Activity Range of Arabidopsis Small RNAs Derived from Different Biogenesis Pathways^,

Several classes of small RNAs are known in plants and accumulating evidence indicates that different classes of small RNAs may function either cell autonomously or noncell autonomously to regulate gene expression. Here, a simple visual assay used to compare the biological activity of small RNAs produced via different biogenesis pathways suggests that trans-acting small interfering RNAs (tasiRNAs) are mobile while confirming that microRNAs (miRNAs) are not.

Small RNAs have several functions in plant cells, with small interfering RNAs (siRNAs) involved in immunity against invading nucleic acids (RNA viruses and transposons) and in silencing of repetitive DNA. In contrast, miRNAs modulate physiological and developmental gene expression, a role also played by a limited number of siRNAs produced from endogenous genes. Transgene-born or viral-induced siRNAs can move from cell to cell. Current evidence suggests that miRNAs are not mobile, while it is unknown if endogenously encoded siRNAs such as tasiRNAs act noncell autonomously. The question of small RNA mobility is biologically critical since these molecules may act as signals.

During RNA silencing, Dicer ribonuclease enzymes produce 21- to 24-nucleotide RNAs from double-stranded RNA (dsRNA) substrates (Baulcombe, 2004; Brodersen and Voinnet, 2006; Mallory and Vaucheret, 2006). The small RNAs guide RNA-induced silencing complexes (RISCs) to suppress gene expression either transcriptionally or posttranscriptionally. In Arabidopsis (Arabidopsis thaliana), miRNAs are generated by DCL1-mediated cleavage of RNA polymerase II-transcribed endogenous precursor pre-miRNAs with distinct expression patterns reflecting their roles in developmental or physiological processes. In contrast, 21-nucleotide siRNAs are derived from DCL4-mediated cleavage of dsRNA precursors of various origins, including RNA viruses and transcribed inverted repeats used in RNAi experiments. One feature of siRNAs is their induction of a cell-to-cell movement of a sequence-specific silencing signal, likely comprised of the siRNAs themselves (Dunoyer et al., 2007). Local cell-to-cell movement of siRNA-derived silencing signals requires DCL4 and an RNA-dependent RNA polymerase (RdRP), RDR2 (Dunoyer et al., 2007). The local silencing signal typically travels five to 15 cells, with movement thought to be via plasmodesmata. In plants, miRNAs act to direct RISC-mediated cleavage of a target mRNA with nearly perfect complementarity into two fragments that are rapidly degraded in the cell (Tang et al., 2003). Due to their role in developmental processes and their molecular similarity to siRNAs, it has been proposed that miRNAs could act as mobile developmental signals (Juarez et al., 2004). However, studies using an miRNA-responsive reporter gene and the ectopic expression of miRNAs suggest cell autonomy (Parizotto et al., 2004; Alvarez et al., 2006).

While many siRNAs are generated against foreign nucleic acids or transcripts generated from repetitive elements and target the sequences from which they are derived, tasiRNAs can silence messages from loci other than those from which they are derived and can act in developmental and physiological programs (Brodersen and Voinnet, 2006; Nagasaki et al., 2007; Nogueira et al., 2007). tasiRNAs are produced from a noncoding RNA polymerase II-derived transcript in a phased manner from one or two initial miRNA target sites. The cleaved single-stranded RNA (ssRNA) is a template for an RdRP, RDR6, and cleavage of its resulting dsRNA in a phased manner by DCL4 yields the tasiRNAs that subsequently target transcripts from unrelated loci (Yoshikawa et al., 2005). A question that remains unanswered for tasiRNAs is whether their effects are autonomous or move from cell to cell.

We compared the phenotypic effects of small RNAs derived via different biogenesis pathways from transgenic synthetic precursors, each targeting the same Arabidopsis phytoene destaurase (PDS) transcripts, loss of which results in photobleaching of green tissues. Different small RNA precursors, miRNA, siRNA, and tasiRNA mimics, were generated as described in Supplemental Materials and Methods S1. The precursors were driven by three promoters with restricted expression patterns: AS1 (expressed in lateral organs), AP3 (expressed in sepal margins), and SUC2 (expressed in phloem companion cells). When driven by the AS1 promoter, both miR-PDS-I and siRNA-PDS caused photobleaching throughout the cotyledons, demonstrating both constructs are functional and can produce relatively equivalent phenotypes (Fig. 1, A and B). When driven with the AP3 promoter, miR-PDS-I and siRNA-PDS result in strikingly different phenotypes (Fig. 1, C and E). In AP3≫miR-PDS-I flowers, sepal margins are photobleached, whereas AP3≫siRNA-PDS flowers exhibited photobleaching throughout the sepals. In SUC2≫siRNA-PDS plants, photobleaching was observed spreading from primary and secondary leaf veins (Fig. 1F) as described previously (Dunoyer et al., 2007). Inclusion of an Op:GUS reporter demonstrated that the siRNA-PDS gene is transcribed only in the phloem and that an siRNA-PDS-induced silencing signal moves five to 15 cells (Fig. 1G). In contrast, no photobleaching was observed along either primary or secondary leaf vascular bundles in SUC2≫miR-PDS-I;GUS plants (Fig. 1H). We infer that all the transgenes are active in these plants based on the GUS staining pattern and photobleaching along sepal margins (due to the presence of an AP3:LhG4 transgene). That a dramatic phenotype can be induced by expressing two different miRNAs with the SUC2 promoter (see below) indicates that miRNAs are efficiently processed in the companion cells of the phloem.

Figure 1.

A and B, Photobleaching is observed throughout the cotyledons of Arabidopsis seedlings with the same AS1:LhG4 driver line with either an Op:siRNA-PDS or an Op:miR-PDS-I responder transgene. C to F, Inflorescences with the same AP3:LhG4 and Op:GUS transgenes, with additional transgenes in D (Op:miR-PDS-I), E (Op:siRNA-PDS), and F (Op:miR-PDS-I+II). Inset shows GUS staining in margins of sepals driven by the AP3 promoter. G and H, Arabidopsis seedlings with the same SUC2:LhG4, AP3:LhG4, and Op:GUS transgenes, with additional transgenes in G (Op:siRNA-PDS) and H (Op:miR-PDS-I). Seedlings are in left images, with GUS staining patterns prior to removal of chlorophyll in right (G) or center (H) images, and following removal of chlorophyll (right in H). Photobleaching in G occurs in five to 15 cells from the vasculature (inset). I, Seedling with SUC2:miR-PDS-I+miR-PDS-II transgene; photobleaching occurs in five to 15 cells from the vasculature (inset).

To stimulate formation of tasiRNAs, we mimicked the two-hit mode for tasiRNA biogenesis whereby the intervening ssRNA of a mRNA cleaved (or bound) at two sites by miRNA action becomes a template for RDR6, or another RdRP, and siRNA biogenesis (Axtell et al., 2006). We engineered a second miRNA, miR-PDS-II, targeting the PDS mRNA 170 bases 3′ to the cleavage site of miR-PDS-I. Expression of miR-PDS-II driven by the AP3 and SUC2 promoters resulted in phenotypes indistinguishable from those conferred by miR-PDS-I. However, when both miR-PDS-I and miR-PDS-II are coexpressed with the same promoter producing simultaneous, dual targeting by both miRNAs (SUC2:miR-PDS-I;miR-PDS-II and AP3≫miR-PDS-I;miR-PDS-II), a phenotype similar to that of SUC2/AP3≫siRNA-PDS is observed, implying the generation of a short-range silencing signal (Fig. 1, F and I). While localized movement of silencing signals was observed with both the inverted-repeat transgene and the dual miRNA cleavage approach, systemic gene silencing characteristic of virus-induced gene silencing was not seen (Baulcombe, 2004).

The primary difference distinguishing the dsRNA precursors produced by transcription of miR-PDS versus siRNA-PDS and the presumed dsRNA ta-siRNA intermediate is the lack of perfect complementarity in the folded pre-miRNA (Fig. 2). This feature could preclude processing by DCL enzymes other than DCL1. Thus one scenario for their different biological activities is that different small RNAs are funneled through different biochemical pathways that are canalized by interactions between specific Dicer enzymes and distinct RISC complexes possibly restricted by subcellular location of processing reactions. For example, DCL1-generated miRNAs would be introduced into RISC complexes whose accessory proteins confer cell autonomy to miRNAs, while DCL4-processed siRNAs would be handed to RISC complexes whose accessory proteins allow for the transmission of mobile silencing signals. Reduction of induction of mobile silencing signals when dsRNA substrates normally processed by DCL4 are shunted through DCL1 (in a dcl2 dcl3 dcl4 background) is consistent with this hypothesis (Dunoyer et al., 2007).

Figure 2.

Cartoons presenting basic steps and activity range of smRNA biogenesis from different sources, including a pre-miRNA (A), an inverted-repeat, hpRNAi transgene (B), a TAS gene mimic induced by artificial miRNA-mediated double cleavage (C), and a TAS3-like tasiRNA (D). A, miRNA biogenesis including miR-PDS-I or miR-PDS-II from our study. The miRNA-encoding gene yields a pri-miRNA whose fold back leads to the formation of a hairpin structure with internal loops and is processed into a pre-miRNA and subsequently into a mature miRNA (Reinhart et al., 2002). The pri-miRNA hairpin is recognized by the RNase III enzyme DCLI and accessory proteins (Park et al., 2002; Papp et al., 2003; Kurihara and Watanabe, 2004; Qi et al., 2005). The miRNA is loaded into AGO1-containing RISC and provides a sequence-specific guide for the RISC to the homologous mRNA leading to mRNA cleavage and degredation, or to translation inhibition. The imperfect homology of the hairpin may preclude processing by DCL enzymes other than DCL1. This DCL1 and AGO1-RISC exclusivity in miRNA biogenesis are potential features that maintain the cell autonomy of miRNA action. In support of this hypothesis, while dsRNA substrates with extensive complementarity (such as those produced by an inverted-repeat, hpRNAi transgene or by an RdRP from a ssRNA template) can be processed by DCL1 in the absence of other Dicer activity (in a dcl234 mutant background; Xie et al., 2004; Gasciolli et al., 2005; Henderson et al., 2006; Dunoyer et al., 2007), introduction of a SUC2:miR-PDS-I transgene into dcl1-7 and dcl1-9 mutant backgrounds did not induce a mobile silencing signal, suggesting that other DCL enzymes may be unable to process miRNA precursors, although the partial loss-of-function nature of these alleles tempers this interpretation (Schauer et al., 2002). B, Transcription of the inverted-repeat siRNA-PDS hpRNA transgene results in a long dsRNA of perfect complementarity that can be a substrate for different DCL enzymes and cleaved by DCL4 to release active siRNAs whose accessory proteins promote mobile silencing (Himber et al., 2003; Dunoyer et al., 2005, 2007; Adenot et al., 2006; Deleris et al., 2006). C, The tandem miRNAs miR-PDS-I;miR-PDS-II target the PDS gene in two different locations. This double cleavage mimics the circumstances by which tasiRNAs are generated from the miR390-flanked TAS3 genes illustrated in D (Peragine et al., 2004; Vazquez et al., 2004; Allen et al., 2005; Borsani et al., 2005; Xie et al., 2005; Adenot et al., 2006; Axtell et al., 2006; Howell et al., 2007; Moissiard et al., 2007). The absence of polyA and 5′ CAP, the presence of the two RISC complexes, or combination of the two cues RdRP to synthesize a complementary RNA strand for the intervening, which is processed by DCL4 into siRNAs whose accessory proteins promote mobile silencing.

Several recent studies have documented the presence of miRNAs in the phloem of several plant species and the translocation of one miRNA, miR399, through a graft junction (Yoo et al., 2004; Lough and Lucas, 2006; Omid et al., 2007; Buhtz et al., 2008; Pant et al., 2008). Expression of either miR-PDS-I or miR-PDS-II, based on two endogenous miRNA backbones (miR164b and miR167a), in the companion cells did not result in detectable movement from the phloem into the surrounding mesophyll. Thus, while the presence of miRNAs in the phloem stream may be explained by their transcription in the phloem, the results presented here indicate that miRNAs transcribed in the phloem are unlikely to be mobile into surrounding cells, unless there is regulated movement of specific miRNAs.

Our results for the miR-PDS genes add to the evidence that miRNAs act autonomously and are therefore unlikely to act as signaling molecules (Parizotto et al., 2004; Alvarez et al., 2006). In contrast, miRNA-induced double cleavage mimicking endogenous tasiRNA biogenesis produced a mobile silencing signal, similar in action to that of siRNAs derived from transcribed hairpins, suggesting endogenous tasiRNAs are mobile and can act as local signaling molecules (Nogueira et al., 2007). Since for both miRNAs and tasiRNAs plants likely coopted an ancient molecular machinery that evolved as a genomic or cellular immune system to regulate endogenous gene expression, it is perhaps the utility of short-range gene silencing signals in particular plant developmental contexts that led to the evolution of tasiRNAs as alternative mediators of endogenous gene silencing to cell-autonomous miRNAs. However, that comparatively few TAS loci exist may reflect that the diffuse and complex outcomes from even a short-range mobile signal mean their value is restricted, while autonomous miRNAs are broadly exploited to clarify gene expression domains. Even so, the ancient evolutionary history of the TAS3 gene tasiRNAs and their targets, which predates the divergence between mosses and vascular plants (Axtell et al., 2007), suggests an important niche role for tasi-mediated regulation and the short-range signal that it produces.

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Materials and Methods S1. Construction of small RNA precursor genes.

Note Added in Proof

Three recent papers (Takeda A, Iwasaki S, Watanabe T, Utsumi M, Watanabe Y [2008] The mechanism selecting the guide strand from small RNA duplexes is different among Argonaute proteins. Plant Cell Physiol [in press]; Mi S, Cai T, Hu Y, Hodges E, Ni F, Wu L, Li S, Zhou H, Long C, Chen S, et al [2008] Sorting of small RNAs into Arabidopsis Argonaute complexes is directed by the 5′ terminal nucleotide. Cell [in press]; Montgomery TA, Howell MD, Cuperus JT, Li D, Hansen JE, Alexander AL, Chapman EJ, Fahlgren N, Allen E, Carrington JC [2008] Specificity of ARGONAUTE7-miR390 interaction and dual functionality in TAS3 trans-acting siRNA formation. Cell [in press]) provide evidence that different AGO (argonaute) proteins preferentially recruit specific small RNAs based on the 5′ terminal nucleotide identity or a specific small RNA (miR390) in the case of AGO7, thus providing a glimpse of how small RNAs are sorted to specific AGO complexes.