The potyviral silencing suppressor HCPro recruits and employs host ARGONAUTE1 in pro-viral functions

In this study, we demonstrate a novel pro-viral role for the Nicotiana benthamiana ARGONAUTE 1 (AGO1) in potyvirus infection. AGO1 strongly enhanced potato virus A (PVA) particle production and benefited the infection when supplied in excess. We subsequently identified the potyviral silencing suppressor, helper-component protease (HCPro), as the recruiter of host AGO1. After the identification of a conserved AGO1-binding GW/WG motif in potyviral HCPros, we used site-directed mutagenesis to introduce a tryptophan-to-alanine change into the HCPro (HCProAG) of PVA (PVAAG) and turnip mosaic virus (TuMVAG). AGO1 co-localization and co-immunoprecipitation with PVA HCPro was significantly reduced by the mutation suggesting the interaction was compromised. Although the mutation did not interfere with HCPro’s complementation or silencing suppression capacity, it nevertheless impaired virus particle accumulation and the systemic spread of both PVA and TuMV. Furthermore, we found that the HCPro-AGO1 interaction was important for AGO1’s association with the PVA coat protein. The coat protein was also more stable in wild type PVA infection than in PVAAG infection. Based on these findings we suggest that potyviral HCPro recruits host AGO1 through its WG motif and engages AGO1 in the production of stable virus particles, which are required for an efficient systemic infection.


Introduction
In nature, plants are constantly challenged by a variety of biotic and abiotic stresses including virus infections. Plant viruses are a ubiquitous and diverse group of parasites that harnesses the host's cellular components in the execution of viral functions including replication, translation and movement. Many plant viruses cause substantial losses in agriculture and are thus a significant risk to global food and feed production [1]. Understanding the importance of virus infections not only in model plants but also in crops and ecosystems has recently been highlighted as the grand challenge of plant virology [2].
To counter virus infections and other pathogens plants must recognize the threat and deploy their defenses accordingly. Vector transmission of the viruses overcome the primary physical barriers of a waxy cuticle and cell wall, and therefore further layers of defense responses aimed at eliminating, attenuating or containing infection are required [3,4]. RNA silencing is a major conserved regulatory and antiviral mechanism, which has been extensively studied in plants (Recently reviewed in [5][6][7]). Thus, the key steps leading to RNA silencing are well known: ARGONAUTE-family (AGO) proteins employ short RNA (sRNA) molecules to selectively destroy or inhibit complementary RNAs resulting in the silencing of the target (reviewed in [8]). RNA silencing enables the plant to combat transgene or viral gene expression but it is also crucial for the regulation of endogenous gene expression for example during development [9].
AGO proteins are effectors of RNA silencing characterized by the presence of hallmark PAZ-and PIWI-domains responsible for RNA-binding capacity and slicer activity, respectively [10]. The plant AGO family is diverse and its members have acquired specialized functions beyond their typecast as effectors of RNA silencing [8]. AGO proteins can, for example, take part in DNA repair, methylation and chromatin remodeling [11][12][13][14]. The model plant Arabidopsis thaliana (hereafter Arabidopsis) has ten AGO genes (AGO1-10) which are grouped based on phylogeny and their preferred selection of sRNA cargoes [15][16][17]. Arabidopsis AGO1 was the first described ARGONAUTE protein and it became the founding member of the family [9]. Originally identified via mutations causing defects in leaf development, AGO1 activity was soon found to be important to the phenomenon of gene silencing [18]. Befitting a canonical AGO protein, AGO1 is an RNA slicer with the ability to use short RNAs as guides to target complementary RNAs either to cleavage or to translational repression [19,20]. Because of their importance in RNA silencing, AGOs are traditionally placed at the frontline of host antiviral defense [3,21,22].
Potyviruses (family Potyviridae) are the largest genus of positive-sense single-stranded RNA viruses in plants [23,24]. Because they are geographically widespread and infect numerous crops, potyviruses are ranked among the top ten economically most important plant viruses [1,25]. Potyvirus genomes are compact, circa 10 kb in length. Their genome organization is conserved and encompasses two open reading frames, which code for eleven viral proteins [24]. The genomic RNA has a poly-A tail and contains a covalently attached viral protein genome-linked (VPg) at the 5'end. Potyviruses are encapsidated into flexible filamentous particles that are transmitted by aphid vectors in a non-persistent manner.
Recently, HCPro has also been shown to be essential for the formation of potyvirus-induced granules, potential shelters for vRNA against the host's silencing machinery [39].
Evidence accumulated over the past years has highlighted the relationships between viral proteins and plant AGOs. Although AGO2 has been established as the major antiviral AGO protein in leaves [40][41][42][43], AGO1 is still an important target of numerous viral silencing suppressors [44][45][46][47][48][49][50][51][52][53]. Forming a link to potyvirus infection, we have previously demonstrated the co-presence of AGO1 and the potato virus A (PVA) silencing suppressor HCPro on Nicotiana benthamiana (hereafter N. benthamiana) ribosomes during PVA infection [54]. This association suggests AGO1 could be involved in the repression of potyviral translation as part of host defense. Moreover, both proteins are integral components of potyvirus-induced granules [39]. Against this background, we set to investigate the relationship between AGO1 and HCPro and its prospective biological importance in more detail.
In this study, we show that the potyviral silencing suppressor HCPro interacts with AGO1 through a conserved WG motif and directs AGO1's association with the viral coat protein CP. Furthermore, we report the relevance of the interaction for the systemic spread of the infection and suggest that AGO1 contributes to the success of potyvirus infection by enabling the accumulation of stable virions.

AGO1 promotes PVA infection and is required for particle accumulation
To clarify the role of AGO1 in potyvirus infection, we first studied the effects of AGO1 scarcity and surplus on the progress of PVA infection in N. benthamiana. The AGO1 gene was transiently knocked down by expressing a hairpin construct targeting its mRNA in N. benthamiana. An empty pHELLSGATE vector (hp-) was used as the control. We used an infectious cDNA (icDNA) clone of PVA tagged with the Renilla luciferase (RLUC) gene for accurate quantitation of PVA gene expression [55]. Henceforth we refer to this virus as PVA WT . PVA WT gene expression was determined by measuring virus-derived RLUC reporter activity from both AGO1-silenced and control samples. The values were normalized to firefly luciferase activity in the same samples. Immunocapture reverse transcription polymerase chain reaction (IC RT-PCR) was used to determine PVA WT particle accumulation in the same samples. In IC RT-PCR capture of virus particles with anti-CP antibodies is followed by the quantification of enclosed vRNA. The vRNA copy number is used to describe the number of intact particles. Although silencing of AGO1 did not significantly alter PVA WT gene expression, it slightly increased the amount of vRNA compared to the control (Fig 1A and 1C) and drastically reduced PVA WT particle accumulation (Fig 1B). The relative amount of PVA WT RNA within particles in AGO1-silenced leaves was 50 fold lower than in the non-silenced controls (Fig 1B). On the other hand, an abundance of AGO1, achieved by transient overexpression, significantly enhanced both PVA WT gene expression and particle accumulation. AGO1 overexpression improved PVA WT gene expression by over two-fold and yielded six times more vRNA within particles compared to the control although vRNA levels were not significantly affected ( Fig  1D-1F). To gain further information about the role of HCPro, we checked the effect of AGO1 silencing and overexpression on the vRNA levels of an HCPro-deficient virus (PVA ΔHCPro ). Higher vRNA levels were measured in AGO1 -silenced samples while an excess of AGO1 reduced the amount of vRNA (Fig 1G and 1H). The hairpin silencing system was judged to provide sufficient silencing of AGO1 as its mRNA levels were less than 20% of those in the It was clear that while AGO1 targeted the HCPro-less virus in a defensive manner it has a positive effect on the infection in the presence of a functional HCPro. AGO1 was especially important for virus particle production as its knock-down nearly abolished particle accumulation.

Identification and site-directed mutation of a putative ARGONAUTEbinding motif conserved in potyviral HCPros
HCPro is renowned for its capacity to interact with a wide range of host factors. As our results hinted AGO1 could benefit the infection instead of countering it, we were interested in exploring the mechanisms and viral factors involved. The potyviral silencing suppressor HCPro provided an obvious starting point as its absence sensitized the virus to AGO1 and we had earlier demonstrated its involvement with both viral translation and AGO1 association (Fig 1, [54]).
In the current study, we used a computational sequence analysis tool [56,57] to search for putative WG/GW -type AGO1 binding domains in HCPro. The search revealed a single WG motif in the central region of PVA HCPro. This motif contains one WG pair at W208 G209 (Fig 2A). Although the motif was predicted to have low compositional compatibility to an ideal AGO-binding domain, a multiple sequence alignment of 113 potyviral polyproteins revealed the WG pair was conserved in 99.1% of the sequences (Fig 2B). Only banana bract mosaic virus (BBrMV, GenBank accession no. YP_001427389) lacked the WG pair (SI Table, S2 Fig). The high degree of conservation of the WG motif in HCPros throughout the genus Potyvirus spoke strongly in favor of its functional importance.
Next, we focused on deciphering the role of the WG motif in the context of the HCPro-AGO1 interaction and its potential significance in potyvirus infection. Site-directed mutagenesis was performed to change the tryptophan residue in the WG pair to alanine (W208A). Corresponding WG motifs in the icDNAs of two potyviruses, PVA and turnip mosaic virus (TuMV) were chosen as targets because this would allow us to compare and contrast the effects of the alanine substitution in two related virus species. The resulting versions with mutated HCPro sequences were named PVA AG and TuMV AG , respectively. Additionally we generated Twin Strep-tag, RFP fusion constructs of PVA HCPro and HCPro AG to facilitate their transient overexpression and detection in confocal fluorescence microscopy and co-immunopurification experiments. For the sake of clarity, these fusion proteins are called HCPro and HCPro AG .
The biological functionality of HCPro AG was evaluated by analysing its complementation capacity by supplementing the HCPro-deficient virus (PVA ΔHCPro ) with either HCPro, HCPro AG or GUS as a negative control. Recovery of viral gene expression to the level of PVA WT confirmed that HCPro AG complemented the expression of PVA ΔHCPro as well as the wild type protein (Fig 3A).
The main potyviral silencing suppressor HCPro is able to rescue the expression of genes targeted to silencing [28,29]. When a hairpin construct was used to prime the plants for silencing RLUC mRNA, transient overexpression of HCPro AG resulted in the nearly complete restoration of RLUC expression from the co-infiltrated monocistronic RLUC mRNA ( Fig 3B). Altogether, the results suggested HCPro AG suppressed silencing similarly to HCPro. Another of HCPro's critical functions in PVA infection biology is to enable and enhance viral translation together with VPg [39]. We investigated the performance of HCPro AG in the translational enhancement of a replication-deficient virus lacking HCPro (PVA ΔGDDΔHCPro ). In this experiment, HCPro, HCPro AG or a GUS control were co-expressed with VPg and PVA ΔGDDΔHCPro translation was measured as virus-derived RLUC activity. Co-expression of VPg with either HCPro or HCPro AG resulted in a circa 30 fold translational enhancement ( Fig 3C). Expression of the GUS control alone or together with VPg did not significantly improve viral translation. Western blot analysis was used to confirm that HCPro and HCPro AG were expressed at similar levels in the experiments ( Fig 3D). Finally, another α-HCPro western blot was carried out from full length PVA WT and PVA AG infected samples to show presence of monomeric HCPro as evidence of intact auto-proteolytic activity of HCPro ( Fig 3E).
Based on these results the disruption of the WG motif did not significantly disturb HCPro's functionality in PVA expression. The mutated version of the protein retained its essential biological functions including complementation capacity, silencing suppression activity and the ability to enhance translation in collaboration with VPg.

A disrupted WG motif compromises HCPro's co-localization and interaction with AGO1
To study the intracellular localization of HCPro/HCPro AG and CFP AGO1 the expressed proteins were visualized in N. benthamiana by confocal microscopy. Unfused RFP and CFP proteins served as controls. At 3 dpi, the epidermal layer of cells was examined under a confocal microscope and fluorescence emission from RFP and CFP was detected in sequential scanning mode. Control experiments showed that in the absence of HCPro the CFP AGO1 signal with GUS was used as the positive control. Firefly luciferase was included for normalization purposes in all sets at OD 600 0.01. Samples were taken at 5 dpi followed by a dual luciferase assay to determine virus-derived RLUC activity. B) Silencing suppression test. A hairpin construct targeting the Renilla luciferase mRNA (hpRLUC) or the empty vector (hp-) were infiltrated at OD 600 0.4 one day prior to the co-infiltration of RLUC and GUS, HCPro or HCPro AG overexpression vectors each at OD 600 0.3. As above firefly luciferase was included in all sets at OD 600 0.01. Samples were taken at 4 dpi. C) Translational enhancement of PVA ΔGDDΔHCPro by the coexpression of VPg and HCPro or HCPro AG . D) The transient expression of HCPro/HCPro AG in the infiltrated leaves was confirmed by western blotting using a mouse monoclonal anti-RFP antibody (top panels). Arrow indicates the Twinstrep-tag RFP-HCPro fusion proteins. Equal loading of the gels was checked by staining the membranes with Ponceau solution (bottom panels). E) PVA WT -and PVA AG -infected leaf samples and a mock sample were subjected to a western blot analysis using anti-HCPro antibodies. The size of the monomeric HCPro is marked by an arrow. An unspecific band is visible in the mock sample, but it is slightly larger than HCPro. In the presence of wild type HCPro, CFP AGO1 fluorescence accumulated predominantly in granular structures that also contained HCPro (Fig 4A). Co-localization analysis revealed that the sequestration of CFP AGO1 into HCPro-containing granules was significantly reduced but not completely abolished in the presence of the HCPro AG mutant. The degree of CFP AGO1 co-localization with HCPro AG was reduced to 45 ± 38% from the 82 ± 18% of with the wild type protein (P-value of T-test 4 x 10 −9 ; in 3 independent experiments) indicating that less CFP AGO1 accumulated to granules formed by HCPro AG than by HCPro.
Thus, according to confocal microscopy results, HCPro and CFP AGO1 co-localized in N. benthamiana cells and the disruption of the WG motif interfered with the degree of co-localization. This observation supported the hypothesis that the motif could influence HCPro and AGO1 interaction.
The HCPro-AGO1 interaction was then studied by affinity purification/co-immunopurification using HCPro or HCPro AG as baits. For the affinity purification experiments, we coexpressed the bait proteins with either CFP AGO1 or CFP as a negative control. All constructs were agroinfiltrated into N. benthamiana plants at OD 600 0.1 and the infiltrated leaves were sampled at 3 dpi. Overexpression of AGO1 in mRNA level is validated in S3B Fig. We conducted affinity purification with cross-linked StrepTactin MacroPrep resin followed by western blotting with anti-HCPro and anti-GFP antibodies to detect the baits and CFP AGO1 target, respectively. The majority of HCPro/HCPro AG eluted in stable high molecular weight complexes that withstood SDS treatment and heat denaturation ( Fig 4B). Some monomeric forms were also present as indicated by a band at~75 kDa in the HCPro + CFP control sample. The western blots clearly showed CFP AGO1 co-immunoprecipitated with HCPro but not with HCPro AG . The anti-GFP western blot detected CFP AGO1 in high molecular weight complexes uniquely in association with the HCPro bait. CFP AGO1 was not detected in the high molecular weight complexes when HCPro AG was used as bait ( Fig 4B). The specificity of the HCPro-AGO1 interaction was supported by the observation that the CFP control alone did not coimmunoprecipitate with HCPro ( Fig 4B). The loss of CFP AGO1 co-immunoprecipitation in HCPro AG compared to the wild type protein indicates the capacity for AGO1 interaction was abolished or at least significantly weakened by the W208A mutation.

Impaired systemic infection in PVA AG and TuMV AG
We next assessed the effects of the W208A mutation on PVA and TuMV infections in planta. PVA WT and PVA AG were infiltrated into N. benthamiana at OD 600 0.05 and TuMV WT /TuMV AG at OD 600 0.5. PVA gene expression was determined as RLUC activity using a dual luciferase assay and the accumulation of TuMV CP was determined with a DAS ELISA assay. Samples were obtained from local leaves at 3 dpi and from emerging systemic leaves at 8 or 14 dpi.
At 3 dpi, PVA AG gene expression levels were similar to that of PVA WT in infiltrated leaves. Likewise, there were no differences in CP accumulation measured from local leaves infiltrated with TuMV WT and TuMV AG (Fig 5A and 5C). As the dual reporter combination in the TuMV plasmids conveniently enables the tracking of viral cell-to-cell movement, we also compared the initial movement of TuMV WT and TuMV AG and found that the mutant spread to neighboring cells as efficiently as the wild type virus. The average number of neighboring cells infected by TuMV WT was 2.5±2.4 and TuMV AG 1.8±1.9 in three independent experiments. Error is denoted as standard deviation and the difference did not have any statistical significance. In the dual reporter system, primarily infected cells emit both mCherry and GFP fluorescence while secondarily infected cells only emit virus-derived GFP fluorescence. The viruses were infiltrated

PLOS PATHOGENS
Potyviral HCPro employs AGO1 in proviral functions at a low OD 600 of 0.005 to start the infection from individual, sparsely located cells and the leaves were then inspected under a confocal microscope at 3 dpi. Primarily infected epidermal cells were identified based on overlapping mCherry and GFP fluorescence and the number of infected neighboring cells emitting only GFP fluorescence was calculated.
Similar gene expression and cell-to-cell movement signified that the local infection processes of PVA and TuMV were undisturbed by the disruption of the AGO-binding motif in HCPro. However, when the infection was allowed to proceed for one week more, we found that the mutant viruses could not establish a systemic infection as well as the wild type viruses. PVA gene expression and TuMV CP accumulation in non-infiltrated systemic leaves was reduced in both PVA AG and TuMV AG compared to the wild type viruses. After 8 dpi PVA AG gene expression reached only 20% of that of PVA WT while TuMV AG CP amount was, at 14 dpi, circa 30% of TuMV WT (Fig 5B and 5D).

Virus particle accumulation depends on the HCPro-AGO1 interaction
As silencing of AGO1 nearly abolished the accumulation of virus particles, we decided to investigate if disruption of the HCPro-AGO1 interaction affected particle production. In many cases phloem-based long-distance spread of virus infections relies on virus particles [58]. Poor systemic infections by the mutant viruses indeed pointed us to this direction.
The effect of the W208A mutation on PVA and TuMV particle production was studied by IC RT qPCR in N. benthamiana infected with either PVA WT /PVA AG or TuMV WT /TuMV AG . In both PVA and TuMV the mutation had a drastic negative effect on virus particle accumulation. The relative amount of PVA AG particles was, on average, only 9% of that of the wild type virus in systemically infected leaves (Fig 6A). Viral particle production was significantly reduced also in TuMV AG , which yielded 18-fold less particles than the wild type virus ( Fig 6B).
Next, we compared the visual appearance of PVA WT and PVA AG -derived particles using transmission electron microscopy. Fully assembled virions were abundant in a PVA WT infection at 9 dpi while only very few short and thin rod-shaped structures were observed in PVA AG samples (Fig 6C and 6D). While these structures could represent virus-like particles, they are unlikely to contain full-length vRNA. They may instead be degraded or otherwise deformed particles.

PVA CP is degraded faster in a PVA AG infection than in the wild type PVA infection due to misassembled particles
The dramatic reduction in the amount of virus particles was intriguing since it was disproportionate compared to the modest decrease in virus-derived gene expression in systemic leaves (comparison with Fig 5B and 5D). Stability of the CP is another indicator of virion formation because free CP that is not assembled into particles is accessible to proteases. Based on the viral gene expression results (Fig 5) we deduced that both wild type and mutant viruses produce CP and hypothesized that if PVA AG is unable to use the available CP in the production of stable viruses, the protein would be more susceptible to degradation. To test this idea we performed a CP stability assay on samples from PVA WT and PVA AG infections. Crude plant sap was incubated for one hour at room temperature followed by western blotting to detect CP and RLUC protein levels. Western blotting revealed the CP amount remained unaltered in the PVA WT were imaged by confocal microscopy at 3 dpi. Representative images from three independent experiments are shown, scale bar is 20 μm. For control experiments with unfused CFP and RFP, see S3B Fig. CFP AGO1 co-immunoprecipitates with HCPro but not with HCPro AG . Agroinfiltrations were done as in the confocal microscopy experiments and samples were taken at 3 dpi. Subsequently HCPro/HCpro AG were affinity purified using MacroPrep strep tactin. Anti-HCPro and anti-GFP antibodies were used to detect HCPro/HCPro AG and co-purified CFP AGO1, respectively. https://doi.org/10.1371/journal.ppat.1008965.g004

PLOS PATHOGENS
Potyviral HCPro employs AGO1 in proviral functions samples whereas CP produced by PVA AG was almost completely degraded during the incubation period (Fig 6E). The result supported the idea that CP produced by PVA AG was vulnerable because it was not incorporated into stable virions. Furthermore, the effect was CP-specific as the level of the RLUC reporter protein remained similar after the 1 h incubation.
Because IC-RT-PCR and electron microscopy of particles require the disruption of cells prior to analysis, we used transmission electron microscopy of thin sections of infected leaves to study particle formation in intact samples. This method revealed abundant stacks of particles in the vicinity of CI pinwheel structures in PVA WT infected cells (Fig 7). The average length of the particles in PVA WT infected tissues was around 700 nm. The accuracy of the estimation was somewhat impaired by the fact that the surrounding structures covered the exact ends of the stacked particles. Although some similar stacks were observed in PVA AG -infected leaves, they were detected only in a subset of infected cells. The stacked PVA AG particles were shorter, some of them only 500 nm, and thinner than wild type reference particles (Fig 7). The

Co-localization and co-immunopurification revealed an HCPro-mediated association between AGO1 and CP
Because CP was more stable in the presence of HCPro than HCPro AG , we decided to study if the intracellular localization of YFP-tagged CP was affected by AGO1, HCPro/HCPro AG or

PLOS PATHOGENS
Potyviral HCPro employs AGO1 in proviral functions their combinations. When all three proteins, YFP CP, CFP AGO1 and RFP-tagged HCPro were co-expressed and leaf samples were inspected under a confocal microscope, we observed that YFP CP and CFP AGO1 co-localized strongly in distinct cytoplasmic aggregates (Fig 8A). These aggregations of YFP CP and CFP AGO1 were not observed when HCPro was replaced by HCPro AG . In fact, their formation was entirely dependent on HCPro as we did not observe any YFP CP/ CFP AGO1 aggregations during co-expression with the RFP control (S3C Fig). YFP alone did not form distinct aggregations but was evenly spread in the cytoplasm and nucleus (S3C Fig). In the presence of HCPro AG both CFP AGO1 and YFP CP were more uniformly distributed throughout the cytoplasm. Analysis of controls revealed that YFP CP forms distinct, although smaller, spots also when co-expressed with the CFP control and HCPro suggesting that the level of endogenous AGO1 could be enough to induce the sequestration of YFP CP in response to HCPro. In line with this, silencing of AGO1 by a hairpin construct significantly decreased the number of cells containing YFP CP foci (Fig 8B).
Some viral CPs are known to interact with AGO proteins through WG/GW motifs [53,59]. According to a sequence search, the PVA CP did not contain any putative AGO1 binding motifs. Nevertheless, we used affinity purification to assess if CP could be co-purified with AGO1 and found that CP and AGO1 indeed co-immunoprecipitate, but only in the presence of HCPro (Fig 8C). In these experiments CP and HCPro, HCPro AG or GUS were transiently co-expressed with either CFP AGO1 as bait or CFP as negative control. Validation of CP expression in the inputs is presented in S5 Fig. Consistent to [34], accumulation of CP is prominent only when HCPro is co-expressed. Samples were obtained at 3 dpi and a GFP trap was used to affinity purify CFP AGO1 and associated protein partners. Western blotting with anti-GFP and anti-CP antibodies was used to detect proteins in the eluates. CFP AGO1 was mainly present in a high molecular weight complex corresponding to results from strep-tag purifications (Figs 4B and 8C). A strong band at circa 27 kDa suggests CFP AGO1 is proteolytically processed to some extent and that the CFP fusion partner can be detached from AGO1. A CP signal was detected when its expression was combined with CFP AGO1 and HCPro. No CP could be detected in the controls lacking either CFP AGO1 or HCPro (Fig 8C). Overall the co-immunopurification results correlated well with the microscopy results although in one biological repeat very little CP co-immunoprecipitated also when HCPro AG was used instead of the wild type protein. Thus it is possible that while HCPro's ability to mediate the association between CP and AGO1 was weakened by the W208A mutation it was not completely disabled.

Discussion
We report here a unique pro-viral role for plant AGO1 in the stability and accumulation of potyvirus particles. In addition, we demonstrate that the viral HCPro and, more specifically, a conserved WG motif therein, is responsible for engaging AGO1 in the promotion of the infection. When the virus lacked HCPro completely, it fell prey to AGO1's defensive activities. This was observed as a decrease in vRNA level responsive to AGO1 overexpression and a corresponding increase while AGO1 was silenced (Fig 1G and 1H). The generation and characterization of a novel HCPro mutant with impaired ability to associate with AGO1 underlined the biological importance of the interaction for virus particle yield. We found that a fully functional connection between HCPro and AGO1 was required for stable encapsidation and, hence, for optimal systemic spread of potyviruses. In support, our results indicated that the viral CP co-localized and co-immunoprecipitated with CFP AGO1 only in the presence of HCPro and was more vulnerable to degradation if the HCPro-AGO1 interaction was compromised. We show that CP's increased vulnerability is due to a destabilization of virus particles

PLOS PATHOGENS
Potyviral HCPro employs AGO1 in proviral functions linked to the mutation in HCPro. Thus, taken together, we propose that the potyviral HCPro recruits AGO1 through its WG motif and employs it in the assembly of stable virus particles.
Most of the viral silencing suppressors interact with AGO1 via a specific motif called WG/ GW. The motif derives its name from characteristic tryptophan (W) -glycine (G) residue pairs and specifically binds the PIWI domain that conveys slicer activity to AGO proteins [63][64][65][66]. Sequence searches revealed a putative AGO-binding WG motif in PVA HCPro and multiple alignment of 113 potyvirus polyproteins indicated the motif was conserved in almost all of the HCPro sequences (Fig 2, S1 Table), BBrMV making the sole exception (S2 Fig). Based on our results, HCPro's silencing suppressor activity appears to be independent of its ability to interact with AGO1.
The potyviral HCPro has been shown to sequester sRNAs of viral origin [42,67]. Moreover, Arabidopsis AGO1 hosted TuMV-derived vsiRNAs only in the presence of a silencing suppression deficient HCPro [42]. The potyviral HCPro can be divided into three regions: the N-terminal region involved in aphid transmission (amino acids 1-100 in PVA HCPro), the functionally rich central region (amino acids 100-300), and the C-terminal serine proteinase region (amino acids 300-457; [68]). The central region contains domains for RNAand siRNA-binding [68][69][70] and it is involved in RNA silencing suppression [71]. Essential functional properties such as systemic movement [36,72], synergistic interaction with other viruses [73,74] and virion formation [34] have been associated with this domain as well. To investigate the possibility of secondary effects of the W208A mutation located in the central region, HCPro AG was subjected to a number of assays designed to test its functionality. It successfully complemented viral gene expression level of the HCPro-less virus, retained silencing suppression capacity in our dsRNA hairpin test and enhanced translation normally in collaboration with VPg. These results emphasize that HCPro AG maintained central functional properties required for stages preceding encapsidation. Structural models of both HCPro WT and HCPro AG were prepared using I-TASSER server for protein structure predictions. The overlay of the predicted structures indicates that HCPro's overall structure along with the known functional motifs were not disturbed due to the W208A mutation (S6 Fig). Taken together, the functional data and the structural model suggest that the W208A mutation disrupts the AGO1 binding site locally without altering other functions of HCPro.
A tobacco etch virus (TEV) mutant called PC22, which has an asparagine changed to serine in the immediate vicinity of the WG-motif of HCPro (NGNFIWGLR!SGNFIWGLR), did not accumulate in systemically infected plants as efficiently as the wild type virus in a competition assay [75]. Nevertheless the HCPro of PC22 had an equally efficient silencing suppression activity as the wild type HCPro [76]. Interestingly, another mutation (RRH234-236AAA) in PPV HCPro located close to the WG motif affected CP stability, but did not affect its RNAsilencing suppression activity [34]. This mutation had an effect on PPV particle formation. Taken together, accumulated evidence strongly suggests the central region of potyviral HCPro close to the AGO1-binding motif is involved in long-distance movement and particle formation. While it is not possible to conclusively extrapolate the role of AGO1-HCPro interaction in PPV and TEV infections, we propose that it is central for these phenomenon in the cases of PVA and TuMV. number of epidermal cells per mm 2 containing YFP CP aggregates. Results are averages of three independent experiments, error bars denote standard deviation ( ��� P<0.001). C) CP co-immunoprecipitates with CFP AGO1 depending on the presence of HCPro. CFP AGO1 and CP were co-infiltrated with HCPro/ HCPro AG or GUS as control. Samples were collected at 3 dpi. GFP trap was used for affinity purification followed by western blot analysis of the eluates to detect CFP AGO1 (top panel) and associated CP (bottom panel). Validation of CP expression in the input is presented in S5 Fig. https://doi.org/10.1371/journal.ppat.1008965.g008

Potyviral HCPro employs AGO1 in proviral functions
The AGO family of RNA silencing effectors holds an established position in plant antiviral defense (reviewed in [8,21]). AGO1 was thought to be the major player in antiviral activities until an investigation of turnip crinkle virus and cucumber mosaic virus infections in mutant Arabidopsis plants revealed that AGO2 provides an important secondary layer of defense [40]. A study of TuMV infection in Arabidopsis revealed AGO1 contributed to antiviral defense mainly in systemically infected inflorescence tissues where its effects blended in with AGO10 activity [42]. Later, in [60], the authors inactivated N. benthamiana AGO2 by Crispr/Cas9 -mutagenesis and showed its importance in defense against potato virus X, TuMV and turnip crinkle virus. In a healthy plant AGO1 is prevalent and downregulates AGO2 expression via miR403 [61]. During an infection, when AGO1 is targeted by viral silencing suppressors, AGO2 expression is upregulated. This consequently increases its impact on host antiviral defense [40]. Interestingly, data obtained from the Arabidopsis-cucumber mosaic virus pathosystem implies AGO1 and AGO2 can also engage in non-redundant co-operation against the virus by specializing in distinct types of vsiRNAs [62].
The interdependent relationship between AGO1 and AGO2 is relevant for a balanced defense response. Thus, we measured the changes in AGO2 mRNA levels in our AGO1 silencing experiments. We found that indeed the expression of AGO2 mRNA was enhanced in AGO1-silenced samples (S1 Fig). It is possible that this results in a shift in the AGO1/AGO2 balance leading to the unmasking of AGO2-dependent defense during PVA infection. From this point of view the pro-viral effect of AGO1 would be indirect because the higher AGO2 levels could explain the negative impact the silencing of AGO1 had on the infection. Although virus-derived gene expression did not decrease significantly during AGO1 silencing, we acknowledge the possible additional effects of increased AGO2 activity in these circumstances.
An important viral strategy to subdue the host's silencing machinery is to encode proteins with silencing suppression activity. Apparently, due to its central position in the RNA silencing mechanism AGO1 has emerged as a major target for viral silencing suppressors. Data gathered from across plant virus families reveals that a wide range of silencing suppressor proteins interfere with host AGO1. For example, the serine proteinase P1 protein of sweet potato mild mottle virus (SPMMV; genus Ipomovirus; family Potyviridae) binds AGO1 via a WG-motif [49]. Instead of HCPro, P1 serves as the RNA silencing suppressor in SPMMV infection [49]. SPMMV HCPro does not contain either the AGO1-binding WG motif or the typical potyviral consensus sequence around it. P1N-PISPO protein encoded by sweet potato feathery mottle virus (SPFMV; genus Potyvirus) contains a WG-motif. The WG motif of PISPO is required for its silencing suppression activity [77], but it is not known if it is able to bind to AGO. On the other hand, SPFMV HCPro has retained the typical potyviral WG-motif consensus sequence. This proposes that the functions of AGO1-HCPro binding in infection in the close relatives of potyviruses, the ipomoviruses, may have been substituted with some alternative mechanisms. Protein p25 of turnip crinkle virus (family Tombusviridae) as well as 2b of cucumber mosaic virus (family Bromoviridae) bind AGO1 in order to inhibit its activity [44,47,49]. Interaction between the P0 protein of beet western yellows virus and cucurbit aphid-borne yellows virus (family Luteoviridae) and host AGO1 leads to AGO1 degradation via autophagy [45,46,50,51]. Likewise potato virus X (family Alphaflexiviridae) p25 and enamovirus (family Luteoviridae) P0 have been suggested to alleviate defensive pressure on the infection by directing AGO1 to degradation [48,52]. We used confocal microscopy-based co-localization studies and affinity purification to show that the potyviral HCPro indeed associates with AGO1. As the association was significantly reduced by the W208A mutation, we propose the two proteins associate with each other in vivo.
In the current study the systemic spread of two potyviruses, PVA and TuMV, was significantly reduced when the W208A mutation was introduced to the conserved WG motif in HCPro. The effect of the mutation on virus particles was more drastic as it nearly abolished the accumulation of virions (Fig 6). We demonstrate the mutation was detrimental to not only the interaction between PVA HCPro and AGO1 but also for enabling the association between AGO1 and CP. We interpret that poor CP stability in a PVA AG infection is indicative of its degradation due to reduced incorporation into particles or instability of the assembled particles. Based on this evidence we suggest that AGO1 makes positive contributions to potyvirus encapsidation under HCPro's coordination and that it has a key role in ensuring the stability of the particles.
The ability to separate AGO1-binding functions from other biochemical properties of the viral silencing suppressor is not a unique feature of HCPro: for example, AGO1-interaction of the cucumber mosaic virus 2b protein has also been shown to be independent of its silencing suppression activity [78]. Likewise, the silencing suppressor activity of the pelargonium line pattern virus (family Tombusviridae) coat protein p37 was independent of its ability to interact with AGO1 and AGO4 in N. benthamiana [59]. According to the study p37's ability to suppress silencing depended foremost on its sRNA-binding capacity. Our data supports the scenario where the lack of HCPro could result in vsiRNA enrichment within AGO1, which consequently targets vRNA as a slicer. Because the HCPro AG mutant retained its capacity to suppress silencing we expect it also sequesters vsiRNAs normally.
Particle production is a crucial step for the spread of the virus as encapsidation ensures the viral genome is protected from degradation. For many plant viruses, particles are the major form of long-distance movement (reviewed in [58]). For potyviruses virions might not be the only option for systemic movement as replication vesicles have been discovered both in the phloem and xylem [79][80][81]. We found that despite extremely low particle amounts detected by IC-RT-PCR and EM both PVA AG and TuMV AG were able to establish systemic infections, albeit with diminished efficiencies (Figs 5 and 6). Our results gave rise to two possibilities: either particles were not formed or the assembled particles were so unstable they were degraded immediately after cell lysis required by IC-RT-PCR and EM analyzes. In the former case, systemic spread could be explained by long-distance transport of vRNA in replication vesicles or as part of ribonucleoprotein complexes instead of as virus particles. The electron micrographs of the infected leaf tissues provided more evidence for the latter option. EM of thin sections of infected leaves showed that PVA AG produced particles which, however, appeared malformed and undersized. We conclude that the lack of the stabilizing effect of CP-AGO1-HCPro interaction makes the particles sensitive to cellular proteases and RNases, leading to disassembly of particles and degradation of CP. We propose that in spite of being distorted the stability of the both PVA AG and TuMV AG particles was enough to support longdistance movement in intact plants. However, the reduced efficiency of long-distance movement may reflect that less PVA AG and TuMV AG particles succeeded in protecting PVA RNA during phloem transport than PVA WT or TuMV WT particles.
The mechanism of potyviral encapsidation and the required components are not yet fully understood. Although the signal for and origin of encapsidation remain obscure, and may rely on protein-protein interactions rather than protein-RNA interaction [82], Gallo et al. [35] recently demonstrated that only replication-competent vRNA is encapsidated. Successful encapsidation of PPV requires CP stabilization by HCPro [34]. Recent research has also proposed an active role for CP in sequestering vRNA from translation to encapsidation [82]. Torrance et al. [83] used atomic force microscopy to study the detailed architecture of potyvirus particles. The high-resolution technique revealed that mature virions can contain a tip structure located at one end. In other plant viruses, such protein complexes have been linked to enhanced particle stability or controlled disassembly [84][85][86]. In PVA and potato virus Y the structures contained HCPro and VPg [83]. The authors proposed the complex forms at the 5' UTR of the vRNA because VPg's is covalently attached to this position. Also the CI, a viral helicase, was later found to associate with one end of potyvirus particles indicating that it could form part of the structure [87]. We have earlier sketched the concept of a virus-induced complex involving HCPro and AGO1 and proposed it could function in relieving translational repression [54]. Based on results presented here, we suggest HCPro not only attracts AGO1 to the complex on vRNA but also serves as a bridge between AGO1 and CP. We hypothesize that the interaction follows vRNA from translation to encapsidation. When correctly assembled the interface between CP and the RNP-complex or the "tip"-structure as in [83,87] needs to be sealed stably and HCPro-AGO1-CP interaction may be crucial in this function. When HCPro's capacity to interact with AGO1 was compromised by the W208A mutation, consequences for the infection were detrimental. We suggest that in PVA AG a weakened interaction between HCPro, AGO1 and CP could produce fragile particles with incomplete tip structures that lack the stabilizing effect of AGO1. The inability of PVA AG and TuMV AG to accumulate normal amounts of particles could be a consequence of this instability. Similar circumstances could occur during the silencing of AGO1 when host AGO1 levels are too low to support stable encapsidation.
In summary, potyviral HCPro's capacity to interact with AGO1 agreed with previous findings linking viral silencing suppressors to AGO proteins. WG/GW motif mimicry among viral silencing suppressors can be advantageous in directing host AGO proteins away from their antiviral functions. Our results, however, point to the direction that HCPro not only tames AGO1's aggressive antiviral activity but also employs it in pro-viral functions such as virus particle stabilization. The high conservation of the AGO1 binding site in HCPro suggests strong evolutionary pressure to maintain this site. The need for stable particles for aphid transmission may act as a driving force here. The finding is an interesting example highlighting the evolutionary arms race between viruses and their hosts. We anticipate that the emerging role for AGO1 in potyvirus infections calls for further studies in order to clarify the ongoing coevolution of plant defense and viral counter defense.

Plants and growth conditions
Nicotiana benthamiana plants were grown in a greenhouse environment under a 16 h light and 8 h dark cycle at 22 o C and 18 o C respectively with a relative humidity of circa 50%. Plants were used in the experiments at the four-to-six-leaf stage.

Construction of recombinant viruses
PVA constructs used in this study are based on the PVA-B11 full-length infectious cDNA (GenBank accession no. AJ296311) expressed under the cauliflower mosaic virus 35S promoter and tagged with the Renilla luciferase reporter containing a plant intron. To enable Agrobacterium-mediated expression in plants all virus constructs were incorporated into the pBIN19 binary vector (GenBank: U09365.1). PVA WT and the replication-deficient PVA ΔGDD were described in [55] while the HCPro-less viruses PVA ΔHCPro and PVA ΔGDDΔHCPro were described in [39].
The Agos computational tool, available online at http://www.combio.pl/agos/, was used to detect putative ARGONAUTE-binding sites in the PVA HCPro sequence [56,57]. The W208A mutation was introduced into the HCPro of PVA using the site-directed mutagenesis method described in [88]. A pGEM-T Easy vector (Promega) carrying the SexAI-NruI -fragment (contains the P1 C-terminal, HCPro, P3 N-terminal) from the PVA icDNA was used as the template in two separate PCR reactions, one with the forward primer 5'-CAACTTGATA GCGGGCGAGAGAG -3´and the other with the reverse primer 5´-CTCTCTCGCCCGCT ATCAAGTTG-3´. The mutated codon is designated in bold. The Phusion high fidelity polymerase (Thermo Fisher Scientific) was used for all PCR-based cloning steps.

Potyviral HCPro employs AGO1 in proviral functions
The presence of the W208A mutation was verified by sequencing and the SexAI-NruI fragment was then used to replace the wild type fragment in PVA icDNA.
TuMV constructs are based on the UK1 strain icDNA expressed under the cauliflower mosaic virus 35S promoter in the pCB301 binary vector backbone (GenBank accession no: EF028235.1). The recombinant TuMV construct carrying the GFP reporter between P1 and HCPro has been described in [89]. Additionally the expression plasmid contains the mCherry reporter gene under an independent 35S promoter within the bordering T-DNA sequences [90].

Overexpression constructs
To generate a transient overexpression construct of TwinStrep-tagged RFP fusion of the PVA HCPro AG the W208A mutation was introduced into a P1-TwinStrep-RFP-HCPro-P3 fragment in pGEM-T Easy (Promega) with site-directed mutagenesis using the oligos described above. The sequence was then PCR amplified with oligos adding an XhoI restriction site and a start codon to the 5´end and a stop codon and a BamHI restriction site to the 3´end (Forward oligo 5´-TCCGCTCGAGATGAAATGGTCTCATCCACAA-3´and reverse 5´-CGCGGAT CCTCATCCAACCCTGTAGTGCT-3´). The PCR product was digested with XhoI and BamHI and ligated into an XhoI-BamHI double-digested interim vector (pHTT690) to provide the 35S promoter and nos terminator. After sequence verification the 35S-TwinStrep-RFP-HCPro AG -nos expression cassette was transferred into the pBIN19 binary vector using HindIII restriction sites.
To generate a YFP-tagged CP, PVA CP gene was amplified from the PVA icDNA with primers incorporating Gateway-compatible attB sequences. The PCR product was cloned into pDONRZeo and finally into destination vectors using Gateway BP and LR reactions, respectively (Thermo Fisher Scientific). pGWB42 was used as the destination vector to generate an N-terminal fusion of the CP. The pGWB series binary expression vectors have been described in [92]. The CFP AGO1 expression construct was as in [39].

Agroinfiltration
All constructs were delivered into Agrobacteria by electroporation. Agrobacterium strain C58C1 was used as the host bacterium for introducing PVA, protein overexpression and silencing constructs into N. benthamiana plants. Agrobacterium strain GV3101 (pMP90) was used as the host bacterium for TuMV constructs. Agrobacteria harboring the desired plasmids infiltrated into N. benthamiana leaves as described previously [54].

Dual luciferase assay
The dual luciferase assay was performed with the Dual-Luciferase Reporter Assay System (Promega E1910/E1960) as in [55].

SDS-PAGE and western blotting
The separation of proteins on SDS-PAGE gels and western blotting were done according to standard procedures. HCPro and CP were detected with polyclonal antibodies produced in rabbit. An HRP-conjugated anti-rabbit antibody was used as the secondary antibody (Promega; W4011). CFP and YFP and their fusions were detected with a mouse monoclonal GFP antibody (Santa Cruz; SC-9996) and RFP with a mouse monoclonal RFP antibody (Signal-Chem; R46-61M-100). RLUC was detected with a mouse monoclonal antibody (Millipore; MAB4400). The secondary antibody for the monoclonal antibodies was an HRP-conjugated anti-mouse antibody (Promega; W4021).

RNA extraction and qRT-PCR
For gene expression analysis total RNA was extracted from 100 mg leaf samples with the Gene Jet Plant RNA Purification kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Prior to cDNA synthesis contaminating T-DNA molecules carrying PVA cDNA and genomic DNA carrying the PP2A and F-box reference genes were removed with RQ1 RNase free DNAseI (Promega) treatment at 37 o C for 30 min. cDNA was synthesized with the RevertAid H Minus First Strand cDNA Synthesis kit (Thermo Fisher Scientific) as recommended by the manufacturer using random hexamers as primers. Quantitative real-time PCR was done using the Maxima SYBR Green qPCR Master Mix (Thermo Scientific). PVA CP was amplified with 5´-CATGCCCAGGTATGGTCTTC-3´and 5´-ATCGGAGTGGTTGCAGTG AT-3´and TuMV CP with 5´-TGGCTGATTACGAACTGACG-3´and 5´-CTGCCTAAA TGTGGGTTTGG-3´. N. benthamiana PP2A and Fbox were used as reference genes [14] and were amplified with 5´-GACCCTGATGTTGATGTTCG-3´and 5´-GAGGGATTTGAAGAG AGATTTC-3´(PP2A) and 5´-GGCACTCACAAACGTCTATTTC-3´and 5´-TGGGAGGC ATCCTGCTTAT-3´(Fbox). Results were normalized to reference gene expression with the ΔΔCq method.

Immunocapture of virions and RT-PCR
For IC RT-PCR tubes were coated with the anti-CP antibody in ELISA coating buffer for 3 h at 37 o C followed by washing with ELISA washing buffer. 100 mg leaf samples were homogenized, resuspended in ELISA sample extraction buffer and allowed to settle on ice for 1 h. Samples were incubated in the CP-antibody-coated tubes overnight at 4˚C. After incubation the tubes were washed twice with ELISA washing buffer, once with 1 × PBS and once with distilled water. cDNA synthesis was primed with random hexamers and the tubes were heated for 5 min at 65 o C to disrupt bound virions. The rest of the components of the RevertAid H minus First Strand cDNA synthesis kit (Thermo Fisher Scientific) were then added and cDNA was synthesized according to the manufacturer's protocol. The cDNA was used as template in qPCR to determine vRNA copy number originating from captured particles. Copy numbers were calculated based on a linear standard generated from serial dilutions of a plasmid template in the same qPCR run.

Affinity purification and co-immunoprecipitation
HCPro/HCPro AG and interacting partners were immunoprecipitated by affinity purification adapted from the method described in [93].
N. benthamiana plants were infiltrated with Agrobacteria carrying HCPro or HCPro AG as bait plus CFP AGO1 as target or CFP as a negative control. Each overexpression construct was infiltrated at OD 600 0.1 and samples were collected at three dpi. 1 g of frozen leaf material was

PLOS PATHOGENS
Potyviral HCPro employs AGO1 in proviral functions ground in liquid nitrogen and resuspended in 3 ml of binding buffer (25 mM Tris-HCl pH 8.0, 550 mM NaCl, 5 mM NaF, 0.5 mM EDTA, 10% Glycerol, 0.1 mM PMSF supplemented with Pierce protease inhibitor tablets). Samples were then cleared by filtering through doublelayered Miracloth followed by a 5 min centrifugation at 5000 × g at 4˚C. To minimize the binding of endogenous biotinylated proteins to the StrepTactin resin, samples were supplemented with 100 μg/ml avidin and incubated for 15 min at 4 o C in rotation. The cross-linked StrepTactin was mixed by vortexing before adding 50 μl of resin per sample. Samples were rotated for 2 h at 4 o C to allow binding of the bait proteins to the resin. Recovered resin was washed thrice in wash buffer #1 (25 mM Tris-HCl pH 8.0, 500 mM NaCl, 5 mM NaF, 0.4 mM EDTA, 0.2% Igepal CA-630, 5% Glycerol, 0.1 mM PMSF) and once in buffer #2 (25 mM Tris-HCl pH 8.0, 150 mM NaCl) using spin column tubes (Costar Spin-X 0.45 um cellulose). The samples were eluted in 25 mM Tris-HCl pH 8, 1% SDS buffer for 5 min at 55 o C and 1400 rpm shaking. The bait protein and co-immunoprecipitated targets were detected from the eluates by SDS PAGE followed by western blotting. CFP AGO1 was affinity purified similarly by using a GFP-trap (Chromotek) according to the manufacturer's instructions.

Confocal and epifluorescence microscopy
Viral and host proteins tagged with fluorescent markers were transiently expressed in N. benthamiana. All constructs were agroinfiltrated at an OD 600 of 0.1 and infiltrated leaves were examined at 3 dpi. The localization of fluorescent fusion proteins was studied with confocal laser scanning microscopy using a Leica TCS SP5II or SP8 instrument (Leica Microsystems). All images were acquired with a 63x water immersion objective and in sequential scanning mode in order to avoid fluorophore bleed-through. YFP and RFP were excited with a 488 nm Argon laser and emission was detected at 522-552 nm and at 566-618 nm, respectively. CFP was excited with a 405 nm laser and emission was detected at 461-502 nm. Leaf samples were mounted upside down on an objective slide with a drop of water and placed under a cover glass.
Co-localization analysis of CFP AGO1 with HCPro and HCPro AG was performed with the ImageJ image analysis software. HCPro-containing granules were selected as ROIs and colocalization threshold function was used to calculate the amount of co-localized pixels from the RFP and CFP channels. Results are given as % of HCPro/HCPro AG co-localizing with CFP AGO1. Epifluorescence microscopy of N. benthamiana epidermal cells was done with a Zeiss Axioskop2 plus instrument using a FITC filter to detect YFP fluorescence.

Transmission electron microscopy
Virus particles were visualized by negative staining followed by transmission electron microscopy with a Jeol JEM-1400 instrument (Jeol Ltd., Tokyo, Japan). 100 mg leaf samples were homogenized and suspended in cold 0.06 M phosphate buffer. While letting the samples clear on ice, carbon-coated EM-grids were coated for 1 h at RT with an anti-CP antibody (PVA) at 1:100 dilution. The grids were then washed once with phosphate buffer and incubated for 5 min with cleared sample lysate. After washing with 20 drops of phosphate buffer the grids were stained for 15 s with 2% uranyl acetate. EM images of the infected leaf tissues were visualized following Lõhmus et al. [94].  ) and HCPro AG (in blue) were independently made using I-TASSER protein prediction tool and server. Models consistent to TuMV HCPro crystal structure (PDB id-3RNV) were overlaid using Chimera protein structure visualization software. Both the structures matched well and positions of several important functional motifs of HCPro (in yellow) were intact. Finally, WG (in green) and its mutated counterpart AG (in red) were also found to be in close vicinity, however, their molecular structure varied as amino acid 'W' was replaced with 'A' (in the box). (TIF) S1