- Top of page
- Materials and Methods
- Supporting Information
Polyploidy is a widespread phenomenon in angiosperms. Allopolyploids, which carry merged genomes from different species, play an emphasized role in genome evolution (Ramsey & Schemske, 1998; Wendel, 2000). Although some de novo-synthesized or newly formed natural allopolyploid species show structural stability (Liu et al., 2001; Baumel et al., 2002; Mestiri et al., 2010), genetic rearrangements in most polyploids occur rapidly (Pires et al., 2004; Pontes et al., 2004; Soltis et al., 2004; Hegarty et al., 2006; Tate et al., 2009; Petit et al., 2010) via prevalent mechanisms, such as crossovers (COs), between homoeologous chromosomes at meiosis (Chen & Ni, 2006; Leitch & Leitch, 2008; Gaeta & Pires, 2010; Szadkowski et al., 2010).
Allopolyploids differ from homoploids in that they possess two copies of each parental genome. This situation cannot be obtained directly by hybridization of two haploid gametes from different species. One mechanism that is known to induce polyploidy in plants is somatic doubling in mitosis (Ramsey & Schemske, 1998; Comai, 2005). Somatic doubling can occur either spontaneously, as in Primula kewensis, or be induced at a high frequency by artificial means (Ramsey & Schemske, 1998 and references therein). A second mechanism by which polyploids may evolve from diploid plants involves the production of unreduced gametes during meiosis (‘unreduced gamete pathway’), that is, gametes with the same number of chromosomes as the parental plant (Bretagnolle & Thompson, 1995; Ramsey & Schemske, 1998). This pathway is now considered to be the main source of polyploids in the wild. One major type of unreduced gamete is the ‘first division restitution (FDR) gamete’, in which the somatic number of chromosomes is conserved (Bretagnolle & Thompson, 1995). These unreduced FDR gametes can transmit the products of meiotic recombination between homologous chromosomes in diploids (d’Erfurth et al., 2008) and, interestingly, the products of nonhomologous chromosome exchange in haploids of allopolyploid plants (Nicolas et al., 2007). It is thus surprising that, despite the prevalent role of unreduced gametes in allopolyploidy and their potential effect on genome structure, all recent experiments studying early polyploidy generations have focused only on resynthesized plants generated by mitotic genome doubling (Song et al., 1995; Comai et al., 2000; Pires et al., 2004; Wen et al., 2008; Chaudhary et al., 2009; Petit et al., 2010). As a consequence, we still do not know whether the somatic vs unreduced gamete pathways have a different and/or specific impact on the genetic modifications that occur at the onset of polyploid formation.
Rapeseed (Brassica napus, AACC genomes, 2n = 38) is a model allopolyploid plant for the study of meiotic-driven genetic changes (see Pires et al., 2004; Gaeta et al., 2007; reviewed in Gaeta & Pires, 2010; Nicolas et al., 2007, 2009; Szadkowski et al., 2010). Its genomes are highly related to those of Brassica rapa (AA genome, 2n = 20) and Brassica oleracea (CC genome, 2n = 18) (U, 1935). F1 hybrids can be obtained by embryo rescue (Wen et al., 2008) and produce resynthesized allopolyploids from somatic doubling. In these resynthesized B. napus, frequent aberrant meiotic configurations, such as multivalents (Attia & Robbelen, 1986b), are found and numerous genetic changes are detected after a few generations even if each chromosome has a homolog (Parkin & Lydiate, 1997; Udall et al., 2005; Gaeta et al., 2007). These changes are commonly described as translocations or rearrangements even though most are the products of meiotic COs (see Gaeta & Pires, 2010). In a recent study, we showed that the very first meiosis of somatically doubled resynthesized B. napus already acts as a genome blender, with many of the genetic exchanges between A and C chromosomes transmitted to progenies (Szadkowski et al., 2010). However, Brassica species are also able to generate viable unreduced gametes (Olsson & Hagberg, 1955; Heyn, 1977), which may therefore suggest that this pathway is involved in natural B. napus formation. In haploids of natural B. napus that share the same AC structure as F1 hybrids, almost all viable gametes are unreduced, and it has been shown that this haploid structure promotes nonhomologous recombination between chromosomes A and C (Nicolas et al., 2007). The consequences of meiosis in newly formed interspecific F1 hybrids have not yet been studied, but, compared with meiosis in somatically doubled F1 hybrids, may have a major effect on the genetic rearrangements that occur. We therefore aimed to unravel the impact of polyploid formation pathways on genetic rearrangements during the formation of new B. napus.
We first used a genome-wide cytogenetic approach to analyze meiosis in an F1 hybrid. Genetic analysis of the homoeologous chromosome pair A1/C1 in the progeny of the F1 hybrid then allowed comparison with previously published data on somatically doubled resynthesized S0 plants which share the same genetic background (Szadkowski et al., 2010). Lastly, to evaluate the potential of unreduced gametes to create an allopolyploid species, a resynthesized S0 plant, generated by crossing the F1 hybrid with the somatically doubled S0 plant, was also studied using the same approaches.
- Top of page
- Materials and Methods
- Supporting Information
Resynthesized B. napus accumulates homoeologous rearrangements and losses throughout generations (Gaeta & Pires, 2010), from as early as the first meiosis (Szadkowski et al., 2010). In this study, we compared the genetic instability of resynthesized B. napus generated via two distinct pathways (unreduced gametes vs somatically doubled S0 plants), using plant material produced from the same progenitor genotypes. Our results indicate that the resynthesized plants produced from these two pathways show different patterns of genetic rearrangements. In turn, these rearrangements are likely to change the pace at which new rearrangements are expected to form and accumulate in the following generations.
We clearly showed that homoeologous exchanges occurred more frequently during meiosis of the F1 hybrid than in the corresponding CD.S0 colchicine-doubled resynthesized plants, and the resulting recombined chromosomes were transmitted to the progeny of the F1 hybrid by unreduced FDR-like gametes. Indeed, in > 80% of progenies from the F1 hybrid, markers were lost on the A1 and/or C1 chromosomes and allele dosage showed that nine of 10 of these resulted from duplication of the corresponding homoeologous region. Even though A1 and C1 accumulated the largest number of meiotic-driven intergenomic exchanges in resynthesized plants (Gaeta et al., 2007; Szadkowski et al., 2010), we still observed a 1.5-fold increase in the number of rearrangements in the progeny of the F1 hybrid. This could be directly correlated with the frequent chiasmatic associations observed at meiosis of the F1 hybrid (90% of chromosomes per cell). Such a high level of CO formation in B. rapa × B. oleracea hybrids has been reported previously (Prakash & Hinata, 1980; Tsunoda et al., 1980; Attia & Robbelen, 1986a), but our GISH-like analysis provided more precise insight into allosyndetic (between A and C genomes) and autosyndetic (within A and C genomes) pairs. We confirmed that allosyndesis is prevalent in this F1 hybrid (> 75% of chromosomes per cell were involved in bivalent or multivalent formation), even though autosyndesis was detected. The frequency of A–C associations was much higher than in allotetraploid resynthesized S0 plants (20.8 times higher), which is expected because competition for CO formation between homologous and nonhomologous chromosomes does not occur in haploid plants. All our results demonstrate that, compared with somatically doubled plants, the first meiosis in an interspecific hybrid can drastically increase the number of gametes carrying homoeologous translocations. This is in agreement with results from selfed populations, where destabilization of the rDNA locus in self-pollinated lines from unreduced gametes was higher than in somatically doubled lines (T. Książczyk et al., unpublished). In a study on Lilium polyploids, unreduced gametes from F1 hybrid plants were also seen to generate translocations, whereas colchicine-doubled allopolyploids did not (Xie et al., 2010).
Gaeta & Pires (2010) recently introduced the ‘polyploid ratchet’, which predicts that translocations in heterozygote plants generated by homoeologous COs at meiosis will increase genomic instability. In a ratchet-like manner, successive generations will accumulate translocations leading to infertile plants [this is not to be confused with the polyploidy ratchet of Meyers & Levin (2006) that examines the evolution of ploidy within a genus]. According to the theory of Gaeta & Pires (2010), the initial translocation frequency is crucial for the polyploid ratchet to occur. Accordingly, the unreduced gamete pathway from the F1 hybrid would increase the pace of the polyploid ratchet compared with the somatic doubling pathway.
Nevertheless, the size of homoeologous rearrangements in a plant is also an important factor that modulates the pace at which new translocations will appear. Indeed, pre-existing small heterozygous translocations are less prone to destabilize meiosis than are large translocations (Szadkowski et al., 2010). Large translocations in various positions along chromosomes may generate more instability because of the increased probability of COs between large homogenized regions. In connection with this, we observed smaller rearrangements in the F1 hybrid progeny than those transmitted to the progeny of the somatically doubled CD.S0 plant. In addition, rearrangement distributions along chromosomes varied between the F1 × Bn and CD.S0 × Bn populations. For example, no translocation encompassed the centromeric region of C1 in the F1 progeny, whereas it was lost in 16.4% plants of the CD.S0 progeny. In our study, most viable gametes from the F1 hybrid carried 19 chromosomes and, in this respect, transmitted the complete parental chromosome set for centromeric regions. In the CD.S0 progeny, mis-segregation of chromosomes at meiosis leads to gametes carrying pairs of homologs, whilst lacking one homoeolog. This therefore favors larger rearrangements that encompass centromeres, as described in Szadkowski et al. (2010). Taken together, these results indicate that the unreduced gamete and somatic pathways generate different genetic configurations in their respective progeny for processing further meioses. This contrasting situation generated at both meiosis in the F1 hybrid and in its colchicine-doubled counterpart provides an interesting context for testing the influence of pre-existing rearrangements on the emergence of new rearrangements after one generation.
One fertile plant (UG.S0) from the cross between plants from the two pathways carried some translocations and probably one short complete distal reciprocal translocation between chromosomes A1–C1, confirmed by segregation analysis in the progeny. Meiosis in this plant was somewhat irregular, with around half of meiotic cells carrying A–C bivalents. In its progeny, a large number of plants carried rearrangements that mainly resulted from COs between A1 and C1 (90% of observed cases). These translocations were significantly smaller than those described in CD.S0 progeny. Pre-existing translocations influence the frequency of newly formed rearrangements. Indeed, if only markers outside translocated regions are considered, the rearrangement frequency still increased (1.29 times more rearrangements than in the progeny of the S0 somatically doubled plant). Contrary to the observations of Udall et al. (2005), many of the de novo rearrangements formed during meiosis of UG.S0 did not occur in the regions flanking this pre-existing translocation. In contrast, we saw an increase in translocation frequencies on opposite chromosomal arms where large pre-existing translocations were not detected. Thus, the size of pre-existing rearrangements appears to have a critical impact on the instability of genetic regions in their vicinity during meiosis. The relationship between translocated chromosomes and homoeologous CO formation may therefore be extremely complex.
In synthetic B. napus, the ‘polyploid ratchet’ was shown to favor elimination of one genome, because homoeologous marker losses were more frequent on the C genome than on the A genome, in a process of rearrangement fixation from generation to generation (Gaeta et al., 2007). In our study, rearrangements generated from either the F1 hybrid or colchicine-doubled S0 plants (published in Szadkowski et al., 2010) were found as frequently on either A1 or C1. In the UG.S0 plant, however, the difference in structure at the top of A1 and C1 was associated with a strong bias in its gamete transmission, with an excess of C1 marker losses and concurrent duplication of the A1 region. The origin of the bias may be either a larger pre-existing C1 translocation or accumulation of small C1 losses that were not seen at our marker density; selection against unfit genetic configurations may also be acting and cannot be excluded. Although, clearly, analysis of a single plant cannot be used to rule out the average behavior of resynthesized allotetraploids from F1 unreduced gametes, this shows that one single generation hedges chromosomal composition in gametes and illustrates how fast biases can emerge in favor of elimination of one genome copy.
Altogether, these results suggest that the first step in the allopolyploid formation pathway is crucial. It is characterized by different sizes and frequencies of rearrangements and may certainly have different consequences on the pace and extent of the polyploid ratchet. Based on our findings, further work on the unreduced gamete pathway will be necessary to fully address the question of its impact on genetic stability after several generations, and perhaps its potential to escape the B. napus polyploid ratchet.
The genetic pathways which formed natural cultivated B. napus are unknown, but its multiple origins have been confirmed (Song & Osborn, 1992; Allender & King, 2010; Cifuentes et al., 2010). As seen in this study, the genetic outcomes of polyploid formation pathways are different and the largely underexplored unreduced gamete pathway mimics that seen in natural B. napus. For example, translocations seen in B. napus varieties are generally smaller than those seen in somatically doubled plants, and are rarely found on several chromosomes at a time within natural B. napus varieties (Udall et al., 2005; Howell et al., 2008). Likewise, our cytogenetic characterization of the UG.S0 plant, in which only a small number of large-scale translocations are detected, is reminiscent of the few small changes observed by Howell et al. (2008) in domesticated B. napus lines. Natural B. napus varieties seem to share more features in common with the newly formed polyploids obtained via unreduced gametes than with the synthetics produced by somatic doubling. It would thus be tempting to hypothesize that domesticated B. napus may have evolved from unreduced gametes. However, domesticated B. napus certainly had a small population size when it was first selected and certainly underwent breeding for fertility from the very first generations. It is thus conceivable that extensive genome rearrangements that may have first been produced would then have been selected against or swept out of the populations over time by genetic drift. Future sequencing projects examining structural diversity in B. napus and its parental species, in combination with further studies on resynthesized B. napus through unreduced gametes, are thus required to better understand the formation of natural B. napus.