Nuclear dispositions of subtelomeric and pericentromeric chromosomal domains during meiosis in asynaptic mutants of rye (Secale cereale L.)

Meiosis is a specialized cell division of sexually reproducing eukaryotes, which halves the somatic chromosome number during gametogenesis and thereby ensures that the chromosome number of an organism does not double at each generation. Integral parts of the process include the recognition, pairing and synapsis of homologues, which in most eukaryotes are pre-requisites for genetic recombination and balanced segregation of half-bivalents at anaphase I. It is clear from several studies that the integrity of these early meiotic phenomena is largely invested in the behaviour and interaction of two chromosomal domains, the centromeres and telomeres, because perturbations of their nuclear dispositions can dramatically impact upon the successful completion of meiosis I (reviewed by Zickler and Kleckner, 1998; Walker and Hawley, 2000). The importance of the centromeric domain is well illustrated in two previous studies showing that, in achiasmate meiosis, centromeric heterochromatin fulfils a role in the pairing of chromosomes and facilitates regular segregation (Dernburg et al., 1996; Karpen et al., 1996).

Because centromeres and telomeres are implicated in the control of chromosome pairing, it is perhaps not surprising that the nuclear dispositions of these domains have received considerable attention, particularly at pre-meiotic interphase and early prophase I when homologues may be making initial contact. The ordered arrangement of these domains at pre-meiotic interphase into a typical bipolar Rabl orientation (Rabl, 1885) is apparent in some organisms, such as fission yeast (Chikashige et al., 1997), budding yeast (Jin et al., 1998), hexaploid bread wheat (Abranches et al., 1998) and other members of the Triticeae (Martinez-Perez et al., 2000). Furthermore, pre-meiotic association of homologues has been observed in Saccharomyces cerevisiae (Weiner and Kleckner, 1994; Loidl et al., 1994) and wheat (Aragon-Alcaide et al., 1997b; Schwarzacher, 1997; Mikhailova et al., 1998; Martinez-Perez et al., 1999), which has been interpreted as an important component of chromosome pairing. The dynamic reorganization of telomeres into a tight cluster bound to the nuclear periphery has also received considerable attention (reviewed by Zickler and Kleckner, 1998; Walker and Hawley, 2000). In maize (Bass et al., 1997), humans (Scherthan et al., 1996), budding yeast (Trelles-Sticken et al., 1999) and other species, the bouquet arrangement of telomeres occurs at the transition between leptotene and zygotene and may be transient or persist throughout zygotene (Scherthan et al., 1996). In maize and humans, where there is no discernible pre-meiotic association of homologues, the clustering of the telomeres of chromosomes is considered to be the first step in the pairing of homologues (Dernburg et al., 1995; Scherthan, 1997; Zickler and Kleckner, 1998; Bass et al., 2000). However, it does not necessarily guarantee subsequent regularity of meiosis, because the bouquet forms in recombination-defective mutants spo11 and Rad50s of yeast (Trelles-Sticken et al., 1999). Hexaploid wheat also forms a bouquet, but this occurs earlier at the onset of meiotic prophase and persists until the end of zygotene (Martinez-Perez et al., 2000).

This investigation compares the nuclear disposition of pericentromeric and subtelomeric domains at pre-meiotic interphase and meiosis I in two asynaptic mutants of rye. It constitutes part of an ongoing research programme to characterize genetically and cytologically 21 existing, spontaneous meiotic mutant stocks (Sosnikhina et al., 1994), with the aims of understanding the genetic control of meiosis, and of ultimately isolating genes responsible for specific meiotic events in this organism. Microsporocytes of asynaptic mutant sy1 form axial elements in the homozygote during leptotene, but these do not assemble into synaptonemal complexes (SCs), with the result that 99% of pollen mother cells (PMCs) have only univalents at metaphase I (Sosnikhina et al., 1992). Homozygotes for asynaptic mutation sy9 also form axial cores but, in contrast to sy1, 33% of PMCs form between 1 and 7 effective bivalents (Sosnikhina et al., 1998). Mutations sy1 and sy9 are not allelic, and the latter is epistatic (Sosnikhina et al., 1998). The more specific aim of this project is to determine whether or not the differences in meiotic phenotypes and epistatic relationship between the two mutants could be interpreted in terms of differences in the disposition and behaviour of pericentromeric and subtelomeric domains.

Asynaptic mutant sy1 was originally isolated from an individual plant of weedy rye (Sosnikhina et al., 1992) and is maintained in an inbreeding population of diploid winter rye (Secale cereale L; 2n=2×=14). Asynaptic mutant sy9 was recovered from an individual plant of the rye variety Vyatka (Sosnikhina et al., 1998) and is propagated in the same way. Homozygotes for sy1 and sy9 were isolated from segregating progenies of selfed heterozygotes and were identified on the basis of their meiotic phenotypes. One plant homozygous for sy1 and three for sy9 were used in this analysis. All four plants had a characteristically high frequency of univalents at metaphase I (Table 1), and other cytological abnormalities were entirely consistent with published observations (Sosnikhina et al., 1992; Sosnikhina et al., 1998). Three fertile plants from families segregating for the mutants were used as wildtype (Sy1- or Sy9-) controls. These had slightly elevated univalent frequencies compared with those in rye taken from populations or varieties, and some minor irregularities as a result of inbreeding (Table 1; Sosnikhina et al., 1994).

Spikelets from meiotic inflorescences were numbered from the base of the spike, fixed individually in a fresh 3:1 (v/v) mixture of ethanol and acetic acid, and stored at −20°C.

The accurate determination of meiotic stage is crucial to the comparisons made between the three phenotypes in this study. Traditional staging methods based upon progress of synapsis are not applicable in this case, because the pairing of chromosomes in asynaptic mutants is compromised. Similarly, stage determination at early meiosis based upon the relative clustering of centromeres and telomeres is inappropriate, because the behaviour and nuclear disposition of these domains are also affected in the mutants. Staging of meiocytes was therefore accomplished by adhering to the following criteria applied equally and without bias to each phenotype:

(1) Immature inflorescences (spikes) of rye have a developmental gradient along their lengths, with younger spikelets towards the base and more advanced stages in the middle. The intervals between adjacent spikelets along the length of the spike are predictable, which has particular use in selecting stages before leptotene. (2) Each spikelet contains three anthers, which are approximately synchronous in development. One of these was staged by standard squashing in 1% aceto-carmine, and recorded for future reference by phase-contrast microscopy under a Zeiss Axioplan microscope equipped with an MC100 camera and black- and-white film. Pre-meiotic interphase is characterized by two or more nucleoli centrally located in the nucleus (Bennett et al., 1973), and the absence of DAPI-positive threads in FISH preparations. The onset of meiotic prophase is typically heralded by the aggregation of nucleoli into one large nucleolus, which adopts a peripheral location in the nucleus. The nuclei contain distinct threads, reflecting the progressive condensation of the leptotene chromosomes. (3) Mononucleate tapetal cells stained with aceto-carmine or DAPI indicate that pollen mother cells are at stages before leptotene. Synchronous division of tapetal cells coincides with transition between leptotene and zygotene, and binucleate tapetal cells correspond with zygotene (Bennett et al., 1973; Bennett et al., 1979; Martinez-Perez et al., 1999).

All of the anthers used in this study have been staged by applying the above criteria, and only individual anthers were prepared for FISH. However, because there is some variation in stage even within the same anther, we have not rigidly assigned a stage to each cell analyzed. Rather, we have defined meiotic stage intervals, which have been applied to the three phenotypes.

Fixed anthers were prepared for FISH following procedures essentially as previously described (Schwarzacher and Heslop-Harrison, 2000; Zhong et al., 1996), with some modifications (Jenkins et al., 2000). Because the method was optimized to preserve chromatin and to increase the resolution of FISH at early stages of prophase I, the salient features of the method are recorded below. Individual anthers were washed for 20 minutes in 10 mM citrate buffer (pH 4.5) before digestion for 2 hours and 15 minutes in an enzyme mixture comprising 0.5% (w/v) cellulase (Calbiochem), 0.65% (w/v) cellulase (Onozuka RS), 10% (v/v) pectinase (Sigma), 0.15% pectolyase Y-23 (Seishin Pharmaceutical Co. Ltd) and 0.15% cytohelicase (Sigma) in 20 mM citrate buffer (pH 4.5). After digestion, the anthers were washed for 15 minutes in 10 mM citrate buffer, followed by sterile distilled water (SDW), before maceration with a fine needle and passage through a pipette tip in 20 μl of SDW. The suspension of anther cells was spun briefly at 13,000 rpm, and the pellet was washed in 10 μl of 45% acetic acid, spun down again and resuspended in 12 μl of fresh 45% acetic acid. 4 μl aliquots of the suspension were spotted onto clean slides and, for squash preparations, a 20×20 mm coverslip lightly applied and excess fixative removed. Following removal of the coverslips by freezing at −80°C, the preparations were post-fixed in a few drops of fresh 3:1 methanol-acetic acid. For spread preparations, the droplets of suspension were surrounded by fresh, ice-cold 3:1 ethanol-acetic acid and the cells allowed to precipitate onto the glass. The spreads were then washed in fresh fixative. Both squash and spread preparations were prepared, because the preservation of 3D organization was considered to be better in the latter. However, it was found subsequently by optical sectioning that nuclei prepared in either way had approximately the same finite depths.

All preparations were dehydrated in 100% ethanol, air-dried and stored at 4°C (short-term) or −20°C (long-term). Before FISH, slides were pre-treated with RNAse A (10 μg/ml) in 2× SSC for 1 hour at 37°C, followed by washing in 2× SSC, refixing for 10 minutes at room temperature in 1% formaldehyde in PBS (pH 7.0), further washing in 2× SSC, dehydration in an ethanol dilution series and air-drying.

pSc200 is a 521 bp insert in pUC18 comprising a 380 bp tandem repeat unit of subtelomeric DNA from rye (Vershinin et al., 1995). The sequence localizes to 18 major subtelomeric sites and four minor sites in the chromosome complement of wild-type rye and the asynaptic mutants. The insert was amplified and labelled by PCR using M13 forward and reverse primers and FluoroRed (rhodamine-4-dUTP) or FluoroLink Cy3-dCTP (Amersham Pharmacia). First, the insert was amplified using the following conditions: 1 cycle of 93°C for 5 minutes, 35 cycles of 94°C for 30 seconds, 55°C for 40 seconds, 72°C for 1 minute 30 seconds, and 1 cycle of 72°C for 5 minutes. Then the insert itself was labelled as follows: 18 cycles of 94°C for 30 seconds, 55°C for 30 seconds, 72°C for 1 minute 30 seconds, and 1 cycle of 72°C for 5 minutes.

CCS1 is a 260 bp region (Aragon-Alcaide et al., 1996) of a centromere-specific clone (Hi-10) originally isolated from Brachypodium sylvaticum (Abbo et al., 1995). It is localized exclusively to the pericentromeric regions of all rye chromosomes. The sequence was amplified by PCR in the presence of digoxygenin-11-dUTP (Roche) as described previously (Aragon-Alcaide et al., 1996). Both probes were purified by precipitation under ethanol.

FISH was performed according to an amalgam of procedures (Leitch et al., 1994; Aragon-Alcaide et al., 1996; Fransz et al., 1996; Zhong et al., 1996; Schwarzacher and Heslop-Harrison, 2000). For clarity, the salient features of the method adopted are given below. The two probes were mixed to a final concentration of 1-2 ng/μl in a hybridization solution containing 50% (v/v) formamide, 2× SSC, 10% (w/v) sodium dextran sulphate, 50 mM phosphate buffer (pH 7.0) and 125 ng/μl sonicated salmon sperm DNA. The mixture was denatured at 70°C for 10 minutes, snap-cooled on ice, pipetted onto the preparations of anther cells and overlaid with plastic coverslips. The slides were sealed into the humid chamber of an Omnislide thermal cycler (Hybaid), denatured at 70°C for 5 minutes and incubated overnight at 37°C. They were then washed stringently in 0.1× SSC at 50°C for 30 minutes and 2× SSC for 15 minutes at room temperature. The preparations were blocked with 5% (w/v) nonfat dry milk/4× SSC solution for 30 minutes at room temperature in the dark, before detection of the digoxygenated pericentromeric sequence by incubation for 1-2 hours at 37°C with sheep anti-digoxygenin antibodies conjugated to fluorescein. After three washes in 0.05% Tween-20 in 4× SSC, the slides were dehydrated in an ethanol dilution series, air-dried and mounted in Vectashield (Vector Laboratories) containing DAPI (20 μg/ml) and propidium iodide (0.4 μg/ml). Fluorescent images were visualized, processed and stored using either a Zeiss Axioplan microscope and MC100 camera, or a Zeiss Axiovert microscope coupled to a Bio-Rad MRC-1024 MPR confocal laser scanning microscope. To be sure of visualizing all FISH signals, the confocal micrographs shown are composites of optical section series through individual nuclei.

Discrimination between meiocytes and tapetal cells in heterogeneous cell populations at this interval was effected as follows. Meiocytes from sister anthers have a characteristic triangular form and granular chromatin when stained with aceto-carmine (Fig. 1A). In FISH preparations stained with DAPI, meiocytes have egg-shaped nuclei with evenly distributed chromatin, whereas tapetal cells have more spherical nuclei with polarized chromatin (Fig. 1B).

To compare more objectively the nuclear dispositions of pericentromeric and subtelomeric domains in the three phenotypes throughout meiotic prophase, the following categories of aggregation were adopted:

(1) The telomeric/centromeric domains are considered ‘clustered’ if they comprise a single, amorphous and indivisible mass closely adpressed to the nuclear periphery at one of the poles of the nucleus. (2) ‘Multiple clusters’ describes 2-5 distinct aggregates in the same region of the nucleus. (3) ‘Grouped’ refers to largely or wholly separate telomeric/centromeric domains that occupy a distinctly polarized region of the nucleus but mainly unattached to the nuclear periphery.

(4) ‘Dispersed’ indicates that largely or wholly separate telomeric/centromeric domains are distributed without obvious pattern throughout the nucleus.

The relative frequencies of the four classes of subtelomeric and two classes of pericentromeric distribution throughout meiotic prophase of the three phenotypes are presented in Fig. 2. It is not considered likely that the patterns of distribution of these domains are disturbed to any great extent by the preparatory techniques, because it has already been shown that clustering is essentially the same in spread cells compared with those in which the 3D order has been preserved (Trelles-Sticken et al., 1999). Furthermore, in the unlikely event that disturbances occur, it is probable that the three phenotypes would be affected in the same way and comparisons would still be valid.

At pre-meiotic interphase in wildtype and both asynaptic mutants, the pericentromeric and subtelomeric domains adopt a typical Rabl orientation, which is consistent with similar observations of other species of the Triticeae (Abranches et al., 1998; Aragon-Alcaide et al., 1997a; Martinez-Perez et al., 1999; Martinez-Perez et al., 2000). In all three phenotypes the pericentromeric regions are clustered at one pole of the nucleus and the subtelomeric regions grouped at the opposite pole (Fig. 3A-C). This distinctive pattern of distribution of these domains is clearly visible in tapetal nuclei also (Fig. 3A-C). At the onset of meiotic prophase, marking the transition between pre-meiotic interphase and leptotene, the subtelomeric but not pericentromeric regions undergo considerable reorganization in wildtype and sy9 to form a tight, polar cluster in all or most nuclei (Fig. 3D,E). By contrast, sy1 fails to form a bouquet in half of its nuclei. Instead, the subtelomeric DNA forms multiple clusters (Fig. 3F), and in 71.4% of the nuclei studied the pericentromeric regions are dispersed or interspersed between the subtelomeric aggregates (Fig. 2). Thus, the sy1 mutation exhibits incomplete penetrance in homozygous condition, since four out of 14 PMC nuclei have aggregations of pericentromeric/subtelomeric DNA comparable with wildtype (Fig. 3F). In all three phenotypes tapetal nuclei retain the Rabl orientation of these domains.

As leptotene progresses, subtelomeric clusters dissipate in to a point at which a large proportion of nuclei contain dispersed sites (Fig. 3G). Pericentromeric clusters become more relaxed and the pericentromeric regions themselves become distended (Fig. 3G). A minority of these regions are clearly associated in parallel pairs in some nuclei, which is possibly indicative of presynaptic alignment (not shown). The phenotype of sy9 at this stage bears some similarity to in that the highest proportion of nuclei has multiple clusters of subtelomeric regions. In sy1 the majority of nuclei have multiple clusters of subtelomeric regions. The pericentromeric sites are loosely grouped at the opposite pole, and comprise pairs of fuzzy dots or short distended regions (Fig. 3I). There is no obvious alignment of pairs of pericentromeric regions as in wildtype, although this may be attributable to the small sample size of nuclei taken. In some cells, 14 discrete signals reflect the asynaptic condition of this mutant.

During zygotene the tight aggregation of subtelomeric DNA lapses in wildtype (Fig. 4A) so that, by pachytene, broken aggregates are dispersed seemingly randomly throughout the nuclei (Fig. 4D). At the latter stage, the centromere regions have a distinctive tripartite structure comprising two condensed pericentromeric subdomains flanking the centromere proper (Fig. 4D). Mutant sy9 has 50% of nuclei with dispersed subtelomeric regions (Fig. 2), as in (41.5%), but retains a sizeable proportion of subtelomeric regions (31.2%), which form multiple clusters at zygotene (Fig. 4B). Fourteen pericentromeric regions can now be easily counted and reveal the same tripartite structure as wildtype (Fig. 4B). However, at a later stage they appear more distended and less discrete (Fig. 4E) than wildtype (Fig. 4D). A large majority (73.3%) of nuclei of sy1 still have multiple clusters of subtelomeric regions at this interval (Fig. 4C,F) with no nuclei showing dispersal of these domains as in wildtype and sy9 (Fig. 4D,E). It appears that the resolution of the bouquet is hampered, with the possible consequence that multiple clusters of subtelomeric regions persist throughout this period. Pericentromeric regions are fuzzy and differ in the degree of distension within the complement (Fig. 4F).

During this period of meiosis, most of the subtelomeric regions of wild-type chromosomes are dispersed throughout the nucleus. Progressive contraction of the seven bivalents fuses the three pericentromeric subdomains into one by the end of diakinesis. All of the nuclei of sy9 have dispersed subtelomeric regions at this stage (Fig. 4G). Mutant sy1 still has multiple clusters of subtelomeric regions, but some are evidently dispersing by this stage. Fourteen pericentromeric regions of sy9 and sy1 can now be easily counted, representing the characteristic number of univalents at this stage of meiosis. The morphology and contraction status of the pericentromeric DNA of the 14 univalents are indistinguishable from wildtype at this stage (Fig. 4H,I).

As expected, wildtype regularly forms seven bivalents at metaphase I, each of which displays discrete pericentromeric and subtelomeric signals (Fig. 5).

The fluorescent labelling of pericentromeric and subtelomeric regions of rye chromosomes has enabled the tracking before and during meiosis of two important domains implicated in the control of chromosome pairing, synapsis and recombination. The results presented above are summarized in Fig. 6, which compares the essential behaviour of the two domains in the three phenotypes. Fig. 6 embodies a large number of observations and represents by necessity the average picture of nuclear organization at the various sub-stages. At pre-meiotic interphase the two domains adopt a typical Rabl configuration, in which the bipolar distribution of centromeres and telomeres reflect the orientation of chromosomes in the previous mitotic anaphase. The beginning of meiosis, however, heralds the tight clustering of telomeres, which become adpressed to the nuclear periphery. The formation of a characteristic bouquet is specific to pollen mother cells and occurs earlier in rye than in other organisms (Zickler and Kleckner, 1998). Indeed, in a diversity of organisms such as maize (Bass et al., 1997; Bass et al., 2000), yeast (Trelles-Sticken et al., 1999) and mammals (Scherthan et al., 1996) this major reorganization of telomeres occurs at the transition between leptotene and zygotene and is coincident with the onset of synapsis. The timing of bouquet formation is similar to allohexaploid bread wheat (Moore, 1998), with the one notable difference that centromeres are not associated in pairs at this stage in rye (Aragon-Alcaide, 1997b; Martinez-Perez et al., 1999; Martinez-Perez et al., 2000). Clearly, the formation of the bouquet at the transition between premeiotic interphase and leptotene is not dependent upon prior association of homologous chromosomal domains. The association of pairs of pericentromeric regions in this material becomes evident later as these domains pre-align or undergo synapsis.

Asynaptic mutant sy1 differs significantly from wildtype in that, at the start of meiotic prophase, subtelomeric DNA fails to fully cluster and the pericentromeric regions disperse prematurely. If, as has been suggested (Zickler and Kleckner, 1998), the clustering of the telomeres and the centromeres is an integral component of homologue recognition, alignment and recombination, their aberrant behaviour at this stage could be directly responsible for the synaptic phenotype of this mutant. However, it has been shown that homologues in yeast will synapse despite the failure of bouquet formation, indicating that the bouquet is not absolutely necessary for meiotic progression in this organism (Trelles-Sticken et al., 2000). It could, of course, be possible that the failure to form a bouquet is a consequence of a lesion in a different meiosis-specific process. For example, the tardy resolution of subtelomeric clusters in sy1 could be symptomatic of the paucity of recombination intermediates, analogous to spo11 (Cao et al., 1990) and Rad50s (Weiner and Kleckner, 1994) of budding yeast. Differences between the nuclear disposition of these domains in wildtype and sy1 could reflect differences in the status of meiotic chromatin organization, although the differences are manifested earlier compared with the chromatin restructuring observed at the leptotene/zygotene transition in maize (Dawe et al., 1994).

Asynaptic mutant sy9 bears closer similarity to wildtype in terms of its timing of formation and resolution of the bouquet. The presence of subtelomeric clusters during early prophase I is inconsistent with previous observations under the electron microscope of axial elements of surface-spread meiocytes, in which no clusters of telomeres were detected (Sosnikhina et al., 1998). A possible explanation for this discrepancy is that telomeres are notoriously difficult to discern at leptotene in these preparations and may have been overlooked. Alternatively, the bouquet may have been at an advanced stage of resolution at the meiotic prophase stages observed. Indeed, in the one microsporocyte nucleus illustrated (Sosnikhina et al., 1998) the telomeres form two distinct aggregates, which would be interpreted as a multiple cluster in the present work. The nuclear disposition of pericentromeric regions of sy9 at pre-meiotic interphase is similar to wildtype. However, at the onset of meiotic prophase the pericentromeric regions become distended and decondensed. It is not known whether this change in morphology compromises the integrity of meiosis from this point onwards, or whether it is a manifestation of failure of another process. It has been speculated that the abnormally diffuse centromeres in the ph1 mutant of wheat could be responsible for altering the way in which homologues interact during meiotic prophase (Aragon-Alcaide et al., 1997b). Subsequently, pairs of centromeres seldom associate, no synaptonemal complexes are formed and the majority of homologues fail to form chiasmata.

We thank Trude Schwarzacher and Graham Moore (John Innes Centre, Norwich) for supplying the subtelomeric and pericentromeric clones, and Steve Taylor for his expert technical assistance with the confocal microscopy. E. I. M. would like to thank J. Hans de Jong and J. Wennekes van Eden for extensive training in FISH during collaboration on an INTAS project in Wageningen. We acknowledge with gratitude the receipt of a Royal Society ex-Agreement award for E. I. M. to visit Aberystwyth, and support from the Russian Foundation for Basic Research (grants 00-04-48522 and 99-04-48182). The initial contact between the two laboratories was sponsored by the Federation of European Genetical Societies.