Defining a Population of Stem-like HumanProstate Cancer Cells That Can Generate andPropagate Castration-Resistant Prostate CancerXin Chen1,2,3, Qiuhui Li1,3, Xin Liu1, Can Liu1, Ruifang Liu1,3, Kiera Rycaj1,3,Dingxiao Zhang1,3, Bigang Liu1, Collene Jeter1, Tammy Calhoun-Davis1, Kevin Lin1,Yue Lu1, Hsueh-Ping Chao1, Jianjun Shen1, and Dean G. Tang1,3,4,5
Abstract
Purpose: We have shown that the phenotypically undifferen-tiated (PSA�/lo) prostate cancer cell population harbors long-termself-renewing cancer stem cells (CSC) that resist castration, and asubset of the cells within the PSA�/lo population bearing theALDHhiCD44þa2b1þ phenotype (Triple Markerþ/TMþ) is capa-ble of robustly initiating xenograft tumors in castrated mice. Thegoal of the current project is to further characterize the biologicproperties of TMþ prostate cancer cell population, particularly inthe context of initiating and propagating castration-resistantprostate cancer (CRPC).
Experimental Design: The in vivoCSC activities weremeasuredby limiting-dilution serial tumor transplantation assays in bothandrogen-dependent and androgen-independent prostate cancerxenograft models. In vitro clonal, clonogenic, and sphere-forma-tion assays were conducted in cells purified from xenograft andpatient tumors. qPCR, Western blot, lentiviral-mediated gene
knockdown, and human microRNA arrays were performed formechanistic studies.
Results: By focusing on the LAPC9 model, we show that theTMþ cells are CSCs with both tumor-initiating and tumor-prop-agating abilities for CRPC. Moreover, primary patient sampleshave TMþ cells, which possess CSC activities in "castrated" cultureconditions. Mechanistically, we find that (i) the phenotypicmarkers are causally involved in CRPC development; (ii) theTMþ cells preferentially express castration resistance and stemcell–associated molecules that regulate their CSC characteristics;and (iii) the TMþ cells possess distinct microRNA expressionprofiles and miR-499-5p functions as an oncomir.
Conclusions: Our results define the TMþ prostate cancer cellsas a population of preexistent stem-like cancer cells that can bothmediate and propagate CRPC and highlight the TMþ cell popu-lation as a therapeutic target. Clin Cancer Res; 1–12. �2016 AACR.
IntroductionHuman tumors are heterogeneous, containing many pheno-
typically and functionally distinct cancer cells, and tumor cellheterogeneity can arise as a result of genetic diversity and/orepigenetic maturation of stem cell–like cancer cells or cancer stemcells (CSC; refs. 1, 2). CSCs in many human solid tumors have
been implicated in tumor initiation, progression, metastasis, andtherapy resistance (3, 4). Like other human cancers, prostatecancer is a heterogeneous malignancy containing phenotypicallydifferentiated cancer cells and immature cancer cellswith stemcellproperties, i.e., prostate CSCs (PCSC; refs. 4–19).
Prostate cancer is one of the most common malignanciesaffecting American males, with an estimated 220,800 new casesand 27,540 cancer-associated deaths in 2015 (20). Localizedprostate cancer at an early stage can be treated by radical prosta-tectomywith a goodprognosis. Advanced prostate cancer patientsaremostly treated by androgen-deprivation therapy (ADT),whichfails eventually leading to thedevelopment of incurable and lethalcastration-resistant prostate cancer (CRPC). Although manymolecular mechanisms have been proposed to explain CRPC,the cellular origin for CRPC remains largely unknown (4).
In the field, the cell-of-origin of primary prostate cancer, i.e.,whether human prostate cancer is derived from basal, luminal, orintermediate (progenitor) cells, has been an area of intensivestudies. In mouse models, several groups have shown that bothbasal and luminalmurine prostatic epithelial cells can function asthe targets of tumorigenic transformation (21). In the humanprostate, the basally localized stem cell–enriched population, i.e.,CD49fhiTrop2þ, can be transformed by AKT, ERG, and androgenreceptor (AR) and can function as the cell-of-origin for humanprostate cancer (22).Whether this cell population can serve as thecell-of-origin for CRPC is presently unknown. Recent evidence
1Department of Epigenetics and Molecular Carcinogenesis, The Uni-versity of Texas MD Anderson Cancer Center, Smithville, Texas.2Department of Oncology, Tongji Hospital, Tongji Medical College,Huazhong University of Science and Technology, Wuhan, China.3DepartmentofPharmacologyandTherapeutics,Roswell ParkCancerInstitute, Buffalo, New York. 4Cancer Stem Cell Institute, ResearchCenter for Translational Medicine, East Hospital, Tongji UniversitySchool ofMedicine, Shanghai, China. 5Centers for Cancer Epigenetics,Stem Cell and Developmental Biology, RNA Interference and Non-coding RNAs, and Molecular Carcinogenesis, The University of TexasMD Anderson Cancer Center, Houston, Texas.
Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).
Corresponding Authors: Qiuhui Li or Dean G. Tang, Department of Pharma-cology and Therapeutics, Roswell Park Cancer Institute, Buffalo, NY 14263.Phone: 716-845-1254; Fax: 716-845-8857; E-mail: [email protected][email protected]
indicates that some PCSCpopulations in hormone-na€�ve prostatecancer xenograft models and patient tumors express low levels ofAR and could potentially function as the cell-of-origin of CRPC(6, 13, 16, 19), and vice versa, several cell populations, such as theCD166þ (23) and N-Cadherinþ (24) cells, become enriched inCRPCmodels. Using a lentiviral reporter tracking system, we haverecently demonstrated that long-term self-renewing and tumor-propagating PCSCs are enriched in the undifferentiated prostatecancer cell population that lacks prostate-specific antigen (PSA)expression (i.e., PSA�/lo; ref. 16). Importantly, the PSA�/lo cellpopulation is intrinsically resistant to androgen deprivation andenzalutamide in vitro (16, 19) and possesses high tumor-regen-erating activity in castrated NOD/SCIDmice (16), suggesting thatit may function as the cellular origin of CRPC. Nevertheless, thePSA�/lo cell population is still heterogeneous and the subset ofcells within this population that actually mediates castrationresistance remains ill defined (16, 19).
Interestingly, cDNA microarray analysis comparing geneexpression in PSAþ and PSA�/lo LAPC9 prostate cancer cellsidentified the ALDHhiCD44þa2b1þ subpopulation, within thePSA�/lo population, that manifested very high tumor-regenerat-ing activity in castrated mice such that as few as 10 ALDHhi
CD44þa2b1þ cells could regenerate an androgen-independenttumor (16). The main goals of the current study are to (i) betterdefine the ALDHhiCD44þa2b1þ cells, which we term the triplemarker-positive or TMþ prostate cancer cell population, withrespect to its long-term tumorigenic potential in castrated hosts;(ii) test the hypothesis that the TMþ cell population can functionas the cell-of-origin forCRPC; (iii) explore thepotentialmolecularmechanisms underlying the CRPC mediated by the TMþ cellpopulation. The results presented here have important transla-tional and clinical significance as they shed light on CRPCevolution at the cellular level.
Materials and MethodsCells, animals, and reagents
PC3, Du145, PPC-1, 22Rv1, and LNCaP cells were obtainedfrom ATCC and cultured in RPMI-1640 plus 7% heat-inactivated
fetal bovine serum (FBS), 10,000 mg/mL streptomycin, and10,000 U/mL penicillin (PS; HyClone). These cell lines wereregularly authenticated by our institutional CCSG Cell LineCharacterization Core using short tandem repeat (STR) analysisand checked to be free of mycoplasma contamination using theAgilent MycoSensor QPCR Assay Kit (cat. #302107) every 8weeks. All cell lines used were passaged fewer than 6 monthsafter receipt. Antibodies used in this study are summarized inSupplementary Table S1. Immunodeficient mice (NOD/SCIDand NOD/SCID-IL2Rg�/�, i.e., NSG) were purchased from theJackson Laboratory, and breeding colonies were maintained inour animal facility core under standard conditions. All animal-relevant studies in the current project have been approved by theMD Anderson Cancer Center IACUC (Institutional Animal Careand Use Committee; ACUF#00000923-RN00).
Limiting-dilution tumor regeneration assays (LDA) and serialtumor transplantations in immunodeficient mice
Basic procedures were previously described (16, 19). For LDA,FACS purified prostate cancer cells were mixed with Matrigeland injected subcutaneously (s.c.) at increasing numbers inNOD/SCID mice. For serial tumor transplantation assays, tumorcells of a specific phenotype were sorted from the first-generation(1�) tumors of the samephenotype for 2� tumor transplantations.Sequential tumor transplantations were carried out using similarstrategies. For tumor studies in castrated mice, male NOD/SCIDor NOD/SCID-IL2Rg�/� (NSG) mice (8–10 weeks) were surgi-cally castrated 1 to 2 weeks prior to tumor cell injections. At theendpoint, tumors were harvested and various parameters wererecorded, including tumor incidence, weight, latency, and images.Tumor-initiating frequency (TIF) was calculated using the Limdilfunction of the Statmod package (http://bioinf.wehi.edu.au/soft-ware/elda/).
Gene knockdown with shRNA-encoding lentiviral vectorsBasic procedures have been previously described (16, 19).
Briefly, we produced lentivirus in 293T packaging cells (Clon-tech),whichwas then titered usingGFPpositivity inHT1080 cells.Prostate cancer cells were infected at a multiplicity of infection(MOI) of 10 to 20 for 48 to 72 hours at 37�C. All GIPZ-shRNA-encoding lentiviral vectors used in the current study were pur-chased fromGEDharmacon (SupplementaryMaterials andMeth-ods). The knockdown effect on the target molecules was assessedby qRT-PCR.
Statistical analysesWe used an unpaired two-tailed Student t test to compare
significance in cell numbers, percentages of CD44þ and/or a2þ
cells, cloning and sphere-forming efficiencies, tumor weights,knockdown efficiency, mRNA levels of multiple genes and otherrelated parameters.We used a c2 test to compare tumor incidence.All resultswere presented asmean� SDwith aP<0.05 consideredstatistically significant.
See also Supplementary Materials and Methods.
ResultsThe TMþ (ALDHhiCD44þa2b1þ) prostate cancer cellpopulation is enriched in experimental CRPC models
In our earlier cDNA microarray analysis, we compared geneexpression profiles between PSA�/lo versus PSAþ LAPC9 prostate
Translational Relevance
Most advanced prostate cancer patients treated with andro-gen-deprivation therapy develop castration-resistant prostatecancer (CRPC). Although many molecular mechanisms havebeen proposed to explain CRPC, the cellular origin for CRPCremains mostly unclear. Systematic studies shown here dem-onstrate that the TMþ (i.e., ALDHhiCD44þa2b1þ) subpopu-lation of the prostate cancer cells preexist in untreated ADxenograft and patient tumors but become greatly enrichedduring castration in vitro and in vivo. Importantly, the TMþ cellscan function not only as CRPC-initiating but also as CRPC-propagating cells, highlighting this population of the cells asnovel therapeutic target in treating prostate cancer. The avail-ability of large numbers of TMþ cells in certain xenografts (e.g.,LAPC9) should allow high-throughput screening to identifynovel drugs targeting this population, which may eventuallyimpact the clinical treatment of some patients with CRPC.
cancer cells and found that PSA�/lo prostate cancer cells over-expressed several dozens of stem cell–related genes, includingCD44, integrin a2, and ALDH1A1 (16). ALDHhiCD44þa2b1þorTMþ LAPC9 cells regenerated much larger tumors whenimplanted in castrated mice than the corresponding ALDHlo
CD44�a2b1�or TM� cells (16), suggesting that TMþ prostatecancer cells may play an important role in CRPC development.
To directly test this suggestion, we established serially passagedandrogen-independent (AI, i.e., castration-resistant) xenograftmodels, including LAPC9, LAPC4, LNCaP, and HPCa101 (25)from their respective androgen-dependent (AD) parental tumors(Fig. 1A). As illustrated in Fig. 1B, both LAPC9 and LAPC4 AItumors showed a prominent upregulation of N-Cadherin, amolecule known to be involved in CRPC (24). In contrast,E-Cadherin levels did not significantly change in AI tumors incomparison to AD tumors (Fig. 1B). Interestingly, the LAPC4 AItumors showed increased AR protein, whereas the LAPC9 AItumors gradually lost AR, similar to earlier reports by others(24, 26). However, both AI tumor models showed decreased
amounts of PSA (Fig. 1B), consistent with our earlier observationsthat castration resistance is associated with decreasing tumor cellPSA levels and increasing PSA�/lo PCSCs (16, 19). Together, theseresults indicate thatwehave successfully established experimentalCRPC models.
Quantitative RT-PCR (qRT-PCR) analysis revealed that theLAPC9 AI tumors expressed significantly higher levels of integrina2, ALDH1A1, and ALDH7A1 (11) mRNAs than AD tumors(Fig. 1C). A trend of increased CD44 mRNA in LAPC9 AI tumorswas also observed (Fig. 1C). Importantly, theCD44 and integrin a2mRNA levels in TMþ LAPC9 cells purified from serially passagedAI tumors were significantly higher than those in the correspond-ing TM� cells (Fig. 1D). FACS analysis, using the gating strategieswe developed, showed that the percentage of TMþ cells dramat-ically increased in serially passaged LAPC9 AI tumors (Supple-mentary Fig. S1).
IHC and IF staining in formalin-fixed paraffin-embedded(FFPE) samples confirmed that CD44þ cells were highly enrichedin LAPC9 AI tumors compared with AD tumor (Fig. 1E and F; and
Figure 1.TMþ cells in AD and AI prostate cancer models. A, strategies in establishing AD and AI prostate cancer lines. B, Western blot analysis of the molecules indicatedin AD and AI LAPC9 and LAPC4 tumors. Du145 and LNCaP cells were used as controls. C, qRT-PCR analysis of mRNA levels for CD44, a2, ALDH1A1, andALDH7A1 in LAPC9 AD and AI tumors. The relative transcript abundance was normalized to GAPDH levels. Error bars, mean� SD. � , P < 0.05. D, qRT-PCR analysisof mRNA levels of CD44, a2 in sorted TMþ and isogenic TM� cells purified from 6� and 15� LAPC9 AI tumors. Bars, mean � SD. � , P < 0.05. E–H, representativeIF images of CD44 (E) and integrin a2 (G) staining in LAPC9 AD and serially passaged LAPC9 AI (e.g., 7� and 13�) tumors. Percentages of CD44þ (F) and a2þ (H)cells were characterized. Bars, mean � SD. ���, P < 0.001. I, TMþ cells were enriched in serially passaged LAPC9 AI tumors. Bars, mean � SD. �� , P < 0.01;��� , P < 0.001. J, TMþ prostate cancer cells were increased in LNCaP and HPCa101 AI tumors. K, TMþ prostate cancer cells were abundant in long-termcultured AI prostate cancer lines.
Defining Castration-Resistant Prostate Cancer Stem Cells
Supplementary Fig. S2A). Recent studies have linked aldehydedehydrogenase (ALDH) activity in PCSCs to prostate cancerdevelopment (10, 11, 19, 27, 28) that might be conferred byseveral isoforms, including ALDH1A1 (10, 27, 28) and ALDH7A1(11). However, whether these ALDH isoforms are expressed inCRPC samples is unclear. We found that the cells with high ALDHactivity were increased in LAPC9 AI tumors (SupplementaryFig. S1, e), and the abundance of ALDH1A1þ and ALDH7A1þ
prostate cancer cells was also higher in AI versus AD tumors(Supplementary Fig. S2B and S2C). In addition, integrin a2þ
cells were increased in both LAPC9 (Fig. 1G and H) and LAPC4(Supplementary Fig. S2D) AI tumor models.
We used FACS analysis to further investigate TMþ cells inseveral different models. In the LAPC9 model, �1.7% TMþ cellswere present in AD tumors, but this percentage graduallyincreased with increasing cycles of castration to nearly 40% at6� generation (Fig. 1I). The TMþ cells were rare in AD LNCaP andHPCa101 tumors, but their abundance still increased in thecorresponding AI tumors (Fig. 1J). Finally, several commonlyused AI prostate cancer cell lines all contained significant percen-tages of TMþ cells (Fig. 1K). Altogether, these results suggest thatthe TMþ prostate cancer cells are enriched in experimental CRPCmodels and are abundantly present in AI prostate cancer cell lines.
TMþ cells in AD tumors possess long-term tumor-propagatingabilities in hormonally intact male mice
The 'gold standard' to measure CSC activity is to determine if acandidate CSC population has the capacity to generate seriallyxenotransplantable tumors in immunodeficient mice at increas-ing cell doses (an assay termed limiting-dilution tumor regener-ation assay or LDA), and if the regenerated tumors phenocopy, atleast partially, parental tumor histopathologically (4, 29). There-fore, LDA and serial transplantation assays are combined in ourstudy to characterize the CSC properties of the TMþ and relatedprostate cancer cell populations.
Because TMþ cells preexist in AD tumors (Fig. 1I–J; Supple-mentary Fig. S1A), we first determined their tumor-initiating andtumor-propagating activities in androgen-proficient hosts, usingstrategies similar to those we previously used to characterize thePSA�/lo cells (16). Thus, TMþ cells and several other cell popula-tions, including ALDHhiCD44�a2b1�, ALDHloCD44þa2b1þ,and TM�, were purified from LAPC9 AD tumors and s.c. injectedin hormonally intact NOD/SCID male mice at increasing doses(from 10 to 10,000 cells; Table 1). Surprisingly, at 1�, althoughTMþ cells exhibited a trend of higher tumorigenicity than corre-sponding TM� cells (TIF 1/1,312 vs. 1/2,972; P ¼ 0.0647), twointermediate cell populations, ALDHhiCD44�a2b1� (TIF 1/315)
Table 1. TIF of subsets of LAPC9 cells in intact and castrated male NOD/SCID mice
and ALDHloCD44þa2b1þ (TIF 1/421), were significantly moretumorigenic than TMþ cells (Table 1). However, as tumor cellsof a specific phenotype were sorted from the 1� tumors of thesame phenotype (e.g., purifying TMþ cells from 1� TMþ cell-derived tumors) for the 2� transplantations in intact male mice,TMþ cells maintained the highest tumorigenicity comparedwith ALDHhiCD44�a2b1�, ALDHloCD44þa2b1þ, and TM�
cells (Table 1). Similarly, TMþ cells were greatly enriched intumor-regenerating capacity compared with all other cell popu-lations at the 3� transplantations in intact male mice (Table 1).Notably, during serial passages, tumors regenerated from TMþ
cells at each generation contained TMþ cells as the minoritywith the majority of cells being non-TMþ cells (SupplementaryFig. S3, left), suggesting that the TMþ cells possess self-renewalpotential in vivo. Interestingly, protein levels of two well-char-acterized differentiation markers, i.e., AR and PSA, were signif-icantly decreased in TMþ cell-derived AD tumors than TM� cell-derived counterparts even after two consecutive passages inintact male mice (Supplementary Fig. S4), supporting thegeneral notion that CSCs are less differentiated (4). Takentogether, these observations demonstrate that TMþ LAPC9 cellsare endowed with CSC (i.e., long-term tumor-propagating)properties in hormone-proficient conditions.
TMþ cells can function as the cell-of-origin for CRPC andpossess long-term CRPC-propagating activity in androgen-ablated male mice
Given that TMþ cells are prominently enriched in AI tumors, weponder a critical question: Can TMþ cells in AD tumors initiatetumor regeneration in androgen-ablated hosts? To answer thisquestion, we purified TMþ and TM� cells from maintenance ofLAPC9 AD tumors and performed LDA by implanting differentdoses of cells into castrated immunodeficient NOD/SCID-IL2Rg�/� (NSG) mice. Remarkably, TMþ cells demonstrated�43-fold higher tumor-initiating capacity in castrated hosts com-pared with TM� cells (TIF 1/49 vs. 1/2,142; Fig. 2A). As few as 100TMþ cells initiated 7/8 tumors, whereas 100 TM� cells did notregenerate a single tumor in castrated hosts (Fig. 2A). Differencesin both tumor incidence and weight between the two groupswere statistically highly significant (Fig. 2A). This experimentprovides direct evidence that the TMþ LAPC9 cells can functionas the cell-of-origin for CRPC.
To determine whether the TMþ cells can in the long termpropagate CRPC in androgen-deficient hosts, we purified TMþ
and TM� cell populations from LAPC9 AI tumors and carried outLDA and serial transplantation assays in fully castrated hosts, i.e.,castrated male NOD/SCID mice also treated with bicalutamide(ref. 16; Fig. 2B andC; Table 1). In the 1� generation, the TMþ cellsdemonstrated much higher tumorigenic potential than TM� cellsin castrated mice (TIF 1/2, 590 vs. 1/21,299, respectively; Fig.2B; Table 1). In the 2� generation tumor experiments, the TMþ
cells, in a cell dose–dependent manner, regenerated robustAI tumors in castrated hosts such that as few as 100 cells generated5/8 tumors, whereas the TM� cells did not generate tumors at 100and 1,000 cell doses (Fig. 2C; Table 1). Overall, the TMþ cellsmanifested significantly higher 2� tumor-regenerating capacitythan the TM� LAPC9 cells (TIF 1/283 vs. 1/13,802, respectively;P¼ 2.23e�09; Table 1). These results suggest that the TMþ LAPC9cells possess long-term CRPC-propagating capability. In fullagreement with this conclusion, both primary and secondaryLAPC9 AI tumors derived from the TMþ cells contained a similar
small population (i.e., 10%–20%) of TMþ cells as the parental AItumors (Supplementary Fig. S3, right), suggesting that the TMþ
cells in AI tumors self-renewed in vivo.To further determine the intrinsic castration resistance in
TMþ prostate cancer cells, we sorted TMþ and non-TMþ (i.e.,TM-depleted) cells from LAPC9 AI tumors, and carried outsimilar LDA in castrated mice. We observed that the TMþ cellsdisplayed much higher tumor-regenerating activity than theisogenic TM-depleted cells (Fig. 2D; Table 1). Likewise, TMþ
cells were able to self-renew for subsequent generations (Table 1).It was apparent that the TMþ cells outcompeted other subpopula-tions (e.g., ALDHhiCD44þa2b1�, ALDHloCD44þa2b1þ) withrespect to tumor-initiating and tumor-propagating abilities incastrated mice (Table 1). In support of the above tumor experi-ments, the TMþ LAPC9 cells purified out from the AI tumorsdemonstrated much higher sphere formation (Fig. 2E) and clo-nogenic (Fig. 2F) activities than the TM� cells in "castrated"culture conditions.
Lastly, we applied similar strategies to study additional AImodels. TMþ PC3 cells were significantly more clonogenic thanTM� PC3 cells in castrated culture conditions (Fig. 2G and H).Importantly, the TMþ PC3 cells manifested much higher tumor-igenic potential than TM� PC3 cells generating more and largertumors (Fig. 2I). Purified TMþ LAPC4 AI cells also exhibitedincreased clonogenicity as compared with TM� LAPC4 AI cellsin "castrated" culture conditions (data not shown). Combined,these results demonstrate that (i) the TMþ cells in LAPC9 ADtumors can function as the cell-of-origin for LAPC9 AI tumors,(ii) the TMþ cells in LAPC9 AI and PC3 models possess greatCRPC-regenerating activity, (iii) the LAPC9 AI TMþ cells possesslong-term CRPC-propagating activities, and (iv) the AI TMþ
Untreated primary human prostate cancer (HPCA) sampleshave TMþ cells thatmanifest CSCproperties in castrated cultureconditions
Next, we explored the potential clinical relevance of the abovefindings in xenograft studies using cells purified from 16 untreat-ed HPCA samples, all of which have >75% tumor involvement(Supplementary Table S2; Supplementary Fig. S5A). Tumor areaswere confirmed by robust expression of AR, PSA, and racemaseand lack of expression of CK5 (Supplementary Fig. S5B). Wefirst purified out HPCA cells as the CD45�Trop2þ epithelial cells(22, 30), which were then analyzed for the abundance of TMþ
cells (Fig. 3A and B). We found that all these untreated primarytumors contained TMþ cells ranging from �0.001% to as highas �14.5% (Fig. 3B). When the TMþ cells were purified fromHPCa201 and cultured in castrated conditions (CDSS þ enza-lutamide), they exhibited much enhanced clonal, clonogenic,and self-renewal capacities compared with TM� cells (Fig. 3C).Notably, the HPCa201 TMþ cells also demonstrated signifi-cantly higher secondary colony-forming activity than the cor-responding TM� cells (Fig. 3C, b). Purified HPCa222 TMþ cellsgenerated much bigger and more clones and colonies thancorresponding TM� cells in castrated culture conditions(Fig. 3D). Importantly, the HPCa222 TMþ cell–derived colo-nies were positive for TMPRSS–ERG fusion (Supplementary Fig.S5C-D), indicating the tumor cell origin of the colonies. TheHPCa 223 TMþ cells similarly demonstrated higher clonal andcolony-forming capabilities than the corresponding TM� cells
Defining Castration-Resistant Prostate Cancer Stem Cells
(Supplementary Fig. S6A). Interestingly, some HPCA samplescontained almost undetectable TMþ cells (Fig. 3B) but con-tained double-, and/or single-positive cells (Fig. 3A; 19). In theHPCA samples analyzed, we observed that double-positive(i.e., ALDHhiCD44þ) HPCA cells were much more clonal andclonogenic than double-negative (i.e., ALDHloCD44�) cellswith significantly increased colony sizes and clone numbers(Supplementary Fig. S6B and S6C). However, HPCa210ALDHhi cells exhibited similar clonal and clonogenic abilitiesas ALDHlo cells (data not shown). Taken together, these resultsindicate that untreated primary HPCA samples also containTMþ cells that are intrinsically resistant to castration.
Phenotypic markers are functionally involved in CRPCdevelopment
Next, we attempted to determine whether the phenotypicmarkers used to define TMþ cells are causally involved in tumorregeneration under AI conditions by performing lentiviral-medi-ated knockdown of integrin a2, CD44, ALDH1A1, and ALDH7A1followed by assessing the impact on tumor regeneration incastrated male NOD/SCID mice (Supplementary Fig. S7; seeSupplementary Fig. S8A and S8B for knockdown efficiency).Knocking down integrin a2 in LAPC9 AI tumor cells decreasedtumor incidence (5/9 or 55.6%) as compared with the control(6/6 or 100%; Supplementary Fig. S7A), whereas knocking down
Figure 2.TMþ cells are intrinsically castration-resistant and can function as the cell-of-origin for CRPC. A, TMþ LAPC9 cells in AD tumors can preferentially regeneratetumors in castrated hosts. TMþ and TM� cells were purified from LAPC9 AD tumors and s.c. injected at increasing numbers in castrated male NSG mice.At the endpoint, tumor weight, incidence, and TIF were recorded. B, TMþ LAPC9 AI cells (i.e., the TMþ cells purified fromAI tumors) aremuchmore tumorigenic thanTM� cells in fully castrated male NOD/SCID mice during primary (1�) tumor transplantation. The P value for TIF was indicated. C, TMþ LAPC9 AI cells canself-renew and maintain enhanced tumorigenic abilities in fully castrated male NOD/SCID mice during 2� tumor transplantation. TIF and its P value were listedin Table 1. D, TMþ LAPC9 AI cells also exhibit higher tumor-regenerating potential in castrated hosts than the corresponding TM-depleted cells. See TIFand its P value in Table 1. E–F, TMþ LAPC9 AI cells display high CSC activities in vitro. For sphere formation assays (E), sorted TMþ and TM� LAPC9 AI cells wereplated in 6-well ULA plates (10k/well) and cultured for 2 weeks. The 1� spheres were harvested and passaged for 2� sphere assays (2k/well) in ULA plates,which were cultured for 3 weeks. For clonogenic assays (F), sorted TMþ and TM� cells were mixed with Matrigel, plated (5k/well) in 12-well plates and cultured for 3weeks. The 1� colonies were passaged for 2� clonogenic assays (1k/well), and the 2� colonies were enumerated after 3 weeks. All assays were conducted in CDSS-containing IMDM medium. Error bars, mean � SD. n ¼ 3 for each condition. G and H, TMþ PC3 cells show high stem/progenitor activities in castrated cultureconditions. For sphere assays (G), TMþ and TM� PC3 cells were plated (1k/well) in ULA plates for 2 weeks. Shown is the bar graph (mean � SD; n ¼ 3). Forclonogenic assays (H), TMþ and TM� PC3 cells weremixedwith Matrigel, plated (1k/well) in 12-well plates and cultured for 2weeks. Sizes of coloniesweremeasured.Both assays were performed in CDSS-containing PrEBM medium plus 10 mmol/L enzalutamide. I, TMþ PC3 cells are much more tumorigenic than TM� cells incastrated male NOD/SCID mice. TMþ and TM� cells were sorted from PC3 cultures and s.c. injected in castrated mice, at the cell doses indicated for tumorregeneration. Tumor weight, incidence, TIF, and P values are shown.
integrin a2 in LAPC4 AI tumor cells resulted in significantlysmaller tumors (Supplementary Fig. S7B). CD44 knockdown inLAPC9 AI tumor cells not only inhibited tumor initiation but alsoreduced tumor burden (Supplementary Fig. S7C). Finally,ALDH1A1 knockdown in LAPC9 AI cells led to reduced spheres(Supplementary Fig. S8C) and a 50% inhibition in tumor weight(Supplementary Fig. S7D), whereas ALDH7A1 knockdown par-tially affected sphere-forming and tumor-initiating capacities inLAPC9 AI cells (Supplementary Fig. S8C; data not shown). Col-lectively, these results indicate that the phenotypic markers arecausally important in imparting the castration-resistant CSCproperties.
TMþ prostate cancer cells have basal cell features andpreferentially express CSC and castration resistance–associatedgenes such as RegIV and SOX9
To further characterize the TMþ cells at the molecular level, wepurified out TMþ and TM� LAPC9 cells fromADand twodifferent
generations of AI tumors and performed qRT-PCR analysis of thegenes associated with luminal/basal cell differentiation, castra-tion resistance, and CSCs (Fig. 4A–C; Supplementary Table S3).This analysis revealed interesting changes in the gene expressionpatterns. In LAPC9 AD tumors, the TMþ cells overall expressedhigher levels of basal cell genes (CK5, CK14, and CD44) whereasthe TM� cells expressed higher levels of luminal genes (AR, PSA,and CK18 with the exception of CK8; Fig. 4A). In LAPC9 AItumors, the TMþ cells still expressed higher levels of basalmarkersCK5 and CK14 but the TM� cells expressed high PSA, CD26 (31),and CK19 mRNA levels while expressing similar levels of otherluminal markers (i.e., AR, CK8, CK18, and Nkx3.1) to TMþ cells(Fig. 4B and C).
The TMþ cells in the 6� generation LAPC9 AI tumors consis-tently expressed higher levels, compared with the TM� cells, ofRegIV, SOX9, and UBE2C, which have been implicated in castra-tion resistance and CRPC progression (32–35), as well as severalCSC-associated molecules, including TGFBR-1, Bcl-2, and Stat3
Figure 3.TMþ cells preexist in primary HPCA samples and manifest CSC traits in castrated conditions in vitro. A, FACS strategy to purify TMþ cells in primary patientsamples. Bulk HPCA cells were first sorted as CD45�Trop2þ epithelial cells (30). ALDHhi HPCA cells were gated as per its negative control (DEAB; not shown), whichwere then sorted for CD44þ and a2b1þ cells based on respective isotype controls (not shown). Shown is a representative patient sample (HPCa134). B, theabundance of TMþ cells in 16 untreated primary HPCA samples. C, TMþ HPCa201 cells display high clonal (a), and clonogenic (b) capacities. For clonal assays (a),sorted cells were plated (1k/well) in PureCol-coated 6-well plates. Images were taken 10 days after plating. For clonogenic assays (b), sorted cells weremixed with Matrigel, plated at increasing doses (500, 1k, 2k/well) in 12-well plates, and cultured for 2 weeks. The 1� colonies were collected and passaged for 2�
clonogenic assays (500, 1k, 2k/well). After 2 weeks, colony size was shown. Data, mean � SD (n ¼ 3). D, TMþ HPCa222 cells possess enhanced clonal (a) andclonogenic (b) capabilities compared with TM� cells. Experimental strategies and conditions were similar to C. Shown are bar graphs for colony numberand size (mean � SD; n ¼ 3) and representative images. In both C and D, cells were cultured in a CDSS-containing medium plus 10 mmol/L enzalutamide.P values are indicated.
Defining Castration-Resistant Prostate Cancer Stem Cells
(Fig. 4C). Preferential expression of RegIV and SOX9 in TMþ cellswas further confirmed in a 15� generation LAPC9 AI tumor (Fig.4D). To determine whether RegIV and SOX9 are causally involvedin conferring on TMþ LAPC9 cells, certain CSC properties andcastration resistance, we knocked downRegIV and SOX9 in LAPC9AI bulk cells (Supplementary Fig. S8D–S8E), which significantlyattenuated their sphere-forming abilities (Fig. 4E). RegIV andSOX9 knockdown in TMþ LAPC9 AI cells also reduced theirclonogenicity in castrated culture conditions (Fig. 4F) and tumorgrowth in castrated mice (Fig. 4G). These data suggest that RegIVand SOX9 regulate the CSC properties of the TMþ LAPC9 AI cells.
Distinct microRNA (miRNA) expression profiles in TMþ LAPC9AI cells and CRPC-promoting effects of miR-499-5p
Our laboratory has been studying the mechanisms wherebymicroRNAs (miRNA) regulate PCSCs and prostate cancer devel-
opment and metastasis (36–38). To understand how miRNAsmight regulate the TMþ castration-resistant prostate cancer cells,we used a miRNA SmartChip that contained >1,000 humanmiRNAs to compare miRNA expression profiles of TMþ withTM-depleted LAPC9 AI cells. We found that 59 miRNAs weresignificantly upregulated, whereas 22 miRNAs were downregu-lated in the TMþ cells (Fig. 5A; Supplementary Table S4). As anexample to establish a cause-and-effect relationship between thedifferentially expressed miRNAs and castration resistance, wefocused on miR-499-5P, which was among the most upregulatedmiRNAs in the TMþ cells (Fig. 5A). miR-499-5P is a relativelynovel and understudied miRNA. A recent paper reported itspotential oncogenic functions in colorectal cancer cells via target-ing FOXO4 and PDCD4 (39). We found that miR-499-5P wasexpressed at higher levels in purified TMþ LAPC9 AI cells (Fig. 5B)and in the LAPC9 AI tumors (Fig. 5C). Overexpression of miR-
BA C
ED
0.055
6/10 (60%)
6/11 (54.5%)
0.113 ± 0.084
0.028 ± 0.021
Incidence Weight (g) P
8/11 (72.7%)
0.0830.070 ± 0.025
LAPC9 AI TM+ cells (4k) s.c injected in castrated mice
GIPZ-NS
RegIV-shRNA
SOX9-shRNA
GIPZ-NS
Brig
htG
FP
RegIV-shRNA
SOX9-shRNA
G
F
Re
lativ
e e
xp
ressio
n
0.0
0.5
1.0
1.5 TM+TM-
LAPC9 AI 15o
**
LAPC9 AI bulk cells
Sph
eres
NS #1 #2 #1 #2
SOX9-shRNA
RegIV-shRNA
0
20
40
60
80
* ***
Co
lonie
s0
200
400
600
800
1,000
Nonsilencing RegIV-shRNASOX9-shRNA
Nonsilencing RegIV-shRNA SOX9-shRNA
GFP
Brig
ht
LAPC9 AI TM+ 1k cells
* * * *
RegIV SOX9 1k 2k
53 days
LAPC9 ADAR PS
A
CK8
CK1
8
CK5
CK1
4
CD
44
Luminal
0
1
2
3
Re
lativ
e e
xpre
ssio
n TM+TM-
Basal
LAPC9 AI 3o
AR
PS
A
CK
19
CK
18
Nkx
3.1
CK
5
0
1
256789
1011
Re
lativ
e e
xpre
ssio
n
TM+TM-
Luminal Basal
LAPC9 AI 6o
AR PSA
CK8
CK5
CK1
4
Nkx
3.1
CD
26
Reg
IVU
BE2C
TGFB
R-1
SOX9
Bcl-2
Klf-4
Stat
3EG
FRE2
F3BM
I-1SO
X2
SCs & Castration Resistance
0
1
2
3
4
5
Luminal Basal
Re
lativ
e e
xpre
ssio
n
** * ** * *
TM+TM-
*
Figure 4.Further molecular and functional characterizations of TMþ cells. A–C, qRT-PCR analysis of differentiation, castration resistance, and stem cell–associatedgenes in purified TMþ and TM� cells from LAPC9 AD (A), AI 3� (B), and AI 6� (C) tumors. D, TMþ cells from LAPC9 AI tumors preferentially express RegIV and SOX9genes. Shown is the bar graph (mean � SD; n ¼ 2; � , P < 0.05). E. RegIV and SOX9 knockdown in bulk LAPC9 AI (19)� tumor cells reduces sphere-formingcapacities under androgen-deprived conditions. Bulk LAPC9 AI tumor cells were infected with two shRNAs, plated (10k/well) in 6-well ULA plates, and cultured inCDSS-containing IMDM medium plus 10 mmol/L enzalutamide. Spheres were counted in 2 weeks. Data, mean � S.D from triplicate plating in each condition.� , P <0.05; �� ,P <0.01. F,RegIV and SOX9 knockdown in purified TMþ LAPC9AI cells attenuates colony formation under castrated conditions. Purified TMþ cells fromLAPC9 AI tumors (19�) were infected with the shRNAs targeting RegIV or SOX9, mixed with Matrigel, and plated at increasing doses (1k and 2k) in 12-well plate.Colonies were counted in 2 weeks. Shown are the bar graph (mean � SD; n ¼ 3; � , P < 0.05) and representative images. G, RegIV and SOX9 knockdown in TMþ
LAPC9 AI cells inhibits CRPC growth. The TMþ cells purified from LAPC9 AI (19�) tumors were infected with the lentiviral vectors (MOI ¼ 20, 72 hours) ands.c. injected in castrated male NOD/SCID mice. Tumor incidence, tumor weight, and P values are indicated. Right, the phase and GFP images ofrepresentative endpoint tumors.
Figure 5.TMþ and TM� LAPC9 AI cells have distinct miRNA expression profiles. A,Wafergen SmartChip humanmiRNA array was performed using TMþ and TM-depleted (i.e.,dep) cells that were purified from serially passaged LAPC9 AI (i.e., 6� and 7�) tumors. Differentially expressed miRNAs were selected based on P < 0.05. 59 miRNAswere significantly upregulated in TMþ cells, whereas 22 miRNAs were downregulated in TMþ cells. miR-499a-5P was highlighted by a red arrow. B, qRT-PCRvalidation of miR-499-5P expression in TMþ and TM� cells sorted from LAPC9 AI 6� and 15� tumors. C, LAPC9 AI tumors display higher miR-499-5Pexpression than LAPC9 AD tumors. Shown is the bar graph (mean � SD; n ¼ 3). D, miR-499-5P overexpression in LAPC9 AD bulk cells promotes their clonogenicactivities in castrated conditions. Bulk LAPC9 AD cells were transfected with miR-499-5P and negative control (NC) mimic (30 nmol/L, 48 hours), plated in6-well ULA plates and cultured in CDSS-containing IMDM medium plus 10 mmol/L enzalutamide. Spheres were counted in 2 weeks. E, miR-499-5P knockdown inLAPC9 AI TMþ cells impairs their clonogenicity under castrated culture conditions. TMþ cells purified from LAPC9 AI tumors were transfected with nontargetingnegative control miRNA inhibitor (anti-negative control) or miR-499-5P inhibitor (anti-miR-499-5P) (20 nmol/L, 72 hours), plated in 6-well ULA plates andcultured for 2weeks. CDSS-containing IMDMmedium plus 10 mmol/L enzalutamidewas used. The bar graph (mean� SD; n¼ 3) and respective images are shown. F,overexpression of miR-499-5P in the TM-depleted LAPC9 AI cells promotes CRPC regeneration. The TM-depleted LAPC9 AI cells were purified from xenografts,transfected with the negative control or miR-499-5P oligos (30 nmol/L), and implanted, at two cell doses, s.c. in castrated male NOD/SCID mice. Tumorswere harvested 56 days after implantation. The difference in TIF was statistically significant. G, overexpression of miR-499-5P antagomirs in the TMþ LAPC9 AI cellsinhibits CRPC growth. The TMþ LAPC9 AI cells were purified from xenografts, transfected with the anti-negative control or anti-miR-499-5P oligos (30 nmol/L),and implanted s.c. in castrated male NOD/SCID mice. Tumors were harvested 53 days after implantation.
Defining Castration-Resistant Prostate Cancer Stem Cells
499-5P in LAPC9 AD tumor cells by transfection of miR-499-5Poligos promoted sphere formation compared with cells trans-fected with negative control oligos (Fig. 5D). In contrast, trans-fection of the anti-miR-499-5P oligos in the TMþ LAPC9 AI cellssuppressed sphere formation (Fig. 5E). Importantly, overexpres-sion of miR-499-5P in the TM-depleted LAPC9 AI cells promotedtumor regeneration in castrated mice (Fig. 5F), whereas over-expression of anti-miR-499-5P in the TMþ LAPC9 AI cells signif-icantly inhibited tumor growth (Fig. 5G). Together, these resultsdemonstrate that the TMþ prostate cancer cells express a uniquemiRNA profile that causally regulates their tumorigenic andcastration-resistant properties.
DiscussionTwo principal mechanisms, i.e., intrinsic and acquired, under-
lie the resistance of cancer cells to both general and targetedanticancer therapeutics (40). Intrinsic resistance implies that priorto treatments, therapy-insensitive cells preexist in the tumor andbecome selected and enrichedduring treatment,whereas acquiredresistance is caused by treatment-induced gene mutations andother adaptive responses. Both mechanisms of therapy resistancehave been reported in clinical tumors (40–42). Here, we provide aprototypical example for a population of cancer cells that canmediate intrinsic therapy resistance.
As early as 1981, Isaacs andCoffey, working on the ratDunningR-3327-H prostate adenocarcinoma (H-tumor) model, per-formed a fluctuation analysis to show that the H-tumor relapseafter ADT was due to continuous proliferative growth of preexist-ing AI prostate cancer cells after castration (43). Exactly what cellpopulation in this model that became selected during castrationwas uncharacterized (43). In 1999, Craft and colleagues, byperforming a similar fluctuation analysis, provided evidence inthe LAPC9 model that ADT selected and clonally expanded thepreexistent AI cells, resulting in outgrowth of CRPC (44). Never-theless, which cell population in LAPC9 mediated the CRPCemergence was unclear. Here, we demonstrate that the TMþ cellpopulation in the LAPC9 model can function as both a cell-of-origin for CRPC emergence and the tumor-propagating popula-tion for the established CRPC.
LAPC9 is a well-known prostate cancer xenograft expressingPSA andwild-typeAR,whichour laboratory has beenusing for thepast 10 years to study PCSCs (6, 7, 16, 19, 36). Our earlier workhas shown that CD44þ prostate cancer cells in several xenograftmodels, including LAPC9, are highly enriched in PCSCs (6).Follow-up work demonstrates that the CD44þa2b1þ LAPC9 cellsare even more tumorigenic than CD44þ cells (7). The currentproject emanated from our recent observations that the PSA�/lo
cell population, which preexists in the AD tumors but becomesenriched in AI tumors, can regenerate larger tumors than PSAþ
cells in fully castrated hosts (16). This study provided the firstevidence for a population of preexisting human prostate cancercells that preferentially generate AI tumors (16). Nonetheless,both PSA�/lo and PSAþ LAPC9 cells regenerated AI tumors withsimilar incidences (16), suggesting that only a subset of thePSA�/lo cells is actually mediating the AI regeneration. Indeed,microarray analysis led to the identification of the TMþ subset thatwas greatly enriched in AI tumor-regenerating activities (16).
By establishing serially passaged AI tumors and by focusing onthe LAPC9 system, we have made many important and novelfindings. First, the TMþ cells in AD tumors possess long-term
tumor-propagating activities in androgen-proficient hosts. Sec-ond, the TMþ cells purified from the LAPC9 AD tumors canpreferentially initiate xenograft tumors in fully castrated hosts,suggesting that this population can preferentially function as thecell-of-origin for AI prostate cancer. Third, the TMþ cells graduallybecome enriched in serially transplantedAI tumors in vivo. Fourth,the TMþ LAPC9 cells purified from the AI tumors can seriallypropagate the xenograft tumors in androgen-ablated hosts com-pared with either TM� or TM-depleted cells, suggesting that theTMþ LAPC9 cells can also function as robust CRPC-propagatingcells. Fifth, the phenotypic markers used to define TMþ cells arefunctionally important for the TMþ cell activities. Finally, the TMþ
LAPC9 cells seem to possess a unique miRNA expression profile.To our knowledge, the (serial) tumor transplantation studiespresented here (Fig. 2; Table 1) represent themost comprehensiveeffort to define a population of human prostate cancer cells thatfunction as the cell-of-origin as well as the tumor-propagatingcells for CRPC.
Is the TMþ cell population unique to the LAPC9model? At leastin one other model, i.e., PC3, which is all PSA�, the TMþ cellsexhibit significantly higher tumor-regenerating activity than thecorresponding TM� cells in castrated animals. It is interesting thatseveral other AI cell lines examined, including PPC-1, Du145, and22Rv1, all contain significant percentages of TMþ cells, which, webelieve, may also possess high AI tumor-regenerating capacitiescompared with the corresponding TM� cells. In several otherxenograftmodels, including LNCaP andHPCa101, the fraction ofTMþ cells in the AD tumors is small, which, nevertheless, alsoincreases in the AI tumors, although the significance of the TMþ
cell fraction in these models remains to be determined. Ofsignificance, however, we have detected the TMþ cells in morethan a dozen of untreated HPCA samples. The relative abundanceof TMþ cells in primary HPCA samples vary widely, but freshlypurified TMþ HPCA cells manifested significantly higher colony-forming abilities than the corresponding TM� cells in castratedculture conditions. Intriguingly, the ALDHhiCD44þ double-pos-itive cells isolated from twoHPCAsamples also display significantcastration-resistant CSC properties. Taken together, our studies inmultiple culture/xenograft models and primary HPCA samplessuggest that (i) the TMþ prostate cancer cell population seems tobe widely present, (ii) the TMþ cells in some tumor systems,exemplified by LAPC9 and PC3, possess great tumor-initiatingand tumor-propagating activities in androgen-deficient hosts, (iii)the TMþHPCA cells in primary tumors also possess great survivaladvantages and colony-forming capabilities under androgen-defi-cient culture conditions, and (iv) in some HPCa samples, theALDHhiCD44þ double-positive cells may behave similarly to theTMþ cells.
What are the TMþ cells? qRT-PCR–based phenotyping suggeststhat the TMþ LAPC9 cells in both AD and AI tumors are basal-like,preferentially expressing CK5, CK14, and CD44 mRNAs and lessdifferentiated expressing lower levels of PSA and CD26 mRNAs.This would be consistent with the fact that this subpopulation ofthe cells was initially uncovered from the PSA�/lo prostate cancercell population (16). This is also consistent with our currentobservations that theAI cell lines, such as PPC-1, PC3, andDu145,all of which are PSA�/lo, have significantly higher percentages ofthe TMþ cells than the AD LAPC9, LAPC4, and LNCaP tumors, inwhich the majority of cells are PSAþ. It should be noted, though,that cell populations that can mediate and propagate CRPC maynot necessarily be all basal-like. For instance, in the BM18
Chen et al.
Clin Cancer Res; 2016 Clinical Cancer ResearchOF10
xenograft model, cells that mediate CRPC prominently expressluminal markers CK18 and NKX3.1 (14).
What confers the TMþ prostate cancer cells the 'hardy' proper-ties to readily regenerate and propagate the AI tumors? Part of theanswer seems to reside with the three defining phenotypic mar-kers, i.e., CD44, integrin a2, and ALDH1A1, whose knockdownall diminished the tumor-regenerating activities of the TMþ cellsin castrated conditions. This is not surprising considering that theCD44 and integrin a2 function as important adhesion andsignaling molecules and that ALDH1A1, and perhaps ALDH7A1,may further extend cell survival by detoxification (e.g., eliminat-ing ROS). Another reason is that the TMþ prostate cancer cellspreferentially express genes associated with (cancer) stem cellsand castration, illustrated by molecules such as Reg IV, SOX9,UBE2C, and TGFBR-1. Knocking down RegIV and SOX9 in theTMþ LAPC9 cells partially inhibits tumor regeneration in andro-gen-deficient hosts, implicating their causal functions in TMþ
cells. Finally, the TMþ cells express a unique profile of miRNAs.Using miR-499-5P as an example, which is overexpressed in theTMþ cells compared with the TM-depleted cells, we show that thismiRNA is also conferring on the TMþ LAPC9 cells castration-resistant properties.OthermicroRNAsuncovered in this screeningmay also likely be involved inmodulating the biologic propertiesof the TMþ prostate cancer cells. Collectively, the results make itclear that multiple molecules and mechanisms likely cometogether to endow the TMþ prostate cancer cells unique capabil-ities to function as both the cell of origin and tumor-propagatingcells for CRPC.
Our recent work has illustrated that the PSA�/lo prostate cancercell population represents a therapeutic target in treating CRPC(16, 19). The current work further pinpoints the TMþ subsetwithin the PSA�/lo population as the driving force of CRPCemergence and propagation. Ongoing work from our laboratoryis focusing on devising strategies to target the highly tumorigenicTMþ cell population andon elucidating the potential relationshipbetween TMþ cells and several other populations of prostate
cancer cells, e.g., CD166þ (23) and N-Cadherinþ (24), that mayalso be able to propagate castration-resistant tumors.
Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.
Authors' ContributionsConception and design: X. Chen, X. Liu, D.G. TangDevelopment of methodology: X. Chen, Q. Li, X. Liu, B. Liu, D.G. TangAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): X. Chen, Q. Li, J. ShenAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): X. Chen,Q. Li, C. Liu, B. Liu, K. Lin, Y. Lu, H.-P. Chao,D.G. TangWriting, review, and/or revision of the manuscript: X. Chen, Q. Li, R. Liu,K. Rycaj, C. Jeter, K. Lin, Y. Lu, D.G. TangAdministrative, technical, or material support (i.e., reporting or organizing data,constructing databases): X. Chen, C. Liu, D. Zhang, T. Calhoun-Davis, D.G. TangStudy supervision: B. Liu, D.G. TangOther (coordinating several collaborative groups): D.G. Tang
AcknowledgmentsThe authors thank Animal Core for animal care andmaintenance, Molecular
Biology Core for help in Wafergen SmartChip human miRNA array, HistologyCore for IHC studies, Ms. P.Whitney for assistance in FACS sorting and analysis,and other lab members for helpful discussions and suggestions.
Grant SupportThis project was supported, in part, by grants from NIH (R01-CA155693),
Department of Defense (W81XWH-13-1-0352 and W81XWH-14-1-0575),CPRIT (RP120380), and the MDACC Center for Cancer Epigenetics (D.G.T).X. Chen and C. Liu were supported, in part, by DOD post-doctoral fellowshipPC141581 and PC121553, respectively. Both J. Shen and Y. Lu were supportedby CPRIT Core Facility Support Award RP120348.
The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received December 10, 2015; revised March 15, 2016; accepted March 27,2016; published OnlineFirst April 8, 2016.
References1. Kreso A, Dick JE. Evolution of the cancer stem cell model. Cell Stem Cell
2014;14:275–91.2. Tang DG. Understanding cancer stem cell heterogeneity and plasticity.
Cell Res 2012;22:457–72.3. KorkayaH,WichaMS. Breast cancer stem cells: we've got them surrounded.
Clin Cancer Res 2013;19:511–3.4. Chen X, Rycaj K, Liu X, Tang DG. New insights into prostate cancer stem
identification of tumorigenic prostate cancer stem cells. Cancer Res 2005;65:10946–51.
6. Patrawala L, Calhoun T, Schneider-Broussard R, Li H, Bhatia B, Tang S, et al.Highly purified CD44þ prostate cancer cells from xenograft human tumorsare enriched in tumorigenic and metastatic progenitor cells. Oncogene2006;25:1696–708.
7. Patrawala L, Calhoun T, Schneider-Broussard R, Tang DG. Hierarchicalorganization of prostate cancer cells in xenograft tumors: the CD44þa2b1þ
cell population is enriched in tumor-initiating cells. Cancer Res 2007;67:6796–805.
8. Li H, Chen X, Calhoun-Davis T, Claypool K, TangDG. PC3 human prostatecarcinoma cell holoclones contain self-renewing tumor-initiating cells.Cancer Res 2008;68:1820–5.
9. Dubrovska A, Kim S, Salamone RJ, Walker JR, Maira SM, Garcia-EcheverriaC, et al. The role of PTEN/AKT/PI3K signaling in the maintenance and
viability of prostate cancer stem-like cell populations. Proc Natl Acad SciU S A 2009;106:268–73.
10. Li T, Su Y, Mei Y, Leng Q, Leng B, Liu Z, et al. ALDH1A1 is a marker formalignant prostate stem cells and predictor of prostate cancer patients'outcome. Lab Invest 2010;90:234–44.
11. Van den Hoogen C, van der Horst G, Cheung H, Buijs JT, Lippitt JM,Guzman-Ramirez N, et al. High aldehyde dehydrogenase activity identifiestumor-initiating and metastasis-initiating cells in human prostate cancer.Cancer Res 2010;70:5163–73.
12. Rajasekhar VK, Studer L, Gerald W, Socci ND, Scher HI. Tumor-initiatingstem-like cells in human prostate cancer exhibit increased NF-kBsignalling. Nat Commun 2011;2:162.
13. Domingo-Domenech J, Vidal SJ, Rodriguez-Bravo V, Castillo-Martin M,Quinn SA, Rodriguez-Barrueco R, et al. Suppression of acquired docetaxelresistance in prostate cancer through depletion of notch- and hedgehog-dependent tumor-initiating cells. Cancer Cell 2012;22:373–88.
14. Germann M, Wetterwald A, Guzman-Ramirez N, van der Pluijm G,Culig Z, Cecchini MG, et al. Stem-like cells with luminal progenitorphenotype survive castration in human prostate cancer. Stem Cells2012;30:1076–86.
15. Mulholland DJ, Kobayashi N, Ruscetti M, Zhi A, Tran LM, Huang J, et al.Pten loss and RAS/MAPK activation cooperate to promote EMT andmetastasis initiated from prostate stem/progenitor cells. Cancer Res 2012;72:1878–89.
www.aacrjournals.org Clin Cancer Res; 2016 OF11
Defining Castration-Resistant Prostate Cancer Stem Cells
16. Qin J, Liu X, Laffin B, Chen X, Choy G, Jeter CR, et al. The PSA-/lo prostatecancer cell population harbors self-renewing long-term tumor-propagatingcells that resist castration. Cell Stem Cell 2012;10:556–69.
17. Yoshioka T, Otero J, Chen Y, Kim YM, Koutcher JA, Satagopan J, et al. b4Integrin signaling induces expansion of prostate tumor progenitors. J ClinInvest 2013;123:682–99.
18. Li X, Liu Y, Chen W, Fang Y, Xu H, Zhu HH, et al. TOP2Ahigh is thephenotype of recurrence and metastasis whereas TOP2Aneg cells representcancer stem cells in prostate cancer. Oncotarget 2014;5:9498–513.
19. Liu X, Chen X, Rycaj K, Chao HP, Deng Q, Jeter C, et al. Systematicdissection of phenotypic, functional, and tumorigenic heterogeneity ofhuman prostate cancer cells. Oncotarget 2015;6:23959–86.
20. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin2015;65:5–29.
21. KwonOJ, Xin L. Prostate epithelial stem andprogenitor cells. Am JClin ExpUrol 2014;2:209–18.
22. Goldstein AS, Huang J, Guo C, Garraway IP, Witte ON. Identification of acell of origin for human prostate cancer. Science 2010;329:568–71.
23. Jiao J, Hindoyan A, Wang S, Tran LM, Goldstein AS, Lawson D, et al.Identification of CD166 as a surface marker for enriching prostate stem/progenitor and cancer initiating cells. PLoS One 2012;7:e42564.
24. Tanaka H, Kono E, Tran CP, Miyazaki H, Yamashiro J, Shimomura T, et al.Monoclonal antibody targeting of N-Cadherin inhibits prostate cancergrowth, metastasis and castration resistance. Nat Med 2010;16:1414–20.
25. Chen X, Liu B, Li Q, Honorio S, Liu X, Liu C, et al. Dissociated primaryhuman prostate cancer cells coinjected with the immortalized Hs5 bonemarrow stromal cells generate undifferentiated tumors in NOD/SCID-gmice. PLoS One 2013;8:e56903.
26. Chen CD,Welsbie DS, Tran C, Baek SH, Chen R, Vessella R, et al. Moleculardeterminants of resistance to androgen therapy. Nat Med 2004;10:33–9.
27. Cojoc M, Peitzsch C, Kurth I, Trautmann F, Kunz-Schughart LA, TelegeevGD, et al. Aldehyde dehydrogenase is regulated by b-catenin/TCF andpromotes radioresistance in prostate cancer progenitor cells. Cancer Res2015;75:1482–94.
28. LeMagnen C, Bubendorf L, Rentsch CA,Mengus C, Gsponer J, Zellweger T,et al. Characterization and clinical relevance of ALDHbright populations inprostate cancer. Clin Cancer Res 2013;19:5361–71.
29. Rycaj K, Tang DG. Cell-of-origin of cancer versus cancer stem cells: Assaysand interpretations. Cancer Res 2015;75:4003–11.
30. Goldstein AS, Drake JM, Burnes DL, Finley DS, Zhang H, Reiter RE, et al.Purification and direct transformation of epithelial progenitor cells fromprimary human prostate. Nat Protoc 2011;6:656–67.
31. Karthaus WR, Iaquinta PJ, Drost J, Gracanin A, van Boxtel R, Wongvipat J,et al. Identification of multipotent luminal progenitor cells in humanprostate organoid cultures. Cell 2014;159:163–75.
32. Gu Z, Rubin MA, Yang Y, Deprimo SE, Zhao H, Horvath S, et al. Reg IV: apromising marker of hormone refractory metastatic prostate cancer.Clin Cancer Res 2005;11:2237–43.
33. Wang Q, Li W, Zhang Y, Yuan X, Xu K, Yu J, et al. Androgen receptorregulates a distinct transcription program in androgen-independent pros-tate cancer. Cell 2009;138:245–56.
34. Cai C, Wang H, He HH, Chen S, He L, Ma F, et al. ERG induces androgenreceptor-mediated regulation of SOX9 in prostate cancer. J Clin Invest2013;123:1109–22.
35. Xie Y, Lu W, Liu S, Yang Q, Carver BS, Li E, et al. Crosstalk between nuclearMET and SOX9/b-catenin correlates with castration-resistant prostatecancer. Mol Endocrinol 2014;28:1629–39.
36. Liu C, Kelnar K, Liu B, Chen X, Calhoun-Davis T, Li H, et al. ThemicroRNAmiR-34a inhibits prostate cancer stem cells and metastasis by directlyrepressing CD44. Nat Med 2011;17:211–5.
37. Liu C, Tang DG. MicroRNA regulation of cancer stem cells. Cancer Res2011;71:5950–4.
38. Jin M, Zhang T, Liu C, Badeaux MA, Liu B, Liu R, et al. miRNA-128suppresses prostate cancer by inhibiting BMI-1 to inhibit tumor-initiatingcells. Cancer Res 2014;74:4183–95.
39. Liu X, Zhang Z, Sun L, Chai N, Tang S, Jin J, et al. MiroRNA-499–5P promotes cellular invasion and tumor metastasis in colorectalcancer by targeting FOXO4 and PDCD4. Carcinogenesis 2011;32:1798–805.
40. Holohan C, Van Schaeybroeck S, Longley DB, Johnston PG. Cancer drugresistance: an evolving paradigm. Nat Rev Cancer 2013;13:714–26.
41. Wilson PM, Danenberg PV, Johnston PG, Lenz HJ, Ladner RD. Standingthe test of time: targeting thymidylate biosynthesis in cancer therapy.Nat Rev Clin Oncol 2014;11:282–98.
42. Nelson WG, Yegnasubramanian S. Resistance emerges to second-genera-tion antiandrogens in prostate cancer. Cancer Discov 2013;3:971–4.
43. Isaacs JT, Coffey DS. Adaptation versus selection as the mechanismresponsible for the relapse of prostatic cancer to androgen ablation therapyas studied in the Dunning R-3327-H adenocarcinoma. Cancer Res1981;41:5070–5.
44. Craft N, Chhor C, Tran C, Belldegrun A, DeKernion J, Witte ON, et al.Evidence for clonal outgrowth of androgen-independent prostate cancercells fromandrogen-dependent tumors through a two-step process. CancerRes 1999;59:5030–6.
Clin Cancer Res; 2016 Clinical Cancer ResearchOF12
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's
.http://clincancerres.aacrjournals.org/content/early/2016/06/24/1078-0432.CCR-15-2956To request permission to re-use all or part of this article, use this link
