Arabidopsis SNI1 and RAD51D regulate both gene transcription and DNA recombination during the defense response

Mutation of RAD51D Suppresses the sni1 and sni1 npr1 Mutant Phenotypes.

Derepression of genes, predominantly NPR1-dependent genes, in the sni1 and sni1 npr1 mutants results in growth retardation (8, 9). The mutant plants are smaller than WT and have narrower leaves (Fig. 2A). We used this phenotype to screen for suppressors of sni1. Full details of the genetic screen are given in supporting information (SI) Text. As shown in Fig. 2A, rad51d restored WT morphology to both sni1 and sni1 npr1. To determine whether rad51d affects PR gene transcription, we examined the expression of an SA-responsive reporter gene BGL2:GUS, a fusion of the BGL2 (PR2) promoter with the Escherichia coli uidA gene encoding the β-glucuronidase (GUS) enzyme. GUS activity was assayed by using histochemical staining, resulting in a blue color. The low constitutive GUS activity detected in sni1 and sni1 npr1 was eliminated in sni1 rad51d and sni1 npr1 rad51d (Fig. 2B). We hypothesized that the NPR1-independent PR gene induction in sni1 npr1 would also be abolished in sni1 npr1 rad51d. Indeed, in the presence of BTH, BGL2:GUS was induced to WT levels in the sni1 rad51d double mutant, whereas in the sni1 npr1 rad51d triple mutant, the reporter gene was not responsive, resembling npr1 plants (Fig. 2C). We then examined the effect of rad51d on another PR gene, PR1, over a range of INA concentrations (Fig. 2D). As observed previously, PR1 expression in sni1 and sni1 npr1 was induced at a 10-fold lower concentration of INA than in WT. In the sni1 rad51d double mutant, a WT pattern of induction was restored. On the other hand, the induction pattern in sni1 npr1 rad51d reverted to that of npr1. The sni1 npr1 rad51d mutant also exhibited other phenotypes characteristic of npr1, including enhanced susceptibility to the bacterial pathogen Pseudomonas syringae pv maculicola ES4326 (Psm ES4326) (Fig. 2E) and a loss of protection against Psm ES4326 after BTH treatment (data not shown). These data clearly demonstrate that rad51d is a true suppressor of sni1. In other words, both the background and the SA-induced NPR1-independent PR gene expression and resistance observed in sni1 and sni1 npr1 require the function of RAD51D.

The rad51d Mutant Has Increased Susceptibility to Pathogen Infection.

A direct role for RAD51D in plant defense was established by infection experiments performed on rad51d. The rad51d single mutant showed a moderate, but significant, increase in susceptibility to Psm ES4326 (Fig. 2F). This effect was augmented in the sni1 and sni1 npr1 backgrounds, where greater differences in Psm ES4326 growth (10- to 70-fold) were observed (Fig. 2 E and G). Using the sensitized sni1 and sni1 npr1 backgrounds revealed a positive role for RAD51D in PR gene expression and defense, which we would not have discovered otherwise. Indeed, except for the moderate disease susceptibility phenotype, rad51d plants are indistinguishable from WT in morphology and expression of the BGL2:GUS and PR1 genes (Fig. 2 A–D).

Map-Based Cloning of RAD51D.

The rad51d mutation was mapped to the RAD51D locus (At1g07745) (SI Fig. 6), which encodes a member of the RecA/Rad51 family of recombination and repair proteins (10, 11). A 7-bp deletion was discovered in exon 4 of this gene (base pair 974–980, relative to the ATG), which would cause a frameshift resulting in truncation of the protein. The identity of the mutation was confirmed by genetic complementation. Two cosmids (cosmids 24 and 84 in SI Fig. 6) containing RAD51D complemented the rad51d mutation, and transgenic plants homozygous for constructs containing the genomic RAD51D sequence or the RAD51D cDNA (in the sni1 rad51d background) regained sni1 leaf morphology and expression of BGL2:GUS (Fig. 3 A and B). Full details of the mapping and cloning are given in SI Text.

Analysis of RAD51D cDNAs showed that, similar to the mammalian orthologs (12), Arabidopsis RAD51D mRNA undergoes alternative splicing (SI Table 1, SI Fig. 7, and SI Text). The most abundant splice variant (≈50% of the transcripts analyzed) encodes the full-length protein, as determined by alignment with mammalian RAD51D (Fig. 3C). The functionality of this transcript was confirmed by genetic complementation when the cDNA was expressed from the constitutive 35S promoter (Fig. 3 A and B).

Yeast Rad51 (ScRad51) is a eukaryotic homologue of the E. coli RecA protein, which plays a central role in DNA recombination and repair (12). In higher eukaryotes such as mammals and Arabidopsis, in addition to RAD51 and the meiosis-specific DMC1, there are five RAD51 paralogs: RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3 (10–16). The RAD51 paralogs show limited sequence similarity (≈15–20% identity) to ScRad51, and their biological functions are poorly understood (13). Arabidopsis RAD51D has only 14% identity to ScRad51; however, phylogenetic analysis indicates that Arabidopsis RAD51D is the ortholog of human RAD51D (27% identity) (10) (SI Fig. 8).

Expression studies of RAD51D using a fusion of the RAD51D promoter to GUS showed detectable levels of GUS staining only in pollen grains (SI Fig. 9 A and B). Expression in leaf tissues is low, but detectable by RT-PCR (data not shown). Despite the high expression of RAD51D in pollen grains, the rad51d mutant has viable pollen (SI Fig. 9D) and is fertile. Furthermore, rad51d partially suppresses the reduced fertility phenotype of sni1 (data not shown).

The rad51d Mutant Is Hypersensitive to DNA Damaging Agents.

Mutants lacking proteins involved in DNA repair often show increased sensitivity to treatments that induce DNA damage (17). Mutations in all of the other Arabidopsis RAD51 paralogs result in increased sensitivity to the DNA-cross-linking agent mitomycin C (MMC) (10, 14). We therefore tested the sensitivity of rad51d to MMC and the γ-ray mimetic bleomycin, which causes double-strand breaks. The rad51d mutant was hypersensitive to both MMC and bleomycin (Fig. 4). DNA cross-linking and double-strand breaks can be repaired through homologous recombination (17), therefore it is likely that RAD51D plays an essential role in this homologous recombination repair pathway.

RAD51D Is Required for Somatic Homologous Recombination.

In plants, both abiotic and biotic stresses can cause an increase in homologous recombination (5). Treatment of Arabidopsis with BTH, INA, or pathogen challenge increases the rate of somatic homologous recombination (18). The cim3 (constitutive immunity 3) mutant, which shows constitutive activation of SAR, also has elevated levels of somatic recombination. The involvement of RAD51D in defense-related gene expression suggests a possible link between induced gene expression and DNA recombination during plant defense. To test this hypothesis, we used plants containing reporter transgenes consisting of two overlapping segments of the GUS gene in direct or inverted orientations, lines 1418 and 1445, respectively (18). Homologous recombination between the overlapping regions results in a functional GUS gene and groups of cells expressing GUS appear as blue sectors after staining. We crossed each reporter line with sni1 npr1 rad51d to generate lines homozygous for the recombination reporter transgene in a homozygous sni1, rad51d, or sni1 rad51d background. We observed that, in the 1445 background, the sni1 mutant had constitutively elevated levels of recombination compared with WT, as indicated by an increased frequency of recombination sectors (Fig. 5A). This increase was not observed in the sni1 rad51d double mutant, indicating that RAD51D is required for the somatic recombination observed in sni1. Similar results were observed for the 1418 reporter (data not shown). Moreover, the rad51d single mutant has fewer sectors than WT under both uninduced and INA-induced conditions (Fig. 5), indicating that the rad51d mutation alone results in a defect in homologous recombination.

Mutant Screen.

To screen for suppressors of sni1 we took advantage of the sni1 morphological phenotype (smaller and narrower leaves, shorter roots, and lower fertility than WT) by looking for mutants that are similar in size to WT plants, in a homozygous sni1 background. Approximately 40,000 homozygous sni1 seed were mutagenized by using fast neutron bombardment, and M₂ plants were examined for suppression of the sni1 morphology. M₃ seeds from each putative mutant were grown on MS media and tested for loss of expression of the BGL2:GUS reporter gene and the PR1 gene. Full details of the mutant screen are given in SI Text.

Double and Triple Mutant Generation.

To generate the sni1 npr1 rad51d triple mutant, sni1 rad51d was crossed to sni1 npr1–1. The rad51d single mutant and the npr1 rad51d double mutant were generated by outcrossing sni1 rad51d to Col BGL2:GUS and npr1–1, respectively. Cleaved Amplified Polymorphic Sequence analysis was used to identify the sni1, npr1 (8), and rad51d mutations (see SI Text for details of sni1 and rad51d markers).

Map-Based Cloning of RAD51D.

Map-based cloning was carried out using standard techniques; full details are given in SI Text.

Pseudomonas syringae Infection.

Whole leaves of 4-week-old soil-grown plants were infiltrated with a Psm ES4326 suspension (OD₆₀₀ = 0.0001) in 10 mM MgSO₄. The infected tissue was harvested immediately after infection and after 3 days. Two 6-mm leaf discs were collected per plant from 4–16 plants. The leaf discs were placed in 500 μl of MgSO₄ in a 96-well plate and ground by using a metal bead in a Geno/Grinder 2000 (Spex CertiPrep; Spex Industries, Metuchen, NJ) at 1,000 rpm for 30 s followed by 1,500 rpm for 10 s. Serial 10-fold dilutions were made in 10 mM MgSO₄ solution. Aliquots (10 μl) were plated in rows onto King's B medium containing 100 μg/ml streptomycin by using an eight-channel multipipettor. The plates were incubated for 2–3 days at room temperature before counting colonies. Statistical analyses were performed using Student's t test of the differences between two means of log-transformed data (29) or by two-way ANOVA using the STATLETS program (StatPoint LLC, Herndon, VA).

MMC and Bleomycin Assays.

Plants were grown on MS media containing MMC (0, 10, 20, or 40 μM) or bleomycin (0, 5, 10, or 20 μg/ml) (Sigma, St. Louis, MO) for 14 days. In the absence of the genotoxic agent, all plants were able to grow true leaves. The percentage of plants with no true leaves was therefore used as an indication of sensitivity.

Somatic Recombination Assays.

The recombination reporter lines 1418 Col and 1445 Col were crossed to the sni1 npr1 rad51d triple mutant containing the BGL2:GUS transgene. F₂ plants containing the recombination reporters were selected by growth on MS media containing 10 μg/ml hygromycin. The presence of the sni1, npr1, and rad51d mutations was determined by CAPS analysis as described in SI Text. The absence of the BGL2:GUS transgene was determined by PCR using the primers NptIIIF (CTTGGGTGGAGAGGCTATTC) and NptIIIR (CTTGAGCCTGGCGAACAGTT) and confirmed in subsequent generations by growth on MS media containing 50 μg/ml kanamycin. The following lines were identified: sni1 1445, rad51d 1445, and sni1 rad51d 1445. The same mutants were generated with the 1418 reporter (data not shown). Only lines homozygous for the recombination reporter transgene and the mutations of interest and lacking the BGL2:GUS transgene were used for recombination assays. For analysis of recombination WT, sni1, and sni1 rad51d plants were grown on MS media with or without 50 mM INA for 2 weeks. Recombination frequencies were determined by histochemical staining for GUS activity, as described (18). The data were analyzed by Student's t test. The experiments were repeated at least three times with similar results.