Meiosis is a process of essential significance for sexual reproduction. During meiosis distribution of genetic material to gametes is associated with recombination which is achieved through crossing over and chromosome segregation. The recombination event (crossing-over) takes place during the first meiotic prophase between non-sister chromatids of homologous chromosomes. It leads to establishing physical links between homologous chromosomes, called chiasmata. Meiotic crossing-over shuffles genetic information, creates new combinations of alleles, and therefore generates genetic variations and drives evolution. The process of recombination during meiosis starts with a programmed DNA double-strand break (DSB). Meiotic DSBs are introduced through the catalytic action of the evolutionarily-conserved SPO11 (Sporulation Protein 11) protein complex which is an enzyme related to type II DNA topoisomerases (Keeney et al., 1997; Robert et al., 2016; Vrielynck et al., 2016). In general, DSBs can be repaired through two major pathways: homologous recombination (HR) and/or non-homologous end joining (NHEJ) (Ohnishi et al., 2009). Programmed DSBs during meiosis are eliminated by HR in the DSBR pathway (Double Strand Break Repair model). The model of DSB repair was first proposed by Szostak and coworkers in the 1980s (Szostak et al., 1983). The extensive studies of this process in Saccharomyces cerevisiae implemented only several alterations to the original model (reviewed in Andersen and Sekelsky, 2010). After introduction of DSB, the DNA ends are resected and long (about 1 kbp) 3′single-stranded overhangs are created, called 3′ssDNA tails (Ohnishi et al., 2009). RAD51 (Radiation sensitive 51) and DMC1 (Disrupted Meiotic cDNA1) recombinases attach to these tails and form nucleoprotein filaments that search for and invade homologous sequences either on a sister chromatid or on a homologous chromosome (Bishop et al., 1992; Shinohara et al., 1992). The latter case may lead to genetic recombination. After invasion on homologous sequence, the next step in the DSBR pathway is establishing the D-loop structure followed by formation of a double Holliday Junction (dHJ) intermediate. Then, the two strands at each HJ are nicked by specific enzymes and ligated. The resolution of dHJ can result in both, crossover and non-crossover repair products (COs and NCOs, respectively) (Andersen and Sekelsky, 2010). DMC1 and RAD51 belong to the same protein family of recombinases, involved in DNA repair through HR, which are related to the bacterial RecA (Bianco et al., 1998). They catalyze the process of pairing and invasion of 3′ssDNA tails formed at the DSB sites into homologous double-stranded DNA. Both of these proteins take part in the meiotic recombination events, however, DMC1 is specific only for cells undergoing meiosis, while RAD51 is ubiquitous and acts also in DSB repair in somatic cells. It is suggested that DMC1 promotes only the CO recombination with the homologous chromosome, which is unique to meiosis, and RAD51 plays its role mainly in sister chromatid exchange or the NCO recombination (Shinohara and Shinohara, 2004; Neale and Keeney, 2006). However, a recent work has shown that in the case of absence of the RAD51-mediated strand exchange activity, the DMC1 activity is sufficient to repair all DSBs during meiosis into both CO and NCO products and it does not affect meiotic crossing-over rates or patterns (Cloud et al., 2012; Da Ines et al., 2013; Singh et al., 2017).

In the plant kingdom, meiosis has been studied to the greatest degree in Arabidopsis thaliana (for review see Mercier et al., 2015). Cereals with large genomes and large chromosomes, such as barley (Hordeum vulgare), are characterized by highly skewed distribution of meiotic crossovers. Consequently, the large sub-centromeric regions, representing substantial proportions of the physical map, are seldom recombined (Higgins et al., 2012; Ramsay et al., 2014). Therefore, the molecular mechanisms underlying meiotic events may be distinct for model Arabidopsis with genome size of ∼135 Mbp (The Arabidopsis Genome Initiative [AGI], 2000) contained within five chromosomes and for barley with genome size of ∼5.3 Gbp contained within seven chromosomes (International Barley Genome Sequencing Consortium Mayer et al., 2012; Mascher et al., 2017). Our knowledge on the DMC1 function in plants comes mainly from studies performed in Arabidopsis. Moreover, its detailed function in DSB repair during meiosis is still extensively discussed. For example, some contradictory reports have appeared in rice (Oryza sativa L.): one, showing that OsDMC1 is required for homologous pairing (Deng and Wang, 2007), and the other, reporting that it is dispensable in this process (Wang et al., 2016), which is different from the role of DMC1 described in other species. These results imply that the function of DMC1 may be distinct in diverse organisms and a direct transfer of knowledge from related species may not be feasible. The recent findings in rice have been obtained studying rice insertion mutants (Wang et al., 2016). Although some in silico studies of DMC1 have been performed in monocot crops, including barley (Barakate et al., 2014), only very recently role of the barley homolog was analyzed in a spontaneous mutant (Colas et al., 2019).

Barley (Hordeum vulgare L.), ranking fourth in production and acreage, belongs to the most important cereal crops worldwide. Here, we present the identification of barley mutants in the DMC1 gene isolated using TILLING strategy in the HorTILLUS population derived from chemical mutagenesis of spring cultivar ‘Sebastian’. Cytological analysis of male meiocytes in the identified dmc1 mutants revealed various abnormalities during meiosis, in anaphase/telophase I and anaphase/telophase II, as well in tetrads. Our results indicate that DMC1 is involved in the DSB repair, crossing-over and chromosome disjunction during meiosis process in barley.

We have performed the analysis of meiosis in the dmc1.b mutant carrying missense mutation leading to the G212S substitution and both the dmc1.c lines (dmc1.c-3041 and dmc1.c-3223) carrying missense mutation causing the R183K change. Observation of meiosis in the mutants revealed that the dmc1.c lines showed differences in chromosomal aberrations frequency when compared to the wild type, whereas the dmc1.b mutant did not show significant disorders. This results are consistent with the in silico analysis of conservation of the substituted amino acid positions and with the SIFT values, which have shown that the dmc1.c mutation should have a more significant impact on protein function than dmc1.b.

Our dmc1.c mutants exhibited very high chromosome aberration frequencies in anaphase/telophase I/II and also very high number of tetrads with micronuclei. We suggest that these changes are the result of defects in DSB repair and anomaly in crossing-over, what strongly confirms that the DMC1 gene is involved in the DSB repair, recombination and chromosome disjunction during meiosis. Both dmc1.c mutants showed disturbances in the chromosome assembling during the metaphase I. Such abnormalities have previously been observed in other species, both in plants and animals. The DMC1-knock-out mice displayed aberrant chromosomal pairing or non-homologous chromosome pairing in spermatocytes (Habu et al., 1996). Our findings are also consistent with the observations of the atdmc1 mutants which exhibited abnormalities in the formation of bivalents and chiasmata (Da Ines et al., 2013). Moreover, in meiotic cells of the rice osdmc1a osdmc1b insertional double mutant, univalents and abnormal number of chromosomes in the metaphase plate during the second meiotic division were observed (Wang et al., 2016). In our study, one mutated line, dmc1.c-3041, formed irregular tetrads containing three or five haploid cells. Similar anomalies were observed in the rice osdmc1a osdmc1b double mutant (Wang et al., 2016). This type of anomaly may be the result of abnormal, uneven segregation of chromosomes to the opposite poles of the cell during meiotic divisions. The dmc1.b as well as both dmc1.c mutant lines show partial sterility of the spikes. Previously reported dmc1 mutants in different species showed sterility or partial sterility. For example, rice OsDMC1-RNAi lines, as well as the insertional osdmc1a osdmc1b double mutant grow normally during their vegetative phase, but they are characterized by total sterility (Deng and Wang, 2007; Wang et al., 2016). The insertional Arabidopsis mutant, atdmc1, produces viable seeds at very low ratio (1.5%) (Da Ines et al., 2013). The DMC1 knock-out mice displayed total sterility (Habu et al., 1996). This suggests that fertility disorders are a common feature of individuals lacking the DMC1 gene in different species. Our dmc1 mutants are not a knock-out type, they carry missense mutations in the analyzed gene and, as it was predicted in our analysis (Figure 1), the substituted amino-acid residues are located in the Rad51 domain, however outside the functional motifs of the DMC1 protein (such as the Walker A and B motifs, and loops 1 and 2), therefore the effect on sterility is not that strong as in other species.

In our study the mutational screening of 5,376 M₂ plants from the HorTILLUS population revealed six independent G/C to A/T mutations within the HvDMC1 gene. Most mutations (88%) found to date in the HorTILLUS population represented this type of transition (calculated based on data of 32 genes TILLed; Szurman-Zubrzycka et al., 2018). Both mutagens used for creation of our TILLING population (MNU and NaN₃) cause such DNA lesions. N-methyl-N-nitrosourea belongs to alkylating agents that are known to alkylate guanine and create O⁶-metG – the lesion with strong mutagenic property (Kleibl, 2002). O⁶-metG mispairs with thymine, which leads to its replacement by adenine in the subsequent replication cycle. If this methylation is induced in a non-transcribed (sense) DNA strand, it leads to G to A transition, whereas if it occurs in transcribed (antisense) DNA strand it results in C to T transition. TILLING populations which were developed after treatment with MNU for Glycine max and Oryza sativa showed 89.4 and 91.7% G/C to A/T transitions, respectively (Cooper et al., 2008; Suzuki et al., 2008). Sodium azide, the other mutagen used for establishing the HorTILLUS population, is mutagenic only for some plant species, among them barley and rice (reviewed in Gruszka et al., 2012). It was used as the only mutagen in other barley TILLING population - TILLMore developed for cultivar ‘Morex’, where it caused mainly G/C to A/T transitions (95.5%, Talamè et al., 2008; Sparla et al., 2014). The mutation density calculated based on mutations found in the HvDMC1 amplicon is 1 per 729 kbp. The average mutation density in the HorTILLUS population is 1 per 477 kbp, however it varies between gene fragments (Szurman-Zubrzycka et al., 2018). The value obtained for HvDMC1 is slightly lower, what could be caused by amplicon base content (G/C – 42%; A/T – 58%).

One very important aspect in terms of functional genetics is the presence of paralogs within the genome that can take over functions of gene of interest (functional redundancy). If there are two or more closely related genes, usually it is necessary to produce individuals with mutations in both or all of paralogs to perform functional analysis. Barley genome is one of the largest diploid genomes sequenced with a haploid genome size of more than 5 Gbp in seven large chromosomes (International Barley Genome Sequencing Consortium Mayer et al., 2012; Mascher et al., 2017). In order to check if there are any paralogs of HvDMC1 in barley genome we have screened its 2nd version that has been recently released (Mascher et al., 2017) with the use of the EnsemblPlants³ and the IPK Barley BLAST⁴ servers. Our analysis indicated that the HvDMC1 gene (HORVU5Hr1G040730) is located on chromosome 5 and has no paralogs in the genome (Supplementary Materials 1, 2), so our dmc1.b and dmc1.c mutants are good tools to study the function of this gene, because the risk of gene redundancy is very low.

An issue which is sometimes raised considering TILLING mutants is that observed phenotype may be caused by other, than analyzed, mutations in the genome (so called background mutations). Taking into consideration the size of barley genome and the overall mutation density found in the HorTILLUS population (ca. 1/500 kbp; Szurman-Zubrzycka et al., 2018), we can assume that each M₂ plant carries more than 10,000 mutations. However, vast majority of them occur in non-coding regions, since genes (annotated coding sequences) make up 1.3% of barley genome (65.3 Mbp; Mascher et al., 2017). So, statistically, the number of mutations in genes equals to ca. 130 and, probably, most of them are silent and/or do not affect the protein function. Therefore, the probability of the presence of other deleterious mutation in a gene related to the same process of interest is very low. Nevertheless, in order to further decrease this probability, we performed backcrosses of the identified mutants with their parent variety ‘Sebastian’, which reduced the number of (putative) background mutations by half. The homozygous mutants selected from the F₂ populations were phenotyped in this study. What is more, we observed that two different dmc1.c mutant lines (dmc1.c-3041 and dmc1.c-3223), that originated from different M₁ plants and possess different mutational background, show similar defects during meiotic divisions. These two lines share only the mutation leading to the R183K substitution in DMC1, whereas any putative background mutations differ between them, which strongly suggests that the identified mutation is responsible for this phenotype.

The role of DMC1 has been validated mostly in model plant species, such as Arabidopsis thaliana and rice. Here, we described functional analysis of HvDMC1 in barley, which belongs to the most important cereal species worldwide. Our barley TILLING population, HorTILLUS, has been used for mutational screening in the HvDMC1 gene. We have identified and characterized a new allele, named dmc1.c, responsible for abnormalities during meiosis. Two mutated lines, from different M₁ plants carrying the same mutation (G2571A that causes the R183K substitution), showed similar defects in this process, which strongly suggests that HvDMC1 is involved in the proper course of meiosis in barley. We conclude that DMC1 is required for DSB repair during meiosis, the process which has yet to be fully elucidated.

All datasets for this study are included in the manuscript and the Supplementary Files.

IS and DG conceived the project and designed the experiments. DG supervised the project. MS-Z, BB, MS-J, and JK conducted the research. MS-Z, IS, and DG wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.