Silencing in trans: position matters in fission yeast

Abstract

The distinction in RNAi-mediated gene silencing between metazoans —which mostly use a post-transcriptional RNAi mechanism— and fission yeast —which use a transcriptional RNAi mechanism— seems to be less clear-cut than previously thought. Robin Allshire's group has recently published in EMBO reports that S. pombe can repress gene expression in trans, which is reminiscent of mammalian post-transcriptional gene silencing.

EMBO Rep (2010) 11: 2 112–118. doi:10.1038/embor.2009.273

RNA interference (RNAi)-mediated gene silencing is a widespread mechanism in eukaryotes that operates at several levels of gene expression (reviewed in Carthew & Southeimer, 2009). In metazoans, small interfering RNAs (siRNAs) derived from longer double-stranded RNAs (dsRNAs) target mRNA for degradation in the cytoplasm through the activation of the RISC complex. In this case, the dsRNA is normally exogenous, possibly of viral origin or introduced experimentally by transfection. Alternatively, nuclear dsRNA synthesis—often arising from hairpin structures in the introns of mRNA encoding genes—can give rise to cytoplasmic microRNAs and similarly activate RISC to either degrade mRNA or block its translation. By contrast, in lower eukaryotes RNAi seems to operate predominantly at the gene level, as nuclear dsRNA directly targets chromatin repression. Thus, in fission yeast RNAi is initiated by the production of siRNAs from dsRNA by the ribonuclease III enzyme Dicer (Dcr1). These siRNAs are loaded onto the Argonaute-containing RNAi transcriptional silencing complex (RITS), which somehow targets nascent transcripts (Buhler et al, 2006) and also recruits RNA-dependent RNA polymerase (Rdp1)-containing complex RDRC (Motamedi et al, 2004). Rdp1 effectively amplifies the Dcr1 dsRNA substrate. The RNAi machinery subsequently recruits the histone methyltransferase (Clr4)-containing complex CLRC to target loci, thereby promoting H3 Lys 9 methylation (Hong et al, 2005). In turn, this forms a platform for heterochromatin binding proteins Swi6 (HP1 homologue), Chp1 and Chp2, creating a nucleation site from where heterochromatin can spread to adjacent regions.

Interestingly, the distinction between metazoans, which seem to predominantly use a post-transcriptional RNAi mechanism, and fission yeast, which uses a more direct transcriptional RNAi mechanism, could be less clear-cut than previously thought. Recent results in mammals suggest a role of RNAi in transcriptional gene silencing through the targeting of promoter regions (Han et al, 2007) and—in the February issue of EMBO reports—the Allshire group showed that Schizosaccharomyces pombe can repress gene expression through a trans effect that is reminiscent of mammalian post-transcriptional gene silencing (Fig 1; Simmer et al, 2010). Both transcriptional gene silencing and post-transcriptional gene silencing are also well known to coexist in plants (Baulcombe, 2004).

...the Allshire group showed that S. pombe can repress gene expression through a trans effect [...] reminiscent of mammalian post-transcriptional gene silencing

Figure 1.

Trans-silencing depends on the position of the target gene. siRNAs generated from a hairpin (shown in red) have the ability to induce silencing through the RNAi pathway; the siRNAs generated are shown below the centromere. Trans-silencing is only efficient if the target gene (shown as a blue arrow) is located near heterochromatic loci, such as centromeres, or positions of convergent transcription on chromosomal arms. CLRC, Clr4-containing complex; RDRC, RNA-dependent RNA polymerase-containing complex; RITS, RNAi transcriptional silencing complex.

Previous studies in fission yeast suggested that RNAi gene silencing is only efficient when the dsRNA that induces the RNAi response comes from the same site as the silenced genes—that is, in cis. When dsRNA was made at a separate location (that is, in trans)—either elsewhere in the genome or from a transformed plasmid—RNAi effects were only detectable in combination with other genetic modifications, such as deletion of eri1 (Buhler et al, 2006) or overexpression of swi6 (Iida et al, 2008). However, the new study by Simmer and colleagues brings further insight into the mechanism of RNAi-mediated gene silencing in trans. In particular, the authors show that exogenous hairpin RNA made in trans induces heterochromatin formation in otherwise unmodified wild-type cells. They then detect secondary siRNAs that are not derived from the hairpin sequence but are still close to the target gene. This class of siRNA is dependent on Rdp1 activity in vivo. Furthermore, they describe the presence of Ago1-associated siRNAs that do not necessarily lead to gene silencing (Simmer et al, 2010).

The ectopic expression of a long GFP hairpin RNA had been shown previously to lead to dsRNA production and post-transcriptional silencing of adh1:gfp mRNA. This silencing had no effect on the transcriptional rate of the target GFP gene, did not lead to heterochromatin formation and occurred in a RITS-independent manner (Sigova et al, 2004). By contrast, tethering of Tas3 (RITS complex) to a target ura4 gene (that is, in cis) was shown to induce the production of siRNA, the recruitment of RNAi components and the cis-silencing of the ura4 target gene. However, a second separate (that is, in trans) ura4 allele was not silenced, suggesting the negative control of trans-silencing. In fact, only the deletion of eri1 allowed partial silencing of the second ura4 copy, suggesting that trans-silencing is influenced by additional factors (Buhler et al, 2006). Another study showed that siRNA produced from long hairpin RNA can induce trans-silencing, but it depends on local chromatin structure and correlates with antisense transcription at the target locus. Furthermore, trans-silencing is dependent on dsRNA synthesis by RDRC, although the production of primary siRNA from a hairpin is independent of Rdp1. This hairpin-induced siRNA trans-silencing was also enhanced by transient overexpression of Swi6 (Iida et al, 2008).

...ectopic expression of ura4⁺ hairpin RNA (U-HP) can induce heterochromatin formation in trans in the absence of any other genetic modifications

The Allshire group has now shown that ectopic expression of ura4⁺ hairpin RNA (U-HP) can induce heterochromatin formation in trans in the absence of any other genetic modifications. The expression of U-HP produces ura4 siRNA but does not silence ura4 at its endogenous position. However, it does lead to significant silencing of ura4 when it is integrated close to a pericentromeric repeat, which is consistent with previous observations (Iida et al, 2008). The expression of a GFP hairpin (GFP-HP) in the arg3::ura4⁺GFP strain also produces specific gfp siRNA, which are loaded onto Ago1 in the RITS complex. Interestingly, both gfp and ura4 siRNAs can be detected, suggesting that secondary siRNA can also be produced in fission yeast in vivo (Fig 1). Sequence analysis revealed that this secondary siRNA pool is dependent on Rdp1 activity, which is consistent with previous data (Iida et al, 2008). Simmer and co-workers also show that hairpin gfp siRNA partly represses the expression of arg3::ura4⁺GFP grown on 5-FOA plates, but does not significantly decrease the steady state RNA levels of ura4 in cells grown in non-selective medium. Similarly, hairpin gfp siRNA can silence the ade6 gene in an ade6–GFP strain. Colonies grown in low adenine give rise to several red spots, confirming partial repression of ade6. These data suggest that, although there is a significant pool of hairpin siRNA associated with the RITS complex, it does not guarantee efficient silencing of target genes. Furthermore, Simmer and colleagues confirm that silencing in trans depends not only on RNAi factors (Dcr1, Ago1, Arb1 and Rdp1), but also on Clr4, chromodomain proteins Chp1, Swi6 and Chp2, and deacetylases Sir2 and Clr3. These data predict that hairpin-derived siRNA induces heterochromatin formation on target genes, which is also confirmed by the presence of methylated H3 Lys 9 and Swi6.

It is interesting that the efficiency of hairpin siRNA-induced trans-silencing depends on the position of the target gene (Fig 1). Endogenous ura4 is a tandem gene and cannot be trans-silenced; however, when ura4 is integrated into arg3, ade6 or trp1 loci—which are all convergent genes—trans-silencing can be partly established (Iida et al, 2008). Furthermore, the integration of ura4 close to pericentromeric regions makes it a strong target for silencing in trans. Convergent genes have been shown previously to result in antisense transcription and transient heterochromatin formation (Gullerova & Proudfoot, 2008), and centromeric repeats are characterized by sense and antisense transcription from convergent promoters, serving as templates for dsRNA synthesis and RNAi-mediated heterochromatin formation. Trans-silencing seems to be more efficient on gene targets that are close to heterochromatic regions. Consistent with this idea, antisense transcription has been detected at the trp1::ura4 locus, which is a strong target for hairpin siRNA-induced silencing (Iida et al, 2008). However, Simmer and colleagues do not detect antisense RNAs at the arg3::ura4 locus, which is only partly silenced by hairpin siRNA, suggesting that antisense transcription is not absolutely required for silencing in trans, although it enhances the trans-silencing effect. Furthermore, close proximity of heterochromatin regions stimulates the trans-silencing of target genes.

...the efficiency of hairpin siRNA-induced trans-silencing depends on the position of the target gene

Overall, this interesting study by the Allshire lab provides new mechanistic insight about trans-silencing in fission yeast, proving the ability of hairpin siRNA to induce the silencing of target genes in wild-type cells. S. pombe clearly has the potential to silence genes in trans; however, this process seems to be restricted by position and might operate at a co-transcriptional rather than post-transcriptional level.