Nucleolar dominance and different genome behaviors in hybrids and allopolyploids
- First Online:
- Cite this article as:
- Ge, X., Ding, L. & Li, Z. Plant Cell Rep (2013) 32: 1661. doi:10.1007/s00299-013-1475-5
- 682 Views
Many plants are allopolyploids with different nuclear genomes from two or more progenitors, but cytoplasmic genomes typically inherited from the female parent. The importance of this speciation mechanism has stimulated the extensive investigations of genetic consequences of genome mergers in several experimental systems during last 20 years. The dynamic nature of polyploid genomes is recognized, and widespread changes to gene expression are revealed by transcriptomic analysis. These progresses show different stabilities of parental genomes and their unequal contributions to the transcriptome, proteome, and phenotype. We review the results in systems where extensive genetic analyses have been conducted and propose possible mechanisms for biased behavior of parental genomes in allopolyploids, including the role of nucleolar dominance. It is hypothesized that the novel ribosomes with rRNAs from uniparental genome and the ribosomal proteins of biparental origins have some impacts on the biased cellular and genetic behaviors of parental genomes in hybrids and allopolyploids.
KeywordsAllopolyploidGenomic stabilitiesNucleolar dominanceCytogeneticsRibosome
Hierarchies of genomic interaction and evolutionary relationships between separated genomes in hybrids
Hierarchies of genome behavior and interactions in hybrids
B > A, A > Ca; B > A&Cb
U > B > D > A > RcB = Hch > R; H > B; H > HchH & Hch > R; Hch > Sad
A. a > A. te
B > Ca; B > A > Cb; A > Cc
Chromosome fragment loss in A. thalianad
Gene/protein expression pattern
C > Aa (in S0); A > Cb (in S5); A > Cc
D > A&Bd; D > A&Be
A. t > A. af
Bidirectional organ specific or genome-wide expression dominanceg, h
Evolutionary relationship between separated genomes
Million years apart
A–C: 2.5–4.2Myaa; B–A&C: 7.9Myab
H to R, A, B, D 11Mya; R to A, B, D 7Mya; SI to A, B, D 3.5Mya; A to B, D 3Mya; A to D 0.5Myac
A. arenosa & A. lyrata ~ A. thaliana: 5– 6 Myad
A to D: 5–10 Myae
C > B > Aa
R > SI > B > A > U > Db
A. l > A. a > A. tc
A > Dd
Numbers of 45S rDNA loci
B (4) > A (2) = C (2)a
U (4) = A (4) > B (2–4) > D (2) = R (2)bHch (4)c
A. a (12) > A. t (4)d
D (6) > A (5)e
Nucleoli, the sites of 45S rRNA gene transcription (the 5.8S, 18S and 28S rRNA) and ribosome assembly, form at the chromosome loci where tandemly repeated 45S rRNA genes are actively transcribed (Pikaard 2000), also called NORs (nucleolus organizing regions). In many interspecific hybrids and their allopolyploid derivatives, rRNA genes inherited from one parent are transcribed whereas those from the other progenitor are silent, although rRNA-coding sequences are essentially identical in species that are able to inbreed (Chen and Pikaard 1997; Pikaard 2000). This epigenetic phenomenon which was first described in plants about 80 years ago (Navashin 1934) is now best known as ND. ND occurs in plants, insects, amphibians and mammals, and is regulated via RNA interference (RNAi) (Preuss et al. 2008; Tucker et al. 2010), cytosine methylation of DNA, histone modification, and chromatin remodeling factors (Preuss and Pikaard 2007; Volkov et al. 2007). A hierarchy of ND (Brassica nigra, BB > Brassica rapa, AA > Brassica oleracea, CC) has been demonstrated in three allotetraploid Brassica species (Chen and Pikaard 1997; Książczyk et al. 2011), and also in synthetic Brassica allohexaploids (2n = 54, AABBCC) (Fig. 1). The mechanisms allowing the discrimination of parental rDNA and the consequences of that discrimination remain poorly understood (Preuss and Pikaard 2007; Volkov et al. 2007), although there is clearly a genetic basis to the dominance hierarchy given the results from hybrids and also chromosome substitution lines which may alter rDNA activity on chromosomes originating from different genomes. The expression of ND has been shown to be developmentally or genetically regulated in Brassica as well as Arabidopsis, and can also be tissue specific (Chen and Pikaard 1997; Chen et al. 1998; Hasterok and Maluszynska 2000). ND establishment in some hybrids and allopolyploids is realized not on their synthesis, but after several generations, and variable expressions ranging from codominance to dominance are observed in populations of the same allopolyploid species (Joly et al. 2004; Pontes et al. 2003). Interestingly, ND also occurs in interpopulation or interecotype hybrids of the same species (Flowers and Burton 2006; Lewis et al. 2004).
Different genome stabilities at chromosome level in hybrids and allopolyploids
Complete or partial chromosome elimination of one parental genome following interspecific hybridizations is a fairly common phenomenon, for which many causes including parent-specific inactivation of centromeres and spatial separation of genomes during mitosis have been proposed (Jones and Pašakinskienė 2005; Sanei et al. 2011). In interspecific cross between Hordeumvulgare and Hordeum bulbosum, the chromosomes from H. bulbosum may be preferentially eliminated during the early divisions of the zygote (Kasha and Kao 1970). Chromosomes that are destined to be eliminated show suppression of the NORs (Finch and Bennett 1983). In H. vulgare × H. bulbosum hybrids, secondary constrictions at the NOR were visible on chromosomes from the genome occupying the central position of the cell, H. vulgare, but no constrictions were seen on the NOR chromosome from the more peripheral genome, H. bulbosum (Schwarzacher et al. 1992; Anamthawat-Jónsson et al. 1993). In mouse and human somatic cell hybrids, the chromosomes of the dominant species NOR locus will be typically retained, while a variable number of chromosomes from the other species will be lost before the karyotype stabilizes (Miller et al. 1976; Reeder 1985). These examples show the different stabilities of parental genomes at chromosome level in hybrids which are possibly correlated with ND.
The different genome stabilities at chromosome level in natural and synthesized Brassica allopolyploids are revealed in their intergeneric hybridizations with another crucifer Orychophragmus violaceus, and the latter act as an inducer for differential chromosome elimination in stable Brassica species. The study of artificial hybrids can give the information unavailable in natural systems, for stable allopolyploids have been established from the crosses between closely related progenitors that can interbreed naturally and the events at initial stages of formation have disappeared.
In intergeneric sexual hybrids of the six cultivated Brassica species in U-triangle (Fig. 1) with O. violaceus (2n = 24) as the paternal parent, parental genome separation in some metaphases and chromosome elimination lead to cells with different numbers of chromosomes (Fig. 1) (Li et al. 1995, 1998; Li and Heneen 1999; Hua et al. 2006; Li and Ge 2007). Among the mosaic of cells observed in the hybrids’ meristems, only those with entire or partial Brassica complements are competitive during plant growth and form gametes. Thus, the progeny of these hybrids contain all or near complete complements of the Brassica chromosomes, but often lack some or all chromosomes from O. violaceus depending on the cross.
Among hybrids between the three Brassica diploids and O. violaceus, only the one with B. oleracea had the expected chromosome number (2n = 21). Those hybrids with B. rapa and B. nigra maintained all or the majority of the chromosomes from Brassica parents, but lost all or most of the chromosomes from O. violaceus (Li and Heneen 1999; Li and Ge 2007; Liu and Li 2007). The result shows that the chromosomes from O. violaceus are only stably maintained in the hybrid cells with the cytoplasm of B. oleracea. Of the hybrids between three allotetraploids and O. violaceus, those with Brassica carinata showed least variations in chromosome complements or genomic compositions (Li et al. 1998; Hua et al. 2006). In B. carinata cells with partial chromosome complements, more chromosomes from B. nigra were maintained than from B. oleracea (Hua et al. 2006), correlating with B. nigra-origin rDNA loci showing ND over B. oleracea-origin rDNA loci (Fig. 1). Hybrids between O. violaceus and B. napus (Li et al. 1995; Li and Ge 2007) or Brassica juncea (Li et al. 1998) showed a wide range of chromosome numbers, those with B. napus can lose chromosomes of B. oleracea-origin besides O. violaceus-origin (Cheng et al. 2002; Hua and Li 2006). The loss of some C-genome chromosomes from the complement of B. napus is recurrently observed in crosses with other crucifers (Chen et al. 2007; Du et al. 2008; Tu et al. 2010), revealing that A-genome chromosomes are preferentially retained and C-genome chromosomes lost. The B. rapa genome segregated from B. napus (Tu et al. 2010) is now being sequenced as the reference for sequence analysis of B. napus in China. These data will be used to investigate the genetic changes that have occurred subsequent to the formation of B. napus about 10,000 years ago (Cho et al. 2009; Cheung et al. 2009). These results reveal the different genome stabilities in three Brassica allotetraploids, which are consistent with ND hierarchy in these Brassica species.
Though the chromosomes from O. violaceus were largely eliminated in its sexual hybrids with B. napus (Li et al. 1995; Hua and Li 2006), the somatic hybrids of the two species contained the sum of parental chromosomes (Zhao et al. 2008). In the somatic hybrids, backcrossing progenies and addition lines (B. napus genomes plus one of three O. violaceus chromosome pairs carrying NORs) (Zhao et al. 2008), transcripts of rRNA genes from O. violaceus, but not from A genome of B. rapa were detected (Ge et al. 2009). Furthermore, O. violaceus also seems to be phenotypically dominant over B. napus (Li and Ge 2007), as Arabidopsis arenosa was dominant over Arabidopsis thaliana in allotetraploid Arabidopsis suecica (Chen 2007). This adds more example in which the chromosome stability and phenotype of the hybrid are biased to the genome showing ND.
Synthetic Brassica allohexaploids
Artificially synthesized Brassica allohexaploids (2n = 54, AABBCC) using the three diploids and allotetraploids of the U-triangle (Fig. 1) express rRNA genes in a hierarchy of ND (B > A > C) (Ge et al. 2009). ND is established upon allohexaploid synthesis and is independent of the origins of the cytoplasm. To further elucidate the mechanisms involved in parental genome separation in Brassica × Orychophragmus hybrids as described above, these synthetic Brassica allohexaploids are pollinated by O. violaceus and the hybrids produced have only incomplete complements of Brassica chromosomes and no intact O. violaceus chromosomes, but some sequence introgression. Of the Brassica chromosomes, those from A and B genomes are preferentially maintained and those from C genome are lost (Ge et al. 2009). Thus, the pattern of maintenance of Brassica chromosomes in these materials indicates a decreasing order of stability B > A > C which is the same pattern as ND.
Genome-specific genetic changes and gene expressions in polyploids
The complex of six cultivated Brassica species in U-triangle (Fig. 1) is a model system for investigations of polyploidy, especially of polyploid crops. Previous studies suggested that the genomes from two ancestral diploids in natural Brasssica allotetraploids have different stabilities, and that cytoplasm has exerted considerable influence on the evolution of nuclear genomes of alloploids (Prakash et al. 2009). Generally, the A genome in B. juncea has remained mostly intact while B genome has changed considerably, the B genome in B. carinata has unchanged, but C genome considerably altered. In B. napus, both A and C genomes have undergone a similar extent of changes. Some controversial cases are given as to the genomic changes in the same allotetraploids, such as B. juncea, which may be attributable to their different origins and also the aspects observed (Axelsson et al. 2000; Lysak et al. 2005). Recent sequence comparative analysis between homoeologous genome segments of B. napus and its progenitor species shows that the C-genome segments are expanded in size relative to their A-genome counterparts in the majority of the regions studied (Cheung et al. 2009). The sequenced genome segments contain similar amounts of coding sequence, but the expanded C-genome segments show a much increased content of transposon-related and noncoding sequences.
The reciprocal synthetics of Brassica allopolyploids have the advantage of precise known progenitors and also cytoplasm backgrounds of two parents, for it has been confirmed that B. nigra and B. rapa have contributed the cytoplasm to natural B. carinata and B. juncea, respectively. B. rapa was suggested as potential plastid genome donor to B. napus (Allender and King 2010). Studies of resynthesized B. napus have found not only wide novel phenotypic variation (Pires et al. 2004a, b), but also rapid genetic and epigenetic changes caused by meiosis-driven genome reshuffling in allopolyploids (Song et al. 1995; Gaeta et al. 2007; Xu et al. 2009; Szadkowski et al. 2010, 2011; Xiong et al. 2011). In reciprocal synthetics of each polyploid (AA.CC—CC.AA and AA.BB—BB.AA), the frequency of genetic change was influenced by the parental origin of the genomes, which was much higher when the genome with ND was provided by the male parent (Song et al. 1995). The progenies of the synthetics with the B. rapa cytoplasm showed an excess of plants without rearrangements and a lower frequency of plants carrying marker loss on A1 chromosome than the one with B. oleracea cytoplasm. By contrast, no difference was found between marker loss frequencies for C1 chromosome in the progenies of reciprocal synthetics. Conversely, the genetic backgrounds on B. oleracea cytoplasm did not influence the frequency of rearrangements (Szadkowski et al. 2011). Genetic changes in the S5 progenies of ~50 resynthesized B. napus lines (CC.AA) occurred more frequently in the C genome (although the difference was only marginally significant), while cytosine methylation changes occurred more frequently to the A genome (Gaeta et al. 2007). In synthesized Brassica hybrids and allotetraploids from pairwise crosses of three cultivated diploids, the level of intra- and intergenomic pairings varied significantly, depending on the genome combinations, cytoplasm background and/or their interaction, and the pairing variations were more extensive within both A and C genome than B genome (Cui et al. 2012). Similarly, parental genome and cytoplasm also affected the genomic, epigenetic and gene expression changes prior to meiosis in the serial hybrids and allotetraploids (Cui et al. 2013).
It has been noted that when the parental diploid species of allopolyploid have highly differentiated cytoplasm, as in B. juncea and B. carinata, the nuclear genomes contributed by the male parents are considerably altered compared to the nuclear genomes of female parents (Prakash et al. 2009). Higher frequencies of genome change in B. juncea than in B. napus are considered to result from the higher degrees of divergence between the parental diploid genomes (Song et al. 1995), or the nuclear-cytoplasmic incompatibility, because of the more distantly related A and B cytoplasmic genomes and the more incompatible nuclear-cytoplasmic genomes in the AB than in AC and CA polyploids. By reanalysis of the results from Song et al. (1995), we found that the genome with a dominant rDNA locus from the male parent showed higher frequency of genetic change than from the female parent. The artificial allohexaploids (AABBCC) with the cytoplasm and rRNA from B. nigra also show more regular meiotic behavior and fertility (Ge et al. 2009).
In the S5 progenies of ~50 resynthesized B. napus lines, DNA fragment losses were significantly correlated with cDNA-AFLP fragment losses but methylation changes were not correlated, suggesting that genomic rearrangements were largely responsible for loss of cDNA-AFLP markers (Gaeta et al. 2007). In synthetic B. napus, of C-genome-specific gene silencing and DNA methylation alterations is significantly greater than A-genome-specific alterations (Xu et al. 2009). Furthermore, a high proportion (25–38 %) of polypeptides displayed quantitative nonadditive expression patterns, with more than 60 % of proteins being expressed in a manner similar to the paternal parent B. rapa (Albertin et al. 2006). But most of these proteins with a nonadditive pattern had additive transcript levels, suggesting that differential protein regulation is mainly governed by posttranscriptional modifications (Marmagne et al. 2010). The alternative splicing changes of large scale were detected in natural and resynthesized B. napus, and were more common than homeolog silencing (Zhou et al. 2011).
Arabidopsis suecica (2n = 4x = 26) is a natural allotetraploid formed 1.2–1.5 Mya between progenitors most closely related to A. thaliana (2n = 10 or 2n = 4x = 20) and A. arenosa (2n = 16) (Kamm et al. 1995; Koch et al. 2000). Extensive investigations of this natural allotetraploid and synthetic mimics reveal that ND of A. arenosa occurs coincidently with the phenotypic and transcriptomic dominance (Comai et al. 2000; Madlung et al. 2005; Chen 2007). The genome-wide analysis of gene expression reveals global down-regulation of the A. thaliana genome in favor of the A. arenosa genome (Wang et al. 2006). Another allotetraploid between A. thaliana and A. lyrata subp. petraea (2n = 16) was found to express the rDNA loci from A. lyrata and is more phenotypically similar to A. lyrata (Beaulieu et al. 2009). In summary, the phenotype and gene expression of these newly formed Arabidopsis hybrids tends to show the same bias as ND.
Wheat and triticale
Natural and synthetic wheat hybrids at different ploidy levels involving the genera of the Tribe Triticeae including Triticum, Aegilops, Hordeum and Secale are well suited for study of genome changes and rDNA dominance. The allohexaploid bread wheat (T. aestivum) (2n = 6x = 42, AABBDD) has multiple NORs, the most active of which are located on chromosomes 1B and 6B and show the expression hierarchy of major NORs: 1B > 6B, though the 1B NOR carry only 1,300 genes, fewer than the 2,700 genes of 6B NOR (Lacadena et al. 1988). In Triticale, the hybrid typically from a cross of tetraploid wheat and rye (Secale cereale), wheat NORs are expressed and the rye NOR, located on the short arm of chromosome 1R, is suppressed. However, in other hybrids and chromosomal introgression lines involving NOR-bearing chromosomes of Aegilops umbellulata, wheat NORs are suppressed (Lacadena et al. 1988).
In newly synthesized wheats (Aegilops–Triticum) and triticale (wheat × rye), rapid genomic changes were found as a general phenomenon (Ozkan et al. 2001; Ma and Gustafson 2008). The overall examination of the published results reveals that the variation in triticale is significantly higher than that observed in synthetic wheat species complexes and also Brassica (Song et al. 1995; Bento et al. 2011). Furthermore, in hexaploid and octoploid triticales, the wheat and rye parental genomes demonstrated dramatic differences regarding sequence changes (Ma and Gustafson 2008), and the rye genome demonstrated a higher level of changes than the wheat genome. As a whole, the compiled data show that restructuring depends on parental genomes, ploidy level, and sequence type [repetitive, low copy, and (or) coding]; and affects preferentially the larger parental genome (comparing DNA contents per haploid genome) to homogenize parental genome size and stabilize the newly formed polyploid species, independently of their maternal or the paternal status (Bento et al. 2011). Therefore, a correlation between parental genome size and level of genomic rearrangements exists in triticales, which is probably related to the cell cycle duration or epigenetic modification of parental genome, including ND.
The comparison of gene expression profiles in synthetic and natural hexaploid wheat with parental Aegilops tauschii and Triticum turgidum lines indicate rapid initiation of differential expression of homoeologous loci and nonadditive gene expression in T. aestivum (He et al. 2003; Bottley et al. 2006; Bottley and Koebner 2008; Pumphrey et al. 2009). Especially, D-genome homoeoalleles were silenced more frequently than those from the A or B genomes. By using whole-genome shotgun sequencing, the bread wheat genome was further revealed to be highly dynamic, with significant loss of gene family members on polyploidization and domestication, and an abundance of gene fragments (Brenchley et al. 2012). But there was no biased gene loss in any of three genomes at the level of functional categorization. Several classes of genes involved in energy harvesting, metabolism and growth are among expanded gene families that could be associated with crop productivity, interestingly including ribosome proteins.
Whole-chromosome aneuploidy was recently shown to be associated with synthetic allohexaploid wheat lines, independently developed using diverse accessions of tetraploid wheat T. turgidum ssp. durum or ssp. carthlicum (2n = 4x = 28, BBAA) and A. tauschii (2n = 2x = 14, DD)(Zhang et al. 2013). Aneuploidy included an unexpected hidden type, which had a euploid chromosome number of 2n = 42, but with simultaneous loss and gain of nonhomeologous chromosomes. Chromosome loss or gain showed genome and chromosome biases, for B genome showed the most lability for aneuploidy, followed by A, but the newly added D subgenome was largely stable in most of the studied lines. Chromosome loss and gain were also unequal across the 21 homologous chromosome pairs, especially 1B which carried most active NORs showed the highest frequency of aneuploidy in one lineage. The genome and chromosome biases of aneuploidy with the order B > A > D were surprising given that genomes B and A, constituting the natural allotetraploid wheat species (T. turgidum), have coexisted for about 0.5 million years (Huang et al. 2002). But the result was consistent with the previous classic work that showed that whenever the D genome of A. tauschii was a constituent genome of a naturally occurring allopolyploid species in the Aegilops–Triticum complex (including T. aestivum), it functioned as the so-called “pivotal genome” that underwent little alteration in terms of its encoded phenotypic traits over evolutionary time, but those of its coexisting genomes were often significantly modified (Zohary and Feldman 1962). More gene assemblies in T. aestivum landrace Chinese Spring were also classified with high overall precision as D-genome derived (33.8 %) than A- (28.3 %) and B-genome derived (29.2 %) (Brenchley et al. 2012). The existence of “pivotal genome” and Ph genes restricting homoeologous pairing in wheat system made the formation and evolution of bread wheat different from other allopolyploid systems, and also weakened the role of ND for genome stabilization in synthetic lines (Zhang et al. 2013).
The five extant Gossypium allotetraploid species, with A and D genome constitution, including two cotton crops Gossypium hirsutum and G. barbadense, probably evolved from a single polyploidy event about 1–2 Mya (Wendel and Cronn 2003). In synthetic cottons that mimic these allotetraploids, generated from diploids most closely related to the original diploid progenitors, there is almost no genetic change (Liu et al. 2001). This contrasts with what was found in synthetic Brassica and wheat allopolyploids.
Reciprocal gene silencing or biased expression of homeologous gene pairs was found both in natural and synthetic allopolyploids (Adams et al. 2003). Likewise, genome-wide expression dominance was found in natural and synthetic cotton, whereby gene expression was either up- or down-regulated to the level of one of the two parents, the same diploid parent could be either the dominant or the recessive genome depending on the specific genomic combination (Rapp et al. 2009; Flagel and Wendel 2010). Unfortunately, there is little rRNA expression information in cotton, though rDNA from four polyploidy lineages has been homogenized to a D-genome repeat type, whereas in a fifth lineage it homogenized to the A-genome repeat type (Wendel et al. 1995). It remains to be shown if there is any correction between ND, sequence colonization (Wendel et al. 1995; Hanson et al. 1998; Zhao et al. 1998; Hanson et al. 2000; Adams and Wendel 2004) and the gene expression dominance. However, it appeared that the “bias or dominance” is an important mechanism for genomic interaction and gene expression regulation during the process of cotton polyploid evolution. The genome sequence of G. hirsutum revealed many nonreciprocal DNA exchanges between subgenomes, with A-genome genes converted to the D-genome state at more than twice the rate as the reciprocal that may have contributed to phenotypic innovation and/or other emergent properties such as ecological adaptation by polyploids (Paterson et al. 2012).
The genus Nicotiana provides an excellent model group for studies on the consequences of polyploidy, since it consists of c. 70 species, and ~40 % of which are allotetraploids derived from six independent polyploidy events over widely different timeframes (thousands to millions of years) (Clarkson et al. 2005; Leitch and Leitch 2008; Clarkson et al. 2010). Nicotiana tabacum (tobacco) (2n = 48, SSTT) is derived from known ancestors, Nicotiana sylvestris (maternal, S-genome donor), and Nicotiana tomentosiformis (paternal, T-genome donor) within the last 200,000 years. The species has the sum of the rDNA loci expected from the numbers of parents, but has a decrease in rDNA copy number relative to expectation. Furthermore, it has evolved its own distinct gene family, which arose by the reorganization of the parental N. tomentosiformis-inherited units followed by their subsequent amplification and homogenization between rDNA loci. The extent of rDNA sequence homogenization varies among different Nicotiana polyploids. Next generation sequencing reveals that the T genome of N. tabacum has experienced greater sequence loss than the S genome, showing preferential loss of paternally derived repetitive DNAs at a genome-wide level (Renny-Byfield et al. 2011). In N. tabacum, homogenized rDNA units are expressed from both S- and T-genome chromosomes, although most repeats are likely to be located on S-genome chromosomes (the stable genome).
There are allopolyploid species in Tragopogon, Spartina, and Senecio that were formed from diploid progenitors in the last 150 years, affording opportunities to investigate recent and recurrent polyploid evolution (Hegarty et al. 2006; Tate et al. 2006, 2009; Chelaifa et al. 2010). Tragopogon mirus and Tragopogon miscellus are allotetraploids (2n = 24) that formed by hybridization between Tragopogon dubius and Tragopogon porrifolius, and T. dubius and Tragopogon pratensis, respectively. All the T. miscellus populations studied but one and T. mirus have T. dubius as the paternal parent. The concerted evolution has partially homogenized thousands of parental rDNA units typically by reducing the copy numbers of those derived from the T. dubius parent, but repeats of T. dubius origin dominate rDNA transcription in most populations, despite their low copy number (Kovarik et al. 2005; Matyásek et al. 2007), i.e., rDNA units that are genetic losers (copy numbers) are epigenetic winners (high expression). Uniparental reductions of homeologous rRNA gene copies occurred in both synthetic and natural populations of Tragopogon allopolyploids (Malinska et al. 2010). Little or no deviation from additivity for chromosome numbers, satellite sequences, and genome size has been reported for the two allotetraploids (Pires et al. 2004b; Tate et al. 2006), but extensive chromosomal polymorphisms between individuals and populations are detected, including variable sizes and expression patterns of individual rDNA loci (Lim et al. 2008). A marked preferential loss of homoeologues to one parent is not found in the two allotetraploids (Buggs et al. 2009). They predominantly show expression of both homoeologs in different tissues, but loss or silencing of T. porrifolius homoeologs are more common than loss or silencing of T. dubius homoeologs in T. mirus, while silencing of the T. dubius homoeolog is more common in T. miscellus (Buggs et al. 2010a, b). However, the chromosomes from T. dubius are recently found to be gained more often, while those from T. pratensis to be lost more often among five out of six natural T. miscellus populations (Chester et al. 2012). This bias against chromosomes of T. pratensis origin appears irrespective of parentage. The contrast between the genomic and chromosomal investigations of this allopolyploid suggests that chromosome gains or losses alone cannot explain the gene copy number bias, pointing to other mechanisms, such as small nonreciprocal exchange or deletions which may be occurring in T. miscellus (Tate et al. 2009; Chester et al. 2012; Buggs et al. 2009, 2010a, b, 2012).
The genus Spartina is one of the few systems in which both the F1 hybrid and the resulting allopolyploid are alive in natural conditions. The homoploid F1 hybrid Spartina × townsendii between Spartina alterniflora as female parent and Spartina maritima gave rise to dodecaploid Spartina anglica (2n = 122–124), which exhibits additivity of the parental genomes (Ainouche et al. 2004), while the maternal genome from S.alterniflora seems to be more prone to genetic changes (Salmon et al. 2005; Parisod et al. 2009). By using Agilent rice oligomicroarrays (Chelaifa et al. 2010), comparisons of the transcriptomic patterns in the leaves of the parents, the hybrid and the allopolyploid illustrate the expression dominance of the maternal parent in the F1 hybrid, and the balancing effect of genome duplication with regard to expression divergence with the parents.
Senecio cambrensis (2n = 6x = 60) between Senecio vulgaris (2n = 4x = 40) as the maternal parent and Senecio squalidus (2n = 2x = 20) has been formed via chromosome doubling in the sterile triploid hybrid Senecio x baxteri. Microarray analysis showed that the expression in S. x baxteri is largely skewed towards a level similar to that in the lower expressing parent S. vulgaris, while the opposite trend is observed in S. cambrensis (Hegarty and Hiscock 2007). The greatest changes in gene expression relative to the parents (‘‘transcriptome shock’’) appeared to be associated with the hybridization step to form S. x baxteri, which further showed that the polyploidization had an ameliorating effect on altered patterns of gene expression experienced by S. x baxteri (Hegarty et al. 2006). Importantly, this altered pattern of gene expression, apparent in first-generation allopolyploids, was preserved in four successive generations of the synthetic allopolyploids and in wild S. cambrensis (Hegarty et al. 2006). It should be interesting to reveal the reasons for the expression difference between the hybrid and allopolyploid.
To summarize, the genomic expression dominance in natural allopolyploids takes two forms, as typified by (1) Arabidopsissueccica, where the allotetraploid has global down-regulation of the genes from the A. thaliana-derived genome in favor of the dominant A. arenosa-derived genome (Wang et al. 2006). (2) Gossypium allopolyploids, where the expression state of the D genome is significantly more often than the A genome does, though this massive expression dominance was approximately equal with respect to direction (up- or down-regulation) (Rapp et al. 2009; Flagel and Wendel 2010). Such an expression dominance is not so obvious in Brassica and Triticum allopolyploids that formed within 10,000 years. In two recently formed Tragopogon polyploids and in S. cambrensis, one parental genome predominates over the other in terms of global pattern of expression and gene retention (Soltis et al. 2012). These results indicate that the genomic dominance is established over time, and its pattern and degree is influenced by the genome combination and ploidy level.
Prospects and conclusions
Though the nucleocytoplasmic interaction hypothesis (Gill 1991) predicted that the nucleocytoplasmic incompatibilities in a hybrid and polyploid may have some adverse effects of the plant development, it does not specify the functioning organelle for the different genomic stabilities. There are also no reports that link mitochondria and chloroplasts with nuclear genome stability. What are the molecular and ultrastructural mechanisms behind the different parental genome behaviors in interspecific hybrids and allopolyploids?
Since rRNAs are involved in translating all mRNA to proteins or peptides, rRNA and rRNA genes potentially also influence aspects of cell behavior (Moss and Stefanovsky 2002). Many other diverse roles have also been suggested for rRNA, including regulation of aspects of the cell cycle, senescence, and transport (Cockell and Gasser 1999; Olson et al. 2000; Guarente 2000; Visintin and Amon 2000; Sirri et al. 2008). The rDNA and nucleolus have played a prominent role in evolving theories of aging, metabolism, cell differentiation, cell cycle control, cancer progression, and gene regulation (Paredes and Maggert 2009). Although most rDNA copies are dispensable for normal plant development (Ide et al. 2010), low numbers of rDNA repeats have been implicated in the release of heterochromatin-induced gene silencing (Paredes and Maggert 2009), in reduced rDNA condensation and cohesion (Kobayashi and Ganley 2005), and increased sensitivity to DNA damage (Mozgova et al. 2010). These results support a theory that redundant copies of rRNA genes play a role in maintaining genome stability (Kobayashi 2008). We speculate that the dominant rRNA genes and the related nucleolus may also affect the cellular behavior of interspecific hybrids and allopolyploids showing ND, and even preferentially stabilize the genome carrying the expressed rDNA loci.
Plant ribosome is composed of four rRNAs, 5.8S, 18S and 28S rRNA which are transcribed from 45S rDNA loci at NORs, and 5S rRNA from 5S rDNA loci at other chromosomal regions, but of many more ribosomal proteins (RPs). There is an evidence that variations in rRNA contribute to ribosome heterogeneity, as rRNAs are transcribed from multigene families (Mauro and Edelman 2002). In the model plant A. thaliana, the ribosome contains 80 distinct RPs encoded by a total of 249 genes (Chang et al. 2005). Therefore, more paralogous genes for RPs are available in allopolyploids. Genetic variants of RPs and their post-translation modification contributed heterogeneity in RPs in one species (Carroll et al. 2008). On the basis of ribosome heterogeneity and the fact that eukaryotic ribosomes differ structurally in different cell types or during different stages of development, the ribosome filter hypothesis proposes that specific mRNA·rRNA and mRNA·ribosomal protein interactions at sites on the ribosomal subunits are important in controlling translation (Mauro and Edelman 2002). Recent studies indicate that the RP gene mutants have a regulatory role in the development of plant (Byrne 2009) and animal (Kondrashov et al. 2011; Topisirovic and Sonenberg 2011). Potentially, functional divergence and expression networks among RP gene paralogs in B. napus are actively involved in differentiation, development, and/or tissue-specific processes (Whittle and Krochko 2009). Therefore, the heterogeneity in ribosome composition resulting from differential expression and post-translational modifications of RPs, rRNA diversity and the activity of ribosome-associated factors may generate ‘specialized ribosomes’ that confer a regulatory control in gene expression (Xue and Barna 2012).
Given the correlation between ND and the biased genome stability and expression in interspecific hybrids and allopolyploids, we speculate that the ribosomes with rRNAs transcribed from the rDNA loci on the dominant genome likely have some impacts on their cellular and genetic behaviors, which potentially results in the genomic expression dominance. As to RPs, the ribosomes probably contain the mixed RPs of both parents, some determined by the genes of one parent and others by another, while there is still no report for the expressions of RPs in hybrids and derived allopolyploids. The types of RPs from two parents possibly vary in different physiological processes or organs, which may contribute to reciprocal silencing or biased expression of homeologous gene pairs in different organs (Adams et al. 2003; Rapp et al. 2009), or to the complex and variable nature of allopolyploidy performance (Chen 2010). As parental discrimination of RPs is unknown at present, the comparison of ribosomal and nucleolar proteome changes between the parents and allopolyploids will elucidate the parental gene expressions for RPs and then the translation control by ribosomes, though this is a vast task for the dynamic nature of the nucleolus (Andersen et al. 2005). Importantly, RPs were among the over-represented categories in expanded gene families in common wheat (Brenchley et al. 2012).
Notably there is a long interval between the discovery of certain cytological phenomena and the elucidation of the molecular mechanisms in hybrids and allopolyploids, such as chromosome elimination (Kasha and Kao 1970; Sanei et al. 2011) and genetic control of homoeologous pairing in wheat (Riley and Chapman 1958; Griffiths et al. 2006). Although the studies of different systems give the evidence which argues for or against the role of ND in affecting the genome behavior, we believe that the way to exploit the molecular links between ND and the bias of parental genome behavior will be long and hard but worth going.
We thank Prof. J. S. (Pat) Heslop-Harrison, University of Leicester and Prof. Andrew R. Leitch, Queen Mary College, University of London, UK, and three anonymous reviewers for their constructive comments to revise the manuscript. This work was supported by the Grants from Natural Science Foundation (30900903, 31071451, 31171583) and ‘973’ project (2011CB109305) of China.