The de novo silencing of transposable elements in plants and animals is mediated in part by RNA‐directed chromatin modification. In flowering plants, AGO4 has been seen as the key argonaute protein in the RNA‐directed DNA methylation pathway that links the plant‐specific RNA polymerase V with the de novo DNA methyltransferase DRM2 (Zhong et al, 2014). Two recent papers in The EMBO Journal strongly implicate a role for the AGO6 protein in the process of de novo silencing.
See also: C‐G Duan et al and
The RNA‐directed DNA methylation (RdDM) pathway is important for the silencing of active and evolutionarily young transposable elements in Arabidopsis (Zhong et al, 2012; Marí‐Ordóñez et al, 2013). Several forward genetic screens for RdDM components have been carried out in Arabidopsis using various reporter gene silencing assays and have identified more than 30 genes that play a role in RdDM, including multiple alleles of two related argonaute genes, AGO4 and AGO6 (Zilberman et al, 2003; Zheng et al, 2007; Eun et al, 2011).
In this issue of The EMBO Journal, Duan et al (2015) address the requirements of AGO4 and AGO6 in the RdDM pathway by conducting whole‐genome bisulfite sequencing in ago4 and ago6 single mutants, and the ago4 ago6 double mutant. They find that 53% of the loci identified in the ago4 ago6 double mutant are hypomethylated to a similar extent in both ago4 and ago6 single mutants. A further 42% of loci identified in the double mutant were also found in ago4 and ago6 single mutants, but hypomethylation was more pronounced in the double mutant. From this, Duan et al conclude that ago4 and ago6 are mutually dependent and therefore function in parallel, rather than in a redundant manner (see Fig 1A).
A plausible explanation for the mutual dependency of AGO4 and AGO6 could be the different expression patterns of AGO4 and AGO6 in Arabidopsis. AGO6 is most highly expressed in meristems and in tissue enriched for dividing cells. AGO4, on the other hand, is expressed throughout the plant, including terminally differentiated tissue like leaves, where AGO6 is not expressed (Zheng et al, 2007; Havecker et al, 2010). It is possible that in ago6 mutants, loss of DNA methylation occurs at an early meristematic stage and thus is maintained in all of the derived cell lineages, including leaves where AGO4 is expressed.
However, over and above this simple model, Duan et al provide an additional insight into the non‐redundant roles of AGO4 and AGO6, by examining their sub‐nuclear localization. They find that AGO4 localization in the nucleoplasm overlaps with RNA polymerase II (Pol II), and not with RNA polymerase V (Pol V) as would be expected from biochemical purification (Zhong et al, 2014). In contrast, AGO6 localization in the nucleoplasm overlaps with Pol V and not Pol II. Consistent with these results, the authors demonstrate that AGO4 can co‐immunoprecipitate Pol II while AGO6 cannot. This suggests that AGO4 and AGO6 may carry out different functions within the nucleus and that both proteins may be required to silence a given transposable element in a single cell. In D. melanogaster, two argonaute‐like PIWI proteins are required for piRNAs to initiate transposon silencing, leading to the famous ping‐pong cycle (Brennecke et al, 2007; Gunawardane et al, 2007). It remains to be determined whether there are any parallels with AGO4 and AGO6.
In another recent publication in The EMBO Journal, McCue et al (2015) provide complementary insight into AGO6 function by using a ddm1 mutant background, where the majority of heterochromatic DNA methylation is lost and re‐activation of transposable elements is widespread (Lippman et al, 2004). The authors focus on the Athila6A retroelement that they had previously shown to acquire asymmetric DNA methylation in the long terminal repeat (LTR) in ddm1, in a manner that depended on AGO6 and the RNA‐dependent RNA polymerase RDR6 (Nuthikattu et al, 2013). The rest of the element loses DNA methylation extensively and is strongly transcribed in ddm1 mutants.
McCue et al performed small RNA sequencing in a ddm1 background using AGO6‐immunoprecipitated RNA and total RNA. They found that AGO6‐immunoprecipitated small RNAs overlap the Athila6A LTR, including the region of the element that is de novo methylated. However, by far the most abundant small RNAs (~90%) targeting Athila6A overlap the 3′ end of the transposable element and correspond to 21 nucleotide easiRNAs which require AGO1 for biogenesis (see Fig 1B) (McCue et al, 2012; Creasey et al, 2014). Thus, again in this system, two argonaute proteins appear to be important, in this case for the de novo silencing of transposons in a ddm1 background.
In their discussion, McCue et al distinguish the canonical RdDM pathway, which requires the plant‐specific RNA polymerase IV (Pol IV) and RDR2, from the RDR6‐RdDM pathway that appears to be important in targeting LTR methylation de novo in ddm1 mutants. McCue et al propose that AGO6 and RDR6 are components of an ancient RdDM pathway that targets actively transcribed elements for DNA methylation. Importantly, McCue et al identify Pol V, but not Pol IV, as interacting genetically with AGO6 to guide LTR methylation, consistent with the immunolocalization results of Duan et al. Taken together, it appears that AGO6, along with co‐operating partners, is a central player in RdDM in Arabidopsis.
- © 2015 The Authors