Inhibition of in vivo  Slicer activity of Argonaute protein 1 by the viral 2b protein independent of its dsRNA‐binding function

Summary

The 2b protein of C ucumber mosaic virus (CMV) has several unique properties, such as targeting to the nucleolus and interaction with both Argonautes (AGOs) and short and long double‐stranded RNA (dsRNA). We have recently uncoupled the domain requirements for dsRNA binding and nucleolar targeting from the physical interactions with AGO proteins, and have found that the direct 2b–AGO interaction is sufficient to inhibit the in vitro  AGO1 Slicer function independent of the other biochemical properties of 2b. Because the AGO binding activity of 2b is not required for its suppressor function in vivo, this raises the question of whether in vivo 2b–AGO interaction is possible to inhibit the in vivo  AGO Slicer function. In this study, by taking advantage of a technology for the production of artificial trans‐acting small interfering RNA (tasiRNA), a process uniquely associated with AGO1‐mediated in vivo  Slicer activity, we demonstrated that the expression of the 2b protein in planta interfered with the production of tasiRNA. Through further detailed analysis with deletion mutants of 2b proteins, we found that the inhibition of in vivo  AGO1 Slicer function required the nucleolar localization signal (NoLS), in addition to the AGO‐binding domain, of the 2b protein. Our finding demonstrates that in vivo 2b–AGO1 interaction is sufficient to inhibit AGO1 Slicer function independent of the dsRNA‐binding activity of the 2b protein.

Introduction

RNA silencing is a conserved pathway that results in the blockage of gene expression in eukaryotic organisms. In plants, post‐transcriptional gene silencing (PTGS) occurs in the cytoplasm and utilizes small RNAs, including small interfering RNAs and microRNAs (siRNAs and miRNAs), which specifically recognize and base pair with targeted RNA sequences (Baulcombe, 2004, 2005, Voinnet, 2008). Small RNAs are processed from long double‐stranded RNAs (dsRNAs) or hairpin precursor RNAs by RNase III‐type enzymes called Dicer‐like ribonucleases (DCL) proteins (e.g. DCL1 and DCL2–4 which generate miRNA and siRNA, respectively). Small RNAs bind to the Argonaute (AGO)‐containing RNA‐induced silencing complex (RISC) which cleaves targeted RNA. AGO proteins are often named Slicer proteins because they cleave target transcripts at the duplex formed with the guide siRNAs or miRNAs (Baumberger and Baulcombe, 2005; Tolia and Joshua‐Tor, 2007; Vaucheret, 2008). In plants, RNA silencing involves an amplification process that requires RNA‐directed RNA polymerases (RDRs) for secondary siRNA production for persistent silencing (Baulcombe, 2007). The biogenesis of trans‐acting siRNAs (tasiRNAs), one type of plant endogenous siRNA, represents a case of an RDR‐dependent secondary siRNA‐producing pathway. For example, tasiRNAs formed from primary TAS1 transcript are initially targeted and sliced by the AGO1–miR173 complex at a 5′ proximal site. The cleaved RNA products are then transcribed by RDR6, leading to dsRNAs that are sequentially processed by DCL4 to yield phased tasiRNAs (Gasciolli et al., 2005; Montgomery et al., 2008a, b).

In addition to the regulation of development, RNA silencing also functions as an antiviral mechanism in plants and invertebrates. To invade their host effectively, viruses have evolved a vast array of proteins, called viral suppressors of RNA silencing (VSR), which target different stages of the silencing process (Agius et al., 2012; Ding and Voinnet, 2007). More than 35 VSR families have been identified so far. The sequestration of siRNAs by siRNA‐binding suppressors is a common way to inhibit RISC assembly (Csorba et al., 2009; Ding and Voinnet, 2007). Other VSRs inhibit RNA silencing through protein–protein interaction. Polerovirus P0 protein has been suggested to target the PAZ domain of AGO1 to direct its degradation (Baumberger et al., 2007; Pazhouhandeh et al., 2006). Turnip crinkle virus P38 and Sweet potato mild mottle virus P1 proteins have been suggested to bind AGO1 through the WG/GW motifs, and inhibit both existing and de novo‐formed AGO1‐containing RISCs (Azevedo et al., 2010; Giner et al., 2010). The 111‐residue 2b protein encoded by Cucumber mosaic virus (CMV) is unique among the known plant and animal VSRs because it interacts directly with both siRNA and AGO proteins and is targeted to the nucleolus (Gonzalez et al., 2010; Goto et al., 2007; Hamera et al., 2012; Lucy et al., 2000; Zhang et al., 2006). Recently, we have identified regions of the 2b protein that determine these biochemical properties and their contributions to suppression activity (Duan et al., 2012). We identified a functional nucleolar localization signal (NoLS), which includes two closely spaced nuclear localization signals (NLS1 and NLS2), located within the 61‐amino‐acid N‐terminal dsRNA‐binding domain (dsRBD) (Fig. 1A), which exhibits high affinity for short (siRNAs) and long dsRNA. Physical interaction of 2b with AGOs requires an essential 33‐residue region C terminal to dsRBD (Fig. 1A), and such an interaction is sufficient to inhibit in vitro AGO1 Slicer activity independent of its dsRNA‐binding activities (Duan et al., 2012). However, this AGO binding or suppression of AGO Slicer activity in vitro is dispensable for the suppression of RNA silencing by the 2b protein (Duan et al., 2012). Taking into account that all assays of 2b protein inhibition of AGO Slicer activities were examined in vitro (Duan et al., 2012; Hamera et al., 2012; Zhang et al., 2006), this raises the question of whether or not in vivo 2b–AGO interaction affects AGO Slicer function, as AGO binding is not required for the suppression activity of 2b in vivo. To address this question, in this study, we analysed, with a series of deletion mutants of 2b proteins using a technology associated with AGO1‐mediated slicing in planta, and provided evidence that the in vivo 2b–AGO1 interaction endows the 2b protein with the ability to inhibit AGO1 Slicer function independent of its dsRNA‐binding activity.

Figure 1.

Inhibition of trans‐acting small interfering RNA (tasiRNA) production by 2b and its derivative mutants. (A) Amino acid sequence and functional domains of 2b protein. Two arginine‐rich putative nuclear localization signals (NLSs) (blue letters), double‐stranded RNA (dsRNA)‐binding domain (dsRBD) and essential Argonaute (AGO)‐binding domain are labelled. The C‐terminal deletion mutant, 2b(1‐61), and two N‐terminal deletion mutants, 2b(38‐111) and 2b(13‐111), and three NLS substitution mutants (blue letters to red letters), 2bN1m, 2bN2m and 2bN1,2m, are shown. (B, C) Effects of 2b, NLSs and deletion mutants on tasiRNA synthesis. 2b, 2b mutant construct (as indicated on top of gels), 35S‐P19 and control vector (V) were co‐infiltrated with 35S‐MIR173 and 35S‐TAS1C constructs into N icotiana benthamiana (B) or RDR6i (C) leaves. The expression of 2b, the 2b mutants, TAS1C mRNA, tasiRNAs and miR173 was detected as described in Fig. 2B.

Results and Discussion

To evaluate the in vivo suppression activity of the 2b protein, we took advantage of a technology for the production of artificial tasiRNAs in Nicotiana benthamiana. The processes of generation of tasiRNAs include miR173‐guided AGO1‐dependent cleavage of TAS1C‐derived transcripts, which are then synthesized to dsRNA by RDR6 and subsequently diced by DCL4 (Fig. 2A) (Montgomery et al., 2008b; Ying et al., 2010). In this assay, 35S‐MIR173, which produces active miR173, and 35S‐TAS1C were co‐infiltrated into N. benthamiana leaves, so that TAS1C transcripts from 35S‐TAS1C were targets of AGO1‐dependent cleavage by miR173. As expected, the production of tasiRNAs was detected in the leaves co‐infiltrated with 35S‐TAS1C and 35S‐MIR173 (Fig. 2B, lane 3), but not in the leaves infiltrated solely with 35S‐TAS1C (Fig. 2B, lane 2). Also, as expected, the production of tasiRNAs was undetectable following the co‐infiltration of 35S‐MIR173 and 35S‐TAS1C in RDR6i plants (Fig. 2B, lane 5), in which Nb‐RDR6 was silenced (Schwach et al., 2005; Ying et al., 2010). RDR6 silencing in RDR6i plants stabilized the 5′ cleavage products of TAS1C transcripts by miR173, which were detected as a discrete band (Fig. 2B, lane 5), rather than as a smear in wild‐type plants (Fig. 2B, lane 3). 5′ Rapid amplification of cDNA ends (RACE) products from the co‐infiltration in RDR6i plants were cloned and sequenced, confirming that miR173‐mediated cleavage of the TAS1C transcripts was at the complementary region of the miR173/TAS1C pairing area (Fig. 2C).

Figure 2.

Detection of inhibitory effects of 2b on Argonaute protein 1 (AGO1)‐mediated cleavage of miR173‐targeted TAS1C transcripts and production of RNA‐directed RNA polymerase 6 (RDR6)‐mediated trans‐acting small interfering RNA (tasiRNA) in vivo. (A) A sketch map of miR173–AGO1‐mediated cleavage and RDR6‐dependent tasiRNA synthesis. DCL4, Dicer‐like ribonuclease 4; RISC, RNA‐induced silencing complex. (B) 2b protein‐inhibited production of tasiRNA. 35S‐2b and control vector (V) were co‐infiltrated with 35S‐MIR173 and 35S‐TAS1C constructs into wild‐type N icotiana benthamiana (Nb) and RDR6i leaves. The expression of 2b, TAS1C mRNA and 5′ cleavage products was detected using 2b‐ and TAS1C 5′‐specific DNA probes (labelled in A). miR173 and tasiRNA accumulation in infiltrated leaves was detected with ³² P‐labelled specific DNA oligonucleotide probes. Methylene blue‐stained rRNA is shown as the mRNA loading control. (C) Analysis of miR173‐mediated cleavage of the TAS1C transcript. The arrows indicate the cleavage sites of the TAS1C RNA detected by 5′ rapid amplification of cDNA ends (5′ RACE). The proportion of cloned 5′ RACE products corresponding to cleavage at a site is shown above the arrows. (D) 2b protein did not reverse double‐stranded RNA (dsRNA)‐induced RDR6 silencing. RNA gel analysis of samples collected from Nb and RDR6i plants infiltrated with vector (V) or 35S‐2b. ³² P‐labelled RDR6 DNA probes were used. The silencing trigger (RDR6 RNAi transcript) detected in leaves infiltrated with 35S‐2b is indicated.

We then examined whether the expression of the 2b protein would interfere with the production of tasiRNA in this system. On co‐infiltration of 35S‐2b/35S‐TAS1C/35S‐MIR173, we found that the expression of 2b inhibited the production of tasiRNAs and resulted in an increased accumulation of TAS1C transcript in the infiltrated leaves (Fig. 2B, lane 4). These results indicated that 2b suppressed the production of tasiRNAs in the infiltrated leaves. Notably, the 5′ cleavage products of TAS1C transcripts were undetectable in the leaves of RDR6i plants that were co‐infiltrated with 35S‐2b (Fig. 2B, lane 6). To determine whether or not the failure to detect the 5′ cleavage products of TAS1C transcripts was caused by the expression of 2b, which reversed the silencing of RDR6 mRNA in RDR6i plants, the accumulation of RDR6 mRNA was examined in 35S‐2b infiltrated leaves. Silencing of RDR6 mRNA in RDR6i plants was not reversed by the expression of 2b, although the accumulation of the silencing trigger (RDR6i transcript) was detected in the 35S‐2b infiltrated leaves (Fig. 2D, lane 4). These results were consistent with a previous report that CMV 2b did not reverse silencing in silenced tissue (Brigneti et al., 1998).

The processes for the synthesis of tasiRNAs, for example AGO1‐mediated slicing of the TAS1C transcripts and RDR6‐mediated synthesis of dsRNAs, might be affected by the 2b protein. To separate the two inhibition processes, we first used 35S‐2b(1‐61), which carries a deletion mutant of 2b that retains the N‐terminal 61 amino acids of dsRBD, but lacks the AGO‐binding domain (Duan et al., 2012), to examine the possibility of whether the 2b dsRNA‐binding activity was sufficient for the inhibition of the production of tasiRNAs in this system. We co‐infiltrated 35S‐TAS1C/35S‐MIR173 with 35S‐2b(1‐61). Co‐infiltration with empty vector or with the tombusviral suppressor P19, which can bind perfectly matched double‐stranded 21‐nucleotide siRNA duplex, but not imperfectly matched miRNA/miRNA* duplexes or long dsRNA (Duan et al., 2012), was used in parallel assays. A slight effect on tasiRNA production was observed in the co‐infiltration with 35S‐P19, relative to co‐infiltration with the vector control (Fig. 3, lanes 1 and 3); however, the 5′ cleavage products of TAS1C transcripts were detected in the leaves of RDR6i plants co‐infiltrated with 35S‐P19 (Fig. 3, lane 5). This result indicated that P19 did not inhibit the process of miR173–AGO1‐mediated cleavage of TAS1C transcripts, a result consistent with the observation that P19 is incapable of binding to single‐stranded miRNA or miRNA/miRNA* duplexes (Duan et al., 2012). When co‐infiltrated with 35S‐2b(1‐61), inhibition of tasiRNA production was observed, which was similar to the co‐infiltration of 35S‐2b in the infiltrated leaves of wild‐type N. benthamiana (Fig. 3, lanes 4 and 2). However, the 5′ cleavage products of TAS1C transcripts were detected in the leaves of RDR6i plants co‐infiltrated with 35S‐2b(1‐61), but not with 35S‐2b (Fig. 3, lanes 5–7). This result indicated that 2b(1‐61) did not inhibit the process of miR173–AGO1‐mediated cleavage of TAS1C transcripts, but could bind the dsRNA derived from the cleaved products to suppress the production of tasiRNAs. This is consistent with the observation that 2b(1‐61) is unable to physically interact with AGO proteins in the absence of the AGO‐binding domain (Duan et al., 2012). Taking into account that no 5′ cleavage products of TAS1C transcripts were detected by the expression of the full‐length 2b protein in RDR6i plants (Fig. 2, lane 6 and Fig. 3 lane 6), our results indicated that 2b suppressed the production of tasiRNAs in the infiltrated leaves, including substantial inhibition of the AGO1‐mediated slicing of the TAS1C transcripts, in addition to its ability to bind to long dsRNA.

Figure 3.

Inhibition of trans‐acting small interfering RNA (tasiRNA) production by 2b, 2b(1‐61) and P19. 35S‐2b, 35S‐2b(1‐61), 35S‐P19 and control vector (V) were co‐infiltrated with 35S‐MIR173 and 35S‐TAS1C constructs into N icotiana benthamiana (Nb) or RDR6i leaves. The expression of 2b and 2b(1‐61) RNA (2b/mut), TAS1C mRNA and 5′ cleavage product was detected as described in Fig. 2B.

The finding that the inhibition of tasiRNA production by the 2b protein included the prevention of miR173‐guided AGO1‐mediated cleavage of TAS1C transcripts allowed us to examine whether 2b was capable of inhibiting in vivo AGO1 Slicer activity in the absence of the long dsRNA‐binding function. We used five 2b mutants that lacked long dsRNA‐binding activity, but retained the ability to physically interact with AGOs and inhibit AGO1 Slicer activity in vitro (Duan et al., 2012) for the co‐infiltration assay. These 2b mutants included: 2b(13‐111), in which the N‐terminal 13 amino acids were deleted, but the intact NoLS was retained; 2b(38‐111), in which the N‐terminal 38 amino acids, including the intact NoLS, were deleted (Fig. 1A); and 2bN1m, 2bN2m and 2bN1,2m, in which NLS1, NLS2 and both NLSs in the NoLS region were mutated, respectively (Fig. 1A). In a previous study, we noted that the accumulation levels of these 2b mutants were lower than that of the wild‐type 2b in the infiltration assay; hence, co‐infiltration with P19 was used to restore their accumulation (Duan et al., 2012). The above co‐infiltration assays showed that P19 had little effect on miR173‐guided AGO1‐mediated cleavage of its precursor TAS1C transcripts (Fig. 3, lane 5). Therefore, co‐infiltration with P19 was used to restore the accumulation of these 2b mutants in this assay (Fig. 1B). The inhibition activity of mutant 2b(13‐111), which retains the intact NoLS (Fig. 1A), was very effective compared with the wild‐type 2b (Fig. 1B, lanes 3 and 2). Partial inhibition activity was observed in leaves co‐infiltrated with 35S‐2bN2m, which contains NLS1 (Fig. 1B, lane 6), whereas little effect on tasiRNA production was observed in leaves co‐infiltrated with 35S‐2b(38‐111), 35S‐2bN1m or 35S‐2bN1,2m (Fig. 1B, lanes 4, 5 and 7). These results indicated that not all of the 2b mutants were able to inhibit AGO1 Slicer activity in vivo, although they all retained the essential AGO‐binding domain and were capable of suppressing AGO1 Slicer function in vitro (Duan et al., 2012).

To further confirm that efficient inhibition of tasiRNA production by 2b(13‐111) resulted from the inhibition of AGO1‐mediated cleavage of TAS1C transcripts, we co‐infiltrated 35S‐TAS1C/35S‐MIR173/35S‐2b(13‐111) with 35S‐P19 in RDR6i plants. Co‐infiltration with 35S‐2b and 35S‐2b(1‐61) in the presence of 35S‐P19 was also carried out in parallel assays. Although the 5′ cleavage products of TAS1C transcripts were detected in RDR6i leaves co‐infiltrated with 35S‐2b(1‐61) and/or 35S‐P19 (Fig. 1C, lanes 3 and 2), no 5′ cleavage products were detected with the expression of 2b and 2b(13‐111) in RDR6i plants (Fig. 1C, lanes 4 and 5). Our results clearly indicated that suppression of the production of tasiRNAs by 2b(13‐111) resulted from the inhibition of AGO1‐mediated slicing of the TAS1C transcripts, rather than from the action of binding to long dsRNA, as observed with 2b(1‐61).

In previous studies, we found that the 2b–green fluorescent protein (GFP) fusion protein was localized in the nucleolus (Duan et al., 2012). The removal of both NLSs in 2bN1,2m or the deletion of the N‐terminal 38 amino acids, including both NLSs of 2b, in 2b(38–111) led to the cytoplasmic localization of the resulting 2b fusion protein. The removal of NLS1 in 2bN1m specifically excluded the protein from the nucleolus, whereas the removal of NLS2 in 2bN2m showed a weak nucleolar presence (Duan et al., 2012). The three 2b mutants, 2b(38‐111), 2bN1m and 2bN1,2m, which failed to target to the nucleolus (Duan et al., 2012), also failed to inhibit AGO1‐mediated slicing of the TAS1C transcripts (Fig. 1B), although they contained the essential AGO‐binding domain (Fig. 1A); in contrast, the weak nucleolar localization of 2bN2m exhibited partial inhibitory activity of AGO1 slicing of the TAS1C transcript (Fig. 1B). These results were consistent with our previous observations that nucleolar targeting of 2b is required for co‐localization of 2b and AGO proteins (Duan et al., 2012). These findings indicated that, in addition to containing the AGO‐binding domain, a functional NoLS is also required for the 2b protein to achieve the inhibition of in vivo AGO1 Slicer activity. Our previous results showed that 2b(13‐111), which retains the intact NoLS, but lacks long dsRNA‐binding activity, is co‐localized with AGO proteins in the nucleus (Duan et al., 2012). In this study, we showed that 2b(13‐111) retained almost full inhibitory ability of in vivo AGO1 Slicer activity (Fig. 1B). Together, these results reveal that 2b mutants which retain the domains essential for AGO‐binding activity and nucleolar targeting, without dsRNA‐binding activity, are capable of inhibiting in vivo AGO1 Slicer activity.

The finding that 2b(1‐61) effectively suppressed tasiRNA production is in agreement with our previous findings that 2b mutants, which retain the activity to bind dsRNAs, but without the AGO1‐binding domain, exhibit efficient suppression of PTGS, and these results suggest that direct 2b–AGO interaction is not essential for 2b suppression of PTGS and RNA‐directed DNA methylation (RdDM) (Duan et al., 2012). With no requirement of binding to AGOs for the silencing suppression activities of 2b in vivo, in vivo 2b–AGO interaction might not inhibit in vivo AGO Slicer function. In this study, we have provided evidence that the 2b protein is able to inhibit in vivo AGO1 Slicer activity in the absence of its dsRBD. These findings further support the idea that 2b may suppress PTGS and RdDM in vivo by the binding and sequestration of siRNA and the long dsRNAs precursor in a process that can be facilitated by its in vivo interactions with and inhibition of the Slicer function of AGO proteins. However, we cannot rule out the possibility that the in vivo 2b–AGO interaction may represent another layer of host anti‐suppression of RNA silencing of the 2b protein, even though binding to the 2b protein may somewhat sacrifice the in vivo activities of AGOs.

Experimental Procedures

Plant material and growth conditions

RDR6i transgenic N. benthamiana has been described previously (Schwach et al., 2005). Transgenic and wild‐type N. benthamiana plants were grown in a glasshouse at 25 °C with 16‐h light/8‐h dark cycles.

Plasmid constructs

pCAMBIA‐MIR173 (35S‐MIR173) and pCAMBIA‐TAS1C (35S‐TAS1C) were constructed by polymerase chain reaction (PCR) amplification of the Arabidopsis MIR173 precursor sequence with the primers MIR173 5′ (GGTCTAGAATAATTAGCAAGTAATAAGG; italic letters for XbaI) and MIR173 3′ (CCGAGCTCATCTGTTATACAACCAAATCC; italic letters for SacI), and of the TAS1C sequence with the primers TAS1C 5′ (GGTCTAGAAAACCTAAACCTAAACGGCTAAG; italic letters for XbaI) and TAS1C 3′ (CCGAGCTCAACAAAACAATGCTGTTTCATC; italic letters for SacI). The PCR fragments were cloned into pGEM‐Teasy vector (Promega, Madison, WI, USA) to obtain pGEM‐MIR173 and pGEM‐TAS1C, respectively. The XbaI–SacI fragments containing MIR173 or TAS1C were inserted into pCAMBIA1300‐221 digested with XbaI and SacI to obtain pCAMBIA‐MIR173 and pCAMBIA‐TAS1C, respectively.

pCAMBIA‐SD2b (35S‐2b) and its derivative mutant constructs, 35S‐2b(1‐61), 35S‐2b(13‐111), 35S‐2b(38‐111), 35S‐2bN1m, 35S‐2bN2m and 35S‐2bN1,2m, and 35S‐P19, used in in vivo Slicer activity assays, were constructed as described previously (Duan et al., 2012).

Transient expression for in vivo inhibition of Slicer activity assays

Five‐week‐old N. benthamiana leaves were infiltrated with a culture of Agrobacterium EHA105 carrying the indicated DNA plasmid or empty vector pBI121‐GUS to an optical density at 600 nm (OD₆₀₀) of 0.5 for infiltration. For the co‐infiltration of 35S‐TAS1C and 35S‐MIR173, cultures of Agrobacterium carrying 35S‐TAS1C and 35S‐MIR173 were adjusted to OD = 0.5, and an equal volume was mixed and used. For in vivo inhibition of Slicer activity assays, 35S‐MIR173 and 35S‐TAS1C were co‐infiltrated with empty vector, 35S‐2b, 35S‐2b(1‐61), 35S‐2b(13‐111), 35S‐2b(38‐111), 35S‐SD2bN1m, 35S‐SD2bN2m and 35S‐SD2bN1,2m, and 35S‐P19, into N. benthamiana (Nb) and Nb‐RDR6i plants. The concentration of each culture was adjusted to OD₆₀₀ = 1.2, 1.2 and 0.6.

RNA extraction and RNA blot analysis

Plant total RNA used for RNA gel blot was extracted by the hot‐phenol method as described previously (Fernandez et al., 1997). For the detection of TAS1C, 2b and RDR6, gene‐specific DNA fragments were labelled with [α‐³²P]‐dCTP using the Rediprime II System. For small RNA blot, total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) and small RNA was obtained via lithium chloride and ethanol precipitation from total RNA. For the detection of tasiRNAs, 30 μg of small RNA were separated on 17% polyacrylamide–8 m urea gels. The tasiRNA‐specific probe, TTCTAAGTCCAACATAGCGTACCTGTCTC, and miR173‐specific probe, GTGATTTCTCTCTGCAAGCGAA, were labelled with [γ‐³²P]‐ATP (Perkin Elmer Life and Analytical Sciences, Waltham, MA, USA) using polynucleotide kinase (NEB).