Retroviruses are commonly considered to be restricted to vertebrates. However, the genome of many eukaryotes contains mobile sequences known as retrotransposons with long terminal repeats (LTR retrotransposons) or viral retrotransposons, showing similarities with integrated proviruses of retroviruses, such as Ty elements in Saccharomyces cerevisiae, copia-like elements in Drosophila, and endogenous proviruses in vertebrates. The gypsy element of Drosophila melanogaster has LTRs and contains three open reading frames, one of which encodes potential products similar to gag-specific protease, reverse transcriptase, and endonuclease. It is more similar to typical retroviruses than to LTR retrotransposons. We report here experiments showing that gypsy can be transmitted by microinjecting egg plasma from embryos of a strain containing actively transposing gypsy elements into embryos of a strain originally devoid of transposing elements. Horizontal transfer is also observed when individuals of the "empty" stock are raised on medium containing ground pupae of the stock possessing transposing elements. These results suggest that gypsy is an infectious retrovirus and provide evidence that retroviruses also occur in invertebrates.

The gypsy element of Drosophila melanogaster is the first retrovirus identified so far in invertebrates. According to phylogenetic data, gypsy belongs to the same group as the Ty3 class of LTR-retrotransposons, which suggests that retroviruses evolved from this kind of retroelements before the radiation of vertebrates. There are other invertebrate retroelements that are also likely to be endogenous retroviruses because they share with gypsy some structural and functional retroviral-like characteristics. Gypsy is controlled by a Drosophila gene called flamenco, the restrictive alleles of which maintain the retrovirus in a repressed state. In permissive strains, functional gypsy elements transpose at high frequency and produce infective particles. Defective gypsy proviruses located in pericentromeric heterochromatin of all strains seem to be very old components of the genome of Drosophila melanogaster, which indicates that gypsy invaded this species, or an ancestor, a long time ago. At that time, Drosophila melanogaster presumably contained permissive alleles of the flamenco gene. One can imagine that the species survived to the increase of genetic load caused by the retroviral invasion because restrictive alleles of flamenco were selected. The characterization of a retrovirus in Drosophila, one of the most advanced model organisms for molecular genetics, provides us with an exceptional clue to study how a species can resist a retroviral invasion.