Nicotiana benthamiana was grown on Rock Fiber blocks (Nittobo, Tokyo, Japan) at 25°C with a 16 h light/8 h dark photoperiod for 2 weeks. Seedlings on the Rock Fiber blocks were transferred to pots filled with vermiculite and grown under the same conditions, being fertilized with 0.1% (v/v) Hyponex solution (Hyponex Japan, Osaka, Japan) once a week. 7–9 week-old plants were used and kept at 25°C after bombardment and agroinfiltration until observation.

The plasmids described below were constructed using conventional or Gateway cloning techniques (Invitrogen, Carlsbad, CA, United States). Coding sequences in the plasmids amplified by polymerase chain reaction (PCR) were confirmed by sequencing.

For construction of pTH2-126K/183K-GFP, the 126K/183K-encoding sequence was firstly amplified by PCR using pTLW3 that is an expression plasmid with wild-type ToMV genomic cDNA (Hamamoto et al., 1993) and a primer set of 126KdTAG/NcoI/F1 (5′-GAACCATGGCATACACACAAACAGC-3′; the NcoI recognition sequence is underlined) and 183KdTAA/NcoI/R1 (5′-CACCCATGGCACAACTAGAGCCATCAAGAA-3′; the NcoI recognition sequence is underlined). The PCR product was digested with NcoI and ligated into an NcoI-cut large fragment obtained from the GFP-encoding plasmid, 35Ω-sGFP(S65T) (Chiu et al., 1996), generating pTH2-126K-GFP.

pART7-NtREM-YFP and pART7-NtREM-DsRed were constructed as follows. Total RNA was extracted from N. tabacum cv Samsun NN using the NucleoSpin RNA Plant Kit (Takara, Shiga, Japan). First strand cDNA was synthesized from the total RNA extract with an oligo-dT primer, and ReverTra Ace reverse transcriptase (Toyobo, Osaka, Japan). To obtain the NtREM cDNA without the termination codon, first PCR was carried out using the first strand cDNA as templates and a primer set of attB1/NtREM/F1 (5′-AAAAAGCAGGCTACCATGGCAGAAGTAGAAGTTAA-3′) and attB2/NtREM/R2 (5′-AGAAAGCTGGGTTAAAACATCCAAGGAGTTTCT-3′). Second, PCR was performed with purified PCR products and attB adaptor primers (Sasaki et al., 2009). The second PCR products were purified and introduced to the entry vector pDONR/Zeo (Invitrogen). The resulting entry plasmid (pDONR/Zeo-NtREMdTAA) was recombined with the destination vectors pART7-GWC-YFP and pART7-GWC-DsRed (Sasaki et al., 2009), generating pART7-NtREM-YFP and pART7-NtREM-DsRed, respectively.

pART7-DsRed-NtREM was constructed as follows. pART7 (Gleave, 1992) was digested with SmaI and ligated with the Gateway cassette A from the Gateway Conversion kit (Invitrogen). The resulting plasmid (pART7-GWA) was linearized by digestion with EcoRI and KpnI. The DsRed cDNA sequence was amplified by PCR using pART7-DsRed (Sasaki et al., 2009) and a primer set of DsRed/EcoRI/F2 (5′-AAAAAGAATTCATGGACAACACCGAGGACGT-3′; the EcoRI recognition sequence is underlined) and DsRed/KpnI/R (5′-AAAAAGGTACCCTGGGAGCCGGAGTGGCGGG-3′; the KpnI recognition sequence is underlined), digested with EcoRI and KpnI, and ligated into the linearized pART7-GWA, generating the destination vector pART7-DsRed-GWA. An entry plasmid pDONR/Zeo-NtREM that contains the NtREM cDNA with the termination codon was obtained in the same way as pDONR/Zeo-NtREMdTAA was made, except that attB2/NtREM/R1 (5′-AGAAAGCTGGGTTTAAAAACATCCAAGGAGTT-3′) was used for first PCR instead of attB2/NtREM/R2. pDONR/Zeo-NtREM was recombined with pART7-DsRed-GWA.

pART7-erGFP was constructed as follows. The erGFP-coding sequence was amplified by PCR using pGLW3-erGFP that is a binary plasmid for inoculation of ToMV-erGFP (Sasaki et al., 2013) and a primer set of HindIII/erGFP (5′-AAAAAAAGCTTATGAAGACTAATCTTTTTCT-3′; the HindIII recognition sequence is underlined) and XbaI/erGFP (5′-TTTTTTCTAGATTAAAGCTCATCATGTTTGT-3′; the XbaI recognition sequence is underlined). The PCR product was digested with HindIII and XbaI and ligated into a large fragment obtained by digestion of pART7 (Gleave, 1992) with HindIII and XbaI, generating pART7-erGFP.

pGWnV3-NtREM and pGWcV-NtREM were generated by recombination between pDONR/Zeo-NtREMdTAA and the destination vectors pGWnV3 and pGWcV, respectively (Nishimura et al., 2015).

pGWnV3-MP and pGWcV-MP were made by recombination of pDONR-ToMV30K (Haque et al., 2008) with the destination vectors pGWnV3 and pGWcV, respectively (Nishimura et al., 2015).

For construction of pGWnV3-126K and pGWcV-126K, first PCR was carried out by using pTH2-126K-GFP and primers, attB1/126kdTAG/F1 (5′-AAAAAGCAGGCTTAACCATGGCATACACACAAACA-3′) and attB2/126kdTAG/R1 (5′-AGAAAGCTGGGTTTTGAGTACCTGCATCTACTT-3′). Second PCR was performed with purified PCR products and attB adaptor primers (Sasaki et al., 2009). The second PCR products were purified and introduced to the entry vector pDONR/Zeo (Invitrogen) to generate pDONR/Zeo-126K. pGWnV3-126K and pGWcV-126K were made by recombination of pDONR/Zeo-126K with the destination vectors pGWnV3 and pGWcV, respectively (Nishimura et al., 2015).

pGreenII-NtREM-DsRed was constructed as follows. After pART7-NtREM-DsRed was digested with NotI, a fragment containing the NtREM-DsRed coding sequence was ligated to the fragment from the binary vector pGreenII 0000² that was digested with NotI, generating pGreenII-NtREM-DsRed.

pGreenII-DsRed was made by digesting pGreenII-NtREM-DsRed with NcoI and self-ligated.

Particle bombardment of N. benthamiana leaves with piL.erG3L (Sasaki et al., 2009), pTH2-126K-GFP (Sasaki et al., 2014), pART7-MP-YFP (Sasaki et al., 2009), and expression plasmids constructed in this study was carried out as described previously (Sasaki et al., 2009). Infiltration of Rhizobium radiobacter (formerly Agrobacterium tumefaciens) strain GV3101(pMP90) to N. benthamiana leaves was performed as described previously (Haque et al., 2008). After agrobacterium transformants carrying pGreenII-DsRed, pGreenII-NtREM-DsRed, or pGLW3-erGFP (Sasaki et al., 2013) were incubated until OD600 reached about 0.6, each of the agrobacterium transformants carrying pGreenII-DsRed or pGreenII-NtREM-DsRed was mixed with that carrying pGLW3-erGFP at a 10:1 ratio. The total OD600 for infiltration was then adjusted to 0.11 by addition of incubation buffer (10 mM MgCl₂, 10 mM MES, and a150 μM acetosyringone).

Leaves of N. benthamiana at 21 h post-bombardment were treated with 50 μM FM4-64 or 0.01% (w/v) aniline blue for staining the plasma membrane or callose, respectively. After 3 h of staining the plasma membrane or callose, the leaves were observed with confocal laser microscopy (LSM710NLO; Carl Zeiss, Jena, Germany).

For subcellular localization and time-lapse analyses, the fluorescence of the target fluorescent proteins was observed at 24 h after bombardment with the LSM710NLO confocal laser microscope (Carl Zeiss). GFP/YFP and DsRed were excited with an argon laser (488 nm) and a diode-pumped solid state laser (561 nm), respectively, and detected by Gallium arsenide phosphide (GaAsP) detectors for each fluorescence.

Aniline blue was excited with a blue diode laser (405 nm) and its fluorescence was detected in the range between 410 and 528 nm. Images were captured in the sequential mode and processed with the software ZEN 2009 (Carl Zeiss).

For measurement of fluorescent infection sites of ToMV-erGFP, GFP was detected at 48 and 72 h after infiltration with an all-in-one fluorescence microscope (BZ-X700; Keyence, Osaka, Japan).

The areas of fluorescent regions were measured using the ImageJ software³.

Preparation of protein samples from agrobacterium-infiltrated or uninfiltrated (healthy) leaves, SDS-PAGE, and western blotting were performed as described previously (Sasaki et al., 2009). DsRed and NtREM-DsRed were detected by using anti-DsRed antibody (Living Colors A.v. Monoclonal Antibody JL-8) (Takara) and stabilized goat anti-mouse HRP-conjugated secondary antibody (Pierce, Thermo Fisher Scientific, Rockford, IL, United States).

We attempted to isolate the cDNA of NtREM1.2 (accession number, EB449751.1) from tobacco cv. Samsun NN. As a result, we obtained two cDNAs (termed NtREMa and NtREMb) with high identity to NtREM1.2. Comparison of the deduced amino acid sequences of NtREMa and NtREMb with NtREM1.2 (Figure 1) revealed 99.0 and 96.1% identity, respectively. Characteristically, NtREMb has two extra amino acids in the N terminal region, which are absent in NtREM1.2 and NtREMa. Through the BLAST search on the Sol Genomics Network database⁴, we found that two tobacco cultivars (TN90 and BX) and N. tomentosiformis (paternal parent of tobacco) had an ORF sequence with 100% identity of NtREMa. On the other hand, homologous ORF sequences with more than 98% but not 100% identity of NtREMb were found in three tobacco cultivars (TN90, BX and K326) and N. sylvestris (maternal parent of tobacco). All of the NtREMb homologs encode proteins with the above-mentioned two extra amino acids, these results suggested that NtREMa and NtREMb in tobacco, which should be orthologous between its parental species, were derived from N. tomentosiformis and N. sylvestris, respectively. NtREMa was selected for further analyses as the closest homolog of NtREM1.2 and renamed Nt(sNN)REM1.2.

In order to confirm the plasma membrane localization of Nt(sNN)REM1.2, we performed a bombardment assay using a cDNA clone of Nt(sNN)REM1.2 fused C-terminally with YFP (NtREM-YFP). Leaves of N. benthamiana were bombarded with the cDNA clone and observed under a confocal microscope at 24 h post-bombardment (hpb). Fluorescence of YFP was observed to distribute almost uniformly along the cell wall and the cell peripheral surface (Figure 1). Co-localization of NtREM-YFP with the plasma membrane marker FM4-64 (red fluorescence) indicated that Nt(sNN)REM1.2 is a plasma membrane-associated protein.

As observed for NtREM-YFP, bombardment with cDNA clones of Nt(sNN)REM1.2 fused with DsRed at the N-terminus (DsRed-NtREM) (Figures 2a,b) or the C-terminus (NtREM-DsRed) (Figure 2c) resulted in similar, uniform distribution of fluorescence on the plasma membrane. In order to assess a role of Nt(sNN)REM1.2 in virus infection, we examined the change in the subcellular localization pattern of NtREM-DsRed upon infection of ToMV-erGFP that encodes the ER-targeting GFP (erGFP) instead of CP (Sasaki et al., 2009). We bombarded N. benthamiana leaves with the cDNA clone of NtREM-DsRed alone or in combination with the infectious clone of ToMV-erGFP. In cells without virus infection, we found occasionally a few small aggregates of NtREM-DsRed (Figure 2d) as well as DsRed-NtREM (Supplementary Figure S1). The average number per 1,000 μm² and size of NtREM-DsRed aggregates in the non-infected cells were 1.7 and 0.11 μm², respectively (Figure 3). In contrast, cells infected with ToMV-erGFP showed a relatively dispersed distribution of NtREM-DsRed and more and larger aggregates of NtREM-DsRed on the plasma membrane. At 24 hpb (early-to-mid infection), cortical ER networks were disrupted and NtREM-DsRed aggregates were associated closely with the coalescent cortical ERs (Figures 2e–h). Cortical ER networks were recovered at 48 hpb (late infection), and these aggregates located in the proximity of tubular ER structures (Figures 2i–l). Detailed measurement indicated that both of the number and the size of NtREM-DsRed aggregates were increased by infection of ToMV-erGFP. The average number per 1,000 μm² and size of NtREM-DsRed aggregates in the virus-infected cells were 3.7 and 0.24 μm², respectively (Figure 3). In addition, we investigated a non-specific effect of erGFP on the subcellular localization of NtREM-DsRed by co-bombardment of the NtREM-DsRed and erGFP cDNAs. Cells co-expressing erGFP showed a similar localization pattern of NtREM-DsRed to that observed in the cells expressing the NtREM-DsRed alone (Figures 2m–o) (Supplementary Figure S2). These results suggested that the virus infection altered specifically the subcellular localization of NtREM-DsRed and promoted the aggregation of NtREM-DsRed at sites close to cortical ERs.

ToMV-erGFP encodes the 126K and 183K replication proteins and MP but not the CP. One or more of these viral proteins were likely to be involved in the altered subcellular localization of NtREM-DsRed in ToMV-infected cells. In order to examine this possibility, we expressed NtREM-DsRed together with fluorescently labeled viral proteins and observed fluorescence at 24 hpb. In this experiment, we constructed a 126K/183K-GFP cDNA clone for a dual expression of the 126K protein and a GFP-fused 183K protein (183K-GFP) that is translated via ribosomal read-through of the stop codon of the 126K gene, and used cDNA clones of a GFP-fused 126K protein (126K-GFP) and a YFP-fused MP (MP-YFP) (Sasaki et al., 2009, 2014).

When NtREM-DsRed was co-expressed transiently with 126K and 183K-GFP (126K+183K-GFP) or 126K-GFP alone in epidermal cells, we found that the number and size of NtREM-DsRed aggregates increased similarly to the case of infection with ToMV-erGFP. The average number per 1,000 μm² and size of the bodies were 5.1 and 0.20 μm² for 183K-GFP, and 4.3 and 0.24 μm² for 126K-GFP, respectively (Figure 3). There was no clear indication of an association of NtREM-DsRed aggregates with 183K-GFP that appeared to be distributed as uniformly as NtREM-DsRed (Figures 4a–c). Meanwhile, such NtREM-DsRed aggregates were shown to locate predominantly along the cytoplasmic distribution of 126K-GFP (Figures 4d–f). However, time-lapse imaging analysis showed that NtREM-DsRed aggregates associate with small 126K-GFP bodies were static while other NtREM-DsRed aggregates appeared to move with 126K-GFP transported by the cytoplasmic streaming (Figure 5). On the other hand, co-expression of MP-YFP also promoted the formation of larger NtREM-DsRed aggregates than those found in cells expressing NtREM-DsRed alone. The average number per 1,000 μm² and size of the aggregates were 2.3 and 0.20 μm² (Figure 3). As reported for TMV MP and StREM1.3 (Amari et al., 2014), there was no perfect co-localization of the MP and the remorin. However, NtREM-DsRed aggregates were sometimes found to be associated closely with filamentous and granular (body) structures of MP-YFP (Figures 4g–i). These collective results suggested that the overexpression of the replication proteins and MP triggered the formation of aggregates of Nt(sNN)REM1.2 on the plasma membrane.

In order to investigate interactions of NtREM with viral proteins, we adopt a recently developed BiFC (bimolecular fluorescence complementation) system that can avoid a self-interaction of the Venus fluorescent protein itself and detect specific protein interaction in planta (Nishimura et al., 2015). pGWnV3 and pGWcV-derived plasmids were used for expression of the N-terminal (nV) and C-terminal (cV) fragments of Venus fused to the C-terminal end of the target proteins, respectively. pGWnV3-NtREM, pGWcV-NtREM, pGWnV3-MP, and pGWcV-MP were used for the expression of NtREM-nV, NtREM-cV, MP-nV, and MP-cV, respectively. After the expression plasmids for BiFC were bombarded to N. benthamiana leaves, fluorescence of Venus was detected at 24 hpb. We observed the uniform distribution of fluorescence on the plasma membrane by co-expression of NtREM-nV and NtREM-cV. This demonstrated that Nt(sNN)REM1.2 interacts with itself (Figure 6a), consistent with previous reports for other remorins (Bariola et al., 2004; Perraki et al., 2012; Tóth et al., 2012). Cells co-expressing NtREM-nV and MP-cV exhibited patchy fluorescent spots (Figure 6b). Similarly, a reciprocal experiment using NtREM-cV and MP-nV demonstrated patchy distribution of fluorescent spots (Figure 6c). These results suggested strongly a direct interaction between Nt(sNN)REM1.2 and ToMV MP. On the other hand, similar to the case of the co-expression with MP-YFP, when NtREM-DsRed was co-expressed with a combination of MP-nV and MP-cV, we found that NtREM-DsRed aggregates were associated closely but not co-localized perfectly with filamentous and granular structures of self-interacting MPs on the peripheral surface (Figures 6d–f). Furthermore, we also examined the interaction of the 126K protein with Nt(sNN)REM1.2 as well as the 126K protein itself by using pGWnV3-126K and pGWcV-126K. We could rarely observe fluorescent, large amorphous structures in cells co-expressing 126K-nV with 126K-cV, while those structures were not associated with NtREM-DsRed aggregates (Figures 6g–i). On the other hand, no fluorescence was observed in the case of 126K-nV and NtREM-cV (data not shown). These observations suggested that the 126K protein might not interact directly with Nt(sNN)REM1.2. Through the BiFC assays, we noticed that the localization patterns of fluorescent MP seemed to be different between MP-YFP and BiFC-combinations. Since the reconstitution of the split Venus protein can stabilize BiFC complexes (Robida and Kerppola, 2009), it was likely that the BiFC-mediated subcellular localizations of MP might be altered depending on the interacting protein.

Since our BiFC analysis indicated that Nt(sNN)REM1.2 could interact directly with MP on the plasma membrane, we examined whether the two proteins are co-localized at plasmodesmata that are lined with the plasma membrane by observing N. benthamiana cells co-expressing NtREM-DsRed and MP-YFP at 24 hpb. Strong spots of MP-YFP, indicative of its plasmodesmal localization, were found on the layer of NtREM-DsRed in the transverse section of the cells (Figures 7a–d). However, there was little accumulation of NtREM-DsRed specifically on those spots. This observation was similar to the previous study showing that TMV MP and StREM1.3 are not co-localized substantially on the plasma membrane (Amari et al., 2014). Then, we further investigated the association of Nt(sNN)REM1.2 and MP with plasmodesmata by co-expressing NtREM-nV and MP-cV and staining the bombarded cells with aniline blue at 24 hpb to visualize callose deposition at plasmodesmata. The Venus fluorescence in the transverse section was detected mostly along the cell wall (Figures 7e–i), confirming the direct interaction between NtREM-nV and MP-cV on a wide area of the plasma membrane. In addition, we occasionally observed fluorescent spots of Venus, which were adjacent to but not co-localized with aniline blue-stained spots (Figures 7g–i). On the other hand, in control cells co-expressing MP-nV and MP-cV, fluorescent spots of Venus were co-localized perfectly with aniline blue-stained spots (Figures 7j–n). Thus, the results from the BiFC analysis suggested that Nt(sNN)REM1.2 and MP might interact at specific sites around plasmodesmata. Although we did not detect any clear accumulation of NtREM-DsRed together with MP-YFP either at or around plasmodesmata (Figures 7a–d), it was possible that BiFC-mediated stabilization of protein–protein interaction (Robida and Kerppola, 2009) might allow us to detect an innately temporary interaction between Nt(sNN)REM1.2 and MP around plasmodesmata.

We examined whether the transient overexpression of NtREM-DsRed by the above-mentioned bombardment method influenced the initial cell-to-cell movement of ToMV-erGFP in N. benthamiana. The percentage of multiple-cell infection sites at 48 hpb and the average number of cells in the multiple cell infection sites were comparable between bombarded tissues overexpressing DsRed and NtREM-DsRed transiently (Supplementary Figure S3), indicating that the overexpressed Nt(sNN)REM1.2 had no effect on the initial movement of the virus. Since the expression of NtREM-DsRed was restricted to bombarded cells only, we further investigated whether the transient overexpression of NtREM-DsRed in the whole leaf tissue by agroinfiltration could interfere with continuous cell-to-cell movement of the virus. In this analysis, N. benthamiana leaves were infiltrated with agrobacterium transformants for the expression of DsRed or NtREM-DsRed, together with that for inoculation of ToMV-erGFP. Confocal microscopic observation confirmed that agrobacterium-mediated expression of NtREM-DsRed also resulted in a uniform distribution on the plasma membrane (Figure 8A). Observation at 48 and 72 hours post-infiltration (hpi) with an all-in-one fluorescence microscope demonstrated that there was no significant increase in the size of fluorescent infection sites at 48 hpi, but there was a statistically significant difference at 72 hpi (Figures 8B,C) (Supplementary Figure S4). Western blot analysis demonstrated that both of the DsRed and NtREM-DsRed accumulated to detectable levels (Figure 8D). Collectively, our data suggested that transient overexpression of NtREM-DsRed had a promotive effect on continuous cell-to-cell movement of ToMV-erGFP in N. benthamiana.