Using a specific recombination assay, we show in the plant Arabidopsis thaliana that AtRad1 protein plays a role in the removal of non‐homologous tails in homologous recombination. Recombination in the presence of non‐homologous overhangs is reduced 11‐fold in the atrad1 mutant compared with the wild‐type plants. AtRad1p is the A. thaliana homologue of the human Xpf and Saccharomyces cerevisiae Rad1 proteins. Rad1p is a subunit of the Rad1p/Rad10p structure‐specific endonuclease that acts in nucleotide excision repair and inter‐strand crosslink repair. This endonuclease also plays a role in mitotic recombination to remove non‐homologous, 3′‐ended overhangs from recombination intermediates. The Arabidopsis atrad1 mutant (uvh1), unlike rad1 mutants known from other eukaryotes, is hypersensitive to ionizing radiation. This last observation may indicate a more important role for the Rad1/Rad10 endonuclease in recombination in plants. This is the first direct demonstration of the involvement of AtRad1p in homologous recombination in plants.
Plants need sunlight for photosynthesis; however, they are thus exposed to the harmful effects of solar ultraviolet (UV) radiation. The two major DNA lesions produced by UV irradiation are cyclobutane pyrimidine dimers (CPDs) and pyrimidine (6,4) pyrimidinone dimers (reviewed by Mitchell and Nairn, 1989; Friedberg et al., 1995). If the lesions are not removed, they both block DNA replication and transcription (Protic‐Sabljic and Kraemer, 1985) and also cause mutations.
Nucleotide excision repair (NER) is responsible for processing DNA lesions that generate major distortion in the helical DNA structure, such as these UV photoproducts and bulky covalent chemical adducts (reviewed by Friedberg et al., 1995). The biochemical steps constituting this repair pathway have been characterized in Escherichia coli, yeast and mammalian systems (Aboussekhra and Wood, 1994; Laat et al., 1999), but much less information is available for plants (for reviews, see Britt, 1999; Tuteja et al., 2001). Following specific recognition of the damage and formation of a pre‐incision complex, a single‐strand oligonucleotide containing the damage site is excised and the resulting gap filled in and sealed (Friedberg et al., 1995). In Saccharomyces cerevisiae, two structure‐specific endonucleases nick the damaged DNA strand, Rad1p/Rad10p 5′ and Rad2p 3′ of the damaged site (Davies et al., 1995). The Rad1p/Rad10p endonuclease consists of a heterodimer of Rad1p and Rad10p (Bailly et al., 1992; Bardwell et al., 1992, 1993) and has been shown to incise DNA specifically at 5′‐double‐strand/3′‐single‐strand junctions (Tomkinson et al., 1993; Bardwell et al., 1994).
Rad1 and Rad10 proteins exhibit significant amino acid conservation with the mammalian Xpf and Ercc1 proteins, respectively (Brookman et al., 1996; Sijbers et al., 1996). The gene RAD1 also has homologues in Schizosaccharomyces pombe (rad16; Carr et al., 1994) and Drosophila melanogaster (MEI‐9; Sekelsky et al., 1995). AtRAD1, a homologue of RAD1, has recently been described as a single‐copy gene in the plant Arabidopsis thaliana (Fidantsef et al., 2000; Gallego et al., 2000; Liu et al., 2000) and shown to be important for the removal of 6‐4 photoproducts from UV‐irradiated plants (Liu et al., 2000). These authors have also shown that the Arabidopsis uvh1 mutant carries a single base pair substitution in the atrad1 gene, leading to mis‐splicing of the atrad1 mRNA in mutant plants (Liu et al., 2000). In vitro assays have confirmed this role in NER and also indicated a possible role for AtRad1 protein (AtRad1p) in the repair of oxidative DNA damage (Li et al., 2002). A single RAD10 homologue is found in the fully sequenced Arabidopsis genome (locus At3g05210), and a RAD10 homologue has also been described in lily (Xu et al., 1998).
The Ercc1/Xpf endonuclease plays an essential role in NER. Ercc1−/− mice show a more severe phenotype than other NER‐deficient mouse models, apparently associated with a premature replicative senescence in the Ercc1−/− cells (Weeda et al., 1997). Moreover, Xpf‐deficient mammalian cells and AtRad1p‐deficient plants, unlike other NER‐deficient cells, are sensitive to crosslinking agents (Westerveld et al., 1984; Hoy et al., 1985; Gallego et al., 2000). Unlike xpf and yeast rad1 and rad10 mutants, the atrad1 mutant (uvh1) is also hypersensitive to ionizing radiation (Harlow et al., 1994; Jenkins et al., 1995; Jiang et al., 1997; reviewed by Paques and Haber, 1997). These observations underline the importance of a potential role for AtRad1p in recombination, as well as its role in NER. A number of reports have shown Rad1p and Rad10p to have a role in mitotic recombination (reviewed by Paques and Haber, 1997). Mutations in RAD1 and RAD10 reduce intra‐chromosomal recombination between directly repeated sequences 8‐fold and also decrease the efficiency of homologous integration of linear DNA fragments and circular plasmids (Klein, 1988; Schiestl and Prakash, 1988, 1990; Prado and Aguilera, 1995). The rad1 and rad10 mutations are epistatic to one another (Schiestl and Prakash, 1988, 1990; Saffran et al., 1994), so the recombination role depends upon the heterodimer formation and most likely on its endonuclease function. Yeast excision repair mutants other than rad1 and rad10 do not show these recombinational phenotypes, further demonstrating the specific role of Rad1p/Rad10p in recombination (Ivanov and Haber, 1995).
Intra‐chromosomal recombination between tandem, direct repeats occurs most efficiently via single‐strand annealing (SSA) following the formation of a DNA double‐strand break (DSB) between, or within, the repeated sequences (discussed by Paques and Haber, 1997). Exonucleolytic resection of the 5′‐ended strands at the DSB produces two complementary, 3′‐ended single strands. These strands anneal and, after excision of any non‐homologous 3′‐ended overhangs and new DNA synthesis, ligation restores two continuous strands. Rad1p and Rad10p are required for the removal of these 3′ non‐homologous tails and for the removal of non‐homologous sequences during DSB‐induced recombination between plasmid inverted repeats (Fishman‐Lobell and Haber, 1992; Ivanov and Haber, 1995; Prado and Aguilera, 1995; Saparbaev et al., 1996; Paques and Haber, 1997; Sugawara et al., 1997; Colaiacovo et al., 1999). Similarly, Ercc1p is required to remove non‐homologous tails from ends‐in gene targeting constructs in CHO cells (Adair et al., 2000; Sargent et al., 2000) and for targeted gene replacement in mouse embryonic stem cells (Niedernhofer et al., 2001). Interestingly, Niedernhofer et al. (2001) suggest that the role of Rad1p/Rad10p in nicking synapsed recombination intermediates leads to the formation of a DSB in the recipient chromosomal DNA, thus permitting insertion of non‐homologous DNA into the chromosome.
As part of our analysis of DNA recombination mechanisms in plants, we present here results of a specific assay showing that AtRad1p plays a role in permitting productive recombination between DNA molecules in A. thaliana in the presence of non‐homologous tails.
Results and discussion
Over the last decade, a number of studies have contributed to our understanding of the apparent conservation of recombination‐associated proteins and recombination mechanisms between plants and other eukaryotes (reviewed in Britt, 1999; Gorbunova and Levy, 1999; Mengiste and Paszkowski, 1999; Vergunst and Hooykaas, 1999; Tuteja et al., 2001). However, the lack of identified mutant plants for recombination functions has meant that experimental verification of the roles of specific proteins in recombination in plants is lagging behind the identification, based on analogy, of the corresponding genes. In the present study, we tested specifically the involvement of the AtRad1p endonuclease in removing non‐homologous overhangs to permit productive homologous recombination between linear DNA molecules in A. thaliana. Our assay involves observing transient expression of recombined plasmids following biolistic transformation of Arabidopsis leaves.
AtRad1p is not needed for homologous recombination in the absence of non‐homologous overhangs
In this ‘no‐tails’ experiment, we addressed the question of whether AtRad1p is required for recombination between overlapping homologous DNA regions in the absence of non‐homologous overhangs (as indicated in Figure 1).
In order to monitor inter‐molecular recombination, plasmids pcw344 and pcw345 were linearized with EcoRI and BsmI, respectively, and used to co‐transform leaves from wild‐type and atrad1 plants (as described in detail in Methods and shown in Figure 1). Plasmid pcw344 contains only the region encoding the C‐terminal region of the β‐glucuronidase (GUS) gene and the nopaline synthetase (NOS) terminator, whereas pcw345 contains the 35S promoter and the region encoding the N‐terminal region of GUS. The overlapping homology region is 590 bp long, which has previously been shown to be sufficient for efficient recombination in analogous assays (Puchta and Hohn, 1991). GUS activity was subsequently detected as blue spots following histochemical treatment. Examples of transformed leaves showing blue GUS spots are shown in Figure 2.
As shown in Tables 1 and 2 and Figure 3, the frequencies of recombination events do not differ significantly between the two lines. Thus AtRad1p is not required for DSB repair by homologous recombination in the absence of non‐homologous overhangs (‘no tails’).
No recombinant spots were detected in control transformations carried out in parallel with the same quantities of the two plasmids individually (data not shown). Transformation efficiency was controlled in parallel by transforming with pGUS23, which contains the entire GUS transcription cassette. In order to produce a quantifiable number of spots, 10‐fold less plasmid DNA was used (5 mg versus 50 mg in the recombination assays). The numbers of GUS+ spots obtained with pGUS23 transformation of atrad1 and wild‐type leaves were similar; thus, these two plant lines were transformed with similar efficiency (Tables 1 and 2).
AtRad1p is required for homologous recombination in the presence of non‐homologous overhangs
In S. cerevisiae, the Rad1p/Rad10p endonuclease is required for the removal of 3′‐ended, non‐homologous single‐stranded tails from SSA intermediates during DSB‐induced recombination between direct repeats (Fishman‐Lobell and Haber, 1992; Ivanov and Haber, 1995; Prado and Aguilera, 1995; Sugawara et al., 1997).
To test the role of AtRad1p in the elimination of non‐homologous tails during SSA, a second recombination test was carried out. atrad1 and wild‐type plants were co‐transformed with pcw344/ScaI and pcw345/SapI (as described in Methods and shown in Figure 1). The overlapping homology region is 590 bp in both cases, to which pcw344/ScaI adds a non‐homologous tail of 816 bp and pcw345/SapI a tail of 233 bp.
The atrad1 mutant plants show significantly (11‐fold) less recombinant spots than the wild‐type plants (‘with tails’, Tables 1 and 2 and Figure 3). Thus, AtRad1p is implicated in DSB repair by homologous recombination in the presence of non‐homologous tails. We note that, although the absence of AtRad1p strongly reduces recombination in the presence of non‐homologous tails, it does not eliminate it altogether in our assay. Thus, a minor, AtRad1p‐independent pathway for these events must exist. Another possible factor contributing to these AtRad1‐independent events might be due to exonucleolytic action on the linear DNA substrates. This possibility is supported by our observation that, in the wild‐type leaves, the ‘with tails’ constructs give more recombinant spots than the ‘no tails’ constructs. Thus, the tails may act as a ‘buffer’ protecting the homologous GUS sequences from exonuclease degradation prior to recombination. That such an exonuclease activity does not affect our conclusions concerning the role of AtRad1p in our experiments is demonstrated by the equivalent recombination efficiencies in the wild‐type and mutant leaves with the ‘no tails’ substrates (Figure 3).
In conclusion, we show here that the Arabidopsis Rad1p homologue acts on recombining DNA molecules to permit productive homologous recombination in the presence of non‐homologous DNA overhangs. We are now testing the implication of AtRad1p in other forms of homologous recombination in Arabidopsis, such as gene targeting and spontaneous and induced intra‐chromosomal recombination.
Strains and growth conditions.
Arabidopsis thaliana were ecotype Columbia (Col). The atrad1 UV‐hypersensitive mutant (uvh1‐1; Harlow et al., 1994) was obtained from the Nottingham Arabidopsis Stock Center (CS3819; NASC: online catalogue available at http://nasc/nott.ac.uk/home.html). After sterilization with sodium hypochlorite (7%, w/v), atrad1 and wild‐type seeds (ecotype col4) were sown on germination medium [Murashige and Skoog basal medium with vitamins (Sigma), supplemented with 30 g/l sucrose, pH 5.8]. Plates were incubated in a growth chamber at 22°C (16 h light, 8 h dark). atrad1 mutants were verified both by Southern hybridization and UV hypersensitivity (data not shown).
As the control for transformation efficiency, we used plasmid pGUS23 (Puchta and Hohn, 1991), which contains the cauliflower mosaic virus (CaMV) 35S promoter, the GUS gene and the NOS transcription terminator.
The plasmid pcw344 was constructed by digesting pGUS23 with NdeI and SnaBI to eliminate the 35S promoter and the GUS N‐terminal region. The resulting plasmid was blunt‐ended with T4 DNA polymerase, purified through low‐melting‐point agarose and re‐circularized with T4 DNA ligase.
The plasmid pcw345 was obtained by SfuI–EcoRI digestion of plasmid pGUS23C1 (a deletion derivative of pGUS23; Puchta and Hohn, 1991), which contains the 35S promoter, the GUS N‐terminal region and the NOS terminator. This eliminated the NOS terminator. The resulting plasmid was purified, blunt‐ended and circularized as above
Biolistic transformation and recombination assay.
Leaves from 3‐ to 4‐week‐old wild‐type and atrad1 mutant plants were placed adaxial side up on solid germination medium in Petri dishes. Gold beads (Sigma; 0.6 μm diameter) were washed in water then rinsed and resuspended in ethanol at 0.1 μg/μl. To 50 μl of beads, 50 μl (50 μg) of plasmid DNA, 250 μl of 2.5 M CaCl2 and 100 μl of 0.1 M spermidine (Sigma) were added. After 5 min on ice, the beads were pelleted with a brief centrifugation and 400 μl of supernatant removed.
For each transformation, 5 μl (10%) of the resuspended, DNA‐coated gold beads were used with a helium particle gun (6 bars, 28 kPa, 13 cm from the leaves; Finer et al., 1992). Leaves were then placed in a growth chamber at 22°C (16 h light, 8 h dark) for 1 or 2 days and the GUS activity assay carried out as described by Jefferson et al. (1987). The number of blue spots per leaf was determined by visual observation using a dissecting microscope.
Transformation efficiency was controlled by transforming wild‐type and mutant plants with 5 μl (5 μg) of pGUS23. The lower quantity of plasmid DNA (5 μg versus 50 μg in the recombination assays) was used to reduce the number of spots per leaf in order to facilitate counting in the controls.
Plasmids pcw344 and pcw345 were linearized with BsmI or EcoRI, respectively, to prepare the ‘no‐tail’ recombination substrate. ‘With tail’ substrates were prepared by digesting pcw344 and pcw345 with ScaI or SapI, respectively. In all cases, the overlapping homology region is 590 bp, to which pcw344/ScaI adds a non‐homologous tail of 816 bp and pcw345/SapI a tail of 233 bp. In both ‘with‐tails’ and ‘no‐tails’ experiments, the transformation was carried out with 25 μl (25 μg) of each plasmid (pcw344 and pcw345).
The data on the number of recombination events per transformed leaf were analysed non‐parametrically (with the kind advice of Dr E.J. Louis, University of Leicester, UK). All leaves analysed from a given line were ranked by the number of blue spots for each set of leaves 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 wild‐type and atrad1 mutants from each type of transformation (Table 2).
We thank Biogemma and the laboratory of Michel Bernard (INRA) for kindly letting us use their helium canons.
- Copyright © 2002 European Molecular Biology Organization