Analysis of two abundant, highly related satellites in the allotetraploid Nicotiana arentsii using double-strand conformation polymorphism analysis and sequencing


  • Roman Matyasek,

    1. Institute of Biophysics, Academy of Sciences of the Czech Republic, v.v.i, Laboratory of Molecular Epigenetics, Královopolská 135, CZ-61265 Brno, Czech Republic
    Search for more papers by this author
  • Jaroslav Fulnecek,

    1. Institute of Biophysics, Academy of Sciences of the Czech Republic, v.v.i, Laboratory of Molecular Epigenetics, Královopolská 135, CZ-61265 Brno, Czech Republic
    Search for more papers by this author
  • Andrew R. Leitch,

    1. School of Biological and Chemical Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS, UK
    Search for more papers by this author
  • Ales Kovarik

    1. Institute of Biophysics, Academy of Sciences of the Czech Republic, v.v.i, Laboratory of Molecular Epigenetics, Královopolská 135, CZ-61265 Brno, Czech Republic
    Search for more papers by this author

Authors for correspondence:
Roman Matyasek
Tel: +420 541 517 230
Ales Kovarik
Tel: +420 541 517 178


  • Allopolyploidy, a driving force in plant evolution, can induce rapid structural changes in parental subgenomes. Here, we examined the fate of homologous subtelomeric satellites in intrasection allotetraploid Nicotiana arentsii formed from N. undulata and N. wigandioides progenitors < 200 000 yr ago.
  • We cloned and sequenced a number of monomers from progenitors and the allotetraploid. Structural features of both cloned and genomic monomers were studied using double-strand conformation polymorphism analysis.
  • Two homologous satellites were isolated from N. undulata (called NUNSSP) and N. wigandioides (NWISSP). While the NUNSSP monomers were highly homogeneous in nucleotide sequences, the NWISSP monomers formed two separate clades. Likewise, the genomic NUNSSP monomers showed less DNA conformation heterogeneity than NWISSP monomers, with distinct conformations. While both satellites predominantly occupy subtelomeric positions, a fraction of the NWISSP repeats was found in an intercalary location, supporting the hypothesis that dispersion prevents the repeats becoming homogeneous. Sequence, structural and chromosomal features of the parental satellites were faithfully inherited by N. arentsii.
  • Our study revealed that intergenomic homogenization of subtelomeric satellite repeats does not occur in N. arentsii allotetraploid. We propose that the sequence and structural divergence of subtelomeric satellites may render allopolyploid chromosomes less vulnerable to intergenomic exchanges.


A significant evolutionary role has been assigned to allopolyploidy in the formation of flowering plant species (Wood et al., 2009). Two ancient whole-genome duplications, one in the common ancestor of extant seed plants and the other in the common ancestor of extant angiosperms, have probably shaped the genomes of modern angiosperms (Cui et al., 2006; Soltis et al., 2009; Jiao et al., 2011). The success of newly formed allopolyploids is partly attributable to their highly plastic genome (Jackson & Chen, 2009) and mating systems (Brennan & Hiscock, 2010; Chapman & Abbott, 2010; McCullough et al., 2010; Nah & Chen, 2010). The consequences of allopolyploidy have been studied at the chromosomal and sequence levels, demonstrating that rapid genetic and epigenetic changes are frequently linked to polyploidy (Doyle et al., 2008; Jones & Hegarty, 2009; Parisod et al., 2010). The nature of genomic changes associated with allopolyploid speciation is still not clearly understood, but several mechanisms have been proposed, including transposon proliferation, chromosomal translocation, repeat loss and gain, gene conversion and homologous recombination (for review, see Leitch & Leitch, 2008; Gaeta & Pires, 2010; Parisod et al., 2010).

Satellite DNA is a nearly universal component of eukaryotic genomes, and it consists of numerous tandem repeats that are arranged head to tail. These repeats are noncoding, late-replicating in the S-phase and mostly located in the constitutive, nontranscribed heterochromatin at centromeric and subtelomeric locations (Charlesworth et al., 1994). Satellites often show enormous variability in nucleotide sequence and copy number, even among closely related species (Miklos, 1985). Therefore, they are frequently utilized for taxonomic and phylogenetic studies and to follow hybridization events between related species (Bachmann et al., 1993; Contento et al., 2005; Sharma & Raina, 2005; Hemleben et al., 2007; Gill et al., 2009). However, the loss of specific parental satellites in hybrids can rapidly occur in allopolyploids by rearrangement and recombination processes (Pestsova et al., 1998; Han et al., 2005; Skalicka et al., 2005). Since satellite DNA can form up to 50% of the total genomic contents (Plohl et al., 2008), changes in unit copy number can significantly influence genome size. Measurements of C-values revealed that many allopolyploids show a reduction in the amount of DNA relative to the sum of their respective diploid progenitors (Leitch & Bennett, 2004). Consequently, it was proposed that the elimination of satellite DNA sequences could be responsible for this reduction, at least to some degree (Koukalova et al., 2010). One of the most widespread characteristics of a satellite is the intrinsically bent structure of the monomers (Fitzgerald et al., 1994; Palomeque & Lorite, 2008). The bending motifs leading to the curvature of the DNA helix axis include periodically spaced clusters of d(A–T)4–6 (Koo et al., 1986), as well as GGC nucleotide motifs (Ussery et al., 1999). The curvature of the DNA organizes tandem repeats in higher-order structures, believed to be important for the tight nucleosome packing in heterochromatin (Radic et al., 1987; Bussiek et al., 2009).

Nicotiana genomes are particularly rich in tandem satellite repeats. At least three unrelated satellite repeats have been characterized in this genus: HRS60 (Koukalova et al., 1989), NTS9 (Jakowitsch et al., 1998) and NTRS (Matyasek et al., 1997). The HRS60 superfamily, with members that occupy subtelomeric regions of many chromosomes, is probably the best characterized and most widely studied group. Members of this superfamily have been isolated from Nicotiana tabacum (Koukalova et al., 1989; Matyasek et al., 1989; Suter-Crazzolara et al., 1995), Nicotiana tomentosiformis (Gazdova et al., 1995), Nicotiana plumbaginifolia (Chen et al., 1997), Nicotiana paniculata (Lim et al., 2005), Nicotiana palmerii, Nicotiana nesophila (Lim et al., 2007a,b; Koukalova et al., 2010), Nicotiana sylvestris, Nicotiana nudicaulis and Nicotiana quadrivalvis (Koukalova et al., 2010).

While most previous studies involved comparisons of satellites from diverged species, in the present study, we addressed the question of satellite evolution in closely related species united in a single taxonomy unit. Lim et al. (2005) previously isolated and cytogenetically characterized a satellite called NUNSSP from N. undulata. In this study, only a few clones were sequenced and no interspecies comparisons were attempted. Therefore, we analysed subtelomeric satellites in Nicotiana section Undulatae that diverged < 7 million yr ago (MYA; Clarkson et al., 2004). Particular attention was paid to the degree of parental satellite recombination in an intrasection N. arentsii allotetraploid. This allopolyploid was formed < 200  MYA by interspecies hybridization of species close to modern N. undulata (mother genome donor) and N. wigandioides (father genome donor) (Goodspeed, 1954; Clarkson et al., 2010). Using double-strand conformation polymorphism (DSCP) analysis, sequencing and restriction fragment length polymorphism (RFLP) analysis, we examined structural features and evolutionary relationships of two distinct satellites native to the related diploid plant species N. undulata and N. wigandioides, respectively; inheritance and homogenization of these satellites in the N. arentsii allotetraploid; and the distribution of related repeats in the section Undulatae.

Materials and Methods

Plant material, DNA extraction and Southern blot hybridization

The following plant materials were used: Nicotiana arentsii Goodsp., Nicotiana undulata Ruiz & Pav. (accession numbers TW12 and TW145, respectively; USDA, Agricultural Research Center, North Carolina State University, Raleigh, NC, USA), Nicotiana wigandioides Koch & Fintelm. (accession Nee et al., 51764, New York Botanical Gardens, Bronx, NY, USA), Nicotiana glutinosa L. and Nicotiana thyrsiflora Good sp. (accession number MWC 12690). The DNA was extracted from fresh young leaves according to Kovarik et al. (2000), digested with restriction endonucleases (5 U μg−1 DNA, twice, for 6 h), fractionated by gel electrophoresis and transferred to Hybond XL membranes (GE-Healthcare, Little Chalfont, UK) using either alkaline capillary transfer (agarose gels) or electrotransfer (polyacrylamide gels, Semi-dry Blotter; Hoefer, Holliston, MA, USA). The membranes were hybridized with 32P-labelled DNA probes (DecaLabel DNA Labeling Kit, MBI Fermentas, Vilnius, Lithuania). Southern blot hybridization was carried out in a 0.25 M sodium phosphate buffer (pH 7.0) supplemented with 7% w/v sodium dodecyl sulphate (SDS) at 65°C (Sambrook & Russell, 2001). The membranes were washed with 2 × SSC (10 × SSC = 1.5 M NaCl, 0.15 M sodium citrate, pH 7.0), 0.1% SDS (twice for 5 min) and then with 0.2 × SSC and 0.1% SDS (twice for 15 min at 65°C). The membranes were exposed to a storage phosphor screen, scanned (Storm, GE-Healthcare), and the signal was quantified using Image Quant (GE-Healthcare).

The following DNA probes were used: NUNSSP – a mixture of NUNSSP monomers (Lim et al., 2005); 5S rDNA – the 5S rDNA gene region of N. tabacum (Fulnecek et al., 1998); 26S rDNA – a 220 bp fragment obtained by PCR amplification of the 3′-end region of the 26S rDNA gene (Lim et al., 2006); and DNA from N. arentsii.

Isolation of satellite sequences

Plant DNA (15 μg) was digested with SspI, size-fractionated on a 7% polyacrylamide gel and stained with ethidium bromide. A band of c. 200 bp of each species was excised, eluted from the gel using the ‘crush and soak’ method (Sambrook & Russell, 2001) and cloned into the pDrive Cloning vector (Qiagen PCR Cloning kit, Qiagen). Before cloning, a single adenine was added to both ends of the DNA using DyNAzyme II DNA polymerase (FINNZYMES, Espo, Finnland). Cloned inserts that strongly hybridized with NUNSSP and DNA from N. arentsii probes were sequenced. Sequence data were submitted to the EMBL Nucleotide Sequence Database with the accession numbers FN594937-594943, FN658772FN658810, and FN667946FN667951. Most clones fell within the c. 180 bp range; one clone (accession FN667951) contained a 348 bp insert composed of 154 bp of the 5′ sequence of the NUNSSP monomer unit, whereas the 3′ part of the insert was a unique sequence, which probably flanked the satellite. This clone was not included in the distance analyses. The satellite monomers from N. undulata were cloned into an EcoRV site of the pZERO/kan vector (Life Technologies, Invitrogen).

Phylogenetic analysis

The sequence distance values, expressed as the number of substitutions per 100 bases, were computed using the DISTMAT program (implemented in Phylip) which calculates the pairwise evolutionary distances between sequences in a multiple alignment (distance matrix). Uncorrected as well as Jukes–Cantor and the Kimura multiple substitution correction method were used for the calculations, but all methods provided similar results. Gaps were retained in the alignments. The weights of indels were set to zero. An output file containing the distance matrix for the set of sequences was used as the input for statistical evaluation using the STATISTICA 6 program.

Nucleotide sequences were aligned using ClustalW multiple alignment (BioEdit Sequence Alignment Editor (Hall, 1999) controlled by eye. Preliminary analyses showed that removing indels from the analyses did not have a great effect on the resulting tree topologies (data not shown), so all indels were retained in further alignments. The final data sets consisted of 70 sequences and 182 aligned characters for the common tree (Supporting Information, Fig. S1a), 28 sequences and 181 aligned characters for the N. arentsii tree (Fig. S1b), 18 sequences and 180 aligned characters for the N. undulata tree (Fig. S1c) and 24 sequences and 182 aligned characters for the N. wigandioides tree (Fig. S1d). The model of molecular evolution that best fitted the final dataset was determined using the Akaike information criterion (AIC and AICc), the Bayesian information criterion (BIC) or the likelihood ratio test (LRT) implemented in the Modeltest program version 3.8, available on (Posada & Crandall, 1998), or in the MrAIC program version 1.4.4, available on The best-fitting models selected for each sequence set are listed in Table S1. The simplest model was used for the final analysis of each set of sequences (Table S2). More complex models were also used, but these provided the same tree with similar posterior probabilities (results not shown). We performed Bayesian (BI) and neighbour-joining (NJ) analyses.

The BI analysis was performed using MrBayes version 3.1.2 (Huelsenbeck & Ronquist, 2001). Two simultaneous Metropolis-coupled Markov chain Monte Carlo analyses with four chains each were run, incrementally heated by a temperature of 0.1 for at least three million generations, and every 100th tree was sampled. The standard deviation of split frequencies (< 0.01) was used as the convergence diagnostic. After stationarity was reached, the first 25% of the trees were discarded as burn-in, and a consensus tree with branch lengths and posterior probability was computed.

The NJ analysis was carried out using Phylip programs (Felsenstein, 1989), The distance data were obtained by running the DNADIST program with bootstrap resampling of 1000 replicates. Models of the nucleotide substitutions used for each set of sequences are listed in Table S2. Distance trees were constructed from distance data (randomized input order) employing a NJ algorithm (Saitou & Nei, 1987) implemented by the NEIGHBOR program. Finally, consensus distance trees, including bootstrap values, were plotted using the CONSENSE and DRAWTREE programs (Phylip) or DENDROSCOPE V 2.7.4., (Huson et al., 2007).

DNA conformation analysis

In order to examine the positions of the bending centre, we used a test similar to the circular permutation analysis of Wu & Crothers (1984). These authors showed that when a DNA molecule is bent, the anomaly in migration will be greater if the bend is in the middle of the molecule than if it is at the end. As the NUNSSP-like monomers are arranged in a tandem order, that is, in a head-to-tail fashion, the restriction of genomic DNA with distinct enzymes, each having a unique cutting site, produces populations of circularly permutated monomers of the same mean length, but with the major bending sites located in different positions along the fragments.

Permutated genomic monomers excised from the plant DNA with SspI or BfaI were Southern blot-hybridized with the NUNSSP probe. Polyacrylamide gel electrophoresis (PAGE) was performed in 1× Tris/borate/EDTA buffer at 210 V using nondenaturing polyacrylamide gel (30 × 16 cm), either 15% (44 : 1) at 5°C or 7% at 50°C, until xylene cyanol dye was c. 1 cm from the lower edge of the gel. In order to analyse the electrophoretic behaviour of sequenced monomers, genomic inserts were excised from pZeroIIKan (Invitrogen) and pDrive (Qiagen) cloning vectors using PstI/NotI and EcoRI enzymes, respectively. After PAGE (15%, 5°C), the monomers were stained with ethidium bromide. The relative mobility of the monomers was expressed as the distance (in pixels) between the start and the corresponding DNA fragment.


Isolation of satellite DNA

In order to isolate the satellite monomers, we employed a classic approach based on the restriction of genomic DNA and the subsequent separation of fragments by gel electrophoresis. Fluorescent bands are produced when there is a conserved restriction site in tandemly arranged high-copy repeats. The DNA of N. wigandioides and N. arentsii was restricted with SspI, which was previously shown to liberate satellite monomers in N. undulata (Lim et al., 2005). After ethidium bromide staining a prominent band of c. 200 bp was revealed in a smeared background (Fig. S2). Because the SspI bands had a similar intensity as those of N. undulata (Lim et al., 2005), all three species probably contain comparable amounts of homologous satellites. The SspI monomers from all three species were cloned and the inserts sequenced. The monomers from N. wigandioides and N. undulata were called NWISSP and NUNSSP, respectively. All but one of the monomers isolated from N. undulata had lengths of 180 bp. By contrast, most of the monomers (79%) isolated from N. wigandioides were 181 bp in length; none of the monomers of this size were found among the N. undulata clones. Variation in size (165–182 bp) in 21% of N. wigandioides clones was caused by single deletion/insertions occurring at varying positions along the unit (Fig. S1d). We concluded that the 181 and 180 bp lengths of the monomers were characteristic lengths of major satellite repeats in N. wigandioides and N. undulata, respectively. Most of the monomers isolated from N. arentsii were 180 bp (46%) or 181 bp (43%), suggesting parental unit additivity in this allotetraploid.

Species-specific distribution, genomic organization and abundance of satellite repeats

In order to study the species-specific distribution of isolated satellite repeats, we carried out Southern blot analysis in section Undulatae using the NUNSSP probe (Fig. 1a). Ladders of regularly spaced hybridization bands were observed in N. undulata, N. wigandioides, N. thyrsiflora and N. arentsii, whereas N. glutinosa DNA produced no signal. In order to reveal potential interspecies site-specific polymorphisms, we carried out Southern hybridization with a wide spectrum of restriction endonucleases (Fig. 1b). All enzymes cutting the satellite monomers yielded regular hybridization ladders corresponding to the tandem arrangement of units. Some sites were more conserved than others, pointing to satellite unit heterogeneities. No significant interspecies variability in restriction profiles was observed. Slightly weaker hybridization signals in the wig lanes could be explained by the decreased homology between the probe and the N. wigandioides satellite (see later). The hybrid origin of N. arentsii was confirmed using a 5S rDNA probe: in the N. arentsii allotetraploid, the probe hybridized to specific fragments corresponding to both parental diploid species, indicating the Mendelian inheritance of 5S rDNA (Fig. 1c). Thus, the 5S repeats showed more contrasting patterns of evolution than the 35S rDNA that has been completely homogenized in this species (Kovarik et al., 2004).

Figure 1.

Southern blot analysis of satellite repeats (a, b) and 5S rDNA (c). (a) Species specificity of NUNSSP-like satellites: glu, Nicotiana glutinosa; und, Nicotiana undulata; are, Nicotiana arentsii; wig, Nicotiana wigandioides; thy, Nicotiana thyrsiflora. (b) Genomic organization of repeats in N. arentsii, N. wigandioides and N. undulata, analysed by the set of restriction enzymes. (c) Hybridization signals of 5S rDNA were additive in the N. arentsii allotetraploid. The blots were hybridized with the NUNSSP (a, b) and 5S rDNA probes (c).

Phylogenetic relationships between satellite repeats

Distinct patterns of divergence may be apparent between a satellite in a diploid progenitor and the corresponding satellite from that progenitor’s subgenome in the allotetraploid. In order to examine these patterns, we carried out multiple comparisons of NUNSSP-like monomers isolated from the N. arentsii allotetraploid (28 clones), the diploid paternal progenitor N. wigandioides (24 clones) and the diploid maternal progenitor N. undulata (18 clones). Three well-supported clades U, WF and WS were visualized on phylogenetic trees constructed using NJ and BI methods (Figs 2, S3). While the N. undulata sequences were highly clustered (U-clade), sequences from N. wigandioides were split into two clades that we called WS and WF. Sequences isolated from N. arentsii occurred within all three clades: 16 sequences clustered with the 180 bp monomers from the N. undulata parent, and seven and five sequences clustered with the WF and WS clades, respectively. No N. arentsii-specific clade was revealed, even after changing the calculation parameters. In general, the NJ and BI phylogeny methods resulted in essentially similar topologies that slightly differed in the degree of resolution and partly differed in nodal support. Importantly, there was a high congruence in the clustering of individual sequences into three major groups. The posterior probabilities of the BI method showed somewhat better support for sub-branches than the NJ method. A few sequences could not be categorized into any of the three groups in the NJ tree.

Figure 2.

Phylogenetic relationships between the satellite monomers. Trees were constructed from aligned sequences (Supporting Information, Fig. S1a) using (a) neighbour-joining and (b) Bayesian algorithms. The sequences from Nicotiana undulata (symbol U followed by the number of the clone), Nicotiana wigandioides (W) and Nicotiana arentsii (A) are labelled in red, blue and green branch marks, respectively. The three main clades, U, WS and WF, are visualized in both trees. Note that while sequences from the diploid parents N. undulata and N. wigandioides were well separated into U and WS + WF clades, respectively, sequences from the N. arentsii allotetraploid intermingled with sequences from both diploids and occurred in all major clades.

Satellite repeat homogeneity was inspected using extensive pairwise comparisons of the monomers (Fig. 3, Table S3). N. wigandioides had the highest intraspecific sequence heterogeneity (Fig. 3 left part, wig/wig comparisons), followed by N. arentsii (are/are comparisons) and N. undulata (und/und comparisons). The relatively low intragenomic sequence distances found in N. arentsii can be explained by the presence of highly homogenous repeats inherited from the N. undulata parent, decreasing the overall sequence distances in the hybrid. As expected, interspecies sequence distance was highest between N. undulata and N. wigandioides (und/wig comparisons). The mean diversity was relatively low within the clades (6–8% sequence distances, Fig 3), whereas it was significantly larger between the clades (14–16%), in accordance with the phylogenic analysis (Fig. 2). Conserved polymorphic nucleotides distinguishing between individual clades are schematically depicted in Fig. 4. The most divergent WS clade accumulated at least six conserved mutations. In short, these results support Mendelian inheritance of the parental satellites in the N. arentsii allotetraploid.

Figure 3.

Comparison of sequence distances among satellite monomers within (box plots 1–3, 7–9 from the left) and between (4–6, 10–12) individual species (the six box plots on the left) and clades (the six box plots on the right). Distance expressed as substitutions per 100 bp (y-axis) was calculated according to the Jukes–Cantor method. The number of pairwise comparisons used for the calculations is given below each box plot. Interclade distances (e.g. U/WF) were significantly larger than intraclade (e.g. U/U) distances (P < 0.001, Mann–Whitney U-test, Supporting Information, Table S3). Statistically insignificant differences in sequence distances (P > 0.05; Table S3) are indicated with a pair (a,a; b,b;…) of small letters placed above box plots compared. Differences of all other pairwise combinations are statistically highly significant (P < 0.001; Table S3).

Figure 4.

Sequence features of satellite monomers within the individual WS, WF and U clades. Red squares, dA tracts; blue squares, dT tracts; violet squares, altering AT tracts; green squares, other sequence motifs specific for WS clade. The motifs characteristic of the WS family are indicated by the following numbers: 1, 2, 5, conserved point mutations; 3, unique T6 tract; 4, TATATA palindrome; 6, AT-tract; 7, TTCTT motif. The black dotted line indicates a single nucleotide deletion. Restriction sites that define SspI- and BfaI-permutated monomers are indicated with arrows. Consensus sequences were used for the graphs.

DNA conformational polymorphisms in NUNSSP-like satellites

Structural properties of the NUNSSP and NWISSP satellites were studied using DSCP analysis. While RFLP analysis detects mutations that only affect rare restriction target sites, DSCP analysis detects when the curvature of the helical axis of double-stranded DNA molecules is altered via almost any mutation (Saad et al., 1994; Argüello et al., 1998). An increased curvature is generally reflected by a slower migration of fragments in nondenaturing polyacrylamide gels (Marini et al., 1982). There was a broad range of mobility between individual clones, indicating variability in their conformation (Figs 5, S4, Table S4): the three slowest-migrating monomers belonged to the WS clade while the five fastest monomers were members of the WF clade. Importantly, the fastest-moving monomer was 181 bp long, indicating that the size differences only had a minor effect on mobility compared with DNA conformation. The NWISSP monomers (WS and WF clades) had a larger conformational variability compared with those of NUNSSP (U clade) (Fig. 5b, Table S5). The difference in the average intraspecific mobility between the 181 bp monomers from N. arentsii and N. wigandioides was statistically insignificant (> 0.1, Table S5b), whereas the difference in the average intraclade mobility between the 181 bp monomers from the WS and WF clades was moderately statistically significant (< 0.05, Table S5b).

Figure 5.

Conformation analysis of cloned NUNSSP-like monomers. 180 and 181 bp inserts from Nicotiana arentsii (are) and Nicotiana wigandioides (wig) were separated on 15% polyacrylamide gel at 5°C and were visualized by ethidium bromide fluorescence (Supporting Information, Fig. S4). (a) Schematic representation of the electrophoretic mobility of individual clones assigned to species of origin (are, wig) as well as to individual phylogenetic clades U, WS and WF. (b) Statistical evaluation of results shown in (a). The mobility (y-axis) is expressed as the distance (in pixels/10) between the electrophoretic start and the corresponding bend. The legend describing the box plots is as shown in Fig. 3. The corresponding descriptive statistics are summarized in Table S5(a) and the tests of differences between the mobility of individual groups of cloned monomers are summarized in Table S5(b). Note that the difference in the average intraspecific mobility between the monomers from N. arentsii and N. wigandioides was statistically insignificant (#, > 0.1), whereas the difference in the average intraclade mobility between the monomers from the WS and WF clades was moderately statistically significant (*, < 0.05).

Satellites of the HRS60 superfamily typically comprise hundreds of thousands of units in the genome (Koukalova et al., 1989). Interpretation of the clone analysis can be hampered by a relatively low number of clones. In order to provide a statistically more relevant view of the satellite unit conformation in individual species, we carried out DSCP analysis on restricted genomic DNA (Fig. 6). Digestion with SspI released most of the satellite monomers, visualized as single sharp bands after PAGE at an elevated temperature (50°C) and Southern blot hybridization with the NUNSSP probe (Fig. 6 lower panels, Table S6). However, when the gels were run at 5°C, the bands were smeared, probably as a result of temperature-sensitive conformation heterogeneity of the monomers (Fig. 6 upper panels, Table S6). The N. undulata signal showed a single peak, whereas the N. wigandioides signal was split into two peaks, indicating the presence of two prominent conformation types, probably corresponding to the WS and WF clades of monomers that differed in average electrophoretic mobility (Fig. 5b, Table S5). N. arentsii showed a bimodal distribution of conformers: a more prominent, fast-migrating fraction most likely originating from the N. undulata genome, and a slowly migrating shoulder resembling that of N. wigandioides. No significant interspecies differences were observed in the mobility of monomers at a low temperature after the digestion of genomic DNA with BfaI (Fig. 6 right panels), probably because the major centres of curvature are located close to the recognition site of these permutated monomers (Fig. 4). It is known that bends towards the end of a molecule do not significantly influence gel–DNA interactions (Marini et al., 1982).

Figure 6.

Conformation analysis of genomic NUNSSP-like monomers. Genomic DNAs of Nicotiana undulata (und), N. wigandioides (wig) and N. arentsii (are) were digested with the indicated restriction endonucleases, separated on polyacrylamide gels at 5°C (upper panels) or 50°C (lower panels) and hybridized with the 32P-labelled NUNSSP probe. Quantitative densitometric evaluations of the hybridization signals are shown to the right of each gel. The descriptive statistics are summarized in Supporting Information, Table S6.

In order to better understand the nature of the distinct electrophoretic behaviour of satellite monomers, we searched for the motifs known to affect DNA curvature. Among others, the short, periodically phased T4–6 tracts were shown to influence electrophoretic mobility most significantly. A specific T6 tract was found in most WS monomers at position 91 (Figs 4, S1). All monomers belonging to the remaining two clades had a shorter T4 tract (WF clade) or no T tract (U clade) at this position. Another T tract specific for the WS clade was found at position 144. We concluded that the WS clade of monomers had a higher frequency of sequence motifs, inducing DNA curvature. These sequence features are in good agreement with the significantly lower average electrophoretic mobility of WS monomers compared with WF monomers (Figs 5, S4, Table S5). Consequently, WS monomers probably correspond to the slower migration of the SspI genomic band in a genomic blot (Fig. 6).


Origin, evolution and distribution of NUNSSP-like sequences

Phylogenetic studies have placed Undulatae as a monophyletic section comprised four diploid (N. undulata, N. wigandioides, N. thyrsiflora, N. glutinosa) and one allotetraploid (N. arentsii) species (Clarkson et al., 2004; Knapp et al., 2004). Our results showed that the NUNSSP-like satellites occur in N. undulata, N. wigandioides, N. thyrsiflora and N. arentsii genomes, whereas they are absent in N. glutinosa. It follows that amplification of repeats arose after the separation of N. glutinosa from the rest of Undulatae. N. glutinosa has been proposed as being a homoploid hybrid between species from sections Tomentosae and Undulatae (Clarkson et al., 2010). Since its genome does not seem to contain either Undulatae-specific satellites (this study) or Tomentosae-specific satellites (Lim et al., 2000), it is most likely that N. glutinosa is a relatively ancient homoploid that has lost much of the parental repeats and possibly evolved new ones. In the future, it will be interesting to analyse these unique satellites (if present) as well as low-copy sequences that mostly appear to be biparently inherited in Nicotiana allotetraploids (Matassi et al., 1991; Intrieri et al., 2008; Fulnecek et al., 2009; Kelly et al., 2010).

Although satellite repeats have frequently been used as suitable cytogenetic markers, their phylogenetic significance has been a topic of considerable debate (Plohl et al., 2008). This is because the rate of their evolution largely varies between loci and species. For example, distinct satellite variants survived speciation events over 540 MYA in certain bivalve species (Plohl et al., 2010). In angiosperms that radiated 140–150 MYA ago, there is no evidence for such conserved satellites since many distinct satellites evolved at the family and genus levels (Hemleben et al., 2007), underlining the high plasticity of plant genomes (Kejnovsky et al., 2009). The HRS60 superfamily comprises > 10 related satellites amplified in different sections and species of Nicotiana (Lim et al., 2006; Koukalova et al., 2010). The divergence between species-specific NUNSSP and NWISSP satellites is lower than between other members of the HRS60 superfamily. This may be explained by the fact that the sequences were mostly isolated from species comprising different sections (Lim et al., 2006; Koukalova et al., 2010), while here, for the first time, satellites were isolated from species of the same section. We can conclude that the divergence between members of the HRS60 superfamily is larger between than within sections. However, one has to bear in mind that a new satellite, differing from the parental one by as much as 30% in sequence, may evolve in < 5 million yr (Koukalova et al., 2010), suggesting that satellite expansions may be tightly linked with speciation processes.

The monophyletic section Undulatae is probably < 7 million yr old, according to molecular dating based on chloroplast and nuclear markers (Chase et al., 2003). It is therefore not surprising that Southern blot analysis failed to reveal any substantial differences in the restriction profiles of satellite DNA in N. wigandioides or N. undulata (Fig. 1b). However, analysis at the haplotype level revealed distinct features of repeats in N. undulata (amplifying NUNSSP) and N. wigandioides (NWISSP) genomes: (1) the NUNSSP satellite is mostly composed of 180 bp units whereas the NWISSP satellite has a dominant monomer of 181 bp; (2) the NWISSP sequences are more heterogeneous, forming two well-supported clades, whereas the NUNSSP sequences are relatively homogeneous (Fig. 2) – in addition, N. wigandioides evolved a new species-specific WS subfamily that had conserved sequence motifs; (3) conformers of NWISSP, particularly in the WS clade, seem to be more curved than the NUNSSP conformers. Retention of diverged repeats in N. wigandioides suggests that the sequence homogenization process is less efficient in this species or that the repeat population undergoes a transition to a new equilibrium. While there was no experimental support for a differential rate of repeat homogenization, the cytogenetic observations (Fig. 7) suggested a correlation between sequence divergence and the mobility of repeats across the chromosomes. The NUNSSP-like satellites predominantly occupy subtelomeric positions, typical for the HRS60 superfamily, in both N. undulata and N. wigandioides. As such, the interchromosomal barrier does not seem to be a limiting parameter to the homogenization process in N. undulata, at least. However, there were also three to four dispersed interstitial sites in N. wigandioides (Fig. 7). Similarly, highly heterogeneous GRS satellites (Lim et al., 2000) occupy a nontypical interstitial position. Also, analysis of human centromeric satellite arrays revealed haplotype differences between the chromosomes (Roizes, 2006). It might be possible that sequence and conformation variability of NWISSP repeats originates from satellite dispersion along the chromosomes, and homogenization pressures responsible for the maintenance of satellite sequence uniformity may differ between different chromosomal compartments. However, the possibility of colocalization of WS and WF repeats cannot be excluded since divergent satellites exist in subtelomeric regions of N. paniculata (Lim et al., 2005) and rye (Evtushenko et al., 2010). In addition, unit diversity has been maintained when moved within the genome in Triticeae (Contento et al., 2005). The interspersion of satellites (Zinic et al., 2000) and the irregular organization of units have been proposed as hallmarks of increased recombination activity of the locus (Kuhn et al., 2009).

Figure 7.

Karyotypes with NUNSSP-labelled chromosomes of the diploid parents and the Nicotiana arentsii allotetraploid. The probe was NUNSSP-labelled with red (diploid species) and green (N. arentsii) fluorochromes. In Nicotiana undulata the chromosomes were stained with a genomic in situ hybridization (GISH) probe (green); N. arentsii and Nicotiana wigandioides chromosomes were stained with 4’,6-diamidino-2-phenylindole (DAPI, in blue). Asterisks indicate chromosomes with probe hybridization to interstitial locations. The image was adapted from Lim et al. (2004) and reproduced with kind permission from John Wiley & Sons.

Distinct structural features of subtelomeric satellites

The divergence of satellite monomers isolated from a single species was relatively low (c. 8%). It was therefore surprising that the conformation analysis revealed astonishing variability in the migration of individual conformers both across and within clades, and only few clones showed similar electrophoretic mobilities in polyacrylamide gels (Figs 5, S4, Table S4). The conformation heterogeneity of satellites was further supported by genomic blots that showed rather diffuse hybridization bands in both diploid species. The reason for such heterogeneity is not fully understood. Permutation analysis (Fig. 6) showed that oligo(dA) tracts are probably a major source of DNA conformation polymorphisms in all three subfamilies, but other motifs, such as the GGC triplets, could also be involved. In accordance, members of the more curved WS clade had more bend-inducing motifs than those of the other two clades (Fig. 4). Since N. arentsii satellites showed conformation features of both parents, it is likely that these were inherited from both parents in the allotetraploid. It is clear that the NUNSSP-like repeats were mostly arranged in tandem arrays of head-to-tail units (Fig. 1). Since the efficiency of homogenization by concerted evolution depends on the physical distance between repeated units (Dvorak et al., 1987), it is possible that conformers of neighbouring units may be structurally more similar to each other than conformers of distant units. The satellites investigated herein occupy, on average, at least 20 chromosomal loci (Fig. 7; Lim et al., 2004, 2005). Consequently, each locus could be composed of units with a distinct curvature, shaping higher-order superhelix structures and influencing the tight packing of DNA and proteins in heterochromatin (Rouleux-Bonnin et al., 2004; Bussiek et al., 2009). Further experiments are required to clarify these issues.

Do subtelomeric satellites contribute to the stabilization of allopolyploid genomes? An hypothesis

Satellite repeats significantly contribute to chromosome divergence as a result of their high abundance and rapid evolution (Plohl et al., 2008). The diploid-like meiotic behaviour of allopolyploids is thought to result from the divergence between homeologous chromosomes, which may already exist and/or be accentuated at the onset of polyploid formation (Le Comber et al., 2010) and involve the rearrangement of large chromosome fragments, or from the activity of Pairing homeologous (Ph) genes (reviewed in (Jenczewski & Alix, 2004; Cifuentes et al., 2010). Indeed, many allopolyploids combining sufficiently similar genomes show intergenomic translocations at their early phase of evolution (Osborn et al., 2003; Lim et al., 2008), while allopolyploids combining divergent genomes appear to maintain intact parental chromosomes over a long evolutionary period of time (Kotseruba et al., 2003; Lim et al., 2005, 2007a,b). However, there is no strict consensus on the contribution of the phylogenetic divergence of parents to the successful formation of allopolyploid species (Buggs et al., 2008; Paun et al., 2009). Historically, Grant (1981) emphasized the role of chromosomal repatterning rather than phylogenetic distance per se. Here, we propose that structural divergence at the critical parts of chromosomes, such as the telomeres and centromeres, is important for stabilization of the allotetraploid nucleus, whereas the divergence of other parts of chromosomes may not play a role or may be less important. The following supports this hypothesis. In Nicotiana, three relatively recent allotetraploids exist, N. arentsii, N. rustica and N. tabacum, which are all believed to have originated < 200 000 yr ago (Clarkson et al., 2004). The phylogenetic distance between progenitor species decreases in the following order: N. tabacum > N. rustica > N. arentsii. Up to nine intergenomic translocations were identified in tobacco (Kenton et al., 1993; Moscone et al., 1996), whereas none were observed in N. rustica or N. arentsii (Lim et al., 2004), suggesting that tobacco has the least stable genome. The progenitors of N. arentsii and N. rustica evolved bulky subtelomeric satellites that were faithfully transmitted to the allotetraploids derived (this study; Lim et al., 2004). These megabase-sized subtelomeric clusters, which significantly differ in sequence or conformation, might contribute to the stabilization of parental chromosomes by preferential recognition of homologous chromosomes in allopolyploid meiosis. By contrast, the lack of subtelomeric satellites in one of the tobacco subgenomes (T), as evidenced by cytogenetic (Kenton et al., 1993), molecular (Horakova & Fajkus, 2000) and genomic (Renny-Byfield et al., 2011) studies, may stimulate homeologous pairing, possibly explaining the frequent intergenomic translocations in the tobacco genome. The second argument supporting a stabilizing role of subtelomeric satellites is based on the differential behaviour of subtelomeric satellites and rDNA repeats. It is known that rDNA, unlike satellite repeats, was largely homogenized in all natural Nicotiana allotetraploids, largely irrespective of age (Kovarik et al., 2004). Chromosome arms bearing active rDNA loci (NORs) at subtelomeric positions usually lack subtelomeric satellites (Lim et al., 2000). Thus, an absence of subtelometric satellites may render NOR-bearing chromosomes more vulnerable to genetic interaction (Kovarik et al., 2008).

We conclude that the positions of divergent (in primary structure or conformation) satellite repeats at critical chromosomal sites such as telomeres are likely to positively influence the survival time of parental chromosomes in the allopolyploid nucleus. Certainly, over longer evolutionary time periods (1 million yr), translocations of satellites to alien chromosomes can occur (Koukalova et al., 2010). In the future, it will be interesting to test this hypothesis in other natural and synthetic systems.


The authors wish to thank Dr Elizabeth McCarthy (University of London, UK), Dr Blazena Koukalova and Hana Srubarova (both from the Academy of Sciences, Czech Republic) for critical reading and suggestions for improving the manuscript. We further thank Dr Martina Talianova (Academy of Sciences) for help with the Bayesian statistics. This work was funded by the Grant Agency of the Czech Republic (206/09/1751, P501/10/0208), MSMT (LC06040) and the Academy of Sciences of the Czech Republic (AVOZ50040507 and AVOZ50040702).