Mutagenesis Advance Access originally published online on March 7, 2007
Mutagenesis 2007 22(3):217-233; doi:10.1093/mutage/gem007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The mutagenic potential of non-homologous end joining in the absence of the NHEJ core factors Ku70/80, DNA-PKcs and XRCC4-LigIV
1Department of Biology and Geography, Institute of Genetics, University of Duisburg-Essen, Universitätsstrasse 5, D-45117 Essen, Germany 2Present address: Department of Pediatric Hematology/Oncology, University Children's Hospital of Essen, Hufelandstrasse 55, D-45122 Essen, Germany 3Present address: Institute of Biochemistry and Molecular Biology 2, University of Düsseldorf Medical School, Universitätsstrasse 1, D-40225 Düsseldorf, Germany 4Present address: Dr Fooke Laboratories GmbH, Mainstrasse 85, D-41469 Neuss, Germany 5Present address: Institute for Genetics, University of Cologne, Zülpicherstrasse 47, D-50674 Köln, Germany
Non-homologous end joining (NHEJ), the major pathway of double-strand break (DSB) repair in mammalian cells, comprises two subpathways: one that requires the three core factors Ku70/80, DNA-PKcs and XRCC4/LigIV (DNA-PK-dependent NHEJ) and the other that is independent of these factors. Using a cell-free NHEJ assay, we have investigated the ability of three Chinese hamster ovary (CHO) mutants deficient in Ku80 (xrs6), DNA-PKcs (XR-C1) and XRCC4 (XR-1) in comparison with CHO-K1 wild-type cells to rejoin non-compatible DSB ends. Both NHEJ efficiency and fidelity are strongly reduced in the mutants with xrs6 and XR-1 exhibiting the strongest reduction and XR-C1 displaying a phenotype intermediate between the wild-type and the other two mutants indicating a non-essential but facilitating role of DNA-PKcs in NHEJ. The decrease in fidelity in the mutants is expressed by an increase of deletion junctions formed at microhomologies (µhom) near the DSB (microhomology-mediated non-homologous end joining: µhomNHEJ). Using a novel µhomNHEJ assay, we show that µhom regions of 6–10 bp that are located directly at the DSB termini strongly enhance the mutagenic µhomNHEJ reaction even in the wild type. Due to its error proneness, DNA-PK-independent µhomNHEJ may actively promote genome instability. It will, therefore, be of increasing importance to examine NHEJ fidelity in the context with tumorigenesis and cellular senescence for which we here provide two efficient and reliable tools.
 |
Introduction |
|---|
 Top Introduction Materials and methodsResultsDiscussionSupplementary dataReferences |
|---|
Double-strand breaks (DSB) in genomic DNA are generated endogenously during normal cellular processes (e.g. replication) and are induced by a variety of genotoxic agents (e.g. ionizing radiation). Due to the damage of both DNA strands, DSB are particularly harmful and lead, if not repaired, to chromosome fragmentation and cell death or, if misrepaired, to mutations and genomic rearrangements which promote tumorigenesis by genome destabilization (1
–3).
In mammalian cells, DSB are removed by two main repair pathways: homologous recombination (HR) and non-homologous end joining (NHEJ). Both mechanisms complement each other, are strictly regulated and highly conserved in evolution from bacteria and yeast to mammals (3,4). As indicated by the name, HR obligatorily requires sequence homology of several hundred base pairs and is able to restore the original sequence at the break by copying the missing sequence from the sister chromatid (5). HR is mediated by the gene products of the RAD52 epistasis group and is the predominant pathway during late S and G2 phase of the mitotic cell cycle. In contrast to HR, NHEJ can principally dispense with sequence homology since it joins two DSB ends directly with each other. In mammalian cells, NHEJ occurs throughout the entire cell cycle but is used preferentially for the removal of DSB arising in G1 and G0 (6). NHEJ is intrinsically error prone because the original sequence is restored only if two compatible ends are precisely ligated. If, however, two non-compatible ends are rejoined, they first have to be processed to yield a ligatable structure. The necessary enzymatic modifications (nucleolytic trimming, gap filling) of the ends may generate base substitutions, insertions and deletions rendering this kind of DSB repair rather inaccurate.
To date, six NHEJ factors have been identified: Ku70, Ku80, the catalytic subunit of the DNA-dependent protein kinase (DNA-PKcs), Artemis, XRCC4 and DNA ligase IV (LigIV). Ku70 and Ku80 form a heterodimer (Ku70/80) that binds to DNA ends (7) and recruits DNA-PKcs, a Ser/Thr kinase of the PI3 family of kinases (8). DNA-PKcs forms a complex with Artemis (DNA-PKcs/Artemis), a nuclease with intrinsic 5'-3'-exonuclease activity. DNA-PKcs phosphorylates Artemis thus activating its endonuclease activity which cleaves DNA structures containing single- to double-strand transitions (e.g. 5'- or 3'-overhangs) (9). In this way, DNA ends can be processed to allow for subsequent ligation which is catalysed by a complex formed between XRCC4 and LigIV (XRCC4/LigIV) (10,11). In addition to these three protein complexes (also called NHEJ core factors), at least two DNA polymerases, polµ and pol
, are involved in NHEJ which catalyse the filling of small gaps existing in the NHEJ intermediates (12,13). Recently, XLF (Cernunnos), a novel NHEJ factor that interacts with the XRCC4/LigIV complex has been identified (14–16). Interestingly, hypomorphic mutations in XLF, Artemis and LigIV lead to a number of rare hereditary disorders characterized by immunodeficiency (Artemis) and/or developmental abnormalities (XLF, LigIV) which underscore the general importance of the NHEJ pathway for genome integrity and development (17,18).
Apart from the reaction that depends on the described NHEJ core factors (DNA-PK-dependent NHEJ with DNA-PK standing for the whole complex of Ku70/80-DNA-PKcs/Artemis-XRCC4-LigIV), several studies have demonstrated the existence of a separate alternative NHEJ pathway that occurs in the absence of these factors (DNA-PK-independent NHEJ) (19–28). Due to its feature to generate small deletions preferentially at sites of microhomology (µhom, 
4 bp), this pathway has often been designated as µhom-mediated non-homologous end joining (µhomNHEJ) (27,29,30). It should be noted that ‘classical’ DNA-PK-dependent NHEJ also uses µhom as an integral part of its mechanism. However, these µhom units are mostly shorter (1–4 bp) than the ones used by µhomNHEJ which also works with 4 bp but has its optimal length at 
6–10 bp as we shall see in this study. In addition, DNA-PK-dependent NHEJ tends to preserve terminal DSB sequences, whereas DNA-PK-independent NHEJ generates deletions at sites of µhom even when these sites are located within the DSB-adjacent double strand. These differences between the two mechanisms, which shall be elucidated in more detail in the Results section, led to the creation of the term µhomNHEJ to designate the error-prone DNA-PK-independent NHEJ pathway.
Deletions generated by µhomNHEJ are also characteristic of single-strand annealing (SSA), an error-prone variant of HR, which requires the Rad52 strand-pairing protein (but not the HR-specific strand-exchange Rad51 protein) and occurs between sequence repeats on the same or two heterologous chromosomes (ectopic recombination) (3,5,31). As is the case for µhomNHEJ, the mutagenicity of SSA is due to the fact that two homology regions interact directly with each other leading to the loss of one repeat unit and the intervening sequence thus giving rise to interstitial deletions or translocations. However, in spite of its similarity to SSA, µhomNHEJ requires shorter stretches of homology [6–20 bp with an optimal length at about versus 30 bp for SSA (32)] and can occur in the absence of RAD52 epistasis group genes (33). Instead, µhomNHEJ apparently involves poly(ADP-Ribose) polymerase-1 (PARP-1) and the XRCC1/LigIII complex as well as Fen-1 which again underscores the distinctness of µhomNHEJ from DNA-PK-dependent NHEJ (19,25,27,28,34,35).
The aim of this study was to investigate the spectra of NHEJ junctions in the presence and absence of each of the three core factors in order to obtain information on the mutagenic potential of the different NHEJ pathways. To do this, we have introduced a number of parameters to define the fidelity of classical DNA-PK-dependent NHEJ which shall be outlined in the Results section. To obtain reliable data on NHEJ fidelity, it is necessary to apply a large spectrum of DNA substrates that make different demands on the NHEJ reaction with respect to the structure and sequence of DSB ends and the position of fortuitous µhom regions. In addition, it is important to employ NHEJ mutant and wild-type cells that are derived from the same cell system to exclude cell-type-specific effects. Here, we have investigated the ability of three Chinese hamster ovary (CHO) cell lines each of which lacks one of the three NHEJ core factors (xrs6: Ku70/80, XR-C1: DNA-PKcs, XR-1: XRCC4/LigIV) to rejoin 15 model DSB in an established cell-free NHEJ system (22,36,37). We show that the overall NHEJ efficiency in these cell lines is, as expected, strongly reduced when compared to the NHEJ-proficient CHO-K1 cell line. A detailed analysis of 1567 cloned junctions derived from the in vitro rejoining of the 15 DSB substrates shows that the mutant cell lines also display strongly reduced NHEJ fidelity which expresses itself by an increase of deletion junctions formed either by blunting (requires no homology) or by µhomNHEJ when short regions of µhom (4 bp) are present in close proximity of the DSB. Using a newly developed µhomNHEJ assay, we investigated the dependence of µhomNHEJ on homology length and showed that 6 bp are optimal for efficiently enhancing µhomNHEJ while simultaneously suppressing DNA-PK-dependent NHEJ even in CHO-K1. Interestingly, xrs6 and XR-1 exhibit the strongest reduction in NHEJ efficiency and fidelity while the DNA-PKcs-deficient XR-C1 cell line displays a phenotype being intermediate between the wild-type and the other two mutants indicating a non-essential role of DNA-PKcs for NHEJ.
To our knowledge, this is the first study which analyses the influence of the three NHEJ core factors on NHEJ fidelity in such detail by using one homogenous cell system, a single well-characterized cell-free extract system and an extensive spectrum of 15 DSB substrates that pose different demands on the NHEJ machinery. Our study provides two easy-to-handle tools to reliably investigate the impact of the fidelity of DNA-PK-dependent NHEJ and DNA-PK-independent µhomNHEJ on genome instability, tumorigenesis and cellular senescence.
|
Materials and methods |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionSupplementary dataReferences |
|---|
Cells and cell culture
The CHO cell lines CHO-K1 and xrs6 (38
) were purchased from the European Collection of Cell Culture (Wiltshire, UK); XR-C1 (39) and XR-1 (40) were generous gifts from M. Z. Zdienicka (Leiden University, the Netherlands) and S. E. Critchlow and S. P. Jackson (Cambridge University, UK), respectively. All four cell lines were grown at 37°C in a humidified 5% CO2 atmosphere in Ham's F-12 medium enriched with 10% fetal calf serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin.
Cell-free extracts
For all experiments described in this study, whole-cell extracts were used which were prepared as described previously (22,37) with the difference that a small-scale protocol requiring less cells was employed. Approximately 1.8 x 108 cells of each cell line were used in each preparation (for three 1.5 ml ultracentrifuge tubes used with adaptors in a Beckman SW60 Ti rotor) to yield 250 µl of extract with a protein concentration ranging between 4 and 8 mg/ml. For most reproducible conditions, extracts from each mutant and its corresponding NHEJ-proficient parent were prepared on the same day using the same freshly made solutions. Stored in 50 µl aliquots in liquid nitrogen, extracts remained active for at least 6–12 months. Prior to use in NHEJ reactions, the extract volume needed was dialysed on microdialysis filters (0.025 µm; Millipore, Schwalbach, Germany) for 30 min at 4°C against freshly prepared reaction buffer (50 mM MOPSO-NaOH pH 7.5; 40 mM KCl; 10 mM MgCl2; 5 mM 2-mercaptoethanol). At least three batches of extract were prepared from each cell line to verify that the observed differences are due to differences between the cell lines rather than differences between extract preparations.
NHEJ substrates
The three substrates (#1–3) for ‘ligation’ of cohesive and blunt ends were derived from the 3-kb pSP65 (Promega, Mannheim, Germany) by linearization with a single restriction enzyme (RE) [(#1) BamH1: 5'-cohesive, (#2) PstI: 3'-cohesive, (#3) SmaI: blunt ends). The 12 substrates (#4–15) for the joining of non-compatible ends were derived from a 4.2-kb-modified pSP65 containing a 1.2-kb -DNA insert between the restriction sites used for substrate preparation (pSP65/, (37,41)). Generation of 3 kb linear plasmid substrates containing two non-compatible ends was controlled by quantitative excision of the 1.2-kb insert. Each substrate was named after the pair of RE used in its preparation [antiparallel ends: (#4) BamH1/Asp716 and (#5) BamH1/SalI: 5'/5', (#6) BstXI/BstXI and (#7) KpnI/PstI: 3'/3', abutting ends: (#8) SmaI/Sal: bl/5', (#9) AvaI/Hind2: 5'/bl, (#10) SmaI/PstI: bl/3', (#11) BamI/PstI: 5'/3', (#12) SacI/SalI: 3'/5', (#13) EcoRI/Hind3: 5'/3', (#14) SacI/Hind3: 3'/5', (#15) AvaI/KpnI: 5'/3']. All substrates were gel purified using an agarose gel extraction kit (Qiagen, Hilden, Germany).
Assay for NHEJ and analysis of products
In standard reactions, 10 ng of RE-linearized plasmid substrate was incubated for 6 h at 25°C in a total volume of 10 µl containing 3–4 µg/µl of extract protein (the concentration of extract protein was always adjusted to the extract with the lowest protein concentration) in reaction buffer supplemented with 1 mM ATP pH 7.5; 200 µM dNTPs (50 µM each) and 50 ng/µl BSA. Reactions were terminated by adjustment to 20 mM Tris–HCl pH 7.5, 10 mM EDTA, 1% sodium dodecyl sulfate (SDS) and incubation at 65°C for 5 min. After digestion with 2 mg/ml proteinase K for 30 min at 37°C and 15 min at 65°C, equivalents of 2 ng substrate DNA were loaded on 1% agarose gels containing 1 µg/ml ethidium bromide (EtBr). Electrophoresis at 10 V/cm was carried out in the presence of 1 µg/ml EtBr to separate open circles (oc) from covalently closed circles (ccc) products (intercalation of EtBr introduces positive supercoil into relaxed ccc products). DNA bands were visualized by in situ gel hybridization using a pSP65-specific probe labelled with [32P]-
-dCTP by random priming (22,37). Reaction products were quantified in a phosphorimaging facility (Packard Bioscience, Dreieich, Germany) as percentage of total radioactivity per lane corrected against background using the OptiQuant 04.00 program (Packard Bioscience). Circular joined products were cloned by transformation of 4 ng equivalents of substrate DNA of each NHEJ sample in Escherichia coli strain DH5 to yield single clones which were purified by miniscale extraction. In order to compensate for the 10- to 100-fold lower transformation efficiency obtained from the three mutants which is due to the strongly reduced levels of circle formation and to obtain sufficiently large numbers of clones for junction analysis, more bacteria derived from mutant samples than from CHO-K1 samples were plated. Clones from ligation substrates (#1–3) were subjected to cleavage with the original RE to check for accurate ligation. Clones from the NHEJ substrates (#4, #6, #9–13 and #15) were subjected to cleavage with the RE indicated in Figure 5 to check for restoration of the corresponding restriction sites. All cloned junctions resistant to RE cleavage were sent to a sequencing service (Seqlab, Göttingen, Germany).
Inhibition of NHEJ by Wortmannin
Directly prior to use, a 1-mM stock of Wortmannin (Sigma, Munich, Germany) in DMSO was freshly diluted with ddH2O to 50 µM. One microlitre of 50 µM Wortmannin or 5% DMSO (positive control), respectively, was added to 7 µl of extract (3–4 µg/µl) and preincubated for 15 min on ice. The NHEJ reaction was initiated by the addition of 2 µl of DNA-ATP-dNTP mix to yield a final concentration of 5 µM Wortmannin or 0.5% DMSO, respectively.
Substrates for µhomNHEJ
The two DNA fragments used as substrates with variable homology (varHom) in the intermolecular µhomNHEJ reaction were derived from pSP65/-varHom (Figure 8A). pSP65/-varHom exists in eight different forms: pSP65/-varHom6–40 all of which were created by the insertion of eight different double-stranded oligonucleotides (Hom6–40) between the EcoRI and the BamH1 sites of pSP65/ (37,41). The sequences of the top (T) and bottom (B) strands of the oligos Hom6 and Hom40, respectively, are given as examples: Hom6-T: 5'-AATTCGGTACCCGG; Hom6-B: 5'-GATCCCGGGTACCG; Hom40-T: 5'-AATTCGGTACC.AA.AC.ATT.CGTCG.AAGGT. CGAGT.CGGAATAGCA.CCCGG; Hom40-B: 5'-GATCCCGGGTGCTATTCCGAC.TCGACC. TTCGA.CGAAT.GT.TTGGTACCG (the common KpnI site (GGTACC) and SmaI site (CCCGGG) are underlined; the intervening sequence of Hom40 is printed in italics, dots indicate the borders of the sequences of the other six oligos Hom10.12.15.20.25.30). These eight plasmids allow the generation of 3 kb plasmid (P) and 1.25 kb phage lambda () fragments both of which are supplemented on one side with a small region of varHom ranging in size from 6 to 40 bp. Cleavage of pSP65/-varHom6–40 with SmaI and HindIII creates the P-fragments, cleavage with KpnI and HindII the -fragments. In this way, the homology regions of 6–40 bp carry exactly the same bl/3'-terminus (Sma/Kpn) configuration in all eight P- and substrates (Figure 8A). In addition to the bl/3'-terminus configuration used here, the use of isoschizomers of SmaI (AvaI) and KpnI (Asp716) permits the creation of three other terminus configurations: bl/5' (Sma/Asp), 5'/5' (Ava/Asp) and 5'/3' (Ava/Kpn). Prior to incubation with extract, all P- and -fragments were gel purified.
Assay for µhomNHEJ and analysis of products
To achieve maximal formation of linear P–-heterodimer products, 15 ng of each gel-purified P- and substrates (molar ratio of 1:2.5) were incubated together in 10 µl standard NHEJ reactions (see above; final concentrations of 3 ng DNA/µl and 3–4 µg extract protein per microlitre). Reaction products were analysed in 1% agarose gels as described for NHEJ assays. For PCR amplification, reaction products were further purified by extraction with phenol and subsequent ethanol precipitation. A set of four primers [#1 (SP6-For): 5'-ACCTTATGTATCA TACACAT; #2 (SP6Rev): 5'-ACACAGGAAACAGCTATGACCA; #3 (-For): 5'-ATCAGCCAG GAGTCCCAAAGAATG; #4 (-Rev): 5'-ATAGTGGATTGCGGTAGTAAA] was used for the amplification of all four P– heterodimers, and P–P- and – homodimers (and circular P- and -monomers) in head-to-tail (H:T) orientation. The P–-heterodimer B (Figure 8A) which alone can form µhomNHEJ products was amplified by primers #1 and #4. An equivalent of 3 ng of substrate input DNA was amplified with 1 U of Taq polymerase (Biolabs, New England Biolabs, Frankfurt a.M. Germany) in a total volume of 50 µl containing Taq buffer provided by the supplier, 2.5 mM MgCl2, 200 µM of each dNTP and 0.25 pmole of each primer. A total of 20 cycles (30 s 94°C, 30 s 56°C, 30 s 72°C) were used with 5 min at 94°C for template denaturation prior to the first cycle and 7 min at 72°C for extended DNA synthesis in the end. Under these conditions, no artifact bands resulting from unreacted substrate input or spurious contamination by circular pSP65/-varHom were produced. One half of the ethanol-precipitated PCR reaction was separated in a 2% agarose gel in the presence of EtBr. The resulting PCR products vary in size from 202 (Hom6) to 270 bp (Hom40) for accurate NHEJ products and from 198 (Hom6) to 232 bp (Hom40) for µhomNHEJ products. For quantification, photographs of the gels were analysed with the OptiQuant 04.00 program (Packard Bioscience).
Western blot analysis
For western blotting, 20 µg of total protein from cell-free extracts (the same ones as used in the NHEJ assays, see above) were separated by SDS–polyacrylamide gel electrophoresis, electroblotted on polyvinylidene fluoride membrane (Millipore) and immunostained. For immune detection of Ku70 (indicative of the Ku70/80 heterodimer), DNA-PKcs and XRCC4, the following antibodies were used: goat anti-Ku70 polyclonal antibody (M-19 sc-1487; Santa Cruz Biotechnology, Santa Cruz, CA; 1:500 dilution), mouse anti-DNA-PKcs monoclonal antibody (Ab-4 cocktail; Lab Visions, Fremont, CA; 1:500 dilution) and rabbit anti-XRCC4 polyclonal antibody (sc-117; Santa Cruz Biotechnology; 1:1000 dilution). To remove unspecific cross reactivity, the anti-XRCC4 antibody was preadsorbed on membranes containing total protein of XR-1 cells. After detection, membranes were reprobed with a mouse anti-/ß-actin monoclonal antibody (pan Ab-5; Lab Visions; 1:500 dilution) as a control for equal loading. The following antibodies were used as secondary antibodies: donkey anti-goat (AP conjugated, Santa Cruz Biotechnology, detection of -Ku70, 1:500 dilution), sheep anti-mouse (HRP conjugated; Amersham Life Science Inc. Europe, Freiburg, Germany; detection of -DNA-PKcs; 1:5000 dilution), donkey anti-rabbit (HRP conjugated; Jackson Immuno Research Europe, Soham, UK; detection of anti-XRCC4; 1:5000 dilution). Either donkey anti-mouse (AP conjugated, Jackson Immuno Research Europe, 1:5000 dilution) or sheep anti-mouse (HRP conjugated, Amersham Life Science Inc., 1:5000 dilution) were used for the detection of anti-/ß-actin. Membranes were blocked in 5% milk in PBS with 0.2% Tween 20 and 0.02% Na-azid for 2 h at room temperature and incubated with primary antibodies diluted in PBS containing 3% BSA and 0.02% Na-azid at 4°C over night in a humid chamber. Probing with secondary antibodies diluted in 5% milk in PBS was performed for 2 h at room temperature. After each incubation with antibodies, membranes were washed for 1 h in 1x PBS with 0.1% Tween 20. HRP-conjugated immunocomplexes were detected using the ECLplus detection system (Amersham Life Science Inc.) according to manufacturers' instructions. AP-conjugated antibodies were detected using NBT/BCIP (Roche Diagnostics, Mannheim, Germany) as described (42).
Immunehistochemistry for 
-H2AX
Cells were grown on coverslips to almost confluence in normal growth medium. After irradiation with 1 Gy of X-rays (Phillips, 3-mm Al filter, 130 kV/16 mA, dose rate 1.2 Gy/min) and indicated repair times, cells were fixed in 4% paraformaldehyde (10 min), washed in PBS (2 x 5min), permeabilized in 0.2% Triton X-100 (10 min), and blocked in PBS with 1% normal goat serum (NGS) and 0.2 M Glycine (4 x 15 min). Coverslips were incubated over night at 4°C with mouse monoclonal anti--H2AX antibody (Vector Laboratories, Newcastle upon Tyne, UK) diluted 1:200 in PBS with 0.5% NGS, washed in PBS with 0.5% NGS (4 x 15 min) and incubated with Alexa-Fluor®594-conjugated goat anti-mouse secondary antibody (Molecular Probes Europe, Leiden, the Netherlands) diluted 1:2000 in PBS with 0.5% NGS (1 h at room temperature). Cells were washed in PBS (2 x 15 min), stained with 4,6 diamidino-2-phenylindole (DAPI) (Roche Diagnostics) 250 ng/ml in PBS; 1 x 15 min), washed in PBS (1 x 15 min) and mounted using Vectashield Mounting Medium For Fluorescence (Vector Laboratories). Cells were viewed in an Olympus BX60 fluorescent microscope equipped with ISIS software (Metasystems, Altlussheim, Germany) and photographed with an IMAC-CCD S30 camera. For quantitative analysis, -H2AX foci per cell nucleus were counted by eye (at least 100 cells per sample) at x100 magnification.
Statistical analysis
To evaluate the qualitative changes in the distribution of µhom utilization in the four different cell lines, allocated samples from each cell line were tested against the distribution of the entire set of 834 inaccurate junctions with Pearson's chi-square test. The null hypothesis assumes that the distribution of the junctions from each mutant resembles the one from the whole population of junctions. The observed chi-square values were found to be 
2 = 5.07 (P = 0.28) for CHO-K1, 2 = 9.76 (P = 0.04) for xrs6, 2 = 18.54 (P = 0.00) for XR-1 and 2 = 1.35 (P = 0.85) for XR-C1. Based on these data it could be shown that CHO-K1 and XR-C1 resemble the whole population (P value > 0.05), while xrs6 differs significantly (P value < 0.05) and XR-1 differs highly significantly (P value < 0.01) compared to the distribution of all junctions.
|
Results |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionSupplementary dataReferences |
|---|
Kinetics of DSB repair is significantly decreased in NHEJ-deficient mutant CHO cells
In this study, we made use of the X-ray sensitive CHO cell lines xrs6, XR-C1 and XR-1 all of which exhibit strongly reduced levels of DSB repair and V(D)J recombination activity. xrs6 and XR-1 are derived from the CHO-K1 wild-type cell line which displays normal radiosensitivity, DSB repair and V(D)J recombination activity (38
,40). xrs6 carries a heterozygous frameshift mutation in the Ku80-encoding XRCC5 gene (the second allele is inactivated by methylation) which leads to the expression of a non-functional 25aa Ku80 truncation. As a result, these cells lack not only Ku80 but also Ku70 and thus are completely devoid of the Ku70/Ku80 heterodimer (43). To exclude spontaneous reversion of xrs6 to Ku80 proficiency by loss of methylation, cells were routinely checked for the absence of Ku70/80 by western blotting (see below). In XR-1, both copies of the XRCC4 gene are completely deleted leading to the absence of XRCC4 protein (and LigIV) so that these cells retain hardly any DSB repair capacity (44,45). XR-C1 is derived from CHO-9 and carries a defect in the XRCC7 gene which leads to the complete absence of DNA-PKcs in these cells (39). In contrast to other DNA-PKcs-deficient rodent cells (murine SCID, V3, irs-20), not only coding but also signal joint formation of V(D)J recombination is defective in XR-C1. Since we did not observe in various different assays any significant differences between XR-C1 and V3 (not shown), another DNA-PKcs-deficient AA8-derived CHO cell line (46), only the results obtained from XR-C1 are presented here. Likewise, we did not observe any differences between CHO-9 cells and CHO-K1 wild-type cells, so that CHO-K1 is used as the NHEJ-proficient control in all experiments presented here. The western blot shown in Figure 1 confirms the mutant nature of the three cell lines by demonstration of the absence of Ku70 (indicative of the Ku70/80 heterodimer) in xrs6, of DNA-PKcs in XR-C1, and of XRCC4 in XR-1 while the NHEJ-proficient CHO-K1 cells contain each of these proteins.
|
In order to investigate the impact of the absence of DNA-PKcs, Ku70/80 and XRCC4/LigIV on global DSB repair in the NHEJ-deficient CHO mutants in comparison to CHO-K1 wild-type cells, we examined the kinetics of the disappearance of
-H2AX foci after X-irradiation. Within seconds after DSB induction, the H2AX histone is phosphorylated by the ATM (ataxia telangiectasia mutated) kinase or DNA-PKcs (47) to yield visible -H2AX foci each of which is assumed to represent a single DSB event making the formation and disappearance of -H2AX foci a suitable measure of DSB repair capacity and radiosensitivity (48). DSB were induced in the three mutants and the wild type by exposure to 1 Gy of X-rays. Prior to irradiation and in a time course from 1 min to 24 h following irradiation, cells were immunostained for -H2AX (Figure 2A). The average number of background (0 Gy) -H2AX-foci (1.5–3.4 foci per cell), as well as the number of foci induced 1 min post-irradiation (44 foci per cell), was almost equal in all four cell lines (Figure 2B). The latter result correlates well with the expected number of 40 DSB produced in a cell per 1 Gy (49). As anticipated, DSB repair is fastest in CHO-K1 wild-type cells reaching almost background levels with 11.8% of foci remaining 24 h post-irradiation (compared to 6.5% at 0 Gy) (Figure 2C). In contrast, DSB repair is slowed 3-fold in XR-C1 (DNA-PKcs) with 36.2% of foci remaining 24 h post-irradiation, and even 5-fold in both xrs6 (Ku70/80) and XR-1 (XRCC4/Lig4) with 55.7 and 57% of foci, respectively, remaining 24 h post-irradiation. This result is similar to the data obtained for mouse embryonic fibroblasts (MEF) deficient in DNA-PKcs and Lig4, respectively (50), with Lig4-deficient MEF (corresponding to XR-1) displaying a stronger defect in DSB repair than DNA-PKcs-deficient MEF (corresponding to XR-C1). To our knowledge, no comparable data are available for Ku70/80-deficient cells in the literature. In summary, the -H2AX assay confirms the mutant nature of the cell lines used here and shows that the DNA-PKcs mutant (XR-C1) displays a significantly weaker repair defect than the Ku70/80- (xrs6) or XRCC4/Lig4 (XR-1) mutant.
|
General assay for NHEJ in vitro
Dependent on the complexity of the DSB to be rejoined, NHEJ comprises different subreactions all of which minimally require the three NHEJ core factors (10
). The simplest NHEJ reaction is the rejoining of complementary cohesive 5'- or 3'-overhangs or blunt ends by ligation using the Ku70/80-DNA-PKcs-XRCC4/LigIV complex. In contrast, the rejoining of non-compatible ends (5'/5', 3'/3', bl/5', bl/3', 5'/3') requires additional modification of DSB ends by nucleolytic trimming (e.g. by Artemis) and/or DNA polymerase-mediated gap filling (e.g. by DNA polµ and/or ) prior to the covalent sealing of remaining nicks by XRCC4/LigIV. To distinguish between the simplest and the more complex substrate types, we use the term ‘ligation substrates’ to describe substrates with compatible ends that can be rejoined by ligation and the term ‘joining substrates’ for substrates with non-compatible ends that require further modification before ligation. As will be outlined below, exact joining of ligation substrates, as is used in many studies (e.g. (25,26)) will not necessarily reveal differences in NHEJ fidelity between mutants defective in different NHEJ core factors. To reliably establish differences in NHEJ fidelity, it is, therefore, necessary to employ an extended spectrum of substrates with non-compatible ends that make different demands on the NHEJ machinery with respect to the structure and sequence of DSB ends.
Plasmids linearized with a single RE represent ligation substrates with compatible cohesive 5'- or 3'-overhangs or blunt ends to measure the efficiency of the Ku70/80-DNA-PKcs-XRCC4/LigIV-mediated ligation reaction in NHEJ. Joining substrates, generated by cleavage with two different REs, contain non-compatible DNA ends that require additional end modifications. End joining in cell-free extracts from Xenopus laevis eggs or mammalian cells converts both substrate types into monomeric oc and ccc and various linear multimers which can be separated in agarose gels (36). Isolation of single NHEJ events for sequence analysis is achieved by cloning of the circular products in E.coli. Using a total of three ligation substrates (#1–3) and 12 joining substrates (#4–15), we have here examined the NHEJ activity of whole-cell extracts from DNA-PKcs-deficient XR-C1, Ku70/80-deficient xrs6 and XRCC4/LigIV-deficient XR-1 cells in comparison with CHO-K1 wild-type cells to investigate the influence of the corresponding NHEJ core factors on the efficiency and fidelity of NHEJ.
Efficiency of NHEJ is decreased in extracts from NHEJ-deficient CHO mutants
Representative examples of the different NHEJ activities of the four cell lines are shown in Figure 3A. Quantification by phosphorimager analysis of the amount of total product (circles + multimers) reveals the differences in NHEJ efficiency for each of the 15 individual substrates which depend on their different terminus configurations and sequences (Figure 3B). In all four cell lines, joining substrates with two non-complementary 3'-overhangs are the ones displaying the lowest efficiency (substrates #6 and #7; to be discussed below). To exclude effects of extract quality on NHEJ efficiency in the four cell lines, different batches of extract from each cell line were tested and confirmed the reproducibility of the observed variations in NHEJ efficiency. As expected, CHO-K1 wild-type cells display the highest NHEJ activity which is reflected by an average total substrate turnover of 40%. The products formed are about equally distributed on circular (oc + ccc) monomers (21%) and linear multimers (19%) (Figure 3C). By contrast, all three mutants exhibit a different product spectrum in which circle formation is almost completely abolished while the formation of multimers is less severely affected. This results in reduced average efficiency of substrate turnover with the DNA-PKcs mutant (XR-C1) exhibiting an intermediate NHEJ efficiency of 23% (1.7-fold reduction versus CHO-K1) followed by the Ku70/80 mutant (xrs6) with 15% (2.7-fold reduction) and the XRCC4/LigIV mutant (XR-1) with 7% (5.7-fold reduction). To verify, that the observed reduction in NHEJ efficiency in the mutants is indeed due to the lack of the corresponding core components in the different extracts, we performed mixing experiments and showed that NHEJ is restored to wild-type levels (including circle formation) in all possible mixtures of mutant extracts (see Supplementary Figure S1 available at Mutagenesis Online). It should be noted that the extent of reduction of the in vitro NHEJ activity corresponds well to the reduction in overall DSB repair established in the -H2AX assay in which XR-C1 also exhibits a milder phenotype than xrs6 and XR-1 (see Figure 2). Our results also demonstrate that, although DNA-PK-dependent NHEJ is compromised in the mutants, they still exhibit significant levels of end joining in the form of linear multimers. This is consistent with the previously proposed existence of alternative ‘backup’ NHEJ pathways that can operate in the absence of the NHEJ core factors (DNA-PK-independent NHEJ) (22,23,27). The decrease of NHEJ efficiency in the mutants compared to the wild type and the variations of NHEJ efficiency between the different mutants are in line with in vivo data obtained from transient transfection experiments performed in various NHEJ-deficient CHO cell lines (24,51). This confirms that our cell-free assay is suitable and reliable to investigate NHEJ reactions in different NHEJ-proficient and -deficient cell lines.
|
The inhibition of DNA-PK-dependent NHEJ by the addition of Wortmannin, a competitive inhibitor of DNA-PKcs, is also accompanied by a complete loss of circle formation in CHO-K1 while formation of linear multimers, although strongly reduced, is still detected (Figure 3D). This result, too, is corroborative of the presence of alternative NHEJ pathways. Inhibition of XR-1 by Wortmannin leads to a further decrease in NHEJ efficiency in this mutant to a level similar to that observed in the inhibited CHO-K1 extract. By contrast, Wortmannin exerts no inhibitory effect on xrs6 and XR-C1. These results are in line with previous reports which show that inhibition of NHEJ by Wortmannin requires the simultaneous presence of both Ku70/80 and DNA-PKcs (25
,26): after treatment with Wortmannin, the Ku70/80/DNA-PKcs complex remains bound to the ends of the substrate DNA (52) unable to facilitate the remaining steps of the reaction which in turn renders the ends inaccessible to alternative NHEJ pathways. Therefore, only the residual small amounts of unbound substrate are accessible to processing by DNA-PK-independent NHEJ (i.e. CHO-K1 and XR-1). By contrast, the lack of either Ku70/80 or DNA-PKcs (i.e. xrs6 and XR-C1) results in failure of Wortmannin inhibition. This is consistent with a role of Ku70/80 and DNA-PKcs in directing end joining to the dominant DNA-PK-dependent pathway: quick and efficient binding of DNA ends by Ku70/80 and DNA-PKcs directs the ends to the rapid DNA-PK-dependent NHEJ reaction and thus blocks the substrate against processing by DNA-PK-independent NHEJ (25,26). Therefore, DNA-PK-independent NHEJ is more efficient in the absence of Ku70/80 (xrs6) and/or DNA-PKcs (XR-C1) but is weaker in the presence of both core factors (XR-1).
In summary, all three mutants display strongly decreased but still substantial NHEJ efficiency (XR-C1 > xrs6 > XR-1). This suggests the existence of at least one alternative DNA-PK-independent NHEJ pathway whose activity is normally suppressed by the dominant DNA-PK-dependent NHEJ pathway. However, in the absence of any of the NHEJ core factors, DNA-PK-independent NHEJ becomes prominent. This switch from DNA-PK-dependent to DNA-PK-independent NHEJ is accompanied by a shift from circles to multimers in the product spectra of the mutants. The fact that multimer formation in the mutants is less affected by the absence of the NHEJ core factors than circle formation could be explained by assuming that these multimers are mostly head-to-head (H:H) or tail-to-tail (T:T) linkages that use complementary (or blunt) ends. However, a substantial fraction of these linear multimers are indeed H:T (see Supplementary Figure S2A available at Mutagenesis Online) and joined in an error-prone way (Supplementary Figure S2B available at Mutagenesis Online). The same phenomenon was observed previously in Xenopus egg extracts upon immunodepletion of Ku70/80 (53). Together with our own results, this shows that the NHEJ activity in the absence of Ku70/80, DNA-PKcs or XRCC4-LigIV is error prone and promotes preferentially the formation of linear multimers (which has been attributed to histone H1 (54)).
Fidelity of NHEJ is decreased in extracts from NHEJ-deficient CHO mutants
In a previous study, we had shown that the Ku80-deficient xrs6 cell line displays, in addition to the decrease in NHEJ efficiency, also a significant reduction in NHEJ fidelity accompanied by an increase of µhomNHEJ which is restored to wild-type levels by complementation with hamster Ku80 cDNA (22). In addition, a number of in vivo studies performed in various NHEJ-deficient CHO mutants using either transient transfection of linear DSB substrates (20,24,51) or chromosomal DSB induced by the I-SceI-nuclease (55–57) showed that xrs6 and XR-1 exhibit reduced NHEJ fidelity with a strong shift towards µhomNHEJ. We therefore asked whether we would be able to reproduce these in vivo observations in our in vitro system and whether it would be possible to establish distinct spectra of junctions for xrs6, XR-1 and XR-C1 which would be comparable to the spectra found in vivo and perhaps characteristic for the defective NHEJ core factor.
In this context, it is important to define the term fidelity of NHEJ. While it is obvious that ‘accurate ligation’ of compatible ends restores the original restriction site used to create the DSB, the definition of accurate joining of non-compatible ends is not self-evident because this necessarily causes a change in the original sequence. Still, general rules for the dominant DNA-PK-dependent NHEJ pathway were established previously because extracts from X.laevis eggs (58) and mammalian cells (36) generate highly reproducible spectra of junctions by two main mechanisms, designated as ‘overlap’ and ‘fill-in’ mechanism (Figure 4). The type of mechanism used is determined by the ends being joined: while the overlap mechanism joins antiparallel (5'/5', 3'/3') overhangs (41), the fill-in mechanism joins abutting (bl/5', bl/3', 5'/3') ends (60). Both mechanisms have the tendency to preserve the original terminal sequences so that even 3'-overhang sequences are not lost but preserved by fill-in DNA synthesis (for details see Figure 4). In the following, all junctions formed by the overlap or fill-in mechanism are therefore defined as ‘accurate’.
|
In the present study, we have analysed the sequences of 1567 cloned junctions obtained by transformation in bacteria of NHEJ products formed from 15 DNA substrates (see Materials and methods section) in the four extracts (total numbers of junctions analysed: CHO-K1: 446, XR-C1: 337, xrs6: 433, XR-1: 309). The 328 junctions derived from joining of ligation substrates with cohesive 5'- (#1) and 3'- (#2) overhangs, and blunt ends (#3) are shown in Figure 5A. Junctions derived from joining of joining substrates with non-compatible ends (#4–15) are subdivided in 435 junctions derived from substrates containing antiparallel ends (#4 and #5: 5'/5', #6 and #7: 3'/3') which are joined by overlap (Figure 5B) and 804 junctions derived from substrates containing abutting ends (#8 and #9: bl/5', #10: bl/3'; Figure 5C, and 5'/3': #11–15 Figure 5D) which are joined by fill in. Note that only the most frequently occurring junctions are shown in Figure 5 (the complete spectra of junctions derived from all 15 substrates can be viewed in Supplementary Figure S3A–O available at Mutagenesis Online). To facilitate a direct comparison between the 15 substrates, the fractions of accurate junctions formed from each substrate in the four cell lines are given in Figure 6A. It is seen that the accuracy values vary considerably between the individual substrates and the four cell lines which will be described below in more detail. The variations in fidelity among the four cell lines are reproducible for different batches of extract (see Supplementary Figure S4 available at Mutagenesis Online). It addition, NHEJ fidelity is restored to wild-type levels in mixtures of different mutant extracts (see Supplementary Table S1 available at Mutagenesis Online) which indicates that the decrease in NHEJ fidelity is indeed due to the lack of the NHEJ core factors in the corresponding mutants.
|
indicating total numbers of junctions analysed in each cell line and underlined percentages the largest fraction of junctions found in each case. The percentages of junctions containing a µhom of 3 or 4 bp are highlighted in white on black ground. Note that only the main junctions are shown in this figure which is the reason why the sum of the single junctions does not equal the total number of junctions analysed.
|
It should be noted that fractions of accurate junctions (as defined in Figure 4), which we here designate as fidelity, is not always 100% in the wild type as one might expect. The fact that NHEJ fidelity of the wild type is sometimes rather low (see below) is on the one hand due to strong variations among the types of ends being joined and on the other hand due to the buffer conditions chosen here. Since previous studies indicated that DNA-PK-dependent NHEJ occurs at low Mg2+ concentrations (0.5 mM), whereas DNA-PK-independent NHEJ requires higher Mg2+ concentrations (
10 mM) (25,62–64), we chose buffer conditions that support the activities of both DNA-PK-dependent and -independent NHEJ in order to facilitate the direct comparison between wild-type and mutant cells (under low Mg2+, mutant cells hardly display any NHEJ activity; data not shown). In fact, the fractions of accurate junctions are nearly 100% in the wild type when low Mg2+ concentrations are used (see Supplementary Figure S5 available at Mutagenesis Online). Under the conditions used here, both NHEJ reactions can occur efficiently in CHO cells which leads to an increase of ‘inaccurate’ junctions and thus to a decrease of the fidelity of the classical DNA-PK-dependent NHEJ below 100% in the wild type. This premise should be taken into account when encountering low fidelity values in some results.
Spectra of junctions formed from ligation substrates with compatible ends (#1–3)
The spectra of junctions obtained from ligation substrates (#1–3; Figure 5A) show that the accuracy of joining of cohesive 5'-overhangs and blunt ends is significantly decreased in xrs6 while the accuracy of joining of cohesive 3'-overhangs remains unaltered in this mutant. By contrast, XR-C1 and XR-1 display wild-type levels of fidelity for all three types of DNA ends (Figure 6B). Considering the fractions of accurately joined compatible 5'-, 3'- and blunt ends [Figure 6A (#1–3) and Supplementary Figure S4 available at Mutagenesis Online], it becomes obvious that only the accuracy of the joining of 5'-cohesive and blunt ends by ligation is dependent on the presence of Ku70/80 while neither DNA-PKcs nor XRCC4-LigIV, although important for ligation efficiency (see Figure 3C), appears to be relevant for fidelity. This result also demonstrates that the use of ligation substrates is not sufficient to detect differences in NHEJ fidelity in different NHEJ mutants.
Spectra of junctions formed from joining substrates with non-compatible ends (#4–15)
To facilitate the analysis of large, statistically relevant numbers of junctions obtained from joining of non-compatible ends, we have designed a series of joining substrates whose non-compatible terminus configurations will reconstitute a restriction site upon ‘accurate joining’ (#4, #6, #9, #10, #11, #15) and, in some cases, ‘inaccurate joining’ (#12, #13). In this way, 50% of all junctions derived from joining substrates can be analysed by simple restriction cleavage which is much quicker and cheaper than sequencing.
With respect to the fidelity of the joining of non-compatible ends, it is seen in Figure 5B–D, that substantial fractions of the junction spectra formed in CHO-K1 display indeed the typical features of overlap (Figure 5B: #4b, #5b, #6a, #7a,b) and fill-in junctions [(a) junctions in Figure 5C and D] and are therefore defined as accurate. This confirms that the rules for accurate joining, originally established for the Xenopus system (see Figure 4), are generally applicable for vertebrate cells.
Spectra of junctions formed by joining of antiparallel ends (substrates #4–7).
The fidelity of the joining of antiparallel 5'/5'- and 3'/3'-overhangs by overlap formation was tested with four substrates (Figure 5B), two with 5'-overhangs (#4 and #5) and two with 3'-overhangs (#6 and #7) of which #4 and #6 form complete overlaps and #5 and #7 terminal overlaps (see Figure 4) (41). The joining of complete overlaps resembles that of cohesive end ligation with the difference that each of the overlaps bears two mismatches in its center. The presence of these mismatches does not disturb the NHEJ reaction so that the complete overlap structures are ‘simply’ ligated with their mismatches being maintained (59). The joining of terminal overlaps with (#7) or without mismatches (#5) is more demanding due to the presence of small gaps that reduce the stability of the overlaps and require the action of a DNA polymerase (probably polµ or pol).
The accuracy of the joining of the complete 5'-overlap (#4) is highest for CHO-K1 (97%) and XR-C1 (79%) but is decreased in xrs6 (26%) and XR-1 (43%) [Figure 5B (#4b) and Figure 6A (#4)]. The higher complexity of the joining of the terminal 5'-overlap is reflected by a slight decrease in fidelity in CHO-K1 (79%) and a strong decrease in the mutants with XR-C1 exhibiting 45% of accurate junctions, xrs6 12% and XR-1 even 0% [Figure 5B (#5b) and Figure 6A (#5)]. This result is in good accordance with data obtained in vivo with the same Bam/Sal terminus configuration showing that overlap junctions are formed in the wild type but are absent in xrs6 and XR-1 (20). Instead of overlap formation, joining in the mutants has shifted to ‘blunting’ in which both 5'-overhangs are filled in to yield blunt ends which are subsequently ligated [Figure 5B (#4a, #5a)]. This competition between overlap and blunting is hardly detected in CHO-K1 (0 and 7%) but is strongest in XR-1 (43 and 85%) followed by xrs6 (26 and 27%) and weakest in XR-C1 (17 and 13%).
When compared to the joining of 5'-overlaps, the fidelity of the joining of 3'-overlaps (#6 and #7) is decreased in CHO-K1 (48 and 36%) while it remains constant in XR-C1 (48 and 68%) [Figure 5B (#6a, #7a,b)]. By contrast, not even a single junction fulfils the criteria of accurate 3'-overlap joining in xrs6 and XR-1 (both 0%) [Figure 6A (#6, #7)]. Instead, most junctions display complete loss of both 3'-overhangs followed by blunt-end ligation [Figure 5B (#6b, #7c,d)]. This reaction is adequate to the blunting of 5'-overhangs described above except for the fact that 3'-overhangs cannot be filled in by a DNA polymerase but must be trimmed by an exo- or endonuclease to render a blunt end. In addition to blunting, a second reaction is seen in XR-1 for substrate #7 where 27% of the junctions display a µhom of 4 bp at their break points. This phenomenon will be discussed below in more detail.
Spectra of junctions formed by joining of abutting ends (substrates #8–15).
As expected, the joining of substrates with abutting bl/5'-, bl/3'- and 5'/3' ends in CHO-K1 is achieved mainly by accurate fill in that conserves the entire sequences of both 5'- and 3'-overhangs by fill-in DNA synthesis (Figure 5C and D). In the case of 3'-overhangs, fill in has to be primed at the partner terminus (blunt end or filled 5'-overhang; see Figure 4) (60). This is thought to be achieved by untemplated addition of a single nucleotide to the blunt end or filled 5'-overhang which, if matching the terminal nucleotide of the 3'-overhang, can serve as a primer for discontinuous DNA synthesis (61).
The accuracy of the fill-in mechanism varies strongly with the type of ends being joined. Substrates containing bl/5' ends [Figure 5C (#8a, #9a)] display the highest accuracy (78 and 79%, respectively) followed by substrates containing 3' ends [Figure 5C (#10a); Figure 5D (#11–15)] which display reduced fidelity (ranging between 17 and 45%; see also Figure 6A). Fidelity of Sac/Sal (#12) reaches even 80% which may be due to variations in extract quality (abundance of NHEJ factors; batch and age of extract) and special features of the terminal sequences. This shows that DNA-PK-dependent NHEJ in CHO-K1 is able to generate accurate junctions as defined in Figure 4 by fill in of 3'-overhangs by discontinuous DNA synthesis that is primed at the partner terminus (60). As mentioned above, the observed strong variations in NHEJ fidelity in the wild type are mainly due to the fact that we have chosen buffer conditions that permit both DNA-PK-dependent and -independent NHEJ to occur at the same time with the latter pathway contributing to the formation of inaccurate junctions and thus to a decrease of the fidelity values in the wild type (see Supplementary Figure S5 available at Mutagenesis Online: fidelity values in the wild type are much higher at low Mg2+ concentration, when mainly classical DNA-PK-dependent NHEJ is active and approaches zero at high Mg2+ concentration, when ‘error-prone’ DNA-PK-independent NHEJ becomes prominent).
Although displaying slightly decreased fidelity for the joining of bl/5' ends, all three mutants are able to perform accurate fill in [Figure 5C (#8a, #9a), Figure 6A]. This is due to the fact that in the case of bl/5' ends, the accurate fill-in mechanism is identical with the blunting reaction which was already observed for the 5'/5' substrates (#4 and #5) where, mainly in xrs6 and XR-1, the accurate overlap mechanism is substituted by blunting [compare Figure 5B (#4a, #5a)]. This indicates that blunting by complete fill in of 5'-overhangs and subsequent blunt-end ligation is an efficient DNA-PK-independent NHEJ reaction in all three mutants meaning that blunting can occur in the absence of Ku70/80, XRCC4-LigIV or DNA-PKcs.
A completely different picture emerges from the fill-in substrates containing 3'-overhangs (bl/3' and 5'/3') [Figure 5C (#10) and Figure 5D (#11–15)]. While XR-C1 is still able to rejoin these substrates by accurate fill in with a fidelity that is only slightly reduced with respect to CHO-K1, neither xrs6 nor XR-1 has formed a single accurate junction [Figure 6A (#10–15)]. Instead, all 3'-overhang sequences are lost entirely. In a significant number of cases, this leads to blunt-end ligation between the trimmed 3'-overhang and the partner blunt end [Figure 5C (#10b)] or the filled partner 5'-overhang [Figure 5D (#11c, #12c, #13b, #14b, #15c)]. This indicates that, in the absence of Ku70/80 or XRCC4-LigIV, the accurate fill-in mechanism is substituted by nucleolytic blunting of 3'-overhangs and suggests that the ability to conserve entire 3'-overhang sequences by discontinuous DNA synthesis is strictly dependent on the presence of Ku70/80 and XRCC4-LigIV but not on DNA-PKcs.
Alternative DNA-PK-independent NHEJ reactions
Blunting.
The analysis of the total junction spectrum shows that the fidelity of the joining of non-compatible ends (substrates #4–15) is generally lower than that of the joining of compatible ends (substrates #1–3) and varies considerably between the individual mutants (Figure 6B). As mentioned above, fidelity of compatible end joining is significantly reduced only in xrs6 (i.e. blunt ends and 5'-overhangs; Supplementary Figure S6 available at Mutagenesis Online) but not in XR-C1 and XR-1. The variation in fidelity appears to be dependent on the types of ends being joined. With respect to non-compatible end joining, XR-C1 rejoins all three types of ends with similar fidelity which is only slightly reduced when compared to CHO-K1 (Figure 6C). This indicates that DNA-PKcs has only limited impact on NHEJ fidelity in CHO cells. By contrast, NHEJ fidelity is strongly reduced in xrs6 and XR-1 with 3' ends exhibiting the biggest effect as not a single 3'-end is accurately rejoined by overlap or fill in in neither of the two mutants (Figure 6C). 5'-ends exhibit a decreased level of fidelity with respect to XR-C1 while blunt ends are rejoined with unaltered fidelity when compared to XR-C1. So far, we can summarize that blunting by fill in of 5'-overhangs and/or exact nucleolytic trimming of 3'-overhangs (–4 nt) and subsequent blunt-end ligation constitute a major alternative DNA-PK-independent NHEJ reaction especially in xrs6 and XR-1.
In addition, all kinds of small deletions (involving 5'-, 3'- and blunt ends) are observed. In this context, we have chosen the term ‘stability’ as the tendency of an end to lose single nucleotides and thus to be involved in the formation of deletions (high stability: few losses, low stability: frequent losses). For the interested reader, a detailed analysis of the stability of the three types of ends (5'-, 3'- and blunt ends) is presented in Supplementary Figure S7A–C available at Mutagenesis Online. The stability of all ends is lowest in xrs6 directly followed by XR-1. By contrast, XR-C1 exhibits only slightly decreased stabilities when compared to CHO-K1. The spectrum of deletion sizes observed in all inaccurate junctions is displayed in Figure 6D. XR-1 (white bars) shows a clear peak of deletions at 4 bp which is indicative of the alternative blunting reaction that is prominent in XR-1. By contrast, xrs6 peaks at larger deletions (>10 bp) which indicates that Ku70/80 is crucial for the inhibition of nuclease degradation of the ends. It should be noted that the deletions in inaccurate junctions obtained in vivo are often larger and more heterogenous (20,51) than the ones obtained in our in vitro system. This might be due to the transfection procedures used in these experiments, because terminal degradation of the substrate molecules in vivo is not necessarily a result of the NHEJ reaction but may occur in the nucleus and/or during passage of the substrate through the cytoplasm. Thus, only a fraction of the substrate molecules undergoing NHEJ would maintain their original terminal structures which in turn would significantly influence the spectra of NHEJ products observed. This problem is minimized in the cell-free assay because NHEJ factors gain access to the substrates directly upon incubation so that possibly observed terminal degradation is indeed the result of the NHEJ process itself. This problem is also minimized in vivo when substrates are cleaved inside the cell by the I-SceI-nuclease. Of the deletions formed by the I-SceI-method in CHO cells, only 40% of junctions have lost >10 bp (65) which resembles more closely the sizes of the deletions found in our in vitro system.
Microhomology-mediated NHEJ.
A most interesting class of inaccurate junctions is deletions displaying µhom at their break points. In the case of substrate #10 (bl/3', Figure 5C), joining of non-compatible ends via blunting (sequence #5: 0/–4) constitutes only a minor fraction in xrs6 and XR-1. Instead, 50% of the junctions in these two mutants display an 8-bp deletion with a 4-bp µhom at its break point (sequence #12: 0/–8, see also the complete spectra of junctions in Supplementary Figure S3A–O available at Mutagenesis Online). This junction type also occurs in XR-C1 (21%) but is absent in CHO-K1. This result shows that DNA-PK-independent NHEJ in the mutants comprises, in addition to blunting, a second reaction type, designated as µhomNHEJ, which creates small deletions that span one of two short homologous sequences and the intervening sequence if present.
The use of µhom appears to be dependent on their location within the terminal sequences [see Figure 5B (#7f); Figure 5C (#10c); Figure 5D (#12d, 13c, 14b, 15f)]. This becomes clearer by a comparison of the junction spectra formed by the Sma/Pst substrate [Figure 5C (#10c)] and the Sac/Hind3 substrate [Figure 5C (#14c)]. In the first case, the GCCC repeat lies within the duplex regions flanking the blunt end on the left side and the 3'-overhang on the right side so that the 3'-overhang can be regarded as a heterologous-intervening sequence. Here, xrs6 and XR-1 display almost equal frequencies of the (0/–8) deletion, whereas XR-C1 has a frequency of 21% and CHO-K1 of 0%. In the second case, the 3'-overhang and the 5'-overhang, respectively, themselves are the repeats. Here, the AGCT repeat can be used only after fill in of the 5'-overhang and maintenance of the 3'-overhang so that the AGCT of the 3'-overhang can subsequently base pair with the TCGA of the filled 5'-overhang. Of course, this (0/–4) deletion could also be explained by a simple blunting reaction in which the 3'-overhang is completely degraded. However, the fact that this type of junction constitutes 93% of all junctions formed in XR-1, whereas the fractions of comparable 0/–4 blunt end junctions formed by XR-1 from other substrates vary only between 15 and 62% [Figure 5C (#10b), Figure 5D (#11c, #12c, #13b, #15c)] argues for an active use of the µhom in this mutant (see also the complete spectra of junctions in Supplementary Figure S3A–O available at Mutagenesis Online).
In order to investigate the use of µhom by DNA-PK-independent µhomNHEJ, we have analysed the frequency of the occurrence of µhom in a total of 834 inaccurate junctions formed from all 15 substrates in the four cell lines (Figure 7; the large fraction of accurate which also can contain µhom patches are not considered in this analysis; a statistical analysis with P values is presented in Materials and methods section). While the observed frequency of inaccurate junctions containing µhom of 0–3 bp is close to the expectation (see Figure 7), the observed frequency of inaccurate junctions with a µhom of 4 bp is significantly increased with respect to the expected frequency of 1% indicating that µhomNHEJ favours longer µhom patches (see below). The use of µhom of 4 bp is highest in XR-1, followed by XR-C1, whereas xrs6 and CHO-K1 display comparable frequencies of µhom use. These results are in good agreement with data obtained in vivo with a 7-bp µhom which showed that µhom use is more efficient in XR-1 than in xrs6 (20). The result that CHO-K1also displays relatively high frequencies of µhom at 4 bp is again due to the fact that we have chosen reaction conditions that permit both DNA-PK-dependent and -independent NHEJ to occur simultaneously, so that a fraction of inaccurate junctions are formed by µhomNHEJ in the wild type as well. This is expected to reflect the situation under physiological conditions where both NHEJ pathways are present. The result that the use of µhom in xrs6 is close to the wild type is due to the fact that xrs6 exhibits the lowest stability of DNA ends (see above). Depending on the location of the µhom (directly at the terminus or in the adjacent double strand), the µhom patch is sometimes lost and thus cannot contribute to the formation of a µhomNHEJ junction [see e.g. Figure 5D (#14b)].
|
