The Foldback-like element Galileo belongs to the P superfamily of DNA transposons and is widespread within the Drosophila genus

Abstract

Galileo is the only transposable element (TE) known to have generated natural chromosomal inversions in the genus Drosophila. It was discovered in Drosophila buzzatii and classified as a Foldback-like element because of its long, internally repetitive, terminal inverted repeats (TIRs) and lack of coding capacity. Here, we characterized a seemingly complete copy of Galileo from the D. buzzatii genome. It is 5,406 bp long, possesses 1,229-bp TIRs, and encodes a 912-aa transposase similar to those of the Drosophila melanogaster 1360 (Hoppel) and P elements. We also searched the recently available genome sequences of 12 Drosophila species for elements similar to Dbuz\Galileo by using bioinformatic tools. Galileo was found in six species (ananassae, willistoni, peudoobscura, persimilis, virilis, and mojavensis) from the two main lineages within the Drosophila genus. Our observations place Galileo within the P superfamily of cut-and-paste transposons and extend considerably its phylogenetic distribution. The interspecific distribution of Galileo indicates an ancient presence in the genus, but the phylogenetic tree built with the transposase amino acid sequences contrasts significantly with that of the species, indicating lineage sorting and/or horizontal transfer events. Our results also suggest that Foldback-like elements such as Galileo may evolve from DNA-based transposon ancestors by loss of the transposase gene and disproportionate elongation of TIRs.

Fig. 1.

Most complete copies of Galileo and 1360 found in this work. (A) Putative complete Galileo copy from the D. buzzatii genome. (B) Most complete copies of Galileo found in the 12 sequenced genomes. (C) Most complete copies of 1360. TIRs are represented as arrows and TPases are represented as gray rectangles. The direct repeats of the TIRs in Dbuz\Galileo are indicated by striped patterns. Dmoj\Galileo internal inverted repeats are represented as little triangles. In D. mojavensis two Galileo copies representative of two subfamilies found in this species are depicted. See SI Table 4 for details.

Fig. 2.

Neighbor-joining phylogenetic tree inferred from the analysis of 29 Galileo copies found in the D. mojavensis genome. The two TIRs of each copy were included in the tree as separate sequences to allow their comparison within and between copies. TIRa is the TIR located at 5′ from the TPase or the first TIR that appears in the contig if the copy could not be oriented. The complete deletion option was used leaving 269 informative sites. Bootstrap values at main nodes are shown. The average pairwise divergence between groups D and E is ≈25%, indicating a divergence time of ≈8 myr, and the average pairwise divergence between these two groups and groups C and F is ≈32%, implying a divergence time of ≈10 myr. The putative chimeric elements with highly divergent TIRs are marked with an arrow. Details of these Galileo copies are given in SI Tables 4 and 5.

Fig. 3.

Comparison of TIR ends. (A) Alignment of 40 bp of the TIR end of Galileo. A consensus sequence was constructed for Galileo TIRs in each TE subfamily and species. (B) Alignment of the 31-bp TIR of 1360. A representative TIR from a single copy of the TE is included. (C) Comparison of the Galileo TIR end with the TIRs of elements 1360 and P. Identical positions in all sequences are shown in black. Sites identical between Galileo and 1360 or P are shown in gray.

Fig. 4.

Neighbor-joining phylogenetic tree constructed with the eight consensus Galileo TPases, eight consensus 1360 TPases, and five TPases from representative P elements. The human P-like THAP9 protein is included as an outgroup. The complete alignment without Gblocks filtering is shown in SI Fig. 6. The tree topology was identical when using maximum likelihood and parsimony methods.