Proposed mechanism for the initiation of transposable element silencing by the RDR6-directed DNA methylation pathway.

The activity of transposable elements is epigenetically suppressed by both transcriptional and post-transcriptional mechanisms. We recently identified a direct connection between the small RNA-mediated post-transcriptional mRNA degradation of actively transcribing transposable elements and the de novo methylation of transposable element DNA, providing a mechanistic link between these two established pathways of transposable element silencing. Here we provide a model for the initiation, establishment and epigenetic maintenance of transposable element silencing that incorporates recent data in this rapidly emerging field.

Transposable Element Silencing by Small Interfering RNAs

In plants, transposable elements (TEs) are epigenetically repressed through transcriptional gene silencing (TGS) via the establishment of RNA-directed DNA methylation (RdDM) (reviewed in refs. 1 and 2). Current models of RdDM rely on transcription of silenced TEs by RNA Polymerase IV (PolIV) and this transcript is quickly acted upon by particular RNA-dependent RNA polymerase and Dicer-like family proteins (RDR2 and DCL3) to generate specifically 24 nucleotide (nt) small interfering RNAs (siRNAs). This size class of siRNA is incorporated into either Argonaute4 (AGO4) or AGO6 to target a nascent scaffolding transcript generated by RNA Polymerase V (PolV) (reviewed in ref. 1). Effective targeting of PolV transcripts results in establishment of DNA and histone methylation, reinforcing the marking of the TE as silenced heterochromatin.

Transcriptionally reactivated TEs are additionally repressed by post-transcriptional gene silencing (PTGS) via mRNA degradation. Like RdDM, PTGS uses RNA-dependent RNA polymerases and Dicer-like proteins, however the individual proteins are different members of these protein families. The RDR6 protein initiates PTGS and the Poll transcript is degraded by DCL2 and DCL4 to generate 21–22 nt siRNAs. These siRNAs are incorporated into AGO1 and target complementary TE mRNAs for cleavage. The role of RDR6 is therefore to supply the mRNA degradation cycle known as RNAi with amplified secondary 21-22 nt siRNAs.

How the TGS and PTGS pathways are interconnected and more specifically how RdDM and TGS can be initiated from an actively transcribing TE, has remained an unanswered question. Within the last year, studies have uncovered a new role for 21–22 nt siRNAs (previously associated with PTGS) in DNA methylation.- The production of TE 21–22 nt siRNAs is dependent on RDR6 and therefore we refer to this newly identified pathway of DNA methylation as RDR6-RdDM to distinguish it from PolIV-RdDM. We recently investigated the role of both RDR6-RdDM and PolIV-RdDM in the epigenetic maintenance of TE TGS, the initiation of TE silencing and the corrective reestablishment of silencing for recently activated TEs. Our data suggests a model of TE silencing that distinctly utilizes both of these individual pathways to effectively function to identify and repress TE activity (Fig. 1).

Figure 1. Model of TE silencing by RDR6-RdDM and PolIV-RdDM. (A) Active transcription of a TE by PolII (green) generates a TE mRNA. Some TE mRNAs are recognized by the RDR6 protein and are subject to PTGS. The 21 and 22 nt siRNA (green and blue) products of this RDR6-dependent degradation can target other mRNAs for degradation through cycles of endogenous RNAi. (B) Some of the 21 and 22 nt siRNAs produced by RNAi can target PolV scaffolding transcripts, initiating de novo DNA methylation in all sequence contexts (CG, red; CHG, blue; CHH, green). This phase represents the initiation of TGS by RDR6-RdDM. (C) Once partially methylated, PolIV transcription of the TE will activate PolIV-RdDM and the production of 24 nt siRNAs (red and orange), which then target PolV transcripts and additional DNA methylation. This phase represents the establishment and reinforcement of TGS and TE silencing. (D) Eventually RdDM is not required to maintain TE silencing, as TE silencing is now only dependent on the maintenance of CG and CHG DNA methylation. The activity of this TE is now fully silenced and this silencing is propagated by trans-generational epigenetic inheritance. (E) Upon entering a new genome, the only way to initiate the silencing of a unique TE (that does not share homology to the TEs already present in that genome) may be through the PTGS and RDR6-RdDM pathways. (F) If a TE enters a genome that shares sequence similarity to the existing TEs within that genome, the homology-sensing mechanism of PolIV-RdDM and 24 nt siRNAs will recognize and efficiently silence the incoming TE.

Proposed Model of Initiation, Establishment and Maintenance of TE Methylation and Silencing

Similar to genes, epigenetically active TEs are transcribed by RNA Polymerase II (PolII) into mRNAs and produce proteins that enable the TE to transpose. This transcriptional activity represents the “default” epigenetic state of TEs. An active TE in a genome may be tolerated for a variable length of time, which depends on how mutagenic and disruptive that TE is. At some point in time, the active TE will be targeted for silencing, which can be triggered through several different mechanisms, depending on if the TE has homology to other silenced TEs in the genome (see below).

One route to trigger silencing of a TE is its recognition by RDR6 and the RNAi pathway. The production of 21–22 nt siRNAs from actively transcribing TEs is dependent on RDR6, DCL2, DLC4 and AGO1 (Fig. 1A)., Incorporation of the DCL2 and DCL4-dependent 21–22 nt TE siRNAs into AGO1 results in the cycle of RNAi and effective PTGS (Fig. 1A). What specific feature of the TE mRNA attracts the RNAi pathway is currently unknown and remains a major focus of future investigation. DCL1 may also have some contribution to the production of these 21–22 nt siRNAs, but this cleavage would likely be RDR6-independent.

The focus of publications in the last year has been on understanding how the RNAi cycle described in Figure 1A initiates RdDM.,, We have discovered that some TE 21–22 nt siRNAs generated by RDR6, DCL2, DCL4 and AGO1 can act in the nucleus to initiate de novo DNA methylation through the RDR6-RdDM pathway (Fig. 1B). The role of these siRNAs in DNA methylation and the dependence of RDR6-RdDM pathway on PolII transcription are key discoveries toward initiation of TE silencing and form the basis of the following working model. This de novoDNA methylation may depend on 21–22 nt siRNAs being incorporated into other AGO proteins that direct DNA methylation, such as AGO2 or AGO6.,,. Wu et al. demonstrated that DNA methylation at non-TE loci triggered by 21–22 nt siRNAs depends on these siRNAs being incorporated into the AGO6 protein, which is normally associated with 24 nt siRNAs. However, RDR6-RdDM could act through a different or multiple AGO proteins. RDR6-RdDM is dependent on a PolV scaffolding transcript and the de novo DNA methyltransferase DRM2 (Fig. 1B)., This aspect of the RDR6-RdDM pathway is identical to the down-stream components and requirements of the PolIV-RdDM pathway (Fig. 1B and C). It is not known how PolV is initially recruited to TEs and future research needs to focus on how particular loci are selected to be transcribed by PolV (potentially targeting only them for RdDM), or if all regions are targeted by low-level or transient PolV transcription.

After the initial round of DNA methylation via RDR6-RdDM, partially methylated TE loci are now a target for PolIV-RdDM (Fig. 1C). As described above, this pathway reinforces methylation at the TE locus through the activity of PolIV transcription, RDR2, DCL3, 24 nt siRNA production and targeting of DNA methylation through AGO4, AGO6, POLV and DRM2 (Fig. 1C and reviewed in ref. 1). The TE may go through multiple rounds of PolIV-RdDM, increasing the DNA methylation at each cycle. In contrast to the initiation phase of DNA methylation and TGS instigated by RDR6-RdDM, PolIV-RdDM represents the establishment phase of dense DNA methylation and TGS (Fig. 1B and C).

After increased PolIV-RdDM methylation, TE loci may exit the RdDM cycle with a highly methylated status. The CHH methylation established by RdDM will be lost or reduced over time as only CG and CHG (symmetrical) DNA methylation are required to maintain silencing of the these TE loci (Fig. 1D). MET1, CMT3 and DDM1 are essential proteins for faithful propagation of symmetrical DNA methylation and TGS (Fig. 1D)., Long TEs and TEs located in heterochromatic regions of the genome comprise this category of deeply silenced TEs. Most Arabidopsis TEs produce 24 nt siRNAs. However, mutations in PolIV-RdDM component proteins result in transcriptional reactivation of only a small percentage of the TEs that are reactivated when symmetrical DNA methylation is lost, indicating that many TEs are still targets of PolIV-RdDM without depending on its function. Alternatively, short TEs that reside near genes are dependent on PolIV-RdDM for constant DNA methylation reinforcement to achieve TGS.

The example of a transcriptionally active TE entering the genome helps illustrate the robustness of the model described above. The incoming TE can be one of two exclusive categories. Either the TE enters the genome via horizontal transfer, in which case the sequence of the new TE might be completely unique compared with existing TEs (Fig. 1E), or through cross-hybridization with a closely related species and hence the TE shares homology with some existing silenced TEs in the genome (Fig. 1F). In the case of homologous TE invasion (Fig. 1F), 24 nt siRNAs matching the TE will recognize and quickly target the TE for silencing via PolIV-RdDM and this cycle will establish dense DNA methylation (Fig. 1C). RDR6-RdDM may also target this incoming active TE, but the cell can more quickly or easily rely on the existing homologous 24 nt siRNAs that act as a cellular surveillance to silence any matching TEs. In the case of the horizontal transfer of a unique TE (Fig. 1E), the plant must rely on RDR6 to recognize the TE mRNA and initiate transcriptional gene silencing via RDR6-RdDM as described above (Fig. 1B). This process may take substantially more time than silencing the TE by existing homologous PolIV-dependent 24 nt siRNAs and therefore the new unique TE may experience generations of activity and proliferation before it can be tamed by RDR6-RdDM and converted into a 24 nt siRNA generating locus that protects the genome from future invasions of similar TEs.