|Home | About | Journals | Submit | Contact Us | Français|
The CD133hi sub-population of prostate epithelial cells has been demonstrated to possess tumor-initiating capacity consistent with that of the cancer stem cell theory. However, the involvement of oncogenes such as c-myc has not been fully elucidated in the CD133hi sub-population.
We have isolated primary prostate cell strains (IBC-10a) and immortalized them by transfection with hTERT. The in vitro and in vivo tumorigenic capacity of isolated CD133hi and CD133lo cells was evaluated with respect to c-myc expression using specific sense and anti-sense oligonucleotides.
Freshly immortalized cells consisted of <3.3% CD133hi/CD24hi sub-population (SP). ‘Prostaspheres’ generated from single CD133hi cells in the presence of EGF consisted of ~10% CD133hi SPs in 12-21 day cultures. A single Prostasphere generated from single CD133hi cells (6-10 cell stage at day 6 injected i.t) produced dysplastic lesions in NOD-SCID mice (n=4/5). Treatment of Prostaspheres from CD133hi SPs in vitro with c-myc or cyclin D1 anti-sense oligonucleotides totally blocked colony forming ability and growth. Furthermore, treatment of fully formed, 6-day Prostaspheres for 48 hr with c-myc anti-sense significantly reduced c-myc expression and their ability to generate lesions in NOD-SCIDs (n=10 Prostaspheres injected i.t./mouse).
These data demonstrate for the first time that a single CD133hi cell is competent to generate Prostaspheres in vitro and that CD133hi Prostaspheres require c-myc to grow and form dysplastic lesions in vivo.
Normal prostatic epithelium consists of three principal cell types; basal, secretory luminal and neuroendocrine. A stem cell model for prostate organization has been postulated whereby a pluripotent stem cell within the basal layer is the progenitor of the terminally differentiated secretory luminal and neuroendocrine cells (1-3). The differentiation of stem cells is most likely a hierarchical process characterized by the presence of intermediate progenitor basal cell types or IBCs (1,4,5).
Aggressive and poorly differentiated tumors contain both proliferative exocrine cells as well as cells with extensive neuroendocrine differentiation, suggesting that tumor initiating cells could be early IBCs (4,6). Progression to androgen independent disease may result from the up-regulation of anti-apoptotic and growth regulatory genes (i.e. c-myc, cyclin-D1 and bcl2) involved in normal prostate development (7) in these cells, allowing them to bypass the need for androgen and/or androgen receptor for growth. Over expression of c-myc, Bcl-2 (8,9), telomerase (10,11), hepatocyte growth factor receptor (c-met), epidermal growth factor receptor (EGFR) and Her 2/neu (12-15) might protect IBCs from apoptotic stimuli and enable cells to survive in an androgen-deprived environment in vivo. Thus, c-myc, Bcl-2 and telomerase may play critical roles in maintaining the stem cell features of IBCs, such as resistance to apoptosis and immortality, while c-met, Her-2/neu, EGFR and other growth factor receptors (cyclin D1, c-sis) might mediate the proliferation of IBCs (16).
In support of this concept, several groups have found that there is an over expression of basal cell genes by androgen-independent tumors which may reflect the expansion of a pre-existing ‘basal-like’ cancer stem cells from the IBC population (12,17). Moreover, hormone ablation leads to the regression of terminally differentiated androgen-dependent cells and to the selection and/or adaptation of self-renewing androgen-independent stem cells which express a ‘basal-cell like’ phenotype (18). These putative stem cells are maturationally arrested and differentiate incompletely or give rise to androgen-independent neuroendocrine cells (19-21). In sum, IBCs, the tentative source of prostate stem cells and transit amplifying cells, possess the phenotype of androgen-independence as do epithelial cells in advanced prostate cancers (1-3). Therefore, the study of IBCs and the putative stem cell sub-populations may prove critical to understanding prostate carcinogenesis and the development of strategies for preventing and managing prostate cancer.
Several groups have independently isolated a population of cells that are enriched for putative prostatic stem cells (19,22). Isolation of CD133+/α2β1-integrin expressing cells was found to enrich for a population of prostate epithelial cells capable of forming acinar-like structures in xenografts (19). These basal cells generated dysplastic lesions in NOD-SCID mice with morphologic and immunohistochemical evidence of prostate-specific differentiation, consistent with a stem cell origin. Using this same phenotype to isolate cells from human tumors, Collins et al. (23) established that these CD133+/α2β1hi cells had a significantly higher tumorigenic capacity than their negative counterparts isolated from the same tumor. The over expression of integrins (α2β1) and cell surface markers (CD44, CD133, and ABCG2) by these basal cells (i.e. IBCs) suggests that these cells are possible adult cancer stem cells. In this regard, Collins et al. (23) and others (24) further suggest that these cells might accumulate mutations that can result in transformation. Whether they are, in fact, transformed by one or more oncogenes and are, in fact, malignant or give rise to ‘bonifide’ tumors still remains to be shown.
Reports indicate that nuclear over expression of c-myc is prominent in early stages of prostatic-intraepithelial neoplasia, primary carcinomas and metastatic disease, suggesting that c-myc plays a critical role in prostate cancer initiation (25,26). However, in the context of identifying and understanding the formation of tumor-initiating cells, little is known regarding the involvement of c-myc. Felsher and Bishop (27) demonstrated that within hematopoietic cells, a tetracycline regulatory system controlling the sustained expression of c-myc resulted in the formation of acute myeloid leukemia and T-cell lymphomas. Similarly, using a liver cancer model, it was demonstrated that c-myc inactivation reversed the pluripotent capacity of tumors and cells differentiated into normal cellular lineages and tissue structures of the liver. This suggests that tumor cells of the liver retain stem cell properties and that aberrant c-myc expression might result in the malignant expansion of normal liver stem cells (28). However, the role of c-myc may be tissue or stem cell specific. In the epidermis, sustained c-myc expression depletes the epidermal stem cell population (29), and in hematopoietic stem cells, the elimination of c-myc expression results in a loss of ability for these cells to undergo differentiation (30). Arnold and Watt (31) suggest that c-myc might induce stem cells to generate transient amplifying cells which proliferate and differentiate. Thus, the c-myc proto-oncogene may somehow control the balance between stem cell self-renewal and differentiation. Clearly, further work is required to elucidate how c-myc might influence stem cell proliferation and, more importantly, how c-myc might promote stem cell dependent tumorigenesis.
To understand the potential role of c-myc in adult cancer stem cells, we have isolated intermediate basal cell strains (IBC-10a) from human prostate. In this paper, we report that IBC-10a cultures consisted of a putative cancer stem cell, <3.3% CD133hi /CD24hi sub population (SP). We discovered that a single CD133hi cell was capable of generating Prostaspheres in vitro and that a single Prostasphere (6-10 cell stage) was capable of generating a dysplastic lesion in NOD-SCID mice (n=4/5). Moreover, anti-sense ‘knock-down’ of c-myc or cyclin-D1 expression blocked colony forming ability, Prostasphere formation in suspension cultures and, more importantly, blocked Prostaspheres from forming dysplastic lesions in NOD-SCIDS. These data demonstrate for the first time that a single CD133hi cell requires c-myc for ‘asymmetric’ division in vitro and the production of pre-malignant lesions in vivo.
The IBC-10a cell strain was isolated from the right peripheral zone of prostate gland containing normal and Gleason score 6 prostate glands (patient age 56). The tissue was minced and tiny pieces (<0.5 mm dia.) allowed to adhere to tissue culture dishes in a few drops of serum free complete Keratinocyte media (CKM) containing EGF and pituitary extract plus 1% penicillin/streptomycin (Invitrogen Inc, Carlsbad CA) for 4-6 hr. Additional CKM was then added and the epithelial cells were cultured as ‘outgrowths’ from the tissue over 2-4 weeks. The cell strain arising from the cultures was passaged ~3 times and immortalized by transduction with a LXSN-hTERT retroviral vector (courtesy of Johng Rhim, Center for Prostate Disease Research, USUHS, Bethesda, MD) using methods previously described (32). The resulting IBC-10a cultures were maintained in CKM and aliquots of the cells frozen at a passage <10 for future studies. Most studies were carried out with cells at passages 5-10. PC-3ML, LNCaP, CPTX-1532, NPTX-1532 and HEK293 cells were maintain in culture according to previously described protocols of ATCC (Amer. Tissue Culture Consortium, Bethesda, MD). Isolation of primary prostate epithelial cells was done so with the approval from Drexel University College of Medicine Institutional Review Board.
Karyotyping by light microscopy imaging of Giemsa stained chromosomal spreads of the parent IBC-10a cells (passages 10 and 20 (n>10 cells/passage)), the pLenti6.2-GW/EmGFP transfected single cell clones at passage 10 (n=10) and the CD133hi SPs isolated from the parent IBC-10a cells (passage 10) was carried out by Hope Punnett (Cytogenetics Lab., St. Christophers Hospital, Drexel University College of Medicine, PA).
CD133hi SPs were grown as suspension cultures in CKM supplemented with 10 ng/ml EGF in 96 well plates pre-coated with 6 mg/ml Polyhydroxyethyl methacrylate (PolyHEMA; Sigma, St Louis, MO). The Prostaspheres generated from single CD133hi cells grown in suspension cultures were deposited by centrifugation (3000 × g for 10 min) in chambers on glass slides for immunolabeling. Prostaspheres and monolayer cultures were fixed with 3% Paraformaldehyde in PBS for 15 min, then 3% Paraformaldehyde in PBS containing 0.01% Triton X-100 for 10 min. Slides were washed 3 times with PBS, blocked in PBS containing 5% goat serum for 1 hr, and then incubated with primary antibodies, followed by 3 washes of PBS and incubation with secondary goat anti-rabbit antibody tagged with Alexofluor-488 or -568 (Invitrogen Inc., Carlsbad, CA). In some cases, Prostaspheres were double labeled with DAPI (Sigma, St. Louis, MO), or Phalloidin-Alexafluor 488 (Invitrogen Inc., Carlsbad, CA).
Flow cytometric analysis was carried out according to published methods (25) utilizing CD133 (Abcam, Cambridge, MA) and CD24 (BD Bioscience, San Jose, CA) rabbit anti-human antibodies coupled with goat anti-rabbit Alexafluor-488 and Alexafluor-568 tagged secondary antibodies to isolated the CD133hi/CD24hi SPs. Unlabeled and Alexafluor-488 or -568 conjugated IgG isotype controls were performed as part the analysis. Alternatively, a CD133/1 (AC133) magnetic bead cell isolation kit (Miltenyi Biotech Inc., Auburn CA) was used to isolate CD133hi cells.
Colony forming assays were carried out by resuspending CD133hi cells in 0.5% agarose that was then layered on top of 1% agarose in CKM. After 6 days incubation in CKM, the colonies (i.e. generated from single cells) were washed with OptiMEM I Reduced Serum Medium (Invitrogen Inc. Carlsbad, CA) and then overlaid with 1ml OptiMEM containing 100 nmol/L oligonucleotide using Lipofectamine 2000 (Invitrogen Inc. Carlsbad, CA) in OptiMEM. The medium was replaced by CKM and colonies maintained with changes of CKM every 3 days for 15 days. In the experiments with the suspension cultures, Prostaspheres at day 3 were incubated in the liposome/oligonucleotide mixture at 37°C for 24 hr then grown in CKM for 9 days. Alternatively, the Prostaspheres at day 12 were treated with the anti-sense oligonucleotide for 24 hr and collected for MTS cell viability assays or injection in NOD-SCIDS. Cell viability was carried out using the MTS CellTiter 96 Aqueous ONE solution (Promega, Madison, WI), according to the protocol of the manufacturer. c-myc and cyclin D1 anti-sense and sense oligonucleotides were purchased from (BioSource Int., Camarillo, CA), Anti-sense c-myc: 5′-GTTAGCGAAGCTCACGTTGAG-3′; Sense c-myc:5′CTCAACGTGAGCTTCGCT-AAC-3′; Anti-sense Cyclin D1: 5′-CGCUGGAGCCCGUGAAAAATT-3′, Sense Cyclin D1: 5′-TCCGCGCGATAGTACGTA-3′; and Scrambled: 5′CAGGTCTTTCATCT-AGAACGATGCGGG-3′.
CD133hi cells were plated by limited dilution as single cells/well and incubated in CKM as suspension cultures for 6 days. Prostaspheres consisting of ~10 cells/Prostasphere were then treated o.n. with 4 μg/ml c-myc sense or c-myc anti-sense oligonucleotide. The CD133hi Prostaspheres were then injected i.t. in 0.1 ml CKM at 1 Prostasphere/mouse with c-myc sense treated Prostaspheres (n=5 NOD-SCID mice) or 10 Prostaspheres/mouse with c-myc anti-sense treated Prostaspheres (n=10 NOD-SCID mice). In control experiments, mice were injected with CD133lo cells intra-thoracic (i.t.), sub-cutaneously (s.c) or intraperitoneal (i.p.) at titers of 1, 10 and 100 × 103 cells/site (n=5 NOD-SCID mice/cell density) and mice kept for 3 mos. Other controls included injection of NOD-SCID mice s.c. with the parent IBC-10a cells (1 × 106 cells/ml, n=5 mice) and incubation for 3 mos. Male NOD-SCIDs, age 8-10 weeks (~35 gm wt) were used in these studies. Upon sacrifice of the mice after 3 mos, the mice were examined by gross dissection for tumors in different organs and any suspicious nodules prepared for histology. Freshly resected tumors from NOD-SCIDS were fixed with 10% Formalin o.n. and sections stained with H&E or labeled with primary antibodies followed by biotinylated secondary antibody and avidin-conjugated horseradish peroxidase (HRP) from the Vectastain ABC system, following the manufacturer's instructions (Vector Laboratories, Burlingame, CA). The HRP substrate used was diaminobenzidine at 0.8 mg/ml, which was enhanced with addition of nickel chloride and cobalt chloride at a concentration of 0.2% and catalyzed by addition of hydrogen peroxide at 0.03%. Alternatively, the sections were labeled with the Vectastain™ universal ABC (Vector Laboratories Inc. Burlingame, CA) according to the manufacture's protocol, utilizing Vector Red substrates. In some cases, sections were counterstained with Hematoxylin. All animal studies were conducted with the approval of the Drexel University College of Medicine IUCAC.
Each antibody was tested by Western blots on IBC-10a cell extracts to ensure it recognized the correct size protein. Then antibody labeling of human prostate sections was carried out for different dilutions of the antibodies using antigen retrieval techniques (i.e. steaming sections in citrate buffer at different pH's 6-10) to establish the optimal conditions for immunolabeling. Antibodies were purchased for CK5 (rabbit anti-human, clone MK5, Berkeley Antibody Co., Berkeley, CA); c-met (R&D Research, Boston, MA), vimentin (R&D Research), CK18 mouse monoclonal anti-human (Clone DC-10, Millipore, Billerica CA); mouse monoclonal p63 (Clone 4A4, Dako, Carpinteria, CA); AR (Clone 441, Santa Cruz, San Diego, CA), rabbit polyclonal CD133 (clone Abcam, Cambridge, MA), and mouse monoclonal c-myc (Clone 8864, Millipore, Billerica CA).
Statistical analysis was performed either with the Student's t-test versus control with P<0.05 considered as the threshold of significance.
The IBC-10a cells were isolated from low-grade prostate cancer tissue (Gleason Score 6) and maintained in CKM at low passage (<passage 20). The cells grew as monolayers, which formed uniform confluent ‘cobblestone’ patterns with a few enlarged flat cells, and elongated spindle shaped cells interspersed in the cultures. Immunofluoresence labeling of the IBC-10a cells revealed that the cultures were largely heterogeneous in expression of most antigens, but this probably reflects the culture conditions rather than the genotypic properties of the cells. For example, the IBC-10a cells expressed CK5, CK18, p63 and c-met but were negative for AR, and vimentin (supplemental fig. 1a, 1b).
Karyotyping of the cultures at low and high passage (i.e. passage 6 and 37) and karyotyping of CD133hi SPs revealed that the cells were uniformly diploid with several chromosomal abnormalities. The karyotype has an unbalanced (i.e. non-reciprocal) 4;8 translocation with loss of the long arm of 4 (loss of 4q) and short arm of 8 (loss of 8p) as well as an unbalanced 13;19, with loss of proximal 13 long arm and terminal 19 short arm, and two extra copies of 20 (fig. 1A). These results were further validated by spectral karyotyping (data not shown). Overall, although the karyotype appears close to normal it may reflect commonly shared characteristics of IBCs.
Immunolabeling and flow cytometric analysis indicated that <3.3% of the IBC-10a cells at passage 6-20 were CD133hi (fig. 1B-F). We isolated the CD133hi cells, utilizing magnetically labeled CD133 MicroBeads (Miltenyi Biotech Inc. Auburn CA). Growth studies revealed that the CD133hi cells formed tiny 4-5 cell and 6-10 cell Prostaspheres or spheroids by 3 days in suspension cultures (n=26 +/− 6 Prostaspheres/100 cells seeded) in CKM. Normally, these Prostaspheres grew to form 6-10 cell spheroids by 6 days and large compact spheroids by ~21 days. In comparison, although CD133lo cells adhered to culture dishes and grew as ‘cobblestoned’ monolayers, they failed to grow Prostaspheres (n = 2 +/−1 spheroid/100 cells seeded), even after prolonged culture intervals (>30 days). This experiment was repeated 3 times with similar results.
We have attempted to identify putative therapeutic targets that might block the proliferation of CD133hi SPs in vitro (and in vivo). Immunolabeling indicated the CD133hi cells normally expressed c-myc. Colony forming assays were carried out in soft agar (0.5% agarose) where ~3000 CD133hi cells were plated/well in 12 well dishes (fig. 2A). Growth was initiated in CKM for 6 days followed by treatment overnight (o.n.) with either anti-sense (a.s) or sense oligonucleotides targeting c-myc and cyclin D1. Cultures were then maintained for ~15 days in CKM. The data showed that the number of CD133hi colonies > 200 μm dia. were significantly reduced in the presence of c-myc a.s. or cyclin D1 a.s. as compared with the sense oligonucleotides, scrambled oligonucleotide or lipofectamine alone after 15 days (fig. 2A). Additionally, the size of the colonies generated in CKM was reduced to <30 μm dia. in the presence of c-myc a.s. and cyclin D1 a.s. compared with colonies grown in cultures treated with the corresponding sense probes (i.e. where colonies averaged >200 μm dia.). Light microscopy studies confirmed that few or no colonies remained in wells treated with c-myc a.s. oligonucleotides compared with an abundance of colonies in wells treated with c-myc sense and scrambled oligonucleotides (fig. 2B-2D). The colony forming assay indicated that the CD133hi IBC-10a SP were tentatively malignant. Furthermore, we found that single CD133hi cells were also capable of forming Prostaspheres in suspension culture.
Since we were able to generate lesions with CD133hi derived Prostaspheres in NOD-SCIDS we wished to determine whether c-myc a.s. blocked growth of Prostaspheres formed by single CD133hi cells in suspension cultures. CD133hi and CD133lo populations were isolated from IBC-10a cells and were plated by limited dilution into low attachment, PolyHEMA-coated, 96 well plates in order to generate Prostaspheres from single cells. We initially found that treatment of the day 3 Prostaspheres o.n. with c-myc or cyclin D1 a.s. completely blocked subsequent Prostaspheres survival and growth in CKM as compared with sense and scrambled oligonucleotide controls (Table 1). In comparison, in the presence of sense and scrambled oligonucleotides (plus lipofectamine) about ‘50-62’ Prostaspheres (>200 μm dia.) per 100 cells seeded were generated in the presence of CKM. Note that in these experiments, day 3 Prostaspheres were treated o.n. with c-myc a.s. and then cultures were grown in fresh CKM for an additional 9 days. Control experiments showed that the CD133lo cells failed to proliferate or produce Prostaspheres in the presence of CKM. We extended these experiments to examine the effects of c-myc a.s. treatment on large well-formed Prostaspheres generated from single CD133hi cells (i.e. from 12 day cultures in CKM supplemented with 10 ng/ml EGF). Immunolabeling indicated that ~30% of the cells in these 12 day Prostaspheres were CD133hi and the majority were c-myc positive (fig. 3A-3B). However, following treatment of 12 day Prostaspheres with c-myc a.s. oligonucleotides for 24 hr (4 μg/ml), we found that the numbers of c-myc positive cells were reduced to only a few cells in each Prostasphere (fig. 3C). In comparison, the sense and scrambled oligonucleotides had little or no effect on c-myc expression (fig. 3A and fig. 3B). More importantly, the numbers of CD133hi cells (red) were not significantly reduced by c-myc a.s. treatment (fig. 3D) and Phalloidin labeling of the actin networks (green) indicated that Prostaspheres remained intact following c-myc a.s. treatment (fig. 3C). Trypan blue staining indicated that although c-myc sense had zero effect on cell viability (fig. 3B), c-myc a.s. treatment was deleterious to cell survival (fig. 3C-3D). MTS assays designed to measure cell viability confirmed the Trypan blue results and showed that treatment of the CD133hi Prostaspheres with c-myc a.s. o.n. significantly reduced cell viability, whereas the c-myc sense and scrambled oligonucleotides did not influence Prostasphere cell survival (fig. 3E). In the above experiments, Prostaspheres were treated o.n. with 4 μg/ml a.s., sense, or scrambled oligonucleotides.
We extended the in vitro assays to assess whether c-myc sense or c-myc a.s. treatment influenced the ability of CD133hi Prostaspheres to form tumors in NOD-SCID mice. Following intra-thoracic (i.t.) injection of c-myc sense treated CD133hi Prostaspheres (one 6 day Prostasphere/mouse i.t.), large dysplastic lesions formed on the surface of the lungs in 4/5 male NOD-SCID mice after 3 months (fig. 4). The lesions varied in size from 0.5 to >2.5cm dia. Prostaspheres treated with c-myc a.s. (4 μg/ml o.n.), uniformly failed to form dysplastic lesions after 3 months (i.e. ten 6 day Prostaspheres/mouse injected i.t., n=10 NOD-SCID mice). Control experiments showed that the freshly isolated CD133lo cells failed to form dysplastic lesions or tumors in NOD-SCIDs following injection i.t., sub-cutaneously (s.c) or intraperitoneal (i.p.) at titers of 1, 10 and 100 × 103 cells/site (n=5 mice/cell density). Likewise, the parent IBC-10a cells failed to form tumors, dysplastic growths or normal glands in male CB17 SCIDs and male NOD-SCIDs following injection s.c., i.t., and i.p. at low or high titers of ~0.01 to 1 × 106 cells/site (n=5 mice/experiment). Figure 4 shows an example of a dysplastic lesion grown from a single CD133hi Prostasphere. The lesion grew on the surface of the lung. The cells grew as compact masses that were highly vascularized and in some regions the cells formed pseudo-glandular structures (see H&E section, arrows). Immunolabeling revealed that the majority of the cells were CK5 positive, and that a small number were CK18 positive. About 50% of the cells were labeled with AR and p63 antibodies. Finally, a small percentage of the cells near the tumor margins were positive for CD133 and c-myc (fig. 4). Interestingly, the cells were not labeled with CD44v6 antibodies (not shown); indicating these cells may not arise within CD133hi derived lesions. In sum, the immunolabeling suggests that the well-differentiated dysplastic lesions derived from CD133hi SP Prostaspheres consist of both CK5/p63 positive basal cells and CK18/AR positive epithelial cell(s).
Despite advances in detection and treatment of prostate cancer, mortality from this disease remains high because current therapies are limited by the emergence of therapy-resistant cancer cells (2,12). It was, therefore, necessary to identify the clonigenic cells with markers that distinguish these cells from other less-tumorigenic cells. This identification has been accomplished, in part, utilizing the NOD-SCID mouse model. Al-Hajj et al. (33) showed that solid tumors generated from human breast cancer cells contained a CD44hi/CD24lo lineage of cells that gave rise to tumors in NOD-SCID mice, indicating they possessed properties of cancer stem cells. Likewise, in studies of prostate cancer, Collins et al. (19,23) showed that human CD133+/α2β1hi SPs preferentially grew Prostaspheres in suspension cultures and that this SP formed dysplastic lesions following injection in NOD-SCIDS (~1000 cells/sites). Richardson et al. (34) further showed that a small population (approximately 1%) of human prostate basal cells expressed the cell surface marker CD133 and that these cells were restricted to the α2β1hi population. These cells were found to possess a high in vitro proliferative potential and could reconstitute prostatic-like acini in immuno-compromised male nude mice. More recently, Miki et al. (32) have identified human telomerase reverse transcriptase (hTERT)-immortalized primary nonmalignant (RC-165N/hTERT) and malignant (RC-92a/hTERT tumor-derived human prostate epithelial cell lines and shown that they retain stem cell properties with a CD133hi/CD44+/2ß1+/34ßE12/CK18+/p63−/androgen receptor (AR)−/PSA− phenotype. These hTERT immortalized cells readily formed “Prostaspheres” in non-adherent culture systems. The CD133hi cells from these immortalized cell lines had high proliferative potential and were able to differentiate into an AR+ phenotype and form dysplastic lesions in NOD-SCIDS mice. Interestingly, Miki et al (32) required at least 100 CD133hi cells to generate lesions in mice, indicating the CD133hi cells had difficulty surviving and/or required CD133lo ‘helper cells’ to establish lesions. Overall, the studies indicated cancer stem cells and/or tumor initiating cells derived from the basal cell layer were responsible for the development of prostate cancer.
In keeping with published studies, we found that Prostaspheres generated from CD133hi cells readily form dysplastic lesions in NOD-SCIDs. We were unable to obtain lesions following injection of 1-10 CD133hi cells, but we consistently obtained dysplastic lesions following injection intra-thoracic of a single tiny Prostasphere generated from a single CD133hi cell (i.e. at the 6-10 cell stage). We believe, therefore, that the CD133hi cells growing in Prostaspheres might rely on the CD133lo cells to survive and establish lesions in vivo. In comparison, we found that the CD133lo cells did not form Prostaspheres and failed to establish lesions in mice unless injected at high titers (>103 cells/mouse). Little is known about the growth requirements of CD133hi cells in vitro or in vivo. We found that CD133hi cells could generate Prostaspheres of varying size by ~6-10 days and EGF alone or in CKM selectively supported the proliferation of CD133hi cells within the Prostaspheres. These Prostaspheres consisted of ~30% CD133hi cells, suggesting ‘asymmetric divisions’ of the CD133hi cells gave rise to the CD133lo cells. We found that the majority of the cells in Prostaspheres normally expressed c-myc. Moreover, anti-sense experiments further revealed that the growth of the CD133hi Prostaspheres required c-myc and cyclin D1 expression. Knock-down of c-myc expression resulted in a significant loss of cell viability and growth. Even though the residual Prostaspheres still contained CD133hi cells, they were Trypan blue positive and failed to form lesions in NOD-SCIDS. We, therefore, suggest that the CD133hi SPs and/or the CD133lo cells present in Prostaspheres might require c-myc for survival and that c-myc plays a key role in their establishment (or maintenance) of dysplastic lesions in vivo. In support of this conjecture, Okita et al.(35) have recently shown that activation of expression of four genes (i.e. oct4, sox2, klf4 and c-myc) can induce pluripotency in mouse skin cells so that they begin functioning as embryonic stem cells. They and others suggest that c-myc could be a master regulator of cell cycle, since c-myc plays a key role in supporting self-renewal of hematopoietic stem cells as a downstream mediator of Notch and HOXB4 (36,37).
Our observations demonstrate that a single CD133hi cell, but not CD133lo cells, are capable of growth in suspension. Moreover, these CD133hi cells proliferate in suspension to form a tight sphere that consists of both CD133hi and CD133lo cells. Interestingly we observed that only CD133hi derived Prostaspheres, and not individual CD133hi cells, were capable of forming dysplastic lesions in NOD-SCIDS. These results suggest that the 3D architecture and the CD133lo cells present in the Prostaspheres may be critical to supporting growth in vivo. In sum, our data are in agreement with the observations of Richardson et al. (34) and support the contention that CD133hi cells might possess tumor initiating ability. Certainly, they must give rise (i.e. by so called ‘asymmetric’ division) to a variety of differentiated cells that are p63, CK5, CK18 and AR positive in vitro and in vivo. We further conclude that the CD133hi cells may be early stage progenitor cells that require c-myc and cyclin-D1 for proliferation and the development and/or maintenance of dysplastic and/or tumorigenic lesions.
Immunofluoresence labeling of IBC-10a cells (passage 11). Cells were fixed in 3% Paraformaldehyde and permeabilized with 10% Triton-X100 in PBS (pH 7.4). Cells were labeled with CK5, CK18 or p63 antibody, followed by Alexafluor-488 conjugated secondary antibody. Shows the majority of the cells were positive for CK5 and CK18, but only ~30% were positive for p63.
Western blots of crude cell extracts from: (lane 1) PC-3ML; (2) HEK293; (3) LNCaP; (4) CPTX-1532; (5) NPTX-1532 and (6) IBC-10a cells with antibodies specific for vimentin, c-met, androgen receptor (AR), CK18 and CK5. Total protein lysate (10 μg/lane) was resolved on 10% SDS-PAGE. Shows PC-3ML, HEK293, CPTX-1532 and LNCaP cells expressed vimentin but IBC-10a and NPTX-1532 cells did not. All the cells except LNCaP expressed c-met, but only LNCaP expressed AR. CK5 was faintly expressed by PC-3ML, but was not expressed by HEK293 or LNCaP cells. CPTX-1532, NPTX-1532 and IBC-10a cells all expressed CK5. All 6 cell lines expressed CK18.
This work was supported by NIH grant NIH-NCI 232216 to M.E.S.
Disclosures: The authors declare no competing interests.