Unexpected variations in posttranscriptional silencing induced by differentially produced dsRNAs

Small RNAs (sRNAs) present in a cell can induce posttranscriptional silencing of complementary genes and methylation of homologous DNA sequences. In plants, sRNAs are generated from various dsRNA precursors. Here we characterized in detail the time course of GFP silencing induced by three silencers with a different mechanism of dsRNA formation, which was based on antisense (AS), unterminated sense (UT) and inverted repeat (IR) RNAs. These silencers were expressed from the estradiol-inducible promoter in homogeneously responding tobacco BY-2 cell lines. This model allowed us to follow many aspects of the PTGS onset, silencing reversibility and the impact of the initial dsRNA level.Detailed analyses of selected lines showed that the strongest and fastest silencing was induced by IR, and this silencing was accompanied by dense DNA methylation of both target GFP and silencer loci. Interestingly, under very weak induction, IR caused discontinuous silencing with clear binary character at the individual cell level. The IR silencing did not induce any transitive sRNAs, unlike AS and UT variants that were accompanied with strong transitivity to 3'UTR of the GFP transcript. In contrast to IR and UT, the AS variant showed strongly deviated sRNA-strand ratio shifted in favor of antisense orientation and surprisingly, this variant was able to induce weak DNA methylation only at the silencer, but not at the target locus. After removal of the induction, both IR and AS variants fully recovered GFP expression, but stable sRNA levels and only partial reactivation of GFP expression were observed in the UT variant, even though the silencer construct was no longer transcribed.Comparison of the GFP silencing induced by the three different sRNA sources in BY-2 cell lines pointed out to several not yet described phenomena and differences between the silencer variants, some of which were truly unexpected. Although a lot is known about how RNAi works, some mechanisms behind the processes we observed are hard to explain which opens up the space for further research.