The yeast HPR1 gene has a functional role in transcriptional elongation that uncovers a novel source of genome instability

The yeas HPR1 gene has a unc ional ole in ansc ip ional elonga ion ha unco e s a no el sou ce o genome ins abili y Sebas ia´n Cha´ ez and And e´s Aguile a 1 Depa amen o de Gene´ ica, Facul ad de Biologı´a, Uni e sidad de Se illa, 41012 Se illa, Spain The yeas HPR1 gene plays an impo an ole in genome s abili y, as indica ed by he obse a ion ha hp 1 mu an s ha e high equencies o DNA epea ecombina ion and ch omosome loss. He e we epo ha HPR1 is equi ed o ansc ip ional elonga ion. T ansc ip ion d i en om cons i u i e and egula ed yeas p omo e s canno elonga e h ough he bac e ial lacZ coding egion in hp 1 D cells, bu p og esses e icien ly h ough o he sequences such as yeas PHO5. We show ha HPR1 is no equi ed o ansc ip ion ac i a ion and ha he p e iously epo ed e ec s o hp 1 D on he ac i a ion o di e en p omo e s is a consequence o he incapaci y o hp 1 D cells o elonga e ansc ip ion h ough lacZ, used as epo e . T ansc ip ional de ec s a e also obse ed in yeas DNA sequences o hp 1 D cells in he p esence o he ansc ip ion elonga ion inhibi o 6-azau acil. In all cases, he blockage o ansc ip ion elonga ion in hp 1 D is associa ed wi h bo h he high equency o dele ions and he inc ease in plasmid ins abili y ha we epo he e. The e o e, in addi ion o he iden i ica ion o a new elemen in ol ed in ansc ip ional elonga ion, ou wo k p o ides e idence o a new sou ce o genomic ins abili y. [Key Wo ds: HPR1; ansc ip ion elonga ion; ansc ip ion-induced ecombina ion; genomic ins abili y; DNA epea s; lacZ epo e ] Recei ed May 21, 1997; e ised e sion accep ed Oc obe 10, 1997. Gene ic ecombina ion is equi ed o DNA epai in mi- osis, o he gene a ion o gene ic di e si y and o p ope educ ional di ision in meiosis. In addi ion, i is esponsible o ch omosomalabe a ionsassocia edwi h some ypes o cance and he edi a y diseases, many o which occu by ecombina ion be ween DNA epea s. The iden i ica ion o he genes and unc ions ha con- olini ia ion o ecombina ionbe ween epea s is he e- o e essen ial o unde s and he o igin o ch omosome abe a ions and genome ins abili y. One impo an ac o a ec ing he equency o ini ia- ion o ecombina ion is ansc ip ional ac i i y. Recom- bina ion is s imula ed by ansc ip ion as i s shown by he iden i ica ion o HOT1, a sequence om he DNA egion o Saccha omyces ce e isiae equi ed o RNA polyme ase I (Pol I)-dependen ansc ip ion (Keil and Roede 1984; Voelkel-Meiman e al. 1987). Dele ions be ween di ec epea s ha e also been shown o be s imula ed by RNA Pol II-media ed ansc ip ion in yeas (Thomas and Ro hs ein 1989). The ele ance o ansc ip ion in homologous ecombina ion has also been epo ed in mammalian cells (Nickolo 1992; Thy- aga ajam e al. 1995). P obably he mos signi ican con- nec ion be ween ansc ip ion and ecombina ion is he one obse ed in immunoglobulin gene ea angemen s (Blackwell e al. 1986; Laus e e al. 1993). O he ex- amples o ansc ip ion-s imula ed ecombina ion ha e been p o ided in S. ce e isiae (Kla e al. 1981; Ne o- Caspi and Kupiec 1994), Schizosaccha omyces pombe (G imm e al. 1991) and phages and bac e ia (Dul and D exle 1988; Vile e e al. 1992). T ansc ip ional ac i i y may induce ecombina ion because unwinding o he DNA duplex, changes in he local supe coiling o dis up ion o ch oma in s uc u e associa ed wi h ansc ip ion could p o ide a be e ac- cessibili y o ecombina ion p o eins o he ansc ibed DNA, could lead o DNA s uc u es suscep ible o DNA b eaks, whe he o no media ed by nucleases, o could acili a e he pai ing eac ion (McCo mack and Thomp- son 1990; D o¨ge 1993; Ko ani and Kmiec 1994). O he explana ionscanno be excluded, howe e , such as a ole o he ansc ip ion machine y in ec ui ing ecombina- ion p o eins o ansc ibed genes ha could be espon- sible o an inc ease in ecombina ion, as is he case o ansc ip ion-coupled nucleo ide excision epai ( o e- iew, see Bas hia e al. 1996; Hoeijmake s e al. 1996). In his sense, he ecen iden i ica ion o he Rad51p e- 1 Co esponding au ho . E-MAIL [email p o ec ed]; FAX (34) 5-4557104. GENES & DEVELOPMENT 11:3459–3470 © 1997 by Cold Sp ing Ha bo Labo a o y P ess ISSN 0890-9369/97 $5.00; www.genesde .o g 3459 Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om combina ional epai p o ein in he human RNA Pol II holoenzyme is no ewo hy (Maldonado e al. 1996). De- spi e ou poo unde s anding o he molecula basis o ansc ip ion-induced ecombina ion, i seems clea ha ansc ip ion is an impo an po en ial sou ce o spon a- neous ecombina ion be ween epea s leading o geno- mic ins abili y and needs o be in es iga ed. One gene pa icula ly ele an o he unde s anding o he pu a i e connec ion be ween ansc ip ion and e- combina ion as well as o he con ol o di ec epea ecombina ion in yeas is HPR1. The impo ance o Hp 1p in genome s abili y is suppo ed by he high e- quencies o ecombina ion and ch omosome loss ob- se ed in null hp 1 D mu an s (Aguile a and Klein 1990; San os-Rosa and Aguile a 1994). Hp 1p has been sug- ges ed o be equi ed o ansc ip ion ac i a ion o egu- la ed p omo e s (Fan and Klein 1994; Zhu e al. 1995). Theiden i ica iono mu a ionsindi e en ansc ip ion ac o s as supp esso s o ei he he he mosensi i i y (Fan e al. 1996; Uemu a e al. 1996) o he hype - ecom- bina ion pheno ype o hp 1 D (Pi ua and Aguile a 1996; San os-Rosa e al. 1996) is consis en wi h a unc ional ole o Hp 1p in ansc ip ion. Ou ecen obse a ions ha high a es o dele ions in hp 1 D cells only occu ed be ween di ec epea s in which de ined egions o he in e ening sequence we e ansc ibed (P ado e al. 1997),sugges ha Hp 1p migh ha e a unc ional olein ansc ip ional elonga ion. To elucida e he unc ion o Hp 1p on ansc ip ion, whe he a he ini ia ion o he elonga ion s age, and o unde s and he causes o induc- ion o genomic ins abili y by hp 1 D ,we unde ook an in i o molecula analysis ha has allowed us o show ha : (1) Hp 1p unc ions in ansc ip ional elonga ion, (2) p e ious esul s sugges ing ha Hp 1p is a gene al ansc ip ional ac i a o (Fan and Klein 1994; Zhu e al. 1995) a e explained by he incapaci y o hp 1 D cells o elonga e ansc ip ion h ough helacZsequenceusedas epo e o gene exp ession and no by an incapaci y o hp 1 D cells o ac i a e ansc ip ion a he ini ia ion s age, and (3) he inc eased le els o di ec epea ecom- bina ion and ch omosome loss obse ed in hp 1 D cells is a di ec consequence o he blockage o ansc ip ion elonga ion. Ou wo k no only shows a unc ional ole o Hp 1p in ansc ip ional elonga ion, bu also sugges s a new model o he ini ia ion o genomic ins abili y. Resul s HPR1 is equi ed o lacZ ansc ip ion Exp ession o lacZ unde he con ol o he GAL1 p o- mo e s ongly depends on HPR1 (Fan and Klein 1994; Zhu e al. 1995) (Fig. 1B). To assess whe he o no lacZ could be exp essed in hp 1 D cells, he lacZ coding egion was placed unde he con ol o wo di e en cons i u- i e p omo e s, TEF2 and ADH1, he la e o which had been shown o p omo e ansc ip ion in he absence o Hp 1p (Zhu e al. 1995). In con as o he expec ed e- sul , lacZ was no exp essed om ei he o he wo p o- mo e s (Fig. 1C,D). The lack o b-galac osidase exp es- sion d i en om he GAL1p omo e in hp 1 D , he e o e, may no be caused by a de ec in ansc ip ional ac i a- ion a he p omo e . One possibili y is ha he lack o b-gal exp ession was caused by a de ec in elonga ion h ough lacZ. To e alua e such a possibili y, he yeas PHO5 gene was placed unde he con ol o he GAL1 p omo e . In his case, acid phospha ase was ac i a ed o wild- ype le els in hp 1 D cells (Fig. 1E). The e o e, hp 1 D has no e ec on ac i a ion o he GAL1 p omo e . This i s wi h he obse a ion ha induc ion o he endog- Figu e 1. Exp ession o lacZ usion cons uc s placed in CEN plasmids in wild- ype and hp 1 D cells. The hp 1 D mu a ion abolished ansc ip ion h ough he bac e ial lacZ coding egion, ega dless o he yeas p omo e om which i was ansc ibed, whe eas i did no a ec ansc ip ion h ough he yeas PHO5 coding egion. (A) The lacZ coding e- gion was used o ei he he egula ed GAL1 p o- mo e o he cons i u i e TEF2 and ADH1 p omo - e s. The PHO5 coding egion was used o he GAL1 p omo e . Numbe s below each cons uc e e o he ansla ion s a o each DNA sequence. Exp es- sion o lacZ and PHO5 was de e mined by b-galac- osidase (B–D) o acid phospha ase (E) ac i i ies, e- spec i ely. Fo he GAL1-d i en exp ession o ei he lacZ(B)o PHO5(C),ei he 2% glucose(Glu, shaded ba s, B,C,E) o 2% galac ose (Gal, open ba s, B,C,E) was added o 16-h mid-log phase cul u es in glyce - ol–lac a e syn he ic medium, and enzyma ic ac i i- ies we e assayed 8 h la e . Cha´ ez and Aguile a 3460 GENES & DEVELOPMENT Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om enous GAL1 gene was no a ec ed by hp 1 D (Fan and Klein 1994). The p e ious wo k was pe o med on usion con- s uc s loca ed in cen ome ic plasmids. Ne e heless, we ha e con i med he same esul s in usion cons uc s loca ed in ch omosomes. Thus, Figu e 2 shows ha whe eas acid phospha ase mainly exp essed by he en- dogenousPHO5gene eachedsimila le els o ac i a ion in bo h wild- ype and hp 1 D cells, exp ession o lacZ unde he same PHO5 p omo e was se iously impai ed in hp 1 D cells. Tocon i m ha heabsenceo b-galac osidaseac i i y o hp 1 D cells was caused by ansc ip ional, a he han pos - ansc ip ional, de ec s o lacZ exp ession, we pe - o med No he n analysis o lacZ unde he GAL1 and PHO5 p omo e s. Figu e 3 shows ha lacZ mRNA d i en om he GAL1 p omo e was accumula ed a high le els in wild ype a e induc ion, whe eas i did no accumula e in hp 1 D (Fig. 3A,B). The No he n analysis o hp 1 D cells (Fig. 3A) shows a band co e- sponding o he ull leng h lacZ mRNA, which appea ed wi h bo h he 58- and 38-end p obes o lacZ, and a smea which p esumably con ained incomple e 58lacZ mR- NAs, as i appea ed wi h he 58-end bu no wi h he 38-end lacZ p obe. A 30 min o induc ion, he o al lacZ mRNA in hp 1 D (mainly incomple e 58-end mRNA) was p oduced in simila amoun s as in wild- ype (mainly ull-leng h mRNA). This was consis en wi h he idea ha he GAL1 p omo e ini ia ed ansc ip ion equally well in bo h wild- ype and hp 1 D ,bu ansc ip ion did no p oceed u he down om he 58-end egion o lacZ in hp 1 D .The dec ease o o al lacZ mRNA obse ed a e 30 min could be explained by he low s abili y o such incomple e ansc ip s ( o e iew, see Ross 1995) and an e en ual educ ion in he e iciency o eini ia- ion o ansc ip ion om he GAL1 p omo e as a con- sequence o a block o ansc ip ion in lacZ. Ins ead, PHO5 mRNA was accumula ed in bo h wild- ype and hp 1 D ,and only ull-leng h mRNA was obse ed (Fig. 3C,D). T ansc ip ion is blocked a he lacZ 58end in hp 1D cells Tocon i m ha ansc ip ion couldbepausedo blocked a he lacZ 58-end egion we pe o med ansc ip ional un-on analysis o GAL1–lacZ in pe meabilized cells (Fig. 4). RNA Pol II was paused o blocked in he i s 170 bp o he lacZ 58end, as deduced om he obse a ion ha unde induc ion condi ions (Gal) he amoun o a- diolabeled mRNA hyb idizing wi h he i s 170 bp o lacZ in hp 1 D was ∼70% o he wild- ype le els (Fig. 4A), bu 17%–39% o he nex six downs eam agmen s. As expec ed, he le els o adiolabeled lacZ mRNA bound o each DNA agmen unde ep ession condi- ions (Glu) was signi ican ly lowe o all DNA ag- men s. The only excep ion was he 58-end 170-bp ag- men (Fig. 4B), which bound o simila amoun s o adio- labeled mRNA unde ep ession and induc ion condi ions in bo h wild- ype (101%) and hp 1 D cells (74%). This implies ha he polyme ase is no mally paused a he 58end o lacZ unde ep ession condi ions as ecen ly epo ed (Akh a e al. 1996), and ha he ini ia ion o ansc ip ion is equally e icien in bo h wild- ype and hp 1 D cells. Acco ding o ou esul s, Hp 1p would be equi ed o allow he RNA Pol II o a el a he down om he lacZ 58-end egion unde ac i a ion condi ions, bu no o he es ablishmen o he RNA Pol II a he 58end, as is also he case o Kin28p, S b2p, o he ca boxy- e minal domain (CTD) o RNA Pol II (Akh a e al. 1996). Na u al egula o y blocks o ansc ip ion desc ibed in se e al euka yo ic genes ( o e iew, see Eick e al. 1994; Ben ley1995)a enea hei 58endsand equi e ansc ip- ional ac i a o s o be bypassed by RNA Pol II. In con- as , he lack o lacZ exp ession om ADH1 and TEF2 p omo e s(Fig.1) sugges s ha he block o ansc ip ion a lacZ in hp 1 D cells is p omo e -independen . To con- i m his, we decided o place he lacZ coding sequence in he UTR o he GAL1–PHO5 cons uc igh a e he PHO5s opcodon (see Ma e ialsandMe hods).Thus, he RNA Pol II ansc ip ion machine y has o ansc ibe he comple e 1.4-kb PHO5 coding sequence be o e en e ing he lacZ sequence. As can be seen in Figu e 5, ull PHO5–lacZ message is p oduced in wild- ype cells bu no in hp 1 D cells. A smea co esponding o incomple e messages was obse ed in hp 1 D cells.Such smea is also obse ed in wild- ype cells, as in he GAL1–lacZ con- s uc (see Fig. 3) sugges ing ha , indeed, elonga ion h ough lacZ is ine icien in yeas . A new un-on analy- sis o his GAL1–PHO5–lacZ usion cons uc showed ha elonga ion was mos ly blocked a he 58-end o he lacZ coding sequence in hp 1 D cells: The RNA Pol II engaged a he 58end o lacZ was 37%–53% o he RNA Pol II engaged a he PHO5 coding sequence (da a no shown). This esul is consis en wi h he un-on analy- sis o he GAL1–lacZ cons uc (see Fig. 4) and con i ms ha he incapaci y o hp 1 D cells o make comple e Figu e 2. Exp ession o he ch omosomally loca ed lacZ (A) and PHO5 (B) coding sequences unde he con ol o he PHO5 p omo e in wild- ype and hp 1 D cells. Fo ep ession (+Pi, shaded ba s) and induc ion (−Pi, open ba s) condi ions see Ma- e ialsand Me hods.Acid phospha aseand b-galac osidasewe e assayed in he same cul u e o each sample. O he de ails as in Fig. 1. T ansc ip ional elonga ion and genome ins abili y GENES & DEVELOPMENT 3461 Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om PHO5–lacZ mRNA (Fig. 5) and lacZ mRNA (Fig. 3) is caused by he incapaci y o elonga e ansc ip ion h ough lacZ, ega dless o i s posi ion ela i e o he p omo e om which i is ansc ibed. The e o e, HPR1 has a unc ional ole in elonga ion, and no in ini ia ion o ac i a ion o ansc ip ion. T ansc ip ional elonga ion blocks as a cause o genome ins abili y To con i m ha he blockage o ansc ip ion elonga ion was esponsible o hehype - ecpheno ypeo hp 1 D ,we inse ed he lacZ and PHO5 coding sequences be ween wo 0.6-kb di ec epea s, immedia ely downs eam o a 38-end unca ed copy o LEU2 and immedia ely up- s eam o a 58- unca ed copy o LEU2. In hese con- s uc s, ansc ip ion o bo h lacZ and PHO5 was ini i- a ed a he unique LEU2 p omo e , as de e mined by RNAse A p o ec ion (da a no shown). Consequen ly, RNA Pol II mus ansc ibe 760 bp o LEU2 be o e en- e ing ei he he lacZ o he PHO5 coding sequences. Acco ding o ou expec a ions, lacZ should cause a e- duc ion in he le el o ull mRNA ini ia ed a he LEU2 p omo e and a s ong hype - ecombina ion pheno ype in hp 1 D cells, which a e bo h consequences o blockage o s alling o ansc ip ion elonga ion a lacZ (L–lacZ cons uc s), bu none o hese pheno ypes should be caused by PHO5 (L–PHO5 cons uc s). The esul s con i med ou p edic ions (Fig. 6). T an- sc ip ion in L–lacZ cons uc s ini ia ed a he LEU2 p o- mo e ups eam o he i s epea , a e sed lacZ and e mina ed a he LEU2 e mina o downs eam o he second epea in bo h wild ype and hp 1 D .T ansc ip le els in hp 1 D we e 12 imes lowe han in wild ype i lacZ was ansc ibed in i s na u al di ec ion, and 200 imes i ansc ibed in he opposi e di ec ion, whe eas ecombina ion equencies we e inc eased 44- and 266- old, espec i ely. In he L–PHO5 cons uc s, bo h he pa e n and he le el o ansc ip s we e iden ical in wild ype and hp 1 D ,and no signi ican di e ence was ob- Figu e 3. No he n analysis o GAL1–lacZ and GAL1–PHO5. The No he n analysis and kine ics o induc ion o lacZ (A,B) and PHO5 (C,D) mRNAs d i en om he GAL1 p omo e is shown. Wild- ype (s,B,D) and hp 1 D (h,B,D) ans o man s we e ob ained om o e nigh cul u es in glyce ol–lac a e syn he ic media lacking u acil and dilu ed in iden ical esh media o an OD 600 o 0.5. Galac ose (Gal) was hen added and samples we e aken o No he n analysis a e di e en imes, as speci ied. Fo ep ession condi ions (Glu) o al RNA was isola ed om mid-log phase cul u es in 2% glucose syn he ic media lacking u acil. The DNA p obes usedwe e(lacZ58end) he 0.5-kbBamHI–HpaI agmen o pLGZ con aining he58endo lacZ;(lacZ38end) he0.4-kbP uII agmen o pLGZ con aining he 58- end o lacZ; (PHO5) he 1.5-kb EcoRI–Ps I in e nal PHO5 agmen o pJDB207–PHO5 (Eco); (ACT1) he 0.6-kb ClaI in e nal ACT1 agmen o plasmid pYA301. (AU) a bi a y uni s. Cha´ ez and Aguile a 3462 GENES & DEVELOPMENT Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om se ed in ecombina ion. In one o ien a ion, ansc ip- ionini ia eda heLEU2p omo e , a e sedPHO5 and e mina ed a he PHO5 e mina o wi h simila e i- ciency in bo h wild- ype and hp 1 D cells (Fig. 6). A weak e ec was obse ed on ecombina ion (six old inc ease). In he opposi e o ien a ion, ansc ip ion e mina ed im- media ely downs eam o he i s LEU2 copy once i encoun e ed he PHO5 ansc ip ion e mina o , wi h simila e iciency in bo h wild- ype and hp 1 D .No e ec was obse ed on ecombina ion. We ha e also obse ed ha he blockage o ansc ip- ion a lacZ is associa ed wi h an inc ease in plasmid loss. The eplica i e pRS416 plasmids con aining ei he GAL1–lacZ (p416GAL1–lacZ)o GAL1–PHO5 (pSCh202) (see Figs. 1 and 2) had simila s abili y in hp 1 D as in wild- ype cells unde ep ession condi ions o ansc ip- ion (55%–62% o he cells con ained he plasmids a e 23–25 gene a ions in nonselec i e glucose-based media). Unde induc ion condi ions, howe e , p416GAL1–lacZ was clea ly uns able in hp 1 D cells. The p opo ion o hp 1 D cells ha con ained p416GAL1–lacZ a e 23–25 gene a ions in nonselec i e galac ose-based media was 37 imes lowe han ha o wild- ype cells, whe eas only a h ee old educ ion was ound o pSCh202 (da a no shown). Genome ins abili y and ansc ip ion de ec s o hp 1D cells a e obse ed in yeas DNA sequences in he p esence o 6-azau acil Because ansc ip ional elonga ion block, hype - ecom- bina ion, and plasmid loss we e only associa ed wi h he bac e iallacZsequencein hiss udy,i wasimpo an o in es iga e he ole o Hp 1p in ansc ip ion and ge- nomeins abili y o yeas DNAsequences. Despi eHPR1 Figu e 4. T ansc ip ional un-on analysis in wild- ype and hp 1 D cells. To al RNA wasisola ed omwild- ypeandhp 1 D cells ans o med wi h single-copy p416GAL1– lacZ plasmid unde induc ion (A) and e- p ession (B) condi ions. Two pe cen -galac- ose o 2% glucose was added o yeas cul- u es in glyce ol–lac a e syn he ic medium a an OD 600 o 0.05, 5 h p io o he un-on analysis. The 0.6-kb in e nal ACT1 ag- men and se en di e en DNA agmen s (1–7) om he lacZ coding egion we e im- mobilized in hybond-N+ il e s. The lacZ egion co e ing each o he se en DNA agmen s used is shown a he bo om. In all cases, he pe cen age o adiolabeled mRNA bound o each lacZ agmen was no malized wi h espec o hei co e- sponding le els in galac ose-g own wild- ype cells, aken as 100% o each. The o i- en a ion o lacZ a ows indica es he di ec- ion o ansc ip ion. As nega i e con ol, we used DNA om Salmonella yphimu- ium (no shown). Figu e 5. (A) No he n analysis o PHO5 mRNA. The DNA p obe used o No he n analysis was he 1.5-kb EcoRI–Ps I in- e nal PHO5 agmen o pJDB207–PHO5 (Eco). A bi a y uni s o mRNA we e calcula ed acco ding o he same s anda ds o all expe imen s. (B) The mRNA alues a e gi en wi h espec o RNA le els (see Ma e ials and Me hods). (s) Wild ype; (h) hp 1 D .O he de ails as in Fig. 3. T ansc ip ional elonga ion and genome ins abili y GENES & DEVELOPMENT 3463 Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om no being essen ial o iabili y, we conside ed he pos- sibili y ha Hp 1p had a gene al ole on ansc ip ion elonga ion o yeas genes, as i is he case o PPR2 en- coding elonga ion ac o TFIIS ( o e iew, see Kane 1994). Because i was shown p e iously ha 6-azau acil (6AU) p oduces a deple ion o UTP and GTP esponsible o a educ ion o ansc ip ional elonga ion e iciency, and ha he g ow h o mu an s in PPR2 was sensi i e o 6AU (A chambaul e al. 1992; Exinge and Lac ou e 1992), we decided o de e mine whe he he ansc ip- ion and gene ic ins abili y pheno ypes con e ed by hp 1 D we e also obse ed in yeas genes in he p esence o 6AU. We ied h ee di e en concen a- ions o 6AU (100, 300, and 1000 µg/ml). Bo h wild- ype and hp 1 D cells, in con as o ansc ip ional elonga ion TFIIS − mu an s, we e able o g ow on SC media con ain- ing he h ee di e en concen a ions, al hough hp 1 D g ow h was mo e se e ely a ec ed han wild- ype g ow h (da a no shown). When we analyzed ansc ip- ion o he yeas PHO5 sequence unde he GAL1 p o- mo e , howe e , he kine ics o accumula ion o PHO5 mRNA was se e ely a ec ed in hp 1 D in he p esence o 6AU (Fig. 7). Thus, he le el o ac i a ion o PHO5 mRNA a 908o induc ion in wild- ype cells was se en- old highe han in hp 1 D cells. PHO5 ansc ip le els inc eased up o six old a 908o induc ion in wild- ype cells,no inc easewas obse ed in hp 1 D cellsin hep es- ence o 6AU. The ACT1 ansc ip le els we e also wo- old lowe in hp 1 D e sus wild ype in he p esence o 6AU, indica ing ha 6AU speci ically impai s ansc ip- ion in hp 1 D cells (da a no shown). To con i m ha he obse ed e ec o 6AU on ansc ip ion in hp 1 D cells was caused by impai men o ansc ip ion elonga ion and no by side-e ec s o 6AU, we de e mined whe he guanine (100 µg/ml) could pa ially e e he e ec o 6AU, as shown p e iously o mu an s o TFIIS and no o 6AU-sensi i e mu an s una ec ed in ansc ip ional elonga ion (A chambaul e al. 1992). Figu e 7 shows ha accumula ion o PHO5 mRNA was ees ablished by guanine in he p esence o 6AU. In addi ion, in he L–PHO5 cons uc con aining he PHO5 ORF be ween he leu2 epea s in he o ien a ion ha is ansc ibed, ecombina ion was enhanced 14- old o e he wild- ype le els in he p esence o 6AU, clea ly abo e he inc ease obse ed wi hou 6AU (5.5- old) (Fig. 8). In he L–PHO5 cons uc wi h PHO5 in he o ien a- Figu e 6. T ansc ip ion and ecombina- ion analysis o di ec epea sys ems ca ying lacZ (3 kb) o PHO5 (1.5 kb) cod- ing egions. Aschemeo hedele ionp od- uc o med by ecombina ion be ween he di ec epea s used is shown (A). The diag am o each di ec epea sys em in- dica es he 0.6-kb epea ed sequences (shaded boxes), he o ien a ion o he lacZ and PHO5 coding egions, he LEU2 p o- mo e (P m) and ansc ip ion e mina o s (Te ), and he ansc ip s d i en om he LEU2 p omo e (a ow), whose 38ends ha e been made o coincide wi h he po- si ion o he co esponding band in each gel (B). To al RNA was isola ed om o e - nigh cul u es in syn he ic media lacking yp ophan. The DNA p obes used in he hyb idiza ion expe imen s we e he 598- bp ClaI–EcoRV LEU2 epea , and he 581- bp ClaI in e nal ACT1 agmen . The an- sc ip co esponding o he LEU2 endog- enous ch omosomal band is indica ed. No ansc ip ini ia es in he in e nal lacZ and PHO5 in e nal sequences as de e - mined wi h speci ic lacZ and PHO5 DNA p obes (da a no shown). Cha´ ez and Aguile a 3464 GENES & DEVELOPMENT Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om ion ha is no ansc ibed, he p esence o 6-AU had no e ec on ecombina ion (Fig. 8). This esul clea ly indi- ca es ha ecombina ion is inc eased only when an- sc ip ion is elonga ed h ough PHO5. Indeed, his in- c ease in ecombina ion co ela ed wi h a sligh dec ease o ansc ip le els in hp 1 D cells in he p esence o 6AU (da a no shown). A simila e ec was obse ed wi h he LA cons uc , con aining he ADE2 gene be ween he leu2 epea s (da a no shown), sugges ing ha he e ec o 6AUon hp 1 D cellsmaybe associa ed wi helonga ion h ough yeas DNA sequences. No inc ease o ecombi- na ion in he L–lacZ cons uc was obse ed by he he addi ion o 6AU (Fig. 8), sugges ing ha ansc ip ion- induced ecombina ion in his cons uc eaches i s maximum le els wi hou 6AU. Finally, he use o cen ome ic plasmid pRS314-LA pe mi ed us o de e mine di ec ly by isual inspec ion he e ec o hp 1 D on genomic ins abili y in he p es- ence o 6AU. Figu e 9 shows ha he ADE2 gene was ex emely uns able in hp 1 D cells ea ed wi h 6AU, as shownby i s ed-sec o ingpheno ypecaused by heAde − seg egan s, in con as o un ea ed hp 1 D cells o wild- ype cells, ega dless o being ea ed wi h 6AU. The e- combina ion equency leading o Ade − ecombinan s in pRS314-LA is oo low (<5 × 10 −3 ) o be de ec ed by his sec o ing assay. The e o e, he obse ed ed-sec o ing pheno ype is mos ly caused by plasmid loss, as con- i med by gene ic analysis. As expec ed, he pheno ype o inc eased plasmid loss obse ed in hp 1 D cells in he p esence o 6AU was e e ed by guanine (Fig. 9), con- sis en wi h being caused by ansc ip ion-elonga ion blockage. These esul s indica e ha when ansc ip ion elonga- ion is impai ed, yeas DNA sequences, and no only bac e ial sequencies such as lacZ, become gene ically uns able in hp 1 D cells. Discussion The main conclusion o his wo k is ha HPR1, o igi- nally iden i ied by he hype - ecombina ion pheno ype con e ed by hp 1 mu a ions, pa icipa es in ansc ip- ional elonga ion. We show ha he block o ansc ip- ional elonga ion p oduced in hp 1 D cells is esponsible Figu e 9. Plasmid ins abili y in he p esence o 6AU. Yeas colonies o URA3 + wild- ype and hp 1 D s ains ans o med wi h cen ome ic plasmid pRS314-LA g owing on SC − T p wi h o wi hou 100 µg/ml o 6AU and wi h o wi hou 100 µg/ml o guanine. Figu e 7. T ansc ip ion analysis o a yeas ORF in he p esence o 6AU. No he n analysis (A) and kine ics o induc ion (B)o PHO5 mRNAs d i en om he GAL1 p omo e in he p esence o 6AU wi h and wi hou guanine a e shown. The URA3 + wild- ype and hp 1 D s ains ans o med wi h pSCh202 we e ob- ained om o e nigh cul u es in glyce ol–lac a e syn he ic- comple e media lacking yp ophan and u acil, and dilu ed in iden ical esh media o an OD 600 o 0.5 wi h 100 µg/ml o 6AU wi h and wi hou 100 µg/ml o guanine. Galac ose was added a e 2 h , and samples we e aken o No he n analysis a e di e en imes, as speci ied. The mRNA alues a e gi en wi h espec o RNA le els (B) (see Ma e ials and Me hods). O he de ails as in Fig. 3. Figu e 8. Recombina ion in he p esence o 6AU. Recombina- ionanalysiso he di ec epea cons uc sL–lacZand L–PHO5 ca ying he PHO5 ORF in bo h possible o ien a ions be ween he leu2 epea s (see Fig. 6). Recombina ion equencies we e de e mined in he URA3 + wild- ype (open ba ) and hp 1 D (shaded ba ) s ains ans o med wi h he app op ia e plasmids, g own in media wi h o wi hou 100 µg/ml o 6AU. T ansc ip ional elonga ion and genome ins abili y GENES & DEVELOPMENT 3465 Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om o he high equency o dele ions be ween epea s and he high equency o plasmid-loss. Ou wo k no only shows a ole o Hp 1p in ansc ip ional elonga ion bu alsoindica es ha ansc ip ional elonga ion blocksmay be an impo an sou ce o genome ins abili y and p o- ides a molecula mechanism o explain ansc ip ion- induced ecombina ion. Hp 1p unc ions in ansc ip ional elonga ion Ou expe imen s indica e ha Hp 1p s imula es an- sc ip ional elonga ion h ough lacZ in i o. Because p e- ious epo s sugges ing ha Hp 1p is equi ed o an- sc ip ion ac i a ion a e based on he analysis o exp es- sion o a lacZ epo e used o di e en yeas egula ed p omo e s (Fan and Klein 1994; Zhu e al. 1995), we should ein e p e hose esul s as being caused by he incapaci y o hp 1 D cells o elonga e ansc ip ion h ough lacZ. This is independen o whe he he p o- mo e s used a e cons i u i e o egula ed (Figs. 1 and 2) and whe he he lacZ sequence is p oximal o dis al o he p omo e om which i is ansc ibed (Figs. 3, 5, and 6). Because no e ec on ansc ip ion was obse ed when he yeas PHO5 ORF was used as epo e (Figs. 1 and 2), we conclude ha Hp 1p is no in ol ed in ansc ip- ional ac i a ion. Ou esul s sol e he pa adox o why ac i a ion o endogenous yeas genes is no a ec ed by hp 1 D (Fanand Klein 1994; Zhu e al. 1995). Whenmino e ec s a e de ec ed, as is he case o SUC2 (Zhu e al. 1995), heya elikely o becausedbya educed e iciency in elonga ion. Ce ainly, ou s udy aises se ious con- ce ns abou using lacZ o s udy ansc ip ion in yeas and in alida es p e ious conclusions sugges ing a ole o Hp 1pin ansc ip ionalac i a iono p omo e s(Zhu e al. 1995). The s ong e ec o hp 1 D on ansc ip ional elonga- ion h ough lacZ is no obse ed in yeas genes such as PHO5 a 30°C (Fig. 3), e en hough a weak e ec is de- ec ed a 37°C (S. Cha´ ez and A. Aguile a, unpubl.). The e ec o hp 1 D on elonga ion o yeas genes may be masked by o he edundan unc ions. Indeed, he obse - a ion ha in he p esence o he ansc ip ional elon- ga ion inhibi o 6-azau acil, PHO5 mRNA accumula- ion is s ongly impai ed in hp 1 D mu an s, a pheno ype pa ially e e ed by guanine,sugges s ha indeedHp 1p has a gene al ole in ansc ip ional elonga ion o yeas genes.Because hp 1 D a e iable in6-azau acil, he ole o Hp 1p in ansc ip ional elonga ion mus be di e en om ha o TFIIS, a ac o equi ed o he elie o an a es ed RNA Pol II ( o e iew, see Reines 1994). The unc ion o Hp 1p canno be ela ed o TFIIS, because TFIIS is inhibi ed by Sa kosyl (Reines 1992), used in ou un-on expe imen s. I ha we e he case, no di e ence be ween wild ype and hp 1 D should ha e been obse ed in he pa e n o he pausing o RNA Pol II a lacZ (Fig. 4).Ins ead, Hp 1p migh be equi ed o p e en ing a es o RNA pol II, as i has been p oposed o TFIIF, elongin, o ELL (ele en-nine een lysine- ich leukemia) (P ice e al. 1989; Aso e al. 1995). We do no know he na u e o he signal on which RNA Pol II is blocked in he absence o Hp 1p. DNA bends o a pa icula ch oma in s uc u e o he bac e ial lacZ sequence ( o e iew, see Kane 1994; Ben ley 1995) may de e mine he need o Hp 1p o i s ansc ip ion, in pa icula a he ups eam 170 nucleo ides whe e he majo hp 1-dependen ansc ip ional block maps (Fig. 4). In his sense, i is no ewo hy ha sin1(sp 2) mu a- ions, a ec ed in a HMG1-like gene, supp ess he an- sc ip ional pheno ype o hp 1 D (Zhu e al. 1995) and ha SIN1-2 and he imbalance o his ones H2A/H2B o H3/ H4 p oduced syn he ic le hali y in hp 1 D (Fan and Klein 1994; Zhu e al. 1995). In addi ion, p omo e -p oximal pausing du ing ansc ip ional elonga ion in he human hsp70 gene depends on nucleosome empla es in i o (B own e al. 1996). O he ypes o signals canno be dis- missed, howe e , such as hose depending on RNA s ems (Reede and Hawley 1996). In any case, he an- sc ip ional block caused by hp 1 D a lacZ is di e en om RNA Pol II ansc ip ional pausing in o he euka y- o ic genes, because he la e is con olled by ansac i- a o s as an in eg al pa o he ini ia ion s ep (Yankulo e al. 1994), whe eas he block a lacZ is independen o (1) i s posi ion ela i e o he p omo e , whe he p oxi- mal o dis al (Figs. 5 and 6), and (2) he ype o p omo e o which i is used, whe he egula ed o cons i u i e (Figs. 1 and 2). In his sense, i is pa icula ly ele an he obse a ion ha he lacZ coding sequence abolishes p o- duc ion o ull-leng h ansc ip when used o he 38 un ansla ed egion (38UTR) o PHO5 (Fig. 5). Finally, ou esul s open he possibili y ha he syn- he ic le hali y o hp 1 op1 mu an s (Aguile a and Klein 1990) could be ela ed o he ole o Hp 1p on ansc ip- ion elonga ion, because opoisome ase I is also needed in RNA Pol II ansc ip ion (Schul z e al. 1992). In his sense, i is impo an o no e ha mu a ions in di e en TOP genes also con e hype - ecombina ion o di e en DNA epea s in yeas (Ch is man e al. 1988; Wallis e al. 1989). An in i o- econs i u ed sys em would be e- qui ed o assess whe he he ansc ip ional blocks caused by hp 1 D is caused by nucleosomes, DNA se- quence, o an RNA seconda y s uc u e. In any case, ou esul s show clea ly ha in i o ansc ip ional elonga- ion is blocked a lacZ in a p omo e -independen man- ne andimpai eda DNAsequencessuchasPHO5in he p esence o 6AU. We ind no e idence o a ole o Hp 1p on ansc ip ion ac i a ion o p omo e s. T ansc ip ional elonga ion blocks lead o genome ins abili y Ou wo k shows ha hype - ecombina ion and plasmid ins abili y in hp 1 D cells is p oduced in associa ion wi h a ansc ip ional elonga ion block. This explains ou p e- ious obse a ion ha hype - ecombina ion was depen- den on ansc ip ion p og ession h ough pa icula DNA egions ha include he bac e ial e and amp se- quences(P adoe al.1997).As onginc easeindele ions be ween epea s and plasmid loss is obse ed in hp 1 D cells only when ansc ip ion is elonga ed h ough he lacZ sequence (Fig. 6) o p og esses h ough yeas DNA Cha´ ez and Aguile a 3466 GENES & DEVELOPMENT Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om sequences (Fig. 7) in he p esence o he ansc ip ional elonga ion inhibi o 6AU. In his sense, i is pa icula ly ele an ha in he absence o 6AU in hp 1 D cells, a weak bu signi ican inc ease in ecombina ion was ob- se edin he L–PHO5cons uc (six old) when ansc ip- ion elonga ed h ough PHO5, bu no e ec was ob- se ed when PHO5 was no ansc ibed. This s eng h- ens ou conclusion ha Hp 1p pa icipa es in ansc ip ion elonga ion h ough yeas DNA sequences. Ou esul s p o ide e idence ha blockage o ansc ip- ion elonga ion leads o genomic ins abili y o DNA e- pea s and o e s an al e na i e model o explain an- sc ip ion-induced ecombina ion o hose p oposed p e- iously (Blackwell e al. 1986; Voelkel-Meiman e al. 1987; S ewa and Roede 1989; Thomas and Ro hs ein 1989; Nickolo 1992). Recombina ion in hp 1 D is di e - en om p e iously epo ed cases o ansc ip ion-in- duced ecombina ion, in which an inc ease in ecombi- na ion is associa ed wi h an inc ease in ansc ip le els. To explain how a ansc ip ional block can induce bo h a dele ion be ween epea s and he loss o a plasmid o ch omosome, we p opose ha a s alled ansc ip ional elonga ion complex may induce DNA b eaks o may cause an a es o he eplica ion o k, ha would be ei he epai ed o bypassed, espec i ely, by ecombina- ional epai (see Fig. 10). Conce ning he possibili y ha genomic ins abili y could a ise as a consequence o he a es o he eplica ion o k a e colliding wi h he blocked RNA Pol II (Fig. 10), i has been shown ecen ly ha DNA eplica ion o ks ansien ly a es a highly ansc ibed DNA egions in yeas (Deshpande e al. 1996), and i is known ha mu a ions in DNA Pol I and Pol III lead o an inc ease in ch omosome loss and e- combina ion (Ha well and Smi h 1985; Aguile a and Klein 1988). I he eplica ion o k collided wi h he blocked RNA Pol II, ei he he 38end o he nascen DNA could in ade he o he DNA- epea egion beyond he block, gene a ing a dele ion by one-ended in asion, o a cu could occu in he empla e leading o a double- s and b eak (Fig. 10e) ha would be epai ed by a dele- ion e en h ough ei he single-s and annealing (Lin e al. 1984) o one-ended in asion (P ado and Aguile a 1995). The la e possibili y would i wi h he ecen obse a ion ha eplica ion a es s cause double-s and b eaks in Esche ichia coli (Michel e al. 1997), and could explain he high equency o ecombina ion induced by con e ging eplica ion and ansc ip ion machine ies (Vile e e al. 1992). Finally, i DNA b eaks induced by ansc ip ional blocks we e no epai ed, hey would cause plasmid o ch omosome loss (Fig. 9; San os-Rosa and Aguile a 1994). Al hough we do no ha e e idence ha he ansc ip- ion machine y i sel could pa icipa e in he ec ui - men o ecombina ion p o eins, simila o wha is be- lie ed o occu in ansc ip ion-coupled nucleo ide exci- sion epai ( o e iew, see Bas hia e al. 1996; Hoeijmake s e al. 1996), his is a possibili y ha canno bedismissed.Thus,some ansc ip ion ac o sp esen in an elonga ing RNA Pol II migh se e o ec ui he e- combina ion machine y ha would epai he DNA b eaks occu ing as a consequence o a ansc ip ional block. In his sense, i is no ewo hy ha he iden i ica- ion o Rad51p in he human RNA Pol II holoenzyme (Maldonado e al. 1996) and he inding ha S b2p and H s1p, wo RNA Pol II gene al ansc ip ion ac o s, a e equi ed o hp 1 D -induced ecombina ion (Pi ua and Aguile a 1996; Pi ua e al. 1997). In summa y, ega dless o he mechanism leading o dele ions in associa ion o ansc ip ional elonga ion blocks, ou wo k shows ha ailu es in ansc ip ional elonga ion may be an impo an sou ce o genome ins a- bili y, as assessed by ecombina ion be ween epea s and plasmid ins abili y. Figu e 10. Al e na i e models o explain induc ion o genomic ins abili y by a ansc ip ional elonga ion block. An elonga ing RNAPol II (a)would be blockeda pa icula DNAsequencesin he absence o Hp 1p in a egion loca ed be ween di ec epea s DR and DR8(b). The RNA Pol II–DNA complex may acili a e DNA b eaks, whe he o no media ed by a nuclease (c)o may impede p og ession o he eplica ion o k (d). The collapsed eplica ion o k could e en ually acili a e he b eak o he em- pla e s and leading o a double s and b eak (e), al hough his s ep migh no be necessa y. F om ei he s ep (c–e), and p esum- ably a e exonuclease diges ion o one DNA s and, s and pai - ing be ween DR8and DR ( ) acili a ed by ei he single-s and annealing, one-ended in asion, o DNA polyme ase s and slip- page would cause a dele ion e en and he loss o he in e en- ing egion (los ) (g), explaining he hype - ecombina ion pheno- ype o hp 1 D .O he wise he DNA molecule, ei he a plasmid o a ch omosome, would be los , explaining he high le els o ch omosome and plasmid loss o hp 1 D cells. T ansc ip ional elonga ion and genome ins abili y GENES & DEVELOPMENT 3467 Cold Sp ing Ha bo Labo a o y P ess on Oc obe 1, 2015 - Published by genesde .cshlp.o gDownloaded om