MITE insertions in Silene latifolia genome, p. 1 Active miniature transposons from a plant genome and its non-recombining Y chromosome

Mechanisms involved in eroding fitness of evolving Y-chromosomes have been the focus of much theoretical and empirical work. Evolving Y-chromosomes are expected to accumulate transposable elements (TEs), but it is not known whether such accumulation contributes to their genetic degeneration. Among TEs, miniature inverted-repeat transposable elements are non-autonomous DNA transposons, often inserted in introns and untranslated regions of genes. Thus, if they invade Y-linked genes and selection against their insertion is ineffective, they could contribute to genetic degeneration of evolving Y-chromosome. Here, we examine the population dynamics of active MITEs in the young Y-chromosomes of the plant Silene latifolia and compare their distribution with those in recombining genomic regions. In order to isolate active MITEs, we developed a straightforward approach based on the assumption that recent transposon insertions or excisions create singleton or low-frequency size-polymorphisms which can be detected in alleles from natural populations. Transposon display was then used to infer the distribution of MITE insertion frequencies. The overall frequency spectrum showed an excess of singleton and low-frequency insertions which suggest that these elements are readily removed from recombining chromosomes. In constrast, insertions on the Ychromosomes were present at high frequencies. Their potential contribution to Ydegeneration is discussed. MITE insertions in Silene latifolia genome, p. 4 INTRODUCTION Transposable elements (TEs) have major roles in genome diversification and expansion. Due to their ability to self-replicate, they can proliferate and reach high copy numbers and, if fixed, can be retained in evolutionary lineages across wide taxonomic groups. However, abundance of TEs in eukaryotic genomes varies over several order of magnitude (WRIGHT and FINNEGAN 2001) and the factors that control their population dynamics are not yet completely resolved. First, TE insertions can be highly deleterious and selective pressures oppose their insertion, and thus accumulation. Deleterious effects on fitness derived from TE insertions are of two main types: insertions within or near genic regions can disturb gene functions by changing reading frames or disrupting regulatory motifs (FINNEGAN 1992), and chromosomal rearrangements can be caused by ectopic exchange between TE copies at non-homologous genomic locations (LANGLEY et al. 1988; MONTGOMERY et al. 1987). In sexually dimorphic organisms whose gender is controlled by sex chromosomes, recombination is suppressed between the Y-chromosome, inherited only by male individuals, and the homologous X-chromosome (or the Z and W chromosomes in species with female heterogamety). Because Y-chromosomes are recombinationally isolated, TE dynamics can be studied in a non-recombining chromosome in an otherwise recombining background. Y-chromosomes should not experience the deleterious effect of TEs causing chromosomal rearrangements due to ectopic exchanges, because meiotic recombination is suppressed. Thus the main deleterious effect of TEs on fitness of Ychromosomes should be insertions affecting functionally important sequences. MITE insertions in Silene latifolia genome, p. 5 Among TEs, miniature inverted-repeat transposable elements (MITEs) are a class of DNA transposons that move by the trans-activity of a transposase encoded by a related transposable element (ZHANG et al. 2001). Unlike other major classes of TE, MITEs are preferentially located in or near genes (BUREAU and WESSLER 1994; FESCHOTTE et al. 2002), and, most likely because of their small size (~100-500 bp), insertions often do not cause major disruption of the genes or their regulation (NAITO et al. 2006). However, some insertions could be highly deleterious (YANO et al. 2000) and if they occur within Y-linked genes, they could contribute to genetic degeneration of an evolving Ychromosome. On Y chromosomes, selection is expected to be ineffective, since recombination is suppressed. Y chromosomes are thus expected to have low effective population size, Ne, due to the “hitch-hiking” effects of selection (selective sweeps, background selection and weak Hill-Robertson interference, reviewed by (CHARLESWORTH and CHARLESWORTH 2000)). This expectation is supported by empirical data showing low silent site diversity of Y-linked genes, compared with their X-linked alleles (MONTELL et al. 2001; ZUROVCOVA and EANES 1999). Reduced Ne should lead to lower efficacy of natural selection, so that mildly deleterious mutations, including MITE insertions (BROOKFIELD and BADGE 1997), should be able to rise to intermediate frequencies, or to fixation. Moderately deleterious MITE insertions may thus contribute to Y chromosome genetic degeneration. It is well known that Y chromosomes and neo-Y-chromosomes undergo genetic degeneration in the long term (CHARLESWORTH and CHARLESWORTH 2000), and that TEs can quickly accumulate on neo-Y-chromosomes (BACHTROG 2003). There are so far no MITE insertions in Silene latifolia genome, p. 6 empirical data on the dynamics of MITEs in newly evolving Y-chromosomes. We here examine the distribution of MITEs in a dioecious plant species, Silene latifolia, whose sex-chromosome system is not older than 5-10 MY. Silene Y-chromosomes are 40% larger than their homologous X-chromosomes, but the two sex chromosomes carry a number of homologous genes (BERGERO et al. 2007), so that it is unlikely that the larger size of the Y is due to a major autosomal translocation (forming a neo-sex chromosome, (STEINEMANN and STEINEMANN 1998)); most likely repetitive DNA has accumulated. For the few sex-linked gene pairs so far known, S. latifolia Y-linked genes also show lower expression levels than their X-counterparts (unpublished work of R. Bergero), suggesting that genetic degeneration is occurring in this species. To study MITE dynamics, sequences of actively transposing elements are needed. These can be found by scanning complete genome sequences (or sequences of large genome regions) for TE insertions (DURET et al. 2000; RIZZON et al. 2002; SURZYCKI and BELKNAP 2000; WRIGHT et al. 2003), but such extensive genomic sequences are difficult to obtain from DNA regions rich in repetitive sequences, such as the Y-chromosomes (FOOTE et al. 1992; HOLT et al. 2002), or are simply not available from non-model species. Furthermore, genome-scan approaches provide no information about the distribution of insertion frequencies, which is needed to test the predictions of the theories outlined above, and it is not always clear whether any given transposable element has recently been active. Indeed, a large fraction of TEs in genomes of higher eukaryotes are probably inactivated copies (fossils) which have lost transposition activity (FESCHOTTE et al. 2002; PACE and FESCHOTTE 2007; SMIT and RIGGS 1996). MITE insertions in Silene latifolia genome, p. 7 To search for actively transposing MITEs we developed an approach based on assuming that recent transposon insertions or excisions create singleton or low-frequency size polymorphisms, which can be detected in surveys of alleles from natural populations. We used this approach to isolate MITE elements from S. latifolia introns, and identified two active subfamilies from this dioecious plant. Transposon display for MITE insertions from both subfamilies was carried out in order to infer their frequency distributions and compare Y-chromosomes with other genome regions sampled from S. latifolia natural populations. MATERIAL & METHODS Plant material One male and one female S. latifolia individuals from each of eight European natural populations (Sup. Table I) were used to investigate intron-size polymorphisms. A transposon display was carried out on 48 individuals (24 females and 24 males) derived from a larger set of natural populations (Supp. Table 1) and a collection of 108 F2 plants. The F2 family derived from a single cross between two F1 plants obtained by crossing two parents, one obtained from a French population, and one from the Netherlands ((BERGERO et al. 2007). As these parent plants were not inbred, the elements used as genetic markers are sometimes heterozygous in one parent, and sometimes in both (other MITE insertions do not segregate in this family). Genomic DNA was obtained from fresh leaves using the FastDNA kit (Q-Biogen) following the manufacturer’s instructions. MITE insertions in Silene latifolia genome, p. 8 Identification of MITEs from intron-size polymorphisms Introns from a set of 19 genes (two introns were analysed per each locus, see Sup. Table 2) were amplified by PCR and size-estimated by standard gel electrophoresis in alleles from natural populations in order to search for large size-polymorphisms (>150 bp) that could result from recent MITE insertion/excision. Loci were chosen to be single-copy or low-copy in order to limit amplification of paralogous genes, which could hinder interpretation of the results. The set also included the Silene sex-linked genes SlSSX/Y, SlX3/Y3, SlX1/Y1 and SlCypX/Y. Primers were designed based on S. latifolia cDNA sequences. Intron positions were inferred according to gene structures reported for putative Arabidopsis thaliana and Oryza sativa orthologues (BERGERO et al. 2007). PCR products from introns showing size-polymorphisms were cloned in a T-tailed pBSKS+ vector (MARCHUK et al. 1991) and sequenced on an ABI3730 sequencer (Applied Biosystem). Alignment of intron size variants was done using the package Sequencher 4.7 (GeneCodes). As MITEs lack tranposase coding sequences, other features were used for their identification. These were the presence of terminal inverted repeats (10-15 bp) at the ends of the insertion, target site duplication (TSD) as reported for other MITEs (WICKER et al. 2007), size in the range of 150-500 bp and extensive secondary structure. The web package MFOLD (ZUKER 2003) was used to infer the DNA folding and secondary structure of putative MITE elements. MITE insertion variants in a mapping family and in natural populations Transposon display (TD) was used to detect segregating MITE elements in a F2 family, and MITE polymorphisms and copy numbers in a set of natural populations. TD is an MITE insertions in Silene latifolia genome, p. 9 AFLP-based technique (CASA et al. 2000; VAN DEN BROECK et al. 1998) which uses a primer annealing to the adapter and one annealing to conserved regions of the TE element. Although MITEs do not have conserved coding sequences, extensive sequence conservation should occur in members of recently active MITE subfamilies. MITEspecific primers were designed to face outwards from and anneal to sub-terminal sequences of the two MITE elements isolated from S. latifolia (SlTo1 and EITRI). The procedures were as outlined in CASA et al. (2004) with the following modifications. Genomic DNA (0.8-1.5 μg) was digested with DpnII for 3h at 37 °C. After inactivation of DpnII by incubation at 65 °C for 10 min, restriction fragments were ligated to a DpnII linker. The ligation reactions containing T4 DNA ligase 200 U (NEB), BamHI 20 U, 12 μM of DpnII adapter, were performed for 12 cycles, each consisting of 30 min at 16 °C and 10 min at 37 °C. The DpnII adapter was obtained by spontaneous annealing at RT of the oligonucleotides TDADA1 (5'-GACAGTTGTGTACCTCGAATG-3') and TDADA2 (5'-GATCCATTCGAGGTACACAACTG-3'). Adapter dimer formation in the ligation reaction (which could significantly decrease the availability of adapters for ligation with genomic DNA) was avoided by designing a 5' GA overhang at one end of the adapter, and the formation of an ex-novo BamHI site at the other end when adapter dimers formed. Dimers were destroyed by adding BamHI to the ligation reaction. The DpnII library was amplified by a first round PCR of 20 cycles using adapterspecific primers (with one specific base) and the MITE-specific external primer (EITRIext, 5'-TAAATAACGTGTCCCGTGTCC-3' and SlTo1-ext, 5'TCCATTCCAATCCATTCCAAGAG-3'). Thermocycling conditions were: 4 min at 94 °C; 20x (40s at 95 °C, 40s at 50 °C, 30s at 72 °C); 5 min at 72 °C. One microliter of the MITE insertions in Silene latifolia genome, p. 10 PCR reaction was used as template in a nested touchdown PCR (DON et al. 1991), using a set of adapter-specific primers with 2 selective bases at the 3' end (a total of 10 adapterspecific primers out of 16 possible combination were used) and a VIC-labelled MITEspecific internal primer ( EITRI-int, 5'-TCCCGTGTCCTAAATTTCATG-3' and SlTo1int, 5'-GAGAGCAAACCAAACACCCC -3'), with the following thermocycling conditions: 2 min at 95 °C, followed by 10 annealing cycles at 0.5 °C decreasing and 18 cycles at 50 °C. The selective amplification was carried out using a hot-start TAQ polymerase (JumpStart, Sigma), which was found to increase product yields, especially for bigger-size bands (>300 bp), and to produce better electropherograms. VIC-labelled PCR products were firstly diluted (1:25) in distilled water, further diluted (1:10) in formamide containing the size standard GeneScan 500-LIZ (Applied Biosystems) and directly separated by capillary electrophoresis on an ABI-3730 DNA analyzer (Applied Biosystem). Transposon insertions were scored by using the software Genemapper v. 3.7 (Applied Biosystem), and their frequencies analysed and plotted using the R statistics package (http://www.r-project.org). MITE sequences from this study were deposited in the GenBank databases (accession nos. EU334132 and EU334133). Data Analyses The average numbers of MITE insertions per haploid genome were estimated from electropherograms obtained from the set of 24 females (thus excluding Y-linked TE insertions whose properties may be unusual). Because of a preponderance of rare insertions, an estimation of insertion frequencies can be obtained from the observed MITE insertions in Silene latifolia genome, p. 11 frequencies of the recessive genotypes (nulls) of each insertion. To obtain the frequency, pi, of insertions at the i th site, we assumed that the population is in Hardy-Weinberg equilibrium, since the species is outcrossing, and used pi = 1-√qi 2 where qi 2 is the frequency of the recessive genotype at this site (with no band on the gel). The expected number of TE insertions per haploid genome is the total number of insertions weighted by their frequencies ‹Ntot› = ∑pi. Indels in the intervening sequence between a MITE insertion and the restriction site will create a new “spurious” polymorphic TE insertion. We estimated the proportion of the observed polymorphisms due to such events by examining segregation of 203 MITE insertions in a F2 progeny made by crossing two F1 individuals from a cross between two outbred natural populations. Pairs of transposon display bands that segregated as alternatives in repulsion were counted as probable insertions that are descended from a single ancestral insertion, but which have undergone indel events since the insertion. RESULTS Intron insertion polymorphisms due to MITE activity A total of 38 primer pairs amplifying intronic regions from 19 loci were used to search for large insertion polymorphisms in European populations of S. latifolia. Sixteen plants were included in the survey, and five of the 38 introns showed size polymorphisms, with size differences larger than 150 bp (Table 1). PCR amplification of intron 5 from a S. latifolia gene (SlAnk) produced a ~ 600 bp amplicon in all individuals, but a single male plant appeared to be heterozygous, having an additional larger PCR product (~1000 bp, Fig 1). A single 407-bp insertion delimited MITE insertions in Silene latifolia genome, p. 12 by 11-bp terminal inverted repeats (with the sequence 5'-CTAGGTAGCAC-3') and 8-bp target site duplications (TSDs) was confirmed in the larger amplicon by sequencing. The TSD has the imperfect palindromic sequence 5'-CTCTTGAG-3'. Excluding the insertion, the 1-Kb product differed from its allelic counterpart in the same plant by one base substitution and a 3-bp indel. These data suggest that the long and short sequences are allelic. Presence of TIRs, TSDs and extensive secondary structure strongly suggests that this insertion is a MITE element. Classification of non-autonomous TEs relies on the TIR motifs and TSD sizes (WICKER et al. 2007). The size of the TSD (8-bp) suggests that this element is either a hAT or a P element, but the TIR motifs known from these two classes of elements were not found.We therefore identify it as the first MITE obtained from our study species, and, given its uncertain classification, we named it EITRI. In contrast to this singleton polymorphic insertion, a size polymorphism due to a singleton excision was observed in a sex-linked gene. PCR amplification of intron 2 of the recently described S. latifolia sex-linked gene SlCypX/Y (BERGERO et al. 2007) produced two amplicons (710bp and 1000 bp). Segregation analysis of these bands clearly showed Y linkage of the longer intron variant (1000 bp). Sequence alignment of these two variants shows an insertion of 290 bp, delimited by a 14-bp inverted repeat, and the inserted sequence exhibits the potential for extensive secondary structure (Fig. 2). Surprisingly, its TIR (5'-GGGGGTGTTTGGTT-3') matches perfectly the TIR region of a Tourist element (Zm20) isolated from Zea mays (BUREAU and WESSLER 1994) and TIR consensus sequence from 21 rice PIF families (ZHANG et al. 2004), probably because of high conservation of the catalytic domain of the transposase. Furthermore, a 3-bp TTA motif flanked this insertion, which is typical for TSDs of Tourist (mPIF) elements MITE insertions in Silene latifolia genome, p. 13 (JURKA and KAPITONOV 2001). We therefore classified this as a Tourist-like element and named it SlTo1. In a sample of eight Y-chromosomes from natural populations, we found a singleton excision with a clearly visible footprint in the SlCypY sequence alignment (Fig. 2). A search for this insertion in Cyp orthologues of other, closely related, dioecious Silene species revealed the same transposable element insertion in the Y chromosomes of S. diclinis and S. dioica; the sequence identity of these MITE insertions was estimated to be 99%. Although there are reported cases of independent insertions in the identical site (WALKER et al. 1997), the most parsimonious explanation for this MITE in the SlCypY gene is that these three sister species split after X-Y recombination stopped in this region, and that this Tourist insertion occurred before this time. This is consistent with the fact that the divergence between the X and Y copies (Ks = 6.1%, from (BERGERO et al. 2007)) is considerably higher than that between these species (Ks = 4.4%). MITE insertions in Silene latifolia genome, p. 14 We detected three further intron-size polymorphisms due to insertions of unknown origins (Table 1). All the polymorphic insertions from these three loci appear to be at low to intermediate frequencies, but none are singleton insertions or excisions. These insertions did not have inverted repeats, nor did analysis of their sequences suggest extensive secondary structure. Thus they are not recognizable MITEs. The origin of these large insertion variants is puzzling. Their intermediate frequencies in natural populations suggest that these are not recent insertions, and this is consistent with the absence of the conserved features of MITE sequences. They could represent relics of MITEs, deleted for part(s) of the sequence, including the TIRs (solo LTRs are known in other plant genomes, and are caused by deletions, DEVOS et al. 2002; 2004), or the TIR sequences may have become unrecognizable due to mutations. Transposon display of genomic and Y-linked MITE insertions A transposon display (TD) analysis of the two MITE elements, EITRI and SlTo1, was carried out on a set of 48 individuals collected from 24 European natural populations (Supplementary Table 1). Using the frequencies of null alleles (see Methods), we estimated an average of 230 copies of EITRI, and 130 copies of the SlTo1 element per haploid genome. Segregation of MITE insertions in a full-sib F2 family allowed us to recognize MITE insertions in the Y-chromosome. 23 EITRI elements and 16 SlTo1 elements showed a clear segregation pattern of complete Y-linkage. From the estimated average number of MITE insertions per Mb in the genome as a whole, we computed the predicted number of Y-linked insertions that should be found on a Y chromosome. Taking into account the physical size of the Y-chromosome (the largest S. latifolia chromosome, MITE insertions in Silene latifolia genome, p. 15 estimated to be 570-Mb, (SIROKY et al. 2001), and assuming a uniform distribution of TE insertions in the S. latifolia genome, there are significantly fewer MITE insertions in the Y chromosome than the expected copy numbers for both the EITRI and SlTo1 subfamilies (49 and 28 respectively, χ = 18.9, P< 0.0001). An excess of singleton or low-frequency MITE genomic insertions was observed from the TD analysis of the 24 female plants sampled from natural populations (Fig 3); only a small fraction (5%) of insertion sites were at medium or high frequencies (pi> 0.3). In contrast, the frequency spectrum for Y-linked insertions was markedly shifted toward high frequencies, with a remarkable paucity of singleton and low-frequency insertions (Fig. 3). Thus, the ratio of fixed to polymorphic Y-linked insertions was greater than the ratio of fixedto polymorphic genomic insertions, due to an excess of fixed Y-linked insertions (3/23 versus 1/1923, for the Y-linked and the other insertions, respectively). A χ test for independence, with Yates’ correction for continuity, showed that the two ratios were highly significantly different (χ = 113.99, p<0.000001).