Deficiency of the eIF4E isoform nCBP limits the cell-to-cell movement of a plant virus encoding triple-gene-block proteins in Arabidopsis thaliana

One of the important antiviral genetic strategies used in crop breeding is recessive resistance. Two eukaryotic translation initiation factor 4E family genes, eIF4E and eIFiso4E, are the most common recessive resistance genes whose absence inhibits infection by plant viruses in Potyviridae, Carmovirus, and Cucumovirus. Here, we show that another eIF4E family gene, nCBP, acts as a novel recessive resistance gene in Arabidopsis thaliana toward plant viruses in Alpha- and Betaflexiviridae. We found that infection by Plantago asiatica mosaic virus (PlAMV), a potexvirus, was delayed in ncbp mutants of A. thaliana. Virus replication efficiency did not differ between an ncbp mutant and a wild type plant in single cells, but viral cell-to-cell movement was significantly delayed in the ncbp mutant. Furthermore, the accumulation of triple-gene-block protein 2 (TGB2) and TGB3, the movement proteins of potexviruses, decreased in the ncbp mutant. Inoculation experiments with several viruses showed that the accumulation of viruses encoding TGBs in their genomes decreased in the ncbp mutant. These results indicate that nCBP is a novel member of the eIF4E family recessive resistance genes whose loss impairs viral cell-to-cell movement by inhibiting the efficient accumulation of TGB2 and TGB3.

Scientific RepoRts | 7:39678 | DOI: 10.1038/srep39678 of viral proteins from MNSV and CMV RNAs 21,22 . Thus, although the roles of some recessive resistance genes have been partially elucidated, understanding of the variety of recessive resistance genes and their roles remains limited.
Alpha-and Betaflexiviridae are groups of flexuous, filamentous viruses that predominantly infect plants, and encode an RNA-dependent RNA polymerase (RdRp), a 30K-type movement protein (MP) or triple-gene-block (TGB)-type MPs, and a coat protein (CP). Some viruses in Alpha-and Betaflexiviridae encode additional proteins in their genome. The most extensively studied of these plant viruses are members of the genus Potexvirus in the family Alphaflexiviridae, which have one genomic RNA with a cap and poly(A) tail [23][24][25] . The 5′ -terminal open reading frame (ORF) encoding RdRp is translated directly from genomic RNA, but the 3′ -proximal ORFs encoding TGB1, TGB2, TGB3, and CP are translated from subgenomic RNAs (sgRNAs) 26,27 , which are generated during virus replication 28 and possess a cap 29,30 and the same 3′ ends as the genomic RNA 25 . TGB1 and CP were shown to be translated from sgRNA1 and sgRNA3, respectively, while TGB2 and TGB3 are translated from sgRNA2 of Potato virus X (PVX) 26,27 .
In this study, we found that an A. thaliana mutant of another member of the eIF4E family gene, nCBP, was resistant to a potexvirus, Plantago asiatica mosaic virus (PlAMV). Cell-to-cell movement of PlAMV was delayed in an ncbp mutant compared to the wild type, although viral accumulation in single cells of the mutant and the wild type did not differ. The accumulation of TGB2 and TGB3 was decreased in the ncbp mutant compared with the wild type.

Results
PlAMV propagation was delayed in ncbp mutants. A. thaliana has three types of eIF4E isoforms, eIF4E, eIFiso4E, and nCBP. Two homologs of eIF4E, eIF4E1b and eIF4E1c, are also encoded by A. thaliana 31 (see Supplementary Fig. S1). To determine whether any of the eIF4E isoforms has a role in the infection cycle of PlAMV, A. thaliana mutant lines bearing a T-DNA insertion or a point mutation in an eIF4E family gene were mechanically inoculated with PlAMV-GFP, a GFP-expressing variant of PlAMV. The inoculated leaves were examined for fluorescence from PlAMV-GFP at 4 days post inoculation (dpi). Infection foci on the inoculated leaves of eif4e, eif4e1b, eif4e1c, and eifiso4e mutants were nearly the same size as those on the A. thaliana ecotype Columbia-0 (Col-0) leaves (Fig. 1a). In contrast, the foci on the inoculated leaves of two ncbp T-DNA insertion lines, ncbp-1 and ncbp-2 (see Supplementary Fig. S2), were noticeably smaller than those of Col-0 (Fig. 1a). To quantify the virus accumulation, total RNA was extracted from each of the inoculated leaves from the mutants and analyzed by quantitative RT-PCR (RT-qPCR) with PlAMV-specific primers 32 . The relative accumulation of PlAMV viral RNA in eif4e, eif4e1b, eif4e1c, and eifiso4e mutants did not differ from that in Col-0, whereas viral RNA in ncbp-1 and ncbp-2 mutants was drastically decreased to approximately 15% and 25% of that in Col-0, respectively (Fig. 1b).
To explore whether PlAMV systemically infects ncbp mutants, we mechanically inoculated ncbp mutants and Col-0 with PlAMV-GFP. At 3 weeks post inoculation (wpi), the GFP signal was observed in stems and upper leaves of Col-0, but no GFP signal was apparent in the upper leaves of the ncbp-1 mutant (Fig. 1c). To quantify the accumulation of viral RNA, we extracted total RNA from non-inoculated upper leaves of the ncbp-1 and ncbp-2 mutants, and Col-0, and amplified viral RNA by RT-PCR. At 3 wpi, PlAMV infected the Col-0 plants systemically, while PlAMV did not propagate in upper leaves of the ncbp-1 and ncbp-2 mutants (see Supplementary Table S1 and Fig. 1d). At 4 wpi, 3 of 5 ncbp-1 plants, and none of the 6 ncbp-2 plants, harbored PlAMV RNA (see Supplementary Table S1). The accumulation of PlAMV in the upper leaves of the ncbp-1 mutants at 4 wpi was lower than in Col-0 plants (see Supplementary Fig. S3), suggesting that PlAMV systemic infection delayed in these mutants.
To confirm that the decreased accumulation of PlAMV in the ncbp mutants was due to the loss of nCBP, we cloned and introduced the genomic DNA sequence of nCBP with its possible promoter and terminator regions into the ncbp-1 mutant. Three independent transgenic lines (#1A, #3E, and #3F) were obtained and analyzed for their expression of nCBP. While we failed to detect the nCBP protein in the ncbp-1 mutant, three transgenic lines expressed it (Fig. 2a). When those transgenic lines were then inoculated mechanically with PlAMV-GFP, all of the nCBP-complemented lines had GFP foci similar in size to those of Col-0 at 4 dpi (Fig. 2b). The RT-qPCR analysis confirmed that the accumulation level of viral RNA in the inoculated leaves did not drastically differ between the transgenic lines and Col-0 (Fig. 2c). These results suggest that nCBP is essential for the efficient accumulation of PlAMV in inoculated leaves. We also confirmed that PlAMV systemically infects these complemented lines at 3 wpi (see Supplementary Table S1).
Cell-to-cell movement of PlAMV is delayed in ncbp mutant. To analyze the efficiency of cell-to-cell movement of PlAMV, we monitored the sizes of PlAMV-GFP infection foci in inoculated leaves of Col-0 and the ncbp-1 mutant. We often observed the fusion of multiple infection foci during the mechanical inoculation assay, which limits our statistical evaluation of the infection foci size (Figs 1a and 2b). Therefore, we utilized a particle delivery system to introduce the PlAMV-GFP cDNA into a single cell. The bombarded leaves were analyzed at 1, 1.5, and 2 days post bombardment. The PlAMV-GFP in the bombarded leaves of Col-0 spread more rapidly than in those of the ncbp-1 mutant (Fig. 3a). To quantitatively analyze viral cell-to-cell movement, we measured the area expressing GFP signals. We found that the GFP-expressing area in the ncbp-1 mutant was drastically smaller than that in Col-0 (Fig. 3b). These results suggest that the cell-to-cell movement of PlAMV was delayed in the ncbp-1 mutant compared to Col-0.
Replication efficiency of PlAMV is not compromised in ncbp-1 mutant at a single-cell level. We next investigated whether the delay of viral cell-to-cell movement in the ncbp mutants was caused by a decreased replication efficiency of the virus. We isolated mesophyll protoplasts from the ncbp-1 mutant and Col-0 and Scientific RepoRts | 7:39678 | DOI: 10.1038/srep39678 transfected them with infectious PlAMV cDNA. Total RNA was extracted from the cells at 0, 12, and 24 hours post inoculation (hpi), and the amount of viral RNA was analyzed by RT-qPCR using the cotransfected GFP gene as an internal control. RT-qPCR analysis revealed that the accumulation level of viral RNA in the ncbp-1 mutant protoplasts was similar to that in Col-0 protoplasts (Fig. 4a). We further performed northern blot analysis to detect viral genomic RNA, sgRNAs, and negative-strand viral RNA that serves as a template for synthesis of the RNA genome. We found that viral genomic RNA, sgRNAs and negative strand RNA accumulated to a similar degree in both the ncbp-1 and Col-0 protoplasts (Fig. 4b, top and middle panels, Supplementary Fig. S4). These results show that nCBP is not an essential factor for PlAMV genomic replication, suggesting that nCBP is involved in viral cell-to-cell movement. Accumulation of TGB2 and TGB3 is decreased in ncbp mutant. Since potexviruses need TGB1, TGB2, TGB3 and CP for their cell-to-cell movement, we analyzed the accumulation of these viral proteins in PlAMV-transfected protoplasts of the ncbp-1 mutant and Col-0. Immunoblotting of the protoplast samples harvested at 3 dpi revealed that the accumulation of RdRp did not differ between the ncbp-1 and Col-0 protoplasts (see Supplementary Fig. S5). However, we could not clearly detect TGB2 and TGB3 proteins in protoplast samples (see Supplementary Fig. S5), possibly because of their low accumulation levels. Therefore, we  Fluorescence images of more than five foci in (a) were processed using ImageJ software v1.40 (NIH) to measure the size of viral infection foci. The sizes are normalized to Col-0 at 1 dpi. Error bars represent standard errors of at least six measurements. Asterisk indicates a significant difference compared with Col-0 (two-tailed Student's t-test, asterisk; P < 0.05, triple asterisk; P < 0.001). Experiments were replicated three times.
used an Agrobacterium-mediated transient expression (agroinfiltration) assay in which infiltrated regions of plant leaves were infected uniformly with virus. Leaves of the ncbp-1 mutant and Col-0 were agroinfiltrated with Agrobacterium carrying infectious PlAMV-GFP cDNA. The leaves were harvested at 4 dpi and the accumulation of viral proteins was analyzed by immunoblotting (Fig. 5a). Since the band for TGB3 exactly overlapped with a nonspecific band, we separated soluble and insoluble fractions (S30 and P30 fraction, respectively) by ultracentrifugation to exclude the nonspecific band based on the fact that the TGB3 protein is membrane-associated 33 . The nonspecific band was detected only in the S30 fraction, enabling us to evaluate the accumulation of TGB3 in the P30 fraction (see Supplementary Fig. S6 and Fig. 5b). Although the accumulation of RdRp, TGB1, GFP-CP and CP was similar between Col-0 and the ncbp-1 mutant plants, that of TGB2 and TGB3 drastically decreased in the ncbp-1 mutant compared to that in Col-0 ( Fig. 5a and b). The intensities of protein bands in each leaf sample were quantified (Fig. 5c). We evaluated the ratio of TGB1, TGB2, TGB3, and CP accumulation relative to that of RdRp, since the amount of viral RNA and RdRp was almost the same between the ncbp-1 mutant and Col-0 protoplasts (see Supplementary Fig. S5 and Fig. 4). Band quantification confirmed that the amounts of TGB2 and TGB3 drastically decreased in ncbp-1 leaves compared to Col-0 leaves, while the relative amount of TGB1 and CP in the ncbp-1 leaves did not differ (Fig. 5c).  To determine which viruses require nCBP for their efficient accumulation, we mechanically inoculated the ncbp-1 mutant and Col-0 with viruses from various families and tested their accumulation. In Alphaflexiviridae, we analyzed two

Discussion
We identified nCBP as a novel recessive resistance gene against plant viruses in Alphaflexiviridae and Betaflexiviridae. This identification revealed that all members of the eIF4E family (eIF4E, eIFiso4E, and nCBP) can act as recessive resistance genes.
Recessive resistance exhibited by eIF4E deficiency is thought to be caused by the specific use of eIF4E family gene products, eIF4E or eIFiso4E, by plant viruses 2 . Generally, plants encode three eIF4E isoforms, namely, eIF4E, eIFiso4E, and nCBP 10 . The lack of eIF4E or eIFiso4E does not influence the viability of plants 16,31,34,35 , presumably due to their redundancy during translation initiation. However, the lack of eIF4E or eIFiso4E does decrease the infectivity of plant viruses; therefore, such viruses are thought to use specific eIF4E family gene products during their infection 2 . The specific use of eIF4E family gene products may be correlated with the unique translation initiation strategies of plant viruses. In fact, a large number of plant viruses do not possess the cap and/or poly(A) structures in their genomic RNA 36 . Because both of these structures play a critical role in translation initiation 37 , these viruses developed unique translation initiation strategies, e.g., recruiting specific eIFs directly to viral RNAs using their own cis-acting RNA elements 36 . Examples include the 3′ cap-independent translation element (3′ CITE) located within the 3′ UTR of viruses in Tombusviridae, Umbravirus, and Luteovirus. Each virus recruits specific eIFs to the 3′ CITE to facilitate the translation of viral proteins 38 . Therefore, eIF4E-mediated recessive resistance is effective against plant viruses lacking cap and/or poly(A) structures 4,19,36 . In this study, we showed that the infection of plant viruses with cap and poly(A) structures was inhibited in ncbp mutant plants. Our results showed that eIF4E family genes can serve as recessive resistance genes against viruses with cap and poly(A) structures.
In this study, we showed that cell-to-cell movement of PlAMV was inhibited in the ncbp mutant (Fig. 3); thus, we further analyzed the role of nCBP during PlAMV infection. Protoplast transfection assays revealed no significant difference between the ncbp-1 and Col-0 cells in the accumulation of PlAMV genomic RNA (Fig. 4). This result indicates that nCBP is not required for viral replication at the single-cell level, including translation of viral RdRp and its viral genomic RNA synthesis. In agreement with this result, the level of RdRp in PlAMV-agroinfiltrated leaves of the ncbp mutant was similar to that of Col-0 (Fig. 5a). However, the accumulation of TGB2 and TGB3 drastically decreased in PlAMV-infiltrated ncbp mutant leaves ( Fig. 5a and b), indicating that decreased accumulation of these proteins might cause the inefficient cell-to-cell movement of the virus. It still remains unclear why the accumulation of TGB2 and TGB3 decreased in the ncbp mutant. Considering that nCBP is a member of eIF4E isoforms, it is attractive to think that the translation of TGB2 and TGB3 from sgRNA2 may be specifically inactivated in the ncbp mutant, becuase sgRNA2 was shown to function as a template for translation of TGB2 and TGB3 in the case of PVX 26,27 . Otherwise, the stability of TGB2 and TGB3 could be reduced in the mutant. However, since sgRNA2 of PlAMV was below the detectable level in our northern blot analysis (Fig. 4b), it remains also possibile that the synthesis and/or stability of sgRNA2 might be affected during PlAMV infection in the ncbp mutant.
In this study, we showed that nCBP-mediated recessive resistance limits viral cell-to-cell movement (Fig. 3). During viral cell-to-cell movement, three potexviral movement proteins, TGB1, TGB2, and TGB3, function in a concerted manner. TGB2 and TGB3 induce formation of ER-derived TGB2/3 vesicles, which are subsequently directed to plasmodesmata (PD) 39,40 . TGB1 accumulates at the PD only when TGB2 and TGB3 are expressed 33,40 . TGB1 and TGB2 can increase the PD size exclusion limit 41,42 . In addition to TGBs, potexviral CP is considered an essential factor to move viral RNA between cells 43,44 . TGB2/3 vesicles, TGB1, CP, and viral RNA form a layered complex at the PD opening with ER membranes, possibly to promote efficient movement of the viral ribonucleoproteins 40 . The lack of any one of these movement-associated proteins should disable cell-to-cell movement of potexviruses. Therefore, the delay of cell-to-cell movement of PlAMV in the ncbp mutant can be explained by the inefficient accumulation of TGB2 and TGB3. Involvement of the eIF4E protein family in translation of movement proteins was reported in CMV 21 . In A. thaliana cum-1 mutant, which possesses a nonsense mutation in the eIF4E coding sequence, inefficient translation of the CMV 3a movement protein resulted in the inhibition of cell-to-cell movement of the virus 21 . The observation that two unrelated viruses, CMV and PlAMV, utilize specific eIFs for the accumulation of MP supports the importance of controlling viral MP accumulation.
To explore the universal role of nCBP in the plant-virus interaction, we inoculated ncbp mutants with viruses from various genera and examined their accumulation levels. We showed that viruses in the genera Potexvirus, Lolavirus, and Carlavirus require nCBP for their accumulation, whereas viruses in the genera Potyvirus and Tobamovirus do not (Fig. 6). One noticeable characteristic common among potexvirus, lolavirus, and carlavirus is that they encode TGB-type MPs. Considering that nCBP was required for the accumulation of TGB2 and TGB3 of PlAMV (Fig. 5), nCBP may also facilitate the accumulation of TGB2 and TGB3 from lolavirus and carlavirus to promote their movement. Since there are some genera of plant viruses other than Alphaflexiviridae and , and YoMV (f), and total RNA was extracted from inoculated leaves at 4 dpi. Virus accumulation was analyzed by quantitative RT-PCR using virus-specific primers. The accumulation of viral RNA was normalized relative to actin mRNA in each sample. The mean level of viral RNA in Col-0 was used as the standard (1.0). Error bars represent standard errors of 12 measurements from three independent experiments and asterisk and double asterisks indicate significant differences compared with Col-0 (one-tailed Student's t-test, asterisk; P < 0.05, double asterisk; P < 0.01). Betaflexiviridae, such as Hordeivirus, Pomovirus, Pecluvirus, and Benyvirus, that encode TGB proteins 26 , it would be interesting to explore whether these viruses require nCBP for their infection. It would also be interesting to explore whether viruses not encoding TGBs in Alphaflexiviridae and Betaflexiviridae are influenced by the nCBP mutation. Moreover, it is possible that the ncbp mutation is a determinant of the actual recessive resistance of crop cultivars against TGB-encoding viruses, in which the responsible genes for the resistance remain unknown. In addition, the artificial introduction of mutations in the nCBP gene may be a novel and robust strategy to provide crops with a virus-resistant trait, similar to eIF4E or eIFiso4E. Antibodies. Anti-TGB2 and TGB3 antisera were raised in rabbits using purified peptides (TGB2, GDNLHALPHGGRY; TGB3, KQTLHHGTQPSTDL) as antigen (eurofinsgenomics, Tokyo, Japan). The nCBP protein was expressed in Escherichia coli, using a pET30a vector, and the purified recombinant protein was used as an antigen. Anti-TGB1, CP and RdRp antibodies were prepared as described previously 32,43,45 . Plasmid construction. An infectious cDNA clone and a GFP-expressing vector of a PlAMV isolate 46 were constructed as described previously 47,48 . For the complementation assay, the nCBP gene with putative promoter and terminator sequences was PCR-amplified with primers Sl-At5g18110-up1374F and Nt-At5g18110-down1011R using total DNA extracted from Col-0 as template (for primer sequences, see Supplementary Table S2). The amplified product was digested with SalI and NotI restriction enzymes and inserted between SalI and NotI sites of pENTA 49 . To produce pFAST01-nCBPg, the region between attL1 and attL2 containing the nCBP expression cassette was cloned into a binary plasmid vector pFAST01 (Inplanta Innovations Inc., Kanagawa, Japan) using Gateway LR Clonase II enzyme mix (Thermo Fisher Scientific, Massachusetts, USA).

Methods
To produce RNA probes for the detection of positive-and negative-stranded viral RNA, the PCR-amplified fragment of the 3′ terminal region of a PlAMV isolate 46 (nucleotides from 5,101 to 6,102) was cloned into pCR-Blunt II-TOPO vector (Thermo Fisher Scientific), generating pCR-Pr-1 (in antisense orientation behind the T7 promoter to produce the negative-stranded RNA detection probe) and pCR-Pr-2 (in sense orientation to produce the positive-stranded RNA detection probe).
Virus isolate and inoculation. Mechanical inoculation with an extract of PlAMV-GFP-infected N. benthamiana plants and agroinoculation of PlAMV-GFP was performed as described previously 50 . CymMV (accession number, LC125633), AltMV 51 , LoLV 52 , YoMV (MAFF number 104033; National Institute of Agribiological Sciences GenBank), PVM (MAFF number 307027), TuMV 53 , and CMV 54 were also used for mechanical inoculation. Rosette leaves of three-week-old A. thaliana were inoculated with extracts of the upper leaves of N. benthamiana or A. thaliana plants, which were inoculated with each virus and infected systemically. Plant transformation. Agrobacterium tumefaciens strain EHA105 carrying pFAST01-nCBPg was used for transformation of A. thaliana by the floral-dip method, as described previously 55 . T1 seeds of transgenic plants were selected by GFP fluorescence expressed from the seed-specific OLE1 promoter encoded by the pFAST01-nCBPg.
Alignment and phylogenetic analysis of nCBP proteins. Sequences of eIF4E family genes, excluding AteIF4E1b and AteIF4E1c, were obtained from EST databases of the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov) and Sol Genomics Network (http://solgenomics.net). Predicted cDNA sequences of AteIF4E1b and AteIF4E1c were obtained from The Arabidopsis Information Resource (TAIR, https://www.arabidopsis.org). Amino acid alignments of the core region 10 (from His-37 to His-200 in Homo sapiens eIF4E) were preformed using ClustalW software. Phylogenetic trees were constructed from nucleotide alignments of the core region using the neighbor-joining and boot-strapping algorithms within the Mega 6.0 software.
Scientific RepoRts | 7:39678 | DOI: 10.1038/srep39678 Northern blot analysis. Total RNA (1 μ g) was analyzed with the digoxigenin (DIG) system (Roche). To produce probes for plus-and minus-strand viral RNA detection, pCR-Pr-1 and pCR-Pr-2 were digested with BamHI restriction enzyme and transcribed with T7 RNA polymerase. The intensities of RNA bands were quantitated using ImageJ software v1.40 (National Institutes of Health).

Particle bombardment. Particle bombardment was performed using a Biolistic PDS 1000/He Particle
Delivery System (Bio-Rad, California, USA), as described previously 56 . The area showing GFP signal was quantitated using ImageJ software v1.40.
Protoplast preparation and transfection. Arabidopsis protoplast preparation and transfection were performed as described previously 57 with modifications. We added 0.1 M mannitol to W5 solution. For virus inoculation, 100 μ g of 35S-driven virus infectious clone was added to 300 μ L of protoplast suspension (5 × 10 6 protoplasts/mL).