Department of Genetics, University of Washington, Seattle, Washington 981 95. Manuscript received August 27, 1992. Accepted for publication October 24, 1992.
Copyright 0 1993 by the Genetics Society of America
Replication-Dependent Sister Chromatid Recombination in rad1 Mutants of Saccharomyces cerevisiae Lisa C. Kadyk’ and Leland H. Hartwell Department of Genetics, University of Washington, Seattle, Washington 981 95 Manuscript received August 27, 1992 Accepted for publication October 24, 1992 ABSTRACT Homolog recombination and unequal sister chromatid recombination were monitored in r a d l - I / radl-1 diploid yeast cells deficient for excision repair,and in control cells, RADllradl-1, after exposure to UV irradiation. In a radl-llradl-1 diploid, UV irradiation stimulated much more sister chromatid recombination relative to homolog recombination when cells were irradiated in the GI or the G2 phases of the cell cycle than was observed in RADllradl-1 cells. Since sister chromatids are not present during G I , this result suggested that unexcised lesions can stimulate sister chromatid recombination eventsduringorsubsequentto DNAreplication. T h e results of mating rescue experiments suggest that unexcised UV dimers do not stimulate sister chromatid recombination during the G2 phase, but only when they are present duringDNA replication. We propose that there are two types of sister chromatid recombination in yeast. In the first type, unexcised UV dimers and other bulky lesions induce sister chromatid recombination during DNA replication as a mechanism to bypass lesions obstructing the passage of DNA polymerase, and this type is analogous to the type of sister chromatid exchange commonly observed cytologically in mammalian cells. In the second type, strandscissions created by X-irradiation or theexcision of damaged bases create recombinogenic sites that result in sister chromatid recombination directly in G2. Further support for theexistence of two types of sister chromatid recombination is the fact that events inducedin radl-llradl-1 were due almost entirely to gene conversion, whereas those in RADllradl-1 cells were due to a mixture of gene conversion and reciprocal recombination.
U
V irradiation induces mostly pyrimidine dimers and other base modifications (RAHN1979) that ;we normally removed very efficiently by excision repair (FRIEDBERG 1985). However, unexcised UVinduced lesions can interfere with the passage of DNA polymerase during replication (e.g., SETLOW,SWENSON and CARRIER 1963; CLEAVER1965; RUPP and HOWARD-FLANDERS 1968; BERGERand EDENBERG 1986) [for a review, see HANAWALT et al. (1979)l or RNA polymerase during transcription (SAUERBIER and HERCULES1978). Since mutants defective for excision repair can continue to grow and divide in the presence of low levels of damage (JAMES, KILBEYand PREFONTAINE 1978), there must exist mechanisms to complete replication and transcription in the presence of unexcised lesions. Replication over UV-induced lesions in excision-repair defective Escherichia coli and mammalian cells is thought to involve the formation of “daughter-strand gaps” due to incomplete replication around pyrimidine dimers (RUPP and HOWARD1968; SEDCWICK1975; LEHMANN1972; FLANDERS CLARKand HANAWALT 1984). In E. coli, completion of replication after theformation of such gaps appears to be dependent on recombinational repair between
’
Present address: Laboratory of Molecular Biology, University of Wisconsin, 1525 Linden Drive, Madison, Wisconsin 53706. Genetics 133: 469-487 (March, 1993)
sister duplexes (CLARKand MARCULIES 1965; HOWARD-FLANDERS and BOYCE1966; RUPP et al. 197 1; GANESAN1974; SMITHand SHARMA1987). UV treatment of mammalian cells also appears to induce replication-dependent sister chromatid exchange (WOLFF,BODYCOTEand PAINTER1974), although some daughter-strand gaps may be formed and resolved by filling-in mechanisms (LEHMANN 1972). Filling-in of daughter strand gaps was also proposed to explain UV mutagenesis in a radl-1 yeast mutant (KILBEY,BRYCHCY and NASIM1978). R A D l is essential for theincision step of the excision repair pathway that is required for removal of most UV damage in yeast (REYNOLDS and FRIEDBERC 1981; WILCOXand PRAKASH198 1). R A D l hasalso been implicated to have a role in some but not all types of spontaneous recombination (SCHIESTL and PRAKASH 1988; KLEIN 1988; THOMAS and ROTHSTEIN1989; ACUILERAand KLEIN 1989). We monitored sister chromatid and homologrecombinationinduced by UV treatment o f R A D I l r a d 1 and r a d l l r a d l strains at different times in the cell cycle. Our evidence shows that r a d l l r a d l mutants undergo high levels of UVinduced sister chromatid recombination that is replication-dependent, whether the lesions are incurred in GI or GP.By contrast, excision proficient R A D l l r a d l
470
Kadyk L. C.
and L. H. Hartwell
cells primarily undergo sister chromatid recombination after UV irradiation in G2,a n d we infer that most G,-induced UV lesions are removed prior to the onset of replication. T h e hypothesis that there a r e two types of sister chromatid recombination is supported by the fact that the S-phase associated sister chromatidrecombinationevents in radllradl cells a r e almost 100% gene conversions, whereasthe sister chromatid recombination inducedin RADllradl cells a r e only 77% gene conversion. We suggest that the UV-induced sister chromatid recombination we observe in radllradl cells is a result of a mechanism used to complete replication in the presence of low levels of unexcised UV lesions. We considerfour models to explain how these events may be induced. MATERIALS AND METHODS Yeast strains and growth conditions: The yeast strains used in this study are listed inTable 1. Strain LK33 (MATa/ matde1::TRPI; radl-l/radl-l) is of hybrid strain background.A control strain proficient for excision repair, LK33R (RADIlradl-I), was constructed from strain LK33 by one-step gene replacement (ROTHSTEIN1983) of the radl-1 allele on one of the two homologous chromosomes with the RADl allele.The replacement was done by cotransformation of a 5.8-kb linear Sal1 fragment bearing the RAD1 gene from plasmid L869 (a generous gift of R. KEIL) with plasmid pYL21, which has LEUl (Hsu and SCHIMMEL 1984) cloned into YEp24 (BOTSTEINet al. 1979). Leu+ colonies were selected, individually patched onto YEPD plates and screened for UV resistance. One such UV-resistant isolate was picked and analyzed by first growing the cells nonselectively to lose pYL21, then transforming the strain according to the method of SCHIESTL and GIETZ(1989) with plasmid pLK22, bearing LEU1 and MATa. The strain was then sporulated and tetrads were shown to segregate 2+:2- for UV resistance. The locus generating UV resistance was shown to be closely linked to the RADl locus by crossing a Rad+segregant from the sporulation of LK33R with a RADl strain and sporulating the resulting diploid; no UV-sensitive haploids resulted from this cross,out of 64 complete tetrads scored, showing that the radiation resistance maps to within 0.78 map unit of the RADl locus. “Mating rescue” experiments were done using haploid LK25-1(radl-I), one of the parents of diploid strain LK33. The mating partners used in these experiments were either two congenic MATa strains, 831 l(RAD1) and 8315 (radlA)of the A364a background, or two noncongenic MATa strains, LK60-15 (RADI) and LK20-1 (radl-1). The rad1 deletion in strain 831 5 was made by a 2-step gene replacement procedure (WINSTON, CHUMLEY and FINK1983) using plasmid pLK23 (see “Plasmids”).The resulting deletion removes 2.1 kb from the middle of the 2.9-kbRADl gene. A deletion of the RAD50 gene in the A364a strain background was constructed by gene transplacement using the 6.2-kb BglII to EcoRI fragment from the plasmid pNKY83, a generous gift of N. KLECKNER. YEPD and dropout media have been described (SHERMAN, FINKand HICKS1981). YM-1is described by HARTWELL (1967). -N medium is yeast nitrogen base without amino acids and ammonium sulfate, 1.6 g/hter; 0.09 M succinic acid;adjusted to pH 5.8with NaOH. To arrest cells in G I using a-pheromone, log phase cells (5-10 X 106/ml) were collected and resuspended at the same concentration
r
in YM-1 adjusted to H 4.0 withHCI, and a-pheromone was added to 2 X 10- molar. After 2 hr of shaking at 30”, the cells were at least 95% unbudded and forming projections. T o arrest cells in GI by nitrogen starvation, cells were grown in YM-1 glucose (2%) tomid-log phase (about 5 x lo6 cells/ml), washedonce with -N medium, resuspended at 5-10 X lo6 cells/mlin -N glucose (2%) plus nutrients required to supplement auxotrophies in the strain (0.002%), and allowed to grow24-48 hr at 23”, until >95% were unbudded. T o arrest cells in G P ,log-phase cellswere plated to about 5 X 10’ cells per plate on YEPD plates containing the microtubule inhibitor drug methylbenzimidazol-2-yl carbamate (MBC), a gift of Du Pont Corp., at a concentration of 100 pg/ml. After 3 hr at 23”, about 95% of cells were arrested in G2, as monitored by the presence of largebudded cells containing a single nucleus. The remainder of the cells appeared unbudded. This arrest regime typically resulted in about 10% cell lethality. Plasmids: Plasmid pLK22 contains LEUl ona 5.1-kb BamHI fragment (HSUand SCHIMMEL 1984), MATa on a 1980), 4.1-kb BamHI fragment (NASMYTH and TATCHELL ARSl on a 0.8-kb EcoRI/HzndIII fragment (STRUHLet al. 1979),and CEN3 on a 1.7-kb HindIII/BamHI fragment (FITZGERALD-HAYES, CLARK andCARBON 1982) in pHSS6 et al. 1986). Plasmid pYL21 is YEp24 with LEU1 (SEIFERT on a 3.2-kb Sau3A fragment cloned into the BamHI site. Plasmid L1025 (a generous gift ofR. KEIL) contains a disruption of the yeast RADl gene on a 4.2-kb Sal1 fragment: a 2.1-kb StuI-ClaI fragment of the RADl gene is removed and replaced with 1.4 kb of DNA that normally flanks the yeast LEU2gene. L1025 was modified by ligating intothe polylinker a 1.2-kb SmaI fragment containing URA3, to create pLK23. Sister chromatid recombination andhomolog recombination substrates: The recombination substrates usedin this paper have been previously described in detail (KADYK and HARTWELL 1992). Homolog recombination wasmeasured by selecting gene convertants between the heteroalleles leul-1 and leul-12on -leu synthetic medium dropout plates. Unequal sister chromatid recombination was measured using a substrate containing 5’ and 3‘ deletion fragments of the ADE3 gene that overlap by 305 bp and which are separated by the URA3 gene (Figure 1A). This substrate, denoted SCR::URA3,was integrated into chromosome I l l between LEU2 and HIS4.All strains bearing SCR::URA3 carry the ade3-130 mutation, a deletion including all of the chromosomal region containing homology with the 305-bp overlap in the sister chromatid recombination substrate, thus precluding the formation of ADE3by recombination atthat locus. Stable ADE3 recombinants can arise from SCR::URA3 by either unequal reciprocal recombination or gene conversion between sister chromatids. Intrachromosoma1 “popout” events result in an extrachromosomal circle bearing ADE3 which lacks a replication origin and is therefore unstable and not selected in our experiments. Cells bearing the ade3-130 mutation are both Ade- and His-. All strains used in this paper are both ade2 and ade3, therefore ADE3 recombinants were selected on -his dropout medium. Unequal gene conversion and reciprocal recombination events in the SCR::URA3 substrate can be distinguished by a nonselective red-white half-sectored colony assay (Figure 1B) based on the changes in pigmentation that result from mutations in the adenine biosynthetic genes (ROMAN 1956). The ade2 ade3-130 strains bearing SCR::URA3 are white; an unequal sister chromatid recombination event results in a red ADE3 ade2 colony. When cells are plated nonselectively, events occurring in the first division after plating will generate red/white half-sectored colonies, representing the
+
+
47 1
Replication-Dependent Recombination TABLE 1
Saccharomyces cerevisiae strains Genotype
Strain
1.K20-1
MATa ade2 ade3-130cyh2 ura3-52 canl sap3 radl-1
1.K2.5-1
MATa SCR::URA3 ade2 ade3-130 leul-I2 cyh2 ura3-52 canl sap3 trpl lys2 radl-1
LK33
MATaSCR::URA3ade2ade3-13Oleul-l2cyh2 ra3-52canl MATA::TRPIade2ade3-1301eul-1
1.KJJK
sap3trpl lys2 r a d l - 1 cyh2 ura3-52canIsap3trplLYS2radl-1 ” ”
MATaSCR::URA3ade2ade3-130leu1-12cyh2ura3-52canlsap3trpl lys2 R A D l MATA::TRPIade2ade3-130leu-1-lcyh2ura3-52canlsap3trplLYS2radl-1 -
1.K60-15
MATa ade2 ade3-130 leul-12cyh2 ura3-52 canl sap3 lys2
1.~87
MATA::TRPI ade2 ade3-I30 leul-Icyh2 ura3-52 canl radl-1 MATa ade2 ade3-130 leul-I cyh2 ura3-52 can1 trpl radl-1
I.KX8
MATA::TRPlade2ade3-130leul-12 cyh2 ura3-52canltrplradl-1 MATa ade3-130 leul-12 cyh2 ura3-52 canl TRPl RADl
“
~
”
ade:!
LK89
MATA::TRPI ade2 -ade3-130 leul-1 cyh2 ura3-52 can 1 trpl radl-1 ade2 ade3-130 leul-1 cyh2 ura3-52 can I t r p l R A D l MATa
LK90
MATA::TRPl ade2 ade3-130 leul-12 cyh2 ura3-52 canl trpl LYS2 radl-1 MATa ade2 ade3-130 leul-12 cyh2 ura3-52 canl trpl lys2 radl-1
8202SCR
MATa SCR::URA3 ade2 ade3-130 leul-1cyh2 ura3-52 canl sap3 trpl
”
” -
8219-1
MATa SCR::URA3 ade2 ade3-130 leul-12 cyh2 ura3-52 canl sap3 rad52-1
8271-1
MATa SCR::URA3 ade2 ade3-130cyh2 ura3-52 canl sap3 trpl radlA
8271-3
M A T a SCR::URA3 ade2 ade3-130 leul-12 cyh2 ura3-52 canl sap3 trpl radlA
8272-2
MATa ade2 ade3-130 ura3-52 canl sap3 rad52-1
8274-1
M A T a SCR::URA? ade2 ade3-130 leul-1 cyh2 ura3-52 canl sap3 trpl radlA rad52
831 1
MATa ade2 ade3-130 leul-1cyh2 ura3-52 sap3
8313-1
M A T a SCR::URA3 ade2 ade3-130 leul-12 cyh2 ura3-52 canl sap3 trpl radlA rad50A
8314-1
M A T a SCR::URA3 ade2 ade3-I30 leul-12cyh2 ura3-52 canl sap3 trpl rad50A
8315
MATa ade2 ade3-130 leul-1cyh2 ura3-52 canl sap3 radlA
All of the strains described above were generated during the course of this study. Strains whose names begin, with “8” are congenic with A364a. Strains beginning with “LK”are hybrids between A364a and another strain background.
two products of the recombination event. Reciprocal exchange events generate aUra- His- white sector, while gene conversion events generate a Ura+,His- white sector. Southern blot analysis of red/white sectored colonies: DNA was isolated from both URA3 and ura3 white halves of redlwhite sectored solonies, digested with the restriction enzyme EcoRI, electrophoresed on a 0.7% agarose gel and blotted onto a nitrocellulose filter (M.S.I.). The blot was probed with plasmid “Yip5 2.0,” containing 2.0 kb from the BglII to the first ClaI site of the ADE3 gene. Different numbers and sizes of restriction fragments were predicted for the URA3 and ura3 strains. All structures were found to be as expected. U V irradiation: Cells to be irradiated were plated at a density of no more than 5-10 X lo7 cells per plate on minimal, complete synthetic or YEPD MBC plates. Irradiation was done using a Hanovia letheray germicidal lamp. The dose rate was monitored using a UVX radiometer (UVP Inc.).In experiments where recombination induction was being monitored, cells were washed off the plates using -N medium, centrifuged, resuspended in a smallvolume (usually 0.5 ml) of -N, sonicated, diluted and plated on the appropriate plates. All manipulations were done under yel-
+
+
low light, and plated cells were kept under foil or in a metal box to prevent photoreactivation. Effect of U V irradiation on mutation rates toHis+ and Leu+:T o determine what fraction of UV-induced His+ and Leu+ colonies in our experiments were not actually recombinants, but UV-induced revertants or suppressors, we constructed radl-llradl-I and RADllradl-1 diploids that were homozygous for ade3-130,leul-1 or leul-12 (diploidsLK87, LK88, LK89 and LK90) and measured the rate of UVinduced His+ and Leu+ colonies in asynchronouslygrowing cultures of these strains. No UV-induced suppressors of the ade3-130 deletion were observed in strain LK87 ( r a d l - I )or strain LK89 ( R A D I l r a d l - I )at any dose; therefore, the rate of reversion must have been at least 100-1000 times lower, depending on the dose, than the rate of induced recombinants in the SCR::URA3 construct in strains LK33 or LK33R. The rate of reversion of leul-I and l e u l - I 2 in radl-11 radl-1 strains LK87 and LK90 over a range of doses was at least 500 times lowerthan the rateof induced recombination between the leu1 heteroalleles in strain LK33 ( r a d l - l l r a d l 1). The rate of UV-induced reversion of leul-I in RADII radl-1 strain LK89 was 75-100 times lower, and of leul-12
472
A.
L. C. Kadyk and L. H. Hartwell
- -
The SCR::LIRA3substrate in chromosomeIII
FIGURE1.-(A) The sister chromatid recombination substrate (SCR::URA3)consists of 5' and 3' deletions of the ADE3 gene, indicated by the open arrows missing the tail or the head, respectively. A- region of overlap between the two deletions is shaded black and is 305 bp. The two deletions are separated by the selectable B. Sectored colony assayto distinguish marker URA3, indicated by a thick line. This substrate reciprocal recombination from gene conversion was integrated near the centromereof chromosome 111 between LEU2 and HIS4. For further details about the construction of this substrate, see KADYK and HARTWELL (1992). (B) The sectoredcolony assay used to distinguish unequal reciprocal recombination from gene conversion events in the SCR::URA3 substrate. By plating cells nonselectively, both halves of the recom" bination events can be recovered, and are recognized + as red-white sectored colonies. White His- half-sectors 0 generated by reciprocal exchangeevents are Ura-, while those resulting from gene conversion events are Sectored Colony: ny: 5' A d e 3
LIRA3
3'8 d e 3
=+-
Wk=
-
I
1
Red, ADE3,
LIRA3
White, ade3, u r d
Red, ADEJ,
White,
in R A D l l r a d l - 1 strain LK88 was 300-1500 timeslower than the rate of induced recombination between the leu1 heteroalleles in strain LK33R ( R A D l / r a d l - I ) . Viability curves: Cells were grown and synchronized as necessary prior to irradiation, then collected by centrifugation and resuspended in a small volume of -N medium. Cells were then sonicated, diluted onto complete synthetic medium plates (3 per dose) and irradiated either with U V or with X-rays to the appropriate dose. Colonies appearing after 3 days of growth at 30" were counted, and viability at a given dose was calculated as the viable cell concentration atthat dose divided by the viablecell concentration of unirradiated cells. Mating rescueexperiments: Cells of strain LK25-1 were grown to 5-10 X lo6 cells/ml, then arrested in either G I or in G p as described above. Half of the cells were irradiated with U V , then washed from plates. These and the unirradiated cells were collected by centrifugation, and resuspended in 0.5 to 1 mlof -N medium. Cells to be used as mating partners were grown to 5-10 X lo6 cells/ml, collected by centrifugation and resuspended in 0.5-1 ml of -N medium. Cell concentrations of each mating partner were estimated by counting a sample in a hemocytometer. Equal numbers (either 5 X 10' or 1 X IO8 cells) of each mating partner were mixed in a tube, then portions of 5 X 1 O7 cells were divided into tubes containing 5 ml of -N medium. The cells in each tube were then collected by filtration onto individual Millipore 47-mm, 0.45-wm GA-6 filters and the filters were placed onto the surface of a YEPD plate for 2.5 hr at 30". Each filter for a given mating was then placed into a 40 ml tube containing 25 ml -N medium to wash the cells; after removing the filters, the cells were collected by centrifugation, resuspended in 0.5 or 1 .0 ml of -N medium, sonicated, diluted and plated onto -ura -trp dropout plates to measure the total number of diploids formed and onto ura -trp -his dropout plates to measure diploids which had undergone a sister chromatid recombination event. Each haploid strain was also plated on complete synthetic medium
URA3
./ clra . +
ade3, LIRA3
plates and on -his dropout plates to measure the spontaneous and induced levels of sister chromatid recombination prior to mating.
RESULTS
U V irradiation of RADllradl-1 and radl-llradl1 strainsproducesdifferentrecombinationalresponses: Weconstructedisogenicdiploidstrains LK33 ( r a d l - l / r a d l - l ) and LK33R (RADllradl-I), in whichunequalrecombinationbetweensisterchromatids could be monitored using theSCR::URA3 substrate (see MATERIALS AND METHODS) and recombination between homologs could be monitored by gene conversion between heteroalleles for leu1 (Zeul-I and leul-12) as described previously (KADYKand HARTWELL 1992). Recombinational repair was studied by arresting growing cells of these strains either in G1 using a-pheromone or in G Zusing MBC, irradiating them on the surface of agar plates with either UV light or X-rays, then collecting and plating them onleu o r -his plates t o select recombinants, or on rich medium to determineviable cellnumber. After 3 days of growth at 3 0 " , recombinationfrequencieswere calculated for each dose. In RADl/radl-I cells, homolog recombination was highly induced following GI irradiation, butless so after G2 irradiation. In contrast, there was a dose-dependent increase in sister chromatid recombination following irradiation in G2 but not following irradiationin GI. The G I irradiated cells did show a slight induction at the first data point, but itjcreasing doses of' damage gave lower levels of induction, thus the induction was not dose-dependent
Replication-Dependent Recombination
473
C.
D.
loo0
lo00
LK33R.RADlIndl
b.
36C1 SCR
100
SCR
100
100
G2
C2 IIR
a
10
10
0
30
60
90
120
UV dose, J/d
0
2
4
6
8
1
0
UV dosc, J/n?
0
2
4
6
8
X-ray dosc, krads
t 0
2
4
6
8
X-ray dose, krads
FIGURE2.-Asynchronously growing cultures of isogenic strains LK33 (rudZ-l/rudl-Z) and LK33R ( R A D Z l r u d l - I ) were arrested in GI with a-pheromone or in GPwith MBC, UV- or X-irradiated at different doses, and monitored for levels of induced ADE3 and LEU1 colonies. Recombination frequencies were calculated as number of recombinants p e r million viable cells. SCR = sister chromatid recombination. HR = homolog recombination. For each data point, n = 2. These data are thesame as those used in Table 2, A and C.
(Figure 2A). RADllradl-I cells were more resistant to UV damage incurred in G2 than that incurred in G1 (Figure 3A). UV irradiation of the r a d l - l l r a d l - 1 strain in GI or G2 produced strikingly different recombinational responses than were seen in the RADllradl-I strain. For example, even at doses high enough to kill 99% of the cells, much less homolog recombination was induced in the r a d l - l l r a d l - 1 strain in G1 or G2 than was induced in the RADllradl-I strain irradiated to a similar level ofviability. In contrast, even at low doses that generated little loss in viability, sister chromatidrecombination was highly induced following UV-irradiation in either G1 or G2 (Figures 2B and 3B). Differences were also observed in the ability of RADIlradl-1 and radl-Ilradl-1 cells to survive UV damage. Not only are r a d l - l l r a d l - 1 cells over 10 times more sensitive to UV damage than are RADII radl-1 cells, but they exhibit no cell cycle difference in their ability to repair UV damage (Figure 3B). r a d l - l l r a d l - 1 strains are unable to remove UV damage by excision repair (REYNOLDS and FRIEDBERC 1981; WILCOXand PRAKASH 1981). In order to determine whether the differences in recombinational repair observed between RADlIradl-1 and r a d l - I / radl-1 strains were due to this defect, rather than to a second function of RAD1 directly affecting recombination, we examined the responses of these two strains to X-irradiation. Unlike UV irradiation, Xirradiation induces primarily single-strand and double-strand breaks, lesions which are not removed by excision repair.X-irradiation of eitherstrainproduced recombinational and survival responses similar to those observed after UV-irradiation of RADllrad 1I cells (Figures 2, C and D, and 3, C and D). In both strains, homolog recombinationwas highly induced in G1 irradiated cells but less highly induced in G2 irra-
diated cells, there was a dose-dependent increase in sister chromatid recombination following irradiation in G2, and G1 irradiated cells were more sensitive to lethal X-ray damagethan were G2 irradiated cells (Figures 2, C and D, and 3,C and D). All these results are consistent with previous observations in Rad+ cells (BRUNBORGand WILLIAMSON1978; KADYK and HARTWELL1992).Since r a d l - l l r a d l - 1 and RADII radl-1 strains have similar responses to X-irradiation, we conclude that the difference in recombinational and survival responses following UV irradiation of radl-Ilradl-I and RADIlradl-I strains is due to the lack of excision repair in radl mutants, rather than to the lack of a second function of the RAD1 gene product in recombination. X-irradiation in GI produceda modest dose-dependent increase in sister chromatid recombination in the radl-llradl-I strain, that was notapparent in the RADllradI-I strain. Therefore, we suggest thatX-irradiation induces some lesions that can be removed by excision repair. Excision repair played no majorrole in the cell's ability to survive X-ray damage (Figure 3, C and D). To express in a quantitative manner the difference in therecombinational responses observed in the r a d l - l l r a d l - 1 and R A D l l r a d l - 1 strains following UV- or X-ray-induced damage, we determinedthe ratios of induced sister chromatid recombination to induced homolog recombination at all doses for each experiment (Table 2).Very different responses to UV damage were observed in the RADllradl-1 and radll / r a d l - l strains. Because the radl-l/radI-1 strain is unable to remove UV damage, a given dose of irradiation produces greater lethality than for theRADII radl-1 strain, which efficiently removes much UV damage. Therefore, we have compared the ratios of induction of sister chromatid recombination to induction of homolog recombination in these two strains
474
L. C. Kadyk and L. H. Hartwell
B.
A.
0
60 120 180 240 W dose, J/m2
D.
C.
0
2
4
6
8
1
0
W dose, J/m2
0
4 8 12 16 X-ray dose, krads
0
4 8 12 16 X-ray dose, krads
FIGURE3.-Asynchronously growing cultures of strain LK33 or LK33R were arrested either in GI using a-pheromone or in GP using MBC (see MATERIALS AND METHODS), diluted onto complete medium plates, and irradiated at different doses with X-rays or U V light. The number of colony-forming units per ml was expressed as a percent of the colony-forming units per ml in the unirradiated culture.
over a similar range ofviability rather than at the same dose. At any given dose following UV-irradiation of the r a d l - l l r a d l - 1 strain, the ratio was at least 10 times higher than the ratio observedin the R A D I I r a d l - 1 strainirradiatedtotheabout same level of viability in the same stage of the cell cycle(Table 2A). For example, cells from the r a d l - I l r a d l - 1 strain irradiated in G1 with 2 J/m2 to 85% survival had an induced sister chromatid recombination to induced homolog recombination ratio of 5.70, whereas cells from the R A D l l r a d l - I strain irradiated in GI with 30 J/m' to 77% survival had a ratio of 0.40. Cells from the r a d l - l l r a d l - 1 strain irradiated in GZwith 2 J/m' to 84% viability hadaratio of 9.48, whereas cells from the R A D l l r a d l - 1 strain irradiated in Gz with 60 J/m2 to 79% viability had a ratio of 0.51. To confirm thatthe r a d l - I l r a d l - 1 and R A D I l r a d l I strains had different responses to UV damage even when irradiated with the same dose, we irradiated both strains with 10J/m2 in G1 and Gz, and measured the levelsof inducedhomologrecombination and sister chromatid recombination. At equivalent doses of UV, the r a d l - l l r a d l - I strain had a ratio of induced sister chromatid recombination to induced homolog recombination that was 13 times higher in GI (9.11 0.7) and 4 times higher in G2 (2.310.6) than the ratio measured in the R A D l l r a d l - I strain atthecorresponding stages (Table 2B). We also performed this experiment using congenic R A D I I R A D I and r a d l h l r a d l A diploids of a different strain background, and obtained similar results (data not shown). In contrast to the results observed after UV irradiation, the ratios of sister chromatid to homolog recombination induced by X-irradiation were very similar in the r a d l - I l r a d l - 1 and R A D l l r a d l - I strains (Table 2C). In addition to the fact that sister chromatid recombination was highly induced at very low doses of damage in the r a d l - l l r a d l - 1 strain,UV-irradiation in G I stimulated as much sister chromatid recombi-
nation per dose of damage as did UV-irradiation in Gz (Figure 2B). In contrast, except at the lowest dose of damage tested, more sister chromatid recombination was stimulated in GZ than in GI after irradiation of the R A D l l r a d l - 1 strain (Figure 2A). Furthermore, sister chromatidrecombination was maximallyinduced at a very low dose following irradiation of the r a d l - l l r a d l - 1 strain in either GI or GZ, resulting in a plateau in the dose-response curve at higher doses, whereas a more gradual increase in the dose-response curve was observed for UV-induced sister chromatid recombination in Gz-irradiated cells of the R A D I I r a d l - 1 strain. Another difference in the response of R A D I l r a d l 1 and r a d l - l l r a d l - 1 cells to UV-irradiation was that G1 and Gz cells of the r a d l - l l r a d l - 1 strain were equally sensitive to killing by UV, whereas GI cells were more sensitive than Gzcells after UV-irradiation of the R A D l l r a d l - I strain. Although UV-induced recombination was greatly affected by thepresence of a r a d l - I mutation, we have observed no significant difference in spontaneous levels of sister chromatid and homolog recombination between the R A D l l r a d I - I and r a d l - l l r a d l I strains. Spontaneous levels of sister chromatid recombination were 25.8 f 6.5 and 27.0 f 9.7 per lo6 viable cells in the r a d l - l l r a d l - 1 and R A D I l r a d l - I diploids, respectively (n = 4). Spontaneous levels of homolog recombination were 21.3 k 3.9 and 30.8 f 11.1 per lo6 viablecellsin the r a d l - l l r a d l - 1 and R A D l l r a d l - I diploids, respectively ( n = 4). A model for replication-dependentsisterchromatidrecombination: T h e above-described differences in UV-induced recombination and survival between r a d l - l l r a d l - l and R A D l l r a d l - I strains led US to hypothesize that UV-induced damage in a r a d l - I / r a d l - 1 mutant stimulates sister chromatid recombination in a replication-dependent manner, such that even damage incurred in G2 does not stimulatesister chromatidrecombination until thenext cellcycle,
475
Replication-Dependent Recombination TABLE 2
Induction of sister chromatid andhomolog recombination in RADIlradl-1 and radl-llradl-1 strains A. Ratio" of UV-induced SCRIHRb in R A D l l r a d l - 1 vs. r a d l - l l r a d l - 1
LK33R ( R A D l l r a d l - I ) GI
Dose (J/m')
57.5 30 60 90 120
SCR
HR
f 10.6 38.0 f 24.0 29.0 f 11.3
143 f 75.0 178 f 11.3 373 f 18.4
G2
Ratio
Viability (W) HR
SCR
77 17 1
37.0 f 2.8 73.0 f143 5.7 68.0 2 15.6 93.0 f 22.6
0.40 f 0.14 0.21 f 0.15 0.08 f 0.03
Ratio
Viability (%)
0.71 f 0.03 0.51 f 0.08 0.39 f 0.04 0.42 f 0.16
90 79 69 40
Ratio
Viability (W)
28.1 f 5.90 9.48 f 0.90 4.23 f 0.68 6.00 f 1.84
90 84 5 1
HR
Ratio
Viability (%)
44.0 f 15.6 101 f 3.5 236 f 38.9
0.42 f 0.15 0.37 f 0.14 0.30 f 0.08
93 79 79
52.0 f 1.4 f 33.9 173 f 22.6 223 f 33.9
LK33 ( r a d l - l / r a d l - l ) GI
Dose U/mP)
1 2 6 10
SCR
HR
163 f 47.4 163 f 9.19 153 f 4.95 143 f 23.3
6.00 f 12.8 28.5 f 6.36 15.0 f 7.07 31.0 f 4.24
G2
Ratio
Viability (W) HR
SCR
90 85 7 1
98.5 f 0.71 109 f 29.0 55.0 f 2.83 60.0 f 9.90
27.0 f 65.7 5.70 f 1.59 10.2 f 4.46 4.60 f 0.15
8. Net UV-induction of HR and SCR by 10 J/mP in strains LK33 and LK33R Stage HR SCR Strain Dose (J/m2)
LK33 LK33R LK33 LK33R
10 10 10
G, GI
10
Gz
127 56 89 18
Gz
3.50 f 0.71 11.5 f 2.12 13.0 f 1.41 10.0 f 1.41
Ratio
14 80 39 30
9.1 0.7 2.3 0.6
C. Ratios of X-ray induced SCR/HR in R A D l l r a d l - 1 vs. r a d l - l l r a d l - I LK33R ( R A D l l r a d l - 1 ) G2
GI
Dose (krad)
2 4 8
SCR
HR
17.5 f 0.71 16.0 f 15.6 7.5 f 13.4
158 f 7.1 393 f 170 808 f 205
(W)
ViabilityRatio"
SCR
0.11 f 0.0008 18.5 75 f 13.4 37.5 44 f 16.3 0.04 f 0.014 0.01 f 0.01 11 71.5 f 30.4 LK33 ( r a d l - l / r a d l - l )
~
GI
Dose (krad)
2 4 8
SCR
HR
9.0 f 5.7 16.0 f 4.2 32.5 f 0.71
143 f 17.0 313 f 45.3 923 f 215
~
~~~~
G2
Ratio
Viability (%)
0.06 f 0.05 0.05 f 0.006 0.04 f 0.007
74 40 11
SCR
10.0 f 7.1 21.0 f 5.7 50.0 2 5.7
HR
Ratio
Viability (%)
23.0 f 8.5 58.0 f 7.1 195 f 12.0
0.43 f 0.15 0.36 f 0.05 0.26 f 0.01
100 79 76
All values listed are number of recombinants per lo6 viable cells. Each data point in parts A a n d C is the average of two independent cultures, andis calculated from the same data presented in Figure 2 (note that the data presented here are net inductions over the uninduced levels). " Ratios are the net induction of sister chromatid recombination divided by the net induction of homolog recombination. The net induction is calculated by subtracting the spontaneous from the inducedlevels. SCR = sister chromatid recombination.HR = homolog recombination.
during or after replication. In contrast,we expect that UV-induced sister chromatidrecombination in a RADllradl-I cell can, and usually does, occur in a replication-independent manner. T h e reasoning behind this hypothesis is the following: (1) Since UVinduced damage cannot be removed in the radl-I/ radl-1 strain, damage incurred in G I must be present during replication. T h e fact that UV damage incurred in GI induced high levels of sister chromatid recombination shows that these events must occur during or after replication, since sister chromatids are not present to recombine in G I . (2) The ratios of UV-
induced sister chromatidrecombination to UV-induced homolog recombination in the radl-llradl-1 strain were higher than in the RADIlradl-I strain. These differences in ratios were due both to higher induction of sister chromatid recombination and to lower induction of homologrecombination in the radl-llradl-1 strain relative totheRADllradl-1 strain. T o explain the lower induction of homolog recombination, we hypothesize that most UV-induced homolog and sister chromatid recombination in RADllradl-I cells is RADI-dependent, probably because theseevents are stimulated by nicks or gaps
476
L. C. Kadyk and L. H. Hartwell
generated during excision repair (FRIEDBERG 1985). T o explain the high induction of sister chromatid recombination in radl-llradl-1 mutants, we hypothesize that an alternative, RADI-independent pathway for UV-induced sister chromatidrecombination is revealed whencells are forced to continue the cell cycle in the presence of DNA damage. This pathway responds primarily to unexcised UV lesions, since the radl-llradl-1 and RADIlradl-I strains had similar recombinational responses to X-ray damage (Figure 2 and Table 2). (3) The unusual plateau observed in the sister chromatid recombination dose-response curves in the radl-llradl-1 strain suggests that the mechanism for this type of recombination is saturated at low doses of damage, unlike the mechanism that is responsible for theX-ray- or UV-induced homolog and sister chromatid recombination in the RADflradl-I strain or the X-ray-induced homolog and sister chromatid recombination in the radl-llradl-1strain. Saturation at low doses might be expected if replication in the presence of DNA damage requires sister chromatid recombination; at doses where the chance of having UV-induced lesions on both strands is high, cells might be unable to complete replication and would therefore die. Consistent with the idea that sister chromatid recombination contributes to theability of the cells to survive in the presence of UV damage is the fact that the U V survival curve in the radl-llradl-1 strain begins to decrease at a higher rate at the same dose where sister chromatid recombination is saturated. (4) The radl-l/radl-l strain was more resistant to X-rayinduced damage in G2 than in G I , but the G I and G2 viability curves were identical after UV damage of that strain.These observations suggest that UV lesions introduced at any stage of the cell cycle had the same fate in theradf-llradf-1 strain. In contrast,the RADIlradl-1 strain must deal with GP induced lesions more efficiently than G1 induced lesions, since Gz cells were more resistant to either UV or X-ray damage than were GI cells. We tested the following predictions of the model that UV-induced sister chromatid recombination in radl-1 cells occurs by a replication-dependent mechanism: (1) UV-induced sister chromatidrecombination in a radl strain is replication-dependent. This prediction was tested using a mating rescue experiment, described below. (2) If UV-induced sister chromatid recombination in a radl strain is replicationdependent, it may occur by a different mechanism than UV-induced sister chromatid recombination in a RADl strain. We used two methods to test this possibility. One was to characterize the fraction of UVinduced events that appear to be reciprocal recombination relative to those that appear to be gene conversion in RADI and rad2 strains, using the sectored colony assay (Figure 1B). The other was to compare
the dependence of UV-induced sister chromatid recombination in RADl andradl-I cells on known recombinational repair genes. Mating rescue experiments indicate that UV-induced sister chromatid recombination in rudl-l cells is replication-dependent: We hypothesized that the UV-induced sister chromatid recombination in radl1 cells was a result of replication over unexcised lesions; to test this hypothesis, we performed the following “mating-rescue” experiments, outlined in Figure 4. We arrested radl-1 (strain LK25-1) in GZ with MBC and UV irradiated them. These cells were then removed from the MBC-containing plates and mated for 2.5 hr at 30” in equal proportions with either of two congenic strains, strain 831 1 (MATa RADI) or strain 8315 (MATa radlA). Because mating occurs in the next G1, before the irradiated cells have replicated their DNA, we predicted that the RADllradl-I diploids formed by mating with strain 831 1(RADI) should be competent to remove the UV-induced lesions by excision repair prior to replication, thus removing thepotentialforinduction of replicationdependent sister chromatid recombination events among those diploids. By contrast, mating with strain 8315 (radlA) would notcomplementthe excision repair defect, and therefore we expected that replication-dependent sister chromatid recombination would be induced in the radl-llradl-1 diploids. We found about 10-fold higher induction of sister chromatid recombination in theradlAlrad1-1 diploids than in the RADIlradl-I diploids, consistent with our hypothesis that theseevents are replication-dependent (Table 3). However, to our surprise, the induced rateof events among the radl Alradl-1 diploids was five to six times higher than the induced rate in the unmated radl-1 haploids (Table 3, experiments 1 , 2 and 5). A similar result was observed when the irradiated cell was mated to a radl-1 rather thana radlA partner (experiments 3 and 4), so this was not an allele-specific effect. In addition,theRADIlradI-1 diploids formed in the mating rescue experimentsstill appeared tohave some induction of sister chromatid recombination, although never as much as in theunmated haploid parent. These results suggested that the yield of UV-induced recombination per unit of unexcised dimer was much higher in newly formed diploids than in haploids. In an attempt to understand the origin of the elevated levelsof sister chromatidrecombination observed among the newly formed diploids when compared to theunmated haploid strain, we tested several hypotheses. T h e experiments described next will rule out several of these hypotheses, and we will be left with the conclusion that, for unknown reasons, unexcised U V lesions are more recombinogenic in newly formed diploids than in unmated haploids. Before
Replication-DependentRecombination
A.
Outline of matingrescueexperiments
* Rate of SCR inradl-1 haploids * Rate of SCR inRADIlrudl-1 diploids
&mGTaRAD1 in 1 ~ a t e t o M A T a r a d 1in~GI
Rate of SCR inradlAlradl-l diploids
MATaradl-~z-
(or, %GI)
B.
+UV
*
Mate to MATaR*l inG1 Mate to MATaradlA in
Rate of SCR inradl-I haploids Rate of SCR inRADllrad1-1 diploids Rate of SCR inradlAhlrad1-1 diploids
Mating withRAD1 orradl cells
uv\
C"M
radl MAT a, G2
t
x radl MAT a
6 G1
I
x RADl MATa
477 FIGURE4.-(A) Outline of mating-rescue experiments that test the replication-dependence of UV-induced sister chromatid recombination in r a d l cells. r a d l - 1 mutant cells were synchronized in the cell cycle using a-pheromone or MBC, and a portion of these cells was UV-irradiated. The unirradiated and irradiated cells were theneither plated immediately tomeasurethe spontaof sister neousandinducedrates chromatid recombination in the r a d l - 1 haploid, or else mated separately to a RAD1 MATa strain and a rad1A MATa strain. ( B ) Illustration of rescue by RAD1 MATa cells. r a d l cells UV-irradiated in G phave sister chromatids at the time of the damage, but are not predicted to undergo sister chromatid recombination at that stage. Irradiated cells mated with RADl cells are able to remove the UV lesions by excision repair prior to the onset of replication, so sister chromatid recombination is not induced, whereas irradiated cells mated with r a d l cells retain the UVinduced lesions during replication.
TABLE 3 Mating rescueof UV-induced SCR SCR frequency per lo6 viable cells
Haploids
-uv
+uv
Newly formed diploids
-uv "
Expt.
1
2 3' 4' 5 6'
7d
Strain'
Arrest
LK25-1 G2 LK25-1 G2 98 LK25-1 13 G2 LK25-1 G2 LK25-1 GI 8202SCR 30 G2 LK25-1 G2
Dose (J/m')
2.0 99 4.0 1.5 1.5 2.0 50 0.0
radl-1 RADllradl-1 radl-l radlA/radl-1 RADllradl-1 radlA/radl-1
19 19
ND
73
14
ND
ND
ND
ND
15
ND 140
ND
12
76 ND68
14
30
ND
ND
Ratiob
23 450 410 20 120 12
43 79 28 48 46 82
350
ND
15
700 450
8.1 5.7 10.6 14.6 9.8 1.5
An outline of these experiments is in Figure 4. In all cases except for experiments 3, 4, and 6 , mating was done with congenic strains 851 1 ( M A T a , R A D l )and 8315 (MATa, r a d l A ) . a Listed is the strain containing the SCR::URA3 construct, which was arrested prior to mating. Inall cases except experiment 7, this strain was the recipient of UV irradiation prior to mating. This is the ratio of induced events measuredin r a d l A / r a d l - 1 diploids to the induced events measuredin R A D l l r a d l - I diploids. ' Mating in experiments 3, 4 and 6 was with the non-congenic strains LK60-15 ( M A T a , R A D I )and LK20-1 (MATa, r u d l - 1 ) . In experiment 7, strain LK25-1 was arrested in G P .then mated to strain8315 ( r a d l A , n o SCR::CIRA3)that had been UV-irradiated with 'rJ/nl' in log phase prior to mating.
digressingintothisseriesofexperiments, we want to emphasize O L I ~conclusion fromthematingrescue experiment.Weconcludethatthemajorityof UV lesions introduced into a r a d l cell in either G1 o r G2 do notbecomerecombinogenicuntiltheynext pass through the S-phase of the cell cycle. Theorigin of highlevels of recombination in newly formed diploids cannot be explained as an
effect of ploidy, extra growth on selective plates amongnewly formed diploids, mating rescue of lethally irradiated cells, heterozygosity for mating type, or a trans-acting effect of UV irradiation followed by mating: O n e possible explanationforthe high levels of sister chromatid recombination among the newly formed diploidsrelative to theirradiated butunmatedhaploidparent was simply that diploids
478
L. C. Kadyk and L. H. Hartwell
were more recombinogenic than haploids. However, we note that direct UV irradiation of diploid LK33 did not induce such high levels of sister chromatid recombination (Figure 2B). A second hypothesis to explain the greater induction of sister chromatidrecombinationobserved amongthe r a d l A / r a d l - 1 diploids thanamongthe UV-irradiated r a d l - 1 haploids is that (1) the newly formed zygotes divided more on the selective plates than the haploids did, and (2) since lesions cannot be removed in r a d l - 1 cells, these extra divisions produced more recombinants. It is known that new UVinduced mutations can arise several generations after the initial lesions are incurred in a r a d l - l strain (JAMES, KILBEYand PREFONTAINE 1978). T o test the prediction that newly formed zygotes divided more on selective plates than didhaploids, we compared the growth of strain LK25-1 plated inGI or G:! on -his plates with the growth of cells from a mating mixture of strains LK25-1 MATa and 8315M A T a on -ura -trp -his dropout plates. Unbudded, small-budded and large-budded cells, zygotes, and 3-4-celled microcolonies were scored microscopically immediately after plating and after 24 hr at 30”. Sixteen or 30% of haploid cells plated in GI or G2, respectively, formed 3-4-celled microcolonies after 24 hr, and theremainder were unbudded or single-budded cells. N o microcolonies were observed with more than 4 cells. In two mating mixtures that were plated, we estimated that immediately after plating about 10-16% of the cells had the morphology characteristic of zygotes. After 24 hr, 15-30% of the cells from the matingmixes had formed 3-4-celled microcolonies. However, no microcolonies were observed with morethan 3-4cells. Thus, zygotes do not appear to divide on the selective plates any more than do haploids. These data do not rule out the possibility that an undetected fraction of zygotes progress further in the cell cycle than haploid cells. Therefore, to determine whether such an undetectedfractioncouldaccountfor the effects we observe, we tested whether a few generations of nonselective growth of cells following DNA damage and prior to plating selectively could significantly increase the observed recombinationfrequency. We treated G2-arrested cells ofstrain LK25-1 ( r a d l - 1 )with either no UV or 2 J/m2 of UV. Both unirradiatedand irradiated cells were then either plated immediately or diluted in rich medium to 10’ cells/ml and allowed to grow 4-5 generationspriorto plating. Unirradiated cells had19 sister chromatidrecombination events per lo6 viable cells when plated immediately and 13 events per lo6 viablecells after outgrowth. Irradiated cells had 73sister chromatid recombination events per lo6 viable cells when plated immediately and 79 events per lo6 viable cells after outgrowth. If no more induced sister chromatid recombination oc-
curred following outgrowth of irradiated cells, it is expected that the measured recombination frequencies would bedecreased by afactor of two after outgrowth, since the two daughter cellsof a largebudded G2 cell result in only one colony-forming unit per sister chromatid recombination eventwhen plated immediately, but result in two colony-forming units per sister chromatid recombination event if allowed to divide before plating. The fact that the measured recombinationfrequency was the same afteroutgrowth as before outgrowth suggests that more recombinantsdidoccur during outgrowth. However, the increase observed is no more than a factorof two over that expected, not enough to account for the high levels of induced sister chromatid recombination observed in the newly formed r a d l - l / r a d l A diploids. The facts that newly formed zygotes don’t appear to divide any more on selective plates than do haploids, and that extra divisions do not greatly elevate the measured recombination frequenciesled us to discard these explanationsfor the high levelsof induction observed in the mating rescue experiments. A third possible explanation for the unexpectedly high levels of induced sister chromatid recombination among newly formed diploids was that the full induction of sister chromatid recombination was not measured in the haploid strain because those cells in the population that incurred high enough doses of DNA damage to inducea high rate of sister chromatid recombination are also more likely to have suffered a recessive lethal mutation. If this idea were true, the full induction might be measured amongthe diploids, since after mating theunirradiatedgenome would complement recessive lethal lesions, thus rescuing that segment of the population with higher recombination frequencies. We therefore tested whether potentially lethally irradiated cells were being “rescued” by mating. Cells of strain LK25-1 were irradiated in GI at several different doses, mated immediately with asynchronously growing cells of strain 8315 ( r a d l A ) and platedforrecombinants.Inaddition,matingefficiency, defined as the number of diploids formed per viable input MATa cell, was monitored. If lethally irradiated cells were being rescued by mating, the measured mating efficiencies per viable cell for irradiated cells would appear higher than for unirradiated cells. It was found that the apparent mating efficiencies per viablecell did increase as haploid viability decreased (Table 4). This result indicates that some inviable haploid cells were being rescued by mating. If one assumes that the increased recombination frequency among the newly formed diploids were due simply to a higher rate of recombination in the subpopulation of rescued cells, it is possible to estimate the rate of recombination in that population. We find that at the lowest doses, the rate of recombination
479
RecombinationReplication-Dependent TABLE 4 Mating rescue of lethally irradiated cells SCR frequency/106 viable cells Dose Strain
LK25-1
(J/m*) Arrest
G1
Spontaneous
0 1
UV-induced
13 79 64 71 93 83
2 3 4
x radlA
21 310 450 410 510
Haploid Estimated Mating viability (%)a efficiency
100 89 615 25 696 9
100 103 125 255 329
rate'
8010 1966
Viability is that measured for the unmated strain, relative to unirradiated control cells. Mating efficiency is expressed as a percent of the mating efficiency of unirradiated cells. Mating efficiency is total number of diploids divided by the number of viable LK25-1 cells put into the mating mixture. Estimated number of recombinants per l o 6 cells among those cells that were rescued from death by mating. It is assumed that a mating efficiency above 100% is due to mating with cells that might have died, and that only these cells have a higher rate of sister chromatid recombination than was observed in an unmated haploid. Thus, for example, the estimated rate among the cells rescued by mating after 2 J/m' of UV irradiation would be x, such that (0.25)(x) (1)(71) = (1.25)(450). a
+
among the lethally irradiated cells would have to be several orders of magnitude higher than in the nonlethally irradiated cells (Table 4). Since the number of recombinogenic lesions per cell should fit the Poisson distribution, it is not feasible that the rescuedcells could have so many more recombinogeniclesions than the other cells. Furthermore, it is not expected that the rate of recombination among the lethally irradiated cells would decrease with increasing doses. Therefore, we discard rescue of lethally irradiated cells as an explanation for thehigh levels of induction of sister chromatid recombinationamong thediploids. A fourth hypothesis is that the stateof being MATal MATa stimulates recombination,thuselevating the levels of UV-induced sister chromatid recombination in both the RADIlradl-1 and the radlAlrad1-1 diploids. In support of this idea is the fact that MATal MATa diploids have higher levels of spontaneous and UV-induced homolog recombination than do MATaI MATa or MATaIMATa diploids (FRIISand ROMAN 1968; KADYKand HARTWELL1992). Since the radlI diploid strain LK33 was MATaIMATa, we had not previously tested the possibility that heterozygosity at the MAT locus might allow higher levels of UV-induced sister chromatidrecombination.In order to test this idea, we transformed strain LK25-1 (MATa, radl-I, SCR::URA3) with pLK22, a plasmid containing MATa, thus creating a MATalMATa haploid. We arrested cells fromstrainsLK25-1 and LK25-1 pLK22 in G P ,irradiated them with 2 J/m2, and plated forrecombinants and viablecells. Almost identical levels of induction were observed in the two strains: the net induction of sister chromatid recombination following a dose of 2 J/m2 in GB for strain LK25-1 was 45/106 viable cells, andfor strain LK25-1 + pLK22 it was 5 111O6 viable cells. We therefore discard the possibility that cells heterozygous at MAT undergo higher levels of UV-induced sister chromatid recombination than those which are homozygous.
+
A fifth possible explanation for the high levels of sister chromatidrecombinationobserved innewly formed diploids was thatthemating process itself stimulates recombination, in eithera damage-independent manner oras a result of a trans-acting effect of mating immediately after irradiation of one mating partner. However, neither diploids newly formed in the absence of UV damage (Table 3, experiments 1 and 5) nor diploids newly formed afterUV irradiation of the mating partner lacking the SCR::URA3 con7) were found to have struct (Table 3, experiment elevated sister chromatidrecombination. We concludethatmating in the absence of DNA damage cannotaccountfor the high levels ofinduction of sister chromatid recombination we observed in the diploid strains, and that damage is required in cis to the recombinationsubstratefor the high levelsof induction to be observed. It may be that the mating process specifically stimulates recombination in damaged DNA or that it alters the distribution of equal and unequalrecombinationevents such thatmore unequal events are detectablefora given dose of damage. UV-inducedsisterchromatidrecombination in rad1 cells results almost exclusively ingene conversion: We examined the relative frequenciesof unequal sister chromatid recombination events that occurred by reciprocal recombination and those that occurred by geneconversion, using the red-white sectoredcolony assay described in Figure 1B. After plating nonselectively, sister chromatid recombination events that occur on the plate appear as red-white half sectors. Whether an event occurs by reciprocal recombination or geneconversion, the red halfof each sector is always ADE3, URA3. However, the white half of the sector will be URA3 if the recombination event is due to gene conversion, and it will be ura3 if the event is due to reciprocal recombination. The sister chromatid recombination events inducedin the rad 1-
L.and C. KadykHartwell L. H.
480 TABLE 5
TABLE 6
Reciprocal recombinationvs. gene conversion in red-white sectored colonies
Dependence of UV-induced SCR in RADl and rad1 cells on RAD52 and RAD 50
White sectors Percent Stage
Dose (J/m')
Strain
URA+
ura-
G2
100
2 2
LK33R (RAD+) LK33 ( r a d l - I ) LK33 ( r a d l - I )
37 48 28
11
GI GP
Strain
Spontaneous
UV-inducedaNet increase
gene conversion
1
77 98
0
100
UV-induced sectored colonies were isolated on low adenine plates, replated onto rich medium or onto -ura dropout plates, and the Ura phenotype was determined for all cells. All red cells were always URA3, whereas white cells were URA3 ifthe sectored colony arose after a geneconversion in the SCR::URA3 substrate, whereas they were ura3 if the colony arose after a reciprocal recombination event.
I strain were 98% gene conversion in G1 and 100% gene conversion in GP; only 1 reciprocal recombination event was observed in 77 total events monitored in G1 and Gz. In contrast, 77% of the recombination events induced in a Gz RADl strain were gene conversions (Table 5). We have previously estimated the spontaneous rate of sister chromatid recombination to be 2.7 X per cell division (KADYKand HARTWELL 1992), about 100-fold lower than the induced rates in this experiment. Therefore, about 1% of the total sectored colonies scored are likely to be due to spontaneous rather than induced recombination. It is possible that thesingle reciprocal recombination event scored in the radl-llradl-1 strain background was a spontaneous event. These data indicate that the sister chromatid recombination events induced by UVin the radl-llradl-1 and RADllradl-1strains occurred at least in part by different mechanisms. T o confirm that the Ura' and ura- phenotypes scored in these experiments were due to gene conversion and reciprocal recombination as depicted in Figure l B , we analyzed 9 URA3 white sectors and 3 ura3 white sectors by probing a Southern blot with a probe containing homology to the ADE3 gene (see MATERIALS AND METHODS). All recombinants had the structure predicted from their genotypes. RAD50 and RAD52 arerequired for most UVinduced sister chromatid recombination in RADl or rad1 strains: If the UV-induced sister chromatid recombination events in radl and RADl strains are mechanistically different, it is possible that these two types of events might not be dependent on the same recombinational repair enzymes. Therefore, we examined the dependence of UV-induced sister chromatid recombination in radl and RADl strains on RAD50 and RAD52, two well studiedgenes whose products are involved in recombinational repair (HAYNES and KUNZ 1981). All the UV-induced sister chromatid recombination in both radlA and RADl strains was found tobe dependent onRAD52, whereas in either RADI or radlA strains only about 50% of
8202SCR RAD+ 26f5.3 5 7 f 10 2 6 .371 8219-1 rad52 3.7 f 1.0 3.4 f 0.4 -0.3 f 1.3 8314-1 rad50A 18 f39 4.9 f 5.0 21 f 3.8 8271-1 r a d l A 14 f 4.0 74 f 3.6 60 2 4.6 LK25-1 r a d l - l 17 f 0.6 120 f 12 100 2 12 8274-1 radlA rad52-1
