We have studied the way in which circular and linear ... the transforming DNA is a nonreplicating hybrid plasmid, ... some; I , restriction cleavage site present only in the vector sequence; .... Xho I LEU2 fragment from pYeleulO (19) inserted at the Sal I site of ... When two restriction enzyme cuts are made within the yeast.
Proc. Nat. Acad. Sci. USA Vol. 78, No. 10, pp. 6354-6358, October 1981 Genetics
Yeast transformation: A model system for the study of recombination (plasmid integration/double-strand break repair/rad52-l mutation/DNA repair synthesis)
TERRY L. ORR-WEAVER*, JACK W. SZOSTAK*t, AND RODNEY J. ROTHSTEINt *Department of Biological Chemistry, Harvard Medical School and Division of Cancer Genetics, Sidney Farber Cancer Institute, Boston, Massachusetts 02115; and
*Department of Micro iology, New Jersey Medical School, Newark, New Jersey 07103 Communicated by Ruth Sager, June 22, 1981
ABSTRACT DNA molecules that integrate into yeast chromosomes during yeast transformation do so by homologous recombination. We have studied the way in which circular and linear molecules recombine with homologous chromosomal sequences. We show that DNA ends are highly recombinogenic and interact directly with homologous sequences. Circular hybrid plasmids can integrate by a single reciprocal crossover, but only at a low frequency. Restriction enzyme digestion within a region homologous to yeast chromosomal DNA greatly enhances the efficiency of integration. Furthermore, if two restriction cuts are made within the same homologous sequence, thereby removing an internal segment of DNA, the resulting deleted-linear molecules are still able to transform at a high frequency. Surprisingly, the integration of these gapped-linear molecules results in replacement of the missing segment using chromosomal information. The final structure is identical to that obtained from integration ofa circular molecule. The integration of linear and gapped-linear molecules, but not of circular molecules, is blocked by the rad52-1 mutation. Consideration of models for plasmid integration and gene conversion suggests that RAD52 may be involved in the DNA repair synthesis required for these processes. Implications of this work for the isolation of integrative transformants, fine-structure mapping, and the cloning of mutations are discussed.
I
B
x
c=
I
-
FIG. 1. Integration of a plasmid by recombination with a yeast chromosome. Two types of events occur at approximately equal frequencies: (A) integration of a single plasmid molecule by reciprocal recombination within the region of homology or (B) integration of two or more plasmid molecules. These events can be distinguished by restriction enzyme mapping. -, chromosomal DNA; --- vector sequences; o, homologous sequences on the plasmid and the chromosome; I , restriction cleavage site present only in the vector sequence; , restriction enzyme site in flanking chromosomal DNA.
MATERIALS AND METHODS Strains. Strain LL20 (a, his3-11,15, leu2-3,112, canl) was constructed by Lester Lau. It is related to strain S288c, transforms efficiently, and was used as a wild-type RADW host. The double-point mutations in his3 and leu2 do not revert at a detectable frequency. The rad52-1 mutation was obtained from strain LP1929-3D, sent by L. Prakash. This strain was derived from a cross with LL20. We performed three additional backcrosses to LL20 to recover rad52-1 in a more nearly isogenic background. Strain D78-3D (a, his3-11,15, leu2-3,112, canl, rad52-1) transforms moderately well (about 'A as efficiently as LL20) and was used in initial experiments. This strain was crossed two more times to LL20-related strains to generate D217-9C (a, asp5, his311,15, leu2-3,112, rad5, rad52-1), which was used to verify the observed phenotypes ofrad52-1. The rad5 mutation confers UV sensitivity but has no effect on plasmid transformation. Media and Genetic Methods. Media, genetic methods, and yeast transformation were as described (12). Rad52-1 was scored after exposure to 50,000 rads of y-irradiation from a 'oCo source. Plasmid vector sequences were scored by colony hybridization as described (2). Plasmid Construction, Transformation, and DNA Purification. Restriction enzymes and T4 ligase were from New England BioLabs, and reactions were done using conditions specified by the supplier. All DNA samples used for yeast transformation were phenol extracted. The integrity of the DNA and completeness of digestion were verified by agarose gel electrophoresis. Purified DNA fragments were isolated from agarose gels by electroelution in an ISCO sample concentrator. Plasmid constructions were performed by incubating a mixture
Transformation of yeast by defined DNA sequences provides a powerful approach to the study of recombination because the transforming DNA is subject to experimental manipulation. If the transforming DNA is a nonreplicating hybrid plasmid, transformation can occur only by the integration of plasmid information into a yeast chromosome by homologous recombination (1-3) (Fig. 1). Genetic analysis of meiotic recombination in lower fungi provided evidence for heteroduplex DNA at recombination sites and led to the development of the Meselson-Radding model (4-8). In this model, recombination is initiated from a nick on one molecule followed by repair synthesis and displacement ofthe nicked strand to promote heteroduplex formation on the homologous duplex. The experiments described in this paper examine both the effect of DNA ends and the role of repair synthesis in plasmid integration by comparing transformation with circular and linear plasmids. The ends of plasmid molecules are extremely recombinogenic and interact directly with homologous chromosomal sequences; doublestranded gaps in linear molecules are repaired during the process of integration. The pleiotropic recombination and repair mutation rad52-1 (9-11) blocks the integration of linear but not circular molecules. We suggest the RAD52 gene product may be involved in the DNA repair synthesis necessary for the integration and repair of linear plasmids, for double-strand break repair, and for gene conversion. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. ยง1734 solely to indicate this fact.
Abbreviations: kb, kilobase pair(s); bp, base pair(s). t To whom reprint requests should be addressed.
6354
Genetics: Orr-Weaver et al.
Proc. Nad Acad. Sci. USA 78 (1981)
of restriction enzyme-digested DNAs in the presence of T4 liovernight at 40C. The DNA was transformed into competent cells of the Escherichia coli strain 5346 (J. Calvo) that had been prepared by Ca2' treatment and stored at -70'C (13). The desired clones were identified by restriction digests of DNA obtained from minipreparations (14). Large-scale plasmid preparations were performed with slight alterations of the procedure of Kupersztoch and Helinski (15). Yeast genomic DNA was purified by the method of Davis et al. (14). Southern blots, nick translations, and hybridizations were done with only minor modifications of published procedures (16-18). RESULTS DNA Ends Are Recombinogenic. We compared the ability of circular and linear molecules to transform yeast by recombination with homologous chromosomal sequences. We constructed a series of plasmids that contain various fragments of yeast chromosomal DNA (Fig. 2). All of these plasmids are incapable of autonomous replication in yeast; consequently, transformation with such plasmids is much less efficient than with replicating vectors because they must integrate by recombination with homologous chromosomal sequences. The efficiency of transformation with integrating plasmids is increased 10- to 1000-fold when the plasmid DNA is first made linear by restriction enzyme digestion within the region of homology to the yeast genome (Table 1). This phenomenon was observed for DNA fragments from the LEU2, HIS3, and rDNA loci. All transformants are mitotically stable. In one transformation experiment, the stimulation observed by restriction enzyme digestion outside the sequence homologous to chromosomal DNA or on the border between yeast DNA and nonhomologous vector DNA is at least an order of magnitude less than that seen when the break is inside the homologous region. In some cases, no stimulation or a decrease is observed (data not shown). Approximately 30% of transformants from circular molecules or linear molecules cut inside yeast DNA lack integrated vector sequences. Such substitution transgase
R
B/Bg Sil
pS32
I .3kb
iN /
R
RSm
R
p
pWJ12 5.3 kb
5.5kb
:X
7.4kb
is3 ,.
V
-B I/
FIG. 2. Plasmids used. pSZ57 has a 2.3-kilobase pair (kb) Sal Xho I LEU2 fragment from pYeleulO (19) inserted at the Sal I site of pBR322. pSZ32 contains the same LEU2 fragment and a 4.5-kb BgI II
fiagment of yeast rDNA
inserted in the BamHI site of
pBR322.
pSZ62 consists of the 1.7-kb BamHI HIS3 fragment (obtained from K. Struhl) inserted into the BamHI site of pBR322. pSZ63 (not shown) is the same as pSZ62 except that the orientation of the HIS3 fragment
pSZ500 is a deletion derivative of pSZ62 in which a 0.8-kb II fiagment, internal to the HIS3 fiagment, has been removed. pSZ64 is a deletion derivative of pSZ63 in which a 0.6-kb Sal I/Xho I fragment has been removed. pWJ12 consists of pSZ62 with the addition of a 1.7-kbEcoRI/BamHI fragment. All of these plasmids are nonreplicating in yeast and transform only by integration. Cleavage sites: B, BamHI; Bg, Bgi H; K, Kpn I; R, EcoRI; S, Sal I; SI, Sac I; SII, Sac IH; Sm, Sma I; X, Xho I. -, pBR322 sequences;- and is reversed.
Bgi
sup3'
-,
yeast sequences.
Table 1. Transformation with linear plasmids Transformants,* Site Plasmid no./pg 0.2 pSZ64 uncut pBR322 3-18 pSZ64 EcoRI Border 9-33 pSZ64 BamHI 510-750 HIS3 pSZ64 Kpn I 5-450 HIS3 pSZ64 Bgl II
pSZ57 uncut pSZ57 BamHI pSZ57 Sal I pSZ57 Kpn I
pBR322 Border LEU2
0.1-0.2 1.4-1.7 0.9-2.9
12-17
6355
Stimulation -
15-90 45-165 2550-3750 25-2250 -
8-14
4-29 60-117
50-90 pSZ32 uncut rDNA 13-96 1100-4800 pSZ32 Sac I rDNA 1100-3300 13-40 pSZ32 Sac II 14-96 1300-4800 pSZ32 Sac I/Sac II rDNA * The efficiency of transformation with linear plasmids is strongly concentration dependent, so high DNA concentrations must be used to observe a large stimulation in frequency. Routinely, 10 ,ug of DNA was incubated with spheroplasts from 7 x 107 cells and dilutions were plated. The nonspecific variation in frequencies between transformations is probably the result of differences in spheroplast preparations. For pSZ64 and pSZ57, ranges are for two transformations; for pSZ32, range is for four transformations.
formants were described by Hinnen et al. (1) and may result from a mitotic gene conversion event between the plasmid and chromosomal gene. By contrast, 70% of transformants from plasmid molecules cut at thejunction ofpBR322 and yeast DNA and 85% of transformants from plasmids cut inside the pBR322 sequences are of the substitution type. DNA ends must interact directly with homologous chromosomal sequences because restriction enzyme cuts in complex hybrid plasmids can target plasmid integration to specific chromosomal sites. pWJ12 (Fig. 2) recombines only at the sup3+ or HIS3 loci because these are the only loci with homology to the plasmid. In transformed strains that have integrated plasmid DNA, the site of integration can be determined by the presence ofcharacteristic restriction fragments detected by Southern blot hybridization analysis ofgenomic DNA. We examined 13 strains containing integrated pWJ12 DNA, generated by transformation of LL2O with the circular plasmid; 10 had the plasmid integrated at sup3+ and 3 had it at HIS3 (Fig. 3 Upper). In contrast, when the transforming DNA was cut in the HIS3 sequence, all transformants (14/14) had pWJ12 integrated at the HIS3 locus (Fig. 3 Lower) and, when the plasmid was cut in the sup3+ region, 11/11 had pWJ12 integrated at the sup3+ locus. The Southern blot demonstrates that the structure ofintegrated transformants derived from linear molecules is identical to that of transformants derived from circular molecules. Double-Stranded Gaps Are Repaired During Integration. When two restriction enzyme cuts are made within the yeast sequence on a plasmid, an internal segment ofDNA is removed, generating a gapped-linear molecule. The absence ofan internal segment of DNA does not affect the ability of gapped-linear molecules to transform at high efficiency. Remarkably, the gap is always repaired during the process of integration, so that the integrated structure is identical 'to that predicted from an intact circular molecule. We have examined the integration ofgapped plasmids at three loci, the HIS3, sup3+, and rDNA regions. Gaps in the HIS3 region of pSZ62 were made by digestion with Xho I/Kpn I [300-base-pair (bp) gap] or Bgl II (800-bp gap). The sup3+ gap ofpWJl2 was between two Sma I sites (1200-bp gap), and the rDNA gap in pSZ32 was between the Sac I and Sac II sites (700-bp gap) (Fig. 2). We transformed yeast with these gapped plasmids, purified 18 stable transformants for each plas-
6356
Genetics: Orr-Weaver et al. 2
1
3
4
5
6
7
Proc. Nad Acad. Sci. USA 78 (1981) 8
9
kfb
s _ ~ _24
-
-
_
-9.2 --7.4
-*
-4.3
1
2
3
4
5
6
7
8
9
_ _ a_ _w sowao
-24
_
-7.4
-
-4.3
FIG. 3. Targeting of plasmid integration. (Upper) Uncut pWJ12 DNA was transformed into strain LL20. Yeast DNA from independent transformants was isolated and digested with Sal I (which cuts in the pBR322 sequence), and Southern blot analysis was performed with pBR322 as hybridization probe. Integration at the sup3 region results in a major band at 9.2 kb and a minor band at 24 kb (lanes 1, 3, 4, 5, 6, and 8), while integration at the HIS3 region results in a major band at 24 kb and a minor band at 4.3 kb (lanes 2, 7, and 9). Multiple tandem integrations are indicated by an additional band at 7.4 kb (lanes 2, 4, and 8). (Lower) Xho I-digested pWJ12 DNA was transformed into strain LL20, and yeast DNA was isolated and analyzed as described above. Integration occurred solely in theHIS3 region (lanes 1-9). Multiple tandem integrations are indicated by the 7.4-kb fragment in the transformants in lanes 2 and 4.
mid, and analyzed their structure by Southern blot hybridization. Approximately 20% of the transformants resulted from substitution events. In those transformants with integrated plasmid DNA, the restriction fragments we observed were the same in every case as those predicted from the integration of intact circular plasmids (data not shown). To prove that the missing DNA had been repaired from chromosomal information, and not from the plasmid DNA restriction fragment from the gap, we purified gapped plasmids in two ways. First, we cut pSZ32 DNA (an rDNA, LEU2 plasmid) with Sac I/Sac II, both of which cut within rDNA sequences. The DNA fragments were separated by agarose gel electrophoresis and the large fragment, the gapped plasmid, was purified from the gel. This DNA was used to transform yeast. LEU2+ transformants were picked and purified, and DNA was prepared from each transformant. Examination of this DNA by Southern blot hybridization revealed in every case the presence of restriction fragments identical to those produced by integration of circular pSZ32 (data not shown). We never observed the smaller fragments predicted from integration of the deletion. In addition, Southern blot hybridization of Sac I-digested DNA showed that the gap had been repaired correctly so as to regenerate this restriction site. A similar experiment was done with the Bgl II gap of pSZ62. This gapped-linear molecule was purified by cloning. pSZ62 was cut with Bgl II and reclosed into a circle, creating the deletion derivative pSZ500. Digestion of pSZ500 with Bgl II creates a fragment identical to the large Bgl II fragment of pSZ62. Transformation of his3- yeast to HIS' with this gapped-linear DNA occurred at high efficiency. In all cases, the structure of the integrated transformants from gapped-linear pSZ500 or circular pSZ62 was identical; the gap was repaired during integration (Fig. 4).
Figs. 3 and 4 also illustrate the phenomenon ofmultiple tandem integration, previously observed with circular pSZ32 molecules (3). Several of the lanes have a third band not present in the remaining lanes. This band corresponds to full-length linear plasmid molecules, derived (as illustrated in Fig. 1) by restriction digestion of a tandem array of integrated plasmid molecules. To eliminate the possibility that this band derives from free circular plasmid molecules, we did an additional Southern blot experiment. Digestion with a restriction enzyme that cuts once in the pBR322 vector sequence generates the unit-length plasmid band in a subset of transformants (Fig. 5 Upper). However, when the enzyme used cuts only in flanking chromosomal DNA, a single fragment is produced that is proportional in size to the number of integrated plasmid copies (Figs. 1 and 5 Lower). It is noteworthy that the unit-length plasmid band found in the pSZ500 transformants always corresponds to molecules of the full plasmid length and not to the gapped-linear size. As every gapped-linear molecule integrated in a tandem array is repaired during integration, multiple tandem integration cannot be due to the integration of ligated concatemers. Effects of rad52-1 on Plasmid Integration. We have studied the effects ofthe pleiotropic recombination and repair deficient mutation rad52-1 on yeast transformation with circular and linear molecules. We examined two rad52 strains having related but different genetic backgrounds. Both strains grew more slowly than the wild-type strain and both required 5-10 times as much DNA from two different replicating plasmids to yield equivalent numbers of transformants. Transformation with circular integrating plasmids also required 5-10 times more DNA with rad52 strains than with wild type to yield similar numbers of transformants. Therefore, RAD52 does not appear to be required for the integration of circular plasmids. In contrast, the rad52-1 mutation blocks the integration of linear and gappedlinear molecules. In these experiments, amounts of DNA were adjusted so that circular molecules gave similar numbers of transformants on wild-type and rad52 strains. Linear molecules gave a large stimulation oftransformation efficiency in wild-type strains, but little or no stimulation was observed in rad52-1 strains (Table 2). In general, the effect of rad52-1 is somewhat greater for gapped-linear molecules (decreases to 1/400 relative to wild type) made with two separate restriction enzyme cuts than for singly cut linear plasmids. This effect of rad52 on transformation with linear integrating molecules is not a general effect (e.g., increased degradation) on all linear molecules. We 2) 3
1
4 5 6
7 8
9 10
'11 ki
_
_
am
_W
_
_
_
_
_
_
_
_
_w
_
_
-12.0
to -61i -41
FIG. 4. Structure of transformants from gapped-linear molecules. Strain LL20 was transformed with circular pSZ62 or Bgl IH-digested pSZ500. The pSZ500 linear molecules are missing an internal segment of the HIS3 region. DNA from independent transformants was digested with EcoRI, which cleaves the pBR322 sequence, prior to electrophoresis for the Southern blot. Integration of circular pSZ62 produces 12.0- and 4.1-kb fagments that hybridize to pBR322 (lanes 1-3). In addition, a 6.1-kb fiagment is present in lanes 1 and 2, resulting from multiple tandem integration. Integration of gapped-linear pSZ500 in all cases yields full-length fragments of 12.0 and 4.1 kb (lanes 4-11). Therefore, the gap is repaired during integration. Lanes 4, 6, 9, and 11 have the additional 6.1-kb band due to multiple tandem integration.
Proc. NatL Acad. Sci. USA 78 (1981)
Genetics: Orr-Weaver et al. 0
1
2
3 4
5
6
7
kb
_
-
wo g
am-up
2
3
_
-
_
-6.1 -4.4
0
_
1
m
4
5
12.3
6
quences by strand invasion and repair synthesis during recombination. Several lines of evidence support these conclusions. Transformation of yeast with certain hybrid plasmids requires plasmid integration by recombination. This recombination event is greatly stimulated by restriction enzyme digestion within regions homologous to chromosomal DNA, and the integration of complex plasmids can be targeted to a specific chromosomal site by digestion within the corresponding region of the plasmid. Moreover, the integration of gapped-linear molecules is always accompanied by repair of the missing information, using chromosomal information as a template. Fig. 6A presents a model for the integration of linear and gapped-linear plasmids. We propose that one 3' end of a linear or gapped-linear plasmid invades a homologous chromosomal sequence. Strand invasion could be facilitated
-49.0 t
W
-28.0 -23.7
-15.3
FIG. 5. Multiple tandem integration of gapped-linear molecules. DNA was prepared from independent transformants of strain LL20 with gapped-linear molecules of pSZ62. The transforming DNA for the strains in lanes 1 and 2 was digested with Kpn I/Xho I; the remaining strains were transformed with BglI -cut DNA. (Upper) Genomic DNA samples were digested withEcoRl. In addition to the two bands present in every lane, lanes 1,3,4, and 6 contain a 6.1-kb band due to multiple tandem integration. Lanes 0 and 7, size standards. (Lower) The same DNA samples cut in the flanking chromosomal DNA by Xba I. Fragment size is proportional to the number of integrated copies: 24 kb for 1 copy, 30 kb for 2 copies, and 36 kb for 3 copies. Numbers on the side are size standard mobilities. All lanes with a 6.1-kb EcoRI fragment in Upper have a larger Xba I fragment in Lower.
examined transformation efficiency with replicating plasmids made linear by restriction digestion within the pBR322 vector sequences. For both wild-type and rad52 strains the level of transformation is 1-8% of that obtained with circular replicating molecules and presumably reflects ligation of linear molecules to circular forms (Table 2). DISCUSSION Our results show that double-strand breaks in DNA are highly recombinogenic and suggest that the broken ends of DNA molecules interact directly with homologous chromosomal se-
by
Site pBR322 pBR322
5
--
3'
ex-
onucleolytic digestion to make the 3' ends single stranded. Repair synthesis initiated from the 3' invading end would result in strand displacement and the enlargement of the D loop on the chromosome. At some point, the D loop will contain singlestranded DNA complementary to the unpaired 3' end at the other end of the linear plasmid. These single-stranded regions could pair, followed by the initiation of repair synthesis from the 3' end of the plasmid strand. Nicking and degradation of the single-stranded DNA generated by continuation of repair synthesis beyond the region of plasmid-chromosomal homology would result in the integration of the plasmid. This model explains both the integration of linear molecules and the repair of missing information in gapped-linear molecules from chromosomal information. In this model, the two ends of a linear plasmid act cooperatively during integration; the repair synthesis initiated after strand invasion of the first end results in more efficient pairing of the second end. Our observation that cuts on the junction give lower stimulation and result in an increased proportion of substitution events suggests that, for integration ofthe repaired plasmid by this concerted action, both ends need to be in the homologous region. If the nicks and single-stranded tails produced during this process are not immediately repaired, they could facilitate the integration of subsequent plasmids into the same site, resulting in the multiple tandem integrations that occur in >50% of the transformants. The integration of linear plasmids cut within the DNA sequences homologous to the yeast chromosome is analogous to double-strand break repair, a process also shown to require RAD52 (11). Repair of double-strand breaks in yeast is frequently associated with reciprocal recombination in diploids and can occur only in G2 of the cell cycle of haploids (20, 21). This requirement for a sister chromatid suggests that doublestrand break repair involves more than a simple religation. The
Table 2. Transformation of a rad52-1 strain
Plasmid CV13 uncutt CV13 BamHI CV13 Pvu II
6357
Strain LL20 (RAD52) Transformants, Linear/circular* no./.g 20,500 0.035 710 0.020 410
Strain D78-3D (rad52-1)
Transformants,
no./pg
Linear/circular*
2500 148 72
0.059 0.029
pSZ32 uncut 14 67 pSZ32 Sac I 11 0.78 46 rDNA 3,090 pSZ32 Sac II 7 0.50 56 rDNA 3,760 5 0.36 34 rDNA 2,290 pSZ32 Sac I/Sac II Strain LL20 transformations were done with 0.1 pg of CV13, 1 pg of cut CV13, and 2 ,ug of pSZ32 samples, and dilutions were plated. Ten-fold more DNA was used in the D78-3D transformations. Results are the average of two experiments. * Ratio of transformants per pug of DNA for the linear and circular forms of the plasmid. t CV13 is a replicating plasmid containing the LEU2 gene and the 2-pUm circle origin of replication.
Genetics: Orr-Weaver et al.
6358
Proc. Nad Acad. Sci. USA 78 (1981)
A
B
A........ x
f t
FIG. 6. Model for integration of a linear (A) or a circular (B) plasmid. Linear or gapped-linear plasmids integrate by strand invasion of the 3' end (.), followed by repair synthesis, which displaces singlestranded DNA homologous to the other 3' end (e) of the molecule. This end pairs and in turn serves as a primerforrepair synthesis. Continued repair synthesis generates a single-stranded region, nicking of which ( t ) leads to integration of the repaired plasmid. Circular plasmids do not require repair synthesis for integration. Strand invasion may be facilitated by a nick. Nicking the D loop produces a Holliday structure that can be resolved by nicking the outer strands ( T ). -, chromosomal DNA; -, plasmid; ----, newly synthesized DNA.
corollary to this in the integration of linear plasmids is the requirement that the break be within a sequence homologous to chromosomal DNA. Our model is formally equivalent to Resnick's model (22) for double-strand break repair. The integration ofcircular plasmids must occur by a different mechanism from that of linear plasmids, as the RAD52 gene product is not required for the integration ofcircular molecules. Our model for the integration of a circular plasmid is shown in Fig. 6B. Strand invasion is initiated from a nick on either chromosomal or plasmid DNA. In this case, however, it is not necessary to enlarge the resulting D loop by repair synthesis and strand displacement. If the D loop is cut, a short region of symmetric heteroduplex can form, terminated by a classical Holliday junction (4). Resolution of the Holliday structure after isomerization can yield the reciprocally recombined form: an integrated plasmid. The major difference in enzymatic terms between these two models is that repair synthesis is not required for the integration of a circle. RAD52 strains may be defective in some aspect ofthis repair synthesis. This conjecture is consistent with the other known phenotypes of the rad52-1 mutation. rad52-1 mutants are deficient in mitotic gene conversion (9, 10), an event in which repair synthesis is postulated to drive heteroduplex formation (8). The mechanism of homothallic mating-type switching may be partially analogous to gene conversion (23, 24); the rad52-1 mutation leads to lethality during attempted mating-type switching (10, 25). There are several significant practical applications of these results. It is possible to obtain high frequency transformation with integrating plasmids by cutting the plasmid DNA within a region of yeast homology. The addition of sonicated E. coli carrier DNA leads to an increase in transformation frequency with low concentrations (0.01-1 ug/ml) of plasmid DNA (unpublished results). The fact that DNA ends are highly recombinogenic can also be used to advantage in the isolation of integrated derivatives of autonomously replicating plasmids. It is often desirable, for purposes of genetic mapping or strain construction, to integrate a complex plasmid (with several different regions of yeast homology) at a specific chromosomal locus. Such a plasmid can often be targeted to the desired site by
cutting with an appropriate restriction enzyme. The plasmid will integrate at the site to which its DNA ends are homologous. In this manner, an easily scored marker can be placed adjacent to a nonselectable locus to permit genetic mapping. The repair of gapped plasmids from chromosomal information can facilitate the cloning of mutant alleles. Transformation with a gapped plasmid, where the gap covers the site ofa chromosomal mutation, gives an integrated plasmid in which the vector sequences are flanked by identical mutant direct repeats. Isolation of the plasmid in E. coli will always result in recovery ofthe mutant allele. For example, transformation ofsup3' yeast with a gapped SUP3-a plasmid led to the recovery in E. coli of a sup3' plasmid (unpublished results). Transformation with gapped plasmids is also useful for fine-structure genetic mapping. As yeast transformants derived from a gapped plasmid are always homozygous for information within the gap, it is easy to tell whether a particular mutation lies within a particular gap. For example, our data map the his3-11 and his3-15 mutations to a region between the 5' end of the gene and the Bgl II site that is within the H1S3 locus. We thank Ray Wu, in whose laboratory this work was initiated, and Burt Beames and Randy Holcombe for technical assistance. We also thank Charles Radding, Cynthia Kenyon, Andrew Murray, Pat Brown, and Stephanie Ruby for helpful comments on the manuscript. T. L.O.W. was supported by National Institutes of Health Training Grant CA 09361. This work was supported by National Institutes of Health Grant GM 27862 to J.W.S. and National Science Foundation Grant PCM8003805 to R.J.R. 1. Hinnen, A., Hicks, J. & Fink, G. (1978) Proc. NatL Acad. Sci. USA 75, 1929-1933. 2. Ilgen, C., Farabaugh, P., Hinnen, A., Walsh, J. & Fink, G. (1979) in Genetic Engineering, eds. Hollaender, A. & Setlow, J. (Plenum, New York), Vol. 1, pp. 117-132. 3. Szostak, J. & Wu, R. (1979) Plasmid 2, 536-554. 4. Holliday, R. (1964) Genet. Res. 5, 282-304. 5. Stadler, D. & Towe, A. (1971) Genetics 68, 401-413. 6. Leblon, G. & Rossignol, J. (1973) MoL Gen. Genet. 122, 165-182. 7. Fogel, S., Mortimer, R., Lusnak, K. & Tavares, F. (1978) Cold Spring Harbor Symp. Quant. Biol 43, 1325-1341. 8. Meselson, M. & Radding, C. (1975) Proc. Natl Acad. Sci. USA 72, 358-361. 9. Prakash, S., Prakash, L., Burke, W. & Montelone, B. (1980) Genetics 91, 31-50. 10. Malone, R. & Esposito, R. (1980) Proc. NatL Acad. Sci. USA 77, 503-507. 11. Resnick, M. & Martin, P. (1976) MoL Gen. Genet. 143, 119-129. 12. Sherman, F., Fink, G. & Lawrence, C. (1977) Methods in Yeast Genetics (Cold Spring Harbor Laboratory, Cold Spring Harbor,
NY).
13. Morrison, D. (1977) J. BacterioL 132, 349-351. 14. Davis, R., Thomas, M., Cameron, J., St. John, T., Scherer, S. & Padgett, R. (1980) Methods EnzymoL 65, 404-411. 15. Kupersztoch, Y. & Helinski, D. (1973) Biochem. Biophys. Res. Commun. 54, 1451-1459. 16. Southern, E. (1975) J. MoL Biol 98, 503-517. 17. Rigby, P., Dieckmann, M., Rhodes, C. & Berg, P. (1977)J. MoL BioL 113, 237-151. 18. Denhardt, D. (1966) Biochem. Biophys. Res. Commun. 23, 641-652. 19. Ratzkin, B. & Carbon, J. (1977) Proc. NatL Acad. Sci. USA 74, 487-491. 20. Brunborg, G., Resnick, M. & Williamson, D. (1980) Radiat. Res. 82, 547-558. 21. Chlebowicz, E. & Jachymezyk, W. (1979) MoL Gen. Genet. 167, 279-286. 22. Resnick, M. (1976) J. Theor. BioL 59, 97-106. 23. Haber, J., Rogers, D. & McCusker, J. (1980) Cell 22, 277-289. 24. Klar, A., McIndoo, J., Strathern, J. & Hicks, J. (1980) Cell 22,
291-298.
25.
Weiffenbach, B. & Haber, J. (1981) MoL Cell BioL 1, 522-534.
