Most of our understanding of Drosophila heterochromatin structure and evolution has come from the annotation of heterochromatin from the isogenic y; cn bw sp strain. However, almost nothing is known about the heterochromatin's structural dynamics and evolution. Here, we focus on a 180-kb heterochromatic locus producing Piwi-interacting RNAs (piRNA cluster), the flamenco (flam) locus, known to be responsible for the control of at least three transposable elements (TEs). We report its detailed structure in three different Drosophila lines chosen according to their capacity to repress or not to repress the expression of two retrotransposons named ZAM and Idefix, and we show that they display high structural diversity. Numerous rearrangements due to homologous and nonhomologous recombination, deletions and segmental duplications, and loss and gain of TEs are diverse sources of active genomic variation at this locus. Notably, we evidence a correlation between the presence of ZAM and Idefix in this piRNA cluster and their silencing. They are absent from flam in the strain where they are derepressed. We show that, unexpectedly, more than half of the flam locus results from recent TE insertions and that most of the elements concerned are prone to horizontal transfer between species of the melanogaster subgroup. We build a model showing how such high and constant dynamics of a piRNA master locus open the way to continual emergence of new patterns of piRNA biogenesis leading to changes in the level of transposition control.RNAi | gene silencing | epigenetics O ver the course of evolution, transposable elements (TEs) have accumulated in the genomes of eukaryotes, where they can account for up to 85% of the DNA (1). Most of these sequences have lost their ability to transpose. They are now stable components of the genomes. Their conservation throughout evolution suggests that they may confer advantageous effects to their hosts. However, transposition of the copies that remain functional could generate deleterious mutations if they were not severely repressed by their host. RNAi, which is a gene-silencing mechanism triggered by small RNAs (reviewed in ref.2), has been identified as being the main cellular machinery involved in the "taming" of TEs (reviewed in refs. 3-5). RNAi pathways involve small RNAs of diverse families. Among them, Piwiinteracting RNAs (piRNAs) have been shown to be involved in TE silencing in the Drosophila ovary. These piRNAs, 23-29 nt long, are bound by the Argonaute proteins Piwi, Argonaute 3, or Aubergine. They are produced by discrete genomic loci named piRNA clusters, which have been described as containing vestiges of TEs (6). One of these loci, the flamenco (flam) locus, extends over 180 kilobases (kb) on the Drosophila X chromosome. It is proximal to the DISCO interacting protein 1 gene (DIP1) and close to pericentromeric heterochromatin. Before the identification of piRNAs, this locus had been shown to regulate the Gypsy retrotransposon (7, 8 (6) showed the potential for the flam cluster to pr...

Transposable elements can invade virgin genomes within a few generations, after which the elements are 'tamed' and retain only limited transpositional activity. Introduction of the I element, a transposon similar to mammalian LINE elements, into Drosophila melanogaster genomes devoid of such elements initially results in high-frequency transposition of the incoming transposon, high mutation rate, chromosomal nondisjunction and female sterility, a syndrome referred to as hybrid dysgenesis (for review, see refs 2-4); a related syndrome has also been described in mammals. High-frequency transposition is transient, as the number of I elements reaches a finite value and transposition ceases after approximately ten generations. It has been proposed that the I elements encode a factor that negatively regulates their own transcription, but evidence for such a mechanism is lacking. Using the hybrid dysgenesis syndrome in Drosophila as a model, we show here that transpositional activity of the I element can be repressed by prior introduction of transgenes expressing a small internal region of the I element. This autoregulation presents features characteristic of homology-dependent gene silencing, a process known as cosuppression. Repression does not require any translatable sequence, its severity correlates with transgene copy number and it develops in a generation-dependent manner via germline transmission of a silencing effector in females only. These results demonstrate that transposable elements are prone to and can be tamed by homology-dependent gene silencing, a process that may have emerged during the course of evolution as a specific defense mechanism against these elements.