Introduction

Miniature inverted-repeat transposable elements (MITEs) represent a heterogeneous group of short non-autonomous mobile DNA elements, considered to be deletion derivates of autonomous DNA transposons. The mechanism of amplification and insertion into genomic DNA causes these elements to be flanked by Target Site Duplications (TSDs), and the elements themselves have Terminal Inverted Repeats (TIRs) of varying lengths that are typically AT-rich. MITEs are short (typically 70–500 bp long) and have been found in high copy numbers in plant (Wessler et al. 1995), fungal (Bergemann et al. 2008) and animal (Kuhn and Heslop-Harrison 2011) genomes. Initially two MITE groups were recognized based on their structural features: Stowaway-like MITEs with TA TSDs and Tourist-like MITEs with TAA TSDs (Jiang et al. 2004b; Zhang et al. 2004). A 128 bp MITE insertion in a mutant maize waxy gene was the first Tourist element discovered (Bureau and Wessler 1992), while a 257 bp element from sorghum was the first element of the Stowaway MITE group (Bureau and Wessler 1994). Later on, MITEs derived by sequence deletion from other DNA transposon families including hAT, CACTA, Mutator, and PiggyBac have been identified in numerous species (Benjak et al. 2009; Kuang et al. 2009; Menzel et al. 2014), making non-autonomous short elements that lack any protein coding domains but retain some sequence characteristics (TSDs and TIRs) of their progenitors enabling classification (Feschotte and Mouches 2000; Yang and Hall 2003; Jiang et al. 2004a, b). As derivatives of active DNA transposons lacking the transposase protein are necessary for their transposition and integration in new sites, mobility of MITEs is enabled by genes in their autonomous partners (Wicker et al. 2007). MITEs are mostly located in euchromatic (gene rich) regions and have played a role in gene regulation, in few cases downregulating the expression of nearby genes (Lu et al. 2012). The role of MITEs in modulation of gene expression and diversification of important crops such as barley (Lyons et al. 2008; Sun et al. 2009) and rice (Lu et al. 2012), as well as in animals such as Xenopus (Hikosaka et al. 2011), is now established. Sun et al. (2009) investigated insertion of MITEs into low-copy-number sequences or genic regions in rice, finding that a nested structure (along with other deletion/insertions) has modified the RPB2 (nuclear RNA polymerase II) gene, and is useful for phylogenetic and phylogeographic analyses in the Hordeum/Elymus group of species in the Triticeae.

The present study aimed to identify MITE-related sequences derived from various DNA superfamilies in Brassica genomes using dot plot comparisons of homoeologous BAC sequences and BLASTN searches. MITE-like sequences were identified and classified by their TSDs, TIRs, overall length and lack of open reading frames to gain an understanding of the diversity of small non-autonomous mobile elements in Brassica. We aimed to analyse MITEs to identify and describe characteristics of known and any novel Brassica MITE families, and then aimed to examine abundance, mobility, amplification and evolutionary history of the elements in various Brassica species and multiple diverse accessions to identify the contribution of individual mobile MITEs to the diversity of Brassica.

Materials and methods

Dot plot and BLASTN approaches for Brassica MITE identification

Pairs of Brassica rapa (AA) and B. oleracea (CC) BAC sequences (Supplementary Fig.) with long stretches of homoeology were identified by dot plots [(using JDotter (Sonnhammer and Durbin 1995) and Dotlet (Junier and Pagni 2000)] programs. Deletion–insertion pairs where one BAC had a sequence fragment that was absent from the other homologue were identified. The junction points in the sequence with the insertion were manually examined for evidence of TSDs and TIRs. Dot plots were also used to identify MITEs with long TIRs (MITEs derived from Mutator-like elements) within a B. rapa or B. oleracea BAC sequence plotted against itself. Homology lines perpendicular and close to the principal diagonal line of homology showed the TIRs, which were further investigated for MITE characteristics. Elements without MITE characteristics, including hAT/CACTA, full transposons and retrotransposons were also detected in the screen but are not analysed here. BLASTN searches (Altschul et al. 1997, 2009) were used to collect MITE homologues/copies from NCBI Brassica nucleotide (nr/nt) collection database (http://www.ncbi.nlm.nih.gov) using dot plot identified (reference) MITEs sequences as query.

MITEs copy number estimation

We analysed MITEs from 62 Mbp of Brassica genomic DNA BAC sequences available in the GenBank nucleotide database (nr/nt collection) at NCBI before March, 2012. The numbers of hits against the reference queries with >70 % query coverage and identity were extrapolated after getting output from BLASTN with copy no. = no. in database × genome size/database size as used for the estimation of MITEs in African mosquitoes (Tu 2001). We did not use shotgun-sequenced whole genomic assemblies, as MITEs, particularly when heterozygous, may be omitted.

MITEs characterization and nomenclature

The MITE-like sequences identified in dot plots were BLASTed against Repbase (Jurka et al. 2005) and TIGR Plants Repeat Database (JCVI) (Ouyang and Buell 2004) for homology-based characterization. Names to elements were given systematically: e.g. BrSTOW1-1, where ‘B’ stands for genus Brassica, second letter (r/o) represents species name (rapa/oleracea), 4 capital letters (STOW/TOUR) indicate Stowaway/Tourist origin of the MITE, the first number indicates the family and number followed by hyphen represents the respective member of that family as recommended for the nomenclature of transposable elements by Capy (2005). For Mutator-like MITEs such as BrMuMITE1-1, ‘Mu’ represents Mutator and in MITEs, whose autonomous counterpart was unclear or TSDs/TIRs were ambiguous were described as unknown MITE families such as BrXMITE1-1, where ‘X’ indicates unknown MITE.

PCR amplification of Brassica MITEs from diverse accessions

DNA from 40 Brassica accessions (Table 1) was used in the present study. Seeds from 32 Brassica accessions were kindly provided by Drs Graham Teakle and Guy Barker (Warwick Research Institute (WRI), Warwick, UK; see Walley et al. 2012). Two B. juncea and a B. carinata accession were collected from the National Agriculture and Research Center (NARC), Islamabad, Pakistan. Seeds for one commercial variety B. juncea (NATCO) accession were bought from an Asian supermarket at Leicester. The DNA from four synthetic allohexaploids (2n = 6x) Brassica (Ge et al. 2009) was provided by Dr. Xian Hong Ge (University of Wuhan, China). DNA was extracted from young leaves with a standard CTAB method (Doyle and Doyle 1990) and used for PCR amplification. Oligonucleotide primers were designed from the regions flanking MITE insertions using Primer3 (http://frodo.wi.mit.edu/primer3/). PCR amplifications were performed using 50–75 ng Brassica genomic DNA in a 15 µl reaction mix containing 2 µl PCR buffer (Kappa, UK), 1.0 mM additional MgCl2, 1 U Taq DNA polymerase (Kappa, UK), 200–250 mM dNTPs and 0.75 µl (10 pmol) of each primer. The thermal cycling conditions were as follows: 3-min denaturation at 94 °C; 35 cycles of 45 s denaturation at 94 °C, 45 s annealing at 52–64 °C (depending on primers) and 1-min extension at 72 °C; a final 3-min extension at 72 °C. PCR products were separated by electrophoresis in 1 % w/v agarose gels with TAE buffer, stained with addition of 1–2 µl ethidium bromide (10 mg/ml) for the detection of DNA bands under UV illumination, and photographed.

Table 1 Brassica accessions used for the study of diversity of MITE elements
Full size table

Fluorescent in situ hybridization

Seeds were geminated for 2–3 days and root tips were used for the preparation of mitotic chromosomes. The complete MITEs including the flanking regions were amplified by PCR and cleaned after gel electrophoresis (Qiagen). DNA was labelled with digoxigenin-11-dUTP or biotin-11-dUTP by random primer labelling (Invitrogen Bioprime) and used as probes. FISH of Brassica chromosomes was performed according to the protocol of Schwarzacher and Heslop-Harrison (2000). The probe mixture contained 50 % (v/v) formamide, 20 % (w/v) dextran sulphate, 2 × SSC, 25–100 ng probe, 20 mg of salmon sperm DNA, 0.3 % SDS (sodium dodecyl sulphate), and 0.12 mM EDTA (ethylenediaminetetraacetic acid). Hybridization and washing were carried out at low stringency (0.1 × SSC at 42 °C). Chromosomes were counterstained with 0.2 mg/ml DAPI (4′, 6-diamidino-2-phenylindole) diluted in McIlvaine’s buffer (pH7) and mounted in antifade solution (Citifluor). Examination of slides was carried out with a Zeiss epifluorescence microscope with single band pass filters and equipped with a CCD camera (Optronics, model S97790). The images were overlayed and optimized in Adobe Photoshop CS using only functions affecting the whole of the cropped image equally.

MITE sequences analysis

GC and AT contents of the MITEs were calculated using online available GC-Calculator (http://www.genomicsplace.com/gc_calc.html). Pictograms or logos of the sequence domains were generated with WebLogo (http://weblogo.berkeley.edu/logo.cgi).

Results

Identification and characterization of 15 MITE families in Brassica

A total of 14 distinctive MITE families, with mobility in the evolutionary period since separation of B. rapa and B. oleracea were identified by dot plot sequence comparisons (Table 2a). An additional MITE family was identified in dot plot analysis by the presence of TIRs within a single BAC (Table 2b). Based on the structural features (TSDs and TIRs) of the known DNA transposons, 14 of the Brassica MITEs were characterized as being derived from Mariner (Stowaway), PIF/Harbinger (Tourist) and Mutator (MuMITE) elements. The derivation of one MITE family exhibiting 3 bp TSDs was not classified due to the lack of any clear marks or strong homology to any known MITE or autonomous partner; we named this exception BoXMITE1. After initial identification of MITEs by dot plot analyses, these MITEs were used as query in BLASTN searches against the Brassica nucleotide collection (nr/nt) Genbank database to collect all homologues (Table 2c) in database. A total of 33 Stowaway, 35 Tourist, 27 Mutator and 5 elements of the novel BoXMITE1 family were chosen for analysis in the 62 Mbp of sequenced BACs. Based on number of MITEs identified in dot plot analysis and BLASTN searches (Table 2a–c), MITEs in whole genome sequences were estimated (Table 3). Structural characteristics including TSDs [TA in Stowaway (Mariner-derived), TAA or TTA in Tourist (PIF/Harbinger-derived), 9 or 10 bp (Mutator-derived), and TTC (BoXMITE1)], short lengths, lack of open reading frames, AT richness (ranging to 80 %, although notably the BoXMITE1 sequence was not AT-rich at 53 %), and high copy numbers confirmed all elements as MITEs (Table 3).

Table 2 MITEs identified from Brassica BAC sequences
Full size table
Table 3 Estimated copy numbers and AT percentage of Brassica MITE families
Full size table

Dot plots (Fig. 1) show the complex range of structures and lack of conservation in length of the TIRs of representatives of the major MITE families. In one Tourist element and two Mutator-derived elements, the TIRs (boxed) span much of the length of the element. Notably, many MITEs included short near duplications within their internal regions, sometimes (BrTOUR3, BoMuMITE3, BoXMITE1) but not in all cases (BrSTOW1, BrTOUR2, BoMuMITE4) representing fragments of the TIRs.

Fig. 1
figure 1

Structural characterization of MITEs in Brassica by dot plot sequence analysis. Each dot plot of a MITE against itself allowed the identification of TIRs at corners (boxed). The central complete diagonal line represents self-homology, while the boxed inverse-diagonal lines show the TIRs; other lines show near-direct repeats in both forward and inverted orientations. Scales in base pairs show the wide range in sizes of elements and length of TIRs

Full size image

We found element-specific motifs in the TIRs of the five Brassica Stowaway and four Tourist MITE families immediately following the TSD insertion site (Fig. 2). For entry into the program, TIRs (Table 2) of each family were aligned to the same length using CLUSTALW with default parameters, and for a very small number of sequences, single bp insertions or extensions were removed (see Table 3 for length ranges): BrSTOW1 and BoTOUR4 had short insertions relative to the other family members, while 4 bp in BoSTOW4 and 5 bp in BrTOUR3 are low information content (low-height) letters representing non-homologous sequence or an insertion. While showing conservation within each of the 15 families, TIRs varied extensively between most families (Fig. 2). Length differed from 11 to 106 bp, some showed near equal AT-CG content while others were AT-rich or even showed gross strand asymmetry in base pair composition (up to 93 % TG rich in one strand of BrTOUR1), and there was a varying degree of conservation of 3′ or 5′ termini or internal domains within individual families. Mutator-derived MITEs showed highly AT-rich regions within TIRs (Table 2) and internal regions.

Fig. 2
figure 2

Sequence logos (pictograms) of Brassica MITE TIRs. The logos were generated with (n) sequences, and letter heights (0–2 bits) indicate the information content of consensus nucleotides at each position in the TIRs of Brassica Stowaway (left) and Tourist (right) MITEs. Lower heights represent non-conserved motifs or insertions within a family. There is little conservation between the families

Full size image

Site-specific insertion polymorphism of MITEs in Brassica germplasm

To investigate the polymorphisms of Brassica MITEs among 40 Brassica accessions (Table 1) from three diploids (AA, BB, CC), three allotetraploids (AABB, AACC, BBCC) and 2 synthetic hexaploid Brassicas (AABBCC, B. napus × B. nigra; B. carinata × B. rapa), primer pairs (Table 4) were designed from sequences flanking the MITEs identified by comparison of homoeologous BAC pairs. The insertion polymorphisms (Table 2a) showed that some families had been active after the evolutionary separation of the genomes.

Table 4 Brassica MITE primers with size of the elements, size of the expected products, names and sequence of primers
Full size table

Stowaway MITE insertion polymorphisms

The presence of the 237–244 bp BrSTOW3-1 elements was tested using primer pair BoSTOW3F/R (Table 4; Fig. 3), with a product size of 512 bp including the MITE element or 272 bp where the element was not present (flanking region with pre-insertion or empty sites). All B. oleracea lines included the element at this site, while the insert was absent in B. rapa and B. nigra accessions. The BrSTOW3-1 element was also present in the allotetraploids B. napus (AACC) and B. carinata (BBCC) and four hexaploid Brassica lines (AABBCC), but absent in B. juncea (AABB), consistent with its presence only in C-genome diploid except in the Pakistani accession line 12 (NARC-II; see “Discussion”). The BoSTOW4-1 (Fig. 3c) insertion proved to be like BrSTOW3-1, specific to the C-genome, with presence of the element indicated by a 500 bp band in all the accessions with a C-genome, and a 273 bp band representing flanking sequences only in the A- and B-genomes.

Fig. 3
figure 3

Insertion polymorphisms of Brassica Stowaway-like MITEs amplified by PCR with flanking primers. a BrSTOW1-1 and b BoSTOW3-1 c BoSTOW4-1. Figures 3–5 show inverted images of ethidium bromide stained PCR-amplified DNA after size separation by agarose gel electrophoresis. Lower numbers (1–40) identify individual lanes for each Brassica accession listed in Table 1. Braces group Brassica species. Black arrowheads (right) upper bands with amplified loci having MITE insertions while lower bands amplify the loci without insertions. Left lane (HP1) is 200 bp marker ladder (Hyperladder I) with band sizes indicated

Full size image

The organization of the element and flanking sequences for BrSTOW1-1 was more complex (Fig. 3a), with only one (cultivar De Rosny) accession of B. oleracea (CC) and B. carinata (BBCC) showing any amplification using the BrSTOW1F/R primer pair, suggesting divergence or loss of the flanking sites in the B- and C-genome ancestor compared to the A-genome. Of species including the A-genome, some showed sites without the inserted element, some showed presence of the Stowaway element (682 bp), and another group, including three of the diploid B. rapa, showed bands associated with both presence and absence of the Stowaway element; surprising in these inbred lines expected to be homozygous. Therefore, it seems that the region flanking the Stowaway element is duplicated in the genome, allowing amplification of sites with and without its insertion, a duplication that is not shared by only one of the 15 A-genome tetraploids. A few faint bands were interpreted as amplification between sites with weak homology to the primers.

Tourist MITE insertion polymorphisms

The primer pair BrTOUR1F/R amplified BrTOUR1-1 products with the MITE insertion from 15 of the 40 Brassica genomic DNA accessions, as well as a shorter primer-related product from all 40 accessions (Fig. 4a). It did not amplify MITEs in the three B-genome accessions, but amplified only from one A-genome accession (cultivar Chinese Wong Bok). Of the six diploid C-genomes, five showed amplification of the MITE elements. Consistent with these results, the AB-genome B. juncea accessions showed no amplification except for the anomalous accession (NARC-II; see “Discussion”), and five of the six B. napus accessions showed amplification. Despite including the C-genome, there was no amplification from the six B. carinata accessions.

Fig. 4
figure 4

Insertion polymorphism of Brassica Tourist-like MITEs amplified by PCR with flanking primers. a BrTOUR1-1, b BrTOUR2-1 and c BrTOUR3-1. See Fig. 3 for explanation

Full size image

The primers BrTOUR2F/R amplified a 510 bp product with the MITE insertions in accessions with the A-genomes including all B. rapa and B. juncea lines, six lines from B. napus and all four hexaploid Brassica lines (Fig. 4b). Of accessions with only B- and/or C-genomes, only one B. carinata (NARC-PK; Pakistani origin) showed amplification of the MITE element. The primers for BrTOUR3-1 did not detect the element in the B- and C-genomes (306 bp band). Bands including the length of the BrTOUR3-1 MITE insertions (564 bp) were found in B. rapa (A) and the tetraploids AABB and AACC Brassicas (Fig. 4c). Mutation of a primer site (or, less likely, evolutionary mobility of the element) may lead to results showing two bands (with and without the element) in four of the six B. rapa accessions, and the absence of lower bands in four of the nine B. juncea accessions (which would be expected to include the empty site from the B-genome).

Mutator MITE insertion polymorphisms

The primers for BrMuMITE1-1 amplified 1016 bp sites where the Mutator-derived MITE was present (Fig. 5a). The primers showed some homology to other sites in the genome and hence weak amplification of other products. One B. rapa accession (San Yue Man) showed a major product corresponding to the 508 bp of the empty site seen in the source BAC sequence, along with a product corresponding to the insertion and an intermediate product, suggesting genomic rearrangement and duplication associated with the MITE element. No strong amplification was seen in any B- or C-genome accessions, and all the tetraploids with the A-genome showed presence of the MITE. BoMuMITE4-2 (Fig. 5b) showed many polymorphisms between accessions and species, with null sites (no product), amplification across empty sites (400 bp product), and amplification with the BoMuMITE4 insertion (990 bp product).

Fig. 5
figure 5

Insertion polymorphism of Brassica Mutator-like MITEs amplified by PCR with flanking primers. a BrMuMITE1-1 and b BoMuMITE4-2. See Fig. 3 for explanation

Full size image

Chromosomes and genomic localization of MITEs in Brassica

We studied the localization and distribution of high-copy-number (Table 3) MITE elements on Brassica chromosomes. The Mutator-derived MITEs were amplified and labelled, before hybridization to chromosomes from the allotetraploid (4x) Brassica species. BrMuMITE1-1 showed dispersed hybridization to the 20 A-genome chromosomes in B. napus (AACC, 2n = 38; Fig. 6a, b) and B. juncea (AABB, 2n = 36; Fig. 6e). In contrast, BoMuMITE4-2 was most abundant in sub-telomeric regions of particular chromosomes from both genomes, with much weaker dispersed hybridization on A-genome chromosomes. In both B. napus (Fig. 6b, c) and B. carinata (BBCC, 2n = 34; Fig. 6f), BoMuMITE4-2 showed colocalization with major and minor rDNA sites (constrictions and weaker DAPI staining). Chromosome number in B. juncea (AABB, 2n = 4x = 36) line NARC-II was confirmed by DAPI staining (Fig. 6d), as PCR results from this accession were anomalous (see above and “Discussion”).

Fig. 6
figure 6

Fluorescent in situ hybridization (green and red signals) showing locations of MITEs on Brassica metaphase chromosomes stained with DAPI (blue). a–c B. napus (2n = 4x = 38 AACC) with (a, b) BrMuMITE1-1 (red) labelling the 20 A-genome chromosomes along most of their length with some stronger sites, and (b, c) BoMuMITE4 -2 (green) labelling about 14 sites near 45S rDNA loci and some dispersed signal primarily on A-genome chromosomes. d A metaphase of B. juncea line NARC-II (PK001325; 2n = 4x = 36 AABB) stained with DAPI; many PCR results from this accession were anomalous. e BrMuMITE1-1 (red) hybridized to a metaphase of B. juncea showing strong hybridization to A-genome chromosomes (excluding some centromeric regions) and very weak hybridization to chromosomes of B-genome origin. f Metaphase chromosomes of B. carinata (2n = 4x = 34 BBCC) showing BoMuMITE4-2 (red). Scale bar 8 µm (color figure online)

Full size image

Structural features of an unknown MITE family in Brassica

We identified a MITE-like element with 3 bp TSDs (nucleotide sequence TTC) and 42 bp imperfect TIRs but no significant homology to any known MITEs. The element is named BoXMITE1-1 and represents a low-copy-number family (BoXMITE1) with only 229 estimated copies within Brassica (A-, C-genomes, Table 3). BoXMITE1-1, the first identified element from the family was found inserted in B. oleracea BAC sequence accession EU642504.1 from 86275 to 86676 bp. Using this as query sequence against the Brassica nucleotide collection database in GenBank, only two complete sequences were retrieved, while searching against Brassica whole genome shotgun contigs (wgs) database, we collected another 3 full length copies (BrXMITE1-2, BrXMITE1-4, BrXMITE1-5). The annotations indicate their localization on chromosome 1, 4 and 7 of B. rapa. The elements of the family range in size from ~308 to 402 bp with 3 bp TSDs with single bp mismatch at variable positions. The TIRs of the family members range from 21 to 42 bp with few bp mismatches. BoXMITE1-1 is flanked by 42 bp, while BrXMITE1-4 is flanked by 21 bp TIRs (Table 3).

Discussion

Our molecular characterization of 15 novel MITE families (Tables 2, 3) from Brassica showed five were derived from Stowaway-like elements, four were Tourist-like, five Mutator-like and one is a novel MITE family (BoXMITE1), whose progenitors were not identified. Except for BrMuMITE3, the first reference member of all these families was identified by dot plot comparison of homologous BAC sequences of diploid Brassica genomes, indicating MITE presence in one species and absence in the other. This strategy identifies elements which have shown mobility since evolutionary separation of the diploid species from a common ancestor, and is not dependent on previous knowledge or identification of sequences related to reference elements. The abundant, high-copy-number elements showed structural characteristics of MITEs (Figs. 1, 2) and TSDs related to the known Stowaway, Tourist and Mutator groups (Tables 2, 3). One novel group, BoXMITE1, had a lower copy number with unusual TIR and TSD structures consistent with its MITE origin. It is notable that all the MITE families show activity since the separation of the species.

Approximately 30,000 MITE-like sequences belonging to the 15 families were estimated to occur in the B. rapa and B. oleracea genomes (Table 3). Around 45,821 MITE sequences belonging to 174 families were identified in B. rapa using MITE Digger and MITES-Hunter programs (Chen et al. 2013). BraSto, a well-characterized Stowaway MITE family was reported with similar abundance to our MITE families in Brassica (Sarilar et al. 2011). The rice genome harbours rather more elements, with ~178,533 MITE-related sequences clustering into 338 families (Lu et al. 2012). A parallel study in the Solanaceae has revealed a high level of MITE diversity among the crops in the family (tomato, potato and tobacco) and 22 families including derivatives of Stowaway, Tourist, hAT, and Mutator-like MITEs (Kuang et al. 2009). Several CACTA and hAT-like non-autonomous families were also investigated from Brassica (not discussed here), while hAT-derived MITEs have been studied independently in various species, e.g. in Beta vulgaris and Musa (Menzel et al. 2012, 2014). Given the activity and polymorphisms of the elements and the presence of LTRs, whole genome assembly approaches (often published without details of parameters) may well collapse or delete MITEs so copy number estimates from whole sequence data may be wildly inaccurate.

Evolution and biodiversity of MITEs: amplification and insertion polymorphisms

MITEs transpose into new sites, with or without replication at variable rates (influenced by genomic stress and hybridization; Madlung and Comai 2004) in different cultivars or genotypes, creating the presence/absence-based polymorphisms (Lyons et al. 2008) as described for retrotransposons in the RBIP analysis (Flavell et al. 1998) and exploited here to identify non-selectively all the MITE families in Brassica (Fig. 1). Compared to a site containing a MITE, an ‘empty site’ may be detected either where no element has been in the genomic sequence, or after a MITE excises and moves, when the empty donor host site exhibits a footprint usually with an extra TSD sequence compared to the locus prior to MITE insertion.

Individual transposons, including MITEs, differ in their conservation and proliferation properties (Kubis et al. 1998; Feschotte and Mouches 2000). High conservation in a genome can indicate recent amplification as a burst, while presence over a wide evolutionary lineage shows ancient amplification (Oki et al. 2008; Zerjal et al. 2012). We exploited the knowledge that the MITEs identified here are evolutionarily active to characterize their presence in diverse Brassica germplasm, and reconstruct lineages. Primers flanking the MITEs were designed from genomic DNA conserved between two species for PCR of genomic DNA, and the insertion polymorphism of Brassica Stowaway, Tourist and Mutator-derived MITEs was observed among 40 cultivars (Figs. 3, 4, 5). The PCR amplification of BrSTOW1-1 in B. rapa and BoSTOW3-1 and BoSTOW4-1 in B. oleracea suggested the conservation of MITEs in A- and C-genomes, with empty sites (shorter products) observed in some lineages (Fig. 3). Similarly, the amplification of Brassica Tourist and Mutator MITEs (Figs. 4, 5) yielded products with and without insert, displaying insertion polymorphisms. The polymorphisms of particular elements enabled identification and differentiation of many cultivars in Brassica; MITE-related molecular markers were used in other plants such as barley (Lyons et al. 2008) and maize (Lu et al. 2012) to study the biodiversity and evolutionary phenomena.

The high-copy-number families (Table 3) related to the two individual MuMITE elements used for PCR amplification in Fig. 5 were used for in situ hybridization to Brassica chromosomes (Fig. 6). The genomic location of the individual elements is unknown, but the families showed contrasting distributions: BrMuMITE1-1 was amplified and dispersed overall A-genome chromosomes. BoMuMITE4-2 was present on both genomes, co-localizing with 45S rDNA sites (despite the AT richness of MITE sequences, while rDNA sequences are GC rich, seen by their weaker DAPI staining on chromosome preparations), and also showed weaker, dispersed hybridization, greater on the C-genome chromosomes. The proliferation and genomic locations of the families (Fig. 6) are consistent with copy number estimates (Table 3), and PCR amplification results of the single family members (Fig. 5), with a contrast between BrMuMITE1-1 (isolated from B. rapa), being largely A-genome-specific, and BoMuMITE4-2 (from B. oleracea) present in the C-genome but being more polymorphic, less genome-specific and apparently targeted to rDNA sites. While transposon association with rDNA is unusual (and indeed, retrotransposons are often excluded from rDNA loci, e.g. Brandes et al. 1997; Kuipers et al. 1998), rDNA-associated SINE localization is reported in Brassica (Goubely et al. 1999). Recently, Eagle and Crease (2012) have reported a DNA transposon associated with complex amplification and rearrangement events in rDNA loci in Daphnia that, like BoMuMITE4, targets rDNA and also occurs in other genomic locations.

Origins and genomic constitution of Brassica accessions

The PCR insertion polymorphism (Figs. 3, 4, 5) gave results that were generally consistent with the presence or absence of polymorphisms in the diploid species and genome constitutions of the tetraploid species. BrSTOW1-1, BrTOUR1-1, and BrTOUR3-1 were polymorphic in diploid genomes, also reflected in the tetraploids, thus supporting their polyphyletic origin (Cifuentes et al. 2010). However, a few accessions analysed here showed results that were not consistent with their morphological identification. In particular, accession NARC-II (line 12; PK-001325) from Pakistan, morphologically and by chromosome number (previously unconfirmed, Fig. 6d) a B. juncea (AABB) accession, showed properties of the presence of the C-genome as well as A- and B-genomes. Other lines from Pakistan, including B. carinata accession NARC-PK (line 36; PK-0085490) (MuMITE4-2, BoTOUR2-1), showed some results that were not consistent with their expected genomic origin. It is notable that there is a long history of interspecific hybridization and intercrossing of Brassica species within Punjab region (Pakistan and India), where these accessions originate (Sikka 1940). It is therefore possible that current accessions have a hybrid ancestry: accession NARC-II (line 12; PK-001325) shows strong evidence for the presence of MITE elements from all three A-, B- and C-genomes. It will be valuable to characterize further these accessions using more genome-specific markers or genome-specific probes for fluorescent in situ hybridization. The exploitation of agronomic and quality characters introgressed from different Brassica genomes is an important target for breeders (Tu et al. 2009; Kumar et al. 2011; Heneen et al. 2012).

In B. juncea, BrTOUR3-1 shows accessions that are identical (‘homozygous’) at all four sites both for presence and absence of the insertion, and with both alleles. It is unexpected that the empty site from the B-genome diploids is not seen in four of the nine B. juncea AB tetraploids (Fig. 4). Both BrSTOW1-1 and BrTOUR1-1 unexpectedly have heterozygosity in the inbred A-genome diploid (as well as in the tetraploids; Figs. 3, 4); given that these are inbred Brassica lines; it is possible that there is duplication of flanking sequences (also possible for the lower band amplified with BrTOUR1-1 primers). As more sequence for the B-genome is obtained, it will be important to identify elements with specificity to this lineage.

Brassica MITEs display high AT-rich regions

One of the typical features of MITEs is the presence of highly AT-rich sequences (e.g. the AhMITEs from Arachis hypogea exhibit an AT content of 70 %; Shirasawa et al. 2012), a characteristic found in all Brassica MITEs. The average AT contents (Table 3) within the Brassica MITE families range from 53 % (BoXMITE1) to 80 % (BoSTOW4, BrMuMITE2).

Conclusions

Our results show that truncated derivatives of various autonomous DNA transposons superfamilies designated as MITEs, detected by bioinformatics and molecular techniques, are evolutionarily active and dispersed in Brassica genomes, and some have shown polymorphisms in different genotypes. Thus, MITEs are playing a role in diversification and evolution of the Brassica genome. The present work identifies the range of MITE families in Brassica and enables their identification, characterization and annotation as well as study of distribution, diversity and mobility. The study of their flanking genomic sequences and insertion polymorphisms, consequent on their transposition activity, suggests that MITE mobility played an important role in mechanism of genome evolution and diversification. MITEs have potential use as gene modifiers or mutagens. The identification of Brassica MITEs will have broad applications in Brassica genomics, breeding, hybridization and phylogeny through their use as DNA markers.