DMC1 attenuates RAD51-mediated recombination in Arabidopsis

Abstract

Ensuring balanced distribution of chromosomes in gametes, meiotic recombination is essential for fertility in most sexually reproducing organisms. The repair of the programmed DNA double strand breaks that initiate meiotic recombination requires two DNA strand-exchange proteins, RAD51 and DMC1, to search for and invade an intact DNA molecule on the homologous chromosome. DMC1 is meiosis-specific, while RAD51 is essential for both mitotic and meiotic homologous recombination. DMC1 is the main catalytically active strand-exchange protein during meiosis, while this activity of RAD51 is downregulated. RAD51 is however an essential cofactor in meiosis, supporting the function of DMC1. This work presents a study of the mechanism(s) involved in this and our results point to DMC1 being, at least, a major actor in the meiotic suppression of the RAD51 strand-exchange activity in plants. Ectopic expression of DMC1 in somatic cells renders plants hypersensitive to DNA damage and specifically impairs RAD51-dependent homologous recombination. DNA damage-induced RAD51 focus formation in somatic cells is not however suppressed by ectopic expression of DMC1. Interestingly, DMC1 also forms damage-induced foci in these cells and we further show that the ability of DMC1 to prevent RAD51-mediated recombination is associated with local assembly of DMC1 at DNA breaks. In support of our hypothesis, expression of a dominant negative DMC1 protein in meiosis impairs RAD51-mediated DSB repair. We propose that DMC1 acts to prevent RAD51-mediated recombination in Arabidopsis and that this down-regulation requires local assembly of DMC1 nucleofilaments.

Fig 1. pRAD51::DMC1g restores fertility of the Arabidopsis dmc1 mutant and induces expression of DMC1 in somatic cells.

(A) Schematic representation of the pRAD51::DMC1g construct. DMC1g indicates DMC1 genomic sequence. Exons are shown as blue rectangles. (B) Wild-type plants have long siliques full of seeds (mean ± S.D: 54.1 ± 0.6 seeds per silique), while dmc1 mutants have very low fertility (mean ± S.D: 3.5 ± 0.25 seeds per silique). Expression of the pRAD51::DMC1g genomic sequence in dmc1 mutants restores fertility. (C) number of seeds per silique in DMC1, dmc1-/-, and three dmc1-/- + pRAD51::DMC1g independent transformants (T1-1, T1-2, and T1-3) showing restored fertility. Mean numbers are indicated ± S.D. n = 4 plants, N = 10 siliques per plant. (D) RT-PCR expression analysis of RAD51, DMC1, RAD54, HOP2 and MND1 in 7-day-old, untreated and MMC-treated (8 h, 20 μM) seedlings expressing or not the pRAD51::DMC1g transgene. Expression was analyzed in wild-type plants and three independent transgenic lines (T1-1, T1-2, and T1-3). Actin is used as a loading control.

Fig 2. Mitomycin C hypersensitivity of transgenic seedlings expressing pRAD51::DMC1g.

(A) No growth difference is observed between WT, RAD51-GFP, RAD51+/- heterozygous and RAD51+/- expressing pRAD51::DMC1g seedlings (two-week-old seedlings grown without MMC). (B) Pictures of two-week-old seedlings grown with 20 μM MMC. WT seedlings (depicted with red square) do not exhibit MMC sensitivity, in contrast to RAD51-GFP seedlings (blue square) and progeny of RAD51+/- transgenic lines expressing pRAD51::DMC1g. Pictures of progeny of two independent RAD51+/- transgenic lines expressing pRAD51::DMC1g are shown (T1-1 and T1-2). As control, WT (red square) and RAD51-GFP plants (blue square) were also grown on each plate. (C) Sensitivity of the seedlings was scored after 2 weeks and the fractions of sensitive plants (plants with less than 4 true leaves) are shown (3 biological repeats, each with N > 45 seedlings). Three independent RAD51+/- heterozygous lines expressing pRAD51::DMC1g (T1-1,T1-2 and T1-3) were tested and all showed strong hypersensitivity to MMC.

Fig 3. DMC1 inhibits RAD51-dependent somatic homologous recombination.

(A) Schematic map of the IU.GUS chromosomal recombination reporter locus. Restoration of the functional GUS gene is visualized as blue spots in seedlings. Both spontaneous (B) and MMC induced (C) IU.GUS recombination rates are strongly reduced in wild-type seedlings expressing pRAD51::DMC1g. Mean numbers of GUS+ recombinant spots per plant ± standard errors of the means (SEM) are indicated for each genotype (n = 3, N > 45 seedlings). **** p < 0.0001 (Mann-Whitney test). The wild-type plants with and without pRAD51::DMC1g are sister plants from a transgenic line segregating for the pRAD51::DMC1g transgene. Analysis were performed using segregating plants from two independent transgenic lines (T1-2 and T1-3). (D) Schematic map of the GU.US chromosomal recombination assay and graph showing mean numbers of GUS+ recombinant spots per plant ± SEM. Three independent GU.US transformants expressing pRAD51::DMC1g (T1-1, T1-2, T1-3) and 2 WT (sister plants from a population segregating the pRAD51::DMC1g transgene) were analyzed (WT for T1-3 is the same as for T1-2). Three biological replicates, each containing at least 50 plants, were performed, except for the WT T1-1 for which only two replicates were performed. ns indicates p > 0.05 (Mann-Whitney test).

Fig 4. RAD51 foci in somatic cells of DMC1 overexpressing seedlings.

(A-C) Immunolocalization of RAD51 in root tip nuclei of wild-type plants expressing pRAD51::DMC1g, fixed just before (A), and 2 (B) or 8 (C) hours after treatment with 30 μM MMC. T1-1 and T1-2 are two independent transgenic lines expressing pRAD51::DMC1g. MMC-induced RAD51 foci are clearly visible in nuclei of the treated plants. DNA is stained with DAPI (blue) and RAD51 foci (detected using an antibody against RAD51) are colored in red. Images are collapsed Z-stack projections of a deconvoluted 3D image stack. (Scale Bars: 5 μm). (D) Quantification of RAD51 foci in root tip nuclei of wild-type and two independent transgenic lines expressing pRAD51::DMC1g (T1-1 and T1-2) before and after MMC treatment. Means ± s.e.m are indicated for each genotype. More than 200 nuclei from at least 3 seedlings were analyzed per genotype.

Fig 5. DMC1 foci are formed in somatic cells upon DNA damage.

(A-B) Immunolocalization of RAD51 and DMC1 in root tip nuclei of wild-type plants expressing pRAD51::DMC1g fixed 8 hours after treatment with 30 μM Mitomycin C. (B) is a magnification of the area framed in (A). DNA is stained with DAPI (blue), RAD51 foci are colored in red and DMC1 in green. Images are collapsed Z-stack projections of a deconvoluted 3D image stack. (Scale bars: 20 μm (A) and 5 μm (B)). (C) Example of RAD51 and DMC1 signal intensity distribution of the total amount of pixels through a section of the depicted stained nucleus (red line). RAD51 is shown in red and DMC1 in green. (Scale bar: 5 μm). (D) Range of Pearson correlation coefficients (PCCs) of RAD51/DMC1-positive foci formed after 8 h of MMC treatment. RAD51/DMC1 partially co-localized in foci with a mean value of PCCs = 0.71 (n = 101).

Fig 6. DMC1-II3A variant does not form focus nor prevent RAD51-mediated recombination in somatic cells.

(A) DMC1g and DMC1-II3A protein are induced by MMC treatment. Total proteins were extracted from 1-week-old seedlings treated or not with 30 μM MMC for 8 hours and DMC1 abundance measured. No DMC1-specific band (37 kDa) is observed in untreated plants in both wild-type and transgenic plants. In contrast, while DMC1 is still absent in wild-type plants after DNA damage treatment, it becomes clearly visible in transgenic plants expressing either pDMC1::DMC1g or pDMC1::DMC1-II3A. (B) Immunolocalization of RAD51 and DMC1 in root tip nuclei of wild-type plants, transgenic plants expressing pRAD51::DMC1g or pRAD51::DMC1-II3A fixed 8 hours after treatment with 30 μM Mitomycin C. DNA is counterstained with DAPI (blue). Images are collapsed Z-stack projections of a deconvoluted 3D image stack. (Scale bar: 20 μm). (C) Expression of DMC1-II3A does not render plant hypersensitive to MMC nor (D) prevent RAD51-dependent somatic homologous recombination. (C) Sensitivity of seedlings was scored after 2 weeks of growth with 20 μM MMC and the fractions of sensitive plants (plants with less than 4 true leaves) are shown (3 biological repeats, each with N > 45 seedlings). Two independent transgenic lines expressing pRAD51::DMC1-II3A (T1-1 and T1-2) were tested and none exhibit MMC hypersensitivity. (D) Mean numbers of GUS+ recombinant spots per plant ± standard errors of the means (SEM) are indicated for each genotype. (p < 0.05; Mann-Whitney test). The wild-type plants with and without pRAD51::DMC1-II3A are sister plants from a transgenic line segregating for the pRAD51::DMC1-II3A transgene.

Fig 7. DMC1-II3A variant does not complement meiosis of dmc1 mutant and does not interfere with RAD51-mediated repair of meiotic DBS.

(A) Schematic representation of the pDMC1::DMC1-II3A construct. Exons are shown as blue rectangles. (B) Expression of the pDMC1::DMC1-II3A in dmc1 mutants do not restore fertility. Left panel: Wild-type plants have long siliques full of seeds, while dmc1 mutants have very low fertility. Right panel shows the number of seeds per silique in DMC1, dmc1 mutant and two dmc1 + pDMC1::DMC1-II3A independent transformants. Mean is shown as black horizontal bar; N = 20 to 25 siliques per plant. (C) DAPI staining of chromosomes during meiosis. Wild-type cells show pairing and synapsis of homologous chromosomes in pachytene, five bivalents at metaphase I and two groups of five chromosomes at anaphase I. dmc1 mutants exhibit defective synapsis but ten intact univalent at metaphase I, which randomly segregate at anaphase I. A similar phenotype is observed in dmc1 plants expressing pDMC1::DMC1-II3A. (Scale Bar: 10 μm). (D) Absence of DMC1 focus in transgenic plants expressing pDMC1::DMC1-II3A. Male meiocytes stained with ASY1 antibody (red) and DMC1 antibody (green). In wild-type meiocyte, numerous DMC1 foci are visible on chromosome axes. In contrast, dmc1 plants expressing pDMC1::DMC1-II3A lack DMC1 focus. (Scale bar: 5μm).

Fig 8. Expression of DMC1-GFP fusion protein inhibits RAD51-mediated repair of meiotic DBS.

(A) DAPI staining of chromosomes during meiosis. Wild-type cells show pairing and synapsis of homologous chromosomes in pachytene, five bivalents at metaphase I and two groups of five chromosomes at anaphase I. rad51 mutants exhibit defective synapsis and extensive chromosome fragmentation. Meiotic spreads in dmc1-/- RAD51+/+ plants expressing DMC1-GFP revealed a dmc1-like phenotype with defective synapsis but ten intact univalents at metaphase I, which randomly segregate at anaphase I. In contrast, extensive fragmentation is observed in both dmc1-/- RAD51+/- and DMC1+/+ RAD51+/- plants expressing DMC1-GFP. (Scale Bar: 10 μm). (B) Quantification of chromosome fragmentation. The genotype of the plants is indicated below each bar and number of cells analyzed is listed above.

Fig 9. Immunolocalization of DMC1 in pDMC1::DMC1-GFP expressing plants.

Male meiocytes were stained with DAPI (blue), anti-AtASY1 antibody (red), and anti-DMC1 antibody (green). Numerous DMC1 foci are visible on the chromosomes in WT plants expressing DMC1-GFP but not in dmc1-/-_DMC1-GFP. (Scale bar = 10 μm).

Fig 10. Hypothetical model for the activity of RAD51 and DMC1 nucleofilaments.

(A) During meiotic recombination, RAD51 and DMC1 polymerize on ssDNA to form nucleofilaments. DMC1 is the active nucleofilament but requires the presence of RAD51 nucleofilaments and additional co-factors. Presence of the DMC1 nucleofilament inactivates the RAD51 nucleofilament either through an unknown signaling pathway, through inhibiting interaction with co-factors or through nucleofilament remodeling. (B) In absence of RAD51, DMC1 does not form nucleofilaments and DSB repair is defective. (C) Absence of DMC1 allows normal polymerization of RAD51 and interaction with co-factors. RAD51 nucleofilaments are thus proficient for DSB repair using sister chromatid templates. This is reminiscent of the situation in mitosis (E) where RAD51 is the sole recombinase. Its polymerization and interaction with co-factors such as RAD54 lead to an active nucleofilament proficient for DSB repair. (D and F) Addition of non-functional DMC1 (e.g. DMC1-GFP) in (D) meiosis or (F) mitosis, leads to RAD51 down-regulation. In mitosis, DMC1 may be kept inactive presumably through absence of meiosis-specific co-factors or chromosomal features. We note that the spatial organization of recombinases and accessory factors depicted in the model is hypothetical and thus other organizations are possible (reviewer in [16,17,25]). In particular, although the symmetrical loading of RAD51 and DMC1 is currently favored, it has been formally demonstrated in yeast only and it is not yet clear whether it applies for other organisms. This is particularly true for Arabidopsis where asymmetric loading has previously been suggested [19].