Article | Open

Inactivation of Pol θ and C-NHEJ eliminates off-target integration of exogenous DNA

  • Nature Communications 8, Article number: 66 (2017)
  • doi:10.1038/s41467-017-00124-3
  • Download Citation
Received:
Accepted:
Published online:

Abstract

Off-target or random integration of exogenous DNA hampers precise genomic engineering and presents a safety risk in clinical gene therapy strategies. Genetic definition of random integration has been lacking for decades. Here, we show that the A-family DNA polymerase θ (Pol θ) promotes random integration, while canonical non-homologous DNA end joining plays a secondary role; cells double deficient for polymerase θ and canonical non-homologous DNA end joining are devoid of any integration events, demonstrating that these two mechanisms define random integration. In contrast, homologous recombination is not reduced in these cells and gene targeting is improved to 100% efficiency. Such complete reversal of integration outcome, from predominately random integration to exclusively gene targeting, provides a rational way forward to improve the efficacy and safety of DNA delivery and gene correction approaches.

Introduction

Gene targeting, the precise engineering of genomes through the homologous integration of exogenous DNA, is hampered by the orders of magnitude higher efficiency of untargeted random integration (RI)1. RI does not rely on sequence homology, happens with no or little detectable sequence preference, results in insertional mutations, and is hence often referred to as “illegitimate recombination”. The mechanism of RI in higher eukaryotes has been enigmatic for decades, because it has evaded rigorous genetic definition. Integration of exogenous DNA at unpredictable positions in the genome can heavily impact on the functioning of the integrated DNA, as well as its genomic environment, and thus presents a safety risk2,3,4. With the advent of CRISPR-Cas9 technology and its seemingly inevitable implementation in clinically relevant gene-correction strategies5, it becomes all the more important to devise strategies to counteract the potentially detrimental outcomes of RI. Efforts to shift the RI/HR balance of exogenous DNA integration towards HR have been only marginally successful6, 7. One reason is the lack of mechanistic understanding of RI relative to HR. From the analysis of integration sites the involvement of DNA topoisomerases has been suggested but not proven8, 9. Because a double-stranded break (DSB) in the chromosome is ultimately required to ligate the incoming DNA, and because DSB induction stimulates RI10,11,12,13, it has been hypothesized that homology-independent DSB end joining (EJ) proteins are involved. In mammalian cells EJ activity is primarily mediated by the C-NHEJ pathway involving Ku70-Ku80, DNA-PKcs, and the LigIV–XRCC4 complex14. Strikingly in mouse embryonic stem (ES) cells, where Ku proteins are not essential, complete inactivation of C-NHEJ does not result in any increase in gene targeting15, 16. A possible explanation is the existence of alternative pathways for EJ. Indeed, when C-NHEJ is suppressed some residual EJ activity remains. This so-called alternative EJ (Alt-EJ) is genetically not well defined, and may be an umbrella term for several distinct auxiliary or context-specific EJ activities. Because the DNA junctions resulting from EJ activities in C-NHEJ impaired cells are characterized by minute stretches of sequence homology, also the term micro-homology-mediated end joining has been used. Proteins implicated in potentially different alternative EJ activities include PARP116, 17, Pol θ18, CtIP19, 20, the MRN21 complex, LigI and LigIII22, 23.

Here we show that the A-family DNA polymerase θ (Pol θ) promotes most of RI events. In the absence of Pol θ, the major DNA double-strand break (DSB) repair pathway, canonical non-homologous DNA end joining (C-NHEJ), mediates residual RI. We detected not a single RI event in cells lacking both Pol θ and C-NHEJ, indicating that if additional pathways of RI exist, their contribution is at least four orders of magnitude lower. Strikingly, we find that the characteristic unfavorable balance between homologously targeted DNA integration and RI in mammalian cells is completely reversed in cells defective for Pol θ and C-NHEJ.

Results

Pol θ and C-NHEJ inactivation prevents random integration

We used CRISPR-Cas9 to inactivate Polq, the gene encoding Pol θ, and key C-NHEJ genes Ku70, Ku80 and LigIV, and a combination of Polq Ku70 and Polq Ku80 in mouse ES cells (Fig. 1a, Supplementary Figs. 1 and 2); two independent knock-out cell lines were generated for each genotype. The observation that single-mutant cells were hypersensitive toward IR validated functional impairment of the encoded proteins. Strikingly, double mutant cells displayed a synergistic increase in IR sensitivity arguing for redundant activities of C-NHEJ and Pol θ on radiation-induced DNA damage (Supplementary Fig. 2). Next, we measured the ability of these knockout cells to form stable puromycin-resistant colonies upon RI of transfected plasmid DNA encoding a puromycin resistance gene (Fig. 1a, b). Confirming previous work15, 16, the frequency of transfected DNA integration was unaffected by C-NHEJ inactivation, however, we found it to be severely reduced in Polq–/– cells (11% of wild-type). Deficiency in nuclear form of the Alt-EJ protein LigIII22,23,24 did not result in RI frequency decrease (Supplementary Fig. 3), which may be due to redundancy of LigIII and LigI as was previously observed in manifestations of Alt-EJ22, 23. Our observation that loss of C-NHEJ does not at all affect RI slightly deviates from an earlier study reporting a modest decrease in RI upon LIG4 depletion25, which could be explained by different cell- or growth characteristics for mouse ES cells versus transformed human somatic cells that could affect usage of EJ pathways: mouse ES cells are primarily in S-phase and are thus perhaps more geared up to repair breaks via HR or alt-EJ. We conclude that RI in mouse ES cells is predominantly resulting from Pol θ-mediated repair and refer to this process as TMEJ (for polymerase Theta-Mediated EJ) to distinguish it from C-NHEJ and to acknowledge the notion that Alt-EJ may also encompass Pol θ-independent repair.

Fig. 1
Fig. 1

RI is reduced in Polq −/− and completely abrogated in Polq −/− Ku70 −/−, Polq −/− Ku80 −/− and Polq −/− LigIV −/− cells. a Immuno-blots confirming the absence of Ku70, Ku80, and LigIV from wild type and Polq −/− ES cells. b Mouse ES cells of indicated genotypes were electroporated with linearized plasmid DNA encoding a puromycin resistance gene and a GFP expression plasmid to measure transfection efficiency. Colony counts were adjusted for transfection and plating efficiency. Means and s.e.m. are plotted. c Polq −/− and wild-type control cells complemented with wild-type and mutant versions of the human POLQ cDNA were transfected with a plasmid encoding a puromycin-resistance gene. After 10 days of puromycin selection CBB-stained plates were photographed. Right-hand panel shows immuno-blots of Pol θ levels in the clones used in the left-hand panel. Arrowhead indicates full-length Flag-Pol θ protein, asterisk indicates non-specific signal

This conclusion is supported by the observation that integration sites in Pol θ proficient cells, but not in Polq–/– cells, frequently resulted in junctions that contained small insertions, which are suggestive of template-based polymerase action primed by the local presence of homologous sequences (Fig. 2 and Supplementary Data 1, 2). Noteworthy, the observed inserts had a striking similarity with those identified at the junctions of translocations that give rise to Ewing Sarcoma26. The occasional presence of templated insertions is a signature feature of Pol θ-mediated repair of DSBs27, 28.

Fig. 2
Fig. 2

Different junctional diversity at RI sites in wild-type and Polq −/− cells. a Schematic representation of junctions found at RI sites: (i) simple deletions that are without insertions (in grey), (ii) complex junctions that contain insertions, which display sequence identity to the transfected DNA but in a reverse complement (RC) orientation (in red), and (iii) complex junctions with untraceable miscellaneous insertions (in pink); *** denotes p < 0.001 in Fischer’s exact test. b Two representative cases of insertion-containing transfected DNA-genome junctions. Junctional sequences that are identical to the transfected (foreign) DNA (in bold) or the mouse genome are marked. Micro-homologous sequences are boxed. Identical sequences in a RC orientation are underlined

Simultaneous inactivation of Pol θ and Ku70, Ku80 or LigIV resulted in a complete inability of cells to integrate transfected DNA (Fig. 1b). We did not recover a single puromycin-resistant colony in multiple experiments involving more than 1.5×108Polq–/–Ku70–/–, 7.5×107Polq–/–Ku80–/– and 2.4×108Polq–/–Lig4–/– cells. No significant differences in transient transfection efficiency, as measured by expression of fluorescent marker 1 day after transfection, was observed between the genotypes (Supplementary Fig. 2b). We conclude that the residual level of RI in Polq–/– cells results from C-NHEJ.

We confirmed that total suppression of RI in Polq–/–Ku70–/– cells was due to loss of Pol θ by complementing Polq–/–Ku70–/– cells with a gene encoding Flag-tagged human Pol θ (Fig. 1c). The wild-type form of human Pol θ restored RI capacity, while mutant versions of the protein with abrogated ATPase or polymerase activity did not.

Pol θ inactivation facilitates gene targeting

To test whether the absence of Pol θ affected the balance between RI and HR-mediated targeted integration we used our previously established Rad54-GFP gene targeting assay29, which measures the ratio between random (GFP negative cells) and targeted integration (GFP positive cells) of the construct targeting exon 4 of the Rad54 locus using FACS (Fig. 3a, b). In Polq–/– cells we observed a profound shift in the shape of the FACS profile from RI peak towards HR peak. Strikingly, in Polq–/–Ku70–/– cells only HR-mediated gene targeting events were detected. The changes in the FACS profiles were associated with a reduction in the total number of integrants (puromycin-resistant cells regardless of GFP status). These results are consistent with a reduction in the efficiency of RI in Polq–/– and complete inactivation of RI in Polq–/–Ku70–/– cells. This outcome also proves that the observed reduction in RI is not resulting from diminished delivery or stability of transfected DNA30. Since induction of DSBs stimulates RI, we next tested whether γ-irradiation affected the HR/RI ratio. Indeed, we observed significant stimulation of RI in wild-type and single mutant cells (Fig. 3b). However, even under these stimulatory conditions, no RI was detected in Polq–/–Ku70–/– cells, and only HR manifested. As most gene targeting strategies currently involve stimulation by DSB induction at the target locus, we tested whether loss of Pol θ and C-NHEJ is also beneficial in the context of CRISPR-Cas9-stimulated gene targeting, and indeed it is (Supplementary Fig. 4)

Fig. 3
Fig. 3

Highly efficient gene targeting in the absence of Pol θ. a Map of the Rad54 locus showing regions of homology with the gene targeting constructs used in the study. The oval on the construct map indicates the location of the cassette containing the downstream part of the Rad54 CDS fused to GFP, and a selection cassette. b FACS frequency histograms from the Rad54-GFP gene targeting assay in unirradiated and irradiated (200 mGy) cells. Marker used to determine the percentage of Rad54-GFP-positive cells is shown. As the shoulders of RI and GT peaks overlap, the number of events within the marker under-reports or over-reports the true fraction of targeted cells, when the distribution is strongly skewed toward GT or RI, respectively. c Gene targeting efficiency at the Pim1 locus (means and s.e.m. are plotted). d Gene targeting efficiency of exon 9 and exon 18 Rad54-GFP knock-in constructs in wild type (filled histograms) and Polq −/− (red outline) cells

.

HR-mediated gene targeting at exon 4 of Rad54 is very efficient: 30–70% compared to ~5% for an average location in mouse ES cells. Therefore, we tested whether the high gene targeting efficiency we observed in Polq–/– cells will be maintained in a more typical scenario. Constructs that target exon 9 or 18 of Rad54 (Fig. 3a) recombine with much lower frequency (<5% in wild type cells, which results in a GFP-positive cell frequency below detection threshold). For these constructs, we also found a profound shift from RI towards HR in Polq–/– cells (Fig. 3d). This phenomenon is not specific for the Rad54 locus, as we detected a three-fold increase in relative gene targeting efficiency (ratio between random and targeted integration frequencies, resulting in puroR+hygroR– and puroR+hygroR+phenotypes, respectively) at another genomic locus, i.e., the Pim1 gene (Fig. 3c). We conclude that Pol θ inhibition is a useful method to dramatically and generally increase gene targeting efficiency.

Pol θ inactivation does not affect homologous recombination

Recent studies suggest that Pol θ inhibits HR by interacting with Rad51 recombinase18, 31. The strong reduction in RI we observed in Polq–/– cells can account for the increase in the HR/RI ratio measured in the Rad54-GFP gene targeting assay, however an increase in HR efficiency may also contribute. We used a direct-repeat GFP (DR-GFP) assay32 to measure DSB-induced gene conversion and found no effect from genetically inactivating Pol θ alone (Fig. 4a). In addition, we found no significant difference in the absolute frequency of targeted integration of the Pim1 targeting construct in Pol θ proficient and deficient mouse ES cells (p = 0.37, n = 4). However, in the DR-GFP assay, the simultaneous inactivation of Pol θ and Ku80 resulted in a higher yield of gene conversion events (Fig. 4a). Thus, a potential increase in HR through the lack of Pol θ is unlikely to contribute to the increased gene targeting efficiency in Polq–/– mouse ES cells, but may be a factor when both C-NHEJ and TMEJ are inactive, as these end joining factors may locally act redundantly to suppress HR.

Fig. 4
Fig. 4

HR frequency and pharmacological manipulation of HR/RI ratio in Polq −/− cells. a DR-GFP reporter constructs were integrated in wild-type and mutant cell lines at the Pim1 locus. Reporter-containing cells were co-lipofected with I-SceI expression construct and RFP expression plasmid. Gene conversion frequency (GFP+) and transfection efficiency (RFP+) was measured by FACS 48 h after transfection. Means and s.e.m. are plotted. b Wild-type (left) and Polq −/− cells were electroporated with exon 9 or exon 18 Rad54-GFP gene targeting constructs and 10 µM NU-7441 (red outline) was added to the media immediately after electroporation and removed 24 h later. The HR/RI ratio was determined by FACS

Pharmacological manipulation to improve gene targeting

Our results described above imply that transient inactivation of RI by small molecule inhibition of TMEJ and C-NHEJ could obviate the need for DSB induction (and thereby avoid its associated side effects) for precise gene targeting in certain genomic engineering scenarios. We tested whether suppressing residual RI activity in Polq–/– cells by NU-7441, a compound implicated in C-NHEJ inhibition by targeting DNA-PKcs activity, increases gene targeting efficiency at Rad54 exons 9 and 18 (Fig. 4b). Indeed, treatment for 24 h after transfection with 10 µM of NU-7441 resulted in a significant increase in the ratio of HR to RI in Polq–/– but not wild-type cells. Thus, our results revealed the essential targets, Pol θ and C-NHEJ, and demonstrate that pharmacological manipulation can provide an opportunity to enhance the efficiency of precise genomic engineering.

Discussion

We established the up until now elusive genetic dependencies of RI in mammalian cells. We demonstrate that exactly two genome maintenance pathways, TMEJ and C-NHEJ, carry out all detectable RI, with TMEJ being the default dominant pathway. Similar observations have now been made in human cells33. It has been known for four decades that transforming cells with exogenous DNA leads to integration at unpredictable positions in the genome34, 35, which hampers clinically relevant gene-correction strategies. The genetic definition of the major promoters of RI of exogenous DNA presented here allows the rational design of methods to improve the efficacy and safety of gene correction approaches for industrial or medical purposes.

Methods

Cell lines

Mouse ES cells were routinely cultured on gelatinized (0.1% gelatin in water) dishes in media comprising 1:1 mixture of DMEM (Lonza BioWhittaker Cat. BE12-604F/U1, with Ultraglutamine 1, 4.5 g/l Glucose) and BRL-conditioned DMEM, supplemented with 1000 U/ml leukemia inhibitory factor, 10% FCS, 1× NEAA, 200 U/ml penicillin, 200 µg/ml streptomycin, 89 µM β-mercaptoethanol. The parental wild-type line (IB10) was a specific pathogen free clonal isolate of the 129/Ola E14 line36.

For the generation of Polq, Ku80 (Xrcc5), and LigIV knockout cell-lines, ES cells were cultured on gelatin-coated plates containing irradiated primary mouse embryonic fibroblasts in knockout Dulbecco’s Modified Eagle Medium (Gibco) supplemented with 100 U/ml penicillin, 100 µg/ml streptomycin, 2 mM GlutaMAX, 1 mM sodium pyruvate, 1× non-essential amino acids, 100 µM β-mercaptoethanol (all from Gibco), 10% FCS and 1000 U/ml leukemia inhibitory factor. IB10 cells were transfected with plasmids expressing Cas9-WT-2A-GFP (pX458 and sgRNAs targeting different sequences in the three genes using Lipofectamine 2000 (Invitrogen)). Cells were seeded at low density and maintained with regular medium changes for 8–10 days until colonies were formed. Colonies were picked and grown in 96-wells format, at (semi) confluence cells were split to two 96-wells plates. One plate was used for sub-culturing of cells; the other plate was used for DNA isolation and PCR analysis of the individual clones. Restriction fragment length polymorphism (RFLP, based on the loss of a unique restriction site) of PCR-products was used to identify clones with a bi-allelic mutation. PCR-products of clones that showed bi-allelic mutation were sent for sequencing to confirm the introduction of a premature stop-codon due to deletions/insertions that cause a frameshift (Supplementary Fig. 1b). Polq Ku80 double knockout cell lines were generated by targeting the Polq gene in the resulting Ku80 deficient cell lines. Experiments described in this study have been done using at least two independent clonally derived knockout lines per gene. An overview of the targeted sequences and the oligonucleotides and restriction sites used for the RFLP analysis can be found in Supplementary Data 3.

Ku70 (Xrcc6) knockout ES cells were generated by CRISPR-Cas9 stimulated gene targeting with a homologous template. The sgRNA recognizing the sequence in exon 4 of Ku70 was cloned into pX459 vector37. The information on oligonucleotides and constructs is provided in Supplementary Data 3 and 4. The donor template was produced by PCR-amplification of two homology arms from IB10 genomic DNA (3.3 and 2.3-kb long), the PGK-hygro selection cassette and the high-copy plasmid backbone pBluescript, and ligated using the Gibson isothermal assembly method38. The construct was partially sequenced to verify the junctions between the selection cassette and homology arms. Oligonucleotides used in cloning and genotyping are listed in Supplementary Data 3 and 4. Homologous template and sgRNA+Cas9 expression constructs were co-electroporated into 1–2×107 IB10 or Polq–/– cells. Selection with 200 µg/ml Hygromycin B (Roche) was started 1 day after electroporation and maintained with regular media changes for 8–10 days until colonies formed. Colonies were picked using a sterile pipette tip under the microscope, dispersed with Trypsin-EDTA, divided for re-plating and for lysis (50 mM KCl, 10 mM Tris–HCl pH 9, 0.1% Triton-X100, 0.15 mg/ml proteinase K, 60 °C 1 h, 95 °C 15 min) followed by genotyping PCR with K70-e34-scrF, K70-e34-scrR and PGK-R1 primers. The absence of the band expected form the wild-type allele was interpreted as an indication of the bi-allelic modification of the locus (~50% clones). A subset of them was expanded, re-genotyped and tested by immuno-blotting.

Immuno-blotting

Whole cell lysates were fractionated by Tris-glycine SDS–PAGE, transferred to nitrocellulose membrane and blotted (5% milk, 0.05% Tween 20 PBS, 4 °C overnight) with the following antibodies: Ku70 (C-19 goat polyclonal, 1:2500, Santa Cruz sc-1486), Ku80 (M-20, 1:2000 Santa Cruz sc-1485), LigIV (1:2000 Abcam ab80514), Flag tag (M2 mAb, 1:5000, Sigma), PARP-1 (C2-10 mAb, 1:5000, ENZO). Uncropped scans of the immuno-blots are show in Supplementary Fig. 5.

Clonogenic survivals

Cells were plated in triplicates in 6-well plates at 100–4000 per well and irradiated using 137Cs (Ku70) or YXlon X-ray generator (Ku80) immediately after seeding. Colonies were allowed to form for 7 days, after which cells were fixed and stained.

DNA constructs

pLPL was derived from the construct loxP-PGK-gb2-neo-polyA-loxP cassette in pGEM-T Easy (a gift from Francis Stewart39) by replacing neomycin phosphotransferase with puromycin N-acetyltransferase CDS. Gene targeting constructs for Rad54 exons 9 and 18 were produced by recombineering with mobile agents39 with a cassette containing PCR-amplified fragment of the previously described exon 4 gene targeting vector,40 encoding part of the RAD54 CDS downstream of the targeted exons, GFP CDS and PGK polyadenylation signal, and the loxP-flanged PGK-EM7-neo dual selection cassette into a BAC bMQ-422M23 Sanger 129/Sv BAC library41. Fragments of the modified BAC were subcloned into high-copy pBluescript vector using recombineering (retrieval). Plasmids SpCas9(BB)-2A-GFP (PX458) and SpCas9(BB)-2A-puro (PX459 v2.0) were a gift from Feng Zhang (Addgene plasmid #48138, #62988). A plasmid based on the pENTR233 vector and encoding human Pol θ was purchased from the ORFeome collaboration collection (Dharmacon), and re-cloned into pAZ125 PiggyBac vector, which was derived from 5′-PTK-3′42 by replacing the transposon cargo (XhoI-SpeI fragment) with two expression cassettes: loxP-flanked PGK→neo and CAG→[ClaI XhoI]~bGHpa (bovine growth hormone polyadenylation signal). BglII-KasI fragment of Pol θ CDS and two PCR-amplified fragments covering parts of Pol θ CDS not included in the BglII-KasI fragment were inserted into pAZ125 using Gibson assembly38. Point mutations to produce D216A, E217A (ATPase-deficient), T2239S, G2240R, R2241K (polymerase-deficient), and the double ATPase+polymerase deficient forms of Pol θ were introduced by PCR amplification with Q5 polymerase (NEB) of the POLQ CDS fragments, and insertion into pAZ125 using Gibson assembly. Two sgRNAs targeting the Rad54 exon 18 region between the homology arms of the corresponding targeting construct, and thus replaced by the GFP-PKG-neo cassette upon recombination, were cloned in pX459.

Gene targeting assays

For Rad54 gene targeting assay 107 were electroporated with 10 µg PvuI-linearized gene targeting construct and plated in 10 gelatinized 10 cm dishes. For experiments involving irradiation, cells were seeded in duplicate, one of which irradiated with 200 mGy using 137Cs source. The following day media was exchanged to selective media (1.5 µg/ml puromycin for exon 4 constructs, 500 µg/ml G418 for exon 9 and 18 constructs). Where indicated, sgRNA+Cas9 expression construct was co-transfected with the gene targeting construct, which in these experiments was not linearized. After 8–10 days of selection, cells were trypsinized, collected by centrifugation, fixed by re-suspending in 1 ml of 1% paraformaldehyde in PBS, incubated for 15 min and mixed with an equal volume of 0.2% Triton X100 solution in PBS to enhance the separation between GFP+ and GFP- peaks. FACS analysis was performed using BD LSR Fortessa instrument. Pim1 gene targeting assay32 was performed using the DR-GFP reporter construct which carries a hygromycin resistance gene that can be expressed from the Pim1 promoter upon targeted integration and a puromycin resistance gene that is under PGK promoter and is expressed after both targeted and random integration. After electroporation cells were divided into two dishes: 10% for puromycin selection (random and targeted integration) and 90% for hygromycin selection (targeted integration). An aliquot was taken to estimate plating efficiency without selection. After 8–10 days of selection colonies were stained and counted. The ratio of hygromycin-resistant to puromycin-resistant colonies was interpreted as relative gene targeting efficiency, while the number of hygromycin-resistant colonies per viable plated cell was defined as absolute targeting frequency.

Random integration assay

Fifteen million cells were electroporated (118 V, 1200 µF, ∞ Ω, exponential decay, GenePulser Xcell apparatus (BioRad), 2 mm cuvette (BTX)) with 15 µg linearized (DraI digestion, followed by phenol extraction and ethanol precipitation) or circular pLPL plasmid containing puromycin N-acetyltransferase under mouse PGK promotor and 15 µl of circular GFP reporter plasmid (pEGFP-N1). For mutant cell lines electroporations were done in duplicates or triplicates to compensate for low plating efficiency. Cells were re-suspended in 5.5 ml media and seeded at 1/5 and 4/5 dilution into 10 cm dishes; 2–4 µl samples of the 1/5 dish were plated in triplicates for plating efficiency estimation; the remainder of the suspension was seeded into a 24-well plate and analyzed by FACS the following day to estimate transfection efficiency. Puromycin selection (1.5 µg/ml) was started the following day and maintained for 8–10 days. Colonies were stained (0.25% Coomassie Brilliant Blue R (CBB), 40% methanol, 10% acetic acid), photographed and counted manually using FIJI CellCounter plugin or with OpenCFU43.

Mapping integration sites

Insertion junctions were amplified and analyzed as described previously44. Briefly, cells were transfected with linear pLPL plasmid PCR-products using Lipofectamine 2000 (Invitrogen). Puromycin (1.5 µg/ml) was added 24  h after transfection and colonies were allowed to grow for 8–10 days. Individual puromycin-resistant clones were isolated and grown on 96-wells plates. At confluency, DNA was isolated from individual wells and plasmid-genome junctions were amplified via TAIL-PCR45. Sequences of plasmid-specific and arbitrary degenerate (AD) oligonucleotides used for TAIL-PCR can be found in Supplementary Data 4. Sanger sequencing was performed using the innermost TAIL-PCR oligonucleotide (pLPL_For-3). Subsequent automated analysis was performed by custom scripts. High-quality DNA stretches (base error quality <0.1, stretch ≥50 nucleotides) were identified within each Sanger sequence and further analyzed. Blast against the mouse genome and the pLPL plasmid PCR-product was performed to identify junctions. For each junction, the type of junction was determined (plasmid product–plasmid product or plasmid product–mouse genome) and whether overlap existed between the two top BLAST hits. If the two top BLAST hits overlapped it indicated the use of (micro)homology. When no overlap between the two top BLAST hits was found, the junction was annotated as a filler. When fillers were large enough (≥5 nucleotides) the origin of the filler was searched for in the pLPL plasmid and in the mouse genome at the location the junction mapped. Total of 100 base pairs at both sides of the mapped location were searched in normal and reverse complement orientation using the longest common substring approach44.

Data Availability

All data generated or analyzed during this study is included in this published article and its supplementary information.

Additional Information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. 1.

    Würtele, H., Little, K. C. E. & Chartrand, P. Illegitimate DNA integration in mammalian cells. Gene Ther. 10, 1791–1799 (2003).

  2. 2.

    Kamthan, A., Chaudhuri, A., Kamthan, M. & Datta, A. Genetically modified (GM) crops: milestones and new advances in crop improvement. Theor. Appl. Genet. 129, 1639–1655 (2016).

  3. 3.

    Yáñez, R. J. & Porter, A. C. Therapeutic gene targeting. Gene Therapy 5, 149–159 (1998).

  4. 4.

    Saito, S. & Adachi, N. Advances in the development of gene-targeting vectors to increase the efficiency of genetic modification. Biol. Pharm. Bull. 39, 25–32 (2016).

  5. 5.

    Liu, J. & Shui, S.-L. Delivery methods for site-specific nucleases: Achieving the full potential of therapeutic gene editing. J. Control Release 244, 83–97 (2016).

  6. 6.

    Würtele, H., Gusew, N., Lussier, R. & Chartrand, P. Characterization of in vivo recombination activities in the mouse embryo. Mol. Genet Genomics 273, 252–263 (2005).

  7. 7.

    Vasquez, K. M., Marburger, K., Intody, Z. & Wilson, J. H. Manipulating the mammalian genome by homologous recombination. Proc. Natl Acad. Sci. USA 98, 8403–8410 (2001).

  8. 8.

    Ikeda, H., Kawasaki, I. & Gellert, M. Mechanism of illegitimate recombination: common sites for recombination and cleavage mediated by E. coli DNA gyrase. Mol. Gen. Genet. 196, 546–549 (1984).

  9. 9.

    Bullock, P., Champoux, J. J. & Botchan, M. Association of crossover points with topoisomerase I cleavage sites: a model for nonhomologous recombination. Science 230, 954–958 (1985).

  10. 10.

    Stoker, M. Effect of x-irradiation on susceptibility of cells to transformation by polyoma virus. Nature 200, 756–758 (1963).

  11. 11.

    Perez, C. F., Botchan, M. R. & Tobias, C. A. DNA-mediated gene transfer efficiency is enhanced by ionizing and ultraviolet irradiation of rodent cells in vitro. I. Kinetics of enhancement. Radiat. Res. 104, 200–213 (1985).

  12. 12.

    Schiestl, R. H. & Petes, T. D. Integration of DNA fragments by illegitimate recombination in Saccharomyces cerevisiae. Proc. Natl Acad. Sci. USA 88, 7585–7589 (1991).

  13. 13.

    Stevens, C. W., Zeng, M. & Cerniglia, G. J. Ionizing radiation greatly improves gene transfer efficiency in mammalian cells. Hum. Gene Ther. 7, 1727–1734 (1996).

  14. 14.

    Pannunzio, N. R., Li, S., Watanabe, G. & Lieber, M. R. Non-homologous end joining often uses microhomology: implications for alternative end joining. DNA Repair 17, 74–80 (2014).

  15. 15.

    Pierce, A. J., Hu, P., Han, M., Ellis, N. & Jasin, M. Ku DNA end-binding protein modulates homologous repair of double-strand breaks in mammalian cells. Genes Dev. 15, 3237–3242 (2001).

  16. 16.

    Domínguez-Bendala, J., Masutani, M. & McWhir, J. Down-regulation of PARP-1, but not of Ku80 or DNA-PKcs’, results in higher gene targeting efficiency. Gene Ther. 30, 389–393 (2006).

  17. 17.

    Mansour, W. Y., Rhein, T. & Dahm-Daphi, J. The alternative end-joining pathway for repair of DNA double-strand breaks requires PARP1 but is not dependent upon microhomologies. Nucleic Acids Res. 38, 6065–6077 (2010).

  18. 18.

    Mateos-Gomez, P. A. et al. Mammalian polymerase θ promotes alternative NHEJ and suppresses recombination. Nature 518, 254–257 (2015).

  19. 19.

    Ahrabi, S. et al. A role for human homologous recombination factors in suppressing microhomology-mediated end joining. Nucleic Acids Res.. doi:10.1093/nar/gkw326 (2016).

  20. 20.

    Taty-Taty, G.-C. et al. Control of alternative end joining by the chromatin remodeler p400 ATPase. Nucleic Acids Res. 44, 1657–1668 (2016).

  21. 21.

    Deriano, L., Stracker, T. H., Baker, A., Petrini, J. H. J. & Roth, D. B. Roles for NBS1 in alternative nonhomologous end-joining of V(D)J recombination intermediates. Mol. Cell 34, 13–25 (2009).

  22. 22.

    Masani, S., Han, L., Meek, K. & Yu, K. Redundant function of DNA ligase 1 and 3 in alternative end-joining during immunoglobulin class switch recombination. Proc Natl Acad Sci USA 113, 1261–1266 (2016).

  23. 23.

    Lu, G. et al. Ligase I and ligase III mediate the DNA double-strand break ligation in alternative end-joining. Proc. Natl Acad. Sci. USA 113, 1256–1260 (2016).

  24. 24.

    Simsek, D. et al. Crucial role for DNA ligase III in mitochondria but not in Xrcc1-dependent repair. Nature 471, 245–248 (2011).

  25. 25.

    Iiizumi, S. et al. Impact of non-homologous end-joining deficiency on random and targeted DNA integration: implications for gene targeting. Nucleic Acids Res. 36, 6333–6342 (2008).

  26. 26.

    Zucman-Rossi, J., Legoix, P., Victor, J. M., Lopez, B. & Thomas, G. Chromosome translocation based on illegitimate recombination in human tumors. Proc. Natl Acad. Sci. USA 95, 11786–11791 (1998).

  27. 27.

    Chan, S. H., Yu, A. M. & McVey, M. Dual roles for DNA polymerase theta in alternative end-joining repair of double-strand breaks in Drosophila. PLoS Genet. 6, e1001005 (2010).

  28. 28.

    Koole, W. et al. A Polymerase Theta-dependent repair pathway suppresses extensive genomic instability at endogenous G4 DNA sites. Nat. Commun. 5, 3216 (2014).

  29. 29.

    Zelensky, A. N. et al. Caffeine suppresses homologous recombination through interference with RAD51-mediated joint molecule formation. Nucleic Acids Res. 41, 6475–6489 (2013).

  30. 30.

    Liang, F. & Jasin, M. Ku80-deficient cells exhibit excess degradation of extrachromosomal DNA. J. Biol. Chem. 271, 14405–14411 (1996).

  31. 31.

    Ceccaldi, R. et al. Homologous-recombination-deficient tumours are dependent on Polθ-mediated repair. Nature 518, 258–262 (2015).

  32. 32.

    Moynahan, M. E., Pierce, A. J. & Jasin, M. BRCA2 is required for homology-directed repair of chromosomal breaks. Mol. Cell 7, 263–272 (2001).

  33. 33.

    Saito, S., Maeda, R. & Adachi, N. Dual loss of human POLQ and LIG4 abolishes random integration. Nat. Commun. 8, 16112 (2017).

  34. 34.

    Maitland, N. J. & McDougall, J. K. Biochemical transformation of mouse cells by fragments of herpes simplex virus DNA. Cell 11, 233–241 (1977).

  35. 35.

    Bacchetti, S. & Graham, F. L. Transfer of the gene for thymidine kinase to thymidine kinase-deficient human cells by purified herpes simplex viral DNA. Proc. Natl Acad. Sci. USA 74, 1590–1594 (1977).

  36. 36.

    Hooper, M., Hardy, K., Handyside, A., Hunter, S. & Monk, M. HPRT-deficient (Lesch-Nyhan) mouse embryos derived from germline colonization by cultured cells. Nature 326, 292–295 (1987).

  37. 37.

    Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. 8, 2281–2308 (2013).

  38. 38.

    Gibson, D. G. et al. Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat. Methods 6, 343–345 (2009).

  39. 39.

    Testa, G. et al. BAC engineering for the generation of ES cell-targeting constructs and mouse transgenes. Methods Mol. Biol. 256, 123–139 (2004).

  40. 40.

    Abraham, J. et al. Eme1 is involved in DNA damage processing and maintenance of genomic stability in mammalian cells. EMBO J. 22, 6137–6147 (2003).

  41. 41.

    Adams, D. J. et al. A genome-wide, end-sequenced 129Sv BAC library resource for targeting vector construction. Genomics 86, 753–758 (2005).

  42. 42.

    Cadinanos, J. & Bradley, A. Generation of an inducible and optimized piggyBac transposon system. Nucleic Acids Res. 35, e87 (2007).

  43. 43.

    Geissmann, Q. OpenCFU, a new free and open-source software to count cell colonies and other circular objects. PLoS ONE 8, e54072 (2013).

  44. 44.

    van Kregten, M. et al. T-DNA integration in plants results from polymerase-θ-mediated DNA repair. Nat. Plants 2, 16164 (2016).

  45. 45.

    Pillai, M. M., Venkataraman, G. M., Kosak, S. & Torok-Storb, B. Integration site analysis in transgenic mice by thermal asymmetric interlaced (TAIL)-PCR: segregating multiple-integrant founder lines and determining zygosity. Transgenic Res. 17, 749–754 (2008).

Download references

Acknowledgements

We thank Robin van Schendel for bio-informatics support. Lig3 deficient cells were kindly provided by Maria Jasin. The work in R.K.’s laboratory was funded by the Cancer Genomics Netherlands Gravitation Program and an ECHO grant from NWO Chemical Sciences. J.S. is funded by a VENI grant from NWO-Life Sciences.

Author information

Author notes

  1. Alex N. Zelensky and Joost Schimmel contributed equally to this work.

Affiliations

  1. Department of Molecular Genetics, Cancer Genomics Netherlands, Erasmus University Medical Centre, Rotterdam,, 3000 CA, The Netherlands

    • Alex N. Zelensky
    •  & Roland Kanaar
  2. Department of Human Genetics, Leiden University Medical Centre, PO Box 9600, Leiden,, 2300 RC, The Netherlands

    • Joost Schimmel
    • , Hanneke Kool
    •  & Marcel Tijsterman

Authors

  1. Search for Alex N. Zelensky in:

  2. Search for Joost Schimmel in:

  3. Search for Hanneke Kool in:

  4. Search for Roland Kanaar in:

  5. Search for Marcel Tijsterman in:

Contributions

A.N.Z. and J.S. designed and performed experiments, interpreted results and wrote the manuscript; H.K. performed experiments; R.K. and M.T. interpreted results and wrote the manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to Roland Kanaar or Marcel Tijsterman.

Electronic supplementary material

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Creative Commons BY

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.