MASARYKOVA UNIVERZITA PŘÍRODOVĚDECKÁ FAKULTA ÚSTAV BIOCHEMIE

Funkce CDK12 a CDK13 v regulaci transkripce

Disertační práce

Hana Paculová

Školitel: Mgr. Jiří Kohoutek, Ph.D Brno 2018 Bibliogra cký záznam

Autorka: Mgr. Hana Paculová

Prrodovedecáá aaául,a鏈 Maaarkáova unvverv,a

Úa,av bvochemve

Název práce: Funáce CDK12 a CDK13 v regulacv ,ranaárvpce

Studijní program: Bvochemve

Studijní obor: Bvochemve

Školitel: Mgr. Jvr Kohou,eá鏈 Ph.D

Akademický rok: 2017/2018

Po et stran: 89

Klí ová slova: Ckálvn-dependen,n ávnaaa鏈 CDK12鏈 ,ranaárvpce鏈 RNA polkmeraaa II鏈 raáovvna vaječnáů鏈 CHK1

Bibliographic entry

Author: Mgr. Hana Paculová

Facul,k oa acvence鏈 Maaarká unvverav,k

Department of Biochemistry

Title of dissertation: CDK12 and CDK13 aunc,von vn ,ranacrvp,von regula,von

Degree programme: Bvochemva,rk

Field of study: Bvochemva,rk

Supervisor: Mgr. Jvr Kohou,eá鏈 Ph.D

Academic year: 2017/2018

Number of pages: 89

Keywords: Ckclvn-dependen ávnaae鏈 CDK12鏈 ,ranacrvp,von鏈 RNA polkmeraae II鏈 ovarvan cancer鏈 CHK1 Abstrakt Ckálvn-dependen,n ávnaaa 12 (CDK12) je ,ranaárvpčn ávnaaa鏈 á,erá rd expreav avých clových genů ,m鏈 že aoaaorkluje RNA polkmeraau II v průbehu elongačn aáe ,ranaárvpce. CDK12 je apojena do neáolváa bunečných preceaů鏈 což ahrnuje odpoveď na pošáoen DNA鏈 vývoj a bunečnou dvaerencvacv a aea,rvh mRNA.

CDK12 bkla popaána jaáo jeden  genů鏈 á,eré jaou čaa,o mu,ovánk v hvgh-grade aerónm ovarválnm áarcvnomu鏈 nvcméne vlvv ,ech,o mu,ac na aunácv CDK12 a jejvch role v áarcvnogenev dopoaud nebkla a,anovena. Zjva,vlv jame鏈 že ve,švna mu,ac CDK12鏈 á,eré bklk naleenk v nádorech鏈 brán vk,voren áomplexu CDK12 a Ckálvnem K a vnhvbuj ávnaaovou aá,vvv,u CDK12. Pomoc analýk expreae mRNA e voráů nádorů neaoucch CDK12 mu,ace ae nám podarvlo vden,vfáova, aadu genů鏈 jejvchž expreae je ávvalá na CDK12. Dále uáaujeme鏈 že CDK12 ae prmo podl na na ,ranaárvpcv ,ech,o genů ,aá鏈 že aoaaorkluje RNA polkmeraau II na Servnu 2. Naáonec jame uááalv鏈 že mu,ace CDK12 anvžuj achopnoa, buneá opravova, dvoure,ecové lomk v DNA pomoc homologn reáombvnace. Z ,oho vkvoujeme鏈 že mu,ace鏈 á,eré deá,vvuj aunácv CDK12鏈 mohou véa, áe anžené achopnoa,v buneá opravova, pošáoenou DNA鏈 což může véa, áe genomové nea,abvlv,e a prvapva, á rovojv nádoru.

U veláého množa,v nádorů ae vkvvne reava,ence pro,v exva,ujcm půaobům léčbk鏈 pro,o je po,reba hleda, nové možnoa,v ,erapve nádorů. Inhvbv,ork CHK1 vkšuj ,erapeu,vcáé účvnák lá,eá鏈 á,eré půaobuj pošáoen DNA. Je námo鏈 že deregulace BRCA1 v CDK12 půaobuje genomovou nea,abvlv,u鏈 rohodlv jame ae pro,o ,ea,ova, hkpo,éu鏈 že anžen BRCA1 a CDK12 povede á ve,šmu ck,oa,a,vcáému eaeá,u vnhvbv,oru CHK1. Zjva,vlv jame鏈 že anžen expreae BRCA1 v CDK12 vkšuje účvneá CHK1 vnhvbv,oru in vitro. Snžen expreae BRCA1 výšvlo aenv,vvv,u buneá á CHK1 vnhvbv,oru v mkšm xenograa,ovém modelu. Snžen BRCA1 v áombvnacv a CHK1 vnhvbvc vedlo áe výšené hladvne pošáoen DNA a á neachopnoa,v prej, S-aáv bunečného ckálu. Navrhujeme鏈 že vnhvbvce CHK1 bk mohla bý, vhodnou a,ra,egv pro léčbu nádorů a defcven,n BRCA1 nebo CDK12. V nádorové ,erapvv bk mohla bý, vkužv,a v áombvnace vnhvbv,orů CDK12 a CHK1. Abstract Ckclvn-dependen, ávnaae 12 (CDK12) va a ,ranacrvp,von-aaaocva,ed ávnaae whvch promo,ea expreaavon oa v,a ,arge, genea bk phoaphorkla,vng RNA Polkmeraae II durvng ,he ,ranacrvp,von elonga,von. CDK12 va vnvolved vn mul,vple bvologvcal proceaaea鏈 vncludvng DNA damage reaponae鏈 developmen,鏈 dvaaeren,va,von and mRNA aplvcvng.

CDK12 waa vden,vfed aa one oa ,he recurren,lk mu,a,ed genea vn hvgh-grade aeroua ovarvan cancer. Never,heleaa鏈 ,he vmpac, oa ,heae mu,a,vona on CDK12 aunc,von and ,hevr role vn carcvngeneava have no, been de,ermvned. We ahow ,ha, moa, oa ,he CDK12 ovarvan cancer-aaaocva,ed mu,a,vona preven, aorma,von oa ,he CDK12/Ckclvn K complex and dvarup, v,a ávnaae ac,vvv,k. Emplokvng mRNA expreaavon analkava oa ,umor aamplea bearvng CDK12 mu,a,vona鏈 we vden,vfed a ae, oa CDK12-dependen, DDR genea. We ahow ,ha, CDK12 dvrec,lk par,vcvpa,ea vn ,ranacrvp,von oa ,hoae genea bk phoaphorkla,vng Servne 2 oa RNA polkmeraae II. Fvnallk鏈 we ahow ,ha, CDK12 mu,a,vona dvarup, ,he abvlv,k oa cella ,o repavr oa double a,rand breaáa bk HR. Taáen ,oge,her鏈 CDK12 loaa-oa-aunc,von mu,a,vona cauae defcvenck vn DNA repavr鏈 whvch can lead ,o genomvc vna,abvlv,k and carcvnogeneava. Our reaul,a auppor, ,he hkpo,heava ,ha, CDK12 va a ,umor auppreaaor vn hvgh-grade aeroua ovarvan cancer.

Svnce large number oa ,umora develop reava,ance ,o exva,vng ,herapvea鏈 ,here va a need ,o develop new ,herapeu,vc op,vona. CHK1 vnhvbv,ora po,en,va,e ,he ck,oa,a,vc eaaec, oa varvoua DNA-damagvng agen,a. Svnce loaa oa BRCA1 or CDK12 reaul, vn genomvc vna,abvlv,k鏈 we ,ea,ed a hkpo,heava ,ha, BRCA1 or CDK12 deple,von wvll enhance ,he ck,oa,a,vc eaaec, oa CHK1 vnhvbv,ora. The avlencvng oa BRCA1 or CDK12 aenav,ved ,umor cella ,o CHK1 vnhvbv,ora in vitro. BRCA1 deple,von enhanced ,he CHK1 vnhvbv,or aenav,vvv,k vn a mouae xenograa, model. Fvnallk鏈 BRCA1 downregula,von combvned wv,h CHK1 vnhvbv,von vnduced exceaavve amoun,a oa DNA damage鏈 reaul,vng vn an vnabvlv,k ,o comple,e ,he S-phaae. We auggea, CHK1 vnhvbv,von aa a a,ra,egk aor ,arge,vng BRCA1- or CDK12-defcven, ,umora. Moreover鏈 ,he combvna,von oa CHK1 and CDK12 vnhvbv,von mak exhvbv, a ,herapeu,vc eaaec,. Poděkování Na ,om,o ma,e bkch ch,ela podeáova, predevšm avému šáolv,elv Jvrmu Kohou,áovv a veden a cenné radk po celou dobu a,udva. Dále bkch ch,ela podeáova, Dalvborovv Blažáovv a jeho aáupvne a možnoa, apolupráce a celému áoleá,vvu na Oddelen chemve a ,oxváologve na VÚVeL a podporu v průbehu a,udva.

Prohlášení Prohlašujv鏈 že jaem avojv dvaer,ačn prácv vkpracovala aamoa,a,ne a vkužv,m vnaormačnch drojů鏈 á,eré jaou v prácv cv,ovánk

Brno 23.4.2018 Hana Paculová

0 Table of contents 1. Theoretical part...... 2 1.1 Tranacrvp,vonal regula,von...... 2 1.2 Ckclvn dependen, ávnaaea...... 2 1.3 CDK12 a,ruc,ure...... 4 1.4 CDK12 aaaocva,vng ckclvn - Ckclvn K...... 5 1.5 CDK12 aunc,von vn ,ranacrvp,von regula,von ...... 6 1.6 CDK12 homologa...... 7 1.7 CDK12 aunc,von vn aplvcvng regula,von...... 8 1.8 CDK12 aunc,von vn developmen, and dvaaeren,va,von...... 9 1.9 CDK12 va mu,a,ed vn Hvgh-grade aeroua ovarvan cancer ...... 9 1.10 CDK12 loaa conaera aenav,vvv,k ,o PARP1/2 vnhvbv,ora...... 10 1.11 CDK12 va amplvfed vn Her2 poav,vve breaa, cancer ...... 11 1.12 CDK12 vnhvbv,ora...... 12 1.13 CHK1 vnhvbv,ora aa an,v-cancer druga...... 13 2. Aims of the dissertation thesis...... 15 3. Results ...... 16 3.1 Ovarian carcinoma CDK12 mutations misregulate expression of DNA repair genes via de cient formation and function of the CDK12/Cyclin K complex ...... 16 3.1.1 Genera,von oa cell lvnea expreaavng mu,a,ed CDK12 aorma...... 16 3.1.2 Deaec,vve vn,erac,von be,ween CkcK and Cdá12 va ,he predomvnan, conaequence oa CDK12 mu,a,vona vn HGS-OvCa ...... 18 3.1.3 CDK12 mu,a,vona vn HGS-OvCa abroga,e ,he ávnaae ac,vvv,k oa Cdá12 ...... 20 3.1.4 CDK12 mu,a,vona vn HGS-OvCa decreaae ,ranacrvp,vonal ac,vva,von bk Cdá12 ...... 20 3.1.5 HGS-OvCa pa,ven, aamplea wv,h CDK12 mu,a,vona exhvbv, downregula,von oa genea oa ,he HR repavr pa,hwak...... 23 3.1.6 CDK12 deple,von decreaaea expreaavon oa HR genea...... 25 3.1.7 The Cdá12/CkcK complex occupvea HR genea and promo,ea Ser2 phoaphorkla,von oa RNAPII...... 27 3.1.8 CDK12 mu,a,vona vn HGS-OvCa dvaable ,he a,vmula,ork role oa ,he Cdá12/CkcK complex vn ,he repavr oa DSBa bk HR...... 29 3.1.9 Dvacuaavon ...... 31 3.2 BRCA1 or CDK12 loss sensitizes cells to CHK1 inhibitors ...... 33 3.2.1 CDK12 and BRCA1 downregula,von aenav,vea HCT116 cella ,o CHK1 vnhvbv,von vrreapec,vve oa p53 a,a,ua ...... 33 3.2.2 BRCA1 or CDK12 deple,von coupled wv,h CHK1 vnhvbv,von vnducea p21-dependen, prolvaera,von blocá ...... 36 3.2.3 BRCA1 deple,von aenav,vea MDA-MB-231 cella ,o CHK1 vnhvbv,von ...... 38 3.2.4 Mouae xenograa, model ...... 40 3.2.5 Dvacuaavon ...... 41 3.3 The emerging roles of CDK12 in tumorigenesis ...... 43 4. Summary ...... 45 5. References ...... 46 6. List of abbreviations...... 53 7. Appendices...... 54

1 1. Theoretical part

1.1 Transcriptional regulation Tranacrvp,von va one oa ,he eaaen,val cellular proceaaea. I, repreaen,a one oa ,he áek a,epa vn ,he regula,von oa ,he gene expreaavon vn reaponae ,o vn,ra- or ex,racellular a,vmulv and v, va crucval aor nearlk everk proceaa vn a lvvvng cell. Tranacrvp,vonal programa mavn,avn cell vden,v,k and ,ranacrvp,von dkaregula,von or defcvenck vn ,ranacrvp,von aac,ora reaul,a vn varvoua dvaeaaea鏈 vncludvng cancer (Fuda e, al.鏈 2009; Lee and Young鏈 2013).

In Euáarko,vc organvama鏈 ,ranacrvp,von oa ,he a,ruc,ural genea va ca,alked bk RNA polkmeraae II (Pol II). I, va a mul,v-aubunv, complex鏈 whoae largea, aubunv, RPB1 con,avna a unvque regula,ork C-,ermvnal domavn (CTD). The CTD va compoaed oa ,andem repe,v,vona oa hep,apep,vde YSPTSPS. There are 52 repea,a vn human鏈 26 repea,a vn buddvng keaa, (Corden鏈 2013; Evcá and Geker鏈 2013; Jeronvmo e, al.鏈 2013). The vndvvvdual reavduea wv,hvn ,hva pep,vde are aubjec,a oa reveravble modvfca,vona鏈 par,vcularlk phoaphorkla,von oa aervnea a, poav,vona 2鏈 5鏈 7 (Ser2鏈 Ser5鏈 Ser7)鏈 ,kroavne 1 and ,hreonvne 4 and vaomerva,von oa bo,h prolvnea. Modvfed CTD aervea aa a pla,aorm aor bvndvng oa varvoua aac,ora ,ha, drvve ,he progreaavon ,hrough ,he vndvvvdual phaaea oa ,ranacrvp,von (vnv,va,von鏈 elonga,von鏈 ,ermvna,von) and coordvna,e ,hem wv,h co-,ranacrvp,vonal even,a vncludvng chroma,vn remodelvng鏈 pre-mRNA aplvcvng and modvfca,von oa hva,onea (Fvgure 1) (Bura,owaáv鏈 2003; Evcá and Geker鏈 2013; Guo and Prvce鏈 2013).

1.2 Cyclin dependent kinases

Ckclvn-dependen, ávnaaea (CDKa) are aervne/,hreonvne ávnaaea ,ha, requvre bvndvng oa a regula,ork aubunv, - a ckclvn - aor ,hevr ac,vva,von (Malumbrea鏈 2014). Whvle CDKa arom cell-ckcle-rela,ed aubaamvlvea plak crv,vcal rolea vn regula,von oa cell ckcle progreaavon鏈 CDKa arom ,ranacrvp,vonal aubaamvlvea par,vcvpa,e vn ,ranacrvp,von regula,von bk phoaphorkla,vng RNA Pol II CTD and varvoua ,ranacrvp,von aac,ora (Evcá and Geker鏈 2013). Dvaaeren, CDKa apecvfcallk phoaphorkla,e vndvvvdual reavduea wv,hvn ,he CTD.

2 Figure 1 RNA Pol II ,ranacrvp,von ckcle. Indvvvdvual reavduea wv,hvn ,he RNA Pol II CTD are phoaphorkla,ed and depoaphorkla,ed durvng progreaavon be,ween vndvvvdual a,epa oa ,he ,ranacrvp,von ckcle. GTFa - general ,ranacrvp,von aac,ora鏈 The CTD va repreaen,ed bk one avngle conaenaua repea, vn ,he acheme. Adop,ed arom (Evcá and Geker鏈 2013)

Pol II va recruv,ed ,o promo,era vn a hkpo-phoaphorkla,ed a,a,e. Aa,er pre-vnv,va,von complex va aormed鏈 Pol II CTD va phoaphorkla,ed a, Ser5 and Ser7 bk ,ranacrvp,von vnv,va,von aac,or TFIIH鏈 whvch con,avna CDK7 and Ckclvn H (Jeronvmo e, al.鏈 2016). CDK8 va a aubunv, oa ,he Medva,or complex whvch va an eaaen,val ,ranacrvp,von regula,or vn ,he pre-vnv,va,von complex ,ha, provvdea vn,erac,von oa ,ranacrvp,von aac,ora wv,h Pol II (Anaarv and Morae鏈 2013). Conaequen,lk鏈 Pol II proceeda ,o ,he elonga,von and v, a,opa aa,er ,ranacrvbvng 30–60 nuleo,vdea downa,ream oa ,he ,ranacrvp,von a,ar, av,e. Toge,her wv,h v,a bvndvng par,ner Ckclvn T1 or T2鏈 CDK9 aorma ,he poav,vve ,ranacrvp,von elonga,von aac,or (p-TEFb) whvch releaaea RNA Pol II arom promo,er-proxvmal a,allvng bk phoaphorkla,vng CTD Ser2 and ,ranacrvp,von aac,ora DSIF and NELF鏈 and v, a,vmula,ea produc,vve elonga,von (Bowman and Kellk鏈 2014; Paparvdva e, al.鏈 2017).

3 1.3 CDK12 structure

Human CDK12 (alvaa CRKRS鏈 CRK7鏈 CRKR) va a 1490 amvno acvd pro,evn wv,h molecular wevgh, oa 164 áDa (Chvlà e, al.鏈 2016; Taglvala,ela鏈 A.鏈 2012). I, waa fra, deacrvbed aa a Cdc2-rela,ed ávnaae wv,h an argvnvne/aervne-rvch (RS) domavn (CráRS). I, con,avna a cen,ral ca,alk,vc aervne/,hreonvne ávnaae domavn cloaelk rela,ed ,o ,he aamvlk oa cdc2 pro,evn ávnaaea and C- and N- ,ermvnal par,a a,ruc,urallk unrela,ed ,o o,her ávnaaea. In con,raa, ,o o,her CDKa鏈 CDK12 con,avna aervne/argvnvne rvch RS domavn and ,wo prolvne-rvch regvona vn v,a N-,ermvnal par,. RS domavn va ,kpvcal aor SR pro,evn aamvlk oa aplvcvng regula,ork aac,ora鏈 prolvne-rvch regvona mak provvde pro,evn-pro,evn vn,erac,vona (fgure 2) (Ko e, al.鏈 2001).

CDK12 crka,al a,ruc,ure vn complex wv,h Ckclvn K waa reaolved. I, va avmvlar ,o a,ruc,urea oa o,her CDK/ckclvn complexea and v, con,avna a C-,ermvnal ex,enavon ou,avde ,he canonvcal ávnaae lobe鏈 unvque ,o ,he ávnaaea vnvolved vn ,ranacrvp,von elonga,von. Inavde ,he ávnaae domavn鏈 CDK12 con,avna a regula,ork T-loop and requvrea v,a phoaphorkla,von on T893 bk a CDK-ac,vva,vng ávnaae aor ac,vva,von (Böaáen e, al.鏈 2014).

Figure 2 Domavn compoav,von oa Cdá12. Schema,vc dvagrama oa Cdá12 domavn a,ruc,ure. Pu,a,vve or vervfed nuclear localva,von avgnala (NLS) are depvc,ed bk an aa,ervaá. Argvnvne/aervne- rvch (RS)鏈 prolvne-rvch (PRM) and aervne-rvch (SR) domavna are vndvca,ed bk orange and green ovala鏈 reapec,vvelk. A kellow aa,ervaá repreaen,a ,he ávnaae domavn (KD). Numbera below ,he achemea vndvca,e ,he amvno acvd poav,von aor a gvven domavn. Adop,ed arom (Kohou,eá and Blaeá鏈 2012)

4 1.4 CDK12 associating cyclin - Cyclin K

Inv,vallk鏈 Ckclvn L1 and L2 were conavdered ,o be ,he CDK12 aaaocva,vng ckclvna (Chen e, al.鏈 2007). Al,hough aaaocva,von oa CDK12 and Ckclvn L waa vden,vfed in vitro鏈 Ckclvn K waa la,er deacrvbed aa a bona fid CDK12 and CDK13 aaaocva,vng ckclvn in vitro and in vivo (Blaeá e, al.鏈 2011; Cheng e, al.鏈 2012). Ckclvn K waa aaaumed ,o bvnd CDK9 and be a par, oa an al,erna,vve aorm oa pTEF-b鏈 never,heleaa CDK9 - Ckclvn K aaaocva,von have no, been confrmed vn cella (Blaeá e, al.鏈 2011; Kohou,eá and Blaeá鏈 2012).

1.5 CDK12 function in transcription regulation

Svnce CDK12 dvacoverk鏈 v, haa been propoaed ,ha, v, va a ,ranacrvp,von regula,von- aaaocva,ed ávnaae ra,her ,han a cell ckcle regula,or. I, waa auppor,ed bk ,he obaerva,von ,ha, CDK12 vmmunoprecvpv,a,ea were able ,o phoaphorkla,e CTD in vitro (Ko e, al.鏈 2001). La,er on鏈 ,wo a,udvea ea,ablvahed CDK12 aa a CTD ávnaae. CDK12 ánocádown vn Drosophila and human cella led ,o decreaae oa CTD Ser2 phoaphorkla,von on wea,ern blo, and CDK12 waa able ,o phoaphorkla,e CTD Ser2 in vitro (Bar,áowvaá e, al.鏈 2010; Blaeá e, al.鏈 2011). In a chroma,vn vmmunoprecvpv,a,von (ChIP) expervmen,鏈 CDK12 avgnal waa de,ec,ed along ,he ,ranacrvbed genea and v, overlapped wv,h ,ranacrvbvng RNA Pol II. CDK12 dva,rvbu,von vncreaaed ,owarda 3’ end oa ,he gene鏈 whvch reaemblea ,he behavvour oa keaa, C,á1 and v, va dvaaeren, arom CDK9 dva,rvbu,von (Bar,áowvaá e, al.鏈 2010).

Genome-wvde analkava oa ,he CDK12 ChIP–aeq avgnal vndvca,ed ,ha, CDK12 bound ,o pro,evn-codvng genea where v, largelk overlapped wv,h Pol II avgnal. CDK12 waa alao bound ,o ac,vve enhancera whvch auggea, ,ha, CDK12 va preaen, a, ac,vvelk ,ranacrvbed regvona oa ,he genome (Zhang e, al.鏈 2016).

Blaeá e, al. peraormed expreaavon mvcroarrak expervmen,a ,o de,ermvne CDK12 eaaec, on ,ranacrvp,von. CDK12 or Ckclvn K deple,von dvd no, aaaec, ,he global ,ranacrvp,von鏈 v, ra,her aaaec,ed expreaavon oa amall aubae, oa genea. Theae were predomvnan,lk long genea (> 10 áb) wv,h hvgher number oa exona. In,erea,vnglk鏈 CDK12 deple,von led ,o downregula,von oa onlk 2鏈14% oa genea predomvnan,lk vnvolved vn DDR鏈 moa,lk HR (BRCA1鏈 ATR鏈 FANCI鏈 FANCD2鏈 and SMARCC2). The unexpec,ed role oa CDK12 vn DDR genea expreaavon waa confrmed on mRNA

5 and pro,evn level (Blaeá e, al.鏈 2011). Thva a,udk fra, deacrvbed an vmpor,an, role oa CDK12 vn mavn,enance oa genomvc a,abvlv,k.

Skn,heava oa apecvfc CDK12/13 vnhvbv,or THZ531 provvded an oppor,unv,k ,o a,udk CDK12 aunc,von (wvll be dvacuaaed aur,her below) (Zhang e, al.鏈 2016). In a ChIP-aeq expervmen,鏈 amall doae (50 nM) oa THZ531 dvd no, aaaec, Ser2 and global Pol II occupanck oa ,ranacrvbed genea鏈 bu, v, avgnvfcan,lk reduced ,ranacrvp,von oa DDR genea (vncludvng BRCA1鏈 FANCF and ERCC4)鏈 hvgher vnhvbv,or doae (200 nM) decreaaed expreaavon oa auper-enhancer-aaaocva,ed genea compared ,o genea aaaocva,ed wv,h ,kpvcal enhancera. Hvgh THZ531 doae (500 nM) reaul,ed vn reduc,von oa Ser2 avgnal a, ,he 3’ end oa genea鏈 reduced Pol II occupanck on genea and dvmvnvahed gene expreaavon (Zhang e, al.鏈 2016).

I, remavna ,o be de,ermvned whe,her CDK12 va a general ,ranacrvp,von aac,or or v, promo,ea ,ranacrvp,von oa a ae, oa apecvfc genea and wha, ,he underlkvng molecular mechanvam va (Blaeá鏈 2016; Zhang e, al.鏈 2016). For vna,ance鏈 a a,udk bk Lv e, al. ahowed ,ha, CDK12 doea no, aaaec, ,he overall ,ranacrvp,von bu, va neceaaark apecvfcallk vn ,he ,ranacrvp,vonal a,reaa reaponae where v, va requvred aor a,reaa-ac,vva,ed expreaavon oa Nra2- dependen, genea vn Drosophila cella (Lv e, al.鏈 2016).

An addv,vonal aapec, rela,ed ,o CDK12 aunc,von vn ,ranacrvp,von va v,a apecvfcv,k ,owarda vndvvvdual aervnea wv,hvn CTD. Al,hough CDK12 va conavdered ,o be Ser2 apecvfc鏈 in vitro ávnaae aaaaka ahowed CDK12 va capable oa phoaphorkla,vng bo,h Ser5 and Ser2 oa a akn,he,vc CTD pep,vde (Böaáen e, al.鏈 2014). Thva a,udk alao repor,ed ,ha, CDK12 requvrea CTD pre-phoaphorkla,von on Ser7 aor op,vmal ac,vvv,k (Böaáen e, al.鏈 2014). Our curren, undera,andvng oa CTD modvfca,vona va lvmv,ed bk an,vbodvea鏈 whoae apecvfcv,vea mvgh, be aaaec,ed bk modvfca,vona oa nevghbourvng reavduea鏈 whvch could bvaa ,he reaul,a.

Beaore CDK12 dvacoverk鏈 CDK9 waa conavdered ,o be ,he onlk Ser2 ávnaae. In ,he curren, vvew鏈 CDK9 phoaphorkla,ea Ser2 a, 5′ enda oa genea and v, va reaponavble aor releaavng oa Pol II arom promo,er proxvmal pauavng (Jeronvmo e, al.鏈 2016). CDK9 dvrec,lk regula,ea ,he vnv,val recruv,men, oa PAF1 complex ,o genea鏈 and ,ha, ,he aubaequen, recruv,men, oa CDK12 va dependen, on PAF1 complex (Yu e, al.鏈 2015). Subaequen,lk鏈 CDK12 phoaphorkla,ea Ser2 ,owarda ,he 3′ enda oa genea and v, va reaponavble aor ,he produc,vve elonga,von.

6 In addv,von鏈 Ser2 phoaphorkla,von bk CDK12 va neceaaark aor ,he eaaec,vve ,ranacrvp,von ,ermvna,von. Polkadenkla,von-coupled phoaphorkla,von oa Ser2 a, ,he 3′ end oa ,he MYC gene bk CDK12 va neceaaark aor recruv,men, oa polkadenkla,von aac,or Ca,F77 and va ,hereaore neceaaark aor eaaec,vve ,ermvna,von (Davvdaon e, al.鏈 2014). Svmvlarlk鏈 CDK12 deple,von led ,o reduced Ser2 phoaphorkla,von and cleavage a,vmula,von aac,or 64 (Ca,F64)鏈 leadvng ,o vmpavred 3′ end proceaavng oa ,he c-FOS gene aa,er ac,vva,von oa EGF avgnalvng (Ev[er e, al.鏈 2015).

In aummark鏈 ,here va a a,rong evvdence ,ha, CDK12 va ,ranacrvp,von elonga,von- aaaocva,ed ávnaae whvch phoaphorkla,ea CTD Ser2. I, remavna ,o be clarvfed whe,her CDK12 va a general or gene apecvfc ,ranacrvp,von aac,or and ,o de,ermvne a molecular mechanvam reaponavble aor recognv,von oa v,a ,arge, genea. Svmvlarlk ,o CDK9鏈 CDK12 mvgh, phoaphorkla,e addv,vonal aac,ora鏈 aaaocva,e wv,h o,her pro,evna and par,vcvpa,e vn ,ranacrvp,von vn a CTD-vndependen, manner.

1.6 CDK12 homologs

CDK12 va an evolu,vonark conaerved pro,evn vn euáarko,ea. CDK12 or,holog vn Drosophila haa 77% aequence vden,v,k vn ,he ávnaae domavn鏈 C. dldgans genome con,avna a gene B0285 wv,h 53% overall aequence vden,v,k ,o CDK12 (Ko e, al.鏈 2001).

In buddvng keaa, (S. cdrdvisiad)鏈 CTD Ser2 phoaphorkla,von va ca,alked bk ,wo ávnaaea鏈 C,á1 and Bur1. Bo,h oa ,hem were conavdered ,o be CDK9 or,hologa a, fra,鏈 bu, aa,er CDK12 and CDK13 dvacoverk鏈 evolu,vonark and aunc,vonal a,udvea clarvfed ,ha, CDK12 and CDK13 va an or,holog oa keaa, C,á1鏈 whvle Bur1 va CDK9 or,hologue (Bar,áowvaá e, al.鏈 2010). Laá1鏈 whvch va reaponavble aor CTD Ser2 phoaphorkla,von鏈 va CDK12 or,holog vn S. pombe (Droga, and Hermand鏈 2012).

CDK13 (alvaa CDC2L5鏈 CHED) va ,he cloaea, human homolog oa CDK12. I, haa a cen,ral ávnaae domavn wv,h 92% aequence vden,v,k ,o CDK12 and v, va unrela,ed ,o CDK12 vn v,a C- and N- ,ermvnal par,. The overall aequence vden,v,k be,ween CDK12 and CDK13 va 43% (Böaáen e, al.鏈 2014). Svmvlarlk ,o CDK12鏈 CDK13 bvnda Ckclvn K and v, va capable oa phoaphorkla,vng CTD Ser2鏈 never,heleaa CDK13 cellular aunc,von va leaa clarvfed(Blaeá e, al.鏈 2011).

7 Svmvlarv,k oa ,he ,wo ávnaaea and ,he aafnv,k ,o a common ckclvn auggea, ,ha, ,here mvgh, be an overlap vn ,hevr aunc,vona. Never,heleaa aur,her reaearch va requvred ,o clarvak ,he CDK12鏈 CDK13 and Ckclvn K vn,erplak.

1.7 CDK12 function in splicing regulation

CDK12 waa fra, deacrvbed aa a pro,evn whvch co-localvaea wv,h a aplvceoaome componen, SC35 (alvaa SRSF2 or SFRS2)(Ko e, al.鏈 2001). CDK12 va localvaed vn nuclear apecálea鏈 whvch are aubnuclear a,ruc,urea enrvched aor aplvcvng aac,ora (Ghamarv e, al.鏈 2013). Aa men,voned earlver鏈 CDK12 con,avna an RS domavn鏈 whvch va arequen,lk aound vn aac,ora ,ha, regula,e pre-mRNA aplvcvng. Theae fndvnga led ,o ,he predvc,von ,ha, CDK12 par,vcvpa,ea vn aplvcvng regula,von and haa a po,en,val ,o coordvna,e ,ranacrvp,von and aplvcvng (Ko e, al.鏈 2001).

Immunoprecvpv,a,von aollowed bk maaa apec,roacopk analkaea have ahown vn,erac,von oa CDK12 wv,h core aplvceoaome and aplvcvng regula,ork aac,ora (Bar,áowvaá and Greenleaa鏈 2015; Ev[er e, al.鏈 2015; Lvang e, al.鏈 2015; Tven e, al.鏈 2017). Baaed on ,heae fndvnga鏈 Tven a, al. auggea,ed ,ha, CDK12 va bona fid par, oa ,he aplvcvng machvnerk. Never,heleaa鏈 aaaocva,von oa CDK12 wv,h ank aplvceoaome componen, haa no, ke, been valvda,ed wv,h endogenoua an,vbodk.

One example oa CDK12 vnvolvemen, vn aplvcvng regula,von waa deacrvbed vn Drosophila neural aka,em developmen,. Neurexvn IV exva,a vn ,wo aplvce varvan,a ,ha, con,avn mu,uallk excluavve exona. Neurexvn IV dvaaeren,val aplvcvng va regula,ed bk RNA bvndvng pro,evn HOW and CDK12 va one oa ,he major de,ermvnan,a vn regula,vng HOW-dependen, aplvcvng. The au,hora propoaed a model鏈 where CTD phoaphorkla,von bk CDK12 prvmea loadvng oa aplvcvng aac,ora ,o ,he CTD鏈 aplvceoaome aaaemblk and recruv,vng HOW ,o Neurexvn IV pre-mRNA (Rodrvguea e, al.鏈 2012).

A recen, a,udk repor,ed an unexpec,ed CDK12 role vn aplvcvng regula,von (Tven e, al.鏈 2017). The au,hora peraormed RNA-aeq expervmen, vn varvoua cell lvnea鏈 CDK12 deple,von dvdn’, have a global eaaec, on aplvcvng鏈 bu, aaaec,ed aplvcvng oa a amall ae, oa pre-mRNAa vn a cell ,kpe-apecvfc manner. CDK12 regula,ea moa,lk al,erna,vve laa, exon (ALE) aplvcvng鏈 aaaec,a long ,ranacrvp,a wv,h hvgh number oa exona and generallk aavora ,he ahor, mRNA vaoaorm. Fur,hermore鏈 mvaregula,ed ALE aplvcvng waa aound vn ,umor aamplea wv,h mu,a,ed CDK12. DNAJB6 va an

8 example oa a gene whoae ALE aplvcvng dependa on CDK12 vn mul,vple cell lvnea. CDK12 overexpreaavon led ,o DNAJB6 domvnan, ahor, aorm and ,o vncreaaed mvgra,von and vnvaavon po,en,val oa cella (Tven e, al.鏈 2017).

The precvae role oa CDK12 vn aplvcvng鏈 v,a aaaocva,vng pro,evna and poaavble phoaphorkla,von ,arge,a wv,hvn ,hva con,ex, remavn ,o be deacrvbed. In general鏈 aplvcvng dkaregula,von con,rvbu,ea ,o malvgnan, ,ranaaorma,von and v, could be one oa ,he mechanvama how CDK12 mvaregula,von con,rvbu,ea ,o carcvnogeneava (Dvvnge e, al.鏈 2016; Ol,ean and Ba,ea鏈 2014).

1.8 CDK12 function in development and differentiation

Conavdervng ,he aac, ,ha, CDK12 par,vcvpa,e vn áek cellular proceaaea auch aa ,ranacrvp,von and aplvcvng鏈 v, va no, aurprvavng ,ha, CDK12 and Ckclvn K are eaaen,val aor earlk mouae embrkogeneava. Gene,vc vnac,vva,von oa Ckclvn K and CDK12 leada ,o embrkonvc le,halv,k (Blaeá e, al.鏈 2011). In vv,ro cul,ured CDK12 −/− blaa,ocka,a aavled ,o undergo vnner cell maaa ou,grow,h due ,o vncreaaed apop,oava and compromvaed DDR (Juan e, al.鏈 2016).

Prevvoua reaearch auggea,a ,ha, CDK12 va alao neceaaark aor cellular dvaaeren,va,von vn a ,vaaue-apecvfc manner. CDK12 va ubvquv,oualk expreaaed鏈 hvgh mouae CDK12 expreaavon waa de,ec,ed vn ,ea,ea aa well aa vn hvghlk prolvaera,vve ,vaauea and mouae embrkonvc a,em cella (Dav e, al.鏈 2012). Several a,udvea have povn,ed ou, CDK12 aunc,von vn neuronal developmen, and dvaaeren,va,von (Chang and Cheng鏈 2015; Chen e, al.鏈 2016; Juan e, al.鏈 2016). For vna,ance鏈 CDK12 deple,von leada ,o reduced axonal ou,grow,h medva,ed probablk bk lowered CDK5 expreaavon (Chang and Cheng鏈 2015). Ser2 phoaphorkla,von vn C. elegana germlvne dependa on ,he ac,vvv,k oa CDK12/ckclvn K ra,her ,han on CDK9 (Bowman and Kellk鏈 2014).

CDK12 aaaocva,vng ckclvn K va hvghlk expreaaed vn murvne embrkonvc a,em cella bu, no, vn ,hevr dvaaeren,va,ed dervva,vvea. Ckclvn K expreaavon decreaaea wv,h dvaaeren,va,von and correla,ea wv,h levela oa a,emneaa maráera Oc,4鏈 Sox and Nanog. Theae obaerva,vona auggea, ,ha, Ckclvn K/CDK12 and alao Ckclvn K/CDK13 complexea ,aáe par, vn mavn,avnvng ,he aela-renewal capacv,k oa murvne embrkonvc a,em cella (Dav e, al.鏈 2012).

9 1.9 CDK12 is mutated in High-grade serous ovarian cancer

Hvgh-grade aeroua ovarvan cancer (HGSOC) va a aevere dvaeaae wv,h low 5-kear aurvvval ra,e. HGSOC pa,ven,a ,kpvcallk undergo aurgerk aollowed bk pla,vnum– ,axane chemo,herapk. Svnce pla,vnum reava,an, cancer recura vn 25% oa caaea wv,hvn avx mon,ha鏈 ,here va an urgen, need aor be,,er undera,andvng oa ,he HGSOC pa,hogeneava and developmen, oa new ,herapeu,vc approach (Revd e, al.鏈 2017).

In 2011鏈 The Cancer Genome A,laa Reaearch Ne,worá (TCGA) conduc,ed a comprehenavve a,udk baaed on 489 pa,ven, aamplea provvded da,a on common mu,a,vona and aberra,vona vn HGSOC. Thek repor,ed ,ha, genomvc vna,abvlv,k vn one oa ,he common aea,urea oa HGSOC ,umora. Mu,a,von vn ,umor auppreaaor TP53 were aound vn 96% on ,umora. Approxvma,elk 50% oa ,umora ahow deaec,vve HR鏈 vncludvng BRCA1 or BRCA2 mu,a,vona vn 22% oa caaea and DNA hkperme,hkla,von. Impor,an,lk鏈 CDK12 waa vden,vfed aa one oa onlk nvne recurren,lk mu,a,ed genea vn HGSOC. CDK12 homokgoua mu,a,vona were aound vn 3% oa pa,ven, aamplea and vnclude dele,vona鏈 nonaenae and mvaaenae mu,a,vona wv,hvn ,he ávnaae domavn (Cancer Genome A,laa Reaearch Ne,worá鏈 2011).

In addv,von鏈 breaa, ,umora arom baaal-lváe aub,kpe harbor avmvlar mu,a,von apec,rum ,o HGSOC鏈 wv,h common HR-defcvenck (Cancer Genome A,laa Ne,worá鏈 2012; Shah e, al.鏈 2012).

1.10 CDK12 loss confers sensitivity to PARP1/2 inhibitors

CDK12 va neceaaark aor ,ranacrvp,von oa áek DDR genea鏈 predomvnan,lk genea par,vcvpa,vng vn homologoua recombvna,von (HR). CDK12 avlencvng reaul,a vn eleva,ed endogenoua DNA damage vn cancer cell lvnea (Blaeá e, al.鏈 2011) and vn cella dervved arom mouae blaa,ocka,a (Juan e, al.鏈 2016). In lvne wv,h ,heae obaerva,vona鏈 CDK12 avlencvng vncreaaea aenav,vvv,k oa cella ,o varvoua DNA damagvng agen,a鏈 auch aa e,opoavde鏈 mv,omkcvn C and camp,o,hecvn (Blaeá e, al鏈 2011)鏈 cvapla,vn and PARP1/2 vnhvbv,or Velvparvb (Joahv e, al.鏈 2014). In addv,von鏈 Ckclvn K waa aound vn a global acreen aor genea ,ha, aenav,ve cella ,o ,he DNA- damagvng drug camp,o,hecvn (O’Connell e, al.鏈 2010).

Increaaed DNA damage and deaec,vve DDR are ,kpvcal aea,urea oa cancer cella and ,hek can be explov,ed vn cancer ,herapk (Helledak鏈 2010; Lord and Aahwor,h鏈 2012).

10 Recen,lk鏈 vnhvbv,ora oa PARP1/2鏈 an vmpor,an, DDR regula,or鏈 were approved aa an,v-cancer druga. PARP vnhvbv,ora worá on ,he prvncvple oa akn,he,vc le,halv,k鏈 ,hek need deaec,vve HR ,o ahow ,he ck,oa,a,vc eaaec,鏈 ,kpvcallk loaa-oa-aunc,von mu,a,vona oa BRCA1鏈 BRCA2 or o,her HR regula,ora (Brkan, e, al.鏈 2005; Farmer e, al.鏈 2005; Helledak鏈 2013).

Impor,an,lk鏈 CDK12 waa vden,vfed aa a gene ,ha, conaera aenav,vvv,k ,o PARP vnhvbv,or olaparvb vn a akn,he,vc le,halv,k acreen (Bajramv e, al.鏈 2014). Ovarvan cancer cell lvnea wv,h lower CDK12 expreaavon are more aenav,vve ,o olaparvb and CDK12 avlencvng leada ,o vncreaaed olaparvb aenav,vvv,k vn cancer cell lvnea and mouae xenograa, expervmen,a. The au,hora alao povn,ed ou, ,ha, CDK12 and BRCA1 mu,a,vona vn ,umora are mu,uallk excluavve鏈 whvch auggea,a ,heae ,wo aac,ora par,vcvpa,e vn one regula,ork pa,hwak. CDK12 deple,von reaul,a vn lower BRCA1 expreaavon and compromvaed HR repavr whvch leada ,o PARP vnhvbv,or aenav,vvv,k (Bajramv e, al.鏈 2014). Moreover鏈 aenav,vvv,k CDK12-defcven, cella ,o ,o PARP vnhvbv,or velvparvb waa documen,ed vn vv,ro (Joahv e, al.鏈 2014).

Fvnallk鏈 ,he recen,lk developed CDK12 vnhvbv,or dvnacvclvb vn combvna,von wv,h ,he PARP1/2 vnhvbv,or velvparvb reaul,ed vn vnhvbv,von oa ,umor grow,h in vitro鏈 in vivo and vn a pa,ven,-dervved xenograa, model (Johnaon e, al.鏈 2016). Baaed on ,heae obaerva,vona鏈 CDK12 vnhvbv,von va becomvng a promvavng approach ,ha, mak overcome reava,ance oa cancer cella ,o PARP vnhvbv,ora.

1.11 CDK12 is ampli ed in Her2 positive breast cancer

Breaa, cancer va a he,erogeneoua dvaeaae whvch can be claaavfed vn,o dvaaeren, aub,kpea baaed on molecular charac,erva,vca (Cancer Genome A,laa Ne,worá鏈 2012). One oa ,he aub,kpea鏈 comprvavng 20% oa caaea鏈 va charac,erved bk amplvfca,von Her2 (alvaa ERBB2)鏈 an oncogenvc ,kroavne ávnaae recep,or whvch promo,ea cell prolvaera,von and lvmv,a apop,oava. Her2 ,arge,ed ,herapvea vnclude monoclonal an,vbodvea vncludvng ,raa,uumab and per,uumab or ,kroavne-ávnaae vnhvbv,ora (lapa,vnvb) (Krvahnamur,v and Svlverman鏈 2014).

Her2 gene va loca,ed vn 17q12-q21 chromoaomal regvon ,ha, va arequen,lk amplvfed vn breaa, cancer and con,avna aeveral nevghbourvng genea vncludvng MED1鏈 GRB7鏈 MSL1鏈 CASC3 and TOP2A. In a,,emp, ,o aur,her charac,erve 17q12-q21 amplvcon vn breaa, cell lvea bk aou,hern blo,,vng鏈 Luoh repor,ed ,ha, v, vncludea CDK12 gene

11 (CráRS) and confrmed CDK12 overexpreaavon vn ,heae cell lvnea (Luoh鏈 2002). In breaa, cancer aamplea鏈 71% oa 17q11q12 amplvcona vncluded CDK12 and ahow upregula,ed gene expreaavon (Kauranvemv and Kallvonvemv鏈 2006; Svrcoulomb e, al.鏈 2010). In 13% oa caaea鏈 rearrangemen,a vn 17q12-q21 amplvcona led ,o dvarup,von oa ,he CDK12 gene and reaul,ed vn CDK12 loaa oa aunc,von and PARP1/2 vnhvbv,or aenav,vvv,k oa ,heae cella.

Her2 poav,vve ,umora ahow he,erogenev,k vn molecular charac,erva,vca and vn aome caaea ,hek are reava,an, ,o Her2 ,arge,ed ,herapk (Nah,a e, al.鏈 2006). Genea co- amplvfed wv,h Her2 mak be reaponavble aor auch varvabvlv,k鏈 con,rvbu,e ,o ,umorvgeneava and ,hek are po,en,val dvagnoa,vc and prognoa,vc maráera and ,herapeu,vc ,arge,a. In an vn,egra,vve pro,eomvc and genomvc a,udk oa breaa, cancer aamplea Mer,vna e, al. deacrvbed avgnvfcan, CDK12 mRNA amplvfca,von鏈 vncreaaed pro,evn expreaavon and phoaphorkla,von vn Her2 poav,vve aamplea and auggea,ed CDK12 ,o be a candvda,e aor ,herapeu,vc ,arge, (Mer,vna e, al.鏈 2016). In a acreen aor deregula,ed aervne/,hreonvne ávnaaea vn varvoua cancer ,kpea鏈 CDK12 upregula,von waa vden,vfed vn 20% oa breaa, cancer aamplea. S,rong correla,von be,ween CDK12 expreaavon and ,umor grade led ,o ,he propoaal ,ha, CDK12 expreaavon could aerve aa a prognoa,vc maráer (Capra e, al.鏈 2006; Taglvala,ela鏈 A.鏈 2012).

1.12 CDK12 inhibitors

I, va now well ea,ablvahed ,ha, dkaregula,von oa CDKa haa a role oa vn malvgnan, ,ranaaorma,von and varvoua CDK vnhvbv,ora have been developed and ,ea,ed aa po,en,val an,v-cancer ,herapeu,vca. However鏈 pan-apecvfc vnhvbv,ora ,ha, ,arge, varve,k oa CDKa鏈 repreaen,ed bk [avopvrvdol and roacovv,vne鏈 were no, aucceaaaul vn clvnvcal ,rvala. Developmen, oa aelec,vve CDK vnhvbv,ora ,arge,vng ckcle CDKa or ,ranacrvp,von-aaaocva,ed CDKa mak provvde be,,er ,herapeu,vc ou,come (Aaghar e, al.鏈 2015). An example oa auch an approach va developmen, oa CDK7 vnhvbv,or THZ1 (Kwva,áowaáv e, al.鏈 2014). I, haa recen,lk been ahown ,o lvmv, ,he ,ranacrvp,von oa aac,ora dependen, on auper-enhancera鏈 among ,hem MYC pro,o-oncogenea (Chvpumuro e, al.鏈 2014)鏈 and v, waa pre-clvnvcallk ,ea,ed aor ,rea,men, oa lung carcvnoma鏈 T-cell acu,e lkmphoblaa,vc leuáemva鏈 and ,rvple nega,vve breaa, cancer (Chrva,enaen e, al.鏈 2014; Kwva,áowaáv e, al.鏈 2014; Wang e, al.鏈 2015).

12 In recen, keara鏈 aeveral a,udvea have repor,ed on CDK12 rolea vn carcvnogeneava and CDK12 haa become a po,en,val ,herapeu,vc ,arge,. Conaequen,lk鏈 a apecvfc CDK12/CDK13 vnhvbv,or THZ531 waa developed (Zhang e, al.鏈 2016). THZ531 a,ruc,ure waa dervved arom prevvoualk deacrvbed CDK7 vnhvbv,or THZ1鏈 explov,vng a,ruc,ural avmvlarv,k be,ween ,heae ávnaaea. TZH1 vnhvbv,a CDK12 a, hvgher concen,ra,vona and v,a bvologvcal eaaec, could be par,lk aacrvbed ,o CDK12 vnhvbv,von. CDK12 and CDK13 cka,evnea 1039/1017 are loca,ed vn a avmvlar poav,von ,o CDK7 cka,evne 312鏈 whvch va covalen,lk bound bk THZ1. Baaed on crka,al a,ruc,ure鏈 ,he au,hora modvfed ,he molecule and acreened aor a compound ,ha, aelec,vvelk bvnda CDK12 and CDK13. THZ531 exhvbv,ed an,vprolvaera,vve and apop,o,vc eaaec, vn Juráa, T-cell acu,e lkmphoblaa,vc leuáemva cella and reduced CTD Ser2 phoaphorkla,von鏈 downregula,von oa DDR and auper-enhancer- aaaocva,ed genea (aa deacrvbed vn de,avl prevvoualk). Never,heleaa鏈 THZ531 ,arge,a bo,h CDK12 and CDK13 and con,rvbu,von oa ,he vndvvvdual ávnaaea ,o obaerved eaaec,a va no, defned (Zhang e, al.鏈 2016).

Dvnacvclvb (SCH 727965)鏈 a aelec,vve CDK1鏈 2鏈 5 and 9 vnhvbv,or鏈 waa a,udved aa a clvnvcal candvda,e baaed on v,a eafcack and ,olerabvlv,k. I, exhvbv,ed an,v- prolvaera,vve eaaec, vn cell lvnea and mouae xenograa,a (Parrk e, al.鏈 2010) and an,v,umor ac,vvv,k vn breaa, cancer pa,ven,a (Mv,a e, al.鏈 2014). A recen, a,udk dvacovered ,ha, dvnacvclvb po,en,lk vnhvbv,a CDK12 wv,h IC50 comparable ,o CDK9. I,a bvologvcal eaaec, could be aacrvbed ,o CDK12 vnhvbv,von鏈 avnce v, reduced CTD Ser2 phoaphorkla,von and HR gene expreaavon vn ,rvple-nega,vve breaa, cancer cell lvnea. Impor,an,lk鏈 dvnacvclvb aenav,ved BRCA wvld-,kpe and BRCA mu,a,ed cella ,o PARP vnhvbv,ora and reveraed bo,h prvmark and acquvred PARP vnhvbv,or reava,ance vn pa,ven,-dervved xenograa, modela (Johnaon e, al.鏈 2016).

In aummark鏈 ,wo CDK12 vnhvbv,ora have been recen,lk akn,heaved. Thek can be uaed ,o vnvea,vga,e ,he CDK12 bvologvcal aunc,von and ,hek repreaen, promvavng an,v-cancer druga鏈 aa avngle agen,a or vn combvna,von wv,h PARP1/2 vnhvbv,ora.

1.13 CHK1 inhibitors as anti-cancer drugs

The checápovn, ávnaae 1 (CHK1) va a aervne/,hreonvne ávnaae whvch par,vcvpa,ea vn DDR regula,von bk mul,vple mechanvama. In reaponae ,o DNA damage鏈 CHK1 va ac,vva,ed bk ATR phoaphorkla,von and promo,ea ,he cell ckcle arrea,.

13 Phoaphorkla,ed CHK1 vnac,vva,ea ,he CDC25 phoapha,aaea whvch are reaponavble aor ac,vva,von oa CDKa neceaaark aor G2/M ,ranav,von (McNeelk e, al.鏈 2014). In addv,von鏈 phoaphorkla,von bk CHK1 ac,vva,ea vmpor,an, DDR aac,ora BRCA2 and RAD51 (Bahaaav e, al.鏈 2008). CHK1 preven,a replvca,von a,reaa bk a,abvlvvng replvca,von aoráa and ,hua regula,ea replvca,von orvgvna frvng (McNeelk e, al.鏈 2014).

Al,hough exceaavve DNA damage and DDR deaec,a are drvvera oa malvgnan, ,ranaaorma,von鏈 druga ,ha, ,arge, apecvfc DDR componen,a can be explov,ed vn cancer ,herapk (Lord and Aahwor,h鏈 2012). CHK1 vnhvbv,ora combvned wv,h DNA- damagvng agen,a have been ,ea,ed aa an,v-,umor agen,a vn ovarvan and breaa, cancer modela (McNeelk e, al.鏈 2014; Thompaon and Eaa,man鏈 2013). Varvoua CHK1 vnhvbv,ora po,en,va,e ck,oa,a,vc eaaec, oa hkdroxkurea鏈 cvapla,vn鏈 ,opovaomeraae vnhvbv,ora (,opo,ecan鏈 vrvno,ecan)鏈 and an,vme,abolv,ea鏈 auch aa gemcv,abvne (Guv e, al.鏈 2011; Kvm e, al.鏈 2015; Ma e, al.鏈 2013; Mon,ano e, al.鏈 2013; Pere e, al.鏈 2006).

Analogvcallk ,o PARP vnhvbv,von鏈 CHK1 vnhvbv,von mvgh, be akn,he,vcallk le,hal wv,h loaa-oa aunc,von mu,a,vona or vnhvbv,von oa cer,avn cellular aac,ora鏈 whvch could be explov,ed vn ,he ,rea,men, oa cancer. For vna,ance鏈 cella defcven, vn IKKε and Fanconv’a Anemva DNA repavr aac,ora were aenav,vve ,o CHK1 deple,von and vnhvbv,von (Chen e, al.鏈 2009; Kvm e, al.鏈 2015).

14 2. Aims of the dissertation

Avma oa ,he dvaaer,a,von were ,o

1) de,ermvne ,he vmpac, oa CDK12 mu,a,vona aound vn HGSOC ,umor aamplea on CDK12 aunc,von and elucvda,e ,he CDK12 role vn HGSOC carcvnogeneava.

2) ,ea, ,he hkpo,heava ,ha, CDK12 and BRCA1 defcvenck aenav,ve cancer cella ,o CHK1 vnhvbv,ora.

15 3. Results

Thva chap,er va dvvvded vn,o ,hree aec,vona (3.1鏈 3.2鏈 3.3) ,o commen, on ,he reaul,a oa ,he ,hree publvahed a,udvea.

3.1 Ovarian carcinoma CDK12 mutations misregulate expression of DNA repair genes via de cient formation and function of the CDK12/Cyclin K complex

Several a,udvea have repor,ed CDK12 mu,a,vona or overexpreaavon vn varvoua cancer ,kpea. Among ,hem鏈 a comprehenavve a,udk on HGSOC bk TCGA vden,vfed CDK12 aa one oa onlk 9 recurren,lk mu,a,ed genea vn ,hva cancer ,kpe and hvghlvgh,ed CDK12 aa a ,umor auppreaaor candvda,e (Cancer Genome A,laa Reaearch Ne,worá鏈 2011). Never,heleaa鏈 ,he vmpac, oa CDK12 mu,a,vona on aorma,von oa CDK12/Ckclvn K complex鏈 expreaavon oa v,a ,arge, genea and ,hevr relevance ,o cancer remavned ,o be de,ermvned.

A large number oa HGSOC ,umor aamplea bear mu,a,ed or epvgenvcallk avlenced HR genea and exhvbv, genomvc vna,abvlv,k whvch underlvea ,umorvgeneava (Cancer Genome A,laa Reaearch Ne,worá鏈 2011). Svnce CDK12 va a ,ranacrvp,von elonga,von aac,or eaaen,val aor expreaavon oa DDR genea鏈 we hkpo,heavaed ,ha, CDK12 loaa-oa- aunc,von mu,a,vona mak compromvae HR whvch mak reaul, vn malvgnan, ,ranaaorma,von. Moreover鏈 v, haa been wvdelk accep,ed ,ha, mu,a,vona vn ,ranacrvp,von aac,ora or ,hevr mvaregula,von con,rvbu,e ,o ,umorvgeneava (Lee and Young鏈 2013).

The avm oa ,hva a,udk waa ,o elucvda,e ,he vmpac, oa HGSOC-rela,ed CDK12 mu,a,vona on CDK12 aunc,von vn ,ranacrvp,von and DDR and ,o clarvak ,he role oa mu,a,ed CDK12 vn ,umorvgeneava.

3.1.1 Generation of cell lines expressing mutated CDK12 forms

The TCGA a,udk on HGSOC provvded exome aequencea aor 316 ,umor aamplea and vden,vfed 9 valvda,ed CDK12 mu,a,vona. Fvve oa ,hem were nonaenae or vnaer,von/dele,von mu,a,vona and aour were mvaaenae mu,a,vona (Table 1). L122aa4*鏈 Q602* and W719* mu,an, aorma lacá ,he ávnaae domavn鏈 vnaer,von

16 mu,a,von E928aa27* leada ,o ,ermvna,von vnavde ,he CDK12 ávnaae domavn. All ,he aour mvaaenae mu,a,vona (R882L鏈 Y901C鏈 K975E and L996F) are av,ua,ed vnavde ,he ávnaae domavn. Svmvlarlk鏈 T1014 Q1016 dele,von va preaen, wv,hvn ,he ávnaae domavn (Cancer Genome A,laa Reaearch Ne,worá鏈 2011).

Table 1 CDK12 mu,a,vona vn HGSOC

Mutation (aa) Mutation (nt) Mutation ID Mutation Validation Type L122fs4* 363_364delCT COSM69050 Deletion - yes Frameshift S363* 1088C>G COSM1324005 Nonsense no Q602* 1804C>T COSM78919 Nonsense yes P653L 1958C>T COSM1324004 Missense no W719* 2156G>A COSM118018 Nonsense yes S785fs4* 2351_2352insA COSM1324003 Insertion - no Frameshift R882L 2645G>T COSM70301 Missense yes Y901C 2702A>G COSM74101 Missense yes E928fs27* 2778_2779insG COSM69187 Insertion - yes Frameshift K975E 2923A>G COSM70302 Missense yes R983_E984>Q 2948_2950delGAG COSM1324002 Deletion - no In frame L996F 2988A>T COSM70303 Missense yes T1014_Q1016del 3036_3044delACAGACCCT COSM69117 Deletion - yes In frame A1174G 3521C>G COSM94100 Missense yes / 1028_1046+14del 33 COSM1324006 Intronic deletion no

For our a,udk鏈 we have choaen W719* and E928aa27* ,o repreaen, dele,von mu,an,a and all aour mvaaenae mu,a,vona (Fvgure 3a).

In order ,o a,udk ,he avgnvfcance prevvoualk deacrvbed CDK12 mu,a,vona鏈 we ea,ablvahed ,hree ae,a oa cell lvnea whvch expreaa Flag-,agged CDK12 con,avnvng ,he vndvvvdual mu,a,vona. Fvra,鏈 we vnaer,ed ,heae mu,a,vona vn,o CDK12-Flag plaamvd uavng ,he Sv,e-Dvrec,ed Mu,ageneava Kv,. Nex,鏈 we vnaer,ed an addv,vonal mu,a,von whvch conaera reava,ance ,o a defned avRNA agavna, CDK12 ,o all plaamvda con,avnvng ,he cancer-rela,ed mu,a,vona. Fvnallk鏈 we ,ranaaec,ed ,hoae plamvda ,o HCT116 cella and aelec,ed aor ,he a,able ,ranaaorman,a. In addv,von鏈 we emploked ,he Flvp-vn T-rex aka,em ,o genera,e a,able cell lvnea expreaavng mu,a,ed CDK12 vn HCT116 and HEK293 cella. The Flvp-vn T-rex aka,em enablea vn,egra,von oa vndvvvdual CDK12 expreaavon vec,ora ,o ,he defned locua鏈 whvch reaul,a vn comparable expreaavon oa CDK12 mu,an,a vn dvaaeren, cell lvnea.

17 3.1.2 Defective interaction between Cyclin K and CDK12 is the pre- dominant consequence of CDK12 mutations in HGSOC

A common aea,ure oa all CDKa va ,ha, ckclvn bvndvng vn requvred aor ,hevr aull ac,vva,von. Aa,er ,he ckclvn va bound鏈 ,he ca,alk,vc clea, va reorven,ed and aubaequen,lk ,hreonvne wv,hvn ,he T-loop va phoaphorkla,ed bk ,he CDK ac,vva,vng ávnaae (Malumbrea鏈 2014).

Fvra,鏈 we wan,ed ,o de,ermvne ,he vmpac, oa CDK12 mu,a,vona on CDK12 abvlv,k ,o bvnd Ckclvn K and aorm a ca,alk,vcallk ac,vve complex (Fvgure 3c). In order ,o do ,hva鏈 we uaed a ae, oa HEK293 cell lvnea expreaavng Flag-,agged CDK12 bearvng vndvvvdual cancer-rela,ed mu,a,vona. We vmmunoprecvpv,a,ed wvld ,kpe or mu,a,ed CDK12 arom ,he whole cell lkaa,ea uavng Flag an,vbodk and de,ermvned Ckclvn K levela vn ,hoae purvfed lkaa,ea bk wea,ern blo,,vng.

Aa expec,ed鏈 we confrmed ,he vn,erac,von oa wvld ,kpe CDK12 wv,h Ckclvn K. In con,raa,鏈 dele,von mu,an,a W719* and E928aa27* were no, able ,o bvnd Ckclvn K鏈 lváelk due ,o ,he mvaavng ávnaae domavn. R882L and K975E mu,an,a vn,erac,ed wv,h Ckclvn K鏈 whvle Y901C and L996F mu,an,a ahowed reduced or no abvlv,k ,o aorm ,he CDK12/Ckclvn K complex reapec,vvelk.

Nex,鏈 we projec,ed ,he mu,a,ed reavduea on prevvoualk repor,ed CDK12 crka,al a,ruc,ure ,o ge, a,ruc,ural vnavgh, ,o ,he poaavble role oa ,heae mu,a,vona (Böaáen e, al.鏈 2014). All ,heae mu,a,vona were loca,ed wv,hvn ,he C-,ermvnal lobe oa ,he CDK12 ávnaae domavn鏈 whvle Ckclvn K vn,erac,a wv,h ,he N-,ermvnal lobe. Thva obaerva,von auggea,a ,ha, alloa,ervc mechanvama are reaponavble aor ,he eaaec, oa ,heae mu,a,vona on ,he CDK12/Ckclvn K complex (fgure 3b).

18 Figure 3 CDK12 mu,a,vona vn HGS-OvCa abroga,e ,he ac,vvv,k oa Cdá12 predomvnan,lk bk vmpavrvng ,he vn,erac,von be,ween Cdá12 and CkcK. (A) Schema,vc depvc,von oa ,he wvld-,kpe and mu,an, Cdá12 pro,evna con,avnvng vndvvvdual CDK12 mu,a,vona analked vn ,hva a,udk. Hvghlk a,ruc,ured ávnaae domavn (KD; red)鏈 argvnvne/aervne rvch regvon (RS; green) and ,wo regvona wv,h prolvne-rvch mo,vaa (PRM1 and PRM2; blue) are depvc,ed. The ruler on ,op vndvca,ea ,he leng,h oa Cdá12 pro,evn vn amvno acvda. Ver,vcal lvnea deno,e av,ea oa vndvvvdual mvaaenae and vnaer,von mu,a,vona. Fvnallk鏈 ‘aa’ a,anda aor a arame-ahva, mu,a,von and ,he aaaocva,ed number vndvca,ea ,he number oa al,ered amvno acvda a, ,he C-,ermvnua oa mu,an, Cdá12 pro,evn (do,,ed). (B) Overall a,ruc,ure oa human Cdá12/CkcK and poav,vona oa ,he mu,a,vona. Cdá12 va ahown aa car,oon repreaen,a,von vn blue and CkcK aa auraace repreaen,a,von vn grek. CDK12 mvaaenae鏈 vnaer,von and vn,ernal dele,von mu,a,vona are loca,ed vn ,he C-,ermvnal lobe oa ,he KD. The mu,a,ed amvno acvd reavduea are hvghlvgh,ed vn red. pT893 va hvghlvgh,ed vn orange. (C) Eaaec,a oa ,he CDK12 mu,a,vona on ,he vn,erac,von be,ween Cdá12 and CkcK. The vndvca,ed wvld-,kpe and mu,an, FLAG epv,ope- ,agged Cdá12 pro,evna (Cdá12-F) were vmmuno-purvfed arom whole cell ex,rac,a (WCEa) oa ,he vndvvvdual HEK 293 Flp-In T-Rex cell lvnea uavng FLAG-M2 agaroae (FLAG IP) and examvned aor ,hevr vn,erac,von wv,h endogenoua CkcK. Levela oa Cdá12-F and CkcK pro,evna vn WCEa (INPUT鏈 5% oa WCEa; ,op) and IPa (FLAG IP; bo,,om) were de,ec,ed bk Wea,ern blo,,vng uavng FLAG and CkcK an,vbodvea. (D) CDK12 mu,a,vona abroga,e ,he ávnaae ac,vvv,k oa Cdá12. The vndvca,ed wvld- ,kpe and mu,an, Cdá12-F pro,evna were vmmuno-purvfed (IP) aa vn panel C and ,he complexea were examvned aor ,hevr ávnaae ac,vvv,k bk vn vv,ro ávnaae aaaak (IVKA) ,oward ,he recombvnan, GST-CTD. Levela oa Ser2-P GST-CTD vaoaorma and vnpu, GST-CTD (30%) were de,ec,ed bk Wea,ern blo,,vng uavng Ser2-P-apecvfc RNAPII and GST an,vbodvea.

19 3.1.3 CDK12 mutations in HGSOC abrogate the kinase activity of Cdk12 I, can expec,ed ,ha, CDK12 mu,a,vona ,ha, dvarup, Ckclvn K vn,erac,von wvll cauae ,he loaa oa ,he CDK12 ávnaae ac,vvv,k ,owarda RNA Poll CTD Ser2. Thua鏈 we emploked in vitro ávnaae aaaaka ,o explore ,he vmpac, oa CDK12 mu,a,vona on v,a ca,alk,vc ac,vvv,k. Svmvlarlk ,o ,he prevvoua expervmen,鏈 we vmmunoprecvpv,a,ed wvld ,kpe and mu,an, CDK12-Flag pro,evna arom ,he whole cell lkaa,ea oa Flvp-vn HEK293 cell lvnea and uaed ,hem vn in vitro ávnaae aaaaka emplokvng recombvnan, GST-CTD pro,evn aa a auba,ra,e. Subaequen,lk鏈 we de,ermvned levela oa phoaphorkla,ed Ser2 uavng a apecvfc an,vbodk bk wea,ern blo,,vng (Fvgure 3d).

Wvld ,kpe CDK12 waa able ,o phoaphorkla,e ,he GST-CTD eaaec,vvelk. In con,raa,鏈 majorv,k oa mu,an, CDK12 aorma aavled ,o phoaphorkla,e ,he GST-CTD鏈 wv,h excep,von oa K975E mu,a,ed pro,evn.

In aummark鏈 CDK12 mu,a,vona ,ha, dvaable aorma,von oa CDK12/Ckclvn K complex alao cauae ,he loaa oa ,he ávnaae ac,vvv,k. R882L mu,a,von dvarup,ed ,he CDK12 ávnaae ac,vvv,k wv,hou, aaaec,vng ,he Ckclvn K vn,erac,von.

3.1.4 CDK12 mutations in HGSOC decrease transcriptional activation by CDK12 Several a,udvea have ea,ablvahed ,ha, CDK12/Ckclvn K complex dvrec,lk par,vcvpa,ea vn ,ranacrvp,vonal regula,von bk phoaphorkla,vng CTD Ser2. I, haa been prevvoualk ahown ,ha, CDK12 va able ,o a,vmula,e ,ranacrvp,vonal ac,vva,von鏈 meaaured bk RNA ,e,hervng gene repor,er aaaak (Blaeá e, al.鏈 2011). Svnce HGSOC-rela,ed CDK12 mu,a,vona lead ,o a compromvaed CDK12 ávnaae ac,vvv,k鏈 we nex, wan,ed ,o de,ermvne whe,her ,heae mu,a,vona have vmpac, on CDK12 abvlv,k ,o ac,vva,e ,ranacrvp,von.

To addreaa ,hva quea,von鏈 we emploked ,he RNA ,e,hervng gene repor,er aaaak whvch va compoaed oa SLIIB-CAT repor,er gene and Rev-CDK auavon gene. Rev a,ronglk vn,erac,a wv,h a naacen, ,ranacrvp, oa SLIIB RNA elemen, and ,e,hera Rev- CDK auavon pro,evn ,o ,he ,ranacrvbvng RNA Pol II鏈 whvch reaul,a vn ,ranacrvp,von oa CAT repor,er gene (fgure 4a) (Coller and Wvcáena鏈 2007; Tvlek e, al.鏈 1992).

In lvne wv,h a prevvoua repor,鏈 wvld ,kpe CDK12 waa able ,o ac,vva,e ,ranacrvp,von (fgure 4b). In con,raa,鏈 all Rev-CDK12 mu,a,ed aorma ahowed decreaaed abvlv,k ,o ac,vva,e ,ranacrvp,von鏈 compared ,o wvld ,kpe. To de,ermvne whe,her CDK12

20 ávnaae ac,vvv,k va crucval ,o ac,vva,e ,ranacrvp,von鏈 we peraormed ,e,hervng gene repor,er aaaak wv,h a ávnaae dead CDK12 mu,an,. Fvra, we genera,ed ,he ávnaae dead CDK12 bk mu,a,vng v,a aapar,vc reavdue 877 wv,hvn ,he DFG mo,va ,o aaparagvne (D877N) and confrmed v,a abolvahed in vitro ac,vvv,k (da,a no, ahown). In ,e,hervng repor,er aaaak鏈 CDK12 bearvng D877N mu,a,von ahowed decreaaed abvlv,k ,o ac,vva,e ,ranacrvp,von鏈 comparable ,o ,he cancer-rela,ed mu,an,a.

In aummark鏈 ,heae reaul,a ahow ,ha, CDK12 mu,a,vona compromvae ,he abvlv,k oa CDK12 ,o ac,vva,e ,ranacrvp,von. In cancer cella鏈 v, mak reaul, vn decreaaed expreaavon oa CDK12 dependen, genea鏈 among ,hem vmpor,an, DDR regula,ora and ,o genomvc vna,abvlv,k. We aound ou, ,ha, CDK12 ávnaae ac,vvv,k va reaponavble aor lower ac,vva,von oa ,he repor,er gene鏈 never,heleaa ,he K975E mu,an,鏈 whvch dvd no, aaaec, ,he CDK12 ac,vvv,k鏈 alao aavled ,o ac,vva,e ,he repor,er gene expreaavon. I, mak auggea, ,ha, K975E mu,an, vmpavra an addv,vonal mechanvam whvch va neceaaark aor ,he ,ranacrvp,vonal ac,vva,von bk CDK12. Moreover鏈 mu,a,ed CDK12 aorma were able ,o a,vmula,e ,ranacrvp,von ,o aome level鏈 whvch mak vndvca,e ,ha, o,her CDK12 aunc,vona ,han ,he ávnaae ac,vvv,k mak a,vmula,e ,he gene expreaavon.

21 Figure 4 CDK12 mu,a,vona vn HGS-OvCa decreaae ,ranacrvp,vonal ac,vva,von bk Cdá12. (A) Schema,vc depvc,von oa ,he he,erologoua RNA ,e,hervng aaaak. (B) CDK12 HGS-OvCa mu,a,vona compromvae a,vmula,von oa ,ranacrvp,von bk Rev-Cdá12. HEK 293 cella were co,ranaaec,ed wv,h pSLIIB-CAT repor,er gene and plaamvda encodvng ,he pro,evna vndvca,ed below ,he graph. Tranacrvp,vonal ac,vvv,vea oa Cdá12-F (whv,e bar 1)鏈 Rev (whv,e bar 2)鏈 ,he mu,an, Rev-Cdá12 chvmeraa (red bara) and ca,alk,vcallk dead Rev-Cdá12 D887N chvmera (green bar) are repreaen,ed aa CAT ac,vvv,vea rela,vve ,o ,he ac,vvv,k oa wvld-,kpe Rev-Cdá12 chvmera (blue bar)鏈 whvch waa ae, ,o 100%. Reaul,a are preaen,ed aa ,he mean ± SD. Levela oa ,he Rev-Cdá12 chvmeraa and endogenoua Cdá12 pro,evn are ahown below ,he graph and were de,ec,ed bk Wea,ern blo,,vng uavng Cdá12 an,vbodk.

22 3.1.5 HGSOC patient samples with CDK12 mutations exhibit downregulation of genes of the HR repair pathway

I, haa been prevvoualk deacrvbed ,ha, CDK12 va neceaaark aor expreaavon oa DDR- aaaocva,ed genea鏈 vncludvng BRCA1鏈 ATR鏈 FANCD2 and FANCI. Never,heleaa鏈 ,hoae fndvnga were baaed on CDK12 avlencvng bk avRNAa vn cell lvnea and mak be aomewha, lvmv,ed (Blaeá e, al.鏈 2011). To addreaa ,he quea,von va DDR genea are downregula,ed vn CDK12-mu,a,ed ,umora鏈 we a,a,va,vcallk analkaed publvclk avavlable da,a arom ,he TCGA a,udk (Cancer Genome A,laa Reaearch Ne,worá鏈 2011). Taávng vn,o accoun, ,ha, approxvma,elk 50% oa HGSOC aamplea ahow HR deaec,a and CDK12 va neceaaark aor expreaavon oa áek HR genea鏈 we hkpo,heaved ,ha, CDK12 mu,a,vona mvgh, lead ,o decreaaed expreaavon oa HR genea vn ,umor aamplea.

We compared mRNA expreaavon levela oa vndvca,ed genea be,ween CDK12 wvld ,kpe ,umor aamplea wv,h ,he aamplea bearvng one oa ,he nvne valvda,ed CDK12 mu,a,vona (Fvgure 5a鏈b). We were able ,o vden,vak nvne DDR genea whvch are avgnvfcan,lk downregula,ed vn CDK12-mu,a,ed aamplea. Onlk ,wo oa ,hem鏈 ATR and FANCI鏈 were prevvoualk deacrvbed aa CDK12 dependen, genea (Blaeá e, al.鏈 2011). Four downregula,ed genea鏈 ATM鏈 CHK1鏈 MDC1 and RAD51D鏈 are aaaocva,ed wv,h HR鏈 ,hree genea (NEK9鏈 ORC3L and TERF2) are unrela,ed ,o HR and vnvolved vn dvaaeren, DNA repavr pa,hwaka (fgure 5a鏈b).

Surprvavnglk鏈 expreaavon oa an vmpor,an, HR regula,or BRCA1 waa no, aaaec,ed vn ,he aamplea bearvng CDK12 mu,a,vona鏈 deapv,e ,he aac, ,ha, BRCA1 va conavdered ,o be a CDK12-dependen, gene (fgure 5c). Svmvlarlk鏈 level oa BRCA2 waa vndependen, oa ,he CDK12 a,a,ua. We were unable ,o analkae expreaavon da,a aor prevvoualk deacrvbed CDK12-dependen, DDR genea FANCD2 and MMS22L becauae ,heae probea were no, vncluded vn ,he TCGA a,udk.

23 Figure 5 The crucval DDR genea are downregula,ed vn HGS-OvCa pa,ven, aamplea wv,h mu,a,vona vn CDK12. (A鏈 B and C) Grapha ahow comparvaona oa rela,vve expreaavon levela be,ween ,he HGS-OvCa aamplea wv,h ,he wvld-,kpe or mu,a,ed CDK12. The vden,v,k oa genea va vndvca,ed on ,op oa each graph. The da,a waa genera,ed uavng ,he aollowvng mvcroarrak probea: ATM (212672 a,)鏈 ATR (209903 a a,)鏈 CHEK1 (205394 a,)鏈 FANCI (213007 a,)鏈 MDC1 (203062 a a,)鏈 RAD51D (209965 a a,)鏈 NEK9 (212299 a,)鏈 ORCL3 (210028 a a,)鏈 TERF2 (203611 a,)鏈 BRCA1 (211851 x a,)鏈 BRCA2 (208368 a a,). Whereaa aamplea wv,h ,he wvld-,kpe CDK12 are plo,,ed aa blacá ,rvanglea鏈 ,hoae con,avnvng vndvvvdual mvaaenae or nonaenae/vndel CDK12 mu,a,vona are depvc,ed aa colored cvrclea or aquarea鏈 reapec,vvelk鏈 aa vndvca,ed bk ,he legend vn ,he ,op rvgh, corner. Reaul,a are preaen,ed aa mean (red lvne) wv,h a,andard error oa ,he mean (SEM) (blacá whvaáera). P-valuea are gvven nex, ,o red aa,ervaáa (*) and ,he number oa aa,ervaáa vndvca,ea ,he degree oa avgnvfcance aa aollowa: * = P ≤ 0.05; ** = P ≤ 0.01; *** = P ≤ 0.001. Panela A and C ahow genea rela,ed ,o ,he HR pa,hwak鏈 whvle o,her aaaec,ed DDR genea are ahown vn panel B. (D and E) Deple,von oa Cdá12 decreaaea ,he mRNA levela oa HR genea vn Caov-3 cella. Rela,vve mRNA levela oa genea vndvca,ed below ,he bara were de,ermvned bk RT-qPCR uavng ,o,al RNA aamplea vaola,ed arom Caov-3 cella ,rea,ed wv,h ,he con,rol (blue bara) or Cdá12 #1 avRNA (red bara). The error bara repreaen, ,he mean ± SD.

24 3.1.6 CDK12 depletion decreases expression of HR genes

Emplokvng mRNA expreaavon analkava鏈 we vden,vfed a new ae, oa genea whvch are downregula,ed vn ,umor aamplea bearvng CDK12 mu,a,vona. Nex, we wan,ed ,o aaaeaa va ,heae genea are CDK12-dependen, vn cell lvne modela. We downregula,ed CDK12 bk avRNA vn an ovarvan cancer cell lvne Caov-3 and colon cancer HCT116 and quan,vfed mRNA levela oa vndvca,ed genea bk RT-qPCR. Aa expec,ed鏈 CDK12 avlencvng led ,o decreaaed expreaavon oa BRCA1鏈 FANCI鏈 FANCD2 and ATR鏈 whvch are genea ,ha, were prevvoualk deacrvbed aa CDK12-dependen, (fgure 5d and 6a) (Blaeá e, al.鏈 2011). Fur,hermore鏈 upon CDK12 ánocádown鏈 we obaerved decreaaed expreaavon oa DDR regula,ork genea ATM鏈 CHK1鏈 MDC1 and RAD51D whvch have no, been prevvoualk deacrvbed aa CDK12-dependen, (fgure 5e and 6b). In addv,von鏈 deple,von oa CDK12 and Ckclvn K bu, no, CDK13 led ,o alvgh,lk decreaaed pro,evn levela oa ATM鏈 CHK1 and Rad51D vn HCT116 cella (fgure 6c).

Overall鏈 our reaul,a vndvca,e ,ha, CDK12 mu,a,vona compromvae v,a ávnaae ac,vvv,k whvch leada ,o decreaaed expreaavon oa aeveral DDR genea鏈 vncludvng vmpor,an, HR regula,ork aac,ora.

25 Figure 6 Deple,von oa Cdá12 decreaaea ,he mRNA and pro,evn levela oa crv,vcal HR genea vn HCT116 cella. (A and B) Rela,vve mRNA levela oa genea vndvca,ed below ,he bara were de,ermvned bk RT-qPCR uavng ,o,al RNA aamplea vaola,ed arom HCT116 cella ,rea,ed wv,h ,he con,rol (blue bara) or Cdá12 #1 avRNA (red bara). The error bara repreaen, ,he mean ± SD. (C) Levela oa pro,evna vndvca,ed on lea, were de,ermvned bk Wea,ern blo,,vng uavng whole cell lkaa,ea oa HCT116 cella ,rea,ed wv,h ,he con,rol鏈 Cdá12 #1鏈 CkcK or Cdá13 avRNA.

26 3.1.7 The Cdk12/CycK complex occupies HR genes and promotes Ser2 phosphorylation of RNAPII

In prevvoua expervmen,a鏈 we vden,vfed a new ae, oa genea whoae expreaavon dependa on CDK12. Conavdervng ,he aac, ,ha, CDK12 va a regula,or oa ,ranacrvp,von鏈 we nex, wan,ed ,o de,ermvne va ,heae genea repreaen, dvrec, CDK12 ,ranacrvp,vonal ,arge,a and va CDK12 promo,ea RNA Pol II Ser2 phoaphorkla,von on ,heae genea.

To addreaa ,hva quea,von鏈 we peraormed ChIP aollowed bk RT-PCR wv,h CDK12鏈 ,o,al/Ser5-P RNAPII and Ser2-P RNA Pol II an,vbodvea on ATM鏈 CHK1鏈 MDC1 and RAD51D genea. ChIPa were peraormed vn HeLa DR-GFP cella ,ranaaec,ed wv,h con,rol or CDK12 avRNAa uavng prvmera covervng ,he ,ranacrvp,von a,ar, av,e (TSS)鏈 ,he vn,ervor oa ,he open readvng arame (ORF) and ,he vvcvnv,k oa ,he polkadenkla,von av,e (pA) oa each gene.

CDK12 occupved all ,hree aaaaked regvona oa genea (Fvgure 8a). CDK12 avgnal waa avgnvfcan,lk enrvched a, all aour aaaaked genea compared ,o vn,ergenvc regvona. Aa expec,ed鏈 CDK12 avlencvng reaul,ed vn v,a reduced occupanck a, genea鏈 whvch confrma apecvfcv,k oa ,he CDK12 avgnala.

CDK12 deple,von led ,o reduced Ser5/,o,al occupanck a, ,he TSS oa all genea (fgure 8b). A poaavble explana,von aor ,hva eaaec, va ,ha, CDK12 mak be capable oa CTD Ser5 phoaphorkla,von鏈 whvch haa been repor,ed in vitro (Böaáen e, al.鏈 2014). Impor,an,lk鏈 aa,er CDK12 ánocádown鏈 Ser2-P occupanck waa reduced a, all regvona (fgure 8c). Theae reaul,a auggea, ,ha, CDK12 va preaen, a, ,he aaaaked genea durvng ,he ,ranacrvp,von鏈 v, phoaphorkla,ea Ser2 oa a ,ranacrvbvng Pol II and ,hereaore DDR genea are dvrec, CDK12 ,ranacrvp,vonal ,arge,a.

27 Figure 8 CDK12/CkcK va preaen, a, ,he novel HR genea ,o promo,e phoaphorkla,von oa ,he CTD oa RNAPII a, Ser2. (A鏈 B and C) Con,rol (blue bara) or CDK12 ánocádown HeLa DR-GFP cella (red bara) were aubjec,ed ,o ChIP-qPCR analkava ,o de,ermvne ,he levela oa Cdá12 (panel A)鏈 ,o,al/Ser5- P RNAPII (panel B) and Ser2-P RNAPII (panel C) occupanck a, RAD51D vn,ergenvc regvon (IR) and ,hree gene-apecvfc regvona aa vndvca,ed below ,he grapha. The vden,v,k oa genea analked va vndvca,ed on ,op oa each graph. Levela oa IgG avgnala a, IR and gene regvona are preaen,ed aa grak bara. Where ,heae levela were avgnvfcan,lk lower ,han ,hoae ob,avned wv,h apecvfc an,vbodk鏈 ,he error bara were omv,,ed aor avmplvcv,k reaaona. Reaul,a are preaen,ed aa percen, oa vnpu, DNA and plo,,ed aa ,he mean ± SD.

28 3.1.8 CDK12 mutations in HGS-OvCa disable the stimulatory role of the Cdk12/CycK complex in the repair of DSBs by HR There va a,rong evvdence ,ha, CDK12 va neceaaark aor expreaavon oa mul,vple DDR genea鏈 par,vcularlk genea vnvolved vn HR repavr pa,hwak. Nex, we aaáed ,he quea,von whe,her CDK12 defcvenck mak compromvae ,he abvlv,k oa cella ,o repavr a double a,rand breaá bk HR. We emploked ,he DR-GFP repor,er aaaak鏈 whvch va commonlk uaed ,o de,ermvne HR level vn cella (Pverce e, al.鏈 1999). Hela DR-GFP cella con,avn an HR repor,er caaae,,e whvch va compoaed oa ,wo dvaaeren,vallk mu,a,ed enhanced green [uoreacen, pro,evn (EGFP) genea. SceGFP con,avna a recognv,von av,e aor ScdI endonucleaae鏈 vGFP va a 5′ and 3′-,runca,ed EGFP aragmen,. None oa ,heae genea are capable oa producvng ,he aunc,vonal EGFP pro,evn. Aa,er SceI va expreaaed鏈 v, vnducea a double a,rand breaá ,o ,he apecvfc av,e wv,hvn ,he SceGFP鏈 whvch can be repavred bk HR uavng vGFP aa a ,empla,e. Ia ,he repavr va aucceaaaul鏈 a aunc,vonal GFP gene va genera,ed and [uoreacence avgnal mak be analked bk [ow ck,ome,rk (fgure 9a).

Aa,er CDK12 downregula,von bk avRNA鏈 we de,ec,ed 3鏈5 aold decreaae vn HR arequenck compared ,o ,he con,rol avRNA. Nex,鏈 we wan,ed ,o de,ermvne ,he vmpac, oa vndvvvdual cancer-rela,ed CDK12 mu,a,vona on HR eafcvenck. In order ,o do ,hva鏈 we peraormed a ,ranaven, expreaavon oa wvld ,kpe or mu,a,ed CDK12-Flag鏈 whvch va vnaenav,vve ,o CDK12 avRNA. Aa,er avRNA ,rea,men,鏈 endogenoua CDK12 waa deple,ed and cella relved on ,he overexpreaaed CDK12鏈 ev,her wvld ,kpe or bearvng one oa ,he depvc,ed mu,a,vona (fgure 9b). Expreaavon oa wvld ,kpe CDK12 waa able ,o reacue 2鏈3 aold HR arequenck over ,he decreaaed level cauaed bk CDK12 deple,von.

However鏈 majorv,k oa CDK12 mu,a,ed aorma dvd no, rea,ore ,he EGFP avgnal鏈 whvch vndvca,ea a compromvaed abvlv,k oa cella ,o repavr double a,rand breaáa bk HR. In,erea,vnglk鏈 one oa ,he CDK12 mu,an, aorma鏈 K975E鏈 dvd no, avgnvfcan,lk reduce ,he HR arequenck compared ,o wvld ,kpe. Thva reaul, auggea,a ,ha, ,here va an aaaocva,von be,ween CDK12 ávnaae ac,vvv,k and abvlv,k oa cella ,o execu,e HR.

Overall鏈 ,heae reaul,a vndvca,e ,ha, CDK12-defcven, cella have vmpavred HR鏈 whvch va conava,en, wv,h prevvoua obaerva,vona (Bajramv e, al.鏈 2014). I, can be auggea,ed ,ha, CDK12 mu,a,vona aound vn HGSOC aamplea con,rvbu,e ,o ,he genomvc vna,abvlv,k vn ,heae ,umora.

29 Figure 9 CDK12 mutations in HGS-OvCa abrogate the ability of Cdk12/CycK to stimulate the repair of DNA double-strand breaks by HR. (A) Schematic representation of the DR-GFP recombination substrate. The defective EGFP genes of the cassette, separated by 3.7 kb, are shown (top). The first one (SceGFP) contains an I-SceI endonuclease site and an in-frame termination codon (white vertical line), while the second one (iGFP) is an internal EGFP fragment, yielding GFP− cells. Induction of DSB by I-SceI triggers the repair of SceGFP by HR using the iGFP sequence as a donor DNA, resulting in wild-type EGFP (green rectangle) and GFP+ cells (bottom). (B) The mutant Cdk12 proteins fail to promote the repair of the DSB by HR. HeLa DR-GFP cells were treated with the control or Cdk12 #2 siRNA and transfected with the I-SceI expression plasmid together with plasmids encoding the wild-type (blue bar) or mutant (red bars) Cdk12-F proteins as indicated below the graph. The HR frequency for each experimental condition is represented as the frequency relative to the one reached by the I-SceI expression in the control siRNA-treated cells, which was set to 100% (green bar). Results are presented as the mean ± SD. Levels of the endogenous Cdk12 and Cdk12-F proteins are shown below the graph and were detected by Western blotting using Cdk12 antibody.

30 3.1.9 Discussion A purpoae oa our a,udk waa ,o de,ermvne ,he vmpac, oa CDK12 mu,a,vona aound vn HGSOC aamplea on CDK12 aunc,von and ,o clarvak ,he role oa CDK12 vn carcvnogeneava. Baaed on bvochemvcal鏈 gene expreaavon and aunc,vonal aaaaka鏈 we ahowed ,ha, HGCSOC-rela,ed CDK12 mu,a,vona are loaa-oa aunc,von mu,a,vona ,ha, compromvae expreaavon oa CDK12-dependen, genea vnvolved vn HR and ,hua mak con,vbu,e ,o carcvnogeneava. Thereaore our a,udk auppor,a ,he hkpo,heava ,ha, CDK12 va a po,en,val ,umor auppreaaor vn HGSOC.

Majorv,k oa ,heae mu,a,vona dvaabled aorma,von oa ,he CDK12/Ckclvn K complex. We demona,ra,ed ,ha, CDK12 mu,a,vona dvaable v,a ávnaae ac,vvv,k鏈 whvch va vn lvne wv,h a prevvoua repor, (Joahv e, al.鏈 2014).

Emplokvng mRNA expreaavon a,a,va,vcal analkava and RT-PCR鏈 we vden,vfed new CDK12 dependen, genea whvch are vmpor,an, regula,ora oa ,he DDR ne,worá. Uavng ChIP expervmen,a鏈 we provvded ,he evvdence ,ha, ,heae genea are dvrec, CDK12 ,ranacrvp,vonal ,arge,a ,ha, requvre CTD Ser2 phoaphorkla,von bk CDK12 durvng ,he proceaa oa ,hevr ,ranacrvp,von. Svmvlar reaul,a were ob,avned vn a recen, a,udk whvch uaed a ChIP-aeq analkava aa,er CDK12 vnhvbv,von.

Never,heleaa鏈 ,he precvae mechanvam oa how CDK12 con,rola gene expreaavon remavna ,o be de,ermvned. Defcven, Ser2 phoaphorkla,von bk CDK12 mak vmpavr recruv,men, oa Ser2–vn,erac,vng aac,ora whvch are vmpor,an, aor ,ranacrvp,von ,ermvna,von and aplvcvng. In addv,von鏈 v, can be expec,ed ,ha, ,here are addv,vonal CDK12 auba,ra,ea ,o CTD and o,her mechanvama bk whvch CDK12 vn[uencea gene expreaavon. Moreover鏈 CDK12 mu,a,vona mak dvarup, v,a vn,erac,vona wv,h varvoua pro,evna鏈 whvch can alao lead ,o dkaregula,von oa ,he gene expreaavon.

The TCGA a,udk on HGSOC repor,ed ,ha, 50% oa ,heae ,umora dvaplak deaec,vve HR due ,o gene,vc and epvgene,vc deaec,a vn ,he HR repavr pa,hwak. Deaec,vve HR leada ,o genomvc vna,abvlv,k whvch con,rvbu,ea ,o malvgnan, ,ranaaorma,von. In our a,udk鏈 we provvded evvdence ,ha, aunc,vonal CDK12 va neceaaark aor eaaec,vve DNA repavr bk HR. We aound ,ha, CDK12 mu,a,vona lvmv, ,he expreaavon oa ATM鏈 CHK1鏈 MDC1 and RAD51D genea鏈 ,ha, are vnvolved vn HR. Surprvavnglk鏈 BRCA1鏈 prevvoualk deacrvbed aa a CDK12-dependen, gene鏈 waa no, downregula,ed vn CDK12-mu,a,ed ,umora. Never,heleaa鏈 vnac,vva,vng mu,a,vona vn BRCA1 and CDK12 were mu,uallk excluavve鏈 whvch vndvca,ea ,ha, ,heae aac,ora are par, oa one regula,ork pa,hwak.

31 Several a,udvea have repor,ed ,ha, CDK12 defcvenck conaera aenav,vvv,k ,o PARP vnhvbv,ora (Bajramv e, al.鏈 2014; Joahv e, al.鏈 2014). Our a,udk auppor,a ,hva fndvng鏈 ,aávng vn,o accoun, ,ha, HR defcvenck va de,ermvnan, aor aenav,vvv,k oa cella ,o PARP vnhvbv,ora. Theae a,udvea open ,he poaavbvlv,k ,ha, ,umora wv,h loaa-oa- aunc,von CDK12 mu,a,vona mak be aenav,vve ,o PARP vnhvbv,ora ,herapk. Impor,an,lk鏈 a recen, evvdence auggea,a ,ha, combvna,von oa CDK12 vnhvbv,von mak reverae reava,ance oa cancer cella ,o PARP vnhvbv,ora (Johnaon e, al.鏈 2016).

32 3.2 BRCA1 or CDK12 loss sensitizes cells to CHK1 inhibitors

Ex,enavve reaearch haa ea,ablvahed ,ha, defcven, DDR and conaequen, genomvc vna,abvlv,k underlve carcvnogeneava (Lord and Aahwor,h鏈 2012; Pearl e, al.鏈 2015). CHK1 va one oa ,he vmpor,an, aac,ora vnvolved vn DDR ne,worá. I, va a aervne/,hreonvne ávnaae whvch va able ,o blocá ,he cell ckcle progreaavon vn reaponae ,o exceaavve DNA damage鏈 v, phoaphorkla,ea and ,hua ac,vva,ea DDR aac,ora BRCA2 and RAD51 and v, can preven, replvca,von a,reaa bk regula,von oa replvca,von orvgvna frvng (McNeelk e, al.鏈 2014; Sørenaen and Skljuåaen鏈 2012; Thompaon and Eaa,man鏈 2013). Conaequen,lk鏈 CHK1 repreaen,a a promvavng ,herapeu,vc ,arge, vn a varve,k oa cancer ,kpea. Several CHK1 vnhvbv,ora have been ,ea,ed vn clvnvcal ,rvala aa an,v-cancer druga鏈 moa,lk vn combvna,vona wv,h DNA- damagvng compounda or an,v-me,abolv,ea (Garre,, and Collvna鏈 2011; Kvm e, al.鏈 2015; Mon,ano e, al.鏈 2013; Pere e, al.鏈 2006).

Prevvoua reaearch haa vden,vfed BRCA1 aa on oa ,he CDK12-dependen, genea and ea,ablvahed ,ha, CDK12 defcvenck leada ,o vmpavred DNA damage repavr and ,o genomvc vna,abvlv,k (Blaeá e, al.鏈 2011). Taávng vn,o accoun, ,ha, bo,h BRCA1 and CDK12 loaa ahow akn,he,vc le,halv,k wv,h PARP vnhvbv,von (Helledak鏈 2010; Lord and Aahwor,h鏈 2013)鏈 we wan,ed ,o ,ea, ,he hkpo,heava ,ha, CDK12 or BRCA1 avlencvng mak aenav,ve cancer cella ,o CHK1 vnhvbv,ora.

3.2.1 CDK12 and BRCA1 downregulation sensitizes HCT116 cells to CHK1 inhibition irrespective of p53 status

In order ,o de,ermvne ,he eaaec, CHK1 vnhvbv,ora on prolvaera,von oa CDK12- or BRCA1-deple,ed cella鏈 we emploked 6-dak prolvaera,von aaaaka. Several a,udvea have repor,ed ,ha, p53 null a,a,ua vncreaaea ,he eaaec,vvv,k oa CHK1 vnhvbv,ora (Carraaaa and Damva鏈 2011; Ma e, al.鏈 2012)鏈 ao we peraormed ,hva expervmen, vn HCT116 cell lvnea wv,h p53 wvld ,kpe or null a,a,ua. In addv,von鏈 ,he aame ,kpe oa expervmen, waa conduc,ed vn an ovarvan cell lvne OVSAHO whvch beara p53 mu,a,von and wvld-,kpe BRCA1 (Domcáe e, al.鏈 2013).

Fvra,鏈 we ,ranaaec,ed cella wv,h a con,rol avRNA or avRNAa agavna, CDK12鏈 CDK13 and BRCA1. To exclude poaavble oaa-,arge, eaaec,a鏈 we uaed ,wo dvaaeren, avRNAa agavna, CDK12. We vncluded CDK13鏈 a ávnaae avmvlar ,o CDK12 wv,h no repor,ed

33 role vn DDR. Nex,鏈 ,ranaaec,ed cella were ,rea,ed wv,h vndvca,ed concen,ra,vona oa ,wo a,ruc,urallk dvaaeren, CHK1 vnhvbv,ora SCH900776 and LY2603618 reapec,vvelk (Guv e, al.鏈 2011; Kvng e, al.鏈 2014). Fvnallk鏈 Ck-QUANT NF aaaak waa uaed aor quan,vfca,von oa ,o,al DNA whvch correaponda ,o ,he cell numbera.

Downregula,von oa BRCA1 and CDK12 led ,o avgnvfca,lk vncreaaed aenav,vvv,k ,o CHK1 vnhvbv,or SCH900776 vn HCT116 cella vrreapec,vve oa p53 wvld ,kpe or null a,a,ua (fgure 10 a鏈b). In con,raa,鏈 no eaaec, waa de,ec,ed vn CDK13-avlenced cella. A avmvlar reaul, waa ob,avned wv,h CHK1 vnhvbv,or LY2603618鏈 whvch haa a dvaaeren, chemvcal a,ruc,ure鏈 and ,hereaore we can exclude oaa ,arge, eaaec,a oa SCH900776 (McNeelk e, al.鏈 2014) (fgure 10 c鏈d). Moreover鏈 aenav,va,von ,o CHK1 vnhvbv,von vn BRCA1- and CDK12-deple,ed cella waa obaerved vn OVSAHO cell lvne (fgure 10e). The eaaec,vve downregula,von oa CDK12鏈 CDK13鏈 ckclvn K鏈 and BRCA1 vn HCT116 cella waa confrmed bk Wea,ern blo,,vng (fgure 10a).

34 A B

C D

E F

Figure 10 Downregula,von oa CDK12 or BRCA1 aenav,vea cella ,o CHK1 vnhvbv,ora. Svx-dak aurvvval curvea oa (a鏈 c) HCT116 p53+/+ or (b鏈 d) HCT116 p53−/− cella or (e) OVSAHO ,ranaaec,ed wv,h vndvca,ed avRNAa (CTRL鏈 CDK12鏈 CDK13鏈 and BRCA1) and ,rea,ed wv,h ev,her ,he CHK1 vnhvbv,or (a鏈 b鏈 e) SCH900776 or (c鏈 d) LY2603618. Cell vvabvlv,k aor each avRNA-,rea,ed cell lvne waa aaaeaaed bk ,he CkQuan, NF áv, and normalved ,o ,he rela,vve grow,h oa cella ,rea,ed wv,h DMSO. Error bara repreaen, SEM aor ,hree vndependen, expervmen,a. CDK12- and BRCA1-avlenced cella are aenav,vve ,o CHK1 vnhvbv,ora (p < 0.001鏈 ANOVA). (a)The eaaec,vve ánocádown oa vndvca,ed pro,evna aa,er avRNA ,ranaaec,vona waa examvned bk Wea,ern blo, analkava. HCT116 p53+/+ and HCT116 p53−/− cella were ,ranaaec,ed wv,h varvoua avRNAa and pro,evn levela were aaaeaaed bk Wea,ern blo, analkava aa,er 72 h. The pro,evn level oa Ckclvn T1 waa uaed aa a loadvng con,rol.

35 3.2.2 BRCA1 or CDK12 depletion coupled with CHK1 inhibition induces p21-dependent proliferation block To gavn undera,andvng oa molecular mechanvam reaponavble aor CHK1 aenav,vvv,k oa BRCA1 and CDK12 avlenced cella鏈 we de,ermvned levela oa DNA damage鏈 apop,oava鏈 au,ophagk鏈 and cell ckcle a,a,ua maráera bk wea,ern blo,,vng. We deple,ed HCT116 cella oa CDK12鏈 CDK13鏈 or BRCA1 bk avRNA and aa,er 24 houra cella were ,rea,ed wv,h CHK1 vnhvbv,or (SCH900776) aor addv,vonal 96h and aubaequen,lk wea,ern blo,,vng analkava wv,h vndvca,ed an,vbodvea waa peraormed.

We de,ec,ed decreaaed CHK1 and BRCA1 levela aa,er CDK12 deple,von鏈 whvch va conava,en, wv,h our prevvoua fndvng ,ha, CHK1 and BRCA1 are CDK12- dependen, gene (fgure 11a) (Blaeá e, al.鏈 2011; Eáumv e, al.鏈 2015). Inhvbv,von reaul,ed vn decreaae oa CHK1 pro,evn level鏈 whvch va conava,en, wv,h our prevvoua repor, (Eáumv e, al.鏈 2015). Nex,鏈 we obaerved decreaaed CHK1 au,ophoaphorkla,von a, aervne 296 aa,er SCH900776 ,rea,men,鏈 whvch vndvca,ea eaaec,vve vnhvbv,von oa CHK1 ávnaae ac,vvv,k . In con,raa, ,o CDK13鏈 CDK12 avlencvng combvned wv,h CHK1 vnhvbv,von led ,o avgnvfcan,lk eleva,ed level oa kH2AX鏈 whvch va a maráer oa ,he DNA damage repavr (Sedelnváova e, al.鏈 2002).

Surprvavnglk鏈 CHK1 vnhvbv,or admvnva,ra,von had no eaaec, on PARP-1 cleavage鏈 whvch va an apop,oava maráer (fgure 11b). In addv,von鏈 we dvd no, de,ec,ed ank avgnvfcan, changea vn levela oa au,ophagk maráera Beclvn and LC3B cleaved produc, (da,a no, ahown).

Nex,鏈 we de,ermvned levela oa ,he cell ckcle progreaavon maráera. CHK1 vnhvbv,von had no eaaec, on p53 level鏈 bu, vn,erea,vnglk鏈 we obaerved p53 eleva,von vn CDK12- deple,ed cella. Level oa CDK vnhvbv,or p27 dvd no, change vn ank ,ea,ed condv,vona. Fvnallk鏈 we examvned ,he a,a,ua p21 (Cvp1/Waa1)鏈 whvch va CDKa vnhvbv,or and v, blocáa ,he cell ckcle progreaavon vn reaponae ,o DNA damage (Wang e, al.鏈 2014). Level oa p21 vncreaaed wv,h CHK1 vnhvbv,or vn a doae-dependen, manner. S,rongea, p21 eleva,von waa de,ec,ed vn BRCA1-deple,ed cella. In,erea,vnglk鏈 CDK12 deple,von reaul,ed vn a robua, p21 vnduc,von vrreapec,vve oa CHK1 vnhvbv,von.

In addv,von ,o regula,vng G1/S progreaavon ,hrough CDK vnhvbv,von鏈 p21 alao vnducea ,he degrada,von oa ,he Re,vnoblaa,oma pro,evn (pRb) whvch va an vmpor,an, cell ckcle regula,or (Broude e, al.鏈 2007). CDK12 deple,von led ,o

36 degrada,von oa pRb regardleaa oa CHK1 vnhvbv,von鏈 and a modera,e eaaec, waa alao obaerved vn BRCA1 deple,ed cella aa,er ,rea,men, wv,h 1μM oa SCH900776.

Taáen ,oge,her鏈 our reaul,a auggea, ,ha, vncreaaed aenav,vvv,k oa CDK12- or BRCA1- deple,ed cella ,o ck,oa,a,vc eaaec, oa CHK1 vnhvbv,von could be aacrvbed ,o vncreaaed DNA damage鏈 whvch leada ,o a robua, eleva,von oa p21 and delaked cell ckcle progreaavon. Reduced cell vvabvlv,k canno, be explavned bk vnduc,von oa apop,oava or au,ophagk.

Figure 11 Impac, oa CDK12 and BRCA1 downregula,von on DDR鏈 apop,oava and cell ckcle. (a) The eaaec,vve ánocádown oa varvoua pro,evna aa,er avRNA ,ranaaec,von鏈 CHK1 vnhvbv,von and DNA damage vnduc,von waa aaaeaaed bk Wea,ern blo, analkava. HCT116 p53+/+ cella were ,ranaaec,ed wv,h con,rol or apecvfc avRNAa (CTRL鏈 CDK12鏈 CDK13鏈 and BRCA1) and 2daka poa,-,ranaaec,von cella were ,rea,ed wv,h 0鏈 0.3鏈 or 1 μM CHK1 vnhvbv,or SCH900776 aor an addv,vonal 96 h. The pro,evn levela oa ,he a,udved pro,evna were elucvda,ed bk Wea,ern blo, wv,h vndvca,ed an,vbodvea. The pro,evn level oa Ckclvn T1 waa uaed aa a loadvng con,rol. (b) S,a,ua oa cellular aac,ora par,vcvpa,vng vn regula,von oa apop,oava and cell ckcle. The pro,evn levela oa PARP鏈 a maráer oa la,e apop,oava鏈 ,umor auppreaaor p53鏈 and ,he cell-ckcle regula,vng pro,evna p21鏈 p27鏈 pRb鏈 and pRB- pSer780 were elucvda,ed bk Wea,ern blo, wv,h vndvca,ed an,vbodvea. The pro,evn level oa Ckclvn T1 waa uaed aa a loadvng con,rol.

37 3.2.3 BRCA1 depletion sensitizes MDA-MB-231 cells to CHK1 inhibition Svnce BRCA1 vnac,vva,von va a common aea,ure oa ,rvple-nega,vve breaa, ,umora (Carek e, al.鏈 2010)鏈 we wan,ed ,o ,ea, va BRCA1 deple,von can po,en,va,e ck,oa,a,vc eaaec, oa CHK1 vnhvbv,or vn a breaa, cancer model.

Fvra,鏈 we genera,ed MDA-MB-231 breaa, cancer cell lvnea wv,h a a,able expreaavon oa ahRNA agavna, BRCA1 (MDA-MB-231 ahBRCA1 #2 and #4). We confrmed ,he eaaec,vve BRCA1 avlencvng on mRNA and pro,evn level (Fvgure 12a).

Nex,鏈 we peraormed a cell vvabvlv,k aaaak ,o aaaeaa ,he ck,oa,a,vc eaaec, oa CHK1 vnhvbv,von aa,er BRCA1 deple,von (Fvgure 12b). Aa,er 3 daka鏈 admvnva,ra,von oa 1uM SCH900776 led ,o a avgnvfcan,lk decreaaed vvabvlv,k oa bo,h BRCA1 avlenced cell lvnea compared ,o ,he con,rol cella鏈 whvch va vn accordance ,o our reaul,a ob,avned vn HCT116 and OVSAHO cell lvnea (Fvgure 10). We wan,ed ,o aur,her examvne ,he deacrvbed eaaec, vn a clonogenvc aaaak (Fvgure 12c). We ,rea,ed ,he cella wv,h vndvca,ed doaea oa CHK1 vnhvbv,or and a,avned ,he cella aa,er 14 daka. BRCA1- deple,ed MDA-MB-231 cella aormed much larger colonvea compared ,o ,he paren,al cella and CHK1 vnhvbv,or admvnva,ra,von draa,vcallk reduced ,he number and ,he ave oa ,he colonvea. To clarvak whe,her ,hva eaaec, could be explavned bk ,he delaked cell ckcle progreaavon鏈 we peraormed analkava oa ,he cell ckcle. We ,rea,ed cella wv,h vndvca,ed doaea oa SCH900776 and aa,er 3 daka we a,avned cella wv,h propvdvum vodvde and analkaed ,hem on bk [ow ck,ome,rk (Fvgure 12d). CHK1 vnhvbv,von had no eaaec, on ,he paren,al cell lvne. In BRCA1-avlenced cella鏈 CHK1 vnhvbv,or admvnva,ra,von led ,o drama,vc vncreaae oa ,he S-phaae popula,von鏈 whvch vndvca,ea arrea, vn ,he cell ckcle progreaavon. The prevvoua reaul, mak vndvca,e ,he preaence oa replvca,von a,reaa vn ,he S-phaae whvch can reaul, vn vncreaaed DNA damage (Gavllard e, al.鏈 2015). To ,ea, ,hva hkpo,heava鏈 we examvned γH2AX levela bk Wea,ern blo,,vng (Fvgure 12e). CHK1 vnhvbv,von led ,o a drama,vc vncreaae vn ,he kH2AX avgnal vn bo,h ahBRCA1 cell lvnea鏈 whvle ,he eaaec, waa onlk modera,e vn ,he paren,al cella. Fvnallk鏈 CHK1 vnhvbv,von led ,o decreaaed level oa ,o,al and phoaphorkla,ed pRB鏈 whvch correaponda ,o ,he da,a ob,avned vn HCT116 cell lvnea.

Taáen ,oge,her鏈 our reaul,a vndvca,e ,ha, BRCA1 deple,von leada ,o vncreaaed aenav,vvv,k oa MDA-MB-231 cella ,o CHK1 vnhvbv,von. CHK1 vnhvbv,or

38 admvnva,ra,von reaul,ed vn vncreaaed DNA damage and ,he cell ckcle arrea, vn ,he S-phaae.

Figure 12 Downregula,von oa BRCA1 aenav,vea MDA-MB-231 cella ,o CHK1 vnhvbv,or. (a) BRCA1 pro,evn levela were evalua,ed vn ahBRCA1 #2 and ahBRCA1 #4 MDA-MB-231 cell lvnea bk Wea,ern blo, analkava. The mRNA levela oa BRCA1 vn ,heae cell lvnea were meaaured bk RT-qPCR (p<0.05鏈 S,uden,’a ,-,ea,). (b) CHK1 vnhvbv,von reduced ,he vvabvlv,k oa MDA-MB-231 cella expreaavng ahRNAa agavna, BRCA1. The grapha ahow ,he reaul,a oa aurvvval aaaaka oa paren,al鏈 ahBRCA1 #2 and ahBRCA1 #4 MDA-MB-231 cella ,rea,ed wv,h 0鏈 0鏈 3鏈 or 1 μM CHK1 vnhvbv,or SCH900776 aor 3 daka. For each cell lvne鏈 ,he cell numbera were normalved ,o ,he rela,vve grow,h oa cella ,rea,ed wv,h DMSO. Error bara repreaen, SEM aor ,hree vndependen, expervmen,a (p<0.05鏈 S,uden,’a ,-,ea,). (c) A 14-dak clonogenvc aaaak ahowed ,ha, ,he combvna,von oa BRCA1 avlencvng and CHK1 vnhvbv,von reducea cell vvabvlv,k. MDA-MB-231 cell lvnea were aeeded and ,rea,ed wv,h ,he vndvca,ed SCH900776 concen,ra,vona. (d) CHK1 vnhvbv,von a,ronglk aaaec,ed ,he cell ckcle progreaavon oa BRCA1- avlenced MDA-MB-231 cell lvnea. Cella were cul,vva,ed and ,rea,ed aa deacrvbed vn (b). Cell ckcle progreaavon waa evalua,ed bk ,he vncorpora,von oa propvdvum vodvde aollowed bk [ow ck,ome,rk. (e) The combvna,von oa BRCA1 avlencvng and CHK1 vnhvbv,von vnducea a a,ronger ac,vva,von oa DDR. Cell were prepared and ,rea,ed aa deacrvbed vn (b). The ac,vva,von oa DDR and vnhvbv,von oa CHK1 bk SCH900776 were valvda,ed bk Wea,ern blo, wv,h vndvca,ed an,vbodvea.

39 3.2.4 Mouse xenograft model

Baaed on our prevvoua reaul,a鏈 we nex, wan,ed ,o elucvda,e ,he eaaec, oa CHK1 vnhvbv,or on BRCA1-avlenced cella in vivo emplokvng mouae or,ho,opvc xenograa, model. MDA-MB-231 paren,al or ahBRCA1 #4 cella were vnjec,ed ,o aa, pada oa

SHO-PrkicSCID HrHr mvce. Aa,er ,umora were aormed鏈 bo,h groupa were ,rea,ed wv,h CHK1 vnhvbv,or or vehvcle. Mvce were con,vnuoualk monv,ored aor ,he developmen, oa a prvmark xenograa, ,umor and aacrvfced when ,umora reached 10% oa bodk wevgh,.

For ,he paren,al con,rol cella鏈 CHK1 vnhvbv,or ,rea,men, dvd no, avgnvfcan,lk aaaec, ,he ,umor volumea (fgure 13a). In,erea,vnglk鏈 ahBRCA1 cella aormed ,umora wv,h hvgher volumea鏈 whvch correaponda wv,h aaa,er grow,h oa ahBRCA1 MDA-MB-231 in vitro. Fvnallk鏈 vn mvce vnjec,ed wv,h ahBRCA1 MDA-MB-231cella鏈 CHK1 vnhvbv,or ,rea,men, reaul,ed vn avgnvfcan,lk decreaaed ,umor volumea (fgure 13b).

Figure 13 Inhvbv,von oa CHK1 preven,a ,umor grow,h vn vvvo. The CHK1 vnhvbv,or SCH900776 decreaaed ,umor grow,h. The (a) paren,al and (b) ahBRCA1 MDA-MB-231 cella were ,ranaplan,ed vn,o ,he mammark pada oa SCID mvce. When ,umor maaa reached a volume oa 0.03 cm3鏈 mvce were ,rea,ed wv,h ev,her a vehvcle aolu,von oa 20% Kollvphor ELP or SCH900776 25mg/ág/dak dvaaolved vn ,he aame 20% Kollvphor aolu,von鏈 everk dak aor 5daka. The grow,h oa ,umor maaa waa ,hen monv,ored over ae, ,vme pervoda. Each da,a povn, repreaen,a ,he mean vncreaae vn ,umor volume aa,er ,he begvnnvng oa ,rea,men, and error bara repreaen, SEM鏈 where n aor each cohor, waa avx anvmala (p < 0.001鏈 S,uden,’a ,-,ea, aor ahBRCA1 SCH900776 va ahBRCA1 vehvcle).

40 3.2.5 Discussion The ,herapeu,vc po,en,val oa CHK1 vnhvbv,ora have been ,ea,ed vn preclvnvcal a,udvea and clvnvcal ,rvala (Carraaaa and Damva鏈 2011鏈 p. 1; McNeelk e, al.鏈 2014; Thompaon and Eaa,man鏈 2013). Svnce CHK1 vnhvbv,ora exhvbv, ,he an,v- prolvaera,vve eaaec, vn combvna,von wv,h DNA damagvng druga鏈 we hkpo,heaved ,ha, compromvaed DDR cauaed bk BRCA1 or CDK12 deple,von wvll have a avmvlar aknergva,vc eaaec,. In haa been prevvoualk demona,ra,ed ,ha, CDK12 regula,ea ,he expreaavon oa a aubae, oa DDR genea鏈 par,vcularlk genea vnvolved vn HR repavr pa,hwak鏈 vncludvng BRCA1 (Blaeá e, al.鏈 2011). Subaequen,lk鏈 BRCA1 and CDK12 defcvenck reaul,a vn genomvc vna,abvlv,k and conaer ,he akn,he,vc le,halv,k wv,h PARP vnhvbv,ora (Bajramv e, al.鏈 2014; Joahv e, al.鏈 2014). In our a,udk鏈 we demona,ra,ed ,ha, ,he BRCA1 and CDK12 avlencvng alao po,en,va,ea ,he an,v– prolvaera,vve eaaec, oa CHK1 vnhvbv,ora.

Our reaul,a vndvca,e ,ha, ,he an,v-prolvaera,vve eaaec, oa CHK1 vnhvbv,or ,rea,men, combvned wv,h BRCA1 or CDK12 defcvenck va no, dependen, on ,he p53 a,a,ua vn HCT116 cella. Thva obaerva,von va con,rark ,o aome prevvoua repor,a鏈 never,heleaa ,hva could be a cell lvne-apecvfc eaaec, (Carraaaa and Damva鏈 2011; Ma e, al.鏈 2012).

We ahowed ,ha, CHK1 vnhvbv,von leada ,o vncreaaed DNA damage meaaured bk ,he vnduc,von oa γH2AX pSer139 vn HCT116 cella鏈 whvch va vn lvne wv,h prevvoua repor,a (Guv e, al.鏈 2011). In HCT116 cella鏈 ,he a,rongea, eaaec, waa ob,avned aor combvna,von oa CHK1 vnhvbv,von and CDK12 avlencvng. Moreover鏈 we de,ec,ed a drama,vc γH2AX pSer139 vnduc,von vn ,he BRCA-avlenced MDA-MB-231 cella aa,er CHK1 vnhvbv,von. Conaequen,lk鏈 CHK1 vnhvbv,von reaul,ed vn a,rong vnduc,von oa ,he cell ckcle regula,or p21鏈 whvch va capable oa cell ckcle arrea, vn reaponae ,o exceaavve DNA damage. Baaed on ,heae da,a鏈 one could apecula,e ,ha, CHK1 vnhvbv,ora ,rvgger vncreaaed DNA damage and replvca,von a,reaa durvng ,he S-phaae oa ,he cell ckcle (Gavllard e, al.鏈 2015)鏈 whvch canno, be repavred vn cella wv,h deple,ed BRCA1 and CDK12 and v, va vncompa,vble wv,h cell prolvaera,von or aurvvval.

In our prevvua a,udk鏈 we have demona,ra,ed ,ha, CHK1 va a CDK12-dependen, gene and v, va avgnvfcan,lk downregula,ed vn HGSOC ,umor aamplea (Eáumv e, al.鏈 2015). Thua CDK12-defcven, cella relk on lower CHK1 level and ,hva can explavn ,he reaaon whk CHK1 vnhvbv,von haa a,ronger eaaec, on vvabvlv,k oa ,heae cella.

41 Fvnallk鏈 emplokvng xenograa, mouae model鏈 we demona,ra,ed ,ha, CHK1 vnhvbv,von lvmv,a ,he ,umor grow,h ob BRCA1-defcven, cella in vivo.

In aummark鏈 we auggea, ,ha, CHK1 vnhvbv,von mvgh, be an eaaec,vve a,ra,egk aor ,arge,vng BRCA1- or CDK12-defcven, ,umora. BRCA1 and CDK12 loaa-oa-aunc,von mu,a,vona mvgh, be conavdered aa maráera oa CHK1 aenav,vvv,k. Moreover鏈 CHK1 vnhvbv,ora could po,en,va,e ,he an,v-cancer eaaec, oa ,he recen,lk developed CDK12 vnhvbv,ora.

42 3.3 The emerging roles of CDK12 in tumorigenesis

An vncreaavng number oa a,udvea have repor,ed on CDK12 mu,a,vona or v,a overexpreaavon vn varvoua cancer ,kpea (Cancer Genome A,laa Ne,worá鏈 2012; Capra e, al.鏈 2006; Eáumv e, al.鏈 2015; Mer,vna e, al.鏈 2016). Subaequen,lk鏈 ,here va an a,,emp, ,o elucvda,e ,he role oa CDK12 vn malvgnan, ,ranaaorma,von鏈 ,o vden,vak ,he underlkvng molecular mechanvama and ,o develop CDK12-baaed ,herapvea (Bajramv e, al.鏈 2014; Johnaon e, al.鏈 2016; Zhang e, al.鏈 2016). We decvded ,o wrv,e a revvew ,o aummarve and dvacuaa curren, ánowledge concernvng CDK12 wv,h reapec, ,o ovarvan and breaa, cancer. In our revvew鏈 we dvacuaaed ,he apparen, con,radvc,von ,ha, dvaaeren, a,udvea deacrvbe CDK12 aa a ,umor auppreaaor or aa a po,en,val oncogene.

In HGSOC鏈 CDK12 va a aubjec, ,o loaa-oa-aunc,von mu,a,vona and haa a role oa a ,umor auppreaaor (Cancer Genome A,laa Reaearch Ne,worá鏈 2011; Joahv e, al.鏈 2014). In ,heae ,umora鏈 CDK12 defcvenck or vnhvbv,von conaer aenav,vvv,k ,o PARP or CHK1 vnhvbv,ora (Bajramv e, al.鏈 2014; Johnaon e, al.鏈 2016).

On ,he o,her hand鏈 CDK12 va amplvfed vn Her2 poav,vve breaa, ,umora. In ,heae ,umora鏈 eleva,ed expreaavon oa CDK12 correla,ea wv,h more aggreaavve ,umor progreaavon (Capra e, al.鏈 2006)鏈 CDK12 va hvghlk ac,vve and v, haa been propoaed aa a druggable ,arge, (Mer,vna e, al.鏈 2016). CDK12 overexpreaavon aaaec,a al,erna,vve aplvcvng oa v,a ,arge, genea and ,hereaore v, can vncreaae ,he vnvaavveneaa oa a breaa, cancer cell lvne (Tven e, al.鏈 2017). Fvnallk鏈 CDK12 vnhvbv,or haa been ahown ,o lvmv, grow,h oa cancer cella (Zhang e, al.鏈 2016). Collec,vvelk鏈 ,heae obaerva,vona auppor, ,he hkpo,heava ,ha, CDK12 haa an oncogenvc po,en,val.

The apparen, con,radvc,von mak be explavned bk ,he poaavbvlv,k ,ha, CDK12 va reaponavble aor expreaavon oa bo,h ,umor auppreaaora and oncogenea (fgure 14).

43 Figure 14 Po,en,val rolea oa CDK12 vn carcvnogeneava. a) CDK12 haa ,he ,umor- auppreaavve proper,vea. CDK12 loaa-oa aunc,von mu,a,vona led ,o decreaaed expreaavon oa HR genea reaul,vng vn genomvc vna,abvlv,k and ,umorvgeneava. CDK12 loaa or vnhvbv,von aenav,vea ,umor cella ,o PARP1/2 vnhvbv,ora. B) CDK12 haa oncogenvc proper,vea. CDK12 amplvfca,von mvgh, lead ,o vncreaaed expreaavon oa varvoua oncogenea and conaequen,lk par,vcvpa,e vn ,umorvgeneava. Thereaore ,arge,vng CDK12 wv,h apecvfc vnhvbv,ora vn ,heae ,umora could be benefcvark aor pa,ven, ,rea,men,

44 4. Summary In mk ,heava I commen,ed on ,he reaul,a oa ,hree publvahed a,udvea. The purpoae oa ,he fra, par, waa ,o de,ermvne ,he vmpac, and avgnvfcance oa CDK12 mu,a,vona鏈 ,ha, were aound vn HGSOC ,umor aamplea. Our reaul,a vndvca,e ,ha, ,heae are loaa- oa-aunc,von mu,a,vona ,ha, preven, CDK12/Ckclvn K complex aorma,von鏈 dvarup, ,he CDK12 ávnaae ac,vvv,k and compromvae ,he abvlv,k oa cella ,o execu,e DNA repavr bk HR. In addv,von鏈 we vden,vfed new CDK12-dependend genea ,ha, par,vcvpa,e HR repavr pa,hwak. Taáen ,oge,her鏈 ,heae reaul,a auggea, ,ha, CDK12 vnac,vva,vng mu,a,vona reaul, vn downregula,von oa mul,vple DNA repavr- aaaocva,ed aac,ora vncludvng HR鏈 whvch mak lead ,o genomvc vna,abvlv,k and ,umorvgeneava. Our reaul,a auppor, recen, fndvnga鏈 ,ha, HR defcvenck cauaed bk CDK12 loaa or vnhvbv,von mak provvde an oppor,unv,k ,o ,rea,men, oa ,heae ,umora wv,h PARP vnhvbv,ora.

In ,he aecond par, we ,ea,ed a hkpo,heava ,ha,鏈 avmvlarlk ,o PARP1鏈 CHK1 vnhvbv,von reaul,a vn ,he reduced vvabvlv,k oa BRCA1- and CDK12- deple,ed cella. Our expervmen,a confrmed ,ha, bo,h BRCA1 and CDK12 avlencvng reaul,a vn vncreaaed CHK1 vnhvbv,or aenav,vvv,k in vitro. Mechanvama underlkvng ,hva eaaec, vnclude p21-dependen, cell ckcle arrea, vn reaponae ,o eleva,ed DNA damage. Impor,an,lk鏈 we ahowed ,ha, cella deple,ed oa BRCA1鏈 whvch va a CDK12- dependen, gene鏈 are aenav,vve ,o CHK1 vnhvbv,or in vivo. Overall鏈 our reaul,a auggea, ,ha, pa,ven,a wv,h BRCA1 or CDK12 defcvenck mak benef, arom CHK1 vnhvbv,or ,rea,men,鏈 never,heleaa more dvrec, evvdence and aur,her a,udvea are neceaaark ,o confrm ,hva hkpo,heava.

In our revvew we aummarved ,he curren, ánowledge and dvacuaaed ,he con,roverak concernvng CDK12 role vn carcvnogeneava. We hvghlvgh,ed ,he aac, ,ha, CDK12 va ev,her a ,umor auppreaaor candvda,e or v, haa proper,vea ,ha, reaemble oncogenea vn dvaaeren, ,kpea oa ,umora. However鏈 v, remavna ,o be de,ermvned whe,her CDK12 va bona fid oncogene.

Overall鏈 our reaul,a provvded be,,er undera,andvng oa CDK12 bvologvcal aunc,vona and mak con,rvbu,e ,o ,he developmen, oa CDK12-baaed cancer ,herapvea.

45 5. References Anaarv鏈 S.A.鏈 Morae鏈 R.H.鏈 2013. Mechanvama oa Medva,or complex ac,von vn ,ranacrvp,vonal ac,vva,von. Cell. Mol. Lvae Scv. 70鏈 2743–2756. Aaghar鏈 U.鏈 Wv,ávewvc鏈 A.K.鏈 Turner鏈 N.C.鏈 Knudaen鏈 E.S.鏈 2015. The hva,ork and au,ure oa ,arge,vng ckclvn-dependen, ávnaaea vn cancer ,herapk. Na, Rev Drug Dvacov 14鏈 130–146. Bahaaav鏈 E.M.鏈 Oveaen鏈 J.L.鏈 Rveaenberg鏈 A.L.鏈 Berna,evn鏈 W.Z.鏈 Haa,k鏈 P.E.鏈 S,ambrooá鏈 P.J.鏈 2008. The checápovn, ávnaaea Chá1 and Chá2 regula,e ,he aunc,vonal aaaocva,vona be,ween hBRCA2 and Rad51 vn reaponae ,o DNA damage. Oncogene 27鏈 3977–3985. Bajramv鏈 I.鏈 Franáum鏈 J.R.鏈 Konde鏈 A.鏈 Mvller鏈 R.E.鏈 Rehman鏈 F.L.鏈 Brough鏈 R.鏈 Campbell鏈 J.鏈 Svma鏈 D.鏈 Rafq鏈 R.鏈 Hooper鏈 S.鏈 Chen鏈 L.鏈 Koarewa鏈 I.鏈 Aaavo,va鏈 I.鏈 Fenwvcá鏈 K.鏈 Na,rajan鏈 R.鏈 Lord鏈 C.J.鏈 Aahwor,h鏈 A.鏈 2014. Genome-wvde proflvng oa gene,vc akn,he,vc le,halv,k vden,vfea CDK12 aa a novel de,ermvnan, oa PARP1/2 vnhvbv,or aenav,vvv,k. Cancer Rea. 74鏈 287–297. Bar,áowvaá鏈 B.鏈 Greenleaa鏈 A.L.鏈 2015. Expreaavon鏈 purvfca,von鏈 and vden,vfca,von oa aaaocva,ed pro,evna oa ,he aull-leng,h hCDK12/CkclvnK complex. J. Bvol. Chem. 290鏈 1786–1795. Bar,áowvaá鏈 B.鏈 Lvu鏈 P.鏈 Pha,nanv鏈 H.P.鏈 Fuda鏈 N.J.鏈 Cooper鏈 J.J.鏈 Prvce鏈 D.H.鏈 Adelman鏈 K.鏈 Lva鏈 J.T.鏈 Greenleaa鏈 A.L.鏈 2010. CDK12 va a ,ranacrvp,von elonga,von-aaaocva,ed CTD ávnaae鏈 ,he me,aoan or,holog oa keaa, C,á1. Genea Dev. 24鏈 2303–2316. Blaeá鏈 D.鏈 2016. Tranacrvp,vonal ávnaaea: Caugh, bk a a,vcák drug. Na,. Chem. Bvol. 12鏈 765–766. Blaeá鏈 D.鏈 Kohou,eá鏈 J.鏈 Bar,holomeeuaen鏈 K.鏈 Johanaen鏈 E.鏈 Hulvnáova鏈 P.鏈 Luo鏈 Z.鏈 Cvmermancvc鏈 P.鏈 Ule鏈 J.鏈 Pe,erlvn鏈 B.M.鏈 2011. The Ckclvn K/Cdá12 complex mavn,avna genomvc a,abvlv,k vva regula,von oa expreaavon oa DNA damage reaponae genea. Genea Dev. 25鏈 2158–2172. Böaáen鏈 C.A.鏈 Farnung鏈 L.鏈 Hvn,ermavr鏈 C.鏈 Merel Schach,er鏈 M.鏈 Vogel-Bachmakr鏈 K.鏈 Blaeá鏈 D.鏈 Anand鏈 K.鏈 Fvaher鏈 R.P.鏈 Evcá鏈 D.鏈 Geker鏈 M.鏈 2014. The a,ruc,ure and auba,ra,e apecvfcv,k oa human Cdá12/Ckclvn K. Na, Commun 5鏈 3505. Bowman鏈 E.A.鏈 Kellk鏈 W.G.鏈 2014. RNA polkmeraae II ,ranacrvp,von elonga,von and Pol II CTD Ser2 phoaphorkla,von: A ,avl oa ,wo ávnaaea. Nucleua 5鏈 224–236. Broude鏈 E.V.鏈 Swva,鏈 M.E.鏈 Vvvo鏈 C.鏈 Chang鏈 B.-D.鏈 Davva鏈 B.M.鏈 Kalurupalle鏈 S.鏈 Blagoaálonnk鏈 M.V.鏈 Ronvnaon鏈 I.B.鏈 2007. p21(Waa1/Cvp1/Sdv1) medva,ea re,vnoblaa,oma pro,evn degrada,von. Oncogene 26鏈 6954–6958. Brkan,鏈 H.E.鏈 Schul,鏈 N.鏈 Thomaa鏈 H.D.鏈 Paráer鏈 K.M.鏈 Flower鏈 D.鏈 Lope鏈 E.鏈 Kkle鏈 S.鏈 Meu,h鏈 M.鏈 Cur,vn鏈 N.J.鏈 Helledak鏈 T.鏈 2005. Specvfc ávllvng oa BRCA2- defcven, ,umoura wv,h vnhvbv,ora oa polk(ADP-rvboae) polkmeraae. Na,ure 434鏈 913–917. Bura,owaáv鏈 S.鏈 2003. The CTD code. Na,. S,ruc,. Bvol. 10鏈 679–680. Cancer Genome A,laa Ne,worá鏈 2012. Comprehenavve molecular por,rav,a oa human breaa, ,umoura. Na,ure 490鏈 61–70. Cancer Genome A,laa Reaearch Ne,worá鏈 2011. In,egra,ed genomvc analkaea oa ovarvan carcvnoma. Na,ure 474鏈 609–615. Capra鏈 M.鏈 Nucvaoro鏈 P.G.鏈 Conaalonverv鏈 S.鏈 Quar,o鏈 M.鏈 Bvanchv鏈 M.鏈 Nebulonv鏈 M.鏈 Boldorvnv鏈 R.鏈 Pallo,,v鏈 F.鏈 Vvale鏈 G.鏈 Gvahvák鏈 M.L.鏈 Drae,,a鏈 G.F.鏈 Dv Fvore鏈 P.P.鏈 2006. Frequen, al,era,vona vn ,he expreaavon oa aervne/,hreonvne ávnaaea vn human cancera. Cancer Rea. 66鏈 8147–8154.

46 Carek鏈 L.鏈 Wvner鏈 E.鏈 Vvale鏈 G.鏈 Cameron鏈 D.鏈 Gvannv鏈 L.鏈 2010. Trvple-nega,vve breaa, cancer: dvaeaae en,v,k or ,v,le oa convenvence? Na, Rev Clvn Oncol 7鏈 683– 692. Carraaaa鏈 L.鏈 Damva鏈 G.鏈 2011. Unleaahvng Chá1 vn cancer ,herapk. Cell Ckcle 10鏈 2121–2128. Chang鏈 T.-Y.鏈 Cheng鏈 P.-L.鏈 2015. Relak oa ckclvn-dependen, ávnaaea vn ,he regula,von oa axonal grow,h. Exp. Neurol. 271鏈 259–261. Chen鏈 C.C.鏈 Kennedk鏈 R.D.鏈 Svdv鏈 S.鏈 Looá鏈 A.T.鏈 D’Andrea鏈 A.鏈 2009. CHK1 vnhvbv,von aa a a,ra,egk aor ,arge,vng Fanconv Anemva (FA) DNA repavr pa,hwak defcven, ,umora. Mol. Cancer 8鏈 24. Chen鏈 H.-H.鏈 Wong鏈 Y.-H.鏈 Genevvere鏈 A.-M.鏈 Fann鏈 M.-J.鏈 2007. CDK13/CDC2L5 vn,erac,a wv,h L-,kpe ckclvna and regula,ea al,erna,vve aplvcvng. Bvochem. Bvophka. Rea. Commun. 354鏈 735–740. h,,pa://dov.org/10.1016/j.bbrc.2007.01.049 Chen鏈 H.-R.鏈 Juan鏈 H.-C.鏈 Wong鏈 Y.-H.鏈 Taav鏈 J.-W.鏈 Fann鏈 M.-J.鏈 2016. Cdá12 Regula,ea Neurogeneava and La,e-Arvavng Neuronal Mvgra,von vn ,he Developvng Cerebral Cor,ex. Cereb. Cor,ex. Cheng鏈 S.-W.G.鏈 Kuká鏈 M.A.鏈 Moradvan鏈 A.鏈 Ichu鏈 T.-A.鏈 Chang鏈 V.C.-D.鏈 Tven鏈 J.F.鏈 Volle,,鏈 S.E.鏈 Grvaf,h鏈 M.鏈 Marra鏈 M.A.鏈 Morvn鏈 G.B.鏈 2012. In,erac,von oa ckclvn-dependen, ávnaae 12/CráRS wv,h ckclvn K1 va requvred aor ,he phoaphorkla,von oa ,he C-,ermvnal domavn oa RNA polkmeraae II. Mol. Cell. Bvol. 32鏈 4691–4704. Chvlà鏈 R.鏈 Guaaan,v鏈 F.鏈 Damva鏈 G.鏈 2016. Role and ,herapeu,vc po,en,val oa CDK12 vn human cancera. Cancer Trea,. Rev. 50鏈 83–88. Chvpumuro鏈 E.鏈 Marco鏈 E.鏈 Chrva,enaen鏈 C.L.鏈 Kwva,áowaáv鏈 N.鏈 Zhang鏈 T.鏈 Ha,hewak鏈 C.M.鏈 Abraham鏈 B.J.鏈 Sharma鏈 B.鏈 Yeung鏈 C.鏈 Al,abea鏈 A.鏈 Pere- A,akde鏈 A.鏈 Wong鏈 K.-K.鏈 Yuan鏈 G.-C.鏈 Grak鏈 N.S.鏈 Young鏈 R.A.鏈 George鏈 R.E.鏈 2014. CDK7 Inhvbv,von Suppreaaea Super-Enhancer-Lvnáed Oncogenvc Tranacrvp,von vn MYCN-Drvven Cancer. Cell 159鏈 1126–1139. Chrva,enaen鏈 C.L.鏈 Kwva,áowaáv鏈 N.鏈 Abraham鏈 B.J.鏈 Carre,ero鏈 J.鏈 Al-Shahrour鏈 F.鏈 Zhang鏈 T.鏈 Chvpumuro鏈 E.鏈 Her,er-Sprve鏈 G.S.鏈 Aábak鏈 E.A.鏈 Al,abea鏈 A.鏈 Zhang鏈 J.鏈 Shvmamura鏈 T.鏈 Capelle,,v鏈 M.鏈 Revbel鏈 J.B.鏈 Cavanaugh鏈 J.D.鏈 Gao鏈 P.鏈 Lvu鏈 Y.鏈 Mvchaelaen鏈 S.R.鏈 Poulaen鏈 H.S.鏈 Area鏈 A.R.鏈 Barbve鏈 D.A.鏈 Bradner鏈 J.E.鏈 George鏈 R.E.鏈 Grak鏈 N.S.鏈 Young鏈 R.A.鏈 Wong鏈 K.-K.鏈 2014. Targe,vng ,ranacrvp,vonal addvc,vona vn amall cell lung cancer wv,h a covalen, CDK7 vnhvbv,or. Cancer Cell 26鏈 909–922. Coller鏈 J.鏈 Wvcáena鏈 M.鏈 2007. Te,hered aunc,von aaaaka: an adap,able approach ,o a,udk RNA regula,ork pro,evna. Me,h. Enkmol. 429鏈 299–321. Corden鏈 J.L.鏈 2013. The RNA Polkmeraae II C-,ermvnal Domavn: Te,hervng Tranacrvp,von ,o Tranacrvp, and Templa,e. Chem Rev 113鏈 8423–8455. Dav鏈 Q.鏈 Lev鏈 T.鏈 Zhao鏈 C.鏈 Zhong鏈 J.鏈 Tang鏈 Y.鏈 Chen鏈 B.鏈 Yang鏈 J.鏈 Lv鏈 C.鏈 Wang鏈 S.鏈 Song鏈 X.鏈 Lv鏈 L.鏈 Lv鏈 Q.鏈 2012. Ckclvn K-con,avnvng ávnaae complexea mavn,avn aela-renewal vn murvne embrkonvc a,em cella. J. Bvol. Chem. 287鏈 25344– 25352. Davvdaon鏈 L.鏈 Munv鏈 L.鏈 Wea,鏈 S.鏈 2014. 3’ end aorma,von oa pre-mRNA and phoaphorkla,von oa Ser2 on ,he RNA polkmeraae II CTD are recvprocallk coupled vn human cella. Genea Dev. 28鏈 342–356. Domcáe鏈 S.鏈 Svnha鏈 R.鏈 Levvne鏈 D.A.鏈 Sander鏈 C.鏈 Schul,鏈 N.鏈 2013. Evalua,vng cell lvnea aa ,umour modela bk comparvaon oa genomvc proflea. Na, Commun 4鏈 2126. Droga,鏈 J.鏈 Hermand鏈 D.鏈 2012. Gene-apecvfc requvremen, oa RNA polkmeraae II CTD phoaphorkla,von. Mol. Mvcrobvol. 84鏈 995–1004.

47 Dvvnge鏈 H.鏈 Kvm鏈 E.鏈 Abdel-Wahab鏈 O.鏈 Bradlek鏈 R.K.鏈 2016. RNA aplvcvng aac,ora aa oncopro,evna and ,umour auppreaaora. Na, Rev Cancer 16鏈 413–430. Evcá鏈 D.鏈 Geker鏈 M.鏈 2013. The RNA polkmeraae II carboxk-,ermvnal domavn (CTD) code. Chem. Rev. 113鏈 8456–8490. Ev[er鏈 T.T.鏈 Shao鏈 W.鏈 Bar,holomeeuaen鏈 K.鏈 Fujvnaga鏈 K.鏈 Jgger鏈 S.鏈 Johnaon鏈 J.R.鏈 Luo鏈 Z.鏈 Krogan鏈 N.J.鏈 Pe,erlvn鏈 B.M.鏈 2015. Ckclvn-dependen, ávnaae 12 vncreaaea 3’ end proceaavng oa grow,h aac,or-vnduced c-FOS ,ranacrvp,a. Mol. Cell. Bvol. 35鏈 468–478. Eáumv鏈 K.M.鏈 Paculova鏈 H.鏈 Lenaav鏈 T.鏈 Poapvchalova鏈 V.鏈 Böaáen鏈 C.A.鏈 Rkbarváova鏈 J.鏈 Brkja鏈 V.鏈 Geker鏈 M.鏈 Blaeá鏈 D.鏈 Barborvc鏈 M.鏈 2015. Ovarvan carcvnoma CDK12 mu,a,vona mvaregula,e expreaavon oa DNA repavr genea vva defcven, aorma,von and aunc,von oa ,he Cdá12/CkcK complex. Nuclevc Acvda Rea. 43鏈 2575–2589. Farmer鏈 H.鏈 McCabe鏈 N.鏈 Lord鏈 C.J.鏈 Tu,,鏈 A.N.J.鏈 Johnaon鏈 D.A.鏈 Rvchardaon鏈 T.B.鏈 San,aroaa鏈 M.鏈 Dvllon鏈 K.J.鏈 Hvcáaon鏈 I.鏈 Knvgh,a鏈 C.鏈 Mar,vn鏈 N.M.B.鏈 Jacáaon鏈 S.P.鏈 Smv,h鏈 G.C.M.鏈 Aahwor,h鏈 A.鏈 2005. Targe,vng ,he DNA repavr deaec, vn BRCA mu,an, cella aa a ,herapeu,vc a,ra,egk. Na,ure 434鏈 917–921. Fuda鏈 N.J.鏈 Ardehalv鏈 M.B.鏈 Lva鏈 J.T.鏈 2009. Defnvng mechanvama ,ha, regula,e RNA polkmeraae II ,ranacrvp,von vn vvvo. Na,ure 461鏈 186–192. Gavllard鏈 H.鏈 Garca-Muae鏈 T.鏈 Aguvlera鏈 A.鏈 2015. Replvca,von a,reaa and cancer. Na,. Rev. Cancer 15鏈 276–289. Garre,,鏈 M.D.鏈 Collvna鏈 I.鏈 2011. An,vcancer ,herapk wv,h checápovn, vnhvbv,ora: wha,鏈 where and when? Trenda Pharmacol. Scv. 32鏈 308–316. Ghamarv鏈 A.鏈 van de Corpu,鏈 M.P.C.鏈 Thongjuea鏈 S.鏈 van Cappellen鏈 W.A.鏈 van Ijcáen鏈 W.鏈 van Haren鏈 J.鏈 Soler鏈 E.鏈 Evcá鏈 D.鏈 Lenhard鏈 B.鏈 Groaveld鏈 F.G.鏈 2013. In vvvo lvve vmagvng oa RNA polkmeraae II ,ranacrvp,von aac,orvea vn prvmark cella. Genea Dev. 27鏈 767–777. Guo鏈 J.鏈 Prvce鏈 D.H.鏈 2013. RNA Polkmeraae II Tranacrvp,von Elonga,von Con,rol. Chem Rev 113鏈 8583–8603. Guv鏈 T.J.鏈 Paruch鏈 K.鏈 Dwker鏈 M.P.鏈 Labrolv鏈 M.鏈 Shanahan鏈 F.鏈 Davva鏈 N.鏈 Tarvcanv鏈 L.鏈 Wvawell鏈 D.鏈 Seghev鏈 W.鏈 Pena[or鏈 E.鏈 Bhagwa,鏈 B.鏈 Wang鏈 W.鏈 Gu鏈 D.鏈 Haveh鏈 Y.鏈 Lee鏈 S.鏈 Lvu鏈 M.鏈 Parrk鏈 D.鏈 2011. Targe,vng ,he replvca,von checápovn, uavng SCH 900776鏈 a po,en, and aunc,vonallk aelec,vve CHK1 vnhvbv,or vden,vfed vva hvgh con,en, acreenvng. Mol. Cancer Ther. 10鏈 591– 602. Helledak鏈 T.鏈 2013. Pu,,vng polk (ADP-rvboae) polkmeraae and o,her DNA repavr vnhvbv,ora vn,o clvnvcal prac,vce. Curr Opvn Oncol 25鏈 609–614. Helledak鏈 T.鏈 2010. Homologoua recombvna,von vn cancer developmen,鏈 ,rea,men, and developmen, oa drug reava,ance. Carcvnogeneava 31鏈 955–960. Jeronvmo鏈 C.鏈 Ba,avlle鏈 A.R.鏈 Rober,鏈 F.鏈 2013. The wrv,era鏈 readera鏈 and aunc,vona oa ,he RNA polkmeraae II C-,ermvnal domavn code. Chem. Rev. 113鏈 8491– 8522. Jeronvmo鏈 C.鏈 Collvn鏈 P.鏈 Rober,鏈 F.鏈 2016. The RNA Polkmeraae II CTD: The Increaavng Complexv,k oa a Low-Complexv,k Pro,evn Domavn. J. Mol. Bvol. 428鏈 2607–2622. Johnaon鏈 S.F.鏈 Cru鏈 C.鏈 Grevaenberg鏈 A.K.鏈 Dua,鏈 S.鏈 S,over鏈 D.G.鏈 Chv鏈 D.鏈 Prvmacá鏈 B.鏈 Cao鏈 S.鏈 Bernhardk鏈 A.J.鏈 Coulaon鏈 R.鏈 Laaro鏈 J.-B.鏈 Kochupuraááal鏈 B.鏈 Sun鏈 H.鏈 Unv,,鏈 C.鏈 Moreau鏈 L.A.鏈 Saroaveá鏈 K.A.鏈 Scal,rv,v鏈 M.鏈 Jurvc鏈 D.鏈 Baaelga鏈 J.鏈 Rvchardaon鏈 A.L.鏈 Rodvg鏈 S.J.鏈 D’Andrea鏈 A.D.鏈 Balmaña鏈 J.鏈 Johnaon鏈 N.鏈 Geker鏈 M.鏈 Serra鏈 V.鏈 Lvm鏈 E.鏈 Shapvro鏈 G.I.鏈 2016. CDK12 Inhvbv,von Reveraea De Novo and Acquvred PARP Inhvbv,or Reava,ance vn BRCA Wvld-Tkpe and Mu,a,ed Modela oa Trvple-Nega,vve Breaa, Cancer. Cell Rep 17鏈 2367–2381.

48 Joahv鏈 P.M.鏈 Su,or鏈 S.L.鏈 Hun,oon鏈 C.J.鏈 Karnv,鏈 L.M.鏈 2014. Ovarvan cancer- aaaocva,ed mu,a,vona dvaable ca,alk,vc ac,vvv,k oa CDK12鏈 a ávnaae ,ha, promo,ea homologoua recombvna,von repavr and reava,ance ,o cvapla,vn and polk(ADP-rvboae) polkmeraae vnhvbv,ora. J. Bvol. Chem. 289鏈 9247–9253. Juan鏈 H.-C.鏈 Lvn鏈 Y.鏈 Chen鏈 H.-R.鏈 Fann鏈 M.-J.鏈 2016. Cdá12 va eaaen,val aor embrkonvc developmen, and ,he mavn,enance oa genomvc a,abvlv,k. Cell Dea,h Dvaaer. 23鏈 1038–1048. Kauranvemv鏈 P.鏈 Kallvonvemv鏈 A.鏈 2006. Ac,vva,von oa mul,vple cancer-aaaocva,ed genea a, ,he ERBB2 amplvcon vn breaa, cancer. Endocr. Rela,. Cancer 13鏈 39– 49. Kvm鏈 M.K.鏈 Jamea鏈 J.鏈 Annunva,a鏈 C.M.鏈 2015. Topo,ecan aknergvea wv,h CHEK1 (CHK1) vnhvbv,or ,o vnduce apop,oava vn ovarvan cancer cella. BMC Cancer 15鏈 196. Kvng鏈 C.鏈 Dva鏈 H.鏈 Barnard鏈 D.鏈 Barda鏈 D.鏈 Clawaon鏈 D.鏈 Bloaaer鏈 W.鏈 Cox鏈 K.鏈 Guo鏈 S.鏈 Marahall鏈 M.鏈 2014. Charac,erva,von and preclvnvcal developmen, oa LY2603618: a aelec,vve and po,en, Chá1 vnhvbv,or. Invea, New Druga 32鏈 213–226. Ko鏈 T.K.鏈 Kellk鏈 E.鏈 Pvnea鏈 J.鏈 2001. CráRS: a novel conaerved Cdc2-rela,ed pro,evn ávnaae ,ha, colocalvaea wv,h SC35 apecálea. J. Cell. Scv. 114鏈 2591–2603. Kohou,eá鏈 J.鏈 Blaeá鏈 D.鏈 2012. Ckclvn K goea wv,h Cdá12 and Cdá13. Cell Dvv 7鏈 12. Krvahnamur,v鏈 U.鏈 Svlverman鏈 J.F.鏈 2014. HER2 vn breaa, cancer: a revvew and upda,e. Adv Ana, Pa,hol 21鏈 100–107. Kwva,áowaáv鏈 N.鏈 Zhang鏈 T.鏈 Rahl鏈 P.B.鏈 Abraham鏈 B.J.鏈 Reddk鏈 J.鏈 Fvcarro鏈 S.B.鏈 Daa,ur鏈 A.鏈 Amallag鏈 A.鏈 Ramaawamk鏈 S.鏈 Teaar鏈 B.鏈 Jenávna鏈 C.E.鏈 Hanne,,鏈 N.M.鏈 McMvllvn鏈 D.鏈 Sanda鏈 T.鏈 Svm鏈 T.鏈 Kvm鏈 N.D.鏈 Looá鏈 T.鏈 Mv,avadea鏈 C.S.鏈 Weng鏈 A.P.鏈 Brown鏈 J.R.鏈 Benea鏈 C.H.鏈 Mar,o鏈 J.A.鏈 Young鏈 R.A.鏈 Grak鏈 N.S.鏈 2014. Targe,vng ,ranacrvp,von regula,von vn cancer wv,h a covalen, CDK7 vnhvbv,or. Na,ure 511鏈 616–620. Lee鏈 T.I.鏈 Young鏈 R.A.鏈 2013. Tranacrvp,vonal regula,von and v,a mvaregula,von vn dvaeaae. Cell 152鏈 1237–1251. Lv鏈 X.鏈 Cha,,erjee鏈 N.鏈 Spvrohn鏈 K.鏈 Bou,roa鏈 M.鏈 Bohmann鏈 D.鏈 2016. Cdá12 Ia A Gene-Selec,vve RNA Polkmeraae II Kvnaae Tha, Regula,ea a Subae, oa ,he Tranacrvp,ome鏈 Includvng Nra2 Targe, Genea. Scv Rep 6鏈 21455. Lvang鏈 K.鏈 Gao鏈 X.鏈 Gvlmore鏈 J.M.鏈 Florena鏈 L.鏈 Waahburn鏈 M.P.鏈 Smv,h鏈 E.鏈 Shvla,vaard鏈 A.鏈 2015. Charac,erva,von oa human ckclvn-dependen, ávnaae 12 (CDK12) and CDK13 complexea vn C-,ermvnal domavn phoaphorkla,von鏈 gene ,ranacrvp,von鏈 and RNA proceaavng. Mol. Cell. Bvol. 35鏈 928–938. Lord鏈 C.J.鏈 Aahwor,h鏈 A.鏈 2013. Mechanvama oa reava,ance ,o ,herapvea ,arge,vng BRCA-mu,an, cancera. Na,. Med. 19鏈 1381–1388. Lord鏈 C.J.鏈 Aahwor,h鏈 A.鏈 2012. The DNA damage reaponae and cancer ,herapk. Na,ure 481鏈 287–294. h,,pa://dov.org/10.1038/na,ure10760 Luoh鏈 S.-W.鏈 2002. Amplvfca,von and expreaavon oa genea arom ,he 17q11 approxvma,elk q12 amplvcon vn breaa, cancer cella. Cancer Gene,. Ck,ogene,. 136鏈 43–47. Ma鏈 C.X.鏈 Cav鏈 S.鏈 Lv鏈 S.鏈 Rkan鏈 C.E.鏈 Guo鏈 Z.鏈 Schavaa鏈 W.T.鏈 Lvn鏈 L.鏈 Hoog鏈 J.鏈 Govaaon鏈 R.J.鏈 Pra,鏈 A.鏈 Aa,鏈 R.L.鏈 Ellva鏈 M.J.鏈 Pvwnvca-Worma鏈 H.鏈 2012. Targe,vng Chá1 vn p53-defcven, ,rvple-nega,vve breaa, cancer va ,herapeu,vcallk benefcval vn human-vn-mouae ,umor modela. J Clvn Invea, 122鏈 1541–1552. Ma鏈 C.X.鏈 Ellva鏈 M.J.C.鏈 Pe,ronv鏈 G.R.鏈 Guo鏈 Z.鏈 Cav鏈 S.-R.鏈 Rkan鏈 C.E.鏈 Cravg Locáhar,鏈 A.鏈 Naugh,on鏈 M.J.鏈 Pluard鏈 T.J.鏈 Brenvn鏈 C.M.鏈 Pvcua鏈 J.鏈 Creeámore鏈 A.N.鏈 Mwandoro鏈 T.鏈 Yarde鏈 E.R.鏈 Reed鏈 J.鏈 Ebber,鏈 M.鏈 Bernard鏈 P.S.鏈 Wa,aon鏈 M.鏈 Dokle鏈 L.A.鏈 Dancek鏈 J.鏈 Pvwnvca-Worma鏈 H.鏈 Fracaaao鏈 P.M.鏈 2013. A phaae II

49 a,udk oa UCN-01 vn combvna,von wv,h vrvno,ecan vn pa,ven,a wv,h me,aa,a,vc ,rvple nega,vve breaa, cancer. Breaa, Cancer Rea. Trea,. 137鏈 483–492. Malumbrea鏈 M.鏈 2014. Ckclvn-dependen, ávnaaea. Genome Bvol. 15鏈 122. McNeelk鏈 S.鏈 Becámann鏈 R.鏈 Bence Lvn鏈 A.K.鏈 2014. CHEK agavn: revvav,vng ,he developmen, oa CHK1 vnhvbv,ora aor cancer ,herapk. Pharmacol. Ther. 142鏈 1–10. Mer,vna鏈 P.鏈 Manv鏈 D.R.鏈 Rugglea鏈 K.V.鏈 Gvlle,,e鏈 M.A.鏈 Clauaer鏈 K.R.鏈 Wang鏈 P.鏈 Wang鏈 X.鏈 Qvao鏈 J.W.鏈 Cao鏈 S.鏈 Pe,ralva鏈 F.鏈 Kawaler鏈 E.鏈 Mund,鏈 F.鏈 Krug鏈 K.鏈 Tu鏈 Z.鏈 Lev鏈 J.T.鏈 Ga,a鏈 M.L.鏈 Wvláeraon鏈 M.鏈 Perou鏈 C.M.鏈 Yellapan,ula鏈 V.鏈 Huang鏈 K.鏈 Lvn鏈 C.鏈 McLellan鏈 M.D.鏈 Yan鏈 P.鏈 Davvea鏈 S.R.鏈 Townaend鏈 R.R.鏈 Sáa,ea鏈 S.J.鏈 Wang鏈 J.鏈 Zhang鏈 B.鏈 Kvnavnger鏈 C.R.鏈 Mearv鏈 M.鏈 Rodrvgue鏈 H.鏈 Dvng鏈 L.鏈 Paulovvch鏈 A.G.鏈 Fenkö鏈 D.鏈 Ellva鏈 M.J.鏈 Carr鏈 S.A.鏈 NCI CPTAC鏈 2016. Pro,eogenomvca connec,a aoma,vc mu,a,vona ,o avgnallvng vn breaa, cancer. Na,ure 534鏈 55–62. Mv,a鏈 M.M.鏈 Jok鏈 A.A.鏈 Mv,a鏈 A.鏈 Sanáhala鏈 K.鏈 Jou鏈 Y.-M.鏈 Zhang鏈 D.鏈 S,a,áevvch鏈 P.鏈 Zhu鏈 Y.鏈 Yao鏈 S.-L.鏈 Small鏈 K.鏈 Bannerjv鏈 R.鏈 Shapvro鏈 C.L.鏈 2014. Randomved phaae II ,rval oa ,he ckclvn-dependen, ávnaae vnhvbv,or dvnacvclvb (MK-7965) veraua capecv,abvne vn pa,ven,a wv,h advanced breaa, cancer. Clvn. Breaa, Cancer 14鏈 169–176. Mon,ano鏈 R.鏈 Thompaon鏈 R.鏈 Chung鏈 I.鏈 Hou鏈 H.鏈 Khan鏈 N.鏈 Eaa,man鏈 A.鏈 2013. Senav,va,von oa human cancer cella ,o gemcv,abvne bk ,he Chá1 vnhvbv,or MK-8776: cell ckcle per,urba,von and vmpac, oa admvnva,ra,von achedule vn vv,ro and vn vvvo. BMC Cancer 13鏈 604. Nah,a鏈 R.鏈 Yu鏈 D.鏈 Hung鏈 M.-C.鏈 Hor,obagkv鏈 G.N.鏈 Ea,eva鏈 F.J.鏈 2006. Mechanvama oa Dvaeaae: undera,andvng reava,ance ,o HER2-,arge,ed ,herapk vn human breaa, cancer. Na,ure Revvewa Clvnvcal Oncologk 3鏈 269–280. O’Connell鏈 B.C.鏈 Adamaon鏈 B.鏈 Lkdeard鏈 J.R.鏈 Sowa鏈 M.E.鏈 Cvccva鏈 A.鏈 Bredemeker鏈 A.L.鏈 Schlabach鏈 M.鏈 Gkgv鏈 S.P.鏈 Elledge鏈 S.J.鏈 Harper鏈 J.W.鏈 2010. A genome- wvde camp,o,hecvn aenav,vvv,k acreen vden,vfea a mammalvan MMS22L- NFKBIL2 complex requvred aor genomvc a,abvlv,k. Mol. Cell 40鏈 645–657. Ol,ean鏈 S.鏈 Ba,ea鏈 D.O.鏈 2014. Hallmaráa oa al,erna,vve aplvcvng vn cancer. Oncogene 33鏈 5311–5318. Paparvdva鏈 N.F.D.S.鏈 Durvale鏈 M.C.鏈 Candurv鏈 F.鏈 2017. The emergvng pvc,ure oa CDK9/P-TEFb: more ,han 20 keara oa advancea avnce PITALRE. Mol Bvoaka, 13鏈 246–276. Parrk鏈 D.鏈 Guv鏈 T.鏈 Shanahan鏈 F.鏈 Davva鏈 N.鏈 Prabhavaláar鏈 D.鏈 Wvawell鏈 D.鏈 Seghev鏈 W.鏈 Paruch鏈 K.鏈 Dwker鏈 M.P.鏈 Doll鏈 R.鏈 Nomevr鏈 A.鏈 Wvndaor鏈 W.鏈 Fvachmann鏈 T.鏈 Wang鏈 Y.鏈 Oa,鏈 M.鏈 Chen鏈 T.鏈 Kvrachmever鏈 P.鏈 Leea鏈 E.M.鏈 2010. Dvnacvclvb (SCH 727965)鏈 a novel and po,en, ckclvn-dependen, ávnaae vnhvbv,or. Mol. Cancer Ther. 9鏈 2344–2353. Pearl鏈 L.H.鏈 Schver鏈 A.C.鏈 Ward鏈 S.E.鏈 Al-Laváanv鏈 B.鏈 Pearl鏈 F.M.G.鏈 2015. Therapeu,vc oppor,unv,vea wv,hvn ,he DNA damage reaponae. Na,. Rev. Cancer 15鏈 166–180. Pere鏈 R.P.鏈 Lewva鏈 L.D.鏈 Beelen鏈 A.P.鏈 Olaanaáv鏈 A.J.鏈 Johna,on鏈 N.鏈 Rhodea鏈 C.H.鏈 Beaulveu鏈 B.鏈 Erna,oaa鏈 M.S.鏈 Eaa,man鏈 A.鏈 2006. Modula,von oa cell ckcle progreaavon vn human ,umora: a pharmacoávne,vc and ,umor molecular pharmacodknamvc a,udk oa cvapla,vn plua ,he Chá1 vnhvbv,or UCN-01 (NSC 638850). Clvn. Cancer Rea. 12鏈 7079–7085. Pverce鏈 A.J.鏈 Johnaon鏈 R.D.鏈 Thompaon鏈 L.H.鏈 Jaavn鏈 M.鏈 1999. XRCC3 promo,ea homologk-dvrec,ed repavr oa DNA damage vn mammalvan cella. Genea Dev. 13鏈 2633–2638. Revd鏈 B.M.鏈 Permu,h鏈 J.B.鏈 Sellera鏈 T.A.鏈 2017. Epvdemvologk oa ovarvan cancer: a revvew. Cancer Bvol Med 14鏈 9–32.

50 Rodrvguea鏈 F.鏈 Thuma鏈 L.鏈 Klgmb,鏈 C.鏈 2012. The regula,von oa glval-apecvfc aplvcvng oa Neurexvn IV requvrea HOW and Cdá12 ac,vvv,k. Developmen, 139鏈 1765– Sedelnváova鏈 O.A.鏈 Rogaáou鏈 E.P.鏈 Panku,vn鏈 I.G.鏈 Bonner鏈 W.M.鏈 2002. Quan,v,a,vve de,ec,von oa (125)IdU-vnduced DNA double-a,rand breaáa wv,h gamma- H2AX an,vbodk. Radva,. Rea. 158鏈 486–492. Shah鏈 S.P.鏈 Ro,h鏈 A.鏈 Goka鏈 R.鏈 Oloumv鏈 A.鏈 Ha鏈 G.鏈 Zhao鏈 Y.鏈 Turaahvvlv鏈 G.鏈 Dvng鏈 J.鏈 Tae鏈 K.鏈 Haaaarv鏈 G.鏈 Baahaaha,v鏈 A.鏈 Pren,vce鏈 L.M.鏈 Kha,,ra鏈 J.鏈 Burlevgh鏈 A.鏈 Yap鏈 D.鏈 Bernard鏈 V.鏈 McPheraon鏈 A.鏈 Shumanaák鏈 K.鏈 Crvaan鏈 A.鏈 Gvulvank鏈 R.鏈 Heravv-Mouaaavv鏈 A.鏈 Roaner鏈 J.鏈 Lav鏈 D.鏈 Bvrol鏈 I.鏈 Varhol鏈 R.鏈 Tam鏈 A.鏈 Dhalla鏈 N.鏈 Wa,aon鏈 P.鏈 Gelmon鏈 K.鏈 Chva鏈 S.鏈 Chvn鏈 S.-F.鏈 Cur,va鏈 C.鏈 Rueda鏈 O.M.鏈 Pharoah鏈 P.D.鏈 Damaraju鏈 S.鏈 Macáek鏈 J.鏈 Hoon鏈 K.鏈 Harávna鏈 T.鏈 Tadvgo,la鏈 V.鏈 Svgaroudvnva鏈 M.鏈 Gaacard鏈 P.鏈 Tla,k鏈 T.鏈 Coa,ello鏈 J.F.鏈 Meker鏈 I.M.鏈 Eavea鏈 C.J.鏈 Waaaerman鏈 W.W.鏈 Jonea鏈 S.鏈 Hun,aman鏈 D.鏈 Hvra,鏈 M.鏈 Caldaa鏈 C.鏈 Marra鏈 M.A.鏈 Aparvcvo鏈 S.鏈 2012. The clonal and mu,a,vonal evolu,von apec,rum oa prvmark ,rvple-nega,vve breaa, cancera. Na,ure 486鏈 395–399. Svrcoulomb鏈 F.鏈 Beáhouche鏈 I.鏈 Fvne,,v鏈 P.鏈 Adélaïde鏈 J.鏈 Ben Hamvda鏈 A.鏈 Bonanaea鏈 J.鏈 Raknaud鏈 S.鏈 Innocen,v鏈 C.鏈 Charaae-Jauaare,鏈 E.鏈 Tarpvn鏈 C.鏈 Ben Aked鏈 F.鏈 Vvena鏈 P.鏈 Jacquemver鏈 J.鏈 Ber,uccv鏈 F.鏈 Bvrnbaum鏈 D.鏈 Chaaaane,鏈 M.鏈 2010. Genome proflvng oa ERBB2-amplvfed breaa, cancera. BMC Cancer 10鏈 539. Sørenaen鏈 C.S.鏈 Skljuåaen鏈 R.G.鏈 2012. Saaeguardvng genome vn,egrv,k: ,he checápovn, ávnaaea ATR鏈 CHK1 and WEE1 rea,ravn CDK ac,vvv,k durvng normal DNA replvca,von. Nuclevc Acvda Rea. 40鏈 477–486. Taglvala,ela鏈 A.鏈 2012. CDK12 va a novel oncogene wv,h clvnvcal and pa,hogene,vc relevance vn breaa, cancer. Dvaaer,a,von ,heava鏈 Unvverav,k oa Mvlan Thompaon鏈 R.鏈 Eaa,man鏈 A.鏈 2013. The cancer ,herapeu,vc po,en,val oa Chá1 vnhvbv,ora: how mechanva,vc a,udvea vmpac, on clvnvcal ,rval deavgn. Br J Clvn Pharmacol 76鏈 358–369. Tven鏈 J.F.鏈 Maloomvan鏈 A.鏈 Cheng鏈 S.-W.G.鏈 Hughea鏈 C.S.鏈 Chow鏈 C.C.T.鏈 Canapv鏈 L.T.鏈 Oloumv鏈 A.鏈 Trvgo-Gonale鏈 G.鏈 Baahaaha,v鏈 A.鏈 Xu鏈 J.鏈 Chang鏈 V.C.-D.鏈 Shah鏈 S.P.鏈 Aparvcvo鏈 S.鏈 Morvn鏈 G.B.鏈 2017. CDK12 regula,ea al,erna,vve laa, exon mRNA aplvcvng and promo,ea breaa, cancer cell vnvaavon. Nuclevc Acvda Rea. 45鏈 6698–6716. Tvlek鏈 L.S.鏈 Madore鏈 S.J.鏈 Malvm鏈 M.H.鏈 Cullen鏈 B.R.鏈 1992. The VP16 ,ranacrvp,von ac,vva,von domavn va aunc,vonal when ,arge,ed ,o a promo,er-proxvmal RNA aequence. Genea Dev. 6鏈 2077–2087. Wang鏈 F.-Z.鏈 Fev鏈 H.鏈 Cuv鏈 Y.-J.鏈 Sun鏈 Y.-K.鏈 Lv鏈 Z.-M.鏈 Wang鏈 X.-Y.鏈 Yang鏈 X.-Y.鏈 Zhang鏈 J.-G.鏈 Sun鏈 B.-L.鏈 2014. The checápovn, 1 ávnaae vnhvbv,or LY2603618 vnducea cell ckcle arrea,鏈 DNA damage reaponae and au,ophagk vn cancer cella. Apop,oava 19鏈 1389–1398. Wang鏈 Y.鏈 Zhang鏈 T.鏈 Kwva,áowaáv鏈 N.鏈 Abraham鏈 B.J.鏈 Lee鏈 T.I.鏈 Xve鏈 S.鏈 Yuugullu鏈 H.鏈 Von鏈 T.鏈 Lv鏈 H.鏈 Lvn鏈 Z.鏈 S,over鏈 D.G.鏈 Lvm鏈 E.鏈 Wang鏈 Z.C.鏈 Iglehar,鏈 J.D.鏈 Young鏈 R.A.鏈 Grak鏈 N.S.鏈 Zhao鏈 J.J.鏈 2015. CDK7-dependen, ,ranacrvp,vonal addvc,von vn ,rvple-nega,vve breaa, cancer. Cell 163鏈 174–186. Yu鏈 M.鏈 Yang鏈 W.鏈 Nv鏈 T.鏈 Tang鏈 Z.鏈 Naáadav鏈 T.鏈 Zhu鏈 J.鏈 Roeder鏈 R.G.鏈 2015. RNA polkmeraae II-aaaocva,ed aac,or 1 regula,ea ,he releaae and phoaphorkla,von oa pauaed RNA polkmeraae II. Scvence 350鏈 1383–1386. Zhang鏈 T.鏈 Kwva,áowaáv鏈 N.鏈 Olaon鏈 C.M.鏈 Dvxon-Claráe鏈 S.E.鏈 Abraham鏈 B.J.鏈 Grevaenberg鏈 A.K.鏈 Fvcarro鏈 S.B.鏈 Elávna鏈 J.M.鏈 Lvang鏈 Y.鏈 Hanne,,鏈 N.M.鏈 Man鏈 T.鏈 Hao鏈 M.鏈 Bar,áowvaá鏈 B.鏈 Greenleaa鏈 A.L.鏈 Mar,o鏈 J.A.鏈 Geker鏈 M.鏈 Bullocá鏈 A.N.鏈 Young鏈 R.A.鏈 Grak鏈 N.S.鏈 2016. Covalen, ,arge,vng oa remo,e cka,evne reavduea ,o develop CDK12 and CDK13 vnhvbv,ora. Na,. Chem. Bvol. 12鏈 876– 884.

51 6. List of abbreviations

CDK – ckclvn-dependen, ávnaae ChIP – chroma,vn vmmunoprecvpv,a,von ChIP – chroma,vn vmmunoprecvpv,a,von - aequencvng CTD – C-,ermvnal domavn DDR – DNA damage reaponae EGFP – enhanced green [uoreacen, pro,evn HGSOC – hvgh-grade aeroua ovarvan cancer HR – homologoua reconbvna,von p-TEFb - ,he poav,vve ,ranacrvp,von elonga,von aac,or Pol II – RNA polkmeraae II RT-qPCR - Quan,v,a,vve reverae ,ranacrvp,von polkmeraae chavn reac,von SD – a,andard devva,von Ser2 – aervne 2 ahRNA - ahor, havrpvn RNA avRNA – ahor, vn,eraervng RNA SLIIB - ,he a,em-loop IIB TCGA - The Cancer Genome A,laa Reaearch Ne,worá TFIIH - Tranacrvp,von aac,or II Human

52 7. Appendices

1) Eáumv鏈 K.M.鏈 Paculova, H.鏈 Lenaav鏈 T.鏈 Poapvchalova鏈 V.鏈 Böaáen鏈 C.A.鏈 Rkbarváova鏈 J.鏈 Brkja鏈 V.鏈 Geker鏈 M.鏈 Blaeá鏈 D.鏈 Barborvc鏈 M.鏈 2015. Ovarvan carcvnoma CDK12 mu,a,vona mvaregula,e expreaavon oa DNA repavr genea vva defcven, aorma,von and aunc,von oa ,he Cdá12/CkcK complex. Nuclevc Acvda Rea. 43鏈 2575–2589.

Peraonal con,rvbu,von: Genera,von oa ,wo ae,a oa nvne cell lvnea expreaavng mu,a,ed CDK12 aorma鏈 RT-PCR analkava and wea,ern blo,,vng鏈 da,a analkava and vn,erpre,a,von鏈 manuacrvp, prepara,von.

2) Paculová, H.鏈 Kramara鏈 J.鏈 Švmečáová鏈 Š.鏈 Fedr鏈 R.鏈 Součeá鏈 K.鏈 Hklae鏈 O.鏈 Paruch鏈 K.鏈 Svoboda鏈 M.鏈 Mva,rá鏈 M.鏈 Kohou,eá鏈 J.鏈 2017. BRCA1 or CDK12 loaa aenav,vea cella ,o CHK1 vnhvbv,ora. Tumour Bvol. 39鏈 1010428317727479.

Peraonal con,rvbu,von: Deavgn oa ,he a,udk鏈 collec,von and aaaemblk oa da,a (prolvaera,von aaaak鏈 wea,ern blo,,vng鏈 cell ckcle analkava鏈 clonogenvc aaaak鏈 prepared cella aor ,he xenograa, expervmen,)鏈 da,a analkava and vn,erpre,a,von鏈 manuacrvp, prepara,von.

3) Paculová, H.鏈 Kohou,eá鏈 J.鏈 2017. The emergvng rolea oa CDK12 vn ,umorvgeneava. Cell Dvv 12鏈 7

Peraonal con,rvbu,von: Manuacrvp, prepara,von

53 Published online 20 February 2015 Nucleic Acids Research, 2015, Vol. 43, No. 5 2575–2589 doi: 10.1093/nar/gkv101 Ovarian carcinoma CDK12 mutations misregulate expression of DNA repair genes via deficient formation and function of the Cdk12/CycK complex

1, 2, 1, 3, Kingsley M. Ekumi †,HanaPaculova †,TinaLenasi ‡,VendulaPospichalova ‡,Christian A. Bosken¨ 4 ,JanaRybarikova2, Vitezslav Bryja3,5 ,MatthiasGeyer4 ,DaliborBlazek2,* and Matjaz Barboric1,*

1Institute of Biomedicine, Biochemistry and Developmental Biology, University of Helsinki, Helsinki FIN-00014, Finland, 2Central European Institute of Technology (CEITEC), Masaryk University, 62500 Brno, Czech Republic, 3Institute of Experimental Biology, Faculty of Science, Masaryk University, 61137 Brno, Czech Republic, 4Center of Advanced European Studies and Research, Group Physical Biochemistry, 53175 Bonn, Germany and 5Institute of Biophysics, Academy of Sciences of the Czech Republic, 61265 Brno, Czech Republic Downloaded from Received July 17, 2014; Revised January 5, 2015; Accepted January 30, 2015

ABSTRACT DNA repair pathways, leading to genomic instability underlying the genesis of the cancer. The Cdk12/CycK complex promotes expression of a http://nar.oxfordjournals.org/ subset of RNA polymerase II genes, including those INTRODUCTION of the DNA damage response. CDK12 is among only nine genes with recurrent somatic mutations in high- Gene transcription by RNA polymerase II (RNAPII) is grade serous ovarian carcinoma. However, the influ- a sophisticated process involving numerous factors that ence of these mutations on the Cdk12/CycK complex enable regulated progression of the polymerase through and their link to cancerogenesis remain ill-defined. sequential stages of the transcription cycle (1). Therein, the C-terminal domain (CTD) of the Rbp1 subunit of Here, we show that most mutations prevent forma- RNAPII, consisting of multiple heptapeptide repeats with by guest on May 25, 2015 tion of the Cdk12/CycK complex, rendering the ki- consensus sequence Tyr-Ser-Pro-Thr-Ser-Pro-Ser, under- nase inactive. By examining the mutations within the goes a dynamic cycle of post-translational modifcations Cdk12/CycK structure, we find that they likely pro- and cis/trans isomerizations, providing a platform for voke structural rearrangements detrimental to Cdk12 mRNA biogenesis and export factors (2). Among vari- activation. Our mRNA expression analysis of the pa- ous modifcations, Ser2 phosphorylation (Ser2-P) of the tient samples containing the CDK12 mutations re- CTD is most strongly linked to productive transcription veals coordinated downregulation of genes critical to and pre-mRNA processing, the steps following promoter- the homologous recombination DNA repair pathway. proximal pausing of RNAPII (3,4). In addition to the well- Moreover, we establish that the Cdk12/CycK com- established P-TEFb kinase, which consists of the catalytic plex occupies these genes and promotes phospho- Cdk9 and regulatory cyclin (Cyc) T subunits, recent ev- idence from fruit fy and human cells indicates that the rylation of RNA polymerase II at Ser2. Accordingly, Cdk12/CycK complex catalyzes the Ser2-P mark as well (5– we demonstrate that the mutant Cdk12 proteins fail 9). Likewise, Ctk1 and Lsk1, the yeast orthologs of Cdk12, to stimulate the faithful DNA double strand break are major Ser2-P kinases in Saccharomyces cerevisiae and repair via homologous recombination. Together, we Schizosaccharomyces pombe,respectively(10,11). Illustrat- provide the molecular basis of how mutated CDK12 ing the signifcance of this novel transcriptional kinase, ceases to function in ovarian carcinoma. We propose genetic inactivation of CycK in mice is embryonic lethal that CDK12 is a tumor suppressor of which the loss- (6), possibly due to the role of Cdk12/CycK in maintain- of-function mutations may elicit defects in multiple ing embryonic stem cell self-renewal (12). In contrast to the broad requirement for P-TEFb in transcription (13,14),

*To whom correspondence should be addressed. Tel: +358 440 525 903; Fax: +358 919 125 444; Email: [email protected] Correspondence may also be addressed to Joint Senior Author Dalibor Blazek. Tel: +420 730 588 450; Fax: +420 549 497 564; Email: dali- [email protected] †The authors wish it to be known that, in their opinion, the frst two authors should be regarded as Joint First Authors. ‡The authors wish it to be known that, in their opinion, these authors should be regarded as Joint Third Authors.

C The Author(s) 2015. Published by Oxford University Press on behalf of Nucleic Acids Research. ⃝ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected] 2576 Nucleic Acids Research, 2015, Vol. 43, No. 5 depletion of Cdk12/CycK in human cells leaves expres- MATERIALS AND METHODS sion of most genes unaltered (6). However, a prominent Cell culture group among the downregulated genes is the one direct- ing DNA-damage response (DDR), which includes many HEK 293, HEK 293 Flp-In T-REx (Life Technologies), core players that ensure genomic stability, such as BRCA1, Caov-3 (ATCC) and HeLa DR-GFP cell lines were main- ATR, FANCI and FANCD2. Accordingly, knockdown of tained in Dulbecco’s Modifed Eagle’s Medium (DMEM) Cdk12/CycK induces DNA damage, as well as sensitivity supplemented with 10% fetal bovine serum (FBS) and 100 to various DNA damaging agents (6). This important func- U/ml penicillin/streptomycin. HCT116 cell line was grown tion of Cdk12/CycK seems to be evolutionarily conserved in DMEM supplemented with 5% FBS. HEK 293 Flp-In as mutations of CTK genes in S. cerevisiae prevent upreg- T-REx cell lines expressing 3X-FLAG peptide or Cdk12- ulation of several DNA repair genes, rendering yeast cells F proteins were generated according to the manufacturer’s incapacitated in the face of genotoxic insult (15,16). instructions (Life Technologies). All cell lines were main- Mutations in factors controlling transcription underlie tained at 37◦C with 5% CO2. Media and supplements were many diseases including cancer (17). While contribution from Sigma-Aldrich. of misregulated transcription elongation to tumorigenesis through P-TEFb has been documented (13,18,19), the rela- Plasmid DNAs and siRNAs tionship between mutations in Cdk12/CycK, altered gene expression and relevance to cancer has not been defned. The plasmids used in this study are listed in Supplementary Importantly, CDK12 is perturbed in several cancers, includ- Table S4. Rev-Cdk12 and Cdk12-F proteins encoded by Downloaded from ing breast, gastric and ovarian cancer. In breast cancer, the pRev-Cdk12 and pcDNA3.1-Cdk12-F expression CDK12 was found to be co-amplifed with ERBB2, the ty- plasmids, respectively, were described previously (6). rosine kinase receptor gene, which is deregulated in about Cdk12 #1 siRNA (sc-44343), CycK siRNA (sc-37600) 20% of breast tumors (20,21). Furthermore, out-of-frame and Cdk13 siRNA (sc-72835) were from Santa Cruz rearrangements of CDK12 were identifed in micropapillary Biotechnology. Cdk12 #2 siRNA has the sense sequence 5′- breast carcinoma, 13% of ERBB2 positive cancers (22) and rCrArGrArUrGrArCrCrCrUrUrGrArArGrCrUrUdTdT- http://nar.oxfordjournals.org/ gastric cancers (23). Finally, recent work by The Cancer 3′. Whereas Cdk12 #1 siRNA was used in Figure 5A and Genome Atlas (TCGA) on the high-grade serous ovarian B, and Supplementary Figure S7, Cdk12 #2 siRNA was carcinoma (HGS-OvCa) has provided the most compelling used in Figure 5C and Supplementary Figure S8. Control evidence for a possible role for mutated CDK12 in cancer siRNAs (sc-37007 and SIC001) were from Santa Cruz (24). Employing whole-exome DNA sequencing, the TCGA Biotechnology and Sigma-Aldrich, respectively. The plas- study reported a catalog of somatic gene mutations for 316 mids encoding the wild-type and mutant Cdk12 proteins HGS-OvCa tumor samples. Whereas TP53 dominated the that were used to generate the HEK 293 Flp-In T-REx mutation spectrum, CDK12 was identifed as one of only cell lines were mutated using QuikChange Lightning by guest on May 25, 2015 eight further genes with statistically recurrent somatic mu- Site-Directed Mutagenesis Kit (Agilent Technologies) to tations. Detailed re-analyses of the CDK12 mutations from become resistant to Cdk12 #2 siRNA using the following the TCGA work and the Catalog of somatic mutations in primer pair: 5′-CAGCAGAACAGACGACACTCGAGG cancer (COSMIC) database defned 7 out of 12 mutations CTTCAAGCACACCAG-3′;5′-CTGGTGTGCTTGAA as homozygous, highlighting CDK12 as a novel candidate GCCTCGAGTGTCGTCTGTTCTGCTG-3′. tumor suppressor in ovarian carcinoma (25). Importantly, approximately half of all HGS-OvCa cases display defects Immunoprecipitation assay and in vitro kinase assay in homologous recombination (HR) (24), the pathway that repairs DNA double-strand breaks (DSBs) most faithfully HEK 293 Flp-In T-REx cell lines expressing 3X-FLAG (26). Since depletion of Cdk12/CycK downregulates many and Cdk12-F proteins upon induction with 1 ␮g/␮l of components of DDR that function in HR (27), it is possible tetracycline for 24 h and transiently transfected HEK that CDK12 mutations in HGS-OvCa could be detrimental 293 cells (Supplementary Figure S2 and Figure 4)were to the effcacy of HR. lysed in buffer C containing 20 mM Tris–HCl (pH.7.9), Despite the available genetic evidence, the signifcance 150 mM NaCl, 10 mM KCl, 1.5 mM MgCl2, 1 mM of CDK12 mutations for the assembly and function of ethylenediaminetetraacetic acid (EDTA), 0.5% NP-40 and Cdk12/CycK remains to be defned. Furthermore, it is un- protease/phosphatase inhibitors (Pierce). Upon centrifu- clear how mutations in this novel transcriptional kinase gation at 20 000 g for 30 min at 4◦C, protein com- might infuence cancerogenesis. To address these questions, plexes containing Cdk12-F proteins were immuno-purifed we focused on CDK12 mutations identifed in HGS-OvCa. from whole cell extracts (WCEs) using 10 ␮l of packed Collectively, our results show that the CDK12 mutations im- FLAG-M2 agarose (Sigma-Aldrich) for 1 h at 4◦C. The pair the transcriptional role of Cdk12/CycK in DNA dam- immuno-purifed complexes were washed 3 with 1 ml of age repair by HR, elucidating an important link between buffer C, eluted by boiling in Sodium dodecyl× sulphate- the non-functional Cdk12 proteins and cancer. polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer, resolved by SDS-PAGE and examined by Western blotting using CycK (NBP1-06519, Novus Biologicals) and FLAG (F3165, Sigma-Aldrich) antibodies. For in vitro ki- nase assays, immuno-purifcations of Cdk12-F-containing complexes were performed as above with the following Nucleic Acids Research, 2015, Vol. 43, No. 5 2577 modifcations. Cell pellets were lysed in buffer B containing (sc-8408), Rad51D (sc-366363) and GAPDH (sc-32233) an- 50 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, tibodies from Santa Cruz Biotechnology. 1% NP-40 and protease/phosphatase inhibitors (Pierce). The complexes were washed two times with buffer B, fol- ChIP-qPCR assay lowed by fve washes with the kinase buffer A containing 50 mM Tris–HCl, 100 mM NaCl, 10 mM MgCl2 and 0.1% The assay was performed as described previously (28) NP-40. The beads with the purifed complexes were resus- with the following modifcations. Hela DR-GFP cells were pended in 30 ␮l of kinase buffer A supplemented with 100 plated in 15 cm plates, transfected with 400 pmol of Cdk12 mM adenosine triphosphate, 1 mM dithiothreitol (DTT) #2 or control siRNA using Lipofectamine RNAiMAX and 300 ng of recombinant Glutatione S-transferase (GST)- reagent (Life Technologies) at 70% confuency, plated in two CTD substrate (9) and incubated for 10 min at room tem- 15 cm plates 24 h later and grown for additional 24 h be- perature, followed by 1 h at 30◦C in Bioer mixing block MB- fore proceeding with the assay. Crude nuclear extract ob- 102 at 300 rpm. Reactions were stopped by boiling in SDS- tained from the confuent 15 cm plate was sonicated in 800 PAGE sample buffer, resolved by SDS-PAGE and analyzed ␮l of RIPA buffer (150 mM NaCl, 1% NP40, 0.5% DOC, by Western blotting using CycK (NBP1-06519, Novus Bio- 0.1% SDS, 50 mM Tris.Cl pH 8.0, 5 mM EDTA pH 8.0) logicals), Ser2-P RNAPII (ab5095, Abcam), FLAG (F3165, 10 for 11 s at output power of 11 Watt. Chromatin was Sigma-Aldrich) and GST (sc-138, Santa Cruz Biotechnol- clari×fed by centrifugation at 13 000 g for 15 min, 400 ␮l ogy) antibodies. of supernatant added to 12 ␮l of antibody-coupled pro- tein G Dynabeads (Life Technologies) and the mixture sup- Downloaded from RNA-tethering gene reporter assay plemented with additional 600 ␮l of RIPA buffer rotated overnight at 4 C. Before adding the chromatin, the beads HEK 293 cells were plated in 12-well format at 100 000 cells ◦ were pre-blocked with bovine serum albumin and salmon per well and transfected with 0.3 ␮g of pSLIIB-CAT and 1.5 sperm DNA overnight at a fnal concentration of 0.2 ␮g/␮l, ␮g of Rev-Cdk12, Rev or Cdk12-F expression plasmids as pre-incubated in 500 ␮lRIPAbufferfor4hwiththeanti- indicated by the FuGENE6 reagent (Roche) After 48 h, cells body and collected by magnet to remove the unbound anti- http://nar.oxfordjournals.org/ were lysed in 125 ␮l of lysis buffer (250 mM Tris–HCl, 0.5% body. We used 3 ␮g of Cdk12 (ab57311), 2 ␮g of total/Ser5- Triton-X, pH 7.5) for 30 min on ice. Finally, 100 ␮l of the P RNAPII (ab5408) and 1 ␮g of Ser2-P RNAPII (ab5095) heat-inactivated cell lysates was supplemented with 50 ␮l of antibodies from Abcam. Normal mouse (sc-2025) and rab- CAT assay buffer (0.5 ␮l of 3H-acetyl CoA (250 ␮Ci, Amer- bit (sc-2027) IgG control antibodies were from Santa Cruz sham), 30 ␮l of 1.66 mg/ml chloramphenicol, 5 ␮l of 250 Biotechnology. Generally, 1/60th of the precipitated ChIP mM Tris (pH 7.5), 14.5 ␮lddH2O)andCATactivitywas sample was used for each qPCR reaction. For input DNA, measured by a scintillation counter. For normalization, to- 2.5% of the cleared chromatin was used and 1/100th of tal protein content in whole-cell lysate was determined using the DNA dissolved in 100 ␮l of water was used for each Bio-Rad DC kit. Experiments were done twice in biologi- by guest on May 25, 2015 qPCR reaction. For sequences of the primers, see Supple- cal triplicates. Statistical analysis was performed using one- mentary Table S6. Samples were analyzed by the Strata- way ANOVA. Values are plotted as mean SD. P-value be- gene Mx3005P real-time polymerase chain reaction system tween groups is statistically signifcant (P±< 0.05). Levels of and FastStart Universal SYBR Green QPCR Master (Rox) Rev-Cdk12 chimeras were detected by Western blotting us- (Roche). Results were obtained from three independent ex- ing Cdk12 (sc-32643, Santa Cruz Biotechnology) antibody. periments. Values are plotted as mean of percentage of input DNA SD. RNAi, RT-qPCR and Western blotting ± Thirty to ffty percent confuent Caov-3 or HCT116 cells DR-GFP assay were transfected with 10 pmol of control (sc-37007), Cdk12 #1, Cdk12 #2, CycK or Cdk13 siRNA using Lipofectamine Hela DR-GFP cells were plated in 6 cm plates at 800 RNAiMax reagent (Life Technologies). Total RNA was iso- 000 and transfected with 100 pmol of control (SIC001) or lated using RNAzol (Molecular Research Center) 72 h post- Cdk12 #2 siRNA using Lipofectamine RNAiMAX reagent transfection. Reverse transcription was performed with M- (Life Technologies). Twenty-four hours after siRNA trans- MLV reverse transcriptase system using random hexamers fection, the cells were co-transfected with 2 ␮g of I-SceI (Life Technologies). For qPCR, 1 ␮l of cDNA was mixed and 3 ␮g of Cdk12-F expression vectors or a parental with 2 Light Cycler 480 Sybr Green Master (Roche) and pcDNA5/FRT/TO/3XFLAG vector as indicated by us- 0.3 ␮M× forward and reverse primer (for sequences of the ing Lipofectamine 3000 reagent (Life Technologies). Af- primers, see Supplementary Table S5) in a fnal volume of ter 48 h, the cells were trypsinized, collected by centrifu- 20 ␮l. Amplifcations were run on the Light Cycler 480II gation at 300 x g for 5 min, resuspended in 500 ␮l of (Roche) using the following cycle conditions: 94◦C for 2 DMEM and analyzed for GFP fuorescence using BD Ac- min (1 cycle); 95◦Cfor30s,60◦Cfor30s,72◦Cfor30s curi fow cytometer. Experiments were performed in biolog- (45 cycles). HPRT expression levels were used for normal- ical triplicates and statistical analysis was done using one- ization and gene expression differences were calculated us- way ANOVA. The error bars represent the mean SD. P- ing the threshold cycle method. Values are plotted as mean value between groups is statistically signifcant (P±< 0.05). SD. Levels of the analyzed proteins in control, Cdk12, Levels of Cdk12-F proteins were detected by Western blot- CycK± and Cdk13 siRNA-treated HCT116 cells were deter- ting using Cdk12 (sc-32643, Santa Cruz Biotechnology) an- mined by Western blotting using ATM (sc-23921), Chek1 tibody. 2578 Nucleic Acids Research, 2015, Vol. 43, No. 5

Statistical analysis of TCGA study Whereas the deletions of the Cdk12 KD brought about by the former mutations are expected to inactivate Cdk12, it is The expression levels of DDR genes in TCGA were the evolutionary conservation of the amino acids altered by downloaded from the Oncomine database (TCGA Ovarian the latter mutations that likely refects their functional im- dataset). The information about CDK12 mutation status portance. In fact, except for the K975 residue which is con- was obtained from the COSMIC database. Data from these served from fruit fies to humans, the R882, Y901 and L996 repositories was retrieved on February 26, 2014. Samples residues are all conserved from budding yeast to humans for which CDK12 mutation status was not available in (Supplementary Figure S1). To get a structural perspective COSMIC were discarded from the TCGA Ovarian dataset on these mutations, we took advantage of the crystal struc- similarly to the fve samples ‘variants of unknown origin’ ture of the human Cdk12/CycK complex encompassing the and ‘previously reported’ (1 sample) resulting in 462 HGS- KD of Cdk12 and the cyclin box domain of CycK (9). An OvCa samples. Out of these samples, there are 453 CDK12 important characteristic of this structure is that CycK in- wild-type samples, four samples with missense mutation in teracts only with the N-terminal lobe of the KD. However, CDK12 and fve samples with nonsense/indel mutation in upon positioning the CDK12 mutations on the structure, CDK12 (see also Supplementary Table S1). Data were ana- we found that the insertion and missense mutations are all lyzed with the Statistica software (Version 12.0) and visual- located within the C-terminal lobe of the KD (Figure 1B). ized using GraphPad Prism 5. Results are expressed as mean This observation suggested that possible effects of the mu- with standard error of the mean. The Shapiro-Wilk test tations on the Cdk12/CycK complex could be exerted via was used to assess the normal distribution of the variables allosteric mechanisms. (P > 0.05). Data without a normal distribution were ana- Downloaded from Activation of any Cdk is a two-step process wherein the lyzed with non-parametric test (Mann–Whitney test), and binding between the Cdk KD and the cyclin reorients the data with a normal distribution were analyzed with para- catalytic cleft of the kinase, setting the stage for its full acti- metric test (unpaired t-test). Unpaired t-test with Welsch’s vation by T loop phosphorylation (29,30). Thus, we next correction was applied for variables with signifcantly differ- asked whether the CDK12 mutations identifed in HGS- ent variances (F-test, P < 0.05). The statistical analysis was OvCa affect the interaction between Cdk12 and CycK (Fig- http://nar.oxfordjournals.org/ conducted at 95% confdence level. Symbols used to express ure 1C). Here, we employed the Flp-In T-REx system to statistical signifcance: non-signifcant (ns) P > 0.05; * generate stable HEK 293 cell lines each expressing the wild- P 0.05; ** P 0.01; *** P 0.001.= = ≤ = ≤ = ≤ type or mutant 3XFLAG epitope-tagged Cdk12 (Cdk12-F) proteins upon tetracycline induction. This system allows ac- Structural analyses curate comparison between the CDK12 mutations because the integration of individual Cdk12 expression vectors to Structural analysis of the CDK12 mutations was performed the defned locus ensures the same genetic background of on the human Cdk12/CycK structure determined at 2.2 A˚ different cell lines. Upon immuno-purifying the wild-type resolution (4NST, (9)). The molecular diagrams were drawn by guest on May 25, 2015 and mutant Cdk12-F proteins from the WCEs, we exam- with PyMOL (http://www.pymol.org/). ined the levels of endogenous CycK in these isolations by Western blotting (Figure 1C, bottom panels). As expected, RESULTS the wild-type Cdk12 interacted with CycK, and the mutant Cdk12 W719* and E928fs27* proteins failed to do so (Fig- Defective interaction between CycK and Cdk12 is the pre- ure 1C, lanes 1–4). In contrast, while the mutant Cdk12 dominant consequence of CDK12 mutations in HGS-OvCa R882L and K975E proteins displayed unperturbed CycK In this study, we focused our attention on nine validated binding, this interaction was greatly reduced or lost in the CDK12 mutations, which were identifed in the TCGA work case of the Y901C or L996F mutations, respectively (Fig- (24), including fve nonsense and insertion/deletion mu- ure 1C, lanes 5–8). We also asked whether the binding be- tations, as well as four missense mutations (Supplemen- tween Cdk12 and CycK is impacted by the panel of CDK12 tary Table S1 and Figure 1A). Among the nonsense muta- mutations found in lung, skin, ovarian, and large intestine tions, the L122fs4* and Q602* are the most drastic, since cancers (23,31,32) (Supplementary Figure S2). Of note, a they yield the mutant Cdk12 proteins without most pre- relative majority of all currently identifed CDK12 muta- dicted regions. Next, the nonsense W719* and the insertion tions in human cancers clusters within the KD, while the E928fs27* mutation terminate Cdk12 at the start and in the rest of them are located within other Cdk12 regions, such as middle of its kinase domain (KD), respectively. Finally, the those with many arginine/serine (RS) residues and proline- four missense mutations, including R882L, Y901C, K975E rich motifs (PRM) (Supplementary Table S2). Thus, we se- and L996F and an internal deletion T1014 Q1016! mu- lected S325N and S352F as the Cdk12 mutations represent- tation are all positioned within the KD of Cdk12. Of ing the RS region, and the G909R and E1024* mutations note, by inspecting the COSMIC database, we found that found within and right after the Cdk12 KD, respectively. among the nine HGS-OvCa patient cases with the mutated While the mutations within the RS region did not show any CDK12, each sample contains one CDK12 mutation (data defects in CycK binding, the mutant Cdk12 G909R protein not shown), providing the rationale for examining the im- lost the ability to bind CycK, and the E1024* mutation de- pact of single CDK12 mutations. creased this interaction considerably (Supplementary Fig- We analyzed W719* and E928fs27* as the two most rep- ure S2B, top panels, lanes 3–6). Collectively, these fndings resentative CDK12 nonsense and insertion mutations, as demonstrate that individual CDK12 mutations in HGS- well as all four CDK12 missense mutations (Figure 1A). OvCa and other cancers that delete or alter the KD can have Nucleic Acids Research, 2015, Vol. 43, No. 5 2579 Downloaded from http://nar.oxfordjournals.org/ by guest on May 25, 2015

Figure 1. CDK12 mutations in HGS-OvCa abrogate the activity of Cdk12 predominantly by impairing the interaction between Cdk12 and CycK. (A) Schematic depiction of the wild-type and mutant Cdk12 proteins containing individual CDK12 mutations analyzed in this study. Highly structured kinase domain (KD; red), arginine/serine rich region (RS; green) and two regions with proline-rich motifs (PRM1 and PRM2; blue) are depicted. The ruler on top indicates the length of Cdk12 protein in amino acids. Vertical lines denote sites of individual missense and insertion mutations. Finally, ‘fs’ stands for a frame-shift mutation and the associated number indicates the number of altered amino acids at the C-terminus of mutant Cdk12 protein (dotted). (B)OverallstructureofhumanCdk12/CycK and positions of the mutations. Cdk12 is shown as cartoon representation in blue and CycK as surface representation in grey. CDK12 missense, insertion and internal deletion mutations are located in the C-terminal lobe of the KD. The mutated amino acid residues are highlighted in red. pT893 is highlighted in orange. (C)EffectsoftheCDK12 mutations on the interaction between Cdk12 and CycK. The indicated wild-type and mutant FLAG epitope-tagged Cdk12 proteins (Cdk12-F) were immuno-purifed from whole cell extracts (WCEs) of the individual HEK 293 Flp-In T-Rex cell lines using FLAG-M2 agarose (FLAG IP) and examined for their interaction with endogenous CycK. Levels of Cdk12-F and CycK proteins in WCEs (INPUT, 5% of WCEs; top) and IPs (FLAG IP; bottom) were detected by Western blotting using FLAG and CycK antibodies. (D) CDK12 mutations abrogate the kinase activity of Cdk12. The indicated wild-type and mutant Cdk12-F proteins were immuno-purifed (IP) as in panel C and the complexes were examined for their kinase activity by in vitro kinase assay (IVKA) toward the recombinant GST-CTD. Levels of Ser2-P GST-CTD isoforms and input GST-CTD (30%) were detected by Western blotting using Ser2-P-specifcRNAPIIandGSTantibodies. detrimental effects on the formation of the Cdk12/CycK We immuno-purifed the wild-type and mutant Cdk12-F- complex. However, some mutations did not interfere with containing complexes from WCEs of the HEK 293 Flp- the binding between Cdk12 and CycK, the presumed oblig- In cell lines as above and used their comparable amounts atory step in the activation of Cdk12, raising a possibility in the kinase assays employing the recombinant GST-CTD that activity of those mutant Cdk12 proteins is not com- protein as a substrate (Figure 1D and Supplementary Fig- promised. ure S2C). Since Cdk12/CycK catalyzes the formation of Ser2-P in human cells (6–8), we followed kinase activity of the Cdk12-F-containing complexes by Western blotting CDK12 mutations in HGS-OvCa abrogate the kinase activ- using the antibody specifc for this RNAPII CTD mark. ity of Cdk12 Compared with our control immuno-purifcation, which To investigate the effects of the mutations on the catalytic was prepared using WCEs of HEK 293 Flp-In cell line ex- pressing 3XFLAG peptide, the wild-type Cdk12-F protein activity of Cdk12, we next performed in vitro kinase assays. 2580 Nucleic Acids Research, 2015, Vol. 43, No. 5 phosphorylated the GST-CTD chimera effectively (Fig- the binding of CycK to Cdk12 but did result in the inac- ure 1D, lanes 1–2). However, except for the mutant Cdk12 tive kinase, indicating that this mutation precludes the acti- K975E protein, the entire set of the proteins containing vation of Cdk12 following CycK binding. In fact, R882 is the CDK12 mutations failed to phosphorylate the GST- one of the three canonical arginine residues in Cdks that are CTD above background levels (Figure 1D, lanes 3–8). Us- coordinated upon activation by the phosphorylated thre- ing the same experimental approach, we also examined cat- onine 893 (pT893) in the T loop via salt bridges (9,29). alytic activity of the panel of Cdk12 proteins carrying the Subsequently, the arginine residues contact other Cdk12 mutations found in various cancers (Supplementary Figure and CycK groups, extending the organizing infuence of S2B). While we observed that the S325N and S352F muta- the phosphate group, which completes reorganization of the tions had no effect, the mutant Cdk12 G909R and E1024* catalytic cleft and substrate recognition site initiated by the proteins displayed impaired kinase activity (Supplementary cyclin binding. Given that the R882L mutation prevents Figure S2C, bottom panels, lanes 3–6). Thus, except for one, the contact to pT893, the subsequent series of events may all the CDK12 mutations that delete or alter the KD of not ensue, leading to inactive Cdk12 (Figure 2D). Finally, Cdk12 ablate the activity of the kinase. Whereas defcient K975 is located in the exposed region of the C-terminal CycK binding correlated completely with the loss of Cdk12 lobe, suggesting no structural role for this residue. That the activity, the R882L mutation inactivated the kinase with- K975E mutation did not show any effects on CycK bind- out impairing CycK binding, suggesting that a subsequent ing and Cdk12 activity but did compromise the function of defect prevents activation of this mutant Cdk12 protein. Cdk12/CycK in cells (see Figures 3, 4 and 6 below) may suggest the failure of our biochemical assays to reveal sub- Downloaded from tle defects of this mutation. Alternatively, K975 could play Insertion and missense CDK12 mutations in HGS-OvCa a stimulatory role in Cdk12-dependent gene expression via likely elicit structural defects that perturb formation and ac- a mechanism that has yet to be determined. tivity of the Cdk12/CycK complex We also attempted to shed light on how the G909R and These fndings prompted us to analyse why the CDK12 mu- E1024* mutations could lead to the defects in CycK bind- tations provoke such profound effects on the formation and ing and the activity of Cdk12. We found that the mutated http://nar.oxfordjournals.org/ activity of the Cdk12/CycK complex (Figure 2). Given that residues are located within the C-terminal lobe and right the mutant Cdk12 W719* protein lacks the entire KD, its after the end of the KD, respectively (Supplementary Fig- failure to bind CycK and phosphorylate the CTD is not sur- ure S3). We cannot offer a probable scenario for the effects prising. Even though we did not examine them experimen- of the G909R mutation. The E1024* mutation, however, tally, the same applies to the rest of the CDK12 nonsense truncates the C-terminal extension of the KD that other- mutations that generate more drastically truncated Cdk12 wise promotes the kinase activity of Cdk12 by stabilizing L122fs*4 and Q602* proteins. However, we were particu- the conformation of the KD (9). Thus, the E1024* muta- larly eager to understand those CDK12 mutations in HGS- tion may impair the latter, leading to the decreased CycK by guest on May 25, 2015 OvCa that alter the KD within its C-terminal lobe but still binding and diminished Cdk12 activity. Overall, we con- compromise the ability of Cdk12 to bind CycK or phospho- clude that the CDK12 mutations within the C-terminal lobe rylate the CTD of RNAPII (Figure 2A). of KD or right after it likely elicit conformational changes For these analyses, we again employed the structural in- in the KD. In turn, the CycK interacting surface in the N- formation of Cdk12/CycK (9). First, the E928fs27* muta- terminal lobe is altered, preventing the interaction between tion, which causes the appearance of 27 irregular residues Cdk12 and CycK, and the subsequent activation of the ki- before reaching the stop codon, truncates Cdk12 in the mid- nase. dle of the C-terminal lobe, most likely resulting in mis- folding of the remaining part of the KD including the N- CDK12 mutations in HGS-OvCa decrease transcriptional terminal lobe, leading to the observed defects in CycK bind- activation by Cdk12 ing and kinase activity. Next, the Y901C mutation substi- tutes the large aromatic residue located in the core of the The Cdk12/CycK complex is a transcriptional kinase, C-terminal lobe and beneath the catalytic center of the ki- which can stimulate expression of target genes (33,34). Since nase with a much smaller cysteine residue, creating a ‘void’ the CDK12 mutations led to such pronounced defects in the in the local environment and prohibiting multiple bonds of biochemical experiments, we next asked whether they also Y901 with the residues nearby (Figure 2B). In turn, the rear- compromise transcriptional activation by Cdk12 (Figure 3). rangements within the C-terminal lobe are likely to follow, Here, we employed a classical RNA tethering gene reporter affecting the conformation of the N-terminal lobe and thus assay, composed of the chimeric Rev-Cdk12 protein and leading to the decreased assembly and activity of this mu- the SLIIB-CAT reporter gene, which has been used to fol- tant Cdk12/CycK complex. Furthermore, the L996F mu- low transcriptional activation by Cdk12 (6) (Figure 3A). tation is located on helix ␣I in the core of the C-terminal Therein, the interaction between Rev and the SLIIB RNA lobe and surrounded by helices ␣E, ␣F and ␣J, forming a element of nascent transcripts tethers the Rev fusion pro- tight four helical bundle (Figure 2C). It is likely that the tein to RNAPII engaged in transcription, resulting in the spacious aromatic ring of this Cdk12 mutation demolishes activation of the SLIIB-CAT reporter gene (35). the helical assembly. In turn, this could provoke misfold- We frst determined whether transcriptional activation by ing of the C- as well as N-terminal lobes, rendering the N- Rev-Cdk12 requires the kinase activity of Cdk12. Hence, we terminal lobe refractory to CycK binding, thus inactivating constructed the mutant Cdk12 D877N protein, in which the the kinase. In contrast, the R882L mutation did not affect aspartic residue within the activation segment ‘DFG’ mo- Nucleic Acids Research, 2015, Vol. 43, No. 5 2581 Downloaded from

Figure 2. Structural analysis of the CDK12 mutations. (A)SchematicrepresentationofsecondarystructureelementswithintheKDofCdk12.Canonical ␣-helices and ␤-strands as well as boundaries of N- and C-terminal lobes are labeled. Positions of the mutated Cdk12 amino acids analyzed in this study http://nar.oxfordjournals.org/ are indicated on top of the schematic. (B)ThehydroxylgroupoftheY901aromaticringmediatesatighthydrogenbond(2.7A)˚ to the carboxyl group of E928. Interestingly, the tyrosine is closely surrounded by two cysteines, C862 and C924, which could make the Y901C mutation unfavorable. Hydrophobic contacts of Y901 are formed to I925 and K861, with P934 stabilizing the conformation of E928. (C)L996residesonhelix␣I. It is surrounded by T1014 and F1019 on helix ␣J, by C922, G923 and L926 of helix ␣F, and M840 and M844 of helix ␣E. The replacement of L996 by the spacious aromatic ring of a phenylalanine might preclude the tight assembly of the four helical bundle. (D) R882 is part of the canonical arginine network in CDKs that stabilize the T-loop upon phosphorylation of the critical threonine. Together with R858 of the HRD motif, R882 mediates salt bridges to the phosphate group of pT893. Its mutation to leucine might prohibit the conformational arrangement of the T loop required for full activation of the kinase. tif was substituted with asparagine, the mutation which can come more active at stimulating gene expression compared generate a kinase-dead form of Cdk (36). Indeed, while the to the same regions that reside in the context of the full- by guest on May 25, 2015 mutation did not affect the interaction between Cdk12 and length Cdk12 protein. CycK, it did abolish the activity of Cdk12 in vitro (Supple- Using this reporter assay, we also examined transcrip- mentary Figure S4A, lanes 1–3). Importantly, it also led to a tional properties of the mutant Cdk12 proteins carrying signifcant decrease in the ability of the chimeric Rev-Cdk12 the CDK12 mutations found in other cancers (Supplemen- D877N protein to activate transcription (Figure 3B, bars 3 tary Figure S4B). Again, all mutant proteins displayed the and 10), setting the stage for examining the effects of the decreased activation of the reporter gene (Supplementary CDK12 mutations. Figure S4B, bars 3–6). Although the activity of the mu- We hypothesized that the CDK12 mutations in HGS- tant Rev-Cdk12 G909R and E1024* proteins was compro- OvCa, which abrogated the formation of the Cdk12/CycK mised the most, the S325N and S352F mutations also de- complex and/or activity of the kinase, would hamper the creased the activity considerably, arguing for a contributing ability of the Rev-Cdk12 chimera to activate the reporter role for the RS region in the stimulation of gene expression gene expression effectively. Indeed, the Rev-Cdk12 chimeric by Cdk12/CycK. Collectively, these fndings demonstrate proteins containing all these CDK12 mutations showed that CDK12 mutations found in HGS-OvCa and other can- compromised activation of the reporter gene compared to cers compromise the ability of Cdk12 to promote gene ex- the activity of the wild-type Rev-Cdk12 chimera (Figure 3B, pression, raising the possibility of decreased transcription bars 3–7 and 9), correlating with our biochemical fndings. of Cdk12-dependent target genes in cancer cells. The mutant Rev-Cdk12 chimera harboring the K975E mu- tation, which did not affect CycK binding and activity of Cdk12, also failed to activate the reporter gene expression HGS-OvCa patient samples with CDK12 mutations exhibit strongly (Figure 3B, bar 8). This result suggests that the downregulation of genes of the HR repair pathway K975E mutation may indeed interfere with an important In a previous study that used a non-ovarian HeLa cell line event, which is in addition to the kinase activity required and RNAi approach, the Cdk12/CycK complex has been for the transcriptional activation by Cdk12/CycK. Of note, found to promote the expression of 30 DDR genes (6). among the mutant proteins, the Rev-Cdk12 719* chimera Among these were key effectors in∼ DNA repair mecha- displayed the highest activity (Figure 3B, bar 4). This result nisms, including BRCA1, ATR, FANCD2 and FANCI pro- could refect that functional regions that remained in this teins, implicating the Cdk12/CycK complex in the mainte- truncated protein, such as the RS region (see below), be- nance of genomic stability. However, whether the CDK12 2582 Nucleic Acids Research, 2015, Vol. 43, No. 5

mutations affect mRNA levels of these and additional DDR genes in HGS-OvCa remains an open question. To ad- dress this issue, we took advantage of the mRNA expres- sion data sets of the HGS-OvCa patient samples reported by the TCGA study (24). We compared gene expression lev- els between the samples that harbor either the wild-type CDK12 or CDK12 with the nine validated mutations (Fig- ure 4). Given that defects within the HR pathway occur in approximately half of all the HGS-OvCa cases (24), and that several identifed Cdk12-dependent DDR genes are re- quired for HR (37), we hypothesized that the CDK12 mu- tations in HGS-OvCa may trigger downregulation of genes of this most faithful DNA repair pathway. Our compari- son of mRNA levels of genes encoding the known com- ponents of HR (for the list of thirty eight examined genes, see Supplementary Table S3) identifed six genes, including ATM, ATR, CHEK1, FANCI, MDC1 and RAD51D, which were signifcantly downregulated (P < 0.05) in tumor sam- ples with mutated CDK12 in comparison to the samples Downloaded from carrying the wild-type CDK12 (Figure 4A, Supplementary Figure S5A, and Supplementary Figure S6). Among these, only ATR and FANCI were defned previously as Cdk12- dependent (6). To our surprise, we found that expression of BRCA1, a key effector of HR and a known Cdk12-

dependent gene, as well as of BRCA2, another important http://nar.oxfordjournals.org/ HR player, was not affected in the HGS-OvCa samples with the CDK12 mutations (Figure 4C). Of note, due to the ab- sence of the array probes for FANCD2 and MMS22L,two known Cdk12-dependent genes implicated in the HR path- way, we were unable to determine whether their expression was compromised by the CDK12 mutations. In addition to the HR genes, depletion of Cdk12/CycK led to decreased expression of many other DDR genes (6). by guest on May 25, 2015 To determine whether this group of genes is also affected by the CDK12 mutations, we analyzed their mRNA levels as above (for the list of twenty three examined genes, see Supplementary Table S3). We found three DDR genes un- related to HR, including NEK9, ORC3L and TERF2, to be signifcantly downregulated (P < 0.05) in the patient sam- Figure 3. CDK12 mutations in HGS-OvCa decrease transcriptional acti- ples with mutated CDK12 (Figure 4B, Supplementary Fig- vation by Cdk12. (A)SchematicdepictionoftheheterologousRNAteth- ure S5B, and Supplementary Figure S6). ering assay. Plasmid reporter pSLIIB-CAT contains modifed HIV1-LTR To confrm that expression of the previously charac- promoter in which the apical region of transactivation response RNA el- terized and the new HR genes identifed above requires ement (TAR) was substituted with 29-nucleotide stem-loop IIB (SLIIB) / subdomain of the HIV-1 Rev response element (RRE), the Rev bind- Cdk12 CycK in human cells of ovarian origin, we lowered ing RNA sequence. The interaction between Rev (pink oval) and SLIIB CDK12 mRNA levels by RNAi in the ovarian carcinoma within the TAR/SLIIB stem-loop structure at the 5′ end of nascent RNA Caov-3 cell line. Indeed, compared to cells treated with tethers the Rev-Cdk12 chimeric protein to RNAPII engaged in transcrip- a non-targeting control short interfering RNA (siRNA), tion, resulting in elevated transcription of CAT reporter gene. CTD of the cells treated with an siRNA specifctoCDK12 displayed biggest RNAPII subunit Rbp1 is represented as a tail of RNAPII (yellow), wherein white circle depicts Ser2 residue to be phosphorylated by Rev- decreased expression of BRCA1, FANCI, FANCD2 and Cdk12 (dashed arrow) and gold circles depict Ser5 and Ser7 residues in ATR, the HR genes previously characterized as Cdk12- an already phosphorylated form. Arrow within HIV1-LTR indicates tran- dependent (6) (Figure 4D). We observed these effects in scription start site. (B) CDK12 HGS-OvCa mutations compromise stimu- the colon cancer HCT116 cell line as well (Supplemen- lation of transcription by Rev-Cdk12. HEK 293 cells were co-transfected with pSLIIB-CAT reporter gene and plasmids encoding the proteins indi- tary Figure S7A). Furthermore, mRNA levels of ATM, cated below the graph. Transcriptional activities of Cdk12-F (white bar 1), CHEK1, MDC1 and RAD51D, the downregulated HR Rev (white bar 2), the mutant Rev-Cdk12 chimeras (red bars) and catalyti- genes in the HGS-OvCa cases with mutated CDK12 that cally dead Rev-Cdk12 D887N chimera (green bar) are represented as CAT have not been previously defned as Cdk12-dependent, also activities relative to the activity of wild-type Rev-Cdk12 chimera (blue decreased upon the knockdown of CDK12 in both Caov-3 bar), which was set to 100%. Results are presented as the mean SD. Lev- els of the Rev-Cdk12 chimeras and endogenous Cdk12 protein± are shown and HCT116 cell lines (Figure 4E and Supplementary Fig- below the graph and were detected by Western blotting using Cdk12 an- ure S7B). Of note, the depletion of Cdk12 and CycK but not tibody. The top asterisk (*) indicates migration of the endogenous Cdk12 Cdk13, which forms the separate Cdk13/CycK complex, protein and the bottom one indicates the position of an unspecifcband led to modest but reproducible reduction of ATM, Chek1 recognized by the Cdk12 antibody, which serves as a loading control. Nucleic Acids Research, 2015, Vol. 43, No. 5 2583 Downloaded from http://nar.oxfordjournals.org/ by guest on May 25, 2015

Figure 4. The crucial DDR genes are downregulated in HGS-OvCa patient samples with mutations in CDK12.(A, B and C)Graphsshowcomparisons of relative expression levels between the HGS-OvCa samples with the wild-type or mutated CDK12. The identity of genes is indicated on top of each graph. The data was generated using the following microarray probes: ATM (212672 at), ATR (209903 s at), CHEK1 (205394 at), FANCI (213007 at), MDC1 (203062 s at), RAD51D (209965 s at), NEK9 (212299 at), ORCL3 (210028 s at), TERF2 (203611 at), BRCA1 (211851 x at), BRCA2 (208368 s at). Whereas samples with the wild-type CDK12 are plotted as black triangles, those containing individual missense or nonsense/indel CDK12 mutations are depicted as colored circles or squares, respectively, as indicated by the legend in the top right corner. Results are presented as mean (red line) with standard error of the mean (SEM) (black whiskers). P-values are given next to red asterisks (*) and the number of asterisks indicates the degree of signifcance as follows: * P 0.05; ** P 0.01; *** P 0.001. Panels A and C show genes related to the HR pathway, while other affected DDR genes are shown in panel B.= (D and≤ E) Depletion= ≤ of Cdk12= decreases≤ the mRNA levels of HR genes in Caov-3 cells. Relative mRNA levels of genes indicated below the bars were determined by RT-qPCR using total RNA samples isolated from Caov-3 cells treated with the control (blue bars) or Cdk12 #1 siRNA (red bars). The error bars represent the mean SD. ± 2584 Nucleic Acids Research, 2015, Vol. 43, No. 5 and Rad51D protein levels (Supplementary Figure S7C). To the former were from 7.6-fold (at MDC1 TSS region) to rule out possible off-target effects of the CDK12 siRNA, 55.7-fold (at RAD51D ORF region) and averaged 15.7-fold we performed RNAi in same cell lines by targeting a differ- (SD 13.2), the latter equaled 3.2-fold. Confrming speci- ent region in CDK12. Again, we observed similar defects on fcity= of the Cdk12 signals, depletion of Cdk12 decreased the HR gene expression (Supplementary Figure S8 and data the occupancy of Cdk12 at the novel genes by an aver- not shown). Finally, to provide additional evidence that the age of 44.8% (SD 14.0), while the levels of Cdk12 at downregulation of all HR genes analyzed here is specifc the intergenic regions= were similar (Figure 5A, compare to the Cdk12/CycK complex, we treated Caov-3 cells with blue and red bars, and Supplementary Figure S9A). To siRNA targeting CCNK and CDK13. Whereas the knock- fnd out whether the Cdk12/CycK complex facilitates Ser2 down of CCNK, the gene encoding CycK, led to decreased phosphorylation at these genes, we next monitored the oc- expression of the HR genes, that of CDK13 did not (Sup- cupancy of total/Ser5-P and Ser2-P forms of RNAPII at plementary Figure S8). Together, these analyses provide im- them in control and Cdk12-depleted cells (Figure 5B and portant evidence that in HGS-OvCa patient samples, the C and Supplementary Figure S9D and E). The knockdown CDK12 mutations lead to decreased expression of a subset of Cdk12 reduced the occupancy of total/Ser5-P RNAPII of DDR genes, particularly those encoding key components by an average of 16.4% (SD 21.3) and primarily at TSSs of the HR pathway. Moreover, considering our biochemical of the genes, where the average= decrease was 34.1% (SD and gene reporter fndings, we conclude that the observed 12.3) (Figure 5B, compare blue and red bars). This ef- downregulation of DDR genes in patient samples is a conse- fect= may refect in part the reported ability of Cdk12/CycK quence of the disabled kinase activity of the mutant Cdk12 to phosphorylate the CTD of RNAPII at Ser5 in vitro (9). Downloaded from proteins. Importantly, however, decrease in the levels of the Ser2-P form of RNAPII was much more pronounced, averaging 48.9% (SD 13.5) (Figure 5C, compare blue and red bars). The Cdk12/CycK complex occupies HR genes and promotes In addition,= this reduction was evident at all three gene Ser2 phosphorylation of RNAPII regions, thus very likely spanning the entire length of the

Given our above fndings on the stimulatory role of genes. We conclude that the HR genes are direct targets of http://nar.oxfordjournals.org/ Cdk12/CycK in the expression of HR genes, it is important the Cdk12/CycK complex, which is present at them to fa- to determine whether these DDR genes are direct targets of cilitate the CTD Ser2 phosphorylation of transcriptionally Cdk12/CycK and whether this complex promotes the lev- engaged RNAPII. els of Ser2-P of RNAPII at them. To address these possibil- ities, we employed the quantitative chromatin immunopre- CDK12 mutations in HGS-OvCa disable the stimulatory role cipitation (ChIP-qPCR) assay to measure the occupancy of of the Cdk12/CycK complex in the repair of DSBs by HR Cdk12, total/Ser5-P RNAPII and Ser2-P RNAPII at se- lect genes at three different regions: the vicinity of tran- In light of our demonstration that the absence of functional by guest on May 25, 2015 scription start site (TSS), the interior of the open reading Cdk12/CycK leads to coordinated downregulation of crit- frame (ORF) and the vicinity of the polyadenylation site ical HR genes, we fnally asked whether the CDK12 muta- (pA) (Figure 5). We analyzed all four of the newly identifed tions disable the repair of DSBs in DNA by HR (Figure 6). HR genes, including ATM, CHEK1, MDC1 and RAD51D, To this end, we employed the DR-GFP assay, a classical as well as ATR and FANCD2 representing the previously chromosomal DSB repair assay that has been widely used established Cdk12-dependent genes. To assess whether the to measure HR in living cells (Figure 6A) (38). The genomic signals at these positions were specifc, we used RAD51D DR-GFP cassette contains two direct repeats of differen- and FANCI intergenic regions which showed minimal dif- tially mutated enhanced green fuorescent protein (EGFP) ferences between the occupancy of Cdk12, both forms of genes both of which fail to produce a functional EGFP RNAPII and IgG antibody control (Supplementary Figure protein. Whereas the upstream repeat has been modifed to S9A). Because of the subsequent functional studies involv- contain the recognition site for the rare-cutting I-SceI en- ing the HR pathway, we performed the ChIP-qPCR assay donuclease and an in-frame termination codon, the repeat using HeLa DR-GFP cell line, a HeLa cell line derivative that follows is an EGFP fragment truncated at the 3′ end. containing one copy of the HR substrate, the direct repeat Expression of I-SceI leads to a DSB in the upstream repeat, green fuorescence protein (DR-GFP) cassette (38). triggering the repair by HR using the wild-type sequence of Following confrmation of successful depletion of Cdk12 the downstream repeat as a DNA donor, resulting in EGFP (Supplementary Figure S9B), we proceeded with the as- positive cells. say on control and Cdk12 knockdown cells. As shown To determine whether the repair of DSB by HR relies on in Figure 5A and Supplementary Figure S9C, Cdk12 is Cdk12/CycK, we used RNAi to deplete endogenous Cdk12 present at all three regions of the newly identifed as well as in HeLa DR-GFP cells. Next, we transiently expressed the the established Cdk12-dependent HR genes. At the novel I-SceI endonuclease in these cells and in those treated with set of genes, Cdk12 enrichment over the intergenic region the non-targeting control siRNA. In agreement with re- ranged from 1.9-fold (at ATM pA region) to 6.1-fold (at cent reports (39,40), depletion of Cdk12 decreased the fre- CHEK1 pA region) and averaged 3.1-fold (SD 1.2). Al- quency of the HR-mediated repair by 3.5-fold (Figure 6B, though these levels were not very robust, they= were spe- bars 1–3). To fnd out whether the CDK12∼ mutations in cifcgiventhattheCdk12enrichmentoverIgGwassub- HGS-OvCa result in defcient HR-mediated DSB repair, stantially higher at genes compared to that at the intergenic we next co-expressed the I-SceI endonuclease together with region (Figure 5A, compare blue and gray bars). Whereas the Cdk12-siRNA resistant wild-type and mutant Cdk12-F Nucleic Acids Research, 2015, Vol. 43, No. 5 2585 Downloaded from http://nar.oxfordjournals.org/ by guest on May 25, 2015

Figure 5. Cdk12/CycK is present at the novel HR genes to promote phosphorylation of the CTD of RNAPII at Ser2. (A, B and C) Control (blue bars) or Cdk12 knockdown HeLa DR-GFP cells (red bars) were subjected to ChIP-qPCR analysis to determine the levels of Cdk12 (panel A), total/Ser5-P RNAPII (panel B) and Ser2-P RNAPII (panel C) occupancy at RAD51D intergenic region (IR) and three gene-specifcregionsasindicatedbelowthe graphs. The identity of genes analyzed is indicated on top of each graph. Levels of IgG signals at IR and gene regions are presented as gray bars. Where these levels were signifcantly lower than those obtained with specifcantibody,theerrorbarswereomittedforsimplicityreasons.Resultsarepresentedas percent of input DNA and plotted as the mean SD. ± 2586 Nucleic Acids Research, 2015, Vol. 43, No. 5

proteins in HeLa DR-GFP cells depleted of the endogenous Cdk12 protein. Importantly, the wild-type Cdk12-F protein rescued the EGFP signal by 2.3-fold over the diminished levels provoked by the Cdk12∼ depletion, reaching 65% of the HR frequency in cells with endogenous Cdk12/CycK (Figure 6B, bars 2–4). In contrast, despite their comparable expression to the wild-type protein, all Cdk12-F proteins carrying the CDK12 mutations but one failed at increas- ing the frequency of HR-mediated repair of the DSB (Fig- ure 6B, bars 5–10). The only exception was Cdk12 carrying the K975E mutation, which showed an 1.4-fold increase in the HR frequency over the background∼ levels (Figure 6B, bar 9). Thus, these functional fndings correlate completely with the damaging impacts of the CDK12 mutations on the activity of the Cdk12/CycK complex that we observed in our gene expression studies. We conclude that the CDK12 mutations found in HGS-OvCa disable the faithful repair of DSBs by HR, likely contributing to the pervasive instability of the genome in this cancer. Downloaded from

DISCUSSION To clarify the relationship between the genetic alterations in Cdk12 and cancerogenesis, we focused here on the mu-

tations identifed in HGS-OvCa. We provide biochemical, http://nar.oxfordjournals.org/ structural, gene expression and functional evidence that the recurrent somatic mutations in CDK12 are not inert. Rather, they are loss-of-function mutations that disable the ability of the Cdk12/CycK complex to promote the expres- sion and function of genes critical to the HR DNA repair pathway.Thus, we uncovered an important mechanism con- necting the defects in Cdk12/CycK to cancerogenesis, sup- porting the notion that CDK12 is a potential tumor sup- by guest on May 25, 2015 pressor that ceases to function in this most deadly form of ovarian cancer. Our fndings on the mechanism by which CDK12 muta- tions incapacitate Cdk12/CycK provide new insights into the deregulation of Cdks in cancer. It is established that hyperactivation of the cell cycle Cdks enable canceroge- nesis (41). Mechanistically, the increased kinase activity stems from overexpression or amplifcation of the Cdks Figure 6. CDK12 mutations in HGS-OvCa abrogate the ability of or their corresponding cyclins. In addition, the same can Cdk12/CycK to stimulate the repair of DNA double-strand breaks by be accomplished by diminished levels of Cdk inhibitors HR. (A)SchematicrepresentationoftheDR-GFPrecombinationsub- (CKI) in cancers or by missense mutations within the KD strate. The defective EGFP genes of the cassette, separated by 3.7 kb, are shown (top). The frst one (SceGFP) contains an I-SceIendonucleasesite of Cdks that disrupt CKI binding. Similarly, transcrip- and an in-frame termination codon (white vertical line), while the second tional Cdks, including the Cdk8/CycC and Cdk9/CycT one (iGFP) is an internal EGFP fragment, yielding GFP cells. The ho- complexes, promote tumorigenesis by elevated kinase ac- mologous EGFP sequences are depicted as black rectangles− and the non- tivity (17). Whereas CDK8 is overexpressed or amplifed repetitive EGFP sequence is depicted as gray rectangle. Induction of DSB in cancers (42), P-TEFb can enable cancerogenesis in con- by I-SceItriggerstherepairofSceGFPbyHRusingtheiGFPsequenceas a donor DNA, resulting in wild-type EGFP (green rectangle) and GFP+ cert with the amplifed c-Myc oncogene (13), by stimulat- cells (bottom). (B)ThemutantCdk12proteinsfailtopromotetherepair ing the improper target gene expression within mixed lin- of the DSB by HR. HeLa DR-GFP cells were treated with the control eage leukemia (MLL) fusion protein complexes (18) or via or Cdk12 #2 siRNA and transfected with the I-SceIexpressionplasmid the direct upregulation of transcription factors promoting together with plasmids encoding the wild-type (blue bar) or mutant (red bars) Cdk12-F proteins as indicated below the graph. The HR frequency the epithelial-mesenchymal transition (19). In contrast, we for each experimental condition is represented as the frequency relative demonstrate that the CDK12 mutations in HGS-OvCa and to the one reached by the I-SceIexpressioninthecontrolsiRNA-treated other cancers disable the kinase activity of Cdk12. Our re- cells, which was set to 100% (green bar). Results are presented as the mean sults on the Cdk12 inactivation in ovarian carcinoma are in SD. Levels of the endogenous Cdk12 and Cdk12-F proteins are shown perfect agreement with the recent report by Karnitz labora- below± the graph and were detected by Western blotting using Cdk12 an- tibody. The top asterisk (*) indicates migration of the endogenous Cdk12 tory (40). Importantly, we further disclose that the loss of protein and the bottom one indicates the position of an unspecifcband the interaction between Cdk12 and CycK is the predom- recognized by the Cdk12 antibody, which serves as a loading control. inant mechanism underlying this defect. Of note, among Nucleic Acids Research, 2015, Vol. 43, No. 5 2587 transcriptional Cdks, Cdk12 and its paralogue Cdk13 are in mRNA levels of Rad51D, a paralogue of Rad51 which mutated most heavily in human cancers (data not shown). plays a role in the assembly and maintenance of the Rad51 We propose that at least with regard to Cdk12, the KD mu- nucleoprotein flament, the critical intermediate for strand tations are selected for during the genesis of human cancers invasion during HR (26). Despite the fact that BRCA1, because of their damaging impacts on the Cdk12 activa- a known Cdk12-dependent gene (6,39,40), is not affected tion process. Further investigations of the non-KD muta- by the CDK12 mutations in HGS-OvCa, we speculate that tions in Cdk12 and of those in other transcriptional Cdks expression of this crucial HR protein may be diminished are needed to understand their mechanisms and contribu- in earlier stages of this cancer containing CDK12 muta- tions to cancer. tions. Importantly, we noted that HGS-OvCa patient sam- Our results from the gene reporter assay and the mRNA ples contain somatic and/or germline mutations in all of expression analysis of HGS-OvCa patient samples indicate the Cdk12-dependent HR genes revealed here. Moreover, that all CDK12 mutations hamper the ability of Cdk12 by inspecting the COSMIC database, we found that mu- to promote target gene expression. Importantly, by using tations in these HR genes are mutually exclusive with the ChIP-qPCR assay, we fnd that all examined HR genes CDK12 mutations (data not shown). This also holds true are direct targets of Cdk12/CycK, at which this transcrip- for the majority of BRCA1 and BRCA2 mutations, which tion elongation-associated kinase facilitates optimal levels are absent in 78% of HGS-OvCa cases that contain CDK12 of RNAPII Ser2 phosphorylation. Together, these fndings mutations (39). Together, these revelations suggest that mu- suggest that rather than through an indirect mechanism, tating CDK12 equips the developing cancer cells with an al- the CDK12 mutations in HGS-OvCa directly downregu- ternative source of defects in the HR repair pathway, which Downloaded from late DDR gene expression by precluding the Cdk12/CycK is achieved by the collective downregulation of critical HR complex to exert its stimulatory action on transcriptionally genes. Finally, three recent studies showed that depletion of engaged RNAPII at the genes. However, further work is Cdk12 sensitizes ovarian cancer cell lines to PARP1/2 in- needed to reveal which steps in the DDR gene expression hibitors (22,39,40). Because this sensitivity is a hallmark of fail to operate due to the inactive Cdk12 proteins. Since the the defective HR pathway, our work is in agreement with knockdown of Cdk12/CycK lowers the levels of nascent these fndings and also supports the possibility that tumors http://nar.oxfordjournals.org/ transcripts of several DDR genes (6), regulates alternative with CDK12 mutations might be particularly susceptible to splicing (43), and interferes with effective 3′-end formation the therapy with these inhibitors. of model pre-mRNAs (8,44), the mutations could hamper Together, our study provides a framework for under- the synthesis and maturation of DDR gene transcripts. Be- standing how mutations in Cdk12/CycK could promote cause all CDK12 HGS-OvCa mutations except one inac- cancerogenesis. By impairing HR, the most error-free cellu- tivate Cdk12 activity in vitro, these possible defects could lar repair mechanism of DSBs, CDK12 mutations may elicit stem from the poorly phosphorylated Ser2 residues of the the usage of error-prone DNA repair pathways, including transcribing RNAPII, preventing effcient recruitment of non-homologous end joining and single-strand annealing by guest on May 25, 2015 elongation and RNA processing factors (3,4). Alternatively, (52). As a result, spontaneous and DNA damage-induced the mutations might antagonize the cooperation between gene and chromosomal aberrations are likely to accumu- Cdk12/CycK and some of its recently reported interacting late, giving rise to the unstable genome, the enabling char- factors (44,45), resulting in the reduction of gene expres- acteristic of any cancer (53). Because HR plays a central sion. role in repairing ICLs that stall replication forks (54), an A previous report has found that half of the HGS-OvCa increase in errors during DNA replication, which is an im- tumor cases display genetic and epigenetic defects in the portant source of genomic instability, can also stem from components of the HR repair pathway (24). Our fndings the HR defciency provoked by the CDK12 mutations. In on the causal relationship between the CDK12 mutations support of this possibility, depletion of Cdk12/CycK sen- and the failure of the HR-mediated DNA repair shed a sitizes cells to the ICL-inducing drug mitomycin C (6), and new light on the source of the HR defects in this can- CycK is one of the major proteins underlying the resistance cer. Effcient HR is mediated by several groups of factors to camptothecin, the drug that creates an obstruction in the including components of the ATM/ATR signaling path- DNA, thereby blocking its replication (55). Finally, unlike way, BRCA1/2 proteins, Fanconi anemia (FA) proteins and ATM, which is activated by DSBs, ATR responds to a broad members of the Rad51 protein family (46–49). Our anal- spectrum of DNA aberrations. In addition to DSBs, it is ysis of the HGS-OvCa patient samples harboring CDK12 activated primarily by replication protein A-coated single mutations demonstrates signifcant decrease in mRNA lev- stranded DNA, an intermediate in many DNA repair path- els of several factors from these groups. In particular, the ways, such as nucleotide excision repair, mismatch repair ATM/ATR signaling pathway, which responds to DNA and base excision repair (50). Thus, we speculate that the damage and transduces signals to downstream effectors loss-of-function CDK12 mutations may trigger malfunc- (50), seems to be predominantly compromised. We fnd tions in a variety of DNA repair pathways, contributing to defcient mRNA levels of ATM, ATR, Chek1 and Mdc1 the remarkable genomic disarray in HGS-OvCa. proteins, out of which only ATR was previously charac- terized as a Cdk12-dependent gene (6). Next, we also re- veal decreased expression of mRNA for FANCI, an FA protein that functions in the repair of DNA inter strand SUPPLEMENTARY DATA crosslinks (ICLs) together with FANCD2 (51), the factor vi- tal to effcient HR. Finally, our analysis shows the decrease Supplementary Data are available at NAR Online. 2588 Nucleic Acids Research, 2015, Vol. 43, No. 5

ACKNOWLEDGEMENT gene-specifcrequirementofRNApolymeraseIICTD phosphorylation for sexual differentiation in S. pombe. Curr. Biol., We thank all the members of Barboric and Blazek lab- 20,1053–1064. oratories for discussions and critical comments on the 12. Dai,Q., Lei,T., Zhao,C., Zhong,J., Tang,Y.Z., Chen,B., Yang,J., Li,C., manuscript; Junya Kobayashi and Kenshi Komatsu for gen- Wang,S., Song,X. et al. (2012) Cyclin K-containing kinase complexes maintain self-renewal in murine embryonic stem cells. J. Biol. Chem., erously providing the HeLa DR-GFP cell line and I-SceI ex- 287,25344–25352. pression plasmid; Heini Hakala and Liisa Kauppi for tech- 13. Rahl,P.B., Lin,C.Y.,Seila,A.C., Flynn,R.A., McCuine,S., Burge,C.B., nical assistance and critical reading of the manuscript and Sharp,P.A. and Young,R.A. (2010) c-Myc regulates transcriptional Petra Ovesna´ for help with statistical analysis of Oncomine pause release. Cell, 141,432–445. data. 14. Chao,S.H. and Price,D.H. (2001) Flavopiridol inactivates P-TEFb and blocks most RNA polymerase II transcription in vivo. J. Biol. Chem., 276,31793–31799. 15. Ostapenko,D. and Solomon,M.J. (2003) Budding yeast CTDK-I is FUNDING required for DNA damage-induced transcription. Eukaryot. Cell, 2, 274–283. Academy of Finland [1263825 to M.B., 1273842 to T.L.]; 16. Winsor,T.S., Bartkowiak,B., Bennett,C.B. and Greenleaf,A.L. (2013) Marsha Rivkin Center for Ovarian Cancer Research [to ADNAdamageresponsesystemassociatedwiththephosphoCTD M.B. and D.B.]; Sigrid Juselius Foundation [4702687 to of elongating RNA polymerase II. PLoS One, 8,e60909. M.B.]; University of Helsinki Three-year Research Grant 17. Lee,T.I. and Young,R.A. (2013) Transcriptional regulation and its [490123 to M.B.]; Project ‘CEITEC-Central-European In- misregulation in disease. Cell, 152,1237–1251. 18. Smith,E., Lin,C. and Shilatifard,A. (2011) The super elongation stitute of Technology’ [CZ.1.05/1.1.00/02.0068 to D.B.]; complex (SEC) and MLL in development and disease. Genes Dev., 25, Downloaded from GACR [14-09979S to D.B., GAP301/11/0747 to V.B.]; 661–672. Deutsche Forschungsgemeinschaft [GE-976/9 to M.G.]; 19. Ji,X., Lu,H., Zhou,Q. and Luo,K. (2014) LARP7 suppresses P-TEFb European Social Fund and the state budget of Czech activity to inhibit breast cancer progression and metastasis. Elife, 3, e02907. Republic ‘Postdoc I’ Grant [CZ.1.07/2.3.00/30.0009 to 20. Benusiglio,P.R., Pharoah,P.D., Smith,P.L., Lesueur,F., Conroy,D., V.P. and V.B.]; DFG Excellence Cluster ImmunoSensation Luben,R.N., Dew,G., Jordan,C., Dunning,A., Easton,D.F. et al.

[EXC 1023 to M.G.]. Funding for open access charge: (2006) HapMap-based study of the 17q21 ERBB2 amplicon in http://nar.oxfordjournals.org/ Academy of Finland [1263825 to M.B.]. susceptibility to breast cancer. Br. J. Cancer, 95,1689–1695. Confict of interest statement. None declared. 21. Sircoulomb,F., Bekhouche,I., Finetti,P., Adelaide,J., Hamida,A., Bonansea,J., Raynaud,S., Innocenti,C., Charafe-Jauffret,E., Tarpin,C. et al. (2010) Genome profling of ERBB2-amplifed breast cancers. BMC Cancer, 10,539. REFERENCES 22. Natrajan,R., Wilkerson,P.M., Marchio,C., Piscuoglio,S., Ng,C.K., 1. Fuda,N.J., Ardehali,M.B. and Lis,J.T. (2009) Defning mechanisms Wai,P., Lambros,M.B., Samartzis,E.P., Dedes,K.J., Frankum,J. et al. that regulate RNA polymerase II transcription in vivo. Nature, 461, (2014) Characterization of the genomic features and expressed fusion 186–192. genes in micropapillary carcinomas of the breast. J. Pathol., 232, 553–565.

2. Eick,D. and Geyer,M. (2013) The RNA polymerase II by guest on May 25, 2015 carboxy-terminal domain (CTD) code. Chem. Rev., 113,8456–8490. 23. Zang,Z.J., Ong,C.K., Cutcutache,I., Yu,W., Zhang,S.L., Huang,D., 3. Zhou,Q., Li,T. and Price,D.H. (2012) RNA polymerase II elongation Ler,L.D., Dykema,K., Gan,A., Tao,J. et al. (2011) Genetic and control. Annu. Rev. Biochem., 81,119–143. structural variation in the gastric cancer kinome revealed through 4. Lenasi,T. and Barboric,M. (2010) P-TEFb stimulates transcription targeted deep sequencing. Cancer Res., 71,29–39. elongation and pre-mRNA splicing through multilateral mechanisms. 24. Cancer Genome Atlas Research Network. (2011) Integrated genomic RNA Biol., 7,145–150. analyses of ovarian carcinoma. Nature, 474,609–615. 5. Bartkowiak,B., Liu,P., Phatnani,H.P., Fuda,N.J., Cooper,J.J., 25. Carter,S.L., Cibulskis,K., Helman,E., McKenna,A., Shen,H., Price,D.H., Adelman,K., Lis,J.T. and Greenleaf,A.L. (2010) CDK12 Zack,T., Laird,P.W., Onofrio,R.C., Winckler,W., Weir,B.A. et al. is a transcription elongation-associated CTD kinase, the metazoan (2012) Absolute quantifcation of somatic DNA alterations in human ortholog of yeast Ctk1. Genes Dev., 24,2303–2316. cancer. Nat. Biotechnol., 30,413–421. 6. Blazek,D., Kohoutek,J., Bartholomeeusen,K., Johansen,E., 26. Jasin,M. and Rothstein,R. (2013) Repair of strand breaks by Hulinkova,P., Luo,Z., Cimermancic,P., Ule,J. and Peterlin,B.M. homologous recombination. Cold Spring Harb. Perspect. Biol., 5, (2011) The Cyclin K/Cdk12 complex maintains genomic stability via a012740. regulation of expression of DNA damage response genes. Genes Dev., 27. Blazek,D. (2012) The cyclin K/Cdk12 complex: an emerging new 25,2158–2172. player in the maintenance of genome stability. Cell Cycle, 11, 7. Cheng,S.W., Kuzyk,M.A., Moradian,A., Ichu,T.A., Chang,V.C., 1049–1050. Tien,J.F., Vollett,S.E., Griffth,M., Marra,M.A. and Morin,G.B. 28. Lenasi,T., Peterlin,B.M. and Barboric,M. (2011) Cap-binding protein (2012) Interaction of cyclin-dependent kinase 12/CrkRS with cyclin complex links pre-mRNA capping to transcription elongation and K1 is required for the phosphorylation of the C-terminal domain of alternative splicing through positive transcription elongation factor b RNA polymerase II. Mol. Cell. Biol., 32,4691–4704. (P-TEFb). J. Biol. Chem., 286,22758–22768. 29. Pavletich,N.P. (1999) Mechanisms of cyclin-dependent kinase 8. Davidson,L., Muniz,L. and West,S. (2014) 3′ end formation of pre-mRNA and phosphorylation of Ser2 on the RNA polymerase II regulation: structures of Cdks, their cyclin activators, and Cip and CTD are reciprocally coupled in human cells. Genes Dev., 28, INK4 inhibitors. J. Mol. Biol., 287,821–828. 342–356. 30. Larochelle,S., Amat,R., Glover-Cutter,K., Sanso,M., Zhang,C., 9. Bosken,C.A., Farnung,L., Hintermair,C., Merzel Schachter,M., Allen,J.J., Shokat,K.M., Bentley,D.L. and Fisher,R.P. (2012) Vogel-Bachmayr,K., Blazek,D., Anand,K., Fisher,R.P., Eick,D. and Cyclin-dependent kinase control of the initiation-to-elongation Geyer,M. (2014) The structure and substrate specifcity of human switch of RNA polymerase II. Nat. Struct. Mol. Biol., 19,1108–1115. Cdk12/Cyclin K. Nat. Commun., 5,3505. 31. Ding,L., Getz,G., Wheeler,D.A., Mardis,E.R., McLellan,M.D., 10. Cho,E.J., Kobor,M.S., Kim,M., Greenblatt,J. and Buratowski,S. Cibulskis,K., Sougnez,C., Greulich,H., Muzny,D.M., Morgan,M.B. (2001) Opposing effects of Ctk1 kinase and Fcp1 phosphatase at Ser et al. (2008) Somatic mutations affect key pathways in lung 2oftheRNApolymeraseIIC-terminaldomain.Genes Dev., 15, adenocarcinoma. Nature, 455,1069–1075. 3319–3329. 32. Kan,Z., Jaiswal,B.S., Stinson,J., Janakiraman,V., Bhatt,D., 11. Coudreuse,D., van Bakel,H., Dewez,M., Soutourina,J., Parnell,T., Stern,H.M., Yue,P., Haverty,P.M., Bourgon,R., Zheng,J. et al. (2010) Vandenhaute,J., Cairns,B., Werner,M. and Hermand,D. (2010) A Nucleic Acids Research, 2015, Vol. 43, No. 5 2589

Diverse somatic mutation patterns and pathway alterations in human Cyclin-dependent kinase 12 increases 3′ end processing of growth cancers. Nature, 466,869–873. factor-induced c-FOS transcripts. Mol. Cell. Biol., 35,468–478. 33. Drogat,J. and Hermand,D. (2012) Gene-specifcrequirementofRNA 45. Bartkowiak,B. and Greenleaf,A.L. (2014) Expression, purifcation, polymerase II CTD phosphorylation. Mol. Microbiol., 84,995–1004. and identifcation of associated proteins of the full length 34. Kohoutek,J. and Blazek,D. (2012) Cyclin K goes with Cdk12 and hCDK12/CyclinK complex. J. Biol. Chem., 290,1786–1795. Cdk13. Cell Div., 7,12. 46. Morrison,C., Sonoda,E., Takao,N., Shinohara,A., Yamamoto,K. 35. Tiley,L.S., Madore,S.J., Malim,M.H. and Cullen,B.R. (1992) The and Takeda,S. (2000) The controlling role of ATM in homologous VP16 transcription activation domain is functional when targeted to recombinational repair of DNA damage. EMBO J., 19,463–471. apromoter-proximalRNAsequence.Genes Dev., 6,2077–2087. 47. Moynahan,M.E., Chiu,J.W., Koller,B.H. and Jasin,M. (1999) Brca1 36. Garriga,J., Mayol,X. and Grana,X. (1996) The CDC2-related kinase controls homology-directed DNA repair. Mol. Cell, 4,511–518. PITALRE is the catalytic subunit of active multimeric protein 48. Nakanishi,K., Yang,Y.G., Pierce,A.J., Taniguchi,T., Digweed,M., complexes. Biochem. J., 319,293–298. D’Andrea,A.D., Wang,Z.Q. and Jasin,M. (2005) Human Fanconi 37. San Filippo,J., Sung,P. and Klein,H. (2008) Mechanism of eukaryotic anemia monoubiquitination pathway promotes homologous DNA homologous recombination. Annu. Rev. Biochem., 77,229–257. repair. Proc. Natl. Acad. Sci. U.S.A., 102,1110–1115. 38. Pierce,A.J., Johnson,R.D., Thompson,L.H. and Jasin,M. (1999) 49. McCabe,N., Turner,N.C., Lord,C.J., Kluzek,K., Bialkowska,A., XRCC3 promotes homology-directed repair of DNA damage in Swift,S., Giavara,S., O’Connor,M.J., Tutt,A.N., Zdzienicka,M.Z. mammalian cells. Genes Dev., 13,2633–2638. et al. (2006) Defciency in the repair of DNA damage by homologous 39. Bajrami,I., Frankum,J.R., Konde,A., Miller,R.E., Rehman,F.L., recombination and sensitivity to poly(ADP-ribose) polymerase Brough,R., Campbell,J., Sims,D., Rafq,R., Hooper,S. et al. (2014) inhibition. Cancer Res., 66,8109–8115. Genome-wide profling of genetic synthetic lethality identifes 50. Marechal,A. and Zou,L. (2013) DNA damage sensing by the ATM CDK12 as a novel determinant of PARP1/2 inhibitor sensitivity. and ATR kinases. Cold Spring Harb. Perspect. Biol., 5,a012716. Cancer Res., 74,287–297. 51. Smogorzewska,A., Matsuoka,S., Vinciguerra,P., McDonald,E.R. 40. Joshi,P.M., Sutor,S.L., Huntoon,C.J. and Karnitz,L.M. (2014) 3rd, Hurov,K.E., Luo,J., Ballif,B.A., Gygi,S.P., Hofmann,K., Downloaded from Ovarian cancer-associated mutations disable catalytic activity of D’Andrea,A.D. et al. (2007) Identifcation of the FANCI protein, a CDK12, a kinase that promotes homologous recombination repair monoubiquitinated FANCD2 paralog required for DNA repair. Cell, and resistance to cisplatin and poly(ADP-ribose) polymerase 129,289–301. inhibitors. J. Biol. Chem., 289,9247–9253. 52. Moynahan,M.E. and Jasin,M. (2010) Mitotic homologous 41. Malumbres,M. and Barbacid,M. (2009) Cell cycle, CDKs and cancer: recombination maintains genomic stability and suppresses achangingparadigm.Nat. Rev. Cancer, 9,153–166. tumorigenesis. Nat. Rev. Mol. Cell Biol., 11,196–207.

42. Firestein,R., Bass,A.J., Kim,S.Y., Dunn,I.F., Silver,S.J., Guney,I., 53. Hanahan,D. and Weinberg,R.A. (2011) Hallmarks of cancer: the next http://nar.oxfordjournals.org/ Freed,E., Ligon,A.H., Vena,N., Ogino,S. et al. (2008) CDK8 is a generation. Cell, 144,646–674. colorectal cancer oncogene that regulates beta-catenin activity. 54. Sirbu,B.M. and Cortez,D. (2013) DNA damage response: three levels Nature, 455,547–551. of DNA repair regulation. Cold Spring Harb. Perspect. Biol., 5, 43. Chen,H.H., Wang,Y.C. and Fann,M.J. (2006) Identifcation and a012724. characterization of the CDK12/cyclin L1 complex involved in 55. O’Connell,B.C., Adamson,B., Lydeard,J.R., Sowa,M.E., Ciccia,A., alternative splicing regulation. Mol. Cell. Biol., 26,2736–2745. Bredemeyer,A.L., Schlabach,M., Gygi,S.P., Elledge,S.J. and 44. Eifer,T.T., Shao,W., Bartholomeeusen,K., Fujinaga,K., Jager,S., Harper,J.W. (2010) A genome-wide camptothecin sensitivity screen Johnson,J.R., Luo,Z., Krogan,N.J. and Peterlin,B.M. (2015) identifes a mammalian MMS22L-NFKBIL2 complex required for genomic stability. Mol. Cell, 40,645–657. by guest on May 25, 2015 TUB0010.1177/1010428317727479Tumor BiologyPaculová et al. 727479research-article20172017

Original Article

Tumor Biology October 2017: 1 –11 BRCA1 or CDK12 loss sensitizes © The Author(s) 2017 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav cells to CHK1 inhibitors https://doi.org/10.1177/1010428317727479DOI: 10.1177/1010428317727479 journals.sagepub.com/home/tub

Hana Paculová1, Juraj Kramara2, Šárka Šimečková3,4, Radek Fedr3,5, Karel Souček3,4,5, Ondřej Hylse5,6, Kamil Paruch5,6, Marek Svoboda7, Martin Mistrík2 and Jiří Kohoutek1

Abstract A broad spectrum of tumors develop resistance to classic chemotherapy, necessitating the discovery of new therapies. One successful strategy exploits the synthetic lethality between poly(ADP-ribose) polymerase 1/2 proteins and DNA damage response genes, including BRCA1, a factor involved in homologous recombination–mediated DNA repair, and CDK12, a transcriptional kinase known to regulate the expression of DDR genes. CHK1 inhibitors have been shown to enhance the anti-cancer effect of DNA-damaging compounds. Since loss of BRCA1 increases replication stress and leads to DNA damage, we tested a hypothesis that CDK12- or BRCA1-depleted cells rely extensively on S-phase-related CHK1 functions for survival. The silencing of BRCA1 or CDK12 sensitized tumor cells to CHK1 inhibitors in vitro and in vivo. BRCA1 downregulation combined with CHK1 inhibition induced excessive amounts of DNA damage, resulting in an inability to complete the S-phase. Therefore, we suggest CHK1 inhibition as a strategy for targeting BRCA1- or CDK12-deficient tumors.

Keywords DNA damage response, BRCA1, CDK12, CHK1 inhibitor, transcription

Date received: 22 April 2017; accepted: 29 July 2017

Introduction Various CHK1 inhibitors have been tested as anti-tumor agents in combination with DNA-damaging agents, such A shared feature of various malignancies is the dysregula- as hydroxyurea, cisplatin, and topoisomerase inhibitors tion of DNA damage response (DDR), which leads to 1 genomic instability. The checkpoint kinase 1 (CHK1) rep- 1 Department of Chemistry and Toxicology, Veterinary Research resents a cellular factor that could be used to target the Institute, Brno, Czech Republic viability of tumor cells with genomic instability.2–4 This is 2Institute of Molecular and Translational Medicine, Faculty of Medicine because CHK1 is involved in numerous essential cellular and Dentistry, Palacky University, Olomouc, Czech Republic processes. For example, a cell responds to DNA damage 3Institute of Biophysics of the Czech Academy of Sciences, Brno,Czech Republic by activating CHK1 through ATR-promoted phosphoryla- 4 Department of Experimental Biology, Faculty of Science, Masaryk tion, which effectively blocks cell cycle progression. Once University, Brno, Czech Republic activated, CHK1 regulates the G2/M checkpoint by inacti- 5International Clinical Research Center, St. Anne’s University Hospital, vating the CDC25 phosphatases that would otherwise Brno, Czech Republic remove the inhibitory phosphates of cyclin-dependent 6Department of Chemistry, CZ Openscreen, Faculty of Science, Masaryk University, Brno, Czech Republic kinases (CDK), which are responsible for the G2/M transi- 7Department of Comprehensive Cancer Care, Masaryk Memorial 3 tion. CHK1 also participates in the DNA damage repair Cancer Institute, Brno, Czech Republic mechanism by phosphorylating, and thus activating, the 5 Corresponding author: repair factors BRCA2 and RAD51. In addition, CHK1 is Jiří Kohoutek, Department of Chemistry and Toxicology, Veterinary integral to the prevention of replication stress, as it stabi- Research Institute, Hudcova 296/70, 621 00 Brno, Czech Republic. lizes replication forks and regulates origin firing.3 Email: [email protected]

Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Tumor Biology

(topotecan, irinotecan), and antimetabolites, such as gem- with loss of BRCA1 or 2, are sensitive to inhibitors of citabine.6–10 Several studies have reported that the anti- PARP1/2, a protein involved in DNA repair. As with the tumor effect of CHK1 inhibitors is determined by p53 loss of BRCA1/2, the loss or inhibition of CDK12 sensi- status, with p53-deficient cells more responsive to CHK1 tizes cells to PARP inhibitors, which have recently been inhibitor treatment.2,11 However, other authors have approved for the treatment of ovarian cancer.21,41 reported that CHK1 inhibitors decrease cellular viability Nevertheless, research has reported that certain tumors irrespective of p53 suppressor status,6 showing that CHK1 have become resistant to PARP inhibitors as a result of inhibitors strongly potentiate the effects of DNA-damaging restored HR capacity, altered non-homologous end-joining agents in p53−/− cells. These results suggest that patients (NHEJ) capacity, decreased levels or activity of PARP1, with p53-mutated tumors could benefit from treatment and/or decreased intracellular availability of PARP inhibi- approaches that include CHK1 inhibition.11,12 The identifi- tors.25 Thus, novel alternatives to PARP1 inhibitors are cation of cellular factors that, when combined with CHK1 necessary for further patient treatment. inhibitors, confer synthetic lethality could strengthen the Although the genomic instability that results from DDR portfolio of combinatory treatments with CHK1 inhibitors deficiency often drives tumor development, it also pro- or potentially lead to the discovery of monotherapy vides a great opportunity for cancer treatment.26 The loss approaches for tumors with relevant mutations. Certain of BRCA1 and CDK12 function most likely potentiates scenarios have been described recently; for example, cells the anti-tumor effects of PARP1/2 inhibitors by crippling deficient in Fanconi anemia genes are hypersensitive to HR-mediated DNA repair, with this mechanism a perfect CHK1 inhibition, and a novel essential interplay between example of the concept of synthetic lethality. Since the loss CHK1 and I-kappa-B kinase epsilon was observed in ovar- of BRCA1 compromises DDR and leads to replication ian cells.13,14 stress and DNA damage,27 we hypothesized that BRCA1- The CDK12 regulates the elongation phase of transcrip- or CDK12-deficient cells will extensively rely on the tion by phosphorylating the C-terminal domain of RPB1, a S-phase-related kinase activity of CHK1 for survival. In subunit of RNA polymerase II (RNAPII).15–18 We previ- this study, we demonstrate that silencing BRCA1 or ously identified CDK12 as a cellular factor that orches- CDK12 indeed sensitizes cancer cells to CHK1 trates the expression of several key DDR genes, for inhibitors. example, BRCA1, ATR, ATM, FANCI, and FANCD2.16 By regulating DDR genes, CDK12 consequently affects Materials and methods homologous recombination (HR)-mediated DNA repair. A downregulation of CDK12 leads to increased endoge- Synthesis of CHK1 inhibitors nous DNA damage, DDR activation, and pronounced sen- sitivity to DNA-damaging agents.16,19 Cancer-associated The racemic CHK1 inhibitor SCH900776 (Merck, CDK12 mutations are predominantly located within the Darmstadt, Germany) was prepared in-house through a 28,29 kinase domain and result in a catalytically inactive pro- previously published route. The enantiomers were sep- tein.20 Based on these observations, CDK12 has been sug- arated by high-performance liquid chromatography ® gested to be a tumor suppressor candidate.19–21 (HPLC) with a chiral stationary phase (Chiralcel OJ™ The breast cancer–associated gene 1 (BRCA1) tumor column (Daicel Corporation, Tokyo, Japan), diameter suppressor protein is a central component of several dis- 21 mm, length 250 mm; mobile phase: n-hexane/ethanol tinct protein complexes that are vital to HR-mediated DNA 80:20 + 0.5% diethylamine, flow: 20 mL/min). The desired damage repair, cell cycle checkpoints, and transcriptional active R-enantiomer of SCH900776 eluted faster (reten- regulation.22,23 BRCA1 is inheritably mutated in about 9% tion time: 10:04 min) than the inactive S-enantiomer and 13% of unselected women with newly diagnosed tri- (retention time: 13:07 min). LY2603618 was purchased ple-negative breast cancer and ovarian cancer (respec- from Selleckchem (Houston, TX, USA; cat. no. S2626). tively). If metastatic, these patients have generally very unfavorable prognosis and currently are candidates for tar- Cell culture geted drug therapy, such as poly(ADP-ribose) polymerase (PARP) inhibitors.24–26 The loss of BRCA1, caused by HCT116 p53+/+ and p53−/− cells were a kind gift from B. homozygous mutations, reduces the ability of cells to carry Vogelstein.30 The cells were cultivated in Dulbecco’s out HR-mediated DNA repair, resulting in cellular genomic Modified Eagle’s Medium (DMEM; Sigma-Aldrich, instability.24 Interestingly, BRCA1 mutations are mutually D6429, Darmstadt, Germany) medium supplemented with exclusive with CDK12 mutations, which suggests that 5% fetal bovine serum (FBS; Sigma-Aldrich, F0804) at CDK12 belongs to the same HR-mediated DNA damage 37°C. MDA-MB-231 cells were obtained from the repair pathway as BRCA1.21 American Type Culture Collection (ATCC, Rockville, HR deficiency presents an opportunity for cancer treat- MD, USA) and were cultivated in DMEM medium sup- ment. Tumors exhibiting HR deficiency, especially those plemented with 10% FBS at 37°C with 5% CO2. Paculová et al. 3

Proliferation assays Reverse transcription–polymerase HCT116 cells were transfected with the following small chain reaction interfering RNAs (siRNAs; Santa Cruz Biotechnology, HCT116 cells were transfected with control or specific Dallas, TX, USA): CTRL A (sc-37007), BRCA1 (sc-29219), siRNAs using lipofectamine RNAiMAX (Invitrogen, CDK12 (sc-44531), CDK13 (sc-72836), and CDK12_2 13778150). After 72 h, cells were harvested, and total RNA (Sigma-Aldrich, SIHK0490) using Lipofectamine RNAi was isolated with RNAzol (Molecular Research Centre, MAX (Invitrogen, 13778150, Carlsbad, CA, USA). Viable RN190, Cincinnati, OH, USA). Reverse transcription (RT) cells were counted after 24 h and equal cell concentrations and quantitative polymerase chain reaction (qPCR) were were seeded into 96-well plates. After an additional 24 h, performed according to the method described by Blazek cells were treated with a CHK1 inhibitor, either SCH900776 et al.16 Changes in gene expression were calculated using or LY2603618, in dimethyl sulfoxide (DMSO) for 6 days. the comparative threshold cycle method with Hypoxanthine The medium was exchanged after 48 and 96 h, and fresh Phosphoribosyltransferase 1 (HPRT) to normalize for vari- inhibitors were added to the medium at these time points. ations in RNA input. The following primers were used for For each siRNA, the cell viability was assessed with the PCR: CDKN1A—forward: CTGGAGACTCTCAGGGT CyQuant NF Cell Proliferation Assay Kit (Invitrogen) and CGAAA and reverse: GATTAGGGCTTCCTCTTG and normalized to the relative growth of cells treated with HPRT—forward: 5′-CCAGACAAGTTTGTTGTAGG DMSO. All experiments were performed three times in ATATGCCCTTGAC-3′ and reverse: 5′-ACTCCAGATG triplicates. TTTCCAAACTCAACTTGAACTCTC-3′. MDA-MB-231 cells were seeded at equal concentra- tions into six-well plates. After 24 h, they were exposed to DMSO, along with either 0.3 or 1 µM SCH900776. After Plasmids 72 h, the cells were trypsinized and counted with a hemo- The pLKO.1 shBRCA1-2 and pLKO.1 shBRCA1-4 plas- cytometer. Cell viability was normalized to relative growth mids, which were used to generate stable BRCA1 knock- of cells treated with DMSO for each short hairpin RNA down cell lines, were part of the MISSION library (shRNA). The experiments were performed three times in (Sigma-Aldrich; construct numbers TRCN0000244985 duplicates. and TRCN0000244987). The lentiviral packaging plas- mids pMD2.G and psPAX2 were purchased from Addgene (Cambridge, MA, USA; Plasmids numbers 12259 and Western blot 12260). Cells were lysed in lysis buffer (100 mM Tris, pH 7.4, 1% sodium dodecyl sulfate (SDS), 10% glycerol), sonicated, Generation of MDA-MB-231 shBRCA1 cell lines and protein concentrations were assessed by the bicin- choninic acid (BCA) assay. Laemmli buffer (3×) was then Lentiviral transduction was used to generate MDA-MB-231 added, and lysates were boiled for 5 min at 100°C. Cell cell lines that harbor a stable shRNA knockdown of lysates were separated with electrophoresis employing BRCA1. Lentivirus production and transduction was per- 8%–15% gels and then wet blotted to a nitrocellulose mem- formed according to the method described by Tiscornia brane (GE Healthcare, Amersham, #10600008, Little et al.31 Briefly, lentiviruses were generated by co-transfect- Chalfont, UK). Individual proteins were detected with spe- ing 293T cells with 4 µg of pMD2.G, 7 µg of psPAX2, and cific antibodies: BRCA1 (Santa Cruz, sc-6954), CDK12 9 µg of a lentiviral plasmid of interest using the CaPO4 (Cell Signaling, #11973, Danvers, MA, USA), CDK13 precipitation method. Next, 6–8 h post-transfection, cells (rabbit serum produced in-house), Cyclin K (Santa Cruz, were washed with pre-warmed phosphate-buffered saline sc-376371), p53 Pantropic Ab-6 (Millipore, OP43, (PBS) and the medium was changed. Supernatant contain- Billerica, MA, USA), Cyclin T1 (Santa Cruz, sc-8127), ing lentiviruses was collected 48 h later and supplemented CHK1 (Cell Signaling, #2360), CHK1-pSer296 (Cell with 4 µg/mL polybrene (Sigma-Aldrich, 107689). Target Signaling, #2349), PARP (Cell Signaling, #9542), γH2AX cells were transduced at multiplicities of infection (MOIs) pSer 139 (Biolegend, 613402, San Diego, CA, USA), p21 of 1–10. The medium was changed 24 h post-transduction, (Santa Cruz, sc-397), p27 (Santa Cruz, sc-528), pRb (Cell and the cells were selected with 1 µg/mL puromycin. Signaling, #9309), and pRb-pSer780 (Cell Signaling, Resistant colonies were evaluated for expression of shRNA #8180). Anti-rabbit and anti-mouse secondary horseradish and the consequent reduction in BRCA1 protein levels. peroxidase (HRP)-linked antibodies were obtained from GE Healthcare (NA934V, NA931V), and anti-goat anti- Clonogenic assay body was obtained from Sigma-Aldrich (A5420). The immunoreactive bands were visualized using the Western MDA-MB-231 cells were seeded at equal concentrations blot Luminol reagent (Santa Cruz Biotechnology, SC-2048). into six-well plates (150 cells/well) and, after 24 h, treated 4 Tumor Biology with 0, 0.3, or 1 µM SCH900776 for 14 days. Medium was Results exchanged and fresh inhibitors were added every 3 days. After 2 weeks of treatment, the cells were fixed with 4% CDK12 and BRCA1 downregulation sensitizes paraformaldehyde and stained with Gram I Solution HCT116 cells to CHK1 inhibition irrespective (PENTA s.r.o., cat. no. 14600-11000, Prague, Czech Republic). of p53 status CDK12 regulates BRCA1 expression, and a loss of Cell cycle analysis BRCA1 results in increased DNA damage and replication stress.16,27 Thus, we hypothesized that cells with depleted HCT116 or MDA-MB-231 cells were trypsinized, washed BRCA1 or CDK12 should be extensively dependent on the in PBS, and fixed with ice-cold 70% ethanol overnight. S-phase-related function of CHK1. We tested how the Cells were then washed in PBS and stained with Vindelov CHK1 inhibitors SCH900776 and LY2603618 affect the solution (1M TrisHCl, pH 8.0; 0.1% Triton-X100; 10 mM proliferation of BRCA1- and CDK12-silenced tumor cells. NaCl; propidium iodide, 50 µg/mL; RNAse A 50 Kunitz U/mL) Whether the functional status of p53 influences the anti- for 30 min at 37°C. proliferative effect of CHK1 inhibitors remains an open question, so we employed a pair of HCT116 cell lines with Xenograft model p53 null and p53 WT status.32 In addition to CDK12 we validated the effects of CHK1 inhibitors on cells with All animal procedures were performed in strict accord- silenced CDK13, another kinase binding Cyclin K that can ance with the Guide for the Care and Use of Laboratory phosphorylate RNAPII,33 but that has no impact on Animals and approved by the Institutional Animal Care BRCA1 levels. and Use Committee. Tumor xenografts were generated We performed 6-day proliferation assays to assess the 6 by injecting 2 × 10 MDA-MB-231 parental or shBRCA relationship between CDK12 or BRCA1 deficiency and #4 breast cancer cells, in PBS with a final volume of CHK1 inhibition (Figure 1). The siRNA-transfected cells 100 µL, into the left mammary fat pads of Ctrl: SHO- were treated with different concentrations of CHK1 inhibi- SCID Hr Prkdc Hr mice (Charles River Laboratories, tor SCH900776. We used two distinct siRNAs against Wilmington, MA, USA). In total, 12 mice were injected CDK12, a pool of three siRNAs (CDK12), and a single with MDA-MB-231 parental or shBRCA #4 cells and sequence different from the other three (CDK12_2). half of them (six mice) from particular transplantation CDK12 and BRCA1 silencing significantly sensitized both was treated with the vehicle or CHK1 inhibitor p53+/+ and p53−/− cell lines to the CHK1 inhibitor SCH900776. Drug treatment began once a tumor had SCH900776. In contrast, CDK13 silencing had no effect on 3 reached 0.03 cm . The CHK1 inhibitor SCH900776 was sensitivity to SCH900776 (Figure 1(a) and (b)). To avoid dissolved in 20% Kolliphor ELP (Sigma-Aldrich, 30906) the risk that the effect of SCH900776 on cell proliferation and administered intraperitoneally at a final concentra- is compound-specific (i.e. that it employs unknown off- tion of 25 mg/kg for 5 days. This dose was selected based target effects), we tested the effect of another CHK1 inhibi- on a previously published study that investigated the tor, LY2603618, which has a different chemical structure,34 6 same CHK1 inhibitor. Two-dimensional calipers were using the same setup as in Figure 1(a) and (b). This experi- used to measure tumor volumes during and after the ment provided similar results as those obtained when treatment period, and volume was calculated based on SCH900776 was used (Figure 1(c) and (d)). It is important 2 9 the equation: π/6 × length × width . Data were normal- to note that sensitivity to both tested CHK1 inhibitors was 3 ized to the starting tumor volume of 0.03 cm . The whole independent of the p53 status. In order to demonstrate that experiment was carried out in two independent replicates. effect of CHK1 inhibitor is not limited only to the colorec- Increase in tumor volume after the treatment was statisti- tal cell line HCT116, the same type of experiment as in cally analyzed by Student’s t-test. Figure 1(a) was performed with an ovarian cancer cell line OVSAHO, which bears p53 mutation and wild-type 35 Statistical analysis BRCA1. Sensitization to CHK1 inhibitor after BRCA1 and CDK12 silencing was observed similar to HCT116 cell All data were statistically analyzed and visualized using line (Supplementary Figure S1). Prism5 (GraphPad Software, Inc., La Jolla, USA). Results The effective downregulation of CDK12, CDK13, cyc- were expressed as mean with standard error of the mean lin K (associating partner of CDK12 and CDK13), and (SEM). Viability assays data were fitted to sigmoidal dose BRCA1 in the cells was verified by Western blot analysis response curve and were analyzed with analysis of vari- (Figure 1(e)). As expected, the protein levels of cyclin K ance (ANOVA) test, and data from other experiments were and BRCA1 decreased after CDK12 downregulation, analyzed by Student’s t-test. Symbols used to express sta- which corresponds to our previous observations.16,20. tistical significance are as follows: *p ≤ 0.05; **p ≤ 0.01; Cyclin T1, an associating partner of CDK9 kinase involved and ***p ≤ 0.001. in transcription elongation, was used as a loading control. Paculová et al. 5

Figure 1. Downregulation of CDK12 or BRCA1 sensitizes HCT116 cells to CHK1 inhibitors. Six-day survival curves of (a, c) HCT116 p53+/+ or (b, d) HCT116 p53−/− cells transfected with various siRNAs (CTRL, CDK12, CDK13, and BRCA1) and treated with either the CHK1 inhibitor (a, b) SCH900776 or (c, d) LY2603618. Cell viability for each siRNA-treated cell line was assessed by the CyQuant NF kit and normalized to the relative growth of cells treated with DMSO. Error bars represent SEM for three independent experiments. CDK12- and BRCA1-silenced cells are sensitive to CHK1 inhibitors (p < 0.001, ANOVA). (e) The effective knockdown of indicated proteins after siRNA transfections was examined by Western blot analysis. HCT116 p53+/+ and HCT116 p53−/− cells were transfected with various siRNAs and protein levels were assessed by Western blot analysis after 72 h. The protein level of Cyclin T1 was used as a loading control.

BRCA1 or CDK12 depletion coupled with yH2AX36 and CHK1 autophosphorylation at serine 296, CHK1 inhibition induces p21-dependent an indication of activated CHK1 kinase.14 proliferation block BRCA1 and Cyclin K levels once again decreased fol- lowing CDK12 knockdown (Figure 2(a)). As expected, We also investigated endogenous DNA damage, apoptosis, treatment with CHK1 inhibitor SCH900776 led to a notice- autophagy, and cell cycle status to gain more insight into able decrease in the detected p-S296 CHK1 signal at all the molecular mechanisms responsible for the enhanced tested conditions (Figure 2(a)), and also induced CHK1 cytostatic effect of CDK12 or BRCA1 downregulation degradation in a dose-dependent manner, which is consist- coupled with CHK1 inhibition. HCT116 cells were ent with published research.14 Importantly, the yH2AX sig- depleted of CDK12, CDK13, or BRCA1 by siRNAs and nal, which reflects the amount of endogenous DNA damage, were then, 24 h post-transfection, exposed to a CHK1 significantly increased after SCH900776 treatment. This inhibitor (SCH900776) for an additional 96 h. Samples effect was exacerbated in CDK12- and BRCA1-depleted were collected and assessed for the DNA damage marker cells, but not in CDK13 cells (Figure 2(a)). 6 Tumor Biology

Figure 2. Impact of CDK12 and BRCA1 downregulation on DDR, apoptosis and cell cycle. (a) The effective knockdown of various proteins after siRNA transfection, CHK1 inhibition and DNA damage induction was assessed by Western blot analysis. HCT116 p53+/+ cells were transfected with control or specific siRNAs (CTRL, CDK12, CDK13, and BRCA1) and 2 days post-transfection cells were treated with 0, 0.3, or 1 μM CHK1 inhibitor SCH900776 for an additional 96 h. The protein levels of the studied proteins were elucidated by Western blot with indicated antibodies. The protein level of Cyclin T1 was used as a loading control. (b) Status of cellular factors participating in regulation of apoptosis and cell cycle. The protein levels of PARP, a marker of late apoptosis, tumor suppressor p53, and the cell-cycle regulating proteins p21, p27, pRb, and pRB-pSer780 were elucidated by Western blot with indicated antibodies. The protein level of Cyclin T1 was used as a loading control.

Together, these results verified that our experimental the status of the p21 (Cip1/Waf1), a protein that is a promi- setting was successfully able to inhibit CHK1 and con- nent inhibitor of CDKs, can induce cell cycle blockade, firmed our hypothesis that CHK1 is necessary for the and is known to respond to CHK1 inhibition.14 Indeed, p21 effective repair of endogenous DNA damage, especially in protein levels increased proportionally to CHK1 inhibition cells lacking functional components of DDR such as under all tested conditions, with BRCA1-depleted cells CDK12 or BRCA1. showing the strongest induction (HCT116 p53+/+) (Figure Next we focused on induction of apoptosis, which was 2(b), lanes 3 and 12). Interestingly, a robust induction of examined by the presence of cleaved PARP and Caspase 3, p21 was also observed upon CDK12 downregulation, even a commonly used markers of apoptosis. Surprisingly, without CHK1 inhibitor treatment (Figure 2(b), lanes 4–6). SCH900776 treatment of CDK12-depleted cells only The p21 induction was detected in HCT116 p53−/− cells moderately increased PARP-1 cleavage (Figure 2(b)). In as well (Supplementary Figure S3). On the contrary, the addition, no cleaved Caspase 3 was detected after transfec- expression levels of a related cellular inhibitor of CDKs, tion with any siRNAs or after administration of CHK1 the p27 protein, did not change following CHK1 inhibitor inhibitor (Supplementary Figure S2). Furthermore, we treatment, implying a specific induction of p21 in response were interested whether other types of cell death that might to CDK12 downregulation (Figure 2(b)). play a role in viability of CDK12-silenced cells, therefore In addition to regulating G1/S progression through CDK the protein levels of specific autophagy markers were eval- inhibition, p21 also induces the degradation of another cell uated. As depicted in Supplementary Figure S2, no signifi- cycle regulator, the Retinoblastoma protein (pRb).37 cant change in the protein level of Beclin, factor associated Therefore, we evaluated the levels and phosphorylation sta- with vesicle-trafficking during autophagy, after CHK1 tus of pRb. Silencing of CDK12 led to degradation of pRb inhibitor administration in combination with any of the regardless of CHK1 inhibition, and a moderate effect was tested siRNAs was identified. Also, no detectable change also observed in BRCA1 depleted cells after treatment with in protein level of the cleaved product of LC3B protein 1 µM of SCH900776 (Figure 2(b)). In addition, the cell was detected in any tested conditions. cycle status was examined by propidium iodide staining Because decreased cell viability cannot be entirely followed by flow cytometry. Depletion of CDK12 together explained by apoptosis or autophagy, we elucidated cell with administration of CHK1 inhibitor led to significant cycle progression. To corroborate this further, we checked prolonged G1 phase and shortened S phase of the cell cycle Paculová et al. 7 in comparison to CTRL-, CDK13-, or BRCA1-silenced reproducible, increase in γH2AX levels was observed in cells (Supplementary Figure S4). both shBRCA1 cell lines in comparison to the parental Based on these data, we conclude that the enhanced cells (Figure 3(e), lanes 1, 4, and 7). Interestingly, CHK1 cytostatic effect of CHK1 inhibition in CDK12- or inhibition led to a dramatic increase in the yH2AX signal BRCA1-depleted HCT116 p53+/+ cells is a result of in both shBRCA1 cell lines, while the parental cells only increased DNA damage, which leads to a robust induction demonstrated a moderate dose-dependent γH2AX response of p21 and delayed cell cycle progression. (Figure 3(e)). As was earlier observed in the HCT116 cells, CHK1 inhibition led to a reduction in CHK1 levels and a dose-dependent decrease in total and phosphorylated pRb BRCA1 depletion sensitizes MDA-MB-231 cells levels (Figure 3(e)). to CHK1 inhibition In summary, BRCA1 downregulation also sensitizes MDA-MB-231 breast cancer cells to CHK1 inhibitors. As Next we tested whether BRCA1 deficiency can sensitize in HCT116 cells, treatment with a CHK1 inhibitor induced other cancer models to the chemical inhibition of CHK1 excessive DNA damage followed by arrest in the S-phase. kinase. Since BRCA1 deregulation or loss-of-function mutations are common characteristics of ovarian and Xenograft mouse model breast cancers,38,39 we chose to manipulate MDA-MB-231 breast cancer cells, which have mutated form of p53 and Based on our in vitro results, we decided to assess the in normal BRCA1 status, prior to the administration of a vivo therapeutic effects of CHK1 inhibitors by employing a CHK1 inhibitor. The therapeutic potential of CHK1 inhibi- mouse orthotopic xenograft model of the MDA-MB-231 tors has been already tested in triple-negative breast can- parental and shBRCA1 #4 human breast carcinoma cells in cer (TNBC) cell lines and pre-clinical mouse models, but the fat pads of SHO-PrkdcSCID HrHr mice. Mice were con- the status of BRCA1 has not been modified.11 tinuously monitored for the development of a primary xen- First, we generated two MDA-MB-231 cell lines with ograft tumor and sacrificed when tumors reached 10% of stable expression of BRCA1 shRNA (MDA-MB-231 body weight. Besides the observation that animals trans- shBRCA1 #2 and #4). Successful BRCA1 depletion at the planted with either parental or shBRCA1 #4 MDA-MB-231 protein and mRNA level was confirmed by Western blot cells all formed tumors (100% tumor growth), mice injected and RT-PCR, respectively (Figure 3(a)). The effect of with shRBCA1 cells showed a slightly higher proportion of BRCA1 downregulation and CHK1 inhibition on cellular tumor volume, reflecting faster growth of shBRCA1 cells viability was assessed by a survival assay that employed (Figure 3(c)). Furthermore, the growth rate and final size of the same setup as for HCT116 cells (Figure 3(b)). the tumors differed distinctly between populations after Treatment with 1 µM of SCH900776 led to a severe reduc- CHK1 inhibitor treatment. The parental MDA-MB-231 tion in cell viability, decreasing the cell counts of both control cells formed significant tumors over the course of depleted cell lines by 70% when compared to the parental 44 days, but treatment with a CHK1 inhibitor did not sig- non-transfected cells (Figure 3(b)). We examined the nificantly affect the size of the tumors when compared to effect of CHK1 inhibition on cell viability in control animals (Figure 4(a)). In contrast, the animals MDA-MB-231 (shBRCA1 #2 and #4) cell lines further by injected with shBRCA1 MDA-MB-231 cells developed performing a clonogenic survival assay, which better significantly larger tumors than the control group, and reflects the effects of long-term exposure (Figure 3(c)). treatment with a CHK1 inhibitor significantly decreased After 14 days of cultivation, we noticed that the colonies tumor size (Figure 4(b)). of both untreated shBRCA1 cell lines were larger than those of the parental cell line, suggesting faster cell cycle Discussion progression and higher mitotic potential. This is in line with the observation that BRCA1 loss accelerates the CHK1 inhibitors represent a promising cancer therapy growth of cancer cells.40 However, treatment with a CHK1 approach.3 Since the anti-cancer effect of CHK1 inhibitors inhibitor (1 µM) noticeably reduced colony size in both is potentiated by DNA damaging drugs, we hypothesized shBRCA1 cell lines and also led to dramatic decrease in that impaired DDR will have a similar synergistic effect. In cell counts (Figure 3(c)) our previous study, we demonstrated that CDK12 regulates Moreover, the 3-day inhibition of CHK1 had no appar- the transcription of certain DDR genes, particularly HR ent effect on cell cycle in the parental MDA-MB-231 cell genes (including BRCA1), and is necessary for maintaining line, whereas the shBRCA1 clones experienced a promi- genomic stability.16 In line with this observation, the loss of nent increase in the S-phase cells, suggesting major prolif- either BRCA1 or CDK12 is a prerequisite for sensitizing eration arrest (Figure 3(d)). cancer cells to PARP1/2 inhibitors.21,41 In this study, we We then tested how BRCA1 downregulation enhances demonstrate that the loss of BRCA1 or CDK12 also poten- endogenous DNA instability. A rather moderate, yet tiates the anti-proliferative effect of CHK1 inhibitors. 8 Tumor Biology

Figure 3. Downregulation of BRCA1 sensitizes MDA-MB-231 cells to CHK1 inhibitor. (a) BRCA1 protein levels were evaluated in shBRCA1 #2 and shBRCA1 #4 MDA-MB-231 cell lines by Western blot analysis. GAPDH was used as a loading control. The mRNA levels of BRCA1 in these cell lines were measured by RT-qPCR (p < 0.05, Student’s t-test). These experiments confirmed effective BRCA1 downregulation by shRNAs in these cell lines. (b) CHK1 inhibition reduced the viability of MDA-MB-231 cells expressing shRNAs against BRCA1. The graphs show the results of survival assays of parental, shBRCA1 #2 and shBRCA1 #4 MDA-MB-231 cells treated with 0, 0, 3, or 1 μM CHK1 inhibitor SCH900776 for 3 days. For each cell line, the cell numbers were normalized to the relative growth of cells treated with DMSO. Error bars represent SEM for three independent experiments (p < 0.05, Student’s t-test). (c) A 14-day clonogenic assay showed that the combination of BRCA1 silencing and CHK1 inhibition reduces cell viability. MDA-MB-231 cell lines were seeded and treated with the indicated SCH900776 concentrations. The experiment was performed three times in duplicates. BRCA1 silencing combined with CHK1 inhibition reduces cell viability. (d) CHK1 inhibition strongly affected the cell cycle progression of BRCA1-silenced MDA-MB-231 cell lines. Cells were cultivated and treated as described in (b). Cell cycle progression was evaluated by the incorporation of propidium iodide followed by flow cytometry. Error bars represent SEM from three experiments. (e) The combination of BRCA1 silencing and CHK1 inhibition induces a stronger activation of DDR. Cell were prepared and treated as described in (b). The activation of DDR and inhibition of CHK1 by SCH900776 were validated by Western blot with antibodies against γH2AX pSer139 and CHK1, respectively. The dose-dependent degradation of total pRb and pRb phosphorylated on serine 780 following CHK1 inhibition was examined by Western blot. Cyclin T1 was used as a loading control.

Our results show that the anti-proliferative effect of the induction of γH2AX pSer139 in HCT116 cells.6,14 The CHK1 inhibitor treatment combined with BRCA1 or strongest effect was obtained when CHK1 inhibition was CDK12 deficiency is comparable in both cell lines regard- combined with CDK12 silencing. Interestingly, in contrast less of p53 status. Previous reports have shown that CHK1 to the results obtained in HeLa cells,16 we did not observe inhibition leads to increased DNA damage by measuring increased γH2AX Ser139 phosphorylation upon CDK12 Paculová et al. 9

Figure 4. Inhibition of CHK1 prevents tumor growth in vivo. The CHK1 inhibitor SCH900776 decreased tumor growth. The (a) parental and (b) shBRCA1 MDA-MB-231 cells were transplanted into the mammary pads of SCID mice. When tumor mass reached a volume of 0.03 cm3, mice were treated with either a vehicle solution of 20% Kolliphor ELP or SCH900776 25 mg/kg/day dissolved in the same 20% Kolliphor solution, every day for 5 days. The growth of tumor mass was then monitored over set time periods. Each data point represents the mean increase in tumor volume after the beginning of treatment and error bars represent SEM, where n for each cohort was six animals (p < 0.001, Student’s t-test for shBRCA1 SCH900776 vs shBRCA1 vehicle). depletion in HCT116 cells. This may be partially due to deficiency suggesting that CDK12 functions in additional differences between these cell lines. However, we observed parts of the DNA repair machinery.44 From the clinical robust γH2AX pSer139 induction in the MDA-MB-231 point of view, about 9% and 13% of unselected women BRCA-silenced cells after CHK1 inhibition. CHK1 inhibi- with newly diagnosed triple-negative breast cancer and tion was previously reported to induce the cell cycle regu- ovarian cancer (respectively) have an inheritable BRCA1 lator p21.14 We observed a particularly robust p21 increase mutation. If metastatic, these patients have generally very in BRCA1-silenced cells. CDK12 silencing (regardless of unfavorable prognosis and currently are candidates for CHK1 inhibition) resulted in increased apoptosis, which targeted drug therapy, such as PARP inhibitors.24–26 We was consistent with result obtained from CDK12 inhibi- confirmed the additive effect of BRCA1 loss-of-function tion.42 Based on the data obtained from MDA-MB-231 and CHK1 on cell proliferation in vitro and in vivo. A cells, one could speculate that CHK1 inhibitors trigger xenograft model was employed to evaluate whether increased DNA damage and replication stress during the CHK1 inhibition confers an anti-proliferative effect on S-phase of the cell cycle, which is incompatible with cell tumor growth in vivo. CHK1 inhibitor administration had proliferation or survival. Moreover, the potential of CHK1 no substantial impact on the parental MDA-MB-231 cells, inhibitors to weaken tumor growth has been reported by which was in sharp contrast to significant decrease in several groups, but, as of yet, the synergy between impaired tumor mass observed in BRCA1-silenced cells receiving CDK12 function and CHK1 inhibition to counteract tumor CHK1 inhibitors (Figure 4). Recent studies have clearly progression has not been investigated.6,9,11 shown that patients can develop a resistance to PARP1/2 It has been demonstrated that CDK12 regulates the inhibitors over the course of PARP inhibitor therapy.45 expression of DDR genes including CHK1.20 Importantly, Since many factors, such as HR and NHEJ status, as well high-grade serous ovarian tumors bearing CDK12 muta- as the level, activity, or intracellular concentration of tions have reduced CHK1 expression.20 Therefore, we PARP proteins, can influence the efficacy of PARP inhibi- speculate that viability of CDK12-depleted cells rely tors, it is vital to identify different conditions that can extensively on the residual activity of CHK1 making these either re-sensitize tumor cells to PARP inhibitors or ena- cells sensitive to lower doses of CHK1 inhibitors. ble the use of additional strategies that target the systems Several studies have demonstrated that CDK12 regu- necessary for cell survival to treat resistant tumors. The lates the expression of DDR genes (BRCA1, FANCI, anti-proliferative effect of CHK1 inhibitors in BRCA1- FANCD2, and ATM).16,20,42 Nevertheless, the precise deficient (HR compromised) tumor cells differs from that mechanism of this regulation has not yet been described. of PARP1 inhibitors; hence, it seems rational to investi- CDK12 loss might downregulate DDR genes directly gate how using CHK1 inhibitors as a second round of through the role of CDK12 in the transcription of therapy will affect patients who have tumors resistant to these genes, as has been suggested by numerous PARP inhibitors due to the restoration of HR. studies.16,19,20,42,43 In summary, we have found that CHK1 inhibition is a In addition, a recent publication reported that genomic promising strategy for targeting BRCA1- or CDK12-deficient instability in ovarian tumors with a loss of CDK12 has a cells. We propose that BRCA1 and CDK12 deficiency should specific pattern of defective HR caused by BRCA1/2 be considered a CHK1 sensitivity biomarker candidate. The 10 Tumor Biology cell cycle arrest triggered by recently developed specific 5. Bahassi EM, Ovesen JL, Riesenberg AL, et al. The check- CDK12 inhibitors is in line with our presented observations.42 point kinases Chk1 and Chk2 regulate the functional Moreover, combination therapy with PARP1/2 inhibitors and associations between hBRCA2 and Rad51 in response to a CDK12 inhibitor conferred a strong anti-proliferative effect DNA damage. Oncogene 2008; 27: 3977–3985. in breast cancer cells.41 Our results provide promising evi- 6. Guzi TJ, Paruch K, Dwyer MP, et al. Targeting the replica- tion checkpoint using SCH 900776, a potent and function- dence for the combinatory effect of CDK12 and CHK1 inhib- ally selective CHK1 inhibitor identified via high content 41–43 itors in treating cancer patients. screening. Mol Cancer Ther 2011; 10: 591–602. 7. Kim MK, James J and Annunziata CM. Topotecan syner- Acknowledgements gizes with CHEK1 (CHK1) inhibitor to induce apoptosis in The authors certify that they have NO affiliations with, or involve- ovarian cancer cells. BMC Cancer 2015; 15: 196. ment in, any organization or entity with any financial or non- 8. Ma CX, Ellis MJ, Petroni GR, et al. A phase II study of financial interest (such as personal or professional relationships, UCN-01 in combination with irinotecan in patients with affiliations, knowledge or beliefs) in the subject matter or materi- metastatic triple negative breast cancer. Breast Cancer Res als discussed in this manuscript. The authors would like to thank Treat 2013; 137: 483–492. Dr Sabina Sevcikova and members of the J. Kohoutek laboratory 9. Montano R, Thompson R, Chung I, et al. Sensitization of human for critical discussion and their suggestions during the preparation cancer cells to gemcitabine by the Chk1 inhibitor MK-8776: of this manuscript, Dr Jiri Jarkovsky for statistical analyses, Prof. cell cycle perturbation and impact of administration schedule Jiri Bartek for helpful discussion, D. Blazek for providing us with in vitro and in vivo. BMC Cancer 2013; 13: 604. OVSAHO cells and Dr Miroslav Machala for the sharing antibod- 10. Perez RP, Lewis LD, Beelen AP, et al. Modulation of cell ies that were used in the Western blot analyses. cycle progression in human tumors: a pharmacokinetic and tumor molecular pharmacodynamic study of cisplatin plus Declaration of conflicting interests the Chk1 inhibitor UCN-01 (NSC 638850). Clin Cancer Res 2006; 12: 7079–7085. The author(s) declared no potential conflicts of interest with respect 11. Ma CX, Cai S, Li S, et al. Targeting Chk1 in p53-deficient to the research, authorship, and/or publication of this article. triple-negative breast cancer is therapeutically beneficial in human-in-mouse tumor models. J Clin Invest 2012; 122: Funding 1541–1552. 12. Chen Z, Xiao Z, Gu WZ, et al. Selective Chk1 inhibitors The author(s) disclosed receipt of the following financial support differentially sensitize p53-deficient cancer cells to cancer for the research, authorship, and/or publication of this article: This therapeutics. Int J Cancer 2006; 119: 2784–2794. study was supported by the Internal Grant Agency of the Ministry 13. Chen CC, Kennedy RD, Sidi S, et al. CHK1 inhibition as of Health of the Czech Republic, grant NT14599-3/2013 (M.S. a strategy for targeting Fanconi Anemia (FA) DNA repair and J.K.); the Ministry of Health of the Czech Republic, grant pathway deficient tumors. Mol Cancer 2009; 8: 24. 16-34152A (J.K.); the Ministry of Agriculture, grant RO0517; the 14. Kim MK, Min DJ, Wright G, et al. Loss of compensatory European Regional Development Fund, project no. LQ1605 from pro-survival and anti-apoptotic modulator, ikkepsilon, the National Program of Sustainability II (MEYS CR) (K.S., K.P., sensitizes ovarian cancer cells to CHEK1 loss through an and O.H.); the Ministry of Health of the Czech Republic, grant increased level of p21. Oncotarget 2014; 5: 12788–12802. 15-33999A (M.S., K.P., and K.S.); CZ-OPENSCREEN: National 15. Bartkowiak B, Liu P, Phatnani HP, et al. CDK12 is a tran- Infrastructure for Chemical Biology, LM2015063 (K.P.); the scription elongation-associated ctd kinase, the metazoan Czech National Program of Sustainability, LO1304 (J.K. and ortholog of yeast Ctk1. Genes Dev 2010; 24: 2303–2316. M.M.); and the Norwegian Financial Mechanism 2009-2014 16. Blazek D, Kohoutek J, Bartholomeeusen K, et al. The cyclin Project, Contract 7F14061 (J.K.). K/Cdk12 complex maintains genomic stability via regula- tion of expression of DNA damage response genes. Genes Supplementary material Dev 2011; 25: 2158–2172. Supplementary material is available for this article online 17. Cheng SW, Kuzyk MA, Moradian A, et al. Interaction of cyclin-dependent kinase 12/CrkRS with cyclin K1 is required for the phosphorylation of the C-terminal domain References of RNA polymerase II. Mol Cell Biol 2012; 32: 4691–4704. 1. Hanahan D and Weinberg RA. Hallmarks of cancer: the 18. Kohoutek J and Blazek D. Cyclin K goes with Cdk12 and next generation. Cell 2011; 144: 646–674. Cdk13. Cell Div 2012; 7: 12. 2. Ma CX, Janetka JW and Piwnica-Worms H. Death by 19. Joshi PM, Sutor SL, Huntoon CJ, et al. Ovarian cancer- releasing the breaks: CHK1 inhibitors as cancer therapeu- associated mutations disable catalytic activity of CDK12, a tics. Trends Mol Med 2011; 17: 88–96. kinase that promotes homologous recombination repair and 3. McNeely S, Beckmann R and Bence Lin AK. CHEK again: resistance to cisplatin and poly(ADP-ribose) polymerase revisiting the development of CHK 1 inhibitors for cancer inhibitors. J Biol Chem 2014; 289: 9247–9253. therapy. Pharmacol Ther 2014; 142: 1–10. 20. Ekumi KM, Paculova H, Lenasi T, et al. Ovarian carcinoma 4. Thompson R and Eastman A. The cancer therapeutic poten- CDK12 mutations misregulate expression of DNA repair tial of chk1 inhibitors: how mechanistic studies impact on genes via deficient formation and function of the Cdk12/ clinical trial design. Br J Clin Pharmacol 2013; 76: 358–369. Cyck complex. Nucleic Acids Res 2015; 43: 2575–2589. Paculová et al. 11

21. Bajrami I, Frankum JR, Konde A, et al. Genome-wide pro- 34. Wang FZ, Fei HR, Cui YJ, et al. The checkpoint 1 kinase filing of genetic synthetic lethality identifies CDK12 as a inhibitor LY2603618 induces cell cycle arrest, DNA dam- novel determinant of PARP1/2 inhibitor sensitivity. Cancer age response and autophagy in cancer cells. Apoptosis 2014; Res 2014; 74: 287–297. 19: 1389–1398. 22. Deng CX. BRCA1: cell cycle checkpoint, genetic instabil- 35. Domcke S, Sinha R, Levine DA, et al. Evaluating cell lines ity, DNA damage response and cancer evolution. Nucleic as tumour models by comparison of genomic profiles. Nat Acids Res 2006; 34: 1416–1426. Commun 2013; 4: 2126. 23. Jiang Q and Greenberg RA. Deciphering the BRCA1 tumor 36. Sedelnikova OA, Rogakou EP, Panyutin IG, et al. suppressor network. J Biol Chem 2015; 290: 17724–17732. Quantitative detection of (125)IdU-induced DNA double- 24. Helleday T. Homologous recombination in cancer devel- strand breaks with gamma-H2AX antibody. Radiat Res opment, treatment and development of drug resistance. 2002; 158: 486–492. Carcinogenesis 2010; 31: 955–960. 37. Broude EV, Swift ME, Vivo C, et al. P21(Waf1/Cip1/Sdi1) 25. Konecny GE and Kristeleit RS. PARP inhibitors for mediates retinoblastoma protein degradation. Oncogene BRCA1/2-mutated and sporadic ovarian cancer: current prac- 2007; 26: 6954–6958. tice and future directions. Br J Cancer 2016; 115: 1157–1173. 38. Cancer Genome Atlas Network. Comprehensive molecu- 26. Bartek J, Mistrik M and Bartkova J. Thresholds of replica- lar portraits of human breast tumours. Nature 2012; 490: tion stress signaling in cancer development and treatment. 61–70. Nat Struct Mol Biol 2012; 19: 5–7. 39. Cancer Genome Atlas Research Network. Integrated 27. Gaillard H, Garcia-Muse T and Aguilera A. Replication genomic analyses of ovarian carcinoma. Nature 2011; 474: stress and cancer. Nat Rev Cancer 2015; 15: 276–289. 609–615. 28. Dwyer MP, Paruch K, Labroli M, et al. Discovery of 40. Mullan PB, Quinn JE and Harkin DP. The role of BRCA1 in pyrazolo[1,5-a]pyrimidine-based CHK1 inhibitors: a tem- transcriptional regulation and cell cycle control. Oncogene plate-based approach—part 1. Bioorg Med Chem Lett 2011; 2006; 25: 5854–5863. 21: 467–470. 41. Johnson SF, Cruz C, Greifenberg AK, et al. CDK12 29. Labroli M, Paruch K, Dwyer MP, et al. Discovery of inhibition reverses de novo and acquired PARP inhibi- pyrazolo[1,5-a]pyrimidine-based CHK1 inhibitors: a tem- tor resistance in BRCA wild-type and mutated mod- plate-based approach—part 2. Bioorg Med Chem Lett 2011; els of triple-negative breast cancer. Cell Rep 2016; 17: 21: 471–474. 2367–2381. 30. Bunz F, Dutriaux A, Lengauer C, et al. Requirement for 42. Zhang TH, Kwiatkowski N, Olson CM, et al. Covalent tar- p53 and p21 to sustain g2 arrest after DNA damage. Science geting of remote cysteine residues to develop CDK12 and 1998; 282: 1497–1501. CDK13 inhibitors. Nat Chem Biol 2016; 12: 876. 31. Tiscornia G, Singer O and Verma IM. Production and puri- 43. Blazek D. Transcriptional kinases caught by a sticky drug. fication of lentiviral vectors. Nat Protoc 2006; 1: 241–245. Nat Chem Biol 2016; 12: 765–766. 32. Dai Y, Chen S, Kmieciak M, et al. The novel Chk1 inhibitor 44. Popova T, Manie E, Boeva V, et al. Ovarian cancers har- MK-8776 sensitizes human leukemia cells to HDAC inhibi- boring inactivating mutations in CDK12 display a distinct tors by targeting the intra-S checkpoint and DNA replica- genomic instability pattern characterized by large tandem tion and repair. Mol Cancer Ther 2013; 12: 878–889. duplications. Cancer Res 2016; 76(7): 1882–1891. 33. Greifenberg AK, Honig D, Pilarova K, et al. Structural and 45. Ricks TK, Chiu HJ, Ison G, et al. Successes and challenges functional analysis of the CDK13/Cyclin K complex. Cell of parp inhibitors in cancer therapy. Front Oncol 2015; Rep 2016; 14: 320–331. 5: 222. Paculová and Kohoutek Cell Div (2017) 12:7 DOI 10.1186/s13008-017-0033-x Cell Division

REVIEW Open Access The emerging roles of CDK12 in tumorigenesis Hana Paculová and Jiří Kohoutek*

Abstract Cyclin-dependent kinases (CDKs) are key regulators of both cell cycle progression and transcription. Since dysregula- tion of CDKs is a frequently occurring event driving tumorigenesis, CDKs have been tested extensively as targets for cancer therapy. Cyclin-dependent kinase 12 (CDK12) is a transcription-associated kinase which participates in various cellular processes, including DNA damage response, development and cellular differentiation, as well as splicing and pre-mRNA processing. CDK12 mutations and amplification have been recently reported in different types of malig- nancies, including loss-of-function mutations in high-grade serous ovarian carcinomas, and that has led to assump- tion that CDK12 is a tumor suppressor. On the contrary, CDK12 overexpression in other tumors suggests the possibil- ity that CDK12 has oncogenic properties, similarly to other transcription-associated kinases. In this review, we discuss current knowledge concerning the role of CDK12 in ovarian and breast tumorigenesis and the potential for chemical inhibitors of CDK12 in future cancer treatment. Keywords: CDK12, RNA pol II, Suppressor, Oncogene, Dinaciclib, THZ531

Background cellular physiology. Consequently, deregulation of these Cyclin-dependent kinases (CDKs) are principal regula- processes drives cancer onset and progression [4]. tors of various cellular processes. Tey are divided into Transcription factors are frequently mutated in cancer two subfamilies: cell cycle-associated CDKs (CDK1, 2, 4, cells and represent typical oncogenes and tumor sup- 6), which directly regulate progression through individ- pressors. Tese mutations lead to alterations in the gene ual cell cycle phases, and transcription-associated CDKs expression programs and might create dependency on (CDK7, 8, 9, 11, 12, 13), which regulate gene transcrip- certain transcriptional regulators making cancer cells tion. Tese kinases phosphorylate the C-terminal domain addicted to their activities [5]. Such a phenomenon is (CTD) of Rbp1, the largest subunit of RNA polymerase II called “Transcriptional Addiction”, and it provides oppor- (RNA pol II) as well as various transcription regulatory tunities for novel therapeutic interventions in cancer [5]. factors. Since CDKs are frequently dysregulated in tumor CDK12 is a transcription-associated CDK that phos- cells, therefore they are attractive therapeutic targets for phorylates the CTD of RNA pol II and it is essential for a broad spectrum of tumors [1, 2]. DNA damage response (DDR), splicing, and differentia- Eukaryotic transcription is very complex and tightly tion [6]. CDK12 mutations as well as overexpression have regulated. Essential cellular processes, including differ- been reported in various malignancies. Subsequently, two entiation and response to extracellular stimuli, depend CDK12 inhibitors were developed independently. Tey on regulation at the transcriptional level [3]. In addition, will be instrumental for studying physiological function precise coordination of transcription with other events, of CDK12 and could be tested as anti-cancer drugs [7, such as mRNA processing, splicing, chromatin remod- 8]. In this review, we summarize current knowledge con- eling, and modification of histones is crucial for normal cerning CDK12 role in tumorigenesis and its therapeutic potential.

*Correspondence: [email protected] Department of Chemistry and Toxicology, Veterinary Research Institute, Hudcova 296/70, Brno 621 00, Czech Republic

© The Author(s) 2017. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/ publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Paculová and Kohoutek Cell Div (2017) 12:7 Page 2 of 10

CDKs in transcription regulation displays high sequence identity to CDK12 kinase domain, Transcription is a complex process coordinated by but its sequence is unrelated on C- and N- terminus. numerous factors. Posttranslational modifications (par- CDK13 associates with Cyclin K and forms a separate ticularly phosphorylation) of RNA pol II CTD constitute complex [12]. Similarly to CDK12, CDK13 is capable one of the crucial mechanisms of transcription regulation of phosphorylating CTD [12]. Nevertheless, CDK13 is [3, 4]. RNA pol II is a multi-subunit complex responsible much less studied than CDK12 and its function is less for RNA synthesis of eukaryotic genes coding most pro- clear. Due to the sequence similarity, one can expect a teins and small RNAs. Modifications of CTD, which con- redundancy or overlap in functions of these two kinases. sists of repeated YSPTSPS heptapeptides, form patterns In yeast (S. cerevisiae), there are two kinases capable that enable specific binding of various factors coordinat- of phosphorylating Ser2 of CTD, Ctk1 and Bur1. Before ing transcription as well as co-transcriptional pre-mRNA CDK12 and CDK13 were discovered, it was assumed that processing [3, 4]. Phosphorylation of serines, threonine CDK9 is the only metazoan orthologue of Ctk1 and Bur1. and tyrosine, among other post-translational modifica- Based on evolutionary and functional evidence Bar- tions, is essential for the transition between individual tkowiak et al. identified Drosophila CDK12 and human phases of transcription, such as initiation, elongation, CDK12 and CDK13 as an ortholog of yeast Ctk1, while termination, and individual steps of pre-mRNA process- Bur1 is CDK9 orthologue [11, 15] Lsk1, less studied Ser2 ing. Te majority of kinases phosphorylating CTD of CTD phosphorylating protein, is CDK12 ortholog in S. RNA pol II belong to the group of transcription-related pombe [11]. CDKs. CDK7 is a part of transcription initiation factor TFIIH. CDK9, responsible for the activity of P-TEFb, is CDK12 in transcription regulation associated with early elongation [3, 4]. Similarly to CDK9, CDK12 is associated with transcrip- tion elongation and is able to phosphorylate the RNA CDK12 structure, homologs and associating cyclin pol II CTD serine at position 2 (Ser2) in Drosophila In 2001, a new transcription-related kinase was discov- [11] and in human cells [12, 13]. However, downregula- ered. It was described as a Cdc2-related kinase with an tion of CDK12 activity does not affect the global tran- arginine/serine-rich (RS) domain (CrkRS). Te CrkRS scription rate, and when CDK12 is depleted from cells, consists of 1490 amino acids and has a kinase domain, transcription of a unique subset of genes is altered (inter- proline-rich regions and a serine-rich domain, which is estingly mainly the genes necessary for DDR). Depletion typical for splicing factors from the SR protein family [9]. of CDK12 and Cyclin K results in decreased expression Later on, cyclins L1 and L2 were described as CDK12- of long genes (> 10 kb) and genes with higher number associating cyclins and the CrkRS was renamed to CDK12 of exons. Observations have led to the hypothesis that [10]. Nevertheless, the cloning of Drosophila CDK12 CDK12 is a kinase that promotes transcription of a set of led to identification of cyclin K as a bona fide CDK12- specific genes [12]. Supporting this hypothesis, CDK12 associating cyclin. Te cyclin K/CDK12 complex was was found to be necessary for the expression of Nrf2- demonstrated to phosphorylate the CTD of RNA pol II dependent genes in Drosophila cells, where CDK12 does in vitro, and CDK12 was established as a CTD kinase not affect the overall transcription but is rather involved [11]. Te association between cyclin K and CDK12 was in the transcriptional stress response [16]. further confirmed by mass spectrometry and immu- Nevertheless, recent studies provide arguments against noprecipitation in mammalian cells [12, 13]. Addition- a hypothesis that CDK12 is a gene-specific CDK kinase. ally, the ability of CDK12 to phosphorylate the CTD of According to this concept, CDK12 is actively recruited RNA pol II was clearly demonstrated [12]. Analogically to the body of transcribed genes by fully operated Pol to CDK9, CDK12 is expected to phosphorylate additional II associated factor 1 (Paf1) right after paused RNA pol substrates other than CTD, such as transcription or splic- II is released into the productive elongation phase [17]. ing factors, which may be critical for CDK12 role in regu- Also, when specific anti-CDK12 antibody was used in a lation of transcription and related processes. Additional ChIP-Seq experiment, it was found that CDK12 binds CDK12 phosphorylation targets and related biological to promoters and bodies of protein-coding genes and to functions remain to be determined. Te crystal structure active transcription enhancers, with the ChIP-seq sig- of CDK12 was later described, opening new possibili- nal overlapping with the RNA pol II signal [8]. Te same ties to study its function and its potential as a drug tar- group developed a specific CDK12 inhibitor and identi- get and to develop specific CDK12 inhibitors [14]. Te fied CDK12-responsive genes in microarray experiments. closest CDK12 human homologue is CDK13 (also known Critically, a lower dose of the CDK12 inhibitor reduced as CDC2L5, CHED). It contains a kinase domain which transcription of core DDR genes (including BRCA1, Paculová and Kohoutek Cell Div (2017) 12:7 Page 3 of 10

FANCF and ERCC4) and higher inhibitor dose decreased kinase-specific for a unique set of genes, such as DDR expression of super-enhancer-associated genes compared or super-enhancer-associated genes? Similarly to CDK9, to genes associated with typical enhancers [8]. CDK12 could bind and phosphorylate also additional An additional aspect related to CDK12 function in factors and thus regulate transcription employing CTD transcription is the specificity of CDK12 in relation to the phosphorylation independent mechanisms. specific serine within CTD. In the classical view, CDK12 phosphorylates CTD Ser2 in vivo and in vitro. In contrast CDK12 function in splicing to CDK9, CDK12 is responsible for Ser2 phosphorylation Since its discovery, it has been known that CDK12 co– on the 3′ prime end of genes [8]. In an attempt to clarify localizes with SC35 (also known as SRSF2 or SFRS2), a CDK12 substrate preference, the Geyer group performed spliceosome component, and contains the RS domain in vitro kinase assays and immunoprecipitation experi- which is typical for RNA-interacting and splicing fac- ments. Tey described CDK12 ability to phosphorylate tors [9]. Supporting CDK12 involvement in RNA-splicing both CTD Ser2 and Ser5; however, it needs pre-phos- machinery, three studies independently identified sev- phorylation of CTD on Ser7 for optimal activity [14]. eral factors of the splicing apparatus and components of Nonetheless, inhibition of CDK12 in HeLa cells led to a nuclear speckles to be putative CDK12-associating part- severe block of cell growth but global phosphorylation of ners based on mass spectrometry analyses [18, 22, 23]. individual CTD related serines was affected only moder- Nevertheless, most of these associations have not yet ately [18]. Finally, CDK12 inhibition leads to a reduction been confirmed by immunoprecipitation with endog- in global Ser2 but not Ser5 or Ser7 phosphorylation of enous CDK12. One example of CDK12 involvement the CTD [8]. To date, CTD substrate specificity studies in splicing is based on the observation that depletion have been dependent on antibodies against specific CTD of CDK12 leads to dysregulated alternative splicing of modifications, such as H5, H14, and antibodies prepared serine/arginine splicing factor1 (SRSF1). In addition to by the Eick group [19]. Te specificity of these antibodies its association with splicing factors, CDK12 was dem- seems to be affected by neighboring modifications [18], onstrated to co-immunoprecipitate proteins of exon which could bias the results and make it challenging to junction complexes and RNA-binding proteins [22]. In draw conclusions. A cellular system suitable for mass addition, Drosophila CDK12 is involved in the alterna- spectroscopy analyses of CTD of RNA pol II modifica- tive splicing of neurexin IV in coordination with mRNA– tions was recently developed. It provides a promising tool binding protein HOW during Drosophila nervous system for elucidating the specificity of CDK12 and other CDKs development [24]. in relation to individual CTD serines [20]. A recent study confirmed that CDK12 is involved in In addition to regulating the elongation phase of tran- splicing. Te authors described interaction of CDK12 scription, CDK12 participates in transcription termina- with spliceosome components and splicing regulatory tion. Polyadenylation-coupled phosphorylation of Ser2 factors using immunoprecipitation followed by mass at the 3′ end of the MYC gene by CDK12 is necessary spectroscopy [25]. In RNA-seq experiments, Tien et al. for recruitment of polyadenylation factor CstF77 and is described a new role of CDK12 in splicing: CDK12 reg- therefore necessary for effective transcription termina- ulates alternative last exon splicing, gene- and cell type- tion [21]. Similarly, CDK12 depletion leads to reduced specific specialized type of alternative splicing. In breast Ser2 phosphorylation and cleavage stimulation factor 64 cancer cells, depletion or overexpression of CDK12 leads (CstF64), thereby leading to impaired 3′ end processing to altered alternative last exon splicing of a subset of of the c-FOS gene after activation of EGF signaling [22]. genes and may contribute to tumorigenesis [25]. Tese two studies clearly demonstrated how regulation Despite a growing body of evidence supporting CDK12 of transcription is coupled to pre-mRNA processing by involvement in splicing, the precise role of CDK12 in this recruiting different factors to modified CTD. process as well as other co-transcriptional events is yet to It is evident that transcription elongation and ter- be elucidated. Description of additional binding factors mination serve as regulatory steps in gene expression. and new potential phosphorylation substrates may clarify Dysregulation of these processes may alter levels of the precise function of CDK12 in this process. CDK12 tumor suppressors or oncogenes and possibly result in could form a functional link between transcription regu- tumorigenesis. Nevertheless, the exact role of CDK12 lation and co-transcriptional pre-mRNA splicing. Alter- in transcription regulation is not fully understood. native splicing affects a large number of transcripts in Te fundamental question remains unanswered as to mammals and provides regulation for the majority of cel- whether CDK12 affects transcription globally or if it is lular processes. Aberrant splicing of various regulatory Paculová and Kohoutek Cell Div (2017) 12:7 Page 4 of 10

factors also leads to tumorigenesis [26–28] providing one a compromised ability to effectively execute HR [36]. of the explanatory roles for CDK12 deficiency in tumori- CDK12 −/− cells derived from mouse blastocysts show genesis via splicing dysregulation. decreased expression of DDR genes and increased levels of DNA damage [33]. Administration of CDK12 inhibi- CDK12 in development tor THZ531 also reduces expression of DDR-associated Even though CDK12 is ubiquitously expressed, the genes [8]. CDK12 protein level differs in particular tissues. High However, a recent study pointed out that the DNA human CDK12 levels can be found in testes, ovaries, leu- damage pattern in ovarian tumors with CDK12 loss is kocytes and adrenal gland, measured by mRNA levels different than in the case of loss of HR-associated genes. [29]. High mouse CDK12 protein levels are in testes as Samples bearing inactive CDK12 have been shown to well as in highly proliferative tissues and mouse embry- contain large tandem duplications rather than markers of onic stem cells [30]. Tis suggests that CDK12 could have impaired HR [37]. tissue-specific roles in cellular commitment and differen- Cyclin K (in addition to BRCA1) was found in a global tiation. Several studies have pointed out CDK12 function screen for genes sensitizing cells to the DNA-damaging in neuronal development and differentiation [31–33]. drug camptothecin [38]. Furthermore, downregulation of For instance, depletion of CDK12 (and CDK13) leads to CDK12 leads not only to spontaneous cell death but also reduced axonal outgrowth mediated probably by lowered to sensitization of cells to various DNA-damaging agents CDK5 expression [32]. Ser2 phosphorylation in C. ele- such as etoposide, mitomycin C and camptothecin [12]. gans germline depends on the activity of CDK12/cyclin K Also, mutant, inactive forms of CDK12 sensitize cancer rather than on CDK9 [34]. cells to cisplatin [39]. CDK12 deficiency sensitizes cells As is evident from mouse studies, both CDK12 and its to inhibitors of PARP1/2, an important factor involved in associating cyclin K are essential for early embryogenesis DNA repair (discussed further in more detail) [7, 39, 40]. in mice. In vitro cultured CDK12 −/− blastocysts fail Despite the fact that CDK12 role in DDR is not yet fully to undergo inner cell mass outgrowth due to increased understood, it is clear that CDK12 is necessary for main- apoptosis and impaired repair of DNA damage [33]. taining genomic stability and functional DDR, particu- CDK12 associating cyclin K is highly expressed in larly HR promoted DNA damage repair. Impaired DDR murine embryonic stem cells but not in their differenti- and accumulation of DNA damage is a typical hallmark ated derivatives. Te cyclin K protein level decreases with of cancer [41], which directly links CDK12 deficiency to differentiation and correlates with levels of Oct4, Sox tumorigenesis. and Nanog proteins known to be necessary for maintain- ing stemness. Tese observations suggest that cyclin K/ Transcriptional CDKs and cancer CDK12 and also cyclin K/CDK13 complexes take part in CDKs are principal regulators of the cell cycle and con- maintaining the self-renewal capacity of murine embry- sequently participate in control of cell proliferation. Each onic stem cells [35]. CDK plays a distinct role in this process and is activated One of the features of various tumors is stemness and in coordination with multiple factors. Dysregulation cell dedifferentiation. Since CDK12 maintains a dediffer- of CDKs is a feature typical of a large number of tumor entiated state in mouse embryonic stem cells, one could types. Hence, CDKs are attractive targets for cancer ther- envision a scenario where CDK12 maintains the dedif- apy and numerous CDK inhibitors have been synthesized ferentiated state of cancer stem cells. High CDK12 activ- and tested as anti-cancer drugs [1, 42]. In addition to cell ity would therefore accelerate tumor progression and cycle-associated CDKs, transcription-associated CDKs therapy resistance, paradoxically to its proposed role as a have emerged as prospective therapeutic targets, exploit- tumor suppressor. ing the so-called transcriptional addiction, According to this concept, cancer cells depend on dysregulated tran- CDK12 role in DNA damage response scriptional programs maintained by principal transcrip- Even though the exact function of CDK12 is not fully tional regulators, among them transcription-associated understood, it is evident that it plays a significant role CDKs [5]. An increasing number of studies have pointed in DDR by affecting the expression of genes involved out the connection between individual transcriptional in homologous recombination (HR) promoted DNA CDKs and cancerogenesis [5]. damage repair and probably also other repair pathways CDK9, a kinase responsible for the activity of posi- [7, 12, 36]. Consequently, CDK12 silencing results in tive transcription elongation factor (P-TEFb), regulates increased endogenous DNA damage [12]. Cells express- transition from the initiation to the productive elonga- ing catalytically inactive mutant forms of CDK12 exhibit tion phase of transcription. Overexpression of oncogenic Paculová and Kohoutek Cell Div (2017) 12:7 Page 5 of 10

transcription factor c-Myc leads to increased activity of a patient-derived xenograft model. Tese findings fore- CDK9 and enhances the current transcriptional program see a possibility to use a CDK12 inhibitor to sensitize or by stimulating RNA pol II elongation [43]. Consequently, reverse PARP1/2 inhibitor resistance in tumors [7]. CDK9 inhibitors have been shown to limit proliferation CDK7 inhibitor THZ1 has recently been shown to limit in Myc-overexpressing liver cancer cells [44] and B cell the transcription of factors dependent on super-enhanc- lymphoma [45]. CDK9 is activated in different types of ers, among them MYC proto-oncogenes, and it has been leukemia. MLL, a histone methyltransferase, frequently pre-clinically tested for treatment of lung carcinoma [43], fuses with components of the super-elongation complex T-cell acute lymphoblastic leukemia [44], and triple nega- to form oncogenic factors which activate P-TEFb and tive breast cancer [45]. THZ1 inhibits CDK12 at higher promote transcription. Inhibition of CDK9 subsequently concentrations and its biological effect could be partly limits proliferation of these cells [46]. ascribed to CDK12 inhibition [52]. After CDK13 amplification was described in hepatocel- A selective and potent CDK12/13 inhibitor TZH531 lular cancer, CDK13 was consequently proposed to be an was recently developed [8]. Te attempt to synthesize oncogene [47]. A high level of CDK11 in breast cancer a CDK12 inhibitor was based on THZ1, which cova- correlates with clinicopathological parameters. CDK11 lently binds cysteine 312 located on an extension of downregulation limits cell proliferation and migration in the CDK7 kinase domain. CDK12 and CDK13 possess breast cancer cell lines. Targeting CDK11 has been pro- cysteines 1039/1017 in a similar extension close to the posed for breast cancer treatment [48]. kinase domain. Te authors exploited structural dif- In conclusion, overstimulation of transcription-asso- ferences between CDK7 Cys312 and CDK12/13 Cys ciated CDKs promotes proliferation of various cancer 1039/1017 and screened for an inhibitor specific solely types. Analogously, cancer cells may depend on CDK12 for CDK12/13. THZ531 selectively inhibits CDK12/13 and thus it can serve as a therapeutic target. Importantly, activity 50 times more efficiently than CDK7 or CDK9. CDK12 overexpression has been documented in breast In cells, THZ531 induced apoptosis, inhibited elonga- tumors [49, 50]. tion of genes and led to reduced expression of DDR and super-enhancer dependent genes. THZ531 exhibited an CDK12 inhibitors antiproliferative effect in Jurkat T-cell acute lymphoblas- Te anti-tumor potential of various CDK inhibitors has tic leukemia cells [8]. Yet, respective contributions of been tested in clinical trials. In addition to the pan-selec- CDK12 and CDK13 to observed biological effects are not tive inhibitors such as flavopiridol and roscovitine, inhib- known [40]. itors showing specificity for individual CDKs have been In summary, CDK12 inhibitors show promise as anti- developed targeting cell cycle- as well as transcription- cancer drugs, either as a stand-alone treatment or in linked CDKs [1]. Taking into account that CDK12 plays a combination with other compounds such as PARP1/2 critical role in multiple cellular processes and is mutated inhibitors. or overexpressed in various types of cancer, CDK12 inhi- bition emerges as a favorable strategy for cancer treat- CDK12 mutations in high-grade serous ovarian ment. Two studies recently described CDK12 inhibitors cancer (HGSOC) with different chemical structure and specificity range. Genomic instability is a typical feature of various malig- Initially, dinaciclib (SCH 727965) was described as nancies. Mutations in DDR genes resulting in accumula- a potent inhibitor of CDK2, CDK5, CDK1 and CDK9 tion of DNA damage are often driving progressive events exhibiting an anti-proliferative effect in various cell in cancerogenesis. Defective DNA repair machinery lines [51]. Johnson et al. (2016) discovered that dinaci- results in accumulation of mutations and accelerated clib potently inhibits also CDK12 with IC50 comparable cancer transformation and progression [41]. to CDK9. Dinaciclib administration, similarly to CDK12 HR deficiency and genomic instability are characteris- silencing, leads to reduced expression of HR genes and tic for about 50% of HGSOC [53]. HR-associated genes reduced RNA pol II CTD Ser2 phosphorylation, and the such as BRCA1 or BRCA2 are mutated most frequently effects of dinaciclib are thus reminiscent of CDK12 inhi- and represent typical tumor suppressors. In addition bition. Moreover, BRCA1 wild-type cells treated with to p53, BRCA1 and BRCA2, CDK12 is one of only nine dinaciclib exhibit compromised HR, which conveys a recurrently mutated genes; it is mutated in about 3% of sensitivity to PARP1 inhibitors. Importantly, combined HGSOC cases [53]. Additional studies have described treatment with PARP1/2 inhibitor veliparib and CDK12 these mutations in more detail. Mostly they are homozy- inhibitor dinaciclib efficiently inhibited tumor growth in gous point mutations in the CDK12 kinase domain Paculová and Kohoutek Cell Div (2017) 12:7 Page 6 of 10

leading to the loss of CDK12 function [36]. Mutant to PARP1/2 inhibitors. CDK12 loss in a tumor could CDK12 forms have compromised ability to phospho- serve as another marker for treatment with PARP1/2 rylate RNA pol II CTD, and cells display impairment in inhibitors or additional inhibitors of DDR network, as HR promoted DNA repair. Tis is caused predominantly well as with other DNA-damaging compounds. by an inability to bind cyclin K [36, 39]. Furthermore, two studies have pointed out that in patients samples, CDK12 dysregulation in breast cancer BRCA1, BRCA2 and CDK12 mutations were mutually In addition to HGSOC, several studies have shown dys- exclusive [40, 54]. Tis observation strongly suggests that regulation of CDK12 in individual subtypes of breast BRCA1 and CDK12 participate in one regulatory path- cancer. way and supports the hypothesis that CDK12 controls Triple-negative breast cancer (TNBC) samples contain expression of BRCA1 and other DDR genes. Tis sug- mutational spectrum similar to ovarian cancer. Tese gestion can be further supported by the fact that more of tumors do not amplify any characteristic receptor (ER, the key DDR proteins were observed to be deregulated in HER2, or PR) but display mutations in DDR genes that patient tumor samples bearing CDK12 mutations [36]. promote genomic instability. Recurrently mutated genes Considering these observations, CDK12 was suggested to include p53 (80% of cases) and BRCA1 (30% of cases) be a tumor suppressor. [59]. CDK12 mutations were identified in 1.5% of TNBC cases [59, 60]. TNBC patients with defective HR (includ- CDK12 loss confers sensitivity to PARP1/2 ing loss-of-function mutated CDK12) may benefit from inhibitors treatment using PARP1/2 inhibitors [59]. Despite its contribution to tumor promotion, genomic A large number of breast tumors are dependent on an instability also provides an opportunity for cancer ther- overexpressed estrogen receptor (ER) and therefore its apy. Inhibitors of PARP1, a protein which participates in targeted inhibition is used for counteracting the tumor DDR, require defective HR for their anti-cancer activ- growth [61]. CDK12 silencing modifies the sensitivity of ity. As defective HR is common in some tumors, PARP1 ER-positive cells to tamoxifen, a drug blocking ER sign- inhibition is becoming synthetically lethal to such cells aling. CDK12 downregulation activates the mitogen- [41]. Hence, loss-of-function mutations of HR regula- activated protein kinase (MAPK) pathway, which in turn tors BRCA1 and BRCA2 are markers of application of leads to loss of ER dependency and causes resistance to PARP1/2 inhibitors-based therapy [55, 56]. Although tamoxifen [62]. PARP1/2 inhibitors have recently been translated into Another subtype of breast cancer is characterized by clinics, certain tumors develop resistance to PARP1/2 amplification of oncogene HER2 (also known as ERBB2 inhibitors, and new strategies for restoring PARP1/2 or EGFR2), a tyrosine kinase receptor, which stimulates inhibitors sensitivity are needed [57]. cell proliferation and inhibits apoptosis [59]. In breast A synthetic lethality screen determined CDK12 to be cancer, HER2 is a part of the frequently amplified and one of the additional genes conferring sensitivity to the overexpressed 17q12-q21 locus [63]. In addition to HER2, PARP1/2 inhibitor olaparib [40]. Ovarian cancer cell lines 17q12-q21 amplicon commonly contains several neigh- with lower expression of CDK12 are more sensitive to boring genes including MED1, GRB7, MSL1, CASC3 and olaparib treatment, and downregulation of CDK12 leads TOP2A [50, 59]. Interestingly, the HER2 amplicon also to increased olaparib sensitivity. Te therapeutic effect contains the CDK12 gene in 71% of cases [50, 64]. of olaparib on CDK12-silenced tumor cells was con- HER2-amplified tumors may benefit from therapy firmed in vivo in xenograft experiments [40]. Increased based on usage of antibodies against HER2 receptor sensitivity of CDK12-compromised cells to cisplatin, (trastuzumab, pertuzumab) or tyrosine-kinase inhibitors the alkylating agent melphalan, and the PARP1 inhibi- (lapatinib) [59]. In addition to HER2, overexpression of tor veliparib was observed in a CDK12-silenced ovarian co-amplified genes might also have an impact on breast cancer cell line [39]. In addition, Her2-positive breast cancer development. Additional genes involved in the cancer cells with downregulated CDK12 display sensitiv- 17q12-q21 amplicon might therefore be oncogene can- ity to PARP1/2 inhibitors [58]. Finally, the CDK12 inhibi- didates [63]. Moreover, amplification of additional genes tor dinaciclib in combination with the PARP1/2 inhibitor in the 17q12-q21 locus might be responsible for resist- veliparib resulted in inhibition of tumor growth in vitro, ance of certain tumors to HER2-targeted therapy. Tese in vivo and in a patient-derived xenograft model [7]. oncogenic factors, including CDK12, represent poten- Consistent with the fact that CDK12 is necessary for tial druggable targets [64]. Mertins and colleagues ana- expression of HR genes, loss of CDK12 confers sensitivity lyzed the proteome of breast cancer samples and found Paculová and Kohoutek Cell Div (2017) 12:7 Page 7 of 10

CDK12 amplification on mRNA and protein level as well In a different context, CDK12 has properties that as increased CDK12 phosphorylation in HER2-amplified resemble oncogenes. CDK12 amplification, resulting tumors [50]. Tese observations indicate elevated CDK12 in its overexpression, correlates with more aggressive activity in these tumors, and CDK12 has been proposed tumor progression in HER2-positive breast cancers [49]. as an additional druggable target in HER2-amplified CDK12 is highly active in these tumors, and it has been breast tumors [50]. In parallel, the diFiore group per- proposed as a druggable target in HER2-amplified breast formed a broad screen of serine/threonine kinases with cancer [50]. Tis idea might be supported by the fact that altered expression in several human cancers, among CDK12 inhibition limits the growth of cancer cells [8, them breast cancer. Tey found that CKD12 is upregu- 66]. Te concept of transcriptional addiction describes lated in HER2–positive breast cancer samples and dem- dependency of cancer cells on a certain transcriptional onstrated a strong correlation between CDK12 level and regulator, which maintains an altered transcriptional pro- high tumor grade. Tey also proposed that a high CDK12 gram [5]. In line with this concept, inhibitors of CDK12- level could serve as a prognostic marker [49]. related transcriptional kinases CDK7 and CDK9 reduce In 13% of cases, rearrangements in 17q12-q21 ampli- proliferation of cancer cells and are being tested as anti- cons lead to disruption of the CDK12 gene and resulted tumor drugs. Similarly to CDK7, CDK12 inhibitors limit in CDK12 loss of function and PARP1/2 inhibitor sen- the expression of super-enhancer-associated oncogenic sitivity of these cells. Tis observation suggests that the transcriptional factors [8, 67]. Tis suggests that certain subset of HER2-amplified patients with disrupted CDK12 tumors might be transcriptionally addicted to CDK12 could benefit from PARP1/2 inhibitor treatment [58]. and so CDK12 inhibition might be a promising anti-can- Additionally, a fusion form of the likely nonfunctional cer strategy. A recent study described that CDK12 over- CDK12 gene was also found in a micropapillary breast expression affects alternative last exon splicing. Terefore cancer sample [58]. CDK12 overexpression can increase the invasiveness of An increasing number of studies describe loss of func- a breast cancer cell line, by decreasing the expression of tion or amplification of CDK12 in breast cancer sam- the long isoform of DNAJB6 [25]. ples. Loss of function in CDK12 may lead to genomic Taken together, the dual role of CDK12 in cancerogen- instability and be predictive of PARP1/2 inhibitor treat- esis could be explained by the fact that CDK12 is essential ment. Tumors displaying CDK12 amplification, on the for expression of both tumor suppressors and oncogenes, other hand, may be dependent on its overexpression and and it participates in multiple cellular processes. In dif- CDK12 may provide a new therapeutic target for breast ferent tissues and cell types, the process of tumorigenesis malignancies. depends on amplification of particular oncogenes or loss of suppressors. Consequently, CDK12 can act as a tumor Is CDK12 a tumor suppressor or does it have suppressor or its amplification can contribute to cancero- oncogenic properties? genesis depending on cellular context (Fig. 1). Loss of tumor suppressors and addiction to oncogenes are mechanisms driving cancer onset and progression. A growing number of studies describe mutations or ampli- Conclusions fication of the CDK12 gene in tumor samples. Tese data CDK12 is a transcription-associated CDK essential for may seem contradictory, since CDK12 functions either multiple cellular processes, including splicing, differen- as a tumor suppressor or it has features that resemble an tiation and DDR. Despite the fact that CDK12 has been oncogene. extensively studied, our understanding of its functions In HGSOC and TNBC, CDK12 is a tumor suppres- remains limited. In vitro models will be instrumental to sor. CDK12 is necessary for expression of DDR genes identify CDK12-associating factors and additional kinase and it is essential for HR mediated DNA repair [12, 50]. targets and to elucidate whether it is a general transcrip- Consequently, CDK12 loss leads to increased genomic tional regulator or a specific factor for particular sets instability, which is a typical feature of these tumors and of genes. Mouse models recapitulating CDK12 loss or represents one of the hallmarks of cancer progression gain of function will be illustrative in studying particu- [65]. CDK12 loss-of-function mutations or inhibition lar aspects of diseases and development. Elucidation of confers sensitivity of cells to PARP1/2 inhibitors [7, 39, CDK12 functions would lead to better assessment of its 40]. roles during tumorigenesis. Paculová and Kohoutek Cell Div (2017) 12:7 Page 8 of 10

ab Tumor suppressor Oncogene

CDK12

CDK12 CDK12 CDK12

CDK12 loss or inhibition CDK12 amplification, overexpression

oncogene HR genes oncogene oncogene

PARPi synthetic lethality CDK12 inhibitor monotherapy

Fig. 1 Role of CDK12 in cancer. a CDK12 has the tumor-suppressive properties. CDK12 loss-of function mutations lead to decreased expression of HR genes resulting in genomic instability and tumorigenesis. CDK12 loss or inhibition sensitizes tumor cells to PARP1/2 inhibitors. b CDK12 has oncogenic properties. CDK12 amplification might lead to increased expression of various oncogenes and consequently participate in tumorigen- esis. Therefore targeting CDK12 with specific inhibitors in these tumors could be beneficiary for patient treatment

CDK12 mutations and amplification in tumors have Authors’ contributions HP and JK wrote the manuscript. Both authors read and approved the final been documented in an increasing number of studies. manuscript. Recently developed CDK12 inhibitors constitute not only powerful research tools but also promising anti-cancer Acknowledgements drugs. CDK12 inhibitor monotherapy could be useful for The authors would like to thank prof. BM Peterlin, Dr. Sabina Sevcikova and cancer patients tumors with overexpressed and activated members of the J. Kohoutek laboratory for critical discussion and their sugges- CDK12. CDK12 inhibition increases sensitivity of cells to tions during the preparation of this review. PARP1/2 inhibitors, thus presenting a potential strategy Competing interests for targeting PARP1/2-resistant tumors. The authors certify that they have no affiliations with, or involvement in, any organization or entity with any financial or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the Abbreviations subject matter or materials discussed in this manuscript. cdk: cyclin-dependent kinase; CTD: C-terminal domain; RNA pol II: RNA polymerase II; DDR: DNA damage response; CrkRS: Cdc2-related kinase with Availability of data and materials an arginine/serine-rich (RS) domain; HR: homologous recombination; P-TEFb: Not applicable. positive transcription elongation factor b; Paf1: pol II associated factor 1; Nrf: nuclear respiratory factor; CstF77: cleavage stimulation factor 77 kDa Consent for publication subunit; CstF64: cleavage stimulating 34 factor 64; SC35: SR splicing factor 2; The authors agree with publishing this manuscript. SFRS1: serine/arginine splicing factor1; c-MYC: MYC proto-oncogene; HGSOC: high-grade serous ovarian cancer; PARP: poly (ADP-ribose) polymerase; TNBC: Ethics approval and consent to participate triple-negative breast cancer; ER: estrogen receptor; HER-2: human epidermal Not applicable. growth factor receptor 2. Paculová and Kohoutek Cell Div (2017) 12:7 Page 9 of 10

Funding 20. Schüller R, Forné I, Straub T, Schreieck A, Texier Y, Shah N, et al. This study was supported by the Ministry of Health of the Czech Republic, Heptad-specific phosphorylation of RNA polymerase II CTD. Mol Cell. Grant 16-34152A and by the Ministry of Agriculture, Grant RO0517 to J.K. 2016;61:305–14. 21. Davidson L, Muniz L, West S. 3′ end formation of pre-mRNA and phosphorylation of Ser2 on the RNA polymerase II CTD are reciprocally Publisher’s Note coupled in human cells. Genes Dev. 2014;28:342–56. Springer Nature remains neutral with regard to jurisdictional claims in pub- 22. Eifler TT, Shao W, Bartholomeeusen K, Fujinaga K, Jäger S, Johnson JR, lished maps and institutional affiliations. et al. Cyclin-dependent kinase 12 increases 3′ end processing of growth factor-induced c-FOS transcripts. Mol Cell Biol. 2015;35:468–78. Received: 10 July 2017 Accepted: 16 October 2017 23. Liang K, Gao X, Gilmore JM, Florens L, Washburn MP, Smith E, et al. Char- acterization of human cyclin-dependent kinase 12 (CDK12) and CDK13 complexes in C-terminal domain phosphorylation, gene transcription, and RNA processing. Mol Cell Biol. 2015;35:928–38. 24. Rodrigues F, Thuma L, Klämbt C. The regulation of glial-specific splic- ing of Neurexin IV requires HOW and Cdk12 activity. Dev Camb Engl. References 2012;139:1765–76. 1. Asghar U, Witkiewicz AK, Turner NC, Knudsen ES. The history and future 25. Tien JF, Mazloomian A, Cheng S-WG, Hughes CS, Chow CCT, Canapi LT, of targeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug et al. CDK12 regulates alternative last exon mRNA splicing and promotes Discov. 2015;14:130–46. breast cancer cell invasion. Nucleic Acids Res. 2017;45:6698–716. 2. Bruyère C, Meijer L. Targeting cyclin-dependent kinases in anti-neoplastic 26. David CJ, Manley JL. Alternative pre-mRNA splicing regulation in cancer: therapy. Curr Opin Cell Biol. 2013;25:772–9. pathways and programs unhinged. Genes Dev. 2010;24:2343–64. 3. Eick D, Geyer M. The RNA polymerase II carboxy-terminal domain (CTD) 27. Oltean S, Bates DO. Hallmarks of alternative splicing in cancer. Oncogene. code. Chem Rev. 2013;113:8456–90. 2014;33:5311–8. 4. Jeronimo C, Bataille AR, Robert F. The writers, readers, and func- 28. Sveen A, Kilpinen S, Ruusulehto A, Lothe RA, Skotheim RI. Aberrant RNA tions of the RNA polymerase II C-terminal domain code. Chem Rev. splicing in cancer; expression changes and driver mutations of splicing 2013;113:8491–522. factor genes. Oncogene. 2016;35:2413–27. 5. Bradner JE, Hnisz D, Young RA. Transcriptional addiction in cancer. Cell. 29. Aken BL, Achuthan P, Akanni W, Amode MR, Bernsdorff F, Bhai J, et al. 2017;168:629–43. Ensembl 2017. Nucleic Acids Res. 2017;45:D635–42. 6. Chilà R, Guffanti F, Damia G. Role and therapeutic potential of CDK12 in 30. Xiang X, Deng L, Zhang J, Zhang X, Lei T, Luan G, et al. A distinct expres- human cancers. Cancer Treat Rev. 2016;50:83–8. sion pattern of cyclin K in mammalian testes suggests a functional role in 7. Johnson SF, Cruz C, Greifenberg AK, Dust S, Stover DG, Chi D, et al. CDK12 spermatogenesis. PLoS ONE. 2014;9:e101539. inhibition reverses de novo and acquired PARP inhibitor resistance in 31. Chang TY, Cheng PL. Relay of cyclin-dependent kinases in the regulation BRCA wild-type and mutated models of triple-negative breast cancer. of axonal growth. Exp Neurol. 2015;271:259–61. Cell Rep. 2016;17:2367–81. 32. Chen HR, Juan HC, Wong YH, Tsai JW, Fann MJ. Cdk12 regulates neuro- 8. Zhang T, Kwiatkowski N, Olson CM, Dixon-Clarke SE, Abraham BJ, Greifen- genesis and late-arising neuronal migration in the developing cerebral berg AK, et al. Covalent targeting of remote cysteine residues to develop cortex. Cereb Cortex. 2017;27:2289–302. CDK12 and CDK13 inhibitors. Nat Chem Biol. 2016;12:876–84. 33. Juan H-C, Lin Y, Chen H-R, Fann M-J. Cdk12 is essential for embryonic 9. Ko TK, Kelly E, Pines J. CrkRS: a novel conserved Cdc2-related protein development and the maintenance of genomic stability. Cell Death Dif- kinase that colocalises with SC35 speckles. J Cell Sci. 2001;114:2591–603. fer. 2016;23:1038–48. 10. Chen H-H, Wang Y-C, Fann M-J. Identification and characterization of the 34. Bowman EA, Kelly WG. RNA polymerase II transcription elongation and CDK12/cyclin L1 complex involved in alternative splicing regulation. Mol Pol II CTD Ser2 phosphorylation: a tail of two kinases. Nucl Austin Tex. Cell Biol. 2006;26:2736–45. 2014;5:224–36. 11. Bartkowiak B, Liu P, Phatnani HP, Fuda NJ, Cooper JJ, Price DH, et al. 35. Dai Q, Lei T, Zhao C, Zhong J, Tang Y, Chen B, et al. Cyclin K-containing CDK12 is a transcription elongation-associated CTD kinase, the metazoan kinase complexes maintain self-renewal in murine embryonic stem cells. ortholog of yeast Ctk1. Genes Dev. 2010;24:2303–16. J Biol Chem. 2012;287:25344–52. 12. Blazek D, Kohoutek J, Bartholomeeusen K, Johansen E, Hulinkova P, Luo 36. Ekumi KM, Paculova H, Lenasi T, Pospichalova V, Bösken CA, Rybarikova Z, et al. The cyclin K/Cdk12 complex maintains genomic stability via J, et al. Ovarian carcinoma CDK12 mutations misregulate expression of regulation of expression of DNA damage response genes. Genes Dev. DNA repair genes via deficient formation and function of the Cdk12/CycK 2011;25:2158–72. complex. Nucleic Acids Res. 2015;43:2575–89. 13. Cheng SWG, Kuzyk MA, Moradian A, Ichu TA, Chang VCD, Tien JF, et al. 37. Popova T, Manié E, Boeva V, Battistella A, Goundiam O, Smith NK, et al. Interaction of cyclin-dependent kinase 12/CrkRS with cyclin K1 is Ovarian cancers harboring inactivating mutations in CDK12 display a required for the phosphorylation of the C-terminal domain of RNA poly- distinct genomic instability pattern characterized by large tandem dupli- merase II. Mol Cell Biol. 2012;32:4691–704. cations. Cancer Res. 2016;76:1882–91. 14. Bösken CA, Farnung L, Hintermair C, Merzel Schachter M, Vogel-Bachmayr 38. O’Connell BC, Adamson B, Lydeard JR, Sowa ME, Ciccia A, Bredemeyer AL, K, Blazek D, et al. The structure and substrate specificity of human Cdk12/ et al. A genome-wide camptothecin sensitivity screen identifies a mam- Cyclin K. Nat Commun. 2014;5:3505. malian MMS22L-NFKBIL2 complex required for genomic stability. Mol 15. Bartkowiak B, Greenleaf AL. Phosphorylation of RNAPII: to P-TEFb or not Cell. 2010;40:645–57. to P-TEFb? Transcription. 2011;2:115–9. 39. Joshi PM, Sutor SL, Huntoon CJ, Karnitz LM. Ovarian cancer-associated 16. Li X, Chatterjee N, Spirohn K, Boutros M, Bohmann D. Cdk12 is a mutations disable catalytic activity of CDK12, a kinase that promotes gene-selective RNA polymerase II kinase that regulates a subset of the homologous recombination repair and resistance to cisplatin and transcriptome, including Nrf2 target genes. Sci Rep. 2016;6:21455. poly(ADP-ribose) polymerase inhibitors. J Biol Chem. 2014;289:9247–53. 17. Yu M, Yang W, Ni T, Tang Z, Nakadai T, Zhu J, et al. RNA polymerase II- 40. Bajrami I, Frankum JR, Konde A, Miller RE, Rehman FL, Brough R, et al. associated factor 1 regulates the release and phosphorylation of paused Genome-wide profiling of genetic synthetic lethality identifies CDK12 RNA polymerase II. Science. 2015;350:1383–6. as a novel determinant of PARP1/2 inhibitor sensitivity. Cancer Res. 18. Bartkowiak B, Yan C, Greenleaf AL. Engineering an analog-sensitive CDK12 2014;74:287–97. cell line using CRISPR/Cas. Biochim Biophys Acta. 2015;1849:1179–87. 41. Lord CJ, Ashworth A. The DNA damage response and cancer therapy. 19. Hintermair C, Heidemann M, Koch F, Descostes N, Gut M, Gut I, et al. Nature. 2012;481:287–94. Threonine-4 of mammalian RNA polymerase II CTD is targeted by 42. Lapenna S, Giordano A. Cell cycle kinases as therapeutic targets for Polo-like kinase 3 and required for transcriptional elongation. EMBO J. cancer. Nat Rev Drug Discov. 2009;8:547–66. 2012;31:2784–97. Paculová and Kohoutek Cell Div (2017) 12:7 Page 10 of 10

43. Lin CY, Lovén J, Rahl PB, Paranal RM, Burge CB, Bradner JE, et al. 56. Farmer H, McCabe N, Lord CJ, Tutt ANJ, Johnson DA, Richardson TB, et al. Transcriptional amplification in tumor cells with elevated c-Myc. Cell. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic 2012;151:56–67. strategy. Nature. 2005;434:917–21. 44. Huang C-H, Lujambio A, Zuber J, Tschaharganeh DF, Doran MG, Evans MJ, 57. Montoni A, Robu M, Pouliot É, Shah GM. Resistance to PARP-inhibitors in et al. CDK9-mediated transcription elongation is required for MYC addic- cancer therapy. Front Pharmacol. 2013;4:18. tion in hepatocellular carcinoma. Genes Dev. 2014;28:1800–14. 58. Natrajan R, Wilkerson PM, Marchiò C, Piscuoglio S, Ng CKY, Wai P, et al. 45. Gregory GP, Hogg SJ, Kats LM, Vidacs E, Baker AJ, Gilan O, et al. CDK9 Characterization of the genomic features and expressed fusion genes in inhibition by dinaciclib potently suppresses Mcl-1 to induce durable micropapillary carcinomas of the breast. J Pathol. 2014;232:553–65. apoptotic responses in aggressive MYC-driven B-cell lymphoma in vivo. 59. Network Cancer Genome Atlas. Comprehensive molecular portraits of Leukemia. 2015;29:1437–41. human breast tumours. Nature. 2012;490:61–70. 46. Kryštof V, Baumli S, Fürst R. Perspective of cyclin-dependent kinase 9 60. Shah SP, Roth A, Goya R, Oloumi A, Ha G, Zhao Y, et al. The clonal and (CDK9) as a drug target. Curr Pharm Des. 2012;18:2883–90. mutational evolution spectrum of primary triple-negative breast cancers. 47. Kim H-E, Kim D-G, Lee KJ, Son JG, Song M-Y, Park Y-M, et al. Frequent Nature. 2012;486:395–9. amplification of CENPF, GMNN and CDK13 genes in hepatocellular carci- 61. Hart CD, Migliaccio I, Malorni L, Guarducci C, Biganzoli L, Di Leo A. Chal- nomas. PLoS ONE. 2012;7:e43223. lenges in the management of advanced, ER-positive, HER2-negative 48. Zhou Y, Han C, Li D, Yu Z, Li F, Li F, et al. Cyclin-dependent kinase 11(p110) breast cancer. Nat Rev Clin Oncol. 2015;12:541–52. (CDK11(p110)) is crucial for human breast cancer cell proliferation and 62. Iorns E, Martens-de Kemp SR, Lord CJ, Ashworth A. CRK7 modifies the growth. Sci Rep. 2015;5:10433. MAPK pathway and influences the response to endocrine therapy. Car- 49. Capra M, Nuciforo PG, Confalonieri S, Quarto M, Bianchi M, Nebuloni M, cinogenesis. 2009;30:1696–701. et al. Frequent alterations in the expression of serine/threonine kinases in 63. Kauraniemi P, Kallioniemi A. Activation of multiple cancer-associated human cancers. Cancer Res. 2006;66:8147–54. genes at the ERBB2 amplicon in breast cancer. Endocr Relat Cancer. 50. Mertins P, Mani DR, Ruggles KV, Gillette MA, Clauser KR, Wang P, et al. Pro- 2006;13:39–49. teogenomics connects somatic mutations to signalling in breast cancer. 64. Sircoulomb F, Bekhouche I, Finetti P, Adélaïde J, Ben Hamida A, Bonansea Nature. 2016;534:55–62. J, et al. Genome profiling of ERBB2-amplified breast cancers. BMC Cancer. 51. Parry D, Guzi T, Shanahan F, Davis N, Prabhavalkar D, Wiswell D, et al. 2010;10:539. Dinaciclib (SCH 727965), a novel and potent cyclin-dependent kinase 65. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. inhibitor. Mol Cancer Ther. 2010;9:2344–53. 2011;144:646–74. 52. Kwiatkowski N, Zhang T, Rahl PB, Abraham BJ, Reddy J, Ficarro SB, et al. 66. Bartkowiak B, Greenleaf AL. Expression, purification, and identification Targeting transcription regulation in cancer with a covalent CDK7 inhibi- of associated proteins of the full-length hCDK12/CyclinK complex. J Biol tor. Nature. 2014;511:616–20. Chem. 2015;290:1786–95. 53. Cancer Genome Atlas Research Network. Integrated genomic analyses of 67. Wang Y, Zhang T, Kwiatkowski N, Abraham BJ, Lee TI, Xie S, et al. CDK7- ovarian carcinoma. Nature. 2011;474:609–15. dependent transcriptional addiction in triple-negative breast cancer. Cell. 54. Carter SL, Cibulskis K, Helman E, McKenna A, Shen H, Zack T, et al. 2015;163:174–86. Absolute quantification of somatic DNA alterations in human cancer. Nat Biotechnol. 2012;30:413–21. 55. Bryant HE, Schultz N, Thomas HD, Parker KM, Flower D, Lopez E, et al. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP- ribose) polymerase. Nature. 2005;434:913–7.

Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to find the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript at www.biomedcentral.com/submit