We have shown previously that Pten deletion leads to the expansion of subset of prostate cancer cells positive for CK5 and p63. Although this subpopulation may be involved in tumor initiation or progression, studies to date have not functionally validated this hypothesis. Using in vitro sphere-forming assay and in vivo prostate reconstitution assay, we show here the presence of a tumor-initiating subpopulation in the Pten prostate cancer mouse model. Specifically, we show that the Lin−Sca-1+CD49fhigh (LSC) subpopulation overlaps with CK5+;p63+ cells and is significantly increased during prostate cancer initiation and progression and after castration. Mutant spheres mimic the structural organization of the epithelial compartment in the Pten-null primary tumor. Sorted LSC cells from either Pten-null spheres or primary tumors are able to regenerate prostate epithelial structure with cancerous morphology, closely mimicking that of primary cancers. Therefore, the LSC subpopulation is capable of initiating a cancerous phenotype that recapitulates the pathology seen in the primary lesions of the Pten mutant prostate model. [Cancer Res 2009;69(22):8555–62]
- prostate cancer
- stem cell
Prostate cancer is the most common male malignancy and a leading cause of death in men in the western world (1). Although hormone ablation therapy is the typical mode of treatment for progressive disease, this therapy frequently fails when the disease advances to become castrate resistance. One theory accounting for the initiation and progression of prostate cancer as well as castration resistance is the presence of a rare subpopulation of transformed stem cells, often called cancer stem cells.
The presence of normal stem cells in the rodent prostate gland is well supported by androgen cycling experiments, resulting in continuous depletion and reconstitution of the prostatic epithelium (2, 3). The murine prostate epithelial compartment consists of p63/CD49/CK5+ basal and CK8+ luminal epithelial cells (4) as well as Syn/ChromA+ neuroendocrine cells (5). These cell types differ in their proliferation/differentiation potentials and their response to androgen ablation.
Although the cytosolic markers are critical in identifying different cell types in situ in the prostatic epithelial compartment, the lack of cell surface markers for prospective cell isolation has hampered the identification and functional tests for stem/progenitor cells. Through a series of systematic studies, we and others have identified and validated the usefulness of markers such as stem cell antigen-1 (Sca-1; refs. 6, 7), CD49f (4, 6), CD117 (6), and Trop2 (8) for enriching murine stem/progenitor cell activity both in vitro in sphere-forming analysis and in vivo in prostate reconstitution assays. Sca-1+CD49fhigh enrichment, in conjunction with hematopoietic and endothelial lineage (CD45+CD31+Ter119+) depletion, has lead to the identification of the Lin−Sca-1+CD49fhigh (LSC) subpopulation. LSC and Lin−Sca-1+CD133+CD44+CD117+ subpopulations are enriched in the proximal region of normal prostate and enhanced on androgen withdrawal (4, 6). Moreover, both subpopulations have been reported to contain sufficient progenitor activity for the regeneration of normal prostatic acini when grafted in conjunction with inductive urogenital mesenchyme (4, 6).
Although the aforementioned studies have identified cell surface markers for enriching stem/progenitor cells from the normal murine prostate, relatively few markers have been identified in the context of prostate cancer. Prostate cancer cell lines sorted for high expression of CD44 have been associated with enhanced expression of “stemness” markers including BMI, β-catenin, SMO, and Oct 3/4 (9). Moreover, CD44+α2β1+CD133+ subpopulations obtained from human tissue have enhanced capacity for in vitro serial passaging, although these subpopulations showed no correlation with tumor grade (10). CD133+ has been used to identify subpopulations in hTERT immortalized human prostate epithelial cell lines with higher progenitor function (11). Recently, CD133+ was shown to identify a minor population in human cell lines with stem-cell like qualities and the capacity to produce progeny with neuroendocrine, transit-amplifying, and intermediate cell characteristics (12).
Loss of PTEN is associated with prostate cancer initiation and metastasis (13). Previously, we have shown that prostate-specific deletion of Pten leads to invasive prostate cancer, mimicking many aspects of human disease (14). During prostate cancer progression, there is expansion of CK5+, p63+, and BCL2+ cells in the proximal regions of the dorsolateral lobes (15), regions known to enrich in stem/progenitor activities in the normal prostate glands. We also showed that Pten deletion regulates basal cell proliferation and expansion (15). Collectively, these observations suggest that CK5+;p63+ subpopulation may associate with prostate cancer initiation and progression in the Pten-null prostate cancer model.
Our current study aims to identify the potential tumor-initiating cells in the Pten-null prostate cancer model. To do this, we have taken a multipronged approach, including (a) in vitro sphere-forming analysis on sorted subpopulations for their stem/progenitor activities, (b) a sphere-mediated in vivo tissue reconstitution assay for their tumorigenic capacities, and (c) in vivo tissue regeneration assays, using the sorted subpopulations from primary cancers, to evaluate their tumor-initiating activities. Results derived from these complementary analyses are consistent and support the notion that the LSC subpopulation in the Pten-null prostate cancer model carries tumor-initiating activity.
Materials and Methods
Mouse strains, tissue collection, and reconstitution
Mutant mice with prostate-specific deletion of Pten were generated as described previously under a mixed background (14). To generate ROSA26LoxpSTOPLoxp-LacZ;Pb-Cre+;PtenL/L mice, Pten mutant mice were crossed to the ROSA26LoxpSTOPLoxp-LacZ line (16). For clonality analysis, Pb-Cre+;PtenL/L mice were crossed with either β-actin GFP [C57BL/6-TgN(ACTb-EGFP)1Osb] or β-actin dsRED [Tg(ACTB-DsRed.MST)1Nagy/J], purchased from The Jackson Laboratory. No obvious phenotype changes were detected on Pten conditional knockout mice when crossed to these reporter mice (data not shown). For in vivo reconstitution experiments, C57BL/6 female mice were used to obtain embryonic day 16/17 urogenital sinus mesenchyme (UGSM). Severe combined immunodeficient (SCID) mice were used for s.c. inoculations and prostate reconstitution and purchased from The Jackson Laboratory. Prostate tissue was harvested from 6- to 8-week-old donor mice and dissected to include all prostate lobes (including the proximal prostate infiltrating the urethra) but exclude the seminal vesicle and bladder as described previously (4, 17, 18). To remove gonadal androgens, surgical castrations were carried out on Pb-Cre+;PtenL/L mice. All animal housing, breeding, and surgical procedures were done under the regulation of the Division of Laboratory Animal Medicine at the University of California at Los Angeles.
Prostate sphere assay
Prostate spheres were cultured and passaged as described previously (18).
Immunofluorescence and immunohistochemistry analyses
Tissue analysis was carried out using standard techniques as described previously (14). To stain prostate spheres, sphere cultures (Matrigel/sphere/prostate epithelial growth medium) were digested using 1 mg/mL dispase solution (Invitrogen) for 30 min at 37°C and subsequently incubated for 2 min with formalin. Spheres were washed in 70% ethanol and embedded in 20 to 30 μL histogel followed by standard paraffin processing and sectioning. For LacZ analysis, cultured spheres were treated with X-Gal solution for 4 to 6 h followed by fixation and embedding in histogel and paraffin. Areas positive for β-galactosidase activity stained blue. Sections of X-Gal–treated spheres were subsequently used for immunohistochemical analysis. CK5, CK8, p63, Ki-67, and CD49f immunodetection were carried out as described previously (4, 15).
Cytospin and staining
LSC+ and LSC− subpopulations were separated by fluorescence-activated cell sorting (FACS) as described previously (4) and diluted to 5 × 104/mL and 100 μL diluted cells were cytospun onto glass slides using Cytospin3 (Shandon). Cells were then fixed with cold methanol, allowed to air dry, and stained using standard immunocytochemistry techniques.
FACS and analysis
Prostate lobes were digested in 1 mg/mL dispase solution (Invitrogen) for 30 min at 37°C and single-cell suspensions were stained with lineage markers (CD31, CD45, and Ter119), Sca-1, and CD49f as described previously (4). Cell sorting and analysis was done using the BD FACSvantage (BD Biosciences).
In vivo regeneration of prostate tissue
UGSM was dissected and cultured as described previously (17, 19). For prostate regenerations from prostate spheres, 1 × 105 UGSM cells were combined with 4 × 103 sphere cells in 50% Matrigel/prostate epithelial growth medium and injected s.c. into a SCID mouse. For regenerations from primary prostate cells, 1 × 105 UGSM cells were combined with 4 × 103 LSChigh, LSClow, or total unsorted cells in conjunction with or without total (Lin−), control prostate epithelium. Grafts were harvested 6 to 8 weeks after injection.
Relative abundance of LSC subpopulation correlates with progression in the Pten mutant prostate cancer model
Previously, we have shown that the LSC subpopulation carries prostate stem/progenitor cell activity in WT murine prostate (4). To test the functional significance of the LSC subpopulation in the etiology of prostate cancer, we set out to first determine the relative abundance of these cells in the Pb-Cre+;PtenL/L prostate cancer model using age-matched and genetic background–matched Pb-Cre−;PtenL/L mice as controls. Pb-Cre+;PtenL/L and Pb-Cre−;PtenL/L mice are herein to be called mutant and control mice, respectively.
Pten mutant mice develop prostate cancer with well-defined kinetics: hyperplasia at 4 weeks, prostate intraepithelial neoplasia at 6 weeks, and invasive adenocarcinoma at 9 weeks (14). Given this, we considered whether disease progression could associate with the relative LSC content by evaluating the mutant prostate at prostate intraepithelial neoplasia (7 weeks) and advanced cancer (20 weeks) stages with age-matched and genetic background–matched controls. Consistent with our previous study (4), we were able to detect 4 ± 0.41% LSC+ cells in controls at age 7 weeks and 4 ± 0.8% LSC+ cells at age 20 weeks. On the other hand, we observed 1.6-fold (P < 0.002) and 3.3-fold (P < 0.0005) more LSC+ cells at prostate intraepithelial neoplasia (6.5 ± 0.75%) and cancer (13 ± 1.91%) stages in the mutant prostates (Fig. 1A). Therefore, LSC+ cell content correlates with disease progression in the Pten-null prostate cancer model.
Similar to majority of human prostate cancers, Pten mutant prostate glands do respond to androgen ablation, as indicated by increased cell death and reduced gross size, but are able to regrow after prolonged castration (14). Therefore, we asked whether repopulation of prostatic epithelium in surgically castrated mice is correlated with enhanced stem/progenitor cell content. Age-matched and genetic background–matched control and mutant mice were either castrated or left intact at 6 weeks and then analyzed for LSC content at age 10 weeks, the time point when we observed castrate-resistant prostate cancer.5 Castration of mice at 6 weeks and evaluation for LSC content at 10 weeks indicated a significant increase in the LSC+ subpopulation in both control (Fig. 1B, open columns; P < 0.005) and mutant (Fig. 1B, filled columns; P < 0.002) prostates. This suggested that androgen ablation may enhance the percentage of LSC subpopulation in both control and mutant prostates.
Our previous study showed that Pten deletion leads to expansion of a subset of prostate cancer cells positive for the basal epithelial markers such as CK5 and p63 (15), similar to the markers used to identify the LSC subpopulation by FACS analysis (4, 7). We then considered whether enhanced LSC content would correlate with these lineage marker-positive cells (15). Immunohistochemistry analyses showed that castration leads to a significant increase in p63+ cells in the control prostate, although most of the p63+ cells were not proliferative based on the lack of costaining with Ki-67+ (Supplementary Fig. S1, arrow in top left). In contrast, many of the p63+ cells in castrated, mutant prostates were also Ki-67+, indicating that PTEN loss may promote more prostate stem/progenitor cells to enter cell cycle after castration (Supplementary Fig. S1, arrows in bottom left), similar to what we have observed in the intact prostate (15). Interestingly, the location of p63+ cells also changes: from basal in the control gland to both basal and luminal in the mutant prostate gland (Supplementary Fig. S1, left), suggesting that castration may not only alter the relative content of stem/progenitor cells but also change the microenvironment or niche of the stem/progenitor cells in the Pten-null prostate model. Importantly, we validated that CD49f FACS marker can colocalize with CK5+ cytosolic marker near the basement membrane in control prostates (4) and both basal and luminal epithelial compartments in the mutant prostate specimens (Supplementary Fig. S1, arrowhead in right).
Although the above study shows that CK5+/p63+ basal cells colocalize with CD49f+ cells, a critical marker used in LSC isolation, more quantitative measurement is needed to correlate in situ tumor tissue analysis with that of in vitro stem/progenitor cells assays. For this, we sorted LSC+ and LSC− subpopulations from the WT and mutant mice, confirmed by PCR genotyping (Supplementary Fig. S2B), cytospun onto the glass slides, and costained for p63 and/or CK5 basal and 4′,6-diamidino-2-phenylindole nuclear markers (Fig. 1C; Supplementary Fig. S2A). p63+/CK5+, p63+/CK5−, or p63−/CK5+ cells were then quantified and presented as the percentages of 4′,6-diamidino-2-phenylindole–positive viable cells within the LSC+ or LSC− subpopulation. More than 80% of the WT (92.0 ± 0.51%) and mutant (84.1 ± 2.5%) LSC+ cells were positive for both p63 and CK5 cytosolic markers. In contrast, <11% of WT (2.8 ± 0.16%) and mutant (10.5 ± 1.17%) LSC− cells were p63+/CK5+ (Fig. 1B). Interestingly, however, mutant LSC− cells contained more p63+/CK5+ cells than WT LSC− cells at a statistically, significant level. This quantitative analysis gives us confidence that we can directly correlate our in situ pathohistologic analysis to LSC-mediated stem/progenitor cell studies.
Pten mutant prostate spheres are larger and more heterogeneous than control spheres
Our previous analysis suggests that LSChigh and LSClow fractions represent basal and luminal compartments of normal prostate epithelium, respectively (4). We have also shown that the LSChigh subpopulation carries greater sphere-forming activity under the defined in vitro culture condition (4, 18). We then asked whether the LSChigh subpopulation of mutant prostates would carry similar sphere-forming activity as their control counterpart. As shown in Fig. 2A, LSChigh cells isolated possess the majority of sphere-forming activities (P < 0.005) defined as sphere diameter >50 μm for both control (84.5 ± 7.7%) and mutant (86.5 ± 8.2%) cells (ref. 19; Fig. 2A, left). Conversely, the majority of LSClow cells formed structures of <50 μm (Fig. 2A, right). Interestingly, when age-matched and genetic background–matched control and mutant prostates were compared, we observed ∼1.5-fold (P < 0.05) higher sphere-forming activity in the mutant LSChigh cells (Fig. 2A, left).
Morphologic comparison indicates that spheres derived from the mutants affirm greater size distribution and heterogeneity when compared with spheres propagated from the control prostatic epithelium (Fig. 2B, left; Supplementary Fig. S3A). Whereas most control spheres consisted of double-layered or “canalized” spheres, mutant spheres vary in size and morphology (Supplementary Fig. S3A; data not shown). Prostate spheres derived from the mutants (ages 6-8 weeks) were also statistically bigger, residing more frequently in larger diameter categories (P < 0.05 for 101-150 and 151-200 μm and P < 0.02 for 251-300 and 301-350 μm), similar to our previous observations of Pten-null neurospheres (20, 21). To further validate that PTEN intrinsically controls sphere size, we took the advantage of incomplete Pten deletion in 2- to 4-week-old prostate due to different levels of Cre expression in various prostate lobes (14, 22). By crossing ROSA26-LacZ reporter mice (16) with Pb-Cre+;PtenL/L mice, we used LacZ+, as measured by X-gal staining, to mark Pten-null cells in our sphere culture assay as both LacZ reporter activation and Pten gene deletion are controlled by the same Cre transgene (Supplementary Fig. S3B). Under clonal plating conditions (18), the number and size of LacZ+ and LacZ− spheres can be accurately measured within the same culture dish. Using this approach, we validated that PTEN intrinsically controls prostate sphere size (Fig. 2B, right). The majority of LacZ− WT spheres were found to be within the 50 to 100 μm diameter range, whereas LacZ+ Pten-null spheres were found to be distributed in higher numbers throughout the 50 to 200 μm ranges (Fig. 2B, right). To validate that prostate spheres did not arise from cells that carry heterozygous Pten deletion, we isolated individual spheres and conducted PCR genotype analysis (Supplementary Fig. S4A), similar to our previous study of neurospheres (21). Of 30 spheres analyzed, 26 of 30 carried homozygous deletion of floxed Pten exon 5 alleles as indicated by the Δ5 band and concordant loss of the loxp band. The remaining four spheres showed the floxed Pten band. No spheres analyzed were Pten heterozygous.
Similar to our quantitative LSC measurements, we observed enhanced sphere-forming activity when the mutants were castrated (Fig. 2C, left; P < 0.005), further supporting that in vivo LSC content, as measured by FACS analysis (Supplementary Fig. S1A), is related to in vitro sphere-forming potential. Interestingly, addition of androgen (1 nmol/L R1881) led to increased sphere number in the control culture (P < 0.05) but had little influence on the mutant sphere number obtained from intact mice (P > 0.05), consistent with the notion that PTEN can control p63+ basal stem/progenitor cell proliferation in the absence of androgen (Fig. 2C, right; Supplementary Fig. S1B).
Deletion of Pten promotes progenitor expansion in prostate spheres
The finding that mutant LSChigh cells formed larger spheres than their control counterparts suggests that Pten deletion may lead to the formation of spheres with an increased content of proliferating cells, similar to Pten-null neurospheres (21). To explore this hypothesis, we carried out immunohistochemical analysis on individual spheres propagated from control and Pten mutant prostates (Fig. 3A). Whereas control spheres were typically composed of a single layer of p63+ basal epithelium with limited proliferative activity (Fig. 3A, left), mutant spheres showed marked expansion of p63+ cells (Fig. 3A, middle), similar to the primary cancer (Fig. 3A, right). When costained with Ki-67, a cell proliferation marker, we found that the majority of Ki-67+ cells in the mutant spheres were also p63+. These data indicate that mutant spheres recapitulate phenotypes associated with the basal compartment of primary cancer in Pten mutant prostate. To support this observation using a more quantitative assay, we then analyzed the number of p63+ and Ki-67+ cells per sphere, normalized by sphere diameter (Ô). Quantitative analysis of P0 cultures generated from total cells of ROSA26-LacZ;Pb-Cre+;PtenL/L prostates indicated 1.7- and 1.9-fold enhancements for p63+ (P < 0.05) and Ki-67+ (P < 0.05) index, respectively, in mutant spheres (left). Similarly, 2.2- and 2.4-fold increases in p63+ (P < 0.02) and Ki-67+ (P < 0.05) cells, respectively, was observed in LacZ+ spheres compared with LacZ− spheres (Fig. 3B, right). Of total spheres counted in this assay, 15.2 ± 4.2% were observed to be LacZ−. Thus, mutant spheres contained more p63+ and Ki-67+ cells than control spheres. Because prostate spheres are clonally derived (Supplementary Fig. S5), these data suggest that Pten-null sphere cells have a higher proliferative index than control sphere cells.
Our previous studies also indicate that WT spheres derived from the LSC+ subpopulation can be serially passaged, a measurement for stem/progenitor self-renewal capacity, and those LSC− cells cannot (18). Thus, we asked whether Pten loss leads to any change in the self-renewal capacity of its stem/progenitor cells. Plating LSChigh control and mutant cells resulted in 3.3% to 5.0% and 4.0% to 8.0% sphere-forming activities during P0-P3 passaging, respectively. Conversely, LSClow fractions yielded inefficient sphere-forming activities in both control and mutant cultures. Moreover, LSClow subpopulation could not be serial passaged (Fig. 3C).
Cells derived from Pten mutant spheres can recapitulate in vivo primary cancer morphology
We then asked whether sphere cells derived from mutants were capable of tissue regeneration and propagating an in vivo cancer phenotype akin to the Pten prostate cancer model. For this, control and mutant LSChigh sorted epithelial cells, from mice ages 6 to 8 weeks, were expanded to first passage (P1) spheres. Spheres were then dissociated to single cells and grafted s.c. in combination with E16/17 UGSM (ref. 18; Fig. 4A, top). Each graft consisted of 4 × 103 sphere-dissociated cells and 1 × 105 UGSM cells, mixed in Matrigel/prostate epithelial growth medium and injected s.c. on the dorsal flank of a SCID mouse. After 8 weeks, examination of grafts by a pathologist, derived from control prostate spheres, revealed prostatic acini lined by a single epithelial layer, a phenotype consistent with normal prostatic structure (Fig. 4A, left). Grafts derived from mutant spheres, however, generated multilayered AR+ abnormal structures (Fig. 4A, right), similar to phenotypes observed in primary Pten-null prostate (Fig. 4B, bottom). Significant expansion of cells positive for progenitor markers (p63/CK5) was also observed in mutant sphere grafts in conjunction with activated phosphoinositide 3-kinase signaling as measured by P-AKTS473 (Fig. 4B, top). Moreover, increased cell proliferation, which was observed in the mutant spheres, was also preserved in the sphere cell–derived grafts in comparison with control sphere grafts (13 ± 4% Ki-67+ for mutant and 1% Ki-67+ for control; P < 0.05). The observed p63 expansion and activated phosphoinositide 3-kinase signaling effectors in mutant grafts are also consistent with the phenotypes observed in primary tumors (Fig. 4B, bottom).
LSChigh subpopulation from Pten mutants is both necessary and sufficient to initiate tumorigenesis in vivo
Because cells derived from Pten mutant prostate spheres are capable of regenerating cancerous glandular structures, we then investigated whether the LSC cells that form spheres are the cells-of-origin for cancer in the Pten-null mouse model. To do this, we compared total, LSChigh, and LSClow cells from intact mutant prostates at ages 6 to 8 weeks. Total or sorted subpopulations were combined with WT UGSM cells and 50% Matrigel/prostate epithelial growth medium and inoculated s.c. onto SCID mice (Fig. 5A). Histologic analysis revealed that grafts with total mutant epithelium generated adenocarcinoma-like structures as determined by a pathologist (Fig. 5A, left). Grafts in the LSClow group, on the other hand, were composed predominately of stroma morphology, contributed by the UGSM. Although we could discern small clusters of cells that appear to remain viable, there was little evidence of ductal regeneration and no cancer present. To confirm that the lack of regeneration capacity in LSClow grafts is not due to experiment failure, LSClow mutant cells were mixed with total epithelial cells from WT prostate. In this control experiment, the only ductal structure obtained was from WT epithelial cells (Supplementary Fig. S6). Importantly, grafts from the LSChigh subpopulation yielded cancerous structures (Fig. 5A, right) that contained AR+ cells with activated P-AKTS473 levels, thereby confirming that cells were null for Pten activity (Fig. 5B, top). Pten mutant LSChigh cell-derived grafts also showed expansion of Ki-67+ cells (15.8 ± 0.7% Ki-67+ in mutant grafts versus 2.0 ± 1% in controls; P < 0.05) and p63+ cells compared with grafts derived from control LSChigh cells. Collectively, these data show that LSChigh cells isolated from Pten mutants are sufficient to propagate cancer when subject to in vivo regeneration assays, whereas the LSClow subpopulation has significantly less in vivo tumorigenic potential.
The Pten prostate cancer model mimics many aspects of human prostate cancer biology, including the invasive and progressive nature of the disease and the alteration of disease relevant gene expression (14, 15, 23). In this study, we showed that the majority of cancer-initiating activity is contained within the LSChigh subpopulation within Pten-null epithelium. Our study validated that structural and functional differences between normal and Pten-null prostate glands can be recapitulated within the LSC subpopulation by in vitro sphere culture and in vivo regeneration assays using cells derived from either sphere culture or freshly isolated from primary cancers. An important finding from our study is that the basal-like cells, as defined by LSC FACS analysis, of Pten-null prostate cancer model contains the majority of tumor-initiating function, whereas the more differentiated luminal component has very little such activity. The lack of tumor-initiating activity of mutant LSClow subpopulation is likely attributed to the lack of progenitor/stem cells, similar to the WT LSClow subpopulation (4). Consistent with our previous study that p63 and CK5 basal cell markers are separated in a minor population of cells within or immediately above the basal compartment of the Pten-null prostate gland (23), we observed more LSC+ cells from the mutant that are negative for p63/CK5 than its WT counterpart (10.3% versus 2.6%), suggesting that Pten deletion may lead to the expansion of the potential intermediate or transit amplifying cells, which may also be capable of cancer initiation. Moreover, it is possible that not all cells within LSC+ pool have the same potential for cancer initiation. It is clear, however, that the majority of normal and oncogenic progenitor function is contained within the LSC+ population as defined by the LSC immunophenotype.
The capacity of CD49fhigh to enrich for basal subpopulations is not restricted to prostate but has also been shown in skin (24) and mammary epithelium (25). For example, using mammary fat pad transplantation assays, CD49fhigh and CD49flow subpopulations were shown to be critical in determining the presence or absence of progenitor function, respectively (25). Interestingly, gene expression analysis showed CD49f to be highly expressed in tumor-initiating CD44+/CD24− subpopulation (26), supporting the potential role of CD49f+ cells in human mammary gland carcinogenesis.
Although we have shown that the LSChigh subpopulation contains cells with the capacity for regeneration of an oncogenic phenotype, there are several important questions that remain to be answered. Is a single cell from the Lin−Sca-1+CD49f high subpopulation capable of multilineage differentiation and reconstituting a normal prostate or tumorigenic prostate? Are such reconstituted oncogenic structures capable of serial transplantation and therefore have the ability of self-renewal? Several studies of this type have been carried out using WT cells, including the mammary gland (27) and murine prostate (6). Using a single cell from the Lin−CD29highCD24+ subpopulation, an entire mammary gland with normal function was regenerated (27). As a corollary, it is interesting that Lin−CD29highCD24+ subpopulation is also significantly increased in the MMTV-Wnt1 breast cancer model, thereby implying a functional interaction between Wnt signaling and mammary stem cells (27). It has also been reported that a single cell from the Lin−Sca-1+CD133+CD44+CD117+ subpopulation can generate WT prostatic acini, which can be antagonized using an anti-CD117 monoclonal antibody (6). Whether a single cancer stem cell can reconstitute oncogenic structure has yet to be determined.
That the LSC population contains the majority of sphere and regenerative potential provides motivation for targeted therapy of cancer initiation in murine prostate cancer models. Moreover, it is conceivable that in human disease antiandrogen therapy selects for subpopulations of oncogenic progenitor cells with “stemness” qualities similar to those found in the LSC subpopulation. Although it is tempting to directly extrapolate findings from mice to humans, it is important to consider critical anatomic, biological, and marker expression differences between these systems. To date, few surface markers have been shown to be functionally relevant to both mouse and human prostate cancer-initiating cells. Thus, to more accurately bridge findings obtained using murine cancer models to that of human disease, it will be important to consider how progenitor cell expansion in the basal compartment in the murine prostate may relate to human disease, in which the cell-of-origin for resistant disease may occur by clonal expansion and/or derivation from a particular or multiple cell compartments. Thus, common cell surface markers that can be used to identify the tumor-initiating subpopulations from mouse and humans prostate cancers will be critical.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: NIH grants F32 CA112988-01 (D.J. Mulholland), K99CA125937 (L. Xin), and R01 CA107166 and RO1 CA121110 (H. Wu) and Prostate Cancer Foundation (D. Lawson, H. Wu, and O. Witte).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank our colleagues in our laboratories for helpful comments and suggestions.
- ©2009 American Association for Cancer Research.