Genetics, Vol. 163, 1097-1108, March 2003, Copyright © 2003

A High-Copy-Number CACTA Family Transposon in Temperate Grasses and Cereals

Tim Langdona, Glyn Jenkinsb, Robert Hasterokc, R. Neil Jonesb, and Ian P. Kinga
a Institute of Grassland and Environmental Research, Aberystwyth, Ceredigion, SY23 3EB, United Kingdom,
b Institute of Biological Science, University of Wales, Aberystwyth, SY23 3DD, United Kingdom
c Department of Plant Anatomy and Cytology, Silesian University, Jagiellonska 28, PL-40-032 Katowice, Poland

Corresponding author: Tim Langdon, Plas Gogerddan, Aberystwyth, Ceredigion, SY23 3EB, United Kingdom., tlangdon{at}hgmp.mrc.ac.uk (E-mail)

Communicating editor: V. L. CHANDLER


*  ABSTRACT
*TOP
*ABSTRACT
*MATERIALS AND METHODS
*RESULTS
*DISCUSSION
*LITERATURE CITED

A lineage of CACTA family transposons has been identified in temperate grasses and cereals, and a full-length representative of the subfamily from Lolium perenne has been sequenced. Both the size and internal organization of the L. perenne element are typical of other CACTA family elements but its high copy number and strong conservation are unexpected. Comparison with homologs in other species suggests that this lineage has adopted a distinct and novel evolutionary strategy, which has allowed it to maintain its presence in genomes over long periods of time.


TRANSPOSABLE elements were first discovered in maize >50 years ago, using classical genetic and cytological methods (reviewed by GIERL et al. 1989 Down). Elements identified in those early studies, such as Ac/Ds and En/Spm, have now been characterized in detail at the molecular level and found to be mobilized by a cut-and-paste mechanism. This mechanism not only allows movement into new sites but also, depending on the success of the transposition event and its timing relative to DNA replication, may increase or decrease the copy number of the element. The propagation of these DNA, or class II, transposable elements ("transposons") contrasts strongly with that of RNA, or class I, transposable elements ("retroelements"), where a static parental element generates copies at new sites via an RNA intermediate. In the case of the most sophisticated of the retroelements, retrotransposons, each new copy is potentially as active as the parent, leading to the possibility of exponential increases in copy number. In plants, particularly the cereals, massive increases in retrotransposon abundance do appear to have occurred during recent evolution, so that these elements make up 50% or more of the genomes of species such as maize, wheat, and barley (KUMAR and BENNETZEN 1999 Down). Not surprisingly, it appears that retrotransposons tend to accumulate outside genes; intergenic "islands" representing multiple sequential insertions have been characterized in detail in maize (SANMIGUEL et al. 1996 Down, SANMIGUEL et al. 1998 Down) and may be representative of the typical distribution of these elements in large grass genomes. This distribution may well be driven by active targeting of insertion by the elements themselves (SUONIEMI et al. 1997 Down). In contrast, transposons may show a preference for insertion into "open" chromatin in or around coding sequences (CRESSE et al. 1995 Down), increasing the potential phenotypic impact of individual element movements. The relatively low copy numbers seen for plant transposons, ranging from 10 or fewer (e.g., Ac, En) to hundreds (active Mutator), can then be hypothesized to be a consequence of higher levels of selection acting against them.

Transposons first identified in maize are now known to be members of families with representatives in a wide range of species. Ac belongs to the hAT superfamily (ATKINSON et al. 1993 Down), which is found in both animals and plants, while En belongs to the CACTA superfamily (UPADHYAYA et al. 1985 Down), so far found only in plants. This last family is named after the conserved bases found at the beginning of the terminal inverted repeats (TIRs), which define the ends of transposons. CACTA family members have relatively short TIRs, ranging from 13 bp in Spm to 26 bp in sorghum Candystripe (CHOPRA et al. 1999 Down); TIRs are a similar size in most other elements but are several hundred bases long in Mutator. All full-length transposons must encode a transposase to carry out the essential cut-and-paste reaction, but at least one additional protein is also frequently found, whether expressed from a second promoter or from alternative splicing events. The use of multiple proteins can lead to complex regulation of transposition, which has been best characterized in the founding member of the CACTA family, Spm. Spm encodes two proteins, TnpA and TnpD, expressed from a single transcription unit (MASSON et al. 1991 Down). TnpD is the better conserved of the two but its exact function has yet to be demonstrated. TnpA has been shown to bind a 12-bp motif that is highly repetitive in the subterminal regions of the element (TRENTMANN et al. 1993 Down). Its binding can lead to either activation or repression of Spm transcription, depending on the methylation status of the promoter, and it is also believed to play a key role in establishing active transposition complexes, which require the presence of both proteins as well as the TIRs and a minimal number of TnpA target motifs (FEDOROFF et al. 1995 Down; RAINA et al. 1998 Down). The proteins may be provided in trans from intact (autonomous) elements, allowing the movement of internally deleted or otherwise noncoding (nonautonomous) elements, provided that they retain the necessary cis-acting target sequences. Other CACTA elements have also been found to encode two proteins and to have comparable arrangements of repeated subterminal motifs, although there is no conservation of motif sequence in elements from divergent species and only limited similarity of TnpA-like proteins in the DNA-binding domain (DBD; CHOPRA et al. 1999 Down).

To date there have been few comparative studies of plant transposons and their evolution as attention has focused on understanding the activity of specific elements and, more recently, in optimizing their use as genetic tools. In addition, a number of characterized elements show very restricted distributions, with homologs being found only in related species. In some cases the elements are even restricted to specific lines within a species, suggesting that they are transient or dynamic genomic components. However, until recently most plant transposons have been identified on the basis of a phenotypic effect, which may have produced a bias for "aggressive" subfamilies that are in the process of colonizing contemporary genomes. Sequencing projects for rice and Arabidopsis indicate that transposons are a significant component even of these small genomes. Given the wide range of effects on chromosome structure and interaction that may be mediated by these elements, it is of interest to use a comparative approach to understand the fate of transposon populations over evolutionarily significant periods. We have identified a CACTA transposon family that has homologs in a wide range of temperate grasses and cereals, and we have begun its characterization with this aim. Surprisingly, the family appears to maintain a high copy number in divergent species, suggesting a successful alternative evolutionary strategy to the "boom-and-bust" pattern implied in previously described distributions.


*  MATERIALS AND METHODS
*TOP
*ABSTRACT
*MATERIALS AND METHODS
*RESULTS
*DISCUSSION
*LITERATURE CITED

Genetic and DNA stocks:
Plant stocks are maintained at Aberystwyth. The Lolium perenne Lp10 x L. temulentum hybrid was described previously (JENKINS et al. 2000 Down). Genomic DNAs were extracted from leaf material. {lambda}-Libraries were obtained from Stratagene (La Jolla, CA). DNA weight markers are a 1-kb ladder (Life Technologies).

PCR:
The following oligonucleotides were used: TCMK, CCCCCTTGGTTGTGCATGAA; TNC, AAGCCACCTTSGATGYCCCAT; TIR1H, GGCAAAGCTTACACTACTAGGAAAAGGCT; PKW1, GGTTTTAGGGTGGCGCAACCGAARTGGGA; YNL1, GGAATTCAACCAGGGGGGSAGRTTGTA; BTPO3A, GTTCCCGGAGTCTACRACGACTGG; BTPO3B, AACTASCATCACATAGARGAG; BTPO3C, CGATATCGATCATCTAAACTACCATCACAT; and EAR1, TTCGATTCACACTACTAGGAAAA. Conditions used for PCRs described in the text were as follows: initial TnpD-like gene amplification, TCMK/TNC primers, 94° x 1 min, 48° x 2 min, 68° x 3 min followed by 27 cycles of 94° x 30 sec, 50° x 30 sec, 72° x 90 sec; amplification of subterminal regions, TIR1H/YNL1 (5') or TIR1H/PKW1 (3'), (first cycle TIR1H oligo only, 94° x 1 min, 50° x 1 min, 72° x 10 min, subsequent 27 cycles plus YNL1 or PKW1 oligo, 94° x 30 sec, 60° x 30 sec, 68° x 5 min); amplification of RNaseH region, BTPO3A/BTPO3B, BTPO3A/TIR1H, BTPO3C/PKW1, 30–33 cycles of 94° x 30 sec, 60° x 30 sec, 72° x 4 min. Genomic template DNAs were at a final concentration of 5 ng/µl, dNTPs at 1.5 mM, MgCl2 at 5 mM, and Taq DNA polymerase (2.5 units/100-µl reaction) and buffer were as supplied (Roche, Indianapolis).

Fluorescence in situ hybridization (FISH):
Methods were as described previously (JENKINS et al. 2000 Down). Probes were made by asymmetric labeling of PCR products (TNC primer, Tpo1-lp1 template or rye genomic clone template for Lolium and rye TnpD-like probes, respectively; PKW1 primer, Tpo1-lp1 template for TnpA-like probe; EAR1 primer, Tpo1-lp1 template for right-hand subterminal probe).

Sequence analysis:
Analysis was carried out using the GCG10 package (Pharmacopeia). Database searches were carried out using BLAST as implemented by the National Center for Biotechnology Information (http://www4.ncbi.nlm.nih.gov/BLAST/index.html). Sequence display was made with GeneDoc (http://www.psc.edu/biomed/genedoc/). Phylogenetic trees were derived by ClustalX (THOMPSON et al. 1997 Down). Secondary structure predictions were made using the method of Zucker (http://bioinfo.math.rpi.edu/?mfold/dna/form1.cgi).

Hybridization and screening conditions:
Standard (high-stringency) hybridizations were carried out in Church buffer (0.25 M NaCl, 0.25 M phosphate, 10% polyethylene glycol, 7% SDS, 1 mM EDTA, pH 7) at 60° for 4–16 hr. Standard washing conditions were 2 x 20 min, 0.2 x SSC, and 0.5% SDS at 60°. Low-stringency hybridizations and washes were carried out with the same conditions but with the temperature reduced to 50°. The washes correspond to ~Tm - 14° or Tm - 24° for the Tpo1-lp1 TnpD-like probe (high and low stringencies, respectively). {lambda}-plaques were transferred to Hybond-NX membranes according to the manufacturer's instructions from 9-cm plates containing ~1000 plaques each. Probes were made by radiolabeling of PCR products, and filters were stripped before hybridization by washing with 0.2 M NaOH at 60°. Initial screening of the L. perenne library was with the TCMK/TNC PCR product (see above); subsequent sequential screens to estimate deletion frequency were made with probes derived from the TIR1H/YNL1 product [Tpo1-lp1, nucleotides (nts) 1–2362, labeled with TIR1H only or YNL1 only] or TIR1H/PKW1 product (7686–12,485, labeled with BTPO3C only or PKW1 only). The presence of multiple common sequences in {lambda}-clones was established by probing XbaI-digested DNAs with PCR products derived from the following regions of Tpo1-lp1 (number of positive cross-hybridizing clones indicated in brackets): 850–1250 (10/28), 1500–1900 (15/28), 11,500–11,900 (13/28), and with random-labeled 1.1-kb XbaI fragment (3700–4800, 24/28). Interspecific hybridizations were made with TCMK/TNC PCR probe from L. perenne Tpo1-lp1 (nts 2344–2694) or from Hordeum vulgare genomic DNA; equal amounts of XbaI-digested DNA were loaded for each species.


*  RESULTS
*TOP
*ABSTRACT
*MATERIALS AND METHODS
*RESULTS
*DISCUSSION
*LITERATURE CITED

Identification of transposon fragments:
In the course of a project characterizing the rye B chromosome an amplified fragment length polymorphism fragment amplified from Lindstrom wheat was sequenced and found to have significant similarity to the TnpD gene of Spm of maize and its equivalents in Tdc1 (carrot) and Tam1 (snapdragon; TBLASTX P values, 3e-15, 8e-13, and 4e-18, respectively). Oligonucleotides were designed on the basis of the most highly conserved motifs found in these alignments and used for PCR from genomic DNA of a variety of grasses (see MATERIALS AND METHODS). Products of the expected size were amplified, cloned, and sequenced from Lolium, barley, rye, maize, and diploid wheat and oat species (Fig 1). All are closely related and form a distinct lineage of the CACTA superfamily (Fig 2), which we have called Tpo1 (Transposon, Poaceae) to reflect its wide distribution. Species- and lineage-specific polymorphisms are seen in the Tpo1 sequences, indicating that the elements have continued to evolve during recent species divergence. The majority of these polymorphisms either do not alter the predicted peptide sequence or result in conservative amino acid changes, suggesting that the elements are still functional in contemporary genomes and do not represent ancient "fossils."



View larger version (213K):
In this window
In a new window
Download PPT slide
 
Figure 1. Tpo1 represents a distinct lineage of CACTA family transposons. Sequence alignment of TnpD-like sequences recovered from temperate Poaceae by PCR with TCMK and TNC primers is shown. Av-eriX, Avena eriantha; Av-satX, A. sativa; Av-strX, A. strigosa; Ho-vulX, H. vulgare; Ae-sqX, Aegilops squarosa; Trit-mX, Triticum monococcum; Trit-spX, T. speltoides; Trit-uX, T. urartu; Sec-cX, Secale cereale; Lol-perX, L. perenne; Lol-temX, L. temulentum.



View larger version (27K):
In this window
In a new window
Download PPT slide
 
Figure 2. Neighbor-joining tree showing relationship of CACTA family TnpD-like sequences. Species key is as for Fig 1, plus: Gm-tgm5, soy Tgm5 (RHODES and VODKIN 1988 Down); At-BAC, Arabidopsis thaliana accession AP000411; Am-tam1, snapdragon Tam1 (NACKEN et al. 1991 Down); Zm-shoot, maize Shooter-1 (PANAVAS et al. 1999 Down); Dc-tdc1, carrot Tdc1 (OZEKI et al. 1997 Down); Bn-1, Brassica napus, accession no. AJ245479; Sb-bac, sorghum accession no. AF114171; Os-1, rice accession no. AC027660; Os-2, rice accession no. AC027661. Tree is corrected for multiple substitutions based on 1000 bootstrap replicates (values given at nodes) rooted on Tgm5.

Isolation and characterization of a Lolium Tpo1 element:
A genomic library of L. perenne line Lp10, cloned in the {lambda}FIXII vector, was screened with L. perenne Tpo1 transposase fragments derived by PCR as above (MATERIALS AND METHODS). Numerous positive clones were found (see below), of which 28 were purified and used for further characterization. A candidate full-length element, designated Tpo1-lp1, was identified in {lambda}26 following Southern analysis, and the entire insert from this clone was subcloned and sequenced (Fig 3).



View larger version (35K):
In this window
In a new window
Download PPT slide
 
Figure 3. Organization of the {lambda}26 insert. (Top) The Tpo1-lp1 element is 12,485 bp long, bounded by 17-bp TIRs (arrowheads). A highly repeated 16-bp motif is found in three regions (ovals, number of repeats with >=75% identity indicated by numbers above). Three reading frames have been identified (cross-hatched boxes: homologies indicated below, orientation above). Repeated regions are indicated [27- and 32-bp tandem repeats (nts 1409–1489 and nts 6347–6466) indicated by vertical lines, 200-bp tandem repeat (nts 10,981–11,372) indicated by diagonal hatched boxes]. (Bottom) TnpA-like DBD homologies. Sequence key is as for Fig 1, plus: Tpo1-lp1, Lolium clone (this work); Sb-Cs1, sorghum Candystripe-1 gene 2 (CHOPRA et al. 1999 Down); Os-3, rice accession no. AP002484; Os-4, rice accession no. AQ273132.

In this clone, TIR sequences are present some 12.5 kb apart; the TIR itself is 17 bp long (CACTACTAGGAAAAGGC) and is highly similar to those of sorghum Candystripe and maize Spm (CACTATGTGAAAAAAGCTTA and CACTACAAGAAAA), while the transposon sequences flanking the TIR at either end contain multiple repeats of a 16-bp motif (TAGCAGTGGCGCACCA) distributed in both orientations over regions of several hundred base pairs, an organization typical of "cut-and-paste" transposons. The TIRs appear to be flanked by 3-bp duplications of host DNA, as seen at other CACTA family insertion sites. An open reading frame that is highly similar to Spm TnpD is found in the 5' half of the element, while a second gene related to Spm TnpA and Candystripe gene 2 appears to be present downstream of this in the same orientation. The putative second gene contains a region similar to the TnpA DBD (Fig 3); similarities outside of this region are weak, however. Between the two genes lies a central noncoding region that contains a complex arrangement of repeated motifs, including the only other occurrences of the 16-bp motif outside of the flanking regions; three repeats are found between nucleotides 5038 and 5111. As the 16-bp motif is likely to be bound by the TnpA-like protein, assuming similar behavior to Spm, these central repeats may be involved in autoregulation of TnpA-like gene expression. Three other notable duplications are seen. Both upstream and downstream of the TnpD-like gene are regions containing three short (27 and 32 bp, respectively) direct repeats, while downstream of the second gene lies a larger (~200 bp) direct repeat. Finally, a short region at the 3' end of the element is derived from an RNaseH-like gene. Such genes are typically found in retroelements and have not previously been described in DNA transposons. This region was characterized further and is described in more detail below.

Tpo1 is present at high copy number in L. perenne:
Screens of the Lolium {lambda}-library with the Tpo1 transposase PCR products yielded positive clones at a rate of ~1/50, potentially equivalent to 1 element/Mb or 5000 elements/genome. This number is well in excess of the tens of copies seen for maize elements such as Ac/Ds and En/Spm and even of the hundreds seen for "high copy" lines of Mutator. We therefore confirmed this number with further experiments. First, we confirmed that positive plaques represented bona fide transposon sequences by probing Southerns of the 28 purified clone DNAs. All of the clones hybridized to two or more nonoverlapping Tpo1-lp1 probes and 23 of the clones showed significant cross-hybridization to more than one region of the Tpo1-lp1 element (MATERIALS AND METHODS); moreover, the same prominent genomic restriction fragment (which does not overlap with the probe used for clone identification) is also found in most clones (Fig 4, details below). Second, we confirmed that library complexity had not been reduced during amplification. Restriction of clones with XbaI and NciI indicated that while some common fragments were present, no two clones were identical (Fig 4). Third, we confirmed that elements contained within the clones were also distinct by direct sequencing of PCR products of the transposase region; each of 12 clones analyzed was unique (not shown). Finally, we carried out FISH with the TnpD probe on a L. perenne x L. temulentum hybrid (Fig 5). Approximately 2500 discrete signals are seen to be distributed relatively evenly across both genomes (see below), indicating that the high copy number does not result from passive amplification within any particular chromosomal domain.



View larger version (111K):
In this window
In a new window
Download PPT slide
 
Figure 4. A 1.1-kb XbaI fragment is highly conserved in L. perenne {lambda}-clones. (Top) Ethidium-bromide-stained gel of XbaI-digested DNAs (W, weight markers; G, L. perenne genomic DNA). (Bottom) Southern hybridization of same gel, probed with Tpo1-lp1 1.1-kb XbaI fragment. Arrowheads, 1.1-kb fragments.



View larger version (58K):
In this window
In a new window
Download PPT slide
 
Figure 5. FISH of TnpD-like region to meiotic prophase chromosomes of a Lolium hybrid (L. perenne x L. temulentum). (Top) Overlaid confocal laser scanning microscope sections allow display of both entire genomes. (Bottom) A magnified view shows that the pattern of Tpo1 insertion may be similar in both species (indicated).

Conservation of the Tpo1 family in L. perenne:
At the resolution provided by the extended meiotic prophase chromosomes used for the FISH analysis, there often appear to be conserved patterns of Tpo1 signals in the two Lolium genomes (Fig 5, bottom). These two species are sufficiently divergent that genomic in situ hybridization may be used to distinguish each chromosome set (JENKINS et al. 2000 Down), and even given typical transposition rates of <=10-4/element/generation, it is surprising that such patterns should have been preserved. The most obvious explanation for this observation is that many elements are ancient and inactive. However, if only TIRs and a subset of subterminal repeats are required in cis, nonautonomous elements undergoing inactivation by gradual sequence degeneration may be expected to continue to show movement mediated by autonomous elements for long periods of time. Open reading frames may therefore be expected to show considerable divergence in a population of "fixed" transposons that have been inactivated by sequence degeneration. This does not appear to be the case. Our preliminary characterization of the L. perenne {lambda}-clones indicated that not only were the Tpo1 sequences sufficiently well conserved to allow cross-hybridization under stringent conditions, but also some restriction sites were also maintained in the majority of clones. This is most strikingly demonstrated by the conservation of a 1.1-kb XbaI fragment found within the TnpD-like gene (Fig 4). Of 24 cross-hybridizing clones, 15 maintain both XbaI sites.

An alternative mechanism that could rapidly generate a population of inert elements stems from the susceptibility of transposons to deletion during gap repair of abortive transposition events. Deletions occur frequently in the Ac/Ds system, for example, and account for the large excess of nonautonomous Ds elements over the intact Ac parent. We therefore screened {lambda}-clones for the presence of significant numbers of deletion derivatives. To avoid the bias inherent in using clones preselected to contain a specific Tpo1 region, we probed filters of a sample of the unpurified genomic library directly. A total of 137 plaques cross-hybridized to one or more Tpo1-lp1 fragments (Fig 6). Only 15 of these (11%) gave a pattern that was unambiguously derived from a rearranged or divergent genomic element. A single plaque displayed the pattern expected for the largest internal deletion detectable with these probes, whereas two plaques hybridized only to the TnpD-like probe, a pattern that is not generated by internal deletion, suggesting that random processes may be as important as gap repair in generating Tpo1 derivatives. As there may be some bias in library representation of genomic sequences and/or rearrangement of clones, we also carried out FISH with TnpA-like and right-hand subterminal probes from Tpo1-lp1 (Fig 7). The majority of targets showed coincident labeling with both probes, with a minority (~10%) labeled with the subterminal region only. Occasionally only TnpA-like hybridization was seen; as with the {lambda}-screens, this suggests that a low level of random rearrangement or divergence acts on the elements. Again, there was no evidence for a significant excess of deletion derivatives, and it seems unlikely that the proportion seen is abundant enough to inactivate the full-length elements by a simple titration mechanism.



View larger version (29K):
In this window
In a new window
Download PPT slide
 
Figure 6. Internal deletions of Lolium Tpo1 elements are relatively rare; a summary of {lambda}-library hybridization to Tpo1-lp1 probes is shown. Hybridization patterns are indicated by solid (positive) or open (negative) boxes at appropriate positions within the Tpo1 "map" (not to scale). The first five patterns may be generated by the presence of part of an intact genomic element in the phage clone and contain the most frequent classes; the final four can be generated only by rearranged or divergent elements and are less abundant.



View larger version (53K):
In this window
In a new window
Download PPT slide
 
Figure 7. Tpo1-lp1 TnpA- and 3' subterminal probes map to coincident sites by FISH. (A) Subterminal sequences detected by biotin labeling. (B) TnpA-like sequences detected by digoxigenin labeling. (C) Composite of A and B. A–C detail the meiotic prophase of the L. perenne x L. temulentum hybrid; arrows indicate occasional sites detected by a single probe (red, TnpA-like only; green, 3' subterminal only).

We also used PCR to specifically amplify Tpo1 subterminal regions from L. perenne, using a TIR-specific oligonucleotide in combination with oligonucleotides based on conserved peptide motifs within the TnpD- and TnpA-like reading frames. Products recovered were predominantly of the size predicted for Tpo1-lp1, despite the large size of these products (2.2 and 4 kb, respectively), which would have allowed preferential amplification of deletion templates. It therefore appears unlikely that small deletions of essential subterminal sequences occurred at significant levels. No products representing large internal deletions (i.e., having TIRs at both ends) were recovered.

Tpo1 conservation in other genomes:
The distribution of Tpo1 elements in other genomes was examined by Southern hybridization. Under standard conditions, a probe derived from the Tpo1-lp1 1.1-kb XbaI fragment hybridizes predominantly to a single XbaI fragment in the L. perenne genome (Fig 8A), although a small number of minor bands are also present. Some cross-hybridization to barley, rye, wheat, and oat is seen. In these species the signal is predominantly "smeared" but relatively strong bands are visible in oat and, to a lesser extent, in rye. Reducing the stringency of hybridization enhanced the relative strength of these signals but did not generate significant additional bands (Fig 8B). Bands became detectable in the maize DNA, however. "Universal" fragments of ~10 kb, seen in all genomes, including Arabidopsis, probably represent spurious hybridization to the rDNA repeat unit, as their positions correspond to bands visible in the ethidium-stained gel (Fig 8C). As the region of Tpo1-lp1 used in these experiments may not be highly conserved, the filter was rehybridized with PCR product amplified from the barley genome, using the oligonucleotides and conditions first employed to define the Tpo1 family. Again, multiple bands were seen (Fig 8D) but cross-hybridization was predominantly concentrated in a small number of bands, particularly in oat and wheat, where a single major band is found. In barley itself, one band is significantly stronger although at least eight other bands are seen, while in rye multiple bands of an equivalent intensity are seen. These patterns, and the strength of the cross-hybridization, suggest that Tpo1 elements are present at high copy number in other large-genome Poaceae species and that these may be organized into relatively homogeneous subfamilies in at least some species.



View larger version (120K):
In this window
In a new window
Download PPT slide
 
Figure 8. Tpo1 cross-hybridizing sequences in other species. (A) Tpo1-lp1 TnpD-like probe, high stringency, short exposure. (B) Tpo1-lp1 TnpD-like probe, low stringency, long exposure. (C) Ethidium-bromide-stained gel. (D) Barley TnpD-like probe, high stringency, long exposure. Lanes: 1, A. thaliana; 2, Brachypodium distachyon; 3, rice; 4, maize; 5, L. perenne; 6, barley; 7, rye; 8, wheat; 9, oat; 10, weight markers.

The number of potential Tpo1 elements present in the barley genome was estimated by screening a representative genomic library (Stratagene) with a TnpD-like probe. Cross-hybridizing clones were present at a frequency of ~1/100. This is comparable with the density of Tpo1 elements identified in the L. perenne library (see above) and indicates that Tpo1 copy number in barley is also >1000 copies per haploid genome.

Finally, the hybridization experiments suggest that rye contains more Tpo1-related sequences and a greater variety of subfamilies than do other Triticeae species. Rye is unusual in containing more heterochromatin than related species; it also frequently supports the presence of a supernumerary, or B, chromosome, which we have previously shown to be likely to have arisen by rapid evolution from A genome components (LANGDON et al. 2000A Down). It was therefore of interest to examine whether Tpo1 elements may play a structural role in these rearrangements. FISH with Tpo1 probes derived from both rye and barley genomes indicated that related sequences were relatively evenly distributed over both A and B genomes, and there was no evidence for an increased concentration in heterochromatic or supernumerary chromatin (rye probes, Fig 9; barley probes not shown).



View larger version (94K):
In this window
In a new window
Download PPT slide
 
Figure 9. Tpo1 elements are abundant and evenly dispersed on rye A and B chromosomes. The rye TnpD-like sequence was used as a FISH probe on mitotic (root tip) chromosomes. Arrowheads, B chromosomes.

An RNaseH domain in the Tpo family:
An interrupted reading frame at the 3' end of Tpo1-lp1 shows significant homology with a conserved motif of RNaseH genes (CERRITELLI et al. 1998 Down; Fig 10). No requirement for RNAseH activity is expected for transposon function and no such genes have been detected in other elements. However, examples of transposons acquiring host genes or other mobile elements are known (CHOPRA et al. 1999 Down; TAKAHASHI et al. 1999 Down), and RNaseH genes are expected to be among the most numerous in grass genomes, as they are contained within retrotransposon polyprotein reading frames. The Tpo1-lp1 sequence may therefore represent a chance rearrangement or insertion without functional significance. It appears to be derived from an ancient event, however, as cross-hybridizing sequences are found in 11 of the isolated {lambda}-clones, and PCRs with oligonucleotides anchored within the motif region and the downstream TIR or TnpA-like gene amplify products of the predicted size from Tpo1 homologs in barley (MATERIALS AND METHODS). In addition, a number of expressed sequence tag (EST) database entries for barley have significant similarity to this region. These entries fall into two classes: the first is highly similar to the Tpo1-lp1 clone from 11,258 to 11,522 (88% consensus identity) but is unrelated upstream of this region (accession nos. AW982961, AW983166, and BE216619), while the second has a shorter region of similarity to Tpo1-lp1 but in addition shares an upstream region of ~180 bp, which is highly similar to a barley Tpo1-related genomic database entry (EST accession nos. AW983299 and BE060537; genomic accession no. U76261). Downstream of the region of common similarity with the ESTs, the barley genomic sequence and Tpo1-lp1 show extensive colinearity of sequence motifs and TIRs, indicating that the barley sequence contains the 3' end of a Tpo1-like element (Fig 11; see DISCUSSION). The simplest interpretation of the EST similarities is that each class represents a processed transcript from one or another of two Tpo1 subfamilies; alternatively, the junctions between common and unrelated sequences may result from further genomic rearrangements. The ESTs are unlikely to produce functional peptides as key codons are mutated in some clones, while translation would need to be initiated within the unrelated upstream regions. The probability that the motif represents a vestige of a historical rearrangement is supported by the presence of polyadenine stretches in the two genomic clones at positions similar to the 3' ends of the ESTs, suggesting that an RNaseH pseudogene may have been inserted in an ancestral Tpo1 element prior to Lolium barley divergence.



View larger version (47K):
In this window
In a new window
Download PPT slide
 
Figure 10. A region of Tpo1-lp1 is derived from an RNaseH gene. Hv-estX, barley ESTs (accession nos. BE454664, BE216613, and AW982961); Yeast, Saccharomyces cerevisiae (Q04740); Human (AAC09261); Mouse (NP035405); Chick (BF723936); Dros, Drosophila melanogaster (AF032921); T-bruc, Trypanosoma brucei (AAC47537); C-fasc, Crithidia fasciculata (Q07762).



View larger version (61K):
In this window
In a new window
Download PPT slide
 
Figure 11. The organization of Tpo1 termini is highly conserved despite selective sequence divergence. Tpo1-lp1 sequence alignment with barley database entry (accession no. U76261) is shown. Uppercase, presumed functional motifs: TIRs in white text and shaded; universal motif (TGGCGCACCA consensus) is in white text and black; Lolium-specific motif (TAGCAG) is underlined and shaded; barley-specific motif (ATACTAA) is underlined and italicized.


*  DISCUSSION
*TOP
*ABSTRACT
*MATERIALS AND METHODS
*RESULTS
*DISCUSSION
*LITERATURE CITED

The Tpo1 family described here represents a new branch of the plant CACTA transposon superfamily. Its members are unusual in consistently having a high copy number in species with large genomes, although the L. perenne subfamily, which has been most extensively characterized, appears to be conventional in its internal organization and coding content. The single anomalous feature identified so far, an RNaseH-like peptide motif, appears to represent a historical accident without current functional significance, although its presence in elements from divergent species provides evidence of common descent. Here, we discuss some of the properties of the Tpo1 family that may have led to its abundance.

One of the most striking aspects of the L. perenne Tpo1 elements is the lack of the deletion derivatives that are a common feature of other transposon families. These derivatives appear to be created during repair of the double-strand breaks (DSBs) caused by element excision by a pathway involving extension of the DSB free ends on homologous templates provided by sister chromatids and/or sequences on other chromosomes (RUBIN and LEVY 1997 Down). A characteristic of this pathway is that DNA synthesis is frequently interrupted, allowing slippage and other forms of mispairing and strand exchange, which facilitates annealing of the extended ends of the broken chromosome and repair of the DSB but reduces the probability of regenerating an intact copy of the excised sequence. Recently YAMASHITA et al. 1998 Down demonstrated that the formation of unusually strong secondary structures by the subterminal regions of Tam3 may interfere with DSB repair by this route, blocking the creation of detectable nonautonomous deletion derivatives and maintaining the extremely high level of homogeneity seen for the Tam3 subfamily. Subterminal secondary structure appears to play a widespread role in determining the efficiency of gap repair of transposons as there is a correlation across a number of families, from both animals and plants, between predicted free energies of single-stranded end sequences and the tendency of the family to increase in copy number (YAMASHITA et al. 1999 Down).

Tpo1-lp1 is predicted to have relatively strong subterminal secondary structures, comparable with those of Tam3 (-81 kcal and -79 kcal for first 5' and last 3' 300 bp, compared to -127 kcal and -83 kcal, respectively) and higher than those of Spm (-47 kcal and -70 kcal) or other families analyzed by YAMASHITA et al. 1999 Down. There is some empirical support for these predictions in that relevant regions of Tpo1-lp1 generate strong stops in standard sequencing reactions (not shown), as has also been reported for Tam3. Although we do not yet know the fate of DSBs created by Tpo1 excision, the lack of Tpo1 deletion derivatives is consistent with a protective role for the elements' ends. The importance of the organization of these regions is also indicated by a comparison of the Tpo1-lp1 3' sequence with the barley database entry mentioned above.

The two sequences are colinear over a 1-kb region, extending from the putative RNaseH motif to the downstream TIR, and are clearly related but show interesting patterns of divergence. In particular, a symmetrical core region of the Tpo1-lp1 subterminal motifs (TGGTGCGCCA) shows strong conservation of both sequence and organization, leading to similar secondary structure predictions for each sequence, while the remaining part of the Tpo1-lp1 motif (TAGCAG) has been largely replaced by the sequence ATACTAA in the barley entry. This last replacement represents not only a divergent sequence but also a change in the motif's size (i.e., 7 rather than 6 bases). Such a radical change at each of 13 motifs implies a strong selection, which has, moreover, maintained a very similar organization despite the obvious potential for replication slippage or aberrant recombination to act at the repeated motifs. As the motifs are expected to provide binding sites for protein(s) involved in generating the transposition complex, we cannot rule out the possibility that selection is entirely directed by the constraints of the complex's architecture, but it appears highly likely that, as with Tam3, Tpo1 elements are protected against the most significant cause of structural rearrangement. Tam3, however, has a low copy number (~60; MARTIN and LISTER 1989 Down), typical of most transposon families and almost two orders of magnitude below that seen for Tpo1 in large cereal genomes, a discrepancy that does not correlate with genome size (estimated 1C value of 1568 Mb for the Tam3 host, Antirhhinum majus, compared to 4851 Mb for L. perenne and 5439 Mb for H. vulgare; BENNETT et al. 2000 Down).

Another striking aspect of the Tpo1 distribution is that it appears to have a relatively even copy number across divergent species, whereas the majority of transposon families described to date shows great variation even within a species. In the most extreme cases (for example, Mutator or P), this variation may be ascribed to a recent aggressive colonization coupled with an almost replicative mode of transposition, which rapidly generates large numbers of nonautonomous elements. Both the damage caused to the host genome by transposition events and the load imposed on autonomous elements by the increasing number of derivatives are expected to create an unstable situation in which either the transposon family or the active host lines may become extinct, leading to a continuous boom-and-bust pattern of colonization. However, while evolution of strong subterminal secondary structures may have allowed the Tpo1 ancestor to reduce the rates of both processes and hence to adopt a less erratic but more persistent method of colonization, if the current copy numbers reflect a slow increase since a common ancestor, then multiple subfamilies forming a star phylogeny might be expected, with many inactive and degenerate elements blurring the distinction between groups. While our analysis of TnpD divergence provides some evidence for such a phylogeny within the Triticeae, it is clear that there are distinct lineages for Lolium and oat. In addition, the presence of high-copy and lineage-specific subgroups within such divergent species as oat, wheat, and Lolium suggests, rather, that the majority of Tpo1 elements are in some way generated from relatively recent progenitors, whether by active transposition or by a more passive process such as gene conversion. The conservation of copy number could then reflect a genome's capacity to support the family, with subfamilies competing for available sites. Such a competitive scenario would also seem to be supported by the degree of selection implied by the Tpo1-lp1/barley subterminal divergence.

It is tempting to speculate that the Tpo1 family has acquired a mechanism to direct transposition to specific sites within the genome, as this could account for both the host's ability to tolerate large numbers of elements (assuming nongenic or redundant sites are occupied) and competition between elements (assuming a limited number of sites are available). Examples in a wide range of species of mobile elements target particular genomic niches to minimize host disruption, and we have described elsewhere evidence that a family of retrotransposons has successfully adopted this strategy in grasses and cereals by active colonization of centromeres (LANGDON et al. 2000B Down). It is also striking that some cereal miniature inverted-repeat transposable element (MITE) families appear to be specifically clustered around intergenic matrix attachment regions (MARs; AVRAMOVA et al. 1998 Down; TIKHONOV et al. 2000 Down). MITEs are likely to be nonautonomous transposons, several hundred base pairs long and frequently present in thousands of copies. The association with MARs may be functional, possibly based on the strong intrinsic potential of MITEs to adopt secondary structures, suggesting that they may even carry out a useful role for the host (as, for example, in the case of the Drosophila retroelements that replace conventional telomeres; reviewed in PARDUE et al. 1996 Down). It is likely that insertion specificity is involved, given that other MITE families show preferential localization in genic sequences (ZHANG et al. 2000 Down). Directed transposition by Tpo1 is, therefore, a possibility that will be examined in future work; however, even if it occurs, this seems unlikely to provide the full explanation for the observed distribution. First, insertion at specific locations may be expected to result in "nests" of transposons of varying ages, similar to the pattern described for intergenic retrotransposons in maize (SANMIGUEL et al. 1996 Down, SANMIGUEL et al. 1998 Down), which does not appear to have occurred. Second, while insertion directed to phenotypically neutral locations avoids one cause of host load, the problem of large numbers of synchronous excisions remains and may even be accentuated if all internally similar elements are also embedded in similar chromosomal contexts.

One resolution to this problem would be if Tpo1 elements are now at a sufficiently high density in the host genomes that they interact via complexes assembled at their subterminal motifs before excision occurs (i.e., if interelement interactions frequently replace the intraelement interactions seen in conventional transposition). Such interactions have been found occasionally in other systems and can lead to hybrid element insertion, element replacement, and compound movements (reviewed in GRAY 2000 Down). In the first two cases elements would appear to persist in conserved genomic locations although the predominant subfamilies may change over time and flanking sequences may be eroded (INGRAM et al. 1997 Down). DSBs and new insertions into host genes would be avoided, while erosion of flanking DNA by elements inserted into intergenic regions could provide some benefit in counteracting retrotransposon colonization. An evolutionarily stable situation may then develop. We are not aware of any precedent for this explanation, but it is likely that most transposon colonizations would prove lethal to the host long before the genome became saturated with sufficient elements. The unusually high proportion of retrotransposons in grass genomes suggests a possible mechanism by which the initial Tpo1 colonization could have rapidly passed through this lethal stage. Briefly, by inserting in an active retrotransposon, the Tpo1 ancestor could have become dispersed to large numbers of intergenic regions at a rate far higher than could be achieved by conventional transposition and within a timescale that would leave all elements essentially identical in their potential to assemble transposition complexes. As all elements may then become active under similar conditions (e.g., at a particular transposase concentration), interelement interactions may be favored. Beyond a threshold, it may become advantageous to the host to increase element numbers so that exclusively interelement interactions occur and so that further chimeric retrotransposition or appropriate recombination in progeny is favored. Ultimately, sufficient elements may be established that the host's genes are protected while turnover in the transposon population continues, with ancestral sequences being replaced by newer, more aggressive variants. This model, though speculative, gives rise to a number of testable predictions (for example, polymorphism of elements at conserved sites, deletion of flanking sequences), which we are currently investigating.


*  FOOTNOTES

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY089999 and AF492361, AF492362, AF492363, AF492364, AF492365, AF492366, AF492367, AF492368, AF492369, AF492370, AF492371, AF492372, AF492373, AF492374, AF492375, AF492376, AF492377, AF492378, AF492379, AF492380, AF492381, AF492382, AF492383. Back


*  ACKNOWLEDGMENTS

We thank Joe Gallagher for providing the L. perenne library, Chris Pollock for helpful discussions, and Steve Taylor for help with microscopy. This work was supported by BBSRC AFD grant D10173 to I.P.K., G.J. and R.N.J.

Manuscript received January 26, 2001; Accepted for publication November 21, 2002.


*  LITERATURE CITED
*TOP
*ABSTRACT
*MATERIALS AND METHODS
*RESULTS
*DISCUSSION
*LITERATURE CITED

ATKINSON, P. W., W. D. WARREN, and D. A. OBROCHTA, 1993  The Hobo transposable element of Drosophila can be cross-mobilized in houseflies and excises like the Ac element of maize. Proc. Natl. Acad. Sci. USA 90:9693-9697.[Abstract/Free Full Text]

AVRAMOVA, Z., A. TIKHONOV, M. S. CHEN, and J. L. BENNETZEN, 1998  Matrix attachment regions and structural colinearity in the genomes of two grass species. Nucleic Acids Res. 26:761-767.[Abstract/Free Full Text]

BENNETT, M. D., A. V. COX and I. J. LEITCH, 2000 Angiosperm DNA C-values database. http://www.rbgkew.org.uk/cval/database1.html.

CERRITELLI, S. M., O. Y. FEDOROFF, B. R. REID, and R. J. CROUCH, 1998  A common 40 amino acid motif in eukaryotic RNases H1 and caulimovirus ORF VI proteins binds to duplex RNAs. Nucleic Acids Res. 26:1834-1840.[Abstract/Free Full Text]

CHOPRA, S., V. BRENDEL, J. B. ZHANG, J. D. AXTELL, and T. PETERSON, 1999  Molecular characterization of a mutable pigmentation phenotype and isolation of the first active transposable element from Sorghum bicolor. Proc. Natl. Acad. Sci. USA 96:15330-15335.[Abstract/Free Full Text]

CRESSE, A. D., S. H. HULBERT, W. E. BROWN, J. R. LUCAS, and J. L. BENNETZEN, 1995  Mu1-related transposable elements of maize preferentially insert into low copy number DNA. Genetics 140:315-324.[Abstract]

FEDOROFF, N., M. SCHLAPPI, and R. RAINA, 1995  Epigenetic regulation of the maize Spm transposon. Bioessays 17:291-297.[Medline]

GIERL, A., H. SAEDLER, and P. A. PETERSON, 1989  Maize transposable elements. Annu. Rev. Genet. 23:71-85.[Medline]

GRAY, Y. H. M., 2000  It takes two transposons to tango—transposable-element-mediated chromosomal rearrangements. Trends Genet. 16:461-468.[Medline]

INGRAM, G. C., S. DOYLE, R. CARPENTER, E. A. SCHULTZ, and R. SIMON et al., 1997  Dual role for fimbriata in regulating floral homeotic genes and cell division in Antirrhinum. EMBO J. 16:6521-6534.[Medline]

JENKINS, G., J. HEAD, and J. W. FORSTER, 2000  Probing meiosis in hybrids of Lolium (Poaceae) with a discriminatory repetitive genomic sequence. Chromosoma 109:280-286.[Medline]

KUMAR, A. and J. L. BENNETZEN, 1999  Plant retrotransposons. Annu. Rev. Genet. 33:479-532.[Medline]

LANGDON, T., C. SEAGO, R. N. JONES, H. OUGHAM, and H. THOMAS et al., 2000a  De novo evolution of satellite DNA on the rye B chromosome. Genetics 154:869-884.[Abstract/Free Full Text]

LANGDON, T., C. SEAGO, M. MENDE, M. LEGGETT, and H. THOMAS et al., 2000b  Retrotransposon evolution in diverse plant genomes. Genetics 156:313-325.[Abstract/Free Full Text]

MARTIN, C. and C. LISTER, 1989  Genome juggling by transposons: Tam3-induced rearrangements in Antirrhinum majus. Dev. Genet. 10:438-451.[Medline]

MASSON, P., M. STREM, and N. FEDOROFF, 1991  The Tnpa and Tnpd gene-products of the Spm element are required for transposition in tobacco. Plant Cell 3:73-85.[Abstract/Free Full Text]

NACKEN, W. K., R. PIOTROWIAK, H. SAEDLER, and H. SOMMER, 1991  The transposable element Tam1 from Antirrhinum majus shows structural homology to the maize transposon En/Spm and has no sequence specificity of insertion. Mol. Gen. Genet. 228:201-208.[Medline]

OZEKI, Y., E. DAVIES, and J. TAKEDA, 1997  Somatic variation during long term subculturing of plant cells caused by insertion of a transposable element in a phenylalanine ammonia-lyase (PAL) gene. Mol. Gen. Genet. 254:407-416.[Medline]

PANAVAS, T., J. WEIR, and E. L. WALKER, 1999  The structure and paramutagenicity of the R-marbled haplotype of Zea mays. Genetics 153:979-991.[Abstract/Free Full Text]

PARDUE, M. L., O. N. DANILEVSKAYA, K. LOWENHAUPT, F. SLOT, and K. L. TRAVERSE, 1996  Drosophila telomeres: new views on chromosome evolution. Trends Genet. 12:48-52.[Medline]

RAINA, R., M. SCHLAPPI, B. KARUNANANDAA, A. ELHOFY, and N. FEDOROFF, 1998  Concerted formation of macromolecular suppressor-mutator transposition complexes. Proc. Natl. Acad. Sci. USA 95:8526-8531.[Abstract/Free Full Text]

RHODES, P. R. and L. O. VODKIN, 1988  Organization of the Tgm family of transposable elements in soybean. Genetics 120:597-604.[Abstract/Free Full Text]

RUBIN, E. and A. A. LEVY, 1997  Abortive gap repair: underlying mechanism for Ds element formation. Mol. Cell. Biol. 17:6294-6302.[Abstract]

SANMIGUEL, P., A. TIKHONOV, Y. K. JIN, N. MOTCHOULSKAIA, and D. ZAKHAROV et al., 1996  Nested retrotransposons in the intergenic regions of the maize genome. Science 274:765-768.[Abstract/Free Full Text]

SANMIGUEL, P., B. S. GAUT, A. TIKHONOV, Y. NAKAJIMA, and J. L. BENNETZEN, 1998  The paleontology of intergene retrotransposons of maize. Nat. Genet. 20:43-45.[Medline]

SUONIEMI, A., D. SCHMIDT, and A. H. SCHULMAN, 1997  BARE-1 insertion site preferences and evolutionary conservation of RNA and cDNA processing sites. Genetica 100:219-230.[Medline]

TAKAHASHI, S., Y. INAGAKI, H. SATOH, A. HOSHINO, and S. IIDA, 1999  Capture of a genomic HMG domain sequence by the En/Spm-related transposable element Tpn1 in the Japanese morning glory. Mol. Gen. Genet. 261:447-451.[Medline]

THOMPSON, J. D., T. J. GIBSON, F. PLEWNIAK, F. JEANMOUGIN, and D. G. HIGGINS, 1997  The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876-4882.[Abstract/Free Full Text]

TIKHONOV, A. P., J. L. BENNETZEN, and Z. V. AVRAMOVA, 2000  Structural domains and matrix attachment regions along colinear chromosomal segments of maize and sorghum. Plant Cell 12:249-264.[Abstract/Free Full Text]

TRENTMANN, S. M., H. SAEDLER, and A. GIERL, 1993  The transposable element En/Spm-encoded TNPA protein contains a DNA binding and a dimerization domain. Mol. Gen. Genet. 238:201-208.[Medline]

UPADHYAYA, K. C., H. SOMMER, E. KREBBERS, and H. SAEDLER, 1985  The paramutagenic line Niv-44 has a 5 kb insert, Tam-2, in the chalcone synthase gene of Antirrhinum-Majus. Mol. Gen. Genet. 199:201-207.

YAMASHITA, S., T. MIKAMI, and Y. KISHIMA, 1998  Tam3 in Antirrhinum majus is exceptional transposon in resistance to alteration by abortive gap repair: identification of nested transposons. Mol. Gen. Genet. 259:468-474.[Medline]

YAMASHITA, S., T. TAKANO-SHIMIZU, K. KITAMURA, T. MIKAMI, and Y. KISHIMA, 1999  Resistance to gap repair of the transposon Tam3 in Antirrhinum majus: a role of the end regions. Genetics 153:1899-1908.[Abstract/Free Full Text]

ZHANG, Q., J. ARBUCKLE, and S. R. WESSLER, 2000  Recent, extensive, and preferential insertion of members of the miniature inverted-repeat transposable element family Heartbreaker into genic regions of maize. Proc. Natl. Acad. Sci. USA 97:1160-1165.[Abstract/Free Full Text]




This article has been cited by other articles:


Home page
DNA ResHome page
M. Tomita, K. Shinohara, and M. Morimoto
Revolver is a New Class of Transposon-like Gene Composing the Triticeae Genome
DNA Res, February 1, 2008; 15(1): 49 - 62.
[Abstract] [Full Text] [PDF]


Home page
Proc. Natl. Acad. Sci. USAHome page
K. M. Devos, J. Ma, A. C. Pontaroli, L. H. Pratt, and J. L. Bennetzen
Analysis and mapping of randomly chosen bacterial artificial chromosome clones from hexaploid bread wheat
PNAS, December 27, 2005; 102(52): 19243 - 19248.
[Abstract] [Full Text] [PDF]


Home page
Proc. Natl. Acad. Sci. USAHome page
X. Zhang and S. R. Wessler
Genome-wide comparative analysis of the transposable elements in the related species Arabidopsis thaliana and Brassica oleracea
PNAS, April 13, 2004; 101(15): 5589 - 5594.
[Abstract] [Full Text] [PDF]