The rice (Oryza sativa L.) spontaneous mutant Ossds-1 was isolated from an indica rice, Zhongxian 3037. The F2 and F3 mapping populations were generated by crossing the Ossds-1^± heterozygous plants with a japonica cultivar, Zhonghua 11. The other two mutant alleles, Ossds-2 and Ossds-3, both were spontaneous mutants arose in tissue culture of Nipponbare. The meiotic mutant Osrad51c has been reported previously (Tang et al., 2014). The Ossds-1 Osrad51c double mutant was generated by crossing the two heterozygous Ossds-1^± and Osrad51c^±, and further identified from the F2 progeny. All plant materials were grown in paddy fields in the summer in Beijing or in the winter in Hainan.

Total 861 sterile plants segregated from the F2 and F3 mapping populations were used for isolation the target gene. Sequence-tagged site (STS) markers were developed according to sequence differences between the japonica variety Nipponbare and the indica variety 9311, using the data published on the NCBI website (http://www.ncbi.nlm.nih.gov). All primers are listed in Supplemental Table 1.

In the first exon, a 336-bp fragment of OsSDS cDNA sequence was chosen and amplified with the primers SDS-RNAi-F (adding a BamHI site) and SDS-RNAi-R (adding a SalI site) (Supplemental Table 1). RNAi vector construction and transformation were performed as described (Wang et al., 2009).

The complementary plasmid was constructed by cloning the 10.8 kb OSJNBa0081P02 genomic DNA fragment containing the entire OsSDS coding region into the pCAMBIA-1300 vector. A control plasmid, containing 7.8 kb of the truncated OsSDS gene was also constructed. Both of these plasmids were transformed into EHA105 and then into embryonic calli of OsSDS^± plants. The genotypes of the transgenic plants were further identify using the Primers SDS-JD (Supplemental Table 1).

Total RNA was extracted from the root, internode, leaf and panicle of Zhongxian 3037. Real-time PCR analysis was performed using the Bio-Rad CFX96 real-time PCR instrument (Bio-Rad, http://www.bio-rad.com/) and EvaGreen (Biotium, http://www.biotium.com/). The RT-PCR was carried out using the gene-specific primer pairs SDS-RT-F and SDS-RT-R for OsSDS. The primers Ubi-RT-F and Ubi-RT-R for ubiquitin were used as an internal control for the normalization of RNA sample. The results were analyzed using OPTICON MONITOR 3.1 (Bio-Rad). Each experiment had three replicates.

Total RNA was extracted from the panicle of Zhongxian 3037. The 3′ RACE and 5′ RACE were performed according to the protocol of the kit (3′-Full RACE Core Set and 5′ -Full RACE Core Set; Takara, http://www.takara-bio.com/). 3′ RACE was carried out using primers 3R-1F, 3R-2F, 3R-3F, and adaptor primer (P-ada). During 5′RACE, the RNA was reverse transcribed with 5′ (P)-labeled primer (SDS-4Rb); the first and second PCRs were performed using two sets of OsSDS specific primers (5R-1 and 5R-2). The 3′ RACE-PCR and 5′ RACE-PCR products were cloned and sequenced. OsSDS amino acid sequence translation and alignment were completed with the Vector NTI 11.5 (Invitrogen). OsSDS gene structure diagram was generated from GSDS (http://gsds.cbi.pku.edu.cn/index.php).

Young panicles with appropriate size of both Ossds mutants and wild type were collected, fixed in Carnoy's solution (ethanol:glacial acetic, 3:1) and stored at −20°C. Microsporocytes at the appropriate meiotic stage were squashed and stained with acetocarmine. After washing the chromosome preparations with 45% acetic acid and freezing them in liquid nitrogen, the coverslips were quickly removed with a razor blade and the slides harbored with samples were dehydrated through an ethanol series (70, 90, and 100%) for 5 min each and finally air-dried. Chromosomes on slides were counterstained with 4, 6-diamidinophenylindole (DAPI) in an anti-fade solution (Vector Laboratories, Burlingame, CA, USA). Finally, images were captured under the ZEISS A2 fluorescence microscope with a micro CCD camera (Zeiss, http://www.zeiss.de/en).

Fresh young panicles (40–60 mm) were fixed with 4% (w/v) paraformaldehyde for 30 min at room temperature. Anthers with appropriate stage were squashed into one drop of 1×PBS solution added on a slide. Then, covering the slide with a coverslip and pressing it with appropriate strength, the slide together with the coverslip was frozen thoroughly in liquid nitrogen. After quickly prizing up the coverslip, the slide was dehydrated through an ethanol series (70, 90, and 100%). The following immunolocalization procedure was performed as described (Tang et al., 2014).

The polyclonal antibodies against γ-H2AX, OsMSH5, OsREC8, PAIR2, PAIR3, ZEP1, OsMER3, and OsZIP4 used in this study have been described previously (Wang et al., 2009, 2010; Shao et al., 2011; Shen et al., 2012; Zhang et al., 2012; Luo et al., 2013; Miao et al., 2013).

We identified a spontaneous mutant exhibiting complete sterility in a rice field of Zhongxian 3037. From the heterozygous plant progeny related to this mutation, the normal fertile plants and the sterile plants were segregated in a 3:1 ratio, indicating that it was a single recessive mutation (χ² = 0.57; P > 0.05). The mutant plant was normal during vegetative growth and could not be distinguished from the wild type based on plant morphology (Supplemental Figure 1A). However, when come into reproductive stage, its spikelets exhibited complete sterility (Supplemental Figure 1B). So we further examined the mature pollen viability of the mutant by staining with 1% iodine potassium iodide solution (I₂-KI) (Supplemental Figures 1C,D). Only empty and shrunken pollen grains were observed in the mutant plant, indicating that microspores of the mutant are all abnormal and inviable. Moreover, when pollinated with wild-type pollens, the mutant spikelets still did not set any seeds, indicating that its female gametes were also affected.

To isolate the mutated gene, we constructed a population by crossing heterozygous plants with a japonica cultivar Zhonghua 11. A total of 861 sterile plants segregated from the F2 and F3 populations were used for mapping the target gene. Linkage analysis initially mapped the target gene onto the long arm of chromosome 3, which subsequently further delimited to a 130 kb region. Within this region, we found one candidate gene (Os03g12414) annotated as a putative cyclin with high similarity with SOLO DANCERS (SDS) gene in Arabidopsis. Thus, this candidate gene was chosen to be amplified and sequenced. Sequencing analysis revealed that there was a transversion happened from base C to T in the first exon, which produces a premature termination codon (TGA) and causes early termination (Figure 1A). We named the mutant here Ossds-1 and suspected the mutation of Os03g12414 leading to the sterile phenotype. We also obtained two other alleles arose from tissue culture of Nipponbare, named Ossds-2 and Ossds-3, respectively. They all showed the same defects as Ossds-1. Sequencing analysis showed that there were a transversion from base G to C at the third exon causing corresponding amino acid A replaced by P in Ossds-2 and two bases (GC) deletion at the fifth exon resulting in frame-shift mutation causing a premature termination codon (TGA) in Ossds-3 (Figure 1A). As earlier termination close to the start codon in Ossds-1 compared with the other mutants, it was selected for the subsequent experiments described blow. To further confirm the mutant phenotype was resulted by the mutation of OsSDS gene, RNA interference (RNAi) approach was carried out to down-regulate SDS in rice variety Yandao 8. We got 27 transgenic plants with 19 plants exhibited complete sterility. Additionally, the transformation of plasmid pCAMBIA-1300 containing the whole OsSDS gene was successful in rescuing the sterility of the mutant plants, just as respected. These results strongly confirmed that the mutation of OsSDS gene led to the sterile phenotype as described above.

Expression of OsSDS was also analyzed by real-time PCR (RT-PCR). As shown in Supplemental Figure 2, transcripts were all detected at root, internode, leaf and panicle, indicating that OsSDS is not a meiosis-specific gene.

The full-length cDNA of OsSDS gene was obtained by performing RT-PCR combined with 5′ and 3′ rapid amplification of cDNA ends PCR (RACE-PCR) using specific primers. We found that the OsSDS cDNA is comprised of 2786 bp containing an open reading frame (ORF) of 1410 bp and 1376 bp 5′ and 3′ untranslated regions (UTRs). The OsSDS cDNA sequence obtained is consistent with one mRNA (GenBank: AK065907.1) from the public network database (http://www.ncbi.nlm.nih.gov/). As shown in Figure 1A, the OsSDS gene contains seven exons and six introns. The predicted 469 amino acid protein of OsSDS shares as low as 30.6% identity with SDS in Arabidopsis, but with high similarity at the C-terminal (Figure 1B). Compared with the dicots Arabidopsis, OsSDS shares more than 60% identity with those in monocots, such as Zea mays, Sorghum bicolor, and Brachypodium. Conserved domain searching in NCBI revealed there are two conserved domains in SDS, namely, cyclin box fold domain (residues 276–366) and cyclin C-terminal domain (residues 378–419) (Figure 1B). Blast searching in NCBI revealed that SDS is a plant specific protein and owns one single copy in the plant kingdom.

To clarify whether the sterile phenotype of Ossds is caused by meiosis defects, chromosome behaviors at different stages of pollen mother cells (PMCs) from both wild type and Ossds-1 mutants were investigated. In wild type, chromosomes began to condense and became visible as thin thread-like structures at leptotene. After that, homologous chromosomes started to pair at zygotene. Fully synapsis between homologs formed at pachytene (Figure 2A). After further chromosome condensation, 12 bivalents were clearly observed at diakinesis (Figure 2B). Accompanying with the spindle installation, these bivalents aligned on the equatorial plate at metaphase I (Figure 2C). Then, homologous chromosomes separated and moved to the two opposite poles from anaphase I to telophase I (Figures 2D,E). During meiosis II, sister chromatids of each chromosome separated from each other and finally tetrads formed (Figure 2F).

In Ossds-1 PMCs, chromosome behaviors were similar to the wild type from leptotene to zygotene. However, obvious abnormities were first observed at pachytene stage, where the Ossds-1 mutant shows severely defects in homologous chromosome pairing and synapsis, and fully synapsed homologs were never observed (Figure 3A). Due to the lack of chromosome pairing, only univalents were observed at diakinesis (Figure 3B). During metaphase I, the univalents were unable to align on the equator plate (Figure 3C). From anaphase I to telophase I, they randomly segregated into two daughter cells (Figure 3D). In meiosis II, dyads and tetrads always exhibited different sizes caused by uneven chromosome segregation (Figures 3E,F). Cytological observation of meiocytes from the other two alleles, Ossds-2 and Ossds-3, as well as OsSDS RNAi plants showed the same meiotic defects as described in Ossds-1 (Supplemental Figure 3). Therefore, we proposed that the sterility of Ossds was caused by uneven segregation of homologous chromosomes without pairing and recombination during propose I.

The above cytological observation indicated Ossds showed similar defects with the loss of function of CRC1, a meiotic DSB formation protein in rice (Miao et al., 2013). We wondered whether OsSDS is also required for meiotic DSB formation. Phosphorylation of histone H2AX occurs within a few minutes after a DSB initiated in mitosis (Banath et al., 2010), and the same kind phosphorylation takes place during the meiotic DSB formation (Dickey et al., 2009). To verify this hypothesis, a dual-immunostaining experiment was carried out utilizing antibodies specific for γH2AX and OsREC8 raised from rabbit and mouse, respectively, in the meiocytes both from wild type and Ossds-1. OsREC8, one of the key cohesion proteins in rice meiosis, is localized onto meiotic chromosomes from leptotene to metaphase I (Wang et al., 2009). Here, we used it as a biomarker to indicate the rice meiotic chromosomes in the prophase I specifically. Results showed that numerous dot and patchy immunosignals of γH2AX were detected at zygotene in wild type (Figure 4A). However, no γH2AX immunosignals was detected in Ossds-1 meiocytes at the corresponding stage (Figure 4B), showing that OsSDS is essential for meiotic DSB formation.

During meiosis, all DSBs will be repaired by the repair system for maintaining genome stability. The loss function of repair proteins always results in chromosome fragmentation. OsRAD51C has been proved to be involved in meiotic DSB repair in rice (Tang et al., 2014). To verify this speculation, the Ossds-1 Osrad51c double mutant was generated by crossing the two heterozygous Ossds-1^± and Osrad51c^±, and further identified from their F2 progeny. In the Osrad51c mutant meiocytes, cytological observation shows 24 irregularly univalents appeared at diakinesis (Figure 5A). These univalents did not align well along the equatorial plate, and several chromosome fragments started to be shown at metaphase I (Figure 5B). Numerous chromosome fragments detained at the equatorial plate, while those massive chromosome bodies with centromeres moved into the two opposite poles at anaphase I (Figure 5C). While in the Ossds-1 Osrad51c meiocytes, chromosome behaviors were very much similar to Ossds-1 at the corresponding stages (Figures 5D–F). Together with the above γH2AX immunostaining data, we demonstrated very strong evidence that OsSDS is essential for DSB formation during rice meiosis.

In rice, there are three axis-associated proteins OsREC8, PAIR2 and PAIR3 have been reported playing important roles in SC assembly (Nonomura et al., 2006; Shao et al., 2011; Wang et al., 2011). Except for those axial elements, ZEP1, the central element of SC, has also been identified (Wang et al., 2010). To investigate what kind of SC installation defects happened due to the loss function of OsSDS, we conducted immunodetection experiments using antibodies against PAIR2, PAIR3, and ZEP1 in Ossds-1 microsporocytes. The results showed that OsREC8 was normally localized onto chromosomes in Ossds-1 meiocytes. Moreover, both PAIR2 and PAIR3 were overlapped very well with OsREC8 at zygotene, indicating their normal localization along the chromosome axis (Figures 6A,B). However, ZEP1 signals always appeared as short dots or discontinuous lines at early prophase I (Figure 6C), showing the deficient central element installation of SC in the mutant. Thus, the meiotic chromosome axes normally formed but SC installation failed in Ossds.

Meiotic homologous recombination finally falls into two recombination pathways by forming two type crossovers, interference-sensitive COs (class I) and interference-insensitive COs (class II). ZMM proteins are closely associated with class I crossovers formation (Chen et al., 2008; Shinohara et al., 2008). In rice, OsMSH5, OsMER3 and OsZEP4 are three ZMM proteins participating in the class I COs formation (Wang et al., 2009; Shen et al., 2012; Luo et al., 2013). To determine whether the defective OsSDS also affects the localization of OsMSH5, OsMER3, and OsZEP4, dual immunolocalization were carried out using antibodies against OsMSH5, OsMER3, and OsZEP4. In the wild-type microsporocytes, immunostaining experiments showed that OsMSH5, OsMER3, and OsZEP4 were observed as punctuate foci on chromosomes at zygotene (Figures 7A–C), and these foci persisted until late pachytene stage (Wang et al., 2009; Shen et al., 2012; Luo et al., 2013). However, no obvious signal foci of OsMSH5, OsMER3, and OsZEP4 were observed on chromosomes in Ossds-1 microsporocytes at the corresponding stage (Figures 7D–F), indicating that OsSDS is critical for the localization of OsMSH5, OsMER3 and OsZEP4.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.