Sequence of events leading to near-complete genome turnover in allopolyploid Nicotiana within five million years


Author for correspondence: A. R. Leitch Tel: +44 207882 5294 Fax: +44 208983 0973 Email:


  • • Analyses of selected bacterial artificial chromosomes (BACs) clones suggest that the retrotransposon component of angiosperm genomes can be amplified or deleted, leading to genome turnover. Here, Nicotiana allopolyploids were used to characterize the nature of sequence turnover across the whole genome in allopolyploids known to be of different ages.
  • • Using molecular-clock analyses, the likely age of Nicotiana allopolyploids was estimated. Genomic in situ hybridization (GISH) and tandem repeat characterization were used to determine how the parental genomic compartments of these allopolyploids have diverged over time.
  • • Paternal genome sequence losses, retroelement activity and intergenomic translocation have been reported in early Nicotiana tabacum evolution (up to 200 000 yr divergence). Here it is shown that within 1 million years of allopolyploid divergence there is considerable exchange of repeats between parental chromosome sets. After c. 5 million years of divergence GISH fails.
  • • This GISH failure may represent near-complete genome turnover, probably involving the replacement of nongenic sequences with new, or previously rare sequence types, all occurring within a conserved karyotype structure. This mode of evolution may influence or be influenced by long-term diploidization processes that characterize angiosperm polyploidy–diploid evolutionary cycles.


Studies on genome evolution using DNA sequence-based analyses are limited by the number of taxa for which extensive sequence data are available and by the scale of the task (number of base pairs to be characterized). Consequently, whole-genomic characteristics are typically inferred from the detailed analysis of subregions of the genome. This approach has been informative, revealing for example that the replacement time for long-terminal repeats (LTR) retroelements in Oryza is < 6 million years (Myr; Ma et al. 2004). However when similar analyses were focused on the centromere region of two species of rice (Ozyza japonica and Ozyza indica), important differences emerged. At the centromeric region, LTR retrotransposon accumulation was greater, and centromere sequence divergence involved homogenization and slower rates of deletion (Ma & Bennetzen, 2006). Therefore high-resolution sequence analysis of subregions is vulnerable to some misinterpretation when considering the whole genome since different subregions and/or sequences can evolve differently.

Genomic comparisons of allopolyploid species and diploids most closely related to the diploid progenitors provide an alternative method to study genome divergence. These comparisons reveal a variety of events associated with allopolyploid divergence, including, for example, the presence or absence of chromosomal translocations, satellite repeat divergence, and rDNA silencing, epigenetics and homogenization (for reviews see Adams & Wendel, 2005; Soltis et al., 2004; Comai, 2005). Synthetic allopolyploids that mimic the natural species can be made, and studies of these also reveal much variation in early generations, for example much genetic change in allopolyploids of Nicotiana (Skalicka et al., 2005), Triticum (Ozkan et al., 2001) and Brassica (Song et al., 1995) but not in Gossypium (Liu et al., 2001). Allopolyploids are ideally suited for the study of genome divergence because there is a definable point in time when the allopolyploid species formed and from which genome divergence can be measured. The often contradictory data reported may arise because studies are targeted at allopolyploids of different genera and ages. We present here an analysis of allopolyploidy genome divergence in species of genus Nicotiana by comparing four polyploids that formed over widely different yet known time-frames. In Nicotiana there are synthetic allopolyploids to study early genetic change (Skalicka et al., 2003, 2005; Lim et al., 2006b) and natural allopolyploid species of different ages and with extant species closely related to the diploid progenitors for comparisons (Chase et al., 2003; Clarkson et al., 2004).

Here, insights into global genome divergence are described by: examining over what evolutionary time-scales genomic in situ hybridization (GISH) remains effective and; analysing rates of replacement of tandem repeats. This is done by comparing GISH patterns and repeat sequence distribution in Nicotiana allopolyploids ranging over timescales from 10°yr to 106–7 yr. DNA sequences of the nuclear gene glutamine synthase were used to identify the likely diploid species that gave rise to the allopolyploids (Clarkson, 2006) and plastid DNA sequences to indicate the maternal parents (Clarkson et al., 2004). Nicotiana phylogenetic trees were calibrated using divergence of endemic species from their mainland relative on two sets of volcanic islands of known geological ages (Clarkson et al., 2005).

Patterns of genome divergence were compared in four allopolyploids (all 2n = 4x = 48) of various ages: (1) synthetic Nicotiana tabacum in its first generation (TR1B), made by hybridizing Nicotiana sylvestris × Nicotiana tomentosiformis (Lim et al., 2006b); (2) N. tabacum that is < 200 000 yr old, derived from the same progenitor species used to make the synthetic tobacco (Clarkson et al., 2005); (3) Nicotiana quadrivalvis estimated to be c. 1 Myr old and with likely diploid progenitors related to ♀Nicotiana obtusifolia (2n = 2x= 24) × ♂Nicotiana attenuata (2n = 2x = 24); (4) Nicotiana nesophila that formed c. 4.5 Myr ago (Clarkson et al., 2005) with likely progenitors related to ♀N. sylvestris × ♂N. obtusifolia (Clarkson et al., 2005). In these polyploids, we compare the fate of genomic DNA over nearly 5 Myr of divergence (six orders of magnitude).

Materials and Methods

Plant material

Plant material used was: (1) Synthetic N. tabacum L. (tobacco) TR1B (Lim et al., 2006b) prepared from the cross F ♀N. sylvestris‘Seita’ (Institut du Tabac, Bergerac, France) × ♂N. tomentosiformis NIC 479/84 (IPG, Gatersleben, Germany) and chromosomally doubled with oryzalin via tissue culture; (2) N. tabacum acc. 095-55 (KW1, Royal Botanic Gardens, Kew, UK); (3) N. quadrivalvis Prush (TW18, USDA, North Carolina State University, USA, and NIC 904750042, The Botanic Gardens of Radboud University, Nijmegen, the Netherlands); (4) N. nesophila I. M. Johnston (NIC 974750097, The Botanic Gardens of Radboud University, Nijmegen, the Netherlands); (5) N. obtusifolia M. Martens & Galeotti (NIC 84750156, The Botanic Gardens of Radboud University, Nijmegen, the Netherlands; and; (6) N. obtusifolia USDA TW98, called Nicotiana palmeri (North Carolina State University, USA). Nicotiana palmeri is a synonym of N. obtusifolia (Wells, 1960).

Isolation and characterization of tandem repeats

Tandem repeats were isolated using modifications of previous methods (Lim et al., 2005). Briefly, genomic DNA from N. nesophila and N. obtusifolia was digested with VspI or TaqI, and the size-fractionated products and bands of c. 180 bp were isolated, ligated into pZeroIIkan vector (Invitrogen, Carlsbad, CA, USA) and transformed into Escherichia coli DH5α and EZ cells, respectively. The clones were screened with 32P-labelled N. nesophila and N. obtusifolia genomic DNA. Clones showing strongest hybridization signals were sequenced and these sequences were aligned using wisconsin gcg and clustal w software.

Southern hybridization was conducted as previously described (Lim et al., 2005): briefly, 1–2 µg of DNA were digested with excess restriction enzymes (2 × 5 units) and the products loaded on 0.9% (w : v) agarose gels. The electrophoretically size-separated fragments were alkali-blotted onto nylon membranes (Hybond XL; GE Healthcare Ltd, Little Chalfont, UK) and hybridized with 32P-labeled DNA probes (DekaLabel kit; MBI, Fermentas, Vilnius, Lithuania). After washing under high stringency, bands were visualized with a PhosphorImager (STORM; Molecular Dynamics, Sunnyvale, CA, USA).

Fluorescent in situ hybridization (FISH) and GISH

Methods were as described previously (Lim et al., 2005). Genomic DNA for GISH was isolated using Qiagen kits (Qiagen Ltd, Crawley, UK). ‘Nick translation’ was used to label all probes with digoxigenin 11-dUTP or biotin 16-dUTP. Slides were denatured in 70% (v : v) formamide in 2 × standard saline citrate (SSC) (0.3 m sodium chloride, 0.03 m sodium citrate) at 70°C for 2 min. Probe hybridization occurred in a mixture containing 4 µg ml−1 labelled probes and 50% (v : v) formamide, 10% (w : v) dextran sulphate, 0.1% (w/v) sodium dodecyl sulphate in 2 × SSC at 37°C for 20 h. After hybridization the slides were washed in 20% (v : v) formamide in 0.1 × SSC at 42°C to give an estimated hybridization stringency of 80–85%. Sites of probe hybridization were detected using 20 µg ml−1 fluorescein-conjugated antidigoxigenin IgG (Roche Pharmaceuticals, Lewis, UK) and 5 µg ml−1 Cy3 conjugated avidin (Amersham Pharmacia Biotech, Cambridge, UK) in 4 × SSC containing 0.2% (v : v) Tween 20 and 5% (w : v) bovine serum albumin. Chromosome counterstaining was with 2 µg ml−1 DAPI (4′,6-diamidino-2-phenylindole) in 4 × SSC, mounted in Vectashield medium (Vector Laboratories, Burlingame, CA, USA). Material was examined using a Leica DMRA2 epifluorescent microscope fitted with an Orca ER and Open Laboratory software (Improvision, Coventry, UK). All images were processed using Adobe Photoshop and treated for colour contrast and brightness uniformly.


Synthetic N. tabacum (1st generation)

As expected, on conducting GISH on synthetic tobacco TR1B, 24 chromosomes were labelled with N. tomentosiformis genomic DNA (biotin-labelled-Cy3 detected, red fluorescence) and 24 chromosomes were labelled with N. sylvestris genomic DNA (digoxigenin-labelled-FITC detected, green fluorescence) (Lim et al., 2006b). The number of rDNA sites was the sum of that found in the diploids (i.e. one site from N. tomentosiformis and three from N. sylvestris). The GISH probes from both parents labelled each rDNA locus in TR1B (Fig. 1a, yellow). No evidence was found for intermixing of genomic DNA, including tandem or dispersed repeats, intergenomic translocations, or any differences in the fluorescent intensity of the probes to each set of parental chromosomes. All these data point to the synthetic tobacco being the genomic sum of its parents, as would be expected for a newly synthesized allotetraploid.

Figure 1.

Fluorescent in situ hybridization (FISH) to root-tip metaphase spreads from Nicotiana species: biotin-labelled cy3-detected probes (orange-red fluorescence), digoxigenin-labelled fluorescein isothiocyanate (FITC)-detected probes (yellow-green fluorescence) and in (d–g) only, 4,6-diamidino-2-phenylindole (DAPI) for DNA (blue fluorescence). Colours on Figs represent the outcome following merging of signals. We compared genomic in situ hybridization (GISH) patterns in four allopolyploids (all 2n = 4x = 48) of known and estimated ages, they were as follows. (a) Synthetic N. tabacum L. (tobacco), at S0 generation derived from ♀N. sylvestris (2n = 2x = 24) × ♂N. tomentosiformis (2n = 2x = 24) (Lim et al., 2006b) probed with N. sylvestris (green) and N. tomentosiformis (red) genomic DNA. There is clear parental genome distinction. (b) Wild N. tabacum acc. 095-55 with parents as in (a) (Murad et al., 2002) and probed as in (a); arrows show recombinant chromosomes. There is clear parental genome distinction. (c) Nicotiana quadrivalvis, with progenitor parents ♀N. obtusifolia (2n = 2x = 24) × ♂N. attentuata (2n = 2x = 24) probed with N. obtusifolia NIC 84750156 (green) and N. attenuata (red) genomic DNA. Subtelomeric sequences, almost certainly of tandem repeat nature and predominantly labelled yellow from N. obtusifolia genomic DNA, are found on N. attenuata-origin chromosomes. Both genomic DNA probes label both sets of chromosomes, indicating considerable exchange of parental DNA has occurred. (d) Nicotiana nesophila with progenitor parents ♀N. sylvestris × ♂N. obtusifolia (Clarkson et al., 2005) probed with N. sylvestris (green) and N. obtusifolia (red) genomic DNA. Yellow signals are derived by labelling with both probes, notably at rDNA loci. (e) Nicotiana obtusifolia NIC 84750156 probed with NPAL. (f) Nicotiana sylvestris probed with HRS60. (g) Nicotiana nesophila probed NNE. Bar, 10 µm.

Natural N. tabacum (< 200 000 yr old)

A result that was superficially similar to synthetic tobacco was observed when conducting GISH on N. tabacum. The 24 chromosomes that label predominately with labelled N. tomentosiformis genomic DNA (orange) and 24 chromosomes that label predominantly with labelled N. sylvestris genomic DNA (greenish yellow) can readily be distinguished (Fig. 1b). However, there are clearly ‘hybrid’ chromosomes (see arrows) that bear intergenomic translocations, some of which are cultivar-specific and others that may be fixed in this species (Lim et al., 2004). Unlike synthetic tobacco, the N. tomentosiformis-labelled chromosomes were less intensively stained than those labelled with N. sylvestris genomic DNA, as previously observed (Papp et al., 1996). In addition, there is some cross-hybridization of probes between parental genomes. This resulted in the genomes appearing yellow-green (instead of green) and orange (instead of red; compare colours in Fig. 1a,b).

Nicotiana quadrivalvis (c. 1.0 Myr old)

Using GISH it was possible to identify the parental genomes of N. quadrivalvis using digoxigenin-labelled N. obtusifolia genomic DNA (yellow-green fluorescence) and biotin-labelled N. attenuata genomic DNA (orange fluorescence; Fig. 1c). However, considerable cross-hybridization of the genomic DNA was observed between chromosome sets, resulting in a much reduced colour distinction compared with that observed for natural or synthetic tobacco. In addition, most chromosomes carried terminal sequences that showed greater similarity with the N. obtusifolia genomic DNA resulting in yellow signals at subtelomeric/telomeric locations. The sequences at this location in Nicotiana are typically tandem repetitive sequences (Lim et al., 2006a).

Nicotiana nesophila (c. 4.5 Myr old)

As reported previously, GISH for N. nesophila failed when genomic DNA probes from the likely progenitor species, N. sylvestris (green) and N. obtusifolia (red) were used (Clarkson et al., 2005). However, the N. obtusifolia genomic DNA probe did label one locus on a pair of chromosomes, and the N. sylvestris genomic DNA probe gave scattered weak signal across most chromosomes. Both probes labelled rDNA loci yellow (Fig. 1d). These data indicate significant divergence of repetitive DNA in N. nesophila relative to both progenitor diploid species.

Tandem repeat divergence in N. nesophila

To further evaluate why GISH failed to N. nesophila the structure and organization of tandem repeats in N. nesophila and the diploid progenitor species were studied. A novel tandem repeat was isolated from the allopolyploid N. nesophila, NNE (GenBank accession number DQ536093, NNEVSP in gene bank) and another from the diploid progenitor N. obtusifolia TW98, NPAL (accession number DQ536094). These were analysed in conjunction with a subtelomeric tandem repeat previously identified in the other diploid progenitor N. sylvestris (HRS60; (Koukalova et al., 1993; Lim et al., 2000)). Interspecies similarities were calculated from consensus sequences built from alignment of at least four independent clones for each species. All repeats shared > 62% sequence identity, and similarity of individual clones within each species was even higher (89–100%).

NPAL is a tandemly organized 180-bp repetitive sequence in N. obtusifolia (close to the paternal genome donor of N. nesophila), but it is absent in N. sylvestris (progenitor of maternal genome) and N. nesophila (Fig. 2). Using FISH it was shown that the sequence is subtelomeric on most chromosomes of N. nesophila (Fig. 1e). HRS60 is a tandemly organized 180-bp repeat that hybridizes primarily to subtelomeric locations in N. sylvestris (Fig. 1f) (Koukalova et al., 1993; Lim et al., 2000). It is absent in N. obtusifolia but occurs in low numbers in N. nesophila (Fig. 2). The tandem repeat NNE, isolated from N. nesophila, is a 180-bp tandem repeat that is predominantly subtelomeric in location in N. nesophila (Fig. 1g) but is absent in the diploid progenitor species (Fig. 2).

Figure 2.

Southern hybridization of genomic DNA digested with restriction enzymes and probed for tandem repeats NNE, HRS60 and NPAL. For detection of NNE repeats, genomic DNA was digested with VspI; for HRS60 and NPAL, BstNI was used.


Nicotiana allopolyploid genome divergence over time

Plant allopolyploid genomes can be used to measure genome dynamism in Nicotiana over time. Nicotiana allopolyploids are ideally suited for this study since it is possible to estimate their likely age of formation from sequence data using molecular clock methods (Clarkson et al., 2005). With allopolyploids there is a fixed point in time when two divergent parental genomes are united into a common nucleus. Differences between parental genomes provide opportunities to determine the rates of homogenization and genome turnover in plants.

In young Nicotiana allopolyploids, GISH clearly discriminates between the two sets of chromosomes. In the first generation of synthetic tobacco, 24 chromosomes derived from each parent were clearly distinguishable, as reported previously (Fig. 1a) (Lim et al., 2006b). There is no evidence from GISH results of any genetic events stimulated by allopolyploidy in this material. However, it is known from studies of third-generation synthetic tobacco (made by Burk, 1973) that there is genetic change even in young synthetic tobacco including: (1) gain of ribosomal DNA (rDNA) loci and distinctive rDNA units (Skalicka et al., 2003), a situation also observed in synthetic Arabidopsis allotetraploids; (2) intergenomic translocations (Skalicka et al., 2003); and (3) loss of some N. tomentosiformis-derived repeats (Skalicka et al., 2005) including retroelements (Petit et al., 2007). Similarly, in synthetic allopolyploids of Triticum and Aegilops there are losses of amplified fragment polymorphic (AFLP) bands that are genome- and chromosome-specific (Ozkan et al., 2001), and these changes are associated with activation and silencing of genes and retroelements (Kashkush et al., 2002, 2003). However, genetic change was not observed in AFLP profiles of synthetic Gossypium allopolyploids (Liu et al., 2001), although epigenetic changes probably promote tissue-specific expression of homoeologues (Adams et al., 2004). Thus, available evidence points to varying degrees of polyploidy-induced genetic changes in the first few allopolyploid generations.

In nature any genetic diversity that is generated by de novo polyploidy might be under selective pressure, and perhaps a polyploid species emerging through the early genetic population bottleneck may be more homogeneous. However, young (< 100 yr old) allopolyploids may still show much genetic diversity. Natural populations of allopolyploids Tragopogon mirus and Tragopogon miscellus show considerable genetic and epigenetic variation in rDNA between populations that may reflect both multiple origins and rapid genome divergence in young establishing populations (Kovarik et al., 2005). Similarly in two distinct populations of the recently formed allopolyploid Senecio cambrensis, one population showed homogenization of rDNA unit types while another population did not (Abbott & Lowe, 2004). In contrast, populations of the recently formed allopolyploid Spartina anglica are much more genetically uniform, perhaps indicating less genome diversity stimulated by the allopolyploid event itself, rapid fixation of genetic change in the early populations or considerable vegetative propagation (Ainouche et al., 2004).

Our analyses of natural tobacco (< 200 000 yr old; Murad et al., 2002) indicated that genetic change observed in synthetic tobacco was ongoing in the divergence of natural tobacco. Genomic in situ hybridization revealed that the parental origin of chromosomes is easily detected, but with numerous intergenomic translocations (Fig. 1b), as found previously (Lim et al., 2004). However, GISH was less effective than in the synthetic tobacco, particularly with the chromosomes derived from the paternal progenitor, N. tomentosiformis (as in the synthetic tobacco studied here). Similarly, sequence-specific amplification polymorphism (SSAP) analysis of retrotransposon populations in tobacco and the progenitor diploids revealed many bands in N. tomentosiformis that are absent in the allotetraploid (Petit et al., 2007). Instability of paternal genomic DNA was also observed in restriction fragment length polymorphism (RFLP) profiles of synthetic Brassica allopolyploids similar to natural B. napus (Song et al., 1995). According to the nuclear–cytoplasmic interaction hypothesis proposed by Gill (1991), adverse interactions between paternal and maternal cytoplasm in a hybrid/allopolyploid can lead to paternal genome instability. If this is true for N. tabacum then the preferential losses of paternally derived repeats has continued well beyond the first few generations and the ‘genomic shock’ triggered by de novo allopolyploid-stimulated genetic change.

In N. quadrivalvis (estimated to be 1 Myr old), the effectiveness of GISH is much reduced, with clear evidence of considerable mixing of parental DNA (Fig. 1c). In addition, GISH with N. quadrivalvis labels the subtelomeric chromatin predominantly with N. obtusifolia (yellow) genomic DNA, showing that the repeats of N. attenuata origin have been largely replaced by repeats of N. obtusifolia origin. In the even older allopolyploid N. nesophila (c. 4.5 Myr), GISH fails (Fig. 1d) with little signal at the stringency used (estimated to give labelling at > 85% sequence identity). The absence of signal over the majority of the chromosomes indicates more than mixing of genomic sequences between parental genomes, as in N. quadrivalvis, but large-scale, genome-wide divergence of sequences in the allopolyploid. In support of this hypothesis, a tandem repeat (NNE) was identified that has evolved de novo or amplified from subthreshold abundance during the divergence of N. nesophila. This satellite has 62% and 72% sequence identity with related sequences in the progenitor diploids (NPAL and HRS60, respectively) and probably replaced pre-existing sequences using homogenization mechanisms. This satellite type probably evolved as a predominately subtelomeric repeat in the common ancestor of Nicotiana section Repandae species, N. nesophila, N. stocktonii and N. repanda. The mechanism of replacement is presumably by homogenization of subtelomeric repeats, as reported for Nicotiana section Alatae (Lim et al., 2006a).

Unlike satellite replacement in N. nesophila, the more recent allopolyploids N. tabacum, N. rustica and N. arentsii (all < 200 000 yr old) have tandem repeats, as found in the diploid progenitors (Lim et al., 2000, 2004, 2005). It is predicted that N. quadrivalvis (c. 1 Myr divergence) has similar or identical repeats to its diploid parents (because GISH labels most subtelomeric heterochromatin), although the distribution of GISH signal suggests mixing of repeats between parental chromosome sets. A few unlabelled subtelomeric DAPI bands (data not shown) indicate evolution of variant sequences. Thus, in Nicotiana allopolyploids, sequence mobility and evolution of new repeats is apparent at c. 1 Myr divergence, and new variants can be fixed at c. 5 Myr divergence.

These data are summarized in Fig. 3. They reveal that within 5 Myr of Nicotiana allopolyploid genome evolution the following sequence of events occurred: first, loss of paternally derived repeats (synthetic N. tabacum) that continued for many thousands of generations (natural N. tabacum); then sequence mixing became apparent, perhaps by homogenization (N. quadrivalvis) over time-frames of c. 1 Myr; and finally replacement of parental DNA with newly evolved or massively amplified subsets of genomic DNA over a time-frame of < 5 Myr. These changes can be considered as steps towards diploidization of the genome. Perhaps ongoing genome diploidization proceeds by a reduction in chromosome number, as seen in species belonging to Nicotiana section Suaveolentes, which formed c. 10 Myr ago and has species that range in chromosome number from 2n = 4x = 48 (expected) to 2n = 4x = 36.

Figure 3.

Scenario describing time-course of evolutionary events in Nicotiana allopolyploids. 1, Melayah et al. (2004); 2, Skalicka et al. (2005); 3, Petit (2007); 4, Skalicka et al. (2003); 5, Lim et al. (2006b); 6, Dadejováet al. (2007); 7, Kovarik et al. (2004); 8, Lim et al. (2004); 9, Clarkson et al. (2005); 10, Knapp et al. (2004); 11, Reported here; 12, I. J. Leitch (unpublished), 13, Goodspeed (1954).

Near-complete genome turnover

The approach presented here gives a global view of genome divergence, using the fate of two independent genomes united and evolving together in the polyploid nucleus. It provides high coverage and analyses simultaneously tandem and dispersed elements and heterochromatic and euchromatic sequences. Nevertheless, resolution is low, restricted to overall genomic similarities at > 1 Mb scales and labelling sequences of > 1 kb (Gill & Friebe, 1998). By contrast, sequence-based methods provide higher resolution but they do not give the same global view of the genome.

Genomic in situ hybridization is an overall measure of hybridization efficiency across the genome. The question arises as to whether we should just consider loss of GISH labelling in older allopolyploids as sequence divergence by point mutation or by genome ‘turnover’, driven by processes such as insertions, deletions and homogenization (unified by mechanisms of recombination). Retroelements and tandem repeats together contribute the majority of the plant genome. Detailed analyses of particular chromosomal regions in plants indicated that recombination shuffles small sections of DNA at surprisingly high frequencies (Ma et al., 2004). Indeed Ma et al. (2004) estimated that noncentromeric retroelements in rice had a half-life of < 6 Myr and were deleted at < 8 Myr. In maize there may be an even faster divergence of the genome as there is considerable genetic variability between inbred lines (Brunner et al., 2005). These data reveal that high levels of deletions and amplifications of the genome act to replace DNA. Our analysis of tandem repeats (i.e. NPAL) suggests replacement of one repeat type with another within 5 Myr of genome divergence, probably using mechanisms of concerted evolution (Lim et al., 2006a). In addition, over this time-frame GISH largely failed, perhaps indicating genome-wide sequence replacement. The observation that genomic DNA mixing (N. quadrivalvis) over 1 Myr precedes novel sequence replacement (N. nesophila) within 5 Myr is comparable to observations in Gossypium allopolyploids in which intergenomic invasions involving rDNA (Wendel et al., 1995) and certain classes of transposable elements (Zhao et al., 1998) are observed over 1–2 Myr of divergence. Thus, available data indicate replacement of sequence, rather than sequence mutation, as primarily important in eroding the efficacy of GISH. For these reasons we favour the term ‘genome turnover’.

Studies of orthologous centromeric regions of Oryza indica and Oryza japonica indicated higher rates of LTR retrotransposon accumulation and lower rates of deletion, leading to a higher abundance of the most ancient sequences at the centromere (Ma & Bennetzen, 2006). However, in other species satellite sequences at the centromeres are thought to evolve particularly rapidly (e.g. in Oryza and Zea), perhaps because of selective advantage incurred through centromere-spindle attachment at cell division (centromere drive hypothesis) (Dawe, 2005). If a different rate of divergence of sequences occurred at the centromeric region of Nicotiana, one might expect different GISH-labelling patterns in this region. However, no evidence was found that the centromeric regions were behaving any differently from the majority of the genome.

The question arises as to whether the rate of repetitive DNA divergence is greater in allopolyploids compared with diploids. In species of diploid Nicotiana section Alatae, subtelomeric satellite repeats are being replaced over time scales of < 3 Myr (Lim et al., 2006a). In Brassicaceae, Arabidopsis arenosa and Arabidopsis thaliana are thought to have separated 5 Myr ago and have centromeric satellites with 73% sequence identity (Hall et al., 2005). Our observation that HRS60 and NPAL are replaced by NNE within 5 Myr in the allopolyploid N. nesophila (all sequences sharing 72% sequence identity) suggests that the speed of satellite divergence is not materially different in allopolyploids and diploids (i.e. satellites sharing the same nucleus following allopolyploidy are not evolving more quickly than similar repeats in diploids). These satellites from Nicotiana and Arabidopsis are apparently evolving more rapidly than two centromeric satellites in the allopolyploid soybean (Glycine max) that distinguish two chromosome sets and may have their origin in the original diploid parents c. 15 Myr ago (Walling et al., 2006, and S. Jackson unpublished data cited in Udall & Wendel, 2006). Primate satellites may also be evolving more slowly since human and chimpanzee satellites share high sequence identity (84–91%), after 5 Myr of divergence (Baldini et al., 1991; Glazko & Nei, 2003), the same period of time as the age of N. nesophila in which new satellites have evolved. Thus, it is likely that the rate of divergence of repetitive sequences is influenced by the character of the genome as well as the nature of the sequence.

There is a long-standing and ongoing debate about the role of epigenetic chromatin remodelling mechanisms (e.g. RNAi, histone acetylation and chromatin methylation) in stabilization of repetitive sequences, including tandem and dispersed, particularly retroelement repeats, in plants (Lippman et al., 2004). Chromatin-stabilizing mechanisms are clearly not operating over timescales of 5 Myr in Nicotiana or else GISH would be effective. Instead, we predict that these timescales are sufficient for near-complete genome turnover. Effectiveness of chromosome painting using degenerate oligonucleotide primed polymerase chain reaction (DOP-PCR) probes in mammals that have diverged over at least an order of magnitude longer time-periods indicates no such genome instability. These results may indicate that genomes of plants and mammals have fundamentally different overall rates of genome divergence.


The authors thank NERC, the Czech Academy of Sciences, Grant Agency of Czech Republic (521/04/0775 and 204/05/0687) and the British Council Alliance programme for support. We thank B. Koukalova for assistance and three anonymous referees for helpful comments.