The love life of a rose. A commentary on: ‘Asymmetrical canina meiosis is accompanied by the expansion of a pericentric satellite in non-recombining univalent chromosomes’

Abstract

This article comments on:

Jana Lunerová, Veit Herklotz, Melanie Laudien, Radka Vozárová, Marco Groth, Aleš Kovařík and Christiane M. Ritz, Asymmetrical canina meiosis is accompanied by the expansion of a pericentromeric satellite in non-recombining univalent chromosomes in the genus Rosa, Annals of Botany, Volume 125, Issue 7, 4 June 2020, Pages 1025–1038, https://doi.org/10.1093/aob/mcaa028

The study of Arabidopsis has reshaped our understanding of plant science and may lead some to question whether the axiom ‘exceptions help to prove the rule’ still holds true for modern science. For those in doubt, the study by Lunerová et al. (2020) in this issue provides ample reassurance.

There are few more curious biological phenomena than the sexual reproduction system of dogroses (Rosa sect. Caninae). Dogroses are mostly uneven polyploids (usually pentaploid: 2n = 5x = 35) but reproduce by seed. The need for meiotic pairing means it is unusual for uneven polyploids to set seed reliably. Those species that do so often deploy some form of apomixis to circumvent female meiosis. Dogroses provide a rare exception.

A century ago, Täckolm (1920) first noted that dogroses exhibit a strange form of unbalanced meiosis which generates tetraploid eggs but haploid pollen. Fusion of these asymmetrical gametes yields pentaploid offspring (Fig. 1). Chromosomes belonging to two genome sets pair normally during both male and female meiosis. Chromosomes of the remaining three sets fail to pair during male meiosis and are lost prior to pollen formation. In female meiosis, however, all univalents are retained and co-transmitted into a tetraploid egg cell (Fig. 1). This idiosyncratic reproductive system has importance far beyond its obvious strangeness.

Fig. 1.

Schematic illustrating chromosomal inheritance within the pentaploid dogrose Rosa canina (2n-5x-35). Somatic cells in the parent plant includes five genomes containing seven chromosomes. Two of these genomes (coloured in blue) pair to form bivalents during meiosis whereas the remaining three genomes (coloured orange, green and red) fail to pair and form univalents. During male meiosis, chromosomes of the three univalent-forming genomes are lost to produce a haploid pollen grain. In female meiosis, one of the two pairing genomes is passed to the egg cell but all chromosomes of the univalent-forming genomes are also transmitted to yield a tetraploid egg cell. Fusion of egg and sperm results in a zygote with all five genomes.

The presence in one organism of two genomes that undergo conventional meiotic recombination and three that do not provides scope to address some fundamental evolutionary questions. Among these, most intense interest has focused on comparing the genetic variability and expression profiles of the recombining (bivalent-forming) genomes with those of the non-recombining (univalent-forming) genomes.

Lim et al. (2005) contrasted meiotic pairing in pollen mother cells of pentaploid Rosa canina with a gynogenetic polyhaploid (2n = 4x = 28) formed from an unfertilized egg cell of the same species (using irradiated pollen). Fluorescence in situ hybridization (FISH) using 18–26S and 5S probes allowed differentiation of six chromosome types on the basis of signal strength and position. Meiotic pairing in the pentaploid included bivalents between chromosome pairs sharing the same signal profile. The polyhaploid largely failed to pair, suggesting the pentaploid has only two truly homologous genomes.

The authors noted that the 21 univalent-forming chromosomes routinely fail to pair and so do not recombine. They predicted that ‘the long-term evolutionary consequence for the univalents is likely to be genetic degradation through accumulated mutational change as in the mammalian Y chromosome’. The absence of clear structural differences between pairing and non-pairing chromosomes was interpreted as suggesting a recent origin for the system. Kovarik et al. (2008) later compared cloned rDNA sequences in three dogroses (R. canina, R. rubiginosa and R. dumalis). They found six major rDNA variant families in each species (denoted α, β, γ, δ, ε and ω). Pollen only contains recombining genomes. Significantly, internal transcribed spacers (ITS) from pollen samples included more β sequence families but fewer γ and δ sequences than somatic tissues. They thereby provided clear evidence of divergence between genome types. Khaitova et al. (2010) subsequently found rDNA from recombining genomes dominated expression in all five pentaploid dogroses studied. This complied with the predicted functional degradation of the non-recombining genomes. However, a more complex picture emerged from later works. Ritz et al. (2011) found little evidence of differential expression in the expression of nuclear ribosomal ITS or of the single-copy gene cGAPDH. They even reported that another gene (LEAFY) showed increased expression from non-recombining genomes. In a similar vein, Vogt et al. (2015) failed to uncover significant differences in either genome-wide DNA methylation or expression levels between the recombining and non-recombining genomes. More recently, Herklotz et al. (2018) studied the expression and composition of rDNA in two dogrose species (R. canina and R. inodora) and in species that represent the reciprocal crosses between them (R. dumalis and R. agrestis). They found ‘canina’ ribotypes from recombining genomes of both parent species invariably dominated expression, again consistent with rDNA silencing of non-recombining genomes. Thus, overall, there seems patchy evidence for functional degradation of the non-recombining genomes.

The study by Herklotz et al. (2018) also provided some unexpected findings based on cytological comparisons between their hybrids. In R. canina, 18S and 5S FISH signals co-located on one bivalent between the recombining genomes but the signals were on different chromosomes in the non-recombining genomes. This arrangement was in contrast to R. inodora, where the 18S and 5S loci occurred on separate bivalents but were co-located on some non-recombining chromosomes. This was also true in R. dumalis and implies ‘canina type’ chromosomes (or at least the one carrying the nuclear organizer region) may not pair at meiosis in all species. The authors therefore hypothesized that the two major lineages of dogroses originated from reciprocal crossing between Rubiginosa and Caninae ancestors, followed by an independent origin of the unbalanced meiosis system, with either Rubiginosa-type or Canina-type chromosomes forming the recombining genomes. Like so many other questions, further progress relied partly on improved discrimination between recombining and non-recombining karyotypes.

The study by Lunerová et al. (2020) provides the first survey for satellite regions in dogroses and helps to directly address this need. The discovery of the CANR4 satellite provides a useful platform for further study but also yielded new insights itself. Southern blotting and sequence cluster analysis revealed the satellite to be Rosa-specific, ubiquitous in the genus although variable between species and most abundant in dogroses. There are two major CANR4 clades, with most species containing variants of both. Clade 1 is most abundant but is also least variable in dogroses. FISH also revealed that CANR4 signal is strictly pericentromeric. In R. canina, CANR4 was detected in two of the seven bivalents of recombining genomes but also in eight of 15 univalents from the non-recombining genomes. Thus, when combined with 18S and 5S probes, CANR4 greatly enhances the capability to differentiate chromosomes and genome types.

Increased abundance of CANR4 in the non-recombining genomes is of particular interest. The authors put forward two possible explanations. First, that the observed patterning was fixed at the time of speciation. In this scenario, divergence between genomes simply reflects differences between progenitor species. The main weakness in this supposition lay in the apparent absence of extant diploid dogroses with a sufficiently high abundance of CANR4. The authors therefore favoured the premise that CANR4 abundance expanded in the non-recombining chromosomes after hybridization, as has been reported in apomictic systems (Belyayev et al., 2018). Lunerová et al. (2020) speculated that such expansion could occur through mitotic sister chromatid exchange or transposable element activity. This aspect clearly requires further study but, whatever the mechanism, the work confirms that considerable structural divergence has occurred between recombining and non-recombining chromosomes since speciation.

Outside those working on dogroses, the prospect of a deeper understanding of the evolutionary importance of meiotic recombination is likely to appeal to many. My own curiosity gravitates towards the co-migration of the unpaired univalents during female meiosis. Female meiosis is notoriously difficult to study but the presence of the CANR4 probe greatly enhances our ability to track individual chromosomes and so provides grounds for cautious optimism.