4826–4833 Nucleic Acids Research, 2001, Vol. 29, No. 23
© 2001 Oxford University Press
Restriction enzymes increase efficiencies of illegitimate DNA integration but decrease homologous integration in mammalian cells Palaniyandi Manivasakam, Jiri Aubrecht, Samy Sidhom and Robert H. Schiestl* Department of Cancer Cell Biology, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA Received August 8, 2001; Revised and Accepted October 18, 2001
ABSTRACT Mammalian cells repair DNA double-strand breaks by illegitimate end-joining or by homologous recombination. We investigated the effects of restriction enzymes on illegitimate and homologous DNA integration in mammalian cells. A plasmid containing the neoR expression cassette, which confers G418 resistance, was used to select for illegitimate integration events in CHO wild-type and xrcc5 mutant cells. Co-transfection with the restriction enzymes BamHI, BglII, EcoRI and KpnI increased the efficiency of linearized plasmid integration up to 5-fold in CHO cells. In contrast, the restriction enzymes did not increase the integration efficiency in xrcc5 mutant cells. Effects of restriction enzymes on illegitimate and homologous integration were also studied in mouse embryonic stem (ES) cells using a plasmid containing the neoR gene flanked by exon 3 of Hprt. The enzymes BamHI, BglII and EcoRI increased the illegitimate integration efficiency of transforming DNA several-fold, similar to the results for CHO cells. However, all three enzymes decreased the absolute frequency of homologous integration ∼2-fold, and the percentage of homologous integration decreased >10-fold. This suggests that random DNA breaks attract illegitimate recombination (IR) events that compete with homology search. INTRODUCTION In the eukaryotic genome, DNA double-strand breaks (DSBs) can occur during cellular processes such as DNA repair, recombination and replication; the early prophase of meiosis, V(D)J recombination or as the result of exposure to DNA damaging agents. The repair of DSBs is important for the maintenance of genomic integrity and cellular survival because unrepaired DSBs will result in the loss of genetic information,
which may be lethal. Inappropriate repair of DSBs can result in chromosomal rearrangements, which can lead to cell transformation. Saccharomyces cerevisiae cells repair DSBs by homologous recombination, mainly mediated by gene products in the RAD52 pathway. In the absence of RAD52, yeast cells repair DSBs by non-homologous end joining (NHEJ), which requires the Ku homologous proteins. NHEJ has been studied in yeast amongst other approaches by transformation with a linear DNA molecule that contains a selectable marker, which does not have any homology to the genome. Schiestl and Petes (1) used a BamHI fragment containing the URA3 marker to transform a yeast strain in which the entire URA3 gene had been deleted. After addition of the BamHI restriction enzyme to the transformation mixture the efficiency of integration increased several-fold and integrations occurred into BamHI sites. These events were designated restriction enzyme-mediated integration (REMI) events. This study was extended to investigate how two different compatible and non-compatible ends are repaired (2), by restricting the DNA with one enzyme and adding another enzyme to the transformation mixture. Mammalian cells repair DSBs both by homologous recombination and by illegitimate recombination (IR). However, homologous integration frequencies are 100–1000-fold less frequent than illegitimate integration (3,4), the major obstacle for gene targeting in mammalian cells. The molecular mechanism of recombination in mammalian cells is under intensive investigation. Different complementation groups of ionizing radiation-sensitive rodent cell mutants have been identified and three of them designated X-ray repair cross-complementing (XRCC) groups 5, 6 and 7 (5). The genes defective in groups 5 and 7, deficient in the mutants xrcc5 and xrcc7, respectively, encode components of a DNA-dependent protein kinase (DNA-PK), a complex possessing DNA end-binding and protein kinase activity (6–8). The DNA-binding subunit of DNA-PK is a heterodimer of 70- and 80-kDa subunits, named Ku70 and Ku80, respectively. Ku80 is deficient in group 5 cells (9). Ku is an abundant nuclear protein identified originally as an autoantigen from various autoimmune patients (10). The Ku protein binds to free double-stranded DNA ends with 5′- or
*To whom correspondence should be addressed at present address: Departments of Pathology and Environmental Health, UCLA School of Medicine and Public Health, 650 Charles E. Young Drive South, 71-295 CHS, Los Angeles, CA 90095, USA. Tel: +1 310 267 2087; Fax: +1 310 267 2578; Email:
[email protected] Present addresses: Palaniyandi Manivasakam, CombinatoRx, 650 Albany Street, Boston, MA 02118, USA Jiri Aubrecht, Pfizer Global Research and Development, Groton, CT 06340, USA The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors
Nucleic Acids Research, 2001, Vol. 29, No. 23 4827 3′-single-stranded protruding (PSS) ends or blunt ends, nicked DNA, and duplex DNA ending with stem–loop structures (10), which suggests a possible role in DNA repair and recombination. These mutants are defective in DSB rejoining and have been shown to possess defects in components of DNA-PK. They also exhibit defects in the rejoining steps in V(D)J recombination, a site-specific recombination process required for rearrangement of DNA to generate variability in genes encoding immunoglobulins. Several investigators have utilized restriction enzymes to study the effect of DSB repair in mammalian cells. It has been shown that introduction of restriction enzymes into mammalian cells, either by electroporation or by calcium precipitation, produces cell killing, chromosomal aberrations, gene amplification and increased mutation frequency at several loci [for review (11)]. Co-transfection with restriction enzymes HindIII and XbaI induces DNA integration into mouse Ltk– cells (12). Inducible expression of EcoRI in CHO cells showed toxicity and chromosomal aberrations (13). Investigators showed that intrachromosomal homologous recombination can be induced by electroporation with restriction enzymes or intracellular expression of restriction enzymes (14,15). Using a loss of function assay in the APRT gene, Sargent et al. (16) showed that DSBs created by I–SceI induced homologous recombination 100-fold between repeats, but IR was stimulated 1000-fold. This indicates that the preferential mode of DSB repair in mammalian cells is NHEJ. In the present study, we investigated the effects of restriction enzymes on illegitimate integration and homologous gene targeting in mammalian cells. We found that restriction enzymes increased the frequency of illegitimate integration, which was dependent on the Ku80 function. Surprisingly, we also found that the restriction enzymes decreased the frequency of homologous integration, indicating a direct competition between the two pathways. MATERIALS AND METHODS Plasmids Plasmid PMA159 was derived from pUC and contains the neoR gene. This plasmid was constructed in two steps. Digestion of pBK-CMV plasmid (from Stratagene) with AvrII restriction enzyme liberated the 1.6 kb neoR expression cassette. BamHI linkers were ligated to the ends of the fragment after filling in the 5′ single-stranded ends with Klenow. neoR was inserted into the BamHI site of pUC. After partial digestion, the 5′ end of the BamHI site was eliminated by filling in the single-stranded ends and ligating the blunt ends. Plasmid J3N contains the neoR gene flanked by exon 3 of the Hprt gene. A 7 kb BamHI fragment containing exons 2 and 3 of the Hprt gene was ligated into pUC19 missing an EcoRI site. The neoR expression cassette obtained by AvrII digestion was blunt ended with Klenow, and XhoI linker was attached. The neoR fragment containing XhoI ends was ligated into the XhoI site, which resides in exon 3 of the Hprt gene, yielding plasmid J3N. This plasmid has homology to Hprt of 4.5 and 2.5 kb flanking the neoR fragment. J3NB plasmid was constructed by filling in two BglII sites (located 1 and 3.5 kb from the 5′ end), J3NR by filling in the two EcoRI sites (located 4 and 5.5 kb from the 5′ end) and J3NH by filling in the HindIII site (located 3.7 kb
from the 5′ end) in plasmid J3N. Plasmid pcDNA3.1/His/lacZ (Invitrogen, Carlsbad, CA) control vector has been used for transient transfection to determine the effect of restriction enzymes on DNA uptake. Cell lines CHO cells (CHO-K1; ATCC) and the xrs5 mutant cells (obtained from Penny Jeggo, MRC Cell Mutation Unit, University of Sussex, UK) were grown in Cellgro DMEM (Mediatech, Herndon, VA) supplemented with proline (3.7 mM), 10% fetal bovine serum (Life Technologies, Rockville, MD), 100 U/ml penicillin and 100 µg/ml streptomycin in a humidified air incubator at 37°C, 5% CO2 and 95% humidity. Embryonic stem (ES) cells (ES-D3; ATTC) were cultivated on monolayers of irradiated feeder cells (STO-TN) in DMEM supplemented with 10% fetal bovine serum (Life Technologies), LIF (100 U/ml), 0.2 mM mercaptoethanol (Life Technologies), 100 U/ml penicillin and 100 µg/ml streptomycin (Mediatech) in a humidified air incubator at 37°C, 5% CO2 and 95% humidity. The feeder cell line was developed from embryonic fibroblast cells (STO; ATTC). That feeder cell line was transfected with plasmid PMA159 and selected in the presence of G418 (Life Technologies). One stable transfected clone of STO fibroblast cells (STO-TN) conferring resistance to 1 mg/ml G418 was selected for preparation of feeder cells. The STO-TN cell line carries a mutated hprt gene and neoR expression cassette that allows simultaneous selection with 6-thioguanine (6-TG) and G418. A 90% confluent culture of STO cells was trypsinized, suspended in DMEM and irradiated with a 50 Gy dose of γ irradiation with a 60Co γ-ray source at a dose rate of 12.2 cGY/s. After irradiation, the cells were frozen until further use. The frozen feeder cells were thawed 3–10 days before seeding of ES cells, then plated at a density of 2.5–4 × 106 cells per 10 cm dish and grown to near confluency. Electroporation of cells CHO or xrs5 cells were grown to 90% confluence in 225 cm2 culture flasks (Corning, Corning, NY) and trypsinized using trypsin/EDTA (Mediatech). The detached cells were collected in 10 ml of fresh culture medium, washed twice with phosphatebuffered saline (PBS) and resuspended in electroporation buffer (272 mM sucrose, 7 mM Na2HPO4, 1 mM MgCl2, pH 7.4). Approximately 3–5 × 106 cells in 0.8 ml of electroporation buffer were mixed with 15 µg of linearized plasmid (PMA159), 30–480 U of restriction enzyme or an appropriate amount of restriction enzyme buffer and restriction enzyme storage buffer. The electroporation was performed using Bio-Rad Gene Pulser at 0.3 kV, 960 µF and 200 Ω. The electroporated cells were diluted with fresh culture medium, plated on to three 10 cm dishes (Falcon) and incubated. After 24 h, the medium was supplemented with G418 to a final concentration of 750 µM. To determine the viability after electroporation, 100 and 500 cells were plated on to two 10 cm dishes. To minimize the occurrence of satellite clones the cultures were incubated without disturbance for 11–14 days. The colonies of surviving cells were stained using Giemsa stain (Sigma, St Louis, MO). ES cells were grown to confluence on 10 cm dishes covered with STO-TN feeder cells. After trypsinization the cells were resuspended in fresh medium and washed twice in PBS. Approximately 8–12 × 106 cells were electroporated in PBS using Bio-Rad Gene Pulser at 0.3 kV and 500 µF. The
4828 Nucleic Acids Research, 2001, Vol. 29, No. 23
electroporated ES cells were seeded on three 10 cm dishes covered with STO-TN feeder cells. After 24 h of incubation, the medium was supplemented with 500 µg/ml G418. The surviving colonies of stable integrants including random and homologous recombination events were counted after 10 days. At that time, the medium was replaced with one containing 30 µM 6-thioguanine (TG) (Sigma). Surviving colonies were counted after 5–7 days. Transient transfection to determine DNA uptake Plasmid pcDNA3.1/His/lacZ (Invitrogen) was digested with BglII, and cells were prepared and electroporated with the linear plasmid as described above in the presence or absence of the BglII enzyme. The control in the absence of the enzyme contained an equivalent amount of the enzyme storage buffer. The cells were then incubated overnight in the presence of X-gal. Cells were counted and screened for blue color under the microscope. Southern blot analysis Stable transformants were obtained by transforming the cells with PMA159 plasmid, which was restricted with BamHI and transfected with or without BamHI enzyme or with the plasmid digested with KpnI and transfected with or without KpnI. Individual clones were expanded by growing in 10 cm dishes. The genomic DNA was isolated and Southern blot analyses were performed according to standard procedures using the 1.6 kb neoR gene as probe. RESULTS We investigated the effects of restriction enzymes on illegitimate and homologous DNA integrations in mammalian cells. To study IR, a plasmid containing the neoR selection marker was used. A DNA fragment containing the neoR gene was cloned into the BamHI site of pUC plasmid and one BamHI site was eliminated by filling in and religation. All multicloning sites are unique in this plasmid. The plasmid was linearized with various restriction enzymes and transfected in the presence of enzyme into the cells by electroporation. Initially we determined the effects of different concentrations (30, 120 and 480 U per transfection vial) of the BamHI enzyme on the frequency of integration. The same increase in the efficiency of integration was seen with all three concentrations, and thus, 30 U were used for all further experiments. Since restriction enzyme buffers can produce chromosomal aberrations (17,18), the same concentration of restriction enzyme buffer and storage buffer as used for the restriction digests was used in all controls. The restriction enzymes BamHI, BglII and EcoRI, which produce 5′ PSS ends, and KpnI, which produces 3′ PSS ends, increased the efficiency of illegitimate integration 2.5–5-fold (Table 1) in CHO cells. PstI, SmaI and HindIII restriction enzymes, producing 3′ PSS ends, blunt ends and 5′ PSS ends, respectively, did not induce REMI. Addition of PstI decreased the integration efficiency 2-fold. We wanted to determine whether this decrease was due to possible toxic effects of the enzyme. The cells were transfected with three different concentrations: 30, 120 and 480 U for each of the enzymes used in this study. As a control, PvuII, which is known to be a more potent inducer of toxic effects and chromosomal aberrations in mammalian cells than enzymes
producing PSS ends (19), was used. Except for PvuII, which showed a 50% reduction in colony forming unit (CFU), none of the other enzymes showed any significant level of reduction in CFU with 30 U (data not shown). At the highest concentration of 480 U, the enzymes BglII and PvuII, but none of the other enzymes used including PstI, decreased the CFU significantly (data not shown). These results indicate that the restriction enzymes that were used for the REMI experiment were not toxic to the mammalian cells at the concentration used for integration studies (30 U). REMI events in yeast usually integrate into restriction sites in the genome by micro-homology-mediated integration, and both restriction sites at the ends of the integrating DNA are maintained (1,20). This facilitates detection of such an event simply by digestion of the genomic DNA with the same restriction enzyme that was used for REMI, and performing a Southern blot that reveals the fragment size of the integrating vector. We performed Southern blot analysis with twelve clones obtained after a transformation with the BamHIdigested plasmid PMA159 in the presence of BamHI (data not shown). This experiment indicates that a single copy integration event recreated both flanking BamHI sites in only 1 out of 12 transformed clones. In the control in which no enzyme was added, none of the events was flanked by BamHI sites. In the experiment with KpnI, 1 out of 10 events recreated the flanking KpnI sites, and in the control, none of the events recreated both KpnI sites. Control cells as well as restriction enzyme-treated cells showed multiple fragments of various sizes, suggesting multiple copies at a single site and/or multiple integration events at multiple sites. It has been reported that radiosensitive cells are more susceptible to the toxic effects and chromosome aberrations caused by restriction enzymes (21). We used the xrs5 cell line that is deficient in Ku80 to study the toxicity and integrationmediating effect of restriction enzymes. Approximately 55% of the cells survived with 30 U of the BamHI enzyme, while only 14% survived with 60 U (Table 2). In the presence of 30 and 60 U of the BamHI enzyme, REMI efficiency decreased to ∼25 and 8%, respectively, in xrs5 cells, which is a greater decrease than caused by toxicity alone (Table 2). Thus, Ku80 seems to be required for REMI. Next, we investigated the effects of restriction enzymes on the efficiency of gene targeting in ES cells. For this study, we constructed a plasmid containing the neo R gene as a selection marker flanked by the Hprt gene (Fig. 1). This plasmid can be used to select for illegitimate integration as well as for gene targeting. Integration of the plasmid will give rise to G418-resistant clones. Homologous integration results in disruption of the Hprt gene, which can be selected for with 6-TG. As the hprt locus is hemizygous in XY ES cells, targeted clones can be isolated by direct selection in 6-TG for lack of hprt function. We tested the effect of BamHI, BamHI-E77K, BglII, EcoRI and HindIII restriction enzymes on gene targeting in ES cells. In the presence of BamHI, BglII and EcoRI the total number of integration events increased 2–4-fold as seen in CHO cells (Table 3). Surprisingly, however, in the presence of enzyme, the actual number of homologous integration events decreased >2-fold, and the percentage of homologous integration as a fraction of all integration events decreased down to >10-fold (Table 3). The addition of HindIII, on the other hand, did not increase the
Nucleic Acids Research, 2001, Vol. 29, No. 23 4829
Table 1. Effect of different restriction enzymes on REMI in CHO cells Plasmid digested with
Enzyme added
Integration (fold over control)
BamHI
–
1
BamHI
BamHI
3.74 ± 0.28
BglII
–
1
(177)
BglII
BglII
5.49 ± 0.49
(928)
EcoRI
–
1
(223)
–
EcoRI
EcoRI
2.40 ± 0.24
(479)
**
PstI
–
1
(935)
–
PstI
PstI
0.46 ± 0.05
(420)
**a
SmaI
–
1
(254)
–
SmaI
SmaI
0.84 ± 0.05
(233)
–
KpnI
–
1
(445)
–
KpnI
KpnI
5.35 ± 0.96
(1679)
**
1
(650)
–
1.16 ± 0.08
(719)
–
HindIII HindIII
HindIII
No. of clones
Significance
(298) (1009)
**
**
aThe significance level for PstI indicates a significant decrease in transfection efficiency whereas all the other significance levels represent a significant increase. *, significant at 5% level; **, significant at 1% level, determined with the t-test. The enzymes used to digest the plasmid are in the left column and the enzymes added to the transformation mixture during transformation are in the next column. See Materials and Methods for details. There were no sites on the linear plasmid for enzymes added to the transformation mixture. The integration efficiency is represented as fold increase over the control experiments. The values are represented as the mean of 4–15 independent experiments ± standard deviation.
Table 2. Effect of BamHI restriction enzyme on REMI and colony-forming efficiency of xrs5 and CHO cells Cell line
CHO
xrs5
BamHI units
Number of G418-resistant clones
Colony-forming efficiency
DNA + buffer
DNA + BamHI
DNA + buffer
DNA + BamHI
0
61 ± 0.7
64 ± 1.4
115 ± 7.1
124 ± 12.0
30
61 ± 6.4
116 ± 4.2
134 ± 5.7
127 ± 2.1
60
61 ± 2.1
128 ± 3.5
150 ± 2.1
109 ± 2.1
120
59 ± 0.7
110 ± 2.8
131 ± 12.0
118 ± 2.8
0
33 ± 2.8
37 ± 9.1
54 ± 2.1
57 ± 4.9
30
35 ± 7.1
9 ± 2.1
50 ± 0.7
32 ± 4.9
60
33 ± 4.9
3 ± 1.4
50 ± 5.7
8 ± 2.8
120
32 + 1.4
1 ± 0.7
37 ± 2.8
2 ± 0.7
Plasmid PMA159 was digested with BamHI and linear DNA was electroporated into xrs5 and CHO cells. Integrants were selected in the medium containing G418. For the colony-forming efficiency, after electroporation, appropriate dilutions were made and plated without any selection. DNA with buffer is a control, without DNA no G418-resistant colony was obtained. The values are the mean of two experiments ± range.
efficiency of illegitimate integration and also did not decrease the efficiency of homologous integration. It was theoretically possible that the restriction enzymes, which increased the frequency of illegitimate integration, somehow inhibited uptake of the DNA into the cells. This would potentially cause a reduction in the frequency of homologous recombination, and the measured frequency of REMI events would be an underestimate. To address this question, we performed a control experiment in which the linear
plasmid pcDNA3.1/His/lacZ containing the lacZ gene was used. This plasmid was co-transfected into cells with or without the BglII enzyme and the frequency of blue cells was counted under the microscope. This frequency was 1.8% in the presence of the enzyme and 1.3% in its absence. The difference was not significant. Thus, the restriction enzyme did not inhibit DNA uptake. Since restriction enzymes show non-specific weak DNA binding (22), restriction enzymes could possibly bind to the
4830 Nucleic Acids Research, 2001, Vol. 29, No. 23
Figure 1. Substrate for homologous and illegitimate integration. A 7 kb BamHI genomic fragment of Hprt containing exon 2 and 3 and introns 2 and 3 was cloned into the BamHI site of PUC. The NEO gene was inserted into exon 3 (see Materials and Methods for details). Black boxes represent Hprt sequences and gray boxes indicate neo and PUC sequences. A homologous recombination event is shown on the left, upon which the phenotype of the cells changes to G418 and 6-TG resistance. On the right, an IR event is shown, which renders the cells G418 resistant but they remain 6-TG sensitive.
transforming DNA, and in some way interfere with homology search or otherwise inhibit homologous integration. To address this possibility we used purified BamHI-E77K protein, which binds to the BamHI site but cleaves DNA at a rate 1000-fold lower than that of wild-type enzyme (22). The BamHI-E77K protein did not increase the efficiency of illegitimate integration (Table 3) and did not appreciably decrease the frequency of homologous gene targeting. Thus, only enzymes that increase the efficiency of illegitimate integration also lead to a decrease in the frequency of gene targeting. DISCUSSION Efficiency of REMI events with different enzymes Co-transfection of the restriction enzymes BamHI, BglII, EcoRI and KpnI increased the efficiency of linearized plasmid integration up to 5-fold in CHO cells irrespective of the nature of the integrating DNA fragment ends (5′ or 3′ PSS or blunt ends). PstI, SmaI and HindIII restriction enzymes did not induce REMI. In addition, the enzymes BamHI, BglII and EcoRI increased integration efficiencies of transforming DNA in mouse ES cells. These results suggest that the restriction enzymes are able to enter the nucleus and produce DNA breaks. The repair process may lead to insertion of the DNA substrate into these break sites. REMI has been characterized previously in yeast (1,2) and there are several differences between the current data and the previously published yeast data. First, in yeast, in the majority of REMI events the restriction sites at both sides of the integrated plasmid are recreated, indicating insertion of the plasmid into genomic restriction sites without modification of the ends by simple annealing and ligation. In mammalian cells, however, we find that the restriction sites are absent in the majority of integration events. This is likely to be due to modification of the ends before or during end joining. This is in agreement with the previously published data (23,24). In
mouse cells, processing of the ends is in agreement with exonuclease activity degrading the ends before integration (23). In human cells, filling of PSS ends as well as loss of one to several hundred nucleotides was found in 24 out of 25 events during end joining (24). In fact, Derbyshire et al. (24) isolated such an end-joining activity tightly associated with the human homologous pairing activity and an intrinsic 3′–5′ exonuclease activity. The restriction enzymes BamHI, BglII and KpnI mediated integration events in yeast (2) as well as mammalian cells. HindIII was not active in yeast or mammalian cells in our experiment. In mouse Ltk– cells, however, 100 U of HindIII increased DNA integrations 4-fold (12), indicating that the 30 U of HindIII we used might have been below the detection threshold for an effect of HindIII’s activity. EcoRI did not increase the efficiency of DNA integration in yeast (2) whereas it was active in mammalian cells. The fact that three out of 10 integration events, after transformation in the presence of EcoRI, in yeast were flanked by EcoRI sites suggests that EcoRI might possess low activity to mediate integrations in yeast, which is not sufficient to raise the integration efficiency (2). In Dictyostelium, BamHI, EcoRI, Sau3A, ClaI and BglII catalyzed integration of plasmids containing pyr5-6, a homolog of the yeast URA3 gene (25,26). In Cochliobolus heterostrophus, HindIII was used to tag the TOX1 locus with hygB (27). Thus, HindIII catalyzed an increase in integration events in Cochliobolus, but not in mammalian cells or yeast. These data indicate that the inability to raise the frequency of integrations in our experiments may not be an intrinsic property of some restriction enzymes, but may rather depend on cellular environment, and the ability to enter cells or different transformation conditions. The crystal structures of several enzymes have been identified and the active sites of BamHI, EcoRI and EcoRV are structurally similar (28,29). However, their protein sequences are unrelated in most cases. Thus, within different cellular environments, some enzymes may be better substrates for degradation through proteases than others. Some enzymes may not enter the mammalian cells, or even if they enter the cells, they may not be active in the nucleus because their ionic requirements for activity are not met. In yeast, although digestion of a plasmid with Asp718 and addition of BglII restriction enzyme to the transformation mixtures significantly increased the efficiency of integration, the integrating DNA was not inserted into the restriction sites (2). This suggested that the DNA breaks caused by the enzymes open up the chromatin locally, which may lead to a higher accessibility of the genomic DNA for integration. This may also be true for mammalian cells, which would also lead to the lack of recreation of the restriction sites flanking the integrated vector. Electroporation of blunt end-producing endonucleases, PvuII, EcoRV and StuI, caused toxicity and induced small deletions of 1–36 bp, insertions, and combinations of insertions and deletions at the cleavage sites (30). Overexpression of EcoRI enzyme in CHO cells revealed that 80–90% of the surviving cells had chromosomal aberrations if the restriction enzyme is overexpressed in vivo for 45 min (13). Even though we did not see any toxic effect, we do not rule out the possibility of chromosomal aberrations in the transfected cells. Brenneman et al. (15) report that the plating efficiency of human cells decreased by 80–90% after treatment of cells with
Nucleic Acids Research, 2001, Vol. 29, No. 23 4831
Table 3. Effect of restriction enzymes on illegitimate and homologous integration of transfected DNA in ES cells Enzyme added
G418-resistant clones
J3N
–
221 ± 38
1
9.8 ± 5.1
4.43 ± 2.30
–
BamHI
893 ± 324
4.0 ± 1.4**
4.4 ± 1.3
0.49 ± 0.15
0.11 ± 0.05**
HindIII
254 ± 31
1.1 ± 0.13
9.7 ± 0.6
3.8 ± 0.23
0.86 ± 0.056
BamHI-E77K
269 ± 21.1
1.2 ± 0.1
10.5 ± 2.2
3.9 ± 0.82
0.89 ± 0.019
–
253 ± 71
1
8.9 ± 4.7
3.44 ± 1.32
–
BglII
925 ± 404
3.8 ± 1.8**
3.1 ± 1.6
0.37 ± 0.18
0.11 ± 0.06**
–
197 ± 98
1
7.5 ± 4.4
3.72 ± 1.01
–
EcoRI
347 ± 178
2.1 ± 0.75**
4.9 ± 2.7
1.51 ± 0.40
0.44 ± 0.19*
J3NB
J3NR
Fold increase
6-TG-resistant clones
Homologous integration (%) +enzyme/–enzyme fractiona
Plasmid
aThis is the effect of the enzyme on the frequency of homologous integration events. All the plasmids were digested with BamHI and precipitated with ethanol. The linearized plasmid was electroporated into ES cells with enzyme or with shipping buffer alone in the controls (see details in Materials and Methods). Stable G418-resistant colonies expressing the integrated neo R-containing plasmid result in 6-TG clones by integrating into the genomic HPRT locus. The values represent the mean and standard deviation of eight different cultures, except for HindIII (three different cultures) and BamHI-E77K (two cultures) where the mean ± range is given. **, significant at 1% level; *, significant at the 5% level. Significance was determined using the t-test for all enzymes except for BamHI-E77K.
higher doses of restriction enzymes. Thus, it is likely that we would also have seen toxicity at higher doses. Ku80 is required for REMI in mammalian cells Addition of restriction enzymes during transformation with xrs5 cells showed no increase in the efficiency of DNA integration but the colony-forming efficiency was reduced considerably. At the highest concentration, most of the cells were killed (Table 3). This suggests that Ku proteins are required for the integration of a vector into DSBs produced by restriction enzymes. xrs5 cells are deficient in Ku80 function. In mammalian cells, Ku proteins are needed for DSB repair and V(D)J recombination (6,7). Extrachromosomal homologous recombination and gene targeting by homologous recombination are not affected in cells lacking Ku80 (31,32); however, end joining is affected in those cells (33). Thus, Ku protein appears to play a critical role only in the illegitimate end-joining pathway but not in homologous recombination. Ku-deficient cells show more DNA degradation at free ends of plasmids (33). This might account for a reduced frequency of REMI in our experiments. The degradation of ends should, however, affect both spontaneous illegitimate integration and REMI to the same extent. Since we found that REMI was completely abolished, and spontaneous illegitimate integration only decreased to ∼60% (Table 2), Ku is specifically implicated in DSB-induced illegitimate integration. For spontaneous integration events, an alternative pathway, however, may exist in agreement with the data of Liang et al. (32). Other studies report a defect in plasmid integration in the xrs6 mutant, which is dependent on the DNA concentration. At low concentration, very little or no difference was observed between mutant and wild-type cells, but at high plasmid concentration a 5–10-fold difference was found (34). In that study, the polybrene and calcium phosphate methods were used, which are quite different from the electroporation method that we used, which makes a comparison of the DNA concentrations difficult.
Effect of restriction enzymes on gene targeting We determined the effect of the BamHI enzyme on the frequency of homologous integration of an Hprt fragment flanked by BamHI sites. In the presence of restriction enzymes BamHI, BglII and EcoRI the fraction of gene-targeting events among the overall number of integrants decreased up to 10-fold in ES cells. It is, however, well documented that DSBs can stimulate homologous recombination in yeast and mammalian cells. DSBs and gaps in the region of homology carried by targeting plasmid stimulate recombination 33–140-fold in the Hprt locus of ES cells (35). Homologous recombination rates at a given locus between homologs in somatic mammalian cells are generally low: 10–8–10–5 per cell generation (36). Frequencies of gene targeting are also low (37). Homologous intrachromosomal recombination between partially duplicated Hprt genes of the human fibrosarcoma line HT1080 increased by treatment with XbaI or I–SceI restriction endonucleases (15). The recombination frequency increased only if the restriction sites are present within the repeats. Furthermore, expression of I–SceI enhanced homologous recombination several thousand-fold between repeats at integrated defective copies of the neomycin gene when the I–SceI site was present in one of the repeats (32). Since BglII and EcoRI have sites in the target, but those sites have been removed from the targeting vector, we expected that the frequency of homologous integration would be enhanced. On the contrary, we found that the frequency of homologous integration was 2-fold reduced after co-transformation of restriction enzymes. The result that the frequency of homologous recombination was decreased 2-fold after co-transformation with restriction enzymes is only possible if 100% of the cells that received the transforming plasmid also received the restriction enzyme. If some cells receive the plasmid alone, without the restriction enzyme, this decrease is an underestimate of the true effect. The degree of this underestimate depends on the number of cells co-transfected. For instance, if 50% of the cells received the plasmid alone the frequency of gene targeting needs to be
4832 Nucleic Acids Research, 2001, Vol. 29, No. 23
completely abolished in the remaining 50% of the cells that were co-transfected with both the restriction enzyme and the plasmid to result in an ∼2-fold overall decrease. Most previous experiments suggest that the illegitimate endjoining mechanism is more active than homologous recombination in mammalian cells. For instance, DSBs created in mouse cells are repaired preferentially by illegitimate end joining rather than by targeted homologous recombination with an exogenous donor sequence (38). Furthermore, in a tandem duplication of the APRT gene expression of I–SceI stimulates homologous recombination ∼100-fold; however, IR is stimulated >1000-fold (16). Our results show that an increase in DSB-induced illegitimate integration actively reduced the frequency of homologous integration. Assuming 40% AT content, there may be an estimated 324 000 BamHI sites and ∼280 000 BglII and EcoRI sites in the mammalian genome. The results suggest that DSBs at these random sites in the genome actively compete as substrates for integration and effectively reduce the frequency of homologous integration. This could be because homology search takes longer in mammalian cells than illegitimate integration or that free DNA fragments are actively channeled towards DSBs for their repair, preventing them from undergoing homology search. Since functions involved in IR bind to the ends of DNA fragments such as the Ku proteins (6,7), such binding might already predetermine the fate of the ends by channeling them into the illegitimate end-joining pathway. Thus, it is possible that only a minority of DNA molecules are available for homologous integration. In the absence of chromosomal DSBs as substrate, more plasmid molecules will ultimately undergo homology search. Studies with Xenopus nicely demonstrate a developmental change in the efficiency of end joining to homologous recombination. Stage VI oocytes are proficient in homologous recombination and devoid of DNA end joining (39–41) while the mature eggs show predominantly illegitimate end joining (42,43). This demonstrates that efficiencies of homologous recombination to IR can be changed by developmental factors. We have shown that restriction enzymes can increase the frequency of illegitimate integration in mammalian cells, which is dependent on Ku80. Furthermore, restriction enzymes that increased the frequency of illegitimate integration, at the same time, reduced the frequency of gene targeting, suggesting that the two pathways are directly competing for the recombination substrate. Characterizing the genetically, endogenously and possibly exogenously introduced factors modifying both recombination pathways in this competition may lead to better understanding of this process and possibly to the advancements of gene targeting. ACKNOWLEDGEMENTS We thank the members of the Schiestl laboratory for helpful suggestions and discussion. We also thank Penny Jeggo (MRC Cell Mutation Unit, University of Sussex, UK) for the generous gift of the xrs5 CHO mutant cells and Nobuyo Meada (University of North Carolina, Chapel Hill, NC) for the mouse Hprt fragment. Supported by grant No. CN-83B from the American Cancer Society and Research Career Development Award ES00299 from the National Institute of Environmental Health Sciences to R.H.S.
REFERENCES 1. Schiestl,R.H. and Petes,T.D. (1991) Integration of DNA fragments by illegitimate recombination in Saccharomyces cerevisiae. Proc. Natl Acad. Sci. USA, 88, 7585–7589. 2. Manivasakam,P. and Schiestl,R.H. (1998) Nonhomologous end-joining during restriction enzyme-mediated DNA integration in Saccharomyces cerevisiae. Mol. Cell. Biol., 18, 1736–1745. 3. Zhang,H., Hasty,P. and Bradley,A. (1994) Targeting frequency for deletion vectors in embryonic stem cells. Mol. Cell. Biol., 14, 2404–2410. 4. Thomas,K.R., Folger,K.R. and Capecchi,M.R. (1986) High frequency targeting of genes to specific sites in the mammalian genome. Cell, 44, 419–428. 5. Thompson,L.H. and Jeggo,P.A. (1995) Nomenclature of human genes involved in ionizing radiation sensitivity. Mutat. Res., 337, 131–134. 6. Taccioli,G.E., Gottlieb,T.M., Blunt,T., Priestley,A., Demengeot,J., Mizuta,R., Lehmann,A.R., Alt,F.W., Jackson,S.P. and Jeggo,P.A. (1994) Ku80: product of the XRCC5 gene and its role in DNA repair and V(D)J recombination. Science, 265, 1442–1445. 7. Smider,V., Rathmell,W.K., Lieber,M.R. and Chu,G. (1994) Restoration of X-ray resistance and V(D)J recombination in mutant cells by Ku cDNA. Science, 266, 288–291. 8. Kirchgessner,C.U., Patil,C.K., Evans,J.W., Cuomo,C.A., Fried,L.M., Carter,T., Oettinger,M.A. and Brown,J.M. (1995) DNA-dependent kinase (p350) as a candidate gene for the murine SCID defect. Science, 267, 1178–1183. 9. Rathmell,W.K. and Chu,G. (1994) A DNA end-binding factor involved in double-strand break repair and V(D)J recombination. Mol. Cell. Biol., 14, 4741–4748. 10. Mimori,T. and Hardin,J.A. (1986) Mechanism of interaction between Ku protein and DNA. J. Biol. Chem., 261, 10375–10379. 11. Obe,G., Johannes,C. and Schulte-Frohlinde,D. (1992) DNA double-strand breaks induced by sparsely ionizing radiation and endonucleases as critical lesions for cell death, chromosomal aberrations, mutations and oncogenic transformation. Mutagenesis, 7, 3–12. 12. Yorifuji,T. and Mikawa,H. (1990) Co-transfer of restriction endonucleases and plasmid DNA into mammalian cells by electroporation: effects on stable transformation. Mutat. Res., 243, 121– 126. 13. Morgan,W.F., Fero,M.L., Land,M.C. and Winegar,R.A. (1988) Inducible expression and cytogenetic effects of the EcoRI restriction endonuclease in Chinese hamster ovary cells. Mol. Cell. Biol., 8, 4204–4211. 14. Rouet,P., Smih,F. and Jasin,M. (1994) Introduction of double-strand breaks into the genome of mouse cells by expression of a rare-cutting endonuclease. Mol. Cell. Biol., 14, 8096–8106. 15. Brenneman,M., Gimble,F.S. and Wilson,J.H. (1996) Stimulation of intrachromosomal homologous recombination in human cells by electroporation with site-specific endonucleases. Proc. Natl Acad. Sci. USA, 93, 3608–3612. 16. Sargent,R.G., Brenneman,M.A. and Wilson,J.H. (1997) Repair of site-specific double-strand breaks in a mammalian chromosome by homologous and illegitimate recombination. Mol. Cell. Biol., 17, 267–277. 17. Zhang,S.Z. and Dong,W.F. (1987) Chromosomal aberrations induced by the restriction endonucleases Eco RI, Pst I, Sal I and Bam HI in CHO cells. Mutat. Res., 180, 109–114. 18. Vasudev,V. and Obe,G. (1987) Effect of heat treatment on chromosomal aberrations induced by the alkylating agent trenimon or the restriction endonuclease Alu I in Chinese hamster ovary (CHO) cells. Mutat. Res., 178, 81–90. 19. Natarajan,A.T. and Obe,G. (1984) Molecular mechanisms involved in the production of chromosomal aberrations. III. Restriction endonucleases. Chromosoma, 90, 120–127. 20. Schiestl,R.H., Dominska,M. and Petes,T.D. (1993) Transformation of Saccharomyces cerevisiae with nonhomologous DNA: illegitimate integration of transforming DNA into yeast chromosomes and in vivo ligation of transforming DNA to mitochondrial DNA sequences. Mol. Cell. Biol., 13, 2697–2705. 21. Bryant,P.E. (1988) Use of restriction endonucleases to study relationships between DNA double-strand breaks, chromosomal aberrations and other end-points in mammalian cells. Int. J. Radiat. Biol., 54, 869–890. 22. Xu,S.Y. and Schildkraut,I. (1991) Cofactor requirements of BamHI mutant endonuclease E77K and its suppressor mutants. J. Bacteriol., 173, 5030–5035. 23. Henderson,G. and Simons,J.P. (1997) Processing of DNA prior to illegitimate recombination in mouse cells. Mol. Cell. Biol., 17, 3779–3785.
Nucleic Acids Research, 2001, Vol. 29, No. 23 4833
24. Derbyshire,M.K., Epstein,L.H., Young,C.S., Munz,P.L. and Fishel,R. (1994) Nonhomologous recombination in human cells. Mol. Cell. Biol., 14, 156–169. 25. Kuspa,A. and Loomis,W.F. (1992) Tagging developmental genes in Dictyostelium by restriction enzyme-mediated integration of plasmid DNA. Proc. Natl Acad. Sci. USA, 89, 8803–8807. 26. Kuspa,A. and Loomis,W.F. (1994) REMI-RFLP mapping in the Dictyostelium genome. Genetics, 138, 665–674. 27. Lu,S., Lyngholm,L., Yang,G., Bronson,C., Yoder,O.C. and Turgeon,B.G. (1994) Tagged mutations at the Tox1 locus of Cochliobolus heterostrophus by restriction enzyme-mediated integration. Proc. Natl Acad. Sci. USA, 91, 12649–12653. 28. Winkler,F.K., Banner,D.W., Oefner,C., Tsernoglou,D., Brown,R.S., Heathman,S.P., Bryan,R.K., Martin,P.D., Petratos,K. and Wilson,K.S. (1993) The crystal structure of EcoRV endonuclease and of its complexes with cognate and non-cognate DNA fragments. EMBO J., 12, 1781–1795. 29. Newman,M., Strzelecka,T., Dorner,L.F., Schildkraut,I. and Aggarwal,A.K. (1994) Structure of restriction endonuclease BamHI and its relationship to EcoRI. Nature, 368, 660–664. 30. Phillips,J.W. and Morgan,W.F. (1994) Illegitimate recombination induced by DNA double-strand breaks in a mammalian chromosome. Mol. Cell. Biol., 14, 5794–5803. 31. Morrison,C. and Wagner,E. (1996) Extrachromosomal recombination occurs efficiently in cells defective in various DNA repair systems. Nucleic Acids Res., 24, 2053–2058. 32. Liang,F., Romanienko,P.J., Weaver,D.T., Jeggo,P.A. and Jasin,M. (1996) Chromosomal double-strand break repair in Ku80-deficient cells. Proc. Natl Acad. Sci. USA, 93, 8929–8933.
33. Liang,F. and Jasin,M. (1996) Ku80-deficient cells exhibit excess degradation of extrachromosomal DNA. J. Biol. Chem., 271, 14405–14411. 34. Jeggo,P.A. and Smith-Ravin,J. (1989) Decreased stable transfection frequencies of six X-ray-sensitive CHO strains, all members of the xrs complementation group. Mutat. Res., 218, 75–86. 35. Valancius,V. and Smithies,O. (1991) Double-strand gap repair in a mammalian gene-targeting reaction. Mol. Cell. Biol., 11, 4389–4397. 36. Bollag,R.J., Waldman,A.S. and Liskay,R.M. (1989) Homologous recombination in mammalian cells. Annu. Rev. Genet., 23, 199–225. 37. Hasty,P., Rivera-Perez,J., Chang,C. and Bradley,A. (1991) Target frequency and integration pattern for insertion and replacement vectors in embryonic stem cells. Mol. Cell. Biol., 11, 4509–4517. 38. Lukacsovich,T., Yang,D. and Waldman,A.S. (1994) Repair of a specific double-strand break generated within a mammalian chromosome by yeast endonuclease I-SceI. Nucleic Acids Res., 22, 5649–5657. 39. Carroll,D. (1983) Genetic recombination of bacteriophage lambda DNAs in Xenopus oocytes. Proc. Natl Acad. Sci. USA, 80, 6902–6906. 40. Carroll,D., Wright,S.H., Wolff,R.K., Grzesiuk,E. and Maryon,E.B. (1986) Efficient homologous recombination of linear DNA substrates after injection into Xenopus laevis oocytes. Mol. Cell. Biol., 6, 2053–2061. 41. Goedecke,W., Vielmetter,W. and Pfeiffer,P. (1992) Activation of a system for the joining of nonhomologous DNA ends during Xenopus egg maturation. Mol. Cell. Biol., 12, 811–816. 42. Pfeiffer,P. and Vielmetter,W. (1988) Joining of nonhomologous DNA double strand breaks in vitro. Nucleic Acids Res., 16, 907–924. 43. Rusconi,S. and Schaffner,W. (1981) Transformation of frog embryos with a rabbit β-globin gene. Proc. Natl Acad. Sci. USA, 78, 5051–5055.
