Copyright 0 1995 by the Genetics Society of America
Recombination of Ty Elements in Yeast Can Be Induced by a Double-Strand Break Anat Parket, Ori Inbar and Martin Kupiec Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Ramat Aviv 69978, Israel Manuscript submitted October 20, 1994 Accepted for publication January 27, 1995 ABSTRACT The Ty retrotransposons are the main family ofdispersed repeated sequences in the yeast Saccharomyces cerevisiae. These elements are flanked by a pair of long terminal direct repeats (LTRs). Previous experiments have shown that Ty elements recombine at low frequencies, despite the fact that they are present in 30 copies per genome. This frequency is not highly increasedby treatments that cause DNA damage, such as UV irradiation. In this study, we show that it is possible to increase the recombination level of a genetically markedTy by creating a double-strand break in it. This breakis repaired by two competing mechanisms: one of them leaves a single LTR in place of the Ty, and the other is a gene conversion event in which the marked Ty is replaced by an ectopically locatedone. In a strain in which the marked Ty has only one LTR, the double-strand break is repaired by conversion. We have also measured the efficiency of repair and monitored the progression of the cells through the cell-cycle. We found that in the presence of a double-strand break in the marked Ty, a proportion of the cells is unable to resume growth.
and PETES1988a,b;PARKETand KUPIEC1992). All these HE Ty retrotransposons constitute the main family results seem to indicate that ectopic recombination beof dispersed repeated sequences in the yeast Sactween Tys may follow different, special rules. charomyces cerevisiue. They are 6-kblong elements, are In this study, we investigate the consequences of crepresent in 30-40 copies per haploid genome and comating a defined damage (a double-strand break, DSB) prise about 1-2% of the yeast genome (for a review, in a marked Ty element. The DSB is created in vivo see BOEKEand SANDMAYER 1991). Repeated sequences by the yeast HO site-specific endonuclease, which is can be found in the genomes of all eukaryotic organnormally involved in mating-type switching (STRATHERN isms and represent a potential source for karyotypic et al. 1982; KOSTRIKENet al. 1983; KLAR et al. 1984; KOinstability: reciprocal recombination between homoloSTRIKEN and HEFFRON 1984). This enzyme introduces gous sequences located at nonhomologous positions a DSB at a specific sequence present at the MAT locus, in thegenome(ectopicrecombination) can lead to chromosomal aberrations and cell death (JINKS-ROB- thus initiating a conversion event with homologous seERTSON and PETES 1986; LICHTEN et al. 1987; LICHTEN quences foundin the silent cassettes HML or HMR (for reviews see HABER 1992; HERSKOWITZ et al. 1992). The and HABER 1989). The Ty family of yeast may serve as HO cut site (HOcs) recognition sequence is also present a good model system to study the interactions between at the silent loci; at these locations, however, the senaturally occurring repeatedsequences and the mechaquence is protected from cleavage by a specific chromanisms that prevent karyotypic instability. Previousstudtin configuration (KLAR et al. 1981; NASMYI-H 1982). ies using genetically marked Ty elements have shown The HO endonuclease has been previously used to that despite their high copy number, ectopic gene congenerate DSBs at different genomic sites. When a DSB version (nonreciprocalrecombination; PETES et al. is created between directly repeated sequences, it is 1991) between Tys occurs at relatively low rates (lop6efficiently repaired, mainly by a nonconservative mechlO”/locus/generation). No reciprocal recombination anism that deletes the intervening sequences and one could bedetected associated with these conversion of the repeats (NICKOLOFF et al. 1986, 1989;ROTHSTEIN events (KUPIECand PETES1988a,b), in contrast to results obtained with other sequences (JINKS-ROBERTSON et al. 1987; RAY et al. 1988; RUDINet al. 1989; OZENBERG and ROEDER1991;FISHMAN-LOBELL et al. 1992; FISHand PETES 1986;LICHTEN et al. 1987; LICHTEN and f i MAN-LOBELL and HABER 1992; PLESSIS et al. 1992; SuBER 1989). In addition, the rate of ectopic recombinaGAWARA and HABER 1992). This mechanism has been tion of Tys isnot highly induced by either DNA damage shown to involve single-strand DNA degradation and it (PARKET and KUPIEC1992) or entry to meiosis (KUPIEC is probably similar to the single-strand annealing mechanism (SSA) described in other systems (LINet al. 1990; Corresponding author: Martin Kupiec, Department of Genetics, SKMARYON and CARROLL1991;FISHMAN-LOBELL et al. 50, University of Washington, Seattle, WA 98195. E-mail:
[email protected] 1992; FISHMAN-LOBELL and HABER 1992: SUGAWARA and Genetics 140: 67-77 (May, 1995)
A.
Parket, 0. Inbar and M. Kupiec
68
1992). In addition, the HO endonuclease has been shown to induce allelic (KOLODKIN et al. 1986; MCGILLet al. 1993) and ectopic (RAY et al. 1989; FAIRHEAD and DUJON1993) recombination. A DSB that cannot be repaireddue to lack ofhomology in the genome (KLAR et al. 1984; UVEH et al. 1989; BENNETT et al. 1993; FAIRHEAD and DUJON1993) or defects in the recombination machinery (e.g., rad52 cells; MALONE and ESPOSITO 1980; WEIFFENBACH and HABER 1985; RAVEH et al. 1989; WHITEand HABER 1990; SUCAWARA and HABER 1992) will lead to cell death. Because Ty recombination isusuallylow and does not seem to respond to treatments that inducerecombination of other sequences, we tested whether a DSB in a Ty is able to induce Ty recombination. We show that a DSB in a marked Ty element is efficiently repaired by recombination. Two different recombinational products are obtained. These products seem to be theresult of two competing mechanisms, one that requires interactions between the Ty long terminal repeats, and one that requires ectopic conversion between Tys. HABER
MATERIALSANDMETHODS Media, growth, and general procedures: Yeast cells were grown at 32” in either YPD (1%yeast extract, 2% Bacto peptone, 2% dextrose), WGly (1%yeast extract, 2% Bacto peptone, 3% v/v glycerol), YPGal (1% yeast extract, 2% Bacto peptone, 2% galactose), SD (0.67% yeast nitrogen base, 2% dextrose, with the appropriatenutrients added),or SGal (0.67% yeast nitrogen base, 2% galactose, with the appropriate nutrientsadded; SHERMAN et al. 1986). BactoAgar (1.8%) was added for solid media. CAN medium is SDcomplete without arginine, plus 40 pg of canavanine sulfate (Sigma) per milliliter. The adenine concentration in CAN plates was kept low (12 mg/liter) to allow proper red pigmentation of Ade- colonies. 5FOA medium is SDcomplete with 50 mg of uracil and 0.85 g of 5-fluoro-orotic-acid per liter (BOEKE et al. 1984). Standard molecular biology procedures such as cloning, restriction enzyme analysis and Southern blot analysis were done as described by SAMBROOK et a1 . (1989).Yeast molecular biology procedures were done as previously described (PARKET and KUPIEC1992; SILBERMAN and KUPIEC1994). Plasmids: pAP7 was created by digestion of plasmid pM77 (PARKET and KUPIEC 1992) withHpaI and ligation (pM77 carries a lys2:: TylSup allele). The resulting plasmid contains a TylSup with a deletion of the first 815 bp of the Ty (BOEKE et al. 1988), and of 1.7 kb of the flanking LYS2 sequences, in a URA3marked integrative vector. PAPS was constructed by inserting, into the unique Asp718 site of the TylSup (position 3505; BOEKE et al. 1988) in pAP7, a 130-bpHincIl fragment carrying the HO cut site (HOcs) and from plasmid pRKl13 (KOSTRIKENet al. 1983; KOSTRIKEN HEFFRON 1984). The Tyl doubly marked with SUP44 and with the HOcs is called TylSupH. In plasmid pGAL, the 29-bpEcoRI-Hind11 fragment of YCp50 was replaced by a 365-bp fragment carrying the GAL10 promoter. pGALHO is similar to pGAL, except that a 2592bp HindII fragment carrying the HO gene was cloned at the HindII site, creating a HO gene that is regulated by the GAL promoter (HERSKOWITZ and JENSEN 1991). pJH124 (a gift of J. HABER)carries a deletion of the MAT locus replaced by the insertion of a LEU2 marker (RUDIN and HABER 1988).
pRE194 (MELAMEDet al. 1992), containing LYS2sequences, was used as a probe in Southern blot analysis. Yeast strains: All strains in this study were isogenic and derived from AP12. This strain was created in several steps. First, an Ade- CanR derivative of APl (PARKET and KUPIEC 1992; MATa &Z-1 canl-100 ura3-52 lys2:: Tyl h2-3,112 his3&1200 trpldel901) was transformed to Ura+ with plasmid PAPS digested with Hpd. This plasmid carries a TylSupH flanked byLYS2 sequences. Most transformants had a copy of the plasmid integrated at the lys2:: Tyl locus after gap fill in, creating a duplication. Then Ura- segregants were isolated on 5FOA plates. Among the Ura- colonies, those that were Ade+ Can’ were picked and analyzed by Southern blot. They all contained a single copy ofthe complete TylSupH at LYS2. Because these strains were not fully Gal+,one such colony was crossed to W303-1A (MATa a&-21 canl-100 ura3-1 h 2 - 3 , 112 his3-11,15, trpl-1 (RUDINand HABER 1988), and sporulated. A Gal+ haploid, AP12 (MATa ade2-1 canl-100 ura352 lys2:: TylSupHh2-3,112 his3 trpl-l), was chosen among the spores. AP12was transformed with a fragment from plasmid pJH124 (MAT::LEU2, RUDINand HABER 1988) to delete the MAT locus. A Leu+ transformant that showed an “a-like’’ mating phenotype was chosen and called AP34. This strain was transformed with pGALHO to give strain AP35 and with pGAL to give strain AP38. Strain AP36 was created in two steps. An Ade- CanRcolony derived from AP34, that had the TylSupH replaced by a solo LTR was transformed to Ura’ with plasmid pAP8 linearized at the Sac11 site in LYS2. Then Ura- colonies were selected on 5FOA plates; these contained a TylSupH lacking the 5’ LTR, integrated at the LYS2 locus. One such Ura- Ade+ CanR colony was picked and called AP36. This strain was transformed with pGALHO to give AP37 and with pGAL to give AP39. AP35R and AP37R were isolated as red CanRderivatives of AP35 and AP37, respectively. All the relevant chromosomal configurations during the strain construction were confirmed by Southern blot analysis using LYS2specificand SUP4-specific probes (PARKET and KUPIEC 1992; SILBERMAN and KUPIEC1994). PCR All the PCR reactions were carried out in a Minicycler apparatus (MJ Research) starting with a small lump of cells picked with a toothpick from a colony. The cells were transferred to a tube carrying the reaction mix consisting of: 50 mM dNTPs, 200 ng of primers and 1X buffer as recommended by the manufacturer, in a total volume of 50 pl. After celllysis (3 min incubation at 97”),0.3 units of Taq DNA polymerase (Appligene) were added, and the cells were subjected to 30 cycles of: 60 sec at 94”, 60 sec at 54” and 90 sec at 72”. The products were run in 1.2% agarose gels. The following primers were used: M o l , 5’CCAGCGGAATTCCACTTGS’; this sequence overlaps the EcoRI site in the LYS2 gene. M 0 2 , 5’AACTGAGGGGTCClTTCC3’;thissequence is near the BglII site ofLYS2 and in the opposite orientation. M06, 5’GTGATGACAAAACCTClTCCG3’;this sequence is internal to Tyl, at positions 5502-5520 (BOEKE et al. 1988). Under the conditions used, M 0 1 and M02 give a 989-bp fragment when a solo LTR is present, but do not give any band when a whole Ty is present (a 7-kb fragment wouldhave been expected). M 0 1 and M06 give a 876-bp band diagnostic of a Ty element. All the colonies that proved positive for the 876bp band were subjected to Southern analysis (using a LYS2 probe) to distinguish between those that still have the TylSupH, and those in which the marked Ty has been replaced by an unmarked one. Induction of recombination: A single colony was used to inoculate a 5-ml SD-Ura starter. After growth overnight, the starter was used to inoculate a WGly culture, until it reached
Break-Induced Ty Recombination midlogarithmicphase.Thecellswerethencentrifuged, washed once with water, and divided: half of the cells were resuspended in WD, and half in WGal (to).The cells were further incubated, and samples were taken at intervals. For each timepoint, aliquotswere centrifuged and the pellet frozenat -70" for DNA extraction and Southern analysis.In addition, samples were plated, after appropriate dilutions, on WD or CAN plates. Colonies were counted after 3 days. Southern blots were measured witha Fujix BAS1000 phosphorimager. Differences in loading were corrected by dividing the intensity of each band by that of a 4.0-kb band that appears in all the blots (to the right of the Tys in Figures 2B and 4D, not shown). The relative intensitywas then multiplied by the ratio of homology to the probe used. Spontaneous ratesof recombination were measured as described before (PARKETand KUPIEC1992). Cell survival experiments: Six individual colonies each of AP35, AP37, AP38and AP39 were picked from SD-Ura plates was approand dissolved in 1 ml of sterile water. Each colony SD-Ura and fourSGal-Ura priately diluted and plated on four plates. Colonies were counted after 3 and 4 days, respectively. was calculated by dividing the mean The ratio for each culture number of colony-forming units on SGal-Ura, by the mean number of colony-forming units on SD-Ura. The mean of the six ratios thus obtained is presented. The statistical significance was determined by a one-tailed Wilcoxon two-sample test (SOU and ROLF 1981). Cellcycleparameters: During the induction experiment aliquots were taken into a solution of 4% formaldehyde and 0.1 M NaCl. After sonication(10 sec in a Microson sonicator), the cells were counted under the microscope and classified as G1 cells (cells withoutbuds), S cells (bud smaller than twothirds the sizeofthe mother cell) or G2 cells (bud bigger than two-thirds the size of the mother cell). About 300 cells were counted for each timepoint. RESULTS Previous experiments in our laboratory have shown thatthe level of ectopicrecombination involving a marked Ty (TylSup) located at theLYS2 locus on chromosome IZwas not increased by W irradiation or methylmethanesulfonate (MMS) treatment(PARKETand KUPIEC1992). The level of ectopic conversion of the LYS2 locus itself (where the TylSup was located) was induced more than two orders of magnitude in the same cell population, implying that the cells were proficient in the induction of ectopic recombination of non-Ty sequences. These results showed that Tys behaved differently from other DNA sequences in thecell with respect to ectopic recombination. Because the quality and location of the lesions created by UV rays or MMS could not be controlled, we have constructed a system in which a single, specific lesion (a DSB) could be produced in a marked Ty element. Strain AP34 carries a Tyl doubly marked by the and KUPIEC insertion of a SUp4-0 gene (TylSup; PARKET 1992) and a 130-bp fragment containing the recognition site for the yeast HO double-stranded endonucleand H E ~ R O1984; N NICKOLOFF ase ( H O q KOSTRIKEN et al. 1986), inserted80 bp upstream of the SUP4 marker of the TylSup. The resulting doubly marked Ty element is called TylSupH. The HO enzyme is normally involved in mating-type switching; it recognizes and cuts a se-
69
quence present at the MAT locus on chromosome IZZ. In AP34 the whole MAT locus was deleted, so that it will not be affected by expression of the HO gene (HABER 1992; HERSKOWITZ et al. 1992). The strain also carries the ochre-suppressible alleles ade2-1 and cad-100. The presence of the ochre suppressor confers an Ade+ CanS phenotype to the cells (prototrophic for adenine and sensitive to the drug canavanine; see Figure 1).A gene conversion event involving the TylSupH as a recipient, or recombination between the Tyl terminal repeats will render the cells Ade- CanR.These events can be selected for on canavanine-containing plates. Ade- cells accumulate a red pigment and can be easily distinguished from Ade+ (white) cells (PARKET and KUPIEC1992). We next introduced into AP34 the plasmid pGALHO, a URA3 marked centromeric plasmid carrying the HO gene under the control of the GALlO promoter, thus creating strain AP35. When AP35cells are grown in glucosecontaining medium, the HO gene is repressed, but it is expressed in cells grown on galactose-containing medium (Figure 1). Because severalexplanations arepossible for the lack of induction of Ty recombination after DNA damage (including, for example, a special chromatin configuration that precludes damage or the inability of the cells to repair damage in the context of a Ty), we wanted to know whether the endonuclease would be able to cut at thesite in the Ty, or whether the HOcs would be protected (as in the HML and HMR loci; KLAR et al. 1981; NASWH 1982) by some property of the Ty chromatin. If a DSB was introduced in theTy, would it lead to cell death, or would it be repaired? Finally, if the break was repaired, we wanted to determine themechanism of repair. DSB-iiduced T y recombination: AP35 cellswere grown on WGly to midlogarithmic phase. When grown on glycerol, the GALlO promoter is neither repressed nor induced (NICKOLOFF et al. 1989). The culture was then divided into two parts: half was resuspended in WD (glucose-containing medium) and half was resuspended in WGal (galactose containing medium). The cells were incubated further, and at appropriate time intervals samples were plated ontoWD plates. The basal level of recombination after growth in glycerolcontaining medium was relatively high (-0.5%), due probably to leakage of the GALlO promoter; this allowed us to follow the kinetics of TylSupH recombination by scoring the appearence of red colonies (see below). Red colonies can also be created by mutations in other genes of the adeninebiosynthetic pathway, but their frequency is several orders of magnitude lower thantheones observed. Samples werealso taken at different times and subjected to Southern blotanalysis. The results of such an experimentcan be seen in Figure 2. Thirty minutes after transfer to "Gal, a 5.9-kb band appeared and started to accumulate; this band is the result of the DSB created by the HO endonuclease (Fig-
A. Parket, 0. Inbar and M. Kupiec
70
HoaSUP4-0
+ 1 5 7 D
ne
" ade2- 1
canl-100 \
/
Ade+CanS (White colony) 4Growth ongalactose
-DSB
I
adelcanl-100
a
FIGURE I.-Schematic representation of the system used inthisstudy. The Tyis depicted as a rectangle flanked by two triangles (the LTRs). The SUP4-0 and HOcs inserts are indicated. The TylSupHis located at the LYS2 locus (gray rectangle). The SUP4-o is able to suppress the canl-100 and ade2-1 mutations, giving a Ade'CanS phenotype (white colonies). The CALHO construct on the plasmid is expressed on galactose-containing medium, producing HO enzyme (circles) that create a DSB at the HOcs in the Ty. Repair by recombination creates either unmarked Tys or a solo LTR in place of the TylSupH. Cells carrying these configurations at LYS2 produce AdeCan' (red) colonies.
a - I
1
aa a
Repairbyrecombination
/\ Ectopicconversion with unmarked Ty
LTR recombination
1 Plate cells sdo LTR
Adc Can r (Redcolony)
Ads Canr (Redcolony)
ure 2B). At -4 hr the DSB-produced band started to decrease in intensity and two new bands (5.0 and 8.1 kb) appeared. These bands increased in intensity; at -8 hr their rateof accummulation decreased, reaching almost a plateau. These bands representrecombination products oftwo sorts: the 5.0-kb band ("80% of the final products) is caused by the replacement of the TylSupH by a solo LTR the 8.1-kb band (20% of the recombinant products) results from the replacement of the whole TylSupH by an unmarked Ty element. Several proposed mechanisms of homologous recombination betweenLTRs can produce the first product (solo LTR). These include intrachromatidal crossing overbetween LTRs (popout), unequal crossing over between sister chromatids, conversion of the whole Ty by a single LTR donor (ROTHSTEIN et al. 1987) and single-strand annealing (SSA). In SSA, single strand degradation of both strands occurs until homology is reached at the LTRs, followedby annealing of the complementary single-stranded DNAs (LINet al. 1990; MAR-
and CARROLL 1991; FISHMAN-LOBELL et al. 1992; FISHMAN-LOBELL and WER 1992). The second product is the result of an ectopic gene conversion event. This band was first clearly seen 5 hr after transfer to YPGal, an hour after the solo LTR band could be detected. We believe that the apparent delay is due to its low abundance: if the 8.1-kb band represents only 20% of the molecules, its signal wouldbe under background levels at 4 hr. In addition, it shares less homology to theprobethanthe 5.0-kb band. Although reaching different levels, the solo LTR and the conversion final products showed similar kinetics, and arrived to their final levelsat -8 hr. No DSB or recombination product could be seen inglucose-growncells, indicating that the process is dependent on HO activity (Figure 2A). The genetic data shown in Figure 3A are completely consistent with the molecular analysis. Starting SO min after transfer to galactose, there was a sharp increase in the number of red colonies, reaching 90% of the population in the sample plated 7 hr after transfer to YON
Break-Induced Ty Recombination
71
A
AP35 1
80 60
-DsB -LTR
40 20
-CALIBRATION 0 0
B
5
10
Tlme
15
20
15
20
(hours)
AP37
+ -
-+ 2.5 +
20
Conversion
8.1 A
H
.ii 40
0 0
5
10
Time (hours)
H
LTR
5.0
1.o
FIGURE 3.-Kinetics of commitment to recombination. (A) Strain AP35 (TylSupH/pGALHO). The percent of sectored colonies (0)and total red + sectored colonies (m)after growth in YPGal, as well as that of total red and sectored colonies after growth in YPD ( 0 )are shown. (B) Strain AP37 [TylSupH(ALTR)/pGALHO]. Symbols as in A.
0.8 0.6
0.4
0.2
0.0
0
10
5
The
15
20
(hours)
FIGURE 2.-Molecular analysis of DSBinduced Ty recombination. (A) Southern blot analysis of an induction experiment withAP35 (TylSupH/pGALHO). Glu, cellsgrown on YPD [only times 0 and overnight (ON) shown]; Gal, cells grownon "Gal. At different timepoints (in hours) DNA was extracted and digested with HpuI. LYS2 sequences (pRE139; MELAMED et ul. 1992) were used as a probe. The 5.9-kb band indicative of a DSB, as well as the final gene conversion ( G C ) and solo LTR (LTR) products are shown. The band running above the DSB band is of plasmid origin. (B) Schematic representation of the different bands observed in the Southern blot analysis. The region of LYS2 covered by the probe is shaded. H, HpuI. ( C ) Kinetics of DSB creation and repair in AP35. The intensities of the different bands seen in A normalized for amount of DNA loaded (by dividing the intensity of each band to that of a 4kb band common to all lanes), and normalized for homology to the probe, are plotted as relative intensity per lane. 0, parental TylSupH, a, DSB-created band; 0, gene conversion product; @, solo LTR product.
YPGal. No increase in the frequency of red colonies over the basal level was seen in the culture grown in glucose. No loss in viability, measured as colony-forming units, was seen in either culture, althoughit should
be noted that thecells continued to grow, so that death of some proportion of the population could have been compensated by growth of another. Theglucose-grown cells continued to grow logarithmically, whereas the cells transferred to galactosecontaining medium grew more slowly (data not shown). During the first timepoints sectored white/red colonies appeared on YPD plates (Figure 3A). Their level increased at first, reaching -35% of the population after 3 hr, and then it diminished. By 7 hr, only a basal level ofsectors was seen, and remained unchangeduntil the end of the experiment. Full red colonies accumulated continuously throughouttheexperiment.The level ofred and red-sectored colonies reached aplateau at 7 hr (Figure 3A). The sectored colonies could result from recombination events involving one chromatid only (events initiated after DNA synthesis) or resolution of heteroduplex DNA intermediates by replication; part of these events were probably initiated in the liquid culture and completed on the YFD ' plate. Approximately 10% of the colonies obtained after overnight growth in galactose were still white. Analysis of these white colonies showed that most ( 7 4 % of the total colonies) were derived from cells that had lost the HOcarrying plasmid (because no selection was imposed for its presence). The remaining 2% of the cells still carried the plasmid. These are eithercells in whichthe HO enzyme was not expressed at a high enough level, cells carrying a TySupH in whichthe HOB was changed
72
A. Parket, 0. Inbar and M. Kupiec
TABLE 1 Distribution of recombmation products ~~
Time induction after
(h)
Red colonies (%)
LTR" Conversion* Mutation" Total ~~
1.9
AP35 (TylSupH/pGALHO) after HO induction 0 52 0.5 1 3 20 AP35 grown in YPD (no induction) 20 12(80) AP34 (spontaneous) Rate: 2 0.4 X
0.5 26 56
(80)
14(78) 15(75) 16(84) 63(85)
0.5 55(81)
13(20) 5(25) 3(16) 93ll(15)
0 0 0 0 0
65 18 20 19
3(20)
0
15
2(3)
68
4
~
ll(16)
)
74
Colonies in which the TylSup was replacedby a solo LTR numbers in parenthesis are the percentageof colonies analyzed. Colonies in which the TylSup was replaced by an unmarked Ty. could also be Colonies in which Southern blot analysis did not detect changes in the TylSup; these events created by ectopic conversion by the chromosomal sup4 locus. and was no longer recognized by the enzyme, or cells carrying TylSupH in which the HOcs was converted but the S W 4 remained.This last category represented -0.25% of the colonies analyzed, implying that conversion tracts shorter than 80 bp (the distance between the DSB and the HOcs) are very rare. We conclude that in this experiment 2 9 7 % of the cells in which the HO enzyme was expressed went through a recombination event at the TylSupH. Individual Ade- CanR (red) colonies from different timepoints were subjected to PCR or Southern blot analysis to determine the type of event responsible for their appearence. Table 1 shows that 80% of the colonies analyzed were due to events in which a solo LTR replaced the whole TylSup and 20% were conversion events. This ratio remained constantfrom to, where only 0.5% of the population was recombinant up to the end of the experiment,where virtually every cell underwent a recombination event. These results are consistent with the level of final products seen in Southern blots, implying that thesurvival ofthe different types ofrecombinants is similar. Analysis of recombinants obtainedafter overnight incubation inYPD, or spontaneously in AP34 (no HO plasmid), gave a similar ratio of LTR recombination to ectopic conversion (Table l ) , even though the spontaneous rate of TylSupH recombination was much lower(1.9 X lO-'/cell/generation). This may suggest that spontaneous and induced events are produced by a similar recombinational mechanism(s). We conclude thatthe HO endonuclease can efficiently recognize the HOcs when embedded in a Ty element. The enzyme is able to create a DSB in virtually all of the cells, and these DSBs are repaired by means of homologous recombination. Recombination of a Ty carrying a single LTR sequence: Most of the events induced by the HO enzyme (80%) were replacements of the marked Tyby a solo
LTR. To test the relationship between LTR recombination and Ty conversion, we constructed thestrain AP36. This yeast strain is identical to AP34, except that one of the LTRs of the TylSupH has been deleted, so that Ade- CanR (red) colonies can only be obtained by ectopic gene conversion with other Tys (or by very rare mutations in SW4-0). This strain was transformed with pGALHO, thus creating strain AP37, and the experiment was repeated in the same conditions as described above for AP35. The kinetics of appearance of recombinant colonies in both strains is indistinguishable (Figure 3B). Southern analysis of samples taken at different times after induction shows that the DSB appears and accumulates in AP37 withsimilar kinetics to those seen in AP35 (Figure 4). Only one final product is seen, consistent in size with a replacement of the TylSupH by an unmarked Ty element. This band of 8.1 kb starts to appear at 4 hr after transfer to "Gal and accumulates with kinetics identical to those of the final products in AP35, reaching a level comparable with that of the LTR product in that strain (Figure 4C). Southern blot analysis of 37 independent colonies obtained after induction confirmed that allof them were the result of ectopic conversion events (data not shown). As with AP35, no decrease in viability could be seen under the conditionsof this experiment, in which the cells could resume growth after repair. A small percent of cells remained white after HO induction. As with AP35, most had lost the pGALHO plasmid. To get a better sample of colonies that still carried the plasmid, but were white, we plated AP37 cells on SGal-Ura plates. Out of 257 colonies, three were still white:one still carried a Ty with the HOcs fragment and was white probably as a consequence of a mutation in that sequence, andin two colonies (0.78%) the HOcs was converted, but the SUP4 insert remained. We can
Break-Induced Ty Recombination
G lu
73
TABLE 2
Gal
0 ON 0 0.5 1 is 2 2.5 3- 4
5 6.5-8-ox
Plating efficiency of strains on SGal-Ura us. SD-Ura
"
-GC Strain
-DSB
AP35 (AP34/pGALHO) AP38 (AP34/pGAL) "CALIBRATION
pGALH0 j PGkl
---
H
HOC5
SUP4-0
H
I
AP35/AP38:0.66
5
0.83 ?
0.09 0.02
AP37/AP39:0.49
"The ratios were calculated as described in MATERIALS AND The Gal/Gh ratio of AP3.5 is significantly different from that of AP38, and that of AP37 is significantly different from that of AP39 ( P < 0.005).The ratio of AP37 is also different from that of AP35 (P< 0.025).The ratios of AP38 and AP39 are not statistically different from each other. METHODS.
I
SB
0.57 t 0.10 0.86 ? 0.02
Corrected ratio
AP37 IAP36(ALTR)/ AP39 [APSG(ALTR)/ 0.41
B
Gal/Glu ratio"
M
TylSupHALTR
8.6
5.9
H
I
H
Conversion
7 8.1
C
0
5
10
Tlme
15
20
(hours)
FIGURE 4.-Molecular analysis of DSB-induced recombination in a Ty with a single LTR. (A)Southern blot analysis of an induction experimentwith AP37[TylSupHALTR/pGALHO]. (R) Schematic representation of the different bands observed in the Southern blot analysis.(C) Kinetics of DSB creation and repair in AP37. Symbols as in Figure 2.
thus estimate that in AP37 only -0.78% of the conversion tracts are shorter than80 bp, the distance between the HOcs and the SUP4 insert. From the experiments reported we conclude that at least two different mechanisms can repair a DSB in a Ty: one of them leads to ectopic recombination and the other to solo LTR products. These two mechanisms compete for the same substrate(s) and may share certain steps, because it is possible to eliminate the action of one of them (the oneleading to solo LTR products) with only minor effects on kinetics or viability (but see below). Effect of the DSB on cell viability: No loss in viability could be seen in the experiments reportedabove. However, because the cells keep growing during the time course of the experiment, deathof a partof the population might not be detected, due to growth of the remaining cells. The efficiency of the repair mechanisms
can be measured by the ability of the cells to resume growth after the damage has been repaired. To get an accurate estimate of cell viability during the induction of recombination, a plating efficiency experiment was performed. We measured the ability ofAP35 and AP37 to form colonies on galactosecontainingplates, as compared with glucosecontainingones. As controls, we used strains AP38 and AP39, which are AP34 and AP36, respectively, transformed with pGAL (a URA3Ck'N plasmid that has a GAL10 promoter but does notcarry the HO gene). Results are shown in Table 2. Both control strains (AP38 and AP39) gave the same lower plating efficiencies on SGal-Ura plates, as compared with SDUra plates, a result probably reflecting physiological preferences of this particular genetic background, or higher plasmid instability when the GAL promoter is expressed. The ratio Gal/Glu of AP35 was significantly lower than that of AP38, and that of AP37 was even lower. The results obtained with AP37 are statistically different from those of AP35. If we correct for the carbon source factor, we conclude that -34% of the cells of AP35 and 51% of the cells of AP37 are unable to form colonies in galactose when the HO endonuclease is expressed, reflecting a failure to repair the DSB. Effect of the DSB on the cell cycle: When cells are subjected to damage that causes breaks in their chromosomes, they often arrest temporarily in the cell cycle. Checkpoint genes responsible for this delay havebeen characterized (WEINERT and HARTWELL 1988;WEINERT et al. 1994). We thus followed the progression of cultures of AP35, AP37, AP38 and AP39 cells through the cell cycle while growing on YF'Gal. Figure 5 shows the distribution of cells in the cell cycle during aninductionexperiment. Whencells were transferred from YF'Gly to YPGal ( b ) , -30 and 10% of the cells were in the S and G2 stages of the cell cycle, respectively; these proportions remainedunaltered in AP38 and AP39 while the number of cells increased logarithmically. AP37 cells, in contrast, showed a clear accumulationof cells in G2,starting at
74
A. Parket, 0. Inbar and M. Kupiec
A loo
1
0
C 100
B
AP38
,
5 Time (hours)
10
0
"22
5
22
10
T h e (hours)
D
AP35
AP37
loo
;Jr 20 0
. 0
5 Time (hours)
10
. 22
0
5 Tima (hours)
10
..
22
FIGURE 5.-Distribution of cells in the cell cycle. (A) AP38 (AP34/pGAL); (B) "39 [AP36(ALTR)/pGAL];(C) AP35 (AP34/ pGALHO); (D) AP37[AP36(ALTR) /pGALHO]. Percent of cells in G1 ( A ) , S (0), or G2 phase (H) of the cell cycle during growth on WGal, are shown.
2 hr, reaching a plateau (30%) by 3 hr and lasting until 7 hr after transfer to YPGal.At this time the proportion of cells in G2 diminished again. A similar trend could be seen inAP35 cells. AP38 and AP39 did not show this effect, implying that the accumulation of cells in G2 depends on the expression of the HO endonuclease. The difference between AP35 and AP37 probably reflects the relative repair efficiencies in both strains, because the TylSupHinAP35 can be repaired by two mechanisms, whereas that of AP37 can be repaired by only one. We have shown that upon growth on galactose there is a loss in viability and an accumulation of cells in G2. However, it is still possible that these effects are due to the action of the HO endonuclease not in the TylSupH, but in some other site of the genome, such as the HML or HMR loci, or some cryptic HOcs. To rule out this possibility,we repeated theplating and cell cycleexperiments described above with twonew strains: AP35R, and AP37R. These strains were derived, respectively, from AP35 and AP37, and carry the pGALHO plasmid, but their marked TySup lack the HOcs insert. The results obtained were indistinguishable from those of AP38 and AP39, which do not carry the GALinducible HO gene (data notshown). We thus conclude that the viability and cell cycle effects seen in AP35 and AP37 are indeed dueto the DSB created by the HO endonuclease at the TylSupH.
DISCUSSION
Ty elements are naturalrepetitive sequences that can engage in two different types of homologous recombination events: with other Ty elements dispersed in the genome, and between the terminal direct repeats (LTRs;ROEDER and FINK 1982; KUPIEC and PETES 1988a,b; MELAMEDet al. 1992; PARKETand KUPIEC 1992). In this paper we show that a DSB created in a marked Tyis repaired by the same two mechanisms. In both spontaneous and induced events, the ectopic conversion accounts for 20% of the final products produced. The other 80% of the products have the whole Ty replaced by a solo LTR element. This type of event can be createdby a variety ofmechanisms, among them intrachromatidal crossover, unequal crossover and single-strand annealing (SSA). When we deleted one of the LTRs (strain AP37), the DSB was produced and repaired with similar kinetics, but this time only the conversion product was detected. This product accumulated in AP37 with the same kinetics and reached levels similar to those of the combined products seen in AP35, showing that the two mechanisms compete for a common substrate. In the absence of one of them theother can take over and compensate. In a series of studies involving directly repeated LacZ genes on a plasmid (RUDIN and HABER 1988; FISHMANLOBELL et al. 1992; FISHMAN-LOBELL and HABER 1992),
Break-Induced Ty Recombination an HOinduced DSB was repaired by two alternative mechanisms. One mechanism resulted in a gene conversion, in which information was transferred from one copy of the repeated geneon the plasmid to the other. The second mechanism led to the replacement of the repeated genes by a single copy, and was shown to be nonconservative and to include a single-stranded intermediate. After the break, single-strand degradation was observed in a 5' to 3' direction; the deletion was then presumably created by annealing of the exposed regions of homology (SSA LIN et al. 1990, MARYON and CARROLL 1991). our system, the gene involved the whole T ~the ; putative partners for recombination were dispersed in the genome in many different ectopic locations. The two LTRs of the Ty can also engage in direct repeat replacing the whole Tyby a single repeat. Despite the differences, the general kinetics and ratio between the two mechanisms are very similar in both studies, with the solo repeat being the main product (80%). It is therefore probable that the mechanism that repairs the DSBby generating a solo LTR in our system is a SSA one. We will assume in this discussion that the LTR-generating mechanism is SSA, but the same considerations apply to any other mechanism able to create the solo LTR final product. The gene conversion and LTR-generating mechanism compete for repairingthe DSB. The larger amount of solo LTR product, relative to that of the conversion product, indicates that the SSA competes more efficiently for a common substrate. When no SSA is possible, the conversion pathway can take over, repairing most of the breaks with a kinetics similar to the One Seenfor in the mechanisms act independently On the DSB substrate Or whether they share Some to be we have
shown (NEVo-CASP1 and KUPIEC lgg4; LIFR' and M* KUPIEC, data) that mutations in different DNA repair (RAD) genes differentially affect the two pathways. It be pointed Out that in AP35 the ectopic conversion of the Ty, which requires a search for homolO g y throughout the genome, takes approximately the same amount of time as that required for annealing the two located LTRs- This to that, in agreement to previous genetic data, the search for homology is not rate limiting in these conditions UINKSand PETEs l986; LICHTEN and W E R 1989; WALDMANlgg4); AP37 cellsshowed a more Pronounced G2 arrest. It is possible that the search for to the DSB of a TY element isrelatively fast and facilitated by the high number of potential Partners ( M E W E D and KUPIEC 1992; WILSON et al. 1994). Southern analysis shows that the HOcut DNA moled e s diminish while thefinalproducts accumulate. However, an accurate determination ofthe efficiencyof the repair process is not possible, because the cultures
75
continue to grow during the experiment. A better estimate of this efficiency can be obtainedfrom the plating experiment (Table 2). When the cells are challenged to form colonies on galactose in the presence of the p G K H O plasmid, only 66% of the cells in AP35, and only 49% of the cells in AP37 are able to do SO. The higher survival rate of AP35 probably reflects the fact that it has two mechanisms to repair thebreak, whereas AP37 depends on only one. Similar levels of survival have been reported for a DSB-initiated recombination between ectopically located single genes (FAIRHEAD and DUJON1993). Consistent with these results, we can see an accumulation of cellsin G2, probably by checkpoint mechanisms that prevent them from cycling inthe presence Of unrepaired DNA damage (WEINERT and HARTWELL 1988). The fraction of arrested cells is higher in AP37 than in AP35, that either more in the population arrest, Or the Same proportion Of stops for a longer period of time. Cells that are unable to repair the DSB and die probably at G2 permanently. According1Y7 the increase in the Proportion ofG2 cells in AP37 is similar to the increase in cell death in the plating experiment. This is consistent with the possibility that in AP37 more cells arrest terminally in G2. The fact that AP35 and AP37 show the same kinetics of appearance of red colonies reflects the fact that in both strains the commitment to give riseto recombinant colonies when plated is similar. We believe that this commitment is dependent only on the DSB creation: every cell that underwent a DSB will either be repaired efficiently to give a recombinant colony, or will die. This may also explain the difference in the kinetics of accumulation of final products in the Southern blot analysis, as compared with that of appearance of red colonies. The bands representing final products continue to accumulate slowly after 8 hr, whereas at that time the level of red colonies has already reached a plateau. Cells which have not yet completed the recombination event at this time, are already committed to it, and willgive rise, when plated, to red colonies. Previous experiments in our lab have shown that the levelof Ty recombination is notinduced by DNA damage (pARKET and KUPIEC 1992). In this paper, we have shown that it is possible to increase Ty recombination of a marked T~ element by producing a defined lesion (a DSB) in it. Several hypotheses could explain the lack of induction of recombination of T~~by DNA damage: (1) T~ DNA could be resistant to DNA damage, due for example to Some special properties of the Ty chromatin. (2) Tysmay be as sensitive to DNA damage as other regions in the genome, but they may not be repairable by the repair mechanisms of the cell. Any lesion in TY DNA would then be left unremoved, and cause cell death. For example, a special chromatin configuration may preclude the repairenzymes from acting on it. ( 3 ) Lesions in Tys could be repairedby a mechanism that does not yield visible recombinants (such recombinational repair using the sister chromatid as template, or Some form ofnonrecombinational repair).
76
A. Parket, 0.Inbar and M. Kupiec
Without giving a definitive answer, the experiments presented in this paper allow us to distinguish between some of the possibilities mentioned above. Our results show that the HO endonuclease is able to recognize its target site when it is embedded in a Ty and creates a DSB efficiently. The same endonuclease is unable to do so when the same site is in the context of the HM loci, due to a special chromatin configuration (KLAR et al. 1981; NASMYTH 1982). This implies that ifTys do have a special chromatin structure that makes them resistant to DNA damage, its properties are different from those of the HM loci. Of course, we cannot rule out the possibility that the chromatin propertiesof the Ty were changed by the insertion of the HOcs or the SUP4 marker. We have also shown that the DSB is not left unrepaired when it is present in the contextof a Ty element. Even though not all the cells succeed in repairing the break, it is obvious from Southern blot analysis and genetic data that most of the breaks are repaired. The results from the plating experiment imply that cell death can accountfor no more than 50% of the undetected Ty recombination. Thus, it is unlikely that the lack of induction ofTy recombination is due to an intrinsic inability of Tys to be repaired. The third hypothesis proposes that Tys are preferentially repaired by alternative repair mechanisms. It is possible that the successful induction of recombination using the HO endonuclease stems from the fact that it is a continuous source of DNA damage. For example, Tysmay preferentially use their sister chromatids as templates; however, in cells expressing HO continuously at high levels, it will often occur that both chromatids will be broken at the same time, forcing the cells to use an intact ectopic sequence as source of homologous information. In addition, even when the sister chromatid is the preferred source, the HO endonuclease will continue to cut until eventually the break will be repaired by either ectopic conversion or SSA. The same considerations apply if ligation of the broken ends is the mechanism preferred by Tys. Another point to consider is that W or MMS treatments cause damage all over the genome, and may create a cellular response different from that generated by a single DSB at a specific location. Nonrecombinogenic repair mechanisms may be induced by W but not by a DSB (for a discussion see SILBERMAN and KUPIEC1994). Finally, the presence in the cell of linear Ty cDNA molecules able to recombine with chromosomal copies of Tys (MELAMED et al. 1992) may also have a selective effect on Tys. It is interesting to note that spontaneous and HOinduced recombination events show the same distribution between conversion and solo LTRs; it seems that the low level of spontaneous Ty recombination might be due to low levels of initiation by DSBs. It is possible that while a DSB in a Ty leads to repair by recombination, other types of lesions (such as pyrimidine dimers,
or single-stranded nicks and gaps), when present in Tys, may be repaired without being converted to DSBs. Tys have several unique characteristics that may differentiate them from other sequences in the genome: they are present in large numbers (30 per haploid genome; BOEKEand SANDMEYER 1991); they have a structure that includes two direct repeats (BOEKEand SANDMEYER 1991); they are heavily transcribed (Ty RNA represents 10% of the total poly(A) R N A , ELDERet al. 1981); being retrotransposons, Tys are probably s u b jected to transposition regulation (BOEKE and SANDMEYER 1991); and Tys create cDNA copies that previously have been shown to participate in homologous recombination with other Tys in the genome(MELAMED et al. 1992). We do not know whether one or several of these properties is important in determining the different response of Tys to DNA damage. We thank RM
