Homoeologous Chromosome Sorting and Progression of Meiotic Recombination in Brassica napus: Ploidy Does Matter!

Comparisons of meiosis in near-isogenic allohaploid and euploid lines of B. napus reveal that the mechanism that promotes efficient chromosome sorting in euploids is adjusted to promote crossover formation between homoeologs in allohaploids. This suggests that, in contrast to other polyploid species, in B. napus, chromosome sorting depends on context.

Abstract

Meiotic recombination is the fundamental process that produces balanced gametes and generates diversity within species. For successful meiosis, crossovers must form between homologous chromosomes. This condition is more difficult to fulfill in allopolyploid species, which have more than two sets of related chromosomes (homoeologs). Here, we investigated the formation, progression, and completion of several key hallmarks of meiosis in Brassica napus (AACC), a young polyphyletic allotetraploid crop species with closely related homoeologous chromosomes. Altogether, our results demonstrate a precocious and efficient sorting of homologous versus homoeologous chromosomes during early prophase I in two representative B. napus accessions that otherwise show a genotypic difference in the progression of homologous recombination. More strikingly, our detailed comparison of meiosis in near isogenic allohaploid and euploid plants showed that the mechanism(s) promoting efficient chromosome sorting in euploids is adjusted to promote crossover formation between homoeologs in allohaploids. This suggests that, in contrast to other polyploid species, chromosome sorting is context dependent in B. napus.

INTRODUCTION

Meiosis is essential to the life cycle of all sexual eukaryotes; this specialized type of cell division not only ensures fertility and genome stability throughout sexual life cycles, but also generates diversity within species by creating new chromosome/allele combinations. All these outcomes depend on the formation of meiotic crossovers (COs), one of the products of meiotic double-strand break repair occurring during prophase I. At least two classes of COs coexist in plants (Mézard et al., 2007). The first class (class I COs) is dependent on the ZMM (for Zip, Msh, Mer) group of proteins (Osman et al., 2011) and gives rise to interfering COs, which are further apart from each other than would be expected if they were independent. The second class (class II COs) results in noninterfering COs and requires the METHYL METHANESULFONATE and UV SENSITIVE81 and METHYL METHANESULFONATE SENSITIVE proteins (Osman et al., 2011). Due to their role as a physical link between chromosomes, at least one obligate CO is required between pairs of homologs (Jones 1984), which probably arise from the class I pathway (Chelysheva et al., 2010). When this obligate CO is abolished, chromosomes segregate randomly, resulting in aneuploid gametes, aneuploid progenies, and reduced fertility (Martinez-Perez and Colaiácovo, 2009). The same negative outcome occurs when COs are formed between multiple and/or nonhomologous chromosomes, a situation that is more likely to occur in polyploid species (reviewed in Grandont et al., 2013).

In polyploid species, every chromosome has more than one potential partner, which can either be identical (when the polyploid results from the doubling of a single diploid genome, i.e., autopolyploidy) or slightly more divergent (when the polyploid has an interspecific hybrid origin, i.e., allopolyploidy). In any case, the different sets of chromosomes of a polyploid species must be sorted out during meiosis to produce balanced gametes. As a consequence, most allopolyploid species have inherited or evolved recombination-modifying loci that suppress CO formation between the “homoeologous chromosomes” inherited from the different parental species (Grandont et al., 2013, and references therein). The most well characterized of these loci is Pairing homeologous1 (Ph1) from wheat (Triticum aestivum; Griffiths et al., 2006), which was shown to affect CO formation both between homoeologs (Riley and Chapman, 1958; Greer et al., 2012, and references therein) and homologs (Lukaszewski and Kopecký, 2010).

In the past 10 years, Brassica napus has emerged as another model for studying natural variation in CO frequency between either homoeologs or homologs. B. napus (AACC, 2n=38) is a recent allotetraploid species that originated from multiple interspecific hybridization events between ancestors of Brassica oleracea (CC, 2n=18) and Brassica rapa (AA, 2n=20) (Allender and King, 2010, and references therein). These two parental species share an ancestral whole-genome triplication that occurred soon after the divergence of the Brassica and Arabidopsis thaliana lineages (Lysak et al., 2005). It has long been known that B. napus has a diploid-like meiotic behavior, although a small number of homoeologous exchanges, usually referred to as translocations, have been detected in several B. napus cultivars (Osborn et al., 2003; Udall et al., 2005; Howell et al., 2008). However, their frequency remains very low, especially when compared with the rate measured in newly created synthetic B. napus allotetraploids (i.e., colchicine-doubled B. rapa × B. oleracea hybrids). The enhanced genomic stability of natural B. napus could indicate that this species has inherited or evolved a locus (or loci) suppressing/reducing CO formation between homoeologous chromosomes. B. napus allohaploids also show significant variation in meiotic behavior at metaphase I, with two main meiotic phenotypes identified across a wide range of varieties (Nicolas et al., 2009; Cifuentes et al., 2010, and references therein). This variation relies on polygenic control, with one major quantitative trait locus, named Pairing regulator in Brassica napus (PrBn), being far more influential than the others (Cifuentes et al., 2010, and references therein).

Our current understanding of CO variation in B. napus is mainly based on cytological observations at metaphase I and genetic surveys of intergenomic exchanges in the progenies of B. napus allohaploids. As these approaches focused on the final products of meiotic recombination (chiasmata and genetic exchanges), they did not provide comprehensive information on the way(s) homologous bivalents form in euploid B. napus or the causes for the observed difference in CO rate between allohaploid plants. We do not even know whether the same strategy for homologous bivalent formation is implemented in B. napus euploid accessions originating from different polyploidization events (Cifuentes et al., 2010). The aim of this study, therefore, was to specifically address these questions. For this, we set up a thorough comparative analysis of meiosis in euploid and allohaploid plants chosen to be representative of (1) the different origins of B. napus and (2) the dichotomy of meiotic phenotypes observed among B. napus allohaploids. We investigated the formation, progression, and completion of several key hallmarks of meiosis, including sister chromatid cohesion, chromosome axes, the synaptonemal complex, meiotic recombination, in order to determine the extent to which these factors contribute to the observed differences or vary with the different origins of B. napus. Altogether, our results demonstrate that homologous and homoeologous chromosomes are sorted at an early stage of prophase I in two B. napus genotypes, which otherwise show varying numbers of class I COs. However, and contrary to other polyploid species, the mechanisms regulating efficient chromosome sorting in euploids do not suppress CO formation between homoeologs in the near isogenic allohaploids.

RESULTS

B. napus Euploids Show a Diploid-Like Meiotic Behavior

We first compared progression of male meiosis in Darmor-bzh and Yudal euploid genotypes (with 2n=38 chromosomes, i.e., AACC), which are representative of the two main B. napus gene pools (Harper et al., 2012). Staining chromosomes with 4′,6-diamidino-2-phenylindole (DAPI) showed that meiosis is very regular in Darmor-bzh and Yudal euploids (Figure 1) and very similar to meiosis in Arabidopsis (Ross et al., 1996). Briefly, meiotic chromosomes first condensed at leptotene (Figures 1A and 1K). At zygotene, we observed several close alignments of chromosomes (Figures 1B and 1L), which are diagnostic for the establishment of the synaptonemal complex. At pachytene, all chromosomes were closely aligned with one another (Figures 1C and 1M), indicating that homologous chromosomes were fully synapsed (see below). The synaptonemal complex then disassembled and chromosomes condensed until discrete separate bivalents became visible at diakinesis (Figures 1D and 1N). At metaphase I, we always observed 19 bivalents in the two genotypes with readily discernible chiasmata, the cytological manifestation of meiotic COs (Figures 1E and 1O). Based on chromosome shape (Jones 1984), we estimated a mean cell chiasmata frequency (n = 20 pollen mother cells [PMCs]) of 35 in both Darmor-bzh and Yudal (Table 1). The second meiotic division then took place (Figures 1G to 1I and 1Q to 1S). It resulted in four balanced sets of 19 chromatids (Figures 1J and 1T) that invariably led to tetrad formation (n = 250 and 251 cells for Darmor-bzh and Yudal euploids, respectively; Supplemental Figure 1). In both Darmor-bzh and Yudal, pollen grains were 100% viable (Supplemental Figure 1), demonstrating that meiosis in B. napus euploids leads to faithful chromosome segregation.

Figure 1.

Meiosis in Euploid B. napus.

DAPI staining of pollen mother cells during meiosis of Darmor-bzh ([A] to [J]) and Yudal ([K] to [T]) euploids. Leptotene ([A] and [K]): Chromosomes condense and become visible as unpaired threads. Zygotene ([B] and [L]): Arrows indicate several close juxtapositions of chromosomes that probably mark the initiation of the synaptonemal complex. Pachytene ([C] and [M]): All chromosomes are closely aligned with one another, suggesting that the synaptonemal complex is complete. Diakinesis ([D] and [N]): chromosomes are condensed and form discrete separate bivalents. Metaphase I ([E] and [O]): All bivalents aligned on the metaphase plate. Anaphase I ([F] and [P]): Homologous chromosomes, each composed of two sister chromatids, move to the opposite poles. Chromosome bridges are often observed at this stage but no chromosome fragmentation is observed at telophase I ([G] and [Q]). Metaphase II ([H] and [R]). Anaphase II ([I] and [S]): Individual chromatids segregate to the spindle poles. Late anaphase II ([J] and [T]): Cells contain four sets of 19 chromatids. Bars = 5 μm.

Table 1. Progression of Meiotic Recombination in B. napus Euploids (AACC; 2n=38) and Allohaploids (AC; n = 19).

	PMCs at Pachytene		PMCs at Diplotene		PMCs at Diakinesis			PMCs at Metaphase I
Genotypes	No. of PMCs	Avg. No. of HEI10 Focia	No. of PMCs	Avg. No. of HEI10 Foci	No. of PMCs	Avg. No. of MLH1 Foci	Avg. No. of HEI10 + MLH1 Foci	No. of PMCs	Avg. No. of Bivalents	Avg. No. of Chiasmata
Darmor-bzh euploids	13	27.46 (16–36)	7	23.86 (20–28)	33	–	34.82 (25–45)	20	19	35.45
Yudal euploids	37	28.05 (18–38)	11	29.09 (25–36)	29	–	31.00 (21–38)	20	19	35.05
		n.s.		n.s.			**
Darmor-bzh allohaploids	25	24.20 (14–40)	–	–	207	18.34 (3–39)	6.24 (1–10)b	19	6.67 (5–8)	8.92
Yudal allohaploids	71	16.41 (7–28)	–	–	134	10.27 (4–25)	2.47 (1–5)b	20	5.00 (3–8)	5.70
		**				**	**			**

Values in parentheses indicate the range of variation. n.s., not significantly different (Student’s t test). **, Significantly different (Student’s t test).

^aThese values were obtained for PMCs showing only distinct HEI10 foci.

^bThese values were obtained for 17 and 30 PMCs for Darmor-bzh and Yudal allohaploids, respectively.

B. napus Allohaploids Show Impaired Meiosis but Still Produce a Few Viable Pollen Grains

The early stages of meiosis in allohaploid plants (with 19 chromosomes, i.e., AC) appeared similar to those described in euploids. The first noticeable difference occurred at pachytene (Figures 2A and 2J). Both unaligned regions and stretches of juxtaposed regions between pairs of nonhomologous chromosomes were observed in all PMCs at this stage (hereafter referred to as pachytene-like PMCs) in both Darmor-bzh (n = 52) and Yudal (n = 59) allohaploids, suggesting partial synapsis (see below).

Figure 2.

Meiosis in Allohaploid B. napus.

DAPI staining of pollen mother cells during meiosis of Darmor-bzh ([A] to [I]) and Yudal ([J] to [S]) allohaploids. Pachytene ([A] and [J]): The presence of chromosomes that are not closely juxtaposed with one another (arrows) suggests that the synaptonemal complex is incomplete. Diakinesis ([B] and [K]). Metaphase I ([C] and [L]) with variable numbers of bivalents and univalents. Anaphase I ([D] and [M]): Nonhomologous chromosomes, each composed of two sister chromatids, are separated. At this stage chromosome bridges are often observed as in euploids. Telophase I ([E] and [N]): Two groups of chromosomes are observed, indicating that univalents moved to one or the other pole of the cell. Metaphase II ([F] and [O]). Anaphase II ([G] and [P]): Individual chromatids segregated to the spindle poles resulting in the formation of different kinds of meiotic products, including unbalanced tetrads ([H], [Q], and [R]), triads (I), as well as dyads (S). Bars = 5 μm.

At diakinesis (Figures 2B and 2K), the chromosomes looked diffuse, as if they were less condensed than in euploids, and they were often entangled, especially in Yudal allohaploids. At metaphase I (Figures 2C and 2L), both univalents (i.e., chromosomes that failed to form chiasmata) and bivalents were observed, with variable numbers between genotypes. The number of chiasmata was also significantly higher in Darmor-bzh (9.25 on average) than in Yudal (5.70 on average) allohaploids (P = 0.0001; Table 1), which is consistent with Cifuentes et al. (2010). In contrast to our observations in euploids, these chiasmata were formed between nonhomologous A and C chromosomes because allohaploids do not contain homologous chromosomes.

Regardless of the number of chiasmata formed, in allohaploid telophase I nuclei (Figures 2E and 2N), the chromosome composition was unequal, which, in most cases, led to unbalanced tetrads (Figures 2H, 2Q, and 2R) or polyads (when laggard chromosomes lead to the formation of additional nuclei) and, thus, to unviable pollen grains (Supplemental Figure 2). Unexpectedly, meiosis in Darmor-bzh and Yudal allohaploids also produced varying numbers of triads (Figure 2I) and dyads (Figure 2S); these were the minority in Darmor-bzh allohaploids (∼5%), but they represented half the products formed during meiosis in Yudal allohaploids (Supplemental Figure 2). Immunolocalization of α-tubulin (a component of the meiotic spindle) showed that the dyads and triads resulted from the formation of parallel and/or tripolar spindles during metaphase II (Supplemental Figure 3), which regrouped the products of the first division. Thus, the resulting microspores contained the 19 chromosomes of the basic B. napus chromosome set and probably generated the viable pollen grains observed in the two allohaploid genotypes (see pollen colored in red in Supplemental Figure 2).

Chromosome Axes Are Correctly Formed in Both Euploid and Allohaploid B. napus

Next, we investigated sister chromatid cohesion and assembly of meiosis-specific chromosome axes in B. napus, as these two meiotic landmarks are essential for CO formation. For this, we coimmunolocalized the meiosis-specific cohesin RECOMBINATION8 (SYN1/REC8) and the axis-associated protein ASYNAPTIC1 (ASY1), which are both required for wild-type levels of COs in Arabidopsis (Armstrong et al., 2002; Chelysheva et al., 2005).

REC8 and ASY1 staining was normal in both euploid and allohaploid B. napus (Supplemental Figures 4 and 5). The two proteins were correctly loaded on chromatin at leptotene and progressively formed a continuous signal as the chromosomes condensed and synapsed. The two signals colocalized along the entire length of the chromosome axes, whether they were synapsed or not in pachytene or pachytene-like cells. The signals persisted at diakinesis and metaphase I in both allohaploid and euploid B. napus.

These results suggest that sister chromatin cohesion and chromosome axes are correctly established and maintained in the two genotypes. As no difference was observed between Darmor-bzh and Yudal, the difference in meiotic behaviors observed at metaphase I in allohaploids is not due to an obvious defect in chromosome axis formation. Chromosome shape (based on DAPI staining) and ASY1 staining were also instrumental for discriminating between the different prophase I-like stages in allohaploids (Supplemental Figure 6). The chromosomes in early zygotene PMCs had a compact appearance and were strongly labeled with the anti-ASY1 antibody, whereas in pachytene-like PMCs, chromosomes looked less compact and a continuous ASY1 signal was observed. The chromatin in late diplotene PMCs was very diffuse and the ASY1 signal began to disappear from the chromosome axes. These criteria were used in the rest of our study.

The Extent and Nature of Synapsis Is Different in B. napus Euploids and Allohaploids

We then used two complementary approaches to gain a better understanding of synaptonemal complex formation in B. napus euploids (Figure 3). First, to monitor the progression of synapsis, we immunolocalized ZIPPER1 (ZYP1), a component of the transverse filament of the synaptonemal complex in A. thaliana (Higgins et al., 2005). Second, we analyzed surface-spread prophase I nuclei with electron microscopy to examine whether synaptic multivalents are formed prior to or at pachytene.

Figure 3.

Synaptonemal Complex Formation in B. napus Euploids.

(A) and (B) Chromosomes at pachytene were labeled with ASY1 (red) and ZYP1 (green) antibodies in Darmor-bzh (A) and Yudal (B) PMCs, respectively. An overlay of the ASY1 and ZYP1 signals is shown (merge). Bars = 10 μm.

(C) and (D) Electron micrographs of silver stained late-zygotene nucleus in Darmor-bzh (C) and pachytene nucleus in Yudal (D) showing only synaptic bivalents. Bars = 5 μm.

(E) to (G) High-magnification electron micrographs of quadrivalents in pachytene nuclei of Darmor-bzh ([E] and [F]) and Yudal (G) euploids with one ([E] and [G]) and two (F) synaptic partner switches and their corresponding schematic drawings. Arrows indicate synaptic partner switches. IL, interlocking. Bars = 2 μm.

The ZYP1 signal overlapped with the ASY1 signal along the entire length of the chromosomes in both Darmor-bzh and Yudal euploids (Figures 3A and 3B). This indicates that full synapsis does occur at pachytene in the two genotypes (n = 15 and n = 18 PMCs, respectively), even if the signal in Yudal was never as continuous as in Darmor-bzh. Electron microscopy observations confirmed that synapsis is complete in the two genotypes with no apparent difference between them (59 and 43 nuclei from late zygotene to pachytene were reconstructed for Darmor-bzh and Yudal euploids, respectively). In the two genotypes, 19 synapsed bivalents, which most likely correspond to homologous pairs, were observed only in ∼50% of nuclei (Figures 3C and 3D, Table 2). In the other ∼50% of nuclei, one (40% of nuclei) or two (∼10% of nuclei) synaptic quadrivalents, formed by the association of four chromosomes joined at different points, were observed, with no difference between the two genotypes (Figures 3E to 3G, Table 2; Supplemental Figure 7). The majority of these synaptic quadrivalents had only one synaptic partner switch (Figures 3E and 3G), and only 15% had two synaptic partner switches (Figure 3F). Most synaptic partner switches were preferentially located near the chromosome ends (subterminal or distal position), while a few occurred in more interstitial regions. We never observed a synaptic partner switch within the central part of the chromosomes. When synaptonemal complexes were measured in nuclei showing no or little distortion using electron microscopy (i.e., no interlock, no synaptonemal complex fragmentation, and no foldback loop), no significant difference in synaptonemal complex length between Darmor-bzh and Yudal euploids was found (532 ± 20 versus 503 ± 25 μm, respectively; Student’s t test, P = 0.389).

Table 2. Synaptic Associations in B. napus Euploids (AACC; 2n=38) at Late Zygotene-Pachytene.

	Number of PMCs with			Total No. of PMCs
Genotypes	19 II	17 II + 1 IV	15 II + 2 IV
Darmor-bzh euploids	28 (47.5%)	24 (40.7%)	7 (11.8%)	59
Yudal euploids	19 (44.2%)	20 (46.5%)	4 (9.3%)	43

II, bivalents; IV, quadrivalents. Values in parentheses indicate the proportion of cells in the different classes.

Synaptonemal complex formation in Darmor-bzh and Yudal allohaploids was examined using the same two approaches as for the euploids (Supplemental Figure 8). This confirmed that synapsis is never complete in allohaploids. Although several continuous stretches of ZYP1 signal were observed in both Darmor-bzh and Yudal allohaploids, most of these tracts were shorter than the chromosome length and were systematically accompanied by unsynapsed chromosome axes labeled by ASY1, but not ZYP1 (n = 16 and n = 15 PMCs, respectively). Nevertheless, electron microscopy did reveal the presence of one or two completely synapsed bivalents (Supplemental Figure 8) in ∼50% of prophase I nuclei, amid partially synapsed or completely unsynapsed chromosomes. These complete synaptic bivalents appeared to be as frequent in Darmor-bzh as in Yudal allohaploids (n = 46 and n = 35 PMCs, respectively).

The absence of complete synapsis made comparison between Darmor-bzh and Yudal allohaploids difficult; cell-to-cell differences could reflect either differences in the extent of synaptonemal complex completion or differences in the stages that were observed. With this limitation in mind, no obvious difference in synaptic behavior was found between Darmor-bzh and Yudal allohaploids with either immunolocalization or electron microscopy.

The Dynamics of Class I CO Formation Is Different between the Two Genotypes

We monitored the progression of meiotic recombination by immunostaining HUMAN ENHANCER OF INVASION10 (HEI10), a key protein that is unique because it allows the progressive transition from early recombination intermediates into final class I COs to be scrutinized (Chelysheva et al., 2012).

HEI10 showed the same dynamic localization pattern during prophase I in B. napus euploid plants as in A. thaliana (Chelysheva et al., 2012) and rice (Oryza sativa; Wang et al., 2012). At zygotene, multiple HEI10 foci were present along chromosome axes (Figure 4A; Supplemental Figure 9A). During pachytene, the number of HEI10 foci dropped dramatically, changing from a zygotene-like pattern (Figure 4B; Supplemental Figure 9B) until only a few distinct foci persisted on chromatin (Figure 4C; Supplemental Figure 9C). At diplotene (Figure 4D; Supplemental Figure 9D), distinct HEI10 foci were still present on the chromosomes where they systematically colocalized with MUT L HOMOLOG1 (MLH1) foci, a marker of class I COs (Chelysheva et al., 2010) at diakinesis (Figure 4E; Supplemental Figure 9E). Thus, HEI10 staining in B. napus euploids is progressively restricted to the sites where class I COs mature.

Figure 4.

Progression of Class I CO Formation in Darmor-bzh Euploids.

(A) to (D) Immunolocalization of ASY1 and HEI10 at zygotene (A), pachytene ([B] and [C]), and diplotene (D). Chromosomes were labeled with DAPI (white, false color), and HEI10 (green) and ASY1 (red) antibodies. An overlay of the three signals is shown (merge).

(E) Immunolocalization of MLH1 and HEI10 at diakinesis. Chromosomes were labeled with DAPI (white, false color), and HEI10 (red) and MLH1 (green) antibodies. The overlay of three signals is shown (merge).

Bars = 5 μm.

This HEI10 dynamic pattern was observed in both Darmor-bzh and Yudal euploids. However, two important differences were found. First, the pronounced decline in HEI10 foci was even stronger in Yudal than in Darmor-bzh (Figure 5). In Darmor-bzh euploids, 82% (60 out of 73) and 46% (6 out of 13) PMCs still showed numerous HEI10 foci along chromosome axes at pachytene and diplotene, respectively (Figure 5B; Supplemental Figure 10). By contrast, in Yudal euploids, the vast majority of PMCs (95% at pachytene, 37 out of 39; 100% at diplotene) showed only distinct HEI10 foci (Figure 5G). No significant difference was found when the number of HEI10 foci was compared in Darmor-bzh and Yudal pachytene PMCs that showed only distinct HEI10 foci (Student’s t test; P = 0.7); only a moderate difference was found at diplotene (P = 0.007).

Figure 5.

The Dynamics of Class I CO Formation Is Different between Darmor-bzh and Yudal Euploids.

The proportion of PMCs showing either a high number of HEI10 foci along the entire length of chromosome axes (in green) or a small number of distinct HEI10 foci on chromosome axes (in orange) is shown for Darmor-bzh euploid ([A] to [D]) and Yudal euploid ([E] to [H]). Bars = 5µm.

Second, the mean number of foci where HEI10 and MLH1 colocalized at diakinesis was higher in Darmor-bzh (∼35 foci; n = 33 PMCs) than in Yudal euploids (∼31 foci; n = 29 PMCs). This small difference was statistically significant (Student’s t test; P = 0.0014), suggesting that Darmor-bzh and Yudal differ slightly in their propensity to form interfering COs. Indeed, Darmor-bzh euploids also showed more bivalents with two, three, or more MLH1 foci than Yudal (Fisher’s exact test, P = 1.65 10⁻⁵; Supplemental Table 1).

We used the same approaches to compare the progression of meiotic recombination in Darmor-bzh and Yudal allohaploids (Figure 6; Supplemental Figure 11) and observed essentially the same HEI10 patterns as in the corresponding euploids. Of note, we observed the same difference in HEI10 dynamics at pachytene (Figure 5). In Yudal allohaploids, distinct HEI10 foci typical of the late HEI10 pattern were seen in 90% (71 out of 79) of pachytene-like PMCs, whereas these were observed in only 25% (25 out of 98) of PMCs in Darmor-bzh allohaploids. The largest HEI10 foci observed in B. napus allohaploids were never found on unsynapsed regions even at the end of pachytene (Figure 6C; Supplemental Figure 11C). This suggests that the formation of distinct HEI10 foci occurs only on synapsed regions.

Figure 6.

Progression of Class I CO Formation in Darmor-bzh Allohaploids.

(A) to (C) Immunolocalization of ASY1 and HEI10 at zygotene (A) and pachytene ([B] and [C]). Chromosomes were labeled with DAPI (white, false color), and HEI10 (green) and ASY1 (red) antibodies. An overlay of the three signals is shown (merge).

(D) Immunolocalization of HEI10 at diplotene. Chromosomes were labeled with DAPI (white, false color) and HEI10 (green) antibodies. An overlay of the two signals is shown (merge).

(E) Immunolocalization of MLH1 at diakinesis. Chromosomes were labeled with DAPI (white, false color) and MLH1 (green) antibodies. The overlay of two signals is shown (merge).

(F) Immunolocalization of MLH1 and HEI10 at diakinesis. Chromosomes were labeled with DAPI (white, false color), and HEI10 (red) and MLH1 (green) antibodies. The overlay of three signals is shown (merge). Arrows indicate stand-alone MLH1 foci, arrowheads indicate HEI10 foci, and asterisks show the colocalization of MLH1 and HEI10 foci.

Bars = 5 μm.

Despite these similarities, we found three differences between euploids and allohaploids. First, in allohaploids but not euploids, the mean number of HEI10 foci in pachytene-like PMCs was significantly higher in Darmor-bzh than in Yudal (24.2 versus 16.4; P < 0.001). This was mainly due to a significant decrease in the number of HEI10 foci in Yudal allohaploids compared with Yudal euploids (Table 1; P < 0.001), whereas the estimates were the same in Darmor-bzh allohaploids and euploids (Table 1; P = 0.19).

Second, almost twice as many MLH1 foci were seen in Darmor-bzh than Yudal allohaploids (on average, 18 versus 10 foci, respectively; Student’s t test, P < 0.001; Table 1, Figure 6E; Supplemental Figure 11E), while the variation was much smaller between euploids. Indeed, the distribution of MLH1 foci per cell fit a Poisson distribution in the two allohaploids but not in the euploids, in which an excess of counts near the mode and large deficiencies at the extremes of the distribution were found (Supplemental Figure 12). This strongly suggests that the distribution of MLH1 foci was random between PMCs in allohaploids, while it was constrained in euploids; this also explains why data counts were more widely dispersed in allohaploids than in euploids (Table 1).

Finally, coimmunolocalization of MLH1 with HEI10 revealed that the two proteins only rarely colocalized at diakinesis in allohaploids (Figure 6F), but systematically did so in the corresponding euploids (Figure 4E; Supplemental Figure 9E). Distinct and separate MLH1 and HEI10 foci were repeatedly observed at diakinesis in three independent experiments, in both Darmor-bzh and Yudal allohaploids. While the identity of the recombination intermediates marked by the distinct and separate MLH1 and HEI10 foci remain unknown, the two proteins colocalized in more foci in Darmor-bzh than in Yudal allohaploids (6.2 versus 2.5; P < 0.001). Thus, a significant difference was found between the two allohaploids regardless of which criteria were used to assess class I CO formation (number of HEI10 foci, number of MLH1 foci, number of MLH1+HEI10 foci; Table 1).

Finding a Partner to Recombine with: Chromosome- and Genotype-Specific Effects

Finally, we characterized recombination between individual chromosomes during meiosis in allohaploid B. napus. For this, BAC fluorescence in situ hybridization (FISH) experiments were performed using some of the BAC clones identified by Xiong et al. (2010), which specifically and simultaneously label pairs of homoeologous chromosomes/regions. We first focused on the A1-C1 pair of homoeologs (Figure 7), which are collinear along their entire length (Parkin et al., 2005). A1 and C1 were frequently held together by COs in Darmor-bzh and Yudal allohaploids (Figures 7A and 7B, Table 3), with no significant difference between the two genotypes (P = 0.25). However, in Darmor-bzh allohaploids, chiasmata bound the two arms (thus, at least two COs) of a slightly higher proportion of bivalents than in Yudal allohaploids (72.3% versus 60.4%). In the few PMCs in which A1 and C1 did not recombine together at metaphase I in Darmor-bzh (10 PMCs out of 129; 7.8%) or Yudal (nine PMCs out of 57; 15.8%) allohaploids, the chromosomes formed either a bivalent with another chromosome (Figure 7C) or remained univalent (Figure 7D).

Figure 7.

BAC FISH Survey of CO Formation for a Given Pair of Homoeologous Chromosomes (A1-C1) in B. napus Allohaploids.

Chromosomes were stained with DAPI (white, false color), and FISH was performed using two BACs that specifically and simultaneously identify A1 and C1 chromosomes (red and green). Metaphase I PMCs showing either only one bivalent labeled, indicating that A1 and C1 recombined together ([A] and [B]) or showing two labeled entities indicating that A1 and C1 did not recombine together ([C] and [D]). In (C), two bivalents are labeled, indicating that A1 and C1 recombined with other chromosomes (arrows). In (D), two univalents are labeled, indicating that A1 and C1 did not recombine. Bars = 5 µm.

Table 3. Homoeologous Bivalent Formation in B. napus Allohaploids (AC; n = 19) at Metaphase I.

	Frequencies of PMCs with a Bivalent between
Genotypes	A1-C1	A10-C9	A3-C3	A7-C6
Darmor-bzh allohaploids	92.8% (119/129)	80% (16/20)	53% (18/34)	40% (8/20)
Yudal allohaploids	84.2% (48/57)	35% (7/20)	71.4% (20/28)	50% (10/20)

Values in parentheses indicate the number of PMCs over the total number of PMCs observed.

We then extended our BAC FISH survey to three other pairs of chromosomes that share homoeology only along one of their arms, i.e., the long arms of A3/C3, the long arms of A7/C6, and the long arm of A10/short arm of C9 (Parkin et al., 2003, 2005; Xiong et al., 2010). Every pair showed a distinct and specific pattern at metaphase I (Supplemental Figure 13): A10 and C9 recombined together much more frequently in Darmor-bzh than in Yudal allohaploids (Table 3; P = 0.0095); A7 and C6 recombined together in the same proportion of PMCs in Darmor-bzh and Yudal allohaploids (Table 3; P = 0.7512); A3 and C3 recombined together a little more often in Yudal than in Darmor-bzh allohaploids (Table 3; P = 0.1915). We were also able to distinguish a series of other bivalents, for which only one operating chromosome was identified by the probes we used (e.g., A3-C_x, C9-A_x…). These additional bivalents more frequently involved A3 and/or C3 in Darmor-bzh allohaploids and A10 and/or C9 in Yudal allohaploids. This suggests that chromosome- and genotype-specific factors can change the odds of forming a CO between a given pair of homoeologous regions irrespective of the overall difference in CO rate between Darmor-bzh and Yudal allohaploids.

DISCUSSION

Early Homoeologous Chromosome Sorting in B. napus Euploids

Cytological diploidization of allopolyploid species can be achieved in different ways, depending on the stage when homologous and homoeologous chromosomes are successfully split apart (Jenkins and Rees, 1991). In this study, we showed that homoeologous chromosomes are sorted early on during prophase I in B. napus, synapsis being mostly restricted to homologs by late zygotene-pachytene in both Darmor-bzh and Yudal euploids. We observed no more than one or two synaptic multivalents in half of the PMCs observed at these stages (Table 2), the origin of which can be interpreted in two different ways.

The most straightforward hypothesis is that the synaptic multivalents result from synaptonemal complex formation between homoeologous regions. In that case, B. napus would be like wheat, Aegilops, and Festuca polyploids, in which varying numbers of homologous and homoeologous chromosomes are partially synapsed during prophase I (Hobolth, 1981; Thomas and Thomas, 1993; Cuñado et al., 1996). Given that, in plants, the synaptonemal complex does not form when strand invasion is impaired (Osman et al., 2011), the presence of homoeologous synaptic associations most likely indicates that strand exchanges are initiated between homoeologs in B. napus euploids. This is consistent with the demonstrated propensity for homoeologous chromosomes to recombine with one another in allohaploids (Figures 2 and 5). However, given the limited extent to which synaptic multivalents persisted to pachytene in B. napus euploids, it appears that most interhomoeolog recombination intermediates abort early and are probably redirected into intersister or noncrossover pathways (Hunter and Kleckner, 2001). Evidence for very short nonreciprocal exchanges between homoeologous sequences, which possibly originated from meiotic noncrossovers, was recently obtained in cotton (Gossypium hirsutum), another bivalent-forming allopolyploid species (Salmon et al., 2010; Flagel et al., 2012). Further studies are needed to demonstrate that similar events occur in B. napus.

As an alternative, the synaptic multivalents observed at pachytene may result from the presence of translocated A/C chromosome(s), in which one region has been replaced by its homoeolog that is thus duplicated within the genome. In that case, which was first advocated by Osborn et al. (2003), the synaptonemal complex could be formed only between homologous segments, albeit carried by different chromosomes; this would mean that B. napus homoeologous regions are sorted at an even earlier stage, or even more efficiently than envisaged above. This is exactly what happens in oat (Avena sativa), Avena marrocana, and Allium montanum, in which the synaptonemal complexes appear to be confined to homologous bivalents from zygotene onwards (Loidl, 1988; Jones et al., 1989). Although to date no A/C translocations have been described in Darmor-bzh and Yudal, their existence would help explain some of our BAC FISH observations; because the chance for CO formation is higher between duplicated homologous segments than between homoeologous regions, translocated A/C chromosome(s) (e.g., presumably A3/C3 in Yudal) are expected to recombine much more systematically than nonrecombinant chromosomes (Table 3). Thus, more work is needed to confirm the presence of translocated A/C chromosome(s) in Darmor-bzh and Yudal and their possible consequence on the synaptonemal complex.

Regardless of the origin of the synaptic multivalents, our results suggest that, in B. napus euploids, homoeologous chromosomes are sorted prior to, or during, formation of stable strand exchanges, which are thought to occur during zygotene concomitant with synaptonemal complex formation (Hunter and Kleckner, 2001; Tiang et al., 2012). The remaining multivalents are then eliminated before diakinesis, as if a second layer of control suppresses CO formation within the “illegitimately” synapsed regions. This latter adjustment probably remains an error-prone process and a few COs can be expected to occasionally form between homoeologs in euploids, in the regions where synapsis first took place. Our results suggest that these are fairly rare events (Table 1), although we have no evidence that COs never occur between homoeologs in B. napus euploids.

Chromosome Sorting Is Context Dependent in B. napus

In wheat and oat, the sorting of homoeologous chromosomes in euploids is paralleled by an almost complete suppression of CO formation between homoeologs in the corresponding allohaploids (Riley and Chapman, 1958; Gauthier and McGinnis, 1968). In these plants, even if chromosomes have no choice but to recombine with their homoeologs, they are prevented from doing so by loci responsible for the cytological diploidization of the euploid forms (Griffiths et al., 2006; Greer et al., 2012, and references therein). Our results show that the situation is strikingly different in B. napus; the homoeologous synaptic and chiasmatic associations that are suppressed in euploids become dominant in allohaploids. This indicates that the mechanism(s) responsible for the early sorting of homoeologous chromosomes in euploid B. napus do(es) not suppress CO formation between homoeologous chromosomes in allohaploids. This does not mean, however, that there is no sorting process operating in B. napus allohaploids. On the contrary, COs are preferentially formed between homoeologs in these plants and more occasionally between the duplicated blocks inherited from the Brassica ancestral whole-genome triplication (i.e., paleologs) (Nicolas et al., 2009, and references therein). Thus, it is as if a flexible but efficient sorting mechanism is adjusted to promote CO formation between the closest available matches (homologs in euploids and homoeologs in allohaploids), disregarding the other potential partners (homoeologs in euploids; paleologs in allohaploids).

Our detailed comparison of meiosis in euploid versus allohaploid B. napus thus suggests that the threshold for committing a pair of chromosomes to form a CO is not fixed once and for all in B. napus, but depends on the operating chromosomes. As mentioned above, this requires that only a fraction of recombination intermediates is set to become mature, a decision that cannot be made only locally (e.g., based on the stability of DNA heteroduplexes) because otherwise the intermediates that are committed to form COs in allohaploids could hardly be aborted in euploids. Instead, homology appears to be processed along the entire chromosome length. This destabilizes the weakest or less common nascent recombination intermediates formed between the most divergent chromosomes and promotes the strongest or more numerous intermediates that were formed, from the outset, between the least divergent chromosomes. As the outcome of this competition depends on the operating chromosomes, the most likely winners are expected to change with context.

The mechanism(s) for this “context-dependent chromosome sorting” is (are) unknown but appear(s) to be as efficient or work with almost the same level of stringency in Darmor-bzh and Yudal (see below). Indeed, no clear difference was observed between the two genotypes, which showed almost the same proportion of pachytene PMCs with 19 synapsed bivalents and the same frequency of synaptic multivalents (Table 2). Why, then, are there different numbers of meiotic COs between homoeologous chromosomes in allohaploid plants (Table 1)?

Variation in Class I CO Numbers between B. napus Varieties

In this study, we first demonstrated that the spatial-temporal localization of HEI10, a ZMM protein essential for CO formation in plants, is the same during prophase I in B. napus euploids as in A. thaliana and rice (Chelysheva et al., 2012; Wang et al., 2012). Thus, as with these two species, the dynamic relocation of HEI10 in B. napus probably reflects the progressive channeling of recombination intermediates into the ZMM CO pathway (see Chelysheva et al., 2012 for details). We then showed that this progression varies from one genotype to another. The transition from early to late HEI10 staining appeared sharper in Yudal than in Darmor-bzh (Figure 5). Finally, immunolocalization of MLH1 demonstrated that Darmor-bzh and Yudal form different numbers of class I COs (Table 1). Although it remains to be established whether the different patterns and/or chronology of HEI10 relocation contribute to this variation in class I COs, our results suggest that there are more COs in Darmor-bzh than in Yudal, all things being equal. Does it follow that allohaploids with higher CO frequencies have originated from allotetraploid accessions also showing higher numbers of homologous COs?

Interestingly, B. napus allohaploids displayed the same zygotene-to-diplotene patterns of HEI10 staining as the corresponding euploids (Figure 6; Supplemental Figure 11). This suggests that the progression of early meiotic recombination is essentially the same whether recombination intermediates are formed between homologs or homoeologs (provided, however, that there is no homolog to compete with them; see above). However, (late) distinct HEI10 foci were only found on synapsed regions in allohaploids (Figure 6C; Supplemental Figure 11C), suggesting that these are probably dependent on synaptonemal complex formation or conversely required to nucleate synaptonemal complex (see Reynolds et al. [2013] for comparable results in mice). Even more strikingly, we observed that the number of MLH1 foci per cell fit a Poisson distribution in the two allohaploids, as if they were distributed at random between PMCs. As first proposed by Jones (1967), this distribution may actually reflect a breakdown of the process(es) regulating the distribution of MLH1 foci at the bivalent level. In line with this hypothesis, we observed that only a small fraction of HEI10 and MLH1 foci colocalized at diakinesis in B. napus allohaploids (Figure 6F), while this was systematic in euploids (Figure 4E; Supplemental Figure 9E). The nature of these separate HEI10 and MLH1 foci is unknown, but it is probably no coincidence that they were found in plants showing partial synapsis (Qiao et al., 2012, and references therein). As recently proposed for the tomato (Solanum lycopersicum) asynaptic mutant as1 (Qiao et al., 2012) or haploid Arabidopsis (Cifuentes et al., 2013), the “stand-alone” MLH1 foci could in fact mark the locations where COs eventually failed or occurred between sister chromatids. As there is bias against forming COs between sister chromatids (at least in haploid Arabidopsis; Cifuentes et al., 2013), it is possible that a fraction of late recombination intermediates are still in the process of being resolved at diakinesis in B. napus allohaploids, resulting in some HEI10 foci persisting longer than usual on chromosomes without producing the conditions required for MLH1 loading. Finally, separate HEI10 and MLH1 foci may reflect an aberrant turnover of meiotic proteins in allohaploid plants, with MLH1 loading and off-loading irrespective of HEI10.

Regardless of the causes and consequences of separate HEI10 and MLH1 foci, there were significantly more HEI10 foci, more MLH1 foci, and more “MLH1+HEI10” foci in Darmor-bzh allohaploids than Yudal allohaploids (Table 1; Supplemental Figure 12). Variation in class I CO rates may thus explain a significant proportion of the variation in total CO rate observed between Darmor-bzh and Yudal allohaploids. It is of note, however, that the mean number of distinct HEI10 foci is significantly lower at pachytene in Yudal allohaploids compared with Yudal euploids (16.4 versus 28.0 foci; P < 0.001; Table 3), while it is almost the same in Darmor-bzh allohaploids and euploids (24.20 versus 27.46 foci; P = 0.19; Table 3). This suggests that fewer double-strand breaks are committed to form COs between homoeologous chromosomes in Yudal allohaploids than in euploids (or in Darmor-bzh). Thus, our results point toward a possible difference in the sensitivity to chromosome divergence between the two genotypes.

In fact, it may be pointless to consider suppression of COs between homoeologs and variation in CO number between homologs as separate processes in allopolyploid species. There is some indication, for instance, that Ph1, the main locus responsible for the cytological diploidization of wheat (Riley and Chapman, 1958; Griffiths et al., 2006; Greer et al., 2012, and references therein), affects recombination between dissimilar homologs (Lukaszewski and Kopecký, 2010, and references therein), suggesting that this locus regulates some basic mechanism of chromosome recognition (Greer et al., 2012). Whether the same holds true for PrBn and B. napus is an avenue worth exploring.

METHODS

Plant Material and Growth Conditions

The production of allohaploid plants from Brassica napus cv Darmor-bzh and Yudal was described by Cifuentes et al. (2010).The plants were cultivated in a greenhouse or growth chamber under a 16-h-light/8-h-night photoperiod, at 22°C day and 18°C night, with 65% humidity.

Cytology

Male meiotic spreads for DAPI staining were prepared as described by Chelysheva et al. (2013) from buds fixed in Carnoy’s fixative (absolute ethanol:acetic acid, 3:1, v/v).

Final male meiotic products were observed by toluidine blue staining as described by Azumi et al. (2002), and mature pollen grain viability was estimated as described by Alexander (1969). Images were taken with a Leica Diaplan bright-field microscope.

Immunolocalization of α-Tubulin

Anthers dissected from fixed buds at the appropriate meiotic stage were stained according to a protocol adapted from Mercier et al. (2001). Buds were incubated for 5 min in citrate buffer, pH 6.0, and rinsed in water. After a first digestion of 1 h in digestion mix (0.3% [w/v] cellulase RS, 0.3% [w/v] pectolyase Y23, and 0.3% [w/v] cytohelicase in citrate buffer), one anther was placed on a polysin slide, dissected in water, squashed, and fixed in nitrogen. The released cells were immobilized with a thin layer of 1% gelatin (w/v), 1% agarose (w/v), and 7% glucose, incubated for 5 min in citrate buffer, pH 6.0, and digested for a further 1 h in digestion mix at 37°C. After three rinses in 0.1% PBS-T (10 mM sodium phosphate, pH 7.0, 143 mM NaCl, and 0.1% Triton X100), cells were incubated for 1 h and 30 min in 1% PBS-T (10 mM sodium phosphate, pH 7.0, 143 mM NaCl, and 1% Triton X100) at room temperature and rinsed two times with 0.1% PBS-T. Cells were incubated overnight at 4°C with an anti-α-tubulin mouse monoclonal antibody (Sigma-Aldrich) diluted 1/325 in PBS-T-BSA (1% BSA in 0.1% PBS-T). After three rinses with 0.1% PBS-T, cells were incubated for 2 h and 30 min in fluorescein isothiocyanate–labeled secondary antibody (labeled goat anti-mouse IgG Alexa fluor 488) diluted 1/100 in PBS-T-BSA at 37°C. After three rinses in 0.1% PBS-T, the cells were mounted in Vectashield antifade medium (Vector Laboratories) with 2 μg/mL DAPI.

Immunolocalization of Meiotic Proteins

Coimmunolocalization of ASY1/REC8, HEI10/ASY1, and MLH1/HEI10 was performed on meiotic spreads prepared as described previously for DAPI staining. Immunolabeling was performed as described by Chelysheva et al. (2013). The coimmunolabeling of ASY1/ZYP1 was performed as described by Chelysheva et al. (2007) from fresh buds. The anti-ASY1 polyclonal antibody has been described elsewhere (Armstrong et al., 2002). It was used at a dilution of 1:250. The anti-REC8 polyclonal antibody (Cai et al., 2003) was used at a dilution of 1:250. The anti-HEI10 polyclonal antibody (Chelysheva et al., 2012) was used at a dilution of 1:200. The anti-MLH1 antibody (Chelysheva et al., 2012) was used at a dilution of 1:20 (purified serum). The anti-ZYP1 polyclonal antibody (Higgins et al., 2004) was used at a dilution of 1:500.

BAC FISH Experiment

BAC FISH experiments on meiotic spreads (see above) were performed as described by Lysak et al. (2006) with the following modifications.

Probe Labeling

To label A1/C1, A3/C3, A7/C6, and A10/C9 homoeologous chromosomes, five BACs from Brassica rapa (KbrB036M17 and KbrB052L10 for A1/C1; KbrH117M18 for A3/C3; KbrB21P15 for A7/C6; and KbrH80A08 for A10/C9) that were shown to be specific for A1 in B. rapa (Kim et al., 2009) and expected to hybridize to homoeologous regions of the C genome of B. napus (Xiong and Pires, 2011) were used. The five BACs were labeled using Biotin-Nick Translation Mix (Roche) and DIG Nick Translation Mix (Roche) as described by Lysak et al. (2006) or random priming with Alexa 488-5-dUTP and biotin-14-dUTP (Invitrogen, Life Technologies) (Suay et al., 2013). The labeled BACs were then precipitated and the dry pellet dissolved in 10 μL HB50 (50% deionized formamide, 2× SSC [0.30 mM NaCl and 0.030 M Na3-citrate], and 50 mM sodium phosphate, pH 7.0) and 10 μL SD20 (20% dextran sulfate in HB50).

In Situ Hybridization

The probe DNA was prepared by adding ∼100 ng each labeled BAC to 9 μL HB50 and 9 μL SD20. The probe DNA was then denatured at 92°C for 6 min and stored at 70°C until use. One hundred microliters of RNase was added to the meiotic spreads and incubated at 37°C for 60 min. Slides were then rinsed in 2× SSC at room temperature for 2 × 5 min. All washing steps were performed in Coplin jars. If chromosomes/nuclei appeared to be covered by cytoplasm after checking slides in a phase contrast microscope, 100 μL pepsin (100 µg/mL in 0.01 n HCl) was added and the slide was incubated at 42°C until cytoplasm was no longer visible. Care was taken not to treat the slide with pepsin for too long, to avoid degradation of chromatin structures. The slides were then postfixed in 4% formaldehyde in 1× PBS at room temperature for 10 min, rinsed in 2× SSC at room temperature twice for 5 min each, dehydrated through an ethanol series (70, 90, and 100%) for 1 to 3 min each, and air-dried overnight. Then, 100 μL 70% formamide was added and slides were heated at 70°C for 2 min. After a maximum of 2 h of air-drying, the slides were dehydrated with a cold ethanol series (70, 90, and 100%) for 1 min each.

Finally, the denaturated probe DNA was added on the slides, which were then covered with a 22 × 22 cover slip. The slides were placed in a moist chamber and incubated overnight at 55°C.

Fluorescence Detection

Detection of hybridized probes was performed as described by Lysak et al. (2006).

Fluorescence Microscopy

Observations were made using a Leica DM RXA2 microscope or a Zeiss Axio Imager 2 microscope; photographs were taken using a CoolSNAP HQ camera (Roper) driven by OpenLAB 4.0.4 software or a Zeiss camera AxioCam MR driven by Axiovision 4.7. All images were further processed with OpenLAB 4.0.4, Axiovision 4.7, or Adobe Photoshop 7.0 (Adobe Systems).

Electron Microscopy

The synaptonemal complex spreads were prepared according to the protocol described by López et al. (2008) with minor modifications: 0.05% (v/v) Triton X-100 + 0.1% (v/v) Lipsol detergent was included in the swelling medium. The slides were stained with aqueous AgNO₃ (40%, w/v) at 45°C and examined under an electron microscope (Jeol 1010). Micrographs were composed using Photoshop CS4 software (Adobe Systems). A total of 46 and 35 nuclei from allohaploid and 59 and 43 from euploids, respectively, were photographed (captured nuclei); the remaining nuclei exhibited basically the same features as those selected. A complete reconstruction of allohaploid captured nuclei was not possible because the axial/lateral elements were partially destroyed in the asynaptic regions.

Statistical Analysis

Student’s t test and χ² analyses were performed using the PROC FREQ and PROC TTEST procedure of SAS, respectively (SAS Institute Inc., 1999).

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: KBr036M17, AC189325; KBrB052L10, AC189386; and KbrH117M18, AC146875.

Supplemental Data

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

• Supplemental Figure 1. Male Meiotic Products in Brassica napus Euploids.

• Supplemental Figure 2. Male Meiotic Products in Brassica napus Allohaploids.

• Supplemental Figure 3. Meiotic Spindle Orientation at Metaphase II in Brassica napus Allohaploids.

• Supplemental Figure 4. Sister Chromatid Cohesion and Meiotic Chromosome Axes in Brassica napus Euploids.

• Supplemental Figure 5. Sister Chromatid Cohesion and Meiotic Chromosome Axes in Brassica napus Allohaploids.

• Supplemental Figure 6. Chromosome Shape and ASY1 Staining Allow Staging of Prophase I Pollen Mother Cells in Allohaploids.

• Supplemental Figure 7. Electron Micrographs of Silver-Stained Pachytene Nuclei in Brassica napus Euploids.

• Supplemental Figure 8. Synaptonemal Complex Formation in Brassica napus Allohaploids Pollen Mother Cells at Pachytene.

• Supplemental Figure 9. Progression of Class I Crossover Formation in Yudal Euploids.

• Supplemental Figure 10. Immunolocalization of HEI10 and ASY1 in Darmor-bzh Euploids in Diplotene Pollen Mother Cells Showing a Multitude of HEI10 Foci along Chromosome Axes.

• Supplemental Figure 11. Progression of Class I Crossover Formation in Yudal Allohaploids.

• Supplemental Figure 12. Distribution of the Number of MLH1 Foci per Cell in Euploid and Allohaploid Brassica napus.

• Supplemental Figure 13. BAC FISH Survey in Brassica napus Allohaploids: Meiotic Crossover Formation between Homoeologous Chromosome Pairs A10/C9, A3/C3, and A7/C6, Respectively.

• Supplemental Table 1. Distribution of MLH1/HEI10 Foci per Bivalent in Brassica napus Euploids.

Glossary

CO

crossover

DAPI

4′,6-diamidino-2-phenylindole

PMC

pollen mother cell

FISH

fluorescence in situ hybridization