The expression of AURKA is androgen regulated in castration-resistant prostate cancer

1 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 www.na u e.com/scien i ic epo s The exp ession o AURKA is and ogen egula ed in cas a ion- esis an p os a e cance Ka i Ki inummi 1, Al onso U banucci1,4,5,6,7, Ka i Leinonen1, Teu o L. J. Tammela2, Ma i Annala1, William B. Isaacs3, G. S e en Bo a1, Ma i Nyk e 1 & Tapio Visako pi1 Al hough second gene a ion endoc ine he apies ha e signi ican ly imp o ed su i al, cas a ion- esis an p os a e cance (CRPC) cells a e e en ually able o escape a ailable ho monal ea men s due o eac i a ion o and ogen ecep o (AR) signaling. Iden i ica ion o no el, non-classical and d uggable AR- a ge genes may p o ide new app oaches o ea CRPC. Ou p e ious analyses sugges ed ha Au o a kinase A (AURKA) is egula ed by and ogens in p os a e cance cells ha exp ess high le els o AR. He e, we p o ide u he e idence ha AURKA is signi ican ly o e exp essed in AR-posi i e CRPC samples ca ying ampli ica ion o AR gene and/o exp essing AR in high le els. We also demons a e and ogen-induced AR binding in he in onic egion o AURKA. The exp ession o AURKA is inc eased upon and ogen s imula ion in LNCaP-ARhi cells ha exp ess high le els o AR. The g ow h o he cells was also signi ican ly inhibi ed by an AURKA speci ic inhibi o , alise ib (MLN8237). Toge he , hese indings sugges ha he exp ession o AURKA is egula ed by and ogen in p os a e cance cells ha highly exp ess AR, emphasizing i s po en ial as a he apeu ic a ge in pa ien s wi h CRPC. And ogens and and ogen ecep o (AR) a e known o be impo an d i e s o p os a e cance p og ession1,2. AR is a ligand-dependen ansc ip ion ac o ha is a membe o he s e oid ho mone ecep o amily, and i binds o and ogen esponsi e elemen s in DNA o egula e and ogen esponsi e genes3. The cu en s anda d ea men o ad anced p os a e cance is and ogen-dep i a ion he apy (ADT)1,3. E en ually, mos ADT- ea ed pa ien s de elop cas a ion- esis an p os a e cance (CRPC), which exp esses unc ional AR. Mos CRPC pa ien s espond o second-line ho monal he apy, such as enzalu amide o abi a e one, o a limi ed pe iod1,3–5. Almos all p os a e cance s a e AR-posi i e2. The incidence o de no o AR-nega i e small cell ca cinomas wi h signs o neu oendoc ine di e en ia ion has been epo ed o be be ween 0.5 o 2% o all p os a e cance s6,7. In addi ion, some AR-posi i e p os a e cance s ansdi e en ia e du ing ADT o an AR-nega i e, neu oendoc ine ype o p os a e cance (NEPC). I has been sugges ed ha 10 o 25% o ad anced AR-posi i e p os a e cance s will become AR-nega i e NEPC2,7. We ha e ecen ly demons a ed ha he equency o AR-nega i e cance s is 1.5% in locally ecu en CRPC and 7% in CRPC me as ases8. AR-posi i e p os a e cance s and AR-nega i e NEPCs seem o ha e, a leas in some cases, he same clonal o igin because hey sha e he same molecula al e a- ions, such as ERG ea angemen s6,7,9,10. Au o a kinase A (AURKA) is a se ine- h eonine kinase ha unc ions in mi o ic spindle o ma ion and ch o- mosome seg ega ion11–16. I has oncogenic p ope ies when abe an ly exp essed, inducing aneuploidy and cell ans o ma ion11,13,15,16. Du ing mi osis, AURKA localizes o he cen osomes and mi o ic spindle poles, and i associa es wi h o he co-ac i a o s ha de e mine i s exac unc ion14,16. AURKA has been shown o be highly exp essed, especially in AR-nega i e NEPC6 and in basal cell-like b eas cance s17. 1BioMediTech Ins i u e, Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Fimlab Labo a o ies, Tampe e Uni e si y Hospi al, Tampe e, Finland. 2Depa men o U ology, Tampe e Uni e si y Hospi al and Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland. 3The James Buchanan B ady U ological Ins i u e, Johns Hopkins Uni e si y School o Medicine, Bal imo e, Ma yland, 21287, USA. 4Cen e o Molecula Medicine No way, No dic Eu opean Molecula Biology Labo a o y Pa ne ship, Fo ksningspa ken, Uni e si y o Oslo, Oslo, No way. 5Depa men o Molecula Oncology, Ins i u e o Cance Resea ch, Oslo Uni e si y Hospi al, Oslo, No way. 6Depa men o Tumo Biology, Ins i u e o Cance Resea ch, Oslo Uni e si y Hospi al, Oslo, No way. 7K.G. Jebsen In lamma ion Resea ch Cen e, Uni e si y o Oslo, Oslo, No way. Ka i Ki inummi and Al onso U banucci con ibu ed equally o his wo k. Co espondence and eques s o ma e ials should be add essed o K.K. (email: ka i. [email p o ec ed]) Recei ed: 12 June 2017 Accep ed: 6 Decembe 2017 Published: xx xx xxxx OPEN www.na u e.com/scien i ic epo s/ 2 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 We ha e p e iously ound ha AURKA is up egula ed in p os a e cance cells ha o e exp ess AR (VCaP and LNCaP cells s able- ans ec ed wi h AR) unde DHT s imula ion and is o e exp essed in CRPC18,19. Addi ionally, ch oma in immunop ecipi a ion sequencing (ChIP-seq) s udies by us and o he s ha e indica ed ha LNCaP and VCaP cells ha e a pu a i e and ogen ecep o binding si e (ARBS) in he p omo e and in onic egion o AURKA20–22. A ecen s udy by Pome an z e al.23 sugges ed ha he in onic ARBS o AURKA is a p os a e cance -speci ic AR binding e en . He e, we alida ed AURKA as a no el, non-classical and ogen esponsi e gene in CRPC cells highly o e exp essing AR. We also s udied he exp ession o AURKA in clinical p os a e cance samples and i s associa ion wi h he AR exp ession le els. Finally, we s udied he e ec o AURKA speci ic inhibi- ion in CRPC cells highly exp essing AR. Resul s To ind no el, p os a e cance speci ic clinically ele an AR- a ge genes, we in eg a ed ea lie published da a based on gene exp ession18 in clinical p os a e cance specimens wi h AR-ChIP-seq da a18–22,24 (Supplemen a y Figu eS1). In ou ea lie wo k18,we iden i ied a se o 54 genes which we p io i ized based on associa ion o hei exp ession wi h disease ou come and p esence o p os a e cance speci ic AR binding si es (ARBSs) (Supplemen a y TableS1). Gene-wise Kaplan-Meie e-analysis o he Taylo e al.24 da ase showed associa ion o posi i e exp ession wi h ou come o en genes wi h AURKA showing bes associa ion (Supplemen a y TableS1). AURKA was ound o be o e exp essed in p os a e cance , especially in CRPC specimens (Supplemen a y Figu eS2). We ha e also p e iously shown ha he exp ession o AURKA is inc eased wi h and ogen s imula ion in LNCaP-ARhi cells exp essing high le els o AR18 (Supplemen a y Figu eS2), and ChIP-seq analyses ha e indica ed a pu a i e p os a e cance speci ic ARBS in he p omo e as well as in he in onic egion o he gene20–23 (Supplemen a y TableS1, Supplemen a y Figu eS3). I has been p e iously shown ha AURKA is o e exp essed, especially in AR-nega i e NEPCs6,7,25. In e es ingly, howe e , we ound ha i s exp ession migh be egula ed by and ogens in AR-o e exp essing p os- a e cance cells18 (Supplemen a y Figu eS2). The e o e, we ini ially in es iga ed whe he AURKA is di ec ly egula ed by and ogen in high AR-exp essing (LNCaP-ARhi, desc ibed in e .18) cells unde dihyd oxy es os- e one (DHT) s imula ion using cycloheximide (CHX) o s op ansla ion, he eby inhibi ing all downs eam ansc ip ion e en s a e s imula ion. We g ew bo h LNCaP-ARhi cells and he emp y ec o ans ec ed con ol LNCaP (LNCaP-pcDNA3.1) cells wi h and wi hou 10 nM DHT in he p esence and absence o 10 µM CHX o 8 and 12 h. As expec ed, he exp ession o a well-known AR a ge gene KLK3 (alias PSA) simila ly inc eased in he p esence and absence o CHX in bo h LNCaP-ARhi and emp y ec o ans ec ed LNCaP-pcDNA3.1 cells (Fig.1a,b). Whe eas AURKA exp ession was signi ican ly inc eased in he same condi ions in LNCaP-ARhi cells, Figu e 1. AR o e exp ession sensi izes AURKA unde and ogen egula ion. LNCaP-cells we e s ably ans ec ed o o e exp ess 8- o 10- old highe w -AR (LNCaP-ARhi) compa ed o ha in emp y ec o LNCaP- pcDNA3.1 cells. Ba plo s show he ela i e exp ession quan i ied wi h qRT-PCR and no malized o TBP. Rela i e KLK3 exp ession in (a) LNCaP-pcDNA3.1 and (b) LNCaP-ARhi cells and ela i e AURKA exp ession in (c) LNCaP-pcDNA3.1 and (d) LNCaP-ARhi a e 8 and 12 h o ea men wi h DHT (10 ng) wi h and wi hou CHX (10 µg). *P < 0.05 and **P < 0.01, unpai ed es . www.na u e.com/scien i ic epo s/ 3 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 while i was no inc eased in LNCaP-pcDNA3.1 cells in he p esence and absence o CHX (Fig.1c,d). Since AURKA exp ession was inc eased despi e he p esence o CHX in LNCaP-ARhi alone, he da a clea ly sugges ha o e exp ession o AR sensi izes AURKA o become di ec ly and ogen- egula ed. To u he alida e he di ec egula ion o AURKA by and ogens, we pe o med adi ional ChIP-qPCR o ARBSs seen in he AURKA p omo e and in onic egion ha ha e been ecapi ula ed in se e al AR ChIP-seq da ase s published by us and o he s20–23 (Supplemen a y Figu eS3). Fi s , we alida ed AR binding o he well-known enhance egion o KLK3 in bo h a con ol LNCaP-pcDNA3.1 cell line and in wo independen AR-o e exp essing LNCaP-AR cell lines ea ed o 2 h wi h 1 nM and 100 nM DHT. We ound signi ican en ich- men o AR in all h ee cell lines wi h bo h DHT concen a ions, as expec ed (Fig.2a). Nex , we measu ed AR binding on he AURKA p omo e and in onic egions. The AR-binding on p omo e egion o AURKA was no signi ican ly induced by and ogens in any o he h ee cell lines (Fig.2b). Howe e , AR binding in he in onic egion was signi ican ly inc eased in AR-o e exp essing cells unde bo h 1 and 100 nM DHT s imula ion com- pa ed o he emp y ec o ans ec ed LNCaP-pcDNA3.1 cells(Fig. 2c). These esul s a e conco dan wi h he and ogen-induced AURKA ansc ip ion in he p esence o CHX in LNCaP-ARhi cells (Fig.1d) and suppo he di ec egula ion o AURKA in cells ha highly exp ess AR. To con i m he ac i i y o he AR binding, we pe o med a ansien luci e ase epo e assay in LNCaP-ARhi cells unde di e en DHT concen a ions. We ans ec ed he cells wi h AURKA p omo e alone and he in onic egions in bo h he + and – o ien a ions in he same cons uc . Emp y luci e ase ec o and KLK3 enhance luci - e ase ec o cons uc s we e used as nega i e and posi i e con ols, espec i ely (Fig.3a). No and ogen s imula ed luci e ase ac i i y was ound in he cells ans ec ed wi h AURKA p omo e cons uc alone (Fig.3a), whe eas he luci e ase ac i i y o he pGL3-AURKAp omo e -enhance (+)-LUC and PSA ec o s we e signi ican ly inc eased Figu e 2. And ogen ecep o (AR) binds o AURKA. (a) AR binding o he KLK3-enhance (n = 1), (b) AURKA-p omo e and (c) in onic egion unde DHT s imula ion (n = 2). T adi ional AR-ChIP-qPCR was pe o med a e ho mone s a a ion o 4 days ollowed by ea men wi h 1 and 100 nM DHT s imula ion o LNCaP-pcDNA3.1 (con ol) and wo independen AR-o e exp essing cell lines (LNCaP-ARmo and LNCaP- ARhi). Ba s show en ichmen o AR binding exp essed as he % o inpu . *P < 0.0232 and **P < 0.0107, unpai ed es . Figu e 3. Ac i i y o ARBS in p omo e and in onic egions o AURKA. A luci e ase assay was pe o med in LNCaP-ARhi cells wi h di e en DHT concen a ions (0, 10 o 100 nM). Cells we e ans ec ed wi h pGL3-basic-LUC (LUC), pGL3-PSA5.8-LUC (PSA), pGL3-AURKAp omo e -LUC (AURKA) and pGL3- AURKAenhance -LUC (AURKA enh.), and luci e ase ac i i ies we e no malized o he enilla con ol plasmid (LUC). (a) Ac i i y o AR binding o he AURKA p omo e and posi i e (PSA) as well as nega i e con ol egions wi h di e en and ogen s imula ion. (b) Ac i i y o AR binding o he AURKA enhance and p omo e wi h enhance egion and posi i e (PSA) and nega i e con ol (pGL3) egions wi h di e en and ogen s imula ion. *P < 0.05; n.s. = no signi ican . The means ± s.d. a e shown. www.na u e.com/scien i ic epo s/ 4 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 when DHT was added (Fig.3b). No change was ound in he ac i i y o pGL3-AURKAp omo e -enhance (−)-LUC ec o . The da a sugges ha he in onic ARBS o AURKA is needed o induce and ogen egula ion in cells exp essing high le els o AR. Nex , we alida ed ou p e ious RNA-seq da a19 and s udied AURKA exp ession in eshly ozen clinical p os a e cance coho wi h adi ional qRT-PCR. P e iously, AURKA exp ession has been shown o inc ease mo e han ou - old, especially in he neu oendoc ine ype o PC (NEPC) ully nega i e o AR6,26, which was also obse ed in ou da a (Supplemen a y Figu esS4 and S5). Howe e , ou s udy aim was o ocus on he non-classical and ogen egula ion o AURKA in a high AR-posi i e cell con ex (Supplemen a y Figu eS6). The e o e, all AR-nega i e NEPC ype cance specimens we e excluded om ou analysis (Supplemen a y Figu esS4 and S5). The AURKA exp ession was signi ican ly highe in bo h ho mone naï e PC and CRPC (P = 0.038 and P = 0.002, espec i ely, Fig.4a). We also obse ed a posi i e co ela ion be ween AR and AURKA exp ession in ou AR-posi i e RNA-seq da a ( = 0.751, p < 0.0001, Fig.4b) as well as in he Taylo e al.24 mic oa ay da a se ( = 0.576, p < 0.0001, Supplemen a y Figu eS7a) and 17 LuCaP-xenog a s ( = 0.523, P = 0.03, Supplemen a y Figu eS7b) a e excluding AR-nega i e NEPC cases. In a bigge coho o p os a e cance specimens24, he uppe qua ile o AURKA exp ession in 127 ho mone-naï e, p os a ec omy- ea ed cases was also signi ican ly associ- a ed wi h biochemical ecu ence (P = 3.09e-5, Supplemen a y Figu eS7c). Subsequen ly, we s udied AURKA p o ein exp ession in clinical samples wi h immunohis ochemis y (IHC) in FFPE issue samples ep esen ing 106 ho mone naï e p os a ec omies, 126 locally ecu en CRPC samples and 104 CRPC me as ases om 31 pa ien s who died o p os a e cance . The s aining was classi ied as posi i e when he s aining in ensi y was 1 o highe in mo e han 1% o he cells in he pa icula specimen and nega i e when he s aining in ensi y was ze o in all o he cells o i only less han 1% o he cells we e posi i e (Fig.4c). All posi i e samples showed he e ogeneous s aining o AURKA, and only 5% o all samples had mo e han 50% AURKA posi i e cells. In local ecu en CRPC specimens and CRPC me as ases, he equency o AURKA pos- i i i y was signi ican ly highe compa ed o ho mone naï e p os a e cance samples (P < 0.0001 and P < 0.0001, espec i ely, Table1, Fig.4d). Posi i e AURKA p o ein exp ession was ound in 12% (13/106) o ho mone-naï e p os a ec omy cance s, 47% (59/126) o local CRPC samples and 22% (23/104) o CRPC me as ases (Table1, Fig.4d). Because he numbe o dis inc me as ases a ied signi ican ly pa ien -wise, we also analyzed how many pa ien s had a leas one AURKA posi i e me as asis. In ou coho , 23/31 pa ien s had a leas h ee me as ases. Al hough he numbe o AURKA posi i e me as ases was lowe han he locally ecu en CRPC samples (22% and 47%, espec i ely), he numbe o pa ien s who had a leas one AURKA posi i e me as asis was simila , as seen in locally ecu en CRPC specimens o single pa ien s. The e o e, o all CRPC pa ien s wi h h ee o mo e Figu e 4. AURKA exp ession and co ela ion in clinical p os a e cance specimens. (a) AURKA exp ession (qRT-PCR) in an independen p os a e cance coho o 8 BPH, 27 ho mone-naï e p os a ec omy p os a e cance pa ien s and 16 CRPC specimens. *P = 0.038 and **P = 0.002, unpai ed es . (b) AURKA co ela ion wi h AR in 39 AR-posi i e p os a e cance specimens19. (Pea son) = 0.751, p < 0.0001. (c) Rep esen a i e AURKA IHC s ains showing nega i e (uppe panel) and posi i e s ains wi h > 1% (middle) and ~80% (bo om) o p os a e cance cells. (d) P opo ions o AURKA posi i e samples in ho mone un ea ed p os a e cance , local CRPC, me -CRPC specimens and CRPC pa ien s. (e) Kaplan-Meie analysis o biochemical ecu ence (BCR) in ho mone naï e p os a e cance pa ien s. AURKA posi i e (wi h > 1% o he umo cells, ed line) pa ien s had a signi ican ly wo se BCR a e. P- alue = 0.017, Man el-Cox es . www.na u e.com/scien i ic epo s/ 5 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 me as a ic specimens (12/23), 52% had a leas one AURKA posi i e me as asis. All pa ien s wi h h ee o mo e sepa a e me as ases in di e en o gans had AURKA nega i e me as ases. In he ho mone-naï e p os a ec omy- ea ed pa ien s, AURKA exp ession was ound o be signi ican ly asso- cia ed (P = 0.017, Man el-Cox es ) wi h poo p og ession ee-su i al (Fig.4e). The AURKA p o ein exp es- sion le els we e no associa ed wi h any common clinicopa hological a iables, such as he age a diagnosis, p ima y PSA le els, pT-s age o Gleason sco e (P = 0.126, P = 0.948, P = 0.777 and P = 0.204, espec i ely; χ2- es and unpai ed - es ; Table1). AURKA posi i i y was signi ican ly associa ed wi h MKI67 (Ki-67) s ain- ing (P = 0.030), bu i was no associa ed wi h ERG, EZH2, PTEN o SPINK exp ession o wi h he TP53-copy numbe (P = 0.7324, P = 0.934, P = 0.508, P = 0.349 and P = 1.000, espec i ely; Table1). By con as , CRPC me as ases had highe a ia ion in he AR exp ession le el, and he co ela ion o AURKA posi i i y was signi i- can ly associa ed wi h AR exp ession le els (P = 0.005, χ2- es ) as well as wi h ERG posi i i y (P = 0.004, χ2- es ; Table2). AURKA posi i i y was also associa ed wi h he Ki-67 and EZH2 le els (P = 0.025 and 0.035, espec- i ely; Mann-Whi ney U es ; Table2). Simila o ho mone naï e p os a e cance specimens, no associa ion was ound wi h PTEN, TP53 o SPINK exp ession (P = 0.420, P = 0.096 and P = 0.474, espec i ely; Fishe ’s exac es ; Table2). In locally ecu en CRPC samples, AURKA p o ein exp ession was signi ican ly associa ed wi h EZH2 exp ession (P = 0.032, Mann-Whi ney U es , Table2). Finally, we wan ed o es he e ec o AURKA inhibi ion in high-AR exp essing p os a e cells (Fig.5) unde di e en concen a ions o Au o a kinase inhibi o , alise ib (MLN8237), which is 200- old mo e speci ic agains AURKA han AURKB. Fi s , o iden i y he op imal concen a ion o inhibi g ow h, we es ed inc eas- ing concen a ions (0–1000 nM) o alise ib and measu ed he g ow h o high AURKA exp essing PC-3 cells15 (Supplemen a y Figu eS8). The g ow h inhibi ion was 100% in PC-3 cells wi h a 100 nM o highe concen a- ion o MLN8237. Thus, we ea ed LNCaP-pcDNA3.1 as well as LNCaP-ARmo and LNCaP-ARhi cells wi h 10 and 100 nM concen a ions o alise ib o cause nea maximal g ow h inhibi ion wi hou o e ea men . All cell lines esponded o 100 nM alise ib (Fig.5) in a simila manne han AR-nega i e PC-3 cell line (Supplemen a y Va iable AURKA exp ession PNega i e Posi i e P os a ec omy specimens, n (%) 93 (88) 13 (12) Locally ecu en CRPCs, n (%) 67 (53) 59 (47) Me as asized CRPCs, n (%)a 81 (78) 23 (22) Me as asized CRPC pa ien s, n (%)a11 (48) 12 (52) <0.0001 P os a ec omy specimens: Gleason sco e, n (%)a <737 (92) 3 (8) 7 42 (87) 6 (13) >78 (73) 3 (27) 0.204 pT S age, n (%)a pT2 59 (88) 8 (12) pT3 31 (86) 5 (14) 0.777 AR exp ession, n (%)b weak o mode a e 13 (87) 2 (13) s ong 54 (83) 11 (17) 0.504 ERG exp ession, n (%)b nega i e o weak 48 (60) 8 (62) mode a e o s ong 32 (40) 5 (38) 0.734 PTEN exp ession, n (%)b nega i e o weak 40 (56) 7 (70) mode a e o s ong 31 (44) 3 (30) 0.508 p53 Copy numbe , n (%)b no mal 48 (84) 4 He e ozygous dele ion 911.000 SPINK exp ession, n (%)b nega i e o weak 64 (84) 11 (100) mode a e o s ong 12 (16) 0 (0) 0.349 PSA ng/mL (mean +/− SD)c14.3 (10.1) 15.4 (12.4) 0.948 Age (mean +/− SD)c63.3 (4.8) 60.7 (5.9) 0.126 Ki-67 (mean+/− SD)c5.4 (3.2) 7.7 (4.7) 0.030 EZH2 (mean +/− SD)c22.9 (15.2) 23.2 (16.9) 0.934 Table 1. Associa ion o clinicopa hological a iables. AURKA posi i i y was associa ed wi h clinico- pa hological a iables and s aing o AR, Ki67, EZH2 and PTEN exp ession le els in AR posi i e p os- a e cance samples (NEPCs we e excluded om he analysis) aχ2 es . bFishe ’s exac es . cUnpai ed es . www.na u e.com/scien i ic epo s/ 6 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 Figu eS8). Howe e , LNCaP-ARhi cells, exp essing he highes le els o AR, we e signi ican ly mo e sensi i e o AURKA inhibi ion because hei g ow h was al eady educed in he en- old lowe concen a ion o alise ib compa ed o he LNCaP-pcDNA3.1 con ol cells (Fig.5c and d). The modes 2- o 3- old inc ease in he AR le els obse ed in LNCaP-ARmo cells was no su icien o signi ican ly sensi ize cells o AURKA inhibi ion (Fig.5c). Discussion Ou da a indica e ha AURKA is an and ogen- egula ed AR a ge gene. The egula ion o AURKA speci ically occu s in he con ex o highly AR exp essing CRPC cells. We demons a e ha AR binds o he egula o y egions o he AURKA gene, and he AURKA ansc ip is up egula ed in hese cells. The inc ease in AURKA exp ession by and ogens was also p esen despi e he use o CHX ha s ops ansla ion and inhibi s possible seconda y down- s eam e ec s o AR. We also alida ed he sugges ed p os a e cance speci ic AR binding by Pome an z e al.23 in o he in onic egion o AURKA by ChIP-qPCR. The in onic AR binding was also esponsi e o enzalu a- mide ea men in he p esence o DHT in highly AR exp essing VCaP cells (27, Supplemen a y Figu eS3), and ou luci e ase assay showed ha he in onic elemen was esponsi e o he DHT s imula ion. These esul s show ha AURKA is an and ogen-inducible gene in high AR exp essing, and ogen-sensi i e p os a e cance cells. We also epo a ansc ip ional le el co ela ion o AURKA and AR in clinical AR-posi i e p os a e cance specimens in wo independen p os a e cance coho s19,24. In CRPC me as ases, AURKA posi i i y was signi - ican ly associa ed wi h high AR p o ein exp ession when AR-nega i e NEPC specimens we e excluded om he analysis. AURKA exp ession was also associa ed wi h ERG posi i i y in hese samples. The associa ion wi h Va iable AURKA exp ession PNega i e Posi i e Me -CRPCs: AR exp ession, n (%)a weak 31 (44) 4 (20) mode a e 30 (43) 7 (35) s ong 9 (13) 9 (45) 0.005 ERG exp ession, n (%)b nega i e 37 (60) 4 (20) posi i e 25 (40) 16 (80) 0.004 PTEN exp ession, n (%)b nega i e 42 (68) 11 (55) posi i e 20 (32) 9 (45) 0.420 p53 exp ession, n (%)b nega i e 40 (62) 16 (84) posi i e 25 (38) 3 (16) 0.096 SPINK exp ession, n (%)b nega i e 54 (87) 16 (80) posi i e 8 (13) 4 (20) 0.474 Ki-67 (mean +/− SD)d9.3 (9.5) 14.5 (9.3) 0.025 EZH2 (mean +/− SD)d27.9 (22.4) 39.6 (20.4) 0.035 Local CRPCs AR exp ession, n (%)a weak 1 (4) 1 (6) mode a e 9 (39) 6 (38) s ong 13 (57) 9 (56) 0.073 ERG exp ession, n (%)b nega i e 10 (38) 9 (56) posi i e 16 (62) 7 (44) 0.344 PTEN exp ession, n (%)b nega i e 6 (60) 7 (58) posi i e 4 (40) 5 (42) 1.000 SPINK exp ession, n (%)b nega i e 12 (86) 8 (67) posi i e 2 (14) 4 (33) 0.365 Ki-67 (mean +/− SD)d14.4 (9.1) 18.1 (11.1) 0.318 EZH2 (mean +/− SD)d39.2 (29.6) 58.9 (21.9) 0.032 Table 2. Associa ion o AR, EZH2, Ki67, PTEN and SPINK wi h AURKA in CRPC specimens. AURKA posi i i y was associa ed wi h s aining o AR, Ki67, EZH2 and PTEN exp ession le els in AR posi i e CRPC samples (NEPCs we e excluded om he analysis) aχ2 es . bFishe ’s exac es . cUnpai ed es . www.na u e.com/scien i ic epo s/ 7 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 ERG could be explained in clinical samples by he binding o ERG o he AURKA p omo e o enhance egion, p omo ing AURKA exp ession. I has been shown in VCaP cells ha ERG binds o he AURKA p omo e egion 18 h a e DHT ea men 28. In he same s udy, no binding was obse ed 2 h a e DHT ea men o wi hou DHT s imula ion28. He e, we used LNCaP-ARhi cells ha a e ully nega i e o ERG exp ession. The e o e, in ou in i o s udies, he inc eased and ogen s imula ed AURKA exp ession canno be explained by he ERG ansc ip- ional egula ion. I has been shown ha AURKA can unc ion as an oncogene by enhancing he ch omosome ins abili y when i is o e exp essed16,29. Du ing mi osis and cy okinesis, AURKA associa es wi h cen osomes and mic o ubules, a ec ing mi o ic spindle o ma ion and ch omosome seg ega ion16,29. AURKA o e exp ession is associa ed wi h cen osome ampli ica ion and abe an mi o ic di isions, leading o aneuploidy29. In ou s udy, he equency o AURKA posi i e s aining was signi ican ly inc eased in locally ecu en CRPC as well as me as a ic CRPC cells compa ed o p os a ec omy specimens, which is conco dan wi h ea lie s udies ha we e pe o med by o he s6,25,30. The equency o all me as asized CRPC samples (n = 104) om 31 indi idual pa ien s was lowe , which was simila o local CRPC specimens. Howe e , ou local CRPC specimens ep esen indi idual (single) pa ien s and a e hus no di ec ly compa able wi h me as asis specimens. A a simila a e as ha seen in local CRPC spec- imens, AURKA-posi i e me as ases we e ound in 52% o he men who died o p os a e cance . He e ogenei y in AURKA exp ession in CRPC me as ases om he same pa ien s has no p e iously been epo ed. Di e en AURKA le els seen in di e en p os a e cance me as ases may indica e a iabili y in and ogen sensi i i y in he di e en me as ases. P e iously, AURKA o e exp ession was epo ed in high-g ade PINs, sugges ing ha i may be an ea ly e en in he de elopmen o p os a e cance 13,30. We ound ha AURKA exp ession co ela es wi h he AR le el in ho mone-naï e p os a e cance specimens, sugges ing he AURKA exp ession may play a ole in he ea ly p o- g ession o p os a e cance . In ag eemen wi h he indings by Fu ukawa e al.30 and Bel an e al.6, we obse ed a signi ican ly sho e biochemical ecu ence a e in AURKA-posi i e p os a e cance specimens. He e we also demons a ed ha LNCaP-ARhi cells we e signi ican ly mo e sensi i e o AURKA inhibi- ion han cells exp essing no mal o modes ly inc eased AR le els. In e es ingly, Shu e al.31 demons a ed ha AURKA can phospho yla e and ac i a e AR, leading o a mo e ac i e, po en o m o AR. Recen esul s by Jones e al.32 also show ha AURKA deple ion educes AR-V exp ession as well. The e o e, he da a by us and o he s sugges ha he e may be a posi i e eedback loop be ween AR (bo h ull-leng h and AR-Vs) and AURKA. On he o he hand, using naï ely exp essing AR cells, a s udy by Sa ka e al.33 ound ha AURKA is in ol ed in p o- eosomal deg ada ion o AR, he eby sugges ing ha inhibi ing Au o a ac i i y would p omo e p os a e cance Figu e 5. Rela i e g ow h o LNCaP-AR cells ea ed wi h he AURKA-speci ic inhibi o , alise ib. The ela i e g ow h o (a) LNCaP-pcDNA3.1, (b) LNCaP-ARmo, and (c) LNCaP-ARhi cells. All cells we e equally seeded on 12-well pla es and ea ed wi h ehicle (DMSO) o wi h 10 and 100 nM alise ib (MLN8237) in iplica e. The a ea o he a ached cells in each well was measu ed daily and di ided by he mean om day 1. (d) The ela i e cell iabili y o LNCaP-pcDNA3.1, LNCaP-ARmo, and LNCaP-ARhi cells a day 4 measu ed by Alama blue and no malized o day 1 o each cell line. NT = no ea men . *p < 0.05, **p < 0.01, and ***p < 0.001, unpai ed es . www.na u e.com/scien i ic epo s/ 8 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 g ow h by inc easing AR le els and signaling. Thus, al hough mo e da a should be ga he ed o suppo he use o AURKA inhibi o s in p ima y umou s, whe e AR o e exp ession is no common, exis ing AURKA inhibi o s could be use ul in ea ing CRPC pa ien s o e exp essing ull-leng h AR wi h o wi hou AR-Vs, a inding ha should be es ed in u u e clinical ials. Ou indings on AURKA a e in line wi h ou p e ious da a on he CIP2A oncogene34. We ound ha he exp ession o CIP2A is highly inc eased in AR-nega i e NEPC. Addi ionally, pa adoxically, AR-o e exp essing CRPCs exp ess signi ican ly mo e CIP2A han ho mone-naï e PCs, al hough he exp ession le els a e lowe han hose in NEPCs. In addi ion, he exp ession o CIP2A is inc eased wi h DHT s imula ion o LNCaP-ARhi cells. O no e, he inc eased exp ession o CIP2A o AURKA upon and ogen s imula ion is no as high as wi h classical and ogen s imula ed genes, such as TMPRSS2 and he KLKs (such as PSA). Taken oge he , hese indings sugges ha oncogenes ha a e up egula ed in NEPC can also be up egula ed, al hough o a lesse ex en , in AR-posi i e CRPC. This implies ha AR-posi i e CRPC and NEPC a e no en i ely di e en o ms o he disease; ins ead, he AR-posi i e CRPC may acqui e NEPC-like pheno ypes. In conclusion, we ha e demons a ed ha AURKA is o e exp essed no only in NEPC bu also in he mos common o m o AR-posi i e CRPC. The exp ession o AURKA is and ogen- egula ed in such cells, and AURKA inhibi ion supp esses he g ow h o CRPC cells ha exp ess high le els o AR. Fu he s udies a e needed o de e - mine whe he CRPC pa ien s would bene i om combined AURKA inhibi ion ea men and ADT. Ma e ials and Me hods Clinical umo samples. The use o clinical ma e ial was app o ed by he e hical commi ee o he Tampe e Uni e si y Hospi al (TAUH, Tampe e, Finland) and he Na ional Au ho i y o Medicolegal A ai s and he Johns Hopkins Medicine Ins i u ional Re iew Boa d (au opsy samples). W i en in o med consen was ob ained om he subjec s. All me hods we e pe o med in acco dance wi h he ele an guidelines and egula ions. Rep esen a i e egions o o malin- ixed pa a in-embedded (FFPE) issue blocks we e chosen o issue mic oa - ay and cons uc ed as p e iously desc ibed8. P os a ec omy specimens. One hund ed six FFPE p os a e cance samples om consecu i e p os a ec o- mies we e ob ained om TAUH. The clinicopa hological a iables o he coho a e gi en in Table1. P og ession was de ined as a p os a e-speci ic an igen (PSA) alue o 0.5 ng/mL o mo e in wo consecu i e measu emen s o he eme gence o me as ases. Fi y-one pe cen o he pa ien s expe ienced p og ession. Locally ecu en CRPC specimens and CRPC me as ases. One-hund ed and se en y-se en FFPE samples o locally ecu en cas a ion- esis an p os a e cance (CRPC) om ansu e h al esec ion o he p os- a e (TURP) we e ob ained om he TAUH8. Nine y FFPE me as ases we e ob ained om 31 men who died o CRPC and unde wen au opsy as pa o he p ojec o Elimina e Le hal P os a e Cance (PELICAN) apid au opsy p og am a he Johns Hopkins Au opsy S udy o Le hal P os a e Cance 35. Du ing he cou se o ea men o me as a ic p os a e cance , all subjec s ecei ed and ogen dep i a ion he apy ei he wi h an LHRH analogue o an o chiec omy. Many also in e mi en ly ecei ed one o mo e an iand ogens du ing hei disease cou se. F eshly ozen clinical p os a e specimens. F esh- ozen issue specimens om 12 benign p os a e hype plasia (BPH) cases, 27 ho monally un ea ed p os a e cance s and 14 local CRPCs we e acqui ed om Tampe e Uni e si y Hospi al (Tampe e, Finland). Un ea ed p os a e cance samples we e ob ained by adical p os a ec omy and locally ecu en CRPC samples by TURP. Samples we e snap- ozen and s o ed in liquid ni ogen. All samples con ained a minimum o 70% cance ous o hype plas ic cells. P os a e cance cell lines and xenog a s. PC-3 and LNCaP p os a e cance cell lines we e ob ained om he Ame ican Type Cell Collec ion (Manassas, VA, USA). LNCaP-ARhi, LNCaP-ARmo and LNCaP-pcDNA3.1 cell lines ha e been p e iously es ablished in ou labo a o y18. Se en een p e iously es ab- lished LuCaP-se ies xenog a s (LuCaP 23.1, 23.8, 23.12, 35, 41, 58, 69, 70, 73, 77, 78, 81, 86.2, 92.1, 96, 105 and 115)36 we e p o ided by P o . Robe L. Vessella (Uni e si y o Washing on, Sea le, WA, USA). Ch oma in immunop ecipi a ion. Ch oma in immunop ecipi a ion (ChIP) was pe o med as p e- iously desc ibed37. B ie ly, cells we e ho mone s a ed o 4 days and subsequen ly ea ed wi h DHT a di e en concen a ions o 2 h. Cells we e hen ixed, pelle ed and lysed. The ch oma in was immunop ecip- i a ed wi h 10 µg o no mal abbi IgG (San a C uz Inc., San a C uz, CA, USA) o 10 µl o AR an ibody21,22. Immunop ecipi a ed DNA was pu i ied using a QIAgen mini ki (Qiagen, Dusseldo , Ge many) and ampli ied using he ollowing p ime s: o he AURKA p omo e , e e se: ACCAGCTACTCTCCCCGTGT and o wa d: CGGTAGATTGGGCAGGATT; o he AURKA enhance , e e se: CCCTTTCCGTGCTGTATTTC and o wa d: TGGCAATACTCCATCCACTCT; and o KLK3 enhance , e e se: CCAGAGTAGGTCTGTTTTCAATCC and o wa d: TGGGACAACTTGCAAACCTG. Plasmid cons uc s and cloning. ARBS in he AURKA p omo e egion was cloned in o he pGL3-Basic ec- o (P omega) con aining he luci e ase gene and mul iple cloning egion. PCR was pe o med o no mal human leukocy e DNA o ampli y he ARBS AURKA p omo e . Ampli ied ARBS AURKA p omo e was cloned be ween NheI and BglII si es. The p ime s used o PCR we e as ollows: o wa d p ime ATGCAGAGGCTAGCTAAGGACT con aining he NheI si e, and e e se p ime CGCCACTGAGATCTCCCCCACG con aining he BglII si e (unde - lined le e s indica e mu a ed componen s and bold le e s indica e he es ic ion enzyme si e). ARBS in he AURKA enhance egion we e cloned in o he pGL3-Basic ec o di ec ly ups eam o he AURKA p omo e sequence. PCR was done o no mal human leukocy e DNA o ampli y he AURKA enhance ARBS sequence. Ampli ied ARBS AURKA enhance was cloned be ween KpnI and www.na u e.com/scien i ic epo s/ 9 SCIENTIFIC REPORTS | (2017) 7:17978 | DOI:10.1038/s41598-017-18210-3 MluI si es in bo h o ien a ions. The p ime s used o PCR we e as ollows: o wa d p ime ( + o ien a ion) GGGAGATCAGAGGTACCGTGCATTCTTAT con aining a KpnI si e; e e se p ime ( + o ien a ion) AGTGGCCAAAGTTGCAACGCGTTGAACCACAAAATAA con aining a MluI si e; o wa d p ime (- o ien- a ion) CCCAGTGGCCAAGGTACCAATGATCTGAACCA con aining a KpnI si e; and e e se p ime (- o ien- a ion) ATTCTTATCACATACGCGTGATGTAAACTCTTA con aining a MluI si e (unde lined le e s indica e mu a ed componen s and bold le e s indica e he es ic ion enzyme si e). Luci e ase assay. LNCaP-ARhi cells we e g own in RPMI-1640 phenol- ee medium (Biowhi ake , Lonza) wi h 10% cha coal/dex an- ea ed (CCS) FBS (HyClone, The mo Scien i ic) and 1% Glu amine (In i ogen Inc.) o ou days be o e ans ec ions and DHT s imula ion. Cells we e pla ed on a 48-well pla e (100000–120000 cells/well). On he nex day, hey we e ansien ly ans ec ed wi h pGL3-basic (#E1751, P omega, Madison, USA), pGL3-PSA5.8-LUC ( ecei ed om p o esso Jo ma Pal imo, Uni e si y o Eas e n Finland, Finland37), pGL3-AURKAp omo e -LUC, pGL3-AURKAp omo e -AURKAenhance (+)-LUC o pGL3-AURKAp omo e -AURKAenhance (-)-LUC. Each ans ec ion was pe o med in ou eplica es. Fo no - maliza ion, con ol plasmid exp essing he Renilla luci e ase sequence was co- ans ec ed in o he cells. Fo Cos-7 cells, AR was co- ans ec ed wi h he abo emen ioned ec o s. T ans ec ions we e done wi h ei he Lipo ec amine 2000 (Li e Technologies) o je PEI (Polyplus T ans ec ion) ans ec ion eagen acco ding o he manu ac u e ’s ins uc ions. DHT s imula ion (0, 1, 10 o 100 nM) was added a e ans ec ion. A e 24 h o ans ec ion, he luci e ase ac i i y was measu ed wi h he Dual-Glo Luci e ase assay sys em (P omega) acco ding o he manu- ac u e ’s ins uc ions. Luminescence was measu ed wi h En ision Mul ilabel Reade (Pe kinElme ). Measu ed luminescence le els we e no malized o he enilla luminescence le els. G ow h analysis. PC-3, LNCaP-pcDNA3.1 and LNCaP-ARhi cell lines we e i s equally seeded (15,000 cells/well) on 24-well pla e as ou eplica es and we e allowed o adhe e o e nigh . On day 1, he cells we e ea ed wi h DMSO ( ehicle) o wi h app op ia e concen a ions (1–1000 nM) o MLN8237 (Selleck Chemicals LLC) in DMSO. All wells we e scanned daily using he Su eyo So wa e (Objec i e Imaging L d.) wi h a came a (Imaging Inc., Canada) a ached o he Olympus IX71 (Olympus, Tokyo, Japan) mic oscope, and he a ea o he a ached cells in each well was coun ed by analysis wi h ImageJ So wa e (Wayne Rasband, Na ional Ins i u es o Heal h, Be hesda, MD). Finally, he measu ed g ow h a ea o each well was di ided by he mean a ea o day 1 o each ollowing day. Immunohis ochemis y. An ibodies agains AURKA (1:50, NCL-L-AK2, No acas a), ERG (EPR3864, Epi omics, Inc.), Ki-67 (MM1, Leica Biosys ems Newcas le L d.), EZH2 (NCL-L-EZH2, No acas a) and AR (1:200, 318, No ocas a Labo a o ies L d.) we e used wi h a Powe Vision + Poly-HRP IHC ki (ImmunoVision Technologies Co.) acco ding o he manu ac u e ’s ins uc ions. The p o ocol has been p e iously desc ibed by Leinonen e al.38 and o AURKA byS a e al.17. Ape io ScanScope XT scanne (Ape io Technologies, Inc.) was used o scan he slides. Sco ing was done in a blinded ashion wi h he use o i ual mic oscope39. Ki-67 and EZH2 sco ing was pe o med wi h a web based applica ion, Immuno a io40,41. AR s aining was e alua ed om 0 o 3, AURKAand ERG s aining was e alua ed as posi i e o nega i e. S a is ical analyses. The Mann-Whi ney U, χ2, Fishe ’s exac and unpai ed es s we e used o analyze he associa ion be ween AURKA p o ein exp ession and clinicopa hological a iables, AR, ERG, Ki-67 and EZH2. Kaplan-Meie su i al analysis and he Man el-Cox es we e used o de e mine he p og ession- ee su i al o pa ien s. The unpai ed - es was used o calcula e he di e ence in p oli e a ion o g ow h cu e analyses. The Mann-Whi ney U es was used o de e mine he signi ican di e ence in luci e ase ac i i y. Da a a ailabili y s a emen . All da a gene a ed o analyzed du ing his s udy a e included in his pub- lished a icle (and i s Supplemen a y In o ma ion iles). Re e ences 1. Wang, Q. e al. And ogen ecep o egula es a dis inc ansc ip ion p og am in and ogen-independen p os a e cance . Cell 138, 245–256 (2009). 2. Te y, S. & Bel an, H. The Many Faces o Neu oendoc ine Di e en ia ion in P os a e Cance P og ession. F on Oncol. 4, 60, h ps:// doi.o g/10.3389/ onc.2014.00060 (2014). 3. Guo, Z. e al. Regula ion o and ogen ecep o ac i i y by y osine phospho yla ion. Cance Cell 10, 309–319 (2006). 4. Sche , H. e al. P os a e Cance Founda ion/Depa men o De ense P os a e Cance Clinical T ials Conso ium. An i umou ac i i y o MDV3100 in cas a ion- esis an p os a e cance : a phase 1-2 s udy. Lance 375, 1437–1446 (2010). 5. de Bono, J. S. e al. COU-AA-301 In es iga o s. Abi a e one and inc eased su i al in me as a ic p os a e cance . N Engl J Med 364, 1995–2005 (2011). 6. Bel an, H. e al. Molecula cha ac e iza ion o neu oendoc ine p os a e cance and iden i ica ion o new d ug a ge s. Cance Disco 1, 487–495 (2011). 7. Nadal, R., Schweize , M., K y enko, O. N., Eps ein, J. I. & Eisenbe ge , M. A. Small cell ca cinoma o he p os a e. Na Re U ol 11, 213–219 (2014). 8. Leinonen, K. A. e al. Loss o PTEN is associa ed wi h agg essi e beha io in ERG-posi i e p os a e cance . Cance Epidemiol Bioma ke s P e 22, 2333–2344 (2013). 9. Sa amäki, O. R. e al. TMPRSS2:ERG usion iden i ies a subg oup o p os a e cance s wi h a a o able p ognosis. Clin Cance Res 14, 3395–3400 (2008). 10. Annala, M. e al. Recu en SKIL-ac i a ing ea angemen s in ETS-nega i e p os a e cance . Onco a ge 6, 6235–6250 (2015). 11. Wa ne , S. L., Bea ss, D. J., Han, H. & Von Ho , D. D. Ta ge ing Au o a-2 kinase in cance . Mol Cance The 2, 589–595 (2003). 12. Buschho n, H. M. e al. Au o a-A o e -exp ession in high-g ade PIN lesions and p os a e cance . P os a e 64, 341–246 (2005). 13. Lee, E. C., F olo , A., Li, R., Ayala, G. & G eenbe g, N. M. Ta ge ing Au o a kinases o he ea men o p os a e cance . Cance Res 66, 4996–5002 (2006).