Yeast spt6-140 Mutation, Affecting Chromatin and Transcription, Preferentially Increases Recombination in Which Rad51p-Mediated Strand Exchange Is Dispensable

Corresponding author: Andrés Aguilera, Departamento de Genética, Facultad de Biología, Universidad de Sevilla, Avd. Reina Mercedes 6, 41012 Seville, Spain., aguilo{at}cica.es (E-mail)

Communicating editor: L. S. SYMINGTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have shown that the spt6-140 and spt4-3 mutations, affecting chromatin structure and transcription, stimulate recombination between inverted repeats by a RAD52-dependent mechanism that is very efficient in the absence of RAD51, RAD54, RAD55, and RAD57. Such a mechanism of recombination is RAD1-RAD59-dependent and yields gene conversions highly associated with the inversion of the repeat. The spt6-140 mutation alters transcription and chromatin in our inverted repeats, as determined by Northern and micrococcal nuclease sensitivity analyses, respectively. Hyper-recombination levels are diminished in the absence of transcription. We believe that the chromatin alteration, together with transcription impairment caused by spt6-140, increases the incidence of spontaneous recombination regardless of whether or not it is mediated by Rad51p-dependent strand exchange. Our results suggest that spt6, as well as spt4, primarily stimulates a mechanism of break-induced replication. We discuss the possibility that the chromatin alteration caused by spt6-140 facilitates a Rad52p-mediated one-ended strand invasion event, possibly inefficient in wild-type chromatin. Our results are consistent with the idea that the major mechanism leading to inversions might not be crossing over but break-induced replication followed by single-strand annealing.

HOMOLOGOUS recombination is an important repair mechanism of DNA breaks in mitosis. Breaks may occur spontaneously as a consequence of failures of either DNA replication or other DNA repair pathways, as well as by the direct damage caused to DNA by radicals, chemicals, or radiations (see AGUILERA et al. 2000 ). In eukaryotes, our understanding of mitotic recombination owes much to the genetic analysis of DNA repair of ionizing radiation and HO-induced double-strand breaks (DSBs) in yeast (see PAQUES and HABER 1999 ). Mutations in genes of the RAD52 epistasis group not only cause sensitivity to ionizing radiations, but also abolish meiotic recombination (see PETES et al. 1991 ). Nevertheless, their effect on mitotic recombination is very heterogeneous. Whereas rad52 abolishes all types of mitotic recombination, with a weak effect on deletions between some direct repeats (MALONE and ESPOSITO 1980 ; JACKSON and FINK 1981 ; see PETES et al. 1991 ; PAQUES and HABER 1999 ), the effect of rad51, rad54, rad55, and rad57 is clear for mitotic gene conversion (see PAQUES and HABER 1999 ) but weak for inversions (RATTRAY and SYMINGTON 1994 ; AGUILERA 1995 ). This weak repeat recombination phenotype of rad51 and related mutant strains served to identify RAD59, a gene controlling most of the recombination occurring in the absence of Rad51p (BAI and SYMINGTON 1996 ). Altogether, these results indicate that mitotic recombination in yeast may occur via different mechanisms that are under specific genetic control.

Genetic and physical analyses of DSB-induced recombination suggest that in mitosis the major mechanism leading to gene conversion is synthesis-dependent strand annealing (SDSA) as proposed for Drosophila and Ustilago (NASSIF et al. 1994 ; FERGUSON and HOLLOMAN 1996 ). This mechanism requires strand exchange and involves heteroduplex formation, leading to gene conversion events unassociated with crossing over. Homologous recombination between heterologous chromosomes in yeast, Drosophila, and mammalian cells occurs mainly in the absence of crossing over (GLOOR et al. 1991 ; NASSIF et al. 1994 ; RICHARDSON et al. 1998 ; VIRGIN et al. 2001 ) and sister-chromatid repair in mammalian cells is primarily gene conversion unassociated with crossing over (JOHNSON and JASIN 2000 ). This view is consistent with the general findings that mitotic recombination events, contrary to meiotic events, are mainly unassociated with crossing over (see PETES et al. 1991 ). Consequently, mitotic and meiotic recombination may proceed via different mechanisms, it being unclear whether mitotic recombination requires Holliday junction formation and resolution as shown in yeast meiosis (SCHWACHA and KLECKNER 1995 ) and as predicted by the DSB recombination repair model of SZOSTAK et al. 1983 .

The observation that an HO-mediated DSB induces the restoration of a full chromosome arm in rad51 strains suggests that in the absence of Rad51p a DSB is repaired via a mechanism of break-induced replication (BIR) in yeast (MALKOVA et al. 1996 ), which resembles the mechanism of recombination-dependent replication proposed for Escherichia coli (ASAI et al. 1994 ; KOGOMA 1996 , KOGOMA 1997 ). A similar mechanism has been invoked for conversion of a long chromosome fragment induced by the HOT1 hotspot of recombination (VOELKEL-MEIMAN and ROEDER 1990 ) and for the process of yeast chromosome fragmentations by linear DNA fragments (MORROW et al. 1997 ).

Although the analysis of DSB-induced recombination in vivo is very useful in the study of recombination, we need to understand spontaneous recombination to have a precise knowledge of the mechanisms of mitotic recombination. In this sense, it has been shown that the incidence of spontaneous recombination at a particular DNA sequence may be stimulated by other DNA processes such as transcription or chromatin modification (see AGUILERA et al. 2000 ). Several in vitro studies have shown an effect of nucleosomes in strand exchange and branch migration catalyzed by bacterial proteins (KOTANI and KMIEC 1994 ; GRIGORIEV and HSIEH 1997 , GRIGORIEV and HSIEH 1998 ). Indirect evidence has been accumulated on the effect of chromatin in the incidence of spontaneous recombination in vivo. Hyper-recombination has been shown in the rDNA region of sir2 mutants of Saccharomyces cerevisiae (GOTTLIEB and ESPOSITO 1989 ) altered in chromatin structure. rDNA silencing and chromatin accessibility respond inversely to Sir2p dosage, indicating a double effect of closed chromatin in reducing both transcription and recombination (FRITZE et al. 1997 ). In addition, it has been shown that HO-induced recombination in MAT sequences requires the RAD51 series of genes only when the MAT genes are both silenced and located in a chromosome (SUGAWARA et al. 1995 ).

Stimulation of recombination by transcription has been shown in phage and bacteria (IKEDA and KOBAYASHI 1977 ; IKEDA and MATSUMOTO 1979 ; DUL and DREXLER 1988 ; VILETTE et al. 1995 ), yeasts (VOELKEL-MEIMAN et al. 1987 ; STEWART and ROEDER 1989 ; THOMAS and ROTHSTEIN 1989 ; GRIMM et al. 1991 ; NEVO-CASPI and KUPIEC 1994 ; SAXE et al. 2000 ) and mammalian cell cultures (NICKOLOFF 1992 ; THYAGARAJAN et al. 1995 ) as well as in V(D)J recombination and class switching (BLACKWELL et al. 1986 ; LAUSTER et al. 1993 ; OLTZ et al. 1993 ; DANIELS and LIEBER 1995 ). Different studies of the yeast hyper-recombination mutations hpr1 and tho2 have led to the proposal that transcription-associated recombination may be caused by transcription-elongation failures leading to recombinogenic substrates (CHAVEZ and AGUILERA 1997 ; PRADO et al. 1997 ; PIRUAT and AGUILERA 1998 ).

In a previous study we measured the recombination rates of nine different mutants affected in SWI/SNF and SPT/SIN genes related to transcription and chromatin structure (MALAGON and AGUILERA 1996 ). We found that spt4 and spt6 mutations confer hyper-recombination of particular DNA repeat constructs. Extensive genetic analyses of different mutations have revealed that Spt4p and Spt6p are functionally related (SWANSON and WINSTON 1992 ; HARTZOG et al. 1998 ). Spt6p is an essential protein involved in chromatin structure and transcription elongation that interacts with the globular domain of the histone H3 (SWANSON et al. 1990 ; BORTVIN and WINSTON 1996 ; HARTZOG et al. 1998 ). The spt6-140 mutation has been shown to alter chromatin structure of yeast cells as well as transcription (BORTVIN and WINSTON 1996 ). Spt4p is a nonessential protein also involved in chromatin structure and transcription (SWANSON and WINSTON 1992 ; MALONE et al. 1993 ). It forms with Spt5p the 5,6-dichloro-1ß-D-ribofuranosylbenzimidazole sensitivity-inducing factor (DSIF) participating in transcription elongation (HARTZOG et al. 1998 ; WADA et al. 1998 ).

To gain more insight into the mechanisms of spontaneous mitotic recombination, we have analyzed at the genetic and molecular levels the recombination events stimulated by spt4 and spt6. We show that spt6-140 and spt4-3 mutations stimulate recombination between inverted repeats primarily by a Rad52p-dependent mechanism that is partially dependent on Rad1p and Rad59p and very efficient both in the absence and the presence of Rad51p, Rad54p, Rad55p, and Rad57p. Our study provides evidence that chromatin alteration, together with transcription impairment, caused by spt6-140, stimulates the incidence of recombination events and their efficiency, in particular in the absence of Rad51p-mediated strand exchange. Recent studies using inverted repeats as recombination substrates indicate that, contrary to the belief that inversions are the result of an intrachromatid crossing over, inversions may be the result of unequal sister-chromatid gene conversion (ROTHSTEIN et al. 1987 ; CHEN and JINKS-ROBERTSON 1998 ) or intrachromatid BIR followed by single-strand annealing (SSA; BARTSCH et al. 2000 ; KANG and SYMINGTON 2000 ). Our results are consistent with the idea that inversions occur primarily by BIR-SSA and we discuss the possibility that the chromatin alteration caused by spt6-140 facilitates one-ended invasion events.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains:
Yeast strains used are listed in Table 1.

Plasmids:
YEp351-RAD51, containing the RAD51 gene in YEp351, was isolated from the MW90 library (WALDHERR et al. 1993 ) by complementation of the methyl-methanesulfonate (MMS)-sensitivity phenotype of the rad51 strain U678.

Plasmid pRS315-RAD51 was constructed by cloning the 2.1-kb BamHI-PstI RAD51 fragment from YEp351-RAD51 into pRS315 opened with BamHI and PstI (SIKORSKI and HIETER 1989 ).

Plasmid p315-rad51-k, carrying the rad51::KanMX4 allele was constructed by replacing in vitro the 0.93-kb StuI-NsiI internal fragment of RAD51 by the 1.5-kb EcoRV-PstI fragment of plasmid pFA6a-KanMX4, containing the KanMX4 deletion cassette (WACH et al. 1994 ).

Plasmid pAAX43-TER, containing the CYC1ter sequences upstream of the his35'-leu2-r construct, was made by inserting the 0.3-kb SmaI-KpnI CYC1 transcription terminator fragment from plasmid p424GAL1 (MUMBERG et al. 1994 ) into the unique EcoRI site of plasmid pAAX43 (AGUILERA and KLEIN 1989A ). The KpnI and EcoRI ends were made blunt with Klenow prior to ligation.

Plasmid pAAX43k, containing the his3-k-leu2-r construct, was made by subcloning the 0.3-kb BssHII-NheI HIS3 fragment carrying the his3-k allele from plasmid pSZ513 (ORR-WEAVER et al. 1988 ) into pAAX43 opened with BssHII and NheI.

To construct the plasmid pBR-536L containing the his3-h LEU2 we first subcloned the 1.97-kb SalI LEU2 fragment from pHK130 (AGUILERA and KLEIN 1989A ) into the XhoI site of pSZ536 containing the his3-h allele, made by filling with Klenow the HindIII site at position +492 (ORR-WEAVER et al. 1988 ). From this new plasmid, pSZ536-LL, we isolated the 3.74-kb BamHI fragment and cloned it in the BamHI site of pBR322, to obtain pBR-536L.

Construction of yeast strains carrying deletions of RAD1, RAD51, RAD52, RAD54, RAD55, RAD57, and RAD59:
Yeast strains carrying the rad51::KanMX4 deletion in which 75.1% (925 bp) of the RAD51 open reading frame (ORF) was removed, were obtained by transforming the yeasts strains M137-11A, M137-11B, and M236-12D with the 2.7-kb EcoRV-XbaI fragment of pRS315-rad51::k.

To delete the RAD1, RAD52, RAD54, RAD55, RAD57, and RAD59 genes we used the short flanking homology method of WACH et al. 1994 . Oligonucleotides used are listed in Table 2. Deletion alleles thus constructed lack 94.4, 94.5, 94.9, 96.4, 95.2, and 94.7% of the RAD1, RAD52, RAD54, RAD55, RAD57, and RAD59 ORFs, respectively. They were made directly in strains M137-11B and M137-11A by transformation and confirmed by showing lack of complementation with previously characterized rad mutants and genetic linkage between the MMS or UV sensitivity and the G418-resistant phenotypes as determined by tetrad analysis, as well as by PCR and Southern analyses (data not shown).

Construction of the chromosomal inverted repeat constructs his3^p::VST and his3^p::TER:
Three chromosomal inverted repeat systems based on identical 3.0-kb repeats have been used in this study. The first one, his3^p::INV, carrying the his35'-leu2-r and his3-kLEU2 inverted repeats, was described previously (AGUILERA and KLEIN 1989B ).

The his3^p::TER inverted repeat construct, identical to his3^p::INV but with the transcription terminator sequence CYC1ter upstream of the his35' copy, was made by transforming the yeast strain M137-11B.F with pAAX43-TER linearized with XbaI and selecting for Ura⁺ colonies able to form papillae on SC-his, to obtain strain M137-11B.ITE.

The his3^p::VST inverted repeat construct, identical to his3^p::INV but containing the his35'-k double mutant allele and the his3-h allele in the repeats (Fig 1), was made by transforming the strain M137-11BHlF with the 3.74-kb BamHI fragment of pBR-536L carrying the his3-h-LEU2 cassette and selecting for Leu⁺ His^- transformants. The new resulting strain, M137-11BhhL, was then transformed with pAAX43K linearized with XbaI and selected for Ura⁺ strains that formed papillae on SC-his, to obtain strain M137-11BVST.

View larger version (33K): In this window In a new window Download PPT slide   Figure 1. Inverted repeat recombination systems his3p::INV and his3p::VST used in this study. They are located at the HIS3 locus on chromosome XV and are made of a 3.0-kb inverted repeat unit (shown as a gray arrow) containing two convergent copies of different alleles of the HIS3 and LEU2 genes. At the bottom of each system, the phenotype of the different recombination products that can be recovered is shown. For each specified recombinant phenotype, the shortest possible extension of the conversion tract is represented as a continous black line and the largest one as a dashed gray line. The values in kilobases indicate the largest possible conversion tract of each type of recombinant. In the case of the his3p::VST system, the His+ Leu- recombinants arise by two independent conversion events, the maximum extensions of which are shown on the right as two independent values.

In all cases, proper gene replacement of each selected transformant was confirmed at the molecular level by PCR analysis of their genomic DNA using the same external primers as those used to determine crossing-over events (see later) and subsequent restriction analysis (see below).

Determination of recombination frequencies:
Recombination frequencies were determined as the median frequency of six independent colonies isolated from YEPD at 30° as described previously (see PRADO et al. 1997 ), including those of the rad55 and rad57 mutants. Median frequencies are the average of the two median values among the six obtained for each experiment.

Detection of inversion events:
To determine the percentage of gene conversion events that were or were not associated with the inversion of the repeat construct we selected independent His⁺ recombinants, whether Leu⁺ or Leu^-, on SC-his. From these recombinants we selected those that were also Leu^-. As can be seen in Fig 1, His⁺ Leu^- recombinants arise as a result of a long gene conversion event covering both the his3-k and the leu2-r sites. To establish whether such gene conversion events were associated with inversion, we performed PCR analysis of 36–80 independent His⁺ Leu^- recombinants using three oligos at the same time: co.A, co.B, and co.C. The oligos co.A and co.C hybridize outside the inverted repeat construct at 60 and 947 bp, respectively, from their proximal repeat ends, and co.B hybridizes with the intervening region located between the repeats at 195 bp from its proximal repeat end. We confirmed that by using single and double combinations of primers the expected bands for crossover and noncrossover events were amplified (data not shown). Each event leads to clearly distinct PCR fragments when the three primers are used (see Fig 4). Whereas noninversion events give a 4.1-kb fragment, amplified by the B and C primers, inversion events give a 3.2-kb fragment, amplified by the A and B primers. The validity of this PCR analysis to detect inversions was confirmed by Southern analysis of genomic DNA fragmented with SalI and probed with the 1.4-kb SalI-XbaI fragment located between the repeats as described previously (AGUILERA and KLEIN 1989B ; see Fig 4). Similar analysis was also performed with His⁺ Leu⁺ events arising by short gene conversion not covering the leu2-r site.

View larger version (37K): In this window In a new window Download PPT slide   Figure 2. Effect of the spt4-3 and spt6-140 mutations on recombination in the chromosomal inverted repeat system his3p::INV in wild type and rad mutants affected in DSB repair mediated by strand invasion. Frequencies of recombination correspond to the median value of two or three fluctuation tests, each one done with six independent colonies. Strains used were the isogenic series of RAD+ and rad- strains M137-11B (SPT6), M137-11A (spt6-140), and M236-12D (spt4-3) listed in Table 1.

View larger version (48K): In this window In a new window Download PPT slide   Figure 3. Effect of spt6-140 on recombination in wild type and rad mutants affected in DSB repair mediated by strand annealing. Frequencies of recombination shown correspond to the chromosomal inverted repeat system his3p::INV in the strains M137-11B (SPT6), M137-11A (spt6), Mr52-2D (rad52), Mr52-12A (rad52 spt6), Mr1-6D (rad1), Mr1-2C (rad1 spt6), Mr59-1B (rad59), Mr59-2B (rad59 spt6), Mr51r59-2D (rad51 rad59), and Mr51r59-1D (rad51 rad59 spt6) as listed in Table 1. Identical results were obtained with the alternative congenic strains Mr52-6A (rad52), Mr52-2B (rad52 spt6), Mr1-2A (rad1), Mr1-6C (rad1 spt6), Mr59-4A (rad59), Mr59-2C (rad59 spt6), Mr51r59-16D (rad51 rad59), and Mr51r59-2C (rad51 rad59 spt6) (data not shown). Other details as in Fig 2.

View larger version (40K): In this window In a new window Download PPT slide   Figure 4. Physical monitoring of inversions in the His+ recombinants obtained from strains carrying the inverted repeat systems of Fig 1. (A) Schematic representation of the inverted repeat system, in which the 3.0-kb repeat unit is shown as a wide gray arrow and the region located between the repeats is shown as a black arrow. The parental and inverted configurations are shown. The corresponding annealing sites for primers co.A (a), co.B (b), and co.C (c) (see Table 2) are shown as thin black arrows. The DNA fragments (4.1 and 3.2 kb) that can be amplified by PCR using the mix of these three primers are shown as thick black lines at the bottom of each configuration. The DNA fragments (23 and 10 kb) that can be detected by Southern analysis of genomic DNA digested with SalI are shown as thick gray lines. The DNA probe used in hybridization experiments is shown as a thick dashed line (P). (B) Comparative PCR and Southern analysis of 16 independent His+ recombinants from the his3p::INV system, showing that both physical assays yield identical results for parental and inverted products.

Manipulation and analysis of nucleic acids:
DNA and RNA manipulations were made following standard procedures (SANTOS-ROSA et al. 1996 ). For Northern analysis, overnight YEPD fresh cultures of each yeast strain were diluted in 10 ml of YEPD to an OD₆₀₀ of 0.1 and cultured for 7 hr to a final OD₆₀₀ of 0.6–0.8. RNA was extracted, run in agarose gel electrophoresis, and analyzed following previously published standard procedures (CHAVEZ and AGUILERA 1997 ). As [³²P]dCTP-labeled DNA probes we used different fragments of pBR322 and HIS3. For detection of rRNA as a measure of the amount of total RNA loaded, we used an rDNA fragment generated by PCR as described previously (CHAVEZ and AGUILERA 1997 ).

Chromatin analysis:
Chromatin analysis was performed from YEPD log-phase yeast cells as described (BERNARDI et al. 1991 ). Chromatin was digested with AvaII and hybridized with PCR-labeled probes. As probes we used 102 bp of the HIS3 5' end obtained with oligos his3.A and his3.B (Table 2) or 181 bp of pBR322 generated with oligos bre.A and bre.B (Table 2).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Hyper-recombination caused by spt6 and spt4 both in the absence and presence of RAD51-mediated strand exchange:
We have previously shown that spt4-3 and spt6-140 mutations differentially increase recombination. Hyper-recombination was observed in some DNA recombination assays but not in others (MALAGON and AGUILERA 1996 ). Using the his3^p::INV inverted repeat construct located on the HIS3 locus on chromosome XV, which contains 3.0-kb inverted repeats carrying two defective his3 alleles (Fig 1), we have shown that spt4 and spt6 increases the frequency of His⁺ recombinants four- and sixfold above wild-type levels, respectively (MALAGON and AGUILERA 1996 ; Fig 2).

To better understand the spt4 and spt6 hyper-recombination phenotypes we have analyzed their dependency on Rad51p. Fig 2 shows that rad51 reduces the level of recombination of both spt4 and spt6 to approximately RAD51 SPT6 wild-type levels. In parallel, recombination in the rad51 background was increased by spt4 and spt6 more than 15 and 18 times the rad51 SPT6 levels, respectively. This result indicates that the recombination events stimulated by spt4 and spt6 are very efficient in the absence of the strand-exchange protein Rad51p. Given the similarity of phenotypes of both mutations we decided to concentrate our efforts on the characterization of the hyper-recombination phenotype of spt6-140.

Genetic analysis of the dependency of the hyper-recombination phenotype of spt6-140 on Rad54p, Rad55p, and Rad57p revealed that the rad54, rad55, and rad57 mutations reduce the levels of recombination of spt6 mutants to approximately RAD SPT6 wild-type levels (Fig 2). In the rad54 background, recombination was increased 19-fold, whereas in rad55 and rad57 recombination was stimulated 3-fold. It is worth noting that rad55 and rad57 have the weakest effect on wild-type recombination at 30° as compared with rad51 and rad54, but the strongest effect on spt6-induced recombination. These results indicate that spt6-induced recombination proceeds very efficiently regardless of whether or not Rad51p, Rad54p, and, to a minor degree, Rad55p and Rad57p are present in the cell.

Recombination events stimulated by spt6-140 are fully dependent on RAD1 and RAD59:
The existence of a RAD52-dependent pathway of homologous recombination that is very efficient in the absence of Rad51p has been previously inferred (RATTRAY and SYMINGTON 1994 ; AGUILERA 1995 ; MALKOVA et al. 1996 ) and shown to be partially controlled by RAD59 (BAI and SYMINGTON 1996 ). In addition, it has been shown that RAD1 controls direct-repeat recombination occurring by SSA (FISHMAN-LOBELL and HABER 1992 ). This reaction presumably occurs in the absence of Rad51p-mediated strand exchange as mutations of either RAD51, RAD54, RAD55, or RAD57 lead to hyper-recombination between direct repeats (MCDONALD and ROTHSTEIN 1994 ; RATTRAY and SYMINGTON 1994 ; LIEFSHITZ et al. 1995 ). Consequently, we decided to determine whether the recombination events stimulated by spt6-140 were dependent on both RAD1 and RAD59. As can be seen in Fig 3, both rad1 and rad59 abolish the hyper-recombination phenotype of spt6. The levels of recombination of the rad1 spt6 and rad59 spt6 double mutants were the same as those of the single rad1 and rad59 mutants.

All events analyzed in this study are likely generated by homologous recombination, as they all strongly depend on RAD52 (Fig 3). In addition, the double mutant combination rad51 rad59 reduces the frequencies of His⁺ recombinants synergistically, consistent with the observation that each gene controls an independent recombination pathway (BAI and SYMINGTON 1996 ). Nevertheless, in this double mutant background spt6-140 was able to cause a 10-fold stimulation of recombination, suggesting that its effect on recombination is not restricted, in any case, to a particular recombination mechanism. Our data, indeed, are consistent with the previous observation that there is still an important amount of Rad52p-dependent events that may occur in the absence of both Rad51p and Rad59p (BAI and SYMINGTON 1996 ).

spt6-140 increases single gene conversions and coconversions associated with inversions:
The his3^p::INV system is based on two genes, HIS3 and LEU2. Recombinants are first selected as His⁺. Subsequently, we can determine whether they are Leu⁺ (the recombination event does not involve a gene conversion of the LEU2 copy) or Leu^- (the event involves conversion of the LEU2 copy toward the leu2-r allele; see Fig 1). Therefore, by analyzing independent His⁺ events we can determine the percentage of recombination events that have occurred by a gene conversion covering >1.2 kb, which is the distance between the his3-k and the leu2-r mutations (Fig 1).

Table 3 shows that whereas 32.4% of His⁺ recombinants are produced by a gene conversion event longer than 1.2 kb in wild-type strains, this value is significantly higher (P < 0.05) in the spt6-140 mutants (47.6%). This suggests that spt6 strongly stimulates coconversion events. Identical results were obtained with the spt4-3 mutation (data not shown), supporting the idea that spt6-140 and spt4-3 confer similar phenotypes of recombination. As can be seen in Table 3, the percentages of coconversion events (42–49%) among the total recombinants remaining in rad51, rad54, rad55, and rad57 mutants are significantly above the wild-type levels. This indicates that the recombination mechanism catalyzed by Rad51p, Rad54p, Rad55p, and Rad57p leads primarily to short conversion tracts. The double mutant combination of spt6 with any of the rad mutations showed a proportion of coconversion events identical to that in the wild-type strains (Table 3). This confirms that the effects of spt6 and the rad mutations are mutually suppressed by each other, in both the frequency and the type of recombination events affected. If we calculate the effect of spt6 on both single conversions (His⁺ Leu⁺) and coconversions (His⁺ Leu^-) by multiplying the frequency of recombination (Fig 2 and Fig 3) by the proportion of each type of event (Table 3) we find that spt6 increases both single gene conversions and coconversions to a similar degree in either wild-type or rad51 backgrounds (Table 4).

Another important feature defining the molecular nature of the recombination event stimulated by spt6 is its level of association with inversions. Since, as previously shown, coconversion events are preferentially associated with inversions as compared to single gene conversion events (AHN and LIVINGSTON 1986 ; AGUILERA and KLEIN 1989B ), it was likely that a major proportion of His⁺ recombinants has undergone an inversion in spt6 strains. Consequently, we determined the proportion of inversions among His⁺ Leu^- coconversion events in spt6 rad mutants.

The percentage of gene conversion events that are associated with inversions can be physically monitored by Southern analysis (Fig 4; AGUILERA and KLEIN 1989B ). In this study, however, we have devised an easier method for inversion detection based on PCR analysis using three DNA primers simultaneously (Fig 4; see MATERIALS AND METHODS). Using 16 randomly chosen independent His⁺ recombinants, we first demonstrated that the results obtained with both Southern and PCR analyses were identical (Fig 4).

Analysis of 36–80 independent recombination events in each case revealed that whereas in wild-type strains 44% of His⁺ Leu^- coconversion events were associated with inversions, this value went up to 67% in spt6 strains (Table 5). A similar pattern was observed when we studied the proportion of inversions in the His⁺ Leu⁺ single conversion events. These events show low proportion of inversions (14.5% in 76 independent events analyzed) in wild-type cells as previously reported (AGUILERA and KLEIN 1989B ). However, this value is significantly higher (P < 0.05) in spt6 cells (36.8% in 68 independent events analyzed). These results are consistent with most events stimulated by spt6 being coconversions associated with inversions and suggest that a great proportion of single His⁺ Leu⁺ recombinants are the result of conversion tracts close to 1.5 kb, maximum possible length for such events (see Fig 1). This preferential effect on long gene conversions associated with inversion is not observed in the rad spt6 double mutant background (Table 3 and Table 5).

To determine the effect of spt6-140 on well-defined short conversion tracts, we constructed the inverted repeat system his3^p::VST, identical to his3^p::INV but containing the his35'-k double mutant allele and the his3-h allele in the repeats (Fig 1). As shown in Fig 1, His⁺ recombinants can be obtained in the his3^p::VST system by a single gene conversion covering at most the 300-bp region downstream of the KpnI^- mutation. This is an important difference from the his3^p::INV system, in which single His⁺ conversions can cover regions as long as 1.5 kb (see Fig 1). We observed that the frequencies of His⁺ recombinants in wild-type cells in the his3^p::VST system are five times below the his3^p::INV system (3 x 10^-6), whereas the spt6 mutation has no stimulatory effect on His⁺ recombination events in the his3^p::VST system (4 x 10^-6). In this system we were not able to detect inversions associated with His⁺ events. This is consistent with the conclusion that only long conversion events are associated with inversions (AGUILERA and KLEIN 1989B ). On the other hand, the frequency of recombination in the his3^p::VST is reduced 10-fold by the rad51 mutation (3 x 10^-7) and increased 10-fold by spt6 in the rad51 background (3 x 10^-6). In the his3^p::VST system His⁺ Leu^- coconversions reflect discontinuous DNA repair events, which indeed are very low (<2%) in accordance with previously reported data for other recombination assays (JUDD and PETES 1988 ; SYMINGTON and PETES 1988 ; AGUILERA and KLEIN 1989A ). These results suggest that short gene conversion events are less affected by the spt6-140 mutation than are long events.

Chromatin structure of the his3^p::INV construct is altered in the spt6-140 mutants:
Since Spt6p interacts with the globular domain of histone H3, and chromatin is altered in spt6 mutants (BORTVIN and WINSTON 1996 ), we decided to determine whether an alteration of the chromatin structure of our repeat systems by the spt6-140 mutation could be associated with the hyper-recombination effect observed in the his3^p::INV system. Micrococcal nuclease sensitivity of this system was determined in wild-type and spt6 strains. As can be seen in Fig 5, no difference was detected between wild-type and spt6 strains at the HIS3 sequences. However, different MNase hypersensitive sites were found in wild-type and spt6 strains in the region downstream of HIS3, between the HIS3 and LEU2 ORFs. This result indicates that chromatin in this region, which covers the his3-k–leu2-r interval, is altered in the spt6-140 mutants. Therefore, it is likely that the hyper-rec phenotype of spt6 is due to an increase in the incidence of recombination caused by chromatin alteration at the his3-k–leu2-r interval.

View larger version (67K): In this window In a new window Download PPT slide   Figure 5. Chromatin analysis of the his3p::INV inverted repeat system in wild-type (M137-11B) and spt6-140 (M137-11A) strains as determined by treatment with increasing amounts of MNase (see MATERIALS AND METHODS). (A) MNase analysis of the 1.68-kb AvaII fragment from the repeat unit containing the his3-k allele. The DNA probe used was the 102-bp HIS3 fragment obtained by PCR with primers his3.A and his3.B in the presence of [-32P]dCTP. (B) MNase analysis of the 2.2-kb AvaII fragment from the repeat unit containing the his3-5' allele. The DNA probe used was the 181-bp pBR322 fragment obtained by PCR with primers bre.A and bre.B in the presence of [-32P]dCTP. As internal size markers DNA digested with AvaII (A), PstI (P), and/or HindIII (H) was used. Schemes of the DNA regions analyzed are shown.

Deletion of the histone genes HTA1-HTB1 confers hyper-recombination in rad51 cells:
If the alteration of chromatin structure in spt6-140 is a cause of hyper-recombination observed in rad51 cells, we wondered whether other mutations affecting chromatin structure, such as the hta1-htb1 (spt11-spt12), which removes one copy of the histones H2A and H2B genes (CLARK-ADAMS et al. 1988 ), also conferred hyper-recombination. We observed that whereas hta1-htb1 conferred wild-type recombination frequencies in a RAD⁺ background (1.4 x 10^-4), it increased recombination fourfold in a rad51 background with respect to single rad51 cells (5 x 10^-5 vs. 1.3 x 10^-6). This result is consistent with our conclusion that at least part of the hyper-recombination phenotype caused by spt6-140 in rad51 cells is linked to chromatin alteration.

Effect of spt6-140 on transcription of the his3 repeats and its relationship with spt6-induced recombination:
Since the spt6 mutations have been shown to have different effects on transcription (SWANSON et al. 1990 ), and transcription, on the other hand, has been reported to be an efficient stimulator of mitotic recombination (see AGUILERA et al. 2000 ), we decided to investigate whether the spt6-140 mutation affected transcription of the his3 repeats. Northern analysis of the his3^p::INV system using different segments of the his3 and pBR322 regions permitted us to detect the 0.7-kb mRNA expected for the his3-k copy and an unexpected and abundant 2.2-kb mRNA covering the his35' truncated copy (Fig 6). Using four different probes covering the upstream pBR322 sequence, we determined by Northern analysis that this 2.2-kb transcript initiates at the bacterial ori sequence and terminates at the transcription terminator of the his35' truncated copy (data not shown). Analysis of the abundance of both the 2.2- and 0.7-kb transcripts revealed that spt6 has no significant effect on the 0.7-kb mRNA whereas it reduces the levels of the 2.2-kb mRNA by approximately threefold (Fig 6). The rad51 mutation has no effect on these transcripts, regardless of whether the strain was SPT6 or spt6. Therefore, spt6 has a specific negative effect on the transcript covering the his3 copy that is acting as a donor of information in the His⁺ conversion events studied. Consequently, we decided to determine whether transcription may modulate the recombination phenotypes caused by spt6-140.

View larger version (46K): In this window In a new window Download PPT slide   Figure 6. Northern analysis of the his3p::INV system. The 2.2- and 0.7-kb RNA, corresponding to the his35' and his3-k alleles, respectively, are drawn as arrows in the upper diagram. The small open box from which the 2.2-kb RNA fragment initiates corresponds to the area of the pBR322 ori region. Total RNA was isolated from YEPD-grown midlog-phase cultures. To map the 2.2-kb RNA fragment we performed different hybridizations using as DNA probes the 366- and 244-bp DraI fragments of HIS3 and the 222-bp HindIII-SspI, 561-bp PstI-SspI, 1166-bp DraI-PvuII, and 641-bp AvaI-PvuII fragments of pBR322 (data not shown). The Northern shown here was made with the first of the probes mentioned and with a 32P-labeled 589-bp internal 28S rRNA fragment obtained by PCR. All RNA levels are given with respect to the rRNA levels in arbitrary units (AU) and normalized with respect to the 2.2-kb mRNA wild-type value (taken as 1). Strains used were M137-11B (WT), M137-11Br51k (rad51), M137-11A (spt6), and M137-11Ar51k (spt6 rad51).

To do so we constructed the his3^p::TER system, identical to his3^p::INV but with the CYCter transcriptional terminator just upstream of the his35' copy. The goal was to impede RNAPII from transcribing through the his35' allele involved in the recombination event. This was confirmed by Northern analysis (data not shown). Fig 7 shows that the wild-type frequency of His⁺ recombinants in this his3^p::TER system is similar to that in the his3^p::INV, implying that transcription elongation through the his35' copy does not lead to a stimulation of His⁺ recombinants in wild-type cells. The spt6 mutation does not change the frequency of recombination or the proportion of coconversion events with respect to wild-type cells. Therefore, in the absence of transcription through the his35' copy, spt6 does not increase recombination above wild-type levels. Consequently, the observed transcriptional defect of spt6-140 at the his35' may be related to the stimulation of recombination. Nevertheless, spt6 also stimulates recombination in the absence of transcription in a rad51 background (21 times the wild-type levels; Fig 7). Therefore, transcription seems to favor the Rad51p-mediated mechanism of recombination stimulated by spt6, but not recombination occurring in the absence of Rad51p. Consistently, rad59 does not reduce recombination in the absence of transcription, unless RAD51 is simultaneously knocked out (Fig 7).

View larger version (26K): In this window In a new window Download PPT slide   Figure 7. Effect of spt6-140 on recombination at the his3P::TER inverted repeat system. This system is identical to his3p::INV but contains the CYC1 transcription terminator (TER) that impedes transcription from going through the his35' allele as determined by Northern analysis (data not shown). The Ite series of congenic strains listed in Table 1 were used. Frequency of His+ recombinants and percentage of His+ Leu- coconversion events among total His+ events were calculated as described in Fig 2 and Table 3, respectively. Details of the inverted repeat diagram as in Fig 1.

We are not aware of any example in which a DNA sequence inserted in another DNA region changes chromatin positioning of distant sequences. Therefore, there is no reason to believe that the CYCter sequence changes the nucleosome positioning of the adjacent HIS3 sequences. Besides, changes are much more unlikely to be produced beyond the HIS3 sequences, where the differences in MNase sensitivity between wild-type and spt6 mutants were detected (Fig 5).

It is worth noting that rad51 reduces recombination 37-fold below wild-type levels in the his3^p::TER system (Fig 7), a stronger effect than that observed in the his3^p::INV system (10-fold, Fig 2) in which transcription is fully active. This may suggest that recombination of nontranscribed DNA may be less efficient in the absence of Rad51p. However, we have not been able to observe a similar effect in other recombination systems (S. GONZÁLEZ-BARRERA and A. AGUILERA, unpublished results).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have shown that spt6-140 and spt4-3 mutations stimulate recombination between inverted repeats by a Rad52p-dependent mechanism in both the absence and the presence of Rad51p, Rad54p, Rad55p, and Rad57p. The major mechanism of recombination stimulated by spt6-140 is dependent on Rad1p and Rad59p, with a large proportion of stimulated events associated with inversions. spt6 preferentially stimulates gene conversion events that may occur with or without Rad51p-mediated strand exchange and that we propose to be BIR. The hyper-recombination phenotype caused by spt6-140 may be related to the effect of spt6-140 on chromatin structure and is partially stimulated by transcription. We believe that such a chromatin alteration causes an increase in the incidence of recombination and favors Rad52p-mediated strand invasion, in particular, in the absence of Rad51p-promoted strand exchange.

The spt6-140 and spt4-3 mutations stimulate RAD1-RAD59-dependent recombination regardless of whether or not it is mediated by Rad51p:
Spt6p and Spt4p are proteins involved in chromatin structure and transcription elongation (SWANSON and WINSTON 1992 ; BORTVIN and WINSTON 1996 ; HARTZOG et al. 1998 ; WADA et al. 1998 ). We show here that recombination in the inverted repeat system his3^p::INV is stimulated 6-fold by the spt6-140 mutation. This increase is up to 19-fold in rad51 and rad54 backgrounds. As can be deduced from values of Fig 2, whereas 90% of wild-type recombination events are Rad51p dependent (100 - [1.3 x 10^-5/1.3 x 10^-4]), this ratio goes down to 71% in spt6 cells (100 - [2.4 x 10^-4/8.3 x 10^-4]). Therefore, spt6-140 has a stronger effect on recombination events occurring in the absence of Rad51p-mediated strand exchange. Similar results were observed in the spt4-3 mutants (Fig 2). Since both the spt6-140 and spt4-3 mutations have identical recombination phenotypes, it is unlikely that the effects of spt6-140 are allele specific but rather they are a consequence of either chromatin or transcription defects, which may be similar in spt4-3 cells.

It is worth noting that in our repeat system rad55 and rad57 have the weakest effect on recombination in SPT6 strains. This may be due to the fact that all our analyses have been done at 30° and these mutations have their strongest effects at lower temperatures (see RATTRAY and SYMINGTON 1995 ). It seems clear, however, that rad55 and rad57 reduces spt6-induced recombination more strongly than rad51 and rad54 (Fig 2). This implies that Rad55p and Rad57p have an important part in the events stimulated by spt6-140. This raises the possibility that the Rad55p/Rad57p heterodimer participates in some events occurring in the absence of Rad51p.

As expected from the previous work of BAI and SYMINGTON 1996 showing that the RAD51-independent recombination in chromosomes is strongly dependent on RAD59, rad59 abolishes the hyper-recombination phenotype of spt6-140 (Fig 3). Therefore, a major proportion of recombination events stimulated by spt6 is mediated by RAD59. Our results (Fig 3 and Table 5) as well as previously reported data (BAI and SYMINGTON 1996 ) suggest, in any case, that Rad52p-dependent recombination may also occur in the absence of both Rad51p and Rad59p.

Consistent with previous observations (AHN and LIVINGSTON 1986 ; AGUILERA and KLEIN 1989B ), long gene conversion events show the highest association with inversions (Table 5). Accordingly, a major fraction of spt6-stimulated recombination events are long gene conversions associated with inversions.

Hyper-recombination conferred by spt6-140 is also abolished in rad1 backgrounds. As both Rad1p and Rad59 have been shown to be required for SSA (FISHMAN-LOBELL and HABER 1992 ; BAI et al. 1999 ; SUGAWARA et al. 2000 ) it is likely that inversions in our repeat systems require SSA (see below).

Association of chromatin alteration and transcription impairment with spt6-induced recombination:
We have observed that the his3-k–leu2-r interval of the his3^p::INV system shows a different MNase sensitivity pattern between wild-type and spt6 cells (Fig 5). Such a chromatin difference was very clear in the lower repeat copy as depicted in Fig 1, which is the one acting as recipient of information in all His⁺ gene conversion events that we selected. Therefore, this chromatin alteration of the inverted repeats (Fig 5) may be responsible for the increase in the occurrence of recombinogenic substrates. A lower protection of DNA from nucleases, free radicals or other agents damaging DNA, or an increase in replication failures may be favored by this altered chromatin state, causing recombinogenic breaks.

In addition, we have shown that spt6 cells show reduced levels of transcription of one of the repeated sequences involved in the recombination event, in particular the his35' sequence (Fig 6). If a terminator is used to impede transcription from entering his35', the spt6 hyper-recombination phenotype is not observed in Rad⁺ strains. Since the his35' sequence is at the elongation part of transcription, the results suggest that transcription elongation is linked to the stimulation of recombination in spt6 cells. The importance of transcription-elongation impairment on the incidence of repeat recombination has been reported previously (CHAVEZ and AGUILERA 1997 ; PRADO et al. 1997 ; PIRUAT and AGUILERA 1998 ). The results of this study are consistent with the findings that Spt6p has a role in transcription elongation (HARTZOG et al. 1998 ). Given the effect of spt6-140 in chromatin structure (Fig 5), it is possible that the transcriptional dependency of the spt6 hyper-recombination phenotype is mediated by chromatin changes associated with transcription. Both the transcription and chromatin effects may be associated in spt6-140 mutants. In any case, the similar increase in recombination observed in spt6-140 rad51 vs. rad51 strains both with and without transcription (21-fold), suggests that transcription in spt6 cells only enhances Rad51p-mediated recombination.

Chromatin alteration may favor recombination in the absence of Rad51p-strand exchange:
The observation that spt6-induced recombination events are very abundant in rad51 cells suggests that recombination occurring in the absence of Rad51p is very efficient in spt6 mutants as compared to wild type. We do not know whether or not this is due to a poor ability of this reaction to proceed through wild-type chromatin. However, our results suggest that, in addition to initiation events, the spt6-induced chromatin alteration may facilitate a recombination reaction that otherwise would require Rad51p, Rad54p, Rad55p, and Rad57p. In this sense, SUGAWARA et al. 1995 , using plasmidic substrates based on silenced and transcriptionally active MAT sequences, showed a much stronger RAD51 dependency for recombination in silenced DNA. Our results would favor the idea that the Rad51p-mediated reaction has evolved to promote recombination in organized or closed chromatin. In the absence of Rad51p-mediated strand exchange, recombination reactions may not be efficient enough in a properly organized eukaryotic chromatin. In favor of the idea that an altered chromatin could facilitate events in which the Rad51p function is dispensable, we have indeed shown that deletion of one copy of the histones H2A-H2B genes shows a fourfold increase in recombination frequencies in a rad51 background. As discussed later, one possibility is that spt6-induced chromatin modification favors strand invasion.

spt6-Stimulated long gene conversions and inversions may occur by BIR:
It has been observed in different mitotic recombination assays that only long gene conversion tracts are preferentially associated with inversions (AHN and LIVINGSTON 1986 ; AGUILERA and KLEIN 1989B ). If inversions were the results of a crossing over, which requires the resolution of a Holliday junction, long gene conversions and inversions would be expected to be dependent on the strand-exchange protein Rad51p and associated proteins Rad54p, Rad55p, and Rad57p. Paradoxically, rad51, rad54, rad55, and rad57 have a weak effect on His⁺ Leu^- coconversions (Table 3 and Table 5; AGUILERA 1995 ) and inversions (RATTRAY and SYMINGTON 1995 ). The easiest way to explain these events in the absence of strand exchange and Holliday junction resolution is via recombination- or break-induced replication (ASAI et al. 1994 ; KOGOMA 1996 ; MALKOVA et al. 1996 ).

The involvement of BIR in the generation of inversions has also been proposed to explain DSB-induced recombination between plasmid-borne inverted repeats (BARTSCH et al. 2000 ; KANG and SYMINGTON 2000 ). In chromosomal inverted repeats, BIR may lead to the duplication of the full construct if replication is primed "inward" with respect to the repeat (Fig 8A) or to the duplication of one repeat if replication is primed "outward" (Fig 8B). Accordingly, two sets of direct repeats are generated on each side of the break, providing the substrate necessary for repair by SSA (Fig 8A). When DNA synthesis occurs inward it will generate inversions depending on the repeats used in SSA. However, when DNA synthesis occur outwards inversions are never formed (Fig 8B). This model offers an alternative explanation to published data showing that long conversion tracts are preferentially associated with inversions (AHN and LIVINGSTON 1986 ; AGUILERA and KLEIN 1989B ), without implying that long heteroduplexes are required for crossing over. The recombination events stimulated by spt6-140 fit the characteristics of conversion tract length and inversion association expected for BIR (Fig 2 and Fig 3; Table 3 and Table 4). Our genetic data showing that spt6-stimulated recombination between inverted repeats is strongly dependent on RAD1 and RAD59, two genes shown to be required for SSA (FISHMAN-LOBELL and HABER 1992 ; BAI et al. 1999 ; JABLONOVICH et al. 1999 ; SUGAWARA et al. 2000 ), support our conclusion that spt6-induced inverted repeat recombination requires SSA. Our results support the proposal that inversions may occur by BIR and SSA (BARTSCH et al. 2000 ; KANG and SYMINGTON 2000 ), extending its validity from plasmid-borne inverted repeats to chromosomal inverted repeats.

View larger version (10K): In this window In a new window Download PPT slide   Figure 8. Model to explain chromosomal inversions by intramolecular homologous recombination in the absence of Holliday junctions. The recombination event is initiated in the repeat region leading to BIR that can synthesize a long DNA region (marked in gray). The recombination outcome would depend on whether invasion and synthesis occur (A) inward (toward the intervening region flanked by the repeats) or (B) outward (away from the repeats). The event produces two sets of direct-repeat units flanking two free double-strand ends, equivalent to a DSB. Repair of the DSB by a resection type of mechanism such as SSA deletes part of the resynthesized DNA restoring the inverted repeat construct in either the inverted or noninverted configuration. This depends on whether the event is inward (A), in which case inversions or no inversions are formed when SSA occurs between the repeats marked as 1 or 2, respectively, or outward (B), in which case inversions are never formed. All lines drawn represent DNA duplexes.

We do not favor unequal sister-chromatid gene conversion (ROTHSTEIN et al. 1987 ; CHEN and JINKS-ROBERTSON 1998 ) as a major mechanism yielding spt6-induced His⁺ Leu^- inversions in our system for two reasons. First, strand exchange is believed to be an essential step in postreplicative repair of DNA breaks, and, to our knowledge, there is no reported data about sister-chromatid exchange occurring in the absence of Rad51p. Second, His⁺ Leu^- events demand in such a model an unequal gene conversion covering a full repeat on one side (to yield a HIS3 leu2-r repeat) but a short DNA fragment on the other side (to maintain a leu2-r mutation). Such asymmetric events are not abundantly observed (CHEN and JINKS-ROBERTSON 1998 ) and cannot easily yield simple His⁺ Leu^- events if they initiate at the his3-k–leu2-r interval in our inverted repeat systems.

All recombination events we detect are RAD52 dependent. Rad52p is able to anneal two complementary ssDNAs in cooperation with replication factor RPA (MORTENSEN et al. 1996 ; SUGIYAMA et al. 1997 ; SHINOHARA et al. 1998 ). Rad52p, therefore, may catalyze the annealing step required for strand invasion. It has been suggested to be a functional homolog of RecT and Redß (KANG and SYMINGTON 2000 ). Strand invasion could occur in concert with strand exchange catalyzed by Rad51p (SHINOHARA et al. 1992 ; SUNG 1994 , SUNG 1997A ; GASIOR et al. 1998 ) together with Rad54p and the Rad55p-Rad57p heterodimer, as determined in vitro (SUNG 1997B ; BENSON et al. 1998 ; PETUKHOVA et al. 1998 ; SHINOHARA and OGAWA 1998 ; MAZIN et al. 2000 ). Interestingly, it has recently been shown that Rad59p also catalyzes strand annealing between complementary ssDNAs (PETUKHOVA et al. 1999 ). We do not know whether Rad59p necessarily defines a recombination pathway. It might promote, together with Rad52p, strand invasion in all recombination events whether or not Rad51p-mediated strand exchange is involved. Rad51p would catalyze the formation of stable heteroduplexes, presumably essential for recombination mechanisms involving Holliday junction formation, but dispensable in one-ended invasion mechanisms such as BIR. In the absence of Rad51p, however, Rad59p might become important for strand invasion, favoring the stabilization of an initial and very short heteroduplex, a reaction that could be facilitated by spt6-140.

It would certainly be interesting to know whether Rad59p acts cooperatively with Rad52p during strand invasion to stabilize a short heteroduplex that would prime DNA synthesis required for BIR (PETUKHOVA et al. 1999 ). Yet, we cannot exclude the additional possibility that the RAD59 dependency of the spt6-induced gene conversions and inversions observed in this study are due to the recently reported involvement of Rad59p in SSA (BAI et al. 1999 ; JABLONOVICH et al. 1999 ; SUGAWARA et al. 2000 ).

In summary, we favor the idea that the chromatin and transcription alterations caused by spt6-140 lead, in addition to a stimulation of initiation events, to a higher efficiency of Rad52p-mediated strand invasion reaction. This reaction may only be efficient in wild-type chromatin