Proc. Nati. Acad. Sci. USA Vol. 88, pp. 7585-7589, September 1991 Genetics
Integration of DNA fragments by illegitimate recombination in Saccharomyces cerevisiae (yeast/nonhomologous recomblnation/restrction enzymes)
ROBERT H. SCHIESTL AND THOMAS D. PETES Department of Biology, University of North Carolina, Chapel Hill, NC 27599
Communicated by David Botstein, May 17, 1991
DNA fragments (generated by BamHI treatABSTRACT ment) with no homology to the yeast genome were transformed into Saccharomyces cerevisiae. When the fragments were transformed in the presence of the BamfH enzyme, they integrated into genomic BamHI sites. When the fragments were transformed in the absence of the enzyme, they integrated into genomic G-A-T-C sites. Since the G-A-T-C sequence is present at the ends of BamiI fragments, this result indicates that four base pairs of homology are sufficient for some types of mitotic recombination.
Illegitimate recombination events join two DNA molecules (or two noncontiguous parts of a single DNA molecule) without the requirement for extended sequence homology. In a number of human genetic defects caused by genomic rearrangements (1-6), either no homology or very limited homology has been detected at the junctions of the rearrangements (7, 8). Similar recombination events have also been characterized in bacteria (9). In most organisms, nonhomologous integration of introduced DNA is more common than homologous integration. For example, in mammalian cells, only one in one thousand transformants contains the transforming DNA in a homologous position (10). In contrast, in Saccharomyces cerevisiae, all transformants analyzed contained the transforming DNA in homologous positions (11, 12). One interpretation of these results is that Saccharomyces cerevisiae has a very efficient mechanism of homologous recombination (integration) and an inefficient mechanism of nonhomologous recombination. Alternatively, it is possible that yeast has no mechanism allowing nonhomologous integration of transforming DNA. Below, we describe evidence indicating that the first of these alternatives is correct. In addition, we show that certain restriction enzymes can be transformed into yeast cells, and catalyze the integration of transforming DNA sequences.
MATERIALS AND METHODS Plasmids. The plasmid pJL202 was constructed by Joachim Li (The Johns Hopkins University) and obtained from Jef Boeke (The Johns Hopkins University). In this plasmid, the 1.1-kilobase (kb) HindIII URA3 fragment was replaced by the HIS3 gene, leaving sequences homologous to the URA3 flanking region on both sides of the HIS3 gene. The URA3 fragment used in our transformation experiments was derived from the plasmid pM20 constructed by Sue Jinks-Robertson (Emory University) and Martin Kupiec (Tel Aviv University). This plasmid was constructed by "filling-in" the cohesive ends of the 1.1-kb HindIII fragment containing the URA3 gene and inserting this fragment into the HincIl site in the polylinker of pUC7. Thus, the URA3 gene 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. §1734 solely to indicate this fact.
7585
in pM20 is flanked by BamHI sites and (outside of the BamHI sites) by EcoRI sites. Strains and Media. The yeast strain OD5 (MATa leu2-3,112 his3-11,15) was obtained from Walter Spevak (University of Vienna). An isogenic strain (RSY12) lacking URA3 sequences was constructed by replacement of the entire open reading frame of URA3 (13) with the HIS3 gene. Plasmid pJL202 (described above) was digested with Xho I and Not I and strain OD5 was transformed to histidine prototrophy. The complete deletion of URAS sequences in strain RSY12 was verified by Southern analysis. Growth, minimal medium, and sporulation medium were prepared as described (14, 15). Genetic and Molecular Techniques. Standard genetic techniques for sporulation, dissection of asci, and determination of phenotypes (15) and standard methods for molecular biology (16) were used except as noted below. Small-scale plasmid preparations from Escherichia coli were carried out as a modification of the boiling method (17). A rapid procedure for the preparation of small amounts of high molecular weight yeast DNA was used (18). The yeast transformation procedure was that described by Schiestl and Gietz (19). Cells from an overnight culture were resuspended in 300 ml of YPAD medium (15) and grown to a density of about 5 x 106 cells per ml. The cells were harvested by centrifugation and resuspended in 10 ml of sterile distilled water. The washed cells were harvested by centrifugation and resuspended in 1.5 ml of sterile TE/LiOAc (prepared by dilution of 1Ox concentrated stocks: 10x TE = 0.1 M Tris-HC/0.01 M EDTA, pH 7.5 and 1Ox LiOAc = 1 M LiOAc adjusted to pH 7.5 with dilute acetic acid). The cells were then incubated for 1 hr at 30'C with constant agitation. About 50 Al of plasmid DNA solution prepared as described below and 20 p.l of a solution of denatured salmon sperm DNA (10 mg/ml) were added to a Microfuge tube containing 200 ,.1 of cells in TE/LiOAc. In experiments 2-4, an aliquot of 10x concentrated stock of restriction enzyme buffer (described below) sufficient to yield a final 1 x concentration was also added. The resulting mixture was incubated for 30 min at 30'C with constant agitation. Sterile 40%o PEG 4000 (1.2 ml of 40%6 polyethylene glycol 4000 in lx TE/l x LiOAc) was then added, and the cells were incubated for 30 min at 30'C with agitation. The suspension was transferred to 420C for 15 min and then spun in a Microfuge for 5 sec. Cells were washed once with 0.5 ml of TE and resuspended in 1 ml of TE; 200 A.l of the suspension was plated on each selection plate. Plates were incubated at 30'C until colonies appeared. For experiment 1 (Table 1; BamHI present in the original transforming solution), 20 pug of CsCl-purified plasmid pM20 was treated with 100 units of BamHI (obtained from either New England Biolabs or Promega) in 150 mM NaCl/10 mM Tris HC, pH 7.9/10 mM MgCl2/1 mM dithiothreitol/100 pug of bovine serum albumin per ml (volume = 50 pl). After the DNA was completely digested, we added 5 p.l of 10x TE buffer and 5 ,1 of 10x LiOAc buffer. This solution was then added to the yeast cells as described above. For experiments
7586
Genetics: Schiestl and Petes
Table 1. Frequency of restriction enzyme-mediated (REM) events Ratio of REM events to total transformants* BamHI removed BamHI BamHI and readded§ absent* presentt Exp. 29/30 1 10/17 0/9 3/10 2 17/19 0/10 3 8/10 7/10 2/20 4
34/46 2/39 40/50 Totals *DNA was isolated from yeast transformants, treated with BamHI, and examined by Southern analysis. Transformants that had a URA3 BamHI fragment of 1.2 kb were considered to be the result of a REM event. tThe BamHI enzyme present in the original digest was included with the DNA during transformation. *The BamHI enzyme was removed by proteinase K/phenol extraction (experiments 2 and 3) or by heat treatment (experiment 4) prior to transformation. §The BamHI enzyme was removed by proteinase K/phenol extraction or by heat treatment. The enzyme was then readded to the purified BamHI fragments prior to transformation.
2 and 3, the amount of DNA and volume of the digestion buffer were scaled up by a factor of 3. After the plasmid DNA was digested, one-third of the sample was removed to be used in the transformation. The remaining two-thirds was precipitated with ethanol and resuspended in 200 Al of 0.01 M Tris'HCl, pH 7.5/0.05 M EDTA/1% SDS/100 ,ug of proteinase K per ml. After a 30-min incubation at 370C, the sample was extracted twice with phenol/chloroform/isoamyl alcohol, 25:24:1 (vol/vol), precipitated with ethanol, washed with 70% ethanol, and air-dried. The pellet was dissolved in water (40 pl). Half was added to a Microfuge tube containing 200 ,1 of yeast cells in TE/LiOAc buffer, carrier (salmon sperm) DNA, 100 units of BamHI enzyme, and the appropriate concentration of BamHI restriction buffer. The other half was added to an identical solution lacking the BamHI enzyme. The remainder of the transformation procedure was performed as described above. In experiment 4, after digestion of the plasmid DNA, the BamHI enzyme was inactivated by heat treatment (70TC for 10 min). The sample was then split in half. One aliquot was used to transform yeast without any further additions; BamHI (100 units) was added to the second aliquot before transformation. From several of the yeast transformants with a URA3 gene inserted at a nonhomologous site, the URA3 gene and its flanking DNA sequences were recovered. Genomic DNA from the transformants was treated with Bgi II and the resulting fragments were inserted into the BamHI site of the plasmid YEplacl81 (20). The pool of recombinant plasmids was transformed into the yeast strain RSY12, and Ura' colonies were selected (21). From the Ura' colonies, plasmid DNA was extracted (22) and used to transform E. coli to ampicillin-resistance. The sequences flanking the URA3 insertions were determined by using 18-base-pair (bp) primers homologous to the insertion 68 bp from the 5' (relative to URA3) junction (5'-CGGAGATTACCGAATCAA) and 42 bp from the 3' junction (5'-GAATCTCGGTCGTAATGA). Sequencing of double-stranded DNA was carried out with Sequenase (23) and a Bio-Rad sequencing cell. The resulting sequence information was employed to design oligonucleotides that could be used to amplify "target" sites (the site without the insertion). DNA was isolated from strain OD5 and, in separate experiments, pairs of primers designed to produce a DNA fragment 150 bp in length were used in the polymerase chain reaction (PCR method as described in ref. 24). Following amplification, the resulting
Proc. Natl. Acad. Sci. USA 88 (1991)
fragments were sequenced directly (protocol of United States Biochemical).
RESULTS Restriction Enzyme-Mediated Integration Events. The plasmid pM20 contains the yeast URA3 gene on a 1.2-kb BamHI fragment inserted into a pUC7 vector. This plasmid was treated with BamHI and transformed (19) in the presence of the BamHI enzyme into a haploid yeast strain (RSY12) that lacks URA3 sequences. About five Ura' transformants per ,gg of transforming DNA were obtained. Thirty were tested for stability of the Ura' phenotype; all were stable (10-7 frequency of Ura- derivatives). Fifteen of the transformants were crossed to a ura3 strain of opposite mating type. Five tetrads were dissected from each sporulated diploid. For all 15 transformants examined, most ofthe five tetrads examined segregated 2+:2-. This segregation pattern indicates that the wild-type URA3 gene is integrated into a single genomic site in each transformant. The same 15 transformants were crossed to a haploid strain with a wild-type URA3 gene located at its normal location (chromosome V). Each diploid strain segregated 4+:0-, 3+:1-, and 2+:2- tetrads in the ratios expected for unlinked markers. Spore viability was excellent (>90%o), indicating that these transformants did not contain chromosomal translocations or large inversions. DNA was isolated from the transformants and examined by Southern analysis. The hybridization probe was pM20. When the DNA was treated with EcoRI (which does not cut within the 1.2-kb URA3 fragment), we found that most transformants contained a single strong band of hybridization that was different in different transformants (Fig. 1). The URA3 gene in OD5 (a strain isogenic with RSY12 that contains URA3 at its normal location) was located on an EcoPJ fragment of 13 kb, as expected from previous studies (25). The URA3 deletion strain RSY12 lacked this band. The observation that most of the transformants contain a single strong band of hybridization at different positions suggests that most of the transformants have a single URA3 insertion and that the position of the insertions in the genome is different in different transformants. r
r
-
--
FIG. 1. Southern blot of DNA from the yeast strain OD5, an isogenic derivative of OD5 that lacks URA3 (RSY12), and Ura+ derivatives of RSY12 (B1-B5, S1-S5, T1-T5, T9, and T10) obtained by transformation with the URA3 BamHI fragment (in the presence of the BamHI enzyme); these transformants were from experiment 1 in Table 1. DNA from all strains was treated with EcoRI (which does not cut within the URA3 BamHI fragment) and hybridized with a 32P-labeled plasmid (pM20) containing URA3. The strain OD5 contains a single 13-kb fragment characteristic of the URA3 gene at its proper position on chromosome V. Strain RSY12 does not contain any strongly hybridizing fragment, and the Ura+ transformants show fragments of different sizes that hybridize to the URA3 probe, indicating that the URA3 fragment integrated into different genomic positions in different transformants.
Genetics: Schiestl and Petes
Proc. Natl. Acad. Sci. USA 88 (1991)
When the DNA of 30 transformants was digested with BamHI and Southern hybridization was carried out, most (29 of 30) had a single fragment of about 1.2 kb that hybridized to URA3 (Fig. 2). This result was unexpected because it indicates that the BamHI sites (G-G-A-T-C-C) at both junctions of the integrated URA3 gene were recreated upon insertion of the transforming fragment into the genome. To determine the mechanism of integration, we sequenced the junctions of nine different URA3 integrated fragments and obtained the "target" sequences (the genomic sequence prior to the integration event) for four of these (Fig. 3). Approximately 200 bp of genomic sequence was obtained for each junction and the GenBank data base was searched for homology. The genomic sequences in pRS137 [representing the insertion in RSY12(B5)] were identical to those of the SGAI gene (26). This "target" gene had aBamHI site at the position of insertion of the URA3 fragment. To test for the generality of this unexpected result, we sequenced three other target sites. We used the flanking sequences to design pairs of oligonucleotides with the appropriate sequences to yield a DNA fragment of about 150 bp containing the "target" after application of the polymerase chain reaction (PCR). Using these primers, we amplified the genomic target sites representing the insertions in the plasmids pRS124, pRS125, and pRS126. Sequence analysis indicated that all three target sites contained a BamHI site at the position of insertion of the URA3 fragment. In summary, in four offour transformants, the insertion of the BamHI URA3 fragment into the genome had the structure predicted for a conservative integration of the fragment into a BamHI site in the genome. To determine whether these integration events were catalyzed by BamHI, we examined the transformation properties of the URA3 fragment in the presence and the absence of BamHI. The efficiency of transformation was about 7-fold higher in the presence of the enzyme than in its absence (101 transformants versus 15 transformants; 20 pug of DNA in each transformation). The cell viability after transformation was unaffected by the presence of the enzyme. DNA samples were prepared from the transformants, treated with BamHI, and examined by Southern analysis. The frequencies of transformants with URA3 genes with two flanking BamHI sites (see Table 1) were: 80% (40 of 50) for BamHI URA3 fragments in the original BamHI digestion buffer (BamHI ~~~-
->
-X~
rZ
r-
'
-
7587
pRS124 RSY12(Bl)
GGCCACCTACACTACCTAG
CCGGTGGATGTGATG
GATCCGTTTAATGCTGATC GCAAATTACGACTAG
pRS125 RSY12(B2)
TTTACGGGATTTATG AAATGCCCTAAATACCTAG
GATCCGTCAGAACAAAAAG GCAGTCTTGTTTTTC
pRS126 RSY12(T3)
TTCATAGGAAAATGG AAGTATCCTTTTACCCTAG
GATCCATCCATCCCAACAA GTAGGTAGGGTTGTT
pRS137 RSY12(B5)
CATCTCTTGGCTCTG GTAGAGAACCGAGACCTAG
GATCCGTTATCTGTTCTGT
pRS138
GGCTTCTCTTTTCCG
GATCCGTGATTGCGGTACG
RSY12(B7)
CCGAAGAGAAAAGGCCTAG
GCACTAACGCCATGC
pRS139
RSY12(S4)
CTCCGGGAGCGGGCG GAGGCCCTCGCCCGCCTAG
GGTGATTGCCGGGTC
pRS 140 RSY12(85)
ACAAAAAAGTGGAGCCTAG
GATCCGCCCGACAATATCT GCGGGCTGTTATAGA
pRS141 RSY12(T5)
GGAAAGAAAATTTCG
CCTTTCTTTTAAAGCCTAG
GATCCGAAAATTTTTGGCG
TGTTTTTTCACCTCG
pRS142
GTGGCTACAGAATGG
RSY12(T6)
CACCGATGTCTTACCCTAG
GCAATAGACAAGACA
GCTTTTAAAAACCGC GATCCTGAATTTATAACTA GACTTAAATATTGAT
FIG. 3. DNA sequences from both junctions of the integrated BamHI URA3 fragments in nine different Ura+ transformants. The integrated URA3 genes from nine restriction enzyme-mediated transformants were cloned and sequenced. The letters and numbers denoting the transformants are indicated below the name of the plasmid containing the integrated gene. Primers homologous to sequences within the insert were used to obtain the sequence of the flanking genomic DNA. The boxed region represents sequences of the URA3 fragment (the 5' end of the gene is on the left side). For the flanking sequences, the gap represents the position of insertion of the URA3 gene. Although only 15 bp of flanking genomic DNA is shown, we obtained about 200 bp of flanking sequences from each junction. In the sequence of the URA3 fragment, 5 bp flanking the restriction site are underlined; in the flanking DNA sequences, bases homologous to these 5 bp are underlined. As discussed in the text, there is significant homology between the bases at the left side of the URA3 fragment and the right junction of the flanking sequences.
present), 5% (2 of 39) for BamHI URA3 fragments in the absence of BamHI enzyme, and 74% (34 of 46) for BamHI URA3 fragments with the BamHI enzyme removed and then added back. These results indicate that the integration of the BamHI URA3 fragment into BamHI sites in the genome is catalyzed by the BamHI restriction enzyme, a restriction
FIG. 2. Southern blot of DNA from yeast strains OD5, RSY12 (lacking URA3 sequences), and Ura+ transformants RSY12 (S6-SI, T1-T10) obtained with the URA3 BamHI fragment in the presence of BamHI enzyme (experiment 1 in Table 1). The genomic DNA was treated with BamHI and hybridized with 32P-labeled plasmid pM20 (containing URA3). The DNA of most of the transformants hybridized to a single fragment 1.2 kb in size, the same size as the original BamHI fragment used in transformation. The size of the genomic BamHI URA3 fragment in OD5 is about 5.5 kb, as expected (25).
enzyme-mediated event. The simplest explanation of the restriction enzymemediated event is that the BamHI enzyme enters the cells, perhaps in association with the transforming DNA. The enzyme cleaves the chromosomal DNA at BamHI sites, the cohesive ends of the URA3 fragment pair with those of the chromosomal DNA, and the resulting nicks are sealed by DNA ligase (Fig. 4a). An observation that complicates this simple model is that sequences adjacent to the BamHI site at the right junction (Fig. 3) appear nonrandomly related to sequences at the left end (the 5' end of URA3) of the BamHI URA3 fragment. Beginning after the terminal C in the BamHI site at the right junction, the loose consensus sequence is: G
7588
Genetics: Schiestl and Petes
(6 of 9 inserts), T (5 of 9), C (4 of 9), and A (5 of 9). The same four bases are present adjacent to the BamHI site at the left a
GATCC G
G URA3 URA3CCTAG
CCTAQG_
b 1
URA3 2
KZ)ATC
Proc. Natl. Acad. Sci. USA 88 (1991)
end of the URA3 fragment (Fig. 3). The plasmid pRS125 has the complete consensus at the 3' junction plus one additional base in common with the URA3 fragment. The likelihood of a random occurrence of this sequence (G-T-C-A-G) at the 3' junction (calculated based on the 39%o G+C content of yeast DNA) is (0.2)3(0.3)2 or 0.00072; the probability of observing this sequence in one of nine transformants is 0.0065. Integration Events in the Absence of BamHI: legitimate Integration. As described above, we found that the integration events of the BamHI URA3 fragments that occurred in the absence of the enzyme usually did not regenerate BamHI sites flanking the insertion (Table 1). To determine the nature of these enzyme-independent integration events, we sequenced three of the integrated fragments. We found that the G-A-T-C sequence at the ends of the URA3 fragment was conserved; however, there was a BamHI site at only one of the six junctions (Fig. 5). When the target sites for two of these integration events were determined, we observed that the BamHI URA3 fragment had integrated into G-A-T-C sequences in the chromosome. Since the G-A-T-C target sequence is the same as that of the cohesive end in BamHItreated DNA, these integration events have the structure expected for a conservative insertion of the BamHI URA3 fragment into genomic DNA with a staggered cleavage of a G-A-T-C site. These results indicate the existence of a recombination mechanism in yeast that utilizes extremely small (4 bp) sequence homologies (see Fig. 4b). Since recombination events involving little or no sequence homology were defined as "illegitimate" recombination (27), we call this type of event "illegitimate" integration.
DISCUSSION I~CA
3
I
Restriction Enzyme-Mediated Events. The observation of restriction enzyme-mediated events indicates that the BamHI restriction enzyme can enter the cell under the conditions used in the yeast transformation protocol. Although the entry of restriction enzymes into S. cerevisiae cells in vivo has not been described previously to our knowledge, it has been shown that, when the gene encoding EcoRI was expressed in yeast, the chromosomal DNA was digested, resulting in cell death (28). A number of studies indicate that restriction enzymes can enter mammalian cells that had been made permeable by electroporation (29). These enzymes
4
FIG. 4. Models for restriction enzyme-mediated (REM) (a) and illegitimate (b) integration events in yeast. (a) REM events. TheBamHI enzyme enters the cell together with the transforming URA3 BamHI DNA fragments. As discussed in the text, there is likely to be a homology search involving the 5' end of the URA3 fragment, followed by a metastable association between the chromosome and the transformingfragment. Following this association, the chromosomal DNA is cleaved by BamHI (which is possibly associated with the transforming DNA), and the cohesive ends of the URA3 BamHI fragment are ligated to the chromosomal BamHI ends. Other related mechanisms are also possible. For example, it is possible that the transforming DNA associates with the chromosomal DNA after the genomic sequences are cut by BamHI. The net result of the REM event is an integrated URA3 gene flanked by BamHI sites. (b) Illegitimate integration. The BamHI URA3 fragment enters the cell in the absence of BamHI. The 5' single-stranded G-A-T-C ends of the URA3 fragment (indicated by thin lines and thin letters) invade the chromosome (indicated by thick lines and bold letters), pairing at chromosomal G-A-T-C sites (step 2). The chromosomal DNA is nicked at two sites (indicated by arrows). The resulting 3' strands derived from the chromosome are ligated to the 5' strands derived from the URA3 fragment (step 3). The net result of this mechanism is an integrated URA3 sequence flanked by G-A-T-C-C and G-G-A-T-C sequences (step 4).
pRS159
RSY12(pM20
GATCAACTTTCAATTCCAA CCAACGATAATAACG TTGAAAGTTAAGGTT GGTTGCTATTATTGCCTAG
BIIB7)
pRS 161 RSY12(pM2O
TACCTACA&TTAAAT ATGGATGTTAATTTACTAG
GATCACAATjTATGGATGC TGTTACATACCTACG
BIIB9)
pRS162 RSY12(pM20
EE7)
GATCTAAATCAGATGCTGC GAAGCTGAACTAAAA ATTTAGTCTACGACG CTTCGACTTGATTTTCTAG
FIG. 5. DNA sequences from both junctions of illegitimate integration events of the BamHI URA3 fragment. The BamHI URA3 fragment was transformed into RSY12 in the absence of BamHI enzyme. The integrated URA3 gene and flanking DNA sequences were cloned from three transformants, and their sequences were analyzed. These sequences are depicted in the same way as those in Fig. 3. The target sites for the first two insertions were analyzed and found to contain a G-A-T-C sequence at the point of insertion. These sequences indicate a conservative integration of the BamHI fragment into a genomic G-A-T-C sequence.
Genetics: Schiestl and Petes reduced the viability of the cells and induced chromosomal aberrations. W. Morgan (personal communication) has observed deletions and rearrangements of restriction fragments contained within plasmids in mammalian cells as a result of the enzyme treatment. In E. coli, it has been shown that expression ofEcoRI can promote site-specific recombination
in vivo (30). The simplest explanation for restriction enzyme-mediated events is that the chromosomal DNA is broken at many BamHI sites by the BamHI enzyme. Although most of these breaks are religated without incorporation of transforming DNA, occasionally the BamHI URA3 gene is inserted into the chromosomal site. Since a single unrepaired chromosome break would be lethal, this mechanism would require yeast to be very proficient in the repair of BamHI-induced breaks, since no decrease in viability was found after transformation in the presence of BamHI. Potential uses of restriction
enzyme-mediated events include: insertional mutagenesis, generation of chromosome aberrations, stimulation of mitotic recombination at specific sites, and improvements in the efficiency of gene targeting in organisms with high levels of nonhomologous integration. In addition, since there is no reason to assume that the uptake of proteins during transformation is limited to restriction enzymes, one might be able to observe the effects of other proteins (topoisomerases, helicases, etc.) on recombination in vivo. Illegitimate Integration Events. Previous transformation studies in yeast indicated that most integration events were the result of recombination involving more than several hundred base pairs of sequence homology between the transforming DNA and the yeast chromosome (11, 12). It has been found that integration was not detectable when the transforming plasmids had only 31 and 41 bp of homology to the target chromosomal locus (31). A low frequency (
Integration of DNA fragments by illegitimate recombination in Saccharomyces cerevisiae (yeast/nonhomologous recomblnation/restrction enzymes)
ROBERT H. SCHIESTL AND THOMAS D. PETES Department of Biology, University of North Carolina, Chapel Hill, NC 27599
Communicated by David Botstein, May 17, 1991
DNA fragments (generated by BamHI treatABSTRACT ment) with no homology to the yeast genome were transformed into Saccharomyces cerevisiae. When the fragments were transformed in the presence of the BamfH enzyme, they integrated into genomic BamHI sites. When the fragments were transformed in the absence of the enzyme, they integrated into genomic G-A-T-C sites. Since the G-A-T-C sequence is present at the ends of BamiI fragments, this result indicates that four base pairs of homology are sufficient for some types of mitotic recombination.
Illegitimate recombination events join two DNA molecules (or two noncontiguous parts of a single DNA molecule) without the requirement for extended sequence homology. In a number of human genetic defects caused by genomic rearrangements (1-6), either no homology or very limited homology has been detected at the junctions of the rearrangements (7, 8). Similar recombination events have also been characterized in bacteria (9). In most organisms, nonhomologous integration of introduced DNA is more common than homologous integration. For example, in mammalian cells, only one in one thousand transformants contains the transforming DNA in a homologous position (10). In contrast, in Saccharomyces cerevisiae, all transformants analyzed contained the transforming DNA in homologous positions (11, 12). One interpretation of these results is that Saccharomyces cerevisiae has a very efficient mechanism of homologous recombination (integration) and an inefficient mechanism of nonhomologous recombination. Alternatively, it is possible that yeast has no mechanism allowing nonhomologous integration of transforming DNA. Below, we describe evidence indicating that the first of these alternatives is correct. In addition, we show that certain restriction enzymes can be transformed into yeast cells, and catalyze the integration of transforming DNA sequences.
MATERIALS AND METHODS Plasmids. The plasmid pJL202 was constructed by Joachim Li (The Johns Hopkins University) and obtained from Jef Boeke (The Johns Hopkins University). In this plasmid, the 1.1-kilobase (kb) HindIII URA3 fragment was replaced by the HIS3 gene, leaving sequences homologous to the URA3 flanking region on both sides of the HIS3 gene. The URA3 fragment used in our transformation experiments was derived from the plasmid pM20 constructed by Sue Jinks-Robertson (Emory University) and Martin Kupiec (Tel Aviv University). This plasmid was constructed by "filling-in" the cohesive ends of the 1.1-kb HindIII fragment containing the URA3 gene and inserting this fragment into the HincIl site in the polylinker of pUC7. Thus, the URA3 gene 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. §1734 solely to indicate this fact.
7585
in pM20 is flanked by BamHI sites and (outside of the BamHI sites) by EcoRI sites. Strains and Media. The yeast strain OD5 (MATa leu2-3,112 his3-11,15) was obtained from Walter Spevak (University of Vienna). An isogenic strain (RSY12) lacking URA3 sequences was constructed by replacement of the entire open reading frame of URA3 (13) with the HIS3 gene. Plasmid pJL202 (described above) was digested with Xho I and Not I and strain OD5 was transformed to histidine prototrophy. The complete deletion of URAS sequences in strain RSY12 was verified by Southern analysis. Growth, minimal medium, and sporulation medium were prepared as described (14, 15). Genetic and Molecular Techniques. Standard genetic techniques for sporulation, dissection of asci, and determination of phenotypes (15) and standard methods for molecular biology (16) were used except as noted below. Small-scale plasmid preparations from Escherichia coli were carried out as a modification of the boiling method (17). A rapid procedure for the preparation of small amounts of high molecular weight yeast DNA was used (18). The yeast transformation procedure was that described by Schiestl and Gietz (19). Cells from an overnight culture were resuspended in 300 ml of YPAD medium (15) and grown to a density of about 5 x 106 cells per ml. The cells were harvested by centrifugation and resuspended in 10 ml of sterile distilled water. The washed cells were harvested by centrifugation and resuspended in 1.5 ml of sterile TE/LiOAc (prepared by dilution of 1Ox concentrated stocks: 10x TE = 0.1 M Tris-HC/0.01 M EDTA, pH 7.5 and 1Ox LiOAc = 1 M LiOAc adjusted to pH 7.5 with dilute acetic acid). The cells were then incubated for 1 hr at 30'C with constant agitation. About 50 Al of plasmid DNA solution prepared as described below and 20 p.l of a solution of denatured salmon sperm DNA (10 mg/ml) were added to a Microfuge tube containing 200 ,.1 of cells in TE/LiOAc. In experiments 2-4, an aliquot of 10x concentrated stock of restriction enzyme buffer (described below) sufficient to yield a final 1 x concentration was also added. The resulting mixture was incubated for 30 min at 30'C with constant agitation. Sterile 40%o PEG 4000 (1.2 ml of 40%6 polyethylene glycol 4000 in lx TE/l x LiOAc) was then added, and the cells were incubated for 30 min at 30'C with agitation. The suspension was transferred to 420C for 15 min and then spun in a Microfuge for 5 sec. Cells were washed once with 0.5 ml of TE and resuspended in 1 ml of TE; 200 A.l of the suspension was plated on each selection plate. Plates were incubated at 30'C until colonies appeared. For experiment 1 (Table 1; BamHI present in the original transforming solution), 20 pug of CsCl-purified plasmid pM20 was treated with 100 units of BamHI (obtained from either New England Biolabs or Promega) in 150 mM NaCl/10 mM Tris HC, pH 7.9/10 mM MgCl2/1 mM dithiothreitol/100 pug of bovine serum albumin per ml (volume = 50 pl). After the DNA was completely digested, we added 5 p.l of 10x TE buffer and 5 ,1 of 10x LiOAc buffer. This solution was then added to the yeast cells as described above. For experiments
7586
Genetics: Schiestl and Petes
Table 1. Frequency of restriction enzyme-mediated (REM) events Ratio of REM events to total transformants* BamHI removed BamHI BamHI and readded§ absent* presentt Exp. 29/30 1 10/17 0/9 3/10 2 17/19 0/10 3 8/10 7/10 2/20 4
34/46 2/39 40/50 Totals *DNA was isolated from yeast transformants, treated with BamHI, and examined by Southern analysis. Transformants that had a URA3 BamHI fragment of 1.2 kb were considered to be the result of a REM event. tThe BamHI enzyme present in the original digest was included with the DNA during transformation. *The BamHI enzyme was removed by proteinase K/phenol extraction (experiments 2 and 3) or by heat treatment (experiment 4) prior to transformation. §The BamHI enzyme was removed by proteinase K/phenol extraction or by heat treatment. The enzyme was then readded to the purified BamHI fragments prior to transformation.
2 and 3, the amount of DNA and volume of the digestion buffer were scaled up by a factor of 3. After the plasmid DNA was digested, one-third of the sample was removed to be used in the transformation. The remaining two-thirds was precipitated with ethanol and resuspended in 200 Al of 0.01 M Tris'HCl, pH 7.5/0.05 M EDTA/1% SDS/100 ,ug of proteinase K per ml. After a 30-min incubation at 370C, the sample was extracted twice with phenol/chloroform/isoamyl alcohol, 25:24:1 (vol/vol), precipitated with ethanol, washed with 70% ethanol, and air-dried. The pellet was dissolved in water (40 pl). Half was added to a Microfuge tube containing 200 ,1 of yeast cells in TE/LiOAc buffer, carrier (salmon sperm) DNA, 100 units of BamHI enzyme, and the appropriate concentration of BamHI restriction buffer. The other half was added to an identical solution lacking the BamHI enzyme. The remainder of the transformation procedure was performed as described above. In experiment 4, after digestion of the plasmid DNA, the BamHI enzyme was inactivated by heat treatment (70TC for 10 min). The sample was then split in half. One aliquot was used to transform yeast without any further additions; BamHI (100 units) was added to the second aliquot before transformation. From several of the yeast transformants with a URA3 gene inserted at a nonhomologous site, the URA3 gene and its flanking DNA sequences were recovered. Genomic DNA from the transformants was treated with Bgi II and the resulting fragments were inserted into the BamHI site of the plasmid YEplacl81 (20). The pool of recombinant plasmids was transformed into the yeast strain RSY12, and Ura' colonies were selected (21). From the Ura' colonies, plasmid DNA was extracted (22) and used to transform E. coli to ampicillin-resistance. The sequences flanking the URA3 insertions were determined by using 18-base-pair (bp) primers homologous to the insertion 68 bp from the 5' (relative to URA3) junction (5'-CGGAGATTACCGAATCAA) and 42 bp from the 3' junction (5'-GAATCTCGGTCGTAATGA). Sequencing of double-stranded DNA was carried out with Sequenase (23) and a Bio-Rad sequencing cell. The resulting sequence information was employed to design oligonucleotides that could be used to amplify "target" sites (the site without the insertion). DNA was isolated from strain OD5 and, in separate experiments, pairs of primers designed to produce a DNA fragment 150 bp in length were used in the polymerase chain reaction (PCR method as described in ref. 24). Following amplification, the resulting
Proc. Natl. Acad. Sci. USA 88 (1991)
fragments were sequenced directly (protocol of United States Biochemical).
RESULTS Restriction Enzyme-Mediated Integration Events. The plasmid pM20 contains the yeast URA3 gene on a 1.2-kb BamHI fragment inserted into a pUC7 vector. This plasmid was treated with BamHI and transformed (19) in the presence of the BamHI enzyme into a haploid yeast strain (RSY12) that lacks URA3 sequences. About five Ura' transformants per ,gg of transforming DNA were obtained. Thirty were tested for stability of the Ura' phenotype; all were stable (10-7 frequency of Ura- derivatives). Fifteen of the transformants were crossed to a ura3 strain of opposite mating type. Five tetrads were dissected from each sporulated diploid. For all 15 transformants examined, most ofthe five tetrads examined segregated 2+:2-. This segregation pattern indicates that the wild-type URA3 gene is integrated into a single genomic site in each transformant. The same 15 transformants were crossed to a haploid strain with a wild-type URA3 gene located at its normal location (chromosome V). Each diploid strain segregated 4+:0-, 3+:1-, and 2+:2- tetrads in the ratios expected for unlinked markers. Spore viability was excellent (>90%o), indicating that these transformants did not contain chromosomal translocations or large inversions. DNA was isolated from the transformants and examined by Southern analysis. The hybridization probe was pM20. When the DNA was treated with EcoRI (which does not cut within the 1.2-kb URA3 fragment), we found that most transformants contained a single strong band of hybridization that was different in different transformants (Fig. 1). The URA3 gene in OD5 (a strain isogenic with RSY12 that contains URA3 at its normal location) was located on an EcoPJ fragment of 13 kb, as expected from previous studies (25). The URA3 deletion strain RSY12 lacked this band. The observation that most of the transformants contain a single strong band of hybridization at different positions suggests that most of the transformants have a single URA3 insertion and that the position of the insertions in the genome is different in different transformants. r
r
-
--
FIG. 1. Southern blot of DNA from the yeast strain OD5, an isogenic derivative of OD5 that lacks URA3 (RSY12), and Ura+ derivatives of RSY12 (B1-B5, S1-S5, T1-T5, T9, and T10) obtained by transformation with the URA3 BamHI fragment (in the presence of the BamHI enzyme); these transformants were from experiment 1 in Table 1. DNA from all strains was treated with EcoRI (which does not cut within the URA3 BamHI fragment) and hybridized with a 32P-labeled plasmid (pM20) containing URA3. The strain OD5 contains a single 13-kb fragment characteristic of the URA3 gene at its proper position on chromosome V. Strain RSY12 does not contain any strongly hybridizing fragment, and the Ura+ transformants show fragments of different sizes that hybridize to the URA3 probe, indicating that the URA3 fragment integrated into different genomic positions in different transformants.
Genetics: Schiestl and Petes
Proc. Natl. Acad. Sci. USA 88 (1991)
When the DNA of 30 transformants was digested with BamHI and Southern hybridization was carried out, most (29 of 30) had a single fragment of about 1.2 kb that hybridized to URA3 (Fig. 2). This result was unexpected because it indicates that the BamHI sites (G-G-A-T-C-C) at both junctions of the integrated URA3 gene were recreated upon insertion of the transforming fragment into the genome. To determine the mechanism of integration, we sequenced the junctions of nine different URA3 integrated fragments and obtained the "target" sequences (the genomic sequence prior to the integration event) for four of these (Fig. 3). Approximately 200 bp of genomic sequence was obtained for each junction and the GenBank data base was searched for homology. The genomic sequences in pRS137 [representing the insertion in RSY12(B5)] were identical to those of the SGAI gene (26). This "target" gene had aBamHI site at the position of insertion of the URA3 fragment. To test for the generality of this unexpected result, we sequenced three other target sites. We used the flanking sequences to design pairs of oligonucleotides with the appropriate sequences to yield a DNA fragment of about 150 bp containing the "target" after application of the polymerase chain reaction (PCR). Using these primers, we amplified the genomic target sites representing the insertions in the plasmids pRS124, pRS125, and pRS126. Sequence analysis indicated that all three target sites contained a BamHI site at the position of insertion of the URA3 fragment. In summary, in four offour transformants, the insertion of the BamHI URA3 fragment into the genome had the structure predicted for a conservative integration of the fragment into a BamHI site in the genome. To determine whether these integration events were catalyzed by BamHI, we examined the transformation properties of the URA3 fragment in the presence and the absence of BamHI. The efficiency of transformation was about 7-fold higher in the presence of the enzyme than in its absence (101 transformants versus 15 transformants; 20 pug of DNA in each transformation). The cell viability after transformation was unaffected by the presence of the enzyme. DNA samples were prepared from the transformants, treated with BamHI, and examined by Southern analysis. The frequencies of transformants with URA3 genes with two flanking BamHI sites (see Table 1) were: 80% (40 of 50) for BamHI URA3 fragments in the original BamHI digestion buffer (BamHI ~~~-
->
-X~
rZ
r-
'
-
7587
pRS124 RSY12(Bl)
GGCCACCTACACTACCTAG
CCGGTGGATGTGATG
GATCCGTTTAATGCTGATC GCAAATTACGACTAG
pRS125 RSY12(B2)
TTTACGGGATTTATG AAATGCCCTAAATACCTAG
GATCCGTCAGAACAAAAAG GCAGTCTTGTTTTTC
pRS126 RSY12(T3)
TTCATAGGAAAATGG AAGTATCCTTTTACCCTAG
GATCCATCCATCCCAACAA GTAGGTAGGGTTGTT
pRS137 RSY12(B5)
CATCTCTTGGCTCTG GTAGAGAACCGAGACCTAG
GATCCGTTATCTGTTCTGT
pRS138
GGCTTCTCTTTTCCG
GATCCGTGATTGCGGTACG
RSY12(B7)
CCGAAGAGAAAAGGCCTAG
GCACTAACGCCATGC
pRS139
RSY12(S4)
CTCCGGGAGCGGGCG GAGGCCCTCGCCCGCCTAG
GGTGATTGCCGGGTC
pRS 140 RSY12(85)
ACAAAAAAGTGGAGCCTAG
GATCCGCCCGACAATATCT GCGGGCTGTTATAGA
pRS141 RSY12(T5)
GGAAAGAAAATTTCG
CCTTTCTTTTAAAGCCTAG
GATCCGAAAATTTTTGGCG
TGTTTTTTCACCTCG
pRS142
GTGGCTACAGAATGG
RSY12(T6)
CACCGATGTCTTACCCTAG
GCAATAGACAAGACA
GCTTTTAAAAACCGC GATCCTGAATTTATAACTA GACTTAAATATTGAT
FIG. 3. DNA sequences from both junctions of the integrated BamHI URA3 fragments in nine different Ura+ transformants. The integrated URA3 genes from nine restriction enzyme-mediated transformants were cloned and sequenced. The letters and numbers denoting the transformants are indicated below the name of the plasmid containing the integrated gene. Primers homologous to sequences within the insert were used to obtain the sequence of the flanking genomic DNA. The boxed region represents sequences of the URA3 fragment (the 5' end of the gene is on the left side). For the flanking sequences, the gap represents the position of insertion of the URA3 gene. Although only 15 bp of flanking genomic DNA is shown, we obtained about 200 bp of flanking sequences from each junction. In the sequence of the URA3 fragment, 5 bp flanking the restriction site are underlined; in the flanking DNA sequences, bases homologous to these 5 bp are underlined. As discussed in the text, there is significant homology between the bases at the left side of the URA3 fragment and the right junction of the flanking sequences.
present), 5% (2 of 39) for BamHI URA3 fragments in the absence of BamHI enzyme, and 74% (34 of 46) for BamHI URA3 fragments with the BamHI enzyme removed and then added back. These results indicate that the integration of the BamHI URA3 fragment into BamHI sites in the genome is catalyzed by the BamHI restriction enzyme, a restriction
FIG. 2. Southern blot of DNA from yeast strains OD5, RSY12 (lacking URA3 sequences), and Ura+ transformants RSY12 (S6-SI, T1-T10) obtained with the URA3 BamHI fragment in the presence of BamHI enzyme (experiment 1 in Table 1). The genomic DNA was treated with BamHI and hybridized with 32P-labeled plasmid pM20 (containing URA3). The DNA of most of the transformants hybridized to a single fragment 1.2 kb in size, the same size as the original BamHI fragment used in transformation. The size of the genomic BamHI URA3 fragment in OD5 is about 5.5 kb, as expected (25).
enzyme-mediated event. The simplest explanation of the restriction enzymemediated event is that the BamHI enzyme enters the cells, perhaps in association with the transforming DNA. The enzyme cleaves the chromosomal DNA at BamHI sites, the cohesive ends of the URA3 fragment pair with those of the chromosomal DNA, and the resulting nicks are sealed by DNA ligase (Fig. 4a). An observation that complicates this simple model is that sequences adjacent to the BamHI site at the right junction (Fig. 3) appear nonrandomly related to sequences at the left end (the 5' end of URA3) of the BamHI URA3 fragment. Beginning after the terminal C in the BamHI site at the right junction, the loose consensus sequence is: G
7588
Genetics: Schiestl and Petes
(6 of 9 inserts), T (5 of 9), C (4 of 9), and A (5 of 9). The same four bases are present adjacent to the BamHI site at the left a
GATCC G
G URA3 URA3CCTAG
CCTAQG_
b 1
URA3 2
KZ)ATC
Proc. Natl. Acad. Sci. USA 88 (1991)
end of the URA3 fragment (Fig. 3). The plasmid pRS125 has the complete consensus at the 3' junction plus one additional base in common with the URA3 fragment. The likelihood of a random occurrence of this sequence (G-T-C-A-G) at the 3' junction (calculated based on the 39%o G+C content of yeast DNA) is (0.2)3(0.3)2 or 0.00072; the probability of observing this sequence in one of nine transformants is 0.0065. Integration Events in the Absence of BamHI: legitimate Integration. As described above, we found that the integration events of the BamHI URA3 fragments that occurred in the absence of the enzyme usually did not regenerate BamHI sites flanking the insertion (Table 1). To determine the nature of these enzyme-independent integration events, we sequenced three of the integrated fragments. We found that the G-A-T-C sequence at the ends of the URA3 fragment was conserved; however, there was a BamHI site at only one of the six junctions (Fig. 5). When the target sites for two of these integration events were determined, we observed that the BamHI URA3 fragment had integrated into G-A-T-C sequences in the chromosome. Since the G-A-T-C target sequence is the same as that of the cohesive end in BamHItreated DNA, these integration events have the structure expected for a conservative insertion of the BamHI URA3 fragment into genomic DNA with a staggered cleavage of a G-A-T-C site. These results indicate the existence of a recombination mechanism in yeast that utilizes extremely small (4 bp) sequence homologies (see Fig. 4b). Since recombination events involving little or no sequence homology were defined as "illegitimate" recombination (27), we call this type of event "illegitimate" integration.
DISCUSSION I~CA
3
I
Restriction Enzyme-Mediated Events. The observation of restriction enzyme-mediated events indicates that the BamHI restriction enzyme can enter the cell under the conditions used in the yeast transformation protocol. Although the entry of restriction enzymes into S. cerevisiae cells in vivo has not been described previously to our knowledge, it has been shown that, when the gene encoding EcoRI was expressed in yeast, the chromosomal DNA was digested, resulting in cell death (28). A number of studies indicate that restriction enzymes can enter mammalian cells that had been made permeable by electroporation (29). These enzymes
4
FIG. 4. Models for restriction enzyme-mediated (REM) (a) and illegitimate (b) integration events in yeast. (a) REM events. TheBamHI enzyme enters the cell together with the transforming URA3 BamHI DNA fragments. As discussed in the text, there is likely to be a homology search involving the 5' end of the URA3 fragment, followed by a metastable association between the chromosome and the transformingfragment. Following this association, the chromosomal DNA is cleaved by BamHI (which is possibly associated with the transforming DNA), and the cohesive ends of the URA3 BamHI fragment are ligated to the chromosomal BamHI ends. Other related mechanisms are also possible. For example, it is possible that the transforming DNA associates with the chromosomal DNA after the genomic sequences are cut by BamHI. The net result of the REM event is an integrated URA3 gene flanked by BamHI sites. (b) Illegitimate integration. The BamHI URA3 fragment enters the cell in the absence of BamHI. The 5' single-stranded G-A-T-C ends of the URA3 fragment (indicated by thin lines and thin letters) invade the chromosome (indicated by thick lines and bold letters), pairing at chromosomal G-A-T-C sites (step 2). The chromosomal DNA is nicked at two sites (indicated by arrows). The resulting 3' strands derived from the chromosome are ligated to the 5' strands derived from the URA3 fragment (step 3). The net result of this mechanism is an integrated URA3 sequence flanked by G-A-T-C-C and G-G-A-T-C sequences (step 4).
pRS159
RSY12(pM20
GATCAACTTTCAATTCCAA CCAACGATAATAACG TTGAAAGTTAAGGTT GGTTGCTATTATTGCCTAG
BIIB7)
pRS 161 RSY12(pM2O
TACCTACA&TTAAAT ATGGATGTTAATTTACTAG
GATCACAATjTATGGATGC TGTTACATACCTACG
BIIB9)
pRS162 RSY12(pM20
EE7)
GATCTAAATCAGATGCTGC GAAGCTGAACTAAAA ATTTAGTCTACGACG CTTCGACTTGATTTTCTAG
FIG. 5. DNA sequences from both junctions of illegitimate integration events of the BamHI URA3 fragment. The BamHI URA3 fragment was transformed into RSY12 in the absence of BamHI enzyme. The integrated URA3 gene and flanking DNA sequences were cloned from three transformants, and their sequences were analyzed. These sequences are depicted in the same way as those in Fig. 3. The target sites for the first two insertions were analyzed and found to contain a G-A-T-C sequence at the point of insertion. These sequences indicate a conservative integration of the BamHI fragment into a genomic G-A-T-C sequence.
Genetics: Schiestl and Petes reduced the viability of the cells and induced chromosomal aberrations. W. Morgan (personal communication) has observed deletions and rearrangements of restriction fragments contained within plasmids in mammalian cells as a result of the enzyme treatment. In E. coli, it has been shown that expression ofEcoRI can promote site-specific recombination
in vivo (30). The simplest explanation for restriction enzyme-mediated events is that the chromosomal DNA is broken at many BamHI sites by the BamHI enzyme. Although most of these breaks are religated without incorporation of transforming DNA, occasionally the BamHI URA3 gene is inserted into the chromosomal site. Since a single unrepaired chromosome break would be lethal, this mechanism would require yeast to be very proficient in the repair of BamHI-induced breaks, since no decrease in viability was found after transformation in the presence of BamHI. Potential uses of restriction
enzyme-mediated events include: insertional mutagenesis, generation of chromosome aberrations, stimulation of mitotic recombination at specific sites, and improvements in the efficiency of gene targeting in organisms with high levels of nonhomologous integration. In addition, since there is no reason to assume that the uptake of proteins during transformation is limited to restriction enzymes, one might be able to observe the effects of other proteins (topoisomerases, helicases, etc.) on recombination in vivo. Illegitimate Integration Events. Previous transformation studies in yeast indicated that most integration events were the result of recombination involving more than several hundred base pairs of sequence homology between the transforming DNA and the yeast chromosome (11, 12). It has been found that integration was not detectable when the transforming plasmids had only 31 and 41 bp of homology to the target chromosomal locus (31). A low frequency (
