Recent Spread of a Retrotransposon in the Silene latifolia Genome, Apart From the Y Chromosome

Abstract

Transposable elements often accumulate in nonrecombining regions, such as Y chromosomes. Contrary to this trend, a new Silene retrotransposon described here, has spread recently all over the genome of plant Silene latifolia, except its Y chromosome. This coincided with the latest steps of sex chromosome evolution in this species.

TRANSPOSABLE elements (TEs) are ubiquitous in pro- and eukaryotic genomes. Plant genomes appear to be particularly littered with various families of retrotransposons, the elements that transpose via an RNA intermediate. For example, >75% of the maize genome and large proportions of genomes of other crops are composed of retrotransposons (Meyers  et al. 2001; Schulman and Kalendar 2005).

The evolutionary dynamics of TEs is dominated by periods of active transposition (Bingham  et al. 1982; Biemont  et al. 1994; Petrov  et al. 1995; Nuzhdin  et al. 1997; Naito  et al. 2006; Bergman and Bensasson 2007), which may lead to dramatic expansion of copy number of the particular element and even to a significant increase in genome size (Neumann  et al. 2006; Hawkins  et al. 2008). The periods of active transpositions are followed by periods of inactivity that may eventually lead to extinction of the particular transposable element. Such “boom and bust” cycles of TE activity may result from a combination of several factors. One reasonably well-documented factor is invasion by a new TE of a species that previously lacked it (Daniels  et al. 1990; Kidwell 1992; Lohe  et al. 1995; Clark and Kidwell 1997; Silva and Kidwell 2004; Sanchez-Gracia  et al. 2005; Diao  et al. 2006). Over 100 horizontal transfer events of TEs of various kinds have been reported for Drosophila alone (Loreto  et al. 2008). The mechanisms of TE horizontal transfer are not known, but at least for long terminal repeat (LTR) retrotransposons it can be hypothesized that horizontal transfer occurs via the mechanisms similar to that of retroviral infections, as LTR retrotransposons are very closely related to retroviruses and are capable of forming virus-like particles (Miyake  et al. 1987).

We describe a relatively recent “burst” of transposition activity of a new LTR retrotransposon in the plant Silene latifolia (Caryophyllaceae) and its close relatives. One of the most interesting features of this species is the presence of sex chromosomes that determine whether the plant develops as a male with XY chromosomes or a female with XX sex chromosomes (Westergaard 1958). The sex chromosomes in Silene evolved relatively recently, probably ∼10⁷ years ago (Filatov and Charlesworth 2002) within a small cluster of dioecious Silene species, (section Elisanthe), while the rest of the genus is mostly nondioecious. S. latifolia sex chromosomes apparently evolved from a single pair of autosomes that ceased recombining with each other along most of their length and began to diverge (Filatov 2005a). This process occurred in at least three steps (Nicolas  et al. 2005; Bergero  et al. 2007), resembling “evolutionary strata” discovered on the human sex chromosomes (Lahn and Page 1999). The X/Y silent divergence in the oldest “stratum” on the S. latifolia sex chromosomes is only ∼15–24% and the youngest stratum shows only 2–4% divergence (Bergero  et al. 2007). Assuming the substitution rate of 1.05 × 10⁻⁸ per silent site per year (Derose-Wilson and Gaut 2007), the age of the oldest part of the S. latifolia sex chromosomes might be ∼10 million years old and the youngest region may be as young as 2 million years old. However, this estimate is rough, as substitution (and mutation) rates may vary between genes and between species and the exact rate of molecular clock in S. latifolia is not known. The S. latifolia X and Y still pair and recombine in male meiosis in a small pseudoautosomal region (PAR) (reviewed in Armstrong and Filatov 2008).

Evolution of sex chromosomes is expected to lead to degeneration of Y-linked genes due to lack of recombination that slows down adaptive evolution (Rice 1987) and exacerbates the processes of genetic hitchhiking, background selection, and Muller's ratchet that lead to accumulation of deleterious mutations and gene loss (reviewed in Charlesworth 1991, 2008). Accumulation of TE insertions and other repetitive DNA is also a typical feature of Y chromosomes as in degenerate Y chromosomes TE insertions are less likely to damage important genes and lack of recombination prevents ectopic exchange between the TE copies (Charlesworth  et al. 1994; Abe  et al. 2005; Steinemann and Steinemann 2005).

Despite its relatively recent origin, S. latifolia Y chromosome shows some signs of genetic degeneration, including drastic reduction of DNA polymorphism (Filatov  et al. 2000; Ironside and Filatov 2005), accumulation of mutations at sites that are known to be important for the functional activity of proteins (Filatov 2005b), and gene expression changes (Marais  et al. 2008). Accumulation of repetitive DNA on the S. latifolia Y chromosome has also been reported (Hobza  et al. 2006; Kejnovsky  et al. 2006; Kazama and Matsunaga 2008). Below we describe the opposite trend—accumulation of a transposable element all over the S. latifolia genome, except the Y chromosome. This unusual genomic distribution of the transposable element provides a useful “negative paint” for the S. latifolia Y chromosome and allows visualizing regions of the Y chromosome that have different evolutionary histories.

Identification and genomic distribution of a new retrotransposon:

As a by-product of a continuing effort to isolate more S. latifolia sex-linked genes, we identified a fosmid clone that gave an unusual FISH signal. The FISH signal was frequent and widely distributed, suggesting the presence of an abundant repeat in this fosmid. However, there was considerably less hybridization to the majority of the Y chromosome. To investigate the cause of this peculiar distribution, we subcloned fragments of the fosmid and used the subclones as probes for FISH. We identified several subclones that resulted in a hybridization pattern similar to the original fosmid (Figure 1, a and b, and, supplemental Figure S1). These clones were sequenced and blast searched against the nr NCBI database. This revealed similarity to various retrotransposons, with the closest match to the Vicia retrotransposon Ogre (GenBank accession AY936172). One of these clones (clone 4.2), contained an open reading frame with a strong match to Ogre reverse transcriptase, and this clone was used for further analyses. Comparative phylogenetic analysis of the protein sequences of the reverse transcriptases of various retrotransposons confirmed that the newly identified element clusters with Ogre retrotransposons from other plant species (supplemental Figure S2). According to the naming convention proposed by Capy (2005), we named the newly identified S. latifolia retrotransposon SlOgre1.

Figure 1.—

Fluorescence in situ hybridization (FISH) with Silene chromosomes counterstained with DAPI (blue). (a–d) FISH with SlOgre1, clone 4.2 (green signal) on metaphase 1 spreads of (a) S. latifolia, (c) S. dioica, and (d) S. marizii. White arrow indicates the XY bivalent. (b) S. latifolia, with green filter only, showing the extent of hybridization at the PAR end of the Y chromosome (white arrowhead). (e) Mitotic metaphase of S. vulgaris probed with SlOgre1, clone 4.2, no hybridization detected. (f–j) Dual probe FISH with S. latifolia chromosomes. SlOgre1, clone 4.2, (green) and SlCypY gene (red). Open (black) arrowhead indicates the PAR region and open arrow points to SlCypY signal on the Y chromosome. (f–h) Mitotic metaphase. (i) Partial mitotic metaphase. (j) Partial meiotic metaphase 1. (f and g) Sequential probing with images merged, some overlap of autosomes and Y is visible. (h–j) Simultaneous probing with clone 4.2 and SlCypY probes. (j) Note SlCyp (red) signal on the X chromosome; due to low X/Y divergence in this gene the SlCypY probe hybridizes with the X-linked homolog of SlCypY gene. Bar, 10 μm is shown at the bottom of each panel. For details of in situ hybridization procedure see legend of supplemental Figure S1.

A plausible explanation of the apparent absence of the SlOgre1 from the Y chromosome may be transposition of that element in S. latifolia females only. Such sex specificity is not uncommon among TEs. For example, the Drosophila melanogaster retrotransposon copia is preferentially transcribed and transposed in male germinal tissues, while another D. melanogaster TE, Doc, transposes only in female germinal tissues (Filatov  et al. 1998; Pasyukova  et al. 1998). In fact, TE expression and transposition can be very specific, depending on various factors such as stress (Grandbastien  et al. 2005) or age (Filatov  et al. 1998).

To test whether there is a difference in transcript abundance between the sexes, we conducted PCR amplification of reverse transcribed total RNA from flower buds and leaves of S. latifolia males and females, but no amplification of the SlOgre1 sequence was observed in either sex. A fragment of SlssX/Y gene was successfully amplified from the same reverse transcribed RNA samples with primers Slss+1 and Slss−2 (Filatov 2005b), which was used as a control for RNA and cDNA quality (data not shown). This region of the SlssX/Y gene contains a short (240-bp-long) intron in genomic DNA, but not in the PCR product amplified from the reverse transcribed RNA samples, demonstrating that the PCR product could not have been a result of contamination of cDNA with genomic DNA. This is consistent with the view that SlOgre1 may not be active any more (see below). The negative result with RT–PCR of testing for SlOgre1 expression does not necessarily mean that this element is completely inactive; it could still be expressed at a low level in some tissues. This has not been investigated further, as the activity/inactivity of the element is not the central question of this work and does not affect our conclusions.

To investigate the distribution of SlOgre1 in S. latifolia relatives, we conducted FISH with clone 4.2 in three other dioecious Silene species that also belong to section Elisanthe, S. dioica, S. diclinis, and S. marizii, as well as with nondioecious S. vulgaris. The distribution of FISH signal in the dioecious species is very similar to that in S. latifolia: the probe paints the autosomes and the X chromosome, but leaves the majority of the Y chromosome “unpainted” (Figure 1, c and d). These species are very closely related and are likely to have separated within the last 1–2 million years, since nuclear DNA divergence at silent sites does not exceed 2% (Ironside and Filatov 2005; Laporte  et al. 2005; Muir and Filatov 2007). Therefore, this result is not entirely unexpected. On the other hand, there is no signal on S. vulgaris chromosomes (Figure 1e), suggesting that SlOgre1 is not present in multiple copies in S. vulgaris. Alternatively, the lack of FISH signal in S. vulgaris could be due to divergence between the S. latifolia and S. vulgaris sequences (∼15% for silent sites, see Filatov and Charlesworth 2002), precluding effective in situ hybridization. However, the low divergence between the copies of the element present within S. latifolia (see below) suggests that SlOgre1 started to actively transpose quite recently in S. latifolia and its relatives (or their ancestor), providing additional support to the view that this TE has not been widespread in the S. vulgaris genome.

The recent spread of the retrotransposon:

The relatively recent origin of S. latifolia sex chromosomes and the lack of accumulation of SlOgre1 on the majority of the Y chromosome provide an evolutionary perspective on the dynamics of the SlOgre1 retrotransposon spread. As the S. latifolia Y chromosome recombined with the X chromosome before cessation of recombination on the Y chromosome, the Y chromosome should contain approximately the same number of copies of the retrotransposon, unless either the copies have been specifically removed from the Y or the transposable elements started to actively spread after the X and Y stopped recombining. To estimate more precisely the timing of spread of SlOgre1 in the genomes of dioecious Silene section Elisanthe, we measured divergence between the copies of SlOgre1 in four Elisanthe species. For this purpose we PCR amplified, cloned, and sequenced a 1.2-kb-long fragment of the SlOgre1 elements from one male individual from each of the four closely related dioecious Silene species: S. latifolia, S. dioica, S. marizii, and S. diclinis. An attempt to amplify the retrotransposon from S. vulgaris with three different pairs of primers was unsuccessful, which is in line with lack of hybridization of the element with S. vulgaris chromosomes.

The phylogeny of the 90 sequenced copies of SlOgre1 elements from four species is shown in Figure 2 and supplemental Figure S3. Interestingly, the sequences do not cluster by species, which supports the view that this element spread before divergence of these species. However, gene flow between these species may also create the same effect. All crosses among S. latifolia, S. dioica, S. marizii, and S. diclinis yield viable and fertile progeny, except S. latifolia × S. diclinis cross, which often yields only few or no viable progeny (Prentice 1978). S. marizii and S. diclinis are Iberian endemics with fairly restricted ranges that do not overlap (Prentice 1976, 1977). Only S. latifolia is present in the ranges [but not in the same locations (D. A. Filatov, personal observation)] of the two Iberian endemics, but hybrids between these species have not been reported in the wild. S. latifolia and S. dioica are widely distributed across Europe but inhabit somewhat different habitats: S. latifolia is common in open fields and along the roads, while S. dioica prefers more shady and wet forest habitats. The two species are known to form hybrids in contact zones, which are often habitats disturbed by human activity (Backer 1948). Thus, it seems likely that until very recently S. latifolia and S. dioica have not actively exchanged genes. Indeed, our multigenic study of DNA polymorphism and divergence between these two species has not provided any evidence for historical long-term gene flow (G. Muir, A. Harper and D. A. Filatov, unpublished results). Thus, lack of species clustering of the SlOgre1 copies is probably due to active transposition prior to speciation in Silene section Elisanthe rather than due to gene flow between these species.

Figure 2.—

Neighbor-joining tree of SlOgre1 sequences from four closely related dioecious Silene species. The branch lengths reflect Jukes–Cantor (Jukes and Cantor 1969) nucleotide distances between the sequences (all sites). For details of PCR amplification and sequencing conditions see legend of supplemental Figure S3. All sequences were deposited into GenBank under accession nos. FJ531702–FJ531791.

The overall shape of the tree suggests that the retrotransposon has spread across the genome in a relatively short period of time and may not be active any more—the internal branches of the tree are quite short, while the external branches are relatively long, so the tree resembles a multirayed star (Figure 2). Fu and Li's D* statistic (Fu and Li 1993) that effectively compares the lengths of external and internal branches of the tree is significantly negative (D* = −4.4, P < 0.02) confirming that the external branches are stretched, relative to the internal branches. If SlOgre1 elements were still actively transposing, one would expect to see a more balanced tree, with many relatively recent branching points (Brookfield 2005; Brookfield and Johnson 2006), which is not the case.

The presence of multiple in-frame stop codons and frameshift deletions also agrees with the lack of recent transposition activity: almost half of the sequences from all four species (40 of 90) contained at least one stop codon in the open reading frame (ORF). Of the remaining 50 sequences without in-frame stop codons, 8 contained frameshift deletions. Given that the sequenced region represented only a small fraction of the element length, it seems likely that most copies of the retrotransposon might have damaged ORFs or encode nonfunctional proteins.

To estimate the age distribution of individual SlOgre1 copies we used the length of external branches for synonymous sites as a proxy for the age of individual retrotransposon copies. The distribution of branch length in all four species peaks at ∼4.5% (Figure 3) and the distributions are not significantly different among the four species (Kolmogorov–Smirnov two sample tests, P > 0.14; also see supplemental Figure S4). Assuming a substitution rate of 1.05 × 10⁻⁸ per site per year (Derose-Wilson and Gaut 2007), the transposition activity of SlOgre1 peaked ∼5 million years ago. However, the true substitution rate of TEs in S. latifolia is not known and may be higher than for other genomic regions as the RNA intermediate and reverse transcription step in the retroelement life cycle may result in a higher mutation rate for such elements, compared to the rest of the genome. In that case the peak of SlOgre1 transposition activity would be more recent.

Figure 3.—

Age distribution of SlOgre1 insertions. Frequency histogram of terminal branch length for synonymous sites (Nei–Gojobori distance (Nei and Gojobori 1986) with Jukes–Cantor correction (Jukes and Cantor 1969), calculated in MEGA4 (Tamura  et al. 2007) in four closely related dioecious Silene species. The external branch lengths were manually typed into Microsoft Excel and the histogram of branch lengths was created using the “histogram” option in the data analysis pack of Excel 2007.

The recent spread of SlOgre1 in Silene section Elisanthe may have followed a horizontal transfer of this element from an unknown plant species. The closest known SlOgre1 relatives are present in legumes, but these sequences are too diverged from SlOgre1 (supplemental Figure S2) to be considered as immediate ancestors. Ogre elements are closely related to the retrotransposon gypsy, which is effectively a retrovirus and has an env gene, encoding a product essential for the ability of the virus-like particles to leave the host cells and infect other cells. Indeed, infectivity of gypsy elements has been demonstrated in Drosophila (Kim  et al. 1994).

The spread of SlOgre1 and sex chromosome evolution in Silene:

Different sections of the S. latifolia Y chromosomes ceased to recombine with the X at different times in the past, creating evolutionary strata (Filatov 2005a; Nicolas  et al. 2005; Bergero  et al. 2007). If SlOgre1 transpositions occur only in females, then this TE is expected to be absent only from the strata on the Y that are older than the time of the element spread. Younger strata will have been pseudoautosomal at the time of most active transpositions of the element; hence they might contain as many copies on the Y chromosome as the corresponding region on the X. To test whether the SlOgre1 is most abundant in the youngest stratum of the S. latifolia Y chromosome, we hybridized preparations of chromosomes with two probes, the clone 4.2 and a probe to a Y-linked gene SlCypY. The divergence between the SlCypX and SlCypY genes is only 4% at silent sites, placing this gene into the youngest stratum (Bergero  et al. 2007). However, it is clear from Figure 1, f–j, that the SlOgre1 FISH signal does not spread as far as the SlCypY gene. This may suggest that the spread of the SlOgre1 element is more recent than the cessation of recombination between the X and Y chromosomes in the most recent stratum. The weak signal of SlOgre1 hybridization with the Y chromosome (Figure 1, a–c) may be due to cross-hybridization with other repeats on the Y, or to the presence of fragments or full-length copies of SlOgre1 on the Y chromosome (perhaps transposed to the Y by other TEs).

Alternatively, the position of the SlCypY gene may not be as close to the PAR as expected on the basis of the position of its X-linked homolog in the genetic map of the X chromosome (Bergero  et al. 2007). The order of S. latifolia Y-linked genes may have been changed due to multiple rearrangements (Bergero  et al. 2008). Such rearrangements have been reported for the Y chromosomes of other species, e.g., while the order of genes on the human X chromosome corresponds to that expected from the evolutionary strata model, the order of the genes on the Y chromosome have been considerably altered (Lahn and Page 1999; Graves 2006). According to our FISH mapping, the SlCypY gene, is located close to PAR of the Y chromosome (Figure 1, f–j) whereas deletion mapping of the Y chromosome tentatively placed SlCypY on the opposite arm (Bergero  et al. 2008). This can be explained if different accessions have rearrangements of the order of Y-linked genes, for which Bergero  et al. (2008) have provided evidence.

Although it is not completely clear how far into the nonrecombining portion of the Y chromosome the SlOgre1 element has spread, this element will still be useful as a negative paint for the dioecious Silene section Elisanthe Y chromosome. For example, it may help to resolve the controversy regarding the presence of a second PAR in S. latifolia. According to an AFLP map of S. latifolia sex chromosomes, recombination between X and Y occurs on both arms of the Y chromosome (Scotti and Delph 2006), suggesting the presence of a second PAR. We could not see any evidence for a second PAR in our accession (IL25, Kidderminster, United Kingdom) using clone 4.2 as a FISH probe, but it is possible that accessions may differ in this respect. This may be tested by analyzing more accessions with the negative paint for the Y chromosome reported in this article. It may also be feasible to use this probe to estimate the relative size of the PAR in S. latifolia. On the basis of our FISH results with the clone 4.2 probe we suggest that the S. latifolia PAR is much less than 10% of the Y length; however, it is difficult to be more precise. Electron microscopy analysis of the PAR region may yield more precise estimates.

Footnotes

Communicating editor: D. Charlesworth

Acknowledgements

We are grateful to Thomas Meagher for providing S. marizii seeds and to Deborah Charlesworth and two anonymous reviewers for helpful comments. The work was funded by a grant from the Biotechnology and Biological Sciences Research Council to D.A.F. and S.J.A.