Tobacco mosaic virus movement protein complements a Potato spindle tuber viroid RNA mutant impaired for mesophyll entry but not mutants unable to enter the phloem

Tobacco mosaic virus movement protein (TMV MP) is essential for virus spread between cells. To accomplish its task, TMV MP binds viral RNA, interacts with components of the cytoskeleton, and increases the size exclusion limit (SEL) of plasmodesmata. Plasmodesmata are gated intercellular channels that allow passage of small molecules and macromolecules, including RNA and protein, between plant cells. Moreover, plasmodesmata are diverse and those connecting different cell types appear to have unique mechanisms to regulate macromolecular trafficking, which likely contributes to the establishment of distinct cell boundaries. Consequently, TMV MP might be competent to mediate RNA transport through some but not all plasmodesmal gates. Due to a lack of viral mutants defective for movement between specific cell types, the ability of TMV MP in this regard is incompletely understood. In contrast, a number of trafficking impaired Potato spindle tuber viroid (PSTVd) mutants have been identified. PSTVd is a systemically infectious non-coding RNA that nevertheless can perform all functions required for replication as well as cell-to-cell and systemic spread. Previous studies have shown that PSTVd employs different structure and sequence elements to move between diverse cell types in host plants, and mutants defective for transport between specific cell types have been identified. Therefore, PSTVd may serve as a tool to analyze the functions of MPs of viral and cellular origin. To probe the RNA transport activity of TMV MP, transgenic plants expressing the protein were inoculated with PSTVd mutants. Remarkably, TMV MP complemented a PSTVd mutant defective for mesophyll entry but could not support two mutants impaired for phloem entry, suggesting it fails to productively interface with plasmodesmata at the phloem boundary and that additional viral and host factors may be required. Consistent with this idea, TMV co-infection, but not the combination of MP and coat protein (CP) expression, was able to complement one of the phloem entry mutants. These observations suggest that phloem loading is a critical impediment to establishing systemic infection that could involve the entire ensemble of TMV proteins. They also demonstrate a novel strategy for analysis of MPs.

Introduction Intercellular transport of macromolecules is crucial for all multicellular eukaryotes, yet our understanding of the principles and mechanisms underlying this process is far from complete and new analytical approaches are needed. Potato spindle tuber viroid (PSTVd) and Tobacco mosaic virus (TMV) continue to be important models for the analysis of RNA trafficking in plants. TMV has a plus-sense single-stranded RNA genome of~6.5 kb that specifies four proteins. Conversely, PSTVd is a circular non-coding RNA of 359 nucleotides, thus all functions required for host plant infection are mediated by sequence and structure elements within the genomic RNA [1][2][3]. Viroids and viruses accomplish systemic infection by two distinct mechanisms; cell-to-cell spread and long-distance transport through the phloem. Spread between cells occurs through plasmodesmata, gated channels that provide cytoplasmic continuity and regulate the transport of macromolecules, including RNA and protein [4][5][6]. While viroid RNAs, with the support of host factors, can traverse plasmodesmata and traffick through the phloem transport system, viruses additionally require specialized movement proteins [7,8]. The first characterized movement protein, TMV 30 kDa protein (TMV MP), enables viral movement between cells by directly binding viral RNA, interacting with components of the cytoskeleton as well as the ER and associated actin network, and increasing the size exclusion limit (SEL) of plasmodesmata [9][10][11][12][13][14]. (SEL is defined by the size of the largest molecules that can pass through plasmodesmata). However, plasmodesmata connecting different cell types have unique structures and gating mechanisms. For instance, different SELs and structures have been observed in particular cell types, such as cells of the Arabidopsis root tip, and specialized plasmodesmata connecting phloem companion cells and sieve elements [15][16][17]. Further, as we have learned from analysis of PSTVd mutants, plasmodesmata between different cell types (including upper epidermis, palisade mesophyll, spongy mesophyll, bundle sheath, and phloem parenchyma) recognize different RNA structure or sequence elements, and requirements for viroid RNA trafficking are often unique and directional [18][19][20][21][22].
Mechanisms of TMV RNA spread between cells, and phloem loading and unloading, are incompletely understood and may involve other viral factors in addition to MP, such as coat protein (CP), replicase, or replication complexes [23][24][25]. Previous studies using MP-expressing transgenic tobacco plants showed that MP can mediate spread of macromolecules such as FITC-dextran between mesophyll and bundle sheath cells [10], but not between bundle sheath and phloem parenchyma [26]. Further, CP is essential for phloem-mediated accumulation of TMV [27,28], suggesting that phloem transport involves encapsidation of viral RNA into virions or a viral RNA-CP complex. However, CP is not required for TMV invasion of phloem parenchyma in inoculated leaves [29], although CP interaction with specific host factors could influence the efficiency of systemic spread [30]. Interestingly, results of a recent study indicate that negative regulation of salicylic acid signaling by CP also promotes long systemic spread [31], suggesting a role independent of its interaction with viral RNA. Additionally, studies have implicated the TMV 123/183 kDa replicase proteins in phloem loading, and these may also participate indirectly by altering the expression of auxin-responsive genes [32]. In any case, plant viruses in general can move between individual epidermal and mesophyll cells within hours while phloem loading takes considerably longer, suggesting the involvement of different host factors and/or a unique structure of plasmodesmata connecting bundle sheath cells and phloem [33]. Therefore, it is reasonable to hypothesize that TMV MP might vary in its ability to support cell-to-cell spread vs. phloem loading of viral RNA. Because genetic studies to address this question require RNAs defective for trafficking between specific cell types, we speculated that PSTVd mutants might prove useful. Further, because TMV MP can increase plasmodesmal SEL and transit itself to adjacent cells in the absence of other viral proteins [10,34], we examined PSTVd trafficking in transgenic plants expressing native TMV MP.
In this study, we show that TMV MP complements a PSTVd mutant that is unable to enter the mesophyll following rub-inoculation of epidermal cells, but cannot complement two mutants that are unable to enter the phloem. In addition, transient expression of CP in MPexpressing plants also failed to enable phloem loading of these mutants, although one was complemented by co-infection with TMV. These results suggest that TMV MP fails to productively interface with plasmodesmata at the phloem boundary, and that additional viral and host factors may be involved in phloem loading. These findings are consistent with the idea that gating systems for phloem entry are more complex than those at other boundaries. They also highlight the potential of PSTVd mutants as genetic tools to study the abilities of viral and endogenous MPs to traffick RNA between specific cell types.

PSTVd mutants and transgenic plants expressing TMV MP
Three previously characterized PSTVd mutants were selected to test the ability of TMV MP to mediate RNA movement between different cell types: U178G/U179G, G76A, and G156A (hereafter GG, 76AU, and 156AU, respectively) [21,22]. The secondary structure of PSTVd contains 27 loops or bulges and 26 base-paired stems (Fig 1). GG disrupts the 3D structure of loop 27, whereas 76AU and 156AU convert essential G-U pairs on two distinct stem regions to A-U. PSTVd GG is unable to enter palisade mesophyll, the layer of cells that directly underlies the upper epidermis, when rub-inoculated to epidermal cells on the upper leaf surface. However, it can spread between all cell types, including from palisade mesophyll cells to upper epidermis, when inoculated by phloem injection [21]. Mutants 76AU and 156AU are defective for phloem entry from bundle sheath cells following rub-inoculation but can exit the phloem and move between all cell types if directly injected into phloem [22]. The PSTVd intermediate strain in which the mutants were constructed infects Nicotiana benthamiana (Nb) but not N.

PLOS PATHOGENS
TMV MP complements a trafficking impaired PSTVd mutant tabacum, while TMV infects both species. Nb plants expressing native TMV MP (Nb+MP) under control of the constitutive Cauliflower mosaic virus 35S promoter were reported in a previous study and shown to complement a TMV mutant (TE1) defective for cell-to-cell movement [35]. The Nb+MP plants do not have a visually obvious phenotype, suggesting that MP does not randomly mobilize RNAs. The presence of MP-specific transcripts in the transgenic Nb+MP plants was confirmed by RT-PCR in 9 randomly selected plants (S1 Fig).
Functional MP expression in Nb+MP plants was confirmed by monitoring systemic spread of trafficking-defective mutant TMV-GFP(TE1), which we constructed by introducing the original TE1 frameshift mutation into a TMV-GFP vector [36,37]. This involved deletion of a single nucleotide (A4931) in the MP coding region, resulting in a truncated protein of 16 amino acids [35]. Six Nb and Nb+MP plants were co-inoculated by agroinfiltration with pCB301 vectors to express the TMV-GFP(TE1) mutant and GUS (control), and as a positive control another six Nb plants were similarly co-inoculated with TMV and TMV-GFP(TE1). At nine days post-inoculation, GFP fluorescence and MP expression were detected in systemic leaves of Nb+MP plants co-inoculated with TMV-GFP(TE1) and GUS, and Nb plants co-inoculated with TMV and TMV-GFP(TE1) (S2 Fig). Because MP expression could occur from the transgene, wild type virus, or mutant virus, presence of the mutant TE1 MP sequence was confirmed by sequencing six mutant progeny from each of these two groups of plants. In contrast, in Nb plants co-inoculated with pCB301 vectors to express TMV-GFP(TE1) and GUS, GFP fluorescence was confined to agroinfiltration sites on lower leaves and was never observed in systemic leaves, indicating that GUS expression was incapable of complementing TMV-GFP (TE1). Thus, transgenic MP expression is sufficient to complement the trafficking-defective TMV-GFP(TE1) mutant.

TMV MP complements a PSTVd mutant deficient for mesophyll entry
The Nb+MP plants and PSTVd mutants described above were used to assess the ability of TMV MP to mediate RNA trafficking between diverse cell types. Full-length in vitro transcripts of wild type (WT) PSTVd and each of the mutants were rub-inoculated onto the upper surfaces of the first two true leaves of Nb and Nb+MP plants. Accumulation of PSTVd RNAs in the inoculated leaves was analyzed in three biological replicates with 4 plants each. Extracts were obtained 10 days post-inoculation (dpi) and levels of circular PSTVd RNA, the functional form of the genome, were determined by RNA blot (Fig 2A). When RNA levels were quantified and compared between Nb and Nb+MP plants, no significant differences in PSTVd accumulation were observed in inoculated leaves following WT, 76AU, and 156AU infection ( Fig  2B), indicating that TMV MP does not generally enhance PSTVd replication or spread. However, PSTVd RNA levels were significantly increased in Nb+MP plants compared to Nb plants

PLOS PATHOGENS
TMV MP complements a trafficking impaired PSTVd mutant after GG infection (Fig 2B), suggesting TMV MP may enable mesophyll entry of the GG mutant, thereby increasing the number of infected cells.
To test this hypothesis, in situ hybridization was performed on transverse sections of leaves from Nb and Nb+MP plants that were rub-inoculated with the GG mutant. Confirming previous results [21], PSTVd GG was observed only in upper epidermal (uEp) cells of Nb plants (Fig 3). In contrast, the GG mutant was detected in all cell types, including uEp, palisade mesophyll (Pm), spongy mesophyll (Sm), and lower epidermis (lEp), in similarly inoculated leaves of Nb+MP plants (Fig 3), allowing us to conclude that TMV MP enables passage of the GG mutant through plasmodesmata connecting upper epidermis and palisade mesophyll. Transverse section in situ hybridization was also performed on plants that were rub-inoculated with the 76AU and 156AU mutants. Unlike the GG mutant, no obvious differences were observed in the cellular distribution of these PSTVd mutants in inoculated leaves of Nb compared to Nb +MP plants (Fig 3).

TMV MP does not complement PSTVd mutants deficient for phloem entry
To test whether TMV MP can mediate PSTVd trafficking from bundle sheath cells to phloem, the ability of WT, GG, 76AU, and 156AU to systemically infect plants was assessed at 10-, 20-, and 30 dpi following rub inoculation of eight Nb and Nb+MP plants. As local infection is prerequisite for systemic infection, the presence of circular PSTVd RNA in rub-inoculated leaves was confirmed at 10 dpi in all cases ( (Fig 4B), suggesting that TMV MP is unable to complement mutants deficient for phloem loading, and specifically for trafficking from bundle sheath cells to phloem [22].

Systemic infection of Nb+MP plants by the GG mutant is not due to reversion
We and others have shown that PSTVd replication occurs with a high error rate, generating large numbers of variants in progeny populations that can be considered quasispecies In vitro transcripts of WT PSTVd and GG, 76AU, and 156AU mutants were rub-inoculated onto the upper surfaces of the first two true leaves of Nb and Nb+MP plants. Inoculated leaves were collected at 10 dpi and pooled. Total RNA samples were analyzed by RNA blot to detect PSTVd RNA (circular form, PSTVd-C), and levels were normalized to ribosomal RNA (rRNA) loading controls. Three biological replicates are indicated by number, and each included four plants. P, positive control, M, Mock. (B) RNA blot signals from (A) were quantified using Quantity One software. For each treatment the Nb+MP group was normalized to the Nb group, which was set to 1. Shown are mean +/-SD values of the three biological replicates. Asterisk indicates significant difference (p < 0.05) by Student's t test. https://doi.org/10.1371/journal.ppat.1011062.g002

PLOS PATHOGENS
TMV MP complements a trafficking impaired PSTVd mutant  [22,38,39]. Thus, it is possible that apparent complementation leading to systemic infectivity of the GG mutant (178G/179G) observed in Nb+MP plants might be the result of reversion to WT and/or functional single mutants (U178G or U179G) [21]. It is also conceivable that additional beneficial mutations could compensate the original mutations and enable systemic infection. To rule out these possible scenarios, full-length progeny genomes from both inoculated (local, L) and systemic (S) leaves of three GG-infected Nb and Nb+MP plants were analyzed by Sanger sequencing.   (Table 1). Since the GG mutant is defective for mesophyll entry in Nb plants systemic infection is impossible, and only genomes from local leaves of Nb plants could be sequenced. Four of the six progeny genomes sequenced maintained the double mutation, and partial reversion (U178G/U179G to U179G) was detected in the remaining two cases. Similar partial reversion was observed in our previous study [21]. In GG-infected Nb+MP plants, six progeny genomes from local leaves and five from systemic leaves were sequenced. Except for a single additional mutation (U283A) detected in one genome from local leaves and one partial reversion (U178G/U179G to U179G) observed in a genome from systemic leaves, all others retained the double mutation (Table 1). In total, the 178G/179G double mutation was maintained in 13 of 17 sequenced genomes, while one acquired a single additional mutation and 3 experienced partial reversion to U179G. Therefore, successful mesophyll entry of the GG mutant in Nb+MP plants, and ensuing systemic infection, is unlikely to be caused by reversion and/or additional mutations, confirming TMV MP mediated complementation.

TMV MP binds wild type and mutant PSTVd RNAs
TMV MP binds viral RNA to enable spread [11,40]. To determine if impaired binding might be responsible for the inability of MP to complement PSTVd 76AU and 156AU, binding was directly tested by electrophoretic mobility shift assay (EMSA). Since PSTVd RNA in solution adopts a structure similar to that observed in plant cells [41], in vitro transcripts were prepared and folded to ensure the formation of homogenous structures. Ethidium bromide staining revealed a single band after 6% native polyacrylamide gel electrophoresis of in vitro transcripts of WT PSTVd and the GG, 76AU, and 156AU mutants, suggesting the formation of homogenous structures in all cases (Fig 5A). TMV MP was expressed as a six histidine, maltose binding protein fusion (His 6 MBP-MP) in E. coli strain BL21 and purified from cell extracts using metal affinity chromatography with Ni-NTA agarose (Fig 5B). MBP was chosen because it often enhances solubility of recombinant proteins and does not bind nucleic acids [42]. To perform EMSA, increasing amounts of protein (0 to 320 ng) were incubated with 1 ng PSTVd in vitro transcripts in 10 μl binding buffer. Following 6% native gel electrophoresis, RNA-protein complexes were visualized by RNA blot with labeled PSTVd probe. EMSA analysis showed that His 6 MBP-MP was able to bind WT PSTVd and all three mutants (Fig 5C). Binding curves were plotted using quantified EMSA data, and no significant differences were observed in relative binding efficiency of His 6 MBP-MP with WT PSTVd and the mutant RNAs. Under the conditions employed, half the RNA was bound at protein concentrations of 0.036 μM (WT),   Fig 5D). This is consistent with a previous study which showed that TMV MP binds with similar affinity to RNA and singlestranded DNA (ssDNA) through binding sites comprising only 4 to 7 nucleotides, but does not bind double-stranded DNA (dsDNA) [11]. Binding specificity of our fusion protein was confirmed by testing 1 ng folded PSTVd RNA transcripts with 40 ng His 6 MBP-MP alone, or in the presence of 100 ng heterologous ssDNA or dsDNA. Excess ssDNA completely abolished binding of His 6 MBP-MP fusion to PSTVd RNAs, while dsDNA had no apparent effect ( Fig 5E). Additional positive and negative control experiments were performed to further test the binding properties of the His 6 MBP-MP fusion protein. TMV MP has been reported to bind a 182 nucleotide single stranded fragment of the d24 mutant of the spinach chloroplast petD gene [11]. Sense and antisense petD-d24 fragments were cloned and in vitro transcripts prepared. Sense and antisense fragments were annealed, and double-stranded petD-d24 (dspetD-d24) RNA was prepared and purified from a native PAGE gel (5%). Then, single-stranded petD-d24 (sspetD-d24) RNA (sense) and dspetD-d24 RNA were used to repeat the competition assay described in Fig 5E. Consistently, binding of His 6 MBP-MP to PSTVd RNAs was abolished by excess sspetD-d24 RNA, but not dspetD-d24 RNA (Fig 5F). Thus, MP does not selectively bind PSTVd RNA. Binding specificity of His 6 MBP-MP was further tested by incubating 1 ng folded sense petD-d24 transcripts with 40 ng His 6 MBP-MP fusion protein alone, or in the presence of 100 ng heterologous ssDNA or dsDNA. His 6 MBP-MP was able to bind the petD-d24 transcript, and binding was efficiently competed by ssDNA but not dsDNA (Fig 5G). We next asked whether binding by the fusion protein was due to the His 6 MBP tag, which was prepared by the same procedure used to express the His 6 MBP-MP fusion protein (Fig 5H). Binding was tested by incubating 1 ng WT PSTVd with 40 ng His 6 MBP alone or in the presence of excess ssDNA or dsDNA. No binding was observed under any condition (Fig 5I). Thus, while the His 6 MBP-MP fusion protein has not been tested for its ability to support movement of TMV MP mutants, it is clearly capable of binding single stranded nucleic acids. Given these results, it is unlikely that the inability of TMV MP to mediate phloem entry of 76AU and 156AU is due to reduced binding.

TMV co-infection, but not TMV MP and CP expression, complements the PSTVd 76AU mutant deficient for phloem entry
It is known that CP is required for systemic infection by TMV [27,28,43]. The PSTVd 76AU and 156AU mutants deficient for phloem entry are capable of spread between other cell types in inoculated leaves, and both are systemically infectious when delivered directly into phloem ethidium bromide staining. Lanes 1-4 contained WT PSTVd, GG, 76AU, and 156AU, respectively. M, size marker. (B) Preparation of His 6 MBP-MP fusion protein (~73 kDa). M, size marker; lane 1, total protein extracted from E. coli strain BL21 prior to IPTG induction; lane 2, total protein following 0.1 M IPTG induction; lane 3, supernatant containing excess His 6 MBP-MP following metal affinity chromatography with Ni-NTA agarose and centrifugation; lane 4, His 6 MBP-MP eluted from the Ni-NTA agarose pellet fraction. Proteins were visualized by Coomassie blue staining following electrophoresis through 10% SDS-polyacrylamide gel. (C) EMSA assay. Folded and unlabeled PSTVd WT and mutant in vitro transcripts (1 ng) were incubated with increasing amounts of His 6 MBP-MP. Binding was analyzed by EMSA through 6% native polyacrylamide gel, followed by RNA blot using labeled PSTVd RNA probe. Three biological replicates were performed, and a representative replicate for each treatment is shown. (D) Binding was quantified by analyzing data from (C) using Quantity One software. Binding curves were plotted and fit using the Hill function of OriginPro 8.  (Fig 6A). These results suggest that the combination of TMV MP and CP is not sufficient to complement the phloem loading defects of these mutants.
TMV encodes 123 kDa and 183 kDa replicase proteins in addition to MP and CP. To test whether these additional viral factors may also participate in RNA trafficking, and specifically phloem loading, non-transgenic Nb plants rub-inoculated with WT PSTVd, 76AU, or 156AU were agroinoculated with a pCB301 vector expressing TMV or GUS (negative control) at three dpi. Symptoms of TMV infection, including stunting and leaf mosaic, were recorded at 10, 20, and 23 dpi (corresponding to 7, 17, and 20 days after agroinoculation with TMV). PSTVd infection of N. benthamiana is asymptomatic, but this species is very sensitive to TMV. Accordingly, plants began to exhibit obvious TMV disease symptoms (stunting and leaf mosaic) by 10 dpi and succumbed to lethal necrosis by 23 dpi (S7A Fig). To further confirm TMV infection, CP transcripts were detected in inoculated leaves of these plants by RT Ten progeny clones from the two 76AU infected Nb+TMV plants (five from each plant) were sequenced at 20 dpi. From one plant, all five progeny clones maintained the original

PLOS PATHOGENS
TMV MP complements a trafficking impaired PSTVd mutant 76AU mutation without additional mutations. From the other, two progeny maintained the original mutation without additional mutations, one maintained the mutation but gained another (76AU/G245A), and two reverted to wild type. These results suggest that TMV infection can complement the phloem loading defect of the 76AU mutant, albeit with reduced efficiency, and implicate TMV replicase proteins in phloem loading. However, TMV proteins cannot support entry of the 156AU mutant into the phloem, suggesting that unknown host factors participate in RNA transport at the bundle sheath-phloem boundary. Alternatively, the PSTVd mutants may be unable to effectively interact with TMV proteins other than MP.

Conclusion
Previous studies showed that expression of TMV MP in tobacco (N. tabacum cv. Xanthi) increases the SEL of plasmodesmata between mesophyll cells mainly in mature leaves [10,44]. In the same transgenic lines, SEL was also increased between mesophyll and bundle sheath cells, but not between bundle sheath and phloem parenchyma cells, even though MP was observed in plasmodesmata of these cell types [26]. This suggests that MP cannot productively interact with plasmodesmata connecting bundle sheath and phloem parenchyma, which might explain why MP alone is unable to complement PSTVd 76AU and 156AU mutants that are deficient for phloem entry. However, MP accumulation in plasmodesmata of bundle sheath and phloem parenchyma cells, and the ability of MP to increase SEL in the N. benthamiana plants used in the present study, have not been examined. It is also possible that TMV requires MP and additional viral and cellular proteins, or utilizes a non-MP mediated mechanism, to load into the phloem. Consequently, we tested and found that co-infection with TMV, but not the combination of MP and CP expression, could complement the 76AU mutant, suggesting a role for TMV replicase proteins in phloem loading. In contrast, the 156AU mutant was not complemented by either condition, implying that TMV proteins alone are insufficient in this case and implicating host factors in phloem loading. Alternatively, PSTVd RNAs may not effectively interact with TMV proteins other than MP. Nevertheless, TMV MP alone is sufficient to mediate transport of GG mutant PSTVd RNA, and by extension TMV RNA, between upper epidermis and palisade mesophyll cells, validating this approach for testing the RNA trafficking functions of MPs.
It should be noted that transgenic Nb plants expressing TMV MP, as well as plants expressing the MP of Red clover necrotic mottle virus (RCNMV), have previously been shown to support cell-to-cell and systemic spread of heterologous viral RNAs, including that of insectinfecting Flock house virus [45]. On the other hand, our results indicate that by itself, TMV MP cannot support systemic movement. Thus, an important advantage of PSTVd, illustrated here, is the availability of RNA mutants defective for directional movement through specific plasmodesmal gates, allowing analysis of MP trafficking function in fine detail. It is also interesting to note that proteins of different viruses can have different activities. For example, both the MP and CP of Brome mosaic virus (BMV) are required for cell-to-cell spread and long distance movement [46].
In summary, this study reports a novel method to analyze the function of movement proteins. Utilizing PSTVd mutants defective for movement between different cell types, the ability of TMV MP to mediate RNA trafficking between upper epidermis and palisade mesophyll, and between bundle sheath and phloem, was tested. We showed that TMV MP enabled mesophyll entry of the GG mutant which is defective for this process, indicating TMV MP is capable by itself of mediating RNA trafficking from epidermal cells to mesophyll cells. In contrast, TMV MP, and the combination of MP and CP expression, failed to support systemic infection by the 76AU and 156AU mutants defective for phloem entry. However, TMV infection

PLOS PATHOGENS
TMV MP complements a trafficking impaired PSTVd mutant complemented 76AU but not 156AU, suggesting the involvement of viral replicase proteins and unidentified cellular factors in phloem entry in addition to MP. These observations confirm that different gating mechanisms exist between different cell types and highlight the complexity of phloem loading compared to other types of cell-to-cell movement, suggestive of a role as gatekeeper to the vascular system.

Plants and inoculation
Nb+MP plants [35], a gift from Dr. James Culver, were grown in The Ohio State University Biotechnology Support Facility along with wild type Nb plants. Plasmid pRZ6-2 containing WT PSTVd was a gift from Dr. Robert Owens. Plasmids carrying the cDNAs of WT PSTVd and three mutants (GG, 76AU, and 156AU) were linearized with Hind III (Promega), followed by in vitro transcription using the MEGAscript T7 Transcription Kit (ThermoFisher Scientific). PCR products containing the T7 promoter region and the 182 bp sense and antisense spinach petD-d24 sequences were also subjected to in vitro transcription using MEGAscript T7 Transcription Kit to prepare petD-d24 RNA. PSTVd and petD-d24 riboprobes labeled with digoxigenin (DIG) were prepared using MAXIscript T3 Transcription Kit (ThermoFisher Scientific) and MAXIscript T7 Transcription Kit (ThermoFisher Scientific), respectively. Templates were plasmid pInter(-) linearized with SpeI and a PCR product containing T7 promoter and petD-d24 antisense sequence, respectively. After purification using MEGAclear Transcription Clean-Up Kit (ThermoFisher Scientific), RNA transcripts (300 ng) were rub-inoculated onto the upper surface of first two true leaves of 2-week-old Nb and Nb+MP plants that were previously dusted with carborundum powder [21].
RNA isolation and blotting. Plant RNA samples were prepared using Trizol reagent (Invitrogen) according to the manufacturer's instructions. Prior to subsequent experiments, RNA samples were incubated with DNase I (Invitrogen) at 37˚C for 2 h to remove genomic DNA. RNA blotting was performed essentially as described [47], except digoxigenin (DIG)labeled probe instead of [α-32 P]-UTP-labeled probe was used in the present study. To prepare (DIG)-labeled PSTVd riboprobes, pInter(-) plasmid was linearized with SpeI (New England Biolabs) [48]. This was followed by in vitro transcription using the MAXIscript T7 Transcription Kit (ThermoFisher Scientific). Signal was captured using myECL Imager (ThermoFisher Scientific).
PSTVd progeny sequencing. Full-length progeny sequences were amplified from greater than full-length reverse transcription products as described [22], using forward primer PSTVd1F (5'-CGGAACTAAACTCGTGGTTCCT-3') and reverse primer PSTVd1S (5'-AGGAACCAACTGCGGTTCCA-3'). RT-PCR amplified progeny products were purified through agarose gel using the Zymoclean Gel DNA Recovery Kit (Zymo Research). Purified products were cloned into PCR2.1 using a TA Cloning Kit (Invitrogen). DNA sequencing was performed at the Genomics Shared Resource of The Ohio State University.
Vector construction. The entire TMV MP ORF was amplified from NB+MP plants and purified following agarose gel electrophoresis using the Zymoclean Gel DNA Recovery Kit. The ORF was then cloned into the pCR8 TOPO gateway vector using the pCR8/GW/TOPO TA Cloning Kit (ThermoFisher Scientific). Through an LR reaction using Gateway LR Clonase Enzyme mix (ThermoFisher Scientific), the MP ORF was cloned into the pDEST-HisMBP vector for expression as a His6 maltose binding protein fusion (His 6 MBP-MP) [49]. Constructs were sequenced to ensure correct sequence and orientation. To construct the His 6 MBP expression vector, the entire MBP cDNA was amplified from pDEST-HisMBP vector. After gel purification, MBP cDNA was cloned into pET-28a expression vector carrying N-terminal His tag through homologous recombination. Through the same method, pCB301 vector carrying fulllength TMV, kindly provided by Dr. Jiejun Peng at Ningbo University, was used as a template to clone TMV CP full length cDNA. CP cDNA was cloned into the pCAMBIA1301 vector to establish CP-expressing vector using T4 DNA Polymerase (ThermoFisher Scientific)-mediated ligation. MYC-expressing pCAMBIA1301 vector, TMV-GFP-expressing pCB301 vector and GUS-expressing pCB301 vector were kindly provided by Dr. Kelei Han at Ningbo University. The TMV-GFP(TE1) mutant was constructed by site-directed mutagenesis. The TMV-GFP (ΔCP) mutant was prepared by homologous recombination.
Protein purification. The pDEST-HisMBP vector carrying TMV MP and pET-28a vector carrying MBP were transfected into E. coli strain BL21 (EMD Millipore). Bacteria were cultivated at 37˚C until a A600 value close to 0.6 was reached. IPTG was added to 0.1 mM and, following incubation for a further 4 h, cultures were centrifuged at 5000 g for 15 min at 4˚C. To extract protein, cell pellets were resuspended in native binding buffer (50 mM NaH 2 PO 4 , 0.5 M NaCl, pH 8.0, 0.1% (v/v) Triton X-100). Lysozyme (1 mg/ml) was then added to the cell suspension. Cell lysis was performed on ice for 30 min. After sonication, cell lysates were centrifuged at 5000 g for 15 min at 4˚C to collect supernatant, which was incubated with Ni-NTA agarose (Qiagen) to adsorb His tagged fusion protein. Ni-NTA agarose and bound protein were collected by centrifugation. After washing the pellet with native washing buffer (50 mM NaH 2 PO 4 , pH 8.0, 0.5 M NaCl, 20 mM imidazole), fusion protein was eluted using native elution buffer (50 mM NaH 2 PO 4 , pH 8.0, 0.5 M NaCl, 0.25 M imidazole). Gel filtration was performed with Sephadex G-25 beads (Sigma-Aldrich) prior to subsequent assays. Proteins were visualized by Coomassie blue staining of a 10% SDS-polyacrylamide (37.5:1) gel.
In competition assays, PSTVd or petD-d24 transcripts (1 ng) were incubated with 40 ng His 6 MBP-MP fusion protein or 40 ng His 6 MBP protein alone, or in the presence of 100 ng ssDNA, or dsDNA. To prepare dsDNA samples, a mixture of pCR8 TOPO TA vector, pGEM-T easy vector, and pDEST-HisMBP vector (1:1:1 ratio, w/w) was first digested with RNase A overnight to remove all potential RNA contamination, followed by three rounds of phenol-chloroform extraction to remove RNase A. To prepare ssDNA samples, the three plasmids were linearized using EcoRI, SpeI, and EcoRI, respectively. Plasmid mixtures were incubated at 98˚C for 10 min, followed by freezing in liquid nitrogen prior to use. To prepare dspetD-d24 RNA, sense and antisense petD-d24 RNAs were mixed (1:1 ratio, w/w). The mixture was denatured at 98˚C for 10 min, followed by slow cooling to room temperature. The resulting dspetD-d24 RNA was purified twice through 5% native PAGE gel electrophoresis to completely remove ssRNA.
In situ hybridization. Preparation of inoculated leaf transverse paraffin sections and in situ hybridization was performed as described [21]. Two inoculated leaves from each of three GG, 76AU, or 156AU mutant infected Nb and Nb+MP plants were used. Two sections were prepared from each leaf, and four visual fields were examined per section.