Portal de Periódicos da CAPES

A novel yeast gene, THO2, is involved in RNA pol II transcription and provides new evidence for transcriptional elongation-associated recombination

1998; Springer Nature; Volume: 17; Issue: 16 Linguagem: Inglês

10.1093/emboj/17.16.4859

1460-2075

José I. Piruat,

RNA Research and Splicing

Article17 August 1998free access A novel yeast gene, THO2, is involved in RNA pol II transcription and provides new evidence for transcriptional elongation-associated recombination José I. Piruat José I. Piruat Search for more papers by this author Andrés Aguilera Corresponding Author Andrés Aguilera Departamento de Genética, Facultad de Biología, Universidad de Sevilla, E-41012 Sevilla, Spain Search for more papers by this author José I. Piruat José I. Piruat Search for more papers by this author Andrés Aguilera Corresponding Author Andrés Aguilera Departamento de Genética, Facultad de Biología, Universidad de Sevilla, E-41012 Sevilla, Spain Search for more papers by this author Author Information José I. Piruat and Andrés Aguilera 1 1Departamento de Genética, Facultad de Biología, Universidad de Sevilla, E-41012 Sevilla, Spain *Corresponding author. E-mail: [email protected] The EMBO Journal (1998)17:4859-4872https://doi.org/10.1093/emboj/17.16.4859 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info We have identified two novel yeast genes, THO1 and THO2, that partially suppress the transcription defects of hpr1Δ mutants by overexpression. We show by in vivo transcriptional and recombinational analysis of tho2Δ cells that THO2 plays a role in RNA polymerase II (RNA pol II)-dependent transcription and is required for the stability of DNA repeats, as previously shown for HPR1. The tho2Δ mutation reduces the transcriptional efficiency of yeast DNA sequences down to 25% of the wild-type levels and abolishes transcription of the lacZ sequence. In addition, tho2Δ causes a strong increase in the frequency of recombination between direct repeats (>2000-fold above wild-type levels). Some DNA repeats cannot even be maintained in the cell. This hyper-recombination phenotype is dependent on transcription and is not observed in DNA repeats that are not transcribed. The higher the impairment of transcription caused by tho2Δ, the higher the frequency of recombination of a particular DNA region. The tho2Δ mutation also increases the frequency of plasmid loss. Our work not only identifies a novel yeast gene, THO2, with similar function to HPR1, but also provides new evidence for transcriptional blocks as a source of recombination. We propose that there is a set of proteins including Hpr1p and Tho2p, in the absence of which RNA pol II transcription is stalled or blocked, causing genetic instability. Introduction Transcription, in addition to its essential and unique role in gene function, may be intimately related with other DNA transactions. A paradigm of this relationship is the eukaryotic transcription factor TFIIH, which contains proteins of the nucleotide excision repair (NER) machinery and has a functional role in both transcription and excision repair (Bhatia et al., 1996; Friedberg, 1996; Hoeijmakers et al., 1996). Although the putative dual function of this factor in transcription and NER is not obvious, it clearly seems to play an important role in transcription-coupled repair. Transcriptional activity has also been shown to be related to mutagenesis. Thus, high rates of mutation have been found in actively transcribed genes in yeast (Datta and Jinks-Robertson, 1995). In addition, the hypermutation mechanism of immunoglobulin genes, essential in the origin of antibody diversity, seems to be associated with transcription (Goyenechea et al., 1997). Although the molecular basis of this association is not yet understood, it clearly indicates an important role of transcription in mutation and repair. A very intriguing relationship has also been observed between transcription and recombination. Recombination has been shown to be stimulated by transcriptional activity both in prokaryotes (Dul and Drexler, 1988; Vilette et al., 1992), and eukaryotes from yeast to mammals. In yeast it has been shown that recombination leading to deletions between direct repeats is stimulated by activation of transcription occurring via the RNA pol I enhancer HOT1 (Voelkel-Meiman et al., 1987; Stewart and Roeder, 1989), RNA pol II-dependent promoters (Thomas and Rothstein, 1989; Grimm et al., 1991; Bratty et al., 1996) or Ty-expression (Nevo-Caspi and Kupiec, 1994). A connection between transcription and recombination has also been provided in mammalian cells for homologous genes (Nickoloff, 1992; Thygarajam et al., 1995) and for immunoglobulin gene rearrangements (Blackwell et al., 1986; Lauster et al., 1993). It is certainly likely that unwinding of the DNA duplex, changes in local supercoiling or remodelling of the chromatin structure associated with transcription might facilitate recombination by improving (i) the accessibility of the recombination-repair proteins, (ii) the formation of transient DNA–protein structures with transcription-associated activities that could initiate recombination, or (iii) the strand exchange reaction (Thomas and Rothstein, 1989; McCormack and Thompson, 1990; Dröge, 1993; Kotani and Kmiec, 1994). Consistent with these hypotheses, it has been shown that mutations in the structural genes for DNA topoisomerases TOP1, TOP2 (Christman et al., 1988) and TOP3 (Wallis et al., 1989), and the genes involved in chromatin structure SIR2 (Gottlieb and Esposito, 1989) or SPT4 and SPT6 (Malagón and Aguilera, 1996) confer hyper-recombination between different types of DNA repeats in yeast. Site-specific recombination is stimulated in vitro by negative supercoiling transiently built by an advancing RNA polymerase (Dröge, 1993) and transcription stimulates in vitro RecA-promoted strand exchange of nucleosomal templates (Kotani and Kmiec, 1994). In addition, it has been shown that meiotic recombination initiates preferentially at promoter regions (Baudat and Nicolas, 1997) or is increased at the DNA binding sites of transcription factors in eukaryotes from yeast (White et al., 1991) to mammals (Shenkar et al., 1991). Altogether these results suggest that there may be different transcription-associated events triggering recombination. We have recently shown that impairment of RNA pol II-dependent transcriptional elongation may induce mitotic recombination between direct repeats in the yeast Saccharomyces cerevisiae. Hpr1p is a protein required for proper transcriptional elongation by RNA pol II. In the absence of Hpr1p, the elongating RNA pol II is presumably stalled or blocked at particular DNA regions (Chávez and Aguilera, 1997), triggering a deletion event by recombination between the flanking repeats (Chávez and Aguilera, 1997; Prado et al., 1997). The intimate relationship between recombination and transcription in hpr1Δ cells is strengthened by the observation that two components of the mediator of the RNA pol II holoenzyme, Srb2p (Koleske et al., 1992) and Hrs1p/Pgd1p (Piruat et al., 1997; Myers et al., 1998) are completely required for hpr1-induced deletions (Piruat and Aguilera, 1996; Santos-Rosa et al., 1996). To gain further information on the mechanism of induction of recombination by transcription-elongation impairment, we have identified two new yeast genes, THO1 and THO2, that suppress the hpr1Δ transcriptional defects by overexpression. In vivo genetic and molecular analyses of the effect of these genes and their corresponding null mutations reveals that Tho2p has similar effects on transcriptional and genetic stability to Hpr1p. Our work not only provides new evidence that transcriptional elongation may be associated with recombination, but suggests that there is a set of proteins required for RNA pol II-transcription, including Hpr1p and Tho2p, the absence of which may cause recombinogenic stalls. In addition, this work confirms our recombination-based approach as a powerful way to identify new genes participating in transcription. Results Isolation of THO1 and THO2 as multicopy suppressors of the ts and transcriptional phenotypes of hpr1Δ To understand the function of HPR1 in transcription and the mechanism by which transcriptional elongation is associated with genetic instability, we searched for genes that suppressed hpr1Δ by overexpression. Such genes might have functions partially related to HPR1. To isolate genes that suppressed the incapacity of hpr1Δ cells to express lacZ, we first selected clones that in multicopy suppressed the thermosensitivity (ts) phenotype of hpr1Δ cells, expecting that they would also suppress their incapacity to express lacZ. We first selected 19 yeast transformants with the MW90 multicopy library that were able to grow at 37°C and, subsequently, confirmed their capacity to express the GAL1–lacZ construct of plasmid p416GAL1lacZ. Deletion analysis and partial DNA sequencing of the 19 clones isolated (see Materials and methods) permitted us to define two previously uncharacterized open reading frames (ORFs), THO1 (16 clones) and THO2 (3 clones) (suppressors of the transcriptional defects of hpr1Δ by overexpression), as multicopy suppressors of hpr1Δ. THO1 corresponds to the previously defined YER063w ORF of chromosome V (DDBJ/EMBL/GenBank accession No. U18813) and THO2 to the YNL139c ORF of chromosome XIV (DDBJ/EMBL/GenBank accession No. Z71416; Mallet et al., 1995), also included in the DDBJ/EMBL/GenBank as RLR1 (Required for LacZ RNA) by R.W.West (Ithaca, NY). THO1 encodes a basic protein of 218 amino acids (theoretical mol. wt 24.1 kDa) with no relevant homology to any other yeast protein. No homologous gene has yet been detected in any other organism. THO2 encodes a protein of 1597 amino acids (theoretical mol. wt 184 kDa), with no relevant homology to any other yeast protein. An homologous ORF of THO2 exists in Schizosaccharomyces pombe (region SPAC22F3.14c). There is no relevant domain to be mentioned from either Tho1p or Tho2p that could give us a clue to their function. As shown in Figure 1A, hpr1Δ cells show 13 and 21 times higher levels of lacZ expression when containing THO1 or THO2 in multicopy, respectively. This is 7.4 and 11.8% of the wild-type levels, respectively, indicating that overexpression of THO1 and THO2 partially suppresses the incapacity of hpr1Δ cells to transcribe through lacZ. However, this partial suppression of the transcriptional phenotype is not accompanied by suppression of the strong hyper-recombination phenotype of hpr1Δ. Recombination of hpr1Δ cells is not significantly changed by the presence of THO1 or THO2 in multicopy (Figure 1B), although it is noteworthy that the latter has levels of recombination 4.5 times below the hpr1Δ levels. These results are indeed consistent with the idea that transcription is defective in hpr1Δ cells transformed with THO1 or THO2 in multicopy. Figure 1.Effect of the overexpression of THO1 and THO2 on lacZ gene expression and DNA repeat recombination. (A) β-galactosidase activities of the hpr1Δ strain AYW3-3D transformed with the multicopy plasmid pLGS5 carrying the GAL1::lacZ construct and either YEp351 (control, YEp[–]), YCpA13 (YCp[HPR1]), pSUP4 (YEp[THO1]) or pSUP38 (YEp[THO2]). Either 2% glucose or 2% galactose was added to 16 h mid-log phase cultures in glycerol-lactate synthetic medium, and enzymatic activities were assayed 8 h later. Only the data of induced expression (2% galactose) are given. Under repression conditions (2% glucose) β-galactosidase values were ∼1 U in all cases (data not shown). Numbers represent the average value of two different transformants in 2% galactose. Standard deviations are indicated as vertical bars. (B) Recombination frequency of the chromosomal leu2-k::ADE2-URA3::leu2-k construct of the same transformants as before with the only exception of the control, that is the untransformed AYW3-3D strain. Recombinants were scored on SC+FOA. Numbers represent the median frequency value obtained from six independent cultures each. Download figure Download PowerPoint The tho1Δ mutation has no effect on transcription and recombination; tho2Δ increases recombination and impedes GAL1–lacZ expression To gain more insight into the biological processes in which THO1 and THO2 could participate in vivo we constructed tho1Δ::TRP1 and tho2Δ::LEU2 deletions by gene replacement (see Materials and methods). Both deletion mutants were viable, implying that neither THO1 nor THO2 is essential. Whereas tho1Δ mutants grow with the same doubling time as the wild type on YEPD at 30°C (85 min), tho2Δ form small colonies and grow with twice the doubling time of the wild-type (180 min). In addition, tho2Δ cells are also thermosensitive for growth (and do not form colonies) at 37°C. The most relevant phenotypes of hpr1Δ are impairment of transcription elongation through lacZ (Chávez and Aguilera, 1997) and increased recombination between repeats (Aguilera and Klein, 1990; Prado et al., 1997). Since the way THO1 and THO2 were isolated suggests that they might have functions partially related to HPR1, we decided to analyze expression of GAL1–lacZ and recombination in tho1Δ and tho2Δ mutants. As can be seen in Figure 2A, lacZ expression occurs at wild-type levels in tho1Δ cells, whereas it was abolished in tho2Δ. In addition, tho1Δ has wild-type levels of recombination in the chromosomal leu2-k::ADE2-URA3::leu2-k repeat construct, whereas tho2Δ shows a strong increase in recombination (2620 times above the wild-type levels), that is 7-fold above even the hpr1Δ levels (Figure 2B). Such a hyper-recombination phenotype is clearly observed as a strong red-sectoring phenotype of the tho2Δ cells containing the leu2-k::ADE2-URA3::leu2-k construct (Figure 2C). Figure 2.Effect of the tho1Δ and tho2Δ mutations on lacZ gene expression and DNA repeat recombination. (A) β-galactosidase activities of the GAL1–lacZ fusion construct in the strains AW33-1B (WT), AW33-1C (hpr1Δ), AW33-2C (tho1Δ), WR-4B (tho2Δ) and AW33-8A (hpr1Δ tho1Δ) transformed with multicopy plasmid pLGSΔ5 under induced conditions of expression (2% galactose). No determination was made for hpr1Δ tho2Δ strains (n.d., not determined). Under repression conditions (2% glucose) β-galactosidase values were ∼1 U in all cases (data not shown). For other details see legend to Figure 1A. (B) Recombination frequency of the chromosomal leu2-k::ADE2-URA3::leu2-k construct in the strains AW33-12A (WT), AW33-9D (hpr1Δ), AW33-1D (tho1Δ), WRA-4D (tho2Δ), AW33-2A (hpr1Δ tho1Δ) and WRA-31B (hpr1Δ tho2Δ). For other details see legend to Figure 1B. (C) Wild-type AW33-12A and tho2Δ WRA-7B strains growing on SC medium with 16 mg/l adenine and 75 mg/l FOA. A strong red-sectored phenotype is observed in the tho2Δ mutant as a consequence of the high frequency of deletions of the ADE2 gene by recombination. Download figure Download PowerPoint These results suggest that Tho2p is functionally related to Hpr1p. Not only does overexpression of Tho2p partially overcome the transcription defects of hpr1Δ, but tho2Δ mutants show similar incapacity to express GAL1–lacZ, increase of DNA-repeat recombination and slow growth and ts phenotypes as hpr1Δ. In addition, hpr1Δ and tho2Δ double mutants do not show a synergistic effect on recombination (Figure 2B), indicating that THO2 and HPR1 act on the same biological process. Interestingly, overexpression of HPR1 does not suppress the transcriptional defects of tho2Δ (data not shown), what might suggest that the Tho2p has a more important role on transcription than Hpr1p. Neither overexpression of HPR1 nor that of THO2 has any effect on lacZ transcription and recombination of wild-type cells (data not shown). THO1 might also have a role in transcription that can partially substitute for that of Hpr1p. However, given the lack of phenotypes of tho1Δ, even in a hpr1Δ background, we decided to concentrate on the in vivo molecular analysis of THO2. THO2 is required for transcription by RNA pol II The incapacity of hpr1Δ strains to express GAL1–lacZ is due to their incapacity to transcribe through lacZ, since hpr1Δ strains are able to activate the GAL1 promoter. Indeed, hpr1Δ cells are able to transcribe a GAL1–PHO5 fusion construct (Chávez and Aguilera, 1997). In order to determine whether the lack of expression of GAL1–lacZ in tho2Δ strains is due to their incapacity to transcribe the lacZ sequence, we compared the expression levels of GAL1–lacZ with those of GAL1–PHO5 located in single-copy centromeric plasmids. As seen in Figure 3, whereas β-galactosidase was very weakly expressed in tho2Δ strains, acid phosphatase reached 25% of the wild-type levels. This result indicates that transcription can initiate at the GAL1 promoter in tho2Δ strains. Therefore, THO2 is required for transcription of the lacZ sequence. However, our result also indicates that tho2Δ confers a significant reduction in the expression of GAL1–PHO5. As expected, no effect of tho2Δ on transcription was detected under repression conditions (data not shown). Figure 3.Expression of lacZ and PHO5 fused to the GAL1 promoter. (A) β-galactosidase activity of isogenic strains WR-4A (WT), U768–4C (hpr1Δ), WR-4B (tho2Δ) and UR-1A (hpr1Δ tho2Δ) transformed with centromeric plasmid p416GAL1lacZ carrying the GAL1–lacZ fusion construct. (B) Acid phosphatase activity of the same strains as before, transformed with centromeric plasmid pSCh202 carrying the GAL1–PHO5 fusion construct. The average value and standard deviation of two different transformants are given. Only the data of induced expression are given. Under repression conditions values were ∼1 U and 5 mU for β-galactosidase and acid phosphatase, respectively. Other details are the same as in Figure 1. Download figure Download PowerPoint Similar levels of expression of GAL1–lacZ and GAL1–PHO5 were observed in tho2Δ and tho2Δ hpr1Δ cells (Figure 3), confirming a lack of synergism of both mutations. To confirm that the absence of lacZ expression and the reduced expression levels of PHO5 was caused by transcriptional rather than post-transcriptional defects, we determined the kinetics of activation of both the lacZ and the PHO5 mRNA by Northern analysis. Figure 4 shows that GAL1-driven lacZ mRNA was not accumulated at all in tho2Δ cells after galactose induction, whereas PHO5 was accumulated up to 18% of the wild-type levels (Figure 4B). These data are consistent with the enzymatic assays (Figure 3), and confirm that tho2Δ cells, as previously shown for hpr1Δ, cannot transcribe through lacZ. In addition, even though transcription occurs through the PHO5 ORF, there is a significant reduction in the level of PHO5 mRNA, which suggests that Tho2p might also be required for transcription of PHO5. Nevertheless, an additional role for Tho2p in initiation at the GAL1 promoter cannot be dismissed in order to explain such reduction in PHO5 mRNA levels. Figure 4.Transcription analysis of GAL1–lacZ and GAL1–PHO5. (A) Northern analysis of lacZ and PHO5 mRNAs driven from the GAL1 promoter in the strains WR-4A (WT) and WR-4B (tho2Δ) transformed with p416GALlacZ and pSCh202. Transformants were obtained from overnight cultures in glycerol-lactate synthetic media lacking uracil and diluted in identical fresh media to an OD600 of 0.5 for wild type and 1.0 for tho2Δ. 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 used were the 0.5 kb BamHI–HpaI 5′ end fragment of lacZ (lacZ), a 1.5 kb EcoRI–PstI internal PHO5 fragment (PHO5) and a 589 bp 28S rRNA internal fragment obtained by PCR (rRNA). (B) Kinetics of induction of GAL1-promoter driven expression of full-length lacZ and PHO5 mRNAs as determined by quantification of the Northern blots. The mRNA values are given with respect to the rRNA levels. AU, arbitrary units. Download figure Download PowerPoint Since all results previously shown refer to GAL1 fusion constructs located in centromeric plasmids, it was important to show that tho2Δ had similar effects on transcription of chromosomal endogenous genes, whether constitutive or regulated. The mRNA driven from the endogenous chromosomal GAL1 gene is also accumulated in tho2Δ cells to 16% of the wild-type levels after induction with galactose (Figure 5), a value similar to those obtained with the episomal GAL1–PHO5 construct. On the other hand, the mRNA driven from the constitutive ACT1 endogenous gene is also significantly reduced, as it only reaches 32–53% of the wild-type levels. These results suggest a general role for Tho2p in transcription of yeast genes, regardless of whether they are located in plasmids or chromosomes, and whether they are constitutive or regulated. Figure 5.Transcription analysis of the endogenous GAL1 and ACT1 genes. (A) Northern analysis of endogenous GAL1 and ACT1 mRNA levels in wild-type WR-4A and tho2Δ WR-4B strains after different times of addition of 2% galactose to 2% glycerol-3% lactate medium. (B) Kinetics of expression as determined by quantification of the previous Northern blots. The DNA probes used were a 0.75 kb PvuII–AvaI GAL1 internal fragment (GAL1) and a 0.55 kb ClaI–ClaI ACT1 internal fragment (ACT1). Other details are the same as in Figure 4. Download figure Download PowerPoint Promoter-independent defects of transcription in tho2Δ cells Known regulatory block- or pausing-sites of transcription in eukaryotic genes are near their 5′ ends and require transcriptional activators to be bypassed by RNA pol II (for review see Eick et al., 1994; Bentley, 1995). Transcriptional elongation impairment in hpr1Δ cells, however, is promoter-independent (Chávez and Aguilera, 1997). To assess whether transcription elongation is defective in tho2Δ cells we determined the effect of tho2Δ on transcription of a GAL1–PHO5–lacZ fusion construct identical to the previously characterized GAL1–PHO5, but with the lacZ coding sequence inserted at the untranslated region (UTR) of PHO5. This means that the RNA pol II has to elongate 1.5 kb of PHO5 sequences before entering lacZ. We have already observed that RNA pol II can elongate through PHO5 to some extent in tho2Δ cells (Figures 3 and 4). However, as can be seen in Figure 6, when lacZ is inserted downstream of PHO5 (Figure 4) full-length PHO5–lacZ mRNA is accumulated in wild-type cells but not in tho2Δ cells after galactose-induction. Since in the GAL1–PHO5 construct PHO5 transcripts could be clearly detected (Figure 4), the incapacity of GAL1–PHO5–lacZ to produce PHO5–lacZ mRNA must be caused by a transcriptional elongation defect at lacZ. Quantification of total mRNA from the GAL1–PHO5–lacZ construct, indeed, shows that tho2Δ cells accumulate up to 12% of the levels of the wild-type (Figure 6B), confirming that transcription proceeds through the PHO5 coding sequence, as expected, but not through lacZ. Therefore, the reduction in mRNA levels observed in tho2Δ cells can be explained by the incapacity of the RNA pol II to transcribe through lacZ, regardless of the distance to the promoter from which it is transcribed. Figure 6.Transcription analysis of the GAL1–PHO5–lacZ fusion construct in wild-type and tho2Δ strains. (A) Northern analysis of the yeast strains RK2-6A (WT)and RK2-6C (tho2Δ) transformed with the centromeric plasmid pSCh212 carrying the GAL1–PHO5–lacZ construct in which the lacZ ORF is inserted at the 3′ UTR of PHO5. (B) Kinetics of induction of expression of full-length PHO5–lacZ and total mRNA as determined by quantification of the Northern blots. The mRNA values are given with respect to the rRNA levels. Other details are the same as in Figure 4. Download figure Download PowerPoint To confirm whether transcription elongation was impaired in tho2Δ cells at the GAL1–PHO5–lacZ, we performed run-on analysis in permeabilized cells. We found very low levels of RNA synthesis at any given place along the PHO5–lacZ fragment, including the 5′ end (Figure 7). Therefore, the strong effect of tho2Δ on transcription under the conditions used is observed even at initiation. It is probable that the negative effect of tho2Δ on transcription initiation impedes detection of any possible effect on elongation. However, given the complete lack of transcript acumulation in the GAL1–PHO5–lacZ construct in tho2Δ cells (Figure 6), it is also possible that elongation through PHO5–lacZ is blocked in the absence of Tho2p. As a consequence, transcription may not be able to reinitiate or to resume upstream of potential stall regions, explaining the very low levels of RNA pol II activity along the whole PHO5–lacZ region observed in the run-on analysis (Figure 7). Therefore, the run-on analysis clearly confirms an important role of Tho2p in transcription, but cannot answer the question of whether Tho2p has a role in transcription elongation and/or initiation. Figure 7.Transcriptional run-on analysis in wild-type and tho2Δ cells. Total RNA was isolated from wild-type RK2-6A and tho2Δ RK2-6C strains transformed with single copy plasmid pSCh212 carrying the GAL1–PHO5–lacZ construct. Two percent galactose was added to yeast cultures in glycerol-lactate synthetic medium at an OD600 of 0.1, 1 h prior to the run-on analysis. A 589 bp 28S rRNA internal fragment, three fragments of PHO5 (lanes 1–3) and three of lacZ (lanes 4–6) were immobilized in hybond-N+ filters. The PHO5–lacZ DNA region covering each of the six DNA fragments used is shown. In all cases, the percentage of radiolabelled mRNA bound to each fragment was normalized with respect to their corresponding levels in wild-type cells, taken as 100%. The orientation of the PHO5–lacZ arrow indicates the direction of transcription. As negative control we used DNA from Salmonella typhimurium (lane S). Experiments using 2% glucose (repression conditions) instead of galactose gave no signal for any of the DNA fragments used in both wild-type and tho2Δ cells (data not shown). Download figure Download PowerPoint Transcriptional elongation impairment causes hyper-recombination in tho2Δ cells We have recently shown that the impairment of transcriptional elongation causes genome instability (high frequencies of recombination and plasmid-loss), as shown in hpr1Δ cells (Chávez and Aguilera, 1997). Since tho2Δ cells show similar hyper-recombination and transcription defects as hpr1Δ, we assessed whether or not both phenotypes were also linked in tho2Δ cells. We determined the effects of tho2Δ on recombination between two 0.6 kb direct repeats. We used three direct-repeat constructs, all of them based on the same 0.6 kb leu2 internal fragment (Chávez and Aguilera, 1997). In these constructs, either the lacZ or PHO5 coding sequences have been inserted between the two direct repeats, inmediately downstream from a 3′-end truncated copy of LEU2 and immediately upstream of a 5′-end truncated copy of LEU2. In the three repeats, transcription is initiated at the unique LEU2 promoter located outside of the repeats, and has to traverse 760 bp of LEU2 before proceeding through lacZ or PHO5. The lacZ is inserted in the same transcriptional orientation as LEU2 (L-lacZ construct), whereas PHO5 is inserted in either the same (L-PHO5) or the opposite (L-PHO5r) orientation. In the latter case, transcription terminates exactly downstream of the LEU2 3′-end truncated repeat, at the terminator of the PHO5 gene (Chávez and Aguilera, 1997). If the strong hyper-recombination phenotype of tho2Δ is associated with defective transcription elongation through lacZ or PHO5, we predict a very strong hyper-recombination phenotype at L-lacZ, weaker at L-PHO5 and much weaker, if any, at L-PHO5r. As can be seen in Figure 8, the results confirmed our predictions. The frequency of recombination in the L-lacZ construct is so high that these strains cannot maintain the duplication of the leu2 fragment (Figure 8B). All cells (100%) lost the duplication. As a consequence, the LEU2-driven mRNA, clearly observed in wild-type cells, is undetectable in tho2Δ cells. In the L-PHO5 construct, the levels of mRNA covering the first leu2 repeat and PHO5 in tho2Δ cells were 6–7% of the wild-type levels, whereas recombination frequencies reach 700 times the wild-type levels (38% of the cells lost the construct). This result confirms that, indeed, transcription elongation is impaired at the PHO5 sequence causing the deletion of the repeat construct, as in L-lacZ. The higher recombination frequencies of L-lacZ versus L-PHO5 in tho2Δ strains (Figure 8B) are consistent with the observation that transcription does not proceed through lacZ, but does it through PHO5 with 25% of the wild-type efficiency (Figures 3 and 4). In the L-PHO5r construct, in which only the 3′-end truncated copy of the LEU2 repeat is transcribed, recombination in tho2Δ strains is 48 times the wild-type levels. The transcript levels in tho2Δ cells is 12% of the wild-type levels. The lower recombination frequencies of L-PHO5r (3.7%) versus L-PHO5 (38%) are consistent with the observation that transcription does not elongate properly through PHO5, causing stronger DNA repeat instability. Indeed, these results suggest that transcription of LEU2 is also defective in tho2Δ cells and responsible for the increase of recombination observed in L-PHO5r. This is consistent with a general role for Tho2p in transcriptional elongation of RNA pol II-transcribed yeast genes. It is important to note that in the direct repeat constructs studied, recombination can initiate only inside the repeat or in the regions flanked by the repeats, but not outside such sequences (Prado and Aguilera, 1995). Thus, a putative defect of transcription at the externally locat