, however, raise the possibility that normal stem cells and cells of origin of cancer, within the basal compartment, are not necessarily the same cells, and the latter may involve rapidly proliferating or transient-amplifying cells within the basal cell fraction, as was originally postulated by Vander Griend et al. [27]. Such phenotypic plasticity was also documented in leukemia, where a reserve of stem cells could be derived from an oncogenetically activated transient-amplifying cell [32].

Most importantly, the finding that cells of origin of prostate cancer reside in the basal compartment is consistent with findings in human and murine tissues by Witte and coinvestigators who used high expression of Trop2 and CD49f to enrich for prostatic basal cells [3, 4, 33]. In addition to this, Wang et al. showed luminal stem cells (CARNs) are also a cell of origin of prostate cancer in mouse prostate [1]. While the existence of CARNs and their susceptibility to malignancy in human prostate is yet to be discerned, it is plausible that prostate cancer may arise from multiple benign cell types. Such a paradigm is proposed in the breast cancer where different cell types can give rise to genetically distinct subtypes of tumors that have vastly different prognosis and treatments [34], although we do not have subclassification of prostate cancer types to correlate with cells of origin at present.

The observation that the CD133+ fraction within prostatic basal cells was largely resistant to tumorigenesis was surprising but was further supported using hESCs. hESCs form teratomas when cultured alone, but in recombination with normal stroma or embryonic mouse mesenchyme, underwent prostatic epithelial differentiation [16]. However, with CAFs, no malignant tissues were detected and unlike inductive mesenchyme, human prostatic CAFs failed to induce fully differentiated functional prostatic epithelium, since glandular structures that were AR-positive failed to secrete PSA protein or contain the full repertoire of epithelial cells. Therefore, it is evident that normal stroma and tumor stroma are not equipotent, and although stromal signaling in prostate cancer is reminiscent of normal developmental programs, there must be some fundamental differences [35]. Nonetheless, recombination with prostatic CAFs, which was sufficient to induce tumors in BPH-1 cells, was not able to induce a malignant phenotype in hESCs.

While other investigators have studied the direct oncogenic activation of epithelial cells, we chose to apply two different stromal-based assays to indirectly test the effects on subpopulations of prostatic epithelia. First, CAFs derived from primary prostate cancer specimens show distinct genotypic and phenotypic differences compared to their nonmalignant matched controls and have been studied extensively in this BPH-1 cell assay [13, 14, 36, 37]. The second model of hormonal carcinogenesis is based on administration of high doses of testosterone in combination with 17β-estradiol. It is noteworthy that although the prostate is exquisitely sensitive to androgens, administration of high doses of T alone in this model is insufficient to drive tumorigenesis. Instead, estrogens are required to act in synergy with androgens to induce tumor formation [38], mediated through estrogen receptor α expressed in the stromal cells [23]. Both of these approaches highlight the active role played by stroma in the process of tumorigenesis in the prostate.

BPH-1 cells have been activated with SV40 T antigen, which results in genetic and epigenetic instability and significant clonal variation [20, 39]. However, the “initiation events” were the basis for their selection as the epithelial source for these studies, since this instability facilitates induction of malignancy in response to CAFs and/or hormonal carcinogenesis; normal epithelial cells from primary specimens are not susceptible to carcinogenesis using these assays [13, 38]. While there are key differences between BPH-1 cells and primary epithelial cells, including the methylation of the CD133, with silencing of up to 50% of BPH-1 cells [39], our data are supported by several previous studies in mice [4] and in primary human cells [3] showing that basal cells are a cell of origin of prostate cancer. In our study, we have extended these initial findings to further subfractionate the basal population and show the CD133 cells are susceptible to tumorigenesis. Although translation to clinical specimens will be important, this concept is supported by Vander Griend et al. [27] based on their studies on CD133 in primary human cells. Hence, our data highlight the potential for tumor formation in CD133− BPH-1 cells using two independent assays, suggesting the loss of CD133 may be a key event in the transition to malignancy.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Overall, our data confer with previous reports that basal cells can act as cells of origin for prostate cancer. These findings do not exclude the possibility that multiple prostate epithelial cells may have cell of origin potential, including luminal cells, which may also be therapeutic targets. Herein, we provide new data that indicates the CD133 fraction within the basal cell population is more highly susceptible to malignant transformation than the CD133+ stem cells, suggesting that there are subfractions of basal cells with differential tumor-forming potential, and future studies should take this information into account. Finally, we used two stromal-based methods of inducing tumorigenicity, and it is possible that some cell types, or subfractions within cell types, may be uniquely susceptible to different oncogenic stimulants. Understanding the fundamental biology of the origins of prostate cancer will underpin the development of novel preventative or therapeutic strategies for clinical development in the future.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We thank Birunthi Niranjan, Hong Wang, Tameeka Hill, Michelle Richards, and Mitchell Lawrence for expert technical assistance, Stuart Ellem for graphical assistance, Ross Snow and Simon Hayward for providing clinical specimens or cell lines, Alan Trounson and Martin Pera for providing hESCs, and Gerald Cunha and Will Ricke for intellectual input. This research was funded by the Peter and Lyndy White Foundation, grants from U.S. Department of Defense (PC073444), and fellowships from U.S. Department of Defense (R.A.T.; PC030097), Prostate Cancer Foundation of Australia (Movember Young Investigator Grant to R.A.T.), and NH&MRC (G.P.R. #43796; R.A.T. #284395).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
SC_11-1179_sm_supplFigure1.tif597KSupplementary Figure 1: Relative expression of SFRP-1 in CAF and NPF lines used in this study. Data represent patient-matched CAF and NPF expression (n=4) as fold change relative to BPH stromal cells collected from n=4 transurethral resection of the prostate (TURP) specimens. Mean ± SEM are represented by box and whisker plot including minimum and maximum values. * p>0.05, as analyzed by paired t test.
SC_11-1179_sm_supplFigure2.pdf785KSupplementary Figure 2: CAFs induced malignancy in α2β1integrinhi/CD133- BPH-1 basal cells regardless of variability in tumor potential across patient samples. Using a CAF line derived from a different patient that was proven to be less tumorigenic in recombination with unsorted (1x105) BPH-1 cells (80% tumor formation; A), we tested the effect on sorted CD133+ or CD133- BPH-1 cells (500 cells per graft) in tissue recombinants sub-renally grafted into SCID mice for 8-12 weeks. Photomicrographs showing immunolocalization of SV40 to identify BPH-1 cells, hematoxylin and eosin staining to determine structural morphology showed small benign cord structures with normal nuclear architecture (B), compared to CAF + CD133- grafts that showed larger squamous mass with abundant abnormal nuclei (C). Epithelial mass within grafts was determined using stereological analysis of SV40-positive cells (D); data were not significantly different between CAF + CD133+ and CAF + CD133- BPH-1grafts. (E): Using the 3-point criteria to define malignancy (and invasive BPH-1 to highlight aggressive tumors) we showed that in the presence of CAFs, only CD133- cells showed any tendency to form tumors similar to unsorted BPH-1 controls. Values represented as mean ± SEM; n=4-5 per group. Scale bar = 500μm (A-C left panel); 125μm (A mid panel); 50μm (A-C right panel).
SC_11-1179_sm_supplFigure3.pdf341KSupplementary Figure 3: Expression of CD133 in human embryonic stem (hES) cells. A representative scatter plot of CD133 immunoreactivity in primary culture of hES 3 cells showing single, live (propidium iodide negative) cells that were gated based on mouse IgG1-APC isotype control. 92.8% of hES 3 cells expressed CD133 using the CD133/1-APC antibody (Miltenyi Biotec).
SC_11-1179_sm_supplFigure4.pdf399KSupplementary Figure 4: Representative images obtained from one individual CAF line which consistently induced predominately mesoderm derived structures from hESCs. (A) Photomicrographs of hematoxylin and eosin staining in CAF + hESCs with this particular line. Whilst glandular structures were observed (arrow; A), the predominant differentiation of hESCs was into bone and cartilage morphology (indicated by *). (B) Photomicrographs showing co-immunolocalization of androgen receptor (AR) and prostate specific antigen (PSA). Similar to other lines, glands expressed AR but not PSA, neither of which were detected in the adjacent mesodermal structures. No evidence of malignancy was present in these grafts. Scale bar = 125μm except 1250μm in upper left panel of A.
SC_11-1179_sm_supplFigure5.pdf700KSupplementary Figure 5: Both CD133+ and CD133- BPH-1 cells survive in vivo when grafted with rat mesenchyme. (A-B): Photomicrographs showing immunolocalization of SV40 to identify BPH-1 cells, hematoxylin and eosin staining to determine benign nuclear morphology. (A): Benign cords with normal nuclear architecture from 500 CD133+ (top panels) and CD133- (lower panels) BPH-1 cells grafted with rat seminal vesicle mesenchyme in SCID hosts for 4 weeks. (B): Benign cords with normal nuclear architecture from 500 CD133+ (top panels) and CD133- (lower panels) BPH-1 cells grafted with rat urogenital sinus mesenchyme in SCID hosts for 16 weeks. Scale bar = 500μm (A-B left panels); 50μm (A-B right panels).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.