Small RNA Duplexes Function as Mobile Silencing Signals Between Plant Cells

See allHide authors and affiliations

Science  14 May 2010:
Vol. 328, Issue 5980, pp. 912-916
DOI: 10.1126/science.1185880

siRNA Movement in Plant Tissues

Long-distance movement of RNA interference (RNAi)–derived signals in plants plays an important role in development and in defense against viral attack. The nature of the signals that spread from cell to cell is not known, although evidence suggests that they are nucleic acids of some sort (see the Perspective by Martienssen). Molnar et al. (p. 872, published online 22 April) and Dunoyer et al. (p. 912, published online 22 April) now show that in Arabidopsis, both exogenous and endogenous small interfering RNAs (siRNAs), rather than their long double-stranded precursor RNAs, are the molecules that transfer information between plant cells. A viral protein that counters RNAi though sequestering siRNAs blocked spreading of a transgene RNAi silencing signal. Furthermore, siRNA-processing enzymes were required in the source, and not the recipient, cells for spreading, and bombardment of plants with double-stranded siRNAs directly showed siRNA spread between cells. Endogenous siRNAs also spread between tissues and were capable of directing DNA methylation of target sequences in distant tissues.


In the plant RNA interference (RNAi) pathway, 21-nucleotide duplexes of small interfering RNA (siRNA) are processed from longer double-stranded RNA precursors by the RNaseIII Dicer-like 4 (DCL4). Single-stranded siRNAs then guide Argonaute 1 (AGO1) to execute posttranscriptional silencing of complementary target RNAs. RNAi is not cell-autonomous in higher plants, but the nature of the mobile nucleic acid(s) signal remains unknown. Using cell-specific rescue of DCL4 function and cell-specific inhibition of RNAi movement, we genetically establish that exogenous and endogenous siRNAs, as opposed to their precursor molecules, act as mobile silencing signals between plant cells. We further demonstrate physical movement of mechanically delivered, labeled siRNA duplexes that functionally recapitulate transgenic RNAi spread. Cell-to-cell movement is unlikely to involve AGO1-bound siRNA single strands, but instead likely involves siRNA duplexes.

In plants, RNAi spreads over long distances through the vasculature and from cell to cell presumably via plasmodesmata, owing to mobile nucleic acid–based signals (1, 2). Plant cell-to-cell RNAi movement has defensive and developmental roles: The spread of viral-derived silencing signals probably immunizes surrounding naïve cells (3), whereas movement of silencing signals from endogenous TRANS-ACTING siRNA (TAS) (siRNA, small interfering RNA) loci might generate target gene–expression gradients allowing organ polarization (4, 5). Movement of small RNA is also suspected to account for epigenetic reprogramming between pollen nuclei (6). Both viruses and TAS loci produce long double-stranded RNA (dsRNA) precursors, subsequently converted into siRNAs by Dicer-like 4 (DCL4), one of four Arabidopsis Dicer-like proteins (3, 7). Although DCL4-dependent siRNAs are popular candidates as cell-to-cell RNAi signaling molecules (5, 8), a role for their precursors, including long dsRNA, cannot be excluded. Consequently, the nature of the mobile, silencing nucleic acid(s) remains unknown (9).

In the experimental SUC:SUL Arabidopsis system, a long-dsRNA–producing transgene is expressed under the phloem-companion cell-specific promoter, SUC2 (8). This triggers non–cell-autonomous RNAi of the ubiquitously expressed endogenous SULFUR (SUL) mRNA, generating a leaf chlorotic phenotype that expands 10 to 15 cells beyond the vasculature (Fig. 1A). The SUL dsRNA is processed into DCL4-dependent 21-nucleotide (nt) and DCL3-dependent 24-nt siRNAs, which are, respectively, mandatory and dispensable for the manifestation of SUL-silencing movement (Fig. 1B) (8). To uncover the SUL-silencing signal’s identity we used the viral silencing suppressor P19, which specifically sequesters 21–base pair (bp) siRNA duplexes, but not their long-dsRNA precursor (10). A SUC:P19 transgene was introduced into SUC:SUL plants. Highly expressing lines displayed no SUL-silencing movement, whereas movement remained unaltered in low-expressing lines (Fig. 1A). As expected, both high and low P19 lines accumulated similar amounts of each SUL siRNA species, as found in SUC:SUL reference plants (Fig. 1, A and B). Immunoprecipitation analyses using transgenic lines expressing different levels of epitope-tagged P19 (SUC:P19HA) revealed that SUL-silencing suppression was correlated with the extent to which 21-nt SUL siRNAs were sequestered by P19 (Fig. 1, C to E). Immunostaining in SUC:P19HA plants and immunofluorescence analyses carried out in independently generated SUC:SUL2/SUC:P19HA lines confirmed that P19HA accumulation is strictly confined within phloem-companion cells (Fig. 1F and fig. S1). The strict cell autonomy of P19 was further confirmed by analysis of independent transgenics in which the protein was driven under a mesophyll-specific promoter (fig. S1). We conclude that suppression of cell-to-cell SUL-silencing movement relies on the dose-dependent and cell-autonomous capacity of P19 to specifically sequester DCL4-dependent 21-bp SUL siRNAs in silencing-incipient cells. Similar conclusions were drawn using SUC:SUL plants expressing companion-cell–specific P21, a distinct silencing suppressor that, like P19, sequesters 21-nt siRNA (fig. S2) (11).

Fig. 1

Companion-cell–specific P19 expression suppresses cell-to-cell SUL-silencing movement in a dose-dependent, cell-autonomous manner. (A) Phenotypes of SUC:SUL plants in wild-type (WT) SUC:P19 high- (lines #4 and #10) or low-expressing (line #9) transgenics. (B) Northern analysis of P19 mRNA, SUL siRNAs, and miR159 in plants depicted in (A). SS, SUC:SUL; mvt, silencing movement observed; rRNA, ribsomal RNA. (C) Phenotypes of high– (#68) and low–SUC:P19HA-expressing (#55) SUC:SUL transgenics. (D) Western and Northern analysis of P19HA (top) and SUL siRNAs (bottom), respectively, in plants depicted in (C). (E) HA epitope–specific immunoprecipitation in SUC:SUL reference plants or in high– and low–SUC:P19HA-expressing SUC:SUL transgenics. Total RNA was extracted from immunoprecipitates (IPs), and low–molecular-weight RNA was subjected to Northern analysis (top). P19HA immunoprecipitation was confirmed by protein blot analysis (bottom). (F) P19HA immunolocalization (Alexa-Fluor488) in transverse sections of SUC:P19HA-expressing SUC:SUL transgenics. Plants expressing a SUC:GUSHA transgene provide a positive control for companion-cell immunolabeling retention. Cell walls were stained with fluorescent brightener 28. Xy, Xylem; Ph, Phloem; M, Mesophyll.

As a second approach to unravel the signal’s identity, we investigated cell-type–specific requirements for DCL4 in the SUL-silencing movement process. A requirement for DCL4 in incipient, dsRNA-producing companion-cells, but not in recipient cells, would support movement of an siRNA signal (9). A DCL4 genomic fragment was cloned under the SUC2 promoter (SUC:DCL4) and transformed into SUC:SUL plants carrying a dcl4-2 null mutation. Without a DCL4 transgene, dcl4-2 plants lack 21-nt SUL siRNAs and, accordingly, display no SUL-silencing movement (Fig. 2, A and B). Lack of DCL4 results in aberrant processing of endogenous TAS precursors into nonfunctional, 22-nt-long siRNAs by DCL2 (Fig. 2B). Consequently, leaves ectopically accumulate high levels of trans-acting siRNA (tasiRNA) targets, including AUXIN RESPONSE FACTOR 3 (ARF3) transcripts, which normally undergo TAS3 tasiRNA-dependent non–cell-autonomous RNAi over several leaf-primordia cell layers (Fig. 2C) (4, 5). DCL4 expression was detected in nearly all independently generated SUC:DCL4/SUC:SUL/dcl4 transgenic lines (fig. S3), and in those lines, 21-nt SUL siRNA accumulation was fully rescued (Fig. 2B). Companion-cell–restricted DCL4 expression also rescued full SUL-silencing movement (Fig. 2A). This result suggests that movement of SUL long dsRNA, if any, cannot account for the SUL-silencing phenotype. Furthermore, 21-nt-long tasiRNA production was also restored in leaves, as was down-regulation of ARF3 and other endogenous tasiRNA targets (Fig. 2, B and C). Consistent with this finding, leaf expression of Argonaute 7 (AGO7), a limiting factor in the TAS3 pathway, is restricted to the vasculature and cells near the abaxial surface (12). Collectively, these data indicate that 21-nt siRNAs account for cell-to-cell movement of exogenous, and possibly endogenous, DCL4-dependent RNAi.

Fig. 2

Companion-cell–specific DCL4 expression in SUC:SUL/dcl4 mutants restores 21-nt SUL siRNA and tasiRNA accumulation, cell-to-cell SUL-silencing movement, and tasiRNA target down-regulation. (A) SUL-silencing phenotype in WT, dcl4, and SUC:DCL4/dcl4 transgenic plants. (B) Northern analysis of SUL siRNAs, miR159, TAS1 siRNA255, and TAS3 tasiRNAs in WT, dcl4, and independent SUC:DCL4/dcl4 transgenics (#9, #11, #12). (C) Quantitative reverse transcription polymerase chain reaction analysis of endogenous TAS1 siRNA255 (At5g18040) and TAS3 tasiRNA (ARF3) target mRNA accumulation. ARF17 mRNA (miR160 target) was used as a nonaffected control. cDNA inputs were normalized to Actin2 mRNA; expression ratios are relative to levels in the SUC:SUL parental line. Data are displayed as averages ± SD (indicated by error bars) (three replicates).

SUL-silencing requires the specific loading of 21-nt SUL siRNA guide strands into AGO1 (8), 1 of 10 Arabidopsis AGO proteins that also effects posttranscriptional silencing with endogenous microRNAs (miRNAs) and tasiRNAs (13). Therefore, we investigated whether 21-nt siRNAs move between cells in association with AGO1. AGO1 immunoprecipitation experiments conducted in independent SUC:SUL lines expressing the SUC:P19HA or SUC:P19 transgenes showed that approximately half of the 21-nt SUL siRNA pool is sequestered away from AGO1 by phloem-specific P19 (Fig. 3A); this sequestration is sufficient to prevent SUL-silencing movement (Fig. 1, A and C). Similar observations were made in SUC:SUL plants expressing SUC:P21 or SUC:P21HA (fig. S2). As the other half of the companion-cell–derived 21-nt SUL siRNA remains loaded into AGO1 (Fig. 3A), it is unlikely that siRNAs move from cell to cell in an AGO1-bound form. Furthermore, AGO1-bound siRNA movement would entail movement of AGO1 itself. To address this issue, a SUC2-driven, functional epitope-tagged AGO1 allele (SUC:FLAG-AGO1) (14) was transformed into SUC:SUL plants carrying either the ago1-12 or ago1-27 hypomorphic mutations, both of which prevent SUL-silencing movement with minimal or no effect on SUL siRNA production (Fig. 3, B and D, and fig. S4) (8, 15). Of the five lines inspected, none recovered the SUL-silencing movement phenotype (Fig. 3B and fig. S4), despite FLAG-AGO1 being appropriately loaded with the 21-nt SUL siRNAs and with endogenous miRNAs (Fig. 3, C and D, and fig. S4). We conclude that AGO1 is required cell-autonomously for the execution of RNAi in incipient and recipient cells. Therefore, DCL4-dependent siRNAs are unlikely to move between cells bound to AGO1.

Fig. 3

Cell-autonomous requirement for AGO1 in SUL silencing. (A) Northern analysis of total RNA and IP fractions of AGO1-bound small RNA in control, SUC:P19-, or SUC:P19HA-expressing SUC:SUL plants (top). AGO1 immunoprecipitation was confirmed by protein blot analysis (bottom). (B) SUL-silencing in SUC:SUL WT, ago1-12, and SUC:FLAG-AGO1/ago1-12 trangenic plants. (C) Western analysis of FLAG-AGO1 in two independent SUC:FLAG-AGO1/ago1-12 transgenic lines. (D) FLAG-specific immunoprecipitation in SUC:SUL WT, ago1-12, and SUC:FLAG-AGO1/ago1-12 transgenic plants and Northern analysis of SUL siRNA and miR159 accumulation in total RNA or FLAG-AGO1 IP (top). Protein blot analysis confirmed FLAG-AGO1 immunoprecipitation (bottom).

Next, we performed experiments to directly monitor siRNA movement in plants. We used particle bombardment to mechanically deliver various RNAi trigger molecules into XD216 transgenic seedlings. XD216 contains a constitutively expressed GREEN FLUORESCENT PROTEIN (GFP) transgene in an sde1/rdr6 null mutant background (16). This mutation ensures that any silencing events monitored in XD216 are attributable to the bombarded material, as opposed to endogenously amplified and RDR6-dependent secondary silencing events (fig. S5) (16). Bombardment of in vitro transcribed and preannealed GFP-derived dsRNA consistently triggered GFP silencing: Four days post-bombardment (dpb), it was manifested as foci, 10 to 15 cells in diameter, that did not expand any further (Fig. 4A). The same observation was made by bombarding a mix of three chemically synthesized, 21-bp GFP siRNA duplexes (Fig. 4B), but not by bombarding siRNAs or dsRNA with GFP-unrelated sequences. Bombarding particles co-coated with a GUS-reporter gene plasmid confirmed that silencing foci resulted from bona fide GFP-silencing movement radiating from primarily delivered, individual cells or small groups of cells (Fig. 4, C and D). The GFP-silencing–movement process was recapitulated upon bombardment of individual siRNA duplexes (Fig. 4, E and F), but not with single-stranded guide or passenger strands (fig. S5). Moreover, those chemically synthesized siRNA duplexes were faithfully and functionally sorted into cognate AGO effector complexes in bombarded and, presumably, surrounding cells (fig. S5).

Fig. 4

Bombarded siRNAs trigger GFP silencing and move from cell to cell. (A and B) GFP-silencing foci on XD216 leaves triggered by GFP-derived dsRNA or a mix of three independent, chemically synthesized siRNAs duplexes. (C and D) An XD216 leaf cobombarded with GFP-targeting siRNA and GUS-expressing plasmid. GFP silencing is visible at 4 dpb (C). GUS staining was done at 4 dpb (D). (E and F) GFP-silencing foci on XD216 triggered by single GFP-targeting siRNA duplex 6768 or 6970. (G and H) Arabidopsis Col-0 leaf bombarded with ALEXA555-labeled siRNA 6768 at 1 hpb (G) and 20 hpb (H). (I to K) Same as in (G) and (H) at 1 hpb (I), 4 hpb (J), and 20 hpb (K) on turnip leaves. (L and M) Coincidence of ALEXA555-labeled siRNA 6768 (L) and GFP silencing (M) on XD216 at 4 dpb. (N and P) Bombarded Col-0 leaf at 1 hpb (N), 8 hpb (O), 20 hpb (P) showing that ALEXA555-labeled siRNA 6768 reaches the vascular system (v). (Q and R) Movement of ALEXA555-labeled siRNA 6768 within the vascular bundle at 1 hpb (Q) and 4 hpb (R). Scale bars, 500 μm for (G), (H), (N), (O), and (P); 100 μm for (I), (J), and (K); 200 μm for (Q) and (R).

To test if GFP-silencing foci in XD216 resulted from direct cell-to-cell movement of the bombarded siRNAs, the passenger-stand in the siRNA 6768 duplex was covalently labeled at its 3′ end with the fluorophore ALEXA555 (17). One hour pb (hpb), the fluorescence was typically concentrated in single cells, or small groups of cells (Fig. 4G), but by 4 hpb, it had radiated from the initially delivered area into the surrounding cells, a pattern unchanged at 20 hpb and beyond (Fig. 4, H to K). This pattern was unlikely to result from movement of free fluorophore or partially digested siRNAs, as the residual fluorescence was precisely superimposed over GFP-silencing foci in many areas of bombarded leaves at 4 dpb (Fig. 4, L and M, and fig. S5). The most straightforward interpretation of these results is that labeled siRNA 6768 had moved from the bombarded cells to the adjacent 10 to 15 cells and triggered RNAi. Bombarding the siRNA 6768 duplex with a 3′ end–labeled guide strand also triggered movement, but GFP-silencing efficacy was reduced, presumably because the ALEXA555 dye prevents optimal loading of guide strands into AGOs. Occasionally, labeled siRNAs reached the vascular system during the cell-to-cell movement process (Fig. 4, N to P). Moreover, singly bombarded phloem cells could clearly transmit siRNAs to adjacent cells within vascular bundles (Fig. 4, Q and R), suggesting that siRNAs may also act as phloem-transported signals for long-distance silencing in Arabidopsis (18, 19).

We have provided genetic evidence that siRNAs are necessary for mobile silencing, while the bombardment experiments show that they are also sufficient for this process. We conclude, therefore, that siRNAs act as silencing signals between plant cells and, possibly, over long distances (19). Although their exact mobile form (that is, protein-bound versus free molecules) awaits further characterization, duplexes, as opposed to single strands, are likely involved. Indeed, movement occurred with preannealed siRNA duplexes containing either a labeled passenger strand or a labeled guide strand. The former is rapidly turned over upon strand separation (20), whereas guide single strands are usually unstable unless loaded into AGOs (21); yet AGO1, the effector of mobile RNAi, functions cell-autonomously (Fig. 3, B to D). Lack of GFP-silencing foci upon bombardment of anti-GFP single strands also supports movement of siRNAs as duplexes (fig. S5), although single strands might be rapidly degraded in bombarded cells or might fail to incorporate into AGO1 in surrounding cells owing to prerequisite strand separation. Movement of 21-bp siRNA duplexes between cells and their inhibition by viral P19 agree with previous results obtained with the P19-producing Cymbidium ringspot virus (CymRSV): P19 was dispensable for CymRSV accumulation within vascular bundles. However, its lack prevented further invasion of the leaf lamina, which, although virus-free, exhibited nucleotide sequence-specific resistance to CymRSV owing to the onset and cell-to-cell movement of a mobile, virus-induced silencing signal (22). In light of the present results, this signal must be the DCL4-dependent 21-bp siRNAs produced from CymRSV-derived dsRNA.

Of the 21- and 24-bp siRNA species, we studied only the former in this experiment, because AGO1-dependent RNAi generates a measurable silencing movement phenotype (that is, lack of SUL or GFP). Yet, there is no reason to exclude mobility of DCL3-dependent 24-nt siRNAs, which mediate locus-specific chromatin modifications upon loading into AGO4 (12). Using a micrografting procedure, we have obtained evidence that Arabidopsis endogenous siRNAs of all size classes are mobile from cell to cell and over long distances (19). Meanwhile, the experimental system described here now provides a handle to dissect the cell biology of siRNA movement in plants.

Supporting Online Material

References and Notes

  1. Research in O.V.’s laboratory is funded by a prize from the Bettencourt Foundation for Life Science Research and a starting grant from the European Research Council “Frontiers of RNAi” ERC 210890. P.D. and G.S. are supported by a research grant from Agence National pour la Recherche (ANR-08-JCJC-0063-01), J.C.C. is supported by grants from the NIH (AI43288) and the NSF (MCB-0618433), and A.T. is supported by a fellowship from the Japanese Society for the Promotion of Science. We thank J. Mutterrer for help with image acquisition, R. Wagner’s team for plant care, and members of O.V.’s laboratory for critical reading of the manuscript.
View Abstract

Navigate This Article