INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Eukaryotic chromatin is a dynamic package of DNA and protein components, and the structure of chromatin can exert a profound influence on DNA metabolic processes. Chromatin remodeling, for example, accompanies transcription and involves the repositioning of nucleosomes as well as covalent histone modifications which alter DNA-histone interactions (50). In addition, regions of the genome can be maintained in a genetically silent state by protein-mediated compaction of DNA into heterochromatin (15). The role of chromatin structure in maintaining the integrity of DNA has been documented in relation to UV-induced damage, where nucleosome positioning influences the pattern of DNA damage as well as the efficiencies of DNA repair processes (41). Finally, the double-strand breaks that initiate meiotic recombination in yeast occur preferentially in nuclease-sensitive regions of chromatin, which generally correspond to promoters (13).

Although DNA metabolic processes are usually considered in isolation, these processes often occur at the same time and can, in principle, exert either inhibitory or stimulatory effects on each other. Such effects can occur through alterations in chromatin structure, direct coupling between metabolic processes, or steric interference of one process with another. Studies of the repair of UV-induced DNA damage via the nucleotide excision repair pathway were the first to demonstrate that the repair response is related to the transcriptional state of the damaged region (37). Specifically, the transcribed strand of an active gene is targeted for repair, which is initiated by the stalling of the RNA polymerase complex at transcription-blocking lesions. Transcription-coupled repair leads to the enhanced repair of transcribed genes relative to nontranscribed genes, and such preferential DNA repair is evolutionarily conserved from bacteria to humans. More recently, it has been demonstrated that the repair of oxidative damage, which generally involves the base excision repair pathway, also can be transcription coupled (6). Not only can transcription facilitate high-fidelity DNA repair processes, but also high levels of transcription have been associated with increased mutation rates in yeast (7, 27) and bacteria (2, 51). This latter phenomenon could be due to enhanced damage of transcriptionally active DNA or could reflect an interference of high levels of transcription with relatively error-free repair pathways. Finally, transcription has stimulatory effects on recombinational DNA repair in Saccharomyces cerevisiae (20, 28, 42), in Schizosaccharomyces pombe (14), and in mammalian cells (29, 30). Transcription-associated recombination has been best characterized in S. cerevisiae and will be the focus of the studies reported here.

Transcription-associated recombination in yeast was first documented through the identification of the HOT1 mitotic-specific recombination hot spot, which corresponds to the promoter elements for rDNA transcription and hence promotes high levels of transcription by RNA polymerase I (20). HOT1 stimulates both intrachromosomal and interchromosomal recombination, and transcription through the recombination substrates is necessary for the stimulatory effect (45). Mutagenesis of HOT1 has indicated a direct link between transcription and mitotic recombination, with mutations that decrease the transcriptional activity of HOT1 also decreasing the associated recombination (17, 40). Although the HOT1 studies clearly have demonstrated that RNA polymerase I transcription can stimulate recombination, the relevance of these studies to mitotic recombination between non-rDNA sequences, which normally are transcribed by RNA polymerase II, is unclear.

To address the relationship of RNA polymerase II transcription to mitotic recombination in yeast, investigators have used the GAL1-10 promoter, which can be highly induced by galactose or can be controlled genetically by the presence or absence of the positive Gal4p and negative Gal80p regulatory proteins (18). Thomas and Rothstein thus used gal10 alleles oriented as direct repeats in isogenic gal4 and gal80 strains to demonstrate an association between high levels of RNA polymerase II transcription and increased levels of mitotic recombination (42). Two features of the gal10 recombination results indicate that the observed effects might be specific to substrates oriented as direct repeats and therefore may not be generally applicable to other types of recombination events. First, elevated levels of recombination were correlated with transcription through plasmid sequences located between the directly repeated gal10 substrates rather than with transcription through the substrates themselves. Second, high levels of transcription stimulated only plasmid loss events and did not elevate simple gene conversion events. One interpretation of these results is that high levels of transcription through the nonyeast, plasmid sequences lead to the formation of recombination-initiating lesions within the plasmid sequences, which then can be repaired using the single-strand annealing (SSA) mechanism of recombination. In SSA, the ends of a double-strand break are processed by a 5'-to-3' exonuclease to yield single-stranded regions on both sides of the break. Annealing of complementary single-stranded regions (the direct repeats) and removal of the single-stranded tails results in loss of the sequences between the direct repeats (10).

Nevo-Caspi and Kupiec also have used the GAL1-10 promoter to examine the effect of transcription on recombination between Ty retrotransposons (28). In this system, induction of transcription by galactose increases the loss of a selectable marker inserted into the transcribed Ty, with marker loss occurring either by gene conversion with an unmarked Ty or by recombination involving the flanking  direct repeats at the ends of the Ty. This latter event is analogous to the plasmid loss events observed with the gal10 direct repeats. Although RNA polymerase II transcription stimulated recombination in the Ty system, it should be noted that aspects of Ty-Ty recombination appear to be unique to these elements (23, 32). In addition, it was not possible to examine the effect of transcription on crossover events. Finally, it has been reported that mitotic recombination between his4 heteroalleles is positively correlated with the activity of the endogenous HIS4 promoter (49).

In addition to the yeast studies documenting an association between high levels of RNA polymerase I or II transcription and elevated mitotic recombination, there are two notable types of yeast mutants that exhibit elevated mitotic recombination rates and may provide information relevant to the mechanism(s) of transcription-associated recombination. First, rDNA recombination is elevated in yeast topoisomerase-deficient top1 and top2 mutants (5, 21), while recombination between the direct repeats that flank the yeast Ty retrotransposon is elevated in a top3 mutant (48). It has been speculated that the increased supercoiling of DNA in toposiomerase-deficient mutants generates recombination-initiating strand breaks, and this basic idea has been extended to explain transcription-associated recombination (see reference 12 for a discussion). In the case of transcription-associated recombination, the supercoiling changes and torsional stress that accompany high levels of transcription (25) similarly could lead to recombination-initiating lesions. The second type of mutant that may provide information relevant to transcription-associated recombination is exemplified by yeast hpr1 mutants, which exhibit a mitotic hyperrecombination phenotype (1). The HPR1 gene product is important in transcript elongation (4), and it has been speculated that elongation problems generate recombination-initiating lesions. It has been further suggested that artificially high transcription of yeast genes that are normally expressed at a low level could lead to elongation problems, even in HPR1 strains (33). An hpr1-like mechanism could therefore account for the HOT1 and GAL1-10 examples of transcription-associated recombination.

To examine the generality of RNA polymerase II transcription-stimulated recombination and to determine whether gene conversion and reciprocal crossover events are similarly affected by transcription, we placed lys2 recombination substrates under the control of either the highly inducible GAL1-10 promoter or the low-level LYS2 promoter. A distinct feature of this system is that transcription of the recombination substrates can be coregulated by fusion to the same promoter or that very different levels of transcription can be achieved by fusion of the substrates to different promoters. Our results obtained with the lys2 substrates extend those obtained in previous studies and provide novel insights into the mechanism(s) of transcription-associated recombination in yeast.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth conditions. YEP medium (1% yeast extract, 2% Bacto Peptone) was used for all nonselective growth. Synthetic complete medium (39) deficient in lysine (SC  Lys) or uracil (SC  Ura) was used to select Lys⁺ and Ura⁺ prototrophs, respectively. Media were supplemented with 2% glucose (YEPD and SCD), with 2% glycerol-2% ethanol (YEPGE and SCGE), or with 2% galactose-2% glycerol-2% ethanol (YEPGGE and SCGGE) for use as carbon sources. Ura segregants were identified on synthetic minimal medium supplemented with 5-fluoroorotic acid (5-FOA) (3). All yeast strains were propagated at 30°C.

Plasmid constructions. All plasmid manipulations were done using standard molecular biological techniques (36). Plasmid pSR244 contains the gal80::HIS3 disruption allele and was constructed by replacing a BglII fragment within the GAL80 coding sequence of plasmid pSR243 (a 3.1-kb GAL80 HindIII fragment in pGEM3) with a 1.7-kb HIS3 BamHI fragment. pSR91 contains a 5.5-kb genomic BamHI fragment carrying URA3 in the BamHI site of pUC7 (note that the BamHI sites in the pUC7 polylinker are flanked by EcoRI sites). pSR88 contains the pBM150 (19) BamHI-EcoRI 0.7-kb GAL1-10 promoter (pGAL) fragment with added PstI linkers cloned into the PstI site of pUC7. All lys2 recombination substrates were derived from plasmids pDP4 and pDP6, which are pUC9 derivatives containing the entire LYS2 gene (11). pSR309 was derived from pDP6 by inserting a 1.2-kb EcoRI URA3-containing fragment (from pSR66, which contains the 1.2-kb HindIII genomic URA3 fragment in the HincII site of pUC7) into the polylinker region downstream of the LYS2 sequences.

lys2Xho

lys2Xho

pLYS-lys25'

pGAL-lys25'

lys25'

pLYS2-lys25'

pGAL-lys25'

pLYS-lys23'

lys23'

lys23'

pGAL-lys23'

pGAL-lys25'3'

pLYS-lys25'3'

Yeast strain constructions. All strains used in this study were derived from haploid strain SJR195 (MATa ade2-101_oc his3200 ura3Nco) by lithium acetate transformation (38). All strain genotypes were confirmed by Southern blot analysis and/or appropriate phenotypic tests. Strains were constructed with various lys2 recombination substrates positioned either as repeats on nonhomologous chromosomes (heterochromosomal repeats) or as direct repeats on the same chromosome (intrachromosomal repeats). Each lys2 recombination substrate was under control of either the low-level LYS2 promoter (pLYS) or the highly inducible GAL1-10 promoter (pGAL). pGAL promoter activity was maintained at a low or high level by placing the lys2 recombination substrates in a GAL80 or gal80::HIS3 background, respectively. One-step disruption of GAL80 was accomplished by transforming cells with NcoI-SmaI-digested pSR244 and selecting His⁺ transformants.

pGAL-lys2Bgl

pLYS-lys2Bgl

lys2Bgl

9

8

lys2Bgl

lys2Bgl

lys2Xho

lys2Xho

pLYS-lys2Xho

pGAL-lys2Xho

10

9

lys25'

pLYS-lys25'

pGAL-lys25'

pLYS-lys23'

pGAL-lys23'

pLYS-lys2Xho

pGAL-lys2Xho

pLYS-lys25'3'

pGAL-lys25'3'

pLYS-lys25'

pGAL-lys25'

pLYS-lys23'

pGAL-lys23'

pLYS-lys23'

pGAL-lys23'

Recombination rate analysis. Recombination rates were determined by the method of the median (24). Briefly, 2-day old colonies were excised from YEPD plates, inoculated into YEPGE liquid medium, and grown for 2 days on a roller drum. Cells were harvested, washed once with sterile H₂O, and resuspended in 1 ml of H₂O. Aliquots (100 µl) of appropriately diluted cells were plated on SCGGE  Lys medium for prototroph selection and on YEPD to determine the numbers of viable cells. Lys⁺ colonies were counted on day 3 after selective plating, and this number was used to determine the total number of recombinants per culture. The median number of Lys⁺ colonies per culture was determined using at least nine independent cultures. The median (or corrected median for the conversion-crossover substrates [see below]) was then used to calculate the recombination rate (number of recombinants per generation). To determine the rates of conversions versus crossovers, the total rate was partitioned using the proportions of gene conversion versus crossover events.

Partitioning of recombinants. At least 50 independent Lys⁺ recombinants from each relevant strain were classified as gene conversion or crossover events. With the heterochromosomal conversion-crossover recombination substrates, the lys23' allele on chromosome V is flanked by ura3/URA3 direct repeats (see Fig. 1A). Recombination between the flanking direct repeats occurs at a high level, generating Ura segregants that can be detected on 5-FOA. A Lys⁺ recombinant generated by a gene conversion event will maintain the ura3/URA3 direct repeats, whereas a Lys⁺ recombinant generated by a crossover event will be accompanied by physical separation of the ura3/URA3 direct repeats (16). Thus, a Lys⁺ recombinant resulting from gene conversion will produce colonies (papillate) at a very high level on 5-FOA while one generated by a crossover event will papillate at a very low level on 5-FOA. Independent Lys⁺ recombinants were isolated and purified, and three colonies of each were checked for the papillation phenotype on 5-FOA. Correlation between the low-versus high-papillation phenotypes and the occurrence of a crossover versus a gene conversion event was confirmed by Southern blot analysis of random recombinants.

lys23'

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Heterochromosomal conversion-crossover assay system. The initial assay system used to examine the impact of transcription on mitotic recombination was composed of lys2 heteroalleles positioned on nonhomologous chromosomes in a haploid yeast strain. As shown in Fig. 1A, the LYS2 locus on chromosome II contained the lys2-Nhe frameshift allele, either under control of its normal promoter (pLYS) or fused to the highly inducible GAL1-10 promoter (pGAL). A second copy of lys2 was placed at the URA3 locus on chromosome V by integrating a plasmid containing a lys2 allele truncated at the 3' end (lys23'). As with the full-length lys2-Nhe allele, two versions of the lys23' allele were constructed, one controlled by pLYS and the other controlled by pGAL. pGAL activity was modulated by constructing pairs of isogenic Gal80⁺ and Gal80 strains that contained or did not contain, respectively, the negative regulatory protein Gal80p. Although galactose is routinely used to induce high levels of pGAL activity, the use of Gal80⁺-Gal80 strain pairs allowed promoter activity to be controlled without changing cellular growth conditions. In all the experiments reported here, cells were grown nonselectively in rich medium containing glycerol and ethanol as carbon sources; glucose was not used in order to avoid catabolite repression of pGAL activity. Under these growth conditions, transcription from pGAL, as assayed by -galactosidase production from a pGAL-lacZ fusion, was approximately 1,000-fold higher in a Gal80 strain than in a Gal80⁺ strain (data not shown). To identify Lys⁺ recombinants, nonselectively grown cells were plated on lysine-deficient medium supplemented with galactose as well as glycerol and ethanol. In the presence of galactose, pGAL is highly active even in the presence of Gal80p, thus allowing the identification of pGAL-LYS2 recombinants as well as pLYS-LYS2 recombinants in Gal80⁺ strains.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Structures of recombination substrates. Open boxes correspond to lys2 sequences (point mutations are indicated with asterisks), and shaded boxes correspond to URA3/ura3 sequences. Thick and thin horizontal lines in panels A to C correspond to chromosomes II and V, respectively. The thick line in panel D corresponds to chromosome V sequences that flank the lys2 direct repeats, and the thin line corresponds to plasmid sequences between the direct repeats. GC, gene conversion; c/o, crossover. (A) Heterochromosomal conversion-crossover substrates; (B) heterochromosomal conversion-only substrates; (C) heterochromosomal crossover-only substrates; (D) intrachromosomal direct-repeat substrates.

lys2-Nhe/lys23'

lys23'

lys23'

lys23'

lys23'

lys23'

lys23'

Heterochromosomal conversion-only assay system. The initial assay system was modified so that only gene conversion events unassociated with crossing over would be detected. As shown in Fig. 1B, the lys2 allele positioned on chromosome V was truncated at both the 5' and 3' ends (lys25'3') instead of only at the 3' end. The lys2 allele at the LYS2 locus on chromosome II was again full length, but the position of the frameshift mutation was changed (lys2Xho allele with a filled-in XhoI site) so that the mutation was roughly in the center of the homology region. Recombination between the lys2Xho and lys25'3' alleles can produce Lys⁺ recombinants via a gene conversion event, with the full-length allele acting as the recipient of genetic information and the truncated allele acting as the donor. Resolution of the event as a crossover, however, will yield a 5'-truncated allele and a 3'-truncated allele as products, neither of which will produce a Lys⁺ phenotype. Thus, the system detects only simple gene conversion events that are unassociated with crossing over. As with the conversion/crossover assay system, the lys2 alleles were fused to either the same promoter or different promoters and the promoter activity of pGAL was regulated by the presence or absence of Gal80p.

pLYS2-lys2Xho

pGAL-lys2Xho

lys2Xho

lys25'3'

lys2Xho

Heterochromosomal crossover-only assay system. To detect only recombination events that are resolved as crossovers, the lys2 allele at the LYS2 locus on chromosome II was truncated at the 5' end (lys25') and used in combination with the standard lys23' allele at the URA3 locus on chromosome V. As illustrated in Fig. 1C, only a crossover event will generate a full-length LYS2 allele. With the exception of the 5' or 3' truncation, the alleles are wild type in sequence, and so repair processes will not influence the detection of recombinants. The substrates were both fused to pGAL, or one was fused to pGAL and one was fused to pLYS, and recombination rates were examined in Gal80⁺ and Gal80 strains. It should be noted that when both substrates are fused to pGAL, it is formally possible to restore the 5' end of the lys25' by a noncrossover (simple gene conversion) mechanism.

lys23'

lys23'

lys25'

Intrachromosomal direct-repeat assay system. The heterochromosomal conversion-crossover assay system was derived by transforming a lys2-Nhe strain with an integrating plasmid containing URA3 and the lys23' allele, with the plasmid being targeted to integrate at the URA3 locus by appropriate enzyme digestion. To directly compare transcriptional effects on heterochromosomal versus intrachromosomal recombination, the plasmid also was targeted to integrate at the LYS2 locus by digestion within the lys23' sequences. As shown in Fig. 1D, this generates lys23'/lys2-Nhe direct repeats, with the direct repeats flanking the remainder of the plasmid sequences, including the selectable URA3 marker. Lys⁺ recombinants that retain the URA3 marker correspond to simple gene conversion events unassociated with crossing over, while loss of the URA3 marker indicates the loss of plasmid sequences. Plasmid sequences can be lost either via a crossover event between the lys2 direct repeats or by an SSA mechanism. As with the heterochromosomal repeats, the direct repeats were fused to either the same promoter (pLYS or pGAL) or different promoters and recombination rates were then measured in Gal80⁺ versus Gal80 strains.

lys23'

6

6

7

6

lys23'

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The impact of RNA polymerase II transcription on mitotic recombination in yeast was examined using lys2 alleles under control of the low-level LYS2 promoter (pLYS) and/or the highly inducible GAL1-10 promoter (pGAL). In most assays, the lys2 alleles were positioned on nonhomologous chromosomes (heterochromosomal substrates) to avoid some of the mechanistic ambiguities associated with repeats positioned on the same chromosome (intrachromosomal substrates). Three types of heterochromosomal assays were used: an assay that detects both crossover- and noncrossover-associated gene conversion events, an assay that detects only gene conversion events that are unassociated with crossing over, and an assay that detects only crossover events (Fig. 1A to C, respectively). In addition to the heterochromosomal assays, an intrachromosomal direct-repeat assay (Fig. 1D) was established to compare and contrast the impact of transcription on recombination between direct repeats with that for repeats on nonhomologous chromosomes. The data obtained using these assays are summarized in Fig. 2. In the discussion that follows, the transcription-associated recombination observed in our system is compared to that observed in HOT1 experiments (RNA polymerase I transcription) and in experiments that utilized gal10 direct repeats (RNA polymerase II transcription). In addition, possible models for transcription-associated recombination are considered in relation to the results reported here.

View larger version (24K): [in this window] [in a new window]   FIG. 2.   Summary of recombination rate data. Substrate sequences are as follows: open boxes correspond to lys2 sequences (asterisks indicate point mutations), solid boxes correspond to pGAL, and shaded boxes correspond to pLYS. Horizontal arrows indicate substrates that are highly transcribed in gal80 strains. Substrates 1 to 4 are heterochromosomal conversion-crossover substrates, substrates 5 to 8 are heterochromosomal conversion-only substrates, substrates 9 to 11 are heterochromosomal crossover-only substrates, and substrates 12 to 15 are intrachromosomal direct-repeat substrates.

The results obtained using the heterochromosomal lys2 conversion-crossover assay demonstrate that high levels of RNA polymerase II transcription can stimulate both gene conversion and crossover events (substrates 1 to 4 in Fig. 2). Our observation that RNA polymerase II-promoted transcription stimulates both gene conversion and crossover events agrees with RNA polymerase I-promoted (HOT1) recombination results obtained using allelic sequences (20). With the HOT1 system, however, the examination of conversion tract lengths led to the suggestion that crossovers might correspond to break-induced recombination (BIR) events, in which a broken end is used to prime DNA replication that extends all the way to the end of the unbroken, template chromosome (47). Because BIR is a nonreciprocal process that replaces all the information distal to the break with information from the invaded chromosome, only one of the exchange products normally associated with a true reciprocal crossover event is produced. In our heterochromosomal assay system, BIR would yield only half of a reciprocal translocation, resulting in what is presumably a lethal aneuploidy. Southern blot analysis of representative Lys⁺ recombinants classified as crossovers based on phenotypic analysis confirmed the presence of both translocation products (data not shown). The data reported here thus demonstrate that transcription-stimulated recombination is indeed associated with the production of reciprocal crossover products.

In addition to demonstrating transcriptional stimulation of both gene conversion and crossover events, experiments with the conversion-crossover lys2 substrates showed that there is a transcription-associated asymmetry in simple gene conversion events. This asymmetry was evident when only one of the substrates was highly transcribed (compare substrates 1 and 2 in Fig. 2). Specifically, transcription of only the full-length lys2-Nhe allele, which acts as the recipient in gene conversion events, stimulated gene conversion 14-fold, whereas transcription of only the truncated lys23' allele, which acts as a donor in gene conversion events, did not significantly stimulate gene conversion. This asymmetry also was evident in strains containing substrates that produce Lys⁺ recombinants only via a gene conversion mechanism (substrates 5 and 6 in Fig. 2). When the recipient allele (lys2Xho) was highly transcribed, gene conversion was stimulated 11-fold; when the donor allele (lys25'3') was highly transcribed, gene conversion was stimulated only 2.6-fold. It is generally assumed that mitotic recombination is limited by the occurrence of initiating DNA lesions and that the allele that suffers the initiating lesion will be the recipient in gene conversion events (31). Our data thus suggest that one effect of high levels of RNA polymerase II transcription is to increase the occurrence of recombination-initiating lesions. We note that a similar gene conversion asymmetry was seen in HOT1 experiments when only one allele was highly transcribed by RNA polymerase I and that a link between transcription and initiating lesions was suggested (46). An alternative explanation for the pattern of transcription-associated gene conversion observed here invokes a strong transcription-associated bias in recognition and/or repair of mismatches present in recombination intermediates. Although this is a formal possibility, it cannot explain the stimulatory effects of transcription on recombination when the substrates contain no potential mismatches (i.e., the crossover-only substrates).

In addition to using lys2 substrates that produce prototrophs only via a simple gene conversion mechanism, we constructed substrates that can produce Lys⁺ recombinants only via a crossover event (substrates 9 to 11 in Fig. 2). As with the gene conversion-only assay, the crossover-only assay revealed an interesting asymmetry when only one of the substrates was highly transcribed. Specifically, high transcription of only the lys23' allele stimulated crossovers 12-fold whereas high transcription of only the lys25' allele stimulated crossovers only 1.5-fold. We suggest that this asymmetry is related to the issue of recombination-initiating lesions (see above) and reflects the distance of the substrate overlap region from pGAL. In the case where recombination is stimulated, the overlap is separated from pGAL by a nonhomologous region; in the case where recombination is not stimulated appreciably, pGAL directly abuts the substrate overlap region. Thus, the further the overlap region is from pGAL, the higher is the probability that a recombination-initiating lesion will occur. Voelkel-Meiman and Roeder observed a similar gradient in HOT1 gene conversion experiments and suggested that the conversion gradient could be related to breaks associated with transcription-induced torsional stress (46). A similar argument could account for the asymmetry observed with our crossover-only substrates.

Although there are clear similarities between the recombination results we obtained with pGAL-promoted transcription and those obtained with HOT1-promoted transcription, there is one notable exception. With all of the heterochromosomal substrates used here, high-level transcription of only one substrate from pGAL had a greater stimulatory effect on recombination than did high-level transcription of both substrates. Although this seems counterintuitive, it suggests that high transcription of a donor sequence may actually inhibit strand invasion. This could be tested by using the HO endonuclease to introduce recombination-initiating double-strand breaks in the recipient substrate and then assaying repair efficiency using donor substrates that are subject to high versus low transcription. In contrast to our results, there appeared to be an additive effect of highly transcribing both substrates in the HOT1 experiments (46). The difference in HOT1-versus pGAL-associated recombination could reflect an inherent difference in RNA polymerase I and RNA polymerase II transcription or a difference in the assay systems used.

The results obtained with the heterochromosomal lys2 substrates contrast with those obtained by the Rothstein laboratory using gal10 direct repeats. With the gal10 direct repeats, high levels of RNA polymerase II transcription stimulated only events involving loss of sequences between the direct repeats (plasmid loss events) and failed to stimulate simple gene conversion events (42). With direct repeats, plasmid loss can be due to a true reciprocal crossover event involving Holliday junction resolution or can result from the alternative, nonconservative mechanism of SSA (see reference 3 for a description of recombination mechanisms). The absence of transcription-stimulated gene conversion events in the gal10 direct-repeat system suggests that most of the transcription-associated recombination events observed in this system corresponded to SSA events. The overlapping RAD52/RAD1 genetic requirements of the plasmid loss events (43), as well as the observation that transcription between the direct repeats rather than through the repeats was correlated with enhanced recombination (42), is consistent with this interpretation. As discussed in more detail below, the necessity of transcribing sequences between the direct repeats probably is related to the direct-repeat hyperrecombination phenotype observed in hpr1 mutants (33).

To more directly compare heterochromosomal and intrachromsomal transcription-stimulated recombination events, the same lys2 substrates were positioned on nonhomologous chromosomes (substrates 1 to 4 in Fig. 2) or as direct repeats on the same chromosome (substrates 12 to 15 in Fig. 2). In considering results obtained with the direct-repeat substrates, it should be borne in mind that pGAL is a bidirectional promoter. Fusion of pGAL to the downstream, full-length lys2-Nhe allele thus not only results in high-level transcription of lys2-Nhe but also probably results in high-level transcription of the upstream plasmid sequences that are between the lys2 substrates (substrates 12 and 14 in Fig. 2). Furthermore, when both lys2 alleles are fused to pGAL (substrate 14 in Fig. 2), the transcription complexes may converge in the region between the substrates. As observed with the heterochromosomal lys2 repeats, high levels of transcription stimulated both simple gene conversion events and plasmid loss events in the lys2 direct-repeat assay. This is in contrast to lack of simple gene conversion stimulation with gal10 direct repeats (42). In addition, an asymmetry in gene conversion events was evident with the lys2 direct repeats, which was similar to that described above for the heterochromosomal repeats. That is, simple gene conversion events were stimulated only when the recipient, full-length allele was highly transcribed (compare direct-repeat substrates 12 and 13 in Fig. 2). The partitioning of substrate 13 recombinants into conversions versus plasmid loss events indicated that the rate of simple conversion events actually decreased when the donor allele was highly transcribed. This was confirmed by direct measurement of the conversion rates for substrates 12 and 13. Although the reason for this behavior is unclear, it suggests that transcription not only increases the frequency of recombination-initiating lesions but also may alter the recombinational repair of lesions generated by other mechanisms. In the case of the lys2 direct repeats, transcription of the region between the substrates may channel lesions into the SSA pathway. In contrast to the results obtained with the heterochromosomal lys2 repeats, high-level transcription of both substrates in the direct-repeat assay stimulated the total rate of recombination to a greater extent than did high-level transcription of just one substrate. This increased stimulation, however, pertained only to plasmid loss events and may be a consequence of convergent transcription complexes. The different behavior of the heterochromosomal and intrachromosomal lys2 repeats under high-transcription conditions is consistent with SSA being a unique and important recombination mechanism for the intrachromosomal direct repeats.

There are two general models that can account for transcription-stimulated recombination. In the first model, transcription alters the repair of the preexisting DNA lesions by specifically channeling them into the recombination repair pathway or by facilitating the process of recombination. For example, there could exist a recombination subpathway that is analogous to transcription-coupled nucleotide excision repair. Although there is no evidence for transcription-coupled recombination, it is intriguing that at least one recombination protein is associated with a eukaryotic transcription complex (26). Alternatively, high levels of transcription might interfere with competing DNA repair pathways and thereby favor recombination as a DNA repair mechanism. Finally, the open chromatin structure associated with transcription could directly facilitate recombination by increasing the efficiency with which a duplex molecule is invaded by a single strand of DNA. In in vitro recombination studies involving interaction between a single-stranded DNA and a duplex DNA molecule, the assembly of nucleosomes on the duplex molecule was found to inhibit the strand exchange reaction (22, 34). The inhibition was relieved, however, if the chromatin template was transcribed (22). As noted above, transcription of the recipient lys2 allele affected gene conversion more than did transcription of the donor allele (11-fold and 2.6-fold, respectively) in our in vivo conversion-only assay. If the primary effect of transcription in our system were to facilitate the invasion of a duplex molecule, one would expect the greatest elevation in gene conversion rate to accompany transcription of the donor sequence. The relatively small stimulation in gene conversion that accompanied high-level transcription of the donor molecule may indeed reflect more efficient invasion of transcriptionally active DNA, with high levels of transcription perhaps replacing the need for one or more recombination proteins. This possibility could be addressed by examining transcription-associated recombination in appropriate yeast mutant strains. In relation to the role of transcription-associated chromatin remodeling in facilitating strand invasion, we note that transcription of a target sequence in mammalian cells has been reported to increase the efficiency of gene targeting (44).

The second general model that has been proposed to explain transcription-stimulated recombination invokes transcription-associated DNA damage. First, through alterations in chromatin structure and nucleosome positioning, high levels of transcription may increase the accessibility of DNA to endogenous damaging agents. In addition, the transient regions of single-stranded DNA in transcription bubbles may be more susceptible to damage than is duplex DNA (2). Enhanced DNA damage indeed has been invoked to explain the elevated mutation rates that accompany high levels of transcription in yeast (27). Second, the process of transcription generates torsional stress on the DNA template, with regions of positive and negative supercoiling accumulating ahead of and behind, respectively, the transcription machinery (25). Torsional stress has been invoked to explain the increase in rDNA recombination that accompanies the loss of topoisomerase I or II activity in yeast (5, 21). In addition, recombination between the long terminal repeats () of the yeast Ty retrotransposon is increased in top3 mutants (48). The polarity we observed with the lys2 crossover-only substrates and that observed with HOT1-stimulated gene conversion (46) would be consistent with a torsional-stress model. The role of torsional stress in transcription-stimulated recombination could be addressed by examining recombination between lys2 substrates in appropriate topoisomerase-deficient yeast mutants.

A third source of RNA polymerase II transcription-associated DNA damage relates to observations made with yeast hpr1 mutants, which were initially identified as hyperrecombination mutants in a direct-repeat assay (1). Chavez and Aguilera have correlated the increase in direct repeat recombination in hpr1 mutants with a failure to efficiently elongate transcription through the plasmid sequences that separate the direct repeats (4). They thus have hypothesized that a stalled transcription complex either directly induces recombination-initiating strand breaks or impedes the DNA replication machinery, thereby stimulating recombinational bypass. Piruat and Aguilera have noted that all examples of transcription-associated recombination in yeast involve high-level transcription of sequences that normally are expressed at a low level in yeast or are foreign to the yeast genome and thus may be mechanistically similar to hpr1-associated recombination (33). Specifically, in the HOT1 system, his4 alleles were fused to the rDNA promoter (20); in the gal10 direct-repeat system, transcription of plasmid sequences (pBR322) between the repeats was correlated with elevated recombination (42); and in our system, lys2 alleles were fused to pGAL. If transcription elongation problems are indeed triggering recombination in these systems, one would predict that transcription-associated recombination should be further elevated in an hpr1 mutant background. In the gal10 direct-repeat system, eliminating Hpr1p and inducing transcription have synergistic effects on recombination (9), suggesting that elongation problems are responsible for much of the transcription-associated recombination observed in this system. With the heterochromosomal repeats, we can eliminate the transcription of nonyeast plasmid sequences as the causative factor, because we saw elevated recombination rates when no plasmid sequences were linked to the highly transcribed lys2 allele (e.g., substrate 1 in Fig. 2). Elimination of Hpr1p in our strains should indicate whether the transcription-stimulated recombination is related to potential elongation problems through the lys2 recombination substrates. Finally, it has been demonstrated that convergence of RNA polymerase III transcription and DNA replication impedes replication fork movement (8). Although it is not known whether pausing of replication forks triggers recombination in yeast, we note that a similar pausing associated with high levels of RNA polymerase II transcription potentially could trigger the transcription-associated recombination observed in our lys2 system.

In summary, the results reported here demonstrate that high levels of RNA polymerase II transcription can stimulate both mitotic gene conversion and reciprocal crossover events. The asymmetry observed in simple (noncrossover) gene conversion events suggests that the primary effect of high-level transcription is to increase the frequency of recombination-initiating lesions, although the data also indicate that transcription may facilitate the invasion of a duplex chromatin template. Therefore, there may be multiple, independent mechanisms that are important for the elevated recombination associated with high levels of transcription in yeast. The assay system developed for this study should be useful for genetically determining the contributions of these mechanisms to transcription-associated recombination.