Unravelling the involvement of cilevirus p32 protein in the viral transport

Citrus leprosis (CL) is a severe disease that affects citrus orchards mainly in Latin America. It is caused by Brevipalpus-transmitted viruses from genera Cilevirus and Dichorhavirus. Currently, no reports have explored the movement machinery for the cilevirus. Here, we have performed a detailed functional study of the p32 movement protein (MP) of two cileviruses. Citrus leprosis-associated viruses are not able to move systemically in neither their natural nor experimental host plants. However, here we show that cilevirus MPs are able to allow the cell-to-cell and long-distance transport of movement-defective alfalfa mosaic virus (AMV). Several features related with the viral transport were explored, including: (i) the ability of cilevirus MPs to facilitate virus movement on a nucleocapsid assembly independent-manner; (ii) the generation of tubular structures from transient expression in protoplast; (iii) the capability of the N- and C- terminus of MP to interact with the cognate capsid protein (p29) and; (iv) the role of the C-terminus of p32 in the cell-to-cell and long-distance transport, tubule formation and the MP-plasmodesmata co-localization. The MP was able to direct the p29 to the plasmodesmata, whereby the C-terminus of MP is independently responsible to recruit the p29 to the cell periphery. Furthermore, we report that MP possess the capacity to enter the nucleolus and to bind to a major nucleolar protein, the fibrillarin. Based on our findings, we provide a model for the role of the p32 in the intra- and intercellular viral spread.


Results
The cileviruses MP is able to complement the cell-to-cell and systemic AMV transport and is not dependent of the CP interaction for these processes. In a previous study, we found the p32 protein co-localized with plasmodesma structures at cell periphery 28 , suggesting that this protein could play a role in viral movement. Here, we evaluate if cileviruses MPs are able to complement the cell-to-cell transport of AMV. The MPs of CiLV-C and CiLV-C2 were inserted into the AMV RNA3 infectious construct that expresses the GFP (pGFP/A255/CP) 29 by exchanging the AMV MP. Two variants of this AMV construct were tested, which differed in the presence or absence at the C-terminus of the heterologous MP, of the C-terminal 44 amino acids (A44) of the AMV MP, a region required for a compatible interaction with AMV coat protein (CP) 9 . In vitro transcripts from these heterologous constructs were mechanically inoculated in leaves of Nicotiana tabacum plants that constitutively express the P1 and P2 subunits of the AMV replicase (P12 plants), to evaluate the cellto-cell movement and compared to the AMV wild type (wt). The P12 plants permit to work only with the AMV RNA 3 or its derivatives, simplifying the analysis. All constructs harboring the A44 residues, resulted in clear fluorescence infection foci at 2 dpi (Fig. 1A, left panels), indicating that the cileviruses MPs are able to complement the cell-to-cell movement of the chimeric construct. The analysis of the area of 80 independent infection foci at 3 dpi revealed bigger infection foci derived from the construct carrying the CiLV-C:A44 MP (average of 3.56 mm 2 ) followed with the CiLV-C2:A44 derivative (2.93 mm 2 ) and the AMV wt (1.92 mm 2 ) (Fig. 1B). The constructs lacking the A44 residues were also competent for the cell-to-cell transport (Fig. 1A, right panels), but significantly less efficient than those harboring the A44 (p-value < 0.05). Thus, we observed a decrease of 1.55 mm 2 in the foci area comparing the CiLV-C:A44 and CiLV-C constructs (3.56 mm 2 versus 2.01 mm 2 ) and 1.55 mm 2 between the CiLV-C2:A44 and CiLV-C2 derivatives (2.93 mm 2 versus 1.38 mm 2 ), at 3 dpi (Fig. 1B). However, both constructs carrying the heterologous MPs non-fused to A44 (CiLV-C or CiLV-C2) generated infection foci similar to those caused by the AMV wt (Fig. 1B), and the t-test did not indicate significant differences at 3 dpi (p-value > 0.05). The observation that both cilevirus MPs were competent for the cell-to-cell transport of AMV, regardless of the fusion of the A44 region, suggests that both proteins could mediate the AMV transport independently of the AMV CP interaction. To discard a putative interaction between cilevirus MPs and the AMV CP, we performed an in vivo protein interaction BiFC assay. Reconstitution of the YFP fluorescence was not detected in all protein pair combinations performed (Fig. S1), suggesting non-heterologous interactions.
Further analyses were addressed to evaluate the capability of the cileviruses MP to support the systemic movement of AMV RNA 3. To do this, we used an AMV RNA3 wild-type clone that does not express GFP (pAL3NcoP3), since the RNA 3 derivatives harboring the GFP reporter gene do not support systemic movement in P12 plants 29 . The distribution of the RNA 3 expressing the heterologous MPs in inoculated and systemic P12 leaves was analyzed by tissue-printing of petiole cross sections, where positive hybridization signal correlated with the presence of the virus in the corresponding leaf, as previously described [30][31][32] . We observed the presence of the viral RNA in inoculated leaves (I) and upper leaves (U) for all heterologous construction expressing the A44 fragment and AMV control ( When the same experiment was performed with the AMV RNA 3 constructs expressing the heterologous MPs non-fused to the A44, CiLV-C MP was not able to support the systemic transport of the hybrid AMV virus, resulting in virus detection only on inoculated leaves (Fig. 1C, line 3). Unlike CiLV-C MP, the construct expressing the CiLV-C2 MP was able to support the systemic transport regardless its fusion with the A44 residues, since the virus was detected in the inoculated and in the U3 and U5 upper leaves (Fig. 1C, line 5 and consistent in all replicates), indicating that this MP allows the systemic AMV transport independently of its AMV CP interaction.
In order to evaluate if other proteins encoded by cileviruses have the ability to complement the AMV viral movement, we inserted the p29, p15, p61 and p24 CiLV-C genes into the AMV RNA 3 constructs harboring the GFP reporter gene. In all cases, only single fluorescent cells were observed on P12 leaves (Fig. 1A), indicating the inability of these proteins to complement the cell-to-cell transport of the AMV RNA 3 derivatives.
The intercellular spread of cileviruses is independent on nucleocapsid assembly. To obtain additional insights about the performance of the cileviruses MP in viral transport, we examined whether the viral complex transported intercellularly by these MPs is dependent on the virus particle formation. For this purpose, we performed a deletion of C-terminal 21 amino acids of the coat protein (CPN199) on the pGFP/ Analysis of the cell-to-cell and systemic transport of the hybrid AMV RNA 3 in which its MP gene was exchanged with the corresponding genes (MPs) of CiLV-C and CiLV-C2 and the genes p29, p15, p61 and p24 of CiLV-C. (A) Infection foci observed in P12 plants inoculated with RNA 3 transcripts from pGFP/ A255/CP derivatives carrying the AMV MP lacking the C-terminal 44 residues (A44) (MP), the AMV MP wild type (MP:A44), and the aforementioned heterologous MPs alone (right panels; MP) or fused to A44 (left panels; MP:A44). The schematic representation shows the GFP/A255/CP AMV RNA 3 29 , in which the open reading frames, represented by large boxes, correspond to the green fluorescent protein (GFP), the movement protein (MP) and the coat protein (CP). Short box corresponds the C-terminal 44 amino acids of the AMV MP, meanwhile arrows represent subgenomic promoters. The numbers after the viral acronym represent the total amino acids residues of the corresponding MP. The NcoI and NheI restriction sites used for insertions of the MPs are indicated, as well as the restriction sites BspHI, PciI and NheI for insertions of the other CiLV-C genes. White bars represent to 500 μm, 2 mm and 10 mm. N.D non-determined. (B) Histograms represent the average of the area in mm 2 of 80 independent infection foci at 2 and 3 days post-inoculation (dpi). Error bars indicate the standard deviation. Student's t-test and significance was set at p < 0.05. The p-values obtained from comparison between pairs of groups are presented. (C) Tissue-printing analysis of P12 plants inoculated with the AMV RNA 3 derivatives showed in A but lacking the 5′ proximal GFP gene. Plants were analyzed at 14 dpi by printing the transversal section of the corresponding petiole from inoculated (I) and upper (U) leaves. Mock corresponds to non-inoculated plant. www.nature.com/scientificreports/ MP:A44/CP chimeric construct. It has been previously demonstrated that an infectious clone expressing the CPN199 is competent for cell-to-cell spread and RNA accumulation, but not for virion encapsidation 33 . Clear infection foci were visualized with the two AMV constructs carrying the cilevirus MPs and the CPN199 (Fig. 2), similar to what is observed for the AMV wt (positive control). The AMV construct carrying the NSm movement protein of tomato spotted wilt orthotospovirus (TSWV; negative control), that was able to transport only encapsidated virions 30 , generated single fluorescent cells (Fig. 2). These results demonstrate that the virus particles are not required for the cell-to-cell transport mediated by cileviruses MPs.
The C-terminal region of cileviruses MP is dispensable for cell-to-cell movement, but its expression is essential for an efficient transport. A common property observed for C-terminal region of the MPs assigned to the 30K superfamily was its dispensability for the cell-to-cell spread 8,9,30,[34][35][36] . Given this context, we analyzed if such property could also be applied to cilevirus MPs. Firstly, we aligned the MPs of CiLV-C and CiLV-C2, observing a high variability at the C-terminal region (between amino acids 160 and 297) (see complete alignment in Fig. S3), with most divergence located in the last C-terminal 36 residues (Fig. 3A, red dotted line). From this analysis, sequential C-terminal deletions of the CiLV-C2 MP were performed until the protein was not competent to support the virus movement. For this purpose, a chimeric RNA 3 AMV harboring GFP reporter gene, was used to evaluate the functionality of the truncated cileviruses MP. C-terminal mutants of the CiLV-C2 MP revealed that deletions up to 70 residues generated MPs competent to support the cell-to-cell movement of the chimeric AMV RNA3 in P12 plants, showing clear fluorescent infection foci (Fig. 3B, T-Δ 222-292 ). However, when the C-terminal 74 amino acids (aa) were deleted (construct R-Δ 218-292 ), only single fluorescent cells were visualized (Fig. 3B), indicating that residues located between R-T ("GMV", see Fig. 3A) seem to be essential for cell-to-cell spread. The same pattern was observed using the CiLV-C MP (Fig. 3B), in which deletion of the C-terminal 69 aa (construct V-Δ 228-297 ) generated a MP competent for the cell-to-cell transport meanwhile deletion of three additional residues (construct G-Δ 225-297 ) rendered a nonfunctional MP. In the next step, we evaluated the efficiency of the functional cilevirus MPs lacking different portions of the C-terminus. To do that, www.nature.com/scientificreports/ we measured the infection foci area at 2 and 3 dpi of the AMV RNA3 chimeric construct carrying the CiLV-C2 MP wt or its derivatives lacking the C-terminal 27 (P-Δ 266-292 ), 38 (E-Δ 255-292 ), 51 (D-Δ 242-292 ) and 61 (K-Δ 232-292 ) residues, respectively. The analysis of the area of 80 independent infection foci revealed a significant decrease (p-value < 0.05, except for mutant E-Δ 255-292 ), mostly evident in a comparison between the MP wt with the construct lacking 61 residues (K-Δ 232-292 ), rendering infectious foci with an area representing 12% of MP wt (average of 0.56 mm 2 versus 4.65 mm 2 at 3 dpi) (Fig. 3C). This finding reveals that although the movement capacity of MP mutants lacking different C-terminal portions is maintained, the removal of this region affects considerably the efficiency of the cell-to-cell transport. Next, we evaluated whether the large cilevirus MPs C-terminal mutants functional for the cell-to-cell transport were still able to support the systemic movement. To do this, the CiLV-CΔ 228-297 and CiLV-C2Δ 222-292 MP mutants lacking the C-terminal 69 and 70 residues, respectively were assayed in the AMV RNA3 wt construct. Tissue-printing analysis reveals no hybridization signal in the petiole of inoculated and all upper P12 leaves, meanwhile the control AMV and the cilevirus MPs wt showed the viral presence in inoculated and upper leaves, as expected (Fig. 3D).
The C-terminal region of MP is necessary for tubule formation and correct association with the plasmodesma. To further characterize the cilevirus MPs, we investigated first the polymerization of tubule structures on the surface of the Nicotiana benthamiana protoplasts. To do this, the MPs carrying a C-terminal eGFP fusion, were transiently expressed in N. benthamiana leaves by agroinfiltration. The protoplasts were purified 24 h post-inoculation (hpi) and the eGFP signal was visualized 16 h after purification. All cilevirus MPs induced the formation of tubular structures on the protoplast periphery (Fig S2a,b), regardless of viral infection. This indicates tubule-forming property for cilevirus MPs.
In the next step, we analyzed if the negative effects in virus transport observed with the MPs lacking the C-terminal could be related to the capacity to generate tubular structures on protoplasts. To this end, the truncated versions of MPs: CiLV-CΔ 228-297 , CiLV-CΔ 225-297 , CiLV-C2Δ 222-292 and CiLV-C2Δ 218-292 , carrying a C-terminal eGFP were assayed in protoplasts as aforementioned. The constructs that generated infectious foci with low movement efficiency (CiLV-CΔ 228-297 and CiLV-C2Δ 222-292 ) showed the GFP signal accumulation in punctuated structures dispersed along the protoplast periphery with the formation of short tubules (see white arrows in Fig. 4a,c and panels i-iii) when compared with wt MPs constructs (Fig. S2); on the other hand, the non-functional constructs in viral transport (CiLV-CΔ 225-297 and CiLV-C2Δ 218-292 ) rendered fluorescent dots and diffuse signal at the protoplast surface but not tubular structures (Fig. 4b,d). These results clearly indicate that the C-terminal region of both MPs influence the tubule formation.
In the next step, we evaluated the implication of the C-terminal region in the MP association with plasmodesmata. To do that, C-terminal MP deletion constructs carrying the GFP fused at its C-terminus were transiently expressed in N. benthamiana leaves. After staining the plasmodesmata (PD) structures with a callose marker (aniline blue), we observed that deletion of 69 residues of the CiLV-C MP (CiLV-CΔ 228-297 ) impairs the correct co-localization ( However, and unlike the CiLV-C MP, the GFP signal derived from the construct CiLV-C2Δ 218-292, rendered a fluorescent punctate signal along the cell periphery ( Fig. 5f) similar to that observed for the CiLV-C2Δ 222-292 MP mutant (Fig. 5e), but with higher disturbance on the interaction with PD (PCC = 0.42), compared with the CiLV-C2Δ 222-292 construct. The co-localization of CiLV-C and CiLV-C2 wt MPs and the corresponding truncated versions with plasmodesmata was further analyzed by measurements of fluorescence intensity across the plasma membrane. Fluorescent GFP signal of CiLV-C and CiLV-C2 wt MPs coincide with all callose marker signal (Fig. 5ai,div); on the other hand, GFP signal derived from the CiLV-CΔ 228-297 , CiLV-CΔ 225-297 , CiLV-C2Δ 222-292 and CiLV-C2Δ 218-292 constructs did not match completely with plasmodesma signal along the plasma membrane ( Fig. 5b ii,ciii,ev,fvi), further suggesting that these deletions alter the correct localization of the MP to the plasmodesma structures. To see this effect clearly, we decided to determine the percentage of callose-stained PD which co-localize to the MP-GFP fusion proteins (Table 1). Thus, the CiLV-C and CiLV-C2 MPs wt were detected in the 92.3% and 93.7% of stained PD, respectively, meanwhile C-terminal MP mutants lacking the C-terminal 69 (CiLV-CΔ 228-297 ), 72 (CiLV-CΔ 225-297 ), or 70 (CiLV-C2Δ 222-292 ), 74 (CiLV-C2Δ 218-292 ) residues were detected in 45.3%, 9.7% or 48.2%, 34.7% of PD, respectively, indicating that the C-terminal region of cilevirus MPs has a positive role in the MP-PD co-localization.

Cileviruses MP interact with the cognate capsid protein (p29) through both N-and C-terminal regions.
Previous results showed that the MP of CiLV-C is able to interact in vivo with the cognate capsid protein 28 . In the next step, we decided to determine by BiFC analysis whether or not the characterized C-terminal region of the cilevirus MPs, dispensable for viral transport, is required to interact with the cognate cilevirus CPs. To do that, wild-type (CiLV-C MP wt and CiLV-C2 MP wt) and mutated MPs lacking the C-terminus (CiLV-C MPΔ 228-297 and CiLV-C2 MPΔ 222-292 ) or the remaining N-terminus (CiLV-C MPΔ 1-227 and CiLV-C2 MPΔ 1-221 ) were co-expressed with their cognate p29 proteins. Reconstitution of the YFP fluorescence was detected in all MPs versions co-expressed with the p29 proteins (Fig S4Aa-f) suggesting positive interactions, whereas no fluorescence was observed when the viral MPs were co-expressed with the unfused CYFP (CYFPcyt) or with the counterpart (C-YFP) targeted to the ER (CYFPer), (Fig. S4Ag-i). We also used the nucleocapsid (N) of TSWV, www.nature.com/scientificreports/ a virus evolutionarily distinct from cileviruses, which was expressed in combination with the cilevirus MPs. In all cases we did not observed reconstitution of the YFP fluorescence ( Fig. S4Ag-i). Co-immunoprecipitation (Co-IP) experiments between CiLV-C and CiLV-C2 MPs and the respective cognate p29 proteins confirmed the MP-p29 interactions (Fig. S4B). Taken together, these results characterize the interaction of MP and p29 and suggest the possible presence of two or more regions of the cilevirus MPs with capacity to interact with the p29 protein.
The MP co-expressed with p29 redirects the p29 from the cytoplasm to cell periphery associating with the plasmodesmata. In the next step, we evaluated if the characterized MP-p29 interaction could modify the intracellular redistribution of both proteins. For this purpose, first we analyzed the intracellular distribution of each protein separately. To do this, binary constructions containing the CiLV-C p29-eGFP or MP-eGFP fusion proteins were used for transient expression in N. benthamiana leaves. The fluorescent signal derived from the p29-eGFP construct accumulated in numerous punctate bodies (Fig. 6A, a, white arrows) and in large inclusions bodies (Fig. 6A, a, red arrows) dispersed throughout the cytoplasm, as reported previously 28 . Callose staining revealed that p29 derived GFP signal was also detected in the 8.3% of PD (Table 1) (Table 1), as reported previously 28 .
In the next step, the subcellular localization of the p29 was evaluated in presence of the MP. To do that, the p29-eGFP construct was co-infiltrated with the MP-HA construct. We observed a clear redistribution of the small punctate p29-eGFP structures from the cytoplasm to the cell periphery, especially along the plasma membrane ( Fig. 6B,a), which co-localized correctly (PCC = 0.79) with the PD (Fig. 6B,b, blue arrow), incrementing the percentage of stained PD which co-localize with the p29-GFP derived signal (8.3% to 58.5%; Table 1). Apparently, the presence of the MP redirected the p29 to the PD structures.
Based on the suggested capacity of the MP to interact with the p29 with both N-and C-termini, we investigated the role of both MP regions on the p29 redistribution. First, we characterized the sub-localization of these regions (MPΔ 1-227 and MPΔ 228-297 ) by fusion the eGFP at its C-termini (MPΔ 228-297 -eGFP and MPΔ 1-227 -eGFP). The N-terminal region of the MP (MPΔ 228-297 -eGFP) accumulated in punctate structures at the cell periphery ( Fig. 6C,a), similar to that observed for the MP wt, but did not coincide completely (PCC = 0.42) with the PD (see Fig. 5b), reducing the percentage of stained PD showing MPΔ 228-297 -eGFP derived signal, when compared to the MP wt (92.3% vs 45.2%; Table 1). Unlike the N-terminal region, the C-terminal 70 residues of the MP (MPΔ 1-227 -eGFP) resulted in a diffuse GFP signal along the plasma membrane (Fig. 6C,b) which did not colocalize with PD (Fig. 6C,c; PCC = − 0.04; Table 1).
Finally, BiFC analysis were performed by transiently co-expression of the p29 protein with the N (MPΔ 228-297 ) or C (MPΔ 1-227 ) MP regions in N. benthamiana leaves. The YFP signal revealed that the MP C-terminal region interacts with the p29 forming a complex distributed in the cytoplasm (red arrow in Fig. 6D,a) and along the cell periphery (blue arrow in Fig. 6D,a enlarged), indicating that this portion of the MP is able to recruit the p29 protein from the cytoplasm to the periphery. The MP C-terminal redistribution furthermore indicates the capability of interactions between p29 and this MP fragment. Callose staining revealed that the cell periphery fluorescence derived from the MP Δ1-227 -p29 complex did not coincide completely with the PD structures (PCC = 0.38) (Fig. 6D,c, blue arrows) allowing its detection in the 28.7% of stained PD (Table 1). On the other hand, we observed that the BiFC signal derived from the MPΔ 228-297 -p29 interaction accumulated in aggregates throughout the cytoplasm (Fig. 6D,b) which do not co-localize with PD (Table 1). No interactions were detected for any of the negative controls analyzed (P29-CYFP + Ncyt, MPΔ 1-227-CYFP + Ncyt and MPΔ 228-297 -CYFP + Ncyt, see Fig S4g-i). Collectively, these results further indicate that, being both N-and C-terminal regions of CiLV-C  , panels a and b, white arrows), which was confirmed with the coexpression with a nucleus marker fused to the mRFP (mRFP-pCB302) (Fig. 7A,d). This result suggests that the nucleus could has an implication on the movement mechanism mediated by cilevirus MPs. The nuclear localization of CiLV-C2 MP was visually more evident when compared with the CiLV-C MP (Fig. 7A, panel a,b). In this sense, the analysis of the fluorescent signal derived from 20 different nuclei from leaves infiltrated with the same adjusted Agrobacterium density (OD 600 = 0.5) showed that CiLV-C2 MP content was significantly higher into the nucleus (GFP intensity signal of 51.21%) than CiLV-C MP (17.11%) (Fig. 7A,c; p-value < 0.05). www.nature.com/scientificreports/ Previous studies have reported that nuclear viral MPs are required for the formation of RNPs, in association with CBs and fibrillarin, and that complexes were essential for the viral spread, especially for the systemic movement [41][42][43][44] . In this sense, we decided to analyze the putative cilevirus MPs-fibrillarin interaction in the nucleolar compartment, by BiFC assay. To this end, fibrillarin constructs carrying C-terminal fusions of the NYFP or CYFP fragments were co-expressed with the CiLV-C and CiLV-C2 MPs constructs containing the YFP counterpart fused at their N-or C-terminus in N. benthamiana leaves. Reconstitution of the fluorescent signal was detected in the nucleolar compartment of several cells (Fig. 7B,a-c), suggesting the capacity of the cileviruses MP to interact with fibrillarin. No interactions were detected for negative controls (Fib2 + Ncyt, Fib2 + NoLS or MP + Ncyt, MP + Cer, MP + TSWV N; Figs. 7c,d, S4g). The controls images are representative of the equivalent interaction observed for both MPs. To confirm this interaction, Co-IP experiments between CiLV-C2 MP and Fib2 were performed. This MP was chosen for Co-IP, since it accumulated in significantly higher levels into the nucleus than CiLV-C MP. Given that the interaction takes place inside the nucleus, two different lysis buffers were used to show the intranuclear location of the MP-Fib2 interaction. The RIPA buffer, containing ionic detergents and active constituents that permit the nuclear membrane disruption 45 , and a non-denaturing lysis buffer, carrying nonionic detergent (NP-40), which does not disrupt the cell nucleus 46 . N. benthamiana leaves were infiltrated with agrobacterium cultures carrying the expression cassettes for the CiLV-C2 MP-HA and Fib2-3xMyc constructs and subjected to the Co-IP assay at 3 days post infiltration. Using the non-denaturing lysis buffer, no band was visualized in the MP-HA + Fib2-3xMyc immunoprecipitated extract after western blot analysis (Fig. 7C, western panel A). However, when the same infiltrated leaves were treated with RIPA buffer, a clear band was observed in the immunoprecipitated extract (Fig. 7C, western panel B), indicating both the capacity of the MP to interact with fibrillarin but also that such interaction occurs in the nucleus.

Discussion
Although cileviruses have been known for decades and reported in more than 50 natural and experimental host species so far 19,47 , nothing is known about their movement machinery. In this work, we reveal in detail the functionality of the p32 protein encoded by CiLV-C and CiLV-C2. Ours findings provide a direct experimental evidence to support the movement function for the p32 protein, which was able to restore cell-to-cell and systemic movement of a MP defective AMV infectious clone. We also tested this ability for all other encoded cilevirus proteins all being unable to restore the AMV transport, resulting in infection of individual cells. Additionally, the CiLV-C p32 was also able to complement in trans the movement-deficiency phenotype of a turnip crinkle virus (TCV) and tobacco mosaic virus (TMV) mutants in N. benthamiana plants 48,49 , indicating the capacity of cilevirus MP to rescue the local movement of viruses that differ in movement transport mechanisms.
In the AMV transport mechanism, the C-terminal A44 residues of AMV MP retain the capacity to interact with the cognate CP, facilitating that heterologous MPs complement the AMV cell-to-cell and systemic movement 9 . This compatibility was previously tested for ilar-, bromo-, cucumo-, como-and orthotospoviruses MPs, which are only functional when carrying the A44 fragment 9,30,31,50 . Here, the MP of cileviruses were competent to complement the cell-to-cell AMV transport regardless of the A44 extension, suggesting that the transport of AMV RNA by this MP is independent of the CP. Additionally, the hypothesis that the movement generated could be the product of a putative interaction between these heterologous MPs with the CP of AMV was excluded from in vivo interaction analysis, further suggesting our inferences above mentioned. In addition, the positive virus transport observed with the AMV RNA3 derivative carrying a CP gene defective in virion assembly (CPN199) 8,33 clearly shows that the cilevirus MPs are able to mediated the transport of other viral complexes different than the virus particles and independent of the CP, at least in the AMV context. The same ability to complement the AMV transport regardless of the A44 extension, was previously demonstrated for the TMV MP in the AMV context 9 . This protein is able to bind nucleic acids non-specifically, to form RNP complexes in association with viral replication proteins and host factors to traffic the RNPs intracellularly to the cell periphery, and to mediate the passage through the PD without the involvement of CP 12 . Here, we do not observe a significant www.nature.com/scientificreports/ www.nature.com/scientificreports/ decrease in movement efficiency comparing the non-fused A44 constructs with the AMV wt. The cileviruses MPs capacity to support the AMV transport without the CP aid strongly suggest that these proteins show a putative high RNA affinity, a feature identified in MPs of various positive stranded-RNA viruses 51 . However, and in spite that the CP was not required for the cilevirus MPs in the AMV context, we cannot exclude the participation of capsid protein in virus transport in the natural CiLV-C infection. Interestingly, the CiLV-C2 MP was able to complement the systemic movement of AMV non-associated with the A44 fragment, despite this portion of the AMV MP is essential for the systemic movement of AMV 8,33,52 . Based on these findings, we can conclude that this protein seems competent enough to transport, throughout the phloem, infective viral complex regardless of the interaction with the coat protein, indicating that virus particles are not required for the AMV systemic transport, at least with this viral MP. Further experiments will be addressed to know which viral complexes are systemically transported with the CiLV-C2 MP.
Previous studies have suggested that under natural conditions of BTVs infection, the cileviruses and dichorhaviruses are not able to reach their plant hosts systemically 19 . Here, our findings clearly indicate that MPs of cileviruses are competent to allow the systemic transport of a movement defective viral clone in Nicotiana tabacum species. In this sense, we speculate that limitation of the cileviruses to reach their hosts systemically in the natural infection process is not due the functional restriction in their MPs, but possibly to some host defense factors. A general mechanism such as RNA silencing is more likely to be responsible for this viral limited movement phenotype 48 . However, further experiments are needed to address this question.
By BiFC analysis, we observed that the nuclear localization of the cilevirus MPs correlated with their capacity to interact with the fibrillarin in the nucleolus compartment. Similar result was observed by Co-IP experiments using the CiLV-C2 MP but only when a RIPA buffer was used which disrupts the nuclear membranes, suggesting that the MP-fibrillarin complex is located mainly in the nuclear extract. The same nuclear localization property was attributed to other viral proteins associated with viral movement. The viral proteins p20 (potexvirus), p2 (tenuivirus), Triple Gene Block 1 (TGB1) (hordeivirus, pomovirus), P7a (betanecrovirus), VPg (potyvirus) and ORF 3 (umbravirus) have been shown to cycle through the nucleus associating with Cajal bodies (for some cases) and fibrillarin, a route required for the formation of RNPs and essential for viral cell-to-cell movement and systemic spread 42,44 . Although speculative, it is tempting to hypothesize that the MP of cileviruses would act similarly favoring the viral movement. In this sense, the higher nuclear localization observed for the CiLV-C2 MP (compared with CiLV-C MP) correlated with a more efficient systemic movement. However, further experiments, using silenced fibrillarin plants, will be addressed to clarify the role of fibrillarin in the viral transport mediated by cilevirus MPs.
Another feature explored in unraveling the performance of the cileviruses MP in viral transport was the ability of MP to induce the formation of tubular structures on the surface of the protoplasts 53 . Sequential deletion in the C-terminal of CiLV-C2 MP showed that, although the absence of this fragment still enables cell-to-cell movement, its removal impairs the correct tubule polymerization (resulting in short tubules) and MP-plasmodesma association, reducing the percentage of PD targeted with cilevirus MPs from 92.3 (wt) to 45.2% (CiLV-C MPΔ 228-297 ). Interestingly, the functional cilevirus MPs, lacking the large C-terminal region, showed a reduced cell-to-cell transport and were unable to support the viral systemic transport, which correlated with the reduced MP-plasmodesma association and/or the short tubular structures. However, this correlation showed herein was not reproduced for other viruses. For instance, the AMV MP mutant (Δ242-256) induced tubular structures in less that 5% of the assayed protoplasts, whereas this mutation permitted wilt type levels of foci formation 8 . www.nature.com/scientificreports/ Other example is the MP of the TSWV, when expressed in TMV system with a deletion of the C-terminal 54 residues rendered a protein unable to generate tubular structures, but still functional with a reduced cell-to-cell transport 54 . For the cilevirus constructs, the absence of tubules or the presence of short-or large tubules resulted in the absence of movement, low efficient movement or highly efficient movement, respectively. The cilevirus MPs tolerate a C-terminal deletion of -70 or -69 amino acids. Similar C-terminal deletions (60 amino acids or less) did not interfere with the movement functions of the 30 kDa MP of TMV, AMV, cucumber mosaic virus (CMV), cowpea chlorotic mottle virus (CCMV) and odontoglossum ring sport virus (ORSV) 8,34,55,56 . Additional deletions (-72 or -74 amino acids) of the cilevirus MPs inhibited the movement, suggesting that the three or four residues indicated here as essential for the movement ("LIV" for CiLV-C MP and "GMVT" for CiLV-C2 MP, Fig. 3A) are possibly responsible to ensure a correct three-dimensional structure critical for MP polymerization and/or tubule formation. However, we do not exclude that the reduced length of MPs could influence the inhibition of movement. The systemic movement was completely impaired when the C-terminal region was removed (CiLV-CΔ 228-297 and CiLV-C2Δ 222-292 ), despite the fact that these constructions were still functional for a reduced cell-to-cell transport. In this sense, the observation that a minimal cell-to-cell speed is required for the virus to reach the upper region of the plant 31,57 , could justify the systemic infection impediment observed with these truncated MPs.
The MP, when co-expressed with the p29 protein resulted in a redistribution of the p29 from cytoplasm to cell periphery. This observation suggests that the MP coordinates intracellular trafficking of its cognate viral coat protein. More significantly, the C-terminus, but not the N-terminus, of the cileviruses MP proved to be the region responsible to orchestrate the trafficking of the p29 to the cell periphery. Intriguingly, deletion of the MP N-terminus completely abolished its localization with the plasmodesma, but not its distribution at the cell periphery, indicating that the PD localization sequence is located at the N-terminus of the cilevirus MPs, as recently identified in the N-terminus of TMV MP 58 . In addition to the MP, the cilevirus CPs were also able to reach the plasmodesmata structures without any other viral protein, a process that was incremented by the presence of MP through its C-terminus. Similar autonomous PD localization has been described for other viral CP, involving post-translational modifications 59 . Altogether, indicate that the C-terminal region of cilevirus MPs is a critical cis element required for both the systemic transport and the incorporation of coat protein in the viral complexes transported through PD. The implication of the C-terminal region of MPs assigned to the 30K family, in viral systemic transport has been previously described 8,36,60 , but also the requirement of the CP, either in form of ribonucleoprotein complex or virus particles, through an interaction to this MP region 9,31,36,52,61 . The results presented herein indicate that the critical region required for the viral systemic transport is only the C-terminus of cilevirus MPs through its role in tubule formation and PD localization, which contribute to a more efficient cell-to-cell transport. The interaction of this region with the cognate CP could be considered a cis element to ensure the transport of the CP required for other viral processes (e.g. vector transmission, RNA protection, etc.) but, apparently, unnecessary for the systemic transport. The open question is if this model could be applied to the rest of MPs assigned to the 30K family.
Cileviruses accumulate and replicate in the cytoplasm, remodeling the membrane of the ER network and generating large viroplasms 28,62,63 . Our previous study has shown that MP and p29 traffic along the ER system, and the MP is integrally associated with the cell membrane 28 . Taken together with the data showed herein, these findings support a model for the role of the cileviruses MP in viral spread (Fig. 8). Our model posits that the MP is translated in the viroplasm and a portion of its population is transported into the nucleolus of the cell to bind to fibrillarin. This points to a scenario in which the MP could manipulate or recruit nucleolar functions to promote the movement, however whether the core complex formed is exported from the nucleus to the cytoplasm to form infective vRNPs and its association with viral movement, requires additional investigation. The p32, a membrane spanning protein, may anchor the vRNP complex to the ER membrane network, which traffics by the ER system to neighbor cells regardless the CP assistance. The observation that p29 has the capacity to interact with the MP and to traffic along the ER, and probably through the actin system 28 , opens an alternative route, in which the RNP complexes, carrying both p29 and MP proteins, could be transported to the cell periphery along the actin filaments. Finally, a part of the MP population is recruited to PD for tubule formation, which facilities the passage of the infective complexes throughout the PD to neighbor cells, a step in which the MP is able to generate independently of viral particle assembly, indicating its capability to transport viral complexes different to virus particles intra and intercellularly. However, the redirection of p29 to the plasmodesma through its interaction with the MP could also be implicated in the initiation of viral replication in the newly infected cells. This model represents a start point for unraveling the movement mechanism of the cileviruses. Organelle markers. For the nucleus subcellular co-localization, the proteins were co-infiltrated with cultures (OD 600 = 0.1) expressing the NLS of SV40 large T antigen (nucleus marker) fused to the red fluorescent protein (λexc = 587 nm; λem = 610 nm), provided by Dr. José Navarro, IBMCP, Valencia, Spain.

Materials and methods
Protoplast preparation and inoculation of P12 plants. All  Tissue-printing assay. Tissue-printing analysis were performed by transversal sections of the corresponding petiole from inoculated (I) and upper (U) P12 leaves at 14 dpi, as performed previously 64  www.nature.com/scientificreports/ from tissue-printing are representatives from three independent assays. The nucleic acids were fixed to the nylon membranes using UV cross-linker (700 × 100 μJ/cm 2 ). Hybridization and detection were conducted as described 65 , using a Dig-riboprobe (Roche Mannheim, Germany) complementary to the AMV 3′ UTR.
Intracellular sublocalization and BiFC assay. All details on protein expression for sublocalization and in vivo protein-protein interaction are presented in Supplementary Materials and Methods.
Confocal laser scanning microscopy. Fluorescence images were captured with the aid of a confocal laser scanning microscope Zeiss LSM 780 model. The excitation and emission captures of the GFP, YFP, mRFP fluorophores and aniline blue fluorochrome were conducted as described 28 . The images were prepared using Fiji Image J program (version 2.0r). For a better interpretation of the co-localization, graphs of fluorescence intensity were generated with Zeiss quantify intensity tools software. GFP intensity signals of 20 distinct nuclei were measured using Image J (version 2.0r) Macros plugin, the calculation area was selected to have GFP signal of similar intensities. A portion of the MP is transported into the nucleolus of the cell to bind to fibrillarin. The possibility to form a complex between MP plus fibrillarin that could exit the nucleus and interacts with vRNA and/or CP (p29) to form infectious vRNPs, is an open question. The MP, a membrane spanning protein, may anchors the vRNP complex to the ER membrane network, which traffics by the ER system to neighboring cells, facilitating the passage through plasmodesmata by the tubule formation. A possible alternative route may be mediated by the capacity of p29 to interact with MP and to associate with actin, thus anchoring the infectious complexes along the microfilaments (MF), guiding the vRNPs throughout the cytoplasm to the cell periphery. The virus particle is not required for the intercellular transport for this MP (indicating its capability to transport viral complexes different to virus particles intra and intercellularly); furthermore, the MP can transport the infectious complex cell-to-cell and systemically independent of the CP assistance. The redirection of p29 by MP to the plasmodesma could also be implicated to initiate viral replication in the newly infected cells. www.nature.com/scientificreports/