ZmRAD51C Is Essential for Double-Strand Break Repair and Homologous Recombination in Maize Meiosis

Radiation sensitive 51 (RAD51) recombinases play crucial roles in meiotic double-strand break (DSB) repair mediated by homologous recombination (HR) to ensure the correct segregation of homologous chromosomes. In this study, we identified the meiotic functions of ZmRAD51C, the maize homolog of Arabidopsis and rice RAD51C. The Zmrad51c mutants exhibited regular vegetative growth but complete sterility for both male and female inflorescence. However, the mutants showed hypersensitivity to DNA damage by mitomycin C. Cytological analysis indicated that homologous chromosome pairing and synapsis were rigorously inhibited, and meiotic chromosomes were often entangled from diplotene to metaphase I, leading to chromosome fragmentation at anaphase I. Immunofluorescence analysis showed that although the signals of the axial element absence of first division (AFD1) and asynaptic1 (ASY1) were normal, the assembly of the central element zipper1 (ZYP1) was severely disrupted. The DSB formation was normal in Zmrad51c meiocytes, symbolized by the regular occurrence of γH2AX signals. However, RAD51 and disrupted meiotic cDNA 1 (DMC1) signals were never detected at the early stage of prophase I in the mutant. Taken together, our results indicate that ZmRAD51C functions crucially for both meiotic DSB repair and homologous recombination in maize.


Introduction
In sexually reproducing eukaryotes, meiosis is a two-step cell division process with one round of DNA replication followed with two consecutive rounds of cell division to ensure the genetic exchange of parents and the formation of haploid generative cells [1,2]. During meiosis I, the physical connections between homologous chromosomes guarantee the proper reductional segregation and create new combinations of alleles [1,3,4].

Characterization of a Sterile Mutant
A sterile maize mutant was originally obtained from a mutant library created by ethylmethane sulfonate (EMS) mutagenesis in the Mo17 inbred line. The mutant was completely male-sterile in tassel ( Figure 1A). When the mutant ear was pollinated with mature pollens from wild type plants, it failed to produce any seeds ( Figure 1B), suggesting that the mutant was also female-sterile. To further examine male sterility, pollen grains from wild type and mutant plants were stained by Alexander solution, a common pollen viability stain [54]. Pollen grains from wild type showed the regular purple round shape ( Figure 1C). In contrast, pollen grains from mutant plants were empty and shrunken, and failed to be stained ( Figure 1D). Moreover, in the progeny of the self-pollinated heterozygous mutant plants, the segregation ratio of fertile (164) to sterile (53) plants fitted the 3:1 ratio (χ 2 =0.0384; p > 0.05), implying that the sterile phenotype of this mutant is caused by a single recessive mutation.

Map-Based Cloning and Characterization of ZmRAD51C
A map-based cloning approach was used to isolate the mutated gene underlying this sterile mutant. An F2 mapping population was constructed by crossing the heterozygous mutant plants with the other inbred line B73. The candidate gene was primarily mapped to a 4.1 Mb region between M217.4 (Contig1365318) and M221.5 (Contig839180) on the long arm of chromosome 3 using 44 F2 mutant segregates showing the sterile phenotype (Figure 2A), then further delimited to a region of 787 kb by using the other 413 F2 mutant plants. Within this region, one candidate gene (Zm00001d044278) was found to harbor a G-to-A substitution at nucleotide position 629 in the coding sequence (CDS; Figure 2B), resulting in an amino acid substitution from Gly to Asp in the conserved RecA-like NTPases domain of a putative RAD51C homolog ( Figure 2C). The phylogenetic tree further suggested that the predicted protein encoded by Zm00001d046970 is the closest homolog of RAD51C in maize (Figures 2 and S1). Therefore, we suspected this amino acid change on the conserved domain resulting from the mutation of ZmRAD51C, which finally led to maize sterility in the mutant. We named this mutant allele as Zmrad51c-1.
To further verify if the Zmrad51c mutation was responsible for the sterile phenotype, we obtained another mutant allele (EMS4-05639c) at the B73 inbred line background from the Maize EMS induced Mutant Database (MEMD) (http://www.elabcaas.cn/memd/) [55]. Sequencing revealed that this mutant allele contains a nucleotide substitution from G to A at the splicing donor site of the 4 th intron ( Figure 2D). The RT-PCR analysis further confirmed that this single nucleotide mutation indeed resulted in the mis-splice of the 4 th intron into the mature mRNA ( Figure 2E), and consequently produced a longer transcript with an in-frame premature stop codon (underlined tga) ( Figure 2D). Therefore, this mutant allele was named as Zmrad51c-2. The homozygous plants of Zmrad51c-2 also exhibited the same abnormal cytological behavior as Zmrad51c-1 ( Figure S2). Taken together, these results indicated that the sterility phenotype of the mutants was a result of the disruption of the ZmRAD51C gene.

Map-Based Cloning and Characterization of ZmRAD51C
A map-based cloning approach was used to isolate the mutated gene underlying this sterile mutant. An F 2 mapping population was constructed by crossing the heterozygous mutant plants with the other inbred line B73. The candidate gene was primarily mapped to a 4.1 Mb region between M217.4 (Contig1365318) and M221.5 (Contig839180) on the long arm of chromosome 3 using 44 F 2 mutant segregates showing the sterile phenotype (Figure 2A), then further delimited to a region of 787 kb by using the other 413 F 2 mutant plants. Within this region, one candidate gene (Zm00001d044278) was found to harbor a G-to-A substitution at nucleotide position 629 in the coding sequence (CDS; Figure 2B), resulting in an amino acid substitution from Gly to Asp in the conserved RecA-like NTPases domain of a putative RAD51C homolog ( Figure 2C). The phylogenetic tree further suggested that the predicted protein encoded by Zm00001d046970 is the closest homolog of RAD51C in maize ( Figure 2 and Figure S1). Therefore, we suspected this amino acid change on the conserved domain resulting from the mutation of ZmRAD51C, which finally led to maize sterility in the mutant. We named this mutant allele as Zmrad51c-1.
To further verify if the Zmrad51c mutation was responsible for the sterile phenotype, we obtained another mutant allele (EMS4-05639c) at the B73 inbred line background from the Maize EMS induced Mutant Database (MEMD) (http://www.elabcaas.cn/memd/) [55]. Sequencing revealed that this mutant allele contains a nucleotide substitution from G to A at the splicing donor site of the 4 th intron ( Figure 2D). The RT-PCR analysis further confirmed that this single nucleotide mutation indeed resulted in the mis-splice of the 4 th intron into the mature mRNA ( Figure 2E), and consequently produced a longer transcript with an in-frame premature stop codon (underlined tga) ( Figure 2D). Therefore, this mutant allele was named as Zmrad51c-2. The homozygous plants of Zmrad51c-2 also exhibited the same abnormal cytological behavior as Zmrad51c-1 ( Figure S2). Taken together, these results indicated that the sterility phenotype of the mutants was a result of the disruption of the ZmRAD51C gene. Utilizing reverse transcription PCR, we obtained the full-length cDNA sequence of ZmRAD51C, which is composed of 8 exons and 7 introns and encodes a protein with 268 amino acids. According to the conserved domain search in NCBI (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi), ZmRAD51C protein contains a conserved RecA-like-NTPase domain, Walker A and B motif, multimer (BRC) interface, and ATP binding sites ( Figure S3). The multiple alignment using the fulllength amino acid sequence of RAD51C in 9 different species revealed that ZmRAD51C is evolutionarily conserved among different species ( Figure S3). The spatio-temporal expression pattern of ZmRAD51C was analyzed by quantitative RT-PCR analyses. The results showed that ZmRAD51C was highly expressed in the developing anther, ear, embryo, and endosperm, but weakly expressed in root, stem, and leaf ( Figure S4).

Abnormal Meiotic Chromosome Behaviors in Zmrad51c
To clarify the reason for the sterility phenotype of Zmrad51c mutants, the meiotic chromosome behaviors in different stages of pollen mother cells (PMCs) were examined in both wild type and Utilizing reverse transcription PCR, we obtained the full-length cDNA sequence of ZmRAD51C, which is composed of 8 exons and 7 introns and encodes a protein with 268 amino acids. According to the conserved domain search in NCBI (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi), ZmRAD51C protein contains a conserved RecA-like-NTPase domain, Walker A and B motif, multimer (BRC) interface, and ATP binding sites ( Figure S3). The multiple alignment using the full-length amino acid sequence of RAD51C in 9 different species revealed that ZmRAD51C is evolutionarily conserved among different species ( Figure S3). The spatio-temporal expression pattern of ZmRAD51C was analyzed by quantitative RT-PCR analyses. The results showed that ZmRAD51C was highly expressed in the developing anther, ear, embryo, and endosperm, but weakly expressed in root, stem, and leaf ( Figure S4).

Abnormal Meiotic Chromosome Behaviors in Zmrad51c
To clarify the reason for the sterility phenotype of Zmrad51c mutants, the meiotic chromosome behaviors in different stages of pollen mother cells (PMCs) were examined in both wild type and Zmrad51c mutants. In wild type meiocytes, the chromosomes began to condense as long thin threads at leptotene ( Figure 3A). Then the homologous chromosomes came close to pairing and synapsis at zygotene ( Figure 3B). During pachytene, homologs became tightly associated and synapsis was completed to form thick thread-like chromosomes ( Figure 3C), which then gradually condensed in late pachytene ( Figure 3D). During diakinesis, the paired chromosomes were highly condensed, forming 10 rod-shape or ring-shape bivalents ( Figure 3E), which then aligned on the equatorial plate during metaphase I ( Figure 3F). Subsequently, homologous chromosomes separated equally and migrated into two opposite poles of cells from anaphase I to telophase I ( Figure 3G), producing a dyad with the same numbers of chromosomes ( Figure 3H). During meiosis II, sister chromatids in the dyad separated and underwent cell division to produce tetrad microspores ( Figure 3I).

Normal Telomere Bouquet Clustering but Deficient Homologous Chromosome Pairing in Zmrad51c
Telomere bouquet clustering is an evolutionarily conserved chromosome arrangement that clusters telomeres together to one side of the nucleus [56]. This specific structure may promote initiation of homologous pairing in early prophase I [57][58][59]. To explore whether mutation of In Zmrad51c meiocytes, chromosome behavior was indistinguishable from wild type in leptotene ( Figure 3J and Figure S2A). However, chromosomes were observed as single threads without obvious pairing and synapsis from zygotene ( Figure 3K and Figure S2B) to pachytene ( Figure 3L and Figure S2C). At late pachytene, chromosomes gradually condensed but became an entangled mass throughout the nucleus ( Figure 3M and Figure S2D), and bivalents were never observed ( Figure 3N and Figure  S2E) at diakinesis. At metaphase I, a few pieces of chromosome fragments were present and broken away from the main chromosome mass ( Figure 3O and Figure S2F). During anaphase I, despite genome fragmentation, the entangled chromosome masses separated randomly, leading to asymmetric migration of the chromosomes to the opposite poles ( Figure 3P and Figure S2G). Meanwhile, chromosome bridges were clearly visible between the separated chromosomes and chromosome fragments were randomly distributed throughout the nucleus ( Figure 3P and Figure S2G. At telophase I, many chromosome fragments were lagged and scattered in the nucleus of the dyad ( Figure 3Q and Figure S2H). After the second meiosis division, several chromosome fragments and micronuclei were always dispersed in tetrads ( Figure 3R and Figure S2I). Therefore, we concluded that the abnormal chromosome synapsis and the appearance of chromosomal fragmentation are responsible for the complete sterility of Zmrad51c mutants. Since Zmrad51c-1 and Zmrad51c-2 exhibited the same defect in the meiotic chromosome behaviors, all subsequent analyses were conducted using Zmrad51c -1 mutant as representative of the Zmrad51c defect.

Normal Telomere Bouquet Clustering but Deficient Homologous Chromosome Pairing in Zmrad51c
Telomere bouquet clustering is an evolutionarily conserved chromosome arrangement that clusters telomeres together to one side of the nucleus [56]. This specific structure may promote initiation of homologous pairing in early prophase I [57][58][59]. To explore whether mutation of Zmrad51c affects telomere bouquet formation, we conducted FISH (fluorescence in situ hybridization) analysis using pAtT4 as a telomere-specific probe in wild type and Zmrad51c-1 meiocytes [60,61]. The results revealed that almost all of the pAtT4 signals were clustered together and attached to the nuclear envelope at early zygotene stage in both wild type ( Figure 4A, n = 48) and Zmrad51c-1 ( Figure 4B, n = 55), indicating that ZmRAD51C is not required for telomere bouquet formation. segregation [62,63]. In wild type, only one paired 5S rDNA signal was detected at pachytene ( Figure  4C, n = 57), indicating that the homologous chromosomes were properly paired. In contrast, two separate 5S foci were consistently observed in nearly all of Zmrad51c-1 meiocytes at this stage ( Figure  4D, n = 60), indicating that homologous chromosome pairing was defective in Zmrad51c-1 strain. Therefore, these results demonstrated that ZmRAD51C is essential for homologous chromosome pairing.

Normal Axial Element Installation but Defective Central Element Installation in Zmrad51c
The synaptonemal complex (SC) is a large protein scaffold connecting homologous chromosomes and is required for meiotic crossover formation to promote genetic information exchange between homologous chromosomes [64]. In order to monitor the installation behavior of the SC in Zmrad51c-1, we conducted immunolocalization using antibodies against AFD1, ASY1, and ZYP1 in wild type and Zmrad51c-1. AFD1, a critical component of the sister-chromatid cohesion complex (cohesion) associated with axial and lateral elements, is required for axial element (AE) To explore whether the homologous chromosome pairing is defective in Zmrad51c-1, FISH analysis using 5S rDNA as a specific probe was conducted in both wild type and Zmrad51c-1 meiocytes. The 5S ribosomal DNA (rDNA) is a tandem repetitive sequence located on the distal regions of the long arm of chromosome 2 and is often used to assess chromosome pairing and segregation [62,63]. In wild type, only one paired 5S rDNA signal was detected at pachytene ( Figure 4C, n = 57), indicating that the homologous chromosomes were properly paired. In contrast, two separate 5S foci were consistently observed in nearly all of Zmrad51c-1 meiocytes at this stage ( Figure 4D, n = 60), indicating that homologous chromosome pairing was defective in Zmrad51c-1 strain. Therefore, these results demonstrated that ZmRAD51C is essential for homologous chromosome pairing.

Normal Axial Element Installation but Defective Central Element Installation in Zmrad51c
The synaptonemal complex (SC) is a large protein scaffold connecting homologous chromosomes and is required for meiotic crossover formation to promote genetic information exchange between homologous chromosomes [64]. In order to monitor the installation behavior of the SC in Zmrad51c-1, we conducted immunolocalization using antibodies against AFD1, ASY1, and ZYP1 in wild type and Zmrad51c-1. AFD1, a critical component of the sister-chromatid cohesion complex (cohesion) associated with axial and lateral elements, is required for axial element (AE) elongation and maturation [65]. In wild type meiocytes, AFD1 behaved as linear signals along the entire chromosome from leptotene ( Figure 5A, n = 11). In Zmrad51c-1 meiocytes, the AFD1 loading pattern was consistent with wild type ( Figure 5B, n = 14), indicating that ZmRAD51C is dispensable for AFD1 assembling. ASY1, an AE component of SC in maize, is the homolog of S. cerevisiae Hop1 and Arabidopsis ASY1, which is critical for homologous recombination and SC assembly [66,67]. In wild type, ASY1 loading appeared as long linear signals associated with chromosomes in zygotene ( Figure 5C, n = 17). Similarly, the ASY1 signals distributed normally onto the chromosomes during zygotene in Zmrad51c-1 mutant ( Figure 5D, n = 13), indicating that ZmRAD51C is not required for AE installation. ZYP1, the central element (CE) component in maize, is a transverse filament (TF) protein that assembles between the AEs [67]. In wild type, ZYP1 signals were completely elongated along the entire length of synapsed chromosomes at pachytene ( Figure 5E, n = 14). In contrast, only several scattered punctate or short stretches of ZYP1 signals were detected in Zmrad51c-1 pachytene meiocytes ( Figure 5F, n = 8). Quantification measurement and statistical analysis of ZYP1 length revealed that the ZYP1 length in Zmrad51c-1 meiocytes was significantly shorter comparing with wild type ( Figure S5). Taken together, these results indicate that ZmRAD51C is crucial for SC installation in maize meiosis.

Normal DSB Formation but Defective Loading of RAD51 and DMC1 in Zmrad51c
In meiosis prophase I, the formation and repair of programmed DSBs is essential for HR [1]. Soon after DSBs are formed, a protein complex initiates histone H2AX phosphorylation (γH2AX) at the damaged site to promote DNA repair [68,69]. Therefore, γH2AX is often used as a biomarker to detect the appearance of DSBs [70]. To verify whether DSBs are formed normally in Zmrad51c-1, an antibody specifically recognizing plant γH2AX was raised in rabbit and used for immunolocalization in wild type and Zmrad51c-1 meiocytes [71]. At zygotene, a substantial number of dot-like γH2AX signals was detected in both wild type ( Figure 6A) and Zmrad51c-1 meiocytes (Figure 6B), and the

Normal DSB Formation but Defective Loading of RAD51 and DMC1 in Zmrad51c
In meiosis prophase I, the formation and repair of programmed DSBs is essential for HR [1]. Soon after DSBs are formed, a protein complex initiates histone H2AX phosphorylation (γH2AX) at the damaged site to promote DNA repair [68,69]. Therefore, γH2AX is often used as a biomarker to detect the appearance of DSBs [70]. To verify whether DSBs are formed normally in Zmrad51c-1, an antibody specifically recognizing plant γH2AX was raised in rabbit and used for immunolocalization in wild type and Zmrad51c-1 meiocytes [71]. At zygotene, a substantial number of dot-like γH2AX signals was detected in both wild type ( Figure 6A) and Zmrad51c-1 meiocytes (Figure 6B), and the number of γH2AX foci in Zmrad51c-1 (n = 12) were not significantly different from wild type (n = 11) ( Figure 6G), suggesting that ZmRAD51C is dispensable for DSB formation.
To test whether ZmRAD51C functions in DSB repair, immunolocalization using antibodies against RAD51 and DMC1 proteins was conducted in wild type and Zmrad51c-1 meiocytes. The loading of RAD51 and DMC1 onto the ssDNA is considered as an important cytological marker of HR-mediated DSB repair in different species [72,73]. In wild type, both RAD51 ( Figure 6C and 6H, n = 12) and DMC1 ( Figure 6E and 6I, n = 13) proteins were appeared as punctuate foci distributed on meiotic chromosomes in zygotene meiocytes. In contrast, no visible RAD51 ( Figure 6D, n = 10) and DMC1 ( Figure 6F, n = 12) foci were detected in Zmrad51c-1 meiocytes at the same stage, indicating that ZmRAD51C is essential for meiotic DSB repair in maize. number of γH2AX foci in Zmrad51c-1 (n = 12) were not significantly different from wild type (n = 11) ( Figure 6G), suggesting that ZmRAD51C is dispensable for DSB formation.
To test whether ZmRAD51C functions in DSB repair, immunolocalization using antibodies against RAD51 and DMC1 proteins was conducted in wild type and Zmrad51c-1 meiocytes. The loading of RAD51 and DMC1 onto the ssDNA is considered as an important cytological marker of HR-mediated DSB repair in different species [72,73]. In wild type, both RAD51 ( Figure 6C and 6H, n = 12) and DMC1 ( Figure 6E and 6I, n = 13) proteins were appeared as punctuate foci distributed on meiotic chromosomes in zygotene meiocytes. In contrast, no visible RAD51 ( Figure 6D, n = 10) and DMC1 ( Figure 6F, n = 12) foci were detected in Zmrad51c-1 meiocytes at the same stage, indicating that ZmRAD51C is essential for meiotic DSB repair in maize.  foci between wild type and Zmrad51c-1 meiocytes. Triple asterisks indicate the statistical significant at p < 0.001 using a two-tailed Student's t-test. NS means no significant difference.

Zmrad51c Mutants Are Hypersensitive to DNA Damage Agent
Previous studies have shown that the DNA damage induced by the mitomycin C (MMC) was exclusively repaired by HR [74,75]. To investigate whether the mutation of ZmRAD51C could result in the defect in DSB repair during mitosis, the MMC was used to treat wild type and Zmrad51c-1 from germination to seedlings. After genotyping and measuring the length of seedlings, we found that the plant height of Zmrad51c-1 was significantly shorter than wild type when growing on 1/2 MS medium containing 10 µg/mL MMC (Figure 7), suggesting that ZmRAD51C may also play an important role in HR repair during mitosis. and DMC1 foci between wild type and Zmrad51c-1 meiocytes. Triple asterisks indicate the statistical significant at p < 0.001 using a two-tailed Student's t-test. NS means no significant difference.

Zmrad51c mutants are hypersensitive to DNA damage agent
Previous studies have shown that the DNA damage induced by the mitomycin C (MMC) was exclusively repaired by HR [74,75]. To investigate whether the mutation of ZmRAD51C could result in the defect in DSB repair during mitosis, the MMC was used to treat wild type and Zmrad51c-1 from germination to seedlings. After genotyping and measuring the length of seedlings, we found that the plant height of Zmrad51c-1 was significantly shorter than wild type when growing on 1/2 MS medium containing 10 µg/mL MMC ( Figure. 7), suggesting that ZmRAD51C may also play an important role in HR repair during mitosis. Figure 7. Zmrad51-1 is hypersensitive to MMC. (A) Wild type and Zmrad51c-1 showed similar height when growing on 1/2 MS medium without MMC for two weeks. (B) Zmrad51-1 seedlings (right) were significantly shorter than wild type (left) on 1/2 MS medium treated with 10 µg/mL MMC for two weeks. (C) Statistical analysis of wild type (n = 10) and Zmrad51-1 (n = 10) plant height. Values are means ± SE. Double asterisks indicates the statistical significance at p < 0.01 using a two-tailed Student's t-test. NS means no significant difference.

Functional Conservation of RAD51C in Different Organisms
Prior to this study, none of the RAD51 paralogues have been functionally characterized in maize yet. However, based on the high sequence identity with Arabidopsis and rice paralogs, we were be able to ascertain that there is a single copy gene encoding each of these five RAD51 paralogs in maize, respectively (Table S3). Meanwhile, the phylogeny analysis unambiguously discriminates that each of RAD51 paralogs fall into distinct clades ( Figure S6), indicating that different RAD51 paralogues have divergent evolutionary origins. (C) Statistical analysis of wild type (n = 10) and Zmrad51-1 (n = 10) plant height. Values are means ± SE. Double asterisks indicates the statistical significance at p < 0.01 using a two-tailed Student's t-test. NS means no significant difference.

Functional Conservation of RAD51C in Different Organisms
Prior to this study, none of the RAD51 paralogues have been functionally characterized in maize yet. However, based on the high sequence identity with Arabidopsis and rice paralogs, we were be able to ascertain that there is a single copy gene encoding each of these five RAD51 paralogs in maize, respectively (Table S3). Meanwhile, the phylogeny analysis unambiguously discriminates that each of RAD51 paralogs fall into distinct clades ( Figure S6), indicating that different RAD51 paralogues have divergent evolutionary origins.
As one of the RAD51 paralogs, RAD51C has been known as an important player in DSB repair via HR in several organisms. In mammalian cells, deficiency in RAD51C results in spontaneous chromosome aberrations, hypersensitivity to DNA damaging agents, and early embryonic lethality [76][77][78][79][80]. In contrast, the loss-of-function of RAD51C in plants led to sterility without disturbing the vegetative growth under normal condition. Previous studies in Arabidopsis and rice demonstrated that both AtRAD51C and OsRAD51C are critically required for DSB repair in a manner downstream of DSB formation [31,40,45,46]. According to our study, the mutation in the maize RAD51C caused the complete sterility in both male and female meiosis, accompanied with an array of abnormal meiotic chromosome behaviors, including the lack of homologous pairing and synapsis, chromosome entanglement and fragmentation, and SC assembly failure. These results demonstrated that the RAD51C is also essential for meiotic DSB repair in maize. In addition, as no any RAD51 or DMC1 signals could be observed once ZmRAD51C disrupted, suggesting that either RAD51C functions by stabilizing the RAD51/DMC1-ssDNA nucleofilaments, or alternatively, RAD51C may operate directly upstream of RAD51 and DMC1 proteins to direct their loading onto chromosomes. Therefore, our findings strengthen the notion that the function of RAD51C in DSB repair is highly conserved among different species from mammal to plant.

Abnormal Chromosome Fragmentation in Zmrad51c
Chromosome fragmentation is a characteristic phenomenon observed in mutants defective in DSB repair machinery. Our results showed that the Zmrad51c mutant phenotype is similar to that of the Osrad51c and Atrad51c mutants, as well as other related mutants such as Osxrcc3 [44], Osrad1 [81], Osrad17 [82], Atrad50 [83], and Atmre11 [84]. However, the severity of the chromosome segregation seems to be much less in Zmrad51c compared to the Osrad51c and Atrad51c mutants. A possible explanation for this discrepancy could be that more or less in maize, the other RAD51 paralogues may function redundant with ZmRAD51C. Meanwhile, the other possibility we cannot exclude is that the other DSB repair pathway, such as non-homologous end-joining (NHEJ) [85] or microhomology-mediated end-joining (MMEJ) [86], may be alternatively motivated in the absence of the HR pathway at a different extent in maize relative to Arabidopsis and rice. In this context, it would be worthy to examine the meiotic alterations after combining the mutation in ZmRAD51C with mutations in genes involved in the NHEJ and MMEJ pathway, which have not been explored yet in maize.

Plant Materials
The Zmrad51c-1 mutant was created by ethylmethane sulfonate (EMS) mutagenesis in the maize inbred line Mo17. A Zmrad51c-2 mutant was obtained from the Maize EMS induced Mutant Database (MEMD, www.elabcaas.cn/memd/) [55]. Plants were cultivated and fertilized in a same manner as conventional maize production field during the growing season.

Map-Based Cloning of Zmrad51c-1
To isolate the Zmrad51c-1 gene, a positional cloning method was conducted. The heterozygote (ZmRAD51C + /Zmrad51c -) in inbred line Mo17 was outcrossed with maize inbred line B73 to generate F 1 , and then self-crossed to create a F 2 segregation population. Completely infertility plants segregated from the F 2 population were used for fine mapping. InDel (insertion-deletion) and SNP markers were designed based on the sequence divergence between the B73 and Mo17 inbred lines [87,88]. Primer sequences used in cloning and mutants genotyping are listed in Table S1.

cDNA Cloning and Real-Time qPCR for Transcript Expression Assay
Total RNA was extracted individually from root, stem, leaf, developing embryo, and endosperm as well as young tassel and young ear. Reverse transcription was conducted using PrimeScriptTM II 1st strand cDNA Synthesis Kit (TaKaRa, Tokyo, Japan) with Oligo-T as primer. The full-length CDS was amplified by the TransStart FastPfu Fly DNA Polymerase kit (TransGen) using 1st strand cDNA as a template. Quantitative real-time PCR analyses were conducted through a 7500 Fast Real-Time PCR System (Applied Biosystem, Foster city, CA, USA) using SYBR green detection protocol (TaKaRa).
Gene-specific primers and ubiquitin gene primers for the control standard were used in the RT-qPCR. The primer sequences used in cDNA synthesis and RT-qPCR are listed in Table S2.

Pollen Viability Observation
Pollen viability was detected by Alexander staining, following previous protocols [54,89]. Mature anthers were dissected from the wild type and Zmrad51c mutants at tasseling stage. Pollen grains were separated from anthers using forceps and blades, then stained with 10% Alexander solution and observed in a Leica EZ4 HD stereo-microscope. The pictures of pollen grains were taken under a Leica DM2000 LED (Leica, Solms, Germany)

Meiotic Chromosome Preparation
Immature tassels of wild type and mutants were harvested and fixed in Carnoy's solution (ethanol:glacial acetic; 3:1) and stored in 70% ethanol at 4 • C. Anthers at proper stages were dissected and squashed in a 45% acetocarmine solution. Slides with chromosomes were frozen in liquid nitrogen, then the coverslip was removed rapidly. Slides were dehydrated in different ethanol concentration (70%, 90%, and 100%) and air-dried. DAPI in an anti-fade solution (Vector laboratories, Burlingame, CA, USA) was used to counterstain chromosomes. Images were captured using a DS-Qi2 Microscope Camera system installed in a Ci-S-FL microscope (Nikon, Tokyo, Japan).

FISH Analysis
FISH analysis was performed, as described previously [48]. The pTa794 clone containing 5S rDNA repeats or pAtT4 clone containing telomeric-specific repeats were used as FISH probes [60,90]. Probes were labeled with fluorophore-dUTP (Roche, Basel, Switzerland) by nick translation. Anti-digoxigenin (Vector Laboratories) was used as the second antibody for detection with a probe of digoxigenin (DIG). Chromosomes were counterstained using Antifade Mounting Medium with DAPI (Vector laboratories). Finally, chromosome images were taken using a DS-Qi2 Microscope Camera system installed in a Ci-S-FL microscope (Nikon).

Fluorescence Immunolocalization Assays
Immunolocalization assays were conducted, as previous described [72]. Fresh young anthers were dissected from immature tassel and fixed in 1× Buffer A containing 4% (w/v) paraformaldehyde for 30 min at room temperature (25 • C), then shake washed in 1× Buffer A at room temperature for twice and stored in 1× Buffer A at 4 • C. Anthers at proper stages were dissected in 1× Buffer A solution to release meiocytes, then squashed and frozen in liquid nitrogen to rapidly remove the coverslip. Subsequently, the slides were incubated with different antibodies diluted in blocking buffer in a humidity chamber at 37 • C for 1 h and washed 3 times in 1× PBS. Alexa Fluor 555-conjugated goat anti-rabbit antibodies were added to the slides. After washed for 3 times in 1× PBS, the slides were counterstained with DAPI in antifade solution. The slides were observed and images were taken by a Ci-S-FL microscope (Nikon) to generate 2D projected images. Surface rendered images were colored and merged by the ImageJ software (https://imagej.nih.gov/ij/index.html).
The polyclonal antibodies against ASY1, ZYP1, and γH2AX were prepared by specific sequence fusion-peptide immunization in rabbit. AFD1, RAD51, and DMC1 antibodies were gifts from Changbin Chen (University of Minnesota, Twin Cities, MN, United State), Wojciech Pawlowski (Cornell University, Ithaca, NY, United State) and Huabang Chen (Institute of Genetics and Developmental Biology, Beijing, China), respectively.

DNA Damage Agent Mitomycin C (MMC) Treatment
Self-pollinated seeds of heterozygous ZmRAD51C-1 +/plants were surface-sterilized in 75% ethanol for 10 min and 5% NaClO for 15 min, and soaked in sterilized water for one day. Then the sterilized seeds were sown on 1/2 MS solid medium containing 0 or 10 µg/mL MMC (Biochempartner, Wuhan, China) and grown in a growth chamber with a photoperiod cycle of 16 h light in 28 • C and 8 h darkness in 25 • C. Genotyping and phenotyping of the seedlings were conducted two weeks after the seeds were sown.

Image Statistical Analysis
To quantitate the ZYP1 length and the γH2AX, RAD51 and DMC1 foci numbers in wild type and Zmrad51c-1, images were analyzed using ImageJ software based on the measuring instructions of NIH (https://imagej.nih.gov/ij/docs/pdfs/ImageJ.pdf). Statistical significance was calculated using two-tailed Student's t-tests. Scatter Plots were drawn by R language using the "ggplot2" package.  Acknowledgments: We thank all the members of our laboratories for helpful discussion and assistance during this research.

Conflicts of Interest:
The authors declare no conflict of interest.