Skip to main content

Genomic neighborhoods for Arabidopsisretrotransposons: a role for targeted integration in the distribution of the Metaviridae



Retrotransposons are an abundant component of eukaryotic genomes. The high quality of the Arabidopsis thaliana genome sequence makes it possible to comprehensively characterize retroelement populations and explore factors that contribute to their genomic distribution.


We identified the full complement of A. thaliana long terminal repeat (LTR) retroelements using RetroMap, a software tool that iteratively searches genome sequences for reverse transcriptases and then defines retroelement insertions. Relative ages of full-length elements were estimated by assessing sequence divergence between LTRs: the Pseudoviridae were significantly younger than the Metaviridae. All retroelement insertions were mapped onto the genome sequence and their distribution was distinctly non-uniform. Although both Pseudoviridae and Metaviridae tend to cluster within pericentromeric heterochromatin, this association is significantly more pronounced for all three Metaviridae sublineages (Metavirus, Tat and Athila). Among these, Tat and Athila are strictly associated with pericentromeric heterochromatin.


The non-uniform genomic distribution of the Pseudoviridae and the Metaviridae can be explained by a variety of factors including target-site bias, selection against integration into euchromatin and pericentromeric accumulation of elements as a result of suppression of recombination. However, comparisons based on the age of elements and their chromosomal location indicate that integration-site specificity is likely to be the primary factor determining distribution of the Athila and Tat sublineages of the Metaviridae. We predict that, like retroelements in yeast, the Athila and Tat elements target integration to pericentromeric regions by recognizing a specific feature of pericentromeric heterochromatin.


Endogenous retroviruses and long terminal repeat (LTR) retrotransposons (collectively called retroelements) generally comprise a significant portion of higher eukaryotic genomes. Dismissed as parasitic or 'junk' DNA, these sequences have traditionally received less attention than sequences contributing to the functional capacity of the organism. This perspective has changed with the completion of several eukaryotic genome sequences. The contributions of retroelements to genome content range from 3% in baker's yeast to 80% in maize [1, 2]. Retroelement abundance has resulted in increased appreciation of the important evolutionary role they play in shaping genomes, fueling processes such as mutation, recombination, sequence duplication and genome expansion [3].

The impact of retroelements on their hosts is not without constraint: the host imposes an environmental landscape (the genome) within which retroelements must develop strategies to persist. Retroelement cDNA insertion directly impacts on the host's genetic material, making this step a likely target for regulatory control. Transposable elements (TEs) in some systems utilize mechanisms that direct integration to specific chromosomal sites or safe havens [4, 5]. For example, the LTR retrotransposons of yeast are associated with domains of heterochromatin or sites bound by particular transcriptional complexes such as RNA polymerase III [69]. These regions are typically gene poor and may enable yeast retrotransposons to replicate without causing their host undue damage [10]. Non-uniform chromosomal distributions are observed in other organisms as well. For example, many retroelements of Arabidopsis thaliana and Drosophila melanogaster are clustered in pericentromeric heterochromatin [11, 12]. However, beyond the yeast model, it is not known whether retroelements generally seek safe havens for integration.

The genome of A. thaliana is ideal for exploring processes that influence the chromosomal distribution of retroelements. A. thaliana retroelement diversity has been analyzed previously, preparing the way for this study [1315]. In contrast to the genomes of Saccharomyces cerevisiae, Schizosaccharomyces pombe and Caenorhabditis elegans, which have relatively few retroelements, A. thaliana has a diverse mobile element population whose physical distribution can be described in detail. Another benefit of A. thaliana stems from the fact that in contrast to most other 'completely sequenced' eukaryotic genomes, the A. thaliana genome sequence better represents chromosomal DNA of all types, including sequences within heterochromatin [11]. Here we undertake a comprehensive characterization of the LTR retroelements in the well characterized genome of A. thaliana to better understand the factors contributing to their genomic distribution.



All reverse transcriptases in the A. thaliana genome were identified by iterated BLAST searches (Figure 1). The query sequences were representative reverse transcriptases from the Metaviridae, Pseudoviridae and non-LTR retrotransposons (Table 1). LTRs (if present) were assigned to each reverse transcriptase using the software package RetroMap (Figure 1, see also Materials and methods). Although the coding sequences of many elements with flanking LTRs were degenerate, they are referred to as full-length or complete elements (FLE) to indicate that two LTRs or LTR fragments could be identified. 5' LTRs from FLEs and published A. thaliana elements were used to identify solo LTRs in the genome by BLAST searches. The final data set consisted of three insertion subtypes: 376 FLEs, 535 reverse transcriptase (RT)-only hits, and 3,268 solo LTRs (Table 2). These sequences comprise 3,951,101 bases or 3.36% of the total 117,429,178 bases in The Institute of Genomic Research (TIGR) 7 January 2002 version of the genome. Overall, chromosomal retroelement content ranged from 2.64% (chromosome 1) to 4.31% (chromosome 3). Chromosome 4 contained the fewest FLEs (53) and solo LTRs (449), whereas chromosome 3 had the most (92 FLEs and 1,053 solo LTRs).

Figure 1

Assembling the retroelement dataset. (a) Flow chart for the generation of the dataset. The shaded region denotes steps coordinated by the RetroMap software. (Eprobe refers to a BLAST query sequence) (b) LTR prediction. The innermost direct repeats identified in sequences flanking the original BLAST hit are assigned as LTRs. The repeats delimit the boundaries of the full-length LTR retrotransposons.

Table 1 Retroelement species used as BLAST probes
Table 2 A. thaliana LTR retroelements by chromosome

Element subtypes (FLE, RT-only and solo LTRs) were sorted into taxonomic groupings using the formal taxonomic nomenclature assigned to retrotransposons [16, 17]. Our analysis identified numerous insertions for both the Pseudoviridae (211 FLE/82 RT-only/483 solo LTRs) and Metaviridae (168 FLE/142 RT-only/2,803 solo LTRs). The non-LTR retrotransposons lack flanking direct repeats, and therefore only reverse transcriptase information is provided in this study; 311 non-LTR retrotransposon reverse transcriptases were identified. Unlike the Pseudoviridae, A. thaliana Metaviridae elements can easily be divided into sublineages, which are referred to as the Tat, Athila and Metavirus elements [14, 18] (Figure 2). Our method identified 42 Tat FLEs, 38 Athila FLEs and numerous divergent Metavirus elements (82 FLE). No evidence was found for BEL or DIRS retroelements.

Figure 2

Arabidopsis thaliana Metaviridae and Pseudoviridae reverse transcriptase diversity. Phylogenetic trees used in this figure are adapted from [14,18]. Each tree is based on ClustalX [56] alignments of reverse transcriptase domains for elements in a given family. Neighbor-joining trees (10,000 bootstrap repetitions) were generated using MEGA2 [57]. The non-LTR retrotransposon Ta11 served as the root for both trees. The three Metaviridae sublineages are boxed.

The Metaviridae make up 2.34% of the A. thaliana genome, whereas the Pseudoviridae represent only 1.25% of the total genomic DNA. This difference is accounted for largely by the longer average size of Metaviridae FLEs (8,952 nucleotides) and solo LTRs (447 nucleotides) when contrasted with the Pseudoviridae FLEs (5,336 nucleotides) and solo LTRs (187 nucleotides) (data not shown). Among the subgroups of the Metaviridae, the average length of Metaviruses is closer to that of the Pseudoviridae than to the mean lengths of the Athila and Tat lineages. The Pseudoviridae are also more uniformly sized than the Metaviridae. A second factor contributing to the abundance of Metaviridae is that they have approximately six times more solo LTRs than the Pseudoviridae, even though numbers of complete elements are similar between families (Table 2). The ratios of solo LTRs to FLEs also clearly differ between the Metaviridae (16.7:1) and Pseudoviridae (2.3:1).

Chromosomal distribution

The distribution of retroelements was examined on a genome-wide basis. Upon mapping the retroelement families onto the A. thaliana chromosomes, the previously noted pericentromeric clustering of TEs was immediately evident (Figure 3) [11]. The Metaviridae appeared to cluster in the pericentromeric regions more tightly than the Pseudoviridae and non-LTR retrotransposons. Distributions of these latter two groups appeared similar, as did the distribution of solo LTRs relative to full-length elements (Figure 4).

Figure 3

Physical distribution of full-length A. thaliana retroelements. The five A. thaliana chromosomes are designated as Ath1-5. Triangles indicate the location of a particular retroelement on the chromosome. Non-LTR retrotransposons are in black, Pseudoviridae in gray, and Metaviridae in white. Vertical bars on the chromosome show the precise location of the retroelement. Regions of heterochromatin are represented as follows: telomeres and NORs (on Ath2 and Ath4) by rounded chromosome ends; centromeres by hourglass shapes; heterochromatic knobs (on Ath4 and Ath5) by narrowed stretches on chromosome bars. The relatively short chromosome 5 knob is barely visible to the right of the centromere. The inset more clearly depicts heterochromatic regions that are obscured by element insertions. Chromosomes are drawn to scale.

Figure 4

Chromosomal distribution of LTRs for the Metaviridae and Pseudoviridae families in A. thaliana. Chromosomes are displayed as in Figure 3. In addition, solo LTRs are drawn as open triangles. The upper chromosome depicts the distribution of Pseudoviridae, the lower the distribution of Metaviridae. In contrast to Figure 3, shading is not used to distinguish between the families.

We assessed statistical support for the apparent clustering of elements by comparing the observed distribution of each lineage to a random uniform distribution model (Table 3). This model assumes that any location in the genome is expected to have a uniform probability of element insertion. This model was rejected by Kendall-Sherman tests of uniformity for every lineage and chromosome combination. All p-values were less than 0.05 and most were less than 0.0001.

Table 3 Comparison of genome localization by retroelement lineage

We next looked at distribution patterns between element families to determine whether they are similar. On the basis of the retroelement distribution maps (Figure 3), we hypothesized that this would not be the case for the Metaviridae because they appeared to be associated with centromeres to a greater degree than the other families. Each family's chromosomal distribution, inclusive of all subtypes (for example, FLE, RT-only and solo LTR), was tested for similarity to the distribution of the other families using a permutation test. With the exception of chromosome 3, the distribution of non-LTR retrotransposons was not significantly different from that of the Pseudoviridae. Comparisons of Metaviridae elements with Psedoviridae and/or non-LTR elements differed significantly (p < 0.05) for all combinations.

To assess whether the Metaviridae sublineages contributed equally to the observed distribution bias, we tested a model wherein the three sublineages (Athila, Tat and Metavirus) were expected to have similar distributions. This appears to be true, as significant differences were not detected on any chromosome for these sublineages. We then checked whether the FLEs, RT-only hits or solo LTRs displayed different distributions from one another within their respective families. No consistently significant trends were observed for the Pseudoviridae or the Metaviridae. Oddly, the Metaviridae solo LTR distribution displayed significant differences from the FLEs and RT-only hits for chromosome 3.

A feature of pericentromeric regions in A. thaliana is that they are heterochromatic, a state required for targeted integration by the yeast Ty5 retroelement [19]. Because of the observed pericentromeric clustering of retrotransposons in A. thaliana, we assessed a simple model that assumes that all elements transpose to heterochromatin (Table 4). There are several genomic regions that are typically considered heterochromatic in A. thaliana - centromeres, knobs (on chromosomes 4 and 5), telomeres and rDNA [2022]. We looked for differences between lineages with respect to whether retroelements were within a heterochromatic region, or, if outside, whether differences existed in distances to the nearest heterochromatic domain. All lineage combinations showed highly significant differences in heterochromatic distributions. In the Metaviridae, the Metavirus elements are less tightly associated with heterochromatin than are Tat and Athila, which did not differ significantly from each other. Element subtypes also differed in their distribution with respect to heterochromatin. The major source of differences was the distribution of solo LTRs in the Metaviridae.

Table 4 Association of retroelements with heterochromatin

Age of insertions

LTR retroelements have a built-in clock that can be used to estimate the age of given insertions. At the time an element inserts into the genome, the LTRs are typically 100% identical. As time passes, mutations occur within the LTRs at a rate approximating the host's mutation rate. LTR divergence, therefore, can be used to estimate relative ages between elements, assuming that all elements share the same probability of incurring a mutation. Although it is possible to estimate ages for non-LTR retrotransposons by generating a putative ancestral consensus sequence and calculating divergence from the consensus, this method is not directly equivalent to estimating ages by LTR comparisons. Therefore, age comparisons were performed only for the LTR retroelement families. Note that the ages depicted in Figure 5 are relative, and we do not claim that a particular element is a specific age in this study. Rather, we focus on whether elements are significantly older or younger than each other.

Figure 5

Relative ages of A. thaliana LTR retroelement lineages. (a) Box-plot showing the age distribution of Pseudoviridae full-length elements contrasted with those of the Metaviridae. The position of the median is shown as a gray bar in the box that delimits the boundaries of the lower and upper quartiles. Data points more than 1.5 times the inter-quartile range above the upper quartile or below the lower quartile are indicated by individual horizontal lines. Ages were calculated as described in Materials and methods. (b) Relative-age box-plots of Metaviridae sublineages. Permutation test p-values for the significance of the displayed age distributions are shown below each box-plot.

Statistically significant age differences were observed among the Pseudoviridae and three Metaviridae sublineages (F = 14.4, df = 3 and 368, p < 0.0001) (Table 5, Figure 5). Overall, the Pseudoviridae are younger than the Metaviridae (t = 5.72, df = 368, p < 0.0001). When the Metaviridae sublineages are considered, it is apparent that the Athila elements are responsible for much of the increased age of this family. The difference between Athila and the other two sublineages is significant, with p = 0.0003 being the highest value for sublineage comparisons. Elements within heterochromatic regions were significantly older than those found outside (F = 17.19, df = 1 and 368, p < 0.0001). There was suggestive evidence that the mean element ages varied among chromosomes (F = 2.73, df = 4 and 368, p = 0.0289). However, all pairwise comparisons between chromosomes failed to yield significant results at the 0.05 level using the Tukey-Kramer adjustment (data not shown).

Table 5 Comparison of LTR retroelement age distributions


Completed genome sequences enable comprehensive analyses of retroelement diversity and the exploration of the impact of retroelements on genome organization. Although most large-scale sequencing projects use the shotgun sequencing method, this method makes it particularly difficult to assemble repetitive sequences and to correctly position sequence repeats on the genome scaffold. Consequently, regions of repetitive DNA such as nucleolar-organizing regions (NORs), telomeres and centromeres tend to be skipped, or are sometimes represented by consensus or sampled sequences. The difficulty of cloning repetitive sequences and the drawbacks noted above result in the under- or misrepresentation of the repetitive content of most genomes. Because retroelements frequently comprise a large proportion of the repetitive DNA, 'completed' genome sequences are typically not ideal for studies of retroelement diversity and distribution on a genomic scale. In contrast to these cases, the A. thaliana genome is reliably sequenced well into heterochromatic regions and work continues to further define these domains [11, 23].

Another factor frustrating comprehensive analyses of eukaryotic mobile genetic elements is the inherent difficulty in annotating these sequences. Many mobile element insertions are structurally degenerate, rearranged through recombination or organized in complex arrays. Software tools and databases such as Reputer [24] and Repbase update [25] have been developed to identify and classify repeat sequences, and these tools have proved helpful in several genome-wide surveys of mobile elements. RECON [26] and LTR_STRUC [27] are software tools that go one step further and consider structural features of mobile elements that can assist in genome annotation. We developed an additional software tool, called RetroMap, to assist in characterizing the LTR retroelement content of genomes. RetroMap delimits LTR retroelement insertions by iterated identification of reverse transcriptases followed by a search for flanking LTRs. The software goes beyond existing platforms and carries out a number of analytic functions, including age assignment, solo LTR identification and visualization of the chromosomal locations of various groups of identified elements on a whole-genome scale.

Data generated by RetroMap are subject to a few caveats. First, because element searches use reverse transcriptase sequences as queries, elements lacking reverse transcriptase motifs (for whatever reason) will not be identified. Second, when RetroMap encounters nested elements, tandem elements, and other complex arrangements, it does not attempt to delimit the element. Rather, the user is notified that a complex arrangement was encountered and the original reverse transcriptase match and any LTR(s) found are logged as separate entities.

For the most part, RetroMap was quite effective in identifying LTR retrotransposon insertions. Our results closely agree with the findings of a parallel study conducted by Pereira [28]. For the Pseudoviridae and two of the three Metaviridae lineages (Tat and Metavirus), we identified 210 and 128 full-length elements, respectively, whereas Pereira recovered 215 and 130 insertions for these respective element groups. The two studies, however, differed significantly in the number of Athila elements identified. We found 38 insertions, whereas Pereira recovered 219. To reconcile these differences, we independently estimated Athila copy numbers by conducting iterative BLAST searches with a variety of Athila query sequences (data not shown). BLAST hits recovered with each query were then mapped onto the genome sequence. As a result of this analysis, we concluded that RetroMap missed many Athila insertions, either because they are highly degenerate or part of complex arrangements. In contrast to Pereira's approach, RetroMap requires that a reverse transcriptase reside between LTRs, and in many cases reverse transcriptases were absent or not detectable in Athila insertions. This can be resolved in future implementations of RetroMap that enable multiple query sequences to be tested. The Athila elements are large, and our underestimate of the number of Athila elements resulted in a corresponding underestimate of the total amount of retrotransposon DNA in the A. thaliana genome. We calculated 3.36% for this value, whereas Pereira calculated 5.60%. Pereira's estimate is likely to be the more accurate of the two.

With the exception of the Athila elements, the observed frequency of insertions in complex arrangements was rare. For example, the Pseudoviridae had only eight nested and five unassignable elements. The small observed number of complex element arrangements in A. thaliana contrasts sharply with observations in grass genomes, where retroelements are usually found in complex nested arrays [29, 30]. This may reflect a difference between species in factors contributing to chromosomal distribution of retroelements, or it may simply be a consequence of the difference in abundance of retroelements between A. thaliana (5.60% of the genome) and grasses (up to 80% of some genomes) [1, 28].

Genomic distribution of A. thalianaretroelements

Our data on the genomic distribution of retroelements can be considered in the light of theoretical work predicting the distribution of TE populations within genomes. These studies largely focus on the effects of selection and recombination on element insertions [31, 32]. Particularly relevant is the recent study by Wright et al. [33], which considers the effects of recombination on the genomic distribution of major groups of mobile elements in A. thaliana (DNA transposons and retroelements). Our analysis extends this work by considering the genomic distribution of specific retroelement lineages. We investigate a model wherein selection and recombination affect element lineages uniformly, and hypothesize that observed deviations in the genomic distribution of specific element lineages reflect unique aspects of their evolutionary history or survival strategies such as targeted integration.

Ectopic exchange model

The ectopic exchange model assumes that inter-element recombination restricts growth of element populations [31]. Elements should be most numerous in regions of reduced recombination such as the centromeres, because of less frequent loss by homologous recombination. A corollary is that element abundance at a genomic location should inversely reflect the recombination rate for that region in the genome. Previous work suggests that this model is not the primary determinant of element abundance in A. thaliana. Wright et al. [33] examined recombination rate relative to element abundance in detail and found that the abundance of most A. thaliana TE families actually had a small but positive correlation with recombination rate, as was also observed in C. elegans [34]. Devos et al. [35] found ectopic recombination to be very infrequent relative to intra-element recombination, suggesting this process is unlikely to have a significant role in explaining the observed A. thaliana retrotransposable element distribution.

The ectopic exchange hypothesis makes two unique predictions for retrotransposons: solo LTRs (a product of recombination) should be observed in higher proportions relative to full-length elements outside of heterochromatin; and heterochromatic elements will show a shift toward greater average age than elements elsewhere in the genome. Our consideration of age assumes that the chance of loss by recombination remains steady or increases with element age. However, old elements will have higher sequence divergence, thereby reducing the likelihood that they will recombine. In considering age, we also assume that all elements evolve at the same rates. This is unlikely to be the case, as local, chromosomal and compartmental locations are increasingly found to have different mutation rates [36, 37].

With respect to the distribution of solo LTRs, our data show exactly the opposite bias predicted by the ectopic exchange model: the ratio of Metaviridae solo LTRs to FLEs in heterochromatin was nearly twice that found outside heterochromatin. The frequency of solo LTRs at the centromeres suggests that homologous recombination, at least over short distances (less than 20 kilobases (kb)), occurs frequently in pericentromeric regions.

While we did observe the predicted shift toward older elements within heterochromatin, the data are not consistent with low rates of recombination as the determinant of retrotransposon accumulation at the centromeres. Within the Metaviridae, for example, the Metaviruses and Tat elements differ significantly in their association with heterochromatin. The ectopic exchange model would predict that the Tat elements should be older; however, these two lineages do not differ significantly in age. Although it is possible that recombinational forces could act differentially on different element sublineages, we view this as unlikely. Rather, forces other than ectopic recombination, such as targeted integration (see below), are responsible for the differential genomic distribution of certain element lineages. This is not to say that ectopic exchange has no role; however, it is unlikely to be the sole or prevailing influence.

Deleterious insertion model

The deleterious insertion model hypothesizes that element insertions are generally harmful to the host, and thus elements accumulate in regions of low gene density, where insertions are least likely to have negative effects on the host. According to this model, abundance of all classes of mobile elements should inversely reflect gene density within the genome. This is supported by the observation that elements are over-represented in gene-poor pericentromeric heterochromatin and are rare over much of the chromosome arms. However, we did not observe an increase in element abundance at other gene-poor heterochromatic regions (such as the telomeres and NORs), which would be predicted by the deleterious insertion hypothesis. This model would also predict that element insertions into gene-rich regions that are tolerated by the host should act as founders or safe havens for future element insertions. This could lead to an ever-expanding area of tightly clustered and frequently nested elements in euchromatin, assuming the overall random insertion rate is greater than the rate of sequence loss through recombination. Nested clusters of elements have been reported in cereals such as maize and barley [29, 30]. In A. thaliana, although numerous potential 'seed' insertion sites are observed along the chromosome arms, we did not detect dense clusters of nested elements at these locations.

In contrast to the deleterious insertion model, it is important to recognize that some element insertions may provide a selective advantage. Studies in C. elegans and rice indicate that many retrotranposons are associated with genes (63% and 20% in these species respectively) [38, 39]. In D. melanogaster, some retrotransposon-gene associations are preserved in diverse natural populations, consistent with the hypothesis that they confer a positive selective advantage [40]. Furthermore, recent analyses in S. pombe suggest that the Tf1 retrotransposons may regulate expression of adjacent genes [41]. We cannot rule out a role for positive selection in the distribution of some A. thaliana mobile elements, but identifying such a role would require a more refined analysis of element distribution and gene associations.

Impact of targeted integration

The observation that many LTR retroelements have non-uniform genomic distributions suggested that targeted integration may be a driver of retroelement distribution patterns [42]. Neither the deleterious insertion nor ectopic recombination models address the situation where some or all elements have evolved the ability to bias their distributions through targeted integration. The LTR retroelements of S. cerevisiae insert preferentially into heterochromatin or sites occupied by RNA polymerase III, and in the evolutionarily distant S. pombe genome, retroelements are located preferentially upstream of genes transcribed by RNA polymerase II [69]. Retroviruses also insert preferentially into transcribed regions, with some retroviruses favoring insertions into promoter regions [4, 43].

Targeted integration could contribute significantly to the chromosomal distribution of A. thaliana retroelements. As in other systems, targeting may occur because elements recognize a specific chromatin state and actively insert into regions with that type of chromatin. A chromatin-targeting model has the following predictions. First, very few elements will be found outside targeted chromatin domains. For example, all heterochromatic regions such as NORs, knobs and telomeres would be occupied by the same lineage of elements if these regions share a chromatin feature recognized by that lineage. Second, different retroelement lineages may be associated with different regions of the genome if they employ different targeting strategies.

The targeting hypothesis is well supported for the Metaviridae, which on a genome-wide basis differ significantly in their chromosomal distribution from the Pseudoviridae and non-LTR retrotransposons. This is particularly true for the Athila and Tat lineages, both of which are tightly associated with pericentromeric regions. Athila and Tat elements are not found in heterochromatin regions around the telomeres, however, suggesting that telomeric and centromeric heterochromatin differ. Targeted integration to pericentromeric heterochromatin may be a general feature of the Metaviridae. Members of the Metaviridae are abundant in pericentromeric heterochromatin in many grass species [44]. Langdon et al. [45] suggested that an evolutionary ancient member of the Metaviridae in cereals targets to centromeric domains. Portions of a maize homolog of this element were found to co-precipitate with the centromere-specific histone CENH3, indicating an association of this element with a particular type of chromatin [46].

The Pseudoviridae and non-LTR retrotransposons differ in their genomic organization from the Metaviridae and are more loosely associated with pericentromeric regions. It may be that these element lineages do not target their integration, or they may recognize other chromosomal features, although we did not observe any association with other genome features or gene classes such as tRNA genes (data not shown). De novo integration events have been mapped on a chromosomal level for two tobacco Pseudoviridae elements in heterologous hosts - Tto1 in A. thaliana and Tnt1 in Medicago trunculata. In both cases these elements integrated throughout the genome, displaying some preference for genic regions [47, 48]. Whether this observed distribution pattern reflects random integration or recognition of some other subtle chromosomal feature remains to be determined. Because we predict that the Metaviridae recognize pericentromeric heterochromatin, an important dataset for analysis will be maps of the various DNA methylation and histone-modification patterns for the full genome. In-depth characterization of the distribution of retroelements relative to chromatin modifications may reveal additional evidence for targeting and help to understand the impact of targeting on genome organization.


Our analysis of the genomic distribution of the A. thaliana LTR retroelements revealed that the distribution of the Pseudoviridae and the Metaviridae is non-uniform and that they tend to cluster at the centromeres. The pericentromeric association of three Metaviridae sublineages (Metavirus, Tat and Athila) was significantly more pronounced than for the Pseudoviridae. Several factors are likely to contribute to the centromeric association of these elements, including target-site bias, selection against euchromatin integration and pericentromeric accumulation of elements due to suppression of recombination. For the Tat and Athila lineages, however, target-site specificity appears to be the primary factor determining chromosomal distribution. We predict that, like retroelements in yeast, the Tat and Athila elements target integration to pericentromeric regions by recognizing a specific feature of pericentromeric heterochromatin.

Materials and methods

RetroMap and the A. thalianaretroelement dataset

Reverse transcriptase amino-acid sequences (as defined by [49], see also Table 1), were used to query a database of A. thaliana chromosomes (TIGR version 7 January 2002) with the tblastn program (E = 1e-10, XML output, filtering disabled) [50]. The resulting search report was imported into RetroMap. RetroMap (to be described in detail elsewhere) provides a graphical user interface (GUI) to interactively characterize LTR retrotransposons in targeted genomes or large genomic contigs (Figure 1a). RetroMap generates a nonredundant set of database hits from BLAST results generated by a given query sequence set. Hits are merged if they directly overlap or if they align to different portions of the same query sequence. In this study, the nonredundant sequences were used to re-query the chromosome database twice more using tblastx (E = 1e-10, XML output, filtering disabled) to identify increasingly divergent or degenerate elements. Unique hits identified in the final round of screening were taken to represent the entire complement of retroelements in A. thaliana.

RetroMap assigns putative LTRs where possible for each reverse transcriptase by comparing 10 kb of DNA from each flank. This is accomplished using Blast2Sequences to identify flanking repeats [51] (Figure 1b). Direct repeats found closest to the reverse transcriptase, larger than 50 bp and less than 5 kb, are considered to be LTRs. Hits with putative LTRs were considered to be full-length elements (FLE) or complete elements. Twenty-six reverse transcriptase hits were excluded from the FLEs owing to difficulty in automatic LTR assignment (13 each from the Pseudoviridae and Metaviridae). Among these were nested elements and tandem elements sharing a LTR. Reverse transcriptases were assigned to a retroelement lineage (Metaviridae, Pseudoviridae or non-LTR retrotransposon) on the basis of their similarity to the diagnostic reverse transcriptase query sequences. Full-length Metaviridae elements were further subdivided into the classic (Metavirus), Tat and Athila groups on the basis of the highest-scoring match in a BLAST database containing the Metaviridae reverse transcriptase sequences described in [18]. Putative complete elements with a predicted reverse transcriptase failing to significantly match any sequence in this database were removed from further consideration as false positives (two cases).

Solo LTRs and solo LTR fragments were identified with blastn (E < 1e-5) using all predicted 5' LTRs of known A. thaliana elements and the FLEs. RetroMap assigns any putative LTR sequence that fails to match or overlap with a predicted FLE LTR as a solo LTR.

Relative age calculation for full-length elements

LTRs are identical at the time of retroelement integration, and so relative element ages were estimated from the percentage of identical residues shared between 5' and 3' LTRs for FLEs. The age formula used was T = d/2k (time (T) = genetic distance (d)/ [2 × substitution rate (k)]), where genetic distance is 1 - (percent identity/100) and the substitution rate is 1.5 × 10-8 [52].

Assignment of heterochromatin boundaries

Chromosome coordinates relative to the left (north) end were used to calculate distances between retroelements and heterochromatic domains. Heterochromatin boundaries were derived from [2022] and include the telomeres, heterochromatic knobs, NORs and centromeres. Chromosome end-coordinates were considered as the telomere boundaries. The A. thaliana NORs are located at the left (north) ends of chromosomes 2 and 4, and as these regions were only sample sequenced, their boundaries were assigned as the left ends of chromosomes 2 and 4. Heterochromatic knobs and pericentromeric regions were assigned as the outermost physical markers delimiting these regions, as determined by the studies listed above.

Statistical tests

A RetroMap-generated datafile was used as the data source for statistical testing. The data file contains chromosomal element coordinates, LTR identity, age and lineage information for all A. thaliana retroelement families by element category: reverse transcriptase only (R), full-length (F), and solo LTR (S).

For each element type and each chromosome, a Kendall-Sherman test [5355] was conducted to determine if the element positions were randomly distributed across chromosomes according to a uniform distribution. A permutation test [55] was used to assess the statistical significance of observed differences in the chromosomal position distributions for each chromosome across various element categories. The multi-response permutation procedures (MRPP) test is briefly described as follows. The average distance between a pair of elements within a category of interest is determined. A weighted sum of these averages over all categories of interest is computed, with each category weighted in proportion to the number of elements in the category. This weighted sum is the observed value of the test statistic. Next, the test statistic is re-computed for each of 10,000 random permutations of the category labels. For each permutation, the observed chromosomal positions of the elements are held constant while the category labels are randomly shuffled. The proportion of the 10,000 permutation-replicated test statistics that are less than or equal to the original observed test statistic serves as an approximate p-value for a test whose null hypothesis is that all element categories of interest have the same chromosomal position distribution. This permutation approach is useful for the chromosomal position data because first, no distributional assumptions are required, second, differences in chromosomal position distributions other than simple location shifts are detectable, and third, the method is not as sensitive to outliers as common parametric approaches.

For FLEs, linear model analyses were used to assess the effects of the factors 'chromosome', 'lineage/sublineage', and 'location' relative to heterochromatin on the response variable 'element age'. F-tests were used to check for interaction between these three factors and to assess the statistical significance of observed differences among the five chromosomes, among the four lineage/sublineage categories (Pseudoviridae and the three Metaviridae sublineages:Athila, Tat or Metavirus), and between elements inside and outside heterochromatin. The square root of age was used as the response variable in the age analysis so that the variance of the response would be roughly constant across categories defined by combinations of chromosome, lineage/sublineage, and location, as required for standard linear model analyses. Outlying observations were present, but the results of the analysis remained essentially the same with or without the outliers. Thus the reported results are based on the full dataset.

Additional data files

The following additional data are available with the online version of this article: a Microsoft Excel spreadsheet of data generated by RetroMap for each retrotransposon insertion identified; the data in this file was used for all statistical analyses (Additional data file 1). The Java application used to generate the LTR and retrotransposon coordinates and to estimate retrotransposon ages (Additional data file 2). To run RetroMap, version 1.3 or higher of the Java Runtime Environment (JRE must be present. To enable searches for LTRs, NCBI's BLAST 2 Sequences must be locally installed.


  1. 1.

    SanMiguel P, Gaut BS, Tikhonov A, Nakajima Y, Bennetzen JL: The paleontology of intergene retrotransposons of maize. Nat Genet. 1998, 20: 43-45. 10.1038/1695.

    PubMed  CAS  Article  Google Scholar 

  2. 2.

    Kim JM, Vanguri S, Boeke JD, Gabriel A, Voytas DF: Transposable elements and genome organization: a comprehensive survey of retrotransposons revealed by the complete Saccharomyces cerevisiae genome sequence. Genome Res. 1998, 8: 464-478.

    PubMed  CAS  Google Scholar 

  3. 3.

    Kumar A, Bennetzen JL: Plant retrotransposons. Annu Rev Genet. 1999, 33: 479-532. 10.1146/annurev.genet.33.1.479.

    PubMed  CAS  Article  Google Scholar 

  4. 4.

    Bushman FD: Targeting survival: integration site selection by retroviruses and LTR-retrotransposons. Cell. 2003, 115: 135-138. 10.1016/S0092-8674(03)00760-8.

    PubMed  CAS  Article  Google Scholar 

  5. 5.

    Sandmeyer S: Integration by design. Proc Natl Acad Sci USA. 2003, 100: 5586-5588. 10.1073/pnas.1031802100.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  6. 6.

    Chalker DL, Sandmeyer SB: Ty3 integrates within the region of RNA polymerase III transcription initiation. Genes Dev. 1992, 6: 117-128.

    PubMed  CAS  Article  Google Scholar 

  7. 7.

    Devine SE, Boeke JD: Integration of the yeast retrotransposon Ty1 is targeted to regions upstream of genes transcribed by RNA polymerase III. Genes Dev. 1996, 10: 620-633.

    PubMed  CAS  Article  Google Scholar 

  8. 8.

    Bowen NJ, Jordan IK, Epstein JA, Wood V, Levin HL: Retrotransposons and their recognition of pol II promoters: a comprehensive survey of the transposable elements from the complete genome sequence off Schizosaccharomyces pombe. Genome Res. 2003, 13: 1984-1997. 10.1101/gr.1191603.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  9. 9.

    Zou S, Ke N, Kim JM, Voytas DF: The Saccharomyces retrotransposon Ty5 integrates preferentially into regions of silent chromatin at the telomeres and mating loci. Genes Dev. 1996, 10: 634-645.

    PubMed  CAS  Article  Google Scholar 

  10. 10.

    Boeke JD, Devine SE: Yeast retrotransposons: finding a nice quiet neighborhood. Cell. 1998, 93: 1087-1089. 10.1016/S0092-8674(00)81450-6.

    PubMed  CAS  Article  Google Scholar 

  11. 11.

    Initiative TAG: Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature. 2000, 408: 796-815. 10.1038/35048692.

    Article  Google Scholar 

  12. 12.

    Adams MD, Celniker SE, Holt RA, Evans CA, Gocayne JD, Amanatides PG, Scherer SE, Li PW, Hoskins RA, Galle RF, et al: The genome sequence of Drosophila melanogaster. Science. 2000, 287: 2185-2195. 10.1126/science.287.5461.2185.

    PubMed  Article  Google Scholar 

  13. 13.

    Terol J, Castillo MC, Bargues M, Perez-Alonso M, de Frutos R: Structural and evolutionary analysis of the copia-like elements in the Arabidopsis thaliana genome. Mol Biol Evol. 2001, 18: 882-892.

    PubMed  CAS  Article  Google Scholar 

  14. 14.

    Peterson-Burch BD, Voytas DF: Genes of the Pseudoviridae (Ty1/copia retrotransposons). Mol Biol Evol. 2002, 19: 1832-1845.

    PubMed  CAS  Article  Google Scholar 

  15. 15.

    Kapitonov VV, Jurka J: Molecular paleontology of transposable elements from Arabidopsis thaliana. Genetica. 1999, 107: 27-37. 10.1023/A:1004030922447.

    PubMed  CAS  Article  Google Scholar 

  16. 16.

    Boeke JD, Eickbush T, Sandmeyer SB, Voytas DF: Pseudoviridae. In: Virus Taxonomy: Eighth Report of the International Committee on Taxonomy of Viruses. Edited by: Fauquet CM. 2004, New York: Academic Press,

    Google Scholar 

  17. 17.

    Boeke JD, Eickbush T, Sandmeyer SB, Voytas DF: Metaviridae. In: Virus Taxonomy: Eight Report of the International Committee on Taxonomy of Viruses. Edited by: Fauquet CM. 2004, New York: Academic Press,

    Google Scholar 

  18. 18.

    Wright DA, Voytas DF: Athila 4 of Arabidopsis and Calypso of soybean define a lineage of endogenous plant retroviruses. Genome Res. 2002, 12: 122-131. 10.1101/gr.196001.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  19. 19.

    Zou S, Voytas DF: Silent chromatin determines target preference of the Saccharomyces retrotransposon Ty5. Proc Natl Acad Sci USA. 1997, 94: 7412-7416. 10.1073/pnas.94.14.7412.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  20. 20.

    Fransz PF, Armstrong S, de Jong JH, Parnell LD, van Drunen C, Dean C, Zabel P, Bisseling T, Jones GH: Integrated cytogenetic map of chromosome arm 4S of A. thaliana : structural organization of heterochromatic knob and centromere region. Cell. 2000, 100: 367-376. 10.1016/S0092-8674(00)80672-8.

    PubMed  CAS  Article  Google Scholar 

  21. 21.

    Copenhaver GP, Nickel K, Kuromori T, Benito MI, Kaul S, Lin X, Bevan M, Murphy G, Harris B, Parnell LD, et al: Genetic definition and sequence analysis of Arabidopsis centromeres. Science. 1999, 286: 2468-2474. 10.1126/science.286.5449.2468.

    PubMed  CAS  Article  Google Scholar 

  22. 22.

    The Cold Spring Harbor Laboratory, Washington University Genome Sequencing Center, and PE Biosystems Arabidopsis Sequencing Consortium: The complete sequence of a heterochromatic island from a higher eukaryote. Cell. 2000, 100: 377-386. 10.1016/S0092-8674(00)80673-X.

    Article  Google Scholar 

  23. 23.

    Mardis E, McPherson J, Martienssen R, Wilson RK, McCombie WR: What is finished, and why does it matter. Genome Res. 2002, 12: 669-671. 10.1101/gr.032102.

    PubMed  CAS  Article  Google Scholar 

  24. 24.

    Kurtz S, Choudhuri JV, Ohlebusch E, Schleiermacher C, Stoye J, Giegerich R: REPuter: the manifold applications of repeat analysis on a genomic scale. Nucleic Acids Res. 2001, 29: 4633-4642. 10.1093/nar/29.22.4633.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  25. 25.

    Jurka J: Repbase update: a database and an electronic journal of repetitive elements. Trends Genet. 2000, 16: 418-420. 10.1016/S0168-9525(00)02093-X.

    PubMed  CAS  Article  Google Scholar 

  26. 26.

    Bao Z, Eddy SR: Automated de novo identification of repeat sequence families in sequenced genomes. Genome Res. 2002, 12: 1269-1276. 10.1101/gr.88502.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  27. 27.

    McCarthy EM, McDonald JF: LTR_STRUC: a novel search and identification program for LTR retrotransposons. Bioinformatics. 2003, 19: 362-367. 10.1093/bioinformatics/btf878.

    PubMed  CAS  Article  Google Scholar 

  28. 28.

    Pereira V: Insertion bias and purifying selection of retrotransposons in the Arabidopsis thaliana genome. Genome Biol. 2004, 5: R79-10.1186/gb-2004-5-10-r79.

    PubMed  PubMed Central  Article  Google Scholar 

  29. 29.

    SanMiguel P, Tikhonov A, Jin YK, Motchoulskaia N, Zakharov D, Melake-Berhan A, Springer PS, Edwards KJ, Lee M, Avramova Z, et al: Nested retrotransposons in the intergenic regions of the maize genome. Science. 1996, 274: 765-768. 10.1126/science.274.5288.765.

    PubMed  CAS  Article  Google Scholar 

  30. 30.

    Wei F, Wing RA, Wise RP: Genome dynamics and evolution of the Mla (powdery mildew) resistance locus in barley. Plant Cell. 2002, 14: 1903-1917. 10.1105/tpc.002238.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  31. 31.

    Langley CH, Montgomery E, Hudson R, Kaplan N, Charlesworth B: On the role of unequal exchange in the containment of transposable element copy number. Genet Res. 1988, 52: 223-235.

    PubMed  CAS  Article  Google Scholar 

  32. 32.

    Charlesworth B, Sniegowski P, Stephan W: The evolutionary dynamics of repetitive DNA in eukaryotes. Nature. 1994, 371: 215-220. 10.1038/371215a0.

    PubMed  CAS  Article  Google Scholar 

  33. 33.

    Wright SI, Agrawal N, Bureau TE: Effects of recombination rate and gene density on transposable element distributions in Arabidopsis thaliana. Genome Res. 2003, 13: 1897-1903.

    PubMed  CAS  PubMed Central  Google Scholar 

  34. 34.

    Duret L, Marais G, Biemont C: Transposons but not retrotransposons are located preferentially in regions of high recombination rate in Caenorhabditis elegans. Genetics. 2000, 156: 1661-1669.

    PubMed  CAS  PubMed Central  Google Scholar 

  35. 35.

    Devos KM, Brown JK, Bennetzen JL: Genome size reduction through illegitimate recombination counteracts genome expansion in Arabidopsis. Genome Res. 2002, 12: 1075-1079. 10.1101/gr.132102.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  36. 36.

    Webster MT, Smith NG, Lercher MJ, Ellegren H: Gene expression, synteny, and local similarity in human noncoding mutation rates. Mol Biol Evol. 2004, 21: 1820-1830. 10.1093/molbev/msh181.

    PubMed  CAS  Article  Google Scholar 

  37. 37.

    Filatov DA, Charlesworth D: Substitution rates in the X- and Y-linked genes of the plants, Silene latifolia and S. dioica. Mol Biol Evol. 2002, 19: 898-907.

    PubMed  CAS  Article  Google Scholar 

  38. 38.

    Ganko EW, Bhattacharjee V, Schliekelman P, McDonald JF: Evidence for the contribution of LTR retrotransposons to C. elegans gene evolution. Mol Biol Evol. 2003, 20: 1925-1931. 10.1093/molbev/msg200.

    PubMed  CAS  Article  Google Scholar 

  39. 39.

    Gao L, McCarthy E, Ganko E, McDonald J: Evolutionary history of Oryza sativa LTR retrotransposons: a preliminary survey of the rice genome sequences. BMC Genomics. 2004, 5: 18-10.1186/1471-2164-5-18.

    PubMed  PubMed Central  Article  Google Scholar 

  40. 40.

    Franchini LF, Ganko EW, McDonald JF: Retrotransposon-gene associations are wide-spread among D.melanogaster populations. Mol Biol Evol. 2004, 21: 1323-1331. 10.1093/molbev/msh116.

    PubMed  CAS  Article  Google Scholar 

  41. 41.

    Schramke V, Allshire R: Hairpin RNAs and retrotransposon LTRs effect RNAi and chromatin-based gene silencing. Science. 2003, 301: 1069-1074. 10.1126/science.1086870.

    PubMed  CAS  Article  Google Scholar 

  42. 42.

    Dimitri P, Junakovic N: Revising the selfish DNA hypothesis: new evidence on accumulation of transposable elements in heterochromatin. Trends Genet. 1999, 15: 123-124. 10.1016/S0168-9525(99)01711-4.

    PubMed  CAS  Article  Google Scholar 

  43. 43.

    Wu X, Li Y, Crise B, Burgess SM: Transcription start regions in the human genome are favored targets for MLV integration. Science. 2003, 300: 1749-1751. 10.1126/science.1083413.

    PubMed  CAS  Article  Google Scholar 

  44. 44.

    Miller JT, Dong F, Jackson SA, Song J, Jiang J: Retrotransposon-related DNA sequences in the centromeres of grass chromosomes. Genetics. 1998, 150: 1615-1623.

    PubMed  CAS  PubMed Central  Google Scholar 

  45. 45.

    Langdon T, Seago C, Mende M, Leggett M, Thomas H, Forster JW, Jones RN, Jenkins G: Retrotransposon evolution in diverse plant genomes. Genetics. 2000, 156: 313-325.

    PubMed  CAS  PubMed Central  Google Scholar 

  46. 46.

    Zhong CX, Marshall JB, Topp C, Mroczek R, Kato A, Nagaki K, Birchler JA, Jiang J, Dawe RK: Centromeric retroelements and satellites interact with maize kinetochore protein CENH3. Plant Cell. 2002, 14: 2825-2836. 10.1105/tpc.006106.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  47. 47.

    d'Erfurth I, Cosson V, Eschstruth A, Lucas H, Kondorosi A, Ratet P: Efficient transposition of the Tnt1 tobacco retrotransposon in the model legume Medicago truncatula. Plant J. 2003, 34: 95-106.

    PubMed  Article  Google Scholar 

  48. 48.

    Okamoto H, Hirochika H: Efficient insertion mutagenesis of Arabidopsis by tissue culture-induced activation of the tobacco retrotransposon Tto1. Plant J. 2000, 23: 291-304. 10.1046/j.1365-313x.2000.00779.x.

    PubMed  CAS  Article  Google Scholar 

  49. 49.

    Xiong Y, Eickbush TH: Origin and evolution of retroelements based upon their reverse transcriptase sequences. EMBO J. 1990, 9: 3353-3362.

    PubMed  CAS  PubMed Central  Google Scholar 

  50. 50.

    Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ: Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 1997, 25: 3389-3402. 10.1093/nar/25.17.3389.

    PubMed  CAS  PubMed Central  Article  Google Scholar 

  51. 51.

    Tatusova TA, Madden TL: BLAST 2 Sequences, a new tool for comparing protein and nucleotide sequences. FEMS Microbiol Lett. 1999, 174: 247-250. 10.1016/S0378-1097(99)00149-4.

    PubMed  CAS  Article  Google Scholar 

  52. 52.

    Koch MA, Haubold B, Mitchell-Olds T: Comparative evolutionary analysis of chalcone synthase and alcohol dehydrogenase loci in Arabidopsis, Arabis and related genera (Brassicaceae). Mol Biol Evol. 2000, 17: 1483-1498.

    PubMed  CAS  Article  Google Scholar 

  53. 53.

    Greenwood M: The statistical study of infectious diseases. J R Stat Soc Ser A. 1946, 109: 85-110.

    PubMed  CAS  Article  Google Scholar 

  54. 54.

    Sherman B: A random variable related to spacings of sample values. Ann Math Stat. 1950, 21: 339-361.

    Article  Google Scholar 

  55. 55.

    Mielke P, Berry K: Permuation Methods: A Distance Function Approach. 2001, New York: Springer Verlag

    Book  Google Scholar 

  56. 56.

    Jeanmougin F, Thompson JD, Gouy M, Higgins DG, Gibson TJ: Multiple sequence alignment with Clustal X. Trends Biochem Sci. 1998, 23: 403-405. 10.1016/S0968-0004(98)01285-7.

    PubMed  CAS  Article  Google Scholar 

  57. 57.

    Kumar S, Tamura K, Jakobsen IB, Nei M: MEGA2: molecular evolutionary genetics analysis software. Bioinformatics. 2001, 17: 1244-1245. 10.1093/bioinformatics/17.12.1244.

    PubMed  CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Daniel F Voytas.

Electronic supplementary material


Additional data file 1: A Microsoft Excel spreadsheet of data generated by RetroMap for each retrotransposon insertion identified; the data in this file was used for all statistical analyses (XLS 2 MB)


Additional data file 2: The Java application used to generate the LTR and retrotransposon coordinates and to estimate retrotransposon ages (JAR 624 KB)

Authors’ original submitted files for images

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Peterson-Burch, B.D., Nettleton, D. & Voytas, D.F. Genomic neighborhoods for Arabidopsisretrotransposons: a role for targeted integration in the distribution of the Metaviridae. Genome Biol 5, R78 (2004).

Download citation


  • Long Terminal Repeat
  • Additional Data File
  • Pericentromeric Region
  • Element Insertion
  • Long Terminal Repeat Retrotransposons