A novel repeat-associated small interfering RNA-mediated silencing pathway downregulates complementary sense gypsy...

Replication of the gypsy endogenous retrovirus involves contamination of the female germ line by adjacent somatic tissues. This is prevented by flam, an as-yet-uncloned heterochromatic pericentromeric locus, at the level of transcript accumulation in these somatic ovarian tissues. We tested the effect of a presumptive RNA silencing mechanism on the accumulation of RNAs produced by constructs containing various gypsy sequences and report that the efficiency of silencing is indeed correlated with the amount of complementary RNAs, 25 to 30 nucleotides in length, in the ovary. For instance, while these RNAs were found to display a three- to fivefold excess of the antisense strands, only the transcripts that contain the complementary sense gypsy sequences could be repressed, indicating that they are targeted at the RNA, not DNA, level. Their size and asymmetry in strand polarity are typical of the novel repeat-associated small interfering RNA (rasiRNA)-mediated pathway, recently suspected to prevent the deleterious expression of selfish DNA specifically in the germ line. Unlike microRNAs (but like rasiRNAs and, surprisingly, siRNAs as well), gypsy rasiRNAs are modified at the 3' end. The rasiRNA-associated protein Piwi (but not Aub) is required for gypsy silencing, whereas Dicer-2 (which makes siRNAs) is not. In contrast, piwi, aub, and flam do not appear to affect somatic siRNA-mediated silencing. The amount of gypsy rasiRNAs is genetically determined by the flam locus in a provirus copy number-independent manner and is triggered in the somatic tissues by some pericentromeric provirus(es), which are thereby able to protect the germ line from retroviral invasion.