Haploid induction (HI) is the critical first step in doubled haploid (DH) technology and plays an essential role in engineering synthetic apomixis. The gene MATRILINEAL (MTL), also referred to as NOT LIKE DAD and PHOSPHOLIPASE A1, is an important gene responsible for HI. MTL encodes a pollen-specific phospholipase and was first reported in maize (Gilles et al., 2017; Kelliher et al., 2017; Liu et al., 2017). It exhibits specific expression patterns during the later stages of pollen development, beginning from the second mitosis (Gilles et al., 2021). A 4-bp insertion in MTL has been found to be responsible for HI in maize, which significantly contributes to the utilization of DH technology in maize breeding programs (Kelliher et al., 2017; Jacquier et al., 2020). Chromosome fragmentation that occurs postmeiosis in the gametophyte has been shown to cause MTL-induced HI (Li et al., 2017), possibly due to an abnormal burst of reactive oxygen species (ROS) during later pollen development (Jiang et al., 2022). MTL shows high conservation among cereals. Mutations in rice OsMTL can induce haploids at rates of 1–15% (Yao et al., 2018; Liu et al., 2024) and have been utilized to engineer synthetic apomixis for fixing heterosis in hybrid rice (Wang et al., 2019; Li et al., 2023). The infrequent occurrence of aneuploids observed within Fix synthetic apomixis progeny may also be attributed to the loss-of-function of OsMTL (Liu et al., 2023). Despite these advancements made through plant research, the complete understanding of MTL's roles remains elusive. This study reveals that disruption of OsMTL leads to increased seed abortion rates and triggers triploid embryos formation in rice. Previously, we successfully edited the OsMTL gene in hybrid rice and observed that the Osmtl mutants exhibited a HIR ranging from 2.5% to 7.8%, along with a seed setting rate of 11.5% upon self-pollination (Wang et al., 2019). Subsequently, further investigations demonstrated that the Osmtl mutants produced numerous aborted seeds (Supporting Information Fig. S1). However, due to genetic segregation within the hybrid background, it is not feasible to evaluate the impact of OsMTL gene mutation on seed formation using progenies derived from these Osmtl mutants. Therefore, we reintroduced the CRISPR/Cas9 vector targeting OsMTL into the hybrid rice variety ChunYou84 (CY84) and obtained three new Osmtl mutation lines (Fig. S2). The seeds of these Osmtl mutants as well as wild-type (WT) CY84 were examined and classified into four distinct types based on their morphology: Em-En type (intact embryo and endosperm), em-En type (lacking embryo but containing endosperm), Em-en type (lacking endosperm but containing embryo), and em-en type (lacking both embryo and endosperm) (Fig. 1a). In WT plants, Em-En and em-en types accounted for 86.5% and 13.2% in average, respectively, while em-En and Em-en types were present at very low percentages (0.1% and 0.2%, respectively). By contrast, the Osmtl mutants showed a marked increase in em-En (1.1%, P < 0.001) and Em-en types (5.0%, P < 0.001), with an average of 13.4% Em-En and 80.5% em-en types (Fig. 1b; Table S1). These results indicate that knockout of the OsMTL gene in rice leads to increased rates of seed abortion, particularly inducing a high proportion of Em-en type seeds. Given the increased proportion of Em-en type seeds observed in Osmtl mutants and the absence of endosperms, we wondered whether this type of seed could germinate into plants. First, we assessed the components and vitality of the Em-en type seeds by using iodine–iodide kalium (1% I2-KI) and 2,3,5-triphenyltetrazolium chloride (0.1% TTC) staining. Compared with the normal Em-En type seeds, in which the endosperm appeared black under I2-KI staining and the embryo appeared red under TTC staining, the em-En type seeds only showed black staining, and the Em-en seeds only exhibited red staining (Fig. 1c,d). These results demonstrate that Em-en type seeds possess viable embryos but lack endosperms. Since the endosperm in angiosperms plays a critical role in nurturing the embryo and providing nutrients for seed germination, we subsequently tested the germination capacity of all four seed types on ½-strength Murashige Table S3). We then cultivated these different ploidy plants in the field and observed that the triploid plants displayed a larger size compared with diploid plants at the heading stage with enlarged glumes and elongated awns but showed pollen sterility (Figs 1h, S5). Subsequently, we explored the mechanism underlying triploid formation. Given that OsMTL is highly and specifically expressed during anther development (Fig. S6), and its orthologs function in pollen mitosis II (a postmeiotic mitotic division), we reasoned that the defect likely resides in the male gametogenesis. Triploids can arise from the fusion of an unreduced (2n) male gamete with a reduced (n) female gamete (Toda Mirzaghaderi Table S4). These increased bicellular pollen with unreduced 2n sperm cells might be the cause of triploid formation. Therefore, we conducted the reciprocal crosses between Osmtl mutant and the WT CY84. However, the reciprocal cross between Osmtl and WT produced only Em-En type seeds, with no Em-en type seeds observed. Notably, all 680 Em-En type seeds obtained from the cross Osmtl (♀) × WT (♂) were diploid, whereas among the 743 hybrid seeds from WT (♀) × Osmtl (♂), 26 were haploid and none were triploid (Table S5). In addition, we conducted another cross between Osmtl and HI285 (a high-efficiency Osmtl-based haploid inducer derived from a distinct genetic background, Liu et al., 2024). After extensive hybridization work, we obtained Em-en type seeds, but only five in number, and still failed to identify triploid individuals (Fig. S8). Given the infrequent occurrence of Em-en type seeds from crosses and the low proportion of triploids observed in Osmtl self-pollinated seeds, we turned to alternative methods to verify the hypothesis that 2n sperm cells lead to triploid formation. In a heterozygous genetic background, unreduced sperm cells generated from meiosis restitution and postmeiotic mitosis restitution will yield triploid offspring with distinct genotypes upon self-fertilization. Here, the inter-subspecific hybrid rice CY84, derived from a cross between the japonica male-sterile line 16A and the indica-japonica intermediate line C84, was used as the background material, providing abundant polymorphic markers for genetic analysis (Wang et al., 2019). Based on this, we performed subgenomic single nucleotide polymorphism (SNP) genotype proportion analysis (SSPA) on the four triploid lines, using the polymorphic SNPs between parental genomes 16A and C84, with two diploid lines (CY84OsMTL-F2-2X and CY84Osmtl-T1-2X) as controls (Methods S1). We found that the distribution of subgenomic allele ratios in all four triploids was exclusively consistent with the PMR model (Figs S9–S14). Complementing this, a genome-wide genetic recombination analysis revealed that each triploid line contained 19–31 crossovers (COs), a range not significantly different from that of the diploid controls (22 and 29 COs, respectively; Fig. 1n; Table S6). These results, taken together, demonstrate that two of the three triploid genomic sets originate from a single pollen grain, conclusively implicating unreduced 2n sperm cell in Osmtl mutants as likely being produced through the PMR process. Additionally, the fusion of a 2n sperm cell with the central cell could potentially generate a tetraploid (4n) endosperm in the Osmtl mutants. We further used SNP markers to assess and verify endosperm ploidy (Methods S1; Fig. S15). Nevertheless, neither diploid (2n) nor tetraploid (4n) endosperm was detected in either Em-En or em-En seeds (Table S7–S8). This absence is likely attributable to the stringent requirement for parental genomic dosage balance during endosperm cellularization (Lu et al., 2012; Butel et al., 2024). Previous study has shown that high-temperature treatment of poplar florets during the postmeiotic mitotic phase can induce triploid formation via PMR mechanism (Dong et al., 2014), and the high-temperature treatment is generally related to ROS burst (Zhao et al., 2023). To investigate whether elevated ROS levels occur in Osmtl, we performed H2DCFDA staining of mature pollen from WT and Osmtl at anthesis, with DAPI used to visualize nuclei. By analyzing c. 100 pollen grains from each material, we observed positive ROS fluorescence in both WT and Osmtl pollen, consistent with the metabolic activity reported in viable pollen (Luria et al., 2019). However, the average ROS fluorescence intensity in Osmtl pollen was significantly higher than in WT (P = 0.0003; Figs 1m,n, S16a,b), demonstrating that the OsMTL mutation leads to abnormal ROS accumulation. Further comparison between bi- and tricellular pollen revealed that bicellular pollen exhibited significantly higher ROS levels than tricellular pollen (P = 0.0039; Fig. S16a,c). These results indicate that mutation of the OsMTL gene leads to abnormal ROS levels in pollen at anthesis, including elevated average ROS levels in both tricellular and bicellular pollen. In summary, this study elucidates that loss of OsMTL function in rice concurrently induces haploid embryo production, increases seed abortion, and unexpectedly generates triploid embryos. Our findings indicate that MTL mutation triggers significant elevated ROS levels at anthesis in rice, which is consistent with earlier reports in maize and wheat (Jiang et al., 2022; Sun et al., 2022). This abnormal ROS burst likely promotes the production of bicellular pollen via the PMR process. Given that ROS act both as cytotoxins and as signaling molecules regulating diverse biological processes (Wang et al., 2024), we propose a mechanistic model in which OsMTL inactivation disrupts redox homeostasis during late pollen development. This disturbance may simultaneously induce chromosome fragmentation (mediating haploid induction) and impair sister chromatid separation, thereby activating PMR and generating specific bicellular pollen with one 2n sperm cell. Although this kind of bicellular pollen presents in both WT and Osmtl mature anthers, the observed triploid progeny in mutants may result from a competitive advantage of the elevated bicellular pollen with one 2n sperm cell within a viability-compromised tricellular pollen pool (Li et al., 2017). Subsequent fertilization of a reduced (n) egg cell by such an unreduced (2n) sperm cell would ultimately trigger a triploid (3n) embryo upon self-pollination (Fig. 1o). Future studies aimed at obtaining genetic materials that exclusively produce 2n sperm cells through PMR will be crucial to validate this model. Additionally, further investigations are warranted to determine whether Osmtl disrupts male–female signaling during pollination and to evaluate the potential utility of this system for triploid breeding. Collectively, our results advance the understanding of the multifaceted roles of MTL in plant reproduction and provide a new perspective on its functions beyond haploid induction. This work was supported by the National Natural Science Foundation of China (32025028, 32188102, and 32372177), the Earmarked Fund for CARS (CARS-01-10), and the Zhejiang Provincial Natural Science Foundation (LZYQ25C130002). We thank the Public Laboratory of China National Rice Research Institute for their technical support. We also thank DeepSeek for language polishing services and confirm that all revised content has been reviewed and approved by the authors. None declared. KW and CL managed the project. FH, CL and YL performed the experiments. FH, CL and TS analyzed the data. CL and FH wrote the manuscript. KW revised the manuscript. FH and CL contributed equally to this work. The whole-genome resequencing data have been deposited in the NCBI SRA database with the accession no. PRJNA1298227. All data supporting the conclusions drawn in our study are fully documented within this article and Supporting Information (including Figs S1–S16 and Tables S1–S8). Fig. S1 Characteristics of different seed types in Osmtl rice. Fig. S2 Genotypes of the Osmtl mutants used in this study. Fig. S3 Germination of different seed types on ½MS medium. Fig. S4 Chromosome integrity analysis of triploid rice. Fig. S5 Floret morphology and pollen fertility in diploid, haploid, and triploid rice. Fig. S6 Expression profiles of the OsMTL gene. Fig. S7 Scanning electron microscopy analysis of pollen from WT, Osmtl, and OsMiMe rice. Fig. S8 Types and proportion of seeds generated from Osmtl (♀) × HI285 (♂) cross. Fig. S9 Subgenomic SNP genotype proportion analysis (SSPA) of CY84OsMTL-F2-2X. Fig. S10 Subgenomic SNP genotype proportion analysis (SSPA) of CY84Osmtl-T1-2X. Fig. S11 Subgenomic SNP genotype proportion analysis (SSPA) of CY84Osmtl-T1-3X-1. Fig. S12 Subgenomic SNP genotype proportion analysis (SSPA) of CY84Osmtl-T1-3X-2. Fig. S13 Subgenomic SNP genotype proportion analysis (SSPA) of CY84Osmtl-T1-3X-3. Fig. S14 Subgenomic SNP genotype proportion analysis (SSPA) of CY84Osmtl-T1-3X-4. Fig. S15 Determination of endosperm ploidy through subgenomic SNP read count ratios. Fig. S16 Detection of reactive oxygen species (ROS) in the pollen of WT and Osmtl at anthesis. Methods S1 Materials and methods. Table S1 Quantification of the different seed types obtained by selfing in WT and Osmtl rice. Table S2 Ploidy determination of seedlings from Em-En seeds of WT and Osmtl rice. Table S3 Summary of whole genome sequencing (WGS) data for each sample. Table S4 Number of different types of pollen from WT and Osmtl rice at anthesis. Table S5 Ploidy determination of seedlings from different crosses. Table S6 Crossover (CO) number of diploids and triploids. Table S7 Ploidy analysis of endosperm from Em-En and em-En seeds using SNP genotyping. Table S8 Primer information of SNP markers used for endosperm ploidy determination. Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.
Hu et al. (Sun,) studied this question.