WO2011038300A1 - Cancer stem cells, kits, and methods - Google Patents

Abstract

The present invention provides a cancer stem cell that has, inter alia, the properties of asymmetrical cell division, tumor initiating capacity, and metastatic competence. This cancer stem cell population is phenotypically characterized by the lack of major histocompatibility antigens and differentiation markers (e.g., cytokeratins, androgen receptor, etc.) and evidence of activation of self-renewal/developmental pathways (e.g., WNT/β-catenin, NOTCH and Hedgehog). The identification of this cancer stem cell population has important clinical implications in diagnostic and predictive laboratory assays, as well as for development of novel therapeutic strategies specifically targeting the cancer stem cell.

Description

CANCER STEM CELLS, KITS, AND METHODS

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] The present invention claims benefit to U.S. provisional application serial no. 61/245,584 filed September 24, 2009, the entire contents of which are incorporated by reference.

FIELD OF THE INVENTION

[0002] This invention is directed to, inter alia, cancer stem cells, kits, and methods that are useful, for example, for cancer diagnosis, prognosis, as well as target discovery and target validation.

INCORPORATION BY REFERENCE OF SEQUENCE LISTING

[0003] This application contains references to amino acids and/or nucleic acid sequences that have been filed concurrently herewith as sequence listing text file "SeqListing0311188ST25.txt", file size of 1.12 KB, created on September 23, 2010. The aforementioned sequence listing is hereby incorporated by reference in its entirety pursuant to 37 C.F.R. § 1.52(e)(5).

BACKGROUND OF THE INVENTION

[0004] Cancer is a term used to define a group of diseases characterized by unregulated cell proliferation, aberrant differentiation, and defective apoptosis. These neoplastic diseases may all have in common an initial transforming event that is manifold in nature (e.g., viral infection, chemical carcinogens, etc.) and impacts an unique tissue cell, the so-called "adult stem cell." It is believed that these transforming events generate a "tumor/cancer stem cell (CSC)" that is responsible for tumor initiation and the hierarchical organization of cancer. However, such cell has not yet been definitively identified nor subsequently characterized.

[0005] Tumors consist of heterogeneous populations of cells that differ in growth capacities, morphology and marker expression. The cancer stem cell hypothesis posits that in addition to CSCs, tumors comprise small population of transit amplifying clonogens, and large sets of differentiated malignant cells (Reya et al., 2008; Jordan et al., 2006; Dalerba et al., 2007; Vermeulen et al., 2008). Cancer stem cells, like their normal stem cell counterparts, should be undifferentiated and display asymmetrical cell division, resulting in self-renewal and production of differentiated clonogens, thus contributing to tumor heterogeneity. In addition, CSCs have been postulated to be refractory to standard chemotherapy treatments, though such treatments frequently result in eradication of the fastest dividing cells, resulting in tumor response, but not in CSC depletion (Gupta er al., 2009; Sharma et al., 2010; Bao et al., 2006; Trumpp et al., 2008; Dean er al., 2005; Costello et al., 2000; Guzman et al., 2002). This model explains why standard human cancer chemotherapy frequently results in initial tumor shrinkage, though most cancers eventually recur due to regeneration by surviving CSCs. Thus, from a clinical point of view, the identification and molecular characterization of CSCs has fundamental implications for cancer diagnosis, prognosis and novel therapeutic approaches, including targeted treatments.

[0006] In solid tumors, CSCs have been previously claimed to be identified by cell surface immunophenotyping and tumor initiating capacity. These putative CSC subpopulations were shown to display a CD44⁺/CD24^"/|0W and αδ-integrin phenotype in breast cancer (Al-Hajj er a/., 2003; Cariati er a/., 2008; Fillmore et al., 2008; Ponti et al., 2005), a CD133⁺/Nestin⁺ phenotype in brain tumors (Singh et al., 2004), and a CD133⁺ phenotype in colon cancer (Ricci-Vitiani et al., 2007; O'Brien et al., 2007), among others. In prostate cancer, a similar subpopulation of putative CSCs that expresses high levels of CD44 has also been described (Collins et al., 2005; Patrawala et al., 2006; Li et al., 2008; Patrawala et al., 2007). Perhaps the most challenging issue facing the field is that the described CSC subpopulations do not always fulfill the classical properties that define "sternness", mainly the ability to generate differentiated progeny by asymmetrical cell division. Furthermore, the existence of the above-mentioned surface phenotype subpopulations vary dramatically in tumor tissue samples from patients with the same histopathological diagnosis. In some cases, these subpopulations are relatively rare, whereas in others they constitute a large fraction of tumor mass (Quintana et al., 2008; Rosen et al., 2009; Chen ef al., 2010; Shmelkov et al., 2008). These observations highlight the complexity of a consistent identification of CSCs.

SUMMARY OF THE INVENTION

[0007] The inventors postulated that CSCs may lack HLA class I expression, and designed a novel approach to select for CSCs based on this feature using a prostate cancer chemoresistance model. The inventors disclose the identification and characterization of a population of tumor cells that fulfill the properties of CSCs: differentiation through asymmetrical division, tumor initiating capacity, chemotherapy resistance and evidence of activation of self-renewal/developmental transcription factors. Also disclosed is that inhibition of Notch and Hedgehog signaling pathways induces CSC apoptosis in vitro, and delays tumor formation in vivo. Further, the inventors find that this CSC population is present and can be isolated in surgical specimens of tumors, e.g., human prostate, breast, colon, lung and bladder carcinomas. [0008] The present invention identifies and characterizes "tumor/cancer stem cells" with characteristics such as self-renewal, asymmetrical cell division proficiency, tumorigenic and metastatic competence, pluripotent differentiation capabilities, and a distinctive phenotype.

[0009] Table 1 summarizes a number of biomarkers that define the cancer stem cell phenotype. This phenotype may include expression of developmental pathways and biomarkers of "embryonic stem cells" and "tissue/adult stem cells," such as certain homeobox regulatory factors (e.g., Sox2, Sox4), stem cell markers (e.g., Seal, Gata2, Gata3, Nestin), transcription factors (e.g., Notch-2, Glil, Gli2, nuclear beta-catenin), asymmetrical cell division (e.g., RMND5A), and membrane transporters associated with multidrug resistance (e.g., MDRI/P-glycoprotein, MRPI, MRP2, MRP3, ABC3). Of major relevance, it is also discovered that these CSC have in common the lack of expression of histocompatibility proteins, including all HLA class I and HLA class II molecules, which contributes to the evasion of host immune- surveillance, thus being critical for the metastatic spread.

[0010] Accordingly, one embodiment of the present invention is a method for isolating a cancer stem cell (CSC) from a population of cancer/tumor cells. This method comprises: (a) obtaining a population of cancer/tumor cells; (b) identifying those cells from the cancer/tumor cell population that are HLA^"; and (c) separating the HLA^" cells from the cancer/tumor cell population, the HLA^" cells being CSCs.

[0011] Another embodiment of the invention is a method for isolating a cancer stem cell (CSC) from a cancer/tumor cell line that is resistant to an agent used to treat cancer. This method comprises: (a) obtaining a cancer/tumor cell line that is resistant to an agent used to treat cancer; and (b) separating HLA^" cells from the cancer cell line, the HLA^" cells being CSCs. [0012] Another embodiment is a method for isolating a cancer stem cell (CSC) from a docetaxel-resistant cancer cell line selected from the group consisting of DU145 and 22RV1. The method comprises separating HLA^* cells from the cancer cell line, the HLA^" cells being CSCs.

[0013] Another embodiment is a method for isolating a cancer stem cell (CSC) from a sample of cancerous material obtained from a subject. The method comprises separating HLA^" cells from the sample, the HLA^" cells being CSCs.

[0014] Another embodiment is an isolated cancer stem cell obtained by any of the methods disclosed herein.

[0015] In another embodiment, the "tumor/cancer stem cell" that was discovered can divide through asymmetrical division, thus generating the "transit amplifying and expanded tumor cell populations" characterized by tissue-specific differentiation properties (e.g., cytokeratin expression, androgen receptor). These latter neoplastic cells proliferate by mitotic cell cycle division and lack tumorigenic capacity.

[0016] In another embodiment, the "cancer stem cell unit" is defined as a micro-anatomical entity composed of "tumor/cancer stem cells" (ranging from 0.5% to 2.5% of neoplastic cells), "transit amplifying and expanding tumor/cancer cells" (comprising 99% to 97.5% of the neoplastic cells), and neo-vascular tumor- associated endothelial capillaries.

[0017] A further embodiment is a mammalian cancer stem cell (CSC) line that is enriched for cells that are HLA Γ and HLA II^*.

[0018] Another embodiment is a human cancer stem cell line that is enriched for cells that are HLA Γ and HLA II^*. [0019] Another embodiment is an isolated, multipotent mammalian cell line that is enriched for cancer stem cells (CSCs) that are surface antigen negative for

HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin.

[0020] Another embodiment is a mammalian cell culture having an enhanced number of cancer stem cells (CSCs), which CSCs are surface antigen negative for

HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin.

[0021] Another embodiment is a kit for providing mammalian cancer stem cells (CSCs), the kit comprising a cell line enriched for mammalian CSCs that are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin, which are packaged in a suitable container.

BRIEF DESCRIPTION OF THE FIGURES

[0022] Figure 1 shows characterization of Docetaxel acquired resistance in hormone independent prostate cancer cells. Figure 1a shows the results of cell viability assays (MTs) in the parental cells (22RVI and DU145) and Docetaxel acquired resistant cells (22RVI-DR and DUI45-DR) treated with increasing doses of Docetaxel. A red line designates the IC50 concentration of the drug (Docetaxel) for sensitive and resistant cells. 22RV1-DR Docetaxel IC50 increased from 25nM to 10μ (400 fold increase) and DU145-DR Docetaxel IC50 increased from 5nM to 1 μΜ (200 fold increase). The left panels of Figure 1b show quantitative analysis of colony formation assays in parental and Docetaxel resistant cells treated with increasing doses of Docetaxel for 24 hours. The right panels of Figure 1 b show representative colony formation assays of cells treated continuously for 21 days with Docetaxel (25nM in 22RV1 cells and 5nM in DU145 cells). No colonies were observed after continuous administration of Docetaxel in the sensitive cells, whereas colonies were observed in the resistant cells. Similar effects were observed after 24 hour exposure to the drug. Figure 1c shows the results of flow cytometry assessment of apoptosis by annexin-V and propidium iodide staining of cells exposed to DMSO (control) and Docetaxel. Figure 1d shows a Western blot analysis of PARP cleavage of various cell types as indicated. Results shown in Figure 1 (c) and (d) demonstrate that the acquired Docetaxel resistance phenotype is linked to a lack of Docetaxel apoptotic response.

[0023] Figure 2 shows phenotypical characterization and tumor initiating capacity of Docetaxel resistant cells. Figure 2a left shows a Venn diagram of genes with at least 2 fold increase (†) or decrease (|) in transcript expression in the process of acquiring Docetaxel resistance in DU145 and 22RV1 cells. Overlapping genes = 247. Figure 2a right is a histogram representing the gene ontology (GO) of the 247 overlapping genes. Categories with statistical significance (p≤ 0.01) are represented. *GO categories related to cell proliferation, cell death and response to drugs. ^*GO categories related to developmental processes. Figure 2b shows the heat map of developmental genes, organized by hierarchical clustering using Cluster and Treeview. Color coding in red indicates high levels of expression and coding in green indicates low levels of expression. The signal values have been log2 transformed. Figure 2c shows immunoblotting and protein quantification of cell lysates of matched parental and Docetaxel resistant cells for epithelial differentiation markers, prostate related markers, MHC class I antigents, WNT/ β-catenin pathway proteins, NOTCH signaling protein and Hedgehog signaling pathway proteins. Figure 2d shows immunofluorescence analyses in DU145 (2d) and 22RV1 (2d con't) parental and Docetaxel resistant cells of the expression and subcellular localization of CKs and HLA class I antigens, as well as various transcription factors (dephosphorylated β-catenin, cleaved NOTCH 2, Gli1 and GH2). Nuclear localization of developmental transcription factors in Docetaxel resistant cells is seen upon comparison to their parental cells. A lack of expression of CKs is shown. Figure 2e provides histograms showing the tumour initiating capacity and tables summarizing tumor initiating capacity and tumor latencies after injection of parental and Docetaxel resistant cells in NOD/SCID mice. Corresponds to p<0.001 and ""corresponds to p <0.05. Figure 2f shows immunofluorescence analyses in DU145 and 22RV1 parental and Docetaxel resistant cells of the expression and subcellular localization of CKs and HLA class I antigens.

[0024] Figure 3 shows identification of a cytokeratin negative subpopulation in parental Docetaxel sensitive cells. Figures 3a and 3b show flow cytometry and immunofluorescence analysis of CK (18 and 19) expression in parental cells (22RVI and DUI45). Plot of flow cytometry analysis shown in column (1) for control (unstained), CK18, CK19 and CK18+19 stained cells are representative of the population. Column (2) shows the quantification of CK- cells in three independent experiments. Column (3) shows immunofluorescence staining for CK 18 (red), CK19 (green) and CK18+19. White arrow indicates a CK negative cell in both parental cell lines. Figure 3c is a schematic representation of the two initial working hypotheses: transition versus cancer stem cell enrichment by Docetaxel.

[0025] Figure 4 shows generation and validation of the plasmid containing the promoter of CK19 driving GFP expression and that chemotherapy enriches for cancer stem cells with a CK-negative/HLA-negative phenotype. Figure 4a shows a schematic illustration of the generated CK19 promoter-GFP reporter construct. A region of 1768 bp corresponding to the human Cytokeratin 19 promoter was amplified by PCR using DU145 cells genomic DNA. The promoter region includes 1142 bp of the 5' UTR region, 480 bp belonging to Exon 1 and 146 bp belonging to Intron 1. The PCR product was cloned into pEGFPNl The resulting construct has the CK19 promoter upstream the GFP protein coding region and regulates its expression. Figure 4b shows immunofluorescence co-staining for CKs (red) and HLA class I (green) of DU145 and 22RV1 parental cells. White arrow in the merge panel points to a cell with a CK-negative/HLA class l-negative phenotype. Representative flow cytometry analysis shows two distinct populations of cells: a major population of CK-positive/HLA-positive cells and a smaller population of cells with a CK-negative/HLA-negative phenotype. Figure 4c shows the results of PCR of sorted GFP+ and GFP- cells stably transfected with the plasmid. PCR confirms the stable integration of the CK19 reporter construct in GFP+ and GFP- DU145 sorted cells. A fragment of the GFP sequence was PCR amplified using genomic DNA as a template from sorted GFP+ and GFP- DU145 cells stably transfected with the reporter plasmid. As a negative PCR control, untransfected parental DU145 cells were used.

[0026] Figure 5 shows chemotherapy enriches for cancer stem cells with a CK19-negative, GFP negative phenotype. Figure 5a shows a schematic representation of the working hypothesis. Figure 5b shows representative serial imaging of stably transfected cells treated with Docetaxel (10 n ) at various time points as indicated. Unsorted DU145-CK19 promoter-GFP stable cells were treated with Docetaxel (10n ) for 48h and filmed by time-lapse microscopy to study their behavior. Serial images of a representative experiment are included where a black arrow points to a GFP- cell that undergoes cell division and survives under chemotherapy, while GFP expressing cells start dying after mitotic arrest. Figure 5c shows representative flow cytometry analysis and quantification of the percentage of GFP- and GFP+ cells surviving after 48 hours of Docetaxel treatment. Cells from panel (c) were analyzed for GFP expression by flow cytometry after 48 hours of 10nM Docetaxel treatment. Representative plots and quantification of three independent experiments show that there is a shift in the percentage of GFP- and GFP+ populations of cells surviving treatment as compared to untreated cells (control). Figure 5d shows a representative colony formation assay and quantification of sorted GFP-/HLA- and GFP+/HLA+ cells continuously treated with Docetaxel (10 nM). Three independent experiments of GFP/HLA sorted DU145- CK19 promoter-GFP stable cells (GFP+/HLA+ and GFP-/HLA-) continuously cultured with 10nM Docetaxel or in absence of the drug (control) were performed. Corresponds to p<0.001.

[0027] Figure 6 shows the characterization of cancer stem cell features: asymmetrical cell division, tumor initiation and differentiation. Figure 6a shows representative serial imaging of stably transfected cells that depicts a GFP- (CK- negative/HLA class l-negative) cell undergoing an asymmetrical division. Unsorted and untreated DU145-CK19 promoter-GFP stable cells were filmed by time-lapse microscopy during 24 hours. Serial images of a representative experiment show a GFP- (CK-negative) cell dividing asymmetrically and producing a GFP+ daughter cell. Figure 6b shows representative flow cytometry analysis and quantification of GFP populations of cells derived from GFP-negative and GFP-positive sorted cells at different time points. DU145-CK19 promoter-GFP stable cells were GFP sorted and GFP+ and GFP- cells were plated separately and grown for different time periods. Representative flow cytometry analysis plot and quantification of three independent experiments of cultured cells derived from GFP- and GFP+ sorted cells at different time points (Day 1 , 2 weeks, 4 weeks). Figure 6c shows a schematic representation of the experimental design and results, as well as immunofluorescence analysis of differentiation markers (GFP, CKs and HLA class I) in a tumor xenograft generated from GFP-negative/CK-negative/HLA class l-negative cells. The schematic representation of the experiment design shows injection of GFP/HLA sorted DU145- CK19 promoter-GFP stable cells into NOD/SCID mice and the resulting tumours. Three independent experiments that included 8 mice for each sorted cell population (e.g., GFP-negative/HLA class l-negative) and cell dilution (10, 100 and 1000 cells) were performed. Representative immunofluorescence analysis of differentiation markers (GFP, CKs and HLA class I) in a xenograft tumour generated from GFP- negative/CK-negative/HLA class l-negative cells is shown. Figure 6d shows tumor latencies and quantitation of tumor initiating capacity after injection of GFP+/HLA+ and GFP-/HLA- sorted cells in NOD/SCID mice. ^*corresponds to p<0.001 and ^**corresponds to p <0.05. Bar corresponds to ΙΟΟμητι.

[0028] Figure 7 shows the identification of cancer stem cells in human primary and metastatic prostate cancer tissues. Figure 7a shows CK18 and CK19 immunohistochemical expression in representative tissue samples from two patients (Patient 1 with matched primary and metastatic tumors). The histograms show the percentage of CK-negative and positive cells in primary (n=6) and metastatic (n=20) tissue samples from independent patients. Figures 7b-7d show quantification of CK- negative and positive cells in the analyzed tissue samples, as well as representative immunofluorescence based co-expression analysis of CKs (CK18+19) and HLA class I antigens (Figure 7b), transcription factors (cleaved Notch-2, active β-catenin, Gli1 and GH2) (Figure 7c) and androgen receptor (AR) (Figure 7d). Nuclear staining of transcription factors is shown in the identified CK-negative/HLA-negative tumour cells. Bar corresponds to ΙΟΟμιη. [0029] Figure 8 shows the clonability capacity of HLA class I sorted cells. Representative dilution (10, 100 and 1000 cells) colony formation assay of DU145 HLA class I sorted cells and quantification of three independent experiments were performed. HLA class l-negative sorted cells showed a statistically significant higher colony formation when compared to HLA-class l-positive sorted cells. Corresponds to p<0.05.

[0030] Figure 9 shows that HLA class l-negative epithelial tumor cells from different fresh human cancer types have tumor initiating capacity in NOD/SCID mice. Figure 9a shows a representative sorting diagram of HLA class l-negative and positive tumor cells. Figure 9b shows tumor initiating capacity and tumor latencies after dilution assays of human prostate cancer HLA sorted cells. Graphs and corresponding tables summarize the tumour initiating capacity and tumour latency of different dilutions (10, 100 and 1000 injected cells) of human prostate cancer HLA sorted cells (HLA- and HLA+) and unsorted cells, directly from fresh human samples (primary injections) and derived xenografts (secondary injections) in NOD/SCID mice. Four mice for each sorted cell population and cell dilution were injected twice in the upper flanks (HLA-negative) and lower flanks (HLA-positive). Figure 9c shows representative tumor xenograft formation in a NOD/SCID mouse injected with 10² HLA class l-negative (up) and HLA class l-positive (down) cells. Tumors arising from the injection were confirmed to be prostate cancer by histological (H&E) and immunofluorescence (CKs, androgen receptor (AR), and prostate specific membrane antigen (PSMA)) studies performed in human primary tumour and arising xenograft tumours from primary and secondary injections. Figure 9d shows histograms representing tumor initiating capacity and tumor latencies after dilution assays of other human cancers (Colon, Lung, Breast and Bladder) HLA sorted cells. Primary and secondary injections of 100 HLA sorted cells from the other human cancer types were done. Four mice for each sorted cell population and cell dilution were injected twice in the upper flanks (HLA-negative) and lower flanks (HLA-positive). Figure 9e shows histological (H&E) characterization of human primary and matched derived xenografts. Corresponds to p<0.001 , **corresponds to p <0.05 and ***corresponds to p>0.05. Bar corresponds to 100 μιη.

[0031] Figure 10 shows in vitro and in vivo effects of NOTCH and Hedgehog pathway inhibition in the identified cancer stem cells. Figure 10a shows representative cell cycle analysis and quantification of the observed sub-G1 effects in parental (DU145 and 22RV1) HLA-negative and positive sorted tumor cells when exposed during 72 hours to Cyclopamine (C), Compound-E (CE) alone or in combination (C+CE). Figure 10b shows representative colony formation assays and quantification of parental (DU145 and 22RV1) sorted HLA-negative and positive tumor cells when exposed continuously to the Dexamethasone (D) and the combination of the same drugs as in Figure 10a. Figure 10c shows tumor initiating capacity and latencies after injection of 10³ 22RV1 and DU145 HLA-negative sorted cells in NOD/SCID mice exposed to vehicle solution (control), Dexamethasone (D), Cyclopamine plus dexamethasone (D+C), DB2 plus dexamethasone (D+DBZ) or in triple combination (D+C+DBZ). Three independent experiments that included 8 mice for each HLA sorted cell line and treatment (e.g., Cyclopamine) were performed. Figure 10d shows tumor initiating capacity and latencies after injection of 10³ HLA- negative sorted cells from human prostate cancer xenografts #5, #9 and #12 in NOD/SCID mice exposed to same drugs and concentrations as in Figure 10c. The experiment included 8 mice for each prostate cancer case and treatment. Corresponds to p <0.05. [0032] Figure 11 shows that tumor cells that lacked cytokeratins displayed a negative AR phenotype.

[0033] Figure 12 shows the reversibility of acquired Docetaxel resistance in prostate cancer cells. Figure 12a shows quantitation of percent cell viability from cell viability assays (MTs) in Docetaxel resistant cells (22RV1-DR and DU-145-DR) and Docetaxel resistant cells cultured without drug during various time periods (4, 8 and 12 weeks). A red line indicates the IC50 concentration of the drug (Docetaxel) for acquired resistant and reversed resistant cells. Acquired Docetaxel resistant cells cultured without drug become progressively more sensitive to Docetaxel, in a time dependent manner. After 12 weeks of drug withdrawal 22RV1-DR Docetaxel IC50 decreased from 10μΜ to 50nM and DU145-DR Docetaxel IC50 decreased from 1 μΜ to 25nM. Figure 12b right panel shows quantitative analysis of colony formation assays of Docetaxel resistant cells and Docetaxel resistant cells cultured without drug treated with increasing doses of Docetaxel for 24 hours. The left panel of Figure 12b shows representative colony formation assay of cells treated continuously with Docetaxel. The results confirm the reversibility of acquired Docetaxel resistance because reversed resistant cells form fewer colonies when treated with Docetaxel.

[0034] Figure 13 shows Docetaxel resistance reversibility linked to a recovery in the differentiated cell phenotype in DU145 and 22RV1 cells. The left panel of Figure 13 shows western blot analysis and the right panels of Figure 13 show histogram protein quantification of the expression of epithelial differentiation markers (CK18 and CK19) and HLA class I antigens in parental sensitive cells, Docetaxel acquired resistant cells and Docetaxel reversed resistant cells. Reversed resistant cells display higher protein expression levels than Docetaxel acquired resistant cells, achieving similar levels than those observed in parental sensitive cells. [0035] Figure 14 shows generation and validation of the plasmid containing the promoter of CK19 driving GFP expression. Figure 14a shows immunofluorescence staining for CK19 (red) and GFP (green) of DU145 parental cells stably transfected with the pCK19-GFP plasmid. A white arrow in the merge panel points to a cell lacking the expression of CK19 and GFP. Flow cytometry quantification confirms that there is a co-expression of endogenous CK and GFP, thus validating the use of GFP expression as a read out of CK expression in this stable cell line. Figure 14b shows immunofluorescence staining for HLA class I (red) and GFP (green) of DU145 parental cells stably transfected with the pCK19-GFP plasmid. A white arrow in the merge panel points to a cell lacking the expression of HLA class I and GFP. Flow cytometry quantification confirms the co-expression of GFP and HLA-class I antigens. Cells that express GFP are HLA class l-positive, whereas cells that do not express GFP are also HLA class l-negative.

[0036] Figure 15 shows tumour initiating capacity of HLA class I sorted cells. Parental DU145 and 22RV1 cells were sorted by HLA marker expression. Two different cell dilutions (10 and 100 cells) of HLA-negative and HLA-positive populations were injected in NOD/SCID mice. Three independent experiments that included 8 mice for each sorted cell line and cell dilution were performed. The graphs of the upper panels and the corresponding tables of the lower panels summarize the tumour initiating capacity and tumour latency, respectively, of these three independent experiments. In both cell lines, only the injection of 10 HLA-negative sorted cells show tumourigenic capacity, whereas 10 HLA-positive cells do not form tumours. Corresponds to p<0.0001.

[0037] Figure 16 shows quantification of GFP/HLA subpopulations in tumour xenografts generated from injection of GFP-negative/HLA-negative sorted cells. Representative flow cytometry analysis plot and quantification of GFP/HLA subpopulations of cells in tumour xenografts show that two distinct populations of cells are observed: a major population of GFP-positive/HLA-positive cells and a smaller population of cells with a GFP-negative/HLA-negative phenotype.

DETAILED DESCRIPTION OF THE INVENTION

[0038] The present invention includes, inter alia, a method for isolating a cancer stem cell (CSC) from a population of cancer/tumor cells. This method comprises (a) obtaining a population of cancer cells, (b) identifying those cells from the cancer/tumor cell population that are HLA^", and (c) separating the HLA^" cells from the cancer/tumor cell population, the HLA^" cells being CSCs.

[0039] In the present invention, "isolation", "isolating" and other like terms with respect to CSCs means separating, in substantially pure form cells that have the CSC phenotype, including being HLA^" from the rest of the cell population of the cancer/tumor. Preferably, the isolated CSC population is at least 50% pure, such as for example, at least 60%, 70%, 80%, or 90% pure. Also preferably, such CSC populations are at least 91 %, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99-99.99% pure, i.e., free from non HLA^" cells.

[0040] As used herein, "a population of cancer cells" is to be broadly construed to include cancer/tumor cell lines, for example those which are readily available, through, e.g., the ATCC, and may be immortal (e.g., may be propagated indefinitely in vitro) and samples of, e.g., non-immortal cancerous material ("non- immortal samples") obtained from a subject, such as a human, mouse, rat or some other research animal. The cancerous material may be cells and/or tissue and/or fluid obtained, for example from a biopsy or other surgical procedure, from, e.g., a solid tumor, a blood-based tumor, or a nervous system tumor. Non-limiting examples of the source of the cancerous material include blood, urine, cerebrospinal fluid, ascites fluid, tumor ascites, and combinations thereof. Thus, a population of cancer cells may be obtained by harvesting cancer cells from a cell culture using well know techniques, including those disclosed in the Examples below. In addition, a population of cancer cells may be obtained through the acquisition of a tumor sample from a cancer patient.

[0041] In the present invention, cancer cells that are HLA^" may be identified using known techniques for sorting cells by particular cell surface markers, or the lack thereof, such as those disclosed in the Examples, e.g., immunofluorescence staining using HLA antibodies. Preferably, such identifying techniques do not render the cells unviable.

[0042] As used herein, "HLA Γ" means having no major histocompatibility complex (MHC) class I molecules. The MHC is a set of molecules displayed on cell surfaces that are responsible for lymphocyte recognition and antigen presentation. The MHC class I molecules present antigen to cytotoxic T-cells. MHC I molecules are found on almost all types of body cells.

[0043] As used herein, "HLA IP means having no major histocompatibility complex (MHC) class II molecules. The MHC class II molecules present antigen to helper T-cells. MHC II molecules are only found on macrophages, dendritic cells and B cells. In the present invention, "HLA " preferably means HLA and HLA II^".

[0044] In the present invention, separating the HLA^" cells from a cancer cell population may be accomplished using any conventional or known technique that sorts or separates cells based on, e.g., expression, or non-expression, of particular cell surface markers and maintains the viability of the cell. Preferably, the separating is carried out by a FACS analysis as disclosed in more detail in the Examples. [0045] In the present invention, a cancer stem cell ("CSC") is a cancer cell that is HLA^". Such cells are further defined by being at least one of CD24^", CD133^", Notch*, Gli1⁺, Gli2⁺, GFAP^", Neurofil^", and cytokeratin^". Moreover, combinations of any or all of the foregoing are also contemplated. Such CSCs are further defined by having at least one of the following additional properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance. Moreover, combinations of any or all of the foregoing are also contemplated. Further properties of CSCs are listed in, e.g., Table 1. The combination of any of the identified characteristics in Table 1 together with HLA^" may be sufficient to identify a CSC cell.

[0046] In the present invention, the cancer cell line may be resistant to an agent used to treat cancer. Methods for generating such resistant cancer cell lines are known in the art and disclosed in further detail in the Examples. The cancer cell lines used in the present invention may be any cancer cell line that is readily available through commercial sources, e.g., ATCC, or created using known methods. The cell lines may be derived from any cancer that afflicts humans, rats, mice, or other research animals. For example, the cancer cell line may be from solid tumors, blood-based tumors, and nervous system tumors. In one preferred embodiment, the cancer cell line is the prostate cancer cell line DU145 or 22RV1.

[0047] In the present invention, as noted previously, the cancer cell lines and the non-immortal sample from a subject may be obtained or derived from any cancer/tumor from humans, rats, mice or other research animals. Non-limiting examples of such cancers/tumors include prostate, breast, colon, lung, and bladder cancers. Further non-limiting examples of other such cancers that are within the scope of the present invention include leukemia, lymphoma, and glioma. In addition, glioblastoma cell lines, such as, e.g., SNB19, U-373-MG and U-343, have a subpopulation of cells that have a unique phenotype characterized by the lack of expression of Glial Fibrillary Acidic Protein (GFAP) and HLA-class I antigens and are also within the scope of the present invention. These cells mimic the epithelial cancer stem population disclosed herein and have the cancer stem cell functional properties of, e.g., asymmetrical cell division, self-renewal and resistance to conventional therapies such as radiation. In the present invention, this population has also been identified in human Glioblastoma tumors. A cancer stem cell population characterized by the lack of HLA class I expression has also been identified, according to the present invention, in sarcomas. This population of cells is present in sarcoma cell lines of different lineages, such as, e.g., osteosarcoma (e.g., MG-63, Saos-2), leiomyosarcomas (e.g., SKN) and liposarcomas (e.g., SW872).

[0048] As disclosed above, the cancer cell line may be resistant to an agent that is used to treat cancer. As used herein, such an agent is to be broadly interpreted and may include any known chemotherapy compound, composition or combination thereof. For example, the agent may be a DNA damaging drug and/or an anti-mitotic agent. Further non-limiting examples of the agent include: microtubulin inhibitors, topoisomerase inhibitors, vinblastine, vincristine, vinorelbine, paclitaxel, mitoxantrone, cisplatin, docetaxel, colchicines analogs, harringtonine, homoharringtonine, camptothecine, camptothecine analogs, podophyllotoxin, and combinations thereof. In a preferred embodiment, the agent is docetaxel.

[0049] Another embodiment of the invention is a method for isolating a cancer stem cell (CSC) from a cancer cell line that is resistant to an agent used to treat cancer. This method comprises obtaining a cancer cell line that is resistant to an agent used to treat cancer and separating HLA^" cells from the cancer cell line, the HLA^" cells being CSCs.

[0050] In this embodiment, the agent is as previously defined herein. Preferably, the agent is docetaxel.

[0051] In this embodiment, the cancer cell line is obtained from a cancer as defined above. Preferably, the cancer cell line is DU145 or 22RV1.

[0052] In this embodiment, after the obtaining step is carried out the cancer cell line is contacted with a labeled, e.g., a fluorescently labeled, antibody against HLA. Then, the separating step is carried out, as disclosed above, using any available cell sorting/separating technique, including FACS analysis to separate those cells that are HLA^" from those that are HLA⁺.

[0053] Another embodiment of the invention is a method for isolating a cancer stem cell (CSC) from a docetaxel-resistant cancer cell line selected from the group consisting of DU145 and 22RV1. This method comprises separating HLA^" cells from the cancer cell line, the HLA^" cells being CSCs. In this embodiment, the compositions and techniques previously defined may be used.

[0054] Another embodiment of the invention is a method for isolating a cancer stem cell (CSC) from a sample of cancerous material that is obtained from a subject. This method comprises separating HLA^" cells from the sample, the HLA^" cells being CSCs. Preferably, the sample is obtained from a fresh biopsy or other surgical sample of, e.g., a cancerous lesion from a subject, such as a human cancer patient. In this embodiment, the separating step is as defined previously. In a preferred aspect of this embodiment, in addition to being HLA^", the CSCs are further defined by being at least one of CD24^", CD133^", Notch*, Gli1⁺, Gli2⁺, GFAP^", Neurofil^", and cytokeratin^". Moreover, combinations of any or all of the foregoing are also contemplated. In another preferred aspect of this embodiment, in addition to being HLA, the CSCs are further defined by having one or more of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance. Moreover, combinations of any or all of the foregoing are also contemplated.

[0055] Another embodiment of the invention is an isolated cancer stem cell or population of cancer stem cells that are obtained by any of the methods disclosed herein.

[0056] Another embodiment of the invention is a mammalian cancer stem cell (CSC) line that is enriched for cells that are HLA Γ and HLA ΙΓ. In this embodiment, "mammalian" refers, preferably to human, mouse, or other research mammal, such as e.g., a rat. As used herein, "enriched" means a cancer cell line that has increased numbers of CSCs relative to a control cell line that has not been treated, with e.g., an agent used to treat cancer. The increased numbers of CSCs may be transient, e.g., some of the CSC may differentiate, or may be of a longer duration. Moreover, such a cell line may or may not be pure, e.g., substantially free of HLA⁺ cells.

[0057] In this embodiment, the mammalian CSC line is further enriched for cells that are at least one of CD24^", CD133^", Notch\ Gli1⁺, Gli2⁺, GFAP^", Neurofil^", and cytokeratin^". Moreover, combinations of any or all of the foregoing are also contemplated. The mammalian CSC line may further be enriched for cells that are defined by having at least one of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance. Moreover, combinations of any or all of the foregoing are also contemplated.

[0058] In this embodiment, the mammalian CSC line may be a cultured cell line, preferably an immortal cell line as defined above.

[0059] In this embodiment, the CSCs of the mammalian cell line, when transplanted, e.g., into a host animal, form a tumor. Such a tumor may be selected from the group consisting of solid tumors, blood-based tumors, and nervous system tumors, including any of the tumors disclosed herein.

[0060] Another embodiment of the present invention is a human cancer stem cell line that is enriched for cells that are HLA Γ and HLA II^". A further embodiment of the invention is an isolated, multipotent mammalian cell line that is enriched for cancer stem cells (CSCs) that are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin.

[0061] A further embodiment of the present invention is a mammalian cell culture having an enhanced number of cancer stem cells (CSCs), which CSCs are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin. In this embodiment, the mammalian cell culture is also enriched for CSCs that are Notch^"*", Gli1⁺, and Gli2⁺. In this embodiment, the cell culture is enriched for CSCs having at least one of the following properties: capability to self- renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance. Moreover, combinations of any or all of the foregoing are also contemplated. [0062] In this embodiment, the mammalian cell culture may be a cultured CSC line or an immortal CSC line as previously defined. Preferably, the CSCs may be from a human, mouse, or another research mammal, such as e.g., a rat.

[0063] The cells from the mammalian cell culture of this embodiment, when transplanted, form a tumor. Such a tumor may be selected from the group consisting of solid tumors, blood-based tumors, and nervous system tumors, including any of the tumor disclosed herein.

[0064] Another embodiment of the invention is a kit for providing mammalian cancer stem cells (CSCs). The kit comprises a cell line enriched for mammalian CSCs that are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin, which CSCs are packaged in a suitable container. In this embodiment, the enriched CSCs are also Notch⁺, Gli1⁺, and GH2⁺. In this embodiment, the enriched CSCs also have at least one of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance. Moreover, combinations of any or ail of the foregoing are also contemplated.

[0065] In this embodiment, the CSCs may be packaged in any form. For example, they may be part of a frozen stock or part of a growing culture.

[0066] In this embodiment, the container may be any appropriate apparatus for shipping and/or storing the CSCs. Such containers are known in the art and include a tissue culture container or a freezer-safe tissue culture container.

[0067] The following terms shall be used to describe the present invention. In the absence of a specific definition set forth herein, the terms used to describe the present invention shall be given their common meaning as understood by those of ordinary skill in the art.

[0068] As used herein, "capability to self-renew" means, e.g., that a CSC has the ability to go through numerous cycles of cell division while maintaining its undifferentiated state.

[0069] As used herein, "asymmetrical cell division" means, e.g., capable of having a cell division that leads to two cells with different properties, such as for example one cell which maintains the CSC phenotype while the other is programmed to, e.g., differentiate.

[0070] As used herein, "tumorigenic capacity" means, e.g., a cell, such as a CSC, that has the ability to generate a tumor when transplanted into a host animal.

[0071] As used herein, "metastatic potential" means, e.g., the ability of a cell, such as a CSC, to move to a secondary location in a body, i.e., metastasize, and generate a tumor.

[0072] As used herein, "multidifferentiation properties" means, e.g., the ability of a cell to differentiate into more than one cell type.

[0073] As used herein, "sensitive to Notch and hedgehog inhibitors" means a cell, such as a CSC, which is modulated by inhibitors of the Notch and hedgehog pathways. Such inhibitors are known in the art. See, e.g., K. Garber, JNCI, 99(17): 1284-1285 (2007) (Notch inhibitors) and Martinson et al., U.S. Patent No. 7,695,965 (hedgehog inhibitors).

[0074] As used herein, "broad chemoresistance" means a cell, such as a CSC that is resistant to a range of chemical agents, such as, e.g., a variety of agents as previously defined herein. [0075] As used herein, "CSC" refers to tumor/cancer stem cells that are characterized by the ability to self renew, asymmetrical cell division proficiency, tumorigenic and metastatic competence, pluripotent differentiation capabilities, and a distinctive phenotype. This phenotype includes expression of developmental pathways and biomarkers of "embryonic stem cells" and "tissue/adult stem cells," such as certain homeobox regulatory factors (e.g., Sox2, Sox4), stem cell markers (e.g., Seal, Gata2, Gata3, Nestin), transcription factors (e.g., Notch-2, Gli1 , GH2, nuclear beta-catenin), assymetrical cell division markers (e.g., RMND5A), and membrane transporters associated with multidrug resistance (e.g., MDRI/P- glycoprotein, MRP1, MRP2, MRP3, ABC3). In addition, these CSC lack expression of histocompatibility proteins, including all HLA class I and HLA class II molecules. These CSC are identified both in vitro (all tumor cell lines analyzed to date) and in vivo (including all murine and human primary and metastatic tumors studied to date). CSCs isolated from human primary tumors display tumorigenic properties when transplanted into immunedeficient mice.

[0076] As used herein, "cancer stem cell unit," also termed "cancer stem cell niche," refers to a micro-anatomical entity composed of CSC (minority population ranging from 0.5% to 2.5% of neoplastic cells in general), "transit amplifying and expanded tumor cells" (comprising the majority of neoplastic cells, from about 99.5% to about 97.5%, in general), and neo-vascular tumor-associated endothelial cells forming a functional structure present in all tumors.

[0077] As used herein, "transit amplifying and expanded tumor cells" refers to those tumor cells in vitro and in vivo that do not divide by assymetrical cell division, have no tumorigenic competence, and have a phenotype distinct from CSC, mainly the expression of HLA molecules and differentiation markers (e.g., cytokeratins, androgen receptor, etc).

[0078] It has been postulated that CSCs may lack HLA class I expression, and a novel approach was designed to select for CSCs based on this feature using a prostate cancer chemoresistance model. A population of tumour cells has been identified and characterized that fulfill the properties of CSCs : differentiation through asymmetrical division, tumour initiating capacity, chemotherapy resistance and evidence of activation of self-renewal/developmental transcription factors. In the model in accordance with the present invention, inhibition of Notch and Hedgehog signaling pathways induces CSC apoptosis in vitro, and delays tumour formation in vivo. Further, this CSC population is present and can be isolated in surgical specimens of human prostate, breast, colon, lung and bladder carcinomas, a fact that provides further consistency to the findings.

[0079] As noted above, two major hypotheses regarding tumor initiation have been postulated. The "stochastic model" predicts that every neoplastic cell can generate an entirely new tumor. Alternatively, the "cancer stem cell model" proposes that tumor cells exist in a hierarchical state, and that only a few specialized cells possess tumor initiating potential. In the present invention, it is documented for the first time the identification and characterization of the human prostate cancer stem cell, as well as those from other human primary solid tumors such as bladder cancer, colon cancer, breast cancer, lung cancer, melanoma and sarcoma, based on at least the following three functional criteria: 1) Self-renewal capacity, 2) differentiation through asymmetrical cell division, and 3) tumor initiating capacity. This cell is also characterized by a chemoresistance phenotype, contributing to the clinical phenomenon of acquired drug resistance. Furthermore, the lack of HLA class I and class II molecules facilitates the phenomenon of metastatic spread by escaping immune-surveillance mechanism of the host. This phenomenon includes the ability of intravascular circulation as a "phantom" cell unidentified by the immune system, as well as early seeding in distant organs.

[0080] In the present invention, exposure to chemotherapy (Docetaxel) was initially used to select for a small population of cells with sternness signature and characteristics. This subpopulation of cells has a unique phenotype characterized by the lack of epithelial differentiation markers and the expression of transcription factors implicated in self renewal and developmental processes. Additional studies revealed that these cells (CSC) lack HLA class I and class II molecules, offering a further tool for their isolation from both in vitro and in vivo tumor models, as well as murine and human primary and metastatic lesions. The in vitro studies clearly demonstrated that the asymmetrical division process generated typical prostate- lineage epithelial differentiated cell (CK-positive phenotype). This phenotype constituted the majority of parenchymal tumor cells in vivo. Evidence was also provided which demonstrate that chemotherapy targets the differentiated tumor cells, whereas the cancer stem cell is resistant to such treatment. Concerning tumor initiation capacity, it was found that this newly defined prostate cancer stem cell (CK- negative/Developmental transcription factors-positive phenotype) can generate tumors in vivo, while the differentiated, expanding cell population lacks such a property. These studies have been extended to the isolation of CSC from human primary solid tumors, including prostate cancer, bladder cancer, colon cancer, breast cancer, lung cancer, melanoma and sarcoma. In all of these tumor types the CSC has tumorigenic capacity. [0081] Finally, the present invention identifies and characterizes this newly defined prostate cancer stem cell in human prostate cancer samples, as well as in other solid tumors, as indicated above. Like in the experimental prostate cancer models, these cells are characterized by a CK-negative/AR-negative phenotype. Moreover, these cells also possess a negative phenotype for both class I and class II major histocompatibility antigens. It was observed that in metastatic prostate cancer lesions these cells account for 1%-2% of the tumor cell population. Taken together, these findings have major clinical implications, since such phenotype and functional properties could explain host permissiveness, escape of host immune-surveillance, and facilitating tumor spread and metastasis. The identification of this cancer stem cell population has important clinical implications in diagnostic and predictive laboratory assays, as well as for development of novel therapeutic strategies specifically targeting the cancer stem cell. In this context, treatment with Notch and Hedgehog inhibitors induces apoptosis in this cancer stem cell population, and abrogates tumor formation in experimental animal models.

[0082] In another embodiment, the present invention provides an isolated cancer stem cell that expresses, or has increased expression as compared to a control cancer cell, one or more of the following biomarkers: (i) Sox2 or Sox4, (ii) stem cell markers such as Seal, Gata2, Gata3, and Nestin, (iii) transcription factors such as Notch-2, Glil, GH2, and nuclear beta-catenin, (iv) assymetrical cell division marker RMND5A, and (v) membrane transporters associated with multidrug resistance such as MDRI/P-glycoprotein, MRPI, MRP2, MRP3, and ABC3. In one example, the control cancer cell can be a transit amplifying and expanded tumor cell. In another embodiment, the above cancer stem cell does not express, or has decreased expression as compared to a control cancer cell, of one or more of the following biomarkers: HLA class I or HLA class II molecules; transduction signaling molecules (e.g., NFkB, EGFR, HSP70, HSP90); Interferon ligands and receptors (such as IFII6, IFNARI, IFNAR2); IL6 and IL8; epithelial differentiation markers (e.g., cytokeratins); tissue specific differentiation markers (e.g., androgen receptor); and adhesion molecules (e.g., e-cadherin). Other negative cell surface markers include CD133, CD44, and CD24.

[0083] In one embodiment, the cancer stem cell described above is HLA class I and class II negative, CD133 negative, CD44 negative, and CD24 negative. In general, the cancer stem cell is capable of asymmetrical cell division or possesses tumor initiating capacity.

[0084] In another embodiment, the present invention also provides a composition comprising the cancer stem cell described herein.

[0085] In another embodiment, there is provided a method of detecting the presence of cancer cells in an individual, comprising the steps of: obtaining one or more tissue samples from the individual; and detecting the presence of the cancer stem cells described herein, wherein detection of the cancer stem cells in the samples indicates the presence of cancer ceils in the individual. The cancer cells can be primary cancer cells or metastatic cancer cells. For example, the cancer cells can be prostate cancer cells, bladder cancer cells, colon cancer cells, breast cancer cells, lung cancer cells, melanoma, or sarcoma. In one embodiment, the cancer stem cells are HLA class I and class II negative, CD133 negative, CD44 negative, and CD24 negative.

[0086] In another embodiment, there is provided a method of predicting the chances of cancer cell metastasis in an individual, comprising the steps of: obtaining one or more tissue samples from the individual; and detecting the presence of the cancer stem cells described herein, wherein detection of the cancer stem cells in the samples indicates a high probability of cancer cell metastasis in the individual. For example, the cancer cells can be prostate cancer cells, bladder cancer cells, colon cancer cells, breast cancer cells, lung cancer cells, melanoma, or sarcoma. In one embodiment, the cancer stem cells are HLA class I and class II negative, CD133 negative, CD44 negative, and CD24 negative.

[0087] In another embodiment, there is provided a method of screening for a candidate anti-cancer agent, comprising the steps of: contacting the cancer stem cells described herein with a candidate compound, and determining the growth of the cancer stem cells in the presence or absence of the compound in vitro or in vivo, wherein a decreased cancer stem cell growth indicates the compound is a candidate anti-cancer agent. In one embodiment, the candidate compound can be a protein, a product secreted by cancer stem cells, or synthetic compounds. In another embodiment, the candidate compound interferes or inhibits signaling through a pathway such as Hedgehog pathway, Notch pathway, or Wnt pathway.

[0088] In yet another embodiment, the candidate compound can be an assymetrical cell division inhibitory agent or an agent capable of disrupting cancer stem cell unit.

[0089] In another embodiment, there is provided a candidate anti-cancer agent identified by the method described above.

[0090] The following examples are provided to further illustrate the methods and compositions of the present invention. These examples are illustrative only and are not intended to limit the scope of the invention in any way. EXAMPLE 1

PROSTATE CANCER CELLS RESISTANT TO DOCETAXEL DISPLAY A DEVELOPMENTAL MOLECULAR SIGNATURE CONSISTENT WITH CELLULAR STEMNESS.

[0091] Docetaxel is an anti-mitotic agent currently used as standard therapy in patients with hormone refractory prostate cancer (Petrylak et al., 2004; Tannock et al., 2004. However, all patients ultimately experience disease progression, and no other treatment controls the disease in this context. Due to the clinical relevance of this resistance phenomenon, an in vitro model of Docetaxel resistance using the well established prostate cancer hormone-independent cell lines DU145 and 22RV1 was generated to characterize the molecular alterations responsible for such an event.

(Figures 1a-d).

[0092] DU145 and 22RV1 cells were exposed to increasing doses of Docetaxel, and the acquired chemoresistance was characterized and confirmed by cell viability, colony formation, annexin V, and poly-(ADP-ribose) polymerase (PARP) cleavage assays. (Figures 1a-d). The generated Docetaxel resistant (DR) cells showed cross-resistance to DNA damaging drugs (Mitoxantrone, Doxorubicin and Cisplatin), as well as other anti-mitotic agents (Vinorelbine and Paclitaxel), consistent with a multidrug resistant phenotype (data not shown).

[0093] Gene expression profiling, using oligonucleotide microarrays, was performed to compare sensitive (DU145 and 22RV1 - parental cells) and acquired resistant (DU145-DR and 22RV1-DR) prostate cancer cells. Genes with at least 2 fold increase or decrease in transcript expression were selected for further analysis. This analysis disclosed 1245 and 990 deregulated genes in DU145-DR and 22RV1- DR cells, respectively, of which 247 genes overlapped (Figure 2a left). Of these overlapping genes, 29.5% were up-regulated and 70.5% were down-regulated. Surprisingly, gene ontology category assessment of biological processes of the commonly deregulated 247 transcripts revealed that, besides the expected enrichment for cell proliferation, cell death, and drug response biological processes, developmental and immune surveillance categories were significantly represented (Figure 2a right). Genes of relevance in prostate cancer biology and sternness that were found up- or down-regulated included: a) epithelial differentiation biomarkers - cytokeratins (CK18 and 19), and prostate specific markers such as androgen receptor (AR), prostate specific antigen (PSA) and prostate specific membrane antigen (PSMA); b) immune-surveillance antigens - major histocompatibility complex class I (MHC class I); and c) sternness signaling pathways -Wnt p-catenin, Notch and Hedgehog. Selected genes of relevance are shown in Tables 1 and 2 below. Transcript levels of certain cytostructural genes and signaling pathways, as well as clinicopathological and phenotypical characteristics of 20 metastatic prostate cancer patients' tissue samples are summarized in Table 3 below.

TABLE 1

Down regulated/ inactive Up regulated/ active

AMIG02

CLDNI (claudin 1)

CLDN4 (claudin 4)

CLDN7 (claudin 7)

CLDN10 (claudin 10)

CLDN11 (claudin 11)

CDH1 (e-cadherin)

CDH2 (n-cadherin)

CDH3 (p-cadherin)

CDH7 (cadherin 7)

CDCP1 (CUB domain

containing protein 1)

ICAM1

ITGA3 (integrin alpha 3)

ITGA5 (integrin alpha 5)

ITGA6 (integrin alpha 6)

JUP Uunction plakoglobin)

LAMA3 (laminin alpha 3)

LAMB3 (laminin beta 3)

LAMC2 (laminin C 2)

SDC1 (syndecaM)

SDC2 (syndecan2)

ZYX (zyxin)

Developmental AXIN1 (negative regulator ABLIM3

W T\ J NOTCH2

DKK1 (negative regulator HES1

WNT) HEY1

WIFI (negative regulator PTCH1

WNT) GLI1

GPRC5A GLI2

GPRC5C BMP1 (Bone morphogenic protein 1)

GATA2

NES (nestin)

NKX3.1

Immunosurveillance HLA-A

HLA-B

HLA-C

HLA-E

HLA-F

HLA-G

HLA-DR

HLA-DQ

HLA-DP

CD59

CD74

MR1 (major

histocompatibility complex

class I) Down regulated/ inactive Up regulated/ active

SECTM1

Cluster differentiation CD133

Markers CD24

CD63

Signal transduction AKAP1 (PKA& PKC) ARHGEF10 (RhoPTTase)

AKAP12 (PKA & PKC) MZF1

ANXA3 (Inositol

phosphate)

CALM1 (calmodulin)

CDS2 (phosphatidyl

inositol)

EFNB2 (ephrin-B2)

EGFR

EHD1 (EGFR substrate)

ETS1

ETS2

FOSL1

HGF (hepatic growth

factor)

HSP90 (Heat shock

protein 90)

HSP70 (Heat shock

protein 70)

IFI16 (interferon, gamma

inducible protein)

IFI35 (interferon, gamma

inducible protein)

IFITM1 (IF inducible

transmembrane protein)

IFITM2 (IF inducible

transmembrane protein)

IFITM3 (IF inducible

transmembrane protein)

IFNAR1 (intereferon

receptor 1)

IFNAR2 (intereferon

receptor 2)

IL6 (interleukin6)

IL6R (interleukin6

receptor)

IL8 (interleukin8)

LIMK1 (LIM domain

kinase 1)

MET (hepatocyte growth

factor receptor)

NMB (neuromedin)

STAT3

Cell Cycle AURKA (Aurora Kinase A) MCM5

G0S2 (G0G1 switch gene) NSL1 (kinetochore

GSN (gelsolin, amyloid)

TABLE 2

Gene name x-fold change p value (t-test)

DU145/DR;

22RV1/DR

Keratin 19 -29.8,-21.5 <0.0001;<0.0001

Keratin 18 -2.7;-2.1 <0.05;<0.05

Androgen receptor NE;-2.2 -;<0.05

Folate hydrolase (prostate- NE;-5.4 -;<0.0001 specific membrane antigen) 1

kallikrein-related peptidase 3 NE;-4.3 -;<0.0001

Major histocompatibility -2,1 ;-5.4 <0.05;<0.0001 complex, class I, A

Major histocompatibility -2.0;-3.9 <0.05;<0.05 complex, class I, B

Major histocompatibility -2.3;-4.4 <0.05;<0.0001 complex, class I, C

Major histocompatibility -2.1 ;-2.6 <0.05;<0.05 complex, class I, E Major histocompatibility -2.4,-1.8 <0.05;<NS complex, class 1, F

Major histocompatibility -2.1 ;-1.7 <0.05;<NS complex, class 1, G

Dickkopf homolog 1 (Xenopus -3.8;-3.2 <0.05;<0.05

laevis)

Notch homolog 2 (Drosophila) +4.2;+2.1 <0.0001 ;<0.05

Patched homolog 1 +3.2;+3.3 <0.05;<0.05

(Drosophila)

GLI family zinc finger 1 +2.2;+2.3 <0.05;<0.05

GLI family zinc finger 2 +2.1 ;+2.0 <0.05;<0.05

Note: NE= Not expressed. NS=Not Significant.

[0094] Genes that were found up or down-regulated in Docetaxel resistant cells (DU145-DR and 22RV1-DR) when compared to their parental sensitive cells DU145 and 22RV1) are summarized in Table 2. Statistical analysis of the mean expression average difference of genes, which show > 2 fold change based on a logarithmic normalization, was done using a t-test between matched sensitive and resistant cells. There is a decrease in the transcription levels of genes involved in epithelial differentiation, prostate specific markers and immune-surveillance markers. On the other hand, gene transcripts of developmental transcription factors are up- regulated.

TABLE 3

Patient Metastatic Hormone CK18+19 % active % % GJi-1 % Gli-2 status β-catenin cleaved

site

NOTCH-2

1 Peritoneum Independent 66.6 77.7 92

+ 2 4 1.3 2

2 Bone Independent : - 33.3 33.3 73 77

+ 0 0 0.5 0.5

3 Brain Dependent = - 86 94 60 65

+ 4.5 36 1.2 2.1

4 Lung Independent■ 75

+ 2 3.5 1.2 1

5 Bone Independent - 66 75 67 67

+ 3 0 .05 .05

6 Epidural Independent l - 77.7 68 66

+ 0.8 201 0.9 0.3

7 Lymph Independent 38

+ 0.1 10.4 0 0 Patient Metastatic Hormone CK18+19 % active % % Gli-1 % Gli-2 status β-catenin cleaved

site

NOTCH-2

Node

8 Bone Independent - 75 52 71 75

+ 4 0 1.8 2.1

9 Lymph Independent Ϊ - 80 65 65 72

+ 2.2

Node 2.3 2.1 2.1

10 Lymph Independent ; - 30 37 55 63

+ 0 0.8 node 0 0.9

11 Testicular Independent 30 87 48 56

+ 0.1 5.3 1.2 1.5

12 Lymph Independent 5 - 40 68 33.3 33.3

+ 0.3

Node 1.5 0 0

13 Lymph Independent \ - 66 54 68 65

+ 0.1

Node 4.5 1.4 1.5

14 Lymph Dependent ^"i - 75 100 63 75

+ 0.1 48 1.8 1.5

Node

15 Lymph Dependent - 50 76 33.3 33.3

+ 0.3 1.3 0

Node 0

16 Lymph Independent - 95 32 97 97

+ 40

Node 0 33 47

17 Lymph Deoendent ¾ - 40 40 55 66

+ 1

Node 0 1.5 2

18 Bone Independent - 87 83 83 86

+ 9.7 4 18 15

19 Lymph Dependent I - 90 86 79 83

+ 40 17 2.3 2.5

Node

20 Lymph Independent - 88 94 71 67

+ 7 35 1.2 0.9

Node

[0095] Protein validation of these genes by western blot (Figure 2c) and immunofluorescence assays (Figures 2d and 2f) confirmed that the acquired resistant cells had a decrease in epithelial differentiation proteins, prostate specific biomarkers and immune surveillance antigens, and displayed an activation (nuclear expression) of developmental signaling pathways (Figure 2d). Activated β-catenin, Gli1 , Gli2 and cleaved NOTCH-2 showed significant increase in protein expression as well as nuclear immunolocalization, when comparing Docetaxel resistant cells to parental sensitive cells, which displayed mainly a cytoplasmic/membranous expression. Furthermore, these phenotype were found to be linked to the chemoresistant phenomenon, since reversed acquired resistant cells recuperate the differentiated phenotype (Figure 13).

[0096] With respect to studies on differentiation, the expression of cytokeratins (CKs) as epithelial markers, as well as prostate-related biomarkers, including AR, PSA and PSMA, were assessed. CKs have been previously reported to be specific for human differentiated epithelial cells, playing a role in the maintenance of cellular integrity while also functioning in signal transduction and cellular differentiation processes (Brulet et al., 1980; Lu et al., 1980; Oshima ef al., 1981; Tesar et al., 2007J. Low molecular weight CKs (e.g., CK18, and CK19) are specifically expressed in luminal normal human prostate cells and prostate cancer, whereas high molecular weight CKs (e.g., CK5 and CK10) are identified in basal normal prostate cells and rarely observed in cancer cell populations (Ali et al., 2008). In the present invention, Docetaxel resistant cells showed a significant decrease in both gene transcription and protein expression of low molecular weight CKs. DU145-DR cells showed a 6.25 and 16.6 fold decrease in the protein expression of CKs 19 and 18 respectively. Similarly, 22RV1-DR showed a 14.3 and 6.7 fold decrease in such CKs when quantified and compared to the sensitive parental cells. Immunofluorescence staining of CK19 and CK18 confirmed their decreased expression in the Docetaxel resistant cells (Figure 2). Moreover, high molecular weight CKs continued to be undetectable in the Docetaxel resistant cells as in their corresponding parental cells (data not shown), indicating that in the process of acquirement of Docetaxel resistance, cells lose epithelial differentiation markers and do not undergo a shift from a luminal (low molecular weight CK) to a basal-like (high molecular weight CK) phenotype. Furthermore, 22RV1 cells, which express prostate-related differentiation markers including AR, PSMA and PSA, showed a dramatic decrease of gene and protein expression levels of such markers when exposed to Docetaxel (Figure 2). 22RV1-DR cells showed a 16.6, 20.0 and 11.1 fold decrease in the protein expression of AR, PSMA and PSA respectively. These results were further confirmed by immunofluorescence analysis (data not shown).

[0097] Regarding mechanisms of immune evasion, the inventors found that Docetaxel resistant cells showed deregulation of MHC class I molecules. It was observed that MHC class I antigens, which are critical for efficient antigen presentation to cytotoxic T lymphocytes and subsequent tumor cell lysis, were down- regulated at both gene transcription and protein level (Figure 2). Gene expression profiling revealed a significant down-regulation in all MHC class I antigens (A, B, C, E, F, G), a fact that was confirmed at the protein level by immunoblotting and immunofluorescence staining of MHC class I antigens A, B, C (Figure 2). Thus, these cells exhibit a phenotype that favors immune evasion, making them undetectable by the host immune system.

[0098] Regarding studies on stem cell phenotype, the identified deregulated expression of WNT/p-catenin, Notch and Hedgehog was characterized. These pathways have been implicated in self-renewal and differentiation of progenitor cells (Katoh et al., 2007; McDonald et al., 2006; van den Brink et al., 2004; Radtke et al., 2006; Leong et al., 2008; and Grigoryan et al., 2008). In the prostate, these signaling pathways play essential roles in developmental patterning, epithelial regeneration, and prostate cancer tumorigenesis (Wang et al., 2006, Karhadkar et al., 2004). Docetaxel resistant cells showed a significant decrease in both gene transcript and protein levels of the WNT inhibitor Dickkopf-I (DKKI), a well known inhibitor of the WNT/p-catenin signaling network (Fedi er al., 1999). This decrease in DKK1 expression was linked to an increase in the expression of de-phosphorylated (active) β-catenin, which is the major key effector of WNT signaling. Immunofluorescence analyses demonstrated that parental Docetaxel sensitive cells displayed a membranous expression of β-catenin, associated to its function as an adhesion molecule, whereas Docetaxel resistant cells showed a pronounced nuclear localization of this protein (Figure 2), reported as necessary for the activation of the canonical WNT signaling pathway (Willert ef a/., 2006). Moreover, Docetaxel resistant cells also exhibited an increase in the NOTCH signaling network. NOTCH2 gene transcript levels were significantly increased in the resistant cells and were linked to an increase in cleaved Notch2 protein expression that was associated with nuclear translocation of the protein, where it exerts its transcriptional activity (Figure 2). Finally, Docetaxel resistant cells had an increased expression of the Hedgehog receptor Patched and the glioma associated oncogene homolog transcription factors, Gli1 and Gli2. These findings were associated with an increased protein expression and nuclear translocation of the above mentioned transcription factors (Figure 2), a condition that has been related to Hedgehog pathway activation. Surprisingly, other reported stem cell surface markers, such as CD44 and CD133, were not found to be up-regulated in this study, as analyzed both at the gene transcript and protein levels (data not shown).

[0099] Based on the fact that the Docetaxel resistant cells exhibited a phenotype consistent with sternness, whether these cells displayed a higher tumor initiating capacity than the parental cells was tested. Subcutaneous injection in NOD/SCID mice of 10² and 10³ Docetaxel resistant cells (22RV1-DR and DU145- DR) gave rise to significantly more tumors than their parental sensitive cells (Figure 2e). Injection of 10² DU145-DR and 22RV1-DR cells induced tumor formation in 83.3±7.8% and 79.0+10.4% of the recipients, respectively; whereas the injection of parental DU145 and 22 V1 cells induced tumor formation in only 38.8+15.2% and 46.0±9.3%, respectively (p<0.0001). Moreover, although there was no difference in tumor formation when 10⁴ parental and resistant cells were injected, tumor latency was significantly shorter for the resistant cells. The tumor latency for DU145 parental cells was 59.2±4.9 days versus 43.2±2.6 days for resistant cells (p<0.0001). Similarly, the tumor latency was significantly (p<0.0001) longer for 22RV1 cells (54.9+1.5 days) as compared to the resistant cells (35.3±1.9 days). Thus, the sternness molecular signature of Docetaxel resistant cells was functionally reinforced by their high tumor initiating capacity.

EXAMPLE 2

IDENTIFICATION AND CHARACTERIZATION OF PROSTATE CANCER STEM-CELLS

[01001 Considering the sternness signature and the higher tumor initiating capacity of the newly generated Docetaxel resistant cells, the inventors investigated whether the chemoresistance phenomenon was due to a transition of sensitive to resistant cells with sternness characteristics, or alternatively if chemotherapy would select for pre-existing prostate cancer stem cells (Figure 3c). Because DU145 and

22RV1 Docetaxel resistant cells commonly displayed down-regulation of CK19 and

CK18, and MHC class I antigens, whether cells with a CK-negative/HLA class I- negative phenotype were already present in the parental lines was determined.

Immunofluorescence staining revealed the presence of a small CK-negative/HLA class l-negative subpopulation in both cell lines, which represented a 2.19 ± 0.95% and 3.58 ± 0.79% of the total population of DU145 and 22RV1 parental cells, respectively when quantified by flow cytometry (Figure 4b). [0101] The inventors then studied whether the identified CK-negative/HLA class l-negative tumor cells could survive Docetaxel exposure, thus being responsible for the acquired chemoresistance phenomenon. For this purpose, DU145 parental cells were stably transfected with a plasmid containing the promoter of CK19 driving the expression of the green fluorescence protein (GFP). (Figures 14a and 14b) DU145 parental cells were stably transfected with a plasmid containing the promoter of CK19 driving the expression of the green fluorescence protein (GFP) (Figure 6a). Co-expression of CK19 and GFP was confirmed by flow cytometry and immunofluorescence (Figure 6b). Cells that expressed CK19 were GFP positive (GFP+), whereas cells that did not express CK19 were GFP negative (GFP-). Furthermore, stable insertion of the promoter construct in CK19/GFP negative cells was confirmed by PCR (Figure 6C). In addition, it was demonstrated that these CK19/GFP negative cells were HLA negative both by flow cytometry and immunofluorescence (Figures 6d).

[0102] Unsorted DU145-CK19 promoter-GFP stable cells were seeded and exposed to Docetaxel (10 nM) and live imaging was performed for periods of up to 48 hours. The resulting movies showed that Docetaxel exposure selected for cells with the GFP-negative phenotype which were able to divide and exit mitosis under therapy, whereas GFP-positive cells died after mitotic arrest (Figure 5b). Flow cytometry analysis revealed that Docetaxel treatment resulted in a significant shift in the proportion of surviving cells based on their respective phenotypes (Figure 5c). While initially cells displaying the GFP-positive phenotype represented 87.5 ± 10.4% of the total population, it was observed that after treatment this phenotype decreased, constituting only 28.3 + 10.6% of the total cell population (p<0.001). In contrast, cells with the GFP-negative phenotype increased proportionally from 6.4 ± 4.3% to 73.1 + 10.2% of the total population after treatment (p<0.001). Colony formation assays of DU145-CK19 promoter-GFP stable cells sorted by GFP/HLA class I expression confirmed these results, since only tumour cells with a GFP- negative/HLA class l-negative phenotype were able to form clones when continuously exposed to Docetaxel. (Figure 5d). Colony formation assays confirmed these results, since only tumor cells with a GFP-negative phenotype were able to form clones when continuously exposed to Docetaxel (Figure 5d). Thus, the results highlight the existence of a cell population with an undifferentiated phenotype (CK-negative/HLA class l-negative) which is spared by standard chemotherapy and could be responsible for tumor relapse.

[0103] It was also demonstrated that the subpopulation of tumor cells with a CK-negative/HLA class l-negative/GFP-negative phenotype had the unique stem cell property to differentiate through asymmetric cell division. When seeding unsorted DU145-CK19 promoter-GFP stable cells, it was observed that through such process GFP-negative cells give rise to daughter cells that enter the program of differentiation (become GFP-positive), while the GFP-negative cell retains its original identity (Figure 6a). In contrast, cells displaying a GFP-positive phenotype divided symmetrically, meaning that all daughter cells continue to possess the same characteristics. Flow cytometry analysis of DU145-CK19 promoter-GFP stable GFP sorted cells cultured during 4 weeks revealed that after this period of time, cells derived from GFP-negative sorted cells were majorly constituted by differentiated (GFP+) cells, whereas cells grown from GFP-positive sorted cells maintained the differentiated phenotype (Figure 6b). Thus this result explains the fact that the CK- negative/HLA class l-negative cells represent a minority of the tumor cell population inside the parental cell lines, whereas large sets of differentiated malignant cells constitute the major part of this total cell population.

[0104] Finally, the functional cancer stem cell property of tumor initiation in the identified subpopulations of cells was addressed. DU145-CK19 promoter-GFP stable cells were sorted by GFP/HLA class I expression and the obtained GFP sorted subpopulations of cells were injected subcutaneously into immunodeficient NOD/SCID mice and only the GFP-negative/HLA class l-negative cells exhibited tumor initiating capacity (Figure 6c). Injection of 10 cells with the GFP-negative phenotype produced tumors in 63.0 ± 14.6 % of recipients while no tumor formation was observed after injection the same amount of GFP-positive cells (Figure 6d).

[0105] In order to further confirm the observed findings in which only the DU145 CK-negative/HLA class l-negative/GFP-negative phenotype cells showed tumor initiating capacity, parental cell lines (DU145 and 22RV1) were sorted based in the expression of surface marker HLA class I antigen and tested their tumor initiating capacity. (Figures 13 and 15). Similar to the results obtained with the GFP- negative cells, only the HLA class l-negative cells exhibited tumor initiating capacity after dilution assays. Injection of 10 HLA class l-negative DU145 and 22RV1 cells produced tumors in 83.3±19.1% and 100% of recipients, respectively, while no tumor formation was observed after 198 days of injection with 10 cells displaying a HLA class l-positive phenotype. Similar results were obtained after serial transplantation from HLA class l-negative generated tumor xenografts (data not shown).

[0106] Thus, the functional cancer stem cell property of tumor initiation was intrinsic to the subpopulation of cells with a CK-negative/HLA class l-negative phenotype. Furthermore, in addition to being the only cells endowed with tumour initiating capacity, HLA class i-negative cells also displayed a statistically significant higher clonability capacity when compared to HLA class l-positive cells (Figure 8). Most importantly, but not surprisingly, tumors generated from xenotransplanted GFP- negative cells displayed a differentiated (GFP-positive/CK-positive/HLA-positive) phenotype (Figure 6c), and retained a small population of GFP-negative/HLA- negative cells that accounted for 3.93±0.85% of the total tumour population (Figure 16).

[0107] Taken together, in vitro and in vivo results identify a unique cell population with a CK-negative/HLA class l-negative phenotype with tumor initiating capacity, which through asymmetrical cell division generates transit amplifying clonogens that overwhelmingly populate the evolving tumor with differentiated cells. Thus, confirming the discovery of a prostate cancer stem cell. Considering all of the results described above, including the sternness signature and the tumor initiating capacity of the newly generated Docetaxel resistant cells, it was hypothesized that chemotherapy induces an enrichment of prostate cancer stem cells (Figure 9a). Since the Docetaxel resistant cells displayed downregulation of epithelial differentiation markers, including CK19 and CK18, the CK-negative population of cells was initially quantified in the parental lines. Flow cytometry analysis revealed that both parental cells possessed scattered CK19 and CK18 negative subpopulations (Figure 6a). 22RV1 cells displayed 3.4% CK19 and 4.7% CK18 negative cells. Similarly, DU145 cells showed 2.1 % CK19 and 2.3% CK18 negative cells. Moreover, 22RV1 had a 2.6% and DU145 had a 1.3% CK negative cells when co-stained with both CKs. Hence the immunofluorescence staining confirmed the presence of both a rare CK negative subpopulation, and a dominant subpopulation co-expressing both CKs. EXAMPLE 3

IDENTIFICATION OF CANCER STEM CELLS IN HUMAN METASTATIC PROSTATE TUMOR SAMPLES

[0108] In view of the above results, the inventors decided to investigate whether the identified prostate cancer stem cell population was present in human prostate cancer tissue samples. Immunohistochemical studies of metastases (n=20) and matched primary (n=6) human prostate cancer tissues revealed a scattered subpopulation of CK-negative tumoral cells (Figure 7a). These cells were also negative for HLA class I antigens (Figure 7b), displayed activation of key developmental transcription factors (Figure 7c) and lacked the expression of prostate-related differentiation markers (Figure 7d) corresponding to the sternness signature previously observed in in vitro studies.

[0109] All human prostate cancer specimens analyzed contained scattered subpopulations of CK-negative (CK18 and CK19) tumor cells, accounting for 0.05% to 0.3% and 0.4% to 1.8% of all tumor cells in primary and metastatic lesions, respectively (Figure 7a). Next, immunofluorescence-based double staining was performed to assess the association between CK expression and the markers of interest. In this analysis, it was consistently observed that CK expression was significantly associated to HLA class I expression (p<0.0001). More specifically, it was observed that the CK-negative tumor population did not express HLA class I antigens in 97.8 ± 0.7% of the cells. Nevertheless, all (100%) of the CK-positive cells displayed a positive HLA class I antigen phenotype (Figure 7b). Furthermore, it was persistently found that CK-negative/HLA-negative tumor cells had a significant (p<0.0001) increase of nuclear expression (activation) of developmental transcription factors when compared to differentiated CK-positive/HLA-positive cells. CK- negative/HLA-negative cells displayed nuclear expression of de-phosphorylated β- catenin in 63.9 + 22.6% of cells, cleaved Notch2 in 72.8 ± 15.1%, Gli1 in 67.5 ± 17.3%, and Gli2 in 67 ± 17.3%, whereas CK-positive/HLA-positive cells expressed nuclear de-phosphorylated β-catenin in only 5.8 ± 11.9% of cells, cleaved Notch2 in 6.7 ± 7.9%, Gli1 in 1.2 ± 7.9%, and GH2 in 1.5 ± 10.6% (Figure 7c). Moreover, it was also observed that the CK-negative/HLA class l-negative tumor cells showed no expression of nuclear AR, whereas CK-positive/HLA-positive cells displayed nuclear AR in 71.8 + 14.3% of the cells (Figure 7d). Thus the fact that prostate cancer stem cells do not display a positive AR phenotype suggests that these cells may not be dependent on a functional AR signaling, which would explain how these cells might be responsible for the observed relapse after hormone-therapy, an issue to be pursued in future studies. Taken together, these results confirm the existence and the ability to identify a subpopulation of prostate cancer stem cells in human prostate cancer tissue samples.

[0110] Given that the cancer stem cell population was identified in both primary and metastatic prostate cancer, the inventors then designed a series of experiments aimed at investigating the tumorigenic ability of such population of cells from fresh human tumor samples. To assess whether prostate cancer cells with such phenotype were responsible for tumor initiation, tumors from 48 patients who underwent radical prostatectomy for primary prostate cancer were obtained. (Table 4). Cells were isolated by flow cytometry based in the expression of the cell surface marker HLA class I (Figure 9a). This analysis showed that all primary prostate cancer samples were mainly constituted by HLA class l-positive cells which accounted for a median of 98.9 ± 0.35% (range 98.7%-99.5%) of the total population whereas only a small percentage of cells median 1.2 ± 0.65% (range 0.5%-1.5%) showed an HLA class l-negative phenotype, being this result in agreement with the above reported immunohistochemical findings. Next, the same number (10, 10² and 10³) of HLA class l-negative, HLA class l-positive and unsorted cells mixed with Matrigel were injected into NOD/SCID mice. Overall, 4 (8.3%) of 48 xenograft tumors developed from the injection of primary prostate cancer cells after a median follow-up time of 34.5 weeks (range 21.0-45.3). 20.8 weeks (range 9.3-39.6), and only the HLA-negative cells could maintain their tumorigenic potential following serial transplantation when compared to the HLA-positive cells (Figure 9b and Table 5 below).

TABLE 4

Clinico-pathological characteristics and percentage of HLA-negative

cells in the 48 injected fresh human primary prostate tumours

15 pT3b,pN0 8 (4+4) 6.2 1.4

16 pT3a,pN0 9 (4+5) 5.9 1.2

17 pT3a,pN0 7 (3+4) 4.5 1.2

18 pT3a,pN0 7 (3+4) 5.92 0.6

19 pT3b,pNl 8 (4+4) 24 1.5

20 pT3b,pN0 7 (3+4) 9.8 1.3

21 pT2c,pN0 7 (3+4) 3.1 1.5

22 pT2c,pN0 6 (3+3) 4.8 1.1

23 pT3a,pN0 7 (3+4) 9.1 1.4

25 pT2c,pN0 7 (3+4) 4.1 0.5

26 pT3a,pN0 9 (5+4) 8.4 1.1

27 pT4,pNl 9 (4+5) 30 0.8

28 pT2c,pN0 7 (3+4) 8 1.5

29 pT2a,pN0 6 (3+3) 2.4 1.2

30 pT2c,pN0 7 (3+4) 5.4 0.6

31 pT2a,pN0 7 (3+4) 1 1.2

32 pT2c,pN0 7 (3+4) 4.32 1.5 33 pT2c,pN0 7 (3+4) 5.93 1.5

34 pT2c,pN0 7 (3+4) 7.1 0.6

35 pTla,pN0 7 (3+4) N/A 0.5

36 pT2c,pN0 7 (3+4) 7 1.4

37 pT2c,pN0 7 (3+4) 6 1.4

38 pT2c,pN0 7 (3+4) 8.8 1.3

39 pT2c,pN0 7 (3+4) 3.9 1.5

40 pT2c,pN0 7 (3+4) 5.4 1.2

41 pT3a,pNl 7 (4+3) 21.4 1.5

42 pT2c,pN0 7 (4+3) 7.4 1.5

43 pT2c,pN0 7 (3+4) 5.52 1.1

44 pT2c,pN0 7 (3+4) 2.24 0.5

45 pT2c,pN0 6 (3+3) 7.1 0.7

46 pT2b,pN0 8 (3+5) 6.22 1.5

47 pT2a,pN0 6 (3+3) 4.6 0.9

48 pT2c,pN0 6 (3+3) 2.5 1.2

Note: Highlighted in grey cases from which tumour xenografts developed

[0111] Table 5 summarizes the tumour initiating capacity measured by tumour incidence (tumours/injections) and tumour latencies in weeks (mean ± SD), when 10, 100 and 1000 HLA class I sorted and unsorted cells from primary prostate cancer tissues were injected. Four mice for each sorted cell population and cell dilution were injected twice in the upper flanks (HLA-negative) and lower flanks (HLA-positive). Unsorted cells from each tumour specimen were also injected.

TABLE 5 Primary injections Secondary injections

Tumours/ Tumour latency Tumours/Injections Tumour latency Injections

Patient HLA class Cells/injection Cells/injection Cells/injection Cells/injection

I

expression

100 1000 100 1000 100 1000 100 1000

# 5 negative 5/8 7/8 20.2±3.4 12.6±3.9 8/8 8/8 19.9±2.7 13.8+2.8 positive 0/8 3/8 - 26.8+4.0 0/8 1/8 - 23.8 unsorted 2/8 5/8 27.8±4.6 21.0±2.5 4/8 7/8 23.413.5 23.116.4

#9 negative 7/8 8/8 15.5±3.2 10.8±1.5 7/8 8/8 14.212.5 10.3+1.5 positive 0/8 1/8 - 22.4 0/8 0/8 - - unsorted 2/8 6/8 22.0+2.8 15.3+3.6 1/8 8/8 23.5 14.7+3.2

#12 negative 6/8 8/8 19.4±2.0 12.6±3.6 8/8 8/8 20.3+4.5 16.612.5 positive 0/8 0/8 - - 0/8 0/8 - - unsorted 1/8 5/8 23.4 22.2+3.6 0/8 3/8 - 19.5+1.5 # 24 negative 3/8 5/8 25.3+4.9 20.9±5.4 0/8 2/8 - 15.2+3.9 positive 0/8 0/8 - - 0/8 0/8

unsorted 0/8 2/8 26.5±3.5 0/8 0/8 - -

[0112] After primary injection of 10³ cells, it was observed that a significant higher tumor initiating capacity was from the HLA-negative sorted cells (87.5±17.6%) when compared to the HLA-positive cells (12.5±17.6%), a fact that was linked to a lower tumor latency in the HLA-negative xenografts (14.2±4.5 versus 24.6±3.1 weeks). Moreover, further dilution into 10² and 10³ cells revealed that only the HLA- negative cells were endowed with tumor initiating capacity. These observations were confirmed by secondary transplantation experiments in which injection of 10² and 10³ sorted cells continued to demonstrate that tumor development was restricted to the HLA-negative subpopulation of cells (Figure 9b). Secondary injections of HLA sorted cells from the rarely xenografts generated from HLA-positive cells confirmed this result, since HLA-positive cells did not possess tumour initiating capacity. Histological and immunohistochemical analyses revealed that tumors derived from HLA-negative sorted cells faithfully reproduced the phenotype of the original primary human tumor (Figure 9c), showing expression of HLA class I antigens in the majority of their tumor cells, as well as epithelial and prostate related markers (CK and AR). Thus these results show that the HLA-negative population is enriched in cells capable of initiating prostate cancer xenografts in NOD/SCID mice, and reproducing the molecular and phenotypic heterogeneicity distinctive of most human cancers.

[0113] Moreover, it appears that this is a universal phenomenon, because the inventors were able to isolate HLA-negative tumor cells from a variety of fresh human solid neoplasms, including colon, breast, lung and bladder carcinomas which are responsible for tumor initiation when serially injected into NOD/SCID mice (Figure 9d, and Table 6 below ). Overall, 4 of 10 (40%) colon, 2 of 10 (20%) lung, 2 o _dde_rf 12 (16.6%) breast and 2 of 18 (11.1%) bladder xenograft tumors developed from injection of fresh human tumor samples. Characteristics of other primary solid tumor types are shown in Figure 7 below. Tumor initiating capacity and tumor latencies of HLA sorted cells from other solid cancer tissue types are shown in Table 8 below.

TABLE 6

Primary injections Secondary injections

Patient HLA Tumors/ Tumor Tumors/ Tumor class I

#3 negative 6/8 16.0±3.1 8/8 18.3±4.6 positive 0/8 - 0/8 -

#5 negative 8/8 8.5±3.3 8/8 7.0±2.5 c positive 2/8 28.5±2.1 0/8 - o #6 negative 8/8 10.3±3.5 8/8 9.9±4.6 o positive 0/8 - 0/8 -

# 9 negative 7/8 10.4±3.9 8/8 11.0±3.6 positive 0/8 - 0/8 -

# 4 negative 5/8 15.3±2.9 8/8 12.9±2.6 c positive 0/8 - 0/8 -

3

_J # 8 ■ icyc ivo 7/8 11.9±3.2 8/8 10.3±2.1

positive 1/8 27.0 0/8 -

** # 2 negative 8/8 6.9±2.9 8/8 6.5±3.3

V)

flj positive 0/8 - 0/8 -

£

m # 8 negative 4/8 17.015.8 6/8 16.8±5.2

positive 0/8 - 0/8 -

# 11 negative 4/8 15.7±5.4 5/8 16.6±3.2

positive 0/8 - 0/8 -

<e # 16 negative 8/8 7.9±2.7 8/8 7.5±3.9 m

positive 3/8 13.7±4.0 0/8 -

TABLE 7

Characteristics of other primary solid tumour types.

1 lung Large cell carcinoma Poorly differentiated pT2,pNl 0.8 lung Adenocarcinoma Moderately differentiated ρΤΙ,ρΝΟ 1.2 lung Squamous cell carcinoma Moderately differentiated ρΤΙ,ρΝΟ 1.0 lung Adenocarcinoma Moderately differentiated pT2,pN0 0.9 lung Adenocarcinoma Moderately differentiated pT2,pNl 1.3 lung Adenocarcinoma Moderately differentiated pT3,i 1.4

9 lung Adenocarcinoma Moderately differentiated pT2,pN0 0.7

0 lung Adenocarcinoma Poorly differentiated pT2,pN0 1.2

breast Ductal adenocarcinoma Moderately differentiated

breast Ductal adenocarcinoma Moderately differentiated pTlc,pN0 1.4

4 breast Ductal adenocarcinoma Poorly differentiated pT2,pNl 1.5

5 breast Ductal adenocarcinoma Moderately differentiated pTlc,pNlc 0.6

6 breast Ductal adenocarcinoma Poorly differentiated pTlc,pNlmi 0.8

7 breast Ductal adenocarcinoma Moderately differentiated sTl c,pN0 1.0

breast Ductal adenocarcinoma Moderately differentiated pTlcpNO 0.6

0 breast Ductal adenocarcinoma Poorly differentiated pT2,pN0 1.4

1 breast Ductal adenocarcinoma Moderately differentiated pT2,pN0 0.9

2 breast Ductal adenocarcinoma Moderately differentiated pTlcNO 1.2

1 bladder Urothelial carcinoma High grade pTa,pN0 0.5

2 bladder Urothelial carcinoma High grade pTla,pN0 0.7

bladder Urothelial carcinoma High grade pT3,pN0 1.4 bladder Urothelial carcinoma High grade pT4a,pN0 1.5

5 bladder Urothelial carcinoma Low grade pTa,pN0 0.9

6 bladder Squamous cell carcinoma Poorly differentiated pT4,pN0 1.5

7 bladder Urothelial carcinoma High grade pTa,pN0 0.6

8 bladder Urothelial carcinoma High grade pT3a,pN0 1.4

bladder Urothelial carcinoma Low grade pTa,pN0 0.7

0 bladder Urothelial carcinoma High grade pT3a,pN0 1.4

2 bladder Urothelial carcinoma High grade pT3a,pN0 1.4

3 bladder Urothelial carcinoma High grade pT3,pN0 0.9

4 bladder Urothelial carcinoma High grade pTl, pNO 0.7

5 bladder Urothelial carcinoma High grade jT3a,pN0 1.3

bladder Urothelial carcinoma High grade pT3a,pN0 1.3 bladder Sauamous cell carcinoma Well differentiated ρΤΙ, ρΝΟ 0.6

Note: Highlighted in grey cases from which tumour xenografts developed

TABLE 8

Tumour initiating capacity and tumour latencies of HLA

sorted cells from other solid cancer tissue types

Primary injections Secondary Injections

Patient HLA class I Tumours/ Injections Tumour latency Tumours/Injections Tumour latency

#3 negative 6/8 16.0±3.1 8/8 18.3±4.6 positive 0/8 0/8

o

υ #5 negative 8/8 8.5±3.3 8/8 7.0±2.5

positive 2/8 28.5±2.1 0/8 negative 8/8 10.3±3.5 8/8 9.9±4.6 positive 0/8 0/8

negative 7/8 10.4±3.9 8/8 1 1.0±3.6 positive 0/8 0/8

negative 5/8 15.3±2.9 8/8 12.9±2.6 positive 0/8 0/8

s

s

-J negative 7/8 1 1.9±3.2 8/8 10.3±2.1 positive 1/8 27.0 0/8

negative 8/8 6.9±2.9 8/8 6.5±3.3 positive 0/8 0/8

« negative 4/8 17.0±5.8 6/8 16.8±5.2 positive 0/8 0/8

# 11 negative 4/8 15.7±5.4 5/8 16.6±3.2 positive 0/8 0/8

Ό

•a

# 16 negative 8/8 7.9±2.7 8/8 7.5±3.9 positive 3/8 13.7±4.0 0/8

[0114] Table 8 summarizes the tumour initiating capacity measured by tumour incidence (tumours/injections) and tumour latencies in weeks (mean ± SD), when 100 HLA class I sorted cells from other primary tumour tissue types were injected. Four mice for each sorted cell population and cell dilution were injected twice in the upper flanks (HLA-negative) and lower flanks (HLA-positive).

[0115] After primary injection, tumor initiating capacity of 10² HLA-negative cells was significantly higher when compared to HLA-positive sorted cells. More importantly, following serial transplantation only the HLA-negative cells retained tumorigenic capacity, whereas HLA class l-positive cells did not. Moreover, secondary injections of HLA-positive cells sorted from xenografts generated from HLA-positive cells did not possess tumour initiating capacity. The generation of tumours from HLA class l-positive cells could be could occur because of possible contamination of HLA class l-negative cells during cell sorting, although it cannot be excluded that HLA class l-positive cells may have a low tumour initiating capacity that cannot be maintained after serial transplantation. Histological analysis of the developed xenograft tumors showed similar morphological characteristics than their corresponding human primary cancers (Figure 9e). Thus, the identification of such HLA-negative population with tumor initiating capacity in human epithelial tumor types in addition to prostate cancer highlights the importance of the inventor's discovery.

[0116] Finally, the tumor initiating capacity of cells from fresh human tissue samples was not related to any of the analyzed clinico-pathological characteristics of the cancer patients, neither associated with the percentage of HLA-negative cells. Tumours with aggressive clinico-pathological characteristics (e.g., high grade, high tumour stage) or tumours with a high number/percentage of HLA-negative cells (e.g., 1.5%) did not possess a significantly higher tumour initiating capacity. Moreover, no association was observed between the percentage of HLA-negative cells and either tumour stage or tumour grade.

EXAMPLE 4

TARGETING THE HUMAN CANCER STEM CELL

[0117] Based on the fact that the identified cancer stem cells exhibited Hedgehog and Notch pathways activation, the inventors next tested whether inhibition of such pathways could impair cancer stem cell homeostasis. For this purpose, the inventors utilized Cyclopamine, a plant derived hedgehog pathway antagonist that acts at the level of Smo (Taipale et al., 2000; Karhadkar et al., 2004; Chen et al., 2002), and Compound E, a highly active gamma-secretase inhibitor that blocks the proteolytic processing of Notch receptors (Seiffert ef al., 2000), for the in vitro experiments. DBZ, a highly active gamma-secretase inhibitor with established in vivo activity, was used as a substitute for Compound E in the in vivo experiments outlined below (van Es et al., 2005).

[0118] Exposure of the CK-negative/HLA-negative cancer stem cell population from DU145 and 22RV1 to each individual inhibitor did not induce any major effect. However, when both agents were administered together a robust inhibition in cell cycle progression with an accumulation of cells in sub-G1 was observed, whereas this effect was minor in differentiated (CK-positive/HLA-positive) cells (Figure 10a). Because dexamethasone is used to reduce gut toxicity of gamma-secretase inhibitors in in vivo experiments (Real et al., 2009), the inventors analyzed the in vitro effects of dexamethasone in combination with the above mentioned inhibitors. This analysis showed that dexamethasone did not change the cell cycle effects of Hedgehog and Notch inhibitors (data not shown). Moreover, the combined effect of these drugs was further confirmed by colony formation assays, since no colony formed after 21 days when Hedgehog and Notch inhibitors were combined (Figure 10b). Next, to determine whether the inhibition of these pathways could affect the tumor initiating capacity of the cancer stem cells in vivo, 10³ DU145 and 22RV1 CK- negative/HLA-negative cells were injected subcutaneously into NOD/SCID mice and treated with vehicle solution (Control), dexamethasone alone, dual combination (e.g. dexamethasone plus Cyclopamine) or triple combination (dexamethasone plus Cyclopamine plus DBZ) of the drugs. Mice treated with the triple combination showed a significant (p<0.05) delay in tumor first palpability of 5.1 weeks in DU145 and 3.8 weeks in 22RV1 xenografts, whereas this tumor delay was not significant in mice treated with Hedgehog or Notch inhibitors alone. (Figure 10c). [0119] Finally, the tumor initiating inhibitory effects of these compounds was tested in xenografts derived from fresh human prostate cancer tissues. 10³ HLA- negative sorted cells from xenografts #5, #9 and #12 were injected into NOD/SCID mice and treated with the same schedules and combinations of Hedgehog and Notch inhibitors, as described above. As observed previously in the cell lines, only the combined treatment with Hedgehog and Notch inhibitors significantly (p<0.05) delayed tumor initiation (Figure 10d). Tumor xenografts in mice treated with the triple combination of drugs were first palpable after 18.3±3.1 , 13.8±2.5 and 20.1±3.3 weeks, compared to 13.8±2.5, 10.3±1.5 and 16.6±2.5 weeks in their corresponding controls.

[0120] Taken together, these results show that the inhibition of these developmental pathways targets the cancer stem cell population delaying tumor initiation. Thus this finding supports the design of clinical trials in prostate cancer patients utilizing combinations that would include targeting of these critical developmental pathways, and also the generation of new compounds that inhibit these pathways with a higher efficacy which could contribute to improve these results.

[0121] Two major hypotheses regarding tumor initiation have been postulated. The "stochastic model" which predicts that every neoplastic cell can generate an entirely new tumor; and the "cancer stem cell model" which proposes that tumor cells exist in a hierarchical state, and that only a few stem cells possess tumor initiating potential. Here the inventors disclose for the first time the identification and functional characterization of a human cancer stem cell, which fulfills the following sternness criteria: 1) self-renewal and differentiation through asymmetrical cell division, 2) tumor initiating capacity, 3) a negative histocompatibility signature, and 4) a multidrug resistance phenotype.

[0122] An HLA-negative phenotype is also shared by embryonic stem cells. It has been reported that human pre-implantation embryos are HLA class I and class II negative (Desoye et al., 1988). This phenomenon precludes rejection based on expression of paternal antigens, until a blood-tissue barrier develops, in this situation being the placenta. In the context of cancer stem cells, such a histocompatibility negative phenotype has major clinical implications, since it explains host mutation permissiveness, as well as tumor spread and metastatogenic capabilities, since cancer stem cells would escape immune-surveillance.

[0123] Concerning tumor initiation capacity, the present invention discloses that only the identified CK-negative/HLA-negative cancer stem cells generate tumors in vivo, while the differentiated, CK-positive/HLA-positive progenies lack such property. However, it appears that these cancer stem cells exhibit genetic memory independent of certain stroma interactions, since subcutaneous injections confer the tissue-of-origin phenotype without the need for orthotopic implantation, a phenomenon that needs to be further investigated. Moreover, due to its homogenous phenotype, the inventors hypothesized that these cells are genetically stable, a property facilitated by their quiescent state and asymmetrical division. This genetic stability would allow for the identification of "driver" mutations, since molecular heterogeneity would be essentially a product of the tumor expanding and differentiated cancer cell populations, and probably not as critical for tumorigenesis. A new molecular classification of human tumors could probably be derived from analysis of such "driver" mutations in these cancer stem cell populations. [0124] In sum, the inventors have identified and characterized a newly defined prostate cancer stem cell in human cancer cell lines and tissue samples. Moreover, this population of cells was isolated using HLA class I surface marker, and its tumor initiating capacity was demonstrated. Further, it was observed that similar CSC populations are also present in human breast, colon, lung and bladder carcinomas, fact that gives further universality to these findings. The discovery of this human cancer stem cell has important clinical implications in diagnostic and predictive laboratory assays, as well as for development of novel therapeutic strategies. In this context, treatment with Notch and Hedgehog inhibitors attenuates tumor formation in experimental animal models.

EXAMPLE 5

MATERIALS AND METHODS

Inhibitors and drugs

[0125] Docetaxel, Mitoxantrone, Doxorubicin, Cisplatin, Vinorelbine, Paclitaxel, Dexamethasone, Cyclopamine and Compound E were obtained from Sigma-Aldrich (St. Louis, MO). DBZ [(2S)-2-[2-(3,5-difluorophenyl)-acetylamino]-N- (5-methyl-6-oxo-6,7-dihydro-5H-dibenzo[b,-d]azepin-7-yl)-propionamide] was obtained from Syncom (Groningen, The Netherlands).

Cell culture, generation and characterization of acquired Docetaxel resistant cells

[0126] Human hormone-independent prostate cancer cell lines, DU-145 and 22RV1, were obtained from American Type Culture Collection (ATCC) and maintained in RPMI 1640 medium (Gibco, Invitrogen Corp., Carlsbad, CA) supplemented with 10% FBS without antibiotics. Cells were grown at 37°C in a humidified atmosphere with 5% C0₂. DU145 and 22RV1 cells were selected in order to generate a prostate cancer Docetaxel resistance model. This selection was based on the fact that both cell lines are hormone-refractory, a condition treated with Docetaxel in the clinical setting, and that they also exhibit distinct hormone-refractory phenotypes. While 22RV1 cells are still dependent of androgen receptor signaling and express prostate specific markers (e.g., PSMA, androgen receptor), DU145 do not. Thus, the generation of acquired Docetaxel resistance in these two phenotypically distinct cell lines facilitated an approach to study the molecular processes involved in the acquisition of Docetaxel resistance. Docetaxel resistant clones, DU-145-DR and 22RV1-DR, were selected by culturing cells with Docetaxel in a dose-escalation manner. Initial culture was at 5 nM Docetaxel. After the sensitive clones were no longer present and the surviving DU- 45 and 22RV1 cells repopulated the flask, the concentration of Docetaxel was increased to 10 nM and subsequently to 25 nM, 50 nM, 100 nM and 250 nM. 22RV1-DR cells were further exposed to 500 nM Docetaxel. After exposure to each increasing dose of Docetaxel, the remaining surviving cells were maintained in culture medium containing the last selection escalating dose of Docetaxel. The last drug selection concentration at which the cells were exposed was 250 nM for DU-145-DR and 500 nM for 22RV1- DR, in order to avoid reversibility of the acquired Docetaxel resistance phenotype. The process of acquired drug resistance took 9 months for DU-145-DR and 6.5 months for 22RV1-DR. In parallel, parental DU-145 and 22RV1 cells were exposed to DMSO (vehicle solution of Docetaxel) in the same dose-escalation manner.

[0127] Cells were exposed to increasing doses of Docetaxel, and the acquired chemoresistance was characterized and confirmed by cell viability, colony formation, annexin V, and poly-(ADP-ribose) polymerase (PARP) cleavage assays (Figures 1a- d). The generated Docetaxel resistant (DR) cells showed cross-resistance to DNA damaging drugs (Mitoxantrone, Doxorubicin and Cisplatin), as well as other antimitotic agents (Vinorelbine and Paclitaxel), consistent with a multidrug resistant phenotype (data not shown). It was also observed that the Docetaxel chemoresistance was reversible (Figures 12a and b). Removal of the drug from culture medium induced a significant decrease in drug resistance. After 12 weeks of culturing Docetaxel resistant cells without drug, cell viability decreased significantly. Docetaxel IC-50 concentrations for DU145-DR cells decreased from 1μΜ to 25 nM and in 22RV1-DR cells IC-50 drug concentrations decreased from 10μΜ to 50nM (Figure 12a). This decrease in docetaxel resistance after drug removal was confirmed by colony formation assays (Figure 12b). Thus, in order to maintain the Docetaxel resistance phenotype, cells were continuously cultured under the last drug selection concentration.

Cell viability and colony formation assays

[0128] Cell viability was analyzed using the Cell titer 96 Aquos Non-Reactive Cell Proliferation Assay (MTS) kit (Promega Corp., Madison, Wl). Cells were seeded at a density of 10⁴ in 96-well culture dishes and 24 hours later medium was removed and replaced with new medium alone (control) or medium containing drugs. After 72 hours, color absorbance was measured on a microplate spectrophotometer (Molecular Dynamics) at 450 nm (test wavelength) and 620 nm (reference wavelength). The percentage of surviving cells was estimated by dividing the A 450 nm - A 620 nm of treated cells by the A 450 nm - A 620 nm of control cells. Clonogenic survival assays in response to drug treatment were performed by plating 10³ cells in 35 mm culture dishes. After 24 hours, cells were left untreated (control) or treated with drugs. Next day, medium was changed and the cells kept growing in fresh medium without any drug or under continuous exposure to drugs. For this continuous exposure experiments, medium plus drugs was replaced every 3 days until clones of drug-resistant cells appeared. Cells were then fixed with 4% paraformaldehyde in PBS, stained with crystal violet solution and formed colonies were visually counted.

Analysis of apoptosis by flow cytometry

[0129] Cells (10⁵) were left untreated (control) or treated with drugs for 72 hours. Adherent and detached cells were pooled, washed and labeled with annexin- V-FITC and propidium iodide using the annexin-V-FLUOS Staining Kit (Roche, Nutley, NJ) according to manufacturer's instructions. Samples were acquired with a FACscan Flow Cytometer (BD Biosciences, San Jose, CA) and analyzed with CellQuest Pro software (BD Biosciences) to determine the percentage of cells displaying annexin V staining.

Cell cycle analysis

[0130] Cells were treated with drugs for 72 hours, harvested and fixed in 70% ethanol and stored at 4°C. Before analysis, cells were washed with PBS, centrifuged and incubated for 30 min at room temperature in a staining solution containing 0.1% Triton X, 0.2 mg/ml RNAse and 0.02 mg/ml propidium iodide. DNA content was acquired with a FACscan Flow Cytometer (BD Biosciences) and analyzed with CellQuest Pro software (BD Biosciences).

cDNA microarray analysis

[0131] 22RV1, 22RV1-DR, DU-145 and DU-145-DR gene expression profiles were analyzed. Total RNA from each sample was isolated by Tryzol (Invitrogen) and purified by RNeasy mini kit and RNase-free DNase set (Qiagen Inc., Valencia, CA) according to the manufacturer's protocols. RNA quality of all samples was tested by RNA electrophoresis and RNA LabChip analysis (Agilent Technologies, Inc., Santa Clara, CA) to ensure RNA integrity. Samples were prepared for analysis with Affymetrix Human U133 Plus 2.0 arrays according to the manufacturer's instructions. Gene expression levels of samples were normalized and analyzed with Microarray Suite, icroDB, and Data Mining tool software (Affymetrix, Santa Clara, CA). The absolute call (present, marginal, or absent) and average difference of 22.215 expressions in a sample, and the absolute call difference, fold change, average difference of gene expression between two or three samples were normalized and identified using this software package. Statistical analysis of the mean expression average difference of genes, which show >2-fold change based on a log normalization, was done using a t test between Docetaxel sensitive and resistant samples. Genes that were not annotated or not easily classified were excluded from the functional clustering analysis.

Gene ontology analysis

[0132] Genes differentially expressed in the Docetaxel resistant cells compared to the parental sensitive cells generated a list of commonly deregulated transcripts. This list was assessed by the DAVID Bioinformatics Resources, a web- based statistical hypergeometric test applied for enrichment analysis of gene ontology (GO) categories, which are, biological process, molecular function, and cellular component. GO categories enriched on the highest hierarchical level (≥ level 5) at statistical significance (p<0.01) were taken into consideration.

Western Blot analysis

[0133] Whole cell extracts were prepared in sample buffer and analyzed by immunoblotting. Primary antibodies against poly (ADP-ribose) polymerase (PARP) (BD Pharmingen, San Jose, CA), cleaved PARP (BD Pharmingen), cytokeratin 19 (Abeam), cytokeratin 18 (Abeam, Cambridge, MA), androgen receptor (Sigma- Aldrich), prostate specific membrane antigen (Abeam), prostate specific antigen (Epitomics, Burlingame, CA), pan-HLA class I (Abeam), DKK1 (Orbigen, BioCarta LLC, San Diego, CA), activated β-Catenin (Millipore, Billerica, MA), β-Catenin (BD Transduction), activated Notch2 (Abeam), PTCH (Abeam), GN1 (Santa Cruz Antibody, Santa Cruz, CA), GH2 (Abeam), and β-Actin (Sigma-Aldrich) were used in immunoblot assays using standard procedures. Protein expression was quantified by comparing band expression using Quantity One software (Bio-Rad, Hercules, CA). Immunohistochemistry and immunofluorescence analyses

[0134] Immunofluorescence analyses were conducted on prostate cancer cell lines and formalin fixed paraffin-embedded tissue sections from human cancers and tumor xenografts. Primary antibodies included a combination of cytokeratin 19 and 18 (Abeam), pan-HLA class I (Abeam), green fluorescence protein (Abeam) and the following transcription factors: active β-Catenin (Millipore), activated Notch2 (Abeam), Gli1 (Santa Cruz), GN2 (Abeam) and androgen receptor (DAKO, Fort Collins, CO). Secondary antibodies used were Alexa Fluor® 594 (Invitrogen) and Alexa Fluor® 488 (Invitrogen). Prostate cancer cells (10⁵) were plated in 35 mm culture dishes and 24 hours later stained by standard immunofluorescence procedures. Tissue sections (5 pm) were deparaffinized and submitted to standard peroxidase based immunohistochemistry and immunofluorescence procedures. Quantification of the expression of cytokeratins, HLA class I antigen, transcription factors and androgen receptor was performed by evaluating tumoral cells. Percentage of positive and negative cells was determined in 10 high power fields. Generation of the cytokeratin 19-green fluorescent protein (GFP) reporter plasmid

[0135] CK19 gene promoter region was amplified from DU145 cells genomic DNA by PCR with specific primer sets (Fw 5'-AACGCATGCTTTGGGGGGATG-3' (SEQ ID NO: 1) and Rv 5'-TCCCCCTTTACTCGGCCCCCAC-3' (SEQ ID NO: 2)) as described previously (Tripathi et a/., 2005. The PCR products were digested with Ase I and Hind III and cloned into pEGFPNI vector (Clontech, Mountain View, CA) previously digested with the same enzymes. As a result, the CMV promoter was removed from the original vector and the GFP expression was under control of the CK19 promoter. The final construct was confirmed by digestion and sequencing analysis. DU145 cells were transfected with pCK19-GFP construct using Lipofectamine Plus 2000 (Invitrogen). After 24 hours, medium was replaced with fresh medium and stably expressing cells selected in the presence of G418 (Invitrogen). Positive clones were confirmed by direct microscopy and immunofluorescence and also by PCR amplification of GFP coding region using specific primers (Fw 5'-TTCCTGCGTTATCCCCTGATTC-3' (SEQ ID NO: 3) and Rv 5'-GCTCCTCCGGCCCTTGCTCACCAT-3' (SEQ ID NO: 4)).

Live Cell Imaging

[0136] Time-lapse videomicroscopy was used to assess asymmetrical cell division and Docetaxel subpopulation sensitivity of DU145 cells stably transfected with the pCK19-GFP promoter. Cells growing in 6-well plates at low confluence were placed in the stage inside an incubator chamber at 37°C, 50% humidity and in an atmosphere of 5% C0₂. Unattended time-lapse movies of randomly chosen GFP+ and GFP- DU145 cells were performed with a Nikon Eclipse Ti inverted microscope. NIS Elements AR (Nikon Inc., Melville, NY) software was used to collect and process data. Imaging was performed using a 10x objective and images were captured using 200-ms exposure times for GFP and 20-ms for bright field every 30 minutes.

Analysis of subpopulations of cells by flow cytometry

[0137] Flow cytometry analysis of subpopulations of prostate cancer cells were carried out following standard procedures. Intracellular CK19 and CK18 expression was performed in single-cell suspensions fixed with 70% ethanol, whereas the expression of cell surface HLA class I and GFP was determined in fresh cell samples (without fixation). Primary antibodies against CK19 (Abeam), CK18 (Abeam), HLA class I (Abeam), HLA class I conjugated to phycoerythrin (Abeam) and GFP (Abeam) were used. Secondary antibodies, when used, corresponded to Alexa Fluor® 594 (Invitrogen) and Alexa Fluor® 488 (Invitrogen). Samples were acquired with a FACscan Flow Cytometer (BD Biosciences) and analyzed with a CellQuest Pro software (BD Biosciences). A minimum of 10⁴ cells were measured per sample.

Mice procedures

[0138] Animal use and care was in strict compliance with institutional guidelines established by the University of Columbia, Institutional Animal Care and Use Committee. Xenograft experiments were performed with 5-6 weeks old mice (NOD.CB17-Prkdc^SCid) obtained from Jackson Laboratories as recipients.

Human primary and metastatic prostate cancer tissue samples

[0139] Formalin-fixed paraffin-embedded human primary and metastatic prostate cancer tissue samples were provided by the tumor bank of Columbia University Cancer Center. All samples were collected under informed consent and under the supervision of the Columbia University Medical Center Institutional Review Board. Tissue sections with cancer were selected by reviewing Hematoxylin & Eosin (H&E) stained slides.

Tumour initiating capacity of cancer cells from prostate cell lines and fresh human samples

[0140] To compare the tumor initiating capacity of Docetaxel sensitive parental cells and Docetaxel resistant cells, GFP positive and GFP negative sorted cells, or HLA-positive and HLA-negative sorted cells, various numbers of cells (e.g. 10, 10², 10³, 10⁴) were subcutaneously injected in 200 μΙ of medium:Matrigel (1 :1) into male mice. GFP cell subpopulations of prostate cancer ceils were sorted following standard procedures. For HLA class I cell isolation, single suspensions of fresh cells where blocked with PBS + FBS 5% and stained with an HLA class I antibody directly conjugated to phycoerythrin (Abeam). To assess the tumor initiating capacity of human cancer cells from fresh tumor tissue samples, portions of tumors were obtained from patients who underwent surgical procedures at Columbia University medical Center through an Institutional Review Board approved protocol. Forty-eight primary prostate cancers, 10 primary colon cancers, 10 primary lung cancers, 12 primary breast cancers and 18 primary bladder cancers were processed. Specimens were mechanically dissociated and filtered to obtain a single-cell suspension and exposed to red cell lysis buffer (Sigma-Aldrich) to remove red blood cells. Cells were stained with directly conjugated fluorescent antibodies to human CD45 (Abeam), human CD31 (eBiosciences) and human HLA-class I (Abeam). For xenograft tumors, primary fluorescent conjugated antibodies to mouse CD45 (eBiosciences), mouse CD31 (Biolegend, San Diego, CA) and human HLA-class I (Abeam) were used to select live human cancer cells. Cells were suspended in 10 Mg/ml DAPI to label dead cells and sorted on FACSAria Cell Sorting System (BD Biosciences). Different dilutions (10, 100 and 1 ,000 injected cells) of human prostate cancer HLA sorted cells (HLA class l-negative and HLA class l-positive) and unsorted cells, and 00 HLA sorted cells from other human cancer samples (primary injections) and derived xenografts (secondary injections) were injected into NOD/SCID mice. Four mice for each sorted cell population and cell dilution were injected. Four injections were performed in each mouse for sorted cells, two in the upper flanks for HLA class l-negative cells and two in the lower flanks for HLA class l-positive cells. Unsorted cells from each tumour specimen were also injected in NOD/SCID mice. Secondary injections of HLA sorted cells were performed from tumours generated from HLA class l-negative sorted cells and the rarely observed tumours originated from the HLA class l-positive fraction of cells. Tumour initiation was measured by tumor incidence (number of tumors/number of injections) and latency (time from injection to first tumor palpability). Tumour formation was evaluated regularly by palpation of injection sites. In cases where a tumor became palpable at only one injection site, that tumor was surgically removed to allow continued evaluation of other injection sites. Mice were monitored for up to 36 weeks. Animals with no sign of tumor burden were also examined on necropsy to confirm that there was no tumor development. Tumors harvested were fixed in formalin, and paraffin sections were made for H&E staining and immunofluorescence studies when necessary.

In vitro effects of Notch and Hedgehog inhibitors

[0141] The in vitro effects of Notch and Hedgehog inhibitors on HLA-negative and HLA-positive sorted cell lines were analyzed by cell cycle and colony formation assays (described above). Cells were exposed to vehicle solution (Control), dexamethasone (1 μΜ), Cyclopamine (1 μΜ), Compound E (1 μΜ) and a dual (e.g. dexamethasone plus Cyclopamine) or triple (dexamethasone plus Cyclopamine plus Compound E) combination of the drugs.

Effects of Notch and Hedgehog inhibitors in tumor initiation

[0142] To analyze whether the inhibition of these developmental pathways could affect the tumor initiating capacity of the cancer stem cells in vivo, 10³ HLA- negative sorted cells from cell lines and human prostate tumor xenografts were inoculated subcutaneously into NOD/SCID mice. Mice were treated with vehicle solution (Control), dexamethasone (15 mg/kg/ip. daily), Cyclopamine (50 pg/kg/sc daily) plus dexamethasone, DBZ (10 pM/kg/ip. daily) plus dexamethasone or a combination of the 3 drugs. Dexamethasone and Cyclopamine were administered continuously; however, DBZ was administered daily (days 1 to 15 every 4 weeks) in order to avoid gut toxicity. For the in vivo cell lines studies, three independent experiments in 8 mice for treatment arm (e.g., Cyclopamine) were performed, whereas for the human prostate tumours 8 mice were included for each treatment arm. Mice were monitored every day until tumors formed. Animals were sacrificed if they showed any evidence of distress or if they lost more than 20% of their original body weight. Generated tumors were harvested and histologically confirmed.

Characterization of the chemo-resistant phenotype

[0143] Regarding studies on differentiation, the expression of cytokeratins (CKs) as epithelial markers was assessed, as well as prostate-related biomarkers, including AR, PSA and PSMA. CKs have been previously reported to be specific for human differentiated epithelial cells, playing a role in the maintenance of cellular integrity while also functioning in signal transduction and cellular differentiation processes. Low molecular weight CKs (e.g., CK 8, and CK19) are specifically expressed in luminal normal human prostate cells and prostate cancer, whereas high molecular weight CKs (e.g., CK5 and CK10) are identified in basal normal prostate cells and rarely observed in cancer cell populations. In the model of the present invetnion, Docetaxel resistant cells showed a significant decrease in both gene transcription and protein expression of low molecular weight CKs. DU145-DR cells showed a 6.25 and 16.6 fold decrease in the protein expression of CKs 19 and 18 respectively. Similarly, 22RV1-DR showed a 14.3 and 6.7 fold decrease in such CKs when quantified and compared to the sensitive parental cells. Immunofluorescence staining of CK19 and CK18 confirmed their decreased expression in the Docetaxel resistant cells (Figure 2). Moreover, high molecular weight CKs continued to be undetectable in the Docetaxel resistant cells as in their corresponding parental cells (data not shown), indicating that in the process of acquirement of Docetaxel resistance, cells lose epithelial differentiation markers and do not undergo a shift from a luminal (low molecular weight CK) to a basal-like (high molecular weight CK) phenotype. Furthermore, 22RV1 cells, which express prostate-related differentiation markers including AR, PSMA and PSA, showed a dramatic decrease of gene and protein expression levels of such markers when exposed to Docetaxel (Figure 2). 22RV1-DR cells showed a 16.6, 20.0 and 11.1 fold decrease in the protein expression of AR, PSMA and PSA respectively. These results were further confirmed by immunofluorescence analysis (data not shown).

[0144] Regarding mechanisms of immune evasion, it was found that Docetaxel resistant cells showed deregulation of MHC class I molecules. In this context, previous work from our group already reported the identification of MHC class I antigens negative tumour cell subpopulations in human primary and metastatic carcinomas (Cordon-Cardo, et al., 1991). In the present study, it was found that MHC class I antigens, which are critical for efficient antigen presentation to cytotoxic T lymphocytes and subsequent tumour cell lysis, were down-regulated at both gene transcription and protein level (Figure 2). Gene expression profiling revealed a significant down-regulation in all MHC class I antigens (A, B, C, E, F, G), a fact that was confirmed at the protein level by immunoblotting and immunofluorescence staining of MHC class I antigens A, B, C (Figure 2). Moreover, Docetaxel resistant cells showed a down-regulation in gene transcript levels of known NK ligands, such as MICA/B, PVR, and PVRL2, as shown in Table 9. Thus, these cells exhibit a phenotype that favors immune evasion, making them undetectable by the host immune system.

[0145] Natural killer (NK) ligands gene expression that were deregulated in Docetaxel resistant cells (DU145-DR and 22RV1-DR) when compared to their parental sensitive cells (DU145 and 22RV1) are summarized in Table 9. Statistical analysis of the mean expression average difference of genes, which show≥ 2 fold change based on a logarithmic normalization, was done using a t-test between matched sensitive and resistant cells. There is a decrease in the transcription levels of most NK ligands genes.

TABLE 9

Gene transcription levels of natural killer (NK) cell ligands.

Gene name x-fold change p value

DU145/DR ;

22RV1/DR

MHC class I polypeptide-related sequence A -1.4; -1.7 NS;NS

MHC class I polypeptide-related sequence -2.2 ; -1.8 <0.05;NS

A///B

MHC class I polypeptide-related sequence B -2.2 ; -1.5 <0.05;NS

Poliovirus receptor -4.1; -2.1 <0.0001;<0.05 poliovirus receptor-related 2 (herpesvirus -2.5;-1.7 <0.05;NS entry mediator B)

Note: NS=Not Significant [0146] Regarding studies on stem cell phenotype, we characterized the identified deregulated expression of WNT/p-catenin, Notch and Hedgehog, which have been implicated in self-renewal and differentiation of progenitor cells (Katoh ef a/., 2007; McDonald et al., 2006; van den Brink et al., 2004; Radtke et al., 2006; Leong et al., 2008; and Grigoryan et al., 2008). In the prostate, these signaling pathways have been reported to play essential roles in developmental patterning, epithelial regeneration, and prostate cancer tumourigenesis (Wang ef a/., 2006; Karhadkar et al., 2004). Docetaxel resistant cells showed a significant decrease in both gene transcript and protein levels of Dickkopf-1 (DKK1), a well known inhibitor of the WNT/p-catenin signaling network. This decrease in DKK1 expression was linked to an increase in the expression of de-phosphorylated (active) β-catenin, which is the major key effector of WNT signaling. Immunofluorescence analyses demonstrated that parental Docetaxel sensitive cells displayed a membranous expression of β-catenin, associated with its function as an adhesion molecule, whereas Docetaxel resistant cells showed a pronounced nuclear localization of this protein (Figure 2), reported as necessary for the activation of the canonical WNT signaling pathway. Moreover, Docetaxel resistant cells also exhibited an increase in the NOTCH signaling network. NOTCH2 gene transcript levels were significantly increased in the resistant cells and were linked to an increase in cleaved Notch2 protein expression that was associated with nuclear translocation of the protein, where it exerts its transcriptional activity (Figure 2). Finally, Docetaxel resistant cells had an increased expression of the Hedgehog receptor Patched and the glioma associated oncogene homolog transcription factors, Gli1 and GH2. These findings were associated with an increased protein expression and nuclear translocation of the above mentioned transcription factors (Figure 2), a condition that has been related to Hedgehog pathway activation. Surprisingly, other reported stem cell surface markers, such as CD44 and CD133, were not found to be up-regulated in our model, as analyzed both at the gene transcript and protein levels (data not shown).

[0147] Furthermore, it was observed that the reversibility of the resistant phenomenon for both DU145-DR and 22RV1-DR cells were linked to an increase in the expression of differentiation markers. Docetaxel reversed resistant cells (cultured without the drug during 12 weeks) showed higher protein expression levels of low molecular weight cytokeratins (CK19 and CK 8) and HLA-class I when compared to Docetaxel acquired resistant cells, achieving levels similar to those observed in parental sensitive cells (Figure 13).

Generation and characterization of an epithelial differentiation reporter model

[0148] DU145 parental cells were stably transfected with a plasmid containing the promoter of CK19 driving the expression of the green fluorescence protein (GFP) (Figure 14a). Co-expression of CK19 and GFP in DU145-CK19 promoter-GFP stable cells was confirmed by immunofluorescence (Figure 14a). Flow cytometry quantification showed two distinct populations of cells, being the majority of cells positive for both GFP and CK19 (94.3±3.8%) and a discrete population of cells negative for both markers (5.6+4.1%). Few scattered cells outside these two main populations were observed which could represent transiting cells from one compartment to the other. Furthermore, stable insertion of the promoter construct in CK19/GFP negative cells was confirmed by PCR (Figure 4c). The expression of HLA class I in DU145-CK19 promoter-GFP stable cells was further characterized (Figure 14b). Not surprisingly, cells that expressed GFP were also HLA-positive (91.6±5.5%) and cells that did not express GFP displayed an HLA-negative phenotype (7.0±4.95). Thus these results validate the use of GFP as a reporter of epithelial differentiation and further demonstrate, as shown previously in Figure 4b, the existence of a subpopulation of cells that lack differentiation markers (CK19/GFP) and HLA class I antigens.

Tumour initiation studies on HLA class I sorted cell lines

[0149] In order to further confirm that HLA-class I expression can be used as a cell surface marker that identifies cells with the cancer stem cell functional property of tumour initiation, parental cell lines DU145 and 22RV1 were sorted for HLA-class I and their tumour initiating capacity tested in NOD/SCID mice (Figure 15). Similar to the results obtained with the DU145-CK19 promoter-GFP stable GFP-negative cells, only the HLA class l-negative cells exhibited tumour initiating capacity after dilution assays. Injection of 10 HLA class l-negative DU145 and 22RV1 cells produced tumours in 83.3± 9.1% and 100% of recipients, respectively, while no tumour formation was observed after 198 days of injection with 10 cells displaying a HLA class l-positive phenotype. Of note, tumour initiating capacity and tumours latencies of HLA class l-negative cells from 22RV1 and DU145 were different, although these differences did not reach statistical significance. The differences between tumourigenic cell lines could be explained by the fact that other molecular pathways may play a role in the engraftment and growth of human cells in mice. Similar results were obtained after serial transplantation from HLA class l-negative generated tumour xenografts (data not shown).

Clonability studies on HLA class I sorted cell lines

[0150] In order to address the clonability capacity of HLA class I sorted cells from DU145 and 22RV1 parental cell lines, we performed dilution colony formation assays. HLA class l-negative cells displayed a statistically significant higher clonability than HLA class l-positive cells. Specifically, HLA class l-negative sorted cells from DU145 generated colonies in 31.6±7.5%, 17.1±3.6% and 22.0+4.9% when 10, 100 and 1000 cells were plated, respectively. In contrast, HLA class l-positive cells generated colonies in 5.0±8.3%, 8.0±3.3% and 5.5±1.5% when 10, 100 and 1000 cells were plated (Figure 8). Similar results were observed with 22RV1 HLA class I sorted parental cells (data not shown).

Identification of prostate cancer stem cells in human tissues

[0151] Immunohistochemical studies of metastases (n=20) and matched primary (n=6) human prostate cancer tissues revealed that all specimens contained scattered eubpopulations of CK-negative (CK18 and CK 9) tumour cells, accounting for 0.05% to 0.3% and 0.4% to 1.8% of all tumour cells in primary and metastatic lesions, respectively (Figure 7a). Immunofluorescence-based double staining was then performed to assess the association between CK expression and the markers of interest. In this analysis, it was consistently observed that CK expression was significantly associated with HLA class I expression (p<0.0001). More specifically, it was observed that the CK-negative tumour population did not express HLA class I antigens in 97.8 ± 0.7% of the cells, whereas all (100%) of the CK-positive cells displayed a positive HLA class I antigen phenotype (Figure 4b). A small population of CK-negative cells that displayed a HLA class l-positive phenotype was identified, which could represent tumour cells that undergo transition from an HLA class I- negative/CK-negative phenotype to a differentiated phenotype. Furthermore, it was consistently found that CK-negative/HLA-negative tumour cells had a significant (p<0.0001) increase of nuclear expression (activation) of developmental transcription factors when compared to differentiated CK-positive/HLA-positive cells. CK- negative/HLA-negative cells displayed nuclear expression of de-phosphorylated β- catenin in 63.9 ± 22.6% of cells, cleaved Notch2 in 72.8 ± 15.1%, Gli1 in 67.5 ± 17.3%, and Gli2 in 67 ± 17.3%, whereas CK-positive/HLA-positive cells expressed nuclear de-phosphorylated β-catenin in only 5.8 ± 11.9% of cells, cleaved Notch2 in 6.7 ± 7.9%, Gli1 in 1.2 ± 7.9%, and GH2 in 1.5 ± 10.6% (Figure 7c). Moreover, we also observed that the CK-negative/HLA class l-negative tumour cells showed no expression of nuclear AR, whereas CK-positive/HLA-positive cells displayed nuclear AR in 71.8 ± 14.3% of the cells (Figure 7d). Thus the fact that prostate cancer stem cells do not display a positive AR phenotype suggests that these cells may not be dependent on a functional AR signaling, which would explain how these cells might be responsible for the observed relapse after hormone-therapy, an issue to be pursued in future studies. Taken together, these results confirm their existence and the ability to identify a subpopulation of prostate cancer stem cells in human prostate cancer tissue samples.

Statistical analyses

[0152] Experimental data is expressed as means ± SD. Statistical analysis by Student's t-test was performed. Values were considered statistically significant at p≤ 0.05.

[0153] All documents cited in this application are hereby incorporated by reference as if recited in full herein.

[0154] Although illustrative embodiments of the present invention have been described herein, it should be understood that the invention is not limited to those described, and that various other changes or modifications may be made by one skilled in the art without departing from the scope or spirit of the invention. Cited Documents

Al-Hajj, M., et al., Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci USA 100, 3983-3988 (2003).

Ali, T. Z. et al., False positive labeling of prostate cancer with high molecular weight cytokeratin: p63 a more specific immunomarker for basal cells. Am J Surg Pathol 32, 1890-1895 (2008).

Bao, S. et a/., Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature 444, 756-760 (2006).

Brulet, P. et al., Monoclonal antibodies against trophectoderm-specific markers during mouse blastocyst formation. Proc Natl Acad Sci USA 77,41 3- 4117 (1980).

Cariati, M. et al., Alpha-6 integrin is necessary for the tumorigenicity of a stem cell-like subpopulation within the MCF7 breast cancer cell line. Int J Cancer 122, 298-304 (2008).

Chen, J. K. et al., Inhibition of Hedgehog signaling by direct binding of cyclopamine to Smoothened. Genes Dev 16, 2743-2748, (2002).

Chen, R. et al., A hierarchy of self-renewing tumor-initiating cell types in glioblastoma. Cancer Cell 17, 362-375 (2009).

Collins, A. T. et al., Prospective identification of tumorigenic prostate cancer stem cells. Cancer Res 65, 10946-10951 (2005).

Cordon-Cardo, C. et al., Expression of HI_A-A,B,C antigens on primary and metastatic tumor cell populations of human carcinomas. Cancer Res 51 , 6372-6380 (1991).

Costello, R. T. et al., Human acute myeloid leukemia CD34+/CD38- progenitor cells have decreased sensitivity to chemotherapy and Fas-induced apoptosis, reduced immunogenicity, and impaired dendritic cell transformation capacities. Cancer Res 60, 4403-4411 (2000).

Dalerba, P. ei al., Cancer stem cells: models and concepts. Annu Rev Med 58, 267-284 (2007).

Dean, M. er al., Tumor stem cells and drug resistance. Nat Rev Cancer 5, 275-284 (2005).

Desoye, G. et al., Lack of HLA class I and class II antigens on human preimplantation embryos. J Immunol 140, 4157-4159 (1988).

Fedi, P. et al., Isolation and biochemical characterization of the human Dkk-I homologue, a novel inhibitor of mammalian Wnt signaling. J Biol Chem 274, 19465-19472 (1999). Fillmore, C. . et al., Human breast cancer cell lines contain stem-like cells that self-renew, give rise to phenotypically diverse progeny and survive chemotherapy. Breast Cancer Res 10, R25 (2008).

Grigoryan, T. et al., Deciphering the function of canonical Wnt signals in development and disease: conditional loss- and gain-of-function mutations of beta-catenin in mice. Genes Dev 22, 2308-2341 (2008).

Gupta, P. B. et al., Identification of selective inhibitors of cancer stem cells by high-throughput screening. Cell 138, 645-659 (2009).

Guzman, M. L. et al., Preferential induction of apoptosis for primary human leukemic stem cells. Proc Natl Acad Sci USA 99,16220-16225 (2002).

Jordan, C. T. et al., Cancer stem cells. N Engl J Med 355, 1253-1261 (2006). Karhadkar, S. S. et al., Hedgehog signalling in prostate regeneration, neoplasia and metastasis. Nature 431, 707-712 (2004).

Katoh, M., Networking of WNT, FGF, Notch, BMP, and Hedgehog signaling pathways during carcinogenesis. Stem Cell Rev 3, 30-38 (2007).

Leong, K. G. et al., The Notch pathway in prostate development and cancer. Differentiation 76, 699-716 (2008).

Li, H. et al., PC3 human prostate carcinoma cell holoclones contain self- renewing tumor-initiating cells. Cancer Res 68, 1820-1825 (2008).

Lu, H. ef al., Type II keratins precede type I keratins during early embryonic development. Eur J Cell Biol 84, 709-718 (2005).

McDonald, S. A. et al., Clonal expansion in the human gut: mitochondrial DNA mutations show us the way. Cell Cycle 5, 808-811 (2006).

O'Brien, C. A. et al., A human colon cancer cell capable of initiating tumor growth in immunodeficient mice. Nature 445, 106-110 (2007).

Oshima, R. G., Identification and immunoprecipitation of cytoskeletal proteins from murine extra-embryonic endodermal cells. J Biol Chem 256, 8124-8133 (1981).

Patrawala, L. et al., Highly purified CD44+ prostate cancer cells from xenograft human tumors are enriched in tumorigenic and metastatic progenitor cells. Oncogene 25, 1696-1708 (2006).

Patrawala, L. et al., Hierarchical organization of prostate cancer cells in xenograft tumors: the CD44+alpha2beta1+ cell population is enriched in tumor-initiating cells. Cancer Res 67, 6796-6805 (2007).

Petrylak, D. P. ef al., Docetaxel and estramustine compared with mitoxantrone and prednisone for advanced refractory prostate cancer. N Engl J Med 351 , 1513-1520 (2004). Petrylak, D. P. et al., P-glycoprotein expression in primary and metastatic transitional cell carcinoma of the bladder. Ann Oncol 5, 835-840 (1994).

Ponti, D. et al., Isolation and in vitro propagation of tumorigenic breast cancer cells with stem/progenitor cell properties. Cancer Res 65, 5506-5511 (2005). Quintana, E. et al., Efficient tumor formation by single human melanoma cells. Nature 456, 593-598 (2008).

Radtke, F. ei a/., From gut homeostasis to cancer. Curr Mol Med 6, 275-289 (2006).

Real, P. J. et al., Gamma-secretase inhibitors reverse glucocorticoid resistance in T cell acute lymphoblastic leukemia. Nat Med 15, 50-58, (2009). Reya, T. et al., Stem cells, cancer, and cancer stem cells. Nature 414, 105- 111 (2001).

Ricci-Vitiani, L. et al., Identification and expansion of human colon-cancer- initiating cells. Nature 445, 111-115 (2007).

Rosen, J. M. et al., The increasing complexity of the cancer stem cell paradigm. Science 324, 670-1673 (2009).

Sanchez-Carbayo, M. et al., Expression profiling of osteosarcoma cells transfected with MDR1 and NEO genes: regulation of cell adhesion, apoptosis, and tumor suppression-related genes. Lab Invest 83, 507-517

(2003) .

Seiffert, D. ei al., Presenilin-1 and -2 are molecular targets for gamma- secretase inhibitors. J Biol Chem 275, 34086-34091 , (2000).

Sharma, S. V. ei al., A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell 141 , 69-80 (2010)

Shmelkov, S. V. et al., CD133 expression is not restricted to stem cells, and both CD133+ and CD133- metastatic colon cancer cells initiate tumors. J Clin Invest 118, 2111-2120 (2008).

Singh, S. K. et al., Identification of human brain tumor initiating cells. Nature 432, 396-401 (2004).

Taipale, J. et al., Effects of oncogenic mutations in Smoothened and Patched can be reversed by cyclopamine. Nature 406, 1005-1009, (2000).

Tannock, I. F. et al., Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N Engi J Med 351 , 1502-1512

(2004) .

Tesar, P. J. et al., New cell lines from mouse epiblast share defining features with human embryonic stem cells. Nature 448, 196-199 (2007). van den Brink, G. R. et al., Indian Hedgehog is an antagonist of Wnt signaling in colonic epithelial cell differentiation. Nat Genet 36,277-282 (2004).

Tripathi, M. K. et al., Negative regulation of the expressions of cytokeratins 8 and 19 by SLUG repressor protein in human breast cells. Biochem Biophys_^ Res Commun 329, 508-515 (2005)

Trumpp, A. et al., Mechanisms of Disease: cancer stem cells-targeting the evil twin. Nat Clin Pract Oncol 5, 337-347 (2008).

van Es, J. H. et al., Notch/gamma-secretase inhibition turns proliferative cells in intestinal crypts and adenomas into goblet cells. Nature 435, 959-963, (2005).

Vermeulen, L. et al., Cancer stem cells-old concepts, new insights. Cell Death Differ 15, 947-958 (2008).

Wang, X. D. et al., Notch signaling is required for normal prostatic epithelial cell proliferation and differentiation. Dev Biol 290, 66-80 (2006).

Willed, K. et al., Wnt signaling: is the party in the nucleus? Genes Dev 20, 1394-1404 (2006).

Wu, M. et al., Imaging hematopoietic precursor division in real time. Cell Stem Cell (2007).

Claims

WHAT IS CLAIMED IS:

1. A method for isolating a cancer stem cell (CSC) from a population of cancer cells comprising:

(a) obtaining a population of cancer cells;

(b) identifying those cells from the cancer cell population that are HLA^"; and

(c) separating the HLA^" cells from the cancer cell population, the HLA^" cells being CSCs.

2. The method according to claim 1 , wherein the population of cancer stem cells is a cancer cell line.

3. The method according to claim 2, wherein the cancer cell line is resistant to an agent used to treat cancer.

4. The method according to claim 2, wherein the cancer cell line is obtained from a cancer selected from the group consisting of a solid tumor, a blood-based tumor, and a nervous system tumor.

5. The method according to claim 2, wherein the cancer cell line is selected from the group consisting of DU145 and 22RV1.

6. The method according to claim 1, wherein the population of cancer cells is a non-immortal sample from a subject.

7. The method according to claim 6, wherein the sample is from a human patient.

8. The method according to claim 6, wherein the sample is obtained from a cancer selected from the group consisting of prostate, breast, colon, lung, bladder, leukemia, lymphoma, glioma, and glioblastoma cancer.

9. The method according to claim 3, wherein the agent is selected from the group consisting of DNA damaging drugs and anti-mitotic agents.

10. The method according to claim 3, wherein the agent is selected from the group consisting of microtubulin inhibitors, topoisomerase inhibitors, vinblastine, vincristine, vinorelbine, paclitaxel, mitoxantrone, cisplatin, docetaxel, colchicines analogs, harringtonine, homoharringtonine, camptothecine, camptothecine analogs and podophyllotoxin.

11. The method according to claim 10, wherein the agent is docetaxel.

12. The method according to ciaim 1 , wherein step (c) is carried out via a fluorescence activated cell sorting (FACS) analysis.

13. The method according to claim 1, wherein the HLA^" cells are further defined by being at least one of CD24-, CD133-, Notch⁺, Gli1⁺, Gli2⁺, GFAP^", Neurofil^", and cytokeratin^".

14. The method according to claim 1 , wherein the HLA^" cells are further defined by having at least one of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance.

15. A method for isolating a cancer stem cell (CSC) from a cancer cell line that is resistant to an agent used to treat cancer comprising:

(a) obtaining a cancer cell line that is resistant to an agent used to treat cancer; and

(b) separating HLA^" cells from the cancer cell line, the HLA^" cells being CSCs.

16. The method according to claim 15, wherein the agent is selected from the group consisting of microtubulin inhibitors, topoisomerase inhibitors, vinblastine, vincristine, vinorelbine, paclitaxel, mitoxantrone, cisplatin, docetaxel, colchicines analogs, harringtonine, homoharringtonine, camptothecine, camptothecine analogs and podophyllotoxin.

17. The method according to claim 16, wherein the agent is docetaxel.

18. The method according to claim 15, wherein the cancer cell line is obtained from a cancer selected from the group consisting of a solid tumor, a blood-based tumor, and a nervous system tumor.

19. The method according to claim 15, wherein the cancer cell line is selected from the group consisting of DU145 and 22RV1.

20. The method according to claim 15, wherein after step (a) the cancer cell line is contacted with a fluorescently labeled antibody against HLA and step (b) is carried out via a fluorescence activated cell sorting (FACS) analysis to separate those cells that are HLA^" from those that are HLA⁺.

21. A method for isolating a cancer stem cell (CSC) from a docetaxel-resistant cancer cell line selected from the group consisting of DU145 and 22RV1 comprising separating HLA^" cells from the cancer cell line, the HLA^" cells being CSCs.

22. The method according to claim 21, wherein the separating step is carried out via fluorescence activated cell sorting (FACS) analysis.

23. A method for isolating a cancer stem cell (CSC) from a sample of cancerous material from a subject comprising separating HLA^" cells from the sample, the HLA^" cells being CSCs.

24. The method according to claim 23, wherein the separating step is carried out via fluorescence activated cell sorting (FACS) analysis.

25. The method according to claim 23, wherein the HLA^" cells are further defined by being at least one of CD24^", CD133^", Notch⁺, Gli1⁺, Gli2⁺, GFAF, Neurofil^", and cytokeratin^".

26. The method according to claim 23, wherein the HLA^" cells are further defined by having at least one of the following properties: capability to seif-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance.

27. The method according to claim 23, wherein the subject is human.

28. An isolated cancer stem cell obtained by the method of any one of claims 1, 15, 21 , and 23.

29. A mammalian cancer stem cell (CSC) line that is enriched for cells that are HLA Γ and HLA II^".

30. The mammalian CSC line of claim 29, which is further enriched for cells that are at least one of CD24^", CD133-, Notch*, Gli1⁺, Gli2⁺, GFAP^", Neurofil^", and cytokeratin^".

31. The mammalian CSC line of any one of claims 29 or 30 , which is further enriched for cells that are defined by having at least one of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance.

32. The mammalian CSC line of claim 29, which is a cultured CSC line.

33. The mammalian CSC line of claim 29, which is an immortal CSC line.

34. The mammalian CSC line of claim 29, wherein the CSCs are human or mouse.

35. The mammalian CSC line of claim 29, wherein the CSCs, when transplanted, form a tumor, which is selected from the group consisting of solid tumors, blood- based tumors, and nervous system tumors.

36. A human cancer stem cell line that is enriched for cells that are HLA I^" and HLA ir.

37 An isolated, multipotent mammalian cell line that is enriched for cancer stem cells (CSCs) that are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin.

38. A mammalian cell culture having an enhanced number of cancer stem cells (CSCs), which CSCs are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin.

39. The mammalian cell culture of claim 38, wherein the cell culture is enriched for CSCs that are Notch⁺, Gli1⁺, and Gli2⁺.

40. The mammalian cell culture of any one of claims 38 or 39, wherein the cell culture is enriched for CSCs having at least one of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance.

41. The mammalian cell culture of claim 38, which is a cultured CSC line.

42. The mammalian cell culture of claim 38, which is an immortal CSC line.

43. The mammalian cell culture of claim 38, wherein the CSCs are human or mouse.

44. The mammalian cell culture of claim 38, wherein the CSCs, when transplanted, form a tumor, which is selected from the group consisting of solid tumors, blood-based tumors, and nervous system tumors.

45. A kit for providing mammalian cancer stem cells (CSCs), the kit comprising a cell line enriched for mammalian CSCs that are surface antigen negative for HLA I, HLA II, CD24, CD133, GFAP, Neurofil, and cytokeratin, which are packaged in a suitable container.

46. The kit of claim 45, wherein the enriched CSCs are Notch⁺, Gli1\ and Gli2⁺.

47. The kit of claim 45, wherein the enriched CSCs have at least one of the following properties: capability to self-renew, undergo asymmetrical cell division, have tumorigenic capacity, have metastatic potential, have multi-differentiation properties, sensitive to Notch and Hedgehog inhibitors, and have broad chemoresistance.

48. The kit of claim 45, wherein the cells are part of a frozen stock.

49. The kit of claim 45, wherein the cells are part of a growing culture.

50. The kit of claim 45, wherein the container is a tissue culture container or a freezer-safe tissue culture container.