Exploiting the ZIP4 homologue within the wheat Ph1 locus has identified two lines exhibiting homoeologous crossover in wheat-wild relative hybrids

Despite possessing related ancestral genomes, hexaploid wheat behaves as a diploid during meiosis. The wheat Ph1 locus promotes accurate synapsis and crossover of homologous chromosomes. Interspecific hybrids between wheat and wild relatives are exploited by breeders to introgress important traits from wild relatives into wheat, although in hybrids between hexaploid wheat and wild relatives, which possess only homoeologues, crossovers do not take place during meiosis at metaphase I. However, in hybrids between Ph1 deletion mutants and wild relatives, crossovers do take place. A single Ph1 deletion (ph1b) mutant has been exploited for the last 40 years for this activity. We show here that chemically induced mutant lines, selected for a mutation in TaZIP4-B2 within the Ph1 locus, exhibit high levels of homoeologous crossovers when crossed with wild relatives. Tazip4-B2 mutant lines may be more stable over multiple generations, as multivalents causing accumulation of chromosome translocations are less frequent. Exploitation of such Tazip4-B2 mutants, rather than mutants with whole Ph1 locus deletions, may therefore improve introgression of wild relative chromosome segments into wheat. Electronic supplementary material The online version of this article (doi:10.1007/s11032-017-0700-2) contains supplementary material, which is available to authorized users.


Introduction
Hexaploid bread wheat (Triticum aestivum) is composed of three related ancestral genomes (A, B and D), each containing seven identical (homologous) chromosome pairs (homologue pairs 1-7). Each homologue pair has a corresponding related homoeologous chromosome pair, possessing similar gene order and content, within each of the other two genomes. Despite the similarity between homoeologues, wheat behaves as a diploid at meiosis, with synapsis and crossovers (COs) only occurring between homologous chromosomes, rather than between homoeologous chromosomes (for example, 1A only pairs with 1A but not with either 1B or 1D, holding true for all seven chromosome groups). This diploid-like behaviour is predominantly controlled by Ph1, a dominant locus on chromosome 5B (Riley and Chapman 1958;Sears and Okamoto 1958).
The Ph1 locus was defined as a deletion phenotype, first described by scoring meiosis in wheat hybrids lacking the whole 5B chromosome (Riley and Chapman 1958;Sears and Okamoto 1958). The effect of deleting the Ph1 locus in wheat itself is to reduce the level of ring bivalents (and increase rod bivalent levels) at metaphase I in all meiocytes, while inducing a low level of multivalents and univalents in some meiocytes (Roberts et al. 1999). Sexual hybridisation between wheat and wild relatives (for example, rye or Aegilops variabilis) produces interspecific hybrids, containing haploid sets of wheat and wild relative homoeologous chromosomes, but exhibiting virtually no COs during meiosis. However, in Ph1 deleted wheat-rye hybrids, an average of 7 COs per cell is observed (Sears 1977;Dhaliwal et al. 1977). These Ph1 deleted hybrids have been used to introgress important traits from wild relatives into wheat. Thus, the absence of the Ph1 locus is scored as a reduction in homologous crossovers at metaphase I in wheat meiocytes and an increase in homoeologous crossovers at metaphase I in wheat-wild relative meiocytes.
Ethylmethane sulphonate (EMS) treatment failed to produce mutants exhibiting the full Ph1 deletion phenotype because of the difficulty with the phenotype screen. This suggests that the Ph1 phenotype most likely results from activity of more than a single gene (Griffiths et al. 2006). However, exploitation of smaller chromosome 5B deletions later characterised and localised the Ph1 locus to a 2.5 MB region on chromosome 5B (Griffiths et al. 2006). More recently, a metaphase I expressing gene within the 2.5 MB region (corresponding to the wheat gene EST BE498862 and the rice gene Os9g30320), and termed C-Ph1 by the authors, has been proposed as the putative Ph1 gene (Bhullar et al. 2014). The assertion was based on the observation of clumping or multiple chromosome associations in wheat meiocytes at metaphase I, with increased plant sterility, using a VIG-based approach rather than deletion analysis. However, previous studies have already shown that the rice homologue Os9g30320 is actually a tapetal cell gene (Jeon et al. 1999). In addition, a wheat paralogue of the C-Ph1 gene (termed Raftin1) has also previously been characterised as a tapetal cell gene (Wang et al. 2003). The expression peak of tapetal cell genes occurs when the tapetum fully forms around the metaphase I meiocytes. Disruption of these genes results in stressed meiocytes, chromosome clumping and male sterility, hence the previous filing of patents for the exploitation of these genes to generate male sterile lines ( P a t e n t s W O 2 0 0 0 0 2 6 3 8 9 A 3 a n d US20040060084). Moreover, two previously described deletions within the 2.5 MB region, also encompassing C-Ph1, did not exhibit a Ph1 mutant phenotype (Al-Kaff et al. 2008). Therefore, either the phenotype of these previously described deletions has been incorrectly scored, or the phenotype of the VIG approach was due to off-target effects (such as disruption of homoeologous and paralogous tapetal cell gene function). The observed VIG phenotype was also noted to be more extreme than when the whole 5B chromosome is deleted (Bhullar et al. 2014).
Deletion analysis defines the Ph1 locus to the part of the 2.5 MB region containing a duplicated chromosome 3B segment carrying heterochromatin and TaZIP4-B2 (originally termed Hyp3, UniProtKB-Q2L3T5), inserted into a cluster of CDK2-like genes interspersed with methyl transferase genes (originally termed SpG, UniProtKB-Q2L3W3) (Griffiths et al. 2006;Al-Kaff et al. 2008;Martín et al. 2017). The contribution of these genes to the Ph1 phenotype is not known; therefore, the effect of these genes on synapsis and crossover required further analysis. Synapsis is a process early in meiosis by which homologues intimately align with each other, forming bivalents held together by a proteinaceous structure named synaptonemal complex (SC). Ultimately, the SC is degraded, so that the bivalents are only held together by chiasmata or COs at metaphase I, allowing their correct segregation. Recently, Ph1 has been shown to have a dual effect on synapsis and CO formation in wheat (Martín et al. 2014;Martín et al. 2017). The effect on synapsis occurs during the telomere bouquet stage, when Ph1 promotes more efficient homologous synapsis, thereby reducing the chance of homoeologous synapsis (Martín et al. 2017). The effect on CO formation happens later in meiosis, when Ph1 prevents MLH1 sites (Double Holliday Junctions marked to become COs) on synapsed homoeologues from becoming COs. However, the molecular mechanism of how the Ph1 locus affects both synapsis and CO is not currently known.
A single Ph1 deletion mutant, developed in hexaploid wheat cv. Chinese Spring (CS) (CS ph1b), has been used by breeding programmes worldwide to introgress wild relative chromosome segments into wheat. A recent programme has successfully introgressed hundreds of wild relative chromosome segments into wheat, exploiting CS ph1b (King et al. 2016). The segments were identified through a combination of both DNA array and cytological-based approaches. Although CS ph1b has been successfully used in breeding programmes, the CS ph1b mutant is reported to accumulate extensive rearrangements reducing fertility (Sánchez-Morán et al. 2001), due to homoeologous synapsis and COs, as visualised by the occurrence of multivalents at metaphase I during meiosis. It would therefore be useful to identify novel wheat Ph1 mutant lines, with reduced homoeologous synapsis and CO at meiosis, but which do exhibit homoeologous COs in hybrids with wild relatives. As described previously, the Ph1 locus, which affects both synapsis and CO, is a complex cluster of CDK2-like and methyl transferase genes containing a ZIP4 paralogue. It has been proposed that Ph1's effect on synapsis is connected to altered Histone H1 CDK2-dependent phosphorylation in the presence and absence of Ph1. Altered phosphorylation affects chromatin structure and delays premeiotic replication, subsequently affecting homologue synapsis, thus allowing homoeologous synapsis to take place (Greer et al. 2012;Martín et al. 2017). Lines carrying mutations in the Ph1CDK2-like homologue in Arabidopsis also exhibit reduced synapsis under specific conditions, suggesting a role for these genes in efficient synapsis (Zheng et al. 2014). We have also previously proposed that the effect of CDK2-like genes on chromatin structure not only affects synapsis, but might also affect the resolution of Double Holliday Junctions (marked by MLH1) as COs (Greer et al. 2012). Okadaic acid treatment affects chromatin structure and can induce homoeologous CO in wheat-wild relative hybrids (Knight et al. 2010). However, given that the locus contains multiple copies of the CDK2-like and methyltransferase genes, it would be a complex and laborious study to identify EMS mutants within these genes and combine possible mutations. The transfer of multiple mutated genes into different elite genetic backgrounds by breeders, for subsequent crossing with wild relatives, would also be laborious. Moreover, as previously indicated, we want to identify wheat mutant lines which exhibit reduced homoeologous synapsis or multivalents at metaphase I; so, the CDK2-like genes would not be the initial candidates for such an approach.
Although there are ZIP4 homologues on group 3 chromosomes, the ZIP4 paralogue (TaZIP4-B2) within the Ph1 locus on chromosome 5B is single copy, compared to the CDK2-like and methyl transferase gene cluster. Moreover, ZIP4 has been shown to have a major effect on homologous COs, but not on synapsis, in both Arabidopsis and rice (Chelysheva et al. 2007;Shen et al. 2012). Knockouts of this gene in diploids usually result in sterility, as elimination of homologous COs leads to metaphase I pairing failure and incorrect segregation. We therefore assessed whether the selection and scoring of Tazip4-B2 EMS mutants would identify wheat lines with reduced homologous CO and minimal homoeologous synapsis and CO (as observed by the occurrence of multivalents at metaphase I), but which exhibit homoeologous COs in hybrids with wild-relatives. We describe the identification of two such lines through this approach.
Both for the RNA-seq analysis and for meiotic studies, the seedlings were vernalised for 3 weeks at 8°C and then transferred to a controlled environmental room until meiosis under the following growth conditions: 16 h/8 h, light/dark photoperiod at 20°C day and 15°C night, with 70% humidity. Tillers were harvested after 6 to 7 weeks, when the flag leaf was starting to emerge, and anthers collected. For each dissected floret, one of the three synchronised anthers was squashed in 45% acetic acid in water and assigned to each meiotic stage by observation under a LEICA DM2000 microscope (LeicaMicrosystems, http://www.leica-microsystems.com/). The two remaining anthers were either frozen in liquid nitrogen and stored at −80°C for RNA-seq analysis or fixed in 100% ethanol/acetic acid 3:1 (v/v) for cytological analysis of meiocytes.

Sample preparation
Anthers from wild-type wheat (WT) and wheat lacking the Ph1 locus (ph1b deletion) were harvested as described in the BPlant material^section. Anthers at late leptotene-early zygotene stage were later harvested into RNA later (Ambion, Austin, TX). The anthers from three plants of each genotype were pooled in a 1.5-ml Eppendorf until reaching 200 to 400 anthers. Once sufficient anthers had been collected, the material was disrupted using a pestle, centrifuged to eliminate the RNA and then homogenised using QIAshredder spin columns (Qiagen, Hilden, Germany). RNA extraction was performed using a miRNeasy Micro Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. This protocol allows purification of a separate miRNA-enriched fraction (used for further analysis) and the total RNA fraction (˃200 nt) used in this study. This process was repeated to obtain three biological samples of each genotype.

RNA-seq library preparation and sequencing
One microgram of RNA was purified to extract mRNA with a poly-A pull down using biotin beads. A total of six libraries were constructed using the NEXTflex™ Rapid Directional RNA-Seq Kit (Bioo Scientific Corporation, Austin, Texas, USA) with the NEXTflex™ DNA Barcodes-48 (Bioo Scientific Corporation, Austin, Texas, USA) diluted to 6 μM. The library preparation involved an initial QC of the RNA using Qubit DNA (Life technologies, CA, Carlsbad) and RNA (Life technologies, CA, Carlsbad) assays as well as a quality check using the PerkinElmer GX with the RNA assay (PerkinElmer Life and Analytical Sciences, Inc., Waltham, MA, USA). The constructed stranded RNA libraries were normalised and equimolar pooled into one final pool of 5.5 nM using elution buffer (Qiagen, Hilden, Germany, Hilden, Germany). The library pool was diluted to 2 nM with NaOH, and 5 μl was transferred into 995 μl HT1 (Illumina) to give a final concentration of 10 pM. Diluted library pool of 120 μl was then transferred into a 200-μl strip tube, spiked with 1% PhiX Control v3 and placed on ice before loading onto the Illumina cBot. The flow cell was clustered using HiSeq PE Cluster Kit v4, utilising the Illumina PE_HiSeq_Cluster_Kit_V4_cBot_recipe_V9.0 method on the Illumina cBot. Following the clustering procedure, the flow cell was loaded onto the Illumina HiSeq2500 instrument following the manufacturer's instructions. The sequencing chemistry used was HiSeq SBS Kit v4 with HiSeq Control Software 2.2.58 and RTA 1.18.64. The library pool was run in a single lane for 125 cycles of each paired end read. Reads in bcl format were demultiplexed based on the 6 bp Illumina index by CASAVA 1.8, allowing for a one base-pair mismatch per library and converted to FASTQ format by bcl2fastq.

Cytological analysis and image processing
A total of five plants per line were examined in a randomised block design. Generally, 2-3 tillers in each plant were analysed to identify meiocytes at meiotic metaphase I. Anthers from Tazip4-B2 Cadenza mutant lines, wild-type Cadenza, Tazip4-B2 mutant line-Ae variabilis hybrids, and Cadenza-Ae. variabilis hybrids were harvested as described in the BPlant materialŝ ection. Cytological analysis of meiocytes was performed using Feulgen reagent as previously described (Sharma and Sharma 2014). Images were collected u s i n g a L E I C

Nucleotide analysis
The regions of each mutation in both mutant Cadenza lines were sequenced to confirm the existence of either missense or nonsense mutations in Cadenza1961 and Cadeza0348, respectively. Wheat leaf tissues from wild type Cadenza and mutant Cadenza lines were harvested at growth stages 3-4 (Feekes scale). DNA was extracted using the CTAB method (Murray and Thompson 1980). Two pairs of primers were designed using the Primer3plus software (Untergasser et al. 2007) based on t h e Z I P 4 s e q u e n c e c a l l e d TRIAE_CS42_5BL_TGACv1_404600_AA1305800 ( h t t p : / / p l a n t s . e n s e m b l . o r g / T r i t i c u m _ aestivum/Info/Index). The primers used were as follows: forward primer: 5′GCCGCCATGACGATCTCCGAG3′ and reverse primer: 5′GGACGCGAGGGACGCGAG3′ for Cadenza1691 and forward primer: 5′GTGTTCCT AATGCTCACAACTC3′ and reserve primer:

Results and discussion
TaZIP4-B2 expression In hexaploid wheat, ZIP4 homologues are located within the Ph1 locus on 5B, and also on chromosomes 3A, 3B and 3D. Before undertaking the targeted induced lesion in genomes (TILLING) mutant analysis, we assessed the expression of TaZIP4-B2 to confirm that the TaZIP4-B2 gene within the Ph1 locus is expressed during meiosis; that it has a higher level of expression than the ZIP4 homologues present on chromosome group 3; and finally, that Ph1 deletion significantly reduces overall ZIP4 expression. At the coding DNA sequence and amino acid levels, TaZIP4-B2 (AA1305800.1) showed 95.3 and 89.2% similarity to TaZIP4-B1 (AA0809860.1), 9 4 . 1 a n d 8 7 . 5 % s i m i l a r i t y t o Ta Z I P 4 -A 1 (AA0645950.1) and 94.4 and 87.2% similarity to Ta Z I P 4 -D 1 ( A A 0 8 8 4 1 0 0 . 1 ) , r e s p e c t i v e l y ( Supplementary Fig. 1). To compare the relative expression of these ZIP4 homologues on chromosomes 3A, 3B, 3D and 5B, RNA samples were collected from anthers of hexaploid wheat (Triticum aestivum cv. Chinese Spring) (WT) and the Ph1 deletion mutant (ph1b) at the late leptotene-early zygotene stage, and six libraries prepared for the RNA-seq study. RNA-seq analysis showed that TaZIP4-B2 exhibited a higher level of expression than the ZIP4 homologues on homoeologous group 3 chromosomes ( Fig. 1; Supplementary Fig. 1). Moreover, TaZIP4-B2 also showed three splice variants ( Supplementary Fig. 2) in contrast to homoeologous group 3 chromosome ZIP4 homologues. One of these splice variants (splice variant 1) accounted for 97% of the TaZIP4-B2 transcripts. As expected, when the Ph1 locus was deleted, expression of TaZIP4-B2 was also eliminated (p < 0.05), but there was no apparent increase in the t r a n s c r i p t i o n o f t h e Z I P 4 h om o l o g u e s o n homoeologous group 3 chromosomes, to compensate for the absence of ZIP4 on chromosome 5B (p > 0.05) ( Fig. 1; Supplementary Table 1). Thus, RNA-seq data revealed that the expression of ZIP4 was derived mainly from the gene (TaZIP4-B2) on chromosome 5B, within the Ph1 locus.
TaZIP4-B2 suppresses homoeologous COs in wheat-Ae. variabilis hybrids The protein-coding sequences of 1200 EMS mutant lines (Rakszegi et al. 2010) from hexaploid wheat cv. Cadenza have been recently sequenced using exomecapture and displayed to allow the identification of millions of mutations in the sequenced genes using the www.wheat-tilling.com database (Krasileva et al. 2017). The mutations identified are accessible using the wheat survey sequence (Marcussen et al. 2014) via this database, which includes their location within the gene, and the predicted effect that each variant has on its protein.
Simply searching the database (www.wheat-tilling.com) reveals those plants possessing mutations in the target genes, as well as a list of all mutations possessed by the plant (Krasileva et al. 2017). We selected seven of the 1200 EMS mutant lines, which possessed potentially interesting mutations within TaZIP4-B2 (Traes_5 BL_9663AB85C.1) (Supplementary Table 2). Five of these mutant lines exhibited regular pairing at meiotic metaphase I, so were not taken further. These lines possessed a missense mutation, which indicates that amino acid changes within ZIP4 in these Tazip4-B2 lines did not affect its function. However, two of the mutant lines (Cadenza1691 and Cadenza0348) showed reduced number of COs in cytological analysis, suggesting that their Tazip4-B2 mutations did exhibit a phenotype. Both lines were selected for wide crossing studies with wild relatives to score the effect of their Tazip4-B2 mutations on homoeologous CO frequency in the wheat Tazip4-B2 mutant-wild relative hybrids, as compared to non-mutagenised wheat-wild relative hybrids. Null segregants were not available, so wild-type Cadenza lines were used as controls. Mutations within TaZIP4-B2 were verified by sequencing, and primers were designed to the mutated regions to follow mutated genes during crossing ( Supplementary Fig. 3 and Materials and Methods). Tazip4-B2, within the Cadenza1691 mutant line, possessed a missense mutation within the Spo22 domain (C to T change leading to an A167V), shown to be important for ZIP4 function (Perry et al.2005).
In addition to these Tazip4-B2 mutations, the two mutant lines also possessed mutations (mostly missense, but also splice or stop codons) within the coding sequences of 106 other shared genes. Sixteen of these genes, including TaZIP4-B2, were located on chromosome 5B. However, none of the genes apart from TaZIP4-B2 were located within the 2.5 MB Ph1 region defined in our previous study (Griffiths et al. 2006;Al-Kaff et al. 2008). Compared to the chromosome 5B deletion mutantwild relative hybrid-no other wheat chromosome deletion mutants have previously been reported as exhibiting a similar level of homoeologous CO formation at metaphase I (Riley and Chapman 1958;Sears 1977). For example, the 3D locus Ph2 exhibits a fourfold lower level of induction compared to Ph1 (Prieto et al. 2005). Equally, deletion of regions of chromosome 5B apart from the 2.5 MB Ph1 region does not result in homoeologous CO formation at metaphase I when the lines are crossed with wild relatives (Roberts et al. 1999;Griffiths et al. 2006;Al-Kaff et al. 2008). Sears (1977) used such crosses between hexaploid wheat cv. Chinese Spring, both in the presence and absence (ph1b deletion) of Ph1, and the wild relative tetraploid Aegilops kotschyi (also termed Ae. variabilis), to show that homoeologous COs are induced when the Ph1 locus is deleted. Interspecific hybrids of the ph1b mutant and wild relatives have been subsequently used in plant breeding programmes for introgression purposes (Sears 1977). Sears (1977) observed one rod bivalent at metaphase I in the presence of Ph1, and 6.35-7.28 rod bivalents in the Ph1 absent hybrids. Chiasma frequency in the hexaploid wheat-Ae. kotschyi or hexaploid wheat-Ae. variabilis hybrids was between 1 and 3 in the presence of Ph1 and 11-14 in the absence of Ph1 (Sears 1977;Farooq et al. 1990;Fernández-Calvín and Orellana 1991;Kousaka and Endo 2012).
In this study, both Tazip4-B2 mutant Cadenza lines, as well as a wild-type Cadenza (TaZIP4-B2), were crossed with Ae. variabilis. The frequency of univalents, bivalents, multivalents and total chiasma frequency was scored at meiotic metaphase I in the resulting F 1 hybrid (Fig. 2). In these hybrids, there were similar numbers of rod bivalents to that reported by Sears (1977), with 6.75 (SE 0.17) (Cadenza1691) and 6.64 (SE 0.18) (Cadenza0348) rod bivalents at metaphase I in the Tazip4-B2 mutants, and 1.48 rod bivalents (SE 0.12) at metaphase I in the wild-type Cadenza. Moreover, the chiasma mean frequency was 1.48 (SE 0.12) in the presence of Ph1 (TaZIP4-B2) and 12.21 (SE 0.19) and 12.23 (SE 0.20) in the Cadenza1691-Ae. Variabilis and Cadenza0348-Ae. variabilis hybrids, respectively. The Fig. 1 Relative expression of ZIP4 homologues in Triticum aestivum cv. Chinese Spring in presence (WT) and in absence (ph1b deletion) of the Ph1 locus obtained by RNA-seq analysis Fig. 2 Homoeologous CO frequency at metaphase I is increased in Tazip4-B2 mutant line-Ae. variabilis hybrids (b, c), in comparison to the wild-type Cadenza-Ae. variabilis hybrid (a). The number of univalents, bivalents, multivalents and chiasma frequency were scored at meiotic metaphase I in Cadenza0000 (TaZIP4-B2) × Ae. variabilis hybrids, and in Cadenza1691 (Tazip4-B2) × Ae. variabilis hybrids and Cadenza0348 (Tazip4-B2) × Ae. variabilis hybrids. The same letter indicates no differences between TaZIP4-B2 (a) and Tazip4-B2 hybrids (b, c) in metaphase I at P < 0.05. Scale bar represents 10 μm for all panels observed chiasma frequencies at metaphase I, in the two Tazip4-B2 mutant line-Ae. variabilis hybrids, are similar to those previously reported at metaphase I in the Ph1 deletion mutant (ph1b)-Ae. variabilis hybrids. Thus, the data indicate that TaZIP4-B2 within the Ph1 locus is likely to be involved in the suppression of homoeologous COs.

Tazip4-B2 mutant Cadenza lines show no multivalents
The frequencies of meiotic associations at metaphase I in hexaploid wheat and the Ph1 deletion mutant (ph1b) have been reported previously (Martín et al. 2014). Martín et al. (2014) observed 20 ring bivalents and one rod bivalent, with a chiasma frequency of 40.97 in the presence of the Ph1 locus. However, the number of ring bivalents decreased to 14.83, with a reduced chiasma frequency of 35.78, while the number of univalents, rod bivalents, trivalents and tetravalents increased to 0.80, 4.73, 0.20 and 0.37, respectively, when the Ph1 locus was absent. The number of univalents, bivalents, multivalents and chiasma frequency at meiotic metaphase I was also scored in both the Tazip4-B2 mutant lines and in the wild-type Cadenza (Fig. 3). The Tazip4-B2 mutant lines exhibited a reduction in the number of ring bivalents at metaphase I, and a slight increase in the number of rod bivalents, from a mean of 1.30 (SE 0.17) in the wild-type Cadenza, to 3.29 (SE 0.16) in Cadenza1691 and 3.63 (SE 0.18) in Cadenza0348. This indicates a slight reduction in homologous COs in these Tazip4-B2 mutant lines. CO frequency was a mean of 40.50 (SE 0.21) in the wildtype Cadenza,38.13 (SE 0.20) in Cadenza1691 and 37.30 (SE 0.23) in Cadenza0348. These observed chiasma frequencies at metaphase I in the two mutant Cadenza lines are again similar to those previously reported at metaphase I in wheat in the absence of the Ph1 locus. However, no multivalents were observed, and there was no significant increase in the number of univalents at metaphase I in the Tazip4-B2 mutant lines, as is normally observed in Ph1 deletion mutants (Roberts et al. 1999).
If Tazip4-B2 mutants had enabled homoeologues to synapse while failing to CO, then a significant increase in univalents would be expected, but this was not observed. This suggests that homoeologous synapsis may not be significantly affected by TaZIP4-B2. On the other hand, the lack of The number of univalents, bivalents and chiasma frequency were scored at meiotic metaphase I in Cadenza0000 (TaZIP4-B2) and in Cadenza1691 (Tazip4-B2) and Cadenza0348 (Tazip4-B2).
Asterisks indicate the presence of rod bivalents in both mutant Cadenza lines. The same letter indicates no differences between TaZIP4-B2 (a) and c) in metaphase I at P < 0.05. Scale bar represents 10 μm for all panels multivalents at metaphase I suggests that both mutant lines will exhibit a reduced level of homoeologous exchange or chromosome translocation to that observed in the CS ph1b mutant. The ph1b mutant line has been reported to accumulate extensive background translocations over multiple generations due to homoeologous synapsis and COs (Sánchez-Morán et al. 2001). Thus, the apparent lack of multivalents in the Tazip4-B2 mutant lines could allow their exploitation for introgression purposes during plant breeding programmes, rather than the current ph1b line.
Thus, seven lines carrying mutations within the TaZIP4-B2 gene were screened for a phenotype with reduced homologous crossover at metaphase I. Of these, two lines were identified with this phenotype, one carrying a nonsense mutation within TaZip4-B2, and the other carrying a mutation in one of the key functional domains of TaZip4-B2. When crossed with Ae. variabilis, both of these lines also exhibited increased homoeologous crossover at metaphase I in the resulting hybrid, suggesting that the two phenotypes were linked. This is consistent with our previous study, which scored for the absence of the Ph1 locus by the occurrence of reduced homologous crossover in wheat and increased homoeologous crossover in wheat-rye hybrids (Roberts et al. 1999;Al-Kaff et al. 2008). Therefore, lines with increased homoeologous crossover were identified without an initial screen for the desired phenotype. Until now, the only way by which homoeologous crossover at metaphase I could be increased to this extent in wheatwild relative hybrids was by deletion of the Ph1 locus, defined as a deletion effect phenotype specific to chromosome 5B. However, an alternative way of reproducing the Ph1 deletion effect would be to use EMS treatment to generate nonsense or truncation mutations in the homoeologous crossover-suppressing gene within the Ph1 locus. Analysis of the 1200-line TILLING population revealed that in any given mutant line, 1.5% of genes will have a truncation allele and 2% a missense allele (Krasileva et al. 2017). Thus, the probability of two mutant lines both sharing a truncation or missense mutation by chance in a second gene is P < 0.0005. The probability that two mutated genes will be located on the same chromosome is extremely low (2.4 × 10 −5 ), and the probability that they will both be located within the Ph1 region is even lower (2.4 × 10 −7 ). Thus, it is extremely unlikely that the increased homoeologous crossover phenotype found in both Tazip4-B2 mutant lines results from a nonsense mutation in a further gene independently linked with Tazip4-B2 within the Ph1 locus.
In terms of follow-on studies based around the observations reported here, we are currently backcrossing (BC) both Tazip4-B2 lines to clean up background mutations, as well as transferring Tazip4-B2 into the highly crossable hexaploid wheat cv. Chinese Spring. We are currently at BC 3 . BC lines will be made available once this exercise is complete. We have also complemented the approach of identifying chemically induced Tazip4-B2 mutants, by exploiting CRISPR to generate a large deletion within TaZIP4-B2. Initial analysis reveals that this Tazip4-B2 line has a similar phenotype to the two Tazip4-B2 Cadenza lines described above. We are currently segregating the transgenes away. TaZIP4-B2 is also being over-expressed to assess whether increased ZIP4 levels reduce homologous crossover. Our current hypothesis is that a higher ZIP4 level is optimal for homologous crossover, and a lower ZIP4 level is optimal for homoeologous crossover. Reducing or increasing ZIP4 around the optimal levels reduces the frequency of crossover. Interestingly, increasing the copy number of 5B chromosomes carrying TaZIP4-B2 reduces homologous crossover (Feldman 1966). In addition, a recent study identified another recombination pathway gene exhibiting dosage-dependent control on crossover (Ziolkowski et al. 2017). Thus, ZIP4 levels optimal for homologous crossover may be too high for homoeologous crossover, while the lower ZIP4 levels optimal for homoeologous crossover are too low for optimum homologous crossover. Finally, we are exploiting deletions of the less complex, homoeologous Cdk/methyl transferase locus on chromosome 5D, to shed further light on the role of these genes in the regulation of chromosome synapsis.
In summary, two Tazip4-B2 mutants were identified through a non-GM route, which can be exploited as an alternative to the CS ph1b mutant. Seeds for both mutants have been deposited with the Germplasm Resource Unit at the John Innes Centre (www.jic.ac. uk/research/germplasm-resources-unit). The accession number of seeds for Cad0348 is W10336 and for Cad1691 is W10337. The seeds for both lines are available on request, free of intellectual property restrictions.
Author's contribution These authors made the following contributions to the manuscript: M-DR, ACM, PS and GM designed the research; M-DR, ACM and JH performed the research; M-RD, ACM, JH, CU, PS and GM analysed the data; M-DR, PS and GM wrote the manuscript.
Compliance with ethical standards This research does not involve human participants or animals.

Conflict of interests
The authors declare that they have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.