Homoeologous synapsis is suppressed during the telomere bouquet stage independently of Ph1
It has been suggested that the Ph1 locus promotes homologous synapsis rather than suppressing homoeologous synapsis. However, the mechanism through which this is achieved is unknown. In most species, including hexaploid wheat, homologues gradually align and synapse during meiosis, while telomeres cluster forming a bouquet at one pole of the nucleus (Zickler and Kleckner 1999; Prieto et al. 2004; Griffiths et al. 2006; Zhang et al. 2014; Klutstein and Cooper 2014). In yeast, it has been proposed that the telomere bouquet can restrict ectopic pairing between homologous sites in non-homologous chromosomes and so promotes homologous synapsis (Niwa et al. 2000; Davis and Smith 2006). If homoeologous pairing is also restricted in hexaploid wheat during the telomere bouquet, this would provide an explanation of how regular bivalent formation between homologues at metaphase I is achieved in wild-type wheat. Given the number and size of wheat chromosomes in hexaploid wheat, it would be very difficult to follow the relative proportion of overall homologous versus homoeologous synapsis during the telomere bouquet stage. However, the use of wheat-rye hybrids facilitates this study. Wheat-rye hybrids possess only homoeologous chromosomes; no homologues are present. Therefore, only homoeologous synapsis can occur. We have previously shown that in these hybrids, homoeologous synapsis is observed at late zygotene and pachytene, after dispersal of the telomere bouquet (Martín et al. 2014). However, it was unclear whether this homoeologous synapsis occurred at the telomere bouquet stage or only later.
Here, we used wheat-rye hybrids to investigate the timing of homoeologous synapsis in relation to the telomere bouquet. We combined FISH, to label the telomeres, with immunolocalisation of the meiotic proteins ASY1 and ZYP1. ASY1 is part of the lateral elements of the synaptonemal complex and is loaded before synapsis. ZYP1 is part of the central element of the synaptonemal complex and is present only where chromosomes are synapsed. For the purpose of this study, we classified the hybrid meiocytes into three groups, based on the number of telomeres detected: 1 to 5 telomere groups, defining the telomere bouquet stage (group 1); 6 to 14 telomere groups, identifying early bouquet dispersal (group 2); and more than 14 telomere groups, identifying when the telomere bouquet has dispersed (group 3). FISH and immunolocalisation analysis show that there are no long tracks of ZYP1 in wheat-rye hybrid meiocytes during the telomere bouquet, whether Ph1 is present or absent (Fig. 1). The little ZYP1 that is present in some of the cells during the telomere bouquet stage is shown by very short tracks of ZYP1 (Online Resource 1), which may indicate an attempt to start the synapsis process, although the polymerisation process fails to progress further. In the presence of Ph1, 87% of meiocytes show these short tracks of ZYP1, with no ZYP1 present in the remaining meiocytes (Online Resource 1). Conversely, when Ph1 is absent, only 17% of meiocytes show short tracks of ZYP1 (Online Resource 1), suggesting slight delay compared to when Ph1 is present. In any case, long tracks of ZYP1 are only observed after the telomere bouquet stage, whether or not Ph1 is present. Homoeologous synapsis thus mostly occurs after the telomere bouquet has dispersed (Online Resource 2), and not during the telomere bouquet stage as happens with homologous synapsis.
Homologous synapsis is delayed in the absence of Ph1
If suppression of homoeologous synapsis during the telomere bouquet is independent of the presence of Ph1, we would expect to observe regular bivalent formation at metaphase I in 100% of meiocytes, independently of the presence of Ph1. However, some multivalents are observed in wheat in the absence of Ph1 (Martín et al. 2014), implying that there is at least some degree of synapsis between homoeologues. Even so, in the absence of Ph1, there are multivalents in less than 50% of meiocytes; and even when multivalents are present, only up to 8 of the 42 chromosomes are involved (Martín et al. 2014). More than half of the meiocytes show regular bivalent configurations, indicating that homoeologous synapsis is mainly prevented in wheat even in the absence of Ph1, consistent with suppression of homoeologous synapsis at the telomere bouquet stage. This suggests, as reported, that Ph1 may promote homologous synapsis rather than preventing homoeologous synapsis, and in the absence of Ph1, disruption of this process may allow a degree of homoeologous synapsis to occur.
In order to assess the dynamics of overall chromosome synapsis in wheat and to identify any difference in the timing of synapsis in the absence of Ph1, we combined telomere labelling and immunolocalisation of ASY1 and ZYP1 proteins. Meiocytes from wheat, in the presence and absence of Ph1, were analysed from the beginning of telomere bouquet formation until the telomeres had completely dispersed. For the purpose of this study, we again divided the meiocytes into three groups, based on the number of telomere groups detected: 1 to 5, defining the telomere bouquet stage (group 1); 6 to 19, identifying early bouquet dispersal (group 2); and 20 or more telomere groups with no ASY1-ZYP1 colocalisation, identifying when the telomere bouquet is dispersed but total synapsis is not yet completed (group 3). Meiocytes with colocalisation of ASY1 and ZYP1 (occurring in wheat after pachytene) were not considered further in the analysis. Analysis suggests that the presence or absence of Ph1 had no effect on the level of synapsis at the telomere bouquet stage (group 1) or during early bouquet dispersal (group 2) (Fig. 2a). However, when telomeres were dispersed (group 3), the level of ASY1 labelling was higher in the absence of Ph1, indicating a lower level of synapsis at that stage (Fig. 2c). By the end of pachytene, most chromosomes were synapsed in the absence of Ph1 (Online Resource 3), indicating eventual completion of synapsis in most meiocytes even when Ph1 was absent. These observations suggest that even if synapsis completes whether or not Ph1 is present, progression of synapsis is slower in the absence of Ph1 (as shown in group 3) with respect to the telomere dynamics.
In order to assess whether pairing was actually delayed in groups 1 and 2, FISH analysis was performed with specific probes to distal sites close to the telomere regions and to more proximal sites (see “Materials and methods” section). Probe Spelt52 produces a single signal in the distal region of 4BS chromosomes. One signal was observed at the telomere bouquet stage (group 1), whether Ph1 was present or absent, indicating that only homologues associate at the telomere bouquet stage, independent of Ph1 (Griffiths et al. 2006) (Fig. 2b). To study the pairing of more proximal chromosome sites, we used the specific probe 4P6, which labels seven interstitial sites on the D genome (Zhang et al. 2004) and which can be used to follow the dynamics of more interstitial sites undergoing later pairing. Meiocytes in which pairing has not occurred will exhibit 14 signals, 7 from each homologue, while meiocytes that have undergone complete regular pairing will exhibit only 7 signals. Therefore, meiocytes with either unpaired or incorrectly paired sites will exhibit between 8 and 14 signals (the probe cannot distinguish between unpaired and incorrectly paired sites). Here, meiocytes in the presence and absence of Ph1 were collected at the stage of telomere bouquet dispersal (group 2, with 6 to 19 telomere groups detected), to identify any delay in pairing. In the presence of Ph1, most of the seven pairs of sites were associated, with 85.7% of them displaying 7 to 10 signals (Fig. 2d; Online Resource 4). In contrast, in the absence of Ph1, only 25% of the meiocytes displayed 7 to 10 signals, the remainder still showing more than 10 signals (Fig. 2d; Online Resource 4). This difference is consistent with a delay in pairing. It has been previously reported that the length of meiotic prophase I is similar whether or not Ph1 is present (Bennett et al. 1971, 1974). Therefore, it is unlikely that the differences in pairing or synapsis related to the absence of Ph1 are due to a gross difference in the timing of meiotic stages.
Thus, both synapsis and FISH analysis suggest that in the absence of Ph1, homologous chromosomes recognise each other during the telomere bouquet, but synapsis progresses more slowly. Since synapsis between homoeologues only occurs after the telomere bouquet, the delay of homologous synapsis in the absence of Ph1 provides an opportunity for homoeologues to synapse after telomere bouquet dispersal. Previous yeast meiotic studies show that delayed pre-meiotic replication can subsequently delay chromosome pairing (Borde et al. 2000; Murakami et al. 2003). We have also previously reported altered pre-meiotic replication and chromatin in the absence of Ph1 (Greer et al. 2012), proposing a possible subsequent effect on chromosome pairing and synapsis. Thus, this may explain the delayed synapsis found in the absence of Ph1.
The level of CO formation in wheat and wheat-rye hybrids in the absence of Ph1 is affected by the nutrient composition in the soil
The above data provide an explanation for how homoeologous pairing and synapsis might occur in hexaploid wheat in the absence of Ph1. However, it does not tell us anything about the effect of Ph1 on CO formation. Fortuitously, our wheat-rye hybrids were grown recently, with and without Ph1, for chromosome doubling experiments at the Institute for Sustainable Agriculture (CSIC), Córdoba, Spain. Plants were grown outdoors in pots and treated with a modified Hoagland solution during their growth, in order to produce healthier plants for chromosome doubling. Surprisingly, when analysing the metaphase I configurations, we observed a higher level of chiasmata in plants lacking Ph1 than had been previously observed in plants grown in a glasshouse or under controlled environmental conditions in Norwich UK. Some Córdoba meiocytes exhibited up to 21 chiasmata at metaphase I in the absence of Ph1, close to that expected from the number of MLH1 sites observed on synapsed homoeologues at diplotene (Martín et al. 2014). We, therefore, tried to replicate this observation in Norwich, in order to determine which of the different Córdoba growth conditions were responsible for the increased level of chiasmata in wheat-rye hybrids lacking Ph1. It is widely accepted in wheat studies that each chiasma represents a CO (Fu and Sears 1973); therefore, we use both terms synonymously. We first investigated whether the modified Hoagland solution or different temperature could be responsible for the increased CO number. We performed the experiments both in a glasshouse and under controlled environmental conditions, with similar results; however, only data generated under controlled environmental conditions are presented here.
To assess the effect of nutrient concentration in the soil on homologous and homoeologous CO frequency in meiotic metaphase I, we added a modified Hoagland solution to the soil in which wheat and wheat-rye hybrids, both lacking Ph1, were growing at 20 °C. Under these controlled environmental conditions, the modified Hoagland solution had little effect on vegetative growth. However, the treatment significantly increased the number of ring bivalents per meiocyte in wheat lacking Ph1 (a mean of 16.45 against the 15.71 without Hoagland), with a corresponding reduction in rod bivalents (3.26 with Hoagland and 4.27 without Hoagland) (Fig. 3). GISH labelling of the different wheat genomes (A, B and D) revealed that most ring bivalents were formed between homologues (Online Resource 5), suggesting an increase in homologous CO frequency. The treatment also increased homoeologue CO frequency in the wheat-rye hybrids lacking Ph1, shown by an increased number of ring bivalents (2.35 with Hoagland and 1.40 without Hoagland) (Fig. 3b). The number of COs increased from a mean of 8.71 (without Hoagland) to 12.09 (with Hoagland) COs per meiocyte, with up to 21 COs scored in some meiocytes. As described previously (Martín et al. 2014), this is close in number to the 21 MLH1 sites observed on the synapsed homoeologues. Thus, both homoeologous and homologous COs can be increased in wheat lacking Ph1 by altering the nutrient composition of the soil in which they are grown. Previous studies have suggested that soil nutrient composition in which plants are grown can affect CO formation (Grant 1952; Law 1963; Bennett and Rees 1970; Fedak 1973; Deniz and Tufan 1998). We therefore also treated wheat-rye hybrids carrying Ph1 with the same nutrient solution; however, no significant increase in CO number was observed (Online Resource 6).
The level of CO formation in wheat and wheat-rye hybrids in the absence of Ph1 is affected by temperature
The addition of the Hoagland solution was thus identified as one of the possible causes for the increase in CO number observed in Cordoba, in wheat-rye hybrids, in the absence of Ph1. We then decided to also assess the possible effect of temperature. Wheat and wheat-rye hybrids in the absence of Ph1 were grown under controlled environmental conditions, at either 13 or 30 °C during the meiotic period, without the addition of modified Hoagland solution. Both temperature treatments had little effect on vegetative growth.
At 13 °C, the total number of COs was significantly increased in both the wheat (Fig. 4a) and the wheat-rye hybrids (Fig. 4b), in the absence of Ph1. The number of COs increased in wheat from a mean of 36.90 at 20 °C to a mean of 37.63 at 13 °C and in wheat-rye hybrids from a mean of 8.71 at 20 °C to a mean of 11.19 at 13 °C. Both genotypes showed an increase in the number of ring bivalents with a corresponding reduction in rod bivalents (Fig. 4).
Conversely, at 30 °C, the total number of COs was significantly decreased in both the wheat (Fig. 4a) and the wheat-rye hybrids (Fig. 4b), in the absence of Ph1. The number of COs decreased in wheat from a mean of 36.90 at 20 °C to a mean of 33.30 at 30 °C and in wheat-rye hybrids from a mean of 8.71 at 20 °C to a mean of 7.64 at 30 °C. The lowering of chiasma frequency at high temperature was accompanied by an increased frequency of univalents and rod bivalents in wheat and an increase of univalents in wheat-rye hybrids, both in the absence of Ph1. Thus, in the absence of Ph1, the temperature at which wheat is grown can affect the frequency of CO, being significantly increased at low temperature (13 °C) and decreased at high temperature (30 °C). These results are consistent with previous studies in wheat, maize, Arabidopsis and barley, reporting that growth temperature can influence the formation of COs during meiosis (Bayliss and Riley 1972; Verde 2003; Francis et al. 2007; Higgins et al. 2012). In contrast, wheat-rye hybrids carrying Ph1 grown at either 20 or 13 °C exhibited 0.59 and 0.77 COs at metaphase I respectively. This level of CO is not significantly great to exploit practically. We did not study the effect of high temperatures on CO frequency in wheat (or wheat-rye hybrids) carrying Ph1, as it has already been reported that such temperatures reduce CO frequency (Bayliss and Riley 1972).
Overall, these results suggest that, in the absence of Ph1, temperature and soil nutrient composition can affect the level of homologous and homoeologous COs in wheat and wheat hybrids. In the wheat-rye hybrids, both treatments lead to an increase in conversion of rod bivalents to ring bivalents at metaphase I. This suggests an increase in COs between synapsed homoeologues, rather than an increased association of ectopic sites. Genetic map analysis of wheat in the presence and absence of Ph1 showed a similar distribution of COs along chromosomes (Dubcovsky et al. 1995), suggesting that the Ph1 locus affects the level of MLH1-dependent class I interfering COs, but not their distribution. Thus, it is unlikely that nutrients and/or temperature treatments lead to an altered distribution of COs along synapsed chromosomes in the absence of Ph1. It is more likely that temperature and one or more components of the modified Hoagland solution affect the MLH1 complex activity on the double Holliday junction or affect the double Holliday junction conformation itself, allowing the junction to resolve as a CO rather than a non-CO. For instance, changes in magnesium concentration have been shown to affect the conformation of double Holliday junctions, which may then influence CO frequency (Yu et al. 2004; Ranjha et al. 2014).
The Ph1 is a complex locus: proposal for the mode of action of the Ph1 locus
Ph1 is a complex locus, and after almost 60 years since its discovery, we still do not know its exact mode of action. Molecular characterisation of the Ph1 locus combined cereal synteny and wheat BAC contiging, with metaphase I analysis of mutants carrying deletions of chromosome 5B (Roberts et al. 1999; Griffiths et al. 2006; Al-Kaff et al. 2008). Finally, it was defined to a region containing a cluster of Cdk2-like and S-adenosyl methionine-dependent methyltransferase (SAM-MTases) genes, with a duplicated segment of heterochromatin from 3B inserted into this cluster. This heterochromatin segment also contains a gene originally designated as hypothetical 3 (Hyp3) (Griffiths et al. 2006; Al-Kaff et al. 2008), which has been now reannotated as ZIP4 (UniProtKB—Q2L3T5).
In terms of the role of Ph1 during meiosis, we now know that Ph1 stabilises polyploidy in wheat through two different mechanisms: by controlling the accuracy of homologous synapsis during early meiosis and by regulating CO formation later in meiosis (Martín et al. 2014). We propose that the effect of Ph1 on synapsis is probably a consequence of a change in chromatin structure produced by the Cdk-like and SAM-MTase cluster. The involvement of Cdk2 in H1 phosphorylation, replication, chromatin condensation and synapsis between non-homologous chromosomes has been widely reported (Krasinska et al. 2008; Viera et al. 2009). We have already shown that the levels of histone H1 phosphorylation, pre-meiotic replication and chromatin structure are all altered by Ph1 (Greer et al. 2012), which may affect chromatin structure and hence the efficiency of synapsis. On the other hand, although ZIP4 has been reported to affect both synapsis and CO formation in yeast (Tsubouchi et al. 2006), the effect in plants is mainly on CO formation (Chelysheva et al. 2007; Shen et al. 2012). Thus, it seems more likely that ZIP4 is involved in the effect of Ph1 on CO formation. However, there are still no data to elucidate how ZIP4 contributes to the effect of Ph1 on the CO process. This question cannot be answered using ZIP4 RNAi-based approaches because this would result in overall ZIP4 activity suppression, leading to sterility (Chelysheva et al. 2007; Shen et al. 2012). However, the question will most likely be answered in the near future, using other approaches such as analysis of the recently available tilling mutants (www.wheat-training.com/tilling-mutant-resources) and other transformation approaches such as the CRISPR genome editing technique.