Homologous recombination is key to the maintenance of genome integrity and the creation of genetic diversity. At the mechanistic level, recombination involves the invasion of a homologous DNA template by broken DNA ends, repair of the break and exchange of genetic information between the two DNA molecules. Invasion of the template in eukaryotic cells is catalysed by the RAD51 and DMC1 recombinases, assisted by a number of accessory proteins, including the RAD51 paralogues. Eukaryotic genomes encode a variable number of RAD51 paralogues, ranging from two in yeast to five in animals and plants. The RAD51 paralogues form at least two distinct protein complexes, believed to play roles in the assembly and stabilization of the RAD51-DNA nucleofilament. Somatic recombination assays and immunocytology confirm that the three ‘non-meiotic’ paralogues of Arabidopsis, RAD51B, RAD51D and XRCC2, are involved in somatic homologous recombination, and that they are not required for the formation of radioinduced RAD51 foci. Given the presence of all five proteins in meiotic cells, the apparent absence of a meiotic role for RAD51B, RAD51D and XRCC2 is surprising, and perhaps simply the result of a more subtle meiotic phenotype in the mutants. Analysis of meiotic recombination confirms this, showing that the absence of XRCC2, and to a lesser extent RAD51B, but not RAD51D, increases rates of meiotic crossing over. The roles of RAD51B and XRCC2 in recombination are thus not limited to mitotic cells.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Homologous recombination (HR) is a ubiquitous and precise DNA repair mechanism essential for maintaining genomic integrity (San Filippo et al., 2008; Heyer et al., 2010). In somatic cells, it acts in the repair of DNA breaks occurring after environmental stress, and in the recovery of stalled or broken replication forks. Physically linking and generating genetic exchanges between homologous chromosome pairs, HR is also essential to ensure proper chromosome segregation and to generate genetic diversity during meiosis in most eukaryotes. Inappropriate recombination can, however, lead to genomic instability, chromosome rearrangements and cell death.
Homologous recombination is initiated by DNA double-strand breaks (DSBs). Resection of the 5′-ended strands at the DSB generates 3′ single-stranded DNA overhangs (ssDNAs) that are bound by replication protein A (RPA), and in turn this is displaced by RAD51 in somatic cells, and RAD51 and DMC1 in meiotic cells. The helical nucleofilament formed by these recombinases on the broken DNA molecule catalyses the invasion of a homologous DNA template sequence by the 3′-ended DNA strand(s), which are extended through DNA synthesis, and finally the joint recombination intermediate is resolved to complete the process (for a review, see San Filippo et al., 2008; Heyer et al., 2010). At DSBs in mitotic G2- and M-phase cells, recombination primarily involves the invasion of the sister chromatid as a template. In meiotic cells, the situation is, however, more complex. Data from budding and fission yeasts show that both sister and non-sister chromatids can be used as templates for repair of meiotic DSB, although the relative use of one or the other remains uncertain (Schwacha and Kleckner, 1994, 1997; Goldfarb and Lichten, 2010; Hyppa and Smith, 2010; Pan et al., 2011; Pradillo and Santos, 2011). In plants, the use of the sister chromatid to repair meiotic DSB is suggested by the lack of chromosome fragmentation in achiasmate Arabidopsis dmc1 mutants (Couteau et al., 1999; Pradillo and Santos, 2011; Kurzbauer et al., 2012; Pradillo et al., 2012), and clearly occurs in the meiosis of Arabidopsis haploids (Crismani et al., 2013). Whether or not this is also so in wild-type (WT) diploid plants is of course uncertain; however, in the case of meiotic crossover formation (and in G1-phase mitotic cells), the template sequence certainly lies on the homologous sister chromosome. The choice of template on the sister or non-sister chromatid is thus a key determinant for the outcome of recombination (for a recent review, see Youds and Boulton, 2011).
In addition to the recombinases RAD51 and DMC1, eukaryotic genomes encode a variable number of RAD51 paralogues. These have been studied in a variety of organisms and are thought to play roles in homology search, and in the assembly and stabilization of the RAD51 nucleofilament, although their exact roles in recombination remain to be defined (for reviews, see Thacker, 2005; Bleuyard et al., 2006; Suwaki et al., 2011). Budding yeast has two RAD51 paralogues, RAD55 and RAD57, which form a heterodimer, and are implicated in mitotic and meiotic homologous recombination, and DNA repair (Krogh and Symington, 2004). In accordance with their role in RAD51 nucleofilament formation, both RAD55 and RAD57 are required for the formation of RAD51 foci in meiosis, although only partially in mitotic cells (Gasior et al., 1998, 2001; Lisby et al., 2004). An elegant recent study casts light on the roles of these proteins in the formation/stability of the RAD51/DNA nucleofilament, showing that the yeast RAD55/RAD57 complex acts to counterbalance the antirecombinase activity of SRS2 helicase (Liu et al., 2011).
Five RAD51 paralogues have been identified in animals and plants. These form two principal complexes (RAD51C/XRCC3 and RAD51B/C/D/XRCC2), and play roles in meiotic and somatic recombination, DNA repair and chromosome stability (for reviews, see Thacker, 2005; Bleuyard et al., 2006; Suwaki et al., 2011). Although their precise roles remain uncertain, a number of studies support the idea that they play both early and late roles in the recombination process (Bleuyard and White, 2004; Bleuyard et al., 2004; Liu et al., 2004, 2007, 2011; Badie et al., 2009; Compton et al., 2010; Chun et al., 2013). As for RAD55 and RAD57 in yeast, the RAD51 paralogues are suggested to act as mediators, facilitating RAD51 loading and stabilizing the RAD51/DNA nucleofilament. The RAD51 paralogues are needed for the formation of radiation-induced RAD51 foci in HeLa cells (Takata et al., 2001), but not spontaneous S-phase RAD51 foci (Tarsounas et al., 2004a). Chinese hamster ovary and mouse embryonic fibroblast cell lines defective in any of the five RAD51 paralogues have defects in the formation of DNA damage-induced RAD51 nuclear foci (Bishop et al., 1998; Tarsounas et al., 2004a; Abe et al., 2005; Smiraldo et al., 2005; van Veelen et al., 2005a,b). The formation of these RAD51 foci is not absolutely dependent on these proteins, however, as seen in chicken DT40 cells, where mutation of any of the RAD51 paralogues significantly reduces, but does not eliminate, the formation of foci (Takata et al., 2000, 2001; Yonetani et al., 2005; Qing et al., 2011). A possible explanation for this variable dependence on the paralogues comes from recent results showing that the mutation of XRCC2 leads to a delay (not absence) in RAD51 focus formation in hamster cells (Tambini et al., 2010), and that si-RNA depletion of RAD51D in human carcinoma cells reduces the numbers of RAD51 foci, whereas no effect was seen for the depletion of XRCC3 (Chun et al., 2013).
A late recombination role of the paralogues is supported by in vitro studies showing that paralogues can bind ‘Y’ DNA structures and Holliday Junctions (HJs) (Yokoyama et al., 2003, 2004; Liu et al., 2004; Compton et al., 2010). The RAD51C/XRCC3 complex is strongly associated with HJ resolution activity, and the expression of RAD51C and XRCC3 in mouse spermatocytes has been associated with crossover sites (Liu et al., 2004, 2007). Defects in both early and late stages of meiosis are seen in mice carrying hypomorphic rad51c alleles (Kuznetsov et al., 2007); however, the lethality of the mutants in vertebrates has hampered the study of their meiotic phenotypes.
Drosophila mutants impaired for the orthologues of RAD51C (spn-D) or XRCC3 (spn-B) are partially sterile and defective for meiotic recombination, but no effects were observed on DNA repair in somatic cells (Ghabrial et al., 1998; Abdu et al., 2003). In plants, Arabidopsis rad51c and xrcc3 mutants complete meiosis but are fully sterile because of genome fragmentation in zygotene/pachytene (Bleuyard and White, 2004; Bleuyard et al., 2004, 2005; Abe et al., 2005; Li et al., 2005).
In this work we present data from the flowering plant, Arabidopsis thaliana, in which the viability of the mutants greatly facilitates in vivo analysis of these proteins. As is the case for vertebrates, Arabidopsis has five RAD51 paralogues (RAD51B, RAD51C, RAD51D, XRCC2 and XRCC3), and yeast two-hybrid studies have confirmed that they form RAD51C/XRCC3 (CX3) and RAD51B/RAD51C/RAD51D/XRCC2 (BCDX2) protein complexes (Osakabe et al., 2002, 2005; reviewed by Bleuyard et al., 2006; Li and Ma, 2006; Osman et al., 2011; Roth et al., 2012). Mutants of any of the five RAD51 paralogues show hypersensitivity to the DNA cross-linking agent Mitomycin C, but not to ionizing radiation (Osakabe et al., 2002; Bleuyard and White, 2004; Abe et al., 2005; Bleuyard et al., 2005; Osakabe et al., 2005; Charbonnel et al., 2011). The Arabidopsis RAD51C and XRCC3 proteins are involved in DNA repair, somatic recombination and are essential for meiotic recombination (Bleuyard et al., 2004, 2005; Abe et al., 2005; Li et al., 2005; Vignard et al., 2007; Da Ines et al., 2012; Roth et al., 2012). Disruption of RAD51B, RAD51D or XRCC2 has no obvious effect on meiosis, however, and these mutants are fully fertile (Bleuyard et al., 2005; Durrant et al., 2007). Finally, two reports have linked RAD51D and XRCC2 to the defence response and developmental processes in Arabidopsis (Durrant et al., 2007; Inagaki et al., 2009).
Understanding the exact roles of the RAD51 paralogues in mitotic and meiotic HR, however, remains elusive. Can individual RAD51 paralogues contribute to HR independently of the others? Why are only RAD51C and XRCC3 essential for meiosis, when all five proteins play key roles in recombination in somatic cells, and high expression levels of RAD51B, RAD51D and XRCC2 have been observed in testis and pollen mother cells, also suggesting a role during meiosis (Cartwright et al., 1998a,b; Tarsounas et al., 2004b; Yang et al., 2011)? We present here an analysis of the roles of RAD51B, RAD51D, and XRCC2 in somatic and meiotic homologous recombination in Arabidopsis. Our data show that the three RAD51 paralogues contribute to somatic recombination to differing extents. Surprisingly, and notwithstanding the fertility of the mutant plants, analyses of meiotic recombination reveal increased rates of crossovers in rad51b and xrcc2, but not in rad51d mutants.
Roles of RAD51B, RAD51D and XRCC2 in somatic homologous recombination
Although all five Arabidopsis RAD51 paralogues are known to be involved in DNA repair (Bleuyard and White, 2004; Abe et al., 2005; Bleuyard et al., 2005; Osakabe et al., 2005; Durrant et al., 2007; Charbonnel et al., 2011; Roth et al., 2012), of the ‘mitotic’ RAD51 paralogues, only RAD51D has been shown to directly play a role in somatic recombination (Durrant et al., 2007).
Somatic HR was monitored in rad51b, rad51d and xrcc2 mutant plants using a previously described in planta recombination assay, consisting of a tandem pair of overlapping and inverted fragments of the β-glucuronidase (GUS) gene separated by a hygromycin-resistant marker (Swoboda et al., 1994; Puchta et al., 1995; Gherbi et al., 2001; Seeliger et al., 2012). We introduced the GUS recombination reporter locus from line 1415 into rad51b, rad51d and xrcc2 mutant plants through crossing and monitored somatic HR frequencies (HRFs) in F3 progeny, homozygous for the GUS recombination substrate and for the mutant rad51 paralogue allele or sister WT plants (Figure 1; Table 1). Spontaneous HR was significantly reduced in all three paralogue mutants compared with WT controls, although to differing extents, with rad51b, rad51d and xrcc2 showing reductions to 50, 75 and 36% of WT levels, respectively. These results with a new rad51d mutant allele (see 'Experimental Procedures' and Figure S1) are consistent with those previously reported for a different rad51d mutant line (Durrant et al., 2007).
Table 1. Frequencies of spontaneous and induced somatic homologous recombination events in rad51b, rad51d and xrcc2 mutants, and in wild-type (WT) control plants
Spontaneous somatic HR
Bleomycin-induced somatic HR
Recombination in the mutants and corresponding WT were compared using non-parametric statistical analysis (Mann–Whitney test).
n, the number of plantlets screened; N, the total number of blue spots (recombination events; m ± SEM, the mean number of recombination events per plant; ratio, means from mutant/means from the corresponding WT.
1.6 ± 0.3
29.6 ± 1.5
1.3 ± 0.2
25.1 ± 1.8
0.7 ± 0.1
6.6 ± 0.5
1.1 ± 0.1
7.3 ± 0.5
0.5 ± 0.1
2.6 ± 0.2
86.5 ± 2.7
2.6 ± 0.2
82.0 ± 2.2
1.9 ± 0.2
60.8 ± 2.3
2.0 ± 0.2
58.2 ± 2.3
3.7 ± 0.3
29.6 ± 1.5
2.5 ± 0.3
34.1 ± 1.8
1.3 ± 0.1
8.2 ± 0.6
0.9 ± 0.2
11.3 ± 0.9
Similar analyses were carried out after treatment of the plantlets with bleomycin, a γ-ray mimetic agent known to induce DSBs. As seen for spontaneous somatic recombination, the number of recombination events per plant is strongly reduced (to 30% of WT) in rad51b and xrcc2 mutants, and to a lesser extent in rad51d mutants (71%; Figure 2; Table 1).
RAD51B, RAD51D and XRCC2 are not required for RAD51 focus formation in somatic cells
Arabidopsis RAD51B, RAD51D and XRCC2 thus play important roles in somatic homologous recombination, but to differing extents. We note furthermore that, although the reduction in recombination rates is clear, bleomycin treatment induces recombination in the absence of any one of these proteins (see above). This prompted us to analyse the formation of radio-induced RAD51 foci. In Arabidopsis, the dose–response relationship for numbers of DSBs per Gy has been established based on the number of radio-induced γ-H2AX foci (Friesner et al., 2005; Charbonnel et al., 2010). We thus used irradiation to analyse the presence of radioinduced RAD51 foci in root-tip nuclei of WT, rad51b, rad51d and xrcc2 plants. We chose a dose of 100 Gy of γ-rays to give an easily quantifiable number of foci. As expected, no foci were detected in the nuclei of non-irradiated plants, whereas nuclei from WT root tips fixed 2 h after 100 Gy of γ-rays present numerous foci, with a mean of eight foci per nucleus (Figure 3). Similar numbers of RAD51 foci were observed in the root tips of irradiated rad51b, rad51d and xrcc2 mutants (Figure 3b), showing that RAD51B, RAD51D and XRCC2 are not essential for the formation of radioinduced RAD51 foci in somatic cells of Arabidopsis.
Enhanced meiotic recombination rates in rad51b and xrcc2 mutants
Meiosis in Arabidopsis rad51b, rad51d and xrcc2 mutants appears normal, with regular chromosome pairing, synapsis and correct segregation, leading to the formation of viable gametes and full fertility (Bleuyard et al., 2005; Osakabe et al., 2005). This contrasts dramatically with the meiotic genome fragmentation and resulting sterility of rad51c and xrcc3 mutant plants (Bleuyard and White, 2004; Bleuyard et al., 2004, 2005; Abe et al., 2005; Li et al., 2005). Given their expression in meiotic tissue and the important roles of all five paralogues in mitotic recombination, it is intriguing that only RAD51C and XRCC3 play roles in the repair of SPO11-induced DSB in the meiotic prophase. We thus asked whether the other three proteins might participate in meiotic recombination, but that their absence results in more subtle mutant phenotypes.
Meiotic recombination rates in these mutants were measured using the fluorescent seed markers developed by the Levy lab (Melamed-Bessudo et al., 2005). This assay permits the determination of meiotic recombination rates by counting seeds expressing both, neither or expressing only GFP or RFP (recombinant seeds). We used the Columbia background line, col3-4/20, in which the two markers are located on the left arm of chromosome 3, at a physical distance of 5.105 Mb and a genetic distance of 16 cM (Melamed-Bessudo et al., 2005). This marker pair gives a genetic map distance (21 cM; Table 2) that is similar to the value of 20 cM recently reported by Yelina et al. (2012), but is higher than that originally reported. Such differences between laboratories can be explained by differing growth conditions, known to influence recombination rates (Melamed-Bessudo et al., 2005; Francis et al., 2007).
Table 2. Meiotic recombination rates between the GFP and RFP protein markers in tester line col3-4/20 in rad51b, rad51d or xrcc2 mutants, compared with the corresponding wild type (WT)
No. of plants
Frequency of recombination ± SEM
Mean frequency ± SEM
For each genotype two independent experiments were performed and pooled data were used to calculate mean frequencies and P-values using the unpaired Student's t-test.
0.209 ± 0.007
0.217 ± 0.005
0.227 ± 0.004
0.241 ± 0.008
0.247 ± 0.007
0.253 ± 0.012
0.195 ± 0.014
0.211 ± 0.006
0.219 ± 0.009
0.185 ± 0.007
0.211 ± 0.008
0.229 ± 0.008
0.198 ± 0.007
0.203 ± 0.003
0.204 ± 0.004
0.306 ± 0.023
0.315 ± 0.01
0.320 ± 0.010
Homozygous rad51b, rad51d and xrcc2 mutants were crossed with line col3-4/20 homozygous for both fluorescent markers. After a series of self-fertilizations and backcrosses (see 'Experimental Procedures'), recombination was measured in self-fertilized mutants and control plants in the same genetic background (Figure 4; Table 2). Meiotic recombination was measured in two independent experiments using between 4 and 15 plants per genotype, and per experiment, representing 8583–23 134 seeds (Table 2). Strikingly, in two independent experiments xrcc2 mutants showed a significant 50% increase in recombination rate in the marked interval (0.315 versus 0.203). The recombination rate was also significantly increased in two independent experiments in rad51b plants, although to a lesser extent (14%; P = 0.0008, unpaired Student's t-test). No significant differences were seen for rad51d, however. The absence of XRCC2, and to a lesser extent RAD51B, thus increases the meiotic crossing-over rate in this chromosomal interval.
Analysis of meiotic recombination rate in xrcc2 mutants using INDEL markers
In order to determine whether or not the meiotic hyper-recombination phenotype of the xrcc2 mutant is limited to the interval tested on chromosome 3, we measured recombination rates in two additional genetic intervals defined by insertion/deletion (INDEL) DNA sequence markers in Columbia/Landsberg hybrids. F2 populations homozygous for the xrcc2 mutation and the corresponding WT plants were derived from a cross between the xrcc2 mutant (Columbia accession) and a WT plant of the Landsberg erecta accession (Figure S2). Pairs of INDEL markers were selected on chromosomes 1 and 4 (Table S1). F2 plants heterozygous for the markers were selected, meiotic segregation of the INDEL markers were analysed and recombination rates were calculated for each interval (Table 3). We observed a significant increase in meiotic recombination in the xrcc2 mutant for both the chromosome-1 and chromosome-4 intervals (40 and 20%, respectively; Table 3). The stimulation of crossing over in the absence of XRCC2 is thus observed in all three chromosomal intervals tested.
Table 3. Meiotic recombination frequency (MRF) in the wild type (WT) and in the xrcc2 mutant, calculated from analyses of INDEL markers in F2 hybrids
Physical distance (Mb)
No. of plants
No. of plants
RAD51B, RAD51D and XRCC2 are required for efficient homologous recombination in somatic cells
All five of the Arabidopsis RAD51 paralogues act in DNA repair (Bleuyard and White, 2004; Abe et al., 2005; Bleuyard et al., 2005; Osakabe et al., 2005; Durrant et al., 2007; Charbonnel et al., 2011; Roth et al., 2012). RAD51C and XRCC3 have been linked to meiotic chromosome stability and pairing (Bleuyard and White, 2004; Bleuyard et al., 2004; Abe et al., 2005; Li et al., 2005; Da Ines et al., 2012), and RAD51C, RAD51D and XRCC3 have been shown to act in somatic homologous recombination (Abe et al., 2005; Durrant et al., 2007; Roth et al., 2012). We show here that the Arabidopsis rad51b and xrcc2 have reduced somatic recombination, and confirm the previously reported reduction in rad51d mutants (Durrant et al., 2007). We note, however, that although significantly reduced, homologous recombination is not completely abolished in these mutants, and can be stimulated with bleomycin treatment. This bleomycin inducibility of recombination contrasts with the more severe phenotype observed for the Arabidopsis brca2 double mutants (Siaud et al., 2004; Seeliger et al., 2012), another mediator regulating RAD51 filament formation and strand exchange (for a review, see Holloman, 2011).
Our analyses suggest that RAD51B, RAD51D and XRCC2 contribute to differing extents to RAD51-mediated recombination, and/or have evolved individual roles. Such differences seem a priori surprising, given the clearly established existence of the two RAD51 paralogue protein complexes (RAD51C/XRCC3 and RAD51B/C/D/XRCC2). They are, however, fully concordant with reports showing that the situation is more complex. In chicken DT-40 cells, the RAD51 paralogues can act separately in different steps of homologous recombination and DNA damage repair (Yonetani et al., 2005). A recent report shows that knock-downs of RAD51B or RAD51C induce cell-cycle arrest, whereas the inhibition of XRCC3 induces mitotic chromosomal aberrations ascribed to defects in HJ processing (Rodrigue et al., 2013). In Arabidopsis, rad51d mutants have enhanced disease susceptibility, and are altered for pathogenesis-related (PR) gene expression, in addition to hypersensitivity to DNA damage and impairment of somatic homologous recombination (Durrant et al., 2007). In human cells, RAD51-dependent homologous recombination is reduced, but not abolished, in the absence of the RAD51 paralogues (Chun et al., 2013). In agreement with this, we also observe γ-ray induced RAD51 focus formation in rad51b, rad51d and xrcc2 mutant plants. We cannot of course exclude that the kinetics of formation, stabilization or function of the RAD51 nucleofilament are altered in these mutants; however, in agreement with results from vertebrate and yeast cells (see the 'Introduction'), our data clearly establish that RAD51B, RAD51D and XRCC2 are not essential for RAD51 nucleofilament formation in Arabidopsis. The Arabidopsis RAD51 paralogues thus clearly play different and multiple roles in recombination and DNA repair.
A role for XRCC2 in meiotic recombination
The absence of detectable meiotic defects in rad51b, rad51d and xrcc2 mutants has resulted in the assumption that these proteins play roles only in mitotic recombination (Osakabe et al., 2002, 2005; Bleuyard et al., 2005; Durrant et al., 2007). They are, however, transcribed in meiotic tissues in animals (Cartwright et al., 1998a,b; Tarsounas et al., 2004b) and in Arabidopsis (Yang et al., 2011). Reasoning that RAD51B, RAD51D and XRCC2 may play roles in meiotic recombination, but that their absence might result in more subtle effects than the genome fragmentation seen in rad51c and xrcc3 mutants, we tested meiotic recombination rates in the Arabidopsis mutants. Using the well-described seed reporter lines from the Levy lab (Melamed-Bessudo et al., 2005; Charbonnel et al., 2011; Pecinka et al., 2011; Melamed-Bessudo and Levy, 2012; Yelina et al., 2012), and INDEL genotyping, we show here that the absence of XRCC2 leads to significant increases in meiotic crossing over in all three tested genetic intervals on three different chromosomes. The absence of RAD51B also gives a meiotic hyper-recombination phenotype, although to a lesser extent than that of xrcc2 mutant plants. We found no effect on meiotic recombination in rad51d mutants, however.
Given their implication in mitotic recombination, it is intriguing that the xrcc2 and rad51b mutants should have meiotic hyper-recombination phenotypes. It is possible that this effect results from changes in gene expression of other recombination genes or from increased numbers of DSBs in the mutants; however, given that meiotic recombination is upregulated by the induction of programmed DSBs, and that crossovers only represent a minority outcome, we do not feel that this explanation is likely. Specifically, in Arabidopsis pollen mother cells there are 250–300 DSBs for nine or 10 chiasmata per meiocyte, with minor variations depending on the accession (Sanchez-Moran et al., 2002; Lopez et al., 2012; Serrentino and Borde, 2012). This high DSB/CO ratio means that the great majority of meiotic recombinational DSB repair does not lead to chiasmata. In support of this, recent reports demonstrate that decreasing or increasing DSBs has little (or no) effect on crossover numbers in yeast, Caenorhabditis elegans and mouse (Martini et al., 2006; Youds et al., 2010; Lange et al., 2011; Rosu et al., 2011; Cole et al., 2012).
Two conditions must be met for a meiotic DSB to lead to an interhomologue crossover: the broken DNA molecule must recombine with a non-sister chromatid and the resulting recombination intermediate must be resolved to give a physical crossing over of flanking sequences. Recombinational repair of the majority of meiotic DSB (non-crossover) thus does not meet one or both of these conditions. That Arabidopsis is able to repair meiotic DSB using sister chromatids is clearly seen in the analysis of haploid meioses by the Mercier lab (Crismani et al., 2013), and recent results confirm the frequent use of sister chromatids for meiotic DSB repair in yeast (Goldfarb and Lichten, 2010; Hyppa and Smith, 2010). The second requirement for interhomologue crossing over is the choice of a pathway, leading to a resolution of the recombination intermediate and resulting in a crossover (CO), rather than a non-crossover (NCO). In yeast this choice is made early in the recombination process, around the time of strand invasion (Hunter and Kleckner, 2001; Bishop and Zickler, 2004; Borner et al., 2004; Youds and Boulton, 2011; Crismani et al., 2012). In mice, however, recent work suggests that the CO/NCO decision occurs after the formation of early recombination intermediates (Cole et al., 2012). The RAD51 and DMC1 recombinases clearly play key roles in these events, with DMC1 believed to promote interhomologue recombination and crossing over, and RAD51 favouring non-crossover outcomes of the repair of the remaining (majority) of DSB. Work in yeast first showed that this preference may result from the restriction of the contribution of RAD51-dependent recombination in meiosis through phosphorylation of the axial element protein Hop1 (Niu et al., 2005; Carballo et al., 2008; Ho and Burgess, 2011) and through the action of HED1 on the recruitment of RAD54 (Tsubouchi and Roeder, 2006; Busygina et al., 2008, 2012; Nimonkar et al., 2012). The achiasmate meiosis with intact univalent chromosomes of Arabidopsis dmc1 mutants contrasts markedly with the meiotic genome fragmentation of rad51 mutants, in which the numbers of meiotic DMC1 foci are also reduced (Couteau et al., 1999; Li et al., 2004; Vignard et al., 2007; Crismani et al., 2012; Kurzbauer et al., 2012; Pradillo et al., 2012). Both the reduction in numbers of DMC1 foci and the meiotic genome fragmentation of rad51 mutants are substantially dependent upon ATR-dependent phosphorylation of the Hop1 orthologue, AtASY1 (Sanchez-Moran et al., 2007; Ferdous et al., 2012; Kurzbauer et al., 2012). DMC1 is thus able to repair meiotic DSBs using non-sister and sister chromatid templates, a conclusion dramatically confirmed in yeast by a recent study showing that DMC1 alone is sufficient for meiotic recombination, and that the requirement for RAD51 is for the protein itself and not for its catalytic strand-exchange activity (Cloud et al., 2012).
In contrast to the effects of helicases such as FANCM, RTEL-1 and SGS1/BLM in driving recombination intermediate structures to non-crossover SDSA and HJ dissolution pathways (Oh et al., 2007; Youds et al., 2010; Crismani et al., 2012; De Muyt et al., 2012; Knoll et al., 2012; Lorenz et al., 2012; Zakharyevich et al., 2012), we speculate that RAD51B and XRCC2 affect the balance of the activities of DMC1 and RAD51 in meiosis through the modulation of the RAD51 nucleofilament and enhancement of its strand exchange activity, thus affecting the outcome of recombination at a given DSB. In their absence, this balance shifts towards crossover outcomes, thus resulting in the subtle increases in meiotic crossing-over rates observed in Arabidopsis rad51b and xrcc2 mutants.
Plant material and growth conditions
The A. thaliana rad51b and xrcc2 mutants used in this work have been described previously (Bleuyard et al., 2005). The T-DNA insertion rad51d-3 allele (SAIL_564_A06) has not been described elsewhere, and was characterized for this study (Figure S1). Mutant seeds from the SAIL collection (Sessions et al., 2002) were obtained from the Nottingham Arabidopsis Stock Centre (http://arabidopsis.info). The junctions of the T-DNA insertion in the RAD51D locus (AT1G07745) were amplified by PCR and verified by DNA sequencing. Homozygous lines were further analysed by RT-PCR, which showed the absence of the RAD51D transcript (Figure S1).
Plants were grown under standard conditions: seeds were stratified in water at 4°C for 2 days and grown in vitro on 0.8% agar plates, 1% sucrose and half-strength MS salts (M0255; Duchefa Biochemie, http://www.duchefa-biochemie.nl), or on soil in a glasshouse with a 16-h light/8-h dark cycle, at 23°C with 45–60% relative humidity.
Somatic homologous recombination assay
To determine the frequency of somatic homologous recombination, seeds were surface-sterilized, stratified at 4°C for 2 days and grown in petri dishes on 0.8% w/v agar, 1% w/v sucrose and half-strength MS salts for 2 weeks. Seedlings were then harvested and incubated in staining buffer containing 50 mm sodium phosphate buffer (pH 7.2), 0.2% Triton X-100 and 2 mm X-Gluc (5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid), dissolved in N,N-dimethylformamide. Plants were then infiltrated under vacuum for 15 min and incubated for 24 h at 37°C. The staining solution was then replaced with 70% ethanol to remove chlorophyll, and the blue spots were counted under a dissecting microscope.
For bleomycin treatment, 1-week-old seedlings were transferred to liquid medium containing half-strength MS salts, 1% sucrose and 0.1 μg ml−1 bleomycin, and grown for another week before staining as described above.
The RAD51 paralogue mutant plants were crossed with homozygous plants derived from the previously described tester line col3-4/20 (Columbia background; Melamed-Bessudo et al., 2005; Melamed-Bessudo and Levy, 2012). The RFP and GFP markers of this tester line are located 5.1 Mb apart on chromosome 3. The tester line was always used as the male parent. All seeds of the F1 generation expressed the RFP and GFP markers. The resulting F1 plants were backcrossed with the mutant lines (BC1 generation). BC1 seeds expressing RFP and GFP markers, and homozygous for rad51b, rad51d or xrcc2 mutations, were selected. Meiotic recombination rates were monitored in self-pollinated seeds of each rad51b, rad51d-3 or xrcc2 homozygous BC1 plant (BC1F2 generation). As a control for the effect of these mutations, the same F1 plants were also backcrossed with WT (RAD51B, RAD51D and XRCC2) Columbia plants, and recombination rates in the WT background were also checked in BC1F2 seeds.
For evaluation of seed fluorescence, dry seeds were photographed using a LEICA MZFLIII stereomicroscope equipped with GFP- and RFP-specific filters and a LEICA DFC420C camera. Fluorescence was then analysed using imagej.
INDEL marker genotyping
Seeds from the F2 populations were surface-sterilized and grown in vitro on half-strength MS/1% sucrose for 2 weeks. Individual seedlings were harvested and samples genotyped either by PCR, followed by analysis on 2% agarose gels, or by the GENTYANE genotyping platform of the INRA UMR 1095 GDEC (Clermont-Ferrand, France). Briefly, markers were amplified using a M13 protocol, products analysed using an ABI 3730XL DNA Sequencer (Applied Biosystems, http://www.appliedbiosystems.com) and data evaluated using genemapper.
Five-day-old seedlings were irradiated with a dose of 100 Gy from a 137Cs source according to the method described by Charbonnel et al. (2010). Preparation and immunostaining of nuclei were performed as previously described (Charbonnel et al., 2010), except that slides were incubated with primary antibody (1:100) for 24 h at 4°C. The RAD51 antibody used in this study has been previously described, and was raised in rabbit (Mercier et al., 2003). All observations were made with a motorized Zeiss AxioImager. Z1 epifluorescence microscope (Carl Zeiss, http://www.zeiss.com) using a PL Apochromat 100 × /1.40 oil objective. Photographs were taken with an AxioCam Mrm camera (Carl Zeiss) and appropriate Zeiss filter sets adapted for the fluorochromes used: filter set 25HE (DAPI) and filter set 38HE (Alexa 488). Image stacks were captured in three dimensions (x, y, z) and further deconvoluted with the deconvolution module of axiovision 4.6.2 (theoretical PSF, iterative algorithm; Carl Zeiss). RAD51 foci were then counted by eye. For presentation, the pictures are collapsed Z-stack projections obtained using the extended-focus module (projection method) of axiovision 4.6.2.
We thank Avraham Levy for sending the meiotic tester line and Chris Franklin for providing the RAD51 antibody. We acknowledge the members of the GENTYANE genotyping platform of the INRA UMR 1095 GDEC for their help with INDEL genotyping and analyses of the data. We thank the members of the recombination group for their help and discussions. This work was financed by a European Union research grant (FP7-KBBE-2008-227190), the Centre National de la Recherche Scientifique, the Université Blaise Pascal, the Université d'Auvergne and the Institut National de la Santé et de la Recherche Médicale. The authors have no known conflict of interest to declare.