OsDMC1 Is Not Required for Homologous Pairing in Rice Meiosis^,

OsDMC1, a conserved recombinase, maintains the stability of the single-end invasion process but is not required for homologous pairing in rice, which is quite different from the DMC1 homologs identified so far in other organisms.

Abstract

Meiotic homologous recombination is pivotal to sexual reproduction. DMC1, a conserved recombinase, is involved in directing single-end invasion between interhomologs during meiotic recombination. In this study, we identified OsDMC1A and OsDMC1B, two closely related proteins in rice (Oryza sativa) with high sequence similarity to DMC1 proteins from other species. Analysis of Osdmc1a and Osdmc1b Tos17 insertion mutants indicated that these genes are functionally redundant. Immunolocalization analysis revealed OsDMC1 foci occurred at leptotene, which disappeared from late pachytene chromosomes in wild-type meiocytes. According to cytological analyses, homologous pairing is accomplished in the Osdmc1a Osdmc1b double mutant, but synapsis is seriously disrupted. The reduced number of bivalents and abnormal OsHEI10 foci in Osdmc1a Osdmc1b establishes an essential role for OsDMC1 in crossover formation. In the absence of OsDMC1, early recombination events probably occur normally, leading to normal localization of γH2AX, PAIR3, OsMRE11, OsCOM1, and OsRAD51C. Moreover, OsDMC1 was not detected in pairing-defective mutants, such as pair2, pair3, Oscom1, and Osrad51c, while it was loaded onto meiotic chromosomes in zep1, Osmer3, Oszip4, and Oshei10. Taken together, these results suggest that during meiosis, OsDMC1 is dispensable for homologous pairing in rice, which is quite different from the DMC1 homologs identified so far in other organisms.

Meiosis involves two successive rounds of nuclear division combined with a single round of DNA replication (Petronczki et al., 2003). To ensure accurate chromosome segregation, crossovers (COs) must form between homologous chromosomes during meiosis I in most eukaryotes. COs result from meiotic homologous recombination (HR), which is initiated by a double-strand break (DSB) produced through the catalysis of conserved SPO11 proteins (Keeney et al., 1997; Youds and Boulton, 2011). In Saccharomyces cerevisiae and Schizosaccharomyces pombe, SPO11 proteins are covalently linked to the 5′ termini of a DSB and released together with a short DNA oligonucleotide. This process is mediated by the Mre11/Rad50/Xrs2-Nbs1 (MRX/N) complex and Sae2/Com1 (Neale et al., 2005; Uanschou et al., 2007; Milman et al., 2009). Then, the MRX/N complex and Exo1 resect the 5′ termini to produce the 3′ end of the DSB in yeast (Farah et al., 2009; Zakharyevich et al., 2010). The 3′ end is initially protected by the replication protein A complex, which is subsequently released when two strand-exchange proteins, Radiation sensitive51 (Rad51) and Disrupted Meiotic cDNA1 (Dmc1), are loaded onto the 3′ end in yeast and human (Fanning et al., 2006; Sung and Klein, 2006).

Rad51 and Dmc1 are two Escherichia coli RecA homologs in yeast, mouse, and human (Bishop et al., 1992; Shinohara et al., 1992; Habu et al., 1996). These proteins catalyze strand exchange, as demonstrated in vitro (Sung, 1994; Li et al., 1997; Sehorn et al., 2004). In most eukaryotes that have been investigated, Rad51 is present in both mitotic and meiotic cells, whereas Dmc1 is found specifically in meiotic cells (Neale and Keeney, 2006; Sung and Klein, 2006). In budding yeast, available evidence indicates that Rad51 activity is attenuated by Hed1 to facilitate interhomolog repair directed by Dmc1 (Tsubouchi and Roeder, 2006). Additionally, Rad51 and Dmc1 form nucleoprotein filaments on single-stranded DNA. These proteins then conduct homology searches and catalyze the formation of homologous joint molecules (JMs). Dmc1 directs JM formation between the interhomolog chromosome, with Rad51 acting as an accessory factor, suggesting that Dmc1 is specifically involved in ensuring the generation of COs. However, Rad51 can direct JM formation between intersister chromosomes in the event that Dmc1 fails to form JMs between interhomologs. Thus, Rad51 may play an important role in a fail-safe mechanism for JM formation in budding yeast (Cloud et al., 2012).

In S. cerevisiae, mutations in DMC1 lead to abnormal synapsis and meiotic arrest (Bishop et al., 1992). In mouse, dmc1^−/− knockouts are sterile due to asynapsis or nonhomologous synapsis in spermatocytes (Pittman et al., 1998; Yoshida et al., 1998). Mutation of AtDMC1 results in asynapsis and random chromosome segregation during meiosis I in Arabidopsis thaliana but does not cause meiotic arrest (Couteau et al., 1999). AtDMC1 promotes interhomolog-biased DSB repair governed by the axial element protein ASY1 (Sanchez-Moran et al., 2007). Furthermore, ATR kinase is involved in regulating AtDMC1 accumulation at meiotic DSB sites. The elimination of ATR allows AtDMC1 to mediate meiotic DSB repair, even in the absence of AtRAD51 (Kurzbauer et al., 2012).

OsDMC1A and OsDMC1B are two conserved rice (Oryza sativa) DMC1 genes with highly similar exonic sequences (Kathiresan et al., 2002). Whether they play functionally redundant roles in meiosis has not yet been confirmed. Moreover, their roles in meiosis remain to be explored. pair2, pair3, Oscom1, or Osrad51c mutants in rice exhibit serious pairing defects; these mutants are referred to as asynaptic mutants (Wang et al., 2011; Ji et al., 2012; Tang et al., 2014). By contrast, zep1, Osmer3, Oszip4, and Oshei10 do not exhibit defective pairing (Wang et al., 2009, 2012; Wang et al., 2010; Shen et al., 2012). Analysis of these mutants has provided clues about the actual role of rice DMC1 in meiosis.

Here, we characterized OsDMC1A and OsDMC1B using a reverse genetic approach, demonstrating their functional redundancy in meiosis. Cytological analysis revealed serious CO formation defects in the Osdmc1a Osdmc1b double mutant. However, homologous pairing was probably accomplished in this mutant. The depletion of OsRAD51C in the Osdmc1a Osdmc1b background resulted in deficient pairing and nonhomologous associations, suggesting that OsRAD51C is not entirely epistatic to OsDMC1. Furthermore, a series of immunolocalization experiments revealed that OsDMC1 was depleted in pair2, pair3, Oscom1, and Osrad51c. By contrast, this protein was loaded normally onto meiotic chromosomes in zep1, Osmer3, Oszip4, and Oshei10. These results imply that rice DMC1 plays a role in meiosis, which has yet to be fully elucidated.

RESULTS

Characterization of OsDMC1A and OsDMC1B

We used the Arabidopsis DMC1 amino acid sequence as a query in a BLAST search against the rice proteome (Klimyuk and Jones, 1997), yielding a list of 53 proteins, which were ranked according to their BLAST scores. Two of these proteins share the highest similarity with AtDMC1 according to BLAST analysis. These proteins, which are encoded by Os12g0143800 and Os11g0146800, were designated OsDMC1A and OsDMC1B, respectively. Both OsDMC1A and OsDMC1B contain 344 amino acid residues and share 99% identity in amino acid sequence. These proteins were collectively designated as OsDMC1 in this study. Multiple sequence alignment of OsDMC1 with its orthologs revealed its high evolutionary conservation in eukaryotes, especially in monocots (Supplemental Fig. S1).

Two Tos17 insertion lines, NF8016 (Osdmc1a) and NE1040 (Osdmc1b), were identified from the public rice insertion line database (Hirochika et al., 1996; Yamazaki et al., 2001). The insertion site within locus OsDMC1A was mapped to exon 10 and that of OsDMC1B to intron 11 (Supplemental Fig. S2). Both Osdmc1a and Osdmc1b exhibited normal vegetative growth and fertility (Fig. 1, A–C). To determine whether these genes are functionally redundant, we generated an Osdmc1a Osdmc1b double mutant by crossing homozygous Osdmc1a and Osdmc1b single mutant plants. The Osdmc1a Osdmc1b plants exhibited wild-type growth and development but were completely sterile (Fig. 1D). Pollen grains from Osdmc1a, Osdmc1b, and wild-type plants were round, while those from Osdmc1a Osdmc1b plants were empty and shrunken (Fig. 1, E–H). In addition, we calculated the rate of seed-setting in these lines. Unlike the 91.25% seed-setting rate in the wild type, 90.16% in Osdmc1a and 89.67% in Osdmc1b, the seed-setting rate was 0% in Osdmc1a Osdmc1b. Thus, OsDMC1A and OsDMC1B may function redundantly in the process of rice reproduction.

Figure 1.

Phenotypic characterization of the Osdmc1a OsDMC1B, OsDMC1A Osdmc1b, and Osdmc1a Osdmc1b mutants. A to D, Comparison of wild-type, Osdmc1a OsDMC1B, OsDMC1A Osdmc1b, and Osdmc1a Osdmc1b mutant plants. E to H, Pollen grains stained with 1% I₂-KI solution in wild-type, Osdmc1a OsDMC1B, OsDMC1A Osdmc1b, and Osdmc1a Osdmc1b plants. Bars = 50 μm.

Real-time RT-PCR revealed that the expression of OsDMC1 was highest in young panicles of the wild type. However, OsDMC1 expression was extremely low in leaves and roots. Moreover, its expression level was near zero in young panicles of Osdmc1a Osdmc1b, suggesting that the aberrant OsDMC1 transcripts were degraded in the double mutant (Supplemental Fig. S3A). An antibody against OsDMC1 was raised in mouse using the conserved sequence between OsDMC1A and OsDMC1B. To detect the specificity of anti-OsDMC1, we performed a western-blot assay. The OsDMC1 antibody could clearly recognize the recombinant protein OsDMC1B with the His tag and OsDMC1 protein in crude extracts from the young panicles of the wild type, Osdmc1a, and Osdmc1b, but no signal was detected in the protein sample of Osdmc1a Osdmc1b (Supplemental Fig. S3B). Thus, anti-OsDMC1 could specifically recognize the native OsDMC1.

OsDMC1 Is Loaded onto Chromosomes during Early Prophase I

To define the spatial and temporal distribution of OsDMC1 during meiosis in rice, we conducted a dual immunolocalization assay in wild-type meiocytes using specific polyclonal antibodies against OsREC8 and OsDMC1 individually. OsREC8, a conserved meiosis-specific component of the cohesion complex, can be used as a marker for meiotic chromosomes (Shao et al., 2011). In the wild type, OsDMC1 signals initially appeared as punctate foci on chromosomes at leptotene (Fig. 2A), gradually appearing as dense dot-like signals on zygotene chromosomes (Fig. 2B). During early pachytene, the signals began to attenuate (Fig. 2C). They rapidly diminished at pachytene (Fig. 2D) and were absent from late pachytene chromosome axes (Fig. 2E). At zygotene, the localization of OsDMC1 in Osdmc1a and Osdmc1b meiocytes was similar to that of the wild type (Supplemental Fig. S4, A and B). Furthermore, no OsDMC1 signals were detected in Osdmc1a Osdmc1b meiocytes (Supplemental Fig. S4C), indicating that Osdmc1a Osdmc1b is a null mutant.

Figure 2.

Dual immunolocalization of OsREC8 and OsDMC1 during prophase I in wild-type meiocytes. A, Leptotene; B, zygotene; C, early pachytene; D, pachytene; E, late pachytene. OsREC8 signals (red) indicate meiotic chromosome axes. OsDMC1 signals are in green. Bars = 5 μm.

The Meiotic Process Is Disturbed in Osdmc1a Osdmc1b Meiocytes

To clarify the cause of sterility in Osdmc1a Osdmc1b, we investigated the chromosome behavior in wild-type and Osdmc1a Osdmc1b meiocytes via 4′,6-diamino-phenylindole (DAPI) staining. In the wild type, meiotic chromosomes appeared as thin threads at leptotene (Supplemental Fig. S5A). The chromosomes further condensed and initiated homologous pairing and synapsis during zygotene (Supplemental Fig. S5B). At pachytene, the chromosomes appeared as thick threads when the synaptonemal complex (SC) was fully assembled between homologous chromosomes (Supplemental Fig. S5C). Paired chromosomes underwent further condensation at diakinesis; 12 bivalents connected by chiasmata were observed at this stage (Supplemental Fig. S5D). At metaphase I, 12 extremely condensed bivalents aligned in an orderly manner on the equatorial plate (Supplemental Fig. S5E). During anaphase I, homologous chromosomes separated and migrated to opposite poles of the cell (Supplemental Fig. S5F). Regular dyads then formed at telophase I (Supplemental Fig. S5G). In the second meiotic division, sister chromatids separated, ultimately giving rise to tetrad microspores (Supplemental Fig. S5, H and I). In Osdmc1a and Osdmc1b, a similar meiotic process was observed, and 12 bivalents could be observed from diakinesis to metaphase I (Supplemental Fig. S6, A–D).

In Osdmc1a Osdmc1b, meiotic chromosomes behaved normally from leptotene to zygotene (Fig. 3, A and B). Almost all homologous chromosomes became aligned well (indicated with arrows), but some regions of the chromosomes could not complete synapsis at pachytene. The empty or bubble-like regions (indicated arrowheads) were observed between each pair of aligned chromosomes (Fig. 3C). Unlike the wild type, which formed 12 bivalents, ∼24 condensed univalent chromosomes were scattered in the mutant nuclei at diakinesis (Fig. 3D). Subsequently, the univalent chromosomes unequally separated to opposite poles of the cell at late metaphase I. Meanwhile, approximately two bivalents aligned to the equatorial plate in some cells (Fig. 3E). The bivalents were divided into two poles at anaphase I (Fig. 3F). At telophase I, unbalanced dyads formed (Fig. 3G). During the second division, different numbers of chromatids were aligned to the two metaphase II equatorial plates (Fig. 3H). Abnormal tetrads with microspores of different sizes were detected after meiosis II (Fig. 3I). These results suggest that meiotic defects lead to the sterility of Osdmc1a Osdmc1b.

Figure 3.

Meiosis in the Osdmc1a Osdmc1b double mutant. A, Leptotene; B, zygotene; C, pachytene. Some fine alignment regions are marked with arrows. The poor alignment regions are indicated with arrowheads. D, Diakinesis; E, metaphase I; F, anaphase I; G, telophase I; H, metaphase II; I, telophase II. Chromosomes were stained with DAPI. Bars = 5 μm.

OsDMC1 Is Not Required for Meiotic DSB Formation

H2AX is an isoform of histone H2A. Upon DSB formation, phosphorylated H2AX (γH2AX) rapidly accumulates around the break site, making γH2AX a useful cytological marker for detecting DSB formation in most eukaryotes (Dickey et al., 2009). Therefore, we conducted dual immunolocalization in wild-type and Osdmc1a Osdmc1b meiocytes using polyclonal antibodies against γH2AX and OsREC8. In wild-type meiocytes, γH2AX signals appeared as dots and patchy areas at zygotene (Supplemental Fig. S7A). In Osdmc1a Osdmc1b, the pattern of γH2AX signals was similar to that of the wild type at a similar stage (Supplemental Fig. S7B). Additionally, to determine whether the DSB number was affected in Osdmc1a Osdmc1b, the intensity of γH2AX signals was detected using IPLab4.0 in wild-type and mutant meiocytes individually. Compared with the wild type, no obvious difference was found (Supplemental Fig. S8), suggesting that DSB formation may occur as normal in the mutant.

DSB End Resection May Occur in Osdmc1a Osdmc1b

Since DSBs formed in Osdmc1a Osdmc1b, we next explored whether DSB end processing was impaired in this mutant. OsCOM1 and OsMRE11 are essential for DSB end resection in rice. OsCOM1 and OsMRE11 foci appear as dense, dot-like signals on chromosomes during early prophase I (Ji et al., 2012, 2013). We therefore conducted immunolocalization analyses of Osdmc1a Osdmc1b meiocytes using OsCOM1 and OsMRE11 polyclonal antibodies. Both of these proteins were loaded onto chromosomes normally during early meiosis (Fig. 4, A and B), indicating that OsCOM1 and OsMRE11 act independently of OsDMC1 during early recombination. Additionally, normal dense dot-like OsRAD51C signals appeared on chromosome axes in Osdmc1a Osdmc1b during zygotene (Fig. 4C), suggesting that DSB end resection can occur in Osdmc1a Osdmc1b.

Figure 4.

Immunolocalization of several meiotic proteins in Osdmc1a Osdmc1b meiocytes. A to C, Immunolocalization of OsCOM1 (A), OsMRE11 (B), and OsRAD51C (C) at zygotene. D, Immunolocalization of PAIR3 at pachytene. OsREC8 signals are in red. All the other protein immunosignals are in green. Bars = 5 μm.

OsDMC1 Is Vital to Synapsis, But Not to Homologous Pairing

We also monitored the assembly status of the SC in Osdmc1a Osdmc1b meiocytes via immunolocalization studies using antibodies against PAIR3, PAIR2, and ZEP1. PAIR3, an axis-associated protein, is essential for homologous pairing (Wang et al., 2010). In Osdmc1a Osdmc1b, the localization of PAIR3 was not affected from leptotene to pachytene, implying its normal installation on chromosome axes independent of OsDMC1 (Fig. 4D). In the wild type, PAIR2 staining showed punctate signals at leptotene. PAIR2 proteins became increasingly depleted from the synaptic chromosomes during zygotene and completely disappeared from chromosomes after the synapsis was completed at pachytene (Wang et al., 2011). ZEP1 can be used to monitor synapsis in rice. The signal of ZEP1 was first detected as punctate foci at leptotene. Then, ZEP1 gradually elongated and formed short linear signals during zygotene. At pachytene, elongated ZEP1 signals appear along paired homologous chromosomes (Wang et al., 2010). Similar to that in the wild type, PAIR2 and ZEP1 proteins were normally loaded on leptotene chromosomes in Osdmc1a Osdmc1b (Fig. 5A). However, during zygotene to pachytene, their localization appeared abnormal. ZEP1 appeared as only a few short lines on chromosomes, and PAIR2 was retained as a linear signal. Upon careful examination, we found that the PAIR2 signals disappeared on regions where ZEP1 was loaded (Fig. 5, B and C, indicated with arrows). Taken together, these results indicate that synapsis is disrupted due to the aberrant functioning of OsDMC1.

Figure 5.

Synapsis is disturbed in Osdmc1a Osdmc1b. A to C, Immunolocalization of PAIR3 (blue), PAIR2 (red), and ZEP1 (green) in Osdmc1a Osdmc1b meiocytes at leptotene (A), zygotene (B), and pachytene (C). Arrows indicate regions where PAIR2 disappeared as ZEP1 is being loaded. Bars = 5 μm.

The bouquet stage occurs during prophase I. At this stage, telomeres attach to the inner nuclear envelope and form a cluster. The bouquet is thought to facilitate homologous pairing (Harper et al., 2004; Zickler, 2006). To monitor bouquet formation during meiocytes, we performed fluorescence in situ hybridization (FISH) analysis in wild-type and Osdmc1a Osdmc1b meiocytes using pAtT4 as a probe, which recognizes telomere-specific sequences (Zhang et al., 2005). In both wild-type and Osdmc1a Osdmc1b meiocytes, telomeres were clustered within a confined region at early zygotene (Fig. 6, A and B), suggesting that bouquet formation occurs independently of OsDMC1.

Figure 6.

OsDMC1 is not essential for bouquet formation and homologous pairing. A and B, Bouquet formation analysis using FISH with the telomere-specific pAtT4 probe in the wild type (A) and Osdmc1a Osdmc1b (B). C to H, Analysis of the chromosome pairing by FISH using probes 5S rDNA, a0018B15, a0083M01, a0088I16, and a0002E05. BAC clones at early pachytene in the wild type (C, E, and G) and Osdmc1a Osdmc1b (D, F, and H). I and J, Analysis of the bivalent formation at metaphase I by FISH using the 5S rDNA probe in the wild type (I) and Osdmc1a Osdmc1b (J). The inset in (J) shows the magnified image of the framed region. Chromosomes (blue) were stained with DAPI. pAtT4, 5S rDNA, a0083M01, and a0002E05 FISH signals are in red. a0018B15 and a0088I16 FISH signals are in green. Bars = 5 μm.

To determine chromosome pairing behavior in Osdmc1a Osdmc1b meiocytes, we first performed FISH analysis on both wild-type and mutant meiocytes using 5s rDNA as a probe; 5s rDNA is a tandemly repetitive sequence that only exists on the short arm of chromosome 11 close to the centromere in rice. During early pachytene, two side-by-side 5S rDNA signal were observed in both the wild type and Osdmc1a Osdmc1b, indicating that homologous pairing might occur at this stage (Fig. 6, C and D). At metaphase I, two obvious 5S rDNA signals were detected on one pair of homologous chromosomes in the wild type (Fig. 6I). By contrast, there were a few bivalents in most Osdmc1a Osdmc1b meiocytes, ∼9.76% (n = 41) of which contained 5S rDNA signals on one bivalent at the corresponding stages (Fig. 6J). Thus, bivalents might occur between homologs in Osdmc1a Osdmc1b.

The pairing status of chromosomes at early pachytene was further investigated by the chromosome-specific FISH with four BAC clone probes, a0018B15, a0083M01, a0088I16, and a0002E05. These four probes all locate on chromosome 11, of which a0018B15 and a0088I16 are on the short arm and are detected as green FISH signals; a0083M01 and a0002E05 are on the long arm and are detected as red FISH signals. In wild-type meiocytes (n = 25), we observed two adjacent green or red signals at early pachytene, indicating the perfect pairing loci of the chromosome arm (Fig. 6, E and G). Similar FISH signals were also observed in most Osdmc1a Osdmc1b meiocytes (n = 53) during the corresponding stage (Fig. 6, F and H). We found that nearly 92.45% of the mutant meiocytes exhibited two closely adjacent green signals when using a0018B15 as a probe of the short arm. Meanwhile, about 88.68% of the mutant cells presented two side-by-side red signals when using a0083M01 as a probe of the long arm (Fig. 6F). Similar results were also obtained using other two BAC probes a0088I16 and a0002E05 (Fig. 6H). Therefore, the homolog pairing is probably achieved independently of OsDMC1.

Crossover Formation Is Basically Suppressed in Osdmc1a Osdmc1b

COs, which are necessary for the formation of stable bivalents, appear as cytological chiasmata at diakinesis. Chiasmata exhibit the same molecular events as COs and can be used to count the number of COs per cell (Jones, 1984). From diakinesis to metaphase I, the main cytological defect in Osdmc1a Osdmc1b was the presence of abundant univalents. However, a few bivalents were occasionally observed during metaphase I (Fig. 3E). In addition, we calculated the rate of bivalent generating in Osdmc1a Osdmc1b. Unlike the 100% bivalent rate in the wild type (n = 146), about 60.15% of observed Osdmc1a Osdmc1b meiocytes (n = 133) produced one to two bivalents at metaphase I (Supplemental Fig. S9). The mean number of bivalents was highly reduced to 1.83 per cell (n = 133) in Osdmc1a Osdmc1b compared with 12 per cell in the wild type. These results imply that OsDMC1 is necessary for the formation of most crossovers.

OsMER3, OsZIP4, and OsHEI10 are ZMM proteins in rice whose defects lead to dramatically reduced numbers of COs. In wild-type meiocytes, OsMER3, OsZIP4, and OsHEI10 produce dense, dot-like signals at early zygotene. The number of OsMER3 and OsZIP4 foci rapidly decreases at late zygotene, with OsHEI10 signals ultimately restricted to prominent foci specifically localized to chiasma sites at late pachytene (Wang et al., 2009, 2012; Shen et al., 2012). In this study, we carried out immunolocalization studies in Osdmc1a Osdmc1b using antibodies against OsMER3, OsZIP4, and OsHEI10. We did not detect obvious differences in the localizations of OsMER3, OsZIP4, and OsHEI10 at zygotene (Fig. 7, A–C). As meiosis progressed, the distribution of OsHEI10 began to change. At late pachytene, the number of prominent, dot-like signals was dramatically reduced in Osdmc1a Osdmc1b compared with the wild type (mean, 23.79; n = 30), and most OsHEI10 signals appeared as sparser foci at this stage (Fig. 7D; mean, 4.87; n = 21). Thus, the aberrant functioning of OsHEI10 might cause the dramatic reduction in the number of COs in Osdmc1a Osdmc1b.

Figure 7.

Immunolocalization of several ZMM proteins in Osdmc1a Osdmc1b meiocytes. A, OsMER3 (green) in a zygotene meiocyte. B, OsZIP4 (green) in a zygotene meiocyte. C, OsHEI10 (green) in a zygotene meiocyte. D, OsHEI10 (green) in a late pachytene meiocyte. PAIR2 signals (red) indicate meiotic chromosome axes. Bars = 5 μm.

The Loading of OsDMC1 May Depend on Factors Necessary for Homologous Pairing

To further explore the role of OsDMC1 in meiosis, we performed immunolocalization studies (using a specific OsDMC1 polyclonal antibody) in pair2, pair3, Oscom1, OsMRE11^RNAi, Osrad51c, zep1, Osmer3, Oszip4, and Oshei10. pair2, pair3, Oscom1, OsMRE11^RNAi, and Osrad51c exhibit defects in homologous pairing (Wang et al., 2011; Ji et al., 2012, 2013; Tang et al., 2014). zep1, Osmer3, Oszip4, and Oshei10 display normal homologous pairing (Wang et al., 2009, 2012; Wang et al., 2010; Shen et al., 2012). OsDMC1 was never detected on zygotene chromosomes in pair2, pair3, Oscom1, OsMRE11^RNAi, and Osrad51c (Fig. 8, A–E), while its localization was not affected at zygotene in zep1, Osmer3, Oszip4, and Oshei10 (Fig. 8, F–I). These results suggest that the function of OsDMC1 depends on meiotic pairing factors.

Figure 8.

Immunolocalization of OsDMC1 in different meiotic mutants. A to I, OsDMC1 signals (green) were detected in pair2 (A), pair3 (B), Oscom1 (C), OsMRE11^RNAi (D), Osrad51c (E), zep1 (F), Osmer3 (G), Oszip4 (H), and Oshei10 (I). OsREC8 signals (red) indicate meiotic chromosome axes. Bars = 5 μm.

OsRAD51C is required for homologous pairing and meiotic DSB repair in rice (Tang et al., 2014). The triple mutant Osdmc1a Osdmc1b Osrad51c was generated by introducing Osdmc1a Osdmc1b into the Osrad51c background. In the triple mutant, the meiotic defects of Osdmc1a Osdmc1b were greatly exaggerated by the presence of Osrad51c, leading to the serious pairing defect at pachytene (Fig. 9, A and E), irregularly shaped univalents at diakinesis (Fig. 9, B and F), and chromosome entanglements at metaphase I (Fig. 9, C and G). Compared with Osrad51c, Osdmc1a Osdmc1b Osrad51c contained chromosome bridges with fewer fragments at anaphase I (Fig. 9, D and H). FISH analysis with the 5s rDNA probe was also carried out in Osdmc1a Osdmc1b Osrad51c. At pachytene, two separated 5s rDNA signals were observed (Supplemental Fig. S10A). At metaphase I, two signals were still separated and observed locating on chromosome entanglements (Supplemental Fig. S10B). In addition, we also determined the intensity of γH2AX signals in the triple mutant. Compared with the wild type, no obvious difference was detected at zygotene in the triple mutant (Supplemental Fig. S8). Taken together, the failure pairing and serious nonhomologous associations occurred, while the DSB formation was not impaired in the triple mutant.

Figure 9.

Abnormal meiotic chromosome behaviors in Osrad51c and Osdmc1a Osdmc1b Osrad51c at pachytene (A and E), diakinesis (B and F), metaphase I (C and G), and anaphase I (D and H). Chromosomes were stained with DAPI. Bars = 5 μm.

DISCUSSION

OsDMC1 Is Not Required for Homologous Pairing, But It Functions in SC Assembly

In this study, we determined that OsDMC1A and OsDMC1B are functionally redundant during meiosis. It is generally believed that bouquet formation is necessary for homologous pairing (Scherthan, 2001). In Osdmc1a Osdmc1b, bouquet formation was similar to that of the wild type, implying that early chromosome alignment does not depend on OsDMC1. Unlike Atdmc1 (Pradillo et al., 2012; Da Ines et al., 2012), mutation of OsDMC1 did not lead to serious pairing defects. In the earlier work, the RNAi to OsDMC1 led to homolog pairing defect (Deng and Wang, 2007). This defect observed in OsDMC1-RNAi lines might be caused by nonspecific knockdown of other genes.

In many eukaryotes, recombination-dependent homologous recognition is an important component of the homologous pairing mechanism. This process relies on DNA/DNA interchromosomal interactions that occur as a result of the initial steps in Spo11-induced DSB repair. After the 5′ DNA ends of DSBs are resected, the resulting 3′ single-stranded DNA can invade the intact DNA duplex of a homologous chromosome, which is catalyzed by Rad51 and Dmc1 recombinases, generating intermediates capable of assessing homology (Naranjo, 2012). Eukaryotic Rad51 and Dmc1 can stabilize strand exchange intermediates in precise three-nucleotide steps. Triplet recognition strictly depends on correct Watson-Crick pairing. Rad51 and Dmc1 can both step over mismatches, but only Dmc1 can stabilize mismatched triplets (Lee et al., 2015). Programmed DSB formation and its end resection probably occur in Osdmc1a Osdmc1b, as γH2AX, OsCOM1, and OsMRE11 were regularly detected at early prophase I. Introducing Osrad51c into the Osdmc1a Osdmc1b background led to serious pairing defects, suggesting that OsRAD51C plays a role in recombination-dependent pairing processes. Taken together, we suspect that other RecA proteins, including OsRAD51C, may serve as directors of homology searches to promote pairing in rice, while OsDMC1 probably helps stabilize the single-end invasion process.

OsDMC1 is not required for the recombination-dependent pairing process, but it does play a role in synapsis. PAIR3, an axial element-associated protein of the SC, is required for synapsis in rice (Wang et al., 2011). In Osdmc1a Osdmc1b, PAIR3 signals were normal during early meiosis I. However, signals from ZEP1, a transverse filament protein of the SC, did not coincide with chromosomes at pachytene in Osdmc1a Osdmc1b. These results suggest that OsDMC1 functions in chromosome synapsis in rice.

Functional Divergence in RAD51-Like Genes in Rice

In addition to RAD51 and DMC1, rice possesses five RAD51 paralogs, including OsRAD51B, OsRAD51C, OsRAD51D, OsXRCC2, and OsXRCC3 (Lin et al., 2006). Of these, mutations in OsRAD51C, OsRAD51D, or OsXRCC3 disturb homologous pairing, causing the production of large chromosome fragments at metaphase I. However, the roles of OsRAD51B and OsXRCC2 in rice during meiosis have not been demonstrated (Byun and Kim, 2014; Tang et al., 2014; Zhang et al., 2015). In Arabidopsis, mutations in AtRAD51C or AtXRCC3 lead to the presence of entangled chromosomes interconnected by chromatin bridges, while AtRAD51B, AtRAD51D, and AtXRCC2 have not been proven to participate in meiosis. These proteins have partially redundant functions in mitotic DNA repair (Pradillo et al., 2014). It seems that the functional divergence also exists among rice RAD51 paralogs.

In budding yeast, Rad51 directs intersister JM formation, representing an important fail-safe mechanism when Dmc1-dependent interhomolog JM formation fails (Cloud et al., 2012). This mechanism allows meiotic DSBs to be repaired, but it ultimately leads to homologous pairing defects. By contrast, we determined that the disruption of OsDMC1 did not affect homologous pairing, but it led to the formation of almost 24 univalents. Therefore, OsRAD51 may direct the meiotic repair of DSBs using the intersister as a template in Osdmc1a Osdmc1b.

Abnormal DSB Repair Mechanisms Might Be Activated in Osdmc1a Osdmc1b Osrad51c

HR, nonhomologous end joining (NHEJ) and ectopic recombination (ER) may be involved in DSBs repair. However, meiotic DSBs are mainly repaired through HR because ER or NHEJ pathways would result in error-prone chromosome associations. NHEJ and ER are usually repressed during meiosis unless HR process is disturbed (Goedecke et al., 1999; Goldman and Lichten, 2000). The NHEJ pathway joins double-strand DNA ends, leading to chromosome associations. Considering aberrant chromosome entanglements in Osdmc1a Osdmc1b Osrad51c, the NHEJ pathway might be activated following loss of OsDMC1 and OsRAD51C. However, it is still not enough to explain why chromosome entanglements occurred in this triple mutant because this kind of defect can also be caused by abnormal ER. ER is often eliminated by recombination-dependent pairing process. In budding yeast, the strand exchange capacity of Rad51 is shut down to facilitate Dmc1-mediated interhomolog recombination (Tsubouchi and Roeder, 2006; Busygina et al., 2012). Moreover, a dmc1 mutation increased unequal sister chromatid recombination (Grushcow et al., 1999; Thompson and Stahl, 1999). Additionally, Rad51 and Dmc1 suppress meiotic ectopic recombination (Shinohara and Shinohara, 2013). We propose that OsRAD51C, together with OsDMC1, might be involved in suppressing ER in meiosis, which could explain why chromosome entanglements were produced in Osdmc1a Osdmc1b Osrad51c, but not in Osdmc1a Osdmc1b or Osrad51c. Therefore, we suspect OsDMC1 plays a duel role in meiotic recombination. This protein may not only be essential for interhomolog recombination, but also be required for the suppression of abnormal DSB repair pathways such as ER and NHEJ when OsRAD51C is nonfunctional.

CO Formation Is Strongly Disrupted in the Absence of OsDMC1

ZMM proteins in all organisms investigated to date play a role in crossover formation, including ZIP1, ZIP2, ZIP3, ZIP4, MSH4, MSH5, and MER3 in budding yeast (Lynn et al., 2007). In this study, COs were almost absent in Osdmc1a Osdmc1b, although the normal localization of OsMER3 and OsZIP4 to meiotic chromosomes occurred, suggesting that the recruitment of both proteins is not dependent on OsDMC1. Other rice RAD51-like proteins may act in recruiting these proteins during meiosis. In meiotic mouse cells, RAD51 interacts with MSH4 (Neyton et al., 2004). In addition, RAD51 and ZIP3 directly interact in yeast (Agarwal and Roeder, 2000). OsHEI10 may specifically promote the early selection of recombination intermediates to become class I COs (Wang et al., 2012). OsDMC1 facilitates the retention of OsHEI10 on meiotic chromosomes, as the localization of OsHEI10 during the later stages of meiosis is disturbed in Osdmc1a Osdmc1b. It is likely that the production of recombination intermediates is depressed in the absence of OsDMC1, ultimately leading to defective class I CO formation.

The results of this study provide new insight into the role of OsDMC1 during rice meiosis. Our findings reveal that OsDMC1 is not required for homologous pairing in rice. However, the specific role of OsDMC1 in suppressing NHEJ and ER requires further study.

MATERIALS AND METHODS

Plant Materials

The rice (Oryza sativa) japonica cultivar Nipponbare was used as the wild type in our study. Two Tos17 insertion lines NF8016 and NE1040 were named as Osdmc1a and Osdmc1b separately. Their seeds were kindly provided by the Rice Genome Resource Center of the National Institute of Agrobiological Sciences. Other mutant materials of the pair2, pair3, Oscom1, OsMRE11^RNAi, Osrad51c, zep1, Oszip4, Osmer3, and Oshei10 were from our previous study. All plant materials were grown in paddy fields. The double mutant Osdmc1a Osdmc1b was made by crossing homozygotes of appropriate mutant lines. The triple mutant Osrad51c Osdmc1a Osdmc1b was generated from crossing the Osdmc1a/+ Osdmc1b/+ double heterozygote and the Osrad51c/+ single heterozygote. The genotypes were identified by PCR to the F2 population derived from selfing F1 plants heterozygous for alleles.

Tos17 Insertion Site Analysis

The Tos17 inserted region of NF8016 was amplified by using a pair of primers TOSP12F (ATTGTTAGGTTGCAAGTTAGTTAAGA) and TOS12R (TCAGGTTCATTTCCGACACA). That of NE1040 was done by primer pairs TOS11F (ATTGTTAGGTTGCAAGTTAGTTAAGA) and TOS11R (CTTTGCCTTTCCTCAGCATC). The amplification products from NF8016 and NE1040 were cloned into pMD18-T vector (Takara) and sequenced separately.

Real-Time PCR for Transcript Expression Analysis

Total RNA was separately extracted from the leaf, root, and panicle of Nipponbare as well as panicle of Osdmc1a Osdmc1b. The Bio-Rad CFX96 real-time PCR instrument was used to perform real-time PCR analysis. EvaGreen was used as the fluorescent dye (Biotium). The real-time PCR was performed with the specific primers DMC1-RT1F (GTCCAAGCAGTACGACGAAG) and DMC1-RT1R (TCTCCAGAGTTTATCCCTTGC) for OsDMC1 as well as ActinF (CTGACAGGATGAGCAAGGAG) and ActinR (GGCAATCCACATCTGCTGGA) for ACTIN. Each experiment was replicated three times. The experimental results were analyzed by Bio-Rad CFX Manager analysis software.

Antibody Production

Total RNAs were extracted from wild-type young (30 to 50 mm) panicles using the TRIzol reagent (Invitrogen). After being treated by DNase I (Invitrogen), they were reversely transcribed to synthesize cDNA using the oligo(dT) primer and Superscript III kit (Invitrogen). A 528-bp fragment of OsDMC1B coding region (GenBank accession number AB079874) was amplified by PCR primer pairs BF with a BamHI site (AGGATCCATGGCGCCGTCCAAGCAG) and BR with an XhoI site (ACTCGAGTCGTTCAGGCCGGAATGTT). The PCR product was then cloned into the pMD18-T vector (Takara) and reinserted into the pGEX4T-2 vector (Amersham). The GST fusion OsDMC1B peptide containing amino acids residues 1 to 176 was expressed and purified (Wang et al., 2009). The polyclonal antibody was obtained with the fusion peptide immunizing a mouse.

Western-Blot Assay

Proteins were extracted from rice panicles with buffer solution containing 50 mm Tris-HCl (pH 7.5), 150 mm NaCl, 0.5% NP40, and 1 mm phenylmethylsulfonyl fluoride. Insoluble materials were removed by centrifugation. The OsDMC1B coding region was cloned into the pET30a vector (Novagen) and purified using a Ni²⁺-charged column (Qiagen). Protein samples were separated by SDS-PAGE on a 12% polyacrylamide gel and electroblotted onto polyvinylidene difluoride membrane (Amersham). Western blots were performed with OsDMC1 antibodies diluted 1/10,000 and anti-mouse IgG antibodies conjugated to horseradish peroxidase (Abcam) diluted 1/20,000. Signals were detected after adding Immobilon ECL (Millipore).

Immunofluorescence Analysis

Fresh young panicles were first fixed in 4% (w/v) paraformaldehyde for 30 min at room temperature. Chromosome preparation and immunofluorescence analysis were conducted according to previous descriptions (Wang et al., 2009). The immunofluorescence analysis used the following primary antibodies: rabbit antibodies to OsREC8 (Shao et al., 2011), PAIR2 (Wang et al., 2011), and mouse antibodies to OsREC8, PAIR3, OsCOM1, OsMRE11, OsRAD51C, ZEP1, OsZIP4, OsMER3, and OsHEI10 individually (Wang et al., 2009, 2011, 2012; Wang et al., 2010; Ji et al., 2012, 2013; Shen et al., 2012; Tang et al., 2014). All above antibodies were previously generated in our laboratory. Additionally, a 561-bp fragment of PAIR3 cDNA (amino acids 248 to 393) was inserted into the vector pGEX4T-2. The guinea pig antibody to PAIR3 was generated with the PAIR3 fusion protein. The second antibodies, including rhodamine-conjugated goat anti-rabbit antibody, fluorescein isothiocyanate-conjugated sheep anti-mouse antibody, and AMCA-conjugated goat anti-guinea pig antibody were used for fluorescence detection.

Cytological Procedures and Data Analysis

Young panicles containing pollen mother cells entering meiosis were fixed in Carnoy’s solution (ethanol:glacial acetic acid, 3:1) and stored at –20°C to be used in the following chromosome spreading. Pollen mother cells at the meiotic stage were squashed on slides and stained with acetocarmine. Coverslips were then covered on the slides. The slides were frequently frozen in liquid nitrogen. They were dehydrated through an ethanol series (70, 90, and 100%) after coverslips were removed. DAPI in an antifade solution (Vector Laboratories) is used to counterstain chromosomes on the slides (Wang et al., 2009).

Fluorescence in Situ Hybridization

FISH analysis was conducted according to Zhang et al. (2005). The pAtT4 clone contains telomeric repeats. The pTa794 clone has 5S rRNA genes from wheat (Triticum aestivum; Cuadrado and Jouve, 1994). Four BAC clone on chromosome 11, a0018B15, a0083M01, a0088I16, and a0002E05, were also used as probes to monitor the chromosome arm pairing. a0018B15 and a0088I16 are located on the short arm. a0083M01 and a0002E05are located on the long arm. Chromosomes were counterstained with DAPI. Original images were observed under Zeiss A2 fluorescence microscope and captured with a DVC1412 CCD camera using software IPLab4.0.

Computational and Database Analysis

Protein sequence similarity searches were performed in the NCBI (www.ncbi.nlm.nih.gov/BLAST). Gene structure schematic diagrams were generated by GSDS (http://gsds.cbi.pku.edu.cn/index.php). The multiple sequence alignment diagram was generated in GenDoc software. All pictures were further modified by Adobe Photoshop CS3.

Accession Numbers

The sequences used for alignment analysis can be found in the GenBank/EMBL data libraries under the following accession numbers: OsDMC1A, AAM76791; OsDMC1B, NP_001065738; ZmDMC1 (Zea mays), AEQ16523; AtDMC1 (Arabidopsis thaliana), AEE76687; PtDMC1 (Populus trichocarpa), EEF00835; ScDMC1 (Saccharomyces cerevisiae), AAB64706; SpDMC1 (Schizosaccharomyces pombe), BAA28671; and MmDMC1 (Mus musculus), BAA10969.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Alignment of different DMC1 orthologs from Oryza sativa, Zea mays, Arabidopsis thaliana, Populus trichocarpa, Saccharomyces cerevisiae, Schizosaccharomyces pombe, and Mus musculus.

• Supplemental Figure S2. Gene structures of OsDMC1A and OsDMC1B.

• Supplemental Figure S3. Osdmc1a Osdmc1b is a null double mutant.

• Supplemental Figure S4. Immunolocalization of OsDMC1 at zygotene in Osdmc1a OsDMC1B, OsDMC1A Osdmc1b, and Osdmc1a Osdmc1b meiocytes.

• Supplemental Figure S5. Meiosis in wild-type meiocytes.

• Supplemental Figure S6. The meiotic behavior in Osdmc1a OsDMC1B and OsDMC1A Osdmc1b meiocytes.

• Supplemental Figure S7. The dual immunolocalization of OsREC8 and γH2AX at zygotene in wild-type and Osdmc1a Osdmc1b meiocytes.

• Supplemental Figure S8. The relative intensity of γH2AX staining signals at zygotene in the wild type, Osdmc1a Osdmc1b, and Osdmc1a Osdmc1b Osrad51c.

• Supplemental Figure S9. Quantification of bivalents at metaphase I in the wild type and Osdmc1a Osdmc1b.

• Supplemental Figure S10. FISH analysis with probe 5S rDNA in Osdmc1a Osdmc1b Osrad51c.

Glossary

CO

crossover

HR

homologous recombination

DSB

double-strand break

JM

joint molecule

SC

synaptonemal complex

DAPI

4′,6-diamino-phenylindole

FISH

fluorescence in situ hybridization

NHEJ

nonhomologous end joining

ER

ectopic recombination