The CRISPR-Cas9 vector for complete knockout of ZEP1 was constructed by the isocaudomer ligation method, as described in Wang C. et al. (2015). The annealed ZEP1 g++/ZEP1 g-- oligonucleotides (Supplementary Table 1) were ligated into the SK-sgRNA vector that digested with AarI. Then, the sgRNA of ZEP1 (digested with KpnI and BglII) was assembled into the pC1300-Actin:Cas9 binary vector (digested with KpnI and BamHI) to obtain the vector pC1300-Actin:Cas9-sgRNA^ZEP1 for generation of zep1 null mutants.

The binary vector pC1300-Actin:Cas9-sgRNA^ZEP1 was introduced into Agrobacterium tumefaciens (Strain EHA105). The Agrobacterium was then transformed into calluses derived from CY84 to generate transgenic lines. Fifteen independent transgenic plants of pC1300-Actin:Cas9-sgRNA^ZEP1 were obtained and grown in summer in the transgenic paddy fields of the China National Rice Research Institute in Hangzhou, China.

Genomic DNA of transgenic plants was extracted from approximately 50mg fresh leaf tissue via the cetyltrimethylammonium bromide (CTAB) method. PCR was performed with KOD FX DNA polymerase (Toyobo, Japan) to amplify the fragments surrounding ZEP1 target site. Primers ZEP1 F and ZEP1 R are listed in Supplementary Table 1. The mutants were detected according to the published Hi-TOM procedure (Liu et al., 2019).

Panicles at meiosis stage were harvested and fixed in Carnoy’s solution (ethanol:glacial acetic acid, 3:1) for more than 24h at room temperature. Microsporocytes were squashed on a slide with a needle and covered with a coverslip. Then, slides were frozen in liquid nitrogen, and the coverslips were removed rapidly with a blade. Chromosomes were stained with 4′,6-diamidino-2-phenylindole (DAPI) in an antifade solution (Vector Laboratories, Burlingame, CA, United States) as described in (Wang et al., 2009). Fluorescence microscopy was conducted using an Olympus BX61 fluorescence microscope fitted with a micro charge-coupled device camera.

Fresh young rice panicles were fixed in 4% (w/v) paraformaldehyde for 45min at room temperature. Anthers at proper meiosis stage were squashed with a needle in PBS solution and covered with a coverslip. After soaking in liquid nitrogen and removing the coverslip, slides were then incubated in a humid chamber at 37°C for 4h with anti-REC8 (Mouse) and anti-ZEP1 (rabbit) polyclonal antibodies (diluted 1:500 in TNB buffer: 0.1M Tris–HCl, pH 7.5, 0.15M NaCl, and 0.5% blocking reagent). After three rounds of washing in PBS, Texas red-conjugated goat anti-mouse antibody and fluorescein isothiocyanate-conjugated sheep anti-rabbit antibody (1:1,000) were added to the slides. The chromosomes were counterstained with DAPI in an antifade solution (Vector Laboratories, Burlingame, CA, United States). Fluorescence microscopy was performed as described above.

Single nucleotide polymorphisms (SNPs) markers were developed on the basis of the whole-genome sequences of C84 and chunjiang16A (Wang et al., 2019). The primers are listed in Supplementary Table 1. The Hi-TOM method was performed for the genotyping as previously described in (Liu et al., 2019).

The analysis of interference strength was described in Wang K. et al. (2015). For each pair of adjacent intervals, the coefficient of coincidence analysis compared the observed frequency of CO occurring in both intervals with the frequency expected if the CO occurred independently in the two intervals, with the interference strength calculated as (1-observed/expected). The expected number of chromosomes with CO occurring in both intervals (XY and YZ) was calculated as: (recombination frequency between X and Y)×(recombination frequency between Y and Z)×(number of chromosomes examined). See the details in Supplementary Tables 2 and 3.

The WT (CY84) and zep1-KO plants were crossed with the indica variety 93-11 to generate the segregation population. The hybrid seeds were soaked in deionized water at 37°C in the dark for 2days and then transferred to a net floating on deionized water for further 5days. The seedlings were cultured in a half-strength Kimura B nutrient solution (pH 5.4; Liu et al., 2020). The genotypes of each plant in the segregation population were identified by Hi-TOM. The genotypes of each female gamete of CY84 and zep1-KO plants were deduced and listed in Supplementary Tables 4–7. Then, they were used for the frequency and distribution of genetic recombination analysis. For each pair of adjacent markers (X and Y), the recombination frequency was calculated as: (number of chromosomes with one CO between X and Y/number of chromosomes examined).

ZEP1 has a full length of 9.478-kbp, distributed in 21 exons and 20 introns and encodes a transverse filament protein containing 869 amino acids (Figure 1A). ZEP1 comprises coiled-coil proteins with globular domains at its N termini from amino acids 1 to 63 and C termini from amino acids 714 to 869. The C-terminal globular domain of 156 residues putatively binds DNA, while the N-terminal globular domain of 63 residues is basic (Wang et al., 2010). The Tos17 insertion mutant lines of ZEP1 used in previous studies are partial loss-of-function mutants (Wang et al., 2010; Wang K. et al., 2015), which still contain the N-terminal residues. To generate zep1 knockout mutants, we employed the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) system to edit ZEP1 and chose the 19th to the 38th nucleotides in the first exon near the initiation codon as our target site (Figure 1A). We successfully constructed the CRISPR-Cas9 vector pC1300-Actin:Cas9-sgRNA^ZEP1 (Figure 1B) and then introduced it into the elite inter-subspecific hybrid rice variety Chunyou84 (CY84) via Agrobacterium. A total of fifteen independent transgenic plants were obtained in the T₀ generation with six homozygous mutations (#1, #4, #7, #9, #11, and #12) and five biallelic mutations (#2, #3, #6, #13, and #14; Figure 1C).

To clearly uncover the editing mutations of ZEP1 on genetic recombination frequency and genetic interference, we selected the 1-bp insertion homozygous mutants for further researches. As shown in Figure 2A, the 1-bp insertion caused premature termination of ZEP1 transcription and ultimately encoded a severely truncated protein containing 48 residues, of which only 12 residues were consistent with that of WT. Because of the basic of ZEP1 N-terminal globular domain, we speculated the 1-bp insertion mutants are ZEP1 knockout lines, which differ from the previous partial loss-of-function zep1 mutants (Wang et al., 2010; Wang K. et al., 2015). To confirm the absolute absence of ZEP1, we conducted immunostaining by using anti-REC8 and anti-ZEP1 antibodies. REC8 is a component of the cohesion complex and is required for sister chromatid cohesion, axial element formation, and homolog pairing and is used as a marker to monitor early meiotic events during prophase I here (Zhang et al., 2006). In the WT plants, ZEP1 signals were robust at pachytene, while in the mutants, no signals except the background were detected in any observed meiocytes (n=500; Figure 2B). These results implied that ZEP1 was completely knocked out in the mutants. Hereafter, we renamed these mutants as zep1-KO.

Fertility is an important agronomic trait for genetic recombination analysis and plant breeding (Wang K. et al., 2015). There were no discernible differences in plant morphology between the WT and the zep1-KO lines (Figure 3A), but the anthers of zep1-KO plants were light white while those of the WT were bright yellow (Figure 3B). We further confirmed the suspected male sterility by 1% I₂-KI solution staining and observed no staining in the pollens from the zep1-KO plants (Figure 3C). Therefore, unlike the previously described partial fertility of the zep1 mutants (Wang et al., 2010; Wang K. et al., 2015), complete knockout of ZEP1 caused absolute male sterility. In contrast to the anthers, the zep1-KO pistils seemed unaffected, similar to those of WT plants (Figure 3B). To determine whether the female gametes were fertile, we pollinated the zep1-KO and the control WT plants with pollen from WT plants. Both the WT and the zep1-KO plants could set seeds (Figure 3D), with a seed-setting rate of 38.1±5.8% and 34.4±8.5% for the WT and zep1-KO, respectively. These results suggested that the female meiocytes of zep1-KO mutants were not severely affected.

The complete male sterility in the zep1-KO plants might result from defects in meiosis. To test this hypothesis, we investigated the chromosome behaviors of male meiocytes by DAPI staining. In WT plants, all 12 bivalents in the pollen mother cells (PMCs) lined up on the equatorial plate at metaphase I (Figure 4A). However, univalents, bivalents, and multivalents were observed in the zep1-KO PMCs, numbering 4.40±1.96, 8.37±1.30, and 1.53±0.97 (n=30), respectively (Figure 4). Therefore, the presence of univalents and multivalents likely caused the male sterility of the zep1-KO lines.

To explore whether the female recombination frequency was affected by disruption of ZEP1, we firstly crossed the WT (the CY84 hybrid) and zep1-KO lines with the indica rice variety 93-11 and then examined the genetic recombination events of WT×93-11 and zep1-KO × 93-11 F₁ individuals via genotyping. To this end, we developed 10 nearly uniformly distributed single nucleotide polymorphism (SNP) markers along each of chromosomes 1 and 8 (Figure 5A) that were polymorphic between the parental lines of the CY84 hybrid. Then, we genotyped 364 individual F₁ progeny of WT×93-11 and 525 individual F₁ progeny of zep1-KO × 93-11 using those markers. After this, we tracked the recombination events that had occurred in the WT or zep1-KO parental plants (CY84 hybrid background) based on a previous method (Wang K. et al., 2015), using our SNP markers to deduce the genotype of each chromatid following meiosis and thus infer the number of COs that had occurred to produce each chromatid. In the WT plants, chromosome 1 chromatids with zero, single, double, triple, and quadruple COs occurred at frequencies of 20, 34, 34, 11, and 1%, respectively. By contrast, the zep1-KO mutants exhibited a decreased frequency of zero (8%), single (25%), and double (28%) COs but a higher frequency of triple (20%), quadruple (14%), quintuple (3%), sextuple (1%), and septuple (1%) COs (Figure 5B). Similar results were detected on chromosome 8, with 19% zero, 34% single 25% double, 16% triple, 4% quadruple, and 2% quintuple COs occurred in zep1-KO mutants, while the frequencies of zero, single, double, and triple COs occurred in WT plants were 23, 49, 25, and 2%, respectively (Figure 5C).

We further calculated the recombination frequency of zep1-KO and WT plants. The average number of COs per chromatid on chromosomes 1 and 8 in WT plants was 1.40 and 1.06, respectively, while that of zep1-KO mutants was significantly higher, with 2.33 and 1.57 COs per chromatid on chromosomes 1 and 8, respectively (Figure 6). The recombination frequency was elevated in all identified intervals in the zep1-KO lines compared to that of the WT, varying from 1.01- to 3.57-fold (Figure 6). In particular, the recombination frequency of the DE interval on chromosome 1 and the EF interval on chromosome 8, both of which span the centromere, was elevated by 1.83- and 3.29-fold, which was higher than the average recombination frequency of entire chromosomes: 1.63- and 1.58-fold for chromosomes 1 and 8, respectively (Figure 6).

To assess whether genetic interference was affected by the null mutation of the ZEP1 gene, we conducted coefficient of coincidence analysis according to the method described for C. elegans (Libuda et al., 2013). In the analysis, the strength of interference (I) equal to 1 is indicative of complete interference, while an I equal to zero is indicative of an absence of interference (Fernandes et al., 2018). In WT plants, the strength of interference (I) in eight adjacent interval pairs of chromosome 1 ranged from 0.37 to 1.00, indicating the existence of strong genetic interference between neighboring COs. By contrast, the zep1-KO mutants showed reduced I for those interval pairs, with I values of −0.18 to 0.14 (Figure 7A). For chromosome 8, the I value of the four measured interval pairs in the WT ranged from 0.48 to 0.71, whereas that in zep1-KO lines ranged from −0.17 to 0.05 (Figure 7B). These results suggest that genetic interference is likely fully eliminated in zep1-KO plants.