We obtained two Zmrad17 mutants from the Maize EMS induced Mutant Database (MEMD)¹ (Lu et al., 2018). All plants were grown in field during the growing season or greenhouse under normal growth conditions. Primer sequences used in genotyping were listed in Supplementary Table S1.

Pollen grains were dissected out of fresh anthers during pollination stage and viability was assessed by 1% I₂-KI staining. Images of stained pollen grains were taken using a Leica EZ4 HD stereo microscope equipped with a Leica DM2000 LED illumination system (Leica, Solms, Germany).

Total mRNA was isolated from root, stem, leaf, developing meiotic ear (1-2cm in length), immature tassel, developing embryo and endosperm (16 days after pollination) of B73 plants with TRIzol (TIANGEN). cDNA synthesis was performed by TaKaRa kits according to manufacturer’s instructions. The entire cDNA was cloned by RACE using the SMART RACE cDNA amplification kit (Clontech). RT-qPCR analysis was performed using the CFX Connect Real-Time PCR System (BIO-RAD). Primer sequences used in RT-qPCR were listed in Supplementary Table S1.

Immature tassels were fixed for 24 h in Carnoy’s solution (ethanol: acetic acid = 3:1, v/v). Then, tassels were stored in 70% ethanol at 4°C. Anthers at meiotic stages were squashed in 45% (v/v) acetic acid solution. Slides with chromosomes were frozen in liquid nitrogen and then cover slips were removed immediately. The slides were dehydrated through an ethanol series (70/90/100%) for 5 min each once. Dried slides were stained with 4′,6-diamidino-2-phenylindole (DAPI) in an antifade solution (Vector). Images were captured using a Ci-S-FL microscope (Nikon, Tokyo) equipped with a DS-Qi2 Microscope Camera system.

The FISH analysis was performed according to protocols described previously (Richards and Ausube, 1988; Li and Arumuganathan, 2001; Wang et al., 2006a; Han et al., 2007; Cheng, 20136a). Two repetitive DNA elements, 5S rDNA repeats (pTa794) and the telomere-specific repeats (pAtT4), were used as probes (Richards and Ausube, 1988). Probes were labeled with digoxigenin by nick translation mix (Roche) and detected with anti-digoxigenin antibody (Vector). Chromosome images were captured under a Ci-S-FL fluorescence microscope (Nikon) equipped with a DS-Qi2 microscopy camera (Nikon, Tokyo, Japan).

Young anthers during meiotic stages were fixed in 4% (w/v) paraformaldehyde for 30 min at room temperature and stored in 1x Buffer A at 4°C. Immunofluorescence was performed as previously described (Pawlowski et al., 2003; Cheng, 2013). The primary antibodies against ASY1, ZYP1, and γH2AX were prepared as described previously (Jing et al., 2019). Antibody against RAD51 was a gift from Wojtek Pawlowski’s Lab at Cornell University. Fluorochrome-coupled secondary antibodies (ABclonal) were used for fluorescence detection. All primary and secondary antibodies were diluted at 1:100. Images of meiocytes were observed and captured using a Ci-S-FL microscope (Nikon) equipped with a DS-Qi2 microscopy camera (Nikon, Tokyo, Japan). The images were captured by software NIS-Elements and colored by the ImageJ software.

The number of chiasmata were quantified for meiocytes at diakinesis. The rod-, ring-and “∞”-shaped bivalents were scored as one chiasma and two, three chiasmata, respectively.

To identify a putative RAD17 gene in maize, the full-length amino acid sequence of the rice RAD17 was used as a query to search in the maize genome database² by BLASTp analysis. We identified only one candidate gene (Zm00001d047946) with the highest similarity to the rice RAD17 (LOC_Os03g13850). Phylogeny analyses revealed that RAD17 homologs formed two distinct clades reflecting the divergence between monocot and dicot plants (Figure 1A). In addition, the multiple sequence alignment of ZmRAD17 amino acid with its orthologs indicated that the RAD17 proteins were conserved in the primary AAA-ATPase domains (Figure 1B). We then investigated the spatio-temporal expression pattern of ZmRAD17 using RT-qPCR analyses. The result showed that ZmRAD17 was highly expressed in the developing tassel, ear, and embryo, but weakly expressed in root, stem, leaf and endosperm (Figure 1C).

The full-length cDNA sequence of ZmRAD17 was isolated by performing rapid amplification of cDNA ends (RACE). It contains 2,089 bp with an open reading frame of 1,851bp and consists of 12 exons and 11 introns (Figure 2A). To characterize biological functions of ZmRAD17, two independent stop codon mutants were obtained from the EMS induced Mutant Database (MEMD) in B73 background (Lu et al., 2018). By conducting locus-specific PCR amplification followed by Sanger sequencing, we confirmed that the stop codon mutation sites are located in the first exon (named as Zmrad17-1) and the eighth exon (named as Zmrad17-2) of ZmRAD17, respectively (Figure 2A). Both Zmrad17 mutants exhibited normal vegetative growth, but partially male-sterile (Figure 2B and Supplementary Figure S1A). KI-I₂ staining displayed that unlike large, round and purple pollen grains of the wild-type (Figure 2C and Supplementary Figure S1B), a proportion of mutant pollen grains were empty, shrunken and unable to stain (Figures 2D,E and Supplementary Figures S1C,D). Surprisingly, when pollinated with pollen grains from wild-type plants, mutant ears exhibited a similar extent of seed setting (Figure 2F and Supplementary Figure S1E). These results indicate that the dysfunction of ZmRAD17 causes effects on male reproductive development, but not on female.

To explore whether pollen abortion is resulted from the defect in male meiosis, chromosome behaviors were investigated in both wild-type and Zmrad17 meiocytes at different stages by staining chromosome spreads with 4′,6-diamidino-2-phenylindole (DAPI). In the wild-type, chromosomes begun to condense and became visible as thin threads structures at leptotene (Figure 3A). Then, homologous chromosomes came close to each other and started to pair and synapsis at zygotene (Figure 3B). During pachytene, chromosomes were fully synapsed to form thick threads (Figure 3C). With chromosomes further condensed, 10 short, rod-like bivalents appeared to scatter in the nucleus at diakinesis (Figure 3D). Once entry into metaphase I, ten bivalents aligned on the equatorial plate in an orderly manner (Figure 3E). At anaphase I, homologous chromosomes separated equally and migrated toward the opposite poles (Figure 3F) forming dyad (Figure 3G). After the second meiotic division, the sister chromatids segregated and ultimately produced tetrad (Figure 3H).

In both of Zmrad17 mutant meiocytes, chromosome behaviors were indistinguishable from the wild-type from leptotene to zygotene (Figures 3I,J and Supplementary Figures S2A,B). However, meiotic abnormalities started to be constantly observed at pachytene, showing abnormal chromosome associations between non-homologous chromosomes (Figure 3K and Supplementary Figure S2C). At diakinesis, although ten bivalents formed, aberrant bridges among bivalents were frequently observed in Zmrad17 meiocytes (Figure 3L, n = 37; Supplementary Figure S2D). Despite all bivalents could be aligned on the equatorial plate during metaphase I, Zmrad17 meiocytes exhibited abnormal bivalent aggregation (Figure 3M and Supplementary Figure S2E). At anaphase I, homologous chromosomes separated with obvious chromosome bridge and chromosome fragmentation (Figure 3N and Supplementary Figure S2F). Chromosome fragments were lagged and scattered randomly within the nucleus at telophase I (Figure 3O and Supplementary Figure S2G). The second meiotic division subsequently underwent and tetrad with micronuclei were formed (Figure 3P and Supplementary Figure S3H). These results suggest that the abnormal chromosome behaviors are responsible for the male sterility of Zmrad17 mutants. Since Zmrad17-1 and Zmrad17-2 exhibited the same defect in the meiotic chromosome behaviors, all subsequent analyses were conducted using Zmrad17-1 mutant as a representative of the Zmrad17 dysfunction.

To evaluate whether DSB formation is defective in Zmrad17 mutant, we performed immunostaining with antibodies against γH2AX and RAD51. γH2AX is a specific histone variant accumulating at damaged sites to promote DSB repair (Hunter et al., 2001; Dickey et al., 2009). Therefore, γH2AX is routinely used as a cytogenetic marker to detect the presence of DSB (Valdiglesias et al., 2013; Geric et al., 2014; Turinetto and Giachino, 2015). Our analysis revealed a substantial amount of dot-like γH2AX signals appeared in both wild-type (Figure 4A, n = 13) and mutant meiocytes at zygotene (Figure 4B, n = 16), suggesting that ZmRAD17 is dispensable for DSB formation. The loading of RAD51 on chromosomes serves as an important marker to monitor HR-mediated DSB repair in many different organisms (Pawlowski et al., 2003). Constantly, we did not observe marked difference in the localization of RAD51 signals between wild-type (Figure 4C, n = 24) and Zmrad17-1 meiocytes at zygotene (Figure 4D, n = 35), suggesting that ZmRAD17 is not crucial for HR initiation. Moreover, the number of chiasmata were counted in both wild type and mutant meiocytes at diakinesis stage using a method described previously (Moran et al., 2001). We found that although aberrant associations among bivalents occurred in Zmrad17-1 meiocytes, the number of chiasmata (Supplementary Figure S3) seemed comparable between wild type (17.08 ± 1.93, n = 24) and mutant (17.11 ± 2.10, n = 19), implying that ZmRAD17 is not critical for CO formation.

Telomere bouquet clustering occurs specifically at early zygotene and is thought to be essential for homologous pairing and synapsis (Bass et al., 1997; Harper et al., 2004). To test whether telomere bouquet formation is affected in Zmrad17-1, we conducted FISH using a telomere specific probe (pAtT4) in both wild-type and Zmrad17 meiocytes. The result displayed that nearly all of telomere signals were clustered and attached to the nuclear envelope in both wild-type (Figure 5A, n = 12) and Zmrad17-1 (Figure 5B, n = 24) meiocytes at zygotene, indicating that ZmRAD17 is not required for telomere bouquet formation.

The 5S ribosomal DNA (rDNA) is a tandemly repetitive sequence located on the long arm of chromosome 2 in maize and is often used to monitor homologous pairing (Li and Arumuganathan, 2001). To examine whether the disruption of ZmRAD17 could impact the homologous chromosome pairing, FISH analysis using 5S rDNA as a probe was conducted. The results showed that only one 5S rDNA signal was constantly detected in both wild-type (Figure 5C, n = 23) and Zmrad17-1 meiocytes (Figure 5D, n = 37) at pachytene, suggesting that ZmRAD17 is not necessary for homologous pairing.

The synaptonemal complex (SC) is a protein scaffold linking homologous chromosomes to promote meiotic crossover formation (Cahoon and Hawley, 2016). To inspect the installation behavior of the SC, we conducted immunolocalization using antibodies against ASY1 and ZYP1 in both wild-type and Zmrad17-1 meiocytes. ASY1, the axial element (AE) component of SC, localizes at chromosome axis (Sanchez-Moran et al., 2007; Sanchez-Moran et al., 2008). In the wild-type, ASY1 loading appeared as continuous linear signals along entire chromosomes at zygotene (Figure 6A, n = 10). Similar pattern of ASY1 distribution was observed in Zmrad17-1 at the same stage (Figure 6B, n = 32), indicating that ZmRAD17 is not required for AE installation.

ZYP1 constitutes the central element (CE) of SC (Higgins et al., 2005; Golubovskaya et al., 2011). At pachytene, ZYP1 signals in wild-type meiocytes formed continuous linear signals along the whole length of synapsed chromosomes (Figure 6C, n = 14). In contrast, although 18.2% of Zmrad17-1 meiocytes showed a similar ZYP1 staining as wild-type, the remaining 81.8% of meiocytes exhibited short stretches of ZYP1 signals in Zmrad17-1 (Figure 6D, n = 88). Taken together, these results indicate that ZmRAD17 is indispensable for SC installation.