flightless i homolog represses prostate cancer progression ...chenguang wang3, and ke chen1,4...

Biology of Human Tumors

Flightless I Homolog Represses Prostate CancerProgression through Targeting AndrogenReceptor SignalingTao Wang1,Wen Song1, Yuan Chen1, Ruibao Chen1, Zhuo Liu1, Licheng Wu1,Mingchao Li1, Jun Yang1, Liang Wang2, Jihong Liu1, Zhangqun Ye1,Chenguang Wang3, and Ke Chen1,4

Abstract

Purpose: Flightless I (FLII), member of the gelsolin superfamilyof actin-remodeling proteins, functions as a transcriptional core-gulator.We aim to evaluate a tumor-suppressive function of FLII inregulating androgen receptor (AR) in prostate cancer progression.

Experimental Design: We examined FLII protein and mRNAexpression in clinical prostate cancer specimens by immunohis-tochemistry. Kaplan–Meier analysis was conducted to evaluatethe difference in disease-overall survival associated with theexpression levels of FLII and AR. Prostate cancer cells stablyexpressing FLII or shRNA knockdown were used for functionalanalyses. Immunoprecipitation, Luciferase reporter, and immu-nofluorescence staining assays were performed to examine thefunctional interaction between FLII and AR.

Results:Our analysis of the expression levels of FLII in a clinicalgene expression array dataset showed that the expression of FLII

was positively correlated with the overall survival of prostatecancer patients exhibiting high levels of AR expression. Exami-nation of protein and mRNA levels of FLII showed a significantdecrease of FLII expression in humanprostate cancers. AR and FLIIformed a complex in a ligand-dependent manner through theligand-binding domain (LBD) of AR. Subsequently, we observeda competitive binding to AR between FLII and the ligand. FLIIinhibited AR transactivation and decreased AR nuclear localiza-tion. Furthermore, FLII contributed to castration-sensitive andcastration-resistant prostate cancer cell growth throughAR-depen-dent signaling, and reintroduction of FLII in prostate cancer cellssensitized the cells to bicalutamide and enzalutamide treatment.

Conclusions: FLII plays a tumor-suppressive role and serves asa crucial determinant of resistance of prostate cancer to endocrinetherapies. Clin Cancer Res; 22(6); 1531–44. �2015 AACR.

IntroductionProstate cancer is one of the most common malignancies and

the second most common cause of cancer-related deaths in men(1, 2). The progression of prostate cancer normally goes fromcastration-sensitive to castration-resistant, inevitably developinghighlymetastatic properties (3, 4). Androgen deprivation therapy(ADT) is the main therapeutic approach for patients with meta-static prostate cancer (5). Prostate cancer patients respond initiallyto ADT, but in later stages the tumor becomes hormone refractoryand more aggressive, eventually leading to poor prognosis (5).

Arrays of treatments, including secondary hormonal therapies, areavailable for the treatment of metastatic prostate cancer andcastration-resistant prostate cancer (CRPC), which show efficacywhen administered with ADT (6). Continuation of ADT is recom-mended for CRPC treatment as therapies are added. New sec-ondary hormonal therapies include abiraterone, targeting theCYP17 enzyme family, and enzalutamide (Xtandi), an androgenreceptor inhibitor with heightened binding specificity (6–8). Theoptimal decision-making process for metastatic CRPC treatmentoption remains unclear, pending further research and experience.

Androgen receptor (AR), a member of nuclear transcriptionfactor family, promotes tumor growth and survival-related genetranscription via androgen-mediated signaling pathways in cas-tration-sensitive prostate cancer (CSPC; ref. 9). The ligand-inde-pendent activation of AR may be initiated by various biologicalterations including gene mutation, gene amplification, ARcoactivator overexpression, and persistent intraprostatic andro-gens, which often leads to the failure of ADT in CRPC (9, 10). InCRPC, prostate cancer cells still rely on intracellular androgensand, to a greater extent, on active AR for growth and survival (11).Therefore, potent antiandrogens that efficiently disrupt the func-tions (signaling) of AR are envisioned to be effective drugs for alltypes of prostate cancers (11). Blocking AR expression and acti-vation has become one of the most effective therapies in clinicalmanagement of this disease (12).

FLII gene, originally identified in Drosophila melanogaster,encodes a protein containing an N-terminal leucine-rich repeat(LRR) domain and a C-terminal gelsolin-like domain (GLD;

1Department of Urology, Institute of Urology, Tongji Hospital, TongjiMedical College, Huazhong University of Science and Technology,Wuhan, Hubei, China. 2Department of Radiology, Tongji Hospital,Tongji Medical College, Huazhong University of Science and Technol-ogy, Wuhan, Hubei, China. 3Key Laboratory of Tianjin Radiation andMolecular Nuclear Medicine, Institute of Radiation Medicine, PekingUnion Medical College & Chinese Academy of Medical Sciences,Tian-jin, China. 4Union Hospital, Tongji Medical College, Huazhong Univer-sity of Science and Technology,Wuhan, Hubei, China.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Corresponding Author: Ke Chen, Department of Urology, Institute of Urology,Tongji Hospital, Tongji Medical College, Huazhong University of Science andTechnology, Jiefang Avenue No.1095, Wuhan 430030, Hubei, China. Phone:027-8366-3405; Fax: 027-8366-3406; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-15-1632

�2015 American Association for Cancer Research.

ClinicalCancerResearch

www.aacrjournals.org 1531

on April 28, 2021. © 2016 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst November 2, 2015; DOI: 10.1158/1078-0432.CCR-15-1632

http://clincancerres.aacrjournals.org/

ref. 13). Gelsolin protein family has previously been identified asa hormone nuclear receptor coactivator (14). LRR of FLII isimplicated in protein–protein interaction and many have iden-tified that these binding partners and GLD of FLII have someunique regulatory function (13). PI3K and small GTPases mayregulate the subcellular localization of FLII between cytoplasmand nucleus (15). FLII has been found to interact with variousproteins important for cellular signaling (15). FLII may functionas a transcriptional coregulator which positively or negativelyregulates the activity of transcription factors (15). FLII can berecruited by hormone-activated nuclear receptors to the promo-ters of target genes to serve as a coactivator of the nuclear receptortranscription complex (14, 16, 17). FLII also negatively regulatesPPARg , b-catenin, and ChREBP-mediated transcription in cancercells (15, 18, 19). It is currently unknownwhether FLII plays a rolein AR function during prostate cancer development.

In this study, we investigated FLII expression and found it to beinversely correlated with PSA levels in prostate cancer. Patientswith higher FLII expression in their tumor have improved overallsurvival. In human prostate cancer cells, FLII formed a complex ina ligand-dependent manner through the ligand-binding domain(LBD) of AR. Our results indicated competitive relationshipbetween FLII and ligand in binding to the AR. FLII repressed theAR expression, inhibited AR transactivation and modulated ARcytoplasm/nucleus translocation. Furthermore, FLII contributesto the human CSPC and CRPC cell growth, invasion, and migra-tion through AR-dependent signaling. Reintroduction of FLII incastration-sensitive and castration-resistant cells sensitized thecells to bicalutamide and enzalutamide treatment. Our findingsrevealed that FLII functions as a novel AR repressor and inhibitsprostate cancer progression.

Materials and MethodsCell culture, plasmid construction, reporter genes, reagents,expression vectors, and DNA transfection

Human prostate cancer LNCaP cells were obtained from ATCCand maintained in RPMI1640 medium (Invitrogen) supplemen-ted with 10% FBS or 10% charcoal/dextran–stripped FBS (cFBS).LAPC-4 andC4-2 wasmaintained in RPMI1640medium (Invitro-gen) supplemented with 10% FBS. pIREpuro-HA-FLII, pIREpuro-FLAG-FLII, pIREpuro-FLAG-FLII-LRR, and pIREpuro-FLAG-FLII-GLD were provided by Dr. Tong (Shanghai Jiao Tong University

School of Medicine, Shanghai, China) (15). GST-FLII-LRR, GST-FLII-GLD-A, and GST-FLII-GLD-B were provided by Dr. Stallcup(University of Southern California, Los Angeles, CA) (16). GAL4-AR and truncated mutants were provided by Dr. Balk (HarvardMedical School, Boston, MA) and Dr. Chang (University ofRochester Medical Center, Rochester, MN) (20, 21). Plentiviral-vector, GIPZ-shFLII-1, and GIPZ-shFLII-4 were purchased fromThermo Scientific. PSA-Luc and ARE4-Luc reporter genes are de-scribed in refs. 9 and 22. LNCaP andC4-2 cells infectedwithGIPZ-Vector, shFLII-1, shFLII-4. Positive cells were selected by FCM.

Tissue samplesHuman prostate cancer tissue arrays were purchased from

Biomax (PR956B). All the detailed information including Pathol-ogy diagnosis, clinical stage, Gleason scores, PSA level, andsurvival data was directly shown online (http://www.biomax.us/tissue-arrays/Prostate/PR956b).

A collection of 57 different kinds of fresh-frozen prostate cancertumor specimens (including 17 preclinical patients) and 20tumor normal adjacent tissues were from Tongji Hospital, TongjiMedical College, Huazhong University of Science and Technol-ogy. All tumor samples were collected immediately after thesurgical removal and snap-frozen in liquid nitrogen. These studieswere carried out in accordance with the approved guidelines. Allthe experimental protocols were approved by the InstitutionalReview Board of the Tongji Hospital, Tongji Medical College,Huazhong University of Science and Technology.

Overall survival analysis in GSE21034 databaseA normalized mRNA expression dataset for prostate adenocar-

cinomawas downloaded from the cBioPortal for cancer genomicsand used to evaluate expression of FLII and AR transcript levels(23). This dataset includes mRNA profiles for 131 primary tumorand 19 metastasis samples. Overall survival analysis was calcu-lated for these transcripts for primary tumor samples. All thedetailed information of prostate cancer patients including pathol-ogy diagnosis, clinical stage, Gleason scores, and survival datacould be directly downloaded as an Excel sheet from the cBio-Portal website (http://www.cbioportal.org/).

Immunohistochemistry and immunofluorescence stainingImmunohistochemical (IHC) analysis of human prostate can-

cer was conducted using a polyclonal FLII antibody. Four-micrometer sections were prepared from paraffin-embeddedC4-2 tumor tissues derived from nude mice, and tissues wereextracted from paraffin. Tumor tissues were stained with primaryantibody including AR (SC-13062, Santa Cruz Biotechnology),FLII (SC-21716, Santa Cruz Biotechnology), and PSA (SC-7316,Santa Cruz Biotechnology), Ki67 (RM-9106-S1, Thermo Scientif-ic). For immunofluorescent staining, cells were fixed in 3.7%paraformaldehyde, permeatedby0.1%TritonX-100, andblockedwith 5% goat serum in PBS for 1 hour at 37�C, followed byincubation with FITC and rhodamine secondary antibodies.

RNA isolation and real-time PCR analysisTotal RNAs were extracted with TRizol reagent according to the

manufacturer's instruction (Invitrogen). To determine the mRNAlevels of FLII, AR, and PSA, total RNAs were reversely transcribedby iScript cDNA Synthesis Kit (Bio-Rad Laboratories), respective-ly, according to the manufacturer's instructions (2).

Translational Relevance

Prostate cancer is the second leading cause of cancer-relateddeath in men worldwide. Castration-resistant prostate canceris an untreatable form of prostate cancer, which bypassed thenormal dependence on androgens for growth and survival.Here, we identified a biochemical and functional link betweenFLII and androgen receptor (AR). Reduced expressionof FLII inadvanced prostate cancer may lead to AR overactivation,tumor growth, and development of castration-resistant phe-notype of late-stage prostate cancer. This study provides, forthe first time, compelling evidence, which rationalizes thetargeting strategy to reestablish the functional interactionbetween FLII and AR signaling in clinical practice to treatprostate cancer.

Wang et al.

Clin Cancer Res; 22(6) March 15, 2016 Clinical Cancer Research1532




Immunoprecipitation and Western blot analysisImmunoprecipitation andWestern blot assays were conducted

in LNCaP and 293T cells as indicated. Cells were pelleted andlysed in buffer (50 mmol/L HEPES, pH 7.2, 150 mmol/L NaCl,1 mmol/L EDTA, 1 mmol/L EGTA, 1 mmol/L DTT, and 0.1%Tween 20) supplemented with a Protease Inhibitor Cocktail(Roche Diagnostics). Antibodies were used for Western blotanalysis. Western blot analysis and immunoprecipitation were:anti-AR (sc-13062, Santa Cruz Biotechnology), anti-FLII (sc-21716, Santa Cruz Biotechnology), anti-PSA (sc-7316, Santa CruzBiotechnology), anti-TMPRSS2 (ab92323, Abcam), anti-HA (Ssc-7392, sc-805, Santa Cruz Biotechnology), anti-Gal4 (sc-577,Santa Cruz Biotechnology), and anti-FLAG (M2 clone, Sigma).

Protein–protein interaction by GST pull-down assaysIn vitro protein–protein interactions were performed as

described previously (24, 25). In vitro translated proteins wereprepared by coupled transcription–translation with a TNT-cou-pled reticulocyte lysate kit (Promega) using plasmid DNA (1.0mg) in a total of 50 mL. GST fusion proteins were prepared fromE. coli. In vitro translated protein (15 mL) was added to 5 mg GSTfusion protein of GST as control in 225 mL binding buffer [50mmol/L Tris-HCl, 120mmol/LNaCl, 1mmol/LDTT, 0.5%NP40,1 mmol/L EDTA, 2 mg/mL leupeptin, 2 mg/mL aprotinin,1 mmol/L phenylmethylsulfonyl fluoride (PMSF), and 2 mg/mLpepstatin] and rotated for 2 hours at 4�C. Glutathione–sepharosebead slurry (50 mL) was added and the mixture was rotated for afurther 30 minutes at 4�C. Beads were washed 5 times withwashing buffer (1 mL), and binding buffer (30 mL) was addedafter the final wash. Sepharose beads were washed 5 times withlysis buffer and boiled in SDS sample buffer, and the proteinsreleased were resolved by SDS-PAGE followed by Western blotanalysis.

Luciferase assaysCells were seeded at a density of 1 � 105 cells in a 24-well cell

culture plate on the day prior to transfection with Superfectaccording to the manufacturer's protocol (Qiagen). For reportergene assays, a dose response was determined in each experimentwith 50 and 200 ng of expression vector and promoter reporterplasmids (0.5 mg). Luciferase activity was normalized for trans-fection efficiency using b-galactosidase reporter as an internalcontrol. The fold effect of expression vector was determined incomparisonwith the value of the empty expression vector cassetteand statistical analyses were performed using the t test (22, 26).

Colony formation assaysCells were plated in triplicates in 3 mL of 0.3% agarose (sea

plaque) in complete growth medium overlaid on 0.5% agarosebase, also in complete growth medium. Two weeks after cellseeding, colonies were visualized after staining with 0.04% crystalviolet inmethanol for 1 to 2 hours. The coloniesmore than 50mmin diameter were counted using anOmnicon 3600 image analysissystem.

Transwell invasion assaysBiocoat Matrigel invasion chamber inserts (BD Biosciences)

were equilibrated for 2 hours at 37�C in serum-freemedium.Cellswere seeded at a density of 2.5 � 104 per well in serum-freemedium in the top chamber of the insert and cells were allowed toinvade through the Matrigel. Medium containing 10% FBS was

added to the bottom chamber as a chemoattractant. After 22hours, cells were washed with PBS and fixed with 4% formalde-hyde/PBS for 15 minutes. After washing with PBS, cells in the topchamber were removed with a cotton swab and cells attached tothe bottom surface of thefilterwere rinsed. Cellswere stainedwithHoechst 33258 (1 mg/mL) for 30 minutes in the dark andvisualized under a fluorescent microscope. Three random fieldswere captured at 10� magnification (n ¼ 3; ref. 27).

Wound healing assaysCells were grown to confluence and wounded by dragging a 10

mL pipette tip through the monolayer. Cells were washed toremove cell debris and allowed to migrate for 12 hours. Imageswere taken at time 0 and 12 hours postwounding under a NikonDiaphot TMD inverted microscope (20�). The relative distancetraveled by the leading edge from0 to 12hours was assessed usingthe Photoshop 7.0 software (n ¼ 6; ref. 27).

In vivo tumor implantationA total of 2 � 106 cells were injected subcutaneously into 4- to

6-week-old castratedmale nudemice purchased fromBeijingHFKBio-technology Co., Ltd. Tumor growth was measured using adigital caliper every 5 days for 4 to 5 weeks. Tumor weight wasmeasured when mice were sacrificed on day 32 after cell implan-tation. Immunohistochemical staining under the standard pro-cedure was conducted as previously described (2, 22).

Statistical analysisAll statistical analyseswere performed using Excel 2010 (Micro-

soft). All in vitro experiments were performed in triplicate and alldata were represented as mean � SD. Statistical analyses wereconducted using the Student t test and Pearson correlation coef-ficient. The significance of each value was determined whenP value was less than 0.05.

ResultsEndogenous expression of FLII is positively correlated tooverall survival of prostate cancer patients

FLII, as a transcriptional coregulator, positively regulated estro-gen receptor (ER) and thyroid hormone receptor (TR; refs. 14,16, 17) and suppressed PPARg , b-catenin, and ChREBP-mediatedtranscription in cancer cells (15, 18, 19). Currently, it wasunknown whether FLII played a role in AR function. We firstperformed in silico analysis for FLII and AR expression on a geneexpression dataset composed of 150 patient samples with follow-up information (Supplementary Fig. S1A; http://www.cbioportal.org/; ref. 23). The Kaplan–Meier analysis was conducted toevaluate the difference in overall survival of patients associatedwith high versus low expression of FLII and AR gene in theirtumor. The median transcript levels of FLII and AR genes wereused to assign the patient samples to either high or low expressiongroup (28). According to Kaplan–Meier analysis, patients withprimary tumors expressing high levels of FLII (n ¼ 70) hadsignificantly higher overall survival (P ¼ 0.049; SupplementaryFig. S1B). Furthermore, in the primary tumors with high ARexpression (n ¼ 58), there was also a significant trend towardimproved overall survival of patients with high expression of FLII(n¼37) comparedwith thosewith lowFLII expression (n¼21) intumors (P ¼ 0.0076; Supplementary Fig. S1C). However, nostatistically significant difference was found between high and

Flightless I Represses Androgen Receptor

www.aacrjournals.org Clin Cancer Res; 22(6) March 15, 2016 1533




low FLII expression groups in tumors with lowAR expression (n¼74; P ¼ 0.7842; Supplementary Fig. S1D), suggesting FLII mayinteract with AR signaling in prostate cancer.

To further confirm the interaction between FLII and AR, weexamined FLII expression using tissue array containing 36 cases ofhuman prostate cancer and 8 adjacent normal tissue specimens(Biomax, PR956B; Fig. 1A). FLII showed whole epithelial celldistribution with stronger staining in adjacent normal tissue (n¼8) compared with prostate cancer tissues (n ¼ 36; Fig. 1B).Furthermore, FLII protein levels were decreased in canceroustissues with moderate expression in early stages (II) and poorexpression in advanced stages (III/IV; Fig. 1B). Kaplan–Meieranalysis was conducted to evaluate the difference between dis-ease-overall survival associated with high and low expressionlevels of FLII. The median FLII expression was used to assign thesamples to high or low group. Patients with tumors expressinghigh FLII (n¼ 18) had significantly higher disease-overall survival(P ¼ 0.045; Fig. 1C), consistent with the result from in silicoanalysis. Prostate-specific antigen (PSA), a clinical biomarker ofprostate cancer, was mainly induced by androgen and regulatedby the AR at the transcriptional level (29). We analyzed therelationship between FLII and PSA. As shown in Fig. 1D, inpatients with high PSA levels (>20 ng/mL), there was a lowerFLII expression in tumor, suggesting an inverse correlationbetween FLII and AR transcriptional activity in clinical prostatecancer specimens.

Next, we analyzed the abundance of FLII mRNA by quanti-tative real-time PCR (qRT-PCR) in prostate cancer tissues. Asshown in Fig. 1B, the mRNA level of FLII was significantlydecreased in prostate cancer tissues (n ¼ 40) compared withnormal prostate tissue (n ¼ 20). Patients with high PSA levels(10–20 or >20 ng/mL) had a lower FLII expression in tumor(n ¼ 40; Fig. 1D), consistent with the result from tissue arrayanalysis. Furthermore, FLII and AR were inversely correlatedin preclinical prostate cancer patients (r ¼ �0.5745, P ¼ 0.0159,n ¼ 1; Supplementary Fig. S2A).

AR-LBD is required for binding to FLIIThe inverse correlation between PSA and FLII raised a possi-

bility that FLII may function through interaction with androgen/AR signaling. To explore this possibility, we first examined theinteraction between FLII andAR. FLAG-FLII wild-type (wt), FLAG-FLII-LRR (residues 1–450), or FLAG-FLII-GLD (residues 451–1268) with HA-tagged AR wt were expressed in HEK 293T cells.HA-ARwas immunoprecipitated from cell lysateswith an anti-HAantibody and analyzed for FLII binding by Western blot analysis.We found that FLAG-FLII, FLAG-FLII-LRR, and FLAG-FLII-GLDcoimmunoprecipitated with HA-AR (Fig. 2A), suggesting eitherdomain is sufficient for AR binding, and further suggesting thatthe presence of both decreases binding, compared with eitheralone. To prove that FLII and AR directly interact, The FLII proteinwas purified as a GST protein from bacteria and AR was expressedby in vitro transcription/translation (IVT). GST pull-down wasconducted and the analysis was completed for FLII-LRR (residues1–494), FLII-GLD-A (residues 495–822), and FLII-GLD-B (resi-dues 825–1268) and their binding for AR. AR bound to FLII-LRRand FLII-GLD-B, but failed to bind the FLII-GLD-A (Fig. 2B).

To identify the domains of AR required for binding to FLII, aseries of Gal4-AR–mutant (residues 11–265, 262–295, 295–550,and 624–919) expression vectors were transfected into HEK 293Tcells together with vector-expressing HA-tagged FLII (HA-FLII).

The N-terminal HA epitope was used to immunoprecipitateHA-FLII. Transfection of HEK 293T cells with a series of ARmutants showed that AR bound to FLII and that the AR-LBD(residues 624–919)was sufficient for binding to FLII (Fig. 2C). Tofurther confirm that FLII only bound AR-LBD, EGFP-AR-DDBD(deletion of DNA-binding domain), and EGFP-AR-DLBD(deletion of LBD) expression vectors were expressed in HEK293T cells. The N-terminal HA epitope was used to immunopre-cipitate FLII. As shown in Fig. 2D, deletion of the AR-LBD abolishAR binding to FLII. These studies demonstrated that the LBD ofAR is required for binding FLII.

DHT treatment abolishes AR and FLII bindingWhen AR was activated by endogenous androgenic ligands,

the ligand–receptor complex translocated from the cytoplasminto the nucleus, and in association with coregulatory factors,bound to specific androgen-responsive elements in the regula-tory regions of AR target genes (30). The AR agonists andantagonists modulated AR function by binding to the AR-LBD(31). Given that AR-LBD was required for binding to FLII, wehypothesized that ligands regulate binding of FLII to AR. Toexplore this possibility, Gal4-AR wt and Gal4-AR-LBD withFLAG-FLII wt, FLAG-FLII-LRR and FLAG-FLII-GLD wereexpressed in HEK 293T cells and treated with DHT. FLAG-FLIIwas immunoprecipitated from cell lysates using an anti-FLAGantibody and analyzed for AR binding by Western blot analysis.As shown in Fig. 3A, ARwas coimmunoprecipitated and detectedin the absence of DHT, while androgen treatment decreased thebinding of FLII to AR, suggesting a potential competitionbetween FLII and DHT for binding AR. Bicalutamide (Casodex),a competitive inhibitor of AR, prevented androgen binding to ARby blocking its binding sites at the target gene promoters (32).To further confirm the potential competition between FLIIand DHT in binding to the AR, we cotransfected HA-FLII wtand Gal4-AR wt into 293T cells treated with DHT and/or bica-lutamide. HA epitope was used to immunoprecipitate FLII. Asshown in Fig. 3B, bicalutamide promoted the binding of FLIIon AR and inhibited the DHT-mediated effects.

To determine whether endogenous FLII associated with AR, weperformed IP-Western blot assay with an antibody directedtoward endogenous AR. AR antibody coprecipitated FLII inLNCaP cells in either presence or absence of DHT and/or bica-lutamide. DHT decreased binding of FLII on AR (Fig. 3C), whilebicalutamide enhanced the binding of FLII and AR and inhibitedthe DHT-mediated effects (Fig. 3C).

FLII regulates AR subcellular localizationThe competition between FLII andDHT in binding to AR raised

a possibility that FLII may suppress AR transactivation throughinterrupting AR nuclear translocation. To test the possibility, weperformed immunofluorescent staining analyses. PC3 cells werecotransfected with HA-FLII and/or FLAG-AR and treated withDHT. The subcellular localization of AR was examined by detect-ing FLAG-AR. In the absence of FLII, the unliganded AR waslocated mainly in the cytoplasm with the ligand-bound ARpredominantly in the nucleus (Supplementary Fig. S3A andS3B). In contrast, overexpression of FLII did not alter the cyto-plasmic localization of AR in the absence of DHT, while the DHT-inducednuclear localizationofARwas decreased (SupplementaryFig. S3A and S3B). Therefore, FLII antagonizes DHT function andpromotes cytoplasmic accumulation of AR.

Wang et al.





A

0

0.04

0.08

0.12

0.16

>200–200–10PSA (ng/mL)

Rel

ativ

e in

tens

ity F

LII

n = 8 n = 10 n = 12

P < 0.005

0

0.2

0.4

0.6

0.8

1.0

1.2

Surv

ival

pro

babi

lity

P = 0.045

Survival time (months)0 20 40 60 80

FLII low (n = 17)FLII high (n = 18)

B

D

C

NORMALNORMALNORMAL

II, T2cN0M0 III, T3bN0M0 IV, T4N0M0

P < 0.001

Rel

ativ

e in

tens

ity F

LII

0

0.04

0.08

0.12

0.16

0.2

QT-

PCR

of F

LII m

RN

A P < 0.01

Normal Primary n = 40n = 20

00.050.1

0.15

0.20.250.3

0.350.4

PSA (ng/mL)

P < 0.005

0

0.05

0.1

0.15

0.2

0.25

>200–200–10n = 15 n = 15 n = 10

QT-

PCR

of F

LII m

RN

A P < 0.005

Figure 1.Expression of FLII was positively correlated to overall survival of prostate cancer patients. A, representative examples of immunohistochemical stainingfor FLII in each of the clinical stages of prostate cancers as indicated. B, quantification of FLII relative intensity for each clinical stage of prostate cancers. FLIIabundance was determined by qRT-PCR between normal (n ¼ 20) and tumorous (n ¼ 40) prostate samples. Data is shown as mean � SEM for N asindicated in the figure in parenthesis as shown. C, Kaplan–Meier analysis showed a significant trend toward improved survival in cases showing high expression(n ¼ 18) of FLII as opposed to cases showing low expression (n ¼ 17) of FLII (P ¼ 0.045). D, immunohistochemical staining was also conducted todetermine FLII expression and PSA levels in the prostate cancer specimens (n ¼ 28). qRT-PCR was conducted to determine FLII expression and PSA levelsin the prostate cancer specimens (n ¼ 40).






HA-FLII

Gal4-AR

α-HA(FLII)

INPUT HA-IP293T

α-Gal4(AR)

V 11-2

6529

5-55

026

2-29

5

wt

624-

919

V 11-2

6529

5-55

026

2-29

5

wt

624-

919

+ + + + + + + + + + + +

LE

C

919

FLIIbinding

+

9191

706 870

AR

919 –∆LBD

∆DBD 919 +

624559

EGFP LBDDBD

wt

A B

HA-AR + +++ + +++α-HA(AR)

α-FLAG(FLII)

FLAG-FLII V wt

LRR

GLD

V wt

LRR

GLD

INPUT FLAG-IP293T

ARbinding

FLAG LRR GLD

1 1268

1268

450

1268451

FLII

+

+

+

LRR

GLD

wt

α-HA(FLII)

αAR

+ +++ + +++HA-FLIIEGFP-AR V w

t

∆LB

D

∆DB

D

V wt

∆LB

D

∆DB

D

INPUT HA-IP

919

265

550919

FLIIbinding

+

––

+

–

11295262

295

Gal4 LBD9191

706 870

AR

wt

559DBD

624

624

D

GST LRR GLD-A12681

494

825 1268

ARbinding

+

GLD-B

495 822GLD-A

LRR

GLD-B

+

–

FLII

VP16-AR + + + + + +

GLD

-A

GLD

-B

LRR

GLD

-A

GLD

-B

LRR

α-AR

α-GST(FLII)

INPUT GST-IP

GST-FLII

Figure 2.AR-LBD is required for binding to FLII. A, schematic representation of AR, FLII, and FLII mutation expression vectors. Immunoprecipitation (IP)–Western blotanalysis was conducted of 293T cells transfected with expression vectors encoding either N-terminal FLAG-tagged AR or HA-tagged FLII expression vectors.B, schematic representation of AR, FLII, and FLII mutation expression vectors. GST pull-down conducted with GST-FLII proteins (LRR encoding AA 1-494,GLD-A encoding 495-822, and GLD-B encoding 825-1268) and IVT-AR. C, schematic representation of FLII, AR, and AR mutation expression vectors.Immunoprecipitation–Western blot analysis was conducted of 293T cells transfected with expression vectors encoding either N-terminal HA-tagged FLII orGal4-tagged AR expression vectors. D, schematic representation of FLII and AR mutation expression vectors. Immunoprecipitation–Western blot analysis wasconducted of 293T cells transfected with expression vectors encoding either N-terminal HA-tagged FLII or EGFP-tagged AR mutation expression vectors.All the data are representative of N ¼ 3 separate experiments.

Wang et al.





α-β-Actin

++++ ++++

–– ++

INPUT

–– ++–DHT ++ –– ++ –

FLAG-IP

α-Gal4 (AR)

293T

α-FLAG(FLII)

Gal4-AR-LBDFLAG-FLII-GLD

α-Gal4 (AR)

α-FLAG(FLII)

α-β-Actin

++++ ++++

–– ++

INPUT

–– ++–DHT ++ –– ++ –

FLAG-IP293T

FLAG-FLII-LRRGal4-AR-LBD

A

FLAG-FLII ++++++ ++++++

Gal4-AR-LBD –– ++ –– –– ++ ––

Gal4-AR-wt –– ++–– –– ++––

α-β-Actin

α-Gal4(AR)

DHT – ++ – – + – ++ – – +

FLAG-IPINPUT293T

α-FLAG(FLII)

00.5

11.5

22.5

3

DHT – + +–Bica – – + +

Fold

cha

ngeP < 0.01

Gal4-AR-wtHA-FLII +++ +

+++ +

B

DHT – + + –– + –Bica – – + + – – +

Gal4-AR-wt ++++ +++++++HA-FLII +++

α-Gal4(AR)

α-HA(FLII)

HA-IPINPUT293T

++++

α-β-Actin

C

α-AR (long exposure)

DHT – + –

R-IgGAR-IPLNCaP

α-AR(short exposure)

+ – + – + – + – +Bica – – ++ – – ++ –– ++

INPUT

α-FLII 00.5

11.5

22.5

3

Fold

cha

nge

DHT – + +–

Bica – – ++

P < 0.01

Figure 3.DHT treatment abolishes AR and FLII binding. A, Gal4-AR wt and Gal4-AR-LBD with HA-FLII wt, HA-FLII-LRR, and HA-FLII-GLD were expressed in 293T cellsand treated with DHT. HA-FLII was immunoprecipitated from cell lysates by an anti-HA antibody and analyzed for AR binding by Western blotting analysis.B, HA-FLII wt and Gal4-AR wt were expressed in 293T cells treated with DHT and/or bicalutamide. HA epitope was used to immunoprecipitate FLII. Relative bindingof AR to FLII was shown as mean � SEM for three separate experiments. C, Immunoprecipitation (IP)–Western blotting in LNCaP cells treated with 10 nmol/LDHTand/or 10 mmol/L bicalutamide. Immunoprecipitation was conducted with an AR antibody to detect endogenous AR. Relative binding of AR to FLII is shownas mean � SEM for three separate experiments.






To further confirm the immunofluorescent staining results,we performed Western blot analyses using cytoplasmic andnuclear fractions of the cell lysates in PC-3 and LNCaP cells.PC3 cells were cotransfected with HA-FLII and/or FLAG-ARand treated with DHT. LNCaP cells stably transduced withlentiviral shRNAs targeting FLII. As shown in SupplementaryFig. S3C, after DHT treatment, the distribution of AR in thenucleus was decreased when FLII was transfected (lane 8)compared with vector control (lane 6). Conversely, depletionof endogenous FLII in LNCaP cells significantly increased thedistribution of AR in the nucleus and decreased the distribu-tion of AR in the cytoplasm (Supplementary Fig. S3D). Takentogether, these results demonstrated that FLII regulatedAR cytoplasm/nucleus localization and may play a role inDHT-regulated AR activation.

FLII suppresses AR expression and transactivation activityTo determine whether FLII regulated AR signaling, we con-

ducted PSA-luc and ARE4-luc luciferase reporter assays. Asshown in Fig. 4A, overexpression of FLII in LNCaP cellsreduced DHT-induced PSA-luc and ARE4-luc activity, andenhanced the effects of bicalutamide on AR. Conversely,knockdown of FLII increased DHT-induced PSA-luc andARE4-luc activity in LAPC-4 and LNCaP cells (Fig. 4B andSupplementary Fig. S4A). Analysis of FLII effects on basal ARfunction showed that FLII has significantly decreased basallevels of PSA-luc and ARE4-luc activity (Fig. 4A and B andSupplementary Fig. S4A).

The expression of endogenous PSA and TMPRSS2 gene iscontrolled by ligand-activated AR. To support the conclusionthat FLII impacts AR function, we used shRNA to knockdownendogenous FLII expression in LNCaP cells (Fig. 4C and D) anddetermined the basal and DHT-regulated PSA and TMPRSS2gene and protein levels by using Western blot and qRT-PCRanalysis. The results show increased basal and DHT-inducedPSA and TMPRSS2 gene and protein levels (Fig. 4C and D).Furthermore, we conducted Western blot analysis in LNCaPand C4-2 cells with stable knockdown of FLII to determine thebasal and DHT-regulated AR protein levels. As shown in Fig. 4Cand Supplementary Fig. S4B, depletion of FLII enhanced ARprotein levels in the presence and absence of ligand, suggestingof the idea that FLII may control AR function by regulatingstability of the AR protein. Accordingly, LNCaP cells stablyknocked down FLII were treated with cycloheximide to inhibitprotein synthesis, and AR protein turnover was analyzed overtime. Compared with the control cells, AR half-life was con-siderably increased in LNCaP cells stably knocked down FLII(Supplementary Fig. S4C). Previously, studies indicate thetargeting of AR for degradation via the ubiquitin–proteasomepathway is mediated in large part by the RING finger–contain-ing E3 ligases (33–35). We used bortezomib, a proteasomeinhibitor, to investigate mechanisms involved in the FLII-dependent AR degradation. The treatment of control LNCaPcells with bortezomib increased accumulation of AR protein(Supplementary Fig. S4D), suggesting the constitutive AR deg-radation. Similarly, the decreased in AR protein levels (causedby forced overexpression of FLII) was reversed upon treatmentwith bortezomib (Supplementary Fig. S4D), implying FLIIcauses the AR degradation in the proteasome. Taken together,these observations suggested that FLII suppresses AR expressionand transactivation activity.

NCoR and SMRTpromote the binding and suppressing effect ofFLII on AR

Previous studies showed that NCoR and SMRT were importantcorepressors of AR and repress AR transcriptional activities (36).Next, we examined the interactions between FLII, AR with NCoRor SMRT. HEK 293T cells were cotransfected with FLAG-taggedNCoR or SMRT vector together with HA-FLII and Gal4-AR. Thecells were treated with DHT for 1 hour. HA-FLII was immuno-precipitated from cell lysates using HA antibody and analyzed forFLII and AR binding by Western blot analysis. As shown inSupplementary Fig. S5A and S5B, an increased interactionbetween FLII and AR in the presence of NCoR or SMRT wasobserved. These results suggested that NCoR and SMRT interactwith FLII and AR. Notably, NCoR and SMRT promoted thebinding of FLII on AR.

Furthermore, the effect of NCoR or SMRT on FLII functionwas examined. As shown in Supplementary Fig. S5C and S5D,overexpression of NCoR or SMRT in LNCaP cells enhancedFLII-repressed PSA and ARE4 reporter activity by 50% followingDHT treatment, suggesting NCoR and SMRT promoted thesuppressing effect of FLII on AR. Taken together, AR corepres-sors (NCoR and SMRT) and AR antagonist (bicalutamide)promoted the binding of FLII on AR, while AR agonist (DHT)decreased this binding, indicating a potential competition ofFLII in binding to the AR.

FLII contributes to prostate cancer cell growth, migration, andinvasion through AR-dependent signaling, and reintroductionof FLII revert castration resistance of prostate cancer cells

AR has been reported to promote cell proliferation, migration,and invasion and plays a critical role in the development ofprostate cancer (37–39). Given that FLII interacted with AR andregulated AR activity, we reasoned that FLII might regulate thegrowth, migration, and invasion of AR-positive prostate cancercells. To test our hypothesis, we conducted MTT, colony forma-tion, transwell invasion, and wound healing assays. As shownin Fig. 5A and Supplementary Fig. S6A and S6B, depletion ofendogenous FLII in LNCaP, LAPC-4, and C4-2 cells increased thecell growth. We also observed that depletion of endogenous FLIIin LNCaP andC4-2 cells increased the cellmigration and invasion(Supplementary Figs. S7A and S8A).

To further investigate whether FLII-mediated inhibition of cellgrowth, migration, and invasion is dependent on AR, we firstknocked down FLII in LNCaP cells followed by DHT treatment.Depletion of FLII increased LNCaP cell growth potential in thepresence of DHT, while androgen deprivation abolished FLII-mediated inhibition of cellular proliferation (Fig. 5A and Sup-plementary Fig. S6C). Furthermore, overexpression of FLII withsimultaneous depletion of AR in LAPC-4 and C4-2 cells inhibitedcell growth in the presence of endogenous AR (Fig. 5B andSupplementary Fig. S6D), while AR knockdown abolished theFLII-mediated repression. Next, similar observation was made intranswell invasion and wound healing assays. Depletion of FLIIincreased LNCaP cell invasion and migration potential in thepresence of DHT, while androgen deprivation abolished FLII-mediated inhibition (Supplementary Figs. S7A and S8A). Con-versely, overexpression of FLII with simultaneous depletion ofAR in C4-2 cells inhibited cell invasion and migration in thepresence of endogenous AR, and AR knockdown abolished theFLII-mediated repression (Supplementary Figs. S7B and S8B).Taken together, these data indicate that FLII inhibits prostate

Wang et al.





A

PSA-Luc

–DHT

PSA-

LUC

act

ivity

(×10

6 )LNCaP cells

FLII++++

P < 0.01

––– ++++Bica –––––––– ++++

01234567

0

0.2

0.4

0.6

0.8

1

1.2

Fold

cha

nge

–DHTFLII

–––Bica ––––

P < 0.01

PSA-LucLNCaP cells

++++––––

0

0.2

0.4

0.6

0.8

1

1.2

Fold

cha

nge

DHTFLII

Bica

P < 0.01

PSA-LucLNCaP cells

Bica++++++++

0

0.2

0.4

0.6

0.8

1

1.2

Fold

cha

nge

DHTFLII

P < 0.01

PSA-LucLNCaP cells

α-FLII

α-AR

α-β-Actin

shFLII-1shFLII-4

DHT

–– +–

–– +– + – – +

– – – + ++

LNCaP cells

α-PSA

α-TMPRSS2

B

D

shFLII-1 + – –+

rela

tive

expr

essi

on

shFLII-4DHT

––– – + – – +

– – – + ++

P < 0.01

PSA

LNCaP cells

0

1

2

3

4

0

0.5

1

1.5

2

2.5

FLII

P < 0.01

shFLII-1shFLII-4

DHT

–– +

–

–– +

– + – – +

– – – + ++

rela

tive

expr

essi

on

LNCaP cells

0

1

2

3

4

shFLII-1 + – –+

shFLII-4DHT

––

– – + – – +

– – – + ++

rela

tive

expr

essi

onTM

PRSS

2 P < 0.01

LNCaP cells

ARE4

-LU

C a

ctiv

ity (×

105 )

–DHTFLII

++++––– ++++–––––––– ++++

P < 0.01

(ARE4)-Luc

0

1

2

3

4

5

LNCaP cells

Fold

cha

nge

–DHTFLII

–––––––Bica

(ARE4)-Luc

P < 0.01

LNCaP cells

0

0.2

0.4

0.6

0.8

1

1.2

0

0.2

0.4

0.6

0.8

1

1.2

++++DHTFLII

P < 0.01

(ARE4)-Luc

LNCaP cells

Fold

cha

nge

0

0.2

0.4

0.6

0.8

1

1.2

++++DHTFLII

P < 0.01

(ARE4)-Luc

LNCaP cells

Fold

cha

nge

––––Bica Bica ++++Bica

C

02

46

810

12

PSA-

LUC

act

ivity

(×10

6 ) PSA-Luc

cellsLAPC-4P < 0.01

shFLII-1shFLII-4

–––DHT +++ +++Bica

– – + –– + –– +– + – – + – – + –

––– ––– ++ +

Scramble –

–

––

–

+ –– – +––

+–

–––

++

––

––+

shFLII-1shFLII-4

–––DHT +++ +++Bica

– – + –– + –– +– + – – + – – + –

––– ––– ++ +

Scramble –

–

––

–

+ –– – +––

+–

–––

++

––

––+0

2

4

6

8

10

12(ARE4)-Luc

cellsLAPC-4

ARE4

-LU

C a

ctiv

ity (×

106 )

P < 0.01

Figure 4.FLII suppresses AR expression and transactivation activity. A and B, androgen-responsive luciferase reporter genes (PSA-Luc, ARE4-Luc) were assessedfor AR activity. LNCaP cells with FLII overexpression were treated with 10 nmol/LDHT and/or 10 mmol/L bicalutamide. LAPC-4 cells with FLII knockdown weretreated with 10 nmol/LDHT and/or 10 mmol/L bicalutamide. C, FLII, AR, PSA, and TMPRSS2 protein levels were determined by Western blot analysis in LNCaP cellswith stable FLII knockdown. LNCaP cells were treated with 10 nmol/LDHT. D, FLII, PSA, and TMPRSS2 mRNA levels were determined by qRT-PCR in LNCaPcells with stably FLII knockdown. LNCaP cells were treated with 10 nmol/LDHT.






cancer cell growth, migration, and invasion through AR-depen-dent signaling.

One of the hallmarks of prostate cancer treatment failure wasthe development of castration resistance to antiandrogen thera-pies. Ligand-independent AR activation contributed to the devel-opment of castration resistance (38, 40). Bicalutamide, a phar-maceutical drug commonly used as an antiandrogen therapy totreat recurrent prostate cancer, was a competitive inhibitor of AR.Bicalutamide prevented androgen binding to AR by blocking itsbinding sites at the target gene promoters (32). Enzalutamide(Xtandi), a novel AR signaling inhibitor, blocked the growth ofCRPC in cellular model systems andwas shown in a clinical studyto increase survival in patients with metastatic CRPC. Enzaluta-mide inhibited multiple steps of AR signaling: binding of andro-gens to AR, AR nuclear translocation, and association of AR withDNA (6). In this study, bicalutamide enhanced the binding of FLIIand AR and inhibited the DHT-mediated effects. FLII alsoenhanced the effects of bicalutamide on AR transactivation activ-ity. To explore whether FLII inhibited ligand-independent AR

activation and reverted castration resistance, we first conductedMTT assays in LNCaP, LAPC-4, and C4-2 cells treated withbicalutamide or enzalutamide. As shown in Fig. 5C, we observedthat knockdown of FLII alleviated inhibition of growth to bica-lutamide and enzalutamide in castration-sensitive LNCaP cellsand overexpression of FLII increased inhibition of growth tobicalutamide and enzalutamide in castration-sensitive LAPC-4cells, indicating that expression of FLII could enhance enzaluta-mide and bicalutamide sensitivity of castration-sensitive prostatecancer cells. As a result, castration-resistant cells were normallyrefractory to bicalutamide treatment (41). We observed thatoverexpression of FLII induced inhibition of growth to bicaluta-mide and enhanced inhibition of growth to enzalutamide incastration-resistant C4-2 cells (Fig. 5D), indicating that expressionof FLII could revert castration resistance of prostate cancer cells.Next, transwell invasion and wound healing assays were per-formed to further characterize these effects of FLII on cell migra-tion and invasion. As shown in Supplementary Figs. S7C and S7Dand S8C and S8D, knockdown of FLII alleviated inhibition of

C

A

P <

0.01

Days

LNCaP cells (MTT)

Fold

cha

nge

+cFBS

+DHT

0

2

4

6

8

10

12

14

16

7654321

PLKOSHFLII-1SHFLII-4PLKOSHFLII-1SHFLII-4

02468

101214161820

654321

VFLIIV/shAR

FLII/shAR

Fold

cha

nge

Days

cells (MTT)C4-2

P <

0.01

Fold

cha

nge P

< 0.

01

0

5

10

15

20

25

30

35

7654321

VFLIIV/shARFLII/shAR

Days

cells (MTT)LAPC-4

B

D

P <

0.01

P <

0.01

Fold

cha

nge

LNCaP cells (MTT)

Days

shCTL

shFLII-1

0

5

10

15

20

25

654321

+Veh+Bica+Enz+Veh+Bica+Enz

cells (MTT)LAPC-4

P <

0.01

+Veh+Bica+Enz

Vector

FLII

Fold

cha

nge

P <

0.01

Days0

5

10

15

20

25

30

654321

+Veh+Bica+Enz

+Veh+Bica

cells (MTT)C4-2

Days0

24

6

810

12

1416

18

654321

+Enz+Veh+Bica+Enz

Vector

FLII

Fold

cha

nge P <

0.01

P <

0.01

P <

0.01

Figure 5.FLII contributes to prostate cancer cell growth through AR-dependent signaling and reintroduction of FLII revert castration resistance of prostate cancercells. A, LNCaP cells with stable FLII knockdown were treated with or without 10 nmol/LDHT. LAPC-4 cells with stable overexpression of FLII and AR knockdown.Cells were analyzed for cell proliferation by MTT assay. B, C4-2 cells with stable overexpression of FLII and AR knockdown. Cells were analyzed for cellproliferation by MTT assay. C, LNCaP cells with stable knockdown of FLII followed by treatment with bicalutamide (10 mmol/L) or enzalutamide (5 mmol/L). LAPC-4cells with stable overexpression of FLII and AR knockdown followed by treatment with bicalutamide (10 mmol/L) or enzalutamide (5 mmol/L). Cells wereanalyzed for cell proliferation by MTT assay. D, C4-2 cells with stable overexpression of FLII followed by treatment with bicalutamide (10 mmol/L) or enzalutamide(5 mmol/L). Cells were analyzed for cell proliferation by MTT.


Wang et al.




migration and invasion to bicalutamide in castration-sensitiveLNCaP cells and conversely overexpression of FLII induced inhi-bition of migration and invasion to bicalutamide in castration-resistant C4-2 cells.

FLII inhibits prostate tumor growth in vivoEncouragedby these observations,we then investigated the role

of FLII in inhibiting prostate tumor growth in vivo. C4-2 cellsstably transduced with lentiviral shRNAs targeting FLII (C4-2/shFLII) were implanted subcutaneously into immunodeficientmice. Tumor size was measured every 5 days with last measuringperformed on day 32 (Fig. 6A). We observed that depletion ofendogenous FLII in C4-2 cells significantly reduced both thetumor size and the tumor weight (Fig. 6A). To determine theexpression of FLII, AR, PSA, and Ki67 in tumor tissues, weconducted qRT-PCR and IHC staining in tumor specimens. Asshown in Fig. 6B, tumors with FLII knockdown had increased PSAand Ki67 mRNA expression, except for AR. Nevertheless, IHCstaining demonstrated that tumors with FLII knockdownincreased AR, PSA, and Ki67 protein expression, suggesting thatFLII may control AR expression by regulating stability of the ARprotein, consistent with our previous studies. Taken together,these observations indicated that reduction of FLII promotedgrowth of prostate cancer cells through enhanced AR signaling.

Next, to demonstrate that FLII-mediated inhibition of cellgrowth is dependent on AR in vivo, we overexpressed FLII withsimultaneous depletion of AR in LAPC-4 cells. Cells wereimplanted subcutaneously into immunodeficient mice andmon-itored for tumor growth by measuring the tumor size and tumorweight. We observed that overexpression of FLII reduced both thetumor size and the tumor weight in the presence of endogenousAR, while AR knockdown abolished the FLII-mediated repression(Supplementary Fig. S9A), suggesting that FLII inhibits prostatecancer cell growth through AR-dependent signaling. Furthermore,to determine the expression of FLII, AR, and PSA in tumor tissues,we conducted qRT-PCR in tumors. As shown in SupplementaryFig. S9B, tumors with FLII overexpression had decreased PSAexpression, while AR knockdown abolished the FLII-mediatedrepression, suggesting that reduction of FLII promotes growth ofprostate cancer cells through enhanced AR signaling.

DiscussionAR, a ligand-activated transcription factor, played a critical role

in the development of prostate cancer (9). When activated by theendogenous androgenic ligands, testosterone (T) and dihydro-testosterone (DHT), AR was phosphorylated, and translocatedfrom cytoplasm into the nucleus to bind to specific androgen-responsive element (ARE) in the regulatory regions of AR targetgenes (30). Androgen action involves dissociation of AR fromHSP complex, homodimerization, nuclear translocation, andbinding to target genes, all these processes can be influenced byAR coregulators (42). AR coregulators, including AR coactivatorsand corepressors, positively and negatively regulate transactiva-tion of AR. Aberrant coregulator function due to a mutation oraltered expression levels may contribute to the progression ofAR-mediated diseases (43). To date,many coregulators of ARhavebeen characterized. Compared with the coactivators, the knowncorepressors are relatively fewer and less well characterized (42).NCoR and SMRT are important corepressors of AR and repress ARtransactivation (36). ARA67/PAT1 binds to ARmultiple domains

and suppresses AR transactivation by interrupting nuclear local-ization of AR (42). FBI-1 has been reported to suppress the ARtranscription activity, which is dependent on the binding to AR(44). Unfortunately the role of AR corepressors during prostatecancer progression was unclear. Herein, we provided evidenceshowing endogenous FLII expression was positively correlatedwith overall survival of prostate cancer patients. Specially, in theprimary tumors with high AR expression, there was a significanttrend toward improved survival of patients with high expressionof FLII compared with those with low FLII expression in tumors.However, no statistically significant differencewas foundbetweenhigh and low FLII expression groups in tumors with low ARexpression. Next, we demonstrated that FLII is dissociated fromAR in a ligand-dependentmanner,which is similar to that of othercorepressors, such as NCoR1 and SMRT. FLII inhibited AR trans-activation and modulated AR cytoplasm/nucleus localizationthrough direct binding to AR-LBD. Furthermore, FLII contributesto the human CSPC and CRPC cell growth, invasion, and migra-tion through AR-dependent signaling. Our data indicate that FLIIfunctions as an AR corepressor and inhibits prostate cancerprogression.

Like the othermembers of the nuclear receptor superfamily, ARhas four functional domains: N-terminal transactivation domain(NTD), the DNA-binding domain (DBD), C-terminal ligand-binding domain (LBD), and hinge region connecting DBD andLBD (45). The LBD of AR is essential for ligand-dependenttranscription. The mutation or deletion of LBD dramaticallyreduces the transcriptional activity of AR in response to androgen.In general, androgen modulates AR cytoplasmic/nuclear translo-cation and function by binding to its LBD (45). Nuclear locali-zation of androgen bound AR is a prerequisite for its transactiva-tion (42) and decrease of AR nuclear translocation leads tosuppression of AR transactivation and androgen-induced cellgrowth (42). Our study showed that LBD of AR is required forbinding FLII. FLII and androgen aremutually exclusive in bindingthe LBD of AR. In addition, FLII inhibited AR cytoplasmic/nucleartranslocation in the presence of its ligand DHT. These resultsrevealed a novel mechanism by which FLII regulates AR transac-tivation through binding to AR-LBD in prostate cancer. FLIIdecreased the binding of DHT to AR, which further inhibitedandrogen receptor cytoplasmic/nuclear localization and its tran-scriptional activity.

NCoR and SMRT are important corepressors of AR and repressAR transactivation (36). In the presence of AR agonist, NCoR andSMRT can be recruited to the PSA promoter and repress the PSAexpression (46). Corepressors of AR that trigger chromatin con-densation and/or chromatin modifications typically recruitHDACs to the AR complex (44). The well-characterized corepres-sors NCoR and SMRT could directly interact with multipleHDACs and recruit them to AR (44, 47). It has been reportedthat FLII negatively regulated b-catenin–mediated transcriptionby disrupting the synergy of FLAP1with b-catenin and p300 (19).Furthermore, the negative regulation of b-catenin, p300, andFLAP1 activity by FLII found that FLII did not bind well top300 KIX domain and did not appear to inhibit FLAP1 andp300 binding (19). They reasoned FLII may exert its negativeinfluence by squelching the activity of FLAP1 and other essentialfactors that bind to FLII (19). It was possible that FLII may recruitnegative regulators, such as HDACs, to the b-catenin/LEF1/TCFtranscription complex (19). In our studies, an increased interac-tion between FLII and AR in the presence of NCoR or SMRT was






α-FLII

α-AR

α-PSA

α-Ki67

shFLII – ++C4-2 tumors

C

B

00.20.40.60.8

11.2

rela

tive

expr

essi

onFL

II

shFLII-1 ––+

+shFLII-4 – –

P < 0.01

shFLII-1 ––+

+shFLII-4 – –

rela

tive

expr

essi

onAR

00.5

11.5

22.5

3

shFLII-1 ––+

+shFLII-4 – –

rela

tive

expr

essi

onPS

A

P < 0.01

0

1

2

3

4

5

rela

tive

expr

essi

onK

i67

shFLII-1 ––+

+shFLII-4 – –

P < 0.01

00.20.40.60.8

11.2

Fold

cha

nge

shFLII-1 ––+

+shFLII-4 – –

P < 0.01

01234

shFLII-1 ––+

+shFLII-4 – –

Fold

cha

nge P < 0.01

FLII

AR

01234

shFLII-1 ––+

+shFLII-4 – –

PSAP < 0.01

01234

Fold

cha

nge

Fold

cha

nge

shFLII-1 ––+

+shFLII-4 – –

Ki67

P < 0.01

00.20.40.60.8

11.2

Vector

shFLII-1

shFLII-3

C4-2 tumorsA

050

100150200250300

6 11 16 21 27 32

VectorshFLII-1shFLII-3

Tum

or s

ize

(mm

3 )

P <

0.0

1

Days0

20406080

100120140160

shFLII-1 ––+

+shFLII-4 – –

Tum

or w

eigh

t (m

g)

P < 0.01

Figure 6.FLII inhibits prostate tumor growth in vivo. A, C4-2 tumors with stable FLII knockdown were injected into nude mice. Tumor size was measured every5 days. The data are shown as mean � SEM for N > 6 separate tumors for each group. Images of tumors dissected from the mice. The tumor size (mm3)was plotted against days post tumor cell implantation. Tumorswereweighted after resection at the end of experiment. B,mRNA levels of FLII, AR, PSA, and ki67 fromtumors were determined by qRT-PCR. C, IHC staining detected the protein expression of FLII, AR, PSA, and Ki67 in C4-2 tumor tissues derived from mice.Data for quantified immunohistochemistry are shown as mean � SEM for N ¼ 4 tumors in each group.

Wang et al.





observed. Overexpression of NCoR or SMRT in prostate cancercells enhanced FLII-repressedPSAandARE-4 reporter activity afterDHT treatment. Thesemultiple lines of evidence suggestingNCoRand SMRT enhanced the binding and suppressing effect of FLII onAR. Further studies facilitate to the development of the interactionbetween FLII and other corepressors (HDAC/NCoR/SMRT).

Hormone-activated nuclear receptors (NR) activate transcrip-tion by recruiting multiple coactivator complexes to the promo-ters of target genes (48). Previous studies showed that FLII directlybinds ER-LBD in a hormone-dependent manner in breast cancer(14, 16, 48). FLII enhancement of ER function was observed onlywhen GRIP1 was coexpressed with FLII, regardless the expressionof additional coactivators such as CARM1 and p300 (48). Thedependency of FLII coactivator function on GRIP1 suggests thatFLII is a secondary coactivator of ER (48). In the current study, FLIIdirectly bound LBDof AR. In contrast to FLII/ER interactionwhereligand enhanced thebinding,DHTdeassociated FLII/AR complex,suggesting DHT and FLII may compete with each other for ARbinding. Our studies indicate that FLII functioned to repress ARactivity in prostate cancer.

ADT is the frontline treatment for patients with advancedprostate cancer including metastatic and castration-resistant dis-eases (5). As the majority of patients eventually develop castra-tion-resistant cancer despite of initial favorable response, identi-fication of molecules and mechanisms involved in castrationresistance is critical for developing effective treatment of thisdeadly disease (3, 4). We observed that knockdown of FLIIalleviated inhibition of growth to bicalutamide and enzalutamidein castration-sensitive LNCaP cells, and overexpression of FLIIinduced inhibition of growth to bicalutamide in castration-resis-tantC4-2 cells. This study revealed a leading functional role of FLIIon AR antagonist (bicalutamide, enzalutamide) underlying thedevelopment of castration and antiandrogen resistance. In themeanwhile, bicalutamide may cooperate with FLII to inhibit ARtransactivation activity.

The new paradigm of treating CRPC with enzalutamide andabiraterone provided opportunity to dissect the drug resistancemechanisms to these agents (49). It was now established that aproportion of patients were primary refractory to therapy ordevelops resistance despite showing an initial response (49).Castration resistance is often marked by a return of AR signaling.The extent to which the spectrum of molecular alterationsacquired after first-line therapy contributed to resistance in this

new treatment setting remains largely uncharacterized (49). Con-stitutively active AR-Vs, such as AR-V7, lacking the LBD have beenimplicated in the pathogenesis of CRPC and in mediating resis-tance to newer drugs that target the androgen axis (50). In ourstudies, FLII inhibited AR transactivation and modulated ARcytoplasm/nucleus localization through direct binding to AR-LBD. AR-LBD is required for binding to FLII, suggesting FLII genetherapy would not be effective in AR-Vs patients.

In summary, we identified a biochemical and functional linkbetween FLII and reduced expression in advanced prostate cancerand AR signaling in prostate cancer Furthermore, loss of FLIIexpression contributes to prostate cancer growth and survival andreintroduction of FLII sensitizes castration-resistant prostate can-cer cells to antiandrogen therapy. This study provides, for the firsttime, compelling evidence, which rationalizes the targeting strat-egy to reestablish the functional interaction between FLII and ARsignaling in clinical practice to treat particular prostate cancerpatients.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: T. Wang, L. Wang, C. Wang, K. ChenDevelopment of methodology: T. Wang, W. Song, Y. Chen, L. Wang, K. ChenAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): T. Wang, W. Song, Y. Chen, Z. Liu, L. Wu, M. Li,J. Yang, L. Wang, K. ChenAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis):T.Wang,W. Song, Y.Chen, L.Wang,C.Wang,K.ChenWriting, review, and/or revisionof themanuscript: T.Wang, R. Chen, L.Wang,J. Liu, Z. Ye, K. ChenAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): T. Wang, L. Wang, C. Wang, K. ChenStudy supervision: T. Wang, K. Chen

Grant SupportThis work was supported by grant from National Natural Sciences Founda-

tion of China (No.81472405, No. 81001133).The costs of publication of this articlewere defrayed inpart by the payment of

page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received July 9, 2015; revised October 14, 2015; accepted October 25, 2015;published OnlineFirst November 2, 2015.

References1. Alberti C. Prostate cancer progression and surroundingmicroenvironment.

Int J Biol Markers 2006;21:88–95.2. Wang T, Liu Z,Guo S,WuL, LiM, Yang J, et al. The tumor suppressive role of

CAMK2N1 in castration-resistant prostate cancer. Oncotarget 2014;5:3611–21.

3. Sung YY, Cheung E. Androgen receptor co-regulatory networks in castra-tion-resistant prostate cancer. Endocr Relat Cancer 2013;21:R1–R11.

4. Heinlein CA, Chang C. Androgen receptor in prostate cancer. Endocr Rev2004;25:276–308.

5. Mazaris E, Tsiotras A. Molecular pathways in prostate cancer. NephrourolMon 2013;5:792–800.

6. Saad F, Fizazi KS. Androgen deprivation therapy and secondary hormonetherapy in the management of hormone-sensitive and castration resistantprostate cancer. Urology. 2015 Aug 14. [Epub ahead of print].

7. Greasley R, KhabazhaitajerM, RosarioDJ. A profile of enzalutamide for thetreatment of advanced castration resistant prostate cancer. Cancer ManagRes 2015;12:153–64.

8. Modena A, Ciccarese C, Fantinel E, Bimbatti D, Tortora G, Massari F.Metastatic castration-resistant prostate cancer: targeting themechanisms ofresistance to abiraterone acetate and enzalutamide. Expert Rev AnticancerTher 2015;15:1037–48.

9. Xu S,Wang T, SongW, Jiang T, Zhang F, Yin Y, et al. The inhibitory effects ofAR/miR-190a/YB-1 negative feedback loop on prostate cancer and under-lying mechanism. Sci Rep 2015;5:13528.

10. Chmelar R, Buchanan G, Need EF, Tilley W, Greenberg NM. Andro-gen receptor coregulators and their involvement in the deve-lopment and progression of prostate cancer. Int J Cancer 2007;120:719–33.

11. Vasaitis TS,Njar VC.Novel, potent anti-androgens of therapeutic potential:recent advances and promising developments. Future Med Chem2010;2:667–80.

12. Maughan BL, Antonarakis ES. Androgen pathway resistance in prostatecancer and therapeutic implications. Expert Opin Pharmacother 2015;16:1521–37.






13. Li J, Yin HL, Yuan J. Flightless-I regulates proinflammatory caspases byselectively modulating intracellular localization and caspase activity. J CellBiol 2008;181:321–33.

14. Archer SK, Behm CA, Claudianos C, Campbell HD. The flightless I proteinand the gelsolin family in nuclear hormone receptor-mediated signalling.Biochem Soc Trans 2004;32:940–2.

15. Wu L, Chen H, Zhu Y, Meng J, Li Y, Li M, et al. Flightless I homolognegatively regulates ChREBP activity in cancer cells. Int J BiochemCell Biol2013;45:2688–97.

16. Jeong KW, Lee YH, Stallcup MR. Recruitment of the SWI/SNF chromatinremodeling complex to steroid hormone-regulated promoters by nuclearreceptor coactivator flightless-I. J Biol Chem 2009;284:29298–309.

17. Jeong KW. Flightless I (Drosophila) homolog facilitates chromatin acces-sibility of the estrogen receptor a target genes in MCF-7 breast cancer cells.Biochem Biophys Res Commun 2014;446:608–13.

18. Choi JS, Choi SS, Kim ES, Seo YK, Seo JK, Kim EK, et al. Flightless-1, a noveltranscriptional modulator of PPARg through competing with RXRa.Cell Signal 2015;27:614–20

19. Lee YH, Stallcup MR. Interplay of Fli-I and FLAP1 for regulation of beta-catenin dependent transcription. Nucleic Acids Res 2006;34:5052–9.

20. Chen S, Gulla S, Cai C, Balk SP. Androgen receptor serine 81 phosphor-ylation mediates chromatin binding and transcriptional activation. J BiolChem 2012;287:8571–83.

21. Wang L, Lin HK, Hu YC, Xie S, Yang L, Chang C. Suppression of androgenreceptor-mediated transactivation and cell growth by the glycogensynthase kinase 3 beta in prostate cells. J Biol Chem 2004;279:32444–52.

22. Wang T, Guo S, Liu Z,Wu L, Li M, Yang J, et al. CAMK2N1 inhibits prostatecancer progression through androgen receptor-dependent signaling.Oncotarget 2014;5:10293–306.

23. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, et al.Integrative genomic profiling of human prostate cancer. Cancer Cell2010;18:11–22.

24. Chen K,Wu K, Cai S, ZhangW, Zhou J, Wang J, et al. Dachshund binds p53to block the growth of lung adenocarcinoma cells. Cancer Res 2013;73:3262–74.

25. WuK, ChenK,WangC, Jiao X,Wang L, Zhou J, et al. Cell fate factorDACH1represses YB-1-mediated oncogenic transcription and translation.Cancer Res 2014;74:829–39.

26. Chen K, Wu K, Gormley M, Ertel A, Wang J, Zhang W, et al. Acetylation ofthe cell-fate factor dachshund determines p53 binding and signalingmodules in breast cancer. Oncotarget 2013;4:923–35.

27. ConnorKM,HempelN,NelsonKK,DabiriG,GamarraA, Belarmino J, et al.Manganese superoxide dismutase enhances the invasive and migratoryactivity of tumor cells. Cancer Res 2007;67:10260–7.

28. Nishimura M, Jung EJ, Shah MY, Lu C, Spizzo R, Shimizu M, et al.Therapeutic synergy between microRNA and siRNA in ovarian cancertreatment. Cancer Discov 2013;3:1302–15

29. Horii K, Suzuki Y, Kondo Y, Akimoto M, Nishimura T, Yamabe Y, et al.Androgen-dependent gene expression of prostate-specific antigen isenhanced synergistically by hypoxia in human prostate cancer cells. MolCancer Res 2007;5:383–91.

30. Koochekpour S. Androgen receptor signaling and mutations in prostatecancer. Asian J Androl 2010;12:639–57.

31. Li H, Ren X, Leblanc E, Frewin K, Rennie PS, Cherkasov A. Identification ofnovel androgen receptor antagonists using structure- and ligand-basedmethods. J Chem Inf Model 2013;53:123–30.

32. Masiello D, Cheng S, Bubley GJ, LuML, Balk SP. Bicalutamide functions asan androgen receptor antagonist by assembly of a transcriptionally inactivereceptor. J Biol Chem 2002;277:26321–6.

33. Lakshmikanthan V, Zou L, Kim JI, Michal A, Nie Z, Messias NC, et al.Identification of betaArrestin2 as a corepressor of androgen receptorsignaling in prostate cancer. Proc Natl Acad Sci U S A 2009;106:9379–84.

34. Kang HY, Yeh S, Fujimoto N, Chang C. Cloning and characterization ofhuman prostate coactivator ARA54, a novel protein that associates with theandrogen receptor. J Biol Chem 1999;274:8570–6.

35. PoukkaH, KarvonenU, YoshikawaN, TanakaH, Palvimo JJ, J€anneOA. TheRING finger protein SNURFmodulates nuclear trafficking of the androgenreceptor. J Cell Sci 2000;113:2991–3001.

36. Hodgson MC, Astapova I, Hollenberg AN, Balk SP. Activity of androgenreceptor antagonist bicalutamide in prostate cancer cells is independent ofNCoR and SMRT corepressors. Cancer Res 2007;67:8388–95.

37. DevlinHL,MudryjM. Progression of prostate cancer:multiple pathways toandrogen independence. Cancer Lett 2009;274:177–86.

38. Stan SD, Singh SV. Transcriptional repression and inhibition of nucleartranslocation of androgen receptor by diallyl trisulfide in human prostatecancer cells. Clin Cancer Res 2009;15:4895–4903.

39. Qu F, Cui X, Hong Y, Wang J, Li Y, Chen L, et al. MicroRNA-185 suppressesproliferation, invasion, migration, and tumorigenicity of human prostatecancer cells through targeting androgen receptor. Mol Cell Biochem2013;377:121–30.

40. Nacusi LP, Tindall DJ. Androgen receptor abnormalities in castration-recurrent prostate cancer. Expert Rev Endocrinol Metab 2009;4:417–22.

41. Dehm SM, Tindall DJ. Ligand-independent androgen receptor activity isactivation function-2-independent and resistant to antiandrogens inandrogen refractory prostate cancer cells. J Biol Chem 2006;281:27882–93.

42. Zhang Y, Yang Y, Yeh S, Chang C. ARA67/PAT1 functions as a repressorto suppress androgen receptor transactivation. Mol Cell Biol 2004;24:1044–57.

43. Felgueiras J, Silva JV, Fardilha M. Prostate cancer: the need for biomarkersand new therapeutic targets. J Zhejiang Univ Sci B 2014;15:16–42.

44. Cui J, Yang Y, Zhang C, Hu P, KanW, Bai X, et al. FBI-1 functions as a novelAR co-repressor in prostate cancer cells. Cell Mol Life Sci 2011;68:1091–103.

45. KangHY, TsaiMY, Chang C, Huang KE.Mechanisms and clinical relevanceof androgens and androgen receptor actions. Chang Gung Med J 2003;26:388–402.

46. Trtkova K, Paskova L,Matijescukova N, StrnadM, Kolar Z. Binding of AR toSMRT/N-CoR complex and its co-operation with PSA promoter in prostatecancer cells treated with natural histone deacetylase inhibitor NaB. Neo-plasma 2010;57:406–14.

47. Guenther MG, Barak O, Lazar MA. The SMRT and N-CoR corepressors areactivating cofactors for histone deacetylase 3. Mol Cell Biol 2001;21:6091–101.

48. Lee YH, Campbell HD, Stallcup MR. Developmentally essential proteinflightless I is a nuclear receptor coactivator with actin binding activity. MolCell Biol 2004;24:2103–17.

49. Nakazawa M, Antonarakis ES, Luo J. Androgen receptor splice variants inthe era of enzalutamide and abiraterone. Horm Cancer 2014;5:265–73.

50. Xu D, Zhan Y, Qi Y, Cao B, Bai S, Xu W, et al. Androgen receptor splicevariants dimerize to transactivate target genes. Cancer Res 2015;75:3663–71.


Wang et al.




2016;22:1531-1544. Published OnlineFirst November 2, 2015.Clin Cancer Res Tao Wang, Wen Song, Yuan Chen, et al. through Targeting Androgen Receptor SignalingFlightless I Homolog Represses Prostate Cancer Progression

Updated version

10.1158/1078-0432.CCR-15-1632doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2015/10/31/1078-0432.CCR-15-1632.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/22/6/1531.full#ref-list-1

This article cites 49 articles, 21 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/22/6/1531.full#related-urls

This article has been cited by 4 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/22/6/1531To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-15-1632

http://clincancerres.aacrjournals.org/content/suppl/2015/10/31/1078-0432.CCR-15-1632.DC1

http://clincancerres.aacrjournals.org/content/22/6/1531.full#ref-list-1

http://clincancerres.aacrjournals.org/content/22/6/1531.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/22/6/1531


flightless i homolog represses prostate cancer progression ...chenguang wang3, and ke chen1,4...

Documents