Disruption of the Arabidopsis AtKu80 gene demonstrates an essential role for AtKu80 protein in efficient repair of DNA double-strand breaks in vivo

Authors


For correspondence (fax: +44 161 275 3938; email: cbray@man.ac.uk)

Summary

Double-strand breaks (DSBs) in DNA may occur spontaneously in the cell or be induced experimentally by γ-irradiation, and represent one of the most serious threats to genomic integrity. Non-homologous end joining (NHEJ) rather than homologous recombination appears to be the major pathway for DSB repair in humans and plants, and it may also be the major route whereby T-DNA integrates into the plant genome during cell transformation. In yeast and mammals, the exposed ends of damaged DNA are bound with high affinity by a dimer of Ku70 and Ku80 proteins, which protects the ends from exonucleases and juxtaposes the two ends of the DSB, independent of sequence homology. Here we report the functional characterization of Ku70 and Ku80 from Arabidopsis thaliana, and demonstrate that AtKu80 and AtKu70 form a heterodimer with DNA binding activity that is specific for DNA ends. An atku80 knockout mutant shows hypersensitivity to the DNA-damaging agents menadione and bleomycin, consistent with a role for AtKu80 in the repair of DSBs in vivo in Arabidopsis.

Introduction

DNA double-strand breaks (DSBs) may be induced by γ-irradiation, radio-mimetic chemicals or by breaks in single-stranded regions of DNA created, for example, during DNA replication (Kuzminov, 2001). If it is not repaired, a single DSB can cause cell death, and organisms have evolved efficient processes to detect and repair these cytotoxic DNA lesions. Where an intact copy of the damaged DNA is available, bacteria and yeast (Saccharomyces cerevisiae) repair DSBs predominantly by homologous recombination (HR) (Haber, 2000). In organisms with larger genomes including Arabidopsis and humans (Homo sapiens), HR is usually rare, and most DSBs are rejoined by non-homologous end joining (NHEJ), largely independent of sequence homology between the DNA ends (Lieber, 1999). NHEJ is intrinsically more error-prone than HR, often producing short deletions and insertions, but in large genomes with repetitive sequence elements, HR-mediated DNA repair may lead to large-scale genetic rearrangements and chromosomal instability (Jeggo, 1997). The NHEJ pathway in plants is of particular interest as it may also be involved in T-DNA insertion during transformation and movement of transposons (Gorbunova and Levy, 1999). Deletions and insertions of random filler DNA are found at the junctions of the inserted and genomic DNA, and T-DNA insertion requires the activity of endogenous plant nuclear DNA repair pathways (Ziemienowicz et al., 2000). Similar mutational repair products are found in plasmids recircularized in tobacco protoplasts and in the mutational repair of a highly specific I-SceI restriction site in a transgene in the tobacco genome (Gorbunova and Levy, 1997; Salomon and Puchta, 1998). Insertions and deletions may occur more frequently in plants than mammals, although large variations also exist between different plant species (Gorbunova and Levy, 1999; Kirik et al., 2000). The NHEJ pathway is required for T-DNA integration in S. cerevisiae, and, in the absence of yeast Ku70, T-DNA integration was not observed (van Attikum et al., 2001).

The pathway of NHEJ in S. cerevisiae and mammals has been extensively characterized over recent years, and many of the molecular components have putative homologues in other eukaryotes (reviewed in Critchlow and Jackson, 1998; Kanaar et al., 1998; Lewis and Resnick, 2000). The first step in DSB repair is recognition and binding of the exposed DNA ends by a heterodimer of Ku70 and Ku80 proteins (Ku) (Dynan and Yoo, 1998). These Ku DNA end-binding proteins co-localize in the nucleus and bind to DNA ends with high affinity (Koike et al., 1998; Wang et al., 1998). Ku causes looping of DNA in vitro, suggesting that it brings free DNA ends into close proximity, consistent with a role in DSB repair (Cary et al., 1997). Ku also protects DNA ends from large-scale degradation by exonucleases, although modification of DNA ends by DNA polymerases and nucleases can occur in the presence of Ku (Lieber, 1999). In vertebrates, a third protein, the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), binds to the Ku dimer forming a protein complex termed DNA-dependent protein kinase (DNA-PK) that has a role in intracellular DNA damage signalling (Smith and Jackson, 1999). Ku may also recruit other components of the NHEJ pathway into the DSB repair complex and may interact with other proteins involved in DNA repair or DNA damage signalling (Lewis and Resnick, 2000; Nick-McElhinny et al., 2000). Mammalian Ku70 interacts with Mre11, a component of the Rad50/Mre11/Nbs1 complex, which has a role in DNA end processing in NHEJ and HR (Goedecke et al., 1999). In addition, Ku binding to DNA ends stimulates eukaryotic DNA ligase activity in vitro (Ramsden and Gellert, 1998). DNA ligase IV (Lig4) and XRCC4 (Lif1) catalyse ligation of the DSB in mammals (and yeast). DNA ligase I in yeast, or DNA ligases I or III in mammals, cannot completely substitute for DNA ligase IV in vivo (Adachi et al., 2001; Teo and Jackson, 1997; Wilson et al., 1997). We have previously isolated and characterized DNA ligase I, DNA ligase IV and XRCC4 in Arabidopsis, and shown that transcription of DNA ligase IV is induced by γ-irradiation (Taylor et al., 1998; West et al., 2000). Recently, Tamura et al. (2002) cloned Arabidopsis Ku70 (AtKu70) and Ku80 (AtKu80) and found constitutive low-level expression in all tissues and more than threefold induction of AtKu70 and AtKu80 expression in protoplasts after exposure to DNA-damaging agents. AtKu70 and AtKu80 proteins are capable of forming a heterodimer which binds to double-stranded but not single-stranded DNA in vitro. Here we extend the characterization of Ku70 and Ku80 in Arabidopsis, and show that both the AtKu70/AtKu80 heterodimer and the AtKu70 protein, but not the AtKu80 protein, can bind to DNA ends. We have also isolated Arabidopsis mutants in which the AtKu80 gene is disrupted, and demonstrate that the Ku complex is required for DSB repair in vivo in Arabidopsis.

Results

Sequence analysis of Arabidopsis Ku70 and Ku80

Arabidopsis Ku80 (accession number AF283758) and Arabidopsis Ku70 (accession number AF283759) cDNAs were cloned and sequenced by Tamura et al. (2002) based on chromosome I genomic sequence information in F6I1 (accession number AC051629) and BAC F21D18 (accession number AC023673), respectively (Arabidopsis Genome Initiative, 2000). ESTs corresponding to AtKu80 were found in root (accession number AI996630) and silique tissue (accession number AV564250) in Arabidopsis, consistent with RT–PCR analysis showing constitutive expression of AtKu80 in all tissues studied. Analysis of the complete Arabidopsis genomic sequence suggested that AtKu80 is a single-copy gene, consistent with Southern analysis of Arabidopsis DNA digests (data not shown). AtKu70 ESTs were found in flower buds (accession number AV533293) and the developing seed (accession number BG459249). Analysis of the Arabidopsis genomic sequence suggested that AtKu70 is a single-copy gene.

AtKu70 interacts with AtKu80

Two approaches were used to investigate whether Arabidopsis Ku70 and Ku80 interact in vitro. AtKu70 and AtKu80 were synthesized by in vitro transcription and translation as [35S]-methionine-labelled c-myc- and hemagglutinin (HA)-tagged proteins, respectively. Labelled proteins were then immunoprecipitated either separately or together with antisera specific to either the c-myc tag or the HA tag as described in the Experimental procedures. AtKu70 was immunoprecipitated by anti-c-myc antisera, and AtKu80 by anti-HA antisera, with little cross-reaction between AtKu80 and the anti-c-myc antisera or AtKu70 and the anti-HA antisera (Figure 1a). When incubated in the presence of AtKu80, AtKu70 was immunoprecipitated by the addition of anti-HA antisera (Figure 1a), indicating that AtKu70 was co-immunoprecipitating together with Ku80 and therefore that Arabidopsis AtKu80 and AtKu70 interact in vitro. Comparable results were obtained when a mix of AtKu70 and AtKu80 was immunoprecipitated with anti-c-myc antisera, with an increase in AtKu80 precipitation observed in the presence of AtKu70, again indicative of AtKu70/AtKu80 interaction (Figure 1a).

Figure 1.

Arabidopsis Ku70 interacts with Arabidopsis Ku80.

(a) Co-immunoprecipitation of AtKu70 and AtKu80. In vitro transcription and translation produced radiolabelled HA-tagged AtKu80 and c-myc-tagged AtKu70. Radiolabelled proteins were incubated separately or together and immunoprecipitated with either anti-HA or anti-c-myc antisera, followed by SDS−PAGE and autoradiography. (b) Interaction between AtKu70 and AtKu80 on affinity-column chromatography. Arabidopsis Ku80 was over-expressed as a poly-histidine-tagged protein in E. coli and purified by nickel affinity chromatography. Purified AtKu80 was bound to 100 ml nickel affinity beads and radiolabelled AtKu70 was applied to the AtKu80 affinity beads or affinity beads alone. A step gradient of 250–1000 mm NaCl was used to wash the affinity beads before a final elution of the histidine-tagged AtKu80 with 1 m imidazole. Fractions were analysed by SDS−PAGE and autoradiography.

These results were confirmed by affinity-column chromatography. The full-length Arabidopsis Ku80 cDNA was over-expressed in E. coli with an N-terminal 6×His tag, and the protein purified to homogeneity by immobilized metal affinity chromatography. Interaction between [35S]-methionine-labelled AtKu70 and recombinant 6×His-tagged Ku80 bound to a nickel affinity column was investigated. Radiolabelled AtKu70 was incubated with AtKu80-bound affinity beads and eluted with a step NaCl gradient, with a final elution with 1 m imidazole to release the bound 6×His protein. Radiolabelled AtKu70 did not bind to the nickel affinity column in the absence of Ku80, whereas AtKu70 binding to AtKu80-bound affinity beads was stable to at least 1 m NaCl and both proteins were only co-eluted from the column in the 1 m imidazole wash (Figure 1b). This is indicative of a strong, stable association between Arabidopsis Ku70 and AtKu80 in vitro, and verifies the previous report of AtKu70/AtKu80 interaction shown by yeast two-hybrid analysis (Tamura et al., 2002).

Ku interacts with DNA ends

The DNA binding properties of over-expressed purified AtKu70 and AtKu80 (Figure 2a) were investigated by gel-shift assays. The Ku dimer bound to a double-stranded oligonucleotide, with approximately 60% of the radiolabelled oligonucleotide (0.1 pmol) bound by 0.1 pmol of Ku dimer (Figure 2b), and binding to DNA was reduced with increasing NaCl concentration (Figure 2c). The oligonucleotide used for binding assays has a four-base 5′ overhang at one end and is blunt-ended at the other end. It is not possible to estimate whether the Ku dimer binds to both DNA ends, or preferentially to one end, although no intermediate-sized products are seen. Further experiments are required to determine the stoichiometry of Arabidopsis Ku binding to DNA ends. AtKu80 did not bind to dsDNA in the absence of AtKu70, but AtKu70 alone bound to dsDNA, with similar levels of binding as the AtKu70/80 dimer (Figure 2d). This result differs from that of Tamura et al. (2002) where AtKu70 did not bind the DNA substrate in the absence of AtKu80. This may reflect differences in the buffer composition, amounts of Ku protein used, or the use of a blunt-ended DNA substrate by Tamura et al. (2002). In addition, the present study found that AtKu70 and AtKu70/80 dimer dsDNA binding activity was abolished by the addition of excess competitor DNA ends but was not affected by ssDNA or closed circular plasmid DNA, indicating specificity of Ku for binding double-stranded DNA ends (Figure 2d).

Figure 2.

Ku binds to DNA ends in vitro.

(a) SDS−PAGE and Coomassie blue staining of over-expressed, purified His-tagged AtKu80 and AtKu70. (b) Purified AtKu70 and AtKu80 were assayed for DNA end-binding activity. Ku (0–1 pmol as indicated) was incubated with radiolabelled double-strand oligonucleotides (0.1 pmol) in the presence of excess competitor closed circular plasmid DNA, and migration of the dsDNA was assayed by gel electrophoresis and analysed using a Phosphorimager. (c) Stability of DNA end-binding with increasing NaCl concentration. Ku (0.1 pmol) was bound to radiolabelled dsDNA (0.1 pmol) in the presence of increasing NaCl concentrations, and DNA migration was analysed as in (b). (d) Specificity of AtKu70 and AtKu80 DNA binding. AtKu70 and AtKu80 were incubated separately and together with radiolabelled dsDNA in the presence of excess ssDNA, dsDNA and closed circular plasmid DNA as indicated, and DNA migration was analysed as in (b).

Identification of a T-DNA insertional mutant in AtKu80

Pooled DNA from Arabidopsis (genotype Wassilewskija (Ws)) transformed with the Agrobacterium tumefaciens 3850:1003 Ti plasmid was obtained from the ABRC stock centre and screened for inserts in the AtKu80 gene (McKinney et al., 1995). A T-DNA insert was found in the ninth exon (of 12) of the AtKu80 gene (Figure 3a,b), and the atku80 mutant plant was isolated by subsequent rounds of PCR screening. A homozygous mutant plant was isolated but the phenotype of the mutant was not obviously different from wild type under normal growth conditions. Southern analysis confirmed that the T-DNA insert was in the AtKu80 gene, and that only one T-DNA insert was present in the genome of the mutant (Figure 3c). The left border region showed a 3 bp deletion of the T-DNA and a 33 bp insertion of filler DNA, including 21 bp duplicated from the AtKu80 gene at the insertion site (highlighted in Figure 3b) and 12 bp of unknown origin (Figure 3b). RT–PCR using RNA isolated from above-ground tissue of flowering Arabidopsis plants could not detect any wild-type AtKu80 transcript in the mutant plant, as expected for an insertional mutant in a single-copy gene (Figure 3d). Mutant transcript was detected by RT–PCR and this included the 5′ end of AtKu80 although no 3′ end of transcript was detected, suggesting that transcription termination occurs within the 17 kb T-DNA insertion (Figure 3e). The mutant transcript, if expressed, could encode 490 of the 621 amino acids of the wild-type protein (marked by an asterisk in Figure 4), although this truncated form of AtKu80 would be unlikely to be functional (see Discussion).

Figure 3.

Genomic organization and site of T-DNA insertion in Arabidopsis Ku80.

(a) The full-length AtKu80 genomic sequence is located on chromosome I BAC F21D18 (accession number AC023673, complementary strand bases 98 008–94 417), and is shown schematically in the 5′ to 3′ direction with exons as open boxes. The T-DNA insertion is shown in the ninth exon of AtKu80. (b) DNA sequence of the insertion site. The Arabidopsis Ku80 nucleotide sequence is shown aligned to the predicted amino acid sequence at the region of T-DNA insertion. The T-DNA sequence is indicated by a black bar and the right border (RB) and left border (LB) regions are indicated. The position of the filler DNA is indicated by the open box, and the duplicated sequence derived from the Ku80 gene is highlighted in grey. (c) Southern analysis of the T-DNA left border. A 497 bp probe corresponding to nucleotides 4–501 of the T-DNA left border region (accession number AF019744) was used in Southern analysis of Arabidopsis WS and atku80 mutant genomic DNA. Genomic DNA was incubated overnight at 37°C with KpnI or HindIII as indicated, and separated on a 0.8% agarose gel, transferred to nylon membrane and analysed by hybridization. (d) RT–PCR analysis of RNA isolated from WS and atku80 mutant Arabidopsis. Oligonucleotide primers located either side of the insertion site were used in RT–PCR to determine the presence of wild-type AtKu80 transcript in WS and atku80 mutant Arabidopsis. (e) RT–PCR analysis of RNA isolated from WS, atku80 mutant and atku80+AtKu80 Arabidopsis plants. Oligonucleotide primers were used in RT–PCR to determine the presence of wild-type or mutant AtKu80 transcript in the Arabidopsis lines WS, atku80 and atku80 complemented with the wild-type AtKu80 gene (atku80+AtKu80) as described in Experimental procedures.

Figure 4.

Alignment of Arabidopsis Ku80, human Ku80, Xenopus laevis Ku80 and S. cerevisiae Ku80.

Protein sequences were aligned using Clustal W (Thompson et al., 1994) and the alignment annotated to give black boxes around identical residues and grey boxes around conservative substitutions aligned in two or more sequences. The heterodimerization domain of human Ku80 (residues 338–592) is indicated by a black bar, and the conserved 12 amino acid C-terminal peptide responsible for interaction with DNA-PK in humans and X. laevis is indicated by a white bar (Gell and Jackson, 1999). An asterisk indicates the position corresponding to the T-DNA insertion site in the genomic AtKu80 sequence (residue 490). Accession numbers are: Arabidopsis Ku80, AF283758; human Ku80, P13010; X. laevis Ku80, AB020609; S. cerevisiae Ku80, Q04437.

Sensitivity of the atku80 mutant to DNA-damaging agents

Homozygous atku80 mutants were compared to wild-type plants with respect to their sensitivity to DNA-damaging agents. atku80 mutants showed hypersensitivity to menadione (2-methyl-1,4-naphthoquinone), a quinone that causes oxidative damage in cells (Reichheld et al., 1999; Shi et al., 1994). DNA damage products in human cells exposed to menadione included single-strand breaks and DSBs, and a strong correlation between the extent of DSB damage and cytotoxicity was found (Nutter et al., 1992). Concentrations of 100 µg ml−1 menadione were required to kill wild-type plants, whereas lethality occurred at concentrations of 20 µg ml−1 in mutant plants, with wild-type plants being largely unaffected at this concentration (Figure 5a). UVC radiation also causes oxidative damage in cells, but the atku80 mutant displays no significant hypersensitivity to UVC compared to wild-type plants (Patrick Gallois, personal communication). Bleomycin is a chemotherapeutic drug that induces single- and double-strand breaks in DNA (Burger et al., 1981). In wild-type Arabidopsis, repair of bleomycin-induced DNA strand breaks was rapid, with no detectable strand breaks present 1 h after treatment (Menke et al., 2001). Wild-type Arabidopsis plants grown on media containing various concentrations of bleomycin showed slowed growth compared to untreated plants, and, at concentrations of 1 µg ml−1 bleomycin, abnormal callus-like growth was observed (Figure 5b). In contrast to wild-type plants, atku80 mutants showed much greater sensitivity to bleomycin, with lethality (judged by bleaching and absence of growth) occurring at concentrations of 1 µg ml−1 compared to 4 µg ml−1 in the wild-type plant (Figure 5b). Complementation of atku80 with the wild-type AtKu80 resulted in wild-type sensitivity to both bleomycin and menadione, demonstrating that the hypersensitivity to these agents was due to a deficiency in AtKu80 activity in the atku80 mutant plants (Figure 6). This is consistent with a key role for AtKu80 in the repair of single- and/or double-strand breaks in vivo.

Figure 5.

Phenotype of the atku80 T-DNA insertional mutant.

(a) Hypersensitivity of atku80 mutant Arabidopsis to menadione. Wild-type and atku80 mutant plants were germinated on MS agar, and 1-week-old plants were transferred to MS agar plates supplemented with menadione at the concentration indicated. Plants were photographed after 3 days. (b) Hypersensitivity of atku80 mutant Arabidopsis to bleomycin. Wild-type and atku80 mutant plants were germinated on MS agar, and 1-week-old plants were transferred to MS agar plates supplemented with bleomycin at the concentration indicated. Plants were photographed after 1 month.

Figure 6.

Complementation of atku80.

Wild-type, atku80 mutant plants and atku80+AtKu80 plants were germinated on MS agar, and 1-week-old plants were transferred to MS agar plates or MS agar plates supplemented with menadione (25 mm) or bleomycin (2 μg ml−1) as indicated. Untreated plants and plants transferred to menadione plates were photographed after 3 days, and plants transferred to bleomycin plates were photographed after 3 weeks.

Discussion

NHEJ is the major pathway of DSB repair in higher plants, although little is known about the mechanism at the molecular level (Britt, 1999). Recent studies in S. cerevisiae and mammals have identified key components in this pathway, and homologues of many of these genes can be found in the Arabidopsis genome sequence, including DNA ligase IV, XRCC4, RAD50, Ku70 and Ku80 (Gallego et al., 2001; Tamura et al., 2002; West et al., 2000). An Arabidopsis homozygous RAD50 mutant was sterile and hypersensitive to the DNA-damaging agent MMS, consistent with the role of human and S. cerevisiae RAD50 in NHEJ and meiotic recombination (Gallego et al., 2001). However, it is apparent that differences exist between the DNA repair pathways of plants and mammals, as RAD50 and DNA ligase IV null mutants are lethal in mammals, in contrast to the situation in Arabidopsis (Gallego et al., 2001; C.E. West, unpublished results).

Arabidopsis Ku70 and Ku80 interact in vitro as shown by co-immunoprecipitation and by AtKu70 binding to AtKu80 immobilized to a column. The AtKu70/AtKu80 complex was very stable, and interaction between AtKu70 and AtKu80 withstood 1 m NaCl, possibly indicating strong hydrophobic interactions between these proteins. Yeast two-hybrid analysis has also been used to demonstrate interaction between AtKu70 and AtKu80 (Tamura et al., 2002). Structural studies suggest that mammalian Ku binds to an exposed DNA end as a heterodimer, and that interaction between two heterodimers brings two DNA ends together (Cary et al., 1997; Walker et al., 2001). However, in the current investigations, it was not possible to distinguish whether Arabidopsis Ku70 and AtKu80 formed a heterodimer or a larger complex.

DNA-PKcs is recruited to the Ku70/Ku80 complex in vertebrates. Recent studies have shown that human DNA-PKcs binds specifically to a 12 amino acid peptide at the C-terminus of Ku80, and this region is highly conserved in vertebrates, but absent from Drosophila melanogaster, yeast and Arabidopsis Ku80 proteins (Figure 4; Gell and Jackson, 1999). Neither yeast nor the Arabidopsis genome contain genes encoding any obvious putative homologues of DNA-PKcs, suggesting that DSB damage-signalling pathways may differ significantly between plants, yeast and vertebrates. The S. cerevisiae cell-cycle checkpoint proteins Rad9 and Mec1, which have homologues in plants and mammals, may play a role in DSB signalling in S. cerevisiae (Lustig, 1999; Mills et al., 1999), and a similar role for these proteins in Arabidopsis is possible.

The Arabidopsis Ku complex bound specifically to DNA ends in vitro, and binding was not reduced in the presence of excess competitor closed circular plasmid DNA. Ku binding may protect DNA ends from exonuclease attack and consequent production of large deletions at DSBs. Ku protein may also play an important role in the juxtaposition of DNA ends, facilitating re-ligation. S. cerevisiae and mammalian Ku mutants show reduced NHEJ activity, and residual DNA repair is more error-prone, including the appearance of large deletions, supporting the view that Ku80 protects DNA ends from exonuclease activity (Boulton and Jackson, 1996a; Boulton and Jackson, 1996b; Liang and Jasin, 1996). However, processing of DNA ends by DNA polymerases and nucleases prior to ligation can occur in the presence of Ku and often exposes short regions (1–4 nucleotides) of microhomology shared between the two DNA ends (Lieber, 1999). The RAD50/MRE11/NBS1 (Xrs2 in S. cerevisiae) complex, which has exonuclease and helicase activity, is involved in processing DNA ends in both HR and NHEJ (Haber, 1998; Lewis and Resnick, 2000). The molecular structure of Ku bound to DNA ends suggests that approximately 12 nucleotides are exposed between two Ku dimers bridging a DSB, and this may facilitate DNA end processing (Walker et al., 2001). Additionally, yeast, mammalian and Arabidopsis Ku heterodimers have themselves been reported to have ATP-dependent helicase activity (Dynan and Yoo, 1998; Tamura et al., 2002), although mutation of the predicted ATPase domain had no effect on the function of Ku80 in mammalian cells (Singleton et al., 1997).

Arabidopsis Ku70 bound to DNA ends in the absence of AtKu80 in the gel mobility shift assay, indicating that AtKu80 is not required for DNA end-binding activity in vitro. This observation contrasts with that of Tamura et al. (2002), where no interaction of AtKu70 alone with blunt-ended dsDNA oligonucleotides was detected. This may reflect a lower affinity of AtKu70 for blunt-ended DNA substrates than for substrates with overhang as used in the present study. Notably, repair of SmaI-generated blunt-ended plasmids in yeast occurred by a Ku-independent pathway (Boulton and Jackson, 1996b). However, AtKu80 may still be required in vivo, possibly involved in stabilizing AtKu70 end-binding, recruitment of other components of the NHEJ pathway to the repair complex, or aligning DNA ends through protein–protein interactions (Nick-McElhinny et al., 2000). Mammalian Ku70 was found to bind DNA ends in the absence of Ku80 in immunoprecipitation experiments but not in gel-shift assays (Wang et al., 1998).

In eukaryotes with larger genomes, HR is usually rare, and NHEJ is the predominant pathway for DSB repair in the nucleus. However, at certain points in the cell cycle or in certain cell types, including the chicken DT40 cell line and embryonic mouse cells, the ratio of HR to NHEJ is much higher, indicating that the relative activity of the two pathways can be regulated. Human Rad52 is required for homologous recombination and binds DNA ends, promoting end-to-end interactions analogous to the action of Ku (Goedecke et al., 1999). The relative abundance in the cell of Ku70 and RAD52 proteins, which are involved in the first steps of NHEJ and HR, respectively, may determine the repair pathway used (Goedecke et al., 1999; Van Dyck et al., 1999). Consistent with this hypothesis, NHEJ-mutant mammalian cells show up-regulation of HR-mediated DSB repair, with the greatest increase in HR found in the Ku70 mutant (Pierce et al., 2001). The ratio of involvement of NHEJ to HR in the integration of homologous sequences into the plant nuclear genome is approximately 1:10−4, and represents a major obstacle to the development of gene targeting and gene replacement technologies in plants (Mengiste and Paszkowski, 1999). T-DNA insertion into host DNA during plant cell transformation utilizes the host cell's DNA DSB repair pathways for integration at one or both T-DNA ends (Puchta, 1998; Salomon and Puchta, 1998). Only with a detailed understanding of the pathways involved in transgene integration will we be in a position to manipulate the fate of introduced DNA in favour of homologous recombination.

The T-DNA insertion identified by PCR screening of the Feldmann lines of transformed Arabidopsis was in the ninth exon of atku80, and, if expressed, the resulting transcript could encode 490 of the 680 amino acids of the wild-type protein (Figure 4). The insertion site corresponds to a region of Arabidopsis Ku80 that aligns to the heterodimerization domain of human Ku80 (indicated by a solid bar in Figure 4). By comparison with human Ku80, the truncated Arabidopsis Ku80 protein would lack the SH3 (secondary homology domain 3) required for dimerization with AtKu70, and would therefore be unlikely to function normally in vivo. Analysis of mammalian Ku70 deletion mutants shows that both heterodimerization and DNA end-binding activities are required for non-homologous end-joining activity (Jin and Weaver, 1997). Analysis of the homozygous Arabidopsis ku80 mutants showed no obvious phenotypic differences under normal growth conditions. However, the atku80 mutant showed hypersensitivity to the DNA-damaging agents bleomycin and menadione. Bleomycin causes single- and double-strand breaks in DNA, and hypersensitivity of the atku80 mutant to bleomycin is consistent with AtKu80 having a role in the Arabidopsis NHEJ pathway. Menadione causes oxidative damage by generation of H2O2 and superoxide, agents which cause strand breaks in DNA, and both single- and double-strand breaks are found in menadione-treated cells (Nutter et al., 1992; Reichheld et al., 1999; Shi et al., 1994). AtKu80 mRNA levels increased threefold after 1 h exposure of Arabidopsis protoplasts to bleomycin (Tamura et al., 2002), although the concentrations of bleomycin (50 µg ml−1) used to elicit this response in protoplasts were 10-fold higher than the dose required for lethality in Arabidopsis plants in our study (Figure 5). The hypersensitivity of atku80 plants to menadione and bleomycin could be complemented by transformation of the mutant plants with the wild-type AtKu80 gene (Figure 6). Complementation of DNA damage hypersensitivity by AtKu80 is the first demonstration that a component of the conserved Arabidopsis NHEJ pathway is required for DNA DSB repair in the intact plant. Further work will investigate whether the ratio of NHEJ to HR and the frequency of gene targeting is altered in the atku80 mutant. In addition, Ku80 has a role in telomere maintenance in yeast and mammals (Bailey et al., 1999; Boulton and Jackson, 1996b), and it will be of interest to determine whether telomere stability is affected in atku80 mutant Arabidopsis plants.

Experimental procedures

Plant material and growth conditions

Arabidopsis (genotype WS) seeds were sown directly onto damp Levington's M2 compost. Light-grown plants were raised in an environmental growth chamber (SGC970/C/HQI, Sanyo-Gallenkamp, Loughborough, UK) under controlled conditions of constant humidity (70%), with a 16 h light/8 h dark cycle at 20°C. Visible light was provided by a combination of high-intensity discharge lamps (Osram, Light Source Supplies, Bishops Stortford, UK) and tungsten lamps. The quantum flux density was measured daily 2 h into the photoperiod and was equivalent to 100 µmol m−2 sec−1 photosynthetically active radiation (PAR) for light-grown tissue.

Molecular cloning

DNA procedures and bacterial manipulations were performed using established protocols (Sambrook et al., 1989) unless otherwise stated. Plasmid DNA was prepared on an analytical scale using Qiagen columns according to the manufacturer's instructions (Qiagen, Crawley, UK).

AtKu70 and AtKu80 were cloned by RT–PCR using AMV reverse transcriptase (Roche, Lewes, UK) using either Taq DNA polymerase or the Expand PCR kit (Roche). RACE–PCR was performed using the 3′/5′ RACE–PCR kit (Roche) according to the manufacturer's instructions. PCR products were cloned using a TOPO-TA cloning kit (Invitrogen, Breda, The Netherlands) and plasmid DNA prepared using spin columns (Qiagen) prior to fluorescent dye-terminator DNA sequencing (Perkin-Elmer, Warrington, UK), and analysed on an Applied Biosystems automated sequencer. Oligonucleotide primers were purchased from Amersham. The E. coli strain TOPO10F′ (Invitrogen, genotype F′ {lacIqTn10(TetR)} mcrA Δ(mrr-hsdRMS-mcrBC)Φ80lacZΔM15ΔlacX74recA1 araD139 Δ(ara-leu)7697 galU galK rpsL (StrR) endA1 nupG) was used for routine cloning.

Over-expression of AtKu70 and AtKu80

Full-length AtKu70 and AtKu80 cDNA were cloned into pET30EkLIC (Novagen, Nottingham, UK), generating both N- and C-terminal 6×His-tagged proteins according to the manufacturer's instructions. Expression of AtKu70 and AtKu80 was induced for 3 h in E. coli strain BL21(DE3)pLysS (Promega, Southampton, UK) by the addition of isopropylthiogalactoside (1 mm). Bacteria were recovered by centrifugation at 6000 g for 10 min, resuspended in RS buffer (50 mm Tris–Cl pH 7.5, 50 mm NaCl, 10 mm imidazole, 5% v/v glycerol, 0.1% v/v Triton-X100) and lysed by freeze-thawing and sonication. The extract was cleared by centrifugation at 13 000 g for 10 min. His-tagged Ku80 was applied to a 5 ml nickel-chelating Sepharose fast-flow affinity chromatography column (Pharmacia) and washed with RS buffer containing 100 mm imidazole and eluted in RS buffer containing 500 mm imidazole. His-tagged protein was detected using anti-6×His antibody (Sigma) followed by alkaline phosphatase-coupled secondary antibody and colour development according to the manufacturer's instructions. Fractions containing the purified protein were desalted into 50 mm Tris–Cl pH 7.5, 0.05% v/v Triton-X100, 5% v/v glycerol using a Sephadex G-25 PD10 column (Pharmacia) and stored at −80°C. Preparation of the Ku dimer was performed by mixing equimolar amounts of AtKu70 and AtKu80 in GMSA buffer (20 mm Tris–Cl pH 7.5, 150 mm KCl, 2 mm dithiothreitol, 5% v/v glycerol, 0.1% v/v Triton-X100, 100 µg ml−1 acetylated BSA) and incubation at 25°C for 10 min.

Gel mobility shift assay (GMSA)

The gel mobility shift assay was performed as described previously (Taylor et al., 2000) with minor modifications. The oligonucleotide substrates were based on previously published sequences (Lim et al., 2001) as follows: LA3, 5′-GGAGCTTT ACTGGGTCATCACCGAGACCTCTTATGTCCATAGCTTCCCTGTTTGCCCATTGCTTTACCCTC; LA4, 5′-GAGGGTAAAGCAATGGGCAA ACAGGGAAGCTATGGACATAAGAGGTCTCGGTGATGACCCAGTAAAG. Oligonucleotide LA3 was phosphorylated with T4 polynucleotide kinase (Roche) in the presence of [γ-32P]ATP (> 4000 Ci mmol−1, ICN, Cosa Mesa, USA), and unincorporated nucleotides were removed by Sephadex-G25 gel filtration (Oligospin columns, Roche). 32P-labelled LA3 (10 pmol) was incubated with LA4 (10 pmol) at 85°C for 10 min and allowed to cool slowly to room temperature to promote annealing. Duplex oligonucleotide was diluted to a final concentration of 100 fmol µl−1 and stored at −20°C. Radiolabelled oligonucleotide (100 fmol) was incubated at room temperature for 20 min with purified recombinant Arabidopsis Ku70 and Ku80 (as indicated in Results) in GMSA buffer to give a final volume of 15 µl. Samples were fractionated through a 5% non-denaturing polyacrylamide gel at 150 V in 1 × TBE (BioRad), and dried gels were analysed using a BAS1800 Phosphorimager (Fuji Photo Film Co., Tokyo, Japan).

Isolation of total RNA

RNA was isolated using either the SV total RNA isolation kit (Promega) or as described previously (West et al., 2000). Fresh tissue (500 mg) was ground in liquid nitrogen and added to 4 ml extraction buffer (100 mm Tris–HCl pH 8.5, 100 mm NaCl, 20 mm EDTA, 1% w/v sodium lauryl sarcosinate, 1.7% w/v diethyldithiocarbamate) and vortexed. The solution was extracted once with water-saturated phenol (Biogene, Kimbolton, Cambridgeshire, UK), and twice with chloroform. Nucleic acids were then precipitated and resuspended in RNAase-free water. LiCl was added to a final concentration of 2 m, and RNA precipitated for 4 h on ice. RNA was recovered by centrifugation (10 000 g for 10 min) and washed with ice-cold RNAase-free water.

Isolation of genomic DNA and Southern analysis

Arabidopsis tissue was harvested and ground to a fine powder in liquid nitrogen. DNA was isolated using the method of Doyle and Doyle (1990), and Southern analysis performed as described previously (Taylor et al., 1998).

Co-immunoprecipitation

AtKu70 and AtKu80 were amplified by PCR using primers that incorporated a T7 RNA polymerase promoter for both genes. In addition, regions encoding N-terminal affinity tags were also incorporated into the primers such that AtKu70 was expressed with a c-myc tag and AtKu80 with a hemagglutinin (HA) tag. PCR products were transcribed and translated in vitro in the presence of [35S]-methionine (Promega), and co-immunoprecipitation was carried out using the Matchmaker Co-IP kit (Stratagene) according to the manufacturer's instructions. Precipitated proteins were analysed by SDS−PAGE and autoradiography.

Column chromatography

Purified, His-tagged AtKu80 was incubated with 100 µl nickel-chelating Sepharose fast-flow affinity resin (Pharmacia) and washed with wash buffer (50 mm Tris–Cl, 150 mm KCl, 0.1% v/v Triton-X100, pH 7.5). In vitro transcription translation mixture (Promega) containing [35S]-methionine-labelled AtKu70 was applied to the column-bound Ku80, and the column was washed with wash buffer containing NaCl (0.2–1 m as indicated). Finally, nickel-bound 6×His-tagged AtKu80 was eluted by addition of 1 m imidazole. Aliquots of washes were analysed by SDS−PAGE and [35S]-labelled protein was detected by autoradiography.

Isolation and complementation of atku80 mutant

Pooled DNA from the Feldmann lines of transformed Arabidopsis plants was screened by PCR using the T-DNA left border primer LB (5′-GATTCTTTTTATGCATAGATGCAC) and the AtKu80-specific primer K6 (5′-CTCCAAGACGCAGCCTTTAC) (McKinney et al., 1995). Expression of AtKu80 in the homozygous mutant plant was analysed by RT–PCR. Primers were selected to amplify the cDNA 5′ to the T-DNA insertion (Ku8, CATTGGAGGTAAAG TAGCATC; Ku5, GCTTTCGCTATGGACCTCAG), across the T-DNA insertion site (primers Ku5 and K6) and 3′ to the T-DNA insertion (K1, 5′-GGCTGCGTCTTGGAGCAGGTCTC; Ku2, 5′-GCTCTCGAGC ATTGACTC). AtMCM1 was used as control gene and amplified using primers MCM1 (5′-GGAGACATCAACATGATG) and MCM2 (5′-GTCGATACCAGCATCCAT). For complementation of atku80, the 5.8 kb genomic region including wild-type AtKu80 was amplified from WS DNA by PCR using primers containing 5′SalI sites (Ku80genF, 5′-TGAATCACTGTTGTCGACCAACTGGT ATGATCTATGTTGTACTCAC; Ku80genR, 5′-TGAATCACTGTTG TCGACCAACTGGTATGATCTATGTTGTACTCAC). The PCR product was cloned into the binary vector pCB1300 and used to transform atku80 plants with Agrobacterium tumefaciens strain GV3100 as described previously (Clough and Bent, 1998).

Acknowledgements

We wish to acknowledge the ABRC and Kenneth Feldmann for provision of the transformed A. thaliana, and Patrick Gallois and Simon Turner for advice and discussion. The financial support of the UK BBSRC/NERC Gene Flow in Plants and Micro-organisms Initiative Grant (grant number 34/G114142) and the European Community (grant number FAIR5-PL97-37311) is gratefully acknowledged.

Ancillary