NGF/g-IFN Inhibits Androgen-Independent ProstateCancer and Reverses Androgen Receptor

Function Through Downregulation of FGFR2and Decrease in Cancer Stem Cells

Wei Chen, Guo-Min Wang, Jian-Ming Guo, Li-An Sun, and Hang Wang

Androgen-independent prostate cancer (AIPC) is difficult to treat. Present study is to explore the inhibitory effectof a cytokine environment on AIPC and its mechanism. We utilized nerve growth factor (NGF)/g-interferon (g-IFN) to change the cytokine environment. Animal models and 2 androgen receptor (AR)-negative prostatecancer cell lines were used to evaluate the effect of NGF/g-IFN. Flow cytometry, immunocytochemistry, westernblotting, Tunel assay, colony formation efficiency, gene microarray, and in vivo bioluminescence were used todiscern the mechanisms within NGF/g-IFN that effect the environment. In vitro, NGF/g-IFN effectively in-hibited the proliferation of AIPC cell lines and promoted the apoptosis of the cancer cells. In vivo, NGF/g-IFNsuppressed the growth and metastasis of a tumor mass that arose from the AIPC cell line. After NGF/g-IFNtreatment, the AR-negative cell lines re-expressed AR and were then able to respond to the androgen. Contraryto expectations, the proliferation of cells was inhibited after dihydrotestosterone was added, and the resultsindicated that NGF/g-IFN decreased the proportion of cancer stem cells. NGF/g-IFN worked mainly throughthe downregulation of fibroblast growth factor receptor 2.

Introduction

Prostate cancer is one of the most commonly diagnosedtumors in males. During the last few decades, androgens

have been regarded as important stimulators of prostatecells, and a main cause of prostate cancer. Thus, androgen-deprivation therapy (ADT) has become a conventional ther-apy for advanced prostate cancer [1,2]. However, most of theprostate cancers become androgen independent after ADT [3],as the tumor becomes more aggressive and readily metasta-sizes [4,5].

ADT can stimulate the formation of androgen-independentprostate cancer (AIPC). Many studies have confirmed thatthe decline of testosterone disturbs the level of certain cyto-kines that may influence the growth of tumors by regulatingmitotic division and differentiation. A rebalance of thecytokine level may be a new therapy for AIPC. Throughscreening, nerve growth factor (NGF) and g-IFN were foundto inhibit prostate cancer cell lines, Du145 and PC-3M.

NGF and g-IFN are notably influenced by testosterone,and the decline of testosterone decreased the levels of NGFand g-IFN. In this study, our results indicated that NGF andg-IFN inhibited AIPC cell lines Du145 and PC-3M in vivoand in vitro. By the dose–effect curve, 50 ng/mL of NGF and1,000 IU of g-IFN will present the optimal effect.

With NGF/g-IFN treatment, previously negative andro-gen receptor (AR) was re-expressed in Du145 and PC-3M.The re-expressed AR can be activated by dihydrotestosterone(DHT). The proportion of the integrinea2b1 + CD44 + cellsdecreased in the 2 cell lines. In previous reports, in-tegrinea2b1 + CD44 + cells proved to be prostate cancer stemcells (CSCs) in prostate cancer tissue and cell lines. Thus, theformation of AIPC may be due to the selective growth ofprostate CSCs, which do not express an AR. CSCs are a smallpopulation of cells in a tumor that have the ability to self-renew and infinitely proliferate. Non-CSCs represent theremainder of the cells in a tumor and possess limited pro-liferation potential.

CSCs resist radiotherapy [6], chemotherapy and hormonetherapy due to their unique characteristics [7–11]. Conven-tional therapy could eliminate most of the non-CSCs in tu-mors, whereas CSCs survive and become the roots ofrecurrence and metastasis. Even so, CSCs need stem cell ni-ches to regulate self-renewal and differentiation. A stem cellniche is defined as the microenvironment surrounding thestem cells.

Cytokines form a special microenvironment that stronglyinfluences the proliferation and differentiation of stem cells[12]. NGF is expressed in prostate tissues and carcinoma[13,14], and promotes the differentiation of various tumors

Department of Urology, Zhongshan Hospital of Fudan University, Shanghai, People’s Republic of China.

STEM CELLS AND DEVELOPMENT

Volume 21, Number 18, 2012

� Mary Ann Liebert, Inc.

DOI: 10.1089/scd.2012.0121

3372

[15–17]. Importantly, NGF is closely correlated with andro-gen levels [18,19]. One study showed that Du145 cells treatedwith NGF re-expressed AR, decreasing cell invasion [20]. Thelevel of g-IFN is also regulated by ADT [21], and it has beenshown that g-IFN stimulates the differentiation of prostatecancer [22]. In the nervous system, g-IFN upregulates thetrkA receptor (one of the NGF receptors) in neuroblastomacells and coordinates with NGF to induce the differentiationof these cells [23].

In the present study, NGF and g-IFN were used eithersolely or jointly to treat the AIPC cell lines, Du145 and PC-3M. The combination of NGF and g-IFN (NGF/g-IFN) ef-fectively inhibited tumor proliferation in vivo and in vitro,and functional AR was re-expressed in the cell lines. Con-trary to expectations, AR induced by NGF/g-IFN had en-tirely different functions. This activation re-expressed ARand suppressed the proliferation of NGF/g-IFN-treated cells.AR activation had the promoting proliferation effect incommon.

After the NGF/g-IFN treatment, a decline in the propor-tion of CSCs and lower expression of fibroblast growth factorreceptor 2 (FGFR2) was detected in the 2 cell lines. The re-sults suggested that NGF/g-IFN may play a role through thedownregulation of CSCs and FGFR2. This study providesa new insight into the treatment of AIPC and the role of ARin various cytokine environments and in differentiated statesof cells.

Materials and Methods

Cell lines and dose–effect experiment

The human AIPC cell lines, Du145 and PC-3M, werepurchased from the Chinese Academy of Sciences (originallyobtained from ATCC). All cells were cultured in the DMEM(Hyclone) supplemented with 10% of fetal bovine serum(Hyclone) at 37�C in a 5% CO2 atmosphere. Once theyreached 80% confluence, the cells were detached with AC-CUTASE (Millipore), seeded into a 96-well plate, and cul-tured in the DMEM supplemented with 10% fetal bovineserum at 37�C with 5% CO2. After 24 h, cells were grouped,and a combination of NGF and g-INF was added. Accordingto the previous study (20), the concentration of NGF was setto 50 ng/mL. To explore the optimum concentration of g-INF, 6 concentration gradients were set. Cells that added50 ng/mL of NGF served as the control. Other than 50 ng/mL of NGF, other gradients, respectively, added were200 IU, 500 IU, 1,000 IU, 2,000 IU, and 4,000 IU of g-INF.Following 7 days of treatment, the cell number was mea-sured using CellTiter-BlueTM reagent (Promega), accordingto the manufacture’s instruction. Briefly, cells were rinsedwith PBS, and 100 mL of Keratinocyte SFM was added intoeach well. Then, 20 mL of CellTiter blue reagent was added.The cells were incubated for 4 h in the dark at 37�C for 6 h.The intensity of fluorescence in each well was detected usinga fluorometer (Infinite 200), and the cell number was calcu-lated according to the standard curve.

Cytokines treatment

According to the results of the dose–effect experiment, thehuman AIPC cell lines, Du145 and PC-3M, were plated into25-cm2 flasks at a concentration of 0.5 · 105 cells/mL, fol-

lowed by the addition of 50 ng/mL of NGF (Promega),1,000 U/mL of g-INF (Peprotech, Inc.) or each 24 h later. Themedium and the cytokines were renewed every 2 days, andcells were harvested after 7 days of treatment.

Proliferation assay

The Du145 and PC-3M cells (150 cells/200 mL) were see-ded into a 96-well plate and cultured in the DMEM sup-plemented with 10% of fetal bovine serum at 37�C with 5%CO2. After 24 h, cells were grouped and either 50 ng/mL ofNGF, or 1,000 U/mL of g-INF, or both were added, whileuntreated cells served as the control. Following 7 days oftreatment, the cell number was measured with the abovemethod.

Immunocytochemistry and in situcell death detection

Cells (200 cells/well) were seeded onto 8-mm coverslipsand placed in a 24-well plate. A previous cytokine environ-ment was used to culture the cells for 7 days, and then cellswere rinsed with PBS and fixed with 4% of paraformalde-hyde. The slides were rinsed again and treated with 1 ·permeabilization buffer (eBiosience). Next, they were wa-shed and labeled with proliferating cell nuclear antigen(PCNA), CK5, CK18, or CK19 monoclonal antibodies (Ab-cam) for 2 h at 37�C, followed by incubation with a PE-labeled secondary antibody. The slides were then incubatedwith DAPI (300 nM; Wako pure chemical industries, LTD.)for 2 min and washed. We used the in situ cell death detec-tion kit fluorescein (Roche Diagnostic) to detect cell apop-tosis, according to the manufacturer’s protocol. An antifadesolution (Millipore) was used to seal the slides, and theywere examined using a fluorescent microscope.

Test of androgen sensitivity

The cytokine-pretreated cells (24 wells/group) were see-ded into a 96-well-plate at a density of 500 cells per well. Toexclude the interference of other cytokines and hormones,cells were cultured in 200 mL of a keratinocyte serum-freemedium without bovine pituitary extract, epidermal growthfactor (EGF), leukemia inhibitory factor, NGF, or g-INF.

In each group, 8 wells of cells were incubated with 20 nMDHT (Sigma), and another 8 wells were added with vehicle.The remaining 8 wells contained 20 nM DHT and 20 nM offlutamide, and on the seventh day, the cell number wasmeasured using Cell Titer Blue� reagent.

Flow cytometry

The treated cells were detached using ACCUTASE andwashed with PBS, then re-suspended, and incubated in thedark with 1:200 of anti-CD44 antibody-PE (Novus Biologi-cals, LLC.) and 1:200 of anti-integrin a2b1 antibody-FITC(Life Span Bioscience) for 10 min. For NGF/g-INF-treatedcells, 1:500 of anti-AR antibody (Abcam) and 1:500 of PE-labeled secondary antibody were then added. The cells wereincubated and rinsed for 3 times, followed by the addition of1:200 of anti-integrin a2b1 antibody-FITC (Life SpanBioscience) for 10 min. The labeled cell suspension wascentrifuged and washed with PBS in triplicate. The

CYTOKINES, CSCS, AND AIPC 3373

centrifuged cells were re-suspended in 500 mL of PBS andwere subjected to flow cytometry (FACSCalibur; BectonDickinson), and then the percentage of positive events in thegated population was measured.

Assay of colony formation efficiency

The untreated and cytokine-treated cells were plated in a24-well plate with 200 cells per well in 500 mL of a kerati-nocyte SFM (keratinocyte serum-free medium containing25 mg/mL of bovine pituitary extract, 10 ng/mL of EGF, and2 ng/mL of leukemia inhibitory factor; GIBCO). The mediumwas renewed every 3 days, and after 30 days, we counted thecolonies that contain 32 cells.

Quantitative real-time PCR array

The total RNA was isolated with TRIzol reagents. Quan-titative real-time PCR was carried out using Signal Trans-duction PathwayFinder PCR Array and CancerPathwayFinder PCR Array (SA Bioscience Corporation) tosimultaneously examine the mRNA levels of 168 genes in-volved in 24 signal transduction pathways. The experimentswere performed in triplicate according to the protocol of themanufacturer. The gene expression profiles of the untreatedand treated cells were compared, and the data were nor-malized for either GAPDH or b-actin levels by the DDCtmethod. Following our reported methods [24], quantitativereal-time PCR was used to examine the expression of 2 iso-forms of the FGFR gene (FGFR2), FGFR2-IIIb (forwardprimer: 5¢-TGCTGGCTCTGTTCAATGTG-3¢; reverse primer:5¢-GGCGATTAAGAAGACCCCTA-3¢) and FGFR2-IIIc (for-ward primer: 5¢-ACACCACGGACAAAGAGATT-3¢; reverseprimer: 5¢-GGCGATTAAGAAGACCCCTA-3¢).

Western blotting analysis

Cells were lyzed, and western blotting analysis was per-formed as previously described [20]. Every analyzed sam-ple contained 40 mg of protein. The dilution of anti-AR,anti-FGFR2, anti-telomerase reverse transcriptase (TERT),anti-actin, anti-GAPDF (Abcam), anti-AKT, and anti-p-AKTantibody (Cell Signaling Technology, Inc.) was 1:1,000.

FGFR2 signaling blockageusing a neutralizing antibody

Du145 and PC-3M cells were plated into 25-cm2 flasks at aconcentration of 0.5 · 105 cells/mL in the DMEM. The me-dium was supplemented with 10% of fetal bovine serum andanti-FGFR2-neutralizing antibody (1: 300; Abcam) at 37�C ina 5% CO2 atmosphere. After 7 days, gene expression changeswere measured using the PCR Array. Meanwhile, the cellnumber, AR expression, and proportion of CSCs weremeasured using a proliferation assay, immunocytochemistry,and flow cytometry.

In vivo tumorigenicity and metastasisin NOD/SCID mice

The PC-3M cells that stably expressed the luciferase gene(PC-3M-luc-C6 cells) were purchased from Xenogen. Cellswere resuspended in PBS at a concentration of 1 · 106 cells/

0.2 mL. NOD/SCID mice (n = 24) were injected with fireflyluciferin (150 mg/kg) by intraperitoneal injection. After 7 to8 min, mice were anesthetized with 3% isoflurane, and then1 · 106 PC-3M-luc-C6 cells were injected subcutaneously intothe right dorsal flanks of each mouse. The images of tumorcells were examined using a luminometer (NightOWL IILB981; Berthold Technologies). After 7 days, the mice wereimaged and divided into 4 groups (n = 6). The differentgroups were subcutaneously injected with PBS, NGF (0.1mg/gbody weight), g-interferon (g-IFN) (2,000 U/g body weight),or both once a day for 10 days. Afterward, the tumors wereimaged every 10 days.

Statistical analysis

Significance was determined by a paired t-test in paireddesigning experiments and 1-way analysis of variance inmultiparameter experiments. A value of P < 0.05 was usedconsidered statistically significant.

Results

Optimum concentration of NGF/c-IFNin inhibiting AIPC cell line

Using the dose–effect experiment, the optimum concen-tration of NGF/g-IFN was determined. According to theprevious report, the optimum concentration of NGF is 50 ng/mL; therefore, the dose of g-IFN was adjusted gradually. Theresults indicated that a combination of 50 ng/mL NGF and1,000 IU g-IFN maximized the inhibition of Du145 and PC-3M (Fig. 1A). When the concentration was increased to50 ng/mL NGF and 4,000 IU g-IFN, the effect of inhibitionwas not promoted.

NGF/c-IFN inhibits the proliferative abilityand colony formation of AIPC cell lines,Du145 and PC-3M

After incubation with NGF/g-IFN for 7 days, the cellcounts of Du145 and PC-3M decreased by 62.8% – 5.14%and 69.0% – 2.03%, respectively (Fig. 1B), and the colonyforming efficiency (CFE) decreased by 55.0% – 10.35% and42.6% – 6.20%, respectively, compared to the untreated cells(Table 1). The number of proliferating cell nuclear antigen(PCNA) + cells decreased by 31.6% and 23.0%, respectively(Fig. 1C), and apoptotic cells increased by 46.6% and 37.8%in NGF/g-IFN-treated Du145 and PC-3M cells (Fig. 2A).

NGF/c-IFN treatment leads to re-expressed AR

After 7 days of treatment with NGF and g-IFN, AR ex-pression in Du145 and PC-3M was determined using im-munohistochemical staining and western blotting. Less than2% of untreated Du145 and PC-3M cells expressed AR (Fig.2B2). After treatment with NGF/g-IFN, the percentage ofAR + cells remarkably increased to 37.4% – 4.6% and31.2% – 5.7% in Du145 (Fig. 2B3) and PC-3M cells, respec-tively., Treatment with NGF alone led to 14.3% – 3.7% and11.0% – 5.6% of AR + Du145 and PC-3M cells, respectively,whereas g-IFN treatment caused 21.3% – 2.2% of Du145 cellsand 15.1% – 5.2% PC-3M to express AR. Western blot anal-ysis confirmed the higher expression of AR in NGF/g-IFN-treated Du145 cells (Fig. 2C).

3374 CHEN ET AL.

AR induced by NGF/c-IFN has anticancer function

Cytokine-treated cells were pretreated with vehicle, DHT,and a combination of DHT and flutamide for 7 days, andthen the proliferation rate was detected. The number ofvehicle-treated cells was set at 100%.

With the DHT addition, the pretreatment with NGF aloneincreased the number of Du145 and PC-3M cells by 36.41%and 34.81%, respectively. The number of g-IFN pretreatedDu145 and PC-3M cells also increased by 35.67% and 32.93%,respectively. Surprisingly, the activation of NGF/g-IFN-induced AR by DHT suppressed proliferation of the tumorcells. Compared with vehicle-treated cells, the number ofDHT-treated cells decreased by 26.62% and 24.28% in Du145and PC-3M cells after 7 days of DHT treatment (Fig. 2D);however, the effect of DHT could be blocked by flutamide(Fig. 2D). Immunohistochemical staining showed that re-expressed AR was located in the cytoplasm and the nucle-olus, and after exposure to DHT, the re-expressed ARtranslocated to the nucleoli (Fig 2. B4).

NGF/c-IFN decreased the proportion of CSCs

After incubation with NGF/g-IFN for 7 days, the per-centage of CD44 + integrina2b1 + decreased by 43.97% –8.20% (Fig. 3A) and 37.69% – 6.62% (Fig. 3B) in Du145 andPC-3M cells. The immunofluorescence results indicated that

integrina2b1 + cells decreased by 34% – 2.3% in NGF/g-IFN-treated Du145 cells (Fig. 3C). CK5 expression is alsodownregulated in the 2 cell lines (date not shown). Untreatedcells can partly form sphere-like colonies, but NGF/g-IFN-treated cells rarely arrange this figure (Fig. 3D). The FCSresults indicated that AR expressed on integrina2b1- cells(Fig. 4A). As suggested in the results, the most of remainingcells may not be CSCs after NGF/g-IFN treatment. Of them,parts of cells are more differentiated and can response toandrogen.

FGFR2, TERT, and Wnt are reducedby NGF/c-IFN treatment

The genes with altered expression in response to NGFand/or g-IFN treatment are listed in Table 2. Among thesegenes, the downregulation of FGFR2 expression was inagreement with the differentiation of the prostate CSCs. InNGF/g-IFN-treated cells, FGFR2 was downregulated by 9.42(Du145) and 7.15 (PC-3M) times compared with untreatedcells. Western blot analysis confirmed the decreased ex-pression of FGFR2 in NGF/g-IFN-treated cells (Fig. 2C). Inall of the real-time PCR super-array microarrays, TERT,wingless-type MMTV integration site family member 1(Wnt1) and Wnt2 were found to be remarkably down-regulated in NGF/g-IFN-treated cells (Table 2).

3.7. Blockade of FGFR2 signaling showeda similar effect as NGF/c-IFN treatment

Compared with controls at 7 days of incubation, thenumber of neutralizing FGFR2 antibody-treated cells de-creased by 54.6% – 6.2% and 47.9% – 6.8% in Du145 and PC-3M cells, respectively. The CFE was decreased by46.2% – 9.6% and 38.7% – 11.0%, respectively. CD44 + in-tegrina2b1 + cells decreased by 32.7% – 6.5% and 27.6% –5.1%, respectively, and AR + cells were 7.4% – 1.6% and4.6% – 0.8% in Du145 and PC-3M cell lines, respectively.

FIG. 1. NGF/g-IFN inhibits the prolifera-tion of AIPC cell lines. (A) The optimumconcentration of NGF/g-IFN was deter-mined using the dose–effect experiment. Thecombination of 50 ng/mL NGF and 1,000 IUg-IFN maximized the inhibition of AIPCcells. An assay was performed in 8 wells pergroup with a standard error shown; 2 aster-isks (**) represent 1-way analysis of variance,P < 0.001. (B) NGF/g-IFN inhibits the prolif-eration of AIPC cells. After 7 days of treat-ment, the number of NGF/g-IFN-treatedcells decreased by 62.8% – 5.14% and69.0% – 2.03%, respectively, in Du145 andPC-3M. An assay was performed in 8 wellsper group with a standard error shown; 2asterisks (**) represent paired t-test, P < 0.001.(C) PCNA expression in DU145 and PC-3Mcells. Blue: DAPI; green: PCNA. AIPC: an-drogen-independent prostate cancer; NGF,nerve growth factor; g-IFN, g-interferon;PCNA, proliferating cell nuclear antigen.Color images available online at www.liebertpub.com/scd

Table 1. Clone-Forming Efficiency

of the Treated Cells

Cells Du145 PC-3M

Untreated 16.4 – 1.7 19.5 – 1.6NGF-treated 14.3 – 1.4 16.6 – 2.3g-IFN-treated 11.5 – 1.2 13.0 – 1.4NGF/g-IFN 7.30 – 1.3 11.2 – 1.6

NGF, nerve growth factor; g-IFN, g-interferon.

CYTOKINES, CSCS, AND AIPC 3375

After the addition of DHT, the proliferation rate of the cellsincreased by 7.4% – 2.7% and 5.7% – 3.4% in both cell lines.

NGF/c-IFN downregulates FGFR2 III-cand inhibits AKT signaling

The expression of FGFR2 isoforms was examined usingreal-time PCR. After NGF/g-IFN treatment, FGFR2 III-b andFGFR2 III-c were downregulated by 0.8-fold and 7.4-fold inDu145 and 1.06-fold and 3.76-fold in a PC-3M cell line, re-spectively (Fig. 4B). With NGF/g-IFN treatment, the phos-phorylation of AKT was inhibited at Serine 473 while totalAKT expression was not significantly altered. (Fig. 4C).

NGF/c-IFN inhibits prostate cancer in vivo

PC-3M-luc-C6 cells (1 · 106), which are stably expressedluciferase, were subcutaneously transplanted into NOD/SCID mice (n = 24). Bioluminescence imaging was used todetect tumor growth after cell transplantation. In the PBS-(n = 6) and NGF- (n = 6) treated groups, the pulmonarymetastasis of cancer cells was detected on day 57 post-

transplantation, while in the NGF/g-IFN- (n = 6) treatedgroup (Fig. 5), only small-size tumors were observed (Fig. 5).This indicated that NGF/g-IFN administration inhibitedprostate tumor growth.

Discussion

The mechanism responsible for changing prostate cancerinto an androgen-independent type remains unclear. Thetreatment of AIPC remains a challenge for urologists. Onepossible explanation for its formation is that the originalminority of CSCs becomes dominant after a long time ofexposure with ADT. Prostatic CSCs neither express AR norrespond to androgen [25,26], but mature luminal cells ex-press AR and can be stimulated by androgen. It is possiblethat castration and ADT can eliminate highly differentiatedcells and CSCs that remain existent. AIPC may be the cellpopulation that contains the majority of undifferentiatedprostate CSCs. The blockage of CSC differentiation may be acause of AIPC formation; therefore, once the differentiationpathway is recovered, CSCs in AIPC would be differenti-ated, and AIPC may be reverted to an androgen-sensitive

FIG. 2. NGF/g-IFN treatment affected cell apoptosis, AR expression and AR function in AIPC cell lines. (A) In situ celldeath detection indicated that apoptotic cells increased by 46.6% and 37.8% in NGF/g-IFN-treated Du145 and PC-3M cells.Blue: DAPI; cyan: apoptotic cells. (B) NGF/g-IFN treatment leads to re-expressed AR. (Black arrows point to AR positive cells)(1) Control Du145 cells where no primary antibody was added. (2) Untreated Du145 cells weakly expressed AR. (3) NGF/g-IFN-treated Du145 cells expressed AR. After 7 days of NGF/g-IFN treatment, AR was found in the cytoplasm and thenucleus. (4) After DHT addition, AR was transferred to the nucleus in NGF/g-IFN-treated Du145 cells. (C) Western blotresults showed AR was highly expressed, and FGFR2 was downregulated in Du145 cells after 7 days of NGF/g-IFNtreatment. (D) NGF/g-IFN-induced AR has anticancer functions. With the DHT addition, pretreatment with NGF aloneincreased the number of Du145 and PC-3M cells by 36.41% and 34.81%, respectively. g-IFN alone also increased the numberof cells by 35.67% and 32.93%. The number of NGF/g-IFN-pretreated Du145 and PC-3M cells decreased by 26.62% and24.28% after 7 days of DHT administration. The effect of DHT could be blocked by flutamide. An assay was performed in 8wells per group with standard error shown, where two asterisks (**) represent paired t-test P < 0.001. AR, androgen receptor;DHT, dihydrotestosterone; FGFR2, fibroblast growth factor receptor 2. Color images available online at www.liebertpub.com/scd

3376 CHEN ET AL.

type and be easily treated. Prostate CSC markers CD44 andintegrina2b1 were highly expressed in the AIPC cancer celllines PC3 and DU145 [27]. In our previous report [24], theproportion of cancer stem-like cells in DU145 is higher thanthat of primary cancer, suggesting that AIPC is associatedwith CSCs.

As the type I AIPC cell lines, Du145 and PC-3M cannotentirely represent AIPC in patients due to a lack in an AR,

but they are the appropriate model since the cell lines scar-cely react to androgen.

After Du145 and PC-3M cells are chronically exposed tothe NGF/g-IFN environment for 1 week, the number ofCD44 + a2b1 + cells dramatically decreased along with CK5 +

cells, suggesting that CSCs were either differentiated or in-duced to apoptotic death. Meanwhile, both the proliferativepotential and CFE declined, and AR was re-expressed in

FIG. 3. NGF/g-IFN decreased the proportion of cancer stem cells. (A) Flow cytometry dotplots showed that CD44 +

integrina2b1 + cells in Du145 cell lines decreased by 43.97% after 7 days of NGF/g-IFN treatment. (B) Flow cytometrydotplots show that CD44 + integrina2b1 + cells in the PC-3M cell line decreased by 37.69% after 7 days of NGF/g-IFNtreatment. (C) Immunofluorescence images indicated that integrina2b1 + cells in the Du145 cell line decreased by 34% – 2.3%after 7 days of NGF/g-IFN treatment. Blue: DAPI; green: intergrina2b1 + (white arrows). (D) Du145 and PC-3M cells can formsphere-like clones and flat clones. (1) Sphere-like clone formed by Du145 cells after 7 days of culture. (2) Flat clone formed byDu145 cells after 12 days of culture. (3) Sphere-like clone formed by PC-3M cells after 7 days of culture. (4) Flat clone formedby PC-3M cells after 12 days of culture. Color images available online at www.liebertpub.com/scd

FIG. 4. NGF/g-IFN induced AR ex-pression in Intergrina2b1- cells. The cy-tokines mainly decreased the expressionof FGFR2 IIIc and resulted in a reductionof phosphorylated AKT at Serine 473. (A)The AR + cells and intergrina2b1 + cellswere in different populations. (B) WithNGF/g-IFN treatment, FGFR2 III-b andFGFR2 III-c were, respectively, down-regulated by 0.8-fold and 7.4-fold inDu145, 1.06-fold and 3.76-fold in PC-3Mcell line. (C) With NGF/g-IFN treatment,the phosphorylation of AKT was in-hibited at Serine 473. Total AKT expres-sion had no significant changes. Colorimages available online at www.liebertpub.com/scd

CYTOKINES, CSCS, AND AIPC 3377

each cell line. Noticeably, AR was only expressed in a2b1-

cells, confirming that AR was only expressed in more dif-ferentiated cells instead of stem cells or progenitor cells.When androgen was added into NGF/g-IFN-treated cells,the proliferation of cells was inhibited rather than increased,a highly unexpected result. These results suggest that theinhibition of cell proliferation was due to the activation ofAR, which plays an antiproliferation role in the particularcytokine environment. However, the function of AR inducedby either NGF or g-IFN alone is apparently different thanwith AR induced by NGF/g-IFN. The mechanisms are un-clear, and there is a possibility that some molecules associ-

ated with AR are altered by the treatment of NGF/g-IFN, butnot by either NGF or g-IFN alone. Nevertheless, this findingdemonstrates that the different cytokine environments mayinduce divergent functions of AR, and may provide a cluefor the mechanisms presented when young males rarelyhave either benign prostatic hyperplasia or prostate cancer.The cytokine environment in a young man is different fromthat in aged males [28]. With aging, changes in the cytokineenvironment may influence differentiation of the prostateand stimulate the function of AR. NGF and IFN are associ-ated with age and androgen levels [19,21]; thus, NGF/g-IFNmay be an important portion of the cytokine environment.

Table 2. Differentially Expressed Genes That Were Screened Out

with Real-Time PCR Microarrays in Du145 and PC-3M

Treatment

Gene NGF/con c-IFN/con NGF/c-IFN/conAnti-FGFR2antibody/con

NGF/c-IFN +

antibody/con

Du145 cellsBcl-2 - 1.63 - 2.56 - 2.13 1.01 1.23CCL20 - 9.375 - 9.42 - 12.73 - 1.94 - 2.82FGFR2 1.04 - 2.65 - 10.42 — —TANK - 36.46 - 20.72 - 30.15 1.17 - 1.01TERT - 3.26 - 2.62 - 4.15 - 4.45 - 6.02WNT1 - 7.57 - 8.16 - 11.1 - 4.96 - 7.11WNT2 - 1.94 - 1.61 - 2.08 - 2.38 1.96

PC-3M cellsBcl-2 - 1.28 - 1.48 - 1.52 1.19 1.42CCL20 - 2.02 - 2.29 - 1.10 1.42 3.18FGFR2 1.12 - 2.16 - 8.75 — —TERT - 2.67 - 1.25 - 2.09 - 1.57 - 1.7WNT1 1.05 - 1.66 - 3.12 - 2.04 - 1.14WNT2 - 2.77 - 2.72 - 5.53 - 2.67 - 2.4

FIG. 5. Bioluminescence imaging of transplanted prostate cancer in vivo. (A) On the seventh day, the mice were injectedwith cytokines for 10 days. Every 10 days, the tumors were detected using bioluminescence imaging, and the exposure timewas 1 min. (B) On the 57th day, pulmonary metastasis of the cancer cells was detected with bioluminescence imaging (theexposure time was 5 min). (C) On the 60th day, pulmonary metastasis of cancer cells was detected in either PBS- or NGF-treated mice. In NGF/g-IFN-treated mice, tumors remained a small mass. (D) Growth carves indicated that NGF/g-IFNinhibited tumor formation in vivo. Color images available online at www.liebertpub.com/scd

3378 CHEN ET AL.

Our in vivo experiments showed that NGF/g-IFN in-hibited tumor growth in NOD/SCID mice, which are im-munodeficient animals without T, B, and NK cells. Theinhibitory effect of NGF/g-IFN is unlikely due to a manip-ulation of the host’s acquired immune system. In this case, acytokine environment may play a more important role ininhibiting tumor growth.

NGF/g-IFN altered the expression levels of certaingenes, making TERT, Wnt1, Wnt2, and FGFR2 noticeable.TERT repairs telomeres to keep the proliferative potential,and is also a pluripotency-associated gene and a stem cellmarker [29,30]. Wnt1 and Wnt2 are members of the wing-less-type MMTV integration-site family, which is associatedwith the maintenance of an undifferentiated status of cells.Activation of Wnt1 induces a complex signaling cascadethat ultimately leads to the increased expression of over 50genes. Wnt and TERT are able to regulate each other[31], and the downregulation of Wnt and TERT may be theresult of differentiation of the prostate CSCs in the AIPCcell lines.

The downregulation of FGFR2 expression is in agreementwith the differentiation of the prostate CSCs, indicating thatNGF/g-IFN worked through the downregulation of FGFR2.FGFR2 is a member of the tyrosine kinase receptor family,which is expressed on the membrane and binds with FGF4,FGF7, FGF8, or FGF10. It plays a crucial role in signaltransduction and development, and either promotes or in-hibits differentiation in tumors [32–35]. To examine theFGFR2 pathway, we used an anti-FG.FR2-neutralizing anti-body to block the function of the FGFR2 receptor. Comparedwith controls, the FGFR2-blocked cells were further differ-entiated and had a lower proliferative potential, similar tothe NGF/g-IFN-treated cells. However, the gene expressionand cell nature of FGFR2-blocked cells were not alwaysconsistent with that of the NGF/g-IFN-treated cells. AR re-expressed in the FGFR2-blocked cells showed no anticancerfunction, suggesting that other mechanisms may also be in-volved in the NGF/g-IFN pathway. Nevertheless, this studysuggested that NGF/g-IFN induced differentiation of CSCsmainly through inhibiting FGFR2 signaling. On the contrary,one study found that in primary prostate cancer cells, FGFR2plays the role as a proliferation inhibitor and as a differen-tiation inductor [36]. We further investigated the isoforms ofFGFR2: the FGFR2 IIIb and IIIc. Results indicated that NGF/g-IFN mainly decreased the expression of FGFR2 IIIc andresulted in a reduction of phosphorylated AKT at Serine 473.These data suggest that the FGFR2 IIIc isoform affectsphosphorylation of AKT and plays a role as a proliferationpromoter and as a differentiation inhibitor. Furthermore, theanticancer functions of NGF/g-IFN may result from multiplegene expression changes, including FGFR2, p-AKT, TERT,Wnt1, and Wnt2 that are induced by cytokines.

Aging [37] and ADT interfere with the androgen level andinternal cytokine environment, and may lead to the prolif-eration of CSCs. The correction of the internal cytokine im-balances and the induction of differentiation of CSCs mayoffer new therapeutic designs for the treatment of AIPC.

Acknowledgments

We thank Prof. Lian-hua Yin, Prof. Si-feng Chen, andDr. Xiao-bo Li for giving critical advice to our study. This

research was supported by the NSFC (Natural ScienceFoundation of China, no. 30901705) and the Doctoral Fund ofMinistry of Education of China (no. 20090071120024).

Author Disclosure Statement

The authors declare that there is no conflict of interest thatcould be perceived as prejudicing the impartiality of the re-search reported.

References

1. Labrie F, A Dupont, L Cusan, J Gomez, J Emond and GMonfette. (1990). Combination therapy with flutamide andmedical (LHRH agonist) or surgical castration in advancedprostate cancer: 7-year clinical experience. J Steroid Biochem37:943–950.

2. Huggins C and CV Hodges. (2002). Studies on prostaticcancer: the effect of castration, of estrogen and of androgeninjection on serum phosphatases in metastatic carcinoma ofthe prostate. J Urol 168:9–12.

3. Macfarlane RJ and KN Chi. (2010). Research in castration-resistant prostate cancer: what does the future hold? CurrOncol Suppl 2:80–86.

4. Scher HI, S Halabi, I Tannock, M Morris, CN Sternberg, MACarducci, et al. (2008). Design and end points of clinicaltrials for patients with progressive prostate cancer and cas-trate levels of testosterone: recommendations of the prostatecancer clinical trials working group. J Clin Oncol 26:1148–1159.

5. Heidenreich A, G Aus, M Bolla, S Joniau, VB Matveev, HPSchmid, et al. (2008). European Association of Urology. EAUguidelines on prostate cancer. Eur Urol 53:68–80.

6. Pajonk F, E Vlashi and WH McBride. (2010). Radiation re-sistance of cancer stem cells: the 4 R’s of radiobiology re-visited. Stem Cells 28:639–648.

7. Silvestre DC, JR Pineda, F Hoffschir, JM Studler, MAMouthon, F Pflumio, et al. (2011). Alternative lengtheningof telomeres in human glioma stem cells. Stem Cells 29:440–451

8. Taipale J and PA Beachy. (2001). The Hedgehog and Wntsignalling pathways in cancer. Nature 411:349–354.

9. Rizzo S, JM Hersey, P Mellor, W Dai, A Santos-Silva, DLiber, et al. (2011). Ovarian cancer stem cell-like side pop-ulations are enriched following chemotherapyy and over-express EZH2. Mol Cancer Ther 10:325–335.

10. Yuki K, A Natsume, H Yokoyama, Y Kondo, M Ohno, TKato, et al. (2009). Induction of oligodendrogenesis in glio-blastoma-initiating cells by IFN-mediated activation ofSTAT3 signaling. Cancer Lett 284:71–79.

11. Korkaya H and MS Wicha. (2009). HER-2, notch, and breastcancer stem cells: targeting an axis of evil. Clin Cancer Res15:1845–1847.

12. Chou YF, HH Chen, M Eijpe, A Yabuuchi, JG Chenoweth, PTesar, et al. (2008). The growth factor environment definesdistinct pluripotent ground states in novel blastocyst-derived stem cells. Cell 135:449–461.

13. DeSchryver-Kecskemeti K, K Balogh and E Neet. (1987).Nerve growth factor and the concept of neural-epithelialinteractions. Arch Pathol Lab Med 111:833–835.

14. Graham CW, JH Lynch and D Djakiew. (1992). Distributionof nerve growth factor like protein and nerve growth factorreceptor in human benign prostatic hyperplasia and pros-tatic adenocarcinoma. J Urol 147:1444–1447.

CYTOKINES, CSCS, AND AIPC 3379

15. Zhu ZW, H Friess, L Wang, FF Di Mola, A Zimmermannand MW Buchler. (2000). Down regulation of nerve growthfactor in poorly differentiated and advanced human esoph-ageal cancer. Anticancer Res 20:125–132.

16. Missale C, F Boroni, M Losa, M Giovanelli, A Zanellato, RDal Toso, et al. (1993). Nerve growth factor suppresses thetransforming phenotype of f human prolactinomas. ProcNatl Acad Sci U S A 90:7961–7965.

17. Missale C, A Codignola, S Sigala, A Finardi, M Paez-Pereda,E Sher, et al. (1998). Nerve growth factor abrogates the tu-morigenicity of human small cell lung cancer cell lines. ProcNatl Acad Sci U S A 95:5366–5371.

18. Tsim TY, EY Wong, MS Leung and CC Wong. (2004). Ex-pression of axon guidance molecules and their related genesduring development and sexual differentiation of the olfac-tory bulb in rats. Neuroscience 123:951–965.

19. Arsenijevic Y and E Tribollet. (1998). Region-specific effect oftestosterone on oxytocin receptor binding in the brain of theaged rat. Brain Res 78:5167–5170.

20. Sigala S, N Tognazzi, MC Rizzetti, I Faraoni, C Missale, EBonmassar, et al. (2002). Nerve growth factor induces the re-expression of functional androgen receptors and p75(NGFR)in the androgen-insensitive prostate cancer cell line DU145.Eur J Endocrinol 147:407–415.

21. Fisher AD, MA Crowe, EM O’Nuallain, ML Monaghan, JALarkin, P O’Kiely, et al. (1997). Effects of cortisol on in vitrointerferon-gamma production, acute-phase proteins, growth,and feed intake in a calf castration model. J Anim Sci 75:1041–1047.

22. Untergasser G, E Plas, G Pfister, E Heinrich and P Berger.(2005). Interferon-gamma induces neuroendocrine-like dif-ferentiation of human prostate basal-epithelial cells. Prostate64:419–429.

23. Shikata A, T Shikata, Y Sotozono, H Hosoi, T Matsumura, TSugimoto, et al. (2000). Neuronal differentiation in humanneuroblastoma cells by nerve growth factor following TrkAup-regulation by interferon-gamma. Med Pediatr Oncol 34:394–401.

24. Wei C, W Guomin, L Yujun and Q Ruizhe. (2007). Cancerstem-like cells in human prostate carcinoma cells DU145: theseeds of the cell line? Cancer Biol Ther 6:763–768.

25. Collins AT, FK Habib, NJ Maitland and DE Neal. (2001).Identification and isolation of human prostate epithelialstem cells based on alpha(2)beta(1)-integrin expression. JCell Sci 114:3865–3872.

26. Collins AT, PA Berry, C Hyde, MJ Stower and NJ Maitland.(2005). Prospective identification of tumorigenic prostatecancer stem cells. Cancer Res 65:10946–10951.

27. Liu AY. (2000). Differential expression of cell surface mole-cules in prostate cancer cells. Cancer Res 60:3429–3434.

28. Stowe RP, MK Peek, MP Cutchin and JS Goodwin. (2010).Plasma cytokine levels in a population-based study: relationto age and ethnicity. J Gerontol A-Biol 65:429–433.

29. Blasco MA. (2007). Telomere length, stem cells and aging.Nat Chem Biol 3:640–649.

30. Flores I, ML Cayuela and MA Blasco. (2005). Effects of tel-omerase and telomere length on epidermal stem cell be-havior. Science 309:1253–1256.

31. Park JI, AS Venteicher, JY Hong, J Choi, S Jun, M Shkreli,et al. (2009). Telomerase modulates Wnt signalling by as-sociation with target gene chromatin. Nature 460:66–72.

32. Harimoto N, K Taguchi, K Shirabe, E Adachi, Y Sakaguchi,Y Toh, et al. (2010). The significance of fibroblast growthfactor receptor 2 expression in differentiation of hepatocel-lular carcinoma. Oncology 78:361–368.

33. Zhang Y, H Wang, S Toratani, JD Sato, M Kan, WLMcKeehan, et al. (2001). Growth inhibition by keratinocytegrowth factor receptor of human salivary adenocarcinomacells through induction of differentiation and apoptosis. ProcNatl Acad Sci U S A 98:11336–11340.

34. Visco V, FA Bava, F d’Alessandro, M Cavallini, V Ziparoand MR Torrisi. (2009). Human colon fibroblasts inducedifferentiation and proliferation of intestinal epithelial cellsthrough the direct paracrine action of keratinocyte growthfactor. J Cell Physiol 220:204–213.

35. Yoshino M, T Ishiwata, M Watanabe, T Matsunobu, O Ko-mine, Y Ono, et al. (2007). Expression and roles of kerati-nocyte growth factor and its receptor in esophageal cancercells. Int J Oncol 31:721–728.

36. Heer R, AT Collins, CN Robson, BK Shenton and HY Leung.(2006). KGF suppresses alpha2beta1 integrin function andpromotes differentiation of the transient amplifying popula-tion in human prostatic epithelium. J Cell Sci 119:1416–1424.

37. Arianayagam R, M Arianayagam, S McGrath and P Rashid.(2010). Androgen deficiency in the aging man. Aust FamPhysician 39:752–755.

Address correspondence to:Dr. Wei Chen

Department of UrologyZhongshan Hospital of Fudan University

No. 180, Fenglin RoadShanghai 200032

People’s Republic of China

E-mail: chen.wei3@zs-hospital.sh.cn

Dr. Guo-min WangDepartment of Urology

Zhongshan Hospital of Fudan UniversityNo. 180, Fenglin Road

Shanghai 200032People’s Republic of China

E-mail: wang.guomin@zs-hospital.sh.cn

Received for publication March 15, 2012Accepted after revision June 25, 2012

Prepublished on Liebert Instant Online June 25, 2012

3380 CHEN ET AL.