Possible new RNA intermediate in RNA silencing

SGS3 is essential for antiviral silencing and the biogenesis of several classes of siRNAs in plants, but until now no biochemical function has been ascribed to it. Both SGS3 and a viral suppressor of RNA silencing have now been shown to selectively bind 5′ overhang–containing dsRNA, implicating this RNA as a new intermediate in the RNA silencing pathway.

RNA silencing controls gene expression and antiviral defense by producing 21- to 24-nucleotide small interfering RNAs (siRNAs) from a double-stranded RNA (dsRNA) precursor. siRNAs loaded in an Argonaute protein then target complementary single-stranded RNA (ssRNA) to guide specific gene silencing. Fukunaga and Doudna now show that SGS3 is an RNA-binding protein that specially favors dsRNA with a single-stranded 5′ overhang¹. Their findings suggest that 5′ overhang–containing dsRNA is a new, previously uncharacterized RNA intermediate in the RNA silencing pathway.

siRNA pathways may generate 5′ overhang–containing dsRNA molecules. For example, slicing of an mRNA or a ssRNA guided by an Argonaute loaded with a microRNA (miRNA) or siRNA will produce 5′ and 3′ cleavage products (Fig. 1). Because the small RNA guide is not cleaved during slicing and may remain bound to the cleavage products, the 5′ cleavage product is predicted to contain a duplex region with 5′ overhangs at both ends. In addition, mRNA transcripts from convergent transcription of either the natural antisense genes or the geminiviral circular DNA genome may form 5′ overhang–containing dsRNA molecules either with or without siRNA-guided cleavage because of their complementary 3′-terminal regions (Fig. 1).

Figure 1: A model for SGS3-mediated generation of siRNAs.

AGO, Argonaute protein; IR, intergenic region of the geminiviral circular ssDNA genome.

SGS3 was initially identified, together with RNA-directed RNA polymerase 6 (RDR6), as an essential component of RNA silencing induced by a sense transgene² in producing mRNA-sense ssRNA. SGS3 and RDR6 are also required for the biogenesis of trans-acting and natural antisense transcript–derived siRNAs (ta-siRNAs and nat-siRNAs)³. Extensive genetic evidence suggests that SGS3 facilitates the conversion of target ssRNA into dsRNA by RDR6, providing a new source for the production of siRNAs³. The biochemical activities of RDR6 have been characterized recently⁴, but the molecular basis for the biochemical function of SGS3 remained unknown³.

Through a number of in vitro RNA-binding assays, Fukunaga and Doudna demonstrate that SGS3 is a dsRNA-binding protein that preferentially recognizes dsRNA substrates containing a 5′ single-stranded overhang at one or both ends¹. The high-affinity binding substrates of SGS3 include RNAs that contain a duplex region of 14–35 base pairs (bp) plus a 5′ single-stranded overhang of 2–12 nucleotides (nt) in length. These dsRNAs can be generated as a byproduct of siRNA cleavage of ssRNA during slicing (Fig. 1). The presence of a 2′-O-methyl modification on the 3′-terminal ribose found on the 3′ end of plant miRNAs and siRNAs, as well as animal endogenous siRNAs and PIWI-interacting RNAs, does not perturb the binding. In contrast, SGS3 does not bind, or binds poorly, to blunt-end dsRNAs and dsRNAs with 3′ overhangs, such as a typical siRNA (19-bp duplex with 2-nt 3′ overhangs), which are generated by Dicer cleavage of dsRNA. Addition of a 3′ overhang to the other end of a high-affinity substrate seems to inhibit SGS3 binding. SGS3 also does not bind ssRNAs, with or without modifications such as a 5′ mono- or triphosphate and 5′ cap. Therefore, it appears that SGS3 recognizes a unique structure at the junction between the ssRNA and dsRNA in these high-affinity substrates (Fig. 1).

The SGS3 protein encodes a zinc-finger domain, three coiled-coil domains and an XS domain specific to plants. From a mutational analysis, the authors found that whereas the C-terminal XS and coiled-coil domains are essential for dsRNA binding, the zinc-finger domain at the N terminus may be required for the substrate specificity of SGS3.

The presence of the SGS3 RNA substrate as an intermediate in RNA silencing pathways is further supported by a similar substrate specificity seen for the geminiviral V2 protein¹. V2 is a known viral suppressor of RNA silencing (VSR) that is encoded by many plant and animal viruses and is required for virus infection^5,6. Specific recognition and binding of 5′ overhang–containing dsRNAs by SGS3 may result in the stabilization of natural antisense transcripts and/or in RDR6 recruitment and conversion of the 5′ single-stranded overhang region in a primer-dependent manner into new dsRNA, to be processed by Dicer into transgene secondary siRNAs, ta-siRNAs, nat-siRNAs and virus-derived siRNAs (Fig. 1). However, V2 binding may sequester this RNA intermediate from binding to SGS3, and prevent amplification of siRNAs, because V2 lacks similarities with SGS3 in that it does not posses either an XS or a coiled-coil domain⁵. Thus, this work also suggests that the new RNA intermediate is a target of VSRs, which were already known to inhibit dicing and siRNA loading by binding and sequestering the long dsRNA and siRNA, respectively⁶. However, V2 may also suppress RNA silencing by direct protein-protein interaction with SGS3, which was demonstrated previously by yeast two-hybrid and FRET microscopy experiments⁵ but was not supported by in vitro protein-binding experiments in the current study¹.

The RDR6 recruitment model presented in Figure 1 is consistent with the recent demonstration of an in vivo interaction between SGS3 and RDR6 (ref. 7). It explains the SGS3- and RDR6-dependent production of transgene secondary siRNAs and ta-siRNAs spread into the 5′ region from the cleavage site, such as TAS3-derived ta-siRNAs. However, transgene secondary siRNAs, ta-siRNAs and nat-siRNAs also spread into the 3′ region from the cleavage site³. Conversion of the 3′ cleavage products into dsRNA is predicted to depend on the primer-independent polymerase activity of RDR6 (ref. 4), which may occur only in the absence of SGS3 given that the presence of the 3′ overhang on the 3′ cleavage products would be inhibitory to SGS3 binding.

The study by Fukunaga and Doudna provides new insights into the biochemical role of SGS3 and points out the exciting possibility that RNA silencing may involve a new RNA intermediate. Future work will be necessary to determine whether 5′ overhang–containing dsRNA is indeed produced in the SGS3-dependent biogenesis of siRNAs.