Chromosomal breaks are repaired by homologous recombination (HR) or non-homologous end joining (NHEJ) mechanisms. The Ku70/Ku80 heterodimer binds DNA ends and plays roles in NHEJ and telomere maintenance in organisms ranging from yeast to humans. We have previously identified a ku80 mutant of the model plant Arabidopsis thaliana and shown the role of Ku80 in telomere homeostasis in plant cells. We show here that this mutant is hypersensitive to the DNA-damaging agent methyl methane sulphonate and has a reduced capacity to carry out NHEJ recombination. To understand the interplay between HR and NHEJ in plants, we measured HR in the absence of Ku80. We find that the frequency of intrachromosomal HR is not affected by the absence of Ku80. Previous work has clearly implicated the Ku heterodimer in Agrobacterium-mediated T-DNA transformation of yeast. Surprisingly, ku80 mutant plants show no defect in the efficiency of T-DNA transformation of plants with Agrobacterium, showing that an alternative pathway must exist in plants.
DNA double-strand breaks (DSBs) represent one of the most dangerous lesions for the cell, potentially affecting genomic integrity; thus, all organisms have highly efficient systems to repair these lesions. Two main pathways have been identified, conserved from yeast to humans: homologous recombination (HR), which involves interaction between homologous sequences; and non-homologous end joining (NHEJ), which acts independently of DNA sequence homology. The choice between these two repair pathways differs between organisms. Yeast cells rely almost entirely on HR, while animal and plant cells preferentially use NHEJ. Proteins involved in these two processes have been identified in yeast and mammals and shown to present a high degree of conservation (see reviews (Britt, 1999; van Gent et al., 2001; Gorbunova and Levy, 1999; Jackson, 2002; Khanna and Jackson, 2001; Mengiste and Paszkowski, 1999; Paques and Haber, 1999; Ray and Langer, 2002; Tuteja et al., 2001; Vergunst and Hooykaas, 1999)). Here, we are particularly concerned with NHEJ. Proteins playing a role in NHEJ include DNA-dependent protein kinase (DNA-PK), Ku70, Ku80, Lig4 and Xrcc4. In mammalian cells, Ku70 and Ku80 have been shown to form a trimeric complex with the catalytic subunit of the DNA-PKc. This last protein has not been identified in yeast or plant cells; however, the Ku70 and Ku80 proteins form a complex in these organisms (Feldmann and Winnacker, 1993; Feldmann et al., 1996; Tamura et al., 2002; West et al., 2002).
Genetic and biochemical evidence support the hypothesis that NHEJ and HR are two independent and competing mechanisms for DSB repair (Essers et al., 2000; Kooistra et al., 1999; Liang et al., 1998; Takata et al., 1998). In vitro studies have demonstrated that the Ku and Rad52 proteins bind to DNA ends, protecting them from exonuclease attack and facilitating their interaction. Based on this observation, it has been proposed that the choice between NHEJ and HR for DSB repair will be defined by which one of these proteins associate to the DNA ends (VanDyck et al., 1999). Interestingly, the RAD52 gene does not appear to be present in plants, nematodes and insects (Iyer et al., 2002).
Illegitimate recombination is the preferred DSB repair pathway in plants and, together with the repair of DSBs, is implicated in integration of T-DNA and transposons in the plant genome (see reviews (Gelvin, 2000; Gorbunova and Levy, 1999; Zupan et al., 2000)). The T-DNA molecule and its integration site in the genome do not share extensive homology. It has been shown that T-DNA can be transferred from Agrobacterium tumefaciens to Saccharomyces cerevisiae, and this model system has been used to examine the involvement of the recipient cell's recombination machinery in T-DNA integration. T-DNA containing homology to the S. cerevisiae genome integrates via HR (Bundock et al., 1995). In contrast, T-DNA lacking homology with the yeast genome integrates at random via illegitimate recombination (Bundock and Hooykaas, 1996). Furthermore, it has been recently shown that T-DNA integration in yeast via illegitimate recombination requires the NHEJ yeast proteins, including the yeast Ku70 homologue (van Attikum et al., 2001). These results suggest that the process of T-DNA integration is defined by the host factors rather than by the T-DNA itself.
We have cloned the Arabidopsis KU80 homologue and characterised a T-DNA insertion ku80 mutant plant. These plants present longer telomeres, demonstrating the role of the Ku80 protein in telomere homeostasis in plant cells (Gallego et al., 2003). The direct implication of the Arabidopsis Ku80 protein in NHEJ, HR or T-DNA recombination remains to be demonstrated. Here, we show that ku80 mutant plants are hypersensitive to methyl methane sulphonate (MMS) and Bleomycin treatment, as previously shown for another ku80 mutant allele and ku70 (Bundock et al., 2002; Riha et al., 2002; West et al., 2002). We show that ku80 mutant cells are deficient in end joining of a linear plasmid DNA, demonstrating the role of the Ku80 protein in NHEJ in plant cells. No change in the rate of intrachromosomal spontaneous HR was observed in ku80 mutant cells as compared to the wild type. Most surprisingly, we found no defect in T-DNA integration in the absence of the Ku80 protein. This result suggests the existence of an alternative mechanism to end joining for T-DNA integration in plant cells.
The Arabidopsis ku80 mutant is hypersensitive to DNA damage
In order to study the role of the Ku80 protein in DNA repair, we compared the sensitivity of wild-type and ku80 mutant plants to MMS. Ten-day-old seedlings were transferred to liquid medium containing MMS. After a week of further growth, ku80 plants show hypersensitivity to MMS relative to wild-type controls at concentrations of 75 and 100 p.p.m. MMS (Figure 1). The Ku80 protein is thus implicated in the repair of MMS-induced DNA damage in Arabidopsis plants. This sensitivity is comparable to that reported by Bundock et al. (2002) for the Arabidopsis ku70 mutant. Studying an independent Arabidopsis ku70 allele, Riha et al. (2002) found MMS sensitivity only for germination, not for the treatment of seedlings. They did, however, find that their ku70 mutant plants were hypersensitive to gamma-irradiation. We have also found that the ku80 mutant plants are hypersensitive to Bleomycin treatment (not shown), in agreement with the data provided by West et al. (2002). We thus conclude that our results confirm a role for the Ku70/Ku80 heterodimer in cellular resistance to DNA breakage in Arabidopsis.
Role of Ku80 protein in NHEJ
To directly test the involvement of the Arabidopsis Ku80 protein in NHEJ recombination, we used an in vivo plasmid re-joining assay that measures the capacity of the plant cells to repair restriction-enzyme-generated DSBs. Similar assays have been used previously in protoplasts and leaves of tobacco (Gorbunova and Levy, 1997; Gorbunova et al., 2000). The plasmid DNA we used contained the green fluorescent protein (GFP) gene under the control of the CaMV 35S promoter (see Experimental procedures). Linear plasmid molecules bearing blunt-ended or cohesive 5′ overhang termini were generated after digestion by the restriction endonucleases EcoRV and BamH1, respectively. EcoRV and BamH1 cut the plasmid DNA between the promoter and the coding sequence of the GFP gene at 116 and 6 bp upstream of the ATG beginning the gfp open-reading frame (ORF), respectively. Protoplasts derived from wild-type and ku80 suspension culture plant cells were transformed with circular or linearised plasmid DNA. Re-circularisation of the linearised plasmid by the plant cells will regenerate an active GFP gene, which is detected by the GFP fluorescence. The number of protoplasts presenting GFP activity after linear plasmid transformation normalised to the number obtained with the circular plasmid reflects the capacity of Arabidopsis cells to mediate repair of the enzyme-generated DSB. In different experiments, the control, uncut circular plasmid DNA gave between 12 and 33% as many gfp+ transformants in the ku80 as in the wild-type protoplasts. The ku80 cells thus appear to be less competent for transformation under these conditions; however, given that variations in transformability of comparable magnitude are commonly seen between different lots of protoplasts, we do not believe any firm conclusion concerning this possibility at this time. Notwithstanding these variations, the relative transformability of the linearised versus circular plasmid DNA remains remarkably constant between experiments, as shown in Figure 2 and Table 1.
Table 1. End joining of linear plasmid DNA
Relative plasmid end-joining efficiency of WT and ku80 mutant cells. Values are expressed as per cent of transformation with respect to circular plasmid DNA. ku80 mutant cells re-join linear plasmid DNA less efficiently than the wild type.
The results in Table 1 and Figure 2 show that the number of protoplasts presenting GFP activity is 2.3–5-fold lower in the absence of the Ku80 protein. Similar results were obtained whether the ends of the DNA molecule to be ligated were blunt-ended or 5′ overhangs. Furthermore, the fact that the BamHI-linearised plasmid (cut 6 bp upstream of the gfp ORF) transforms both wild-type and ku80 mutant cells, with only slightly lesser efficiency than that of the EcoRV-linearised DNA (cut 116 bp upstream of the gfp ORF), indicates that there is little degradation of the linear DNA. These results show that the Arabidopsis Ku80 protein is involved in the repair of DSBs in the absence of homology in Arabidopsis cells.
Role of the Ku80 protein in intrachromosomal recombination
In order to test for an implication of Ku80 in HR in Arabidopsis, mutant ku80 plants were crossed with plants containing the β-glucuronidase (GUS) recombination reporter transgene in either direct- or inverted-repeat configuration. F3 plants were selected as being homozygous for the recombination reporter gene and wild-type or homozygous for the ku80 mutant locus. Recombination frequency was measured as the number of cell sectors expressing GUS activity, defined as the number of blue spots per plant after histochemical staining.
Seeds from two KU80 wild-type plants and two ku80 mutant plants homozygous for the direct- or inverted-repeat configurations of the GUS substrate were sown to measure recombination activity. Blue spots were counted in more than 100 descendant plants of each. Results are presented in Table 2, and the proportions of plants with a given number of blue spots are presented as frequency histograms in Figure 3. A slight stimulation of recombination in ku80 mutant plants of approximately 1.5-fold was seen with both direct- and inverted-repeat tester loci. This difference is, however, not statistically significant except for one pair of the direct-repeat lines (Table 2). We thus conclude that absence of the Ku80 protein causes no significant stimulation of HR between tandem repeats in either direct or inverted orientations.
Table 2. Chromosomal tandem-repeat recombination
Somatic recombination with direct- and inverted-repeat loci. Each experiment was repeated once with different lines of each genotype. The total number of plants tested and recombinant spots, and the mean number of spots per plant, are given. Chi-squared (1-d.f.) values for the null hypothesis that all plants from a given experiment (WT and ku80) are from the same population with respect to the number of blue spots. Non-parametric statistical analysis was carried out as described in Experimental procedures. With the exception of lines WT 197 and ku80 196, no significant differences were seen between WT and mutants.
T-DNA transformation in Ku80-deficient Arabidopsis plants
Arabidopsis plants infected with Agrobacterium integrate T-DNA at random in the plant genome via illegitimate recombination. We have shown above that the Ku80 protein is involved in NHEJ in Arabidopsis plants, which makes it a good candidate to play a role in T-DNA integration via illegitimate recombination. We thus tested the efficiency of T-DNA transformation by floral dip transformation of Ku80-deficient plants as compared to wild-type plants. Seeds were recovered from two independent transformations using Agrobacterium containing a binary vector conferring hygromycin resistance to plant cells. The results in Figure 4 show no significant differences in the efficiency of transformation between wild-type and ku80 mutant plants. Although the relative imprecision of this assay would make it difficult to see small (e.g. twofold) differences in transformability, the data clearly show that the ku80 mutant plants are transformable by Agrobacterium T-DNA, and that this assay gives no evidence for any major differences in transformability of ku80 and wild-type Arabidopsis plants. Thus, under these transformation conditions, Ku80 deficiency does not significantly affect the non-homologous integration of the T-DNA into the genome of Arabidopsis. This contrasts strongly with the Ku dependence of yeast T-DNA transformation previously reported for yeast, in which the authors were not able to isolate any T-DNA transformants in the Ku70 mutant (van Attikum et al., 2001).
The results presented here show that Ku80 protein is involved in the repair of DNA DSBs in the absence of homology in Arabidopsis. This conclusion is in agreement with the essential role of the Ku80 protein in NHEJ in S. cerevisiae and Schizosaccharomyces pombe cells (Boulton and Jackson, 1996; Manolis et al., 2001; Milne et al., 1996). Similar to our observation, S. pombe cells lacking Ku80 are deficient for plasmid ligation presenting either cohesive or blunt-ended extremities, while in S. cerevisiae, only ligation of cohesive ends are Ku80 dependent. We also note that the overall efficiency of joining in Ku80-deficient Arabidopsis cells is only 2.3–5 times lower than in wild-type cells, while it is more than 100 times lower in yeast cells, suggesting the presence of an alternative mechanism for homology-independent end joining in Arabidopsis cells. Although care must be exercised in making such quantitative comparisons between different experimental systems/organisms, the situation thus seems more comparable to that seen in mammals than in yeast. The existence of such a mechanism has also been postulated in mammalian cells based on the efficiency of plasmid end joining in the absence of the Ku80 protein (Kabotyanski et al., 1998; Liang and Jasin, 1996). In mammalian cells, Ku80 is not required for efficient and accurate ligation of DNA termini. As for other NHEJ mutants, mammalian ku80 mutants are hypersensitive to DNA clastogens such as ionising radiation (review (Jeggo, 1998)). The Arabidopsis ku80 mutants are hypersensitive to Bleomycin (West et al., 2002; this work) and MMS (this work), at levels similar to that seen for other radiation-sensitive Arabidopsis mutants, including ku70, rad50 and mre11. (Bundock and Hooykaas, 2002; Bundock et al., 2002; Gallego et al., 2001).
Our results indicate that absence of the Ku80 protein causes no significant stimulation of spontaneous HR between tandem repeats in Arabidopsis. These results contrast with the recombination stimulation that we have previously observed in the absence of the Rad50 protein using the same substrates (Gherbi et al., 2001). Using I-SceI endonuclease-induced HR assay, Pierce et al. (2001) found that ku70 mutant mouse embryonic stem (ES) cell lines show a sixfold increase in recombination over wild-type controls; however, differences in neither sister chromatid exchange nor gene targeting frequencies were found. As argued by these authors, it is possible that the explanation for this comes from the nature of the broken end(s) involved (see review (Cromie et al., 2001)). Thus, when a replication fork encounters a nick, termination of replication will create a break with a single end. The Ku heterodimer would be unable to interact with these single-ended DSBs, and so no effect on spontaneous HR is seen in ku mutants. However, after induction of a DSB by the I-SceI protein, a continuous duplex is broken, producing two free ends, whose repair by NHEJ would involve the Ku70/Ku80 proteins. In the absence of Ku70/80, these lesions would be taken up by an alternative HR process. A study of the DNA-PKcs-defective Chinese hamster cell lines found a threefold increase in HR in a similar test substrate (Allen et al., 2002). However, in this case, a twofold increase in spontaneous HR was also seen. Thus, the relative dependence on different recombination pathways depends on both the presence of competing pathways and the nature of the initiating DNA lesions. It will be interesting to test this hypothesis further in Arabidopsis by comparing the efficiencies of spontaneous and DSB-induced HR in the absence of Ku80 in Arabidopsis cells.
The biology of the Agrobacterium T-DNA transfer and transformation of plant cells has been the subject of recent reviews (Gelvin, 2000; Tzfira and Citovsky, 2002; Zupan et al., 2000)). Although considerable knowledge has been gleaned concerning the involvement of bacterial proteins in the T-DNA transformation process, our understanding of the host-cell contribution to T-DNA integration into the plant genome remains superficial. Ziemienowicz et al. (2000) have shown in vitro that a plant cell factor and not the bacterial VirD2 protein is responsible for T-DNA ligation. By selecting Arabidopsis mutants, which are proficient for transient transformation but deficient for stable transformation by T-DNA, the group of S. Gelvin has identified mutants competent for T-DNA uptake and transport but defective in integration. Two of these have been shown to involve a Histone2A gene and a myb-like transcription factor gene (Mysore et al., 2000b; Nam et al., 1997, 1999; Yi et al., 2002). Interestingly, the Histone2A mutant, rat5, defective in transformation of somatic tissue, is competent for stable transformation of flowers, possibly indicating the presence of alternative pathways in the female gametophyte (Mysore et al., 2000a). An early report of the involvement of two Arabidopsis genes, UVH1 and RAD5, in T-DNA integration (Sonti et al., 1995) has since been challenged by other authors (Nam et al., 1998; Preuss et al., 1999).
Analysis of the structure and location of T-DNAs integrated into the plant genome implicates illegitimate recombination (NHEJ) in this process (reviews (Brunaud et al., 2002; Gelvin, 2000; Tzfira et al., 2000). Furthermore, using Agrobacterium transformation of the yeast S. cerevisiae as a model system for studying T-DNA integration, the group of P. Hooykaas has clearly shown the implication of NHEJ proteins, including the Ku70/Ku80 complex in this process in yeast (Bundock and Hooykaas, 1996; Bundock et al., 1995; van Attikum et al., 2001). Here, we demonstrate the implication of the Arabidopsis Ku80 protein in extrachromosomal NHEJ and the surprising finding that the ku80 mutant plants are not defective in floral-dip T-DNA transformation. The 2.3–5-fold reduction in plasmid end joining that we report here implies the existence of an alternative end-joining pathway in plants. Thus, in contrast to yeast, where NHEJ and T-DNA integration are strongly dependent on the Ku70/80 complex, in Arabidopsis NHEJ is only partially dependent and T-DNA integration independent of Ku80. Either the Ku80-dependent pathway does not act in T-DNA integration or an alternative pathway is able to substitute for it. Studies of plasmid transformation of Ku80-deficient mammalian cells detected no deficiency in chromosomal integration of non-homologous plasmid DNA (Liang et al., 1996). Together, these results show that the Ku80 protein is not essential for random DNA integration after transformation of higher eukaryotic cells, in contrast with the observed dependence of yeast cells on the NHEJ machinery for the integration of T-DNA (van Attikum et al., 2001). During the review of this manuscript, data has been published showing that (a different allele) Arabidopsis ku80 mutant transforms 0.37-, 0.39- and 0.15-fold as efficiently as the wild type by floral dip transformation in three experiments (Friesner and Britt, 2003). These authors also show that an Arabidopsis ligase 4 mutant shows 0.5-fold and no reduction in transformability in two experiments. We have similar data showing no significant effect on floral dip transformability of a ligase 4 mutant in Arabidopsis (unpublished). This report thus strengthens the conclusions presented above concerning the absence of a direct role for ku80 in Arabidopsis floral T-DNA transformation or the presence of an equally (or almost) efficient pathway able to substitute for it. Furthermore, as mentioned above, circularisation of extrachromosomal linear plasmid DNA is strongly dependent on the NHEJ pathway in yeast cells, while no effect has been observed in mammalian cells. These observations support the existence of an alternative, Ku80-independent NHEJ recombination pathway in higher eukaryotes, which could also be involved in DNA integration. We are carrying out further work to clearly determine the roles and contributions of the(se) NHEJ pathway(s) to different chromosomal and extrachromosomal DNA recombination and repair events in plant cells.
Plant growth and culture
All Arabidopsis plants and cells used in this work were of ecotype WS. Arabidopsis thaliana seeds were sown directly into damp compost or MS agar medium and under white light (16-h light/8-h dark) as previously described by Gallego et al. (2001). Callus suspension cultures were initiated and maintained as described previously by Gallego et al. (2001). Characterisation of the ku80 mutant plant has been described elsewhere by Gallego et al. (2003). It was isolated from the Versailles INRA collection (Samson et al., 2002). The T-DNA is inserted into exon 10 of the KU80 gene. Should any protein be produced by this mutant ku80 locus, it would lack the carboxy-terminal 125 amino acids. As the case with the Arabidopsis ku80 allele reported by West et al. (2002), the inserted T-DNA in the ku80 allele reported here interrupts the proposed heterodimerisation domain of the Ku80 protein and any truncated protein produced from this locus is unlikely to be functional. Southern blot, PCR and segregation data show that the T-DNA insertion is present at a single locus and plants homozygous for the T-DNA insertion develop normally and are fertile (Gallego et al., 2003). RT-PCR data confirms the absence of detectable native KU80 mRNA in the mutant plants (data not shown).
MMS and Bleomycin sensitivity
Seedlings, 10 days after sowing on MS agar, were transferred to liquid MS medium supplemented with different concentrations of MMS (Sigma #M4016, St Louis, MO, USA) in six-well microtitre plates. The plates were then incubated as described above, and resistance or sensitivity was scored visually 1 week later. MMS-containing plates were prepared immediately before use. All MMS-contaminated material was quenched after use by soaking in 10% (w/v) sodium thiosulphate.
Plasmid end-joining capacity of the ku80 mutant was tested as follows.
Plasmid 35Ω-sGFP(S65T) (Chiu et al., 1996) (a kind gift from Jen Sheen, MGH, Harvard University, Boston, MA, USA) was linearised by cleavage with either BamHI or EcoRV. Fresh protoplasts prepared from ku80−/− or KU80+/+ suspension cell lines were transformed with 30 µg of either the uncut or cut plasmid DNA by the polyethylene glycol (PEG) transformation protocol (Weigel and Glazebrook, 2002). Twenty-four hours after transformation, the protoplasts were counted under visible light and GFP-expressing protoplasts were counted using a flourescein (FITC) filter set (Zeiss #44, Oberkochen, Germany) on a Zeiss Axioplan 2 Imaging microscope. For each experiment, separate aliquots of the same preparation of protoplasts were transformed with no plasmid, uncut, BamHI-cut and EcoRV-cut plasmid DNA. No GFP-positive protoplasts were found in the no-DNA controls. The proportion of GFP-expressing protoplasts was calculated and normalised to the uncut plasmid control transformations in each case. Results from four independent transformation experiments on different days are presented.
Intrachromosomal recombination assay
The in planta somatic recombination assay and the recombination tester lines containing tandem repeats of partial GUS genes in direct or inverted orientation have been previously described by Gherbi et al. (2001). Seeds from wild-type or ku80 mutant lines, homozygous for the recombination tester locus, were sown, 18-day-old seedlings were stained, and GUS-expressing spots were visualised under a dissecting microscope. The number of spots per plant was determined from two individual populations and more than 100 individual plants in each case. Non-parametric analysis of the data was performed as previously described by Gherbi et al. (2001).
Wild-type and ku80 mutant plants were transformed with A. tumefaciens strain C58 carrying the binary vector Bin-HYG-Tx (which carries a pNOS-Hpt II-pAg7 selection cassette; a kind gift from C. Gatz) by the floral dip method (Clough and Bent, 1998). Seeds were subsequently collected, and approximately 2000 seeds per plate (by weight) were germinated under hygromycin (50 mg l−1) selection on MS agar medium in Petri dishes. Two independent transformations were carried out, and hygromycin-resistant transformed seedlings were counted from each of five plates for each transformation.
We thank Dr J. Sheen for her kind gift of the sGFP(S65T) plasmid. This work was partly financed by a European Community Research grant (QLG2-CT-2001-01397).