Background

Heteromorphic sex chromosomes (XY/ZW) can often be distinguished from autosomes by the absence of recombination in at least a part of their length and Y/W chromosome degeneration [1, 2]. Plants with sex chromosomes have evolved rarely but repeatedly in many plant lineages, and sex chromosomes have reached various levels of differentiation [3]. In Asparagus officinalis and Carica papaya for example, X and Y chromosomes have diverged little and recombine along most of their length [4–6] whereas in Rumex acetosa and Silene latifolia, the sex chromosomes in males are largely non-recombining [7, 8]. In S. latifolia, X and Y chromosomes can recombine only in the regions known as pseudoautosomal regions (PARs). Westergaard [9] originally identified one PAR on each of the q-arms of Silene latifolia sex chromosomes. Later, Lengerova et al. (2003) [8] using fluorescent in situ hybridization (FISH) revealed that the X PAR is located on the p-arm, whereas the PAR on the Y chromosome is located on the q-arm. More recently, Scotti and Delph [10] proposed that PARs exist on both ends of the X and Y chromosomes, similar to the situation in humans [11]. A further similarity to mammalian sex chromosomes is that the S. latifolia sex chromosomes diverged gradually [12], which led to the formation of evolutionary strata. Comparisons between the evolutionarily young S. latifolia sex chromosomes (about 10 million years [12, 13]) and those of eutherian mammals (about 110 million years [14]) have revealed that similar processes are involved in the evolution of sex chromosomes in both animals and plants.

The sex chromosomes of S. latifolia most likely evolved from a single pair of autosomes as previously shown [12, 13], with one autosome of the gynodioecious relative S. vulgaris, a species lacking sex chromosomes, carrying homologues of S. latifolia sex-linked genes [12, 13]. Silene latifolia and S. vulgaris have the same haploid chromosome number (n = 12), but differ substantially in genome size. The Silene latifolia haploid genome is 2646 Mbp in females [15], with the X chromosome being about 400 Mbp in length [16], whereas the haploid genome size of S. vulgaris is 1103Mbp [17] and autosomes are about 100 Mbp long.

In this study we analyzed a part of the S. latifolia PAR located on the p-arm of the X chromosome and on the q-arm of the Y chromosome (henceforth referred to as PAR) and of the corresponding S. vulgaris autosome, in order to study collinearity and divergence between these chromosome parts and to assess whether the S. latifolia PAR has characteristics in common with animal PARs. Furthermore, we investigate whether the S. latifolia size increase relative to S. vulgaris reflects the increase in size of the entire X chromosome or more closely resembles the increase seen in S. latifolia autosomes.

In mammalian genomes, PARs have several interesting properties including increased GC content, higher mutation rates and a level of recombination higher than in the rest of the genome [18, 19] due to the necessity for crossing over in this region [20]. PARs in mice and the human PAR1 appear to serve a critical function in spermatogenesis, as indicated by the fact that their absence prevents X and Y chromosome segregation during male meiosis, causing male sterility [21–23]. However, PARs differ widely in size among mammals (covering about 4 % of the Y chromosome in humans [24, 25] and mice [26], about 8 % in cattle [27] and about 24 % in dogs [28, 29]), with most eutherians sharing the same genes situated closest to the telomere but having the pseudoautosomal boundary (PAB), separating the PAR from the sex-specific part of the sex chromosomes, at variable positions [27]. In mice, the PAB is located in the gene Fxy. Exons 1–3 are located in the X specific part, while exons 4–10 are located in the PAR [30]. The segment of this gene located in the PAR has a higher GC content than its X-specific portion [30, 31].

In order to analyze the S. latifolia PAR and the corresponding region on the S. vulgaris autosome, we first established and screened a bacterial artificial chromosome (BAC) library of S. latifolia with the marker ScOPA09 that has previously been found to be located in the PAR of the closely related dioecious species S. dioica[32] and has successfully been identified and used for mapping S. latifolia sex chromosomes [12]. The marker ScOPA09 is located in the S. latifolia PAR which is known to recombine once per generation in males [33] and makes up about 10 % of the Y chromosome [34, 35]. In S. vulgaris, the marker OPA is lacking. We therefore first sequenced a clone of the S. latifolia BAC library containing the marker ScOPA09. Sequencing was performed by Sanger and 454 pyrosequencing to explore the suitability of different sequencing strategies for BAC assembly. From these sequences we identified new markers and used them to screen the S. vulgaris BAC library for a homologous clone. Both BAC sequences were assembled into >100,000 bp-long scaffolds using GS De Novo Assembler (Roche).

Here we present the results of a genomic comparison between an area located in the PAR of the X chromosome p-arm and in the q-arm of the Y chromosome of the dioecious plant species S. latifolia and its homologous autosomal area in the closely related gynodioecious species S. vulgaris. Our results identify the first physically mapped genes located in the Silene PAR and reveal characteristics of a plant pseudoautosomal region.

Results and discussion

Our study reports the first comparative analysis of a BAC sequence from a plant pseudoautosomal region and the corresponding autosomal area in a related species that lacks sex chromosomes. Comparative mapping of a limited number of sex-linked genes in S. latifolia and autosomal genes in S. vulgaris has previously demonstrated large-scale synteny between the X chromosome of S. latifolia and one S. vulgaris autosome [12, 13]. Our results provide the first evidence for small-scale synteny and strong collinearity at the gene level within a restricted region of the S. latifolia sex-chromosomes, the PAR located in the p-arm of the X chromosome and in the q-arm of the Y chromosome, and the corresponding S. vulgaris autosomes.

BAC sequencing, assembly and annotation

The 454 paired-end sequencing of both a S. latifolia BAC clones containing marker ScOPA09 and of a homologous BAC clone from S. vulgaris gave more than 150,000 reads for each BAC clone. These were assembled into 171,870 bp and 116,096 bp-long scaffolds for S. latifolia and S. vulgaris, respectively.

We found a total of twenty-eight homologous sequences (seventeen different accession numbers) with the A. thaliana proteome. Of these, nine were found in both Silene species, two were identified only in S. vulgaris (one of them twice), and six were found only in S. latifolia. A total of 16 out of 28 sequences are most likely transposable or repeated elements as indicated by their annotations extracted from the protein domain family database ProDom [36] and repeat coverage (Table 1). The repeat coverage is based on BLAST hits with a Silene repeated elements library [37].

Table 1 Putative transposable elements identified in Silene BAC clones
Full size table

Among the nine sequences shared by the two Silene species, the areas matching TAIR accessions AT4G23160 (CRK8), AT2G01050, AT1G43760 and ATMG00860 contain repeated elements. Moreover, in S. vulgaris, the part matching ATMG00860 is contained in a match with the transposon sequence Q3I6J4_SILLA, which also includes the sequence matching accession number AT3G01410 (Table 1). Using annotated Silene transposable elements [34] we found that the region matching CRK8 is similar to a Copia-like retrotransposon [38], and that the region matching Q3I6J4_SILLA is similar to a Retand-like retrotransposon (see Kejnovský et al. (2006) for description [39]). Moreover, the sequences matching AT2G01050, AT1G43760 and CRK8 are found in both scaffolds at different positions, which provide further evidence that these sequences are transposable elements.

The five remaining sequences correspond to new pseudoautosomal genes. They are homologues of the A. thaliana genes ESP1 (AT4G22970), BIP1 (AT5G28540), ACBP1 (AT5G53470), and of genes AT5G53500 and AT5G41970. These latter two genes we named PAR1 and PAR2, respectively (Table 2).

Table 2 Putative genes identified in Silene BAC clones
Full size table

Finally, two other gene sequences, AT3G15000 and AT4G27700, that we named PAR3 and SVA1, were found on the BACs of S. latifolia and S. vulgaris, respectively (Table 2). However, because the BAC sequences only partly overlap, we do not have the homologous copies of these genes in the other species. The PAR3 gene in S. latifolia corresponds to a putative pseudoautosomal gene. The S. vulgaris SVA1 sequence is homologous to a gene coding for a rhodanese protein in A. thaliana (information collected from TAIR [http://arabidopsis.org/]).

Genes located in PARs close to the PAB (less than 50 cM) often present sex-specific expression [40]. Using RNA-seq data from Muyle et al.[41] we found no evidence for sex-biased expression of the genes located in the S. latifolia PAR ( Additional file 1: Table S1).

GC content

In mammalian PARs, high recombination associated with biased gene conversion (BGC) [42–44] results in a high GC content [27, 30, 45, 46]. Comparisons of GC and GC3 content between the PAR and non-PAR regions of the human X chromosome revealed a higher GC and GC3 content in the pseudoautosomal region [42]. Further studies of the human PAR revealed that the GC content decreases from 64 % close to the telomeric region to 55 % in the middle of the PAR and is only 38 % close to the pseudoautosomal boundary (PAB) [19]. Similar declines of the GC content were also found in other mammals, including cattle [27] and murine species [47].

A recent analysis of sequence polymorphisms in plants has revealed that the mating system affects GC content, with a higher content in outcrossing compared to selfing taxa being observed, but this effect is significant only in Poaceae that are known to have unusual GC contents [48]. Whether this effect is due to BGC and why it is observed only in Poaceae, however, is not clear. To date, evidence for BGC in plant sex chromosomes is lacking, but given the relatively small size of the investigated PAR (about 10 % of the Y chromosome [34, 35]) in S. latifolia, which is comparable to PAR size in mammals, and the fact that recombination occurs during meiosis, a higher recombination rate is expected in this region as compared with other regions of the genome.

In contrast to mammals, the Silene latifolia PAR can directly be compared with a homologous autosomal region in a closely related species. If the S. latifolia PAR has an increased GC content compared to autosomes, then this should be detectable in a comparative analysis. We then determined the GC and GC3 contents for each gene (Table 3). A comparison between S. latifolia and S. vulgaris revealed no significant difference (GC content: t = 0.0638, p-value = 0.9507; GC3 content: t = 0.0521, p-value = 0.9597). Moreover we determined the GC and the GC3 content of nine sex-linked genes that had previously been identified and are located in the sex-specific region of the X chromosome [49–57] ( Additional file 2: Table S2). No difference was found in the GC and GC3 content between the newly identified PAR genes and the genes located in the sex-specific region of the X chromosome (t = 0.601, p-value = 0.5656 and t = 0.6295, p-value = 0.5437, respectively). These results indicate that the pattern typical for mammalian PARs is not present in the investigated part of the S. latifolia PAR. This may indicate that the S. latifolia PAR maintains its “autosomal” features, as the sex chromosomes in this species are evolutionarily young. Alternatively, the studied region of the S. latifolia PAR might be close to the pseudoautosomal boundary where the GC content is lower than in more distal PAR areas, as has been found to be the case in most mammals [19, 27, 46]. Indeed, the ScOPA09 marker was estimated to be located 15 cM from the pseudoautosomal boundary (PAB) in S. dioica[32]. We obtained a very similar estimate of 11 cM for S. latifolia in this study (both calculations were done using Kosambi and Haldane mapping functions, for details see Additional file 3: Table S3). Even though we presently do not know the physical distance between the PAB and the BAC clone studied here, the results of our genetic analysis clearly show that all genes identified in this study are located in the PAR, as evidenced by their cosegregation with marker ScOPA09 ( Additional file 4: Table S4). Furthermore, these genes are recombining with the same recombination frequency of 11 % with the PAB ( Additional file 3: Table S3).

Table 3 GC and GC3 content comparison
Full size table

Structure comparison

Large-scale collinearity between the S. latifolia X chromosome and S. vulgaris autosome has repeatedly been reported in studies of S. latifolia sex chromosome evolution [13, 52, 58, 59]. In addition to large-scale collinearity we here report the presence of small-scale collinearity spanning five genes whose linear arrangement is conserved. We also show that the length of these genes is identical between the studied Silene species. Indeed, while we assessed whether the investigated S. latifolia PAR presents signs of chromosome enlargement, because the X chromosome is about four times the size of a S. vulgaris autosome, we analyzed exon and intron lengths of the five genes previously reported (Table 4). We then compared with a Student’s t-test whether the average size of both introns and exons of the different genes between both species were similar and we found no significant difference between S. latifolia and S. vulgaris (t = −0.0817, p-value = 0.9369 and t = −0.005, p-value = 0.9961 for intron and exon comparisons respectively). The conserved intron size may indicate a functional role in gene regulation. Indeed, introns enlarged by repetitive elements were found to affect gene expression in rice [60]. However, substantial differences in length occur in intergenic regions due to transposable element insertions. Figure 1 presents the global alignment of both BAC sequences. Intergenic regions are highly diverged in size and consequently major gaps are visible in the alignment. We considered as intergenic all regions in-between the five genes described in this paper for which there are copies in both Silene species. The total length of the intergenic region in the S. latifolia BAC is 115,909 bp, and 71,866 bp in the S. vulgaris BAC. This difference of 44,043 bp is highly significant (χ² = 30823.1, df = 3, p-value < 2.2e-16) and corresponds to a 61 % increase of the S. latifolia chromosome size as compared to S. vulgaris. This increase is due to the insertion of transposable elements in the S. latifolia PAR region.

Table 4 Comparison of gene sizes
Full size table
Figure 1
figure 1

Alignment of Silene latifolia and S. vulgaris BAC scaffolds. Genes (light blue), transposable elements (dark blue), full LTR transposon sequences (yellow) and uncharacterized nucleotides (light green) are annotated on the BAC sequences (black bold line). Regions of high identity (green), low identity (dark red) and gaps (fine black line) are indicated. The position of marker ScOPA09, a PAR-specific marker used to identify BAC clones located within the S. latifolia PAR, is indicated by a red diamond

Full size image

Microsatellite comparison

We found 577 and 377 microsatellite loci in S. latifolia and S. vulgaris respectively. A comparison of the average proportion of microsatellite loci (mono-, di-, tri- and tetranucleotide microsatellite) between both Silene species using a Student’s t-test revealed no significant difference (t = 0.5461, p-value = 0.5867). A previous analysis of microsatellites in plants revealed a negative correlation between microsatellite frequency and genome size [61]. The Silene latifolia X chromosome is about four times larger than the S. vulgaris autosome. However, our results revealed that the S. latifolia PAR contains a similar density of microsatellite repeats as the S. vulgaris autosomes, suggesting that microsatellites play no or only a minor role in the size increase of the Silene latifolia PAR. Then, we searched for long-mer microsatellite accumulation in S. latifolia and S. vulgaris (see Methods), which are expected to be rare in the PAR [62]. We found one (ATC)10 microsatellite locus in S. latifolia and one occurrence each of (T)35, (ATA)16, (AAG)19 and (TTA)25 in S. vulgaris. The low density of long-mer microsatellites observed here confirms the previously reported paucity of microsatellite repeats on the S. latifolia X chromosome PAR [62].

Transposable element insertion

We found three transposable elements containing long terminal repeats (LTR) in S. latifolia and two in S. vulgaris. The estimates of the invasion of these elements vary from about 17,200 years ago to 766,000 years ago ( Additional file 5s: Table S5). These elements were inserted after the divergence of the Silene sex chromosomes approximately 5 ~ 10 million years ago [12], which may be an indication of highly active transposable elements in the PAR. However, given that we have observed a smaller than expected size increase in the PAR (about 1.6x instead of about 4x), we hypothesize that the enlargement of the X chromosome occurs mainly in the non-PAR areas of the X chromosome and is due to large-scale accumulation of different tandem repeats [63, 64] and retrotransposons [65, 66]. The observed larger size of the studied S. latifolia PAR segment in comparison to the S. vulgaris autosome is close to the approximate difference in size between autosomes in S. latifolia and S. vulgaris and may therefore not reflect the size increase seen in the sex chromosomes.

Conclusions

In this study we present the first analysis of a fragment belonging to the S. latifolia pseudoautosomal region located in the p-arm of the X chromosome and the q-arm of the Y chromosome. The analysis of BAC sequences revealed five new pseudoautosomal genes that are conserved in size and linear arrangement between S. latifolia and S. vulgaris, indicating small-scale gene collinearity between the X chromosome and the corresponding autosomal region. No increase in GC or GC3 content was found in the studied PAR area, indicating that either the evolutionarily young S. latifolia PAR is not GC rich or alternatively, that the studied region is close to the pseudoautosomal boundary, where no increase in GC content is expected. A structural comparison revealed that non-coding regions of the S. latifolia PAR contain multiple transposable and repeated elements and are overall about 61 % longer than in S. vulgaris. This size increase is similar to the size difference between S. latifolia and S. vulgaris autosomes and may therefore reflect a genome-wide, rather than a sex chromosome-specific trend. Our study reports the first comparative analysis of a partial pseudoautosomal region in a plant which we compare to a closely related species lacking sex chromosomes, thereby providing new insights into genome size and sex chromosome evolution in Silene latifolia.

Methods

BAC preparation

S. latifolia and S. vulgaris were grown from seeds in a climate chamber. Fresh leaf material was harvested after initiation of flowering. Silene latifolia leaves were snap frozen in liquid nitrogen, packaged in dry ice and shipped to Amplicon Express, Pullman, Washington, where the BAC library was constructed from high molecular weight (HMW) genomic DNA following the method described by Tao et al. (2002) [67]. The S. vulgaris BAC library was assembled in the Institute of Experimental Botany of the AS CR Laboratory of Molecular Cytogenetics and Cytometry, Olomouc. In summary, DNA was partly digested with Hind III and inserted into the pECBAC1 vector. Ligations were transformed into DH10B E. coli cells (Invitrogen) and plated on LB agar containing appropriate concentrations of chloramphenicol, X-gal and IPTG. Clones were robotically selected with a Genomic Solution G3 and transferred into 384 well plates, grown for 18 h, replicated and frozen at −80°C. In order to identify positions and plate numbers of each clone, they were placed on a grid in duplicate on Hybond N + (Amersham, Biosciences) nitrocellulose membranes in a 4x4 pattern. The membranes were incubated and processed as described by Bouzidi et al. (2006) [68]. The S. latifolia BAC library was arrayed on six membranes of 18,432 colonies and one membrane containing 9,216 clones. The average insert-size of the library is 128 kb. The S. vulgaris BAC library was arrayed on three membranes with 18,432 colonies each. The average insert-size of the library was 110 kb [50]. Probes for radioactive hybridizations were labeled with α32P using the Prime-It II Random Primer Labelling Kit (Stratagene) according to the manufacturer’s protocol. The presence of the marker in positive BACs was verified by PCR. BAC DNA was isolated with the Large Construct Kit (Qiagen). The S. latifolia BAC clone containing the marker ScOPA09 was then sequenced. In order to test which method was most suitable for subsequent BAC assembly, we used three sequencing methods: shotgun Sanger sequencing, 3 kb and 8 kb 454 paired-end pyrosequencing on a GS-FLX machine (Roche). Sanger sequencing was performed by the GATC Biotech laboratory in Konstanz, Germany, and 454 sequencing by the Functional Genomics Center Zurich (FGCZ). As the ScOPA09 marker is not present in S. vulgaris, we developed suitable markers from neighboring loci using the S. latifolia BAC sequence in order to identify a homologous BAC clone in S. vulgaris. Primers used for screening the BAC library are presented in Table 5.

Table 5 PCR primers used to amplify fragments of genes located in the S. latifolia PAR
Full size table

Sequencing, assemblies and annotations

Assembly of the initial shotgun sequences of the S. latifolia BAC lead to the identification of numerous relatively short contigs because of the presence of multiple repeat regions. We therefore tested two different 454 pair-end sequencing methods (3 kb and 8 kb) in order to overcome these problems. Both approaches provided similar results, but because the 8 kb paired-end sequencing method requires more DNA, we used 3 kb paired-end sequencing for the corresponding S. vulgaris BAC clone ( Additional file 6: Table S6).

All 454 sequences were de novo assembled using Roche GS De Novo Assembler with 98 % minimum overlap identity and 60 bp as minimum overlap length. We set the expected depth parameter with reference to the estimated size of the BAC (determined by pulse field gel electrophoresis (PFGE)) and the number of reads sequenced. In addition to 454 sequencing and assembly, we used targeted Sanger sequencing to close gaps remaining in the S. latifolia scaffold after the assembly process. On S. vulgaris we successfully used the Epicentre transposon insertion Kit (EZ1982K) to fill a 987 bp-long gap remaining after assembly. Nevertheless, a few short stretches remain unsequenced after the assembly. Both scaffolds were submitted to GenBank under accession numbers JN574439-JN574440.

The scaffolds were then annotated by similarity using BLASTX [69] with UniProtKB (Swiss-Prot + TrEMBL, 13 July 2010), the Arabidopsis thaliana proteome (TAIR10_20100802) and transposable elements (TAIR9_TE) http://www.arabidopsis.org with an E-value cut-off of 1E-4. We used the exon prediction tool Genscan to identify coding sequences with A. thaliana as the training model [70] and used ProDom [36] and a Silene repeated element database [37] to detect Silene-specific repeats and transposable elements. Transposable element annotations were then completed using annotated S. latifolia repeats [34].

Genetic analysis

In order to verify the pseudoautosomal location of the BAC clone studied here, we performed segregation analyses using nucleotide polymorphisms in the genes ESP1, PAR2, ACBP1 and ScOPA09. First, PCR products of these genes were amplified and sequenced to look for partially sex-linked restriction polymorphisms segregating in the pseudobackcross population RB1. This population was prepared by crossing a female plant from a Swiss population, with a male plant obtained from a cross between a female of an inbred line, U9 (from Utrecht, kindly provided by Prof. Sarah Grant), and male plant from a Swiss population. Putative restriction polymorphisms (CAPS) were verified by restriction analysis, and genotyping was performed in 76 DNA samples available. For the list of primers and restriction enzymes used, see Additional file 7: Table S7. The observed recombination frequencies were used for the calculation of map distances using the onemap package of R [71] with both the Kosambi and the Haldane mapping functions [72, 73].

Structure comparison

Using BLAST [69] annotation results we searched for genes and transposable elements, also using Perfect Microsatellite Repeat Finder (http://sgdp.iop.kcl.ac.uk/nikammar/repeatfinder.html) with default parameters (minimum number of repeats = 3, minimum repeat unit length = 2 and maximum repeat unit length = 100). Mononucleotide repeats were identified manually. Only mononucleotide microsatellites with a minimum repeat number greater than or equal to 12 were counted. We assessed the frequency of all mono-, di-, tri- and tetranucleotide microsatellites and tested whether a smaller density of these microsatellites in S. latifolia than in S. vulgaris was associated with the observed genome size difference between species [61]. Furthermore, we estimated the frequency of long-mer microsatellite stretches (mononucleotides ≥ 30 repeats, dinucleotides ≥ 15 repeats and trinucleotides ≥ 10 repeats) in both species. We used a combination of A. thaliana annotation and Genscan [70] exon prediction to compare the size of intergenic regions and introns, and we measured the percentage identity of exons and introns and the GC content at third codon positions (GC3). In order to take into account the possibility that a gene was truncated due to its position at the beginning or end of a BAC, the gene fragment for which we have coverage in both species was used in calculations, whereas gaps were excluded. Both the GC and GC3 contents of the PAR genes were then compared with the GC and GC3 contents of nine X-linked genes (DD44X SlAP3X, SlX1, SlCypX, SlX3, SlX4, SlX7, SlX9 and SlssX) located in the non-recombining part of the S. latifolia sex chromosomes [49–57]. In order to test whether average GC and GC3 contents differ between species we used Student’s t-tests in R [71].

Transposable element analysis

We identified LTRs of transposable elements using LTR_Finder [74] set to the default parameters. We then aligned paired LTR sequences and determined the number of mutations (substitutions and insertions/deletions) between them. We estimated the age of the LTR invasion using the following equation: N/(2*L*K), where N is the number of base substitutions between the two LTRs, L is the length of the LTR and K the base substitution per site per year [75]. We used a value of K = 23E-9 as the average substitution rate per site per year [76].

Author’s information

Roman Hobza and Alex Widmer equal contributors as senior authors.