The CYVMV infectious clone has been described previously (Jailani et al., 2016). In the silencing suppressor activity test, wild-type Nicotiana benthamiana leaves and the Agrobacterium strain GV3101 were used for agroinfiltration assays as previously described (Zhai et al., 2021). Moderate GUS expression in N. benthamiana leaves was driven by the pG_N/G_C-CBSm promoter (Zhai et al., 2019). The GUS expression can be specifically silenced by the RNAi construct pRNAi-GG-GUS (Yan et al., 2012). The known virus-silencing suppressor HC-Pro was used as a positive control to suppress the GUS silencing effect of pRNAi-GG-GUS. Individual constructs expressing V2, C2, C4, or βC1 were co-agroinfiltrated to test their silencing suppressor activity. Empty vector (EV) was used as a negative control. Agroinfiltration combinations included pG_N/G_C-CBSm-GUS only, pG_N/G_C-CBSm-GUS + pRNAi-GG-GUS, pG_N/ G_C-CBSm-GUS + pRNAi-GG-GUS + HC-Pro, and pG_N/G_C-CBSm-GUS + pRNAi-GG-GUS + V2, C2, C4, or βC1. For vector construction, the coding sequences of V2, C2, C4, or βC1 were amplified by PCR using the Phusion proofreading DNA polymerase (Invitrogen, Carlsbad CA, United States) and virus-specific primers with attB adaptors. Each amplicon was first cloned into the pENTR/D-TOPO Gateway vector and then inserted into pEarleyGate103 (The Arabidopsis Information Resource, Arabidopsis Biological Resource Center) via the LR reaction. For agroinfiltration, Agrobacterium pellet from 50 ml culture was resuspended in the infiltration buffer (50 mM MES pH 5.6, 10 mM MgCl₂, and 150 μM acetosyringone) and adjusted to an OD₆₀₀ of 0.5. Four leaves from the same N. benthamiana plant were infiltrated with different combinations. For each combination, equal volumes of different cultures with same OD₆₀₀ were mixed and used in agroinfiltration. Three biological replicates were taken for each experiment. All the primers used in this study are listed in Supplementary Table 1.

β-Glucuronidase staining was performed as described previously (Jefferson et al., 1987). For the quantification of the GUS enzyme, fresh leaf samples inoculated with different constructs mentioned earlier were collected using a puncher and homogenized to fine powder by grinding in liquid nitrogen. For each sample, 200 μl GUS extraction buffer (50 mM NaHPO₄, pH 7.0, 5 mM dithiothreitol, 10 mM Na₂EDTA, 0.1% sodium lauryl sarcosine, and 0.1% Triton-100) was added, and 100 μl extract was mixed with 40 μl 10 mM fluorescent substrate of 4-methylumbelliferyl β-D-glucoside (MUG) (Sigma Life Science, St. Louis, MO, United States). After incubation at 37°C for 1 h, the reaction was stopped with 1 ml stop buffer (0.2 M sodium carbonate). After mixing well with the stop buffer, 100 μl of mixture was loaded onto a 96-well black microtiter plate (Greiner Bio-One, Monroe, NC, United States). Two replicate wells were carried out for each sample. The methylumbelliferone (MU) (Sigma Life Science, St. Louis, MO, United States) standard curve was included in each plate. Furthermore, 2.5 μM, 625 nM, 156 nM, and 39 nM MU gradients were used to generate the standard curve. The fluorescence intensity of MU was measured in a Molecular Devices SpectraMax M2 microplate reader (Molecular Devices Co., Sunnyvale, CA, United States).

V2, C2, C4, and βC1 genes were cloned into pEarleyGate103 with a C-terminal green fluorescent protein (GFP) fusion tag. These GFP fusion constructs were infiltrated individually or co-infiltrated with the CYVMV infectious clone into N. benthamiana leaves. After 3 days, infiltrated leaves were examined by a Leica confocal laser-scanning microscope to determine the subcellular localizations of the GFP fluorescent signal as described earlier (Sarkar et al., 2021). Leaves were incubated in 5 ng of Hoechst 33342 per ml of 1 × PBS buffer 20 min prior to imaging. A sequential scan was used to visualize the nucleic acid binding of Hoechst with excitation at 405 nm, and emission was collected from 420 to 470 nm; GFP fluorescence was visualized with an excitation of 488 nm and emission collected from 500 to 600 nm. Leaves were incubated in 1% aniline blue in tap water for 10 min to visualize plasmodesmata. A sequential scan was used with excitation of 405 nm and emission from 435 to 480 nm for aniline blue. The bright-field image to show cell walls was collected with a photomultiplier tube (PMT) detector, and GFP fluorescence of V2 CYVMV was imaged as above. In all GFP experiments, at least three individual plants inoculated with corresponding GFP constructs were used for confocal microscopy. For each plant, at least two leaves with at least three different areas from each leaf were chosen for the analysis. A representative image from each treatment was shown.

Yeast two-hybrid (Y2H) assay was described previously (Peng and Neff, 2020). The coding sequences of V2, C2, C4, or βC1 were separately cloned into the Gateway compatible bait vector pBTM116-D9 and the prey vector pACT2-GW, respectively. The prey constructs were used to transform yeast strain A. For each event, a clone with the lowest self-activation was selected and transformed with the corresponding bait construct. The empty bait vector was used as a negative control. Positive protein-protein interactions were demonstrated by the ability of yeast clones to grow in the synthetic dropout IV (SDIV) media with uracil, histidine, leucine, and tryptophan being dropped. The known interaction between CIRCADIAN CLOCK ASSOCIATED 1 and ATAF2 (Peng and Neff, 2020) was used as a positive control. PCR-amplified sequences in all constructs used in this research were verified by sequencing.

The coding sequence of V2 was cloned into pSITE-cEYFP-N1 and pSITE-nEYFP-C1, respectively. Both constructs were then introduced into Agrobacterium tumefaciens GV3101 for N. benthamiana leaf infiltration. After a 48-h incubation, leaves were examined using a Leica TCS SP8 X confocal laser-scanning microscope, and the LAS X software (Leica Microsystems, Wetzlar, Germany) was used for image processing. The protein-protein interaction was indicated by the restoration of yellow fluorescent protein (YFP). YFP fluorescence was visualized with an excitation of 514 nm and emission collected from 530 to 600 nm. The bright-field signal to show the cell walls was collected with a PMT detector.

The pull-down assay was performed as described previously (Peng and Neff, 2021). V1 and V2 proteins were fused with 6 × histidine (His) and maltose-binding protein (MBP) tags on their C- and N-terminals, respectively. Purified V1-His and MBP-V2 were mixed and passed through the amylose resin (New England Biolabs, Ipswich, MA, United States). After washing off unbound proteins, MBP-tagged proteins and their binding proteins were eluted for four times (i.e., E1 to E4) using elution buffer containing 10 mM maltose. As a negative control, purified V1-His and MBP tag protein were mixed and went through the identical process. Eluted proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel and stained with Coomassie brilliant blue R-250.

To create infectious knockout constructs, it is essential to create partial tandem dimeric repeat constructs where the gene to knockout is absent. To develop such constructs, initially, the intergenic region (IR) of the CYVMV was amplified from the genome of CYVMV using primers AR11 and AR12 and cloned into pGreenII vector using HindIII and EcoRI to create a pGreenII-IR construct. Similarly, a V2-IR region was amplified using AR11 and AR15 primers and cloned into the pGreenII vector using HindIII and EcoRI to create a pGreenII-V2-IR construct. These two initial constructs served the partial dimeric portion of the knockout constructs. Furthermore, a V2 deleted portion of CYVMV was amplified using primer combination AR13 and AR14 and cloned into pGreenII-IR using SpeI and NotI to generate V2-knockout construct of CYVMV (CYVMV-KoV2). Similarly, a V1 deleted portion of the CYVMV was amplified using primers AR17 and AR19 and cloned into pGreenII-V2-IR using SpeI and NotI to generate V1-knockout construct of CYVMV (CYVMV-KoV1). Next, a portion of the CYVMV genome containing the IR and C1, C2, and C3 ORFs was amplified using AR17 and AR14 and cloned into pGreenII-IR using SpeI and NotI to generate a double-knockout construct of V1 and V2 ORFs of CYVMV (CYVMV-KoV1-V2). All three knockout constructs were designed to retain the multiple cloning sites of EcoRI, XmaI, SmaI, and SpeI from pGreenII-IR. CYVMV-KoV1-GFP and CYVMV-KoV2-GFP constructs were generated by amplifying the GFP gene using AR20 and AR21 primers and cloning it into EcoRI/SmaI sites of the corresponding knockout constructs. The GFP expression was driven by the viral promoter in the IR region. Primer details are given in Supplementary Table 1. Primer locations and construct maps are illustrated in Supplementary Figure 1.

To understand the replication behavior of these constructs and the symptom phenotype produced by them, all the knockout constructs alone as well as with GFP were agroinfiltrated into N. benthamiana leaves. Inoculation of CYVMV agroinfectious clone served as a control. At 15 dpi, DNA was isolated from the agroinfiltrated tissue and systemically infected leaves. PCR was carried out using a set of abutting primers, i.e., AR40 and AR41, which were used to amplify the viral replicon. GFP epifluorescence was measured under UV light, and the intracellular GFP expression was analyzed using confocal microscopy. Maps of constructs and deletion mutants, episomal replicon released from the plasmid vector backbone, and locations of the abutting primers are shown in Figure 8.

A set of five N. benthamiana plants were agroinfiltrated individually for each of the suppressor protein constructs in pEarleyGate103 vector; their impact on symptom development, if any, was monitored regularly. To assess the transportation of the transcripts of the suppressor gene constructs in the systemically infected leaf tissues, RNA was isolated from the systemically infected leaves of the suppressor construct-inoculated plants, and a reverse-transcriptase PCR (RT-PCR) was performed for pulled samples of three systemically infected leaves of all the five replications corresponding to all of the four suppressor-inoculated plants. The expression of the suppressor proteins in the systemically infected leaves was analyzed through confocal microscopy as described earlier.

Nicotiana benthamiana plants showed the leaf curl symptom 10 days after co-agroinfiltration of a partial tandem repeat construct of CYVMV and a dimeric construct of CroYVMB. V2 (357 bp), C2 (405 bp), and C4 (258 bp) genes from CYVMV and βC1 (357 bp) gene from CroYVMB were amplified (excluding their stop codon) using the DNA isolated from these agroinfiltrated plants. Amplicons were cloned and sequenced to investigate the putative RNA silencing suppressor function of their encoded products. Transient silencing suppression assays were conducted on four leaves of the same N. benthamiana plant. After infiltration with Agrobacterium carrying pG_NG_C-CBSm:GUS (i.e., GUS), the leaves showed a moderate level of GUS expression. In contrast, dramatically reduced GUS staining signals were observed in leaves co-infiltrated with Agrobacterium carrying pG_NG_C-CBSm:GUS and pRNAi-GG-GUS (i.e., GUS-hp), indicating the suppression of the GUS expression by RNA silencing. When added with the third co-infiltrator of V2, C2, C4, or βC1 harbored by pEarleyGate103 (i.e., GUS + GUS-hp + V2, GUS + GUS-hp + C2, GUS + GUS-hp + C4, or GUS + GUS-hp + βC1), GUS signals were restored in all infiltrated leaves, demonstrating the silencing suppressor function of these four proteins (Figures 1A–D). The silencing suppressor HC-Pro (GUS + GUS-hp + HC-Pro) was used as a positive control for the restoration of the GUS signal. In the negative control assay, co-infiltration of the EV pEarleyGate103 did not restore GUS signal that was suppressed by GUS-hp (Figure 1E).

Quantitative GUS expression assays were carried out to determine the silencing suppressor activity of V2, C2, C4, and βC1 (Figures 2A–D). All N. benthamiana leaves infiltrated with pG_NG_C-CBSm:GUS (i.e., GUS) showed similar levels of GUS accumulation, which was significantly reduced by the co-infiltration of pRNAi-GG-GUS (GUS + GUS-hp). Consistent with our qualitative results, additional co-infiltrations of V2, C2, C4 or βC1 (GUS + GUS-hp + Suppressor) all suppressed RNA silencing and thereby restored GUS accumulation, with V2 being the most effective silencing suppressor (∼85% GUS restoration efficiency). The restoration values with C2, C4, and βC1 were around 50–60%. As a positive control, HC-Pro consistently restored the GUS signal (Figures 2A–E). In contrast, EV exhibited no GUS restoration effect (Figure 2E).

The nucleus and cytoplasm localizations of silencing suppressors may indicate their suppression effect related to TGS and PTGS, respectively. Hence, the subcellular localizations of V2, C2, C4, and βC1 were examined using their GFP fusion proteins in the absence and presence of CYVMV. In our assays, enhanced green fluorescent protein (eGFP)-tagged V2 (V2-GFP) alone was found to aggregate as large clumps in various locations within the cytoplasm, mainly surrounding the nucleus (Figures 3A,B). Co-infiltration of V2-GFP and CYVMV showed a weaker GFP signal loosely associated with the cell wall and moved to reach the plasmodesmata (Figures 3C,D). C2-GFP alone diffused in the nucleus, with most of the proteins located in the nucleoplasm (Figures 4A,B). The presence of CYVMV did not dramatically change the subcellular localization pattern of C2-GFP except that its accumulation in the nucleolus was almost abolished (Figures 4C,D). Furthermore, the nucleus of C2-GFP and CYVMV co-infiltrated plants showed a slight dilation in size and deformation of spherical shape. The C4-GFP localization appeared to be consistently in the plasma membrane with or without CYVMV inoculation. CYVMV can induce the aggregation of C4-GFP into small particles within the membrane (Figure 5). Both βC1-GFP and βC1-GFP + CYVMV showed a strong GFP fluorescent signal in the nucleus (Figure 6). Overall, only V2 localization was affected by CYVMV. It is possible that V2 interacts with one or more other viral proteins or with host factors that are altered by infection and even with the CYVMV genome. Such interactions may facilitate V2 relocalization and subsequent cell-to-cell transfer via the plasmodesmata.

Whether the CYVMV silencing suppressors are capable of physically interacting with one another and themselves are yet to be investigated. Virus might use such interactions for their own replication and spread. Thus, potential self- or mutual interactions among V2, C2, C4, and βC1 were tested by targeted Y2H assays (Figures 7A–E). Only the self-interaction of V2 was observed in all combinations (Figure 7A). This result indicates that V2 may fulfill its silencing suppressor function in the form of a homodimers or multimers. C2, C4, and βC1 may suppress RNA silencing individually with no involvement of self-multimerization.

To further confirm the Y2H result, in vivo self-interaction of V2 was tested using bimolecular fluorescence complementation. The full-length V2 gene was cloned to fuse with N- and C-terminal fragments of YFP and co-infiltrated onto the N. benthamiana leaves. Infiltration with V2-cYFP + nYFP-V2 constructs resulted in strong yellow fluorescent signals (Figure 7F). As negative controls, no fluorescence was detected in leaves infiltrated with V2-cYFP + nYFP-empty (Figure 7G) or empty-cYFP + nYFP-V2 (Figure 7H). These results confirmed the self-interaction of V2 in planta.

Since subcellular localization results (Figure 3) suggest that V2 may interact with other viral protein(s) and involve in the CYVMV cell movement, its protein binding capacity was further investigated. In a targeted Y2H assay, CYVMV V1 and V2 proteins physically interact with each other (Figure 7I). The interaction was confirmed by a pull-down assay (Figure 7J). The physical interaction between V1-His and MBP-V2 ensured their co-elution from the amylose resin.

Since V1 and V2 interact, we investigated whether such an interaction is necessary for the cell-to-cell movement of the virus. V1/V2 single and double deletion (CYVMV-KoV1, CYVMV-KoV2, and CYVMV-KoV1-V2) constructs were created to infect N. benthamiana plants followed by molecular assays. All three deletion CYVMVs were detected only in infiltrated leaves (Figure 8) when using the abutting primers designed from C1 ORF. In contrast, wild-type CYVMV was present in both infiltrated and systemically infected leaves (Figure 8A). Abutting primers are a set of divergent primers whose 5′ end has two consecutive nucleotides in two strands. They specifically amplify the circular viral genome from the outward direction (Figure 8). We used such divergent primers to investigate whether the virus mini genomes (deletions of either V1 or V2 or both ORFs) can be released from the vector backbone and start replication as an episome. GFP-fused V1 (CYVMV-KoV1-GFP) and V2 (CYVMV-KoV2-GFP) deletion constructs were created for infiltrating individual N. benthamiana leaves. The GFP expression was monitored through epifluorescence and confocal imaging. Deletion of either V1 (Figures 9A,C) or V2 (Figures 9B,D) restricted GFP signals within the infiltration sites. Taken together, these results above demonstrate that the absence of either V1 or V2 inhibits the cell-to-cell CYVMV movement but not virus replication (Figures 8, 9).

V1 and V2 deletion constructs were complemented with co-inoculation of V2 and V1 constructs driven by the same right-side bidirectional promoter, respectively. Both complementation efforts failed to restore the movement function (data not shown). Since V1 and V2 ORFs were carried by two different constructs for complementation, the C2 protein might be incapable of activating the promoters of individual V1 and V2 ORFs to generate enough transcripts, which could lead to the complementation failure.

Since silencing suppressors often act as symptom determinants when expressed alone, pathogenicity tests were performed for the four CYVMV silencing suppressors. Levels of pathogenicity induced by V2, C2, C4, or βC1 were determined via agroinfiltrating their respective overexpression constructs (fused with eGFP in pEarleyGate103) to N. benthamiana leaves. Five plants were infiltrated for each construct. Four, two, and three out of five infiltrated plants showed variable leaf curl symptoms with regard to V2-GFP, C2-GFP, and C4-GFP constructs, respectively (Table 1 and Figures 10Ai–iii). In contrast, none of the βC1-GFP-infiltrated plants exhibited any symptoms (Table 1 and Figure 10Aiv). All five plants infiltrated with CYVMV and CroYVMB constructs (i.e., positive control) also showed the similar leaf curl symptom (Figure 10Av), which was not observed in the mock infiltrated plants (i.e., negative control) (Table 1 and Figure 10Avi). Similar to the positive control, transcripts of all four silencing suppressor genes could be detected in systemically infected leaves of their respective infiltrated plants regardless of the symptomatic or asymptomatic phenotypes (Figures 10Bi–v). No amplification was obtained from the negative control (Figure 10Bvi). Consistently, the expression of the respective GFP-tagged suppressor proteins in all systemically infected leaves was found by confocal microscopy (Figures 10Ci–iv) on which no wild-type virus was inoculated. As expected, no GFP fluorescence was observed in either positive or negative controls (Figures 10Cv,vi).