Defects in meiotic chromosome segregation lead to unreduced male gametes in Arabidopsis SMC5/6 complex mutants

Abstract

Structural maintenance of chromosome 5/6 (SMC5/6) complex is a crucial factor for preserving genome stability. Here, we show that mutants for several Arabidopsis (Arabidopsis thaliana) SMC5/6 complex subunits produce triploid offspring. This phenotype is caused by a meiotic defect leading to the production of unreduced male gametes. The SMC5/6 complex mutants show an absence of chromosome segregation during the first and/or the second meiotic division, as well as a partially disorganized microtubule network. Importantly, although the SMC5/6 complex is partly required for the repair of SPO11-induced DNA double-strand breaks, the nonreduction described here is SPO11-independent. The measured high rate of ovule abortion suggests that, if produced, such defects are maternally lethal. Upon fertilization with an unreduced pollen, the unbalanced maternal and paternal genome dosage in the endosperm most likely causes seed abortion observed in several SMC5/6 complex mutants. In conclusion, we describe the function of the SMC5/6 complex in the maintenance of gametophytic ploidy in Arabidopsis.

Mutants defective in the SMC5/6 complex often fail to divide chromosomes during meiosis, leading to the production of diploid pollen and subsequently triploid offspring.

 

Introduction

Meiosis is the reductional division that prevents chromosome doubling at every sexual generation. Chromosomes undergo homologous recombination (HR) during the first meiotic division, which creates new combinations of alleles. In the second meiotic division, the number of genome copies per nucleus is reduced to one copy. In most flowering plants, including Arabidopsis (Arabidopsis thaliana), haploid microspores develop into a tricellular microgametophyte (pollen) containing two sperm cells and one vegetative cell (Kawashima and Berger, 2014). In the female sporophyte, three out of the four haploid megaspores die, while the remaining megaspore divides three times into a megagametophyte (embryo sac) consisting of the haploid egg cell, the diploid central cell, and other accessory cells. Seed development starts with double fertilization, whereby the fertilized egg gives rise to the diploid embryo and the fertilized central cell to the triploid endosperm. The endosperm, containing one paternal and two maternal genomes (1p:2m), nourishes and supports embryo growth. However, an unbalanced maternal to paternal genome dosage in the endosperm slows down or even arrests embryo development and compromises seed viability (Scott et al., 1998).

Structural maintenance of chromosomes (SMC) complexes are key factors mediating large-scale chromatin organization in different functional contexts (Jeppsson et al., 2014; Uhlmann, 2016). The cohesin complex is essential for a plethora of processes, including sister chromatid attachment or cis-regulatory loop formation (Nasmyth and Haering, 2009; Rowley and Corces, 2018). The condensin complex compacts chromosomes via an asymmetric loop extrusion (Ganji et al., 2018). The functions of an enigmatic SMC5/6 complex are strongly linked to the maintenance of genome stability (Kegel and Sjögren, 2010; Aragón, 2018; Díaz and Pecinka, 2018). However, recent studies suggest that the SMC5/6 complex is also involved in the regulation of transcription (Decorsière et al., 2016), maintenance of centromere structure (Gómez et al., 2013), suppression of immune responses (Yan et al., 2013), or prevention of disease (Payne et al., 2014; van der Crabben et al., 2016). SMC5, SMC6, and four NON-SMC ELEMENTs 1–4 (NSEs) represent evolutionarily conserved subunits. SMC5 and SMC6 form a heterodimer via their hinge domains, while their head domains are bridged by the NSE1–NSE3–NSE4 subcomplex, which exhibits DNA binding capacity and can open/close the SMC5–SMC6 ring. Furthermore, the SMC5 coiled-coil region serves as a docking platform for the E3 small ubiquitin modifier ligase NSE2 (also named HIGH PLOIDY2 [HPY2] and METHYL METHANE SULFONATE SENSITIVITY 21 [MMS21]). In addition, each major phylogenetic group contains two to three nonconserved subunits implicated in the loading of the SMC5/6 complex onto chromatin: Nse5 and Nse6 together with BRCT-containing protein 1 (Brc1) in fission yeast (Schizosaccharomyces pombe) or Regulator of Ty1 Transposition107 (Rtt107) in budding yeast (Saccharomyces cerevisiae; Pebernard et al., 2006; Leung et al., 2011) and SMC5–6 complex localization factor proteins 1 and 2 (SLF1 and SLF2) in mammals (Räschle et al., 2015). The putative plant functional homologs of NSE5 and NSE6 are ARABIDOPSIS SNI1-ASSOCIATED PROTEIN 1 and SUPPRESSOR OF NPR1-1, INDUCIBLE 1 (SNI1), respectively (Yan et al., 2013).

Besides an essential role in the maintenance of plant genome stability (Mengiste et al., 1999; Watanabe et al., 2009; Diaz et al., 2019), the Arabidopsis SMC5/6 complex controls apical meristem growth (Ishida et al., 2009), suppresses precocious flowering (Kwak et al., 2016) and hyper-immune responses (Li et al., 1999; Yan et al., 2013), and ensures normal gamete and seed development (Liu et al., 2014; Diaz et al., 2019; Zelkowski et al., 2019). While some of these functions have been well described at the phenotypic level through mutant analyses, the underlying mechanisms remain largely unknown. Here, we investigated the mechanism of seed abortion observed in several SMC5/6 mutants (Liu et al., 2014; Diaz et al., 2019), and show here that this phenotype is very likely due to paternally induced genome dosage imbalance in the endosperm. Some abnormal seeds escape this block and develop triploid offspring. Importantly, the defects originate as recombination-independent problems in chromosome segregation during meiosis of SMC5/6 complex mutants.

Results

Paternally inherited nse2 mutations cause abnormal seed development

We used the Arabidopsis NSE2 loss-of-function mutants (Supplemental Figure S1A) nse2-1 (Q115*) and nse2-2 (T-DNA; Ishida et al., 2009). Plants homozygous for either nse2 allele were viable with developmental abnormalities including reduced height, small siliques, and short roots (Supplemental Figure S1, B–D). Dry seeds from wild-type (WT) plants were light brown and of regular shape, while mutant seeds were variable in color, size, and shape, including light/dark brown, large/small, and regular/shrunken (Figure 1A). Thirteen days after controlled manual pollination (DAP), WT plants produced 94.4% normal seeds, 2.8% abnormal seeds, and 2.8% aborted ovules (plants/siliques/cases, n = 7/29/1,424; Figure 1B;Supplemental Table S1). In contrast, nse2-1 and nse2-2 plants bore, respectively, 34.7% and 30.8% healthy-looking seeds, 16.0% and 23.4% abnormal seeds, and almost 49.3% and 45.8% aborted ovules (plants/siliques/cases, n = 7/32/1,343 for nse2-1 and n = 7/29/1,253 nse2-2; Figure 1B;Supplemental Table S1), representing a significant increase in both aborted ovules and abnormal seeds (Fisher’s exact test, P < 0.005 or lower; Figure 1, C and D; Supplemental Table S2). The abnormal seeds were larger with a glossy surface and watery endosperm (Figure 1B). Embryos were fully developed in WT seeds, but arrested between the torpedo and cotyledon stages in abnormal nse2 seeds at 13 DAP (Figure 1E). At later stages, all abnormal seeds turned dark brown and shrunk (Figure 1A).

Figure 1.

Paternally inherited abnormal seed development in nse2 plants. A, Representative dry seeds of wild-type Col-0 (WT), nse2-1, and nse2-2. Note that the seeds from nse2-1 and nse2-2 plants represent a mixture of normal-sized, large, and aborted seeds. Scale bars = 1 mm. B, Dissected siliques 13 DAP. Aborted ovules are marked with yellow asterisks and abnormal seeds (typically larger, white or pale and partially transparent) with yellow arrowheads. Scale bar = 400 µm. C, D, Percentage of normal seeds (NS), aborted ovules (AO), and aborted seeds (AS) in manually pollinated WT, nse2 mutants and their reciprocal crosses. Source values and basic counts are provided in Supplemental Tables S1 and S2. Significance in Fisher’s exact test: ^-P > 0.05, *P < 0.005, **P < 0.00001. E, Representative embryos dissected from WT and nse2-1 seeds 13 DAP. Scale bars = 200 µm. WT seeds contain a mature embryo, while nse2 seeds have embryos arrested at the torpedo to cotyledon stages. F, Differential interference contrast micrographs of cleared nse2-2 ovules. Scale bars = 50 µm. (i), Typical WT embryo sac showing one central cell nucleus (ccn) and one egg cell (ecn) nucleus; (ii), embryo sac with one smaller and two large nuclei; (iii), embryo sac without any nuclei; (iv), ovule without an embryo sac.

To assess the parental contribution to these phenotypes, we performed reciprocal crosses between WT and nse2 plants (Figure 1, B–D; Supplemental Table S1). nse2-1 or nse2-2 plants pollinated by WT produced 2.0% and 3.7% abnormal seeds, respectively, which matched the percentage of 2.8% of abnormal seeds in self-pollinated WT. In contrast, pollination of WT plants with nse2-1 or nse2-2 pollen resulted in 14.4% and 21.5% of abnormal seeds, representing a 7.2- and 5.8-fold increase, respectively, relative to the reciprocal cross. This result indicated that the loss of NSE2 function causes paternally inherited aberrant seed development in Arabidopsis. We observed about 60% of aborted ovules in nse2 plants pollinated by WT, suggesting severe maternal defects in nse2 mutants as well. Therefore, we took a closer look at nse2-2 embryo sacs (n = 55; Figure 1F). We determined that 25.5% of embryo sacs look normal, 20% have three large nuclei, 10.9% have no nuclei, and 43.6% consist of ovules without embryo sac (Figure 1F[i–iv]). These severe defects suggested that either the two polar nuclei do not fuse into a central cell nucleus, that the megagametophytic nuclei degenerates, or that no megagametogenesis takes place. The presence of 45%–50% nondeveloping ovules in nse2 mutants suggested that the defective embryo sacs abort before fertilization. These experiments showed that the paternally induced nse2 defects are at least partially transmissible while the maternally induced defects result in substantial gametophytic lethality. Since abnormal seed development was caused paternally, we focused on male gametophyte development.

nse2 plants produce diploid sperm nuclei

To elucidate the paternally induced abnormal seed development, we first analyzed the viability of WT, nse2-1 and nse2-2 pollen by fluorescein diacetate (FDA) staining (Figure 2A). FDA staining indicated significantly lower pollen viability in nse2-1 (63.6%; n = 503; Supplemental Table S3; Fisher’s exact test, P < 0.001 or lower) and nse2-2 (52.4%; n = 609; Supplemental Table S3; Fisher’s exact test, P < 0.001 or lower) compared to WT (83.4%; n = 1,504) plants, as expected (Liu et al., 2014). However, we noticed surprisingly variable sizes in nse2 pollen, in sharp contrast with the uniform size of WT pollen (Figure 2A). We quantified this observation by measuring pollen area using the same data as for the FDA analysis (Figure 2B). The area of WT pollen ranged from 250 to 500 µm² with a peak at 440.6 µm². The smallest nse2-1 and nse2-2 pollen had an area of about 250 µm², but the largest pollen grains were ∼1,000 µm². In addition, the distribution of nse2 pollen sizes showed two peaks at 484.1 and 737.8 µm² for nse2-1, and 444.5 and 662.4 µm² for nse2-2. Hence, nse2 plants produce two cohorts of differently sized pollen.

Figure 2.

nse2 plants produce diploid pollen. A, Analysis of pollen viability by FDA staining in mature pollen of WT, nse2-1, and nse2-2 plants. Photographs of pollen were taken using differential interference contrast (DIC, pseudo-colored in magenta) while the FDA signals (pseudocolored in green) were observed by epifluorescence microscopy. Green signals indicate viable pollen. Note uniform sizes of pollen grains produced by WT and different sizes of pollen from nse2 mutant plants. Scale bars = 100 µm. B, Violin plots showing the area of individual pollen grains (µm²) in diploid (2×) and tetraploid (4×) WT and diploid nse2 plants. The numbers above violins indicate the peak area. C, Flow cytometric histogram of sperm ploidy in WT, nse2-1, and nse2-2. The x-axis shows relative DAPI intensity and the y-axis particle count. D, Bivariate scatter plot of sperm nuclei from WT and nse2-1 in ProHTR10:HTR10-mRFP (HTR10-mRFP) background. The x-axis shows relative DAPI intensity; the y-axis is the forward scatter parameter, indicating nuclei size. 1C and 2C nuclei populations are marked and the unmarked signals correspond to debris and organelles. 1C and 2C nuclei were sorted onto slides and used for the experiment shown in (E). E, Microscopy validation of enriched populations of sperm nuclei sorted based on their DNA content. Mature pollen of HTR10-mRFP, nse2-1 HTR10-mRFP, and nontransgenic WT were homogenized to eliminate vegetative nuclei, the remaining 1C and 2C nuclei were flow-sorted separately onto microscope slides and analyzed for RFP signals. Scale bar = 1 µm. F, Quantification of sorted nuclei observed from WT 1C, nse2-1 1C, and 2C in the HTR10-mRFP background.

Arabidopsis pollen size increases with nuclear DNA content (De Storme and Geelen, 2011). We thus hypothesized that the larger pollen might be polyploid. To test this hypothesis, we generated autotetraploid (4x) WT Arabidopsis by colchicine treatment and measured its pollen size. Indeed, pollen area ranged from 340 to ∼1,000 µm², with a distribution peak at 661.4 µm² (n = 357) that perfectly matched the large pollen seen in nse2-2, although tetraploid pollen was still smaller than the largest pollen grains of nse2-1 plants. Next, we used flow cytometry to obtain direct evidence of the ploidy of nse2 pollen nuclei. For this analysis, it is essential to notice that Arabidopsis haploid pollen nuclei rest in different stages of cell cycle and therefore differ as to their DNA content (De Storme and Geelen, 2011). Sperm cell nuclei have DNA contents of 1C while vegetative cell nuclei have a 2C nuclear DNA content. We collected mature pollen from WT, nse2-1 and nse2-2 plants, destroyed vegetative nuclei as described (De Storme and Geelen, 2011), and measured ploidy of the remaining fraction. From WT preparations, we detected almost exclusively 1C nuclei, indicating that our experimental conditions effectively eliminated 2C vegetative nuclei. For nse2-1 and nse2-2 preparations, we obtained both 1C and 2C nuclei (Figure 2C), further indicating that nse2 plants produce not only haploid but also diploid sperm cells.

To confirm this result with another approach, we produced a double homozygous line of nse2-1 expressing the sperm nucleus-specific marker line ProHTR10:HTR10-mRFP (Histone three related [HTR10] fused to monomeric red fluorescent protein [mRFP]; (Ingouff et al., 2007) and repeated the analysis. We only detected 1C nuclei (n = 427) in the WT HTR10-mRFP background. In contrast, nse2-1 HTR10-mRFP pollen presented 68.8% of 1C and 31.2% of 2C nuclei (n = 245 and n = 111, respectively; Figure 2D). We then sorted nuclei onto microscopic slides according to their ploidy and inspected them using epifluorescence microscopy. In total, 93.6% of WT HTR10-mRFP and 85.5% of nse2-1 HTR10-mRFP 1C nuclei contained RFP fluorescence (n = 109 and n = 304, respectively; Figure 2, E and F; Supplemental Figure S2A). Importantly, 83.1% of nse2-1 HTR10-mRFP 2C nuclei also showed mRFP fluorescence (n = 207; Figure 2E, EF; Supplemental Figure S2A). Together, these results provided solid evidence that nse2 plants produce diploid sperm nuclei.

Defective meiosis leads to unreduced microspores in nse2 plants

2C nse2 sperm nuclei indicated that the defects originated during microsporogenesis. Therefore, we analyzed the progression of meiosis in WT and nse2-2 pollen mother cells (PMCs). Chromosome spreads stained with 4′,6-diamidino-2-phenylindole (DAPI) revealed no obvious differences between nse2-2 and WT at prophase I. nse2-2 PMCs displayed typical thread-like chromosomes at leptonema, homologous chromosome pairing at zygonema, and full synapsis at pachynema (Figure 3A). Chromosomes began to de-synapse and condense at diplonema and five bivalents were visible at diakinesis and metaphase I. However, despite the presence of five aligned bivalents, the appearance of the chromatin was not normal in 72.0% of nse2-2 metaphase I PMCs (total n = 93). The bivalents were more stretched and elongated than in WT, chromosomes progressively lost compactness, and the constrictions in the chromatin began to be visible as segregation took place at anaphase I (Figure 3A). At almost all anaphase I figures (98.4%; n = 122), nse2-2 chromosomes appeared de-condensed and thread-like chromatin fibers or fragments were visible, spanning the region between the segregating chromosomes (Figure 3A).

Figure 3.

Characterization of male meiosis and meiotic products in nse2. A, First meiotic division. Pachynema: Full synapsis was detected in the WT and nse2-2. Metaphase I: Five bivalents were observed in nse2-2, but chromatin was less condensed than in WT cells, showing constrictions and fragments. Anaphase I: nse2-2 chromosomes appeared de-condensed and thin thread-like fragments of chromatin were visible spanning the region between all the segregating chromosomes (arrows). Telophase I: A barrier formed by multiple chromosome fragments is apparent between the two groups of segregated chromosomes in the mutant (arrows). Scale bars = 5 µm. B, Representative images of the phenotypes in metaphase II and microspores. Metaphase II: WT figures are followed by nonreduced meiocyte (nr), abnormal chromosome segregation (as), and chromosome fragmentation (f). Fragments are marked with black arrows. Microspores: WT tetrad (te), followed by mutant triad (tr), dyad (dy), and monad (mo). Scale bars = 5 µm. C, Quantification of the different phenotypes observed in nse2-2 second meiotic division. Only normal meiotic figures were observed in WT cells. D, Quantification of different meiotic products observed in nse2-2. E, Examples of meiotic products observed in the double homozygous nse2-2 qrt1-4 mutant background. Scale bars = 20 µm. F, Quantification of meiotic products observed from the WT, nse2-1, and nse2-2 in the qrt1-4 background.

There is a cycle of de-condensation and re-condensation during the second meiotic division, and the chromosomes show the maximum degree of condensation in metaphase II of PMCs. During prophase II, two sets of five chromosomes are separated by a clear band of numerous organelles and re-condense before metaphase II in WT plants (Brownfield et al., 2015). At anaphase II, sister chromatids segregated to generate a tetrad with four balanced nuclei (Figure 3B). At prophase II, the frequency of normal-looking cells with two sets of five chromosomes decreased to only 31.3% in nse2-2 (n = 144). The remaining meiocytes showed some of the following problems (Figure 3, B and C). (1) Nuclei containing two sets of homologous chromosomes (nonreduced nuclei). In such PMCs, the organelles were not organized in a defined band and appeared throughout the cytoplasm (19.4%), preventing the formation of two defined nuclei with five chromosomes (Supplemental Figures S3 and S4A). (2) One or more sets of chromosomes displaying chromosome fragmentation (15.3%). And (3) Chromosome bridges and chromatin masses linking them, suggesting unresolved anaphase I defects (34.0%); this last phenotype possibly combined the previous two. As a consequence of such abnormalities, we observed a range of meiotic products (n = 591) consisting of tetrads (49.2%), triads (5.1%), and dyads (45.7%) in nse2-2 plants (Figure 3, B and D). Putative monads were detected but not quantified, as they are hard to distinguish with this method. Instead, we used nse2 qrt1 double mutant plants to quantify monads (Figure 3, E and F). We confirmed the chromosome constitution of the dyads (10 chromosomes in each nucleus) by performing immunolocalization to detect CENTROMERIC HISTONE H3 (CENH3; Supplemental Figure S4B).

To quantify meiotic products in an independent manner, we produced nse2-1 qrt1-4 and nse2-2 qrt1-4 double mutants. The qrt1 mutation causes a stable association of the microspores arising from one meiosis (Preuss et al., 1994). The NSE2 qrt1-4 plants produced on average 3.9 microspores per meiosis (575 microspores/148 meiotic products) and the products included 94.6% tetrads, 2.0% triads, 0.7% dyads, and 2.7% monads (Figure 3F; Supplemental Table S4). We measured an average of 3.2 microspores per meiosis in nse2-1 qrt1-4 (1,042 microspores/322 meiotic products) and only 2.4 microspores per meiosis in nse2-2 qrt1-4 (479 microspores/202 meiotic products). In nse2-1 qrt1-4, there were 57.8% tetrads, 14.0% triads, 19.2% dyads, and 9.0% monads (Figure 3F; Supplemental Table S4). Similarly, we observed a higher frequency of nontetrad meiotic products in nse2-2 qrt1-4 (20.8% tetrads, 17.3% triads, 40.1% dyads, and 21.8% monads). The vast majority of NSE2 qrt1-4 microspores had a normal shape (99.3%) and was viable (95.3%; Supplemental Table S4). nse2-1 qrt1-4 showed 26.7% abnormal, small and shrunken microspores and only 65.0% of pollen was viable (Supplemental Figure S2B and Supplemental Table S2). nse2-2 qrt1-4 showed 26.3% abnormal, small and shrunken microspores and only 40.7% of pollen was viable (Supplemental Table S4).

Altogether, the initial analysis of nse2 meiosis revealed abnormal progression in some cells (there were 31.3% of apparently normal cells) with the defects originating/emerging from the later stages of meiosis I. Importantly, two major problems in chromosome behavior were observed (1) chromosome fragmentation and (2) defects in chromosome segregation.

Chromosome fragments are the consequence of SPO11-induced DNA breaks in nse2 plants

To ascertain whether SPO11 function is involved in the chromosome fragmentation phenotype observed at anaphase I in nse2, we generated the spo11-1-5 nse2-2 double mutant. Homozygous spo11-1-5 plants showed reduced height and partial sterility, phenotypes that became more pronounced in the spo11-1-5 nse2-2 double mutant (Supplemental Figure S5). Meiosis in the spo11 single mutant is characterized by the presence of ten univalents at metaphase I (Grelon et al., 2001). In contrast with nse2, chromosome fragmentation was almost absent, with all ten univalents detected at metaphase I in the double mutant (n = 24; Figure 4A). The suppression of chromosome fragmentation in spo11 nse2 demonstrated that fragments produced by the absence of NSE2 are caused by the failure to repair joint molecules (JMs) generated from SPO11-induced double-strand breaks (DSBs). To further investigate the HR process in nse2, we analyzed RAD51 foci numbers at pachynema and monitored synaptonemal complex formation by detecting its central component ZYP1 (Higgins et al., 2005; Supplemental Figure S6A). Defects in DSB repair during early stages are associated with persistent RAD51 foci on pachytene chromosomes (Wang et al., 2012). The numbers of RAD51 pachytene foci were not altered in nse2 (n = 36 for WT and n = 38 for nse2-2; two-tailed Mann–Whitney–Wilcoxon test, P = 0.891; Supplemental Figure S6B and Supplemental Table S6), supporting the hypothesis that the SMC5/6 complex activity is required after RAD51. We also did not detect a delay in these early prophase I stages according to the quantification of cells with partial synapsis (27.78% in WT, n = 36, versus 13.16% in nse2-2, n = 38; Fisher’s exact test, P = 0.1526; Supplemental Table S7). This result also demonstrated that there is no increased number of DSBs in the mutant and no delay in recombination, confirming a conserved role for the complex in resolving aberrant JMs (Xaver et al., 2013).

Figure 4.

Analysis of male meiosis in nse2 spo11 and nse2 osd1 double mutants. A, Meiotic figures in spo11-1-5 and nse2-2 spo11-1-5 plants. Prophase I to anaphase I show chromosome univalents in both genotypes. Note the absence of chromosome fragments as observed in nse2-2 spo11-1-5 anaphase I and telophase I (Figure 3A). Segregation to more than two poles can be observed in prophase II, resulting in formation of polyads. Scale bars = 5 µm. B, Percentage of meiocytes with the specified phenotypes in spo11-1-5 and nse2-2 spo11-1-5 plants. Source data are in Supplemental Table S5. C, Percentage of meiotic cells with given number of products in spo11-1-5 and nse2-2 spo11-1-5 plants. Source data are in Supplemental Table S9. D, Representative phenotypes of immature meiotic products from osd1-3 and nse2 osd1-3 plants. Microspores: dyad (dy) and monad (mo). Scale bars = 5 µm. E, Quantification of dyads and monads produced in osd1-3 and nse2 osd1-3 double mutants. F, Immunolocalization of α-tubulin. Representative images of telophase I (TI) and telophase II (TII) in the WT and nse2-2. In nse2-2, the microtubule bundles are diffuse, have lower density and a disorganized appearance, with some miss-localized microtubules (arrow). Scale bars = 5 µm.

The problems in chromosome segregation thus appeared to be mostly independent of the meiotic recombination defects in nse2-2. We noticed differences in the percentage of the different meiotic products in the nse2-2 spo11-1-5 double mutant relative to either nse2-2 or spo11-1-5 plants. In the double mutant, the frequency of dyads was lower than in the nse2-2 single mutant (4.02% versus 45.69%, Fisher’s exact test, P < 0.0001; Supplemental Figure S8) and there were polyads (which are not present in nse2). These differences may be explained by the presence of univalents at metaphase I. In addition, the nse2-2 spo11-1-5 double mutant has a higher frequency of dyads (4.02% versus 0.98%), and a lower frequency of polyads (38.69% versus 52.77%) than the spo11-1-5 single mutant (χ² test, P = 0.0014; Supplemental Figure S9), revealing that the formation of unreduced gametes in nse2 is a consequence of problems independent of HR.

Hence, the chromosomal fragmentation phenotype in nse2 can be explained by the role of SMC5 in the resolution of recombination intermediates but does not clarify how the unreduced gametes are produced.

Unreduced gametes arise via an HR-independent process during the meiosis I and II

To better understand the origin of unreduced gametes, we generated the nse2 osd1 double mutant. OMISSION OF SECOND DIVISION (OSD1) is a negative regulator of the Anaphase promoting complex/Cyclosome needed for the second meiotic division, and its absence leads to diploid gametes (Cromer et al., 2012). The nse2 mutant was crossed with heterozygous osd1 plants and meiosis was analyzed in double homozygous F₂ plants. Homozygous osd1-3 plants only produced dyads (n = 199; Figure 4, D and E). In double homozygous nse2-1 osd1-3 and nse2-2 osd1-3 plants, we observed 19.5% and 40.6% monads, respectively (n = 174 and n = 64, respectively; Figure 4, D and E). The presence of monads in the nse2 osd1 double mutant provided strong genetic evidence that the chromosome segregation problems in nse2 plants are due to anaphase I failures. However, it should be noted that nse2 mutants (in the qrt1-4 background) produce not only dyads but also 9%–21.8% monads (Figure 3, E and F), indicating that in about 1 out of 10 meioses in nse2, there is a nonreduction in both anaphases I and II.

Therefore, the problems in the segregation of homologous chromosomes appeared to be the main foundation of the unreduced gametes in nse2. Despite normal-looking five bivalents at metaphase I in nse2 meiocytes, homologous chromosomes did not complete their migration to opposite poles on some occasions, and they were closer than in WT meiocytes. This observation was strongly associated with the improper position of chromosomes relative to the organelle band, which in these meiocytes is not positioned to function as a physical barrier (Supplemental Figures S3 and S4). As a consequence, instead of being equally divided on either side of the band, all chromosomes appeared only on one side at the end of the first meiotic division (Figure 3B, nr). In the second division, the repetition of this abnormal organization will lead to a monad while the normal progression will produce a dyad instead of tetrad.

To determine whether the chromosome segregation abnormalities are associated with problems in spindle organization, we conducted α-tubulin immunolocalization. In general, WT and nse2 PMCs behaved similarly, with microtubules organized as a radial array around each nucleus at prophase I and as polar oriented spindles at metaphase I. However, we observed abnormal spindle structures in mutant meiocytes (Figure 4F). The microtubule bundles had a lower density in nse2 than in WT at telophase I, which was consistent with the appearance of chromosome organization problems. In addition, microtubules were disorganized around the chromosomal fragments and/or displaced sideways to the spindle periphery. At meiosis II, the spindles also showed a reduced number of microtubules and appeared more diffuse in nse2-2 meiocytes (Figure 4F).

Based on these results, we conclude that the defects in chromosome segregation leading to unreduced gametes can occur in both anaphases I and II (being more frequent in the former) and are hallmarked by disorganized microtubules in the spindle and an aberrant position of the chromosomes relative to the organelle band.

nse2 plants produce triploid offspring

Unreduced gametes may lead to a polyploid progeny. Therefore, we analyzed ploidy in the progeny of self-pollinated WT and nse2 plants by flow cytometry. In total, 97.6% WT, 87.6% nse2-1, and 63.4% nse2-2 seeds germinated and grew at least until the cotyledon stage (Table 1 and Figure 5A). All 110 WT plants were diploid. In contrast, 7.6% nse2-1 plants (n = 380) and 8.8% nse2-2 plants (n = 147) were triploid (Table 1 and Figure 5B). We confirmed the triploidy of selected individuals by chromosome counting (Figure 5C). The triploids were sometimes slightly larger or smaller than diploids, but otherwise within the range of nse2 phenotypes (Figure 5A, arrows). Surprisingly, we detected no nse2 aneuploids by flow-cytometric (n = 1,166) or by cytological (n = 6) analysis. We also did not identify any tetraploids, suggesting that the increased ploidy is transmitted through only one parental gamete. However, we cannot fully exclude trisomy with whole or fragmented chromosomes.

Table 1.

Ploidy levels of offspring plants from nse2 and WT parents and their F₁ hybrids. n = number, 2x = diploid, 3x = triploid, 4x = tetraploid.

Mother	Father	Sown (n)	Germinated (n)	Germination rate (%)	2× (%)	3× (%)	4× (%)	Aneuploids (%)
2× WT	2× WT	113	110	97.3	100	0	0	0
2× nse2-1	2× nse2-1	434	380	87.6	92.4	7.6	0	0
2× nse2-2	2× nse2-2	232	147	63.4	91.2	8.8	0	0
WT 2×	2× nse2-1	283	270	95.4	93.3	6.7	0	0
WT 2×	2× nse2-2	252	174	69.0	94.8	5.2	0	0
2× nse2-1	2× WT	86	82	95.3	100	0	0	0
2× nse2-2	2× WT	119	113	95.0	100	0	0	0
4× WT	2× WT	88	64	72.9	0	100	0	0
4× WT	2× nse2-1	299	284	94.3	0	78.4	21.3	0.3
4× WT	2× nse2-2	160	147	91.9	0	57.1	41.5	1.4
2× WT	4× WT	179	10	5.6	0	100	0	0
2× nse2-1	4× WT	112	31	27.7	0	100	0	0
2× nse2-2	4× WT	72	3	4.2	0	100	0	0

Figure 5.

SMC5/6 complex mutants produce triploid offspring. A, Two-week-old in vitro-grown seedlings of the WT, nse2-1, and nse2-2. Triploid seedlings revealed by flow cytometry are indicated by red arrows. Scale bar = 1 cm. B, Examples of flow cytometry histograms showing representative profiles of peaks from diploid (2×) WT and triploid (3×) nse2 plants. C, Representative mitotic prophase figures in 2× WT and 3× nse2-2 plants. Scale bars = 5 µm. D, WT, nse4a-2 and sni1-3 mutant phenotypes. Left column: dry seeds. Scale bars = 1 mm. Right column: 2-week-old in vitro-grown seedlings. Triploid seedlings are indicated by red arrows. Scale bar = 1 cm.

To corroborate the uniparental induction of triploidy, we tested reciprocal crosses between WT and mutant (Table 1; throughout the paper female × male). Ploidy measurements of the F₁ 2x (diploid) nse2 × 2x WT hybrids revealed only diploids. In contrast, there were 6.7% and 5.2% of triploids among F₁ 2x WT × 2x nse2-1 or 2x WT × 2x nse2-2 hybrids, respectively. A fusion of haploid maternal (m) and diploid paternal (p) gametes will lead not only to a triploid embryo, but also to tetraploid endosperm with an imbalanced parental dosage of 2m:2p genomes (Scott et al., 1998). Excess of the paternal genome will result in delayed or halted endosperm cellularization, arrested embryo development, and finally reduced or compromised seed viability (Köhler et al., 2012), which all align with the phenotypes of nse2 mutants (Figure 1A). Consequently, the observed frequencies may be an underestimation because many triploid seeds die. To offer a balanced endosperm environment for diploid pollen, we reciprocally crossed 2x nse2 plants with 4x WT plants. All crosses between 2x maternal and 4x paternal plants produced only few germinating seeds (2x WT × 4x WT, 5.6%; 2x nse2-1 × 4x WT, 27.7%; 2x nse2-2 × 4x WT, 4.2%; Table 1). In contrast, the control cross between 4x WT maternal and 2x WT paternal plants resulted in 72.9% germination and all the plants were triploids (Table 1). The germination rate of seeds derived from the crosses between 4x WT and 2x nse2-1 or 2x nse2-2 was over 90% (94.3%, n = 299, and 91.9%, n = 160, respectively) and hence better than in 4x WT × 2x WT cross (72.7%, n = 88). The F₁ hybrids of 4x WT × 2x nse2-1 (n = 282) included 78.4% triploids, 21.3% tetraploids and 0.3% aneuploids. Similarly, the F₁ hybrids of 4x WT × 2x nse2-2 (n = 147) plants were represented by 57.1% triploids, 41.5% tetraploids, and 1.4% aneuploids (Table 1). Hence, the frequency of viable polyploids derived from 2x nse2 plants can be several-fold increased by crossing with 4x WT mothers, and our results strongly suggest that the triploid nse2 offspring is caused exclusively by unreduced male gametes (diplogametes).

Mutations in NSE4A and SNI1 also lead to triploid offspring

To test whether polyploidy also occurs in other Arabidopsis SMC5/6 complex mutants, we first checked the dry seeds from diploid smc6b-1, nse4a-2, nse4b-1, and sni1-3 plants. The seeds of 2x smc6b-1 and 2x nse4b-1 plants were normal (Supplemental Figure S7). However, seeds from 2x nse4a-2 and 2x sni1-3 plants showed different sizes and shapes, including dark brown, large, and shrunken seeds (Figure 4D). By flow cytometry-based ploidy analysis, all smc6b-1 and nse4b-1 plants (n = 103 and 120, respectively; Table 2) were diploid. However, we observed 9.6% triploid nse4a-2 plants (n = 271). As with nse2, the triploid plants were sometimes differently sized, but generally within the range of standard nse4a-2 phenotypes (Figure 5D, right column, arrows). Analysis of the reciprocal crosses between 2x nse4a-2 and 2x WT revealed that the triploidy is also caused exclusively by the male gamete (Table 2). Since NSE4A is expressed from both maternal and paternal gametophytes (Díaz et al., 2019), this result suggests that the defective maternal gametes do not give rise to viable offspring. Homozygous sni1-3 mutants phenotypically resemble severely affected nse2 plants, but are almost fully sterile. From homozygous 2x sni1-3 parent, we obtained 137 offspring plants, of which 68.6% were triploid (Table 2). The reciprocal crossing of 2x WT × 2x sni1-3 confirmed that all triploids are induced from the paternal side (Table 2).

Table 2.

Ploidy levels of offspring plants from nse4a-2, sni1-3, and WT parents and their F₁ hybrids. n = number, 2× = diploid, 3× = triploid.

Mother	Father	Sown (n)	Germinated (n)	Germination rate (%)	2× (%)	3× (%)
2× WT	2× WT	113	110	97.3	100	0
2× nse4a-2	2× nse4a-2	300	271	90.3	90.4	9.6
2× WT	2× nse4a-2	87	83	95.4	86.7	13.3
2× nse4a-2	2× WT	93	93	100	100	0
2× sni1-3	2× sni1-3	740	137	18.5	31.4	68.6
2× WT	2× sni1-3	160	22	13.8	31.8	68.2
2× sni1-3	2× WT	98	81	82.7	100	0
2× smc6b-1	2× smc6b-1	119	103	86.6	100	0
2× nse4b-1	2× nse4b-1	135	120	88.9	100	0

Discussion

The SMC5/6 complex is an evolutionary conserved ATP-dependent molecular machine involved in the maintenance of nuclear genome stability (Aragón, 2018). Here, we discovered that loss of function in SMC5/6 complex subunits NSE2, NSE4A, and SNI1 lead to defective male meiosis, the formation of diploid microspores and triploid offspring in Arabidopsis. The absence of triploids in smc6b-1 is most likely due to partial functional redundancy among Arabidopsis SMC6 paralogs (Watanabe et al., 2009; Yan et al., 2013). Testing homozygous smc6a smc6b plants is not possible due to their early embryo lethality (Watanabe et al., 2009; Yan et al., 2013). Lack of triploids for nse4b-1 is in agreement with the fact that NSE4B does not appear to be connected to DNA damage repair (Diaz et al., 2019).

We traced the origin of the triploidy defects to male meiosis. At the onset of meiosis, nse2 meiotic figures show normal five bivalents at metaphase I (Figure 3A), indicating the formation of normal crossovers between homologs, ensuring the presence of obligatory chiasma. We further confirmed normal progression of HR and complete synapsis at pachynema (Supplemental Figure S6). From metaphase I onward, we observed two abnormal phenotypes in nse2 mutants. The first phenotype was characterized by chromosome fragmentation during meiosis I, while the second phenotypes were associated with defects in chromosome segregation mainly during meiosis I. The second phenotype is relevant for the formation of unreduced gametes, but both phenotypes will be discussed and visualized as the working model of SMC5/6 complex action in Arabidopsis reproductive development (Figure 6).

Figure 6.

Model for abnormal phenotypes in Arabidopsis SMC5/6 complex mutant meiosis. A, Schematic representations of the PMC, TI and TII and meiotic products. The chains of events are described from left to right. Part of the nse2 meiocytes goes through two standard reductional divisions corresponding to a tetrad with four haploid microspores (n = 1× = 5; green). Occasional nonreduction (NR) in the second division may produce a triad, most of them with one diploid (n = 2× = 10; yellow) and two haploid microspores. A more common defect is chromosome nonreduction in the first meiotic division, observed as grouping of all chromosomes on one side of the organelle band in telophase I. This is followed either by a standard second meiotic division and development of a dyad with two diploid microspores or by a defective second meiotic division giving rise to a monad (n = 4× = 20; red). The rightmost scenario shows that SPO11 activity in meiotic prophase I may lead to extensive chromosome fragmentation or/and abnormal segregation (F/AS) that requires SMC5/6 for repair. Meiocytes with extensively fragmented chromosomes are not viable and this situation is lethal. B, Effects of aberrant nse2 meiosis on seed development. Only embryo sacs with haploid egg cell nuclei (ecn) and diploid central cell nuclei (ccn) are viable in nse2 plants. Depending on the type of microspore, the resulting embryos and endosperm vary in their ploidy. The diploid and the tetraploid microspores lead to endosperm with an unbalanced ratio of the maternal (M) and paternal (P) genome ratio (color scale). This results in the abortion of about 2/3 of the seeds with triploid embryos and the total absence of seeds with a strongly unbalanced hexaploid endosperm.

The nse2 chromosome fragmentation phenotype is associated with frequent bridges, entanglements, and concatenations from metaphase to telophase I (Figure 3A). Here, we confirmed that these problems in meiotic HR are due to the presence of toxic recombination HR intermediates, as evidenced by a drastic reduction in chromosome fragments in the spo11 nse2 double mutant (Figure 4, A and B). We also demonstrated that these recombination failures are not a consequence of a higher number of DSBs or of a delay in the repair of these DSBs, as revealed by the immunolocalization results of RAD51 in pachynema (Supplemental Figure S6). These observations resemble the situation in budding yeast, where the accumulation of HR intermediates in SMC5/6 complex mutants result in JMs (Copsey et al., 2013; Xaver et al., 2013; Menolfi et al., 2015). Arabidopsis meiotic cells with highly fragmented chromosomes are likely to result in nonviable gametes or may not even progress to meiosis II, as suggested by the absence of aneuploid nse2 offspring. In this sense, the meiotic phenotype of nse2 is similar (although more drastic) to that described for nse4a-2 (Zelkowski et al., 2019). nse2 single mutant plants do not create polyads (Figure 3D). Polyads represented 52.8% of meiotic products in spo11 and their frequency is reduced to 38.7% in spo11 nse2 plants (Figure 4, A and C). Therefore, NSE2 acts upstream of SPO11 and partially promotes polyad phenotype of spo11 plants. In summary, the fragmented chromosomes in nse2 meiocytes strongly reduce the number of viable gametes but do not explain the occurrence of unreduced gametes and eventually triploid offspring.

Then, how are the unreduced gametes generated? The most plausible explanation is that in some cells, chromosomes do not separate to two defined poles during meiosis I. The result would be similar to what occurs in a first division restitution (Bretagnolle and Thompson, 1995), but this phenotype does not occur in all nse2 cells and is the consequence of a more complex process. Diplogametes can originate from the omission of the first and/or the second meiotic division. On the one hand, we confirmed that the unreduced gametes in nse2 arise mainly from failures during the first meiotic division by analyzing the osd1 nse2 double mutant (Figure 4, D and E). On the other hand, we also observed about 9%–21.8% of monads in nse2-1 and nse2-2 mutants, respectively, suggesting that both meiotic divisions can be affected. Furthermore, nse2-2 anaphase I chromosomes show reduced chromatin condensation (Supplemental Figure S3). Such phenotype may be related to a loss of chromosome architecture due to the functional interplay between condensin and the SMC5/6 complex, as shown in animals (Hong et al., 2016; Hwang et al., 2017). As a consequence, after the formation of the organelle band (Brownfield et al., 2015), all chromosomes will appear at the same pole during the second meiotic division in some cells (Figure 3B, Supplemental Figure S3). In addition, the absence of two defined poles may occur as a consequence of disrupted organization and orientation of the spindle (Brownfield and Köhler, 2010). In this context, we observed partially abnormal spindle organization in nse2-2 (Figure 4F). It should be stressed that chromosomes and kinetochores play a more important role in spindle morphogenesis in plants compared to animals due to the absence of centrosomes or proper microtubule organizing centers (Zhang and Dawe, 2011). We speculate that the kinetochores built on de-condensed centromeres might not be fully functional in organizing the spindle. Alternatively, the timing of release of cohesion, spindle elongation and spindle disassembly may be deregulated in nse2 because of a delayed anaphase I.

We showed that the diplogametes are produced exclusively by the male organs. Analysis of nse2 ovules revealed the absence of nuclei within embryo sacs and the presence of unfused (presumably polar) nuclei (Figure 1F), suggesting that any failures in female meiosis and/or post-meiotic development will lead to ovule abortion. Hence, megaspores may be more sensitive to the presence of unstable, damaged, incomplete, or excessive genomes. Normally, the number of eggs (and not pollen) determines the number of offspring. Therefore, quality control may be easier to exert through female gametophytic development. From an evolutionary point of view, strictly selected healthy female gametophytes will then be able to recognize less strictly controlled unhealthy male gametes, based on the compromised 2:1 maternal to paternal genome ratio in the endosperm, and will terminate the development of most such seeds (Batista and Köhler, 2020).

The frequency of triploids was around 10% in nse2 and nse4a-2 plants and reached up to about 70% in sni1-3 plants. SNI1 is a DNA binding transcription factor and was originally identified as a suppressor of immune responses in Arabidopsis (Li et al., 1999). SNI1 may be a functional homolog of yeast NSE6 involved in SMC5/6 complex loading to the sites of DNA damage (Yan et al., 2013). Loss of SNI1 may prolong the persistence of specific types of DNA damage and consequently also the frequency of unreduced gametes. There are several other proteins whose loss-of-function mutants produce a high number of diplogametes. TARDY ASYNCHRONOUS MEIOSIS/CYCLIN A1;2 (TAM/CYCA1;2) phosphorylates OSD1. tam and osd1 plants fail to enter meiosis II. Unreduced gametes are produced by both the male and the female meiocytes, as indicated by the offspring ploidy corresponding to about 60% triploids, 25%–40% tetraploid and only 2%–7% diploids (D’Erfurth et al., 2010). Here, we showed a synergistic effect between nse2 and osd1 mutations. Mutations in PARALLEL SPINDLE 1 (PS1) cause abnormal spindle orientation in meiosis, which leads to the production of about 60% dyads in male meiocytes and 30% triploid offspring (d’Erfurth et al., 2008). Somewhat similar are the phenotypes of jason (jas) mutants. jas male meiocytes produce a high number of diplogametes and 29%–40% of offspring were triploid (Erilova et al., 2009; De Storme and Geelen, 2011). SMC5/6 mutants differ from the above-described cases by a severe negative effect on female meiosis, absence of 4x offspring (tam2) and extensive chromosome fragmentation in some meiocytes. Therefore, we propose that the SMC5/6 complex represents a novel diplogamete suppressor pathway. Further experiments will be directed towards unraveling the regulatory network under which the SMC5/6 complex operates in this process.

Materials and methods

Plant materials and growth conditions

All lines used in this study were in the Columbia-0 (Col-0) background. We used the following mutants: nse2-1/hpy2-1 (Ishida et al., 2009), nse2-2/hpy2-2/mms21-2 (SAIL_77_G06), sni1-3 (SAIL_34_D11), smc6b-1 (SALK_101968C), nse4a-2 (GK-768H08), nse4b-1 (SAIL_296_F02), spo11-1-5 (SALK_009440), qrt1-4 (SALK_024104C), and reporter lines ProHTR10:HTR10m-RFP (Ingouff et al., 2007). The osd1-3 mutant was kindly provided by Dr Claudia Köhler; the T-DNA in this mutant is inserted 58 bp after the start codon of the second exon, resulting in a truncated 144-amino acid OSD1 protein (Heyman et al., 2011). Plant zygosity was monitored by PCR (see primers in Supplemental Table S10) and/or mutant phenotype if possible. Double mutants were generated by crossing and selection in the F₂ and F₃ generations. All lines were used as homozygotes unless stated otherwise.

Tetraploid Arabidopsis Col-0 wild-type (4x WT) was generated by submerging 2-week-old in vitro-grown diploid seedlings in 0.1% (w/v) colchicine (Sigma-Aldrich) in the dark at room temperature for 2 h. Subsequently, seedlings were gently washed with copious amounts of tap water, transplanted to soil and grown until maturity. Seeds were collected from individual plants and bigger seeds were selected and propagated into plants for ploidy measurements (see below). One tetraploid individual was self-pollinated to generate 4x WT.

For cultivation in soil, seeds were placed on the surface of soaked soil and stratified for two days at 4°C in the dark before the pots were moved to an air-conditioned chamber with controlled long-day conditions (16-h-light/8-h-dark cycle, 21°C day and 19°C night temperature, 150-μmol photons m⁻² s⁻¹ light intensity provided by Philips fluorescent tube MASTER TL-D 18W/840, catalog NO. ELSZZA0047412). For in vitro growth, Arabidopsis seeds were surface sterilized (70% ethanol with 0.5% Triton X-100 [v/v]) for 10 min and washed three times with sterile water. Air-dried seeds were sown on half-strength Murashige and Skoog agar medium, stratified in the dark for 2 days at 4°C and then cultivated in a climatic chamber (Percival) under 16-h-light/8-h-dark cycle, 21°C day and 19°C night temperature, 150-μmol photons m⁻² s⁻¹ light intensity provided by Philips fluorescent tubes as above.

DNA isolation and PCR

For genotyping of T-DNA mutants, one rosette leaf was mixed with the Phire Plant Direct PCR Kit (Thermo Scientific) dilution buffer and 1 μL was used for PCR according to the kit instructions. Genotyping primers are listed in Supplemental Table S10.

Ploidy measurements and flow cytometry

To minimize any potential selection bias, seeds were collected per silique. The siliques were taken from the central part of the main stem. All seeds per silique were sown and analyzed. To determine somatic ploidy levels, one to two young leaves from a 2-week-old seedling were chopped with a razor blade in 500-μL Otto I solution (0.1 M citric acid, 0.5% [v/v] Tween-20). The suspension of nuclei was filtered through a 50-µm nylon mesh and stained with 1 mL of Otto II solution (0.4 M Na₂HPO₄) containing 2 µg DAPI. Ploidy was analyzed on a Partec PAS I flow cytometer with 2× WT plants used as an external standard. Subsequent nse2 samples were prepared by simultaneous chopping of equal amounts of tissue from both nse2 and WT genotypes. In the mutant samples, aneuploid plants would be detected based on the presence of double peaks.

To determine ploidy levels of male gametes, the samples were prepared as described (Borges et al., 2012). The nuclei-enriched pellet was resuspended in 1 mL of sperm extraction buffer (1.3-mM H₃BO₃, 3.6-mM CaCl₂, 0.74-mM KH₂PO₄, 438-mM sucrose, 5.83-mM MgSO₄, 7-mM MOPS at pH 6) containing 2-µg DAPI. The different nuclei populations were sorted separately on microscope slides by FACSAria (BectonDickinson) flow-sorter. Slides were dried at room temperature for 1 h then mounted with 5-μL Vectashield (Vector Laboratories) and covered with 24 × 40 mm coverslips. The nuclei were then checked under an Olympus IX 83 inverted microscope: at 558/583 nm excitation/emission wavelengths for RFP and at 358/461 nm for DAPI. Images were captured with a HAMAMATSU ORCA-ER digital camera c4742-80 controlled by xCellence rt software (Olympus).

Pollen viability assays and size measurements

For pollen viability analysis, about 1 mL of opened flowers were collected into a 15-mL tube containing 3-mL BK buffer (0.127-mM Ca(NO₃)₂, 0.081-mM MgSO₄, 0.1-mM KNO₃, 15% (w/v) sucrose, and 10-mM MOPS, pH 7.5), and the tube was vortexed for 5 min to release pollen. Subsequently, the tube was centrifuged (2,600g, 5 min), the supernatant was carefully removed and the pollen pellet was carefully resuspended by gentle pipetting in 20-μL FDA buffer mixture (1-μL FDA [Sigma-Aldrich] stock solution [2 mg/mL in acetone] added to 1 mL of BK buffer). The suspension was carefully transferred to a microscope slide and covered with a 24 × 40 mm coverslip. Fluorescein fluorescence was observed after 20 min of staining using an Olympus IX 83 inverted microscope: at 543/620 nm excitation/emission wavelengths and the same region was photographed with differential interference contrast optics to get the number of all pollen grains.

To estimate the diameter of mature pollen, we used the same photos as for FDA analysis. Diameter measurements were done in Fiji/ImageJ on images calibrated using internal standards (Schindelin et al., 2012).

Hoyer’s clearing

Clearing of ovules was performed as described (Liu and Meinke, 1998).

Chromosome spreads and analysis of meiosis

Fixations of flower buds, chromosome spreads, and fluorescence in situ hybridization were carried out as described (Sánchez Moran et al., 2001) with minor modifications included in (Martinez-Garcia and Pradillo, 2017). Data for cytological analyses were collected from at least four plants per genotype. The DNA probes used for the analysis were: 45S ribosomal DNA (rDNA) (pTa71 of Triticum aestivum; Gerlach and Bedbrook, 1979) and 5S rDNA (pCT4.2; Campell et al., 1992).

Immunolocalization procedures were performed with a spreading technique previously described in Armstrong et al. (2002) in the case of RAD51 detection, and a squash technique as described in Manzanero et al. (2000) for α-tubulin and CENH3 detection. The primary antibodies used were: anti- α-tubulin (Merck; mouse, 1:50), anti-CENH3 (kindly provided by Dr Andreas Houben; rabbit, 1:500), anti-ZYP1 (kindly provided by Dr Chris Franklin; rat, 1:500), and anti-RAD51 (provided by Dr Chris Franklin; rabbit, 1:500). Secondary antibodies were: FITC-conjugated anti-mouse (Agrisera; 1:100), anti-rabbit Alexa Fluor 555-conjugated (Invitrogen, Molecular Probes; 1:500), anti-rat Alexa Fluor 555-conjugated (Invitrogen, Molecular Probes; 1:500), and anti-rabbit FITC-conjugated (Sigma-Aldrich; 1:50). Preparations were analyzed with an Olympus BX61 epifluorescence microscope and images were captured with an Olympus DP71 digital camera controlled by DP Controller software version 2.2.1.227 (Olympus).

Bioinformatic tools

Microsoft Office Excel 2016, PowerPoint 2016, GraphPad Prism 8.2.1, ImageJ 1.52p, BioRender, Flowing Software 2.5.1, Adobe Photoshop, and Adobe Illustrator were used for graph and image composition.

Accession numbers

Genes described in this article can be found in the TAIR database under the following accession numbers: NSE2 (At3g15150); NSE4A (At1g51130); NSE4B (At3g20760); SNI1 (At4g18470); SMC6B (At5g61460); SPO11-1 (At3g13170); OSD1 (At3g57860); QRT1 (At5g55590); HTR10 (At1g19890).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1 . nse2 affects plant growth.

Supplemental Figure S2 . nse2 affects pollen development.

Supplemental Figure S3 . Meiosis progression in WT and nse2-2.

Supplemental Figure S4 . Characterization of unreduced meiocytes by FISH and CENH3 inmunolocalization.

Supplemental Figure S5 . Phenotypes of spo11-1-5 and nse2-2 spo11-1-5 plants.

Supplemental Figure S6 . Quantification of RAD51 foci on pachytene meiocytes.

Supplemental Figure S7 . nse4b-1 and smc6b-1 do not affect seed development.

Supplemental Table S1 . Seed phenotypes of self-pollination and reciprocal crossing between nse2 and WT plants.

Supplemental Table S2 . Source data for analysis of seed development after self-pollination and reciprocal crossing between nse2 and WT plants.

Supplemental Table S3 . Source data for analysis of pollen viability by FDA.

Supplemental Table S4 . Quantification of meiotic products.

Supplemental Table S5 . Quantification of meiotic defects in PMCs of nse2-2, spo11-1-5, and nse2-2 spo11-1-5.

Supplemental Table S6 . Source data for comparison of RAD51 foci between WT and nse2-2.

Supplemental Table S7 . Source data for comparison of partial synapsis between WT and nse2-2 cells.

Supplemental Table S8 . Source data for comparing meiotic products between nse2-2 and nse2-2 spo11-1-5.

Supplemental Table S9 . Source data for comparing meiotic products between spo11-1-5 and nse2-2 spo11-1-5.

Supplemental Table S10 . Primers used in this study.

 

A.P., M.P. and F.Y. conceived and designed the study. F.Y., N.F.J., M.T. and M.D. performed experiments. J.V. and P.C. calibrated and maintained flow cytometers and performed flow-sorting. A.P., F.Y., M.P. and N.F.J. analyzed data and interpreted the results. A.P., F.Y. and M.P. wrote the paper. All authors read and approved the final manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell) is: Ales Pecinka (pecinka@ueb.cas.cz).