The experiments in this report provide evidence for new families of insertion elements in the 28S genes of Drosophila
. Segments from R2 and/or R1 elements comprise these insertions, and they are mobilized by hijacking the R2 or R1 retrotransposition machinery. Because these non-autonomous elements rely (as does the R2 element itself) on cotranscription with the 28S gene, they are referred to as SIDEs rather than SINEs. Non-autonomous DNA-mediated transposable element families, such as the miniature inverted-repeat DINE-1 and non-autonomous P elements, have been previously documented in Drosophila
]. The R2 SIDE and R2/R1 hybrid SIDEs along with HeT-A [42
] are, however, the only clear examples of non-autonomous retrotransposons to be found in Drosophila
. Analysis of the SIDEs provides direct support for the model that R2 retrotransposition requires only the 5' end for RNA self-cleavage from a 28S cotranscript and the 3' UTR for binding the R2 protein to initiate TPRT. The discovery of SIDEs mobilized by the R1 machinery also provides strong support for the model [19
] that the R1 protein recognizes the 3' UTR sequences/secondary structure of its RNA to initiate TPRT and thus belongs to the class of stringent non-LTR retrotransposable elements.
Because there is a single lineage of R2 element vertically transmitted in Drosophila
], the levels of divergence between ribozyme sequences (excluding the highly variable J1/2 loop) from different elements can be compared to provide an estimate of the number of independently formed SIDEs and their approximate ages. First, the 25% sequence divergence between the ribozymes from the R2 element and R2 SIDE of D. willistoni
is similar to the divergence between the ribozymes from the R2 elements from D. willistoni
and D. melanogaster
(23%) as well as between D. ananassae
and D. melanogaste
r (28%). Assuming similar levels of constraint on the ribozyme of these elements, this suggests the R2 SIDE lineage is as old as the divergence between species groups within the Sophophora
subgenus, that is, over 40 million years [43
]. Second, the 27% sequence divergence between the R2 and hybrid SIDE ribozymes from D. immigrans
indicates the R2/R1Dim_SIDE lineage also dates back to a comparable time within the Drosophila
subgenus. Third, the lower levels of sequence divergence between the ribozymes from R2/R1Dwi_SIDE and R2Dwi (11%) and between the ribozymes from R2/R1Dfa_SIDE and R2Dfa (10%) suggests both of these hybrid SIDEs have a more recent origin (approximately 20 million years ago). Because D. falleni
and D. willistoni
are in different subgenuses, their hybrid SIDEs arose independently. Finally, because R2/R1Dfa_SIDE and R2/R1Din_SIDE have only 3% sequence divergence, they likely represent the same event in the ancestor of these two closely related species. In summary, the five identified SIDEs in this report appear to have originated in four separate events.
Non-autonomous elements of DNA transposons (for example, miniature inverted-repeat transposable elements (MITEs)) and LTR retrotransposons (for example, terminal-repeat retrotransposons in miniature (TRIMs)) have been found to originate from autonomous elements by internal deletions [6
]. The non-LTR, non-autonomous elements TbRIME and Ag-Sponge also appear to have arisen by internal deletions [49
]. TbRIME is of special interest because it has sequence identity at the 5' end to the ribozyme encoded by L1Tc [31
]. Two potential mechanisms could have formed the Drosophila
SIDEs. First a template jump [35
] during a retrotransposition reaction could have fused the 3' and 5' ends of an R2 element. The R2 5' junctions with upstream snU12 RNA and tRNA sequences shown in Figure and Additional file 2
demonstrates the R2 protein does have the ability to template jump in vivo
. In the case of the hybrid SIDEs, R1 sequences are located downstream of the R2 sequences, therefore, it is the R1 reverse transcriptase that must be postulated as responsible for the jumps. A second more likely possibility for the formation of the SIDEs is that non-homologous recombination within the rRNA gene locus joined the 5' end of R2 to either the 3' end of R2 or the 3' end of R1. Such recombinants could have been the result of DNA repair after retrotransposition events. The R2 machinery has been associated with large deletions of upstream rDNA sequences in D. melanogaster
] and D. simulans
]. Alternatively, the recombinations generating the SIDEs could simply have been aberrant versions of the frequent crossovers that give rise to the concerted evolution of the rDNA locus. Whatever the scenario, it seems unlikely that the SIDEs were formed in their present configuration. All SIDE families appear old, thus there has been ample opportunity for subsequent internal deletions to shorten the SIDEs until only the minimal sequences needed for activity remain.
Based on the sequence conservation of each SIDE, it appears that these elements have recently been active. Since their formation, the ribozymes and 3' ends of the SIDEs appear to be evolving similarly to the corresponding regions of R2 and R1 with two notable exceptions. A highly conserved ‘U’ located in the catalytic region of 18/19 Drosophila
R2 ribozymes as well as in the R2 SIDE itself (pink circle, Figures AAA and D) has been substituted with an ‘A’ in all hybrid R1/R2 SIDEs. This substitution may reflect the difference in the insertion site of the hybrid SIDEs and consequently the upstream 28S sequences that must be cleaved from the cotranscripts. The second exception is a stop codon that is found in J4/2 in 18/19 R2 elements (pink box, Figures AA and D) but not found in any of the five SIDEs. We suggest this stop codon is important in the initiation of translation of the R2 RNA open reading frame by way of an encoded internal ribosome entry site (IRES) [54
], a function obviously not required for RNA arising from the SIDEs.
In general, non-LTR SIDEs appear to be rare. An L1 SIDE has not been observed despite the fact that studies of L1 retrotransposition in cultured cells revealed the generation of chimeric and internally deleted L1 insertions [38
]. The cis preference of the L1 ORF2 protein for its RNA can, however, readily explain the absence of an associated SIDE [56
]. Likewise, our survey of 39 Drosophila
species suggests that the formation of R2/R1 hybrid SIDEs and to a greater extent R2 SIDEs is also rare and/or their survival after formation is unlikely. While there is no evidence that R1 and R2 undergo cis preference, our studies on R2 expression and regulation suggest an explanation for the paucity of R2 associated SIDEs [57
]. Our current model suggests that Drosophila
has the ability to select for transcription a localized region in the rDNA locus that has the lowest level of insertions. Because the SIDEs as well as the R2 elements rely on cotranscription with the 28S gene, their transcription can only occur whenever an rDNA unit with the insertion is located within this transcription domain. Consequently, in order for an R2 SIDE to retrotranspose both a copy of the SIDE and a copy of the autonomous R2 element must be present in the small transcription domain. Because the R2 lineage itself appears somewhat unstable and has been lost in several species of Drosophila
], the survival of an R2 SIDE would be even more tenuous. However, R1 elements have been suggested to contain their own promoter and thus may not need to be within the transcription domain for activity. R1 elements are present in all lineages of Drosophila
and indeed many species have two distinct lineages [21
]. The greater evolutionary stability of the R1 retrotransposition machinery and the independence of transcriptional control of the hybrid SIDE from the autonomous R1 elements may explain why these SIDEs appear to have a greater chance of long-term survival within the locus.