Identification of an ARGONAUTE for antiviral RNA silencing in Nicotiana benthamiana

ARGONAUTE proteins (AGOs) are known to be key components of the RNA silencing mechanism in eukaryotes that, among other functions, serves to protect against viral invaders. Higher plants encode at least ten individual AGOs yet the role played by many in RNA silencing-related antiviral defense is largely unknown, except for reports that AGO1 , AGO2 , and AGO7 play an antiviral role in Arabidopsis . In the plant virus model host Nicotiana benthamiana, Tomato bushy stunt virus (TBSV) P19-suppressor mutants are very susceptible to RNA silencing. Here we report that a N. benthamiana AGO ( NbAGO ) with similarity to Arabidopsis AGO2, is involved in antiviral defense against TBSV. The activity of this NbAGO2 is shown to be directly associated with anti-TBSV RNA silencing, while its inactivation does not influence silencing of transiently expressed transgenes. Thus, the role of NbAGO2 might be primarily for antiviral defense.

Tombusviruses like Tomato bushy stunt virus (TBSV) are well suited to study antiviral RNA silencing because they generate abundant substrates for Dicer to yield high levels of duplex short interfering RNAs (siRNAs) (Molnar et al., 2005;Omarov et al., 2006) and they encode a 19-kDa protein (P19) that is a potent suppressor of RNA silencing (Voinnet et al., 1999;Scholthof,  2006). P19 is used for RNA silencing research in many organisms because it universally blocks this process (Scholthof, 2006) by sequestering 21-bp duplex siRNAs (Vargason et al., 2003;Ye et al., 2003). The proposed model in the context of TBSV infection is that the appropriation of virus-derived siRNAs by P19 prevents their subsequent incorporation into an antiviral RNAinduced silencing complex (vRISC) (Silhavy and Burgyan, 2004;Scholthof, 2006). In support of this we and others have provided direct evidence for a vRISC in Tombusvirus-infected plants, using TBSV P19-mutants and various biochemical approaches (Omarov et al., 2007;Pantaleo et al., 2007). More recently we isolated a sequence-specific vRISC from monocot and dicot plants infected with other viruses (Ciomperlik et al., 2011). Thus, plants have conserved the ability to mount an antiviral defense by activating a discrete and virus sequence-specific vRISC that can be isolated and analyzed in vitro.
The model for RNAi in eukaryotes implies that members of the ARGONAUTE protein (AGO) family members form key catalytic units of RISC to target RNAs for translational repression or cleavage (Baulcombe, 2004). Higher plants encode ten or more AGO genes, but other than a role for AGO4-like proteins in R-gene induced virus resistance (Bhattacharjee et al., 2009) and antiviral RNAi contributions by AGO1 -2, and -7 in Arabidopsis (Morel et al., 2002;Qu et al., 2008;Azevedo et al., 2010;Harvey et al., 2011;Wang et al., 2011) (Jaubert et al., accompanying manuscript), the contribution of AGOs in antiviral silencing is virtually unknown for other plant species (Alvarado and Scholthof, 2009). Nicotiana benthamiana is a wellestablished host for plant-virus research (Goodin et al., 2008) that mounts a biochemically tractable antiviral RNAi response (Omarov et al., 2007;Pantaleo et al., 2007), for which genomic information is rapidly accumulating, and it is susceptible to many more viruses than the genetic plant model Arabidopsis. For instance, Arabidopsis is not susceptible to TBSV although this virus has a vast host range spanning ~20 plant families and ~120 species (Yamamura and Scholthof, 2005); TBSV also replicates in yeast (Panavas and Nagy, 2003). Studies with TBSV and P19 in N. benthamiana have contributed significantly to our understanding of RNA silencing (Silhavy and Burgyan, 2004;Scholthof, 2006;Ding and Voinnet, 2007) and therefore, results obtained with these models systems can be expected to yield novel results of use and guidance to systems beyond Arabidopsis.
Here we report that down-regulating expression of an N. benthamiana AGO with similarity to Arabidopsis AGO2 plays a key and specific role in anti-TBSV RNA silencing.

A role of NbAGO2 In Susceptibility of N. benthamiana to Suppressor-Defective TBSV
To examine a possible role of an AGO2-like candidate in anti-TBSV silencing in N.
benthamiana, we identified a tobacco (N. tabacum) AGO2-homolog by searching the publicly available tobacco sequences for similarity with the 10 and 18 AGOs from Arabidopsis and rice, respectively. This identified a gene that is relatively well conserved in solanaceous plants based on comparisons with tomato and N. tabacum (Fig. 1A). Phylogenetic analysis suggests that like Arabidopsis AGO2, the solanaceous homolog falls within a clade with AGO3 and AGO7 (Fig.   1B).
Using primers based on the identified sequences a ~0.6 kbp AGO2 (NbAGO2-1, accession JF815524) cDNA fragment was amplified from N. benthamiana. Sequencing showed it to be ~96% similar to the N. tabacum sequence, and a ~65% nucleotide and ~50% amino acid identity with Arabidopsis AGO2 (Fig. 1A) genes identified from solanaceous species or when used in nucleotide sequence BLAST queries, minimizing the potential for cross-silencing genes other than NbAGO2-1.
To facilitate the use of Tobacco rattle virus (TRV) mediated VIGS, the fragment of NbAGO2-1 cDNA, referred to hereafter as NbAGO2, was cloned into TV-00, which is an effective silencing vector for AGO genes in N. benthamiana (Jones et al., 2006;Bhattacharjee et al., 2009), to yield TV-NbAGO2 that was used for VIGS. Four weeks post TV-RNA1+TV-NbAGO2 infection, RT-PCR tests using primers specifically designed to only amplify endogenous NbAGO2 mRNA, showed that NbAGO2 expression was silenced ( Fig. 2A). Western blot analyses using antibodies raised against a specific NbAGO2-peptide confirmed the reduced expression of NbAGO2 (Fig. 2B). Even upon prolonged growth no obvious morphological phenotype was discernable and the plants flowered normally (Fig. 2C).
The NbAGO2-silenced plants were then tested for susceptibility to P19-suppressor defective TBSV mutants that can initially infect N. benthamiana but are subsequently effectively silenced resulting in recovery of the plants (Chu et al., 2000;Omarov et al., 2006). The reasoning was that compromised silencing of a crucial antiviral NbAGO would prevent recovery and yield symptoms reminiscent of those observed upon infection with wild-type TBSV. Inoculation of the NbAGO2-silenced plants with wild-type TBSV (T), resulted in a normal progression of a systemic infection and severe symptoms (Fig. 2C). As expected p19-defective TBSV (TdP19) (Scholthof, 2006) was unable to establish a severe systemic infection in various control plants, i.e., those not silenced or silenced with TRV containing other NbAGO inserts or no insert (Supplemental Fig. 1). In contrast, in NbAGO2-silenced plants the TdP19 mutant induced very severe systemic symptoms reminiscent of those caused by the wild-type virus and the plants did not recover (Fig. 2C). These findings strongly suggest that NbAGO2-silencing compensates for show any effect of silencing NbAGO1, or -4 on symptoms or ability to exhibit recovery associated with infection by the P19-defective TBSV (Supplemental Fig.1).

NbAGO2 Mediates RNA Silencing Against TBSV
To determine whether the strong stimulatory effect of NbAGO2 silencing on invasion by the TdP19 mutants is exclusively associated with RNA silencing, and not due to some unspecified effect on the ability of the mutants to systemically invade the plants, we monitored early infection events in inoculated leaves. For this, we developed a versatile, modified TBSV-GFP coat-protein replacement vector (TG) that yields high levels of GFP expression in inoculated leaves. This vector is available in two constructs, the first allows for the generation of in vitro produced infectious transcripts that can be rub-inoculated onto plants, and the second can be agroinfiltrated to launch the infection (Supplemental Fig. 2A). For both systems the vector is also available as a version that is defective for P19 expression (TGdP19) and this leads to dramatic attenuation of GFP expression (Supplemental Fig. 2A). For both, transcript inoculation and agroinfiltration, the absence of P19 can be compensated for by agroinfiltration with T-DNA constructs expressing P19, or by restoring P19 expression from the virus (Supplemental Fig. 2A).
Furthermore, the defect can also be complemented by co-expression of the potyvirus HC-Pro or hordeivirus γ b suppressors of RNA silencing (Ciomperlik, 2008) (Supplemental Fig. 2B).
Therefore, the attenuation of GFP expression associated with TGdP19 infection is due to the absence of suppressor activity and thus strictly caused by RNA silencing.
Inoculation of the control construct TG (encoding P19), by transcript inoculation or agroinfiltration, leads to rapid (visible in 1-2 days) and high levels of GFP expression in inoculated leaves of control plants as well as in NbAGO2-silenced plants (Fig. 3). In comparison, even though inoculation of TGdP19 (not expressing P19) onto control plants may lead to some early GFP accumulation, GFP expression is rapidly silenced (Fig. 3). However, in NbAGO2-

Specificity of NbAGO2 for Antiviral Silencing
To determine whether NbAGO2 has a general role in RNA silencing we monitored its effect on silencing of a transiently expressed transgene. For instance, upon agroinfiltrating a 35S:GFP T-DNA construct into N. benthamiana leaves, GFP expression will occur at readily detectable levels within 2-3 days post-infiltration ( Fig. 4A and C), but within a matter of five-seven days later RNA silencing will result in a loss of fluorescence ( Fig. 4A and B) (Voinnet et al., 2003).
This loss of expression can be suppressed by P19 demonstrating that the effect is due to RNA silencing ( Fig. 4A). To test whether NbAGO2 is involved in this process, we used plants infiltrated three weeks earlier to establish infection with TRV-RNA1 plus either the empty TV-00 vector or TV-NbAGO2, and agroinfiltrated those either with the 35S:GFP or TGdP19 constructs.
As the results in Fig. 4B demonstrate, silencing of 35S:GFP occurs for both treatments.
However, (as in Fig. 3), GFP expression from TGdP19 is high in NbAGO2-silenced plants.
AGO1and AGO10  benthamiana that must be turned over before the action of P19 can affect GFP expression.
Nonetheless, VIGS of NbAGO2 does not noticeably rescue the silencing of this construct at any time point up to six days, indicating that this AGO protein is not likely to be involved in the translational repression of endogenous transcripts.
Collectively, these results demonstrate that NbAGO2 is not involved in transgene-or miRNA induced silencing but instead has a specific role in antiviral silencing.

DISCUSSION
Inactivation or downregulation of AGO2 does not induce a readily identifiable morphological phenotype in Arabidopsis (Vaucheret, 2008) (Jaubert et al., accompanying manuscript) or in N. benthamiana (Fig. 2). Likewise, NbAGO2 appears to be dispensable for transgene silencing and miRNA-mediated translational repression (Fig. 4), or is at least required at much lower threshiold than is necessary for antiviral activities. These observations suggest a role that is distinct from other characterized AGO proteins. One such role is revealed by our results Although hypomorphic Arabidopsis ago1 (Morel et al., 2002;Qu et al., 2008), and null ago7 mutants (Qu et al., 2008)  The fact that TRV-VIGS can be used to induce silencing of an antiviral AGO (NbAGO2) in N.
benthamiana may at first seem counter intuitive. This paradox can be extended to our observations that apparently neither NbAGO1 nor NbAGO4 are required for TRV to induce and maintain VIGS, as was also found by others for NbAGO1 and NbAGO4 (Jones et al., 2006) and AtAGO4 (Dunoyer et al., 2004). However, these observations may in fact underscore the functional specialization of AGO proteins. That is, whereas certain AGO proteins, such as AGO2 may be required for the targeting of viral RNAs, nuclear-encoded mRNAs clearly seem to be directly targeted by AGO1 and/or AGO10 (Brodersen et al., 2008;Mallory and Vaucheret, 2010). Thus, although silencing of AGO2 might result in a loss of targeting of viral RNAs, NbAGO2 does not appear to affect nuclear transcribed genes (Fig. 4), and thus a lack of NbAGO2 would not affect targeting of mRNAs by NbAGO1 or other AGOs.
Although the above scenario possibly explains the differential targeting of mRNAs versus viral RNAs, it still remains curious that TRV-mediated VIGS is not critically influenced by the compromised expression of aforementioned AGOs, including NbAGO2. It is possible that compromised.
The most straightforward interpretation of our results is that NbAGO2 is the core catalytic unit of the vRISC that is activated or assembled upon infection of N. benthamiana with TBSV (Omarov et al., 2007). However, we cannot rule out the possibility that NbAGO2 might have upstream or downstream regulatory roles that influence RNA silencing through a process not directly related to vRISC activity or composition. To address the mechanistic role it will be necessary to determine if and/or how virus infection influences vRISC properties. Regardless of the importance of such future studies, at present we can conclude that a newly identified ARGONAUTE (NbAGO2) is required for antiviral RNA silencing against TBSV in N.

Agroinfiltration to Launch TRV Infection
Agrobacterium cultures containing the TRV-RNA1 (in C58C1) and TRV-RNA2 cassettes were grown overnight (16-22 hrs) at 25-28 C in 5 ml LB medium in presence of Kan 50 . Cells were collected by centrifugation at 3K (Sorvall) for 15 min, and the pelleted cells resuspended in 10 ml 10 mM MgCl 2 . The OD was measured and subsequently adjusted with 10 mM MgCl 2 to 0.5. Five ml of these TRV-RNA2 expressing cells were mixed with one ml of the TRV-RNA1 cultures, and the mix was infiltrated with a needleless syringe into the abaxial side of two N.
benthamiana leaves on young plants at the five-leaf stage. Infiltrated plants were grown in the greenhouse or in the laboratory at 23-25 C daytime temperatures with 14-16 hour daylight.
Plants were grown for four weeks prior to infection with TBSV variants.
Constructs and procedures for co-infiltration of 35:GFP and P19 expressing constructs were described recently (Saxena et al., 2010).

Infection with TBSV
Prior to inoculation with TBSV variants, lower leaves were removed from silenced N.
benthamiana plants, and one of the remaining middle leaves was inoculated by using sap from plants infected with TBSV or corresponding p19 mutants (Omarov et al., 2006;Omarov et al., 2007), or for TBSV-GFP constructs with in vitro generated transcripts or by agroinfiltration.
Construction and detailed properties of the newly generated TBSV-GFP constructs will be described separately (manuscript in preparation). In essence, the TBSV-GFP (TG) variants are based on an infectious clone in which the coat protein open reading frame was largely removed and replaced with gfp. TBSV-GFP-DP19 (TGdP19) contains a p19 start-codon mutation and two downstream stop-codons. Both TG and TGdP19 were used to replace analogous TBSV backbone inserts of pJL54TG (manuscript in preparation) to obtain T-DNA constructs expressing infectious viral RNAs, for agroinfiltration.

Protein analysis
NbAGO2 antibody. An NbAGO2 affinity purified rabbit polyclonal antibody against a specific synthetic NbAGO2 peptide (CLEDPEGKDPPRDVF) was obtained from GenScript (Piscataway, NJ). The antibody was resuspended in water to a concentration of 1 mg/ml and a dilution of 1:3000 was determined to be best for use in Western blots.     rub-inoculated with TGdP19 transcripts (lanes 3 and 6) (txt). Note: TG infections will not yield systemic GFP expression due to rapid accumulation of recombinants (Qiu and Scholthof, 2007).

SDS-PAGE
Comparative protein loading of the samples is shown in the Coomassie stained gel on the bottom.   (Omarov et al., 2007) and parallel experiments with a mutant not expressing P19 (pHS157) yielded identical results. The results on the TV-00 control in C) are shown for both mutants (P19/75-78 on left and pHS157 on the right). As consistently seen with over ten separate experiments, due to a synergistic interaction, the plants infected both with TRV and TBSV mutants are somewhat stunted compared to plants only infected with TRV (-).