It has previously been reported ^[16] that expressed p25 fused with GFP at its N-terminus (GFP-p25) formed rod-like structures in both PVX-infected and uninfected cells. In our experiments, the structures also occurred when GFP fused with p25 at its C-terminus (p25-GFP) was expressed either alone under the control of CaMV 35S promoter or from the PVX-Δp25 vector with the duplicated coat protein subgenomic promoter (Fig. 1A and B). To study the process of formation of the structure, we expressed p25-GFP by agroinfiltration and detected the subcellular distribution of fluorescence 1, 2 and 3 days post infiltration (dpi) under the confocal microscope. At 1 dpi, p25-GFP fluorescence was distributed evenly at the cell periphery and there were also a few sporadic spots in the cytoplasm (Fig. 1C). At 2 dpi, the fluorescence on the cell periphery had diminished, while single fluorescent rods with intense fluorescence had started to appear (Fig. 1D). These rods moved within cells; a series of photos shows one fluorescent rod approaching and merging with a stationary one at 2 dpi (Fig. 1 D1–4). At 3 dpi, the fluorescence on the cell periphery was barely detected, while the RL structure was clearly formed (Fig. 1E). These observations suggest that the RL structure is progressively assembled from p25 protein.

The structure formed from GFP-fused p25 always showed intense fluorescence, and appeared to be composed of thick fluorescent rods when examined at a single focal plane under confocal microscopy. To examine the structure in detail, we decreased the strength of the exciting laser and monitored and recorded the fluorescence at a serial of focal planes and then combined them together. The combined figure showed that the structure was in fact composed of thin filaments (Fig. 1F) even though it looked like a thick rod when it was monitored at a single focal plane with a strong exciting laser. This is consistent with earlier reports from electron microscopy that the p25 formed “beaded sheets” ^[17].

To determine whether the RL structure is associated with the movement or suppression functions of p25, we attempted to interfere with the formation of the RL structure by fusing an ER or NLS location signal at its amino terminal. NLSp25 expressed by agrobacterium remained a silencing suppressor but did not form an RL structure (Fig. 2A). In contrast, ERp25 could still form an RL structure (although its structure was not perfect) but its suppression of RNA silencing was greatly diminished (Fig. 2A, Fig. S1). Furthermore, ERp25, but not NLSp25, recovered the cell-to-cell movement of PVX-GFPΔp25 with the heterologous silencing suppressor p19 (Fig. 2B; the average diameter of thirty infection loci was about 700 µm). These results imply that the ability of p25 to form the RL structure is necessary for its cell-to-cell movement function, but not for its ability to act as a suppressor.

To confirm these findings, twelve single amino acid mutants of p25 reported previously were used for analysis ^[15]. Eleven of the mutants, N94S, P111L, T117A, P122S, K124E, K153E, K153I, T193A, V195M, T214A and Y221H, are deficient in both virus movement and silencing suppression (double functional deficient or DFD mutants), and of these T117A and Y221H are reported to recover the cell-to-cell movement of PVX-GFPΔp25 when co-expressed with another suppressor. The twelfth mutant, A104V, is deficient in movement but not in suppression. We examined the RL structure of each mutant fused with GFP in agrobacterium-infiltrated cells. The RL structure was not formed in cells expressing GFP-fused A104V, which supported the view that formation of the structure was not necessary for p25 to act as a suppressor (Fig. 3). Among the DFD mutants, typical RL structures were seen where GFP was fused to T117A or Y221H, some slender RL structures were seen with P122S, but there were no structures with the other mutants (Fig. 3). Thus, with the exception of P122S, only those DFD mutants with the conditional ability to recover movement function formed the RL structures. This gives further support to the conclusion that the ability of p25 to form the RL structure is necessary for its function in cell-to-cell movement but not for its ability to act as a suppressor (Fig. 3). We then re-investigated the reported inability of mutant P122S to recover the movement of PVX-GFPΔp25 when co-expressed with another suppressor. Results from more than three repeats showed that when P122S was co-expressed with p19, fluorescence was visible in 6–10 cells at almost all loci whereas in the negative control it was limited to a single cell. This indicated that P122S could indeed recover the movement of PVX-GFPΔp25 when co-expressed with the suppressor p19, although the recovery was weak and the helper virus moved only into limited layers of neighboring cells (Fig. 3; the average diameter of thirty infection loci was about 200 µm, compared to the positive control of about 700 µm). Hence, the result from P122S also supported the conclusion above.

The microfilaments and microtubules that form the host cytoskeleton are implicated in the movement of both plant and animal viruses. It has been reported that treatment with the microfilament inhibitor latrunculin B (LatB) severely limited the spread of PVX in plants ^[18]. Since the RL structure is necessary for PVX movement, we next examined the relationship between microfilaments and RL structures. Tobacco epidermal cells were treated with different concentrations of LatB (5 µM, 10 µM and 20 µM) for 3 h before infiltration with the agrobacterium containing the vector expressing GFP-fused p25. At 3 dpi, the treated cells were examined under the confocal microscope. No complete RL structures, but only single fluorescent filaments, were seen in cells treated with LatB at any concentration, while the entire normal-looking RL structure was seen in the control cells treated with DMSO (Fig. 4). The RL structure was not affected by treatment with LatB 3 days after p25 had been expressed, showing that microfilaments were necessary during the formation of the RL structures, but not after they were formed. The microtubule inhibitor oryzalin did not affect the formation of the structure, whether treated before or after p25 expression (Fig. 4) suggesting that microtubules are unnecessary for RL structure formation.

Since both TGBp2 and TGBp3 are known to be necessary for PVX movement, we fused them individually with RFP and co-expressed them with GFP-fused p25 in tobacco epidermal cells by agroinfiltration. Examination of the fluorescence 3 dpi showed that both RFP-fused TGBp2 and TGBp3 formed granules as described in previous reports and that almost all these granules were recruited into the RL structure and not merely attached to the outside (Fig. 5A, B). At high resolution, it could be seen that red granules aligned on the filaments of the RL structure, hinting that p25 was associated with both TGBp2 and TGBp3 (Fig. 5 A, B). Because no detectable interactions have been reported between these proteins in previously reported yeast two-hybrid (YTH) experiments ^[16], we then used bimolecular fluorescence complementation analysis (BiFC) to analyze the interactions between p25 and TGBp2 or TGBp3. BiFC is a powerful tool for studying protein-protein interactions in living cells, and its results are better than YTH in reflecting natural interactions. Interactions between TGBp1 and TGBp2 of the potexvirus BaMV have previously been reported using this technique ^[19]. Here, yellow fluorescence (recorded as green during imaging) was seen at 3 dpi in the cells co-expressing pCV-n(c)YFP-p25 and either pCV-c(n)YFP-TGBp2 or pCV-c(n)YFP-TGBp3 (Fig. 6), but not in the control cells co-expressing pCV-n(c)YFP-p25 and pCV-c(n)YFP-pΔGUS, or pCV-c(n)YFP-TGBp2 and pCV-n(c)YFP-pΔGUS, or pCV-c(n)YFP-TGBp3 and pCV-n(c)YFP-pΔGUS (Fig. 6). This suggests that there are indeed interactions between p25 and both TGBp2 and TGBp3 in plant cells.

Construction of all plasmids followed standard cloning techniques. Escherichia coli strain TG1 was used for transformation ^[24]. All constructed plasmids are shown in Fig. 7. pCV:GFP, pCV:RFP were as described previously ^[25]. p25 which was introduced into pGEM-T (Promega, Madison, WI) with forward primer ([gene: 5′-TCTAGAATGGATATTCTCATCAGTAGTT -3′], Xba I site underlined) and reverse primer ([gene: 5′-GGATCCCTATGGCCCTGCGCGGA -3′], BamH I site underlined), was digested with Xba I and BamH I, and ligated into the pCV:GFP and digested with the same enzymes to generate pCV:p25-GFP. pCV: RFP-TGBp2 and pCV: RFP-TGBp3 were produced with primer pairs ([gene: 5′-GGATCCATGTCCGCGCAGGGCCATA-3′], BamH I site underlined, [gene: 5′-GAGCTCCTAATGACTGCTATGATTGTC-3′], Sac I site underlined) and ([gene: 5′-GGATCCATGGAAGTAAATACATATCTCA-3′], BamH I site underlined [gene: 5′-GAGCTCTCAATGGAAACTTAACCGTTC-3′]Sac I site underlined), respectively. To express p25-GFP in a virus vector, the pGR106-Δp25 vector was used. p25-GFP was amplified from pCV:p25-GFP with primer pair [gene: 5′-ATCGATATGGATATTCTCATCAGTAGTT-3′] (Cla I site underlined) and [gene: 5′-GTCGACTCACTTGTACAGCTCGTCC-3′] (Sal I site underlined), then ligated into the pGR106-Δp25 digested with the same enzymes to generate pGR106-Δp25- p25-GFP.

The BiFC assay used pCV:cYFP and pCV:nYFP ^[25]. The same primers and restriction sites listed above were used in constructing the vectors of p25, TGBp2 and TGBp3. The p25 mutants were obtained by PCR according to the mutant sequences reported ^[15]. The ER location site ([gene: ATGAAGACTAATCTTTTTCTCTTTCTCATCTTTTCACTTCTCCTATCATTATCCTCGGCCGAA]) used in previous reports ^[26] and the widely-used NLS of Simian virus 40 large T antigen ([gene: ATGCCTCCAAAAAAGAAGAGAAAGGTC]) were used for fusion PCR to produce vectors expressing ERp25 and NLSp25.

Subcellular targeting of proteins by fluorescence was explored in Nicotiana benthamiana. N. benthamiana line 16 c was used to analyse gene silencing suppression as previously reported ^[10]. Briefly, agrobacterium mixtures carrying 35S-green fluorescent protein (35S-GFP) and the individual constructs were infiltrated into leaves of 16c plants. GFP fluorescence was viewed under long-wavelength UV light 5 dpi. Agrobacterium strains C58C1 and GV3101 were used for agrobacterium infiltration at OD₆₀₀ = 1.0 except where stated. Equal volumes of individual agrobacterium cultures (OD₆₀₀ = 1.0) were mixed before co-infiltration. The analysis using chemical inhibition of the cytoskeleton was carried out as reported ^[18].

The Leica TCS SP5 (Leica Microsystems, Bannockburn, IL) confocal laser scanning microscope system was used to examine the fluorescence of GFP, RFP and YFP. Unless otherwise stated, fluorescence was monitored at 5 dpi. GFP was excited at 488 nm and the emitted light captured at 505–525 nm; RFP was excited using 543 nm and captured at 590–630 nm; YFP was excited at 514 nm and captured at 555–575 nm. All images were processed using Adobe Photoshop version 7.0 software (Adobe Systems Inc., San Jose, CA).

PVX movement recovery experiments were carried out according to previous reports ^[15]. Briefly, an agrobacterium culture GV3101 with the 35S:PVX-GFPΔp25 (pGR106) was diluted 10,000-fold, then mixed 1∶1∶1 with agrobacterium culture C58C1 carrying 35S:p19 and 35S:p25 mutants.