Background

The long terminal repeats (LTR) retrotransposon life cycle involves a cytosolic reverse-transcription step within a multiproteic core called virus-like particle (VLP), formed by the polymerization of the Group-specific antigen (GAG) proteins, normally encoded in the element itself; for a recent review, see [1]. This GAG protein classically harbors three domains, from external to internal:

  1. 1)

    the matrix domain (MA), for membrane targeting and capsid assembly;

  2. 2)

    the capsid hydrophobic region (CA) and the most conserved part of GAG, in charge of polymerization, and the

  3. 3)

    nucleocapsid (NC), targeting the specific mRNA through the PSI region [1].

In addition, a CCHC zinc-finger motif is located at the C-terminus of the protein, single or twice repeated (or even thrice), and is in charge of the protein-nucleic acid interactions [1]. This protein is theoretically specific of its own RNA, and is an essential and mandatory component of the retrotransposition of LTR retrotransposons. A second open reading frame (ORF), pol, encodes the reverse transcriptase-RNaseH (RT-RNaseH), which drives the synthesis of a double-stranded cDNA from two RNA matrices and the integrase (INT) which allows the insertion of the new cDNA copy. However, in some cases, some non-autonomous elements have been shown capable of hijacking the GAG from other elements [2].

In cultivated Asian rice (Oryza sativa L.), LTR retrotransposons compose at least 20% of the genome (MSUv7.0 reference genome [3], http://rice.plantbiology.msu.edu/index.shtml). The Copia Tos17 element (for Transposon of Oryza sativa 17) was the first identified as active [4] and able to transpose in this genome. Moreover, Tos17 seems to be the most transpositionally competent one in regenerated plants [5].

Two almost identical genomic copies of Tos17 reside in the reference genome (on chromosomes 7 and 10; Figure 1). Only the chromosome-7 copy is transpositionally active (during in vitro culture at least), whereas the other, located on chromosome 10, is inactive, heavily methylated and contains several stop codons and indels in its predicted coding region [6]. This last copy can, however, be reactivated (transcriptionally) in methylation-defective mutants [6]. In the whole Oryza genus, the copy number as well as the location of active copies (if there are any) may differ [7].

Figure 1
figure 1

Dotter alignment of the two Tos17 genomic copies. Horizontally, the copy from chromosome 7 (active copy); vertically, the copy from chromosome 10 (inactive copy). The difference in transcriptional/transpositional activity is due to small mismatches, forming stop-codons and frameshifting, and to a larger deletion in the inactive copy (blue circle). The long terminal repeats (LTR) repeated sequences are circled in red.

Full size image

The Tos17 activation during in vitro culture was widely used in mutagenesis assays, which allowed reverse genetics analyses through the generation of insertional mutants without transformation [8–10]. In the present study, a detailed functional analysis of Tos17 was performed, showing that both genomic Tos17 copies lack a gag ORF, making Tos17 a non-autonomous element requiring an active one in order to ensure its transposition.

Results and discussion

The two Tos17 genomic copies were extracted from their respective location in the rice MSUv7.0 genome, and manually annotated using a series of basic local alignment search tool (BLAST), ProSite and Protein families (Pfam) analyses. A predicted long ORF (from position 659 to 3835, Figure 2A; annotated as the gag-pol ORF [4]) of 1,058 residues can be detected on the active copy (chromosome 7), whereas no apparent ORFs (that is, more than 100 residues starting with Met) exist on the inactive copy. On this long ORF, INT (gag_pre-integrase and rve) and RT (RVT_2) Pfam-A motifs can be easily identified (see Table 1), which suggests that this ORF is the polyprotein (POL) one. However, none of the truly GAG-related motifs, such as CCHC zinc-finger (18 residues) or the UBN2 group (100 to 150 residues), could be identified, and the first confidently identified motif related to the INT (and thus to the pol ORF) in the Pfam database starts at residue 79 (Figure 2A; gag_pre-integrase motif) of this ORF (base 757 of the internal sequence).

Table 1 Tos17 Open reading frame (ORF)2 Pfam motifs
Full size table
Figure 2
figure 2

Annotation of Tos17 and RIRE1 . Long terminal repeats (LTR) are symbolized in dark blue and the open reading frame (ORF) in light blue. (A) Tos17 annotation. The gag_pre-integrase (green), rve (red) and RVT_2 (blue) motif positions are reported on the ORF. (B) RIRE1 annotation. The UBN2_2 (green), Zf-CCHC (red), gag_pre-integrase (blue), rve (yellow) and RVT_2 (purple) motif positions are reported on the ORF.

Full size image

The Pfam analysis was performed on the largest Tos17 ORF.

This ORF was then compared to ORFs from those of the active Copia elements, RIRE1 from Oryza. australiensis[11, 12] [BAA22288; EMBL/GB] (Figure 2B), and Houba from O. sativa (known to be one of the most recently retrotransposed Copia in rice; [13]). As shown on Figure 3, the ORFs aligned on the whole POL part the elements that are compared two by two; the Tos17 ORF, however, lacked the GAG region, while the ORFs from RIRE1 and Houba are also aligned on the GAG part. No GAG-related region can be detected on the whole Tos17 genomic sequences in BLASTx against nr and protein databases (data not shown). Various tBLASTn (protein query versus nucleic database) analyses against the rice EST databases from NCBI were performed, and no ESTs resembling a larger ORF than the ones known were detected. Finally, no other Tos17 gag-like sequence can be amplified in PCR on the NipponBare genomic DNA (data not shown).

Figure 3
figure 3

Structural comparisons. Comparison between the protein structures of (A) RIRE1 and Tos17, (B) Houba and Tos17, and (C) RIRE1 and Houba. In A and B, no GAG region is found in Tos17, homologies being limited to the polyprotein (POL) part.

Full size image

RT-sequence phylogenetic analysis showed that only RN304 and Lullaby are closely related to Tos17[14]. Interestingly, RN304, the closest element to Tos17, is itself also a non-autonomous element also lacking the gag sequence, similar to Tos17, but no information about its transpositional activity is available. The closest complete element to Tos17 (that is, one that harbors a complete gag-pol ORF) is the Lullaby element, recently shown as transitionally active in only some of the regenerated lines in which its expression was detected [14]. Lullaby is a 5′142-long element, and has two copies in the Nipponbare genome, on chromosomes 6 and 9, with only the chromosome-6 copy active [14]. The DNA similarity between Tos17 and Lullaby is 57% at the DNA level (whole element sequence), and 64% at the protein level (gag-pol region sequence). At the DNA level, the similarity is limited to the internal sequence, whereas at the protein level the two POL sequences aligned well. Moreover, the primer binding site (PBS) region, located immediately after the 5′ LTR, and involved in RNA-GAG recognition [1], is almost identical between the two elements (5′-TGGTATCAGAGC(a/t)A(t/-)GGT-3′), starting at positions 126 and 139 for Lullaby and Tos17 respectively. However, no common INT signal (at the 3′-end of the 3′ LTR [1]) is shared between Lullaby and Tos17, highlighting the use of Lullaby GAG by Tos17 only.

Tos17, the most active LTR retrotransposon in cultivated rice, and the most commonly used element as an insertional mutation tool [8], is thus a non-autonomous element, because no gag sequence exists in the Oryza sativa genome, even if Tos17 is able to retrotranspose in this species. The simplest explanation is that Tos17 is coupled with an active LTR retrotransposon for its mobility, and that the former is able to use the gag (and VLP) from the latter. Such hitchhiking implies a structural (same GAG-recognition signals) as well as translational (same time of expression) relationship between Tos17 and its autonomous partner. This association is probably a long-term association, as the structural annotation of the Tos17 elements (Figure 2A) reveals a complete removal of the gag region, without any identifiable remnants, but without damaging any other structural features of the element (LTR, PBS or polypurine tract (PPT)). Indeed, such clean elimination might have occurred during Tos17 evolution, with only elements within this correct deletion selected (able to be correctly expressed and mobilized by its partner), as no other Tos17-like element with gag remnants has been detected.

The use of Tos17 as an insertional tool for reverse genetics is not affected by this non-autonomous state, as long as requested functional and complementation analyses are performed to validate or invalidate the insertion as the real cause of the observed phenotype. The fact that Tos17 is not able to retrotranspose by itself may help to explain the high rate (almost 90%) of morpho-physiological variations untagged by Tos17 (or the transferred T-DNA) observed among regenerated lines ([9]; M Lorieux, unpublished data; B Hsingh, personal communication), which is probably also due to transposition of other elements, as shown previously [5].

Conclusion

Analyses, such as the one described here, highlight the need for a better knowledge of transposable elements (TEs), in order to ensure a better understanding of their effects upon the host genome. In particular, it may be of interest to further study the details of the relationships between the non-autonomous elements and their autonomous counterparts, because existing data suggest that the former are more active than the latter, as shown for BARE2 and Tos17.

Methods

The nucleotidic sequences from genomic copies of each element were launched in Artemis [15], and the ORFs longer than 100 residues were automatically extracted from the element sequences. The ORFs were then scanned online using a combination of Pfam, ProSite and BLASTp analyses [16] with standard parameters. The results were then reported on Artemis, in order to manually reconstruct the complete structure of each element. The LTRs were identified using Dotter [17], and the PBS and PPT were manually determined. The comparison between putative GAG-POL sequences was performed using the Align2Sequence graphical tool from the NCBI, through a BLASTp analysis, for a better presentation. The identity/similarity levels were calculated using the Stretcher program from the EMBOSS suite.