Chromosomal double‐strand DNA breaks must be repaired; in the absence of repair the resulting acentromeric (and telomereless) fragments may be lost and/or the broken DNA ends may recombine causing general chromosomal instability. The Rad50/Mre11/Xrs2 protein complex acts at DNA ends and is implicated in both homologous and non‐homologous recombination. We have isolated a rad50 mutant of the plant Arabidopsis thaliana and show here that it has a somatic hyper‐recombination phenotype in planta. This finding supports the hypothesis of a competition between homologous and illegitimate recombination in higher eukaryotes. To our knowledge, this is the first direct in vivo support for the role of this complex in chromosomal recombination in a multicellular organism and the first description of a mutation of a known gene leading to hyper‐recombination in plants.
Understanding of the mechanisms used by eukaryotic cells to repair DNA double‐strand breaks (DSBs) has greatly advanced in recent years. One of the most notable aspects of this is the demonstration of the involvement of two general classes of recombination processes, those involving DNA sequence homology between the two partners (homologous recombination, HR) and those independent of such homology (non‐homologous recombination or non‐homologous end joining, NHEJ). Specific protein factors have been identified as being involved in one or other of these processes (reviewed by Jeggo, 1998; Tsukamoto and Ikeda, 1998; Paques and Haber, 1999).
The preferred mechanism of recombinational DNA repair differs between organisms. NHEJ is much more frequent than HR in mammalian and plant cells, whereas yeast cells rely almost entirely on homology‐based recombinational DNA repair.
Mutants showing defects in DSB repair and recombination fall mostly into two general classes: those involved primarily in HR (rad51/52/54/55/57/59) or NHEJ (ku, DNA‐PK, lig4, Xrcc4) (reviewed by Critchlow and Jackson, 1998; Paques and Haber, 1999); however, one protein complex, Rad50/Mre11/Xrs2, has been implicated in both HR and NHEJ pathways. Yeast mutants lacking this complex are X‐ray sensitive and show a weak hyper‐HR and strong hypo‐NHEJ phenotype in mitotic yeast cells (Malone and Esposito, 1981; Gottlieb et al., 1989; Malone et al., 1990; Schiestl et al., 1994; Milne et al., 1996; Moore and Haber, 1996; Tsukamoto et al., 1996, 1997; Lewis et al., 1999). In vivo studies of the roles of this complex come essentially from yeast as animal cells carrying knockouts of the RAD50/MRE11 genes are not viable (Xiao and Weaver, 1997; Luo et al., 1999; Yamaguchi‐Iwai et al., 1999).
In recent years, considerable advances have been made in the understanding of recombination processes in plants, pointing to a conservation of recombination and DSB repair mechanisms between plants and animals (reviewed in Britt, 1999; Gorbunova and Levy, 1999; Mengiste and Paszkowski, 1999; Vergunst and Hooykaas, 1999).
We have recently isolated and characterized the Arabidopsis thaliana RAD50 homologue (DDBJ/EMBL/GenBank accession No. AF168748) and shown that rad50 mutant plants are viable, but are sterile and hypersensitive to the radio‐mimetic agent methylmethane sulfonate (Gallego et al., 2001). The fact that this plant rad50 mutant is viable has permitted us to examine in vivo the chromosomal recombination processes of a multicellular organism in the absence of a member of this crucial complex.
Results and Discussion
We have tested the hypothesis that, in plant cells where DSBs are predominantly repaired by NHEJ, inhibition of this pathway will result in stimulation of HR. The frequency of HR was measured using an in planta recombination assay (Swoboda et al., 1994), with a reporter transgene consisting of two overlapping fragments of the GUS (uidA) gene separated by a hygromycin selectable marker. The GUS sequences have an overlap of 618 bp and recombination between the two overlapping DNA sequences can produce a functional GUS gene (Figure 1). Sectors of cells expressing GUS activity are detectable by their blue colour after histochemical staining. Heterozygous rad50 mutant plants and GUS recombination reporter plants were crossed, and the frequency of recombination events was monitored in F3 progeny plants homozygous for the recombination reporter and homo‐ or heterozygous for the rad50 mutation (see Methods). PCR analysis was used to confirm the rad50 genotype of individual plants and examples of this PCR verification are presented in Figure 2.
Mutant rad50/rad50 and heterozygote rad50/RAD50 seedlings, each homozygous for the GUS recombination substrate in either inverted or direct repeat configuration, were stained for GUS activity. Numbers of recombination events per plant were determined visually under a dissecting microscope as blue spots and sectors. Previous work has confirmed that these blue‐staining spots result from recombination (Swoboda et al., 1994). In two experiments, a total of 112 heterozygote (Rad50+) and 120 mutant (rad50−) plants were tested for the direct repeat recombination. Figure 3 shows examples of stained mutant (A–C) and heterozygote (D–F) plants. A single population with 59 heterozygote and 55 mutant plants was tested for inverted repeat recombination. Proportions of plants with a given number of spots per plant are shown in Figure 4 and the data are summarized in Table 1. For both the direct and inverted repeat populations it can be clearly seen that the homozygote rad50 mutant plants have considerably more recombination spots than the heterozygote controls. This increase is 9‐ to 10‐fold in the direct repeat recombination and 8‐fold in the inverted repeat recombination. Statistical analysis of these results confirms in all cases that the heterozygote and mutant plants belong to different populations with respect to the number of recombination events per plant (Table 1; Methods).
Direct repeat structures may recombine productively via gene conversion/reciprocal exchange (GC) or single‐strand annealing (SSA) mechanisms, whilst productive recombination events in the inverted repeats occur through GC. The rad50− mutant plants show an 8‐ to 10‐fold stimulation of recombination in both the direct and inverted repeat tester lines. Thus, the detected increase in recombination in the rad50 mutant plants is presumably due to an increase in GC. This mitotic hyper‐recombination phenotype is reminiscent of that seen in yeast rad50 mutants, which show 5‐ to 10‐fold increased spontaneous mitotic recombination (Malone and Esposito, 1981; Gottlieb et al., 1989; Malone et al., 1990). The inviability of mammalian rad50 and mre11 mutants has precluded such testing of recombination in animal cells (Xiao and Weaver, 1997; Luo et al., 1999), with one exception: using a conditional mre11 mutant, Yamaguchi‐Iwai et al. (1999) have shown that gene targeting is reduced in the absence of Mre11 protein in a conditional mre11−, hyper‐recombinational chicken DT‐40 cell line. Repression of MRE11 expression in these cells leads to an accumulation of chromosomal DSBs, accumulation at the G2/M phase and cell death in 4–5 days. The gene targeting assay was carried out 3 days after MRE11 repression and MRE11 expression re‐induced 12 h later. This assay measures the proportion of stably transformed cells that have integrated the transforming DNA at the homologous chromosomal locus, versus those in which the DNA had integrated elsewhere. Given the presence of significant numbers of random chromosomal DSBs in these dying chicken cells and the highly recombinogenic nature of chromosomal DSBs, we believe it is difficult to make a direct comparison between our chromosomal recombination assay and the gene targeting assay of Yamaguchi‐Iwai et al. (1999). We are currently undertaking work to test directly whether the mitotic hyper‐recombinational phenotype described here for the Arabidopsis rad50 mutant is accompanied by elevated frequencies of gene targeting in these plants.
VanDyck et al. (1999) have recently presented a model for the channelling of recombination into either HR or NHEJ pathways via the competition between the Rad52 and Ku proteins, respectively, for a common substrate (see Haber, 1999). This substrate results from the Rad50/Mre11/Xrs2‐dependent resection of the DSB to produce single‐stranded extensions. Yeast rad50 mutants show increased mitotic recombination (Malone and Esposito, 1981; Gottlieb et al., 1989; Malone et al., 1990) and are deficient in NHEJ (Schiestl et al., 1994; Moore and Haber, 1996; Tsukamoto et al., 1996). In this context, the hypo‐NHEJ and hyper‐HR phenotype of these mutants implies that, at least in yeast, the Ku‐dependent NHEJ pathway has greater dependence on the prior (or concomitant) action of the Rad50/Mre11/Xrs2 complex than the Rad52‐dependent HR pathway. The resulting reduction in the efficiency of NHEJ in the absence of Rad50 would make more substrate available for HR. Our description of a hyper‐recombination phenotype in the Arabidopsis rad50 mutant raises the attractive possibility that this situation also exists in plants and possibly other multicellular organisms.
Recombination tester lines were produced by Agrobacterium‐mediated transformation (Clough and Bent, 1998) of A. thaliana ecotype ‘Columbia’ plants, and two lines homozygous for a single chromosomal copy of the reporter construct were selected (data not shown). The recombination substrates (Tinland et al., 1994) consist of two fragments of the β‐glucuronidase (GUS) gene with 618 bp of overlap, either in direct (pGU.US; line 1406) or inverted orientation (pU′G′.US; line 1415) (Figure 1).
The two GUS tester lines were crossed to the heterozygous rad50/RAD50 Arabidopsis mutant. The mutant rad50 allele in this line results from a T‐DNA insertion into the RAD50 gene and has been described elsewhere (Gallego et al., 2001). Resulting F1 progeny, heterozygous for the rad50 mutant allele (kanamycin resistant) and hemizygous for the GUS recombination tester substrate (hygromycin resistant), were allowed to self‐fertilize to produce F2 seed. Individual F2 plants were allowed to self‐fertilize to produce F3 seed and samples of this seed were screened for the segregation of the kanamycin and hygromycin resistance markers to identify the F2 plants homozygous for the GUS recombination substrate (homozygous for the hygromycin resistance marker) and heterozygous for the mutant rad50 allele (heterozygous for the kanamycin resistance marker).
F3 seed from plants homozygous for the GUS recombination substrate and heterozygous/homozygous for the mutant rad50 allele was allowed to germinate on agar plates containing 50 mg/l kanamycin (the mutant rad50 allele is marked by kanamycin resistance). Kanamycin‐resistant seedlings (RAD50/rad50 and rad50/rad50) were stained for GUS activity at 3–4 weeks of age. The GUS histochemical staining was carried out as described by Jefferson (1987) and Swoboda et al. (1994). The number of blue spots or sectors, each resulting from a recombination event, on each seedling was then determined visually under a dissecting microscope. For each line, ∼50 rad50/RAD50 heterozygotes and 50 rad50/rad50 homozygotes were tested (see Table 1). Initial visual identification of the rad50/rad50 homozygote seedlings at this stage was based on the fact that they are slightly smaller than the heterozygotes and their leaves are rougher. The RAD50 genotype of the individual plants was then confirmed using PCR as described below.
PCR analyses were used to determine the RAD50 genotype of individual plants. Genomic DNA was extracted from a single seedling leaf according to Berthold et al. (1993). PCR and long PCR conditions have been described previously (Gallego et al., 2001). Two primer pairs were used: o96 (5′‐GAGCTGTGAAGCTAGAAAGAATGAACTTGCAGGTG) and o115 (5′‐CCCATCCAGGTTTGTAGTTG), which span the T‐DNA insertion in the mutant rad50 allele and amplify 1.6 or 8 kb fragments from the wild‐type or mutant alleles, respectively; and o96 and a primer in the T‐DNA sequence (o127: 5′‐CTGATACCAGACGTTGCCCGCATAA), which amplify a 500 bp fragment from the mutant rad50 allele. Positions of the primers on the mutant rad50 locus are shown schematically in Figure 2C.
The data on the number of recombination events per plant were analysed non‐parametrically (with the kind advice of Dr E.J. Louis, University of Leicester, UK). All plants analysed from a given line were ranked by the number of blue spots for each set of plants analysed. On the null hypothesis that the plants analysed from each line belong to the same distribution, half should be below the mid‐point of the ranking and half above. χ2 (1 degree of freedom) values were calculated for the heterozygote and homozygotes from each line (see Table 1).
We thank the members of the plant recombination laboratory and BIOMOVE for their helpful discussions and criticisms. The work reported here was partially funded by the European Community research grant BIO4 CT97 2028.
- Copyright © 2001 European Molecular Biology Organization