Sgs1 and Exo1 suppress targeted chromosome duplication during ends-in and ends-out gene targeting

Abstract

Gene targeting is extremely efficient in the yeast Saccharomyces cerevisiae. It is performed by transformation with a linear, non-replicative DNA fragment carrying a selectable marker and containing ends homologous to the particular locus in a genome. However, even in S. cerevisiae, transformation can result in unwanted (aberrant) integration events, the frequency and spectra of which are quite different for ends-out and ends-in transformation assays. It has been observed that gene replacement (ends-out gene targeting) can result in illegitimate integration, integration of the transforming DNA fragment next to the target sequence and duplication of a targeted chromosome. By contrast, plasmid integration (ends-in gene targeting) is often associated with multiple targeted integration events but illegitimate integration is extremely rare and a targeted chromosome duplication has not been reported. Here we systematically investigated the influence of design of the ends-out assay on the success of targeted genetic modification. We have determined transformation efficiency, fidelity of gene targeting and spectra of all aberrant events in several ends-out gene targeting assays designed to insert, delete or replace a particular sequence in the targeted region of the yeast genome. Furthermore, we have demonstrated for the first time that targeted chromosome duplications occur even during ends-in gene targeting. Most importantly, the whole chromosome duplication is POL32 dependent pointing to break-induced replication (BIR) as the underlying mechanism. Moreover, the occurrence of duplication of the targeted chromosome was strikingly increased in the exo1Δ sgs1Δ double mutant but not in the respective single mutants demonstrating that the Exo1 and Sgs1 proteins independently suppress whole chromosome duplication during gene targeting.

1. INTRODUCTION

A linear DNA fragment introduced in the cell of the yeast Saccharomyces cerevisiae undergoes recombination with the homologous region in the genome (homologous recombination, HR). In order to perform gene targeting, yeast cells are transformed with a non-replicative DNA fragments carrying a selectable marker and containing ends homologous to the particular locus in the genome. These transforming DNA fragments are either (i) linear fragments used to replace the targeted sequence (gene replacement) [1] or (ii) linearized plasmids that integrate in the targeted locus (plasmid integration) [2, 3]. Although both gene replacement and plasmid integration require proteins involved in HR [2–12], these two processes are mechanistically quite different [13]. After homologous pairing of the transforming DNA fragment and targeted sequence, the ends of the transforming DNA fragment point away from each other during gene replacement (ends-out gene targeting, EOGT) whereas during plasmid integration they point toward each other (ends-in gene targeting, EIGT).

In EIGT, the double-strand break (DSB) on the linearized plasmid is almost always repaired by HR resulting in targeted plasmid integration, whereas the frequency of random (non-targeted) integration due to illegitimate recombination could be estimated to be as low as 0.1 % [14]. Moreover, illegitimate integration, including homology-assisted illegitimate recombination [15, 16] is the only aberrant event observed so far. Contrary to EIGT, the spectrum of aberrant events during EOGT is wider. Apart from (i) random integration by illegitimate recombination; transformation can result in (ii) integration of the transforming DNA fragment next to the homologous target, generating tandem repeats and (iii) targeted chromosome duplication (TCD) detected in aberrant transformants containing both, allele expected after a successful gene targeting and an untransformed allele (heteroallelic transformants) [14]. Duplication of a targeted chromosome can occur either by extensive DNA synthesis from the 3′-ends of the transforming fragment to the ends of the targeted chromosome by break-induced replication (BIR) [17, 18] or by chromosome missegregation [19].

The fidelity of gene targeting and spectra of aberrant events during EOGT can be influenced by different parameters such as the length of the flanking homologies used for targeting [20–22] and GC content of the targeted region [23]. Although as few as 40 bp at each end ensures gene targeting, the success of targeted events increases with the length of flanking homologies [20–23]. Various studies using different eukaryotic model organisms suggest that the fidelity of EOGT is governed by the same rules. Transformation of the protozoa Trypanosoma brucei frequently occurs due to HR and fidelity of EOGT depends on the length of flanking homology [24–26]. A high fidelity of gene targeting (percentage of a successful gene targeting) has also been reported for the moss Physcomitrella patens [27, 28]. Moreover, when flanking homologies are not of the same length, the transforming DNA fragment integrates next to the sequence sharing longer homology, frequently creating tandem repeats as previously reported in the moss [29], mouse cell lines [30] and yeast [14].

Gene targeting is managed by a number of proteins also involved in repair of chromosomal DSB by HR. DSB repair by HR starts by processing the DNA ends by 5′-3′ nuclease activity producing 3′-single stranded DNA tails. DNA resection is performed by several proteins including the MRX/Sae2 complex, Exo1 exonuclease, Sgs1 helicase and Dna2 endonuclease [31–33]. In the absence of Exo1 and Sgs1 only short 3′ single stranded DNA tails form and extensive resection is prevented. Exo1 is also involved in mismatch repair [34], survival of telomerase deficient cells [35] processing of the stalled replication forks [36] and its inactivation increases the frequency of gap repair and crossing over in the transformation assay [9]. Sgs1, a homolog of Escherichia coli RecQ helicase [37], functions in extensive resection of DNA ends and dissolves double Holliday junctions into non crossovers [38, 39]. Inactivation of the SGS1 gene increases the efficiency of transformation in both EOGT and EIGT assay [12, 40–42]. Langston and Symington [12] proposed that Sgs1 inhibits ends-out recombination not by heteroduplex rejection, as reported during single strand annealing [43], but by unwinding flanking sequences homologously paired with their genomic target. It has been shown that inactivation of both extensive DNA resection pathways, in the exo1Δ sgs1Δ double mutant, results in the increased transformation efficiency, using either the transforming DNA fragments for EOGT or chromosome fragments, presumably due to decreased degradation of the linear DNA [42, 44]. Furthermore, the exo1Δ sgs1Δ double mutant shows an increased frequency of de novo telomere addition [42, 44, 45]. Therefore, inactivation of EXO1 and SGS1 genes could influence not only the efficiency of transformation [41, 42] but also the fidelity of EOGT, as observed by Chung et al [42], and spectrum of aberrant transformation events. Moreover, the proportion of all transformation outcomes, including targeted chromosome duplication, could be influenced by the design of the gene-targeting assay.

DNA replication initiated at origins of replication during S-phase, as well as by BIR at DSBs, requires several DNA polymerases [reviewed in 46]. It has been shown that BIR is dependent on Pol32, a non-essential subunit of the Pol δ polymerase complex [47]. Therefore, the pol32 mutation could be used to distinguish whether chromosome duplications observed during gene targeting occurs by BIR or chromosome missegregation.

In this study we followed transformation efficiency, fidelity of gene targeting and spectra of aberrant events in five EOGT and one EIGT assay. Heteroallelic transformants having both a transformed and an untransformed, targeted allele that arose due to TCD were ubiquitous aberrant events in all EOGT assays but, for the first time, they were also detected in the EIGT assay. We show that the appearance of TCD during EOGT is POL32 dependent pointing to BIR as the underlying mechanism. Inactivation of EXO1 and SGS1 genes synergistically increased the transformation efficiency, during both EOGT and EIGT, as reported previously [41, 42]. However, this effect was associated with a striking decrease in the fidelity of gene targeting due to increase of TCD. In addition, frequency of TCD was not increased in respective single mutants demonstrating that Exo1 and Sgs1 proteins independently suppress gene targeting-associated chromosome duplication.

2. MATERIALS AND METHODS

2.1. Plasmids and the transforming DNA fragments

All plasmids used for yeast transformation are constructed in this study. Plasmid pLUH is a derivate of pLS42 [48] carrying LEU2, URA3 and HIS3 genes and it was used as a circular replicative control in transformation experiments. All other non-replicative (integrative) plasmids used for generation of the DNA fragments for transformation are listed in Table 1. The pGU2 and pAGU2 plasmids differ only by the RG region, the central part of the ARG4 gene (539 bp, EcoRV-Bsp1407I end-filled fragment) inserted in the NaeI site of the plasmid pGU2 backbone creating pAGU2. The plasmid pGU2 is a derivative of the pGEM-7ZF(−) (Promega) plasmid. It is constructed by deleting and end-filling EcoRI-HindIII fragment from the multiple cloning site (MCS) and Kpn2I-MluI region from the backbone. Resulting plasmid was used to create pGU2 by inserting the URA3 fragment in the end-filled BamHI site. The URA3 fragment was S1-treated 1638 bp EcoRI-EcoRI fragment from pCR2.1.TOPO-URA3 which was obtained by insertion of 1621 bp URA3 PCR fragment (primers: ZG1 5′ gggaagacaagcaacgaaac 3′ and ZG4 5′ ttcattggttctggcgaggt 3′) in the plasmid pCR2.1.TOPO.

Table 1. Plasmids carrying DNA fragments used either for a strain construction or as the transforming DNA fragments in gene targeting assays.

Alleles used in EOGT assays are shown in bold. RG region denotes the central part of the ARG4 gene.

Plasmid	Contains yeast regions	Use
pGU2	URA3	Construction of strains used in EOGT assays 1 and 2 (Figure 1A)
pLGU+4*	ura3::NcoI: URA3 gene inactivated by end-filling of the NcoI site, LEU2: fragment inserted in the NaeI site of the plasmid backbone	Construction of strain used in EOGT ends-out assay 4 (Figure 1A)
pAGU2	URA3, RG: central part of ARG4 gene inserted in the NaeI site of the plasmid backbone	Generation of the transforming DNA fragment used in EOGT assays 4 and 5 (Figure 1A)
pAUHI**	ura3::HIS3: HIS3 gene (1771 bp) inserted in the StuI site within the URA3 gene, RG	Construction of strains used in EOGT assay 3 and 5 (Figure 1A)
pAULI**	ura3::LEU2: LEU2 gene (1815 bp) inserted in the StuI site within the URA3 gene, RG	Generation of the transforming DNA fragment used in EOGT assays 1 and 3 (Figure 1A)
pADULI**	ura3Δ::LEU2: EcoRV-StuI region (248 bp) within the URA3 gene is replaced with LEU2 gene (1815 bp), RG	Generation of the transforming DNA fragment used in EOGT assay 2 (Figure 1A)
pAGUS**	URA3: shorter URA3 region replaces PvuII-PvuII region of pAGU2, RG	The transforming DNA fragment used in EIGT assay (Figure 2A) after linearization with SacI within RG region

^*pGU2 derivates;

^**pAGU2 derivates

The plasmid pLGU+4 was constructed by inactivation of the NcoI site in the URA3 gene by cutting and end-filling. Additionally, NcoI-MroI cut and end-filled LEU2 PCR fragment (1815 bp, primers: YUPL5 5′-atataCCATGGcattatttttttcctcaacat-3′ and YDOL3 5′-atataTCCGGAgtgttttttatttgttgtat-3′, containing NcoI and MroI sites, respectively) was inserted in the NaeI site of the plasmid backbone. The pAUHI and pAULI plasmids were constructed by inserting the BamHI-BamHI end-filled HIS3 gene (1771 bp) and NcoI-MroI end-filled LEU2 gene (1815 bp) in the StuI site in the URA3 gene of the plasmid pAGU2, respectively. The plasmid pADULI was constructed by replacing the StuI-EcoRV fragment from the URA3 gene (248 bp) of the plasmid pAGU2 with the NcoI-MroI end-filled LEU2 gene (1815 bp).

Plasmid pAGUS was constructed by replacing PvuII-PvuII fragment (2038 bp) containing 1638 bp URA3 region of the plasmid pAGU2 with the SmaI-SmaI fragment (1104 bp) carrying shorter URA3 region from the plasmid pTZGU [14]. Standard media and procedures were used for the cultivation of the E. coli strains (DH5α and XL1blue) and DNA manipulations [49].

2.2. Yeast strains

S. cerevisiae strains used in this study are derivatives of BY4741 and BY4742 [50] and are listed in Table 2. All modifications of the URA3 region in different strains were made using the plasmids listed in Table 1. Haploid α-mating type strains were transformed with the transforming DNA fragment carrying specific URA3 allele and successful modification was confirmed by Southern blotting. Strains were mated with a-mating type strain and dissected, using only a-mating type haploids in further experiments.

Table 2.

Yeast strains used in this study

Strain name	Genotype	Reference
BY4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0	50
BY4742	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0	50
Y11809	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 exo1::kanMX4	EUROSCARF
Y10775	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 sgs1::kanMX4	EUROSCARF
BYαURA3	MATα his3Δ1 leu2Δ0 lys2Δ0 URA3	This study
BYaURA3	MATa his3Δ1 leu2Δ0 met15Δ0 URA3	This study
BYauN	MATa his3Δ1 leu2Δ0 lys2Δ0 ura3::NcoI	This study
BYauH	MATa his3Δ1 leu2Δ0 met15Δ0 ura3::HIS3	This study
BYa exo1Δ	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 exo1Δ::kanMX4	This study
BYauH exo1Δ	MATa his3Δ1 leu2Δ0 met15Δ0 ura3::HIS3 exo1Δ::kanMX4	This study
BYαuH exo1Δ	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3::HIS3 exo1Δ::kanMX4	This study
BYauH sgs1Δ	MATa his3Δ1 leu2Δ0 met15Δ0 ura3::HIS3 sgs1Δ::kanMX4	This study
BYa exo1Δ sgs1Δ	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 exo1Δ::kanMX4 sgs1Δ::kanMX4	This study
BYaU exo1Δ sgs1Δ	MATa his3Δ1 leu2Δ0 met15Δ0 exo1Δ::kanMX4 sgs1Δ::kanMX4	This study
BYauH exo1Δ sgs1Δ	MATa his3Δ1 leu2Δ0 met15Δ0 ura3::HIS3 exo1Δ::kanMX4 sgs1Δ::kanMX4	This study
BYauH pol32Δ	MATa his3Δ1 leu2Δ0 met15Δ0 ura3::HIS3 pol32Δ::kanMX4	This study

The yeast strain BYαURA3 was constructed by transforming the strain BY4742 with EcoRI-EcoRI fragment carrying URA3 region from pGU2 (1638 bp). The strain BYaURA3 was constructed by crossing BYαURA3 with the strain BY4741, followed by sporulation and dissection. The strain BYauN was constructed by the integration of the StuI linearized plasmid pLGU+4 in the URA3 region of the strain BYaURA3 and subsequent loss of integrated plasmid resulting in cells with ura3::NcoI allele that grow on 5-FOA. Transformation of the strain BYaURA3 with XhoI-SmaI fragment carrying ura3::HIS3 allele (3298 bp) resulted in the construction of the yeast strain BYauH.

Strain Y11809 (exo1Δ) was crossed with BY4742 followed by dissection thus constructing haploid strain BYa exo1Δ. Strains BYauH exo1Δ, BYαuH exo1Δ and BYauH sgs1Δ were constructed by crossing strain BYauH with the strains Y11809 (exo1Δ) or Y10775 (sgs1Δ) and dissection. Haploid yeast strain BYa exo1Δ sgs1Δ was constructed by crossing strains BYa exo1Δ with single mutant Y10775 (sgs1Δ) and dissection while the strain BYauH exo1Δ sgs1Δ was constructed by crossing BYαuH exo1Δ and BYauH sgs1Δ followed by dissection. The strain BYaU exo1Δ sgs1Δ was constructed by transforming the strain BYauH exo1Δ sgs1Δ with URA3 region from the pGU2 plasmid, selection of Ura⁺ His⁻ transformants followed by Southern blot analysis. Since both SGS1 and EXO1 genes were disrupted by kanMX4 cassette, the presence of two disruption cassettes in the exo1Δ sgs1Δ double mutant was confirmed by Southern blotting. BYauH pol32Δ was constructed by transforming BYauH with pol32Δ::kanMX4 cassette amplified by PCR using primers pol32-F (5′-aattctcgatcagtatgcctcaata-3′) and pol32-R (5′-cctgtacccaattttccttttagat-3′) and the disruption was confirmed by PCR. Throughout the work, standard methods for yeast growth and manipulation were used [51].

2.3. Transformation assays

In gene targeting assays presented in Figures 1A and 2A, yeast transformation was done using the lithium acetate procedure omitting carrier DNA [52]. One isolate per strain in wild type and single mutants and two isolates from exo1Δ sgs1Δ background were tested. Transformations were performed in duplicate using the same transforming DNA fragment and batch of cells, and each assay was carried out at least three independent times, transformants obtained in each assay were pooled for analysis.

Figure 1. Ends-out gene targeting (EOGT) experimental system and results.

(A) EOGT assays designed to insert (assay 1), replace (assays 2 and 3) and delete (assays 4 and 5) particular sequence in URA3 region on the chromosome V. In each assay, the transforming DNA fragment, targeted region and outcome of successful gene targeting are shown. The overall length of transforming and targeted DNA are indicated in parentheses. (B) Typical result of Southern blot analysis and schematic explanation of the aberrant transformants obtained in assay 5. Genomic DNA was cut with AseI and hybridized to the labelled URA3 gene. M-DNA of bacteriophage λ cut with HindIII, lane 1 untransformed strain; lines 2–8 aberrant Ura⁺ His⁺ transformants. (C) Fidelity of EOGT and spectra of aberrant events in five assays for wild type and (D) in the assay 5 for pol32Δ mutant. Numbers of transformants analyzed are indicated in parenthesis, Oth represent other uncharacterized or complex aberrant events, p values of significant differences compared to results from (C) assay 1 or (D) wild type are highlighted by the asterisks (*p <0.05, **p≤0.0001).

Figure 2. Ends-in gene targeting (EIGT) experimental system and results.

(A) EIGT assay designed to disrupt ARG4 gene on the chromosome VIII. Linearized plasmid pAGUS, targeted region and successful gene targeting outcome are shown. (B) Typical result of Southern blot analysis and schematic explanation of the aberrant transformants obtained in EIGT assay. Genomic DNA was cut with XbaI and hybridized to the labeled ARG4 gene. M-DNA of bacteriophage λ cut with HindIII, lanes 1–8 aberrant Arg⁺ Ura⁺ transformants. (C) Fidelity of EIGT and spectra of aberrant events in wild type. Numbers of transformants analyzed are indicated in parenthesis.

2.3.1. Spectra of transformation events in ends-out gene targeting (EOGT)

In EOGT assays (Figure 1A), corresponding yeast strains were transformed with 500 ng of the transforming DNA fragment generated by cutting the plasmids pAULI (assays 1 and 3), pADULI (assay 2) and pAGU2 (assays 4 and 5) with HindIII and isolated from the agarose gel (Table 1). To determine the fidelity of EOGT (proportion of targeted events) and spectra of aberrant transformation events in each assay, transformants were first analyzed by replica plating. In assays 1–3 and 4–5, transformants were selected on SC-leu and SC-ura plates, respectively, followed by replica plating to select for transformants with aberrant phenotype (for assays 1–2 Leu⁺ Ura⁺; assay 3 Leu⁺ His⁺, assay 5 Ura⁺ His⁺). Aberrant phenotype was determined for 15/980 (assay 1), 9/258 (assay 2), 48/2287 (assay 3), 6/39 (assay 4) and 55/932 (assay 5) transformants in wild type. In assay 5, 11/395 transformants from exo1 mutant, 7/190 from sgs1, 1431/1451 from exo1 sgs1 double mutant and 14/69 from pol32 background had an aberrant phenotype. Spectra of the transformation events in all transformants with the aberrant phenotype were determined by Southern blotting (Figures S1–S3). Since assay 4 did not allow phenotypic distinction of targeted and aberrant transformation outcomes, all transformants were analyzed by Southern blotting (Figure S3). 2 to 3 representative transformants for each assay, shown by Southern blotting to contain both an untransformed and region after gene targeting (heteroallelic transformants), were further analyzed by pulsed-field gel electrophoresis (PFGE, Figure S4) [14] and flow cytometry (2.5) to confirm TCD and DNA content. In the assay 5, due to the significant increase of the percentage of the aberrant transformation events, 75 of aberrant transformants from the exo1Δ sgs1Δ double mutant were analyzed by Southern blotting, 7 transformants that contained both the untransformed and gene targeting allele were further analyzed by whole chromosome PFGE (Figure 3) and 2 of them were analyzed by flow cytometry (2.5).

Figure 3. Transformation efficiency and the spectra of transformation events in EOGT and EIGT assays for the exo1Δ and sgs1Δ single mutants and the exo1Δ sgs1Δ double mutant.

(A) Relative efficiency of transformation is expressed as the ratio of the number of transformants obtained in a specific assay (EOGT assays 3 and 5 and EIGT assay) and with replicative plasmid pLUH and then normalized to wild type. (B) Genomic DNA of the aberrant transformants from the double exo1Δ sgs1Δ mutant in assay 5 was cut with AseI and (C) PFGE analysis of whole chromosomes of the same samples. Lanes 1 and 11 untransformed ura3::HIS3 strain (His⁺ Ura⁻ phenotype), lanes 2 and 10 gene targeting transformants with URA3 allele (His⁻ Ura⁺ phenotype), lanes 3–9 transformants with the aberrant His⁺ Ura⁺ phenotype containing both ura3::HIS3 and URA3 alleles, DNA was hybridized to the labelled URA3 gene. (D) Fidelity of EOGT and spectra of aberrant events in assay 5 and (E) fidelity of EIGT and spectra of aberrant events in selected mutants. Error bars represent standard deviations, numbers of transformants analyzed are indicated in parenthesis, p values of significant differences compared to wild type strain are highlighted by the asterisks (*p <0.05, **p≤0.0001).

2.3.2. Spectra of transformation events in ends-in gene targeting (EIGT)

In EIGT assay (Figure 2A), strains BY4741 and BYa exo1Δ sgs1Δ were transformed with 200 ng of plasmid pAGUS linearized with SacI within RG region (539 bp central part of the ARG4 gene) and isolated from the agarose gel. In this assay, half of the transformation mixture was plated on SC-ura to allow growth of all transformants (Ura⁺ phenotype) while the other half was plated on SC-ura -arg to select for aberrant transformants only (Ura⁺ Arg⁺ phenotype). In wild type 14/960 while in the double exo1Δ sgs1Δ mutant 482/498 transformants had an aberrant phenotype. As in ends-out assays, all transformants with the aberrant phenotype from the wild type (14) and 40 from the exo1Δ sgs1Δ double mutant were analyzed by Southern blotting, 6 of them by whole chromosome PFGE or PFGE of SacII cut plugs (Figure S5) and 2 by flow cytometry (2.5.).

2.3.3. Transformation efficiency

In order to determine transformation efficiency in DNA repair mutants (Figure 3A) yeast strains were transformed either with 1000 ng of the plasmid DNA (pAULI or pAGU2) cut with HindIII (EOGT assays 3 and 5) or 100 ng of the plasmid pAGUS linearized within RG region with SacI (EIGT assay). To allow the comparison of results from different strains, relative efficiency of transformation was expressed as the ratio of the number of transformants obtained in each assay and with 100 ng of the circular replicative plasmid pLUH.

2.4. Molecular analysis of the transformants

Southern blot and PFGE analysis of genomic DNA were described previously [14, 53]. Conditions for whole chromosome PFGE of wild type transformants were as follows: 0.5x TBE, 1% agarose, 6 V/cm, 24 h, switch time 60–120 s, or in the case of transformants from the exo1Δ sgs1Δ double mutant switch time was changed to 24–205 s. Conditions for separation of DNA fragments from SacII cut plugs were: 0.5x TBE, 1% agarose, 6 V/cm, 11 h, switch time 1–6 s. Molecular analysis was done using DIG-labeled PCR amplified URA3 (primers ZG1 and ZG4), kanMX (primers KanUp 5′-aaaaaataggcgtatcacgag-3′ and KanDn 5′-tcgatgataagctgtcaaac-3′) or ARG4 probe (primers ARG4-F 5′-taacgtcgccatctgctaa-3′ i ARG4-R 5′-catgtcagacggcactcaa-3′).

2.5. Flow cytometry

Determination of DNA content by flow cytometry was done using protocol previously described by Gerstein et al [54] and analyzed by Accuri-6 (Accuri Cytometrics Inc.). For each transformant two independent cell preparations were analyzed and the DNA content was calculated with respect to the non-transformed parental haploid and diploid strains.

2.6. Statistical analysis

Statistical significance was determined after pooling transformants from several independent transformations using two-tailed Pearson’s chi-squared test (χ²; http://in-silico.net/). When gene targeting fidelity was compared between different assays, numbers of transformants due to a successful gene targeting were compared to those from assay 1, or results from mutant strains were compared to wild type results from assay 5 (Figures 1C, 1D, 2C, 3D and 3E; eg. 249 gene targeting transformants /258 total transformants from assay 2 were compared to 965 gene targeting /980 transformants from assay 1). Statistical significance of spectra of aberrant genetic events was determined by comparing number of transformants due to a specific type of an aberrant event among all aberrant transformants with results from assay 1 or for mutants to wild type assay 5 results (eg. 6 TCD events /9 aberrant transformants from assay 2 were compared to 15 TCD events /15 aberrant transformants from assay 1).

3. RESULTS

Ends-out gene targeting (EOGT) is usually used to replace a particular endogenous sequence with the one on the transforming DNA fragment although this approach could also be used to insert or delete various sized sequences from the targeted locus. On the other hand, ends-in gene targeting (EIGT) is used to insert the transforming DNA, most frequently linearized plasmid, into a particular position of the genome. The main aims of this study were to determine the fidelity of gene targeting and the spectra of aberrant transformation outcomes in specific EOGT assays as well as to investigate if targeted chromosome duplication (TCD) occurs during EIGT. To accomplish this, we employed five different EOGT assays (Figure 1A) and one EIGT assay (Figure 2A). EOGT assay 1 was designed to insert 1.81 kb (LEU2 gene) in the targeted locus; in assays 2 and 3 the expected outcomes were replacement of 0.25 kb and 1.77 kb (HIS3 gene) sequences with the LEU2 gene; whereas in assays 4 and 5, 4 bp and a 1.77 kb (HIS3 gene) were deleted from the targeted locus, respectively. Additionally, EIGT assay (Figure 2A) allowed us to select for aberrant transformation outcomes during plasmid integration.

3.1. Transformation efficiencies in EOGT assays

To investigate if the fidelity of EOGT depends on the type of intended genome modification we used 5 specific assays (Figure 1A). Lengths of the upstream and downstream flanking homologies on the transforming DNA fragment in most assays were the same (0.66 kb and 0.50 kb, respectively). Exceptions to this were in assay 2 (carrying 0.41 kb and 0.50 kb upstream and downstream homology, respectively) and assay 4 (with 0.43 kb and 0.73 kb of upstream and downstream homology, respectively). Since flanking homologies were of comparable sizes and significantly above minimal efficient processing segment (MEPS) we expected that the resulting transformation efficiencies were influenced by the design (specificity) of each transformation assay, such as the length of inserted, replaced or deleted sequence from the targeted locus. Although transformation efficiency varied only 3-fold among assays 1 and 5 (Figure S6), it was slightly higher for the insertion assay (1) than for replacement assays (2 and 3), decreasing further in deletion assay (5), consistent with the results of molecular analysis of transformants (see section 3.2.).

3.2. Fidelity of EOGT in wild type

To determine the fidelity of EOGT (proportion of targeted events) and the spectra of aberrant events in five ends-out assays (Figure 1A), transformants obtained in each assay were analyzed as described in section 2.3.1. The typical Southern blot analysis and the schematic explanation of the aberrant transformants are shown in Figures 1B and S1–S4.

The fidelity of EOGT was high in all assays in the wild type, ranging from 84.61 % to 98.47 % (33/39 for assay 4 to 965/980 for assay 1, Figure 1C). However, consistent with the transformation efficiency results (Figure S6), the proportion of successful gene targeting was slightly higher for the insertion (assay 1; 965/980 transformants analyzed, 98.47 %), replacement assay 2 (249/258, 96.51 %) and assay 3 (2238/2287 transformants, 97.86 %, p=0.0424 and p=0.2785 for assay 2 and 3 compared to assay 1, respectively) than for the deletion assay 5 (877/932 transformants, 94.10 % of gene targeting, p<0.0001 compared to assay 1). Moreover, when a longer DNA sequence was inserted, replaced or deleted (assays 1–3 and 5) the fidelity of gene targeting was significantly higher compared to the deletion of only 4 bp from the genome (33/39; 84.62 %, p<0.0001 compared to assay 1) further reinforcing that design of the experimental system per se influences the fidelity of gene targeting. Lower gene targeting fidelity in assay 4 could be explained by the involvement of mismatch repair in the recognition of 4 bp heterologous sequence from NcoI end-filled site in the URA3 gene [55].

Aberrant transformation events obtained in EOGT include (i) heteroallelic transformants having two copies of a targeted chromosome (ii) integration of the transforming DNA fragment next to the targeted region and (iii) illegitimate integration (Figures 1B, S1–S4). Both heteroallelic transformants and integration of the targeted DNA fragment next to the targeted region have already been described in detail in an EOGT assay designed to delete the 6.1 kb Ty1 retrotransposon from the ura3-52 allele [14].

The spectra of aberrant events described here (Figure 1C) clearly shows that, even in wild type, TCD is the dominant class of all aberrant events ranging from 22/48 aberrant transformants in EOGT assay 3 (45.74 %) to 15 TCDs of 15 aberrant transformants in assay 1 (100.00 %). Integration of the transforming DNA next to the targeted locus was observed only in assay 3 (14/48 aberrant events, 29.06 %) and in assay 5 (15/55 aberrant events, 27.27 %, p=0.003 and p=0.004 for assays 3 and 5, compared to 0/15 in assay 1, respectively) where the transforming DNA fragment had either the same length or it was shorter than the targeted sequence, suggesting they occur as a consequence of the hindered ability of the transforming DNA fragment to simultaneously invade both flanking homologies in the genome. This observation is in accordance with the proposed model for single and multiple integration of ends-out transforming fragment next to the targeted sequence [14].

3.3. Fidelity of EOGT in pol32 background

It has been shown that a chromosomal DSB can promote chromosome missegregation during mitosis yielding one nullsomic and one disomic cell [19]. Moreover, we have proposed previously that EOGT can promote TCD either due to BIR or nondisjunction of sister chromatids [14]. To reveal the mechanism underlying TCD, we determined the spectrum of transformation events for EOGT assay 5 in the pol32Δ mutant that is impaired for BIR (Figure 1A, 1D). Although the fidelity of EOGT was significantly lower than in wild type with 55/69 gene targeting events (79.71 %) compared to 877/932 in wild type (94.10 %; p<0.0001), TCDs were abolished (0/14 compared to 30/55 aberrant events in wild type, p=0,0002) confirming they are a consequence of BIR. The observed significant increase in aberrant events was due to illegitimate integration (13/14 compared to 10/55 in wild type, p<0.0001).

3.4. Transformation efficiency and fidelity of EOGT in the exo1 and sgs1 single and the exo1 sgs1 double mutants

Inactivation of both EXO1 and SGS1 genes influences the stability of the transforming DNA fragment resulting in an increased efficiency of transformation [41, 42, 44]. Therefore, we suspected it might also influence the fidelity of EOGT and the spectrum of transformation events. The efficiency of transformation for the exo1Δ and sgs1Δ single and the exo1Δ sgs1Δ double mutants in EOGT assays 3 and 5 was determined (Figure 3A). While deletion of the EXO1 gene had only minor effect on the efficiency of transformation, the inactivation of SGS1 gene resulted in 8.5-fold and 11.5-fold increase in EOGT assays 3 and 5, respectively. At the same time, inactivation of both genes increased the efficiency of transformation in assays 3 and 5 by 54.4-fold and 31.2-fold, respectively (Figure 3A).

The influence of exo1Δ and sgs1Δ mutations on fidelity of gene targeting was analyzed in assay 5 (Figures 3B–3D). While the exo1Δ mutant displayed slightly higher fidelity of EOGT compared to the wild type (384/395 transformants compared to 877/932, p=0.0118), the sgs1Δ mutant was similar to the wild type (183/190 compared to 877/932, p=0.2007). A striking shift toward the aberrant events was observed in the exo1Δ sgs1Δ double mutant compared to wild type, with 98.62 % of all transformants being aberrant (20 gene targeting events of 1451 transformants analyzed, compared to 877/932 in wild type, p<0.0001). Moreover, 71 of 75 (94.62 %) aberrant transformants analyzed in the exo1Δ sgs1Δ double mutant had TCD, compared to 30 of 55 (54.55 %) in wild type (p<0.0001; Figures 3B–3D) and analysis of 2 aberrant TCD transformants confirmed they had haploid DNA content (1.09 and 1.16). To confirm the generality of these results, the percentage of aberrant transformation events in the exo1Δ sgs1Δ double mutant for assays 1 and 2 was determined by replica plating (98.5 % and 97.73 %, respectfully). Therefore, we suggest that due to the inactivation of both extensive resection pathways, the transforming DNA fragment becomes stabilized and engages more frequently in BIR. We could not directly test this hypothesis due to lethality of the pol32Δ sgs1Δ double mutant [56].

3.5. Fidelity of EIGT in wild type and the exo1 sgs1 double mutant

As reported by others and here, deletion of EXO1 and SGS1 genes synergistically increased the transformation efficiency in both EOGT and EIGT [41, 42] (Figure 3A). In the EOGT assays described here, this increase coincides with the striking increase in the proportion of aberrant events, more precisely an increase in TCD (Figures 3B–3D). Therefore, we used an EIGT assay that allows phenotypic distinction between targeted and aberrant transformation events (Figure 2A, 2B). In this assay, targeted plasmid integration results in the disruption of endogenous ARG4 gene and Arg⁻ Ura⁺ phenotype, whereas aberrant events result in an Arg⁺ Ura⁺ phenotype. Since we expected an extremely low proportion of aberrant events in wild type, the fidelity of gene targeting was determined not only by replica plating but also by direct selection of aberrant transformants; one half of the transformation mixture was plated on SC-ura to select for all types of transformation events (both targeted plasmid integration and aberrant events) and the other half on SC-ura-arg plates, to select for aberrant events only. The proportion of Ura⁺ Arg⁻ transformants due to gene targeting was 98.54 % (946/960) and Southern blot analysis revealed that all aberrant transformants from the wild type (14/14) were heteroallelic (Figure 2C).

The percentage of transformants with the aberrant Arg⁺ Ura⁺ phenotype increased to 96.82 % (482/498) in the exo1Δ sgs1Δ double mutant (p<0.0001). It was confirmed by Southern blotting that all analyzed aberrant transformants (40/40) were heteroallelic, containing both an intact ARG4 locus and arg4 disrupted by plasmid integration (Figure S5A). Furthermore, PFGE analysis of 6 heteroallelic transformants showed hybridization only to the chromosome VIII (Figure S5B) while flow cytometry analysis of two transformants with aberrant phenotype revealed haploid DNA content. Additionally, results of restriction analysis of PFGE plugs with SacII (Figure S5C) again revealed the existence of two alleles: one contained untransformed ARG4 allele (35 kb band) while the other was due to either single or multiple plasmid integration in the ARG4 region (plasmid integration band 39.3 kb), confirming that heteroallelic transformants arose due to TCD. These results are in accordance with the results obtained in EOGT, showing that Exo1 and Sgs1 proteins contribute to the suppression of gene targeting-associated chromosome duplications.

4. DISCUSSION

S. cerevisiae is a model eukaryotic organism with extremely efficient HR and high fidelity of gene targeting [22, 57]. Proteins involved in DSB repair by HR in yeast have their functional, and even structural homologues in higher eukaryotes and plants, suggesting that the same rules might govern gene targeting and giving hope for improving gene targeting and gene therapy as well [58–67]. One of the major problems associated with gene targeting in all higher eukaryotes are mistargeted events [28, 68–72]. We had shown previously that mistargeted events occur during EOGT even in S. cerevisiae [14]. These events include: illegitimate integration, integration of the transforming DNA next to the target sequence and TCD. As a continuation of our studies, we have developed several EOGT and one EIGT assay and here we show that gene targeting success is influenced not only by the length of flanking homologies, but also by the type of intended genome modification by EOGT and the overall length of the transforming DNA fragment with the respect to the targeted sequence in the genome.

4.1. Fidelity of EOGT depends on the type of intended genome modification

Successful EOGT requires flanking homologies at the ends of the transforming DNA fragment to target it to the particular region in the genome. Although throughout the literature gene targeting by ends-out recombination is frequently called gene replacement, here we describe five EOGT assays (Figure 1A), used to modify the targeted region in the yeast genome, distinguishing between insertion (assay 1), replacement (assays 2 and 3) and deletion (assays 4 and 5). All the transforming DNA fragments, used in EOGT assays, had flanking homologies above 0.41 kb. Therefore, we might expect that the observed differences in the spectra of transformation events depended on the type of intended genome modification by EOGT, length ratio of the transforming and targeted sequence and heteroduplex structure formed during recombination.

A large number of transformants analyzed for each assay, allowed us to draw several important conclusions and to hypothesize about causes of the observed effects. Although the differences in the fidelity of gene targeting between EOGT assays were small (84.61 % – 98.47 %), most of them are statistically significant (see sections 3.2.–3.4., Figure 1C). Insertion assay 1 showed the highest (98.5 %) while deletion assays (4 and 5) had the lowest fidelity of EOGT (84.6 % and 94.1 %), confirming that even in yeast it is easier to insert than to delete a sequence in the genome, as has been previously observed for gene targeting in human cell culture [73]. Replacement assays (2 and 3), mimic the experimental design used in most laboratories for the construction of knock-out strains during which a part, or the entire ORF, is replaced with selectable marker and both resulted in intermediate targeting fidelity (96.5 % and 97.9 %).

We were not able to distinguish whether gene targeting occurs by assimilation of a single stranded DNA [10] or by two independent strand invasions [11, 12]. However, we hypothesize that a single strand of the transforming DNA fragment can be assimilated in the recipient DNA if it is entirely homologous to the target sequence, more precisely if it does not contain any sequence heterologous to the targeted locus (EOGT assays 4 and 5). On the other hand, if the transforming DNA fragment contains heterologous insertion (EOGT assays 1–3), single stranded DNA cannot be entirely assimilated so gene targeting is a consequence of two independent strand invasions [11, 12].

The lowest fidelity of EOGT achieved in ends-out assay 4 (84.6 % vs. 98.5 %, Figure 1C) is rather a consequence of an interaction between the transforming and the targeted DNA than property of the transforming DNA fragment per se because the use of the same DNA fragment in assay 5 resulted in higher EOGT fidelity. In both assays (4 and 5) the transforming DNA fragment does not contain a heterologous insertion and therefore can undergo single stranded DNA assimilation. However, assimilation of a single strand of the transforming DNA in assay 4 results in formation of heteroduplex DNA containing 4 nucleotide (nt) single stranded loop and it was already shown that short heterologies (up to 14 bp) recognized by mismatch repair proteins [55] are more anti-recombinogenic than the large ones [74–76] such as those that could appear in assay 5. Therefore, efficient single stranded DNA assimilation in the recombination intermediate, followed by a strong antirecombinogenic effect of 4 bp heterology could result in the frequent displacement of the assimilated single stranded DNA and an appearance of a single stranded transforming DNA fragment. This is in agreement with the previous reports showing that single stranded DNA is highly recombinogenic, being less affected by antirecombinogenic effect of point mutations [77], prone to illegitimate recombination and able to induce genome rearrangement [53].

The lower fidelity of EOGT in assay 5, compared to assay 1 (94.1 5 vs. 98.5 %, Figure 1C), could be a consequence of several phenomena. First, it is possible that it is easier to insert than to delete sequence from the genome because genome-borne insertion has a stronger anti-recombinogenic effect than the insertion present on the transforming DNA. Second, if formed in heteroduplex, large single stranded loops are preferentially repaired using a genomic DNA as a template. However, it is also possible that fidelity of gene targeting is influenced by the length ratio between transforming DNA and targeted region. This hypothesis is supported by the fact that in assay 5 the transforming DNA is significantly shorter than targeted region in the genome (1.16 kb vs. 2.93 kb, respectively) hindering simultaneous invasion of both ends of transforming DNA in flanking homologies. Consistently, integration of the transforming DNA next to the target in the genome was detected only in assays 2 and 3, in which targeted homology is disrupted with a large insertion and the transforming DNA fragment is either a little bit shorter or of same length as targeted region (assays 3 and 5). Again in agreement with this hypothesis, when URA3 (1.16 kb) was used to delete a Ty1 element from the ura3–52 allele the fidelity of EOGT was only 60 % and integration of the transforming DNA fragment next to the homology was observed in 10 % of transformants [14]. In that assay the transforming DNA fragment integrated exclusively in the longer flanking homology in the terminator region of the ura3 gene (0.82 kb; N-terminal homology was 0.34 kb). In the EOGT assays used here (Figure 1A), with both flanking homologies of a similar size, integration of the transforming fragment was not biased suggesting that ends-out transforming fragment preferentially integrates in the region of a longer homology rather than in the terminator region.

4.2. BIR-induced targeted chromosome duplication during EOGT and EIGT is suppressed by Exo1 and Sgs1

TCD were detected in all EOGT assays described here (Figure 1C) explaining the observation that approximately 8 % of strains from the yeast knock-out library are aneuploidic for the targeted chromosome [78]. We have to stress that there was an observable decrease in the survival of the aneuploidic transformants during their passaging compared to gene targeting transformants. This is in agreement with physiological changes observed in aneuploidic yeast strains [79, 80].

Although we have not directly demonstrated two copies of targeted chromosome in TCD transformants, we have shown that they contain two targeted regions (transformed and untransformed; Figure 2B, 3B) which cannot be attributed to a duplication of a short chromosome region (Figure S5B and S5C). Moreover, we have shown by Southern blotting of native chromosomes (Figure 3C and S5B) that URA3 and ARG4 gene (for EOGT and EIGT, respectively) hybridize strictly to the targeted chromosome, and no additional molecules were detected. Finally, several TCD transformants (2 from the double exo1Δ sgs1Δ background in each ends-out and ends-in assay and 2–3 per each assay in wild type) were shown to have haploid DNA content confirming that TCD transformants are disomes for targeted chromosome. However, we cannot exclude that some of presumptive TCDs are diploids formed due to whole genome endoduplication prior or during transformation but not by cell fusion because it does not occur during lithium acetate transformation (Table S1) [81]. Additionally, we cannot exclude that a small percentage of TCD transformants were due to spontaneous duplication of a targeted chromosome, followed by successful gene targeting but the absence of such events in the pol32Δ mutant (assay 5; Figure 1D) clearly demonstrates that most, if not all, observed chromosome duplications occur by BIR [47]. It is also possible that some TCDs contain even three copies of targeted chromosome due to transformation fragment mediated chromosome missegregation [19] followed by BIR.

The fidelity of EOGT was significantly lower in pol32Δ mutant than in wild type due to striking increase of illegitimate integration of the transforming DNA. This effect could be consequence of interaction of transforming DNA with single stranded gaps accumulated in pol32Δ [56]. Moreover, it could be enhanced if gene targeting in assay 5 occurs by assimilation of ss DNA (see above) because illegitimate integration of single stranded transforming DNA can be further encouraged due to pairing with microhomology in single stranded gaps.

Due to reduced extensive DNA resection in the exo1Δ sgs1Δ double mutant [31–33] the transforming DNA fragment is more stable resulting in the synergistic increase of the transformation efficiency during both EIGT [41] and EOGT [42]. In assays described here (Figure 3A) this effect is less pronounced and it depends on the type of intended genome modification, being higher for the EOGT assay 3 (replacement assay) than for the EOGT assay 5 (deletion assay). Inactivation of single EXO1 or SGS1 genes had only a minor influence on both the fidelity of gene targeting and the spectra of transformation events compared to the wild type (Figure 3D). However, in the double mutant fidelity of EOGT decreased 65-fold compared to wild type (from 94.10 % in the wild type to 1.38 % in the double mutant).

To determine the influence of Exo1 and Sgs1 proteins on the yield of each type of transformants (Table 3) we normalized numbers of transformants obtained in assay 5 (Figure 3D) according to the transformation efficiencies achieved in a given genetic backgrounds (Figure 3A, assay 5). Data presented in Table 3 suggests that, although the frequency of gene targeting in the exo1Δ sgs1Δ double mutant, as determined by a molecular characterization of the obtained transformants, was only 1.38 % percent, the yield of transformants dropped only to 50 % compared to wild type strain (from 94 to 43, respectively). Interestingly, in the double exo1Δ sgs1Δ mutant there is a striking increase both in the yield of chromosome duplications (from 6/100 in wild type to 2893/3100) and in illegitimate integrations (from 1/100 in wild type to 164/3100, respectively). A similar increase in aberrant events in the exo1Δ sgs1Δ double mutant was also observed in the met17::ADE2/MET17 assay (L.S. Symington, unpublished). Increased yield of integration of transforming DNA fragment next to the targeted region resulting in tandem repeats [14] in sgs1Δ mutant is in accordance with the results obtained by Mukherjee and Storici [82] showing that deletion of SGS1 significantly increases frequency of gene amplification events initiated by small DNA fragments.

Table 3. The yield of each type of transformants in wild type, the sgs1Δ and exo1Δ single mutants and the exo1Δ sgs1Δ double mutant.

First, the numbers of each type of transformation events obtained in wild type in assay 5 were normalized to the total of 100 transformants (Figure 3D). Second, the total number of transformants obtained from each mutant was normalized according to the increased transformation efficiency in each mutant (Figure 3A, assay 5), and finally, yield of each type of transformants in each mutant was calculated using data presented in Figure 3D.

Transformation event		Assay 5
		wild type		exo1Δ		sgs1Δ		exo1Δ sgs1Δ
Successful gene targeting		94		116		1108		43
Aberrant events	Integration next to the targeted region (Int)	6	2	4	2	42	30	3057	0
	Targeted chromosome duplications (TCD)		3		1		6		2893
	Illegitimate integrations (Ille)		1		1		6		164
Total		100		120		1150		3100

Due to lack of normal checkpoint response in the exo1Δ sgs1Δ double mutant we cannot exclude half-crossover on centromere-containing side and BIR toward the telomere while a truncated chromosome could initiate a secondary recombination event [83] or missegregation of sister chromatids due to the bridging activity of the transforming DNA fragment, as proposed by Svetec et al. [14] and Mukherjee and Storici [82] or formation of joint molecules due to inactivation of SGS1 gene [84, 85] followed by successful gene targeting on one chromatid/chromosome.

Chung et al [42] showed that gene replacement increased from 72 % in wild type cells to 99 % in the exo1Δ sgs1Δ double mutant. Discrepancies between the results described by Chung et al [42] and here could be explained either by specificity of EOGT assays or due to the fact that in their experimental system targeted region (THR4 gene) was located on the chromosome III between MAT and HMRa loci. Therefore, dense heterochromatin repressing HMLα and HMRa regions as well as chromatin dynamics that changes the accessibility of these two genes used as donors during MAT switching [86–89] could also affect gene targeting. In agreement with our results, it has already been shown that Exo1 and Sgs1 inhibit BIR in both chromosome-break repair assay [45] and chromosome fragmentation assay with DNA fragments either linearized in vivo or introduced in the cell by transformation [44]. The increase of yield of successful gene targeting in the absence of Sgs1 protein (Table 3) confirms that its 3′–5′ helicase activity can disrupt/abort recombination intermediate [43]. Moreover the increase of TCD in the exo1Δ sgs1Δ double mutant, but not in the respective single mutants, is in accordance with studies showing that Exo1 and Sgs1 proteins independently prevent BIR [44, 45].

TCD has never been reported in EIGT assays. However, it was proposed that during DSB repair it is not necessary that both 3′-ends simultaneously invade in the homologous duplex. Furthermore, it was shown that double stranded gap on a replicative plasmid can be repaired using two target sequences, each one used as a template to extend one 3′-end [90]. Therefore, we presumed that EIGT might also initiate BIR and result in whole chromosome duplication. In the EIGT assay described here (Figure 2A) the frequency of targeted chromosome duplication in wild type was 1.46 %. Furthermore, in accordance to the results obtained in EOGT assays, the absence of Exo1 and Sgs1 proteins in EIGT assay strikingly decreased gene targeting fidelity to only 3.2 % while the increase in the efficiency of transformation was due to increased proportion of TCD. These results clearly show that EIGT, as well as EOGT, can initiate duplication of a targeted chromosome and this aberrant transformation event is suppressed by Exo1 and Sgs1 proteins.

5. CONCLUSION

Precise genetic modification, using either EIGT or EOGT, can be associated with duplication of the targeted chromosome, which can easily go unnoticed if strains are not appropriately checked either by Southern blotting or by PCR. It is necessary to confirm both the presence of an expected, modified allele and the absence of a starting allele. Although it was proposed that the transient inactivation of homologues of EXO1 and SGS1 genes could be used to increase transformation efficiency in higher eukaryotes [41, 42], results presented here show that absence of Exo1 and Sgs1 proteins leads to BIR-induced chromosome duplication during both, ends-out and ends-in gene targeting.

Research highlights.

• Ends-out gene targeting is accompanied by aberrant genetic events

• Targeted chromosome duplications occur even during ends -in gene targeting

• Targeted chromosome duplications arise by break -induced replication

• Exo1 and Sgs1 proteins suppress targeted chromosome duplication

Abbreviations

HR

homologous recombination

EOGT

ends-out gene targeting

EIGT

ends-in gene targeting

DSB

double-strand break

TCD

targeted chromosome duplication

BIR

break-induced replication

bp

base pair

kb

kilobase

nt

nucleotide

PFGE

pulsed-field gel electrophoresis

ss

single stranded