scieee AI-readable full text Open interactive document viewer

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

Chávez de Diego, Sebastián; Aguilera López, Andrés

Abstract

The yeast HPR1 gene plays an important role in genome stability, as indicated by the observation that hpr1 mutants have high frequencies of DNA repeat recombination and chromosome loss. Here we report that HPR1 is required for transcriptional elongation. Transcription driven from constitutive and regulated yeast promoters cannot elongate through the bacterial lacZ coding region in hpr1Δ cells, but progresses efficiently through other sequences such as yeast PHO5. We show that HPR1 is not required for transcription activation and that the previously reported effects of hpr1Δon the activation of different promoters is a consequence of the incapacity of hpr1Δ cells to elongate transcription through lacZ, used as reporter. Transcriptional defects are also observed in yeast DNA sequences ofhpr1Δ cells in the presence of the transcription elongation inhibitor 6-azauracil. In all cases, the blockage of transcription elongation inhpr1Δ is associated with both the high frequency of deletions and the increase in plasmid instability that we report here. Therefore, in addition to the identification of a new element involved in transcriptional elongation, our work provides evidence for a new source of genomic instability.

Full text

The yeast HPR1 gene has a functional role in transcriptional elongation that uncovers a novel source of genome instability Sebastia´n Cha´vez and Andre´s Aguilera 1 Departamento de Gene´tica, Facultad de Biologı´a, Universidad de Sevilla, 41012 Sevilla, Spain The yeast HPR1 gene plays an important role in genome stability, as indicated by the observation that hpr1 mutants have high frequencies of DNA repeat recombination and chromosome loss. Here we report that HPR1 is required for transcriptional elongation. Transcription driven from constitutive and regulated yeast promoters cannot elongate through the bacterial lacZ coding region in hpr1 D cells, but progresses efficiently through other sequences such as yeast PHO5. We show that HPR1 is not required for transcription activation and that the previously reported effects of hpr1 D on the activation of different promoters is a consequence of the incapacity of hpr1 D cells to elongate transcription through lacZ, used as reporter. Transcriptional defects are also observed in yeast DNA sequences of hpr1 D cells in the presence of the transcription elongation inhibitor 6-azauracil. In all cases, the blockage of transcription elongation in hpr1 D is associated with both the high frequency of deletions and the increase in plasmid instability that we report here. Therefore, in addition to the identification of a new element involved in transcriptional elongation, our work provides evidence for a new source of genomic instability. [Key Words: HPR1; transcription elongation; transcription-induced recombination; genomic instability; DNA repeats; lacZ reporter] Received May 21, 1997; revised version accepted October 10, 1997. Genetic recombination is required for DNA repair in mitosis, for the generation of genetic diversity and for proper reductional division in meiosis. In addition, it is responsibleforchromosomalaberrationsassociatedwith some types of cancer and hereditary diseases, many of which occur by recombination between DNA repeats. The identification of the genes and functions that controlinitiation of recombinationbetweenrepeats istherefore essential to understand the origin of chromosome aberrations and genome instability. One important factor affecting the frequency of initiation of recombination is transcriptional activity. Recombination is stimulated by transcription as first shown by the identification of HOT1, a sequence from the rDNA region of Saccharomyces cerevisiae required for RNA polymerase I (Pol I)-dependent transcription (Keil and Roeder 1984; Voelkel-Meiman et al. 1987). Deletions between direct repeats have also been shown to be stimulated by RNA Pol II-mediated transcription in yeast (Thomas and Rothstein 1989). The relevance of transcription in homologous recombination has also been reported in mammalian cells (Nickoloff 1992; Thyagarajam et al. 1995). Probably the most significant connection between transcription and recombination is the one observed in immunoglobulin gene rearrangements (Blackwell et al. 1986; Lauster et al. 1993). Other examples of transcription-stimulated recombination have been provided in S. cerevisiae (Klar et al. 1981; Nevo- Caspi and Kupiec 1994), Schizosaccharomyces pombe (Grimm et al. 1991) and phages and bacteria (Dul and Drexler 1988; Vilette et al. 1992). Transcriptional activity may induce recombination because unwinding of the DNA duplex, changes in the local supercoiling or disruption of chromatin structure associated with transcription could provide a better accessibility of recombination proteins to the transcribed DNA, could lead to DNA structures susceptible to DNA breaks, whether or not mediated by nucleases, or could facilitate the pairing reaction (McCormack and Thompson 1990; Dro¨ge 1993; Kotani and Kmiec 1994). Other explanationscannot be excluded, however, such as a role of the transcription machinery in recruiting recombination proteins to transcribed genes that could be responsible for an increase in recombination, as is the case of transcription-coupled nucleotide excision repair (for review, see Basthia et al. 1996; Hoeijmakers et al. 1996). In this sense, the recent identification of the Rad51p re- 1 Corresponding author. E-MAIL [email protected]; FAX (34) 5-4557104. GENES & DEVELOPMENT 11:3459–3470 © 1997 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/97 $5.00; www.genesdev.org 3459 Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from combinational repair protein in the human RNA Pol II holoenzyme is noteworthy (Maldonado et al. 1996). Despite our poor understanding of the molecular basis of transcription-induced recombination, it seems clear that transcription is an important potential source of spontaneous recombination between repeats leading to genomic instability and needs to be investigated. One gene particularly relevant for the understanding of the putative connection between transcription and recombination as well as for the control of direct repeat recombination in yeast is HPR1. The importance of Hpr1p in genome stability is supported by the high frequencies of recombination and chromosome loss observed in null hpr1 D mutants (Aguilera and Klein 1990; Santos-Rosa and Aguilera 1994). Hpr1p has been suggested to be required for transcription activation of regulated promoters (Fan and Klein 1994; Zhu et al. 1995). Theidentificationofmutationsindifferenttranscription factors as suppressors of either the thermosensitivity (Fan et al. 1996; Uemura et al. 1996) or the hyper-recom- bination phenotype of hpr1 D (Piruat and Aguilera 1996; Santos-Rosa et al. 1996) is consistent with a functional role of Hpr1p in transcription. Our recent observations that high rates of deletions in hpr1 D cells only occurred between direct repeats in which defined regions of the intervening sequence were transcribed (Prado et al. 1997),suggest thatHpr1p mighthave afunctional rolein transcriptional elongation. To elucidate the function of Hpr1p on transcription, whether at the initiation or the elongation stage, and to understand the causes of induction of genomic instability by hpr1 D ,we undertook an in vivo molecular analysis that has allowed us to show that: (1) Hpr1p functions in transcriptional elongation, (2) previous results suggesting that Hpr1p is a general transcriptional activator (Fan and Klein 1994; Zhu et al. 1995) are explained by the incapacity of hpr1 D cells to elongatetranscriptionthroughthelacZsequenceusedas reporter of gene expression and not by an incapacity of hpr1 D cells to activate transcription at the initiation stage, and (3) the increased levels of direct repeat recombination and chromosome loss observed in hpr1 D cells is a direct consequence of the blockage of transcription elongation. Our work not only shows a functional role for Hpr1p in transcriptional elongation, but also suggests a new model for the initiation of genomic instability. Results HPR1 is required for lacZ transcription Expression of lacZ under the control of the GAL1 promoter strongly depends on HPR1 (Fan and Klein 1994; Zhu et al. 1995) (Fig. 1B). To assess whether or not lacZ could be expressed in hpr1 D cells, the lacZ coding region was placed under the control of two different constitutive promoters, TEF2 and ADH1, the latter of which had been shown to promote transcription in the absence of Hpr1p (Zhu et al. 1995). In contrast to the expected result, lacZ was not expressed from either of the two promoters (Fig. 1C,D). The lack of b-galactosidase expression driven from the GAL1promoter in hpr1 D ,therefore, may not be caused by a defect in transcriptional activation at the promoter. One possibility is that the lack of b-gal expression was caused by a defect in elongation through lacZ. To evaluate such a possibility, the yeast PHO5 gene was placed under the control of the GAL1 promoter. In this case, acid phosphatase was activated to wild-type levels in hpr1 D cells (Fig. 1E). Therefore, hpr1 D has no effect on activation of the GAL1 promoter. This fits with the observation that induction of the endog- Figure 1. Expression of lacZ fusion constructs placed in CEN plasmids in wild-type and hpr1 D cells. The hpr1 D mutation abolished transcription through the bacterial lacZ coding region, regardless of the yeast promoter from which it was transcribed, whereas it did not affect transcription through the yeast PHO5 coding region. (A) The lacZ coding region was fused to either the regulated GAL1 promoter or the constitutive TEF2 and ADH1 promoters. The PHO5 coding region was fused to the GAL1 promoter. Numbers below each construct refer to the translation start of each DNA sequence. Expression of lacZ and PHO5 was determined by b-galac- tosidase (B–D) or acid phosphatase (E) activities, respectively. For the GAL1-driven expression of either lacZ(B)orPHO5(C),either2% glucose(Glu, shaded bars, B,C,E) or 2% galactose (Gal, open bars, B,C,E) was added to 16-hr mid-log phase cultures in glycerol–lactate synthetic medium, and enzymatic activities were assayed 8 hr later. Cha´vez and Aguilera 3460 GENES & DEVELOPMENT Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from enous GAL1 gene was not affected by hpr1 D (Fan and Klein 1994). The previous work was performed on fusion constructs located in centromeric plasmids. Nevertheless, we have confirmed the same results in fusion constructs located in chromosomes. Thus, Figure 2 shows that whereas acid phosphatase mainly expressed by the endogenousPHO5gene reachedsimilarlevels ofactivation in both wild-type and hpr1 D cells, expression of lacZ under the same PHO5 promoter was seriously impaired in hpr1 D cells. Toconfirm that theabsenceof b-galactosidaseactivity of hpr1 D cells was caused by transcriptional, rather than post-transcriptional, defects of lacZ expression, we performed Northern analysis of lacZ under the GAL1 and PHO5 promoters. Figure 3 shows that lacZ mRNA driven from the GAL1 promoter was accumulated at high levels in wild type after induction, whereas it did not accumulate in hpr1 D (Fig. 3A,B). The Northern analysis of hpr1 D cells (Fig. 3A) shows a band corresponding to the full length lacZ mRNA, which appeared with both the 58- and 38-end probes of lacZ, and a smear which presumably contained incomplete 58lacZ mR- NAs, as it appeared with the 58-end but not with the 38-end lacZ probe. At 30 min of induction, the total lacZ mRNA in hpr1 D (mainly incomplete 58-end mRNA) was produced in similar amounts as in wild-type (mainly full-length mRNA). This was consistent with the idea that the GAL1 promoter initiated transcription equally well in both wild-type and hpr1 D ,but transcription did not proceed further down from the 58-end region of lacZ in hpr1 D .The decrease of total lacZ mRNA observed after 30 min could be explained by the low stability of such incomplete transcripts (for review, see Ross 1995) and an eventual reduction in the efficiency of reinitiation of transcription from the GAL1 promoter as a consequence of a block of transcription in lacZ. Instead, PHO5 mRNA was accumulated in both wild-type and hpr1 D ,and only full-length mRNA was observed (Fig. 3C,D). Transcription is blocked at the lacZ 58end in hpr1D cells Toconfirmthat transcription couldbepausedor blocked at the lacZ 58-end region we performed transcriptional run-on analysis of GAL1–lacZ in permeabilized cells (Fig. 4). RNA Pol II was paused or blocked in the first 170 bp of the lacZ 58end, as deduced from the observation that under induction conditions (Gal) the amount of radiolabeled mRNA hybridizing with the first 170 bp of lacZ in hpr1 D was ∼70% of the wild-type levels (Fig. 4A), but 17%–39% for the next six downstream fragments. As expected, the levels of radiolabeled lacZ mRNA bound to each DNA fragment under repression conditions (Glu) was significantly lower for all DNA fragments. The only exception was the 58-end 170-bp fragment (Fig. 4B), which bound to similar amounts of radiolabeled mRNA under repression and induction conditions in both wild-type (101%) and hpr1 D cells (74%). This implies that the polymerase is normally paused at the 58end of lacZ under repression conditions as recently reported (Akhtar et al. 1996), and that the initiation of transcription is equally efficient in both wild-type and hpr1 D cells. According to our results, Hpr1p would be required to allow the RNA Pol II to travel farther down from the lacZ 58-end region under activation conditions, but not for the establishment of the RNA Pol II at the 58end, as is also the case for Kin28p, Srb2p, or the carboxy-terminal domain (CTD) of RNA Pol II (Akhtar et al. 1996). Natural regulatory blocks of transcription described in several eukaryotic genes (for review, see Eick et al. 1994; Bentley1995)areneartheir58endsandrequiretranscrip- tional activators to be bypassed by RNA Pol II. In contrast, the lack of lacZ expression from ADH1 and TEF2 promoters(Fig.1) suggests thatthe block oftranscription at lacZ in hpr1 D cells is promoter-independent. To confirm this, we decided to place the lacZ coding sequence in the UTR of the GAL1–PHO5 construct right after the PHO5stopcodon (see MaterialsandMethods).Thus, the RNA Pol II transcription machinery has to transcribe the complete 1.4-kb PHO5 coding sequence before entering the lacZ sequence. As can be seen in Figure 5, full PHO5–lacZ message is produced in wild-type cells but not in hpr1 D cells. A smear corresponding to incomplete messages was observed in hpr1 D cells.Such smearis also observed in wild-type cells, as in the GAL1–lacZ construct (see Fig. 3) suggesting that, indeed, elongation through lacZ is inefficient in yeast. A new run-on analysis of this GAL1–PHO5–lacZ fusion construct showed that elongation was mostly blocked at the 58-end of the lacZ coding sequence in hpr1 D cells: The RNA Pol II engaged at the 58end of lacZ was 37%–53% of the RNA Pol II engaged at the PHO5 coding sequence (data not shown). This result is consistent with the run-on analysis of the GAL1–lacZ construct (see Fig. 4) and confirms that the incapacity of hpr1 D cells to make complete Figure 2. Expression of the chromosomally located lacZ (A) and PHO5 (B) coding sequences under the control of the PHO5 promoter in wild-type and hpr1 D cells. For repression (+Pi, shaded bars) and induction (−Pi, open bars) conditions see Materialsand Methods.Acid phosphataseand b-galactosidasewere assayed in the same culture for each sample. Other details as in Fig. 1. Transcriptional elongation and genome instability GENES & DEVELOPMENT 3461 Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from PHO5–lacZ mRNA (Fig. 5) and lacZ mRNA (Fig. 3) is caused by the incapacity to elongate transcription through lacZ, regardless of its position relative to the promoter from which it is transcribed. Therefore, HPR1 has a functional role in elongation, and not in initiation or activation of transcription. Transcriptional elongation blocks as a cause of genome instability To confirm that the blockage of transcription elongation wasresponsibleforthehyper-recphenotypeofhpr1 D ,we inserted the lacZ and PHO5 coding sequences between two 0.6-kb direct repeats, immediately downstream of a 38-end truncated copy of LEU2 and immediately upstream of a 58-truncated copy of LEU2. In these constructs, transcription of both lacZ and PHO5 was initiated at the unique LEU2 promoter, as determined by RNAse A protection (data not shown). Consequently, RNA Pol II must transcribe 760 bp of LEU2 before entering either the lacZ or the PHO5 coding sequences. According to our expectations, lacZ should cause a reduction in the level of full mRNA initiated at the LEU2 promoter and a strong hyper-recombination phenotype in hpr1 D cells, which are both consequences of blockage or stalling of transcription elongation at lacZ (L–lacZ constructs), but none of these phenotypes should be caused by PHO5 (L–PHO5 constructs). The results confirmed our predictions (Fig. 6). Transcription in L–lacZ constructs initiated at the LEU2 promoter upstream of the first repeat, traversed lacZ and terminated at the LEU2 terminator downstream of the second repeat in both wild type and hpr1 D .Transcript levels in hpr1 D were 12 times lower than in wild type if lacZ was transcribed in its natural direction, and 200 times if transcribed in the opposite direction, whereas recombination frequencies were increased 44- and 266- fold, respectively. In the L–PHO5 constructs, both the pattern and the level of transcripts were identical in wild type and hpr1 D ,and no significant difference was ob- Figure 3. Northern analysis of GAL1–lacZ and GAL1–PHO5. The Northern analysis and kinetics of induction of lacZ (A,B) and PHO5 (C,D) mRNAs driven from the GAL1 promoter is shown. Wild-type (s,B,D) and hpr1 D (h,B,D) transformants were obtained from overnight cultures in glycerol–lactate synthetic media lacking uracil and diluted in identical fresh media to an OD 600 of 0.5. Galactose (Gal) was then added and samples were taken for Northern analysis after different times, as specified. For repression conditions (Glu) total RNA was isolated from mid-log phase cultures in 2% glucose synthetic media lacking uracil. The DNA probes usedwere(lacZ58end) the 0.5-kbBamHI–HpaIfragmentofpLGZ containing the58endoflacZ;(lacZ38end) the0.4-kbPvuIIfragment of pLGZ containing the 58- end of lacZ; (PHO5) the 1.5-kb EcoRI–PstI internal PHO5 fragment of pJDB207–PHO5 (Eco); (ACT1) the 0.6-kb ClaI internal ACT1 fragment of plasmid pYA301. (AU) arbitrary units. Cha´vez and Aguilera 3462 GENES & DEVELOPMENT Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from served in recombination. In one orientation, transcriptioninitiatedattheLEU2promoter,traversedPHO5 and terminated at the PHO5 terminator with similar efficiency in both wild-type and hpr1 D cells (Fig. 6). A weak effect was observed on recombination (sixfold increase). In the opposite orientation, transcription terminated immediately downstream of the first LEU2 copy once it encountered the PHO5 transcription terminator, with similar efficiency in both wild-type and hpr1 D .No effect was observed on recombination. We have also observed that the blockage of transcription at lacZ is associated with an increase in plasmid loss. The replicative pRS416 plasmids containing either GAL1–lacZ (p416GAL1–lacZ)orGAL1–PHO5 (pSCh202) (see Figs. 1 and 2) had similar stability in hpr1 D as in wild-type cells under repression conditions of transcription (55%–62% of the cells contained the plasmids after 23–25 generations in nonselective glucose-based media). Under induction conditions, however, p416GAL1–lacZ was clearly unstable in hpr1 D cells. The proportion of hpr1 D cells that contained p416GAL1–lacZ after 23–25 generations in nonselective galactose-based media was 37 times lower than that of wild-type cells, whereas only a threefold reduction was found for pSCh202 (data not shown). Genome instability and transcription defects of hpr1D cells are observed in yeast DNA sequences in the presence of 6-azauracil Because transcriptional elongation block, hyper-recom- bination, and plasmid loss were only associated with the bacteriallacZsequenceinthisstudy,itwasimportantto investigate the role of Hpr1p in transcription and genomeinstability of yeastDNAsequences. DespiteHPR1 Figure 4. Transcriptional run-on analysis in wild-type and hpr1 D cells. Total RNA wasisolatedfromwild-typeandhpr1 D cells transformed with single-copy p416GAL1– lacZ plasmid under induction (A) and repression (B) conditions. Two percent-galac- tose or 2% glucose was added to yeast cultures in glycerol–lactate synthetic medium at an OD 600 of 0.05, 5 hr prior to the run-on analysis. The 0.6-kb internal ACT1 fragment and seven different DNA fragments (1–7) from the lacZ coding region were immobilized in hybond-N+ filters. The lacZ region covering each of the seven DNA fragments used is shown at the bottom. In all cases, the percentage of radiolabeled mRNA bound to each lacZ fragment was normalized with respect to their corresponding levels in galactose-grown wildtype cells, taken as 100% for each. The orientation of lacZ arrows indicates the direction of transcription. As negative control, we used DNA from Salmonella typhimurium (not shown). Figure 5. (A) Northern analysis of PHO5 mRNA. The DNA probe used for Northern analysis was the 1.5-kb EcoRI–PstI internal PHO5 fragment of pJDB207–PHO5 (Eco). Arbitrary units of mRNA were calculated according to the same standards for all experiments. (B) The mRNA values are given with respect to rRNA levels (see Materials and Methods). (s) Wild type; (h) hpr1 D .Other details as in Fig. 3. Transcriptional elongation and genome instability GENES & DEVELOPMENT 3463 Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from not being essential for viability, we considered the possibility that Hpr1p had a general role on transcription elongation of yeast genes, as it is the case of PPR2 encoding elongation factor TFIIS (for review, see Kane 1994). Because it was shown previously that 6-azauracil (6AU) produces a depletion of UTP and GTP responsible for a reduction of transcriptional elongation efficiency, and that the growth of mutants in PPR2 was sensitive to 6AU (Archambault et al. 1992; Exinger and Lacroute 1992), we decided to determine whether the transcription and genetic instability phenotypes conferred by hpr1 D were also observed in yeast genes in the presence of 6AU. We tried three different concentrations of 6AU (100, 300, and 1000 µg/ml). Both wild-type and hpr1 D cells, in contrast to transcriptional elongation TFIIS − mutants, were able to grow on SC media containing the three different concentrations, although hpr1 D growth was more severely affected than wild-type growth (data not shown). When we analyzed transcription of the yeast PHO5 sequence under the GAL1 promoter, however, the kinetics of accumulation of PHO5 mRNA was severely affected in hpr1 D in the presence of 6AU (Fig. 7). Thus, the level of activation of PHO5 mRNA at 908of induction in wild-type cells was sevenfold higher than in hpr1 D cells. PHO5 transcript levels increased up to sixfold at 908of induction in wild-type cells,no increasewas observed in hpr1 D cellsin thepresence of 6AU. The ACT1 transcript levels were also twofold lower in hpr1 D versus wild type in the presence of 6AU, indicating that 6AU specifically impairs transcription in hpr1 D cells (data not shown). To confirm that the observed effect of 6AU on transcription in hpr1 D cells was caused by impairment of transcription elongation and not by side-effects of 6AU, we determined whether guanine (100 µg/ml) could partially revert the effect of 6AU, as shown previously for mutants of TFIIS and not for 6AU-sensitive mutants unaffected in transcriptional elongation (Archambault et al. 1992). Figure 7 shows that accumulation of PHO5 mRNA was reestablished by guanine in the presence of 6AU. In addition, in the L–PHO5 construct containing the PHO5 ORF between the leu2 repeats in the orientation that is transcribed, recombination was enhanced 14-fold over the wild-type levels in the presence of 6AU, clearly above the increase observed without 6AU (5.5-fold) (Fig. 8). In the L–PHO5 construct with PHO5 in the orienta- Figure 6. Transcription and recombination analysis of direct repeat systems carrying lacZ (3 kb) or PHO5 (1.5 kb) codingregions. Aschemeof thedeletionproduct formed by recombination between the direct repeats used is shown (A). The diagram of each direct repeat system indicates the 0.6-kb repeated sequences (shaded boxes), the orientation of the lacZ and PHO5 coding regions, the LEU2 promoter (Prm) and transcription terminators (Ter), and the transcripts driven from the LEU2 promoter (arrow), whose 38ends have been made to coincide with the position of the corresponding band in each gel (B). Total RNA was isolated from overnight cultures in synthetic media lacking tryptophan. The DNA probes used in the hybridization experiments were the 598- bp ClaI–EcoRV LEU2 repeat, and the 581- bp ClaI internal ACT1 fragment. The transcript corresponding to the LEU2 endogenous chromosomal band is indicated. No transcript initiates in the internal lacZ and PHO5 internal sequences as determined with specific lacZ and PHO5 DNA probes (data not shown). Cha´vez and Aguilera 3464 GENES & DEVELOPMENT Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from tion that is not transcribed, the presence of 6-AU had no effect on recombination (Fig. 8). This result clearly indicates that recombination is increased only when transcription is elongated through PHO5. Indeed, this increase in recombination correlated with a slight decrease of transcript levels in hpr1 D cells in the presence of 6AU (data not shown). A similar effect was observed with the LA construct, containing the ADE2 gene between the leu2 repeats (data not shown), suggesting that the effect of6AUon hpr1 D cellsmaybe associated withelongation through yeast DNA sequences. No increase of recombination in the L–lacZ construct was observed by the the addition of 6AU (Fig. 8), suggesting that transcriptioninduced recombination in this construct reaches its maximum levels without 6AU. Finally, the use of centromeric plasmid pRS314-LA permitted us to determine directly by visual inspection the effect of hpr1 D on genomic instability in the presence of 6AU. Figure 9 shows that the ADE2 gene was extremely unstable in hpr1 D cells treated with 6AU, as shownby its red-sectoringphenotypecaused bytheAde − segregants, in contrast to untreated hpr1 D cells or wildtype cells, regardless of being treated with 6AU. The recombination frequency leading to Ade − recombinants in pRS314-LA is too low (<5 × 10 −3 ) to be detected by this sectoring assay. Therefore, the observed red-sectoring phenotype is mostly caused by plasmid loss, as confirmed by genetic analysis. As expected, the phenotype of increased plasmid loss observed in hpr1 D cells in the presence of 6AU was reverted by guanine (Fig. 9), consistent with being caused by transcription-elongation blockage. These results indicate that when transcription elongation is impaired, yeast DNA sequences, and not only bacterial sequencies such as lacZ, become genetically unstable in hpr1 D cells. Discussion The main conclusion of this work is that HPR1, originally identified by the hyper-recombination phenotype conferred by hpr1 mutations, participates in transcriptional elongation. We show that the block of transcriptional elongation produced in hpr1 D cells is responsible Figure 9. Plasmid instability in the presence of 6AU. Yeast colonies of URA3 + wild-type and hpr1 D strains transformed with centromeric plasmid pRS314-LA growing on SC − Trp with or without 100 µg/ml of 6AU and with or without 100 µg/ml of guanine. Figure 7. Transcription analysis of a yeast ORF in the presence of 6AU. Northern analysis (A) and kinetics of induction (B)of PHO5 mRNAs driven from the GAL1 promoter in the presence of 6AU with and without guanine are shown. The URA3 + wildtype and hpr1 D strains transformed with pSCh202 were obtained from overnight cultures in glycerol–lactate syntheticcomplete media lacking tryptophan and uracil, and diluted in identical fresh media to an OD 600 of 0.5 with 100 µg/ml of 6AU with and without 100 µg/ml of guanine. Galactose was added after 2 hr, and samples were taken for Northern analysis after different times, as specified. The mRNA values are given with respect to rRNA levels (B) (see Materials and Methods). Other details as in Fig. 3. Figure 8. Recombination in the presence of 6AU. Recombinationanalysisof the directrepeatconstructsL–lacZand L–PHO5 carrying the PHO5 ORF in both possible orientations between the leu2 repeats (see Fig. 6). Recombination frequencies were determined in the URA3 + wild-type (open bar) and hpr1 D (shaded bar) strains transformed with the appropriate plasmids, grown in media with or without 100 µg/ml of 6AU. Transcriptional elongation and genome instability GENES & DEVELOPMENT 3465 Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from for the high frequency of deletions between repeats and the high frequency of plasmid-loss. Our work not only shows a role for Hpr1p in transcriptional elongation but alsoindicates that transcriptional elongation blocksmay be an important source of genome instability and provides a molecular mechanism to explain transcriptioninduced recombination. Hpr1p functions in transcriptional elongation Our experiments indicate that Hpr1p stimulates transcriptional elongation through lacZ in vivo. Because previous reports suggesting that Hpr1p is required for transcription activation are based on the analysis of expression of a lacZ reporter fused to different yeast regulated promoters (Fan and Klein 1994; Zhu et al. 1995), we should reinterpret those results as being caused by the incapacity of hpr1 D cells to elongate transcription through lacZ. This is independent of whether the promoters used are constitutive or regulated (Figs. 1 and 2) and whether the lacZ sequence is proximal or distal to the promoter from which it is transcribed (Figs. 3, 5, and 6). Because no effect on transcription was observed when the yeast PHO5 ORF was used as reporter (Figs. 1 and 2), we conclude that Hpr1p is not involved in transcriptional activation. Our results solve the paradox of why activation of endogenous yeast genes is not affected by hpr1 D (Fanand Klein 1994; Zhu etal. 1995). Whenminor effects are detected, as is the case of SUC2 (Zhu et al. 1995),theyarelikely to becausedbya reduced efficiency in elongation. Certainly, our study raises serious concerns about using lacZ to study transcription in yeast and invalidates previous conclusions suggesting a role forHpr1pintranscriptionalactivationofpromoters(Zhu et al. 1995). The strong effect of hpr1 D on transcriptional elongation through lacZ is not observed in yeast genes such as PHO5 at 30°C (Fig. 3), even though a weak effect is detected at 37°C (S. Cha´vez and A. Aguilera, unpubl.). The effect of hpr1 D on elongation of yeast genes may be masked by other redundant functions. Indeed, the observation that in the presence of the transcriptional elongation inhibitor 6-azauracil, PHO5 mRNA accumulation is strongly impaired in hpr1 D mutants, a phenotype partiallyreverted by guanine,suggeststhat indeedHpr1p has a general role in transcriptional elongation of yeast genes.Because hpr1 D areviable in6-azauracil, therole of Hpr1p in transcriptional elongation must be different from that of TFIIS, a factor required for the relief of an arrested RNA Pol II (for review, see Reines 1994). The function of Hpr1p cannot be related to TFIIS, because TFIIS is inhibited by Sarkosyl (Reines 1992), used in our run-on experiments. If that were the case, no difference between wild type and hpr1 D should have been observed in the pattern of the pausing of RNA Pol II at lacZ (Fig. 4).Instead, Hpr1p might be required for preventing arrest of RNA pol II, as it has been proposed for TFIIF, elongin, or ELL (eleven-nineteen lysine-rich leukemia) (Price et al. 1989; Aso et al. 1995). We do not know the nature of the signal on which RNA Pol II is blocked in the absence of Hpr1p. DNA bends or a particular chromatin structure of the bacterial lacZ sequence (for review, see Kane 1994; Bentley 1995) may determine the need for Hpr1p for its transcription, in particular at the upstream 170 nucleotides where the major hpr1-dependent transcriptional block maps (Fig. 4). In this sense, it is noteworthy that sin1(spt2) mutations, affected in a HMG1-like gene, suppress the transcriptional phenotype of hpr1 D (Zhu et al. 1995) and that SIN1-2 and the imbalance of histones H2A/H2B or H3/ H4 produced synthetic lethality in hpr1 D (Fan and Klein 1994; Zhu et al. 1995). In addition, promoter-proximal pausing during transcriptional elongation in the human hsp70 gene depends on nucleosome templates in vitro (Brown et al. 1996). Other types of signals cannot be dismissed, however, such as those depending on RNA stems (Reeder and Hawley 1996). In any case, the transcriptional block caused by hpr1 D at lacZ is different from RNA Pol II transcriptional pausing in other eukaryotic genes, because the latter is controlled by transactivators as an integral part of the initiation step (Yankulov et al. 1994), whereas the block at lacZ is independent of (1) its position relative to the promoter, whether proximal or distal (Figs. 5 and 6), and (2) the type of promoter to which it is fused, whether regulated or constitutive (Figs. 1 and 2). In this sense, it is particularly relevant the observation that the lacZ coding sequence abolishes production of full-length transcript when fused to the 38 untranslated region (38UTR) of PHO5 (Fig. 5). Finally, our results open the possibility that the synthetic lethality of hpr1 top1 mutants (Aguilera and Klein 1990) could be related to the role of Hpr1p on transcription elongation, because topoisomerase I is also needed in RNA Pol II transcription (Schultz et al. 1992). In this sense, it is important to note that mutations in different TOP genes also confer hyper-recombination of different DNA repeats in yeast (Christman et al. 1988; Wallis et al. 1989). An in vitro-reconstituted system would be required to assess whether the transcriptional blocks caused by hpr1 D is caused by nucleosomes, DNA sequence, or an RNA secondary structure. In any case, our results show clearly that in vivo transcriptional elongation is blocked at lacZ in a promoter-independent mannerandimpairedatDNAsequencessuchasPHO5in the presence of 6AU. We find no evidence for a role of Hpr1p on transcription activation of promoters. Transcriptional elongation blocks lead to genome instability Our work shows that hyper-recombination and plasmid instability in hpr1 D cells is produced in association with a transcriptional elongation block. This explains our previous observation that hyper-recombination was dependent on transcription progression through particular DNA regions that include the bacterial tet and amp sequences(Pradoetal.1997).Astrongincreaseindeletions between repeats and plasmid loss is observed in hpr1 D cells only when transcription is elongated through the lacZ sequence (Fig. 6) or progresses through yeast DNA Cha´vez and Aguilera 3466 GENES & DEVELOPMENT Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from sequences (Fig. 7) in the presence of the transcriptional elongation inhibitor 6AU. In this sense, it is particularly relevant that in the absence of 6AU in hpr1 D cells, a weak but significant increase in recombination was observedin the L–PHO5construct(sixfold) whentranscription elongated through PHO5, but no effect was observed when PHO5 was not transcribed. This strengthens our conclusion that Hpr1p participates in transcription elongation through yeast DNA sequences. Our results provide evidence that blockage of transcription elongation leads to genomic instability of DNA repeats and offers an alternative model to explain transcription-induced recombination to those proposed previously (Blackwell et al. 1986; Voelkel-Meiman et al. 1987; Stewart and Roeder 1989; Thomas and Rothstein 1989; Nickoloff 1992). Recombination in hpr1 D is different from previously reported cases of transcription-in- duced recombination, in which an increase in recombination is associated with an increase in transcript levels. To explain how a transcriptional block can induce both a deletion between repeats and the loss of a plasmid orchromosome, we proposethata stalledtranscriptional elongation complex may induce DNA breaks or may cause an arrest of the replication fork, that would be either repaired or bypassed, respectively, by recombinational repair (see Fig. 10). Concerning the possibility that genomic instability could arise as a consequence of the arrest of the replication fork after colliding with the blocked RNA Pol II (Fig. 10), it has been shown recently that DNA replication forks transiently arrest at highly transcribed DNA regions in yeast (Deshpande et al. 1996), and it is known that mutations in DNA Pol I and Pol III lead to an increase in chromosome loss and recombination (Hartwell and Smith 1985; Aguilera and Klein 1988). If the replication fork collided with the blocked RNA Pol II, either the 38end of the nascent DNA could invade the other DNA-repeat region beyond the block, generating a deletion by one-ended invasion, or a cut could occur in the template leading to a doublestrand break (Fig. 10e) that would be repaired by a deletion event through either single-strand annealing (Lin et al. 1984) or one-ended invasion (Prado and Aguilera 1995). The latter possibility would fit with the recent observation that replication arrests cause double-strand breaks in Escherichia coli (Michel et al. 1997), and could explain the high frequency of recombination induced by converging replication and transcription machineries (Vilette et al. 1992). Finally, if DNA breaks induced by transcriptional blocks were not repaired, they would cause plasmid or chromosome loss (Fig. 9; Santos-Rosa and Aguilera 1994). Although we do not have evidence that the transcription machinery itself could participate in the recruitment of recombination proteins, similar to what is believed to occur in transcription-coupled nucleotide excision repair (for review, see Basthia et al. 1996; Hoeijmakers et al. 1996), this is a possibility that cannot bedismissed.Thus,sometranscriptionfactorspresentin an elongating RNA Pol II might serve to recruit the recombination machinery that would repair the DNA breaks occurring as a consequence of a transcriptional block. In this sense, it is noteworthy that the identification of Rad51p in the human RNA Pol II holoenzyme (Maldonado et al. 1996) and the finding that Srb2p and Hrs1p, two RNA Pol II general transcription factors, are required for hpr1 D -induced recombination (Piruat and Aguilera 1996; Piruat et al. 1997). In summary, regardless of the mechanism leading to deletions in association to transcriptional elongation blocks, our work shows that failures in transcriptional elongation may be an important source of genome instability, as assessed by recombination between repeats and plasmid instability. Figure 10. Alternative models to explain induction of genomic instability by a transcriptional elongation block. An elongating RNAPol II (a)would be blockedatparticular DNAsequencesin the absence of Hpr1p in a region located between direct repeats DR and DR8(b). The RNA Pol II–DNA complex may facilitate DNA breaks, whether or not mediated by a nuclease (c)ormay impede progression of the replication fork (d). The collapsed replication fork could eventually facilitate the break of the template strand leading to a double strand break (e), although this step might not be necessary. From either step (c–e), and presumably after exonuclease digestion of one DNA strand, strand pairing between DR8and DR (f) facilitated by either single-strand annealing, one-ended invasion, or DNA polymerase strand slippage would cause a deletion event and the loss of the intervening region (lost) (g), explaining the hyper-recombination phenotype of hpr1 D .Otherwise the DNA molecule, either a plasmid or a chromosome, would be lost, explaining the high levels of chromosome and plasmid loss of hpr1 D cells. Transcriptional elongation and genome instability GENES & DEVELOPMENT 3467 Cold Spring Harbor Laboratory Press on October 1, 2015 - Published by genesdev.cshlp.orgDownloaded from