FIGL1 prevents aberrant chromosome associations and fragmentation and limits crossovers in polyploid wheat meiosis

We analysed ﬁgl1 mutants and virus-induced gene silencing lines cytogenetically. Genetic mapping was performed in the hexaploid.


Introduction
Meiosis is a specialised form of cell division in which DNA replication is followed by two sequential rounds of cell division to halve the chromosome number during gamete formation.During meiotic prophase I, the programmed formation of numerous DNA double-strand breaks (DSBs) catalysed by the SPO11 complex (Keeney et al., 1997;Vrielynck et al., 2016), followed by repair of the breaks by homologous recombination (HR), results in either noncrossovers (NCOs) or crossovers (COs; Börner et al., 2004).COs involve reciprocal genetic exchanges between the maternal and paternal chromosomes, which generate genetic variation and form physical links between homologous chromosomes that are essential for accurate segregation at the first meiotic division.In plants, COs can form via either of two pathways (Osman et al., 2011;Mercier et al., 2015;Wang & Copenhaver, 2018;Desjardins et al., 2020).The class I pathway accounts for c. 85% of all COs in plants and involves the activities of a group of recombination proteins known as ZMMs: Zip2/ SHOC1, Zip3/HEI10, ZIP4, MER3/RCK, MSH4, MSH5 and PTD (Higgins et al., 2004(Higgins et al., , 2005(Higgins et al., , 2008b;;Mercier et al., 2005;Wijeratne et al., 2006;Chelysheva et al., 2007Chelysheva et al., , 2012;;Macaisne et al., 2008Macaisne et al., , 2011)).Class I COs are interference-sensitive leading them to be well-spaced along chromosomes (Jones & Franklin, 2006;Martini et al., 2006).The remaining COs (Class II) are insensitive to interference and are partly dependent on the activity of the MUS81 endonuclease (Berchowitz et al., 2007;Higgins et al., 2008a,b).
HR is accompanied by remodelling of the meiotic chromosomes and the two processes are tightly integrated (Zickler & Kleckner, 2023).Following the premeiotic S-phase, sister chromatids are held together by cohesin proteins (Haering & Jessberger, 2012) and during leptotene become organised into linear looped arrays whose bases are anchored by the elaboration of a proteinaceous chromosome axis (Zickler & Kleckner, 2023).During zygotene, aligned homologous chromosomes are brought into close apposition by the formation of the synaptonemal complex (SC), which has a tri-partite structure, comprising the chromosome axes (now referred to as lateral elements) and crosslinking, transverse filament proteins (Page & Hawley, 2004).At pachytene, the SC is fully polymerised along the chromosomes and during diplotene, when CO formation is complete, the SC disassembles and chromosomes begin to condense.Condensation continues through diakinesis when homologue pairs become apparent as bivalent structures linked by one or more chiasmata, which are the cytological manifestation of COs.By metaphase I, chromosomes are fully condensed and attach to the meiotic spindle in preparation for segregation of the homologues at the first meiotic division.A second nonreductional division separates the sister chromatids producing the four haploid products of meiosis.
Meiotic recombination is tightly controlled at multiple levels, and in plants, the vast majority of initiating DSBs are not repaired as COs, typically resulting in only one to three COs per homologue pair, regardless of chromosome size (Mercier et al., 2015).For example, hexaploid wheat produces c. 2000 DSBs but only c. 42 COs per male meiosis (Gardiner et al., 2019;Osman et al., 2021;Higgins et al., 2022).The constraints on CO frequency impose severe limitations on plant breeding programmes and are exacerbated by localisation bias in some large genome crop species, particularly cereals where the majority of recombination events take place in proximity to the chromosome ends, making it very difficult to introgress useful agronomic traits into more interstitial and centromere proximal regions (Saintenac et al., 2009;Choulet et al., 2014).For example, cytological estimates indicate a 25 : 1 bias towards distal chiasmata in barley and a 9 : 1 bias in hexaploid wheat (Higgins et al., 2012;Osman et al., 2021).The discovery of several recombination-limiting mechanisms in Arabidopsis has therefore generated much interest, suggesting possible routes for modifying CO frequency and/or location in crop species.For example, class I COs are limited by HEI10 dosage (Ziolkowski et al., 2017); HCR1, which encodes the protein phosphatase PPX1, (Nageswaran et al., 2021); and the presence of the SC transverse filament protein, ZYP1a/ZYP1b, (Capilla-Pérez et al., 2021;France et al., 2021).Class II COs are limited by several independent anti-recombination pathways including the FANCM helicase (Crismani et al., 2012;Knoll et al., 2012); RECQ4 helicase (Séguéla-Arnaud et al., 2015); and FIGL1, a conserved AAA-ATPase, (Girard et al., 2015).Furthermore, it appears that simultaneous disruption of multiple recombination-limiting mechanisms in Arabidopsis can produce synergistic effects leading to substantial increases in recombination frequency (Girard et al., 2015;Serra et al., 2018;Fernandes et al., 2018a;Durand et al., 2022).
To date, attempts to reproduce these pro-CO strategies in crop species have had varying success.Mutation of RECQ4 has increased CO frequency in all species tested, including rice, pea, tomato (all in hybrid contexts) and barley, suggesting that this may be a universal tool for increasing recombination frequency in plants (Mieulet et al., 2018;Arrieta et al., 2021).FANCM disruption increased recombination in hybrid rice, hybrid pea and in a pure line of Brassica rapa (Mieulet et al., 2018;Blary & Jenczewski, 2019), while in wheat, FANCM was revealed to play a dual role in promoting class I COs and suppressing class II COs, resulting in an overall 31% increase in CO frequency in fancm hexaploid mutants, suggesting it could be an effective tool for accelerating breeding programmes (Desjardins et al., 2022).In contrast, disruption of FIGL1 (FIGNL1 in rice) resulted in sterility in rice, pea and tomato (Zhang et al., 2017;Mieulet et al., 2018) but maize figl1 mutants remained fertile, similar to the situation in Arabidopsis (T.Zhang et al., 2023).Rice FIGNL1 limits class II COs and inhibits non-homologous chromosome associations, such that fignl1 mutants display abnormal chromosome bridges and fragmentation at the division stages of meiosis, leading to the formation of inviable gametes and accounting for the infertility of these lines (Yang et al., 2022).OsFIGNL1 and AtFIGL1 together with their interacting partners, MEICA1 and FLIP respectively, act at the strand invasion stage of meiosis (Hu et al., 2017;Fernandes et al., 2018b;Yang et al., 2022).Maize FIGL1 also appears to act at this stage but whereas AtFIGL1 negatively regulates recombination, acting antagonistically with BRCA2 to modulate RAD51/DMC1 dynamics (Girard et al., 2015;Kumar et al., 2019), ZmFIGL1 and ZmBRCA2 act co-ordinately to promote CO formation (T.Zhang et al., 2023).
To the best of our knowledge, FIGL1 has not previously been studied in a polyploid crop species.Here, we investigate the effects of disrupting FIGL1 function in wheat using tetraploid (Triticum turgidum) and hexaploid (Triticum aestivum) mutant lines and virus-induced gene silencing (VIGS).We show that wheat FIGL1 limits COs and prevents aberrant chromosome associations and fragmentation.However, whereas mutation of FIGL1 in the tetraploid results in almost complete sterility, wheat hexaploid figl1 plants retain a moderate degree of fertility, suggesting its higher ploidy level may confer resilience to chromosomal damage.Furthermore, genetic mapping in the triple mutant reveals perturbation of the landscape of recombination, which may be useful for plant breeding.

Plant material
We selected mutants from the original annotation of the EMS-induced mutant lines to the Chinese Spring Survey Sequence (CSS; International Wheat Genome Sequencing Consortium, IWGSC, 2014), which was the only genome assembly available at the time (December 2015).As described above, we used the Arabidopsis and rice FIGL1/FIGNL1 sequence to identify the wheat orthologues in the CSS annotation (Traes_5AS_ 68FB50ED1, Traes_5BS_01D245E9C and Traes_5DS_64BDF B2CF) and select mutants for each homoeologue in tetraploid T. turgidum cv 'Kronos' and hexaploid T. aestivum cv 'Cadenza' mutant populations (Krasileva et al., 2017; Supporting Information Table S1).For tetraploid wheat, we selected premature termination codon mutants in TtFIGL1-A1: Kronos2408 (hereafter referred to as K2408) mutated at AA residue 253 and K2273 (at residue 410).Similarly, a premature termination codon mutant and a splice acceptor site variant were selected in TtFIGL1-B1: K2398 (residue 326) and K2644 (exon 7), respectively.The single mutants were crossed together to generate two independent double mutants: Ttfigl1-1 (K2273 × K2644) and Ttfigl1-2 (K2408 × K2398).Ttfigl1-1 was further crossed with Ttmsh5 (K863; Desjardins et al., 2022) to generate a quadruple mutant (Ttfigl1-1 Ttmsh5).For hexaploid wheat, we selected the following premature termination codon mutants: TaFIGL1-A1: Cadenza0654 (hereafter referred to as C0654) mutated at residue 429; TaFIGL1-B1: C0900 (residue 326) and TaFIGL1-D1: C2036 (residue 326).The single mutants were crossed to generate a Tafigl1 triple mutant (C0654 × C0900 × C2036).We confirmed all mutations by genotyping with homoeologue-specific KASP assays using published protocols (Adamski et al., 2021; Table S2).Upon the subsequent publication of the IWGSC RefSeqv1.1 annotations (IWGSC, 2018), the EMS mutants were reannotated using the TraesCS5A02G109500, TraesCS5B02 G110300 and TraesCS5D02G123800 gene models.The A and D genome mutants selected were re-identified in their expected position, but the B genome mutants were not called by the automated pipeline.However, the premature termination codon and splicing acceptor variants were confirmed by Sanger sequencing in lines K2398 and K2644, respectively, confirming that they are bona-fide mutants in TtFIGL1-B1.In all experiments, the original cultivars Kronos and Cadenza were used as wild-type (WT) controls for the tetraploid and hexaploid EMS-induced TILLING mutants, respectively (Krasileva et al., 2017).All lines were obtained from the Germplasm Resources Unit (https://www.seedstor.ac.uk/shopping-cart-tilling.php).Plants were grown in a controlled environment with photoperiod 16 h; temperature 20°C and relative humidity 60%.VIGS experiments were performed in a Level 3 biological containment facility at Rothamsted Research.

Cytological procedures
Meiotic chromosome spreads were prepared from anthers fixed at the required stage of meiosis in cold ethanol : acetic acid in the ratio 3 : 1 by volume.For slide preparation, anthers were digested in 0.3% cellulase, 0.3% pectolyase, 10 mM citrate buffer, pH 4.5 at 37°C for 90 min then three anthers per slide were macerated in 70% acetic acid and incubated for 1 min on a 45°C hotplate before fixing with 130 μl cold 3 : 1 fix.Slides were either viewed directly by applying 5 μg ml À1 4 0 ,6 0 -diamidino-2-phenylindole (DAPI) in Vectashield (Vector Labs) to stain chromosomes or subjected to further processing for immunolocalisation.For immunolocalisation, prepared slides were air-dried and then heated in 10 mM citrate buffer pH 6 in an 850 W microwave for 45 s, taking care not to let the buffer boil.Slides were then immediately transferred to PBST (1× phosphate-buffered saline, 0.1% triton X100) for 10 min.Slides were blocked for 30 min at room temperature by applying 100 μl of 3% BSA (in PBST) to the surface of each slide using parafilm.Immunostaining was carried out using standard procedures described in Armstrong et al. (2002).Primary antibodies were used at the following dilutions in the blocking buffer: anti-AtASY1 guinea-pig or rabbit, 1 : 500 (Armstrong et al., 2002); anti-AtZYP1 guinea-pig or rabbit, 1 : 500 (Higgins et al., 2005;Osman et al., 2018); anti-HvHEI10 guinea-pig, 1 : 250 (Desjardins et al., 2020); and anti-TaCENH3, 1 : 200 (Osman et al., 2021).Secondary antibodies were FITC (green) or Alexa Fluor 594 (red) conjugates (Sigma; Thermo Fisher, Waltham, MA, USA).Material for root tip analysis was generated by germinating seeds on moist filter paper at 20°C in the dark for 3 d.Roots were harvested when they were 1-3 cm long, fixed and used to prepare slides (as described above for anthers, except cellulase and pectolyase concentrations in the digestion mix were reduced to 0.15%).The viability of pollen grains was determined using Alexander staining (Alexander, 1969).

Microscopy
Fluorescence microscopy was carried out using a Nikon Eclipse 90i microscope fitted with a Nikon DS-Qi1Mc camera.NIS ELE-MENTS software (Nikon) was used to capture images as Z-stacks with a 0.2 μM step.Mitotic chromosome, meiotic HEI10 foci and meiotic CENH3 signal counts were carried out using Z-stack images for accuracy.Stained pollen grains were viewed and imaged using a Motic BA310 digital microscope with an integrated 3MP camera and MOTIC IMAGES PLUS v.2.0 imaging software.
Skyfall × Cadenza hexaploid F3 marker-based recombination analysis F 3 mapping populations fixed for either termination mutations (SxC_241; 74 lines) or WT alleles (SxC_123; 90 lines and SxC_322; 95 lines) at FIGL1 homoeologues were developed to assess variation in marker-based recombination frequency.Tafigl1 triple mutant (C0654 × C0900 × C2036) was crossed to Skyfall.F 1 plants were self-pollinated and F 2 siblings fixed for either WT or mutant alleles at the FIGL1 homoeologues were selected and self-pollinated for F 3 seed.Lines of the F 3 populations were genotyped using the Axiom-35K Breeders array (Allen et al., 2017) with genotype calling performed using the AXIOM ANALYSIS SUITE (v.5.2.0.65) and prior model 'Axiom_WhtBrd-1.r4'and DQC cut-off of 0.8.Genotypes for probes with low call rates were manually confirmed before export.Physical positions and chromosome assignment were allocated to each marker by aligning to the IWGSC REFSEQv1.0assembly (IWGSC, 2018) as described in Shorinola et al. (2022).Markers were filtered and removed if the parents were monomorphic, showed skewed segregation (heterozygous < 25% or > 75%), exceeded (> 20%) missing data limit, or markers were dominant (Allele A/B < 5).The remaining markers were allocated as either homozygous Cadenza (A), homozygous Skyfall (B), or segregating (X) and sorted by chromosome and physical position for each population.Markers showing segregation in any of the three populations were aligned (6475 markers) to enable the selection of 10 heterozygous intervals for genetic mapping (839 markers) to assess recombination frequency.All markers within the selected intervals were manually rescored to minimise errors and subsequent effects on genetic mapping.Genetic mapping was performed on each region individually using MSTmap online (http://www.mstmap.org/) with a LOD threshold of 8.After manual re-scoring and an initial round of mapping, additional markers were removed leaving 560 useful markers for the final mapping.For three regions (2A, 2B and 7B), the markers would not link for a complete linkage group for one of the populations and these regions were therefore manually split and re-mapped.The cM distance for each linkage group was used to determine the percentage difference in figl1 (SxC_241) over FIGL1 (mean of SxC_123 & SxC_322).As cM distances may become inflated due to erroneous data points leading to mis-mapping, the data, once sorted into genetic map order, was manually scored for each line as either having no COs, one CO, two or more COs or as miscall/mis-mapped.Real COs were described as where there was a change in allele score for at least three markers in a row, except for markers towards the end of a linkage group, where a single marker change was allowed.Furthermore, to be as inclusive as possible, each line was granted a single erroneous data point.The number of lines within each category was totalled for each region and calculated against the total number for the percentage of lines for each category.The difference between figl1 and FIGL1 populations was calculated.Analysing the percentage of lines that show no COs is potentially a more accurate way of assessing recombination as it will not be influenced by errors in the data and mismapping.

Statistics
HEI10 foci and seed count differences were tested for significance using a Mann-Whitney U-test; chiasma count differences using a pairwise Wilcoxon rank sum test and differences in pollen viability using a chi-squared contingency test.Mapping population differences were tested using ANOVA.

Fertility is reduced in figl1 plants and differs between ploidies
Single homoeologue figl1 mutants identified from the sequenced tetraploid Kronos EMS-induced population (Krasileva et al., 2017) were crossed together to generate two independent double homozygote mutant lines, where A and B FIGL1 homoeologues were both disrupted (Ttfigl1-1 and Ttfigl1-2).Likewise, single homoeologue figl1 mutants from the sequenced hexaploid Cadenza EMS-induced population (Krasileva et al., 2017) were crossed to produce a triple homozygote mutant line, with A, B and D gene copies all disrupted (Tafigl1; Fig. 1b).All mutations generated premature termination codons, apart from one tetraploid B homoeologue splice-acceptor mutation, and all were located upstream (or just inside the upstream flank) of the C-terminal AAA-ATPase and VPS4 domain region so were predicted to be null mutations (Fig. 1a).
All three mutant lines showed normal vegetative growth and produced spikes of normal appearance (Fig. 1c).However, fertility was severely reduced in all cases, particularly in the tetraploid mutant.The WT tetraploid control, Kronos, had a mean seed count of 20 per spike (n = 14).By contrast, Ttfigl1-1 was completely sterile and failed to produce any seeds from 13 spikes (P < 0.0001), while Ttfigl1-2 produced a total of 16 seeds from 34 spikes (P < 0.0001) with only five spikes containing any seeds at all (maximum number per spike was 6; Fig. 1d).Seed count was also reduced in Tafigl1 (mean = 14 AE 2.3 per spike vs 43 AE 1.7 in WT Cadenza, n = 20 (the first 5 spikes from 4 plants), P < 0.0001) but varied between spikes, ranging from 0 to 31 seeds per spike compared with 29-53 in WT (Fig. 1d).Thus, although some spikes produced no mature seeds (but often contained tiny, aborted seeds), others, even within the same plant, contained up to 72% of the WT mean seed number.

FIGL1 disruption results in aberrant chromosome associations and chromosome fragmentation
To investigate the basis of reduced pollen viability and fertility in figl1 mutants, we carried out a cytological analysis of pollen mother cells (PMCs) of Ttfigl1-1, Ttfigl1-2 and Tafigl1 compared with WT.During meiotic prophase I, pairing and recombination of homologous chromosomes in the tetraploid WT leads to the appearance of 14 bivalent structures at metaphase I, with homologues held together by chiasmata, the cytological manifestation of COs (Fig. 2a).This is followed by separation of the homologues at anaphase I and segregation into two cohorts of 14 chromosomes (Fig. 2f).At metaphase I in Ttfigl1-1 and Ttfigl1-2 abnormal interbivalent associations, multivalent structures and, more rarely, chromosome fragments were apparent (Fig. 2b,c).
As chromosomes separated at the first meiotic division, chromosome bridges and more extensive fragmentation were observed Fig. 2 Disruption of wheat FIGL1 results in abnormal chromosome associations and fragmentation.Chromosome spreads of pollen mother cells (PMCs) from double tetraploid and triple hexaploid figl1 mutants and their wild-type (WT) controls at metaphase I (a-e), the first meiotic division (f-j) and following the second meiotic division at the tetrad stage (k-o).Arrows indicate a quadrivalent structure (red), example fragments (yellow), chromosome bridges at the first division (blue) and stray chromatin at the tetrad stage (green).(p-r) Virus-induced gene silencing (VIGS) knockdown of FIGL1 in the hexaploid Bobwhite showing the empty vector control at metaphase I (p), the knockdown line with abnormal chromosome associations at metaphase I (q) and with chromosome bridges and fragments at the first division (r).DNA is stained with 4 0 ,6 0 -diamidino-2-phenylindole (DAPI).Bars, 10 μm.
Because lines derived from the EMS-induced populations have a high background mutational load (Uauy et al., 2017), we independently silenced the TaFIGL1 gene in hexaploid wheat (Bobwhite) using BMSV VIGS to confirm that the cytological phenotype of the EMS-induced mutant lines was due to disruption of FIGL1.Plants infected with the BSMV:TaFIGL1-i construct exhibited abnormal chromosome associations at metaphase I, and 66% (n = 32) of PMCs at the first division contained chromosome bridges and/or fragments (Fig. 2q,r).By contrast, PMCs from control plants infected with the empty virus produced normal discrete bivalents at metaphase I (Fig. 2p).The similarity in cytological phenotype between the VIGS-treated line and the Ttfigl1-1, Ttfigl1-2 and Tafigl1 EMS-induced mutant lines thus confirms the wheat figl1 phenotype.
The unexpected finding that Tafigl1 retained a modest level of fertility, despite exhibiting similar cytological defects to Ttfigl1-1 (and with a higher proportion of affected first-division nuclei), prompted us to examine the next generation.Seeds from selfed Tafigl1 were viable (100% seed germination, n = 30), producing plants with the parental phenotype of abnormal interbivalent associations at metaphase I, chromosome bridges and fragments at anaphase I and abnormal tetrads (Fig. S4a).Similar to Tafigl1, they also went on to produce seeds at a reduced level (0-30 seeds per spike).We also checked for the possibility of aneuploidy amongst the progeny.Chromosome number was difficult to assess directly due to the extensive meiotic chromosome associations, so we used two indirect approaches.First, we counted chromosomes in Z-stack images of mitotic root tip nuclei at prometaphase.This revealed the normal number of 42 chromosomes per nucleus (n = 5; Fig. S4b).Second, we carried out dual localisation of ZYP1 and CENH3 in late pachytene PMCs when chromosomes are expected to be fully paired and synapsed (to be described later).In four of the five plants examined, this revealed the presence of 21 pairs of CENH3 signals associated with linear ZYP1 signal, confirming that nuclei contained the correct number of chromosomes which could pair and synapse apparently normally (Fig. S4c).The remaining plant appeared to contain one 'extra' chromosome, forming 21 pairs of CENH3 signals plus an additional single signal (Fig. S4d).Thus, as far as we could tell, most progeny possessed an identical phenotype to the parent, with only a minority showing evidence of aneuploidy.

FIGL1 limits class II crossover formation
Next, we investigated the effect of FIGL1 disruption on recombination.It was not possible to carry out meaningful chiasma counts in the figl1 EMS-induced mutants due to the extensive metaphase I chromosome associations.We therefore assessed CO  (Chelysheva et al., 2012;Wang et al., 2012;Hesse et al., 2019;Desjardins et al., 2020;Osman et al., 2021).First, we examined chromosome synapsis in Tafigl1 by carrying out dual immunolocalisation of ASY1, which marks the unsynapsed chromosome axes (Armstrong et al., 2002), and the SC transverse filament protein, ZYP1 (Higgins et al., 2005).By late prophase I, linear ZYP1 signal had largely replaced ASY1 signal along the paired chromosomes in WT PMCs, with just a few traces of ASY1 signal remaining (Fig. 3a).A similar pattern of ZYP1 staining was observed in Tafigl1 at late prophase I, confirming the ability of the mutant to pair and synapse, and indicating that wheat FIGL1 is dispensable for this process (Fig. 3d).We then used dual immunolocalisation of ZYP1 and HEI10 at late prophase I to estimate class I CO frequency.HEI10 foci were counted at two stages: late pachytene, when ZYP1 was fully linear throughout the nucleus and HEI10 formed large foci (Fig. 3b,e), and slightly later, in early diplotene, when ZYP1 remained staining the chromosomes which displayed a 'corkscrew-like' morphology (a distinctive feature of wheat hexaploid chromosomes at this stage; Fig. 3c,f).At both stages, the number of HEI10 foci in Tafigl1 and WT was not significantly different (mean foci per PMC at pachytene in Tafigl1 = 45.4AE 7.1 vs 44.9 in WT, n = 40, P = 0.79; mean foci per PMC at diplotene in Tafigl1 = 42.2AE 8.6 vs 40.8 AE 7.4 in WT, n = 30, P = 0.52; Fig. 3g).This indicates that mutation of FIGL1 has no obvious effect on the formation of class I COs.
We then made use of a tetraploid double homozygote msh5 mutant defective in class I CO formation (Desjardins et al., 2020) to genetically investigate class II CO formation in wheat figl1.The Ttmsh5 mutant was crossed with Ttfigl1-1 to generate the quadruple mutant Ttfigl1-1 Ttmsh5, to determine whether mutation of FIGL1 could restore COs in Ttmsh5.Mean chiasma frequency in the WT tetraploid, Kronos, was 26.4 AE 0.17 (Fig. 4a,d), whereas in Ttmsh5, it was 4.7 AE 0.15 (n = 194), compared with 9.9 AE 0.34 (n = 72) in Ttfigl1-1 Ttmsh5 (P < 0.001; Fig. 4b-d).We also observed a corresponding increase in rod and ring bivalents, and a decrease in univalent pairs in the quadruple mutant relative to Ttmsh5 (Fig. 4e).The rescued chiasmata were randomly distributed around the mean and chiasma frequency did not differ significantly from a Poisson-predicted distribution (χ 2 25 ð Þ = 9.9, n = 72, P > 0.05).This indicates that mutation of figl1 can partially rescue the zmm phenotype of msh5 in wheat and implies that the additional chiasmata were formed via the class II pathway.Taken together, the cytological data suggest that wheat FIGL1 has no discernible effect on the formation of class I COs, but limits class II COs.

Genetic mapping reveals changes in the recombination pattern of figl1 mutants
The moderate fertility of Tafigl1 meant that FIGL1 disruption had potential utility for breeding, so it was of interest to carry out recombination mapping in this line, especially as the tetraploid data indicated that disruption of FIGL1 increased class II COs.Molecular marker-based analysis and genetic mapping were performed on hexaploid F 3 populations fixed for either null mutations or WT alleles at FIGL1, as an alternative method of measuring effects on recombination frequency (Fig. 5a).In total, 13 intervals were analysed on 8 chromosomes.Distances based on genetic mapping (cM) were compared for each region between the populations, with eight regions showing increased recombination (> 10%), three regions showing no effect  1a, S3; Fig. 5b,c).There no overall statistical increase in cM distance across intervals (P = 0.17), despite there being an average 52% increase in distance across intervals.This was mostly due to intervals 1B_1, 2A_1 and 2A_2 exhibiting large (> 100%) expansion in the genetic map (Tables 1a, S3; Fig. 5b,c).
To confirm these effects on recombination, and nullify any potential errors from mis-mapping, we also analysed the data by manually scoring the genotype data for lines showing no COs (Table 1b; Fig. 5d,e).Using this method, we compared the percentage of lines from figl1/FIGL1 populations showing no recombination.We discovered that 7 of the 13 intervals had fewer lines showing no COs (> À5%), 4 regions showed no effect (AE5%) and 2 were increased (> 5%).Across all regions, there were 6% fewer NCO lines in the figl1 null mutant population although, again, this was not statistically significant (P = 0.33).These results suggest that figl1 null mutants do not perturb recombination universally but do impact intervals on some chromosomes.

Discussion
We investigated genetic disruption of the conserved AAA-ATPase FIGL1 as a potential strategy for modifying meiotic recombination in wheat, where the generation of new allelic combinations through plant breeding is laborious and timeconsuming, particularly in hexaploid (bread wheat) varieties.
Cytological analysis of wheat figl1 tetraploid double and hexaploid triple mutants, and FIGL1 VIGS knockdown lines, revealed an identical phenotype of abnormal interbivalent chromosome associations at metaphase I, followed by chromosome bridges and fragmentation as homologues separated at anaphase I. Wheat figl1 mutants appeared to undergo normal and full synapsis implying that the interbivalent associations at metaphase I were between non-homologous chromosomes.One possibility Ó 2024 The Authors New Phytologist Ó 2024 New Phytologist Foundation New Phytologist (2024) www.newphytologist.com is that they were simply due to chromosome entanglements or interlocks which failed to be corrected, thus giving rise to mosome bridges at the division stages.Alternatively, interbivalent associations may have been caused by aberrant recombination events, either between homologous sites on homoeologous chromosomes or between ectopic sites.Left unresolved, these would also lead to anaphase I bridges (as indeed would unrepaired/unresolved recombination intermediates between homologous chromosomes, although these would not be apparent at metaphase I).Yet, a third possibility is that chromosome associations arose from defects left over from meiotic S-phase, and in this context, it is worth noting that there have been recent reports of a role for mammalian FIGNL1 in DNA replication (Ito et al., 2023;Q. Zhang et al., 2023), although to date, there is no evidence for such a role in plants.Similar abnormal chromosome associations occur in rice fignl1 mutants and have been reported to be nonhomologous (Zhang et al., 2017;Yang et al., 2022).In Arabidopsis and rice, FIGL1/FIGNL1 acts downstream of DSBs at the strand invasion step of recombination, such that the chromosome associations observed in rice fignl1 are P31 comet -dependent (Girard et al., 2015;Ji et al., 2016;Yang et al., 2022).It therefore seems likely that wheat figl1 chromosome associations are similarly recombination-dependent and represent a form of interaction which has failed to be repaired or resolved before the division stages.However, further study would be required to confirm this and to determine the type of interaction(s) involved.
Given the severe cytological phenotype of rice fignl1 PMCs, it is perhaps unsurprising that pollen was inviable and mutant plants were sterile (Zhang et al., 2017).FIGL1 disruption in other distantly related diploid crops (pea and tomato) also led to sterility, although the cytological basis of this was not reported (Mieulet et al., 2018).Similarly, wheat tetraploid double figl1 mutants were almost completely sterile and produced very few seeds.By contrast, the hexaploid triple mutant retained a The percentage difference was calculated for each analysis, and each region was categorised by its percentage difference as resulting in either, increasing, neutral or decreased recombination.For analysis of cM distance, the following categories were designated: Increasing (> 10%); Neutral (AE10%); Decreasing (> À10%), and for the No-CO analyses: Increasing (> 5%); Neutral (AE5%); Decreasing (> À5%).Note that a decrease in No-COs corresponds to an increase in recombination.ns, nonsignificance.
New Phytologist (2024) www.newphytologist.comÓ 2024 The Authors New Phytologist Ó 2024 New Phytologist Foundation reasonable level of fertility, with individual spikes producing up to 72% of the WT number of seeds, despite having a similar cytological phenotype to the tetraploid wheat double and rice mutants.One possible explanation for this is that the hexaploid mutant is more tolerant of chromosome damage than the tetraploid and diploids, possessing three sub-genomes and therefore a greater chance of segregating a viable complement of chromosomes to at least some of its haploid microspores.Consistent with this interpretation, 36% of pollen in the tetraploid was inviable compared with only 21% in the hexaploid, despite the hexaploid having a higher proportion of cytologically aberrant first-division nuclei (95% compared to 86%).Presumably, any such hexaploid resilience might also be advantageous during the subsequent stages of fertilisation and seed development, which may explain the difference in fertility between the two ploidies.It is also possible that genetic background differences over and above chromosome copy number per se may have contributed to Tafigl1 having a higher fertility than Ttfigl1-1 or Ttfigl1-2.
Despite the overall moderate level of fertility in the hexaploid, there was wide variation in the number of seeds per spike (0-31).However, even in the WT there was considerable variation (29-53 seeds, representing a range of 24), possibly because the process of pollination, fertilisation etc. is somewhat stochastic in nature.The proportion of viable pollen in the hexaploid was relatively high (c.80%).Nevertheless, some of the grains may have contained defects affecting their ability to support normal fertilisation or seed development, either resulting from FIGL1 disruption or, indeed, from other EMS-induced background mutations and this, together with the (currently unexplored) effects of FIGL1 disruption in female meiosis, may have contributed to the increased variability of individual spike seed counts in the mutant compared with the WT.This is consistent with the observation of tiny, aborted seeds in some of the Tafigl1 spikes.Seeds that did make it through to maturity were viable and produced plants of normal appearance with a similar figl1-like cytological phenotype and fertility to the parental line and, although we did observe one example of aneuploidy (one extra chromosome), this phenomenon did not appear to be widespread and is unlikely to preclude the use of FIGL1 disruption for breeding purposes.
To determine whether FIGL1 disruption might be useful for plant breeding, we investigated its effects on recombination.We found no evidence that figl1 mutation had any effect on class I CO frequency, such that HEI10 foci numbers in the triple mutant, Tafigl1, in late prophase I did not differ significantly from WT.As it was not feasible to investigate class II COs directly in the hexaploid using a figl1 zmm sextuple mutant, we assessed them in the tetraploid by creating a figl1-1 msh5 quadruple mutant.This revealed a 2.1-fold increase in the frequency of chiasmata per meiosis, from 4.7 to 9.9, in figl1-1 msh5 compared with a msh5 (class I) double mutant.Rescued chiasmata were randomly distributed around the mean and their frequency did not differ significantly from a Poisson-predicted distribution, as would be expected for class II COs (Higgins et al., 2004(Higgins et al., , 2008a, b;, b;Desjardins et al., 2020).These results are consistent with data from Arabidopsis and rice, where FIGL1 also limits class II COs but does not affect class I COs (Girard et al., 2015;Yang et al., 2022).
Genetic mapping in the hexaploid was performed using Skyfall × Cadenza mapping populations and involved 13 marker intervals across eight different chromosomes.There was no overall statistical increase in recombination based on cM distance, despite a mean interval increase of 52%.However, three of the intervals, 1B_1, 2A_1 and 2A_2, were associated with very large increases in map distance (100-300%), indicating local changes in the recombination landscape.Furthermore, genotypic analysis to identify lines with no-COs largely corroborated the mapping data, indicating that seven intervals showed an increase in recombination (i.e. had fewer no-COs) including 1B_1, 2A_1 and plant species.The presence of unresolved chromosome in wheat and rice and their absence from Arabidopsis and maize figl1/fignl1 mutants implies simple relationship with either ploidy or size.In fact, genetic analysis in Arabidopsis and maize has revealed a complex interplay between FIGL1 and BRCA2, whereby FIGL1 acts as a negative regulator in Arabidopsis HR, antagonising BRCA2 to control RAD51 and DMC1 focus formation (Kumar et al., 2019), and as a positive regulator of recombination in maize, acting co-ordinately with dosage-sensitive BRCA2 (T.Zhang et al., 2023).Notably, the maize figl1 brca2 double mutant has an identical cytological phenotype to wheat and rice figl1/fignl1 mutants, displaying abnormal chromosome associations at metaphase I and bridges/fragmentation at anaphase I (T.Zhang et al., 2023).It would therefore be interesting to investigate the relationship between FIGL1 and other recombination proteins in future wheat studies.
To the best of our knowledge, this is the first study of FIGL1 in a polyploid plant and increases our understanding of recombination in crop species.The discovery that similar cytological abnormalities could impact fertility differently in the tetraploid double and hexaploid triple mutants was unexpected and suggests that the extra ploidy level may afford a degree of resilience to chromatin damage.This could have important implications for other recombinationmodification strategies.The modest fertility of the hexaploid triple mutant and the perturbation of recombination observed in some mapping intervals, despite examining only a limited number of regions, suggests that FIGL1 disruption may be a useful addition to the plant breeding toolkit.The large increases in recombination frequency observed in some intervals could help to speed up the breeding of desired haplotypes, which typically involves screening large populations of plants over many generations, thus reducing time and cost.Perturbation of recombination might also help to reveal inaccessible, potentially useful, genetic variation and disrupt blocks of linkage disequilibrium, freeing agronomically beneficial traits from linked, undesirable ones.

Fig. 1
Fig. 1 Fertility is reduced in wheat figl1 mutants.(a) TtFIGL1 and TaFIGL1 coding region of homoeologues with conserved domains (dark blue boxes) and position/effect of EMS-induced mutations (red bars) indicated.Asterisks indicate a change to a premature termination codon, SA to splice-acceptor site variant.(b) Generation of double (tetraploid) and triple (hexaploid) mutant lines from single figl1 mutants.(c) Comparison of spike morphology in wildtype (WT) and double or triple mutant lines (bars, 1 cm).(d) Scatter plot of seed counts per spike in Ttfigl1-1, Ttfigl1-2 and Tafigl1 compared with Kronos and Cadenza WTs respectively.Mean and SE bars are shown in red and asterisks indicate significance at the 0.0001% level (n = 20).(e) Alexander staining indicates reduced pollen viability in Ttfigl1-1 and Tafigl1 but is more extreme in the tetraploid than the hexaploid.Contents of viable grains stain red, inviable grains are shrunken and unstained (arrows); bars, 50 μm.Bar chart represents WT and mutant sample sizes of 280 for the tetraploid and 700 for the hexaploid.Asterisks indicate significance at the 0.0001% level.

Fig. 5
Fig. 5 Generation of a hexaploid wheat triple figl1 mutant and F 3 mapping populations for genetic marker-based recombination analysis.(a) Crossing scheme for generating the triple mutant Tafigl1 (background Cadenza) and the Skyfall × Cadenza F 3 mapping populations fixed at FIGL1 for either wildtype (AA/BB/DD) or mutant (aa/bb/dd ) alleles.(b, d) Chromosome depictions showing the locations of the 13 regions and effects as analysed by cM distance from genetic mapping (b) or lines with no crossovers (d).Regions are aligned and ordered by IWGSC RefSeqv1.0coordinates (Mbp).Beneficial effects from the figl1 triple mutant population are highlighted in red, neutral regions in grey and detrimental effects in blue.(c, e) Bar charts showing the % differences between figl1/FIGL1 populations ordered by chromosome region.Bars are coloured by designated effectnote that red corresponds to an increase in recombination in each case.Dashed lines represent an increase/decrease of 10% in (c) and 5% in (e).

Table 1
Variation in hexaploid wheat recombination for the 13 selected segregating regions calculated by either the cM mapping distance (a) or the proportion of lines showing no crossovers (b).