The cenH3 assembly factor ZmKNL2 boosts haploid induction in maize

Haploid induction is one of the most widely used breeding techniques in the process of germplasm improvement in maize, and several haploid induction genes such as MTL/ZmPLA1/NLD (Gilles et al., 2017; Kelliher et al., 2017; Liu et al., 2017), ZmDMP (Zhong et al., 2019), ZmPLD3 (Li et al., 2021) and ZmPOD65 (Jiang et al., 2022) have been used in doubled haploid (DH) breeding. In addition to these traditional haploid inducer systems, the production of haploids through centromeric histone H3 (CenH3)-mediated genome elimination has also been promising (Ravi and Chan, 2010). However, due to the complexity of obtaining cenh3 null mutants (Lv et al., 2020; Wang et al., 2021), the application of the centromere-mediated haploid induction system was still limited in crops. A simple strategy for engineering cenH3 will be beneficial to the application of this high-frequency haploid induction system in crops, as well as elucidating the molecular mechanism of haploid induction. KNL2 has been reported to play a key regulatory role in CENH3 deposition at centromeres (Lermontova et al., 2013). In Arabidopsis, the loss of function of KNL2 resulted in the reduced deposition of CENH3 at centromeres, implying that KNL2 affects the function of CENH3 during centromere assembly. Here, we found that the knockout of ZmKNL2, the cenH3 assembly factor in maize, could trigger haploids when used as either the male parent or the female parent during hybridization. Furthermore, the mutation of ZmKNL2 enhanced the HIR by twofold in the presence of zmpld3 (0.96% HIR) or mtl/zmpla1/nld (1.19% HIR). Meanwhile, the similar HI ability of zmknl2 and atknl2 (Ahmadli et al., 2022) indicated the conservation of KNL2 function across eukaryotes, and it would be worthwhile to extend the application of KNL2 to other monocots and dicots. Firstly, we found that there was only one homologous gene in maize by the BLAST search of the full-length amino acid sequence encoded by AtKNL2 in Arabidopsis, and named it ZmKNL2 for further research. The domain analysis showed that ZmKNL2 also had an N-terminal SANTA domain (Figure 1a; Figure S1), which was the most typical feature of KNL2 proteins in different species (Lermontova et al., 2013). We used the clustered regularly interspaced short palindromic repeats/CRISPR-associated protein 9 (CRISPR-Cas9) system to knock out ZmKNL2 in maize LH244 background. Two zmknl2 mutants without the Cas9 element were identified, named zmknl2-1 and zmknl2-2. Both mutants caused frameshift and premature translation termination (Figure 1b; Figure S2). When compared with the wild-type plant, the zmknl2 homozygous mutants did not show any obvious differences in plant phenotypes from the seedling stage to the pollen shedding stage (Figure S3a,b). However, we found the self-pollinated ears of the zmknl2 homozygous mutants showed kernel abortion (Figure S3c), implying the knockout of ZmKNL2 might induce haploid production. As the cenh3 mutant could induce haploids by crossing the wild-type plant as the male parent or the female parent (Marimuthu et al., 2021; Wang et al., 2021), we made the reciprocal crosses between the zmknl2 mutants and the tester ZD958 (A commercially available hybrid which was a non-inducer). After pollination, the harvested ears displayed significant kernel abortion (Figure 1c,d,f). We then screened for haploids in the hybrid progeny via the detection of polymorphic molecular markers, the flow cytometry for ploidy identification and phenotypic observation (Figure S4). The results showed that the zmknl2 mutants had the ability to induce haploids when used as a male or female parent (Figure 1e; Table S1). Meanwhile, there was no significant difference in HIR between zmknl2-1 and zmknl2-2. To investigate the interactions of ZmKNL2 and other HI genes ZmPLD3, MTL/ZmPLA1/NLD and ZmDMP, we crossed the zmknl2-1 to the mutants of ZmPLD3, MTL/ZmPLA1/NLD and ZmDMP, respectively, to generate the double mutants. After the self-fertilization of heterozygous mutants, the screened homozygous mutants zmpld3-zmknl2, mtl-zmknl2 and zmdmp-zmknl2 were used as male parents to pollinate ZD958. The hybrid ears of zmpld3-zmknl2 and mtl-zmknl2 had significantly more aborted kernels and a lower seed setting rate than those of zmknl2-1, whereas no differences were observed between zmdmp-zmknl2 and zmknl2-1 (Figure 1g,i,j). Kernels from the hybrid ears were germinated to examine the HIR of these double mutants. The corresponding results showed that the average HIRs of zmpld3-zmknl2 and mtl-zmknl2 were 2.17% and 2.37%, respectively (Figure 1h; Table S1), which were significantly higher than the single mutant zmknl2-1 and zmpld3 or mtl/zmpla1/nld (Li et al., 2021). While the average HIR of zmdmp-zmknl2 was 0.26% (Table S1), which was no significant difference compared with that of zmknl2-1 and zmdmp (Li et al., 2021). As zmknl2 could enhance HIR in the presence of zmpld3 or mtl/zmpla1/nld, it was speculated that the combination of these three mutants might further increase the HIR. Statistical results showed that the average HIR of zmpld3-mtl-zmknl2 was 4.94%, which was significantly higher than that of zmpld3-mtl, zmpld3-zmknl2 and mtl-zmknl2 (Figure 1h; Figure S5; Table S1). Although the KI staining for pollen fertility showed no significant difference between these mutants and the WT (Figure S6), the synergistic effect suggested that there might be intercommunication in the regulatory mechanisms of these two HI systems. More data would be needed to explore whether the increased frequency of genome elimination triggered by abnormal cenH3 deposition, phospholipase-caused ROS burst and affected membrane electrostatic properties of the pollen endo-plasma membrane (endo-PM) could be the reason for the synergistic effect between zmknl2 and zmpld3 or mtl/zmpla1/nld. The identification of the HI ability of zmknl2 provided a new approach to the proteins related to cenH3 assembly for the deployment of the centromere-mediated HI system, and the synergistic effect between zmknl2 and zmpld3 or mtl/zmpla1/nld further expanded its application in creating new haploid inducers with higher HIR, as the homozygous mutant zmpld3-mtl-zmknl2 was acquirable which meant its practicality and convenience in application. This work was supported by the National Natural Science Foundation of China (32271541; 32272143; 32325041; 32401901; 62031003). The authors declare no competing interests. H.Z., W.S., J.L., Y.L. and M.L. designed the experiments. Y.L., M.L., H.G. and Z.L. performed the experiments. Y.L., M.L., F.X., Y.B. and L.E. analysed the data. Y.L., M.L., H.Z., W.S. and J.L. wrote the manuscript. The data that support the findings of this study are available in the supplementary material of this article. Figure S1 Alignment of KNL2 orthologues in Zea mays and Arabidopsis thaliana. Figure S2 Expression profiles of ZmKNL2. Figure S3 Predicted protein sequence of ZmKNL2 in wild type (WT) and mutants (zmknl2-1, zmknl2-2). Figure S4 Morphological phenotypes of seedlings, mature plants and self-cross ears of wild type (WT) and mutants (zmknl2-1, zmknl2-2). Figure S5 The EnAR and seed setting rate of the progeny of reciprocal cross between mutations (zmknl2-1, zmknl2-2) and ZD958. Figure S6 Verification of the haploid phenotype. Figure S7 Pollen viability of ZmKNL2-related mutants compared to WT. Table S1 HIR data of plants with homozygous mutations in ZmKNL2, ZmPLD3, MTL/ZmPLA1/NLD and ZmDMP. Table S2 Primers used in this study. 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.