Evidence for the somatic transposition of mobile DNAs in life forms other than mammals is mixed. Tc1, the major DNA transposon of Caenorhabditis elegans, is able to transpose in somatic cells [2], whereas in Drosophila species, P elements and the I factor, an L1-like non-LTR retrotransposon, appear limited to germ-cell mobility [3, 4]. Now, Eickbush and Eickbush, in a study published in Mobile DNA [5], find that transposition of another Drosophila autonomous retrotransposon, R2, can occur in somatic cells during early embryonic development. R2 is a non-LTR retrotransposon with many features in common with mammalian L1 elements, most notably its ability to reverse transcribe and integrate in a single step directly into chromosomal DNA [6]. However, R2 differs from L1 in that it inserts at a single site in the 28S rRNA genes whereas L1 can enter the genome at a very large number of essentially random, short consensus sequences. In addition, the endonucleases encoded by the two elements differ in their position within the element and their enzymatic type (restriction enzyme type IIs for R2 and apurinic/apyrimidinic endonuclease for L1).

Using single-step PCR, Eickbush and Eickbush found evidence for 15 somatic early embryonic insertions in 7 of 29 flies studied. This number of somatic insertions is clearly a minimum, and one wonders how many more insertions would have been detected with two-step, nested PCR. The detected insertions were present in multiple tissues of both the adult and larval stages, and had all the characteristics of authentic R2 insertions. They occurred at, or very close to, the 28S rRNA gene insertion site of R2. Many had a few non-templated nucleotides at the 5' end, and all were 5' truncated. Because the same somatic insertions were found in different tissues, the timing of many events could be estimated as early in development, before the differentiation of tissues, including the germ line.

A previous study of 213 R2 insertions in the offspring of a single female fly found 32 new insertions [7]. Twenty-seven of these were clear germline events, occurring in a single fly each. The remaining five appeared to be derived from somatic events because the identical insertion occurred in more than one fly, meaning that one parent was a germline mosaic and had the same insertion present in many but not all germ cells. It was also previously known that R2 retrotransposition occurs much less frequently in males than in females, and now Eickbush and Eickbush suggest that perhaps all of male R2 retrotransposition may be due to germline mosaicism. Thus, it appears that the incidence of somatic R2 retrotransposition is not very dissimilar from that of germline R2 insertion.

Over the past six years substantial evidence for somatic insertion of mammalian L1 elements has accumulated. Muotri et al. [8] found L1 retrotransposition in neuronal precursors in mouse brains, specifically hippocampus, using engineered human L1 transgenes. Coufal et al. [9] used quantitative PCR to expand those data to include increased endogenous L1 insertions in mouse and human brain regions compared to other tissues. Muotri et al. [10] subsequently showed that mice with a knockout for the gene MeCP2, encoding the methyl-CpG-binding protein 2, have more engineered L1 insertions from L1 transgenes in the hippocampus compared with normal mice. They also found that female human patients with the neurodevelopmental disorder Rett syndrome (who have a natural deficiency of MeCP2) have small but significant increases of L1 insertions in their hippocampi. Meanwhile, van den Hurk et al. [11] had demonstrated an example of early embryonic L1 insertion in a human who exhibited both germline and somatic mosaicism for an L1 insertion, Garcia-Perez et al. [12] showed retrotransposition of a transfected L1 in human embryonic stem cells, and Kano et al. [13] showed that human and mouse L1 transgenes produce more retrotransposition events in early embryogenesis in mice and rats than in the germline. Very recently, Baillie et al. [14] have used enrichment of human retrotransposons followed by next-generation sequencing to discover an amazing number (thousands) of somatic retrotransposition events in the hippocampus. Thus, there is no longer any doubt that somatic insertion of L1 occurs in humans and mice.

These findings beg many questions. Transposition of mobile DNA can be destructive to gene function and host organisms possess mechanisms for suppressing transposition, especially in the germline. So why does somatic retrotransposition occur? Why don't host controls on retrotransposition block it in early embryogenesis? From the point of view of the evolution of the animal host, if retrotransposition is likely to occur at all, it is least damaging if it occurs in somatic cells, which are not heritable in the next generation, rather than in the germline, where it may cause mutations that can be passed on to offspring. In addition, somatic insertions are less likely to be immediately detrimental to the host than germline insertions as they affect only a limited number of cells. Thus, if host resources to control retrotransposition are limited, they would best be used in the developing germline.

In the future, we need to learn why various hosts allow somatic insertions for one element but not for another, and what is the frequency of somatic retrotransposition of various mobile DNAs in a variety of organisms and tissue types. We should also learn more about the stages of embryonic development in which most somatic insertions occur, and the role of mobile DNAs, if any, in oncogenesis. Perhaps we will also discover whether controls on transposon mobility in somatic cells in various hosts are similar to those used to control mobility in the germline. In any case, the study by Eickbush and Eickbush [5] provides new evidence for the importance of somatic insertion of L1-like non-LTR retrotransposons.