I-SceI-Mediated Double-Strand Break Does Not Increase the Frequency of Homologous Recombination at the Dct Locus in Mouse Embryonic Stem Cells

Targeted induction of double-strand breaks (DSBs) at natural endogenous loci was shown to increase the rate of gene replacement by homologous recombination in mouse embryonic stem cells. The gene encoding dopachrome tautomerase (Dct) is specifically expressed in melanocytes and their precursors. To construct a genetic tool allowing the replacement of Dct gene by any gene of interest, we generated an embryonic stem cell line carrying the recognition site for the yeast I-SceI meganuclease embedded in the Dct genomic segment. The embryonic stem cell line was electroporated with an I-SceI expression plasmid, and a template for the DSB-repair process that carried sequence homologies to the Dct target. The I-SceI meganuclease was indeed able to introduce a DSB at the Dct locus in live embryonic stem cells. However, the level of gene targeting was not improved by the DSB induction, indicating a limited capacity of I-SceI to mediate homologous recombination at the Dct locus. These data suggest that homologous recombination by meganuclease-induced DSB may be locus dependent in mammalian cells.


Introduction
The natural efficiency of the introduction of defined sequences at specific locations of the mouse genome in embryonic stem (ES) cells by homologous recombination (HR) varies between 10 25 and 10 28 events per treated cell. Such a frequency is too low to consider the iterative introduction of a number of genes of interest at a given locus in standard practice. This problem can be overcome by enhancing recombination reactions at the target site through the induction of a double-strand break (DSB) [1]. Such DSBs can be induced with the yeast mitochondrial I-SceI meganuclease which has an 18-bp recognition site, absent normally in the mammalian genome but that may be added to the genome of recipient cells. In previous studies performed with I-SceI in Chinese hamster ovary (CHO) cells, mouse 3T3 fibroblasts, PCC7-S multipotent cells, and also in several ES cell lines, specific DSBs were shown to stimulate the repair of a tandem duplication by intrachromosomal HR or gene targeting by plasmid-tochromosome HR [2][3][4][5][6][7][8]. More recently, plasmid-mediated gene targeting was achieved in CHO cells after lentiviral delivery of the I-SceI protein [9]. Expression of I-SceI was also shown to be relevant to improve the efficiency of gene targeting in other organisms, including flies and plants [10,11]. Based on these data, it is generally admitted that a system based on the introduction of an I-SceI recognition site close to the locus to be targeted in the genome of recipient ES cells, combined with transient expression of the I-SceI meganuclease to create a DSB, should enhance the introduction of donor sequences at this site.
The Dct gene encodes the dopachrome tautomerase, a melanogenic enzyme. In the embryo, Dct is expressed in pigment cell precursors, i.e. melanoblasts, derived from the neural crest, in cells of the retinal pigment epithelium and in the developing forebrain [12,13]. In the adult epidermis, Dct is expressed in pigment cells at all differentiation states: in stem cells that reside in the bulge region of the hair follicle, in progenitors of the outer root sheath and in melanocytes of the hair matrix [14]. Studies performed in the mouse embryo with a LacZ reporter gene expressed under the control of 3.4 kb of the Dct promoter (Dct-LacZ transgene) depicted LacZ expression in melanoblasts and melanocytes, in the retinal pigment epithelium, forebrain, dorsal root ganglia and caudal nerves [15]. In addition, the Dct-LacZ reporter allowed to monitor cells of the melanocyte lineage in adult mice [16]. Altogether, Dct promoter-driven expression was shown to mimic largely the endogenous expression pattern of the gene. The Dct promoter has thus been used to drive the expression of genes in melanocytes and their precursors in transgenic mice [17][18][19][20][21]. However the use of combination of regulatory region from Dct and the coding regions of exogenous genes, either reporter genes or genes whose function is to be evaluated, can also have large drawbacks. First, several independent lines are required to distinguish the specific expression of the transgene from ectopic expression. Second, the transgene may be expressed in tissues that do not normally express endogenous Dct. This has been repeatedly observed with the Dct promoter [15,17,22]. It is worth noting that Dct knockout mice are viable and fertile, and exhibit no defects, with the exception of a diluted coat colour [23], making Dct an interesting driver to monitor the effects of expression of genes of interest, such as genes that may be involved in melanoma progression [24,25]. Indeed, either homozygous knockin mice or double heterozygotes for a reporter gene and the gene of interest may be studied. We thus became interested by constructing a genetic tool that would allow to insert with a high efficiency any gene of interest in place of the Dct gene.
Our approach relied on the combination of three components: an ES cell line carrying a Dct allele with the I-SceI recognition site, an I-SceI expressing plasmid and a template for the DSB-repair process carrying sequence homologies to the Dct locus. The targeted integration at the Dct locus was tested after transfection of both the I-SceI expressing plasmid and the repair construct in the modified ES cell line. We report here that an I-SceI recognition site embedded within the Dct gene sequence can be cleaved by transiently expressed I-SceI meganuclease in ES cells. We further show, contrary to expectations, and using two different repair vectors, that I-SceI-mediated DSB did not increase the frequency of HR at the Dct locus compared to conventional gene targeting experiments.

Design of Repair Vectors
To insert an I-SceI recognition site at the Dct locus, a replacement vector was constructed. A 6.5 kb SnabI-HincII fragment from pPBSKlB#4 that contains 18 kb of Dct gene (MGI:102563) [30] was inserted into the SpeI site of pL253 to produce pL253-Dct plasmid. To introduce an I-SceI recognition site and a Neo R cassette into the 6.5 kb Dct fragment near the first exon, we took advantage of a unique NheI site located within intron 1, 112 bp downstream of the first exon. A 59 Dct fragment containing the NheI site was amplified and flanked with KpnI and EcoRI sites using the following primers: 59KpnI forward 59-ATAGGTACCTCCCAATTAAGAAGGCATGG- 39 and 59EcoRI reverse 59-GCGGAATTCCGCCTTTCTGAGTGAA-GAG-39. The amplicon was inserted into pCR2.1 plasmid (TOPO TA CloningH, Invitrogen). The I-SceI recognition site was added at the NheI site. Besides, a 39 Dct fragment was amplified and flanked with BamHI and SacII sites using the following primers: 39BamHI forward 59-GTAGGATCCACCTTTGGCTTGTTTGTTGG-39 and 39SacII reverse 59-ATACCGCGGAGGACATGAGAACCC-CAGA-39. The amplicon was inserted into a pCR2.1 plasmid. pSW23 plasmid was digested by KpnI and SacII, and filled in with the three fragments: the 59 KpnI-EcoRI fragment containing the I-SceI site, an EcoRI-BamHI Neo R cassette from pL452 plasmid and the 39 BamHI-SacII fragment. The replacement vector was produced by a recombineering reaction between the modified pSW23 and pL253-Dct plasmid. The replacement vector carries a herpes simplex virus-thymidine kinase (HSV-TK) negative selection cassette downstream of the 6.5 kb Dct fragment.
The construction of HR repair vectors HR1 and HR2 relied on the GatewayH technology (Invitrogen). Entry and destination vectors were produced. pENTR1A entry vector (Invitrogen) contains ccdB flanked with multi-cloning sites (MCS). The SV40 polyadenylation sequence (pA) was inserted at the EcoRV site to give pENTR1ApA. The Lago1 gene, a synthetic CpG-free LacZnls reporter gene that contains a SV40 nuclear localization signal (Invivogen), with its start codon was inserted in pENTR1ApA in place of ccdB (Fig. 1). The first destination vector (DV1) was constructed as follows. Starting from pL253-Dct plasmid, a SexAI-AvrII 400 bp fragment containing the ATG start codon was removed from Dct sequence. A linker made of the following primers was used to fill the gap in Dct sequences: HpaPme forward 59-CTAGGTTAACGTTTAAA-39 and HpaPme reverse 59-CCTGGTTTAAACGTTAAC-39. The linker allowed the insertion of a unique HpaI recognition site into pL253-Dct, giving pL253-Dct-HpaI plasmid. To insert the Neo R cassette into pL253-Dct-HpaI, a recombineering reaction was performed and gave pL253-Dct-HpaI-Neo R plasmid. Finally, a reading frame cassette A (RfA) (GatewayH technology) that contains the Cm R -ccdB cassette flanked by attR1 and attR2 sites, was inserted at the HpaI site into pL253-Dct-HpaI-Neo R to obtain DV1. In DV1, the negative selection cassette HSV-TK from pL253 is downstream of the 39 sequence homology of Dct (Fig. 1). To produce the first repair vector (HR1), the Cm R -ccdB cassette in DV1 was replaced by Lago1 gene using LR reaction (GatewayH technology) (Fig. 1).
To remove the short regions of homology between the HR1 repair vector and the first intron of Dct gene, a second destination vector (DV2) was produced. The H 2 B-mCherry reporter gene with its start codon [31] was inserted in place of ccdB in pENTR1ApA entry vector (Fig. 2). To obtain DV2, a 2.8 kb AvrII-Bsu36I fragment was synthetized. It contains a Neo R cassette framed with FRT sites and 1 kb of Dct genomic sequence. A 1 kb AvrII-Bsu36I fragment was removed from the pL253-Dct-HpaI plasmid and replaced by the 2.8 kb synthetized AvrII-Bsu36I fragment. Then RfA was ligated at the HpaI site. Figure 2 shows the map of DV2. To produce the second repair vector (HR2), the Cm R -ccdB cassette of DV2 was replaced by the H 2 B-mCherry sequences using LR reaction (Fig. 2). The integrity of the repair vectors was verified by sequence analysis.

Homologous Recombination Assay
To insert a unique I-SceI site at the Dct locus, approximately 1.6 6 10 7 CK35 ES cells were electroporated with the NotIlinearized replacement vector. G418 (300 mg/mL) was added 48 h after plating for 12 days and gancyclovir (2 mM) was added 96 h after plating for 4 days. The Neo R cassette was removed using pIC-Cre plasmid [32] in which the transcription of Cre recombinase is driven by a synthetic HSV-TK promoter and enhancer. Fifteen micrograms of pIC-Cre plasmid were electroporated into approximately 1610 7 ES clone 4 cells and the cells were cultured without G418. For the gene targeting with HR1 repair vector, approximately 1.3610 7 MF1 ES cells were electroporated with 20 mg of supercoiled HR1 alone or with 33 mg pCMV-I-SceI [26] or 37.5 mg pCAG-I-SceI. The ratio of expression plasmid to repair vector was 5 to 1. For the experiment using HR2, approximately 1.6610 7 MF1 ES cells were electroporated with 30 mg of supercoiled HR2 plasmid alone or with either 11 mg pCMV-I-SceI or 13 mg pCAG-I-SceI expression plasmids. The ratio of expression plasmid to repair vector was 1 to 1. Approximately 1.6610 7 MF1 ES cells were independently electroporated with 30 mg of NotI-linearized HR2 plasmid, as a control.

DNA Analysis in Selected Clones
Genomic DNAs of ES clones obtained after selection with G418 and gancyclovir were digested with BamHI. Correct gene targeting was analyzed by Southern blot using a 1 kb 59 external probe produced by PCR amplification with the following primers: 5DCTF forward 59-TTGGGGTCAGGGAGATACAG-39 and 5DCTR reverse 59-TGAGCAGCAGTGAAGTTTGG-39.

Ligation-Mediated PCR (LM-PCR) Analysis
Approximately 1.6 6 10 7 MF1 ES cells were electroporated with 50 mg pCMV-I-SceI, pCAG-I-SceI or mock plasmid. Four hours later, genomic DNA was extracted. Two micrograms of genomic DNA from MF1 cells transfected with the mock plasmid were digested with PstI or I-SceI and precipitated. LM-PCRs were performed with these PstIand I-SceI-digested DNAs, and with undigested DNA from MF1 cells transfected with mock plasmid, pCMV-I-SceI or pCAG-I-SceI. The specific LM-C1 primer 59-AATTCTTCAACCGGACAT-39 was used for the first extension. The asymmetrical synthetic double-stranded linker was prepared by hydridization of two oligonucleotides: linkerF forward 59-GCGGTGACCCGGGAGATCTGAATTC-39 and reverse lin-kerR 59-GAATTCAGATC-39. The specific LM-C2 primer 59-CGGACATGCAAATGCACAGGTGAGG-39 was used for a first PCR amplification. The PCR product was subjected to nested PCR with the specific LM-C3 primer 59-CCCTTGGGCA-GACCCAGATGTCACT-39) and linkerF. After agarose gel electrophoresis and alkaline transfer to a nylon membrane, the DNA was hybridized to the specific 36-mer radioactive probe LM 59-CTTCTGAGGAGAGGCGACACTGGTGA-CAAACTGTTA-39.

Results
Our experiments aimed at testing the efficiency of a ready-touse tool to produce ES cells, and eventually mice, carrying any sequence of interest inserted in place of the Dct gene. Our strategy relied on the reported stimulation of gene targeting frequency at a natural locus associated with a DSB induced by the yeast meganuclease I-SceI in ES cells [3]. We performed a two-step experiment. In a first step, the I-SceI restriction site was inserted to the Dct gene in ES cells using conventional gene targeting procedures. The Dct gene carrying a unique I-SceI restriction site was thus considered as a preferential target for HR. In a second step, an I-SceI-expression plasmid was introduced together with a repair vector sharing a 5.9-kb of Dct isogenic DNA by electroporation in the engineered ES cells, and the efficiency of gene targeting at the Dct locus was assayed.

Production of a New Target Allele at the Dct Locus in ES Cells
As a first step, the I-SceI restriction site was inserted within Dct intron 1 in ES cells. A replacement vector containing a unique I-SceI restriction site, a positive selection (Neo R ) cassette flanked by loxP sites, and 1.9 and 4.5 kb of 59 and 39 genomic sequences from the Dct gene was constructed. A negative selection cassette (HSV-TK) was added after the 39 homology arm (Fig. 3A). The replacement vector was linearized and electroporated into CK35 ES cells. The cells were cultured in the presence of G418 and gancyclovir. Out of 107 G418-and gancyclovir-resistant colonies, one clone (ES clone 4) was correctly targeted with the replacement vector as shown by PCR (data not shown), and later confirmed by Southern blot analysis (Fig. 3B). To test whether the meganuclease I-SceI is able to specifically cleave the new Dct I-SceI-Neo allele, genomic DNA of the ES clone 4 was treated with both BamHI and I-SceI restriction enzymes. Southern blot analysis using an external 59 probe revealed the 4.5 kb BamHI-I-SceI distinctive fragment, indicating that the I-SceI site inserted at the Dct locus was indeed cut in vitro by the meganuclease (Fig. 3C).
To delete the Neo R cassette, a Cre recombinase-expressing plasmid (pIC-Cre) was electroporated into clone 4 Dct I-SceI-Neo/+ ES cells. A preliminary experiment indicated that more than 30% of the cells transfected with pIC-Cre plasmid died in presence of G418, presumably because they had lost the Neo R cassette. pIC-Cre plasmid was electroporated into Dct I-SceI-Neo/+ ES cells and the cells were cultured without G418. Twenty-four clones were picked up and their sensitivity was assessed by adding G418 on a duplicate plate: 8 clones were Neo S . All eight clones had lost the Neo R cassette as shown by PCR analysis (data not shown). Two ES clones (MF1 and MF2) were further selected on morphological criteria. Southern blot analysis further confirmed the deletion of the Neo R cassette in MF1 and MF2 clones (Fig. 3D).
To test whether Dct I-SceI/+ MF1 and MF2 clones are able to colonize the germ line, we injected MF1 and MF2 cells into C57BL/6N blastocysts, and thereafter transferred the embryos to pseudo-pregnant females. Twelve and ten chimeras were produced from MF1 and MF2 cells, respectively. Altogether 18 chimeras were more than 95% chimeric, based on their coat colour pattern. Several male chimeras were mated to C57BL/6N females. Half of their progeny was Dct I-SceI/+ , indicating that the

Insertion of Lago1 at the Dct Locus
We wished to repeatedly introduce gene of interest at the Dct locus. As a first attempt, we used the Lago1 gene. The HR1 repair vector contained Lago1, a Neo R cassette framed with loxP sites, two regions of homology with the Dct I-SceI allele, 1.4 and 4.5 kb in length, and a HSV-TK negative selection cassette (Fig. 4A). The construction of the HR1 repair vector relied on the GatewayH technology (see Materials and Methods, and Fig. 1). We assessed the rate of insertion of Lago1 gene at the Dct locus following DSBinduced HR. MF1 ES cells were electroporated with supercoiled HR1 either with or without an I-SceI expressing plasmid. Two different I-SceI-expressing plasmids were tested: (i) pCMV-I-SceI in which I-SceI expression is driven by the cytomegalovirus promoter [3,26], and (ii) pCAG-I-SceI, where I-SceI is expressed under the control of the CAG composite promoter (see Materials and Methods). The cells were exposed to G418 and gancyclovir. A total of 215, 235 and 252 colonies resistant to both antibiotics were obtained when MF1 ES cells were transfected with HR1 alone, and in combination with pCMV-I-SceI or pCAG-I-SceI respec-  (Table 1). For each experiment, 136 colonies were individually picked up and PCR tested. Transfection with either HR1 alone or with HR1 and pCMV-I-SceI gave no targeted colonies. Transfection with HR1 and pCAG-I-SceI gave a positive PCR signal (data not shown), which was confirmed by Southern blot analysis (Fig. 4B and data not shown). Thus gene targeting using the HR1 repair vector and pCAG-I-SceI led to a frequency of HR that could be estimated at 1.4 6 10 27 events per treated cell. This frequency is not higher than that obtained with the conventional gene targeting procedures (generally range between 10 25 and 10 28 events per treated cell).
Elliott and colleagues (1998) reported previously that, in I-SceIinduced gene targeting with a transfected circular plasmid, the majority of recombination events occurred within 100 bp from the cleavage site [33]. Actually, the HR1 repair vector contains two short regions of homology with the targeted Dct I-SceI allele next to the I-SceI site. These regions are shown in Figure 4C. In the Dct I-SceI allele, the first short region of homology is located between exon 1 of the Dct gene and the I-SceI site. It encompasses 58 bp of Dct intron 1 sequence. Still in the Dct I-SceI allele, a second region of homology is located between I-SceI site and the end of loxP site. It encompasses 125 bp of Dct intron 1 sequence and loxP sequence.
In HR1 repair vector, both 58 bp and 125 bp regions are located between the attB2 site and the Neo R cassette. We hypothesized that these homology regions, 183 bp in total length, could be used as an efficient repair template and would produce by HR a recombinant allele harbouring neither a Lago1 gene nor a Neo R cassette (Fig. 4C). Hence, clones that have undergone HR would die in the presence of G418.

Insertion of H 2 B-mcherry at the Dct Locus
We thus decided to remove the two short regions of homology (including the loxP site) and to construct a novel repair vector. Therefore, a second destination vector (DV2) was produced using the H 2 B-mCherry reporter gene [31]. The HR2 repair vector was constructed (Fig. 2). It carries H 2 B-mCherry, a Neo R cassette flanked with FRT sites, two regions of homology with the Dct I-SceI allele, 1.4 and 4.5 kb in length, and a HSV-TK negative selection cassette (Fig. 5A). By contrast with HR1, HR2 displays neither a short region of homology with the Dct gene next to the I-SceI site nor a loxP site.
We assessed the rate of targeted insertion of H 2 B-mCherry at the Dct locus by I-SceI-induced HR. MF1 ES cells were electroporated with supercoiled HR2 plasmid alone or with either pCMV-I-SceI or pCAG-I-SceI expression plasmids (Fig. 5A). As an additional control, MF1 ES cells were electroporated with linearized HR2 plasmid. The cells were cultured in the presence of G418 and gancyclovir. Colony counting revealed 397 and 370 colonies in presence of pCMV-I-SceI and pCAG-I-SceI plasmids respectively, and 653 colonies in the absence of the meganuclease (Table 1). Electroporation with linear HR2 plasmid, representative of a conventional gene targeting experiment, revealed 635 resistant colonies to both antibiotics. For each experiment, 144 colonies were individually picked up, amplified and PCR tested. Electroporation with linear HR2 plasmid resulted in one targeted clone (clone 7E). No targeted clones were seen in the supercoiled HR2electroporated MF1 ES cells. The same results were obtained in the MF1 ES cells electroporated with both supercoiled HR2 and pCMV-I-SceI plasmids. Electroporation with HR2 and pCAG-I-SceI led to one targeted clone (clone 11E). Both conventional (linear HR2) and I-SceI-mediated (supercoiled HR2 and pCAG-I-SceI) gene targeting HR were confirmed by Southern blot analysis (Fig. 5B). Thus gene targeting using the HR2 repair vector and pCAG-I-SceI led to a frequency of HR that could be estimated at 1.6610 27 events per treated cell whereas conventional gene targeting led to a frequency of 2.7610 27 events per treated cell. Therefore, the I-SceI-induced DSB strategy does not seem to improve the frequency of HR at the Dct locus.

Transient Expression of I-SceI Triggers DSB in Dct I-Scei/+ ES Cells
To test whether the meganuclease could indeed trigger DSB in vivo at the Dct locus in Dct I-SceI/+ ES cells, we electroporated I-SceIexpressing plasmids into MF1 ES cells and assayed the DNA lesion at the I-SceI site using a sensitive technique, known as ligationmediated PCR (LM-PCR), that allows the specific detection of  breaks in a defined region of genomic DNA [34]. pCMV-I-SceI, pCAG-I-SceI and a mock plasmid were independently electroporated into MF1 ES cells and four hours later the genomic DNAs were extracted. In a first step, we tested the specificity and sensibility of the LM-PCR on transfected ES cells. We used two sites recognized by restriction endonucleases: (i) the I-SceI site, whose cleavage was under evaluation; (ii) a PstI site at position +52 relative to the I-SceI site. Approximately 2 mg of extracted genomic DNA from mock plasmid-transfected Dct I-SceI/+ MF1 ES cells were digested with PstI or I-SceI restriction endonuclease respectively. Then the digested DNA was heated to allow annealing with a first Dct gene-specific primer (LM-C1) located at position -185 relative to the I-SceI site. This was followed by LM-C1 primer extension that terminated at the site of a break to produce a blunt-ended DNA, which was then ligated to an asymmetrical synthetic double-stranded linker. The newly synthesized DNA molecule was denatured to allow annealing with a second Dct gene-specific primer (LM-C2) located at position -174 relative to the I-SceI site and amplification in a PCR reaction with linker primer. The PCR-amplified products were exponentially amplified by nested PCR using a third Dctgene-specific primer (LM-C3) located at position 2148 relative to the I-SceI site and linker primer, as shown in Fig. 6A. Finally, the PCR products were separated on an agarose gel, alkaline blotted to a nylon membrane, and hybridized with a radioactive probe which does not overlap the primer sequences. PCR products of the predicted sizes, 148 bp for I-SceI digestion and 200 bp for PstI digestion, were seen (Fig. 6B). These data indicate that the LM-PCR technique allowed the specific detection of a cleavage generated in vitro on the genomic DNA from MF1 ES cells.
In a second step, we evaluated the ability of the meganuclease to trigger a DSB in vivo at the Dct locus. Approximately 2 mg of extracted genomic DNA from mock plasmid-transfected MF1 ES cells, pCMV-I-SceI-transfected MF1 ES cells and pCAG-I-SceItransfected MF1 ES cells were directly analyzed by LM-PCR. Figure 6C shows that no DNA lesions occurred at significant level at the I-SceI site in the absence of I-SceI expression. When the genomic DNA from pCMV-I-SceI-transfected MF1 ES cells was used as a template in the LM-PCR reaction, a 148 bp amplification product was detected, showing that expression of the meganuclease in MF1 ES cells triggered DSB at the I-SceI restriction site. LM-PCR in which the genomic DNA from pCAG-I-SceI-transfected cells was used as a template produced similar results (Fig. 6C), suggesting that both I-SceI expression vectors were equally efficient in triggering DSB at the target locus.

Discussion
In this report, we provided evidence that I-SceI-induced DSB in ES cells does not improve the efficacy of the gene targeting methodology at the Dct locus compared to the conventional approach. Electroporation was used to introduce the I-SceI expressing plasmids into ES cells and a low efficiency of transfection could explain these results. However, we were able to detect the expression of enhanced green fluorescent protein (GFP) from Aequora victoria by fluorescence-activated cell sorter (FACS) analysis in 65% of a population of CK35 ES cells electroporated with a plasmid containing the CMV promoter driving the expression of the GFP, indicating efficient electroporation (data not shown). These data agree with a previous report [35]. Both repair vectors (HR1 and HR2) contain 5.9 kb of Dct homology. The same homology was used previously by Guyonneau et al. [23] to inactivate the Dct gene in ES cells, indicating that such a length is efficient for a gene replacement event.
However, we cannot exclude that increasing the length of homology may improve I-SceI-mediated HR at the Dct locus. Since the first repair vector (HR1) contained short regions of homology in the vicinity of the I-SceI site, we hypothesized that these regions were preferentially used to repair the DSB, thus generating homologous recombinant clones that did not integrate the Neo R cassette and died eventually in the presence of G418. Therefore, a second repair vector (HR2) with no homology to the sequence surrounding the I-SceI site was generated, but we still failed to demonstrate improvement of the frequency of gene targeting.
Previous experiments suggested that non-homologous recombination may be more efficient than plasmid-to-chromosome HR to repair a chromosomal DSB introduced by I-SceI. Indeed, when mouse Ltkfibroblasts carrying a selectable herpes simplex virus thymidine kinase (tk) gene mutated by the 18-bp I-SceI site, were electroporated with I-SceI meganuclease and a repair plasmid with the functional tk gene, tk + clones were recovered. However, all analyzed tk + cells contained deletions that restored the reading frame of the tk gene, indicating that the recovery of a functional tk gene did not occur through HR of the integrated tk gene with a transfected tk fragment, but rather via resection and ligation [36]. These data were obtained in mouse Ltk 2 fibroblasts, not in ES cells. This deserves mention since distinct differences in frequencies of targeted integration driven by a DSB among cell types have been reported. HR after cleavage by a zinc-finger nuclease (ZFN) at the CCR5 locus in presence of cognate donor linear and circular episomes was more efficient in a panel of immortalized cell lines from human leukemia than in human stem cells, such as cord blood CD34 + hematopoietic cells and human ES cells [37]. It has also been reported that the rate of ZFN-mediated gene targeting at the Rosa26 locus was higher in primary fibroblasts from adult mice than in murine ES cells [38]. These results suggest that DSBinduced gene targeting may be lower in ES cells than in somatic cells. This contention seems inconsistent with reports showing that HR is the predominant pathway to repair DSBs in ES cells, whereas somatic cells utilize non-homologous end joining (NHEJ) [39,40]. It has also been reported that ES cells that had been allowed to differentiate preferred the error-prone NHEJ pathway to the high-fidelity HR to repair DNA DSBs [39]. Because ES cells and somatic cells are intrinsically different in the extent to which they preserve their genomic integrity [41], it was important to assess that our experiments were made with genuine ES culture rather than differentiated ES culture. We confirmed that the Dct I2SceI/+ CK35 cells are truly pluripotent ES cells, able to colonize the germ line. Furthermore, CK35 ES cells have been previously used to demonstrate highly efficient gene targeting after DSB [3].
We observed that the repair vector was not inserted at the Dct locus in the majority of neomycin-and gancyclovir-resistant clones. One possible explanation was that I-SceI did not cleave its recognition sequence in Dct I2SceI/+ ES cells. However, we showed here the existence of I-SceI-induced DSBs at the Dct locus using LM-PCR. This result does not rule out the possibility that the efficiency of cleavage by I-SceI is an important factor to favor HR at a given individual locus. Recent findings by Daboussi et al. [42] support this conclusion. They found that chromatin accessibility modulates the ability of meganucleases to induce targeted gene modification in human 293-H cells.
Seminal experiments of gene targeting in ES cells based on an I-SceI-induced gene replacement system were first performed with mutated resistance genes integrated in chromosomal sequences [8], and later extended to natural endogenous genes, Hprt, Villin and Dbx1 genes [3,4,43]. Importantly, random integrations could not be detected in the previous studies performed with I-SceI in mouse ES cells when selection strategies were specifically designed to single out the gene targeting events and eliminate the nonrecombinant recombination events [3,4,8]. Altogether, the number of actual genes that have been efficiently targeted following I-SceI-mediated HR in ES cells is still limited. It is widely accepted that the efficiency of conventional HR depends on the target locus. Our results suggest that, similarly, efficiency of gene correction by DSB-induced HR may be highly dependent on the targeted locus. We anticipate that deeper analysis of the meganuclease, repair vector, target locus and cells that do not show enhanced HR by DSB may also shed light on the nature of the factors that contribute to gene targeting in mammalian cells.