Brassica yellows virus P0 protein impairs the antiviral activity of NbRAF2 in Nicotiana benthamiana

The Brassica yellows virus genotype A P0 protein interacts with Rubisco assembly factor 2 in tobacco, affecting its nuclear accumulation and promoting viral infection.


Introduction
Numerous studies have demonstrated that chloroplasts are a common target of many plant viruses. Chloroplasts and their components are involved in viral movement, replication, and symptom development, and also participate in plant defense against viruses (Zhao et al., 2016;Bhattacharyya and Chakraborty, 2018). Several chloroplast proteins have been shown to play negative roles in viral pathogenesis. The coat protein of Alfalfa mosaic virus interacts with the photosystem-II oxygen-evolving complex protein PsbP, and overexpression of PsbP markedly reduces virus accumulation (Balasubramaniam et al., 2014). The cylindrical inclusion protein from Plum pox virus (PPV) binds to the photosystem-I PSI-K protein, which is the product of psaK. PPV infection down-regulates the expression of psaK mRNA in inoculated leaves, and the silencing of psaK leads to a greater accumulation of PPV (Jiménez et al., 2006). Pathogens attempt to intercept chloroplast proteins by interacting and sequestering them in the cytosol before they are imported into the chloroplasts or by affecting their normal activities (Jin et al., 2007;Shi et al., 2007;Balasubramaniam et al., 2014;Kong et al., 2014).
In mammals, the single pterin-4a-carbinolamine dehydratase/dimerization co-factor of hepatocyte nuclear factor 1 (PCD/DCoH) acts as a metabolic enzyme, and has PCD activity in the mitochondria and also acts as a DCoH in the nucleus (Suck and Ficner, 1996). The activities of PCD and DCoH are independent (Rhee et al., 1997). In plants, there are two copies of PCD/DCoH homologs (Naponelli et al., 2008). Type 1 is localized to the mitochondria and has PCD activity, whereas type 2, without PCD activity, is directed to the chloroplast (Naponelli et al., 2008). The Arabidopsis protein Rubisco assembly factor 2 (AtRAF2) (Fristedt et al., 2018), also named SDIRIP1 (Zhang et al., 2015a) or ATP1 (Oh et al., 2017), is a type 2 protein of the PCD/DCoH family (Naponelli et al., 2008). AtRAF2 is nuclear-encoded and localized to the chloroplast (Naponelli et al., 2008;Zybailov et al., 2008). It has been suggested that AtSDIRIP1/AtRAF2 is localized not only to the chloroplast but also to the cell periphery and nucleus (Zhang et al., 2015a). The chloroplast/ nuclear protein AtSDIRIP1/AtRAF2 interacts with the E3 ligase SDIR1 in the cytosol, and is subsequently degraded by it. AtSDIRIP1/AtRAF2 selectively regulates the expression of the abscisic acid (ABA)-responsive transcription factor gene ABA-INSENSITIVE5 to regulate ABA-mediated seed germination and salt-stress responses (Zhang et al., 2015a). The cytosolic RING-type E3 ligase AtAIRP2 also targets AtATP1/AtRAF2 for degradation, and AtAIRP2 and AtSDIR1 play a combinatory role in ABA-and saltstress responses in Arabidopsis (Oh et al., 2017). In addition, Zea mays RAF2 (ZmRAF2) shares a high amino acid sequence identity with AtRAF2 (Naponelli et al., 2008), and both proteins are involved in Rubisco assembly (Feiz et al., 2014;Fristedt et al., 2018). However, in plants, the function of RAF2 in the nucleus remains unknown.
Brassica yellows virus (BrYV), a newly identified polerovirus, infects crucifer crops in China and causes yellowing or leaf-roll symptoms . BrYV has three genotypes, BrYV-A, B, and C Zhang et al., 2014), and the full-length infectious cDNA clones of these three genotypes have been developed successfully (Zhang et al., 2015b). The BrYV P0 is a strong viral suppressor of RNA silencing and interacts with the SKP1 from N. benthamiana . In the present study, we successfully obtained a novel P0 BrA -interacting protein, NbRAF2, which localized to the nucleus, cell periphery, chloroplasts, and stromules. We demonstrated that P0 BrA decreased the nuclear accumulation of NbRA2. The accumulation of BrYV-A increased when NbRAF2 was silenced using Tobacco rattle virus (TRV)-based virus induced gene silencing (VIGS) but decreased when nuclear NbRAF2 was overexpressed. PLRV P0 PL also interacted with NbRAF2 and decreased its nuclear pool, indicating that nuclear NbRAF2 may be a common target of polerovirus P0s.

Plant material and growth conditions
Wild-type Nicotiana benthamiana and transgenic lines with the ferredoxin NADP(H) oxidoreductase transit peptide fused to enhanced green fluorescent protein (FNR-EGFP) (Schattat et al., 2011) were grown at 24 °C with a 16-h light/8-h dark cycle.

Plasmid constructs
All of the primers used in this study are listed in Supplementary  Table S1 at JXB online.
The vectors pGD and pGDG (Goodin et al., 2002) were used for transient expression. The P38 protein encoded by Turnip crinkle virus was cloned into pGD for transient expression. P0 BrA was cloned into to pGD-3Flag, a modified version of vector pGD that has a C-terminal-fused 3×Flag tag. For co-immunoprecipitation (Co-IP) and confocal microscopy, pGD-3G-mCherry was constructed. A DNA fragment of GGG-mCherry was amplified and cloned into the vector pGD to produce pGD-3G-mCherry. Full-length AtRAF2, NbRAF2, and their mutants were independently cloned into the vector pGDGm [a modified version of pGD that allows the production of a C-terminal green fluorescent protein (GFP)-fused protein] and pGD-3G-mCherry. NbSKP1 was cloned into the vector pGDGm. For its subcellular localization, P0 BrA was amplified and cloned into pSuper1300-GFP (Yang et al., 2010). Wild-type BrYV full-length cDNAs were cloned into the pCB301-2 × 35S-MCS-HDV RZ -NOS vector (Yao et al., 2011;Zhang et al., 2015b).

Agrobacterium-mediated transient expression in N. benthamiana
Plasmids were transformed into the Agrobacterium tumefaciens strain EHA105 or C58CI using the freeze-thaw method (Holsters et al., 1978). The recombinant EHA105 or C58CI was grown overnight, resuspended in infiltration buffer (10 mM MgCl 2 , 10 mM MES, and 100 μM acetosyringone), and incubated at room temperature for at least 3 h before infiltration. The A. tumefaciens cultures were infiltrated into N. benthamiana leaves and the infiltrated leaves were detached for the corresponding assays.

Yeast two-hybrid (Y2H) screen and interaction assays
For the Y2H assay, the Clontech Matchmaker GAL4 Two-Hybrid System 3 (Clontech, Mountain View, CA, USA) was used. Proteininteracting screens were performed with the Matchmaker GAL4 two-hybrid system according to the manufacturer's protocol. The full-length of P0 BrA was cloned into pGDBKT7 containing a binding domain (BD) to generate BD-P0 BrA and then transformed into the yeast host strain Y187. The Arabidopsis cDNA library was used to screen P0 BrA -binding proteins. The full-length AtRAF2 was cloned into pGDADT7 containing an activating domain (AD) to generate AD-AtRAF2 and then transformed into the yeast host strain AH109. Co-transformants were plated onto synthetic dropout (SD) media lacking Trp and Leu (SD/−WL) and SD media lacking Ade, His, Trp, and Leu (SD/−AHLW). Full-length NbRAF2 (GenBank accession MG560271), AtRAF2 (GenBank accession AED96036.1), NbRbcL, P0 MA , and P0 Sc were independently cloned into pGDBKT7. Full-length NbRAF2, truncated NbRAF2 mutants, AtRAF2 and AtRAF2 mutants were independently cloned into pGDADT7. Protein interactions were tested using the yeast mating assay. The interaction of NbRAF2 with P0 BrA was identified using the split ubiquitin membrane-bound yeast two-hybrid (MbY2H) system. NbRAF2 was cloned into the MbY2H bait vector pBT3-STE, and P0 BrA was cloned into the prey vector pPR3-N. All combinations were transformed to the yeast strain NMY51 for mating and the transformed yeast cells were then transferred to SD/−WL and SD/−AHLW plates for 3-5 d.

RNA extraction and RT-PCR
Total RNA was extracted using TRIzol Reagent (Invitrogen) according to the manufacturer's protocol. A 4-μg of total RNA was treated with RNase-free DNase I (TaKaRa). First-strand cDNA was synthesized using 2 μg of treated RNA, oligo d(T) primer, or gene-specific primer and M-MLV Reverse Transcriptase (Promega), as instructed by the protocol. The mRNAs of EF1A, NbRAF2, and BrYV-A RNA were determined using specific primers (see Supplementary Table S1).

Fluorescence microscopy
Agro-infiltrated leaf tissue from N. benthamiana was detached 2 d post infiltration (dpi), and confocal laser scanning microscopy was performed using a Leica SP8 laser-scanning microscope. GFP, mCherry, and chloroplast fluorescence were obtained using laser excitation at 488, 552, and 638 nm, respectively. For bimolecular fluorescence complementation (BiFC) assays, NbRAF2 and AtRAF2 were respectively cloned into pSPYNE-35S and pSPYCE-35S (provided by J. Kudla, Universität Münster, Germany). Yellow fluorescent protein (YFP) fluorescence was obtained using 488-nm laser excitation. To determine its distribution, the nuclear-localized NbRAF2 was quantified from Z stacks and expressed as the total number of nuclear-localized NbRAF2 particles per total number of nuclei.

VIGS assay
For the VIGS assay (Liu et al., 2002a), a 360-bp fragment of NbRAF2 and full-length GFP were cloned and inserted into pTRV2. Each of the TRV2-based constructs was transformed into A. tumefaciens strain GV3101. Agrobacterium harboring TRV1-or TRV2-derived vectors were mixed at a 1:1 ratio and infiltrated into N. benthamiana plants.

Nucleocytoplasmic fractionation assay
Nucleocytoplasmic fractionation was performed using a CelLytic TM PN Isolation/Extraction kit following the manufacturer's instructions (Sigma-Aldrich). Total, nuclear, and cytoplasmic proteins were detected by immunoblotting with corresponding antibodies. The PEPC and H3 proteins were used as cytosolic and nuclear markers, respectively.

P0 BrA interacts with NbRAF2
To study the function of the P0 protein, we performed a Y2H screen of an Arabidopsis cDNA library using P0 BrA as the bait to identify P0 BrA -interacting proteins. Among the screened positive clones, one contained an intact ORF that shared 100% identity with AtRAF2 (At5g51110; GenBank accession AED96036.1). To explore the role of interactions between P0 BrA and RAF2 in N. benthamiana, we searched for AtRAF2 homologs in the N. benthamiana genome (http://solgenomics.net). The homolog that we identified encoded a 211amino acid protein, shared approximately 50% identity with AtRAF2 and ZmRAF2 (GenBank accession NP_001144391) (see Supplementary Fig. S1), and was named NbRAF2 (GenBank accession MG560271). However, the traditional directed Y2H assay did not reveal a P0 BrA -NbRAF2 interaction, although it did detect AtRAF2-P0 BrA ( Supplementary  Fig. S2). Hence, we used the split ubiquitin MbY2H system to test for an interaction (McGee, 2006;Brückner et al., 2009). The P0 BrA was cloned into either the MbY2H bait vector to express a TF::Cub::P0 BrA (P0 BrA -Cub)-fusion protein or into the MbY2H prey vector to express a NubG::P0 BrA (NubG-P0 BrA )-fusion protein. NbRAF2 was also cloned into either the MbY2H bait vector to express a TF::Cub::NbRAF2 (NbRAF2-Cub)-fusion protein or into the MbY2H prey vector to express a NubG::NbRAF2 (NubG-NbRAF2)fusion protein. A series of controls were performed. NubG constructs served as test proteins and NubI constructs were used as positive controls. All combinations were transformed into yeast cells for mating, and the transformed yeast cells were transferred to SD/−WL and SD/−AHLW plates for 2-3 d. However, the NubG-NbRAF2/Cub combination also grew on the SD/−AHLW plates (data not shown). Thus, we chose NbRAF2-Cub to identify the interaction. Yeast transformed with NbRAF2-Cub and NubG-P0 BrA grew on selective plates, as did the positive control; however, the negative controls did not (Fig. 1A). Thus, P0 BrA can directly interact with NbRAF2 in yeast.
We used Co-IP assays to further confirm whether P0 BrA interacted with NbRAF2 in plant cells. Flag-tagged P0 BrA (P0 BrA -Flag) was co-expressed with GFP, GFP-tagged NbRAF2 (NbRAF2-GFP), or GFP-tagged NbSKP1 (NbSKP1-GFP) in N. benthamiana leaves through agroinfiltration. The interaction between P0 BrA and NbSKP1 was used as a positive control. At 2 dpi, protein complexes were immunoprecipitated using anti-Flag beads. The P0 BrA -Flag was co-immunoprecipitated with NbRAF2-GFP, but not with GFP (Fig. 1B). These experiments demonstrated that P0 BrA interacts with NbRAF2 in plant cells.

Subcellular localization of NbRAF2
To investigate the subcellular localization of NbRAF2, we generated mCherry-tagged NbRAF2 (NbRAF2-mCherry) (see Supplementary Fig. S3), and then transiently co-expressed it with free GFP in N. benthamiana leaves through agro-infiltration. The infiltrated leaves were collected at 2 dpi and observed with confocal microscopy. The mCherry signal revealed that NbRAF2 was present in the chloroplast, cell periphery, and nucleus of the same cells, similar to the subcellular localization of AtSDIRIP1/AtRAF2. Surprisingly, NbRAF2 was localized to the chloroplast stromules ( Fig. 2A). To confirm this, we transiently expressed NbRAF2-mCherry in FNR-EGFP transgenic N. benthamiana through agro-infiltration. The FNR-EGFP was used here as a stromule-localized marker (Schattat et al., 2011). Confocal microscopy showed that NbRAF2-mCherry could perfectly co-localize with FNR-EGFP in chloroplasts and stromules (Fig. 2B). Thus, NbRAF2-mCherry was localized not only to the nucleus and cell periphery but also to the chloroplasts and stromal fraction. Similarly, AtRAF2 was also localized to the stromules ( Supplementary Fig. S4).

The C-terminal residues of NbRAF2 are required for its self-interaction
The Y2H and BiFC assays showed that NbRAF2 selfinteracted. The Y2H results demonstrated that the combinations of pGBKT7-NbRAF2 and pGADT7-NbRAF2 could not grow on the SD/−AHLW selective plates (Fig. 3A), but NbRAF2 could interact with NbRAF2-ΔTP, a truncated mutant of NbRAF2 that contained a deleted transit peptide (TP) of 1-45 aa in the N-terminus (Figs 2, 3A, Supplementary Fig. S3). For the BiFC assays, NbRAF2 was fused to N-and C-terminal halves of YFP at the N-termini to generate NbRAF2-YN and NbRAF2-YC, respectively. The reconstituted YFP fluorescence was observed in N. benthamiana leaf epidermal cells co-infiltrated with NbRAF2-YN and NbRAF2-YC, which co-localized with chloroplast autofluorescence (Fig. 3B). In addition, YFP signals occurred in the stromules. Fluorescence could not be visualized in the negative controls NbRAF2-YN/YC and YN/NbRAF2-YC ( Supplementary Fig. S5). Arabidopsis and maize RAF2 can form dimers (Valkai, 2004;Feiz et al., 2014), consistent with our observations for NbRAF2. In addition, the C-terminal truncated mutant NbRAF2-ΔCter, which contained a deleted C-terminal 183-211 aa and localized to the nucleus NbRAF2 and P0 BrA using the split ubiquitin membrane-bound yeast twohybrid (MbY2H) system. NbRAF2 was cloned into the MbY2H bait vector pBT3-STE, and P0 BrA was cloned into the prey vector pPR3-N. NubG constructs served as test proteins and NubI constructs were positive controls. All the combinations were transformed into yeast NMY51. Yeast strains were grown on SD/-Leu/-Trp and SD/-Ade/-His/-Leu/-Trp, and maintained at 30 °C for 2-3 d. (B) Co-immunoprecipitation analyses of NbRAF2 and P0 BrA proteins in N. benthamiana leaves. P0 BrA -Flag was co-expressed with GFP, NbRAF2-GFP, or NbSKP1-GFP through agroinfiltration. GFP and NbSKP1 were used as negative and positive controls, respectively. Protein complexes were immunoprecipitated using anti-Flag beads. Immunoprecipitates were assessed with western blotting using anti-GFP and anti-Flag antibodies. and chloroplast ( Supplementary Fig. S3, Fig. 2), failed to interact with full-length NbRAF2 (Fig. 3A, B).
Thus, NbRAF2 self-interacted in chloroplasts and stromules, and the C-terminal residues were required for self-interaction but not for chloroplastic and nuclear localization. Similarly, the C-terminal 192-220 aa of AtRAF2 was required for its self-interaction but not for its chloroplastic and nuclear localization (Supplementary Figs S4, S6).

NbRAF2 co-localizes with P0 BrA in the nucleus and cell periphery
To investigate the subcellular localization of P0 BrA , we constructed a vector expressing the P0 BrA protein with a GFPtag fused to its C-terminus (P0 BrA -GFP). P0 BrA -GFP was co-expressed with free mCherry (which was used as a marker to delineate the nucleus and cytoplasm) in N. benthamiana leaves through agro-infiltration. Confocal microscopy showed that P0 BrA -GFP co-localized with mCherry, suggesting that P0 BrA -GFP can localize to the nucleus and the cell periphery (Fig. 4A). Moreover, P0 BrA -GFP also formed punctate structures in the cytosol (Fig. 4A). To further investigate where P0 BrA and NbRAF2 co-localized in vivo, the P0 BrA -GFP was transiently co-expressed with NbRAF2-mCherry in N. benthamiana leaves through agro-infiltration. Confocal microscopy showed that they co-localized in the nucleus and cell periphery (Fig. 4B).

The nuclear pool of NbRAF2 decreases in the presence of P0 BrA or BrYV-A
Although P0 BrA -GFP co-localized with NbRAF2-mCherry in the nucleus, the cells demonstrating nuclear co-localization were difficult to find. Interestingly, the distribution of NbRAF2-mCherry to the nucleus was significantly reduced in the presence of P0 BrA -GFP compared with the GFP control (Fig. 5A). The relative numbers of nuclear-localized NbRAF2 were calculated based on the confocal images, and there was a reduction of approximately 30% in nuclear-localized NbRAF2 in the P0 BrA -GFP co-expressed cells compared with GFP co-expressed cells. Nucleocytoplasmic fractionation assays were performed to further identify the nuclear accumulation of the NbRAF2 protein. The NbRAF2-mCherry with P0 BrA -GFP or GFP were transiently coexpressed in N. benthamiana leaves, and protein samples were Fig. 3. The C-terminal residues of NbRAF2 are required for its self-interaction. (A) Self-interaction analysis of NbRAF2 in yeast. Activating domain (AD) constructs containing NbRAF2 and its variants were independently transformed with BD-NbRAF2. Yeast cells transformed separately with binding domain (BD) and AD vectors showed no growth on selective plates. Yeast strains were grown on SD/-Leu/-Trp and SD/-Ade/-His/-Leu/-Trp, and maintained at 30 °C for 5-7 d. The yeast two-hybrid results showed that C-terminal residues are required for the self-interaction of NbRAF2. (B) BiFC assays for NbRAF2 self-interaction were performed on N. benthamiana leaves through agro-infiltration. YFP is an indicator of protein-protein interactions. NbRAF2-NbRAF2 interactions are apparent in the chloroplasts and chloroplast stromules, and the self-interaction requires C-terminal residues. (This figure is available in colour at JXB online.) prepared at 2 dpi. A western blot analysis revealed that the nuclear pool of NbRAF2 protein significantly decreased after leaves were inoculated with P0 BrA -GFP (Fig. 5B). However, no significant difference was observed in the total amounts of NbRAF2 protein.
NbRAF2-ΔTP, the TP deletion mutant, localized to the nucleus but not to the chloroplast; however, NbRAF2-ΔCter, the C-terminal residue deletion mutant, localized to the nucleus and the chloroplast (Fig. 2). We tested the effect of P0 BrA on the nuclear accumulation of these two mutants. Confocal microscopy showed that P0 BrA inhibited the accumulation of a nuclear pool of NbRAF2-ΔCter-mCherry, but not that of NbRAF2-ΔTP-mCherry (see Supplementary  Fig. S7). Thus, the P0 BrA -mediated reduction of nuclear NbRAF2 required the dual localization of NbRAF2 in the chloroplast and nucleus.
The effect of BrYV-A on the nuclear accumulation of NbRAF2 was assayed using nucleocytoplasmic fractionation assays and confocal microscopy. NbRAF2-mCherry was co-expressed with the mock or BrYV-A through agroinfiltration in N. benthamiana leaves. The samples were prepared from inoculated leaves at 2 dpi. The nuclear pool of NbRAF2 decreased after BrYV-A infection (Fig. 5C,  Supplementary Fig. S8).

NbRAF2 expression is down-regulated during BrYV-A infection
We investigated whether the NbRAF2 expression level was affected during BrYV-A infection. Total RNA was obtained from BrYV-A-inoculated N. benthamiana leaves at 2 dpi.
Semi-quantitative RT-PCR showed that the mRNA level of NbRAF2 decreased during the BrYV-A infection (Fig. 6A).

Silencing NbRAF2 increases the local accumulation level of BrYV-A in inoculated leaves and enhances systemic infection of the virus
To investigate the possible role of NbRAF2 during BrYV-A infection, we used the TRV-VIGS vector (Liu et al., 2002b) to knock down the expression of NbRAF2. A partial cDNA fragment of NbRAF2 was cloned into the RNA2-derived vector of TRV to generate pTRV2-NbRAF2, and GFP was cloned into pTRV2 to generate pTRV2-GFP, which was used as a negative control. At 2-3 weeks post-inoculation, NbRAF2silenced plants showed leaf chlorosis compared with TRV-GFP-infected control plants (Fig. 6B).The NbRAF2 expression level was then determined with RT-PCR. The NbRAF2 mRNA level was significantly reduced in NbRAF2-silenced plants compared with non-silenced controls (Fig. 6C). The RbcL protein level correspondingly decreased in NbRAF2-silenced plants (Fig. 6D) and RbcL interacted with NbRAF2 (see Supplementary Fig. S9), consistent with the role of ZmRAF2 and AtRAF2 in Rubisco accumulation (Feiz et al., 2014;Fristedt et al., 2018). Subsequently, the NbRAF2-silenced and non-silenced N. benthamiana plants were inoculated with BrYV-A using agro-infiltration. Western blot and RT-PCR analyses showed that BrYV-A CP protein and BrYV-A RNA levels increased in the NbRAF2-silenced plants compared with non-silenced control plants at 2 dpi (Fig. 6D). BrYV-A can cause necrosis in inoculated leaves, but symptoms are not visible in systemic leaves. However, we observed no obvious  Fig. S10). At 14 dpi, BrYV-A was able to spread to the upper non-inoculated leaves of the NbRAF2-silenced plants but not easily into those of non-silenced plants. BrYV-A RNA was detected in the systemic leaves of NbRAF2-silenced and nonsilenced plants (Fig. 6E). RT-PCR detection showed that 31.6% of non-silenced control plants had systemic BrYV-A infection at 14 dpi, but 63.2% of NbRAF2-silenced plants became systemically infected with BrYV-A (Fig. 6E, lower panel). These findings suggested that silencing of NbRAF2 increases the local accumulation of BrYV-A in inoculated leaves and promotes the systemic infection of the virus. Thus, our results showed that NbRAF2 negatively regulates the BrYV-A infection. The localization of NbRAF2 when co-expressed with GFP or P0 BrA -GFP. Images were taken at 2 d post-inoculation (dpi). The scale bars represent 20 μm. Z stacks of cells were imaged. Values for relative nuclear NbRAF2 are presented in the table, and are calculated as the ratio of the number of nuclear-localized NbRAF2 to the total number of nuclei, for each combination as determined by confocal microscopy. (B, C) Western blotting of NbRAF2 in nuclei-enriched and nuclei-depleted fractions, and total protein when co-expressed with GFP or P0 BrA -GFP (B), and when co-expressed with Mock or BrYV-A (C) in N. benthamiana leaves. Protein samples were prepared from plant leaves at 2 dpi. All fractions were subjected to immunoblot analyses and were assessed using the corresponding antibodies. PEPC and histone H3 were used as cytosolic and nuclear markers, respectively. (This figure is available in colour at JXB online.)

Overexpression of nuclear RAF2 enhances resistance to BrYV-A
Silencing NbRAF2 increased BrYV-A accumulation, and the nuclear accumulation of NbRAF2 decreased during BrYV-A infection (Figs 5C, 6D). Therefore, we investigated the function of nuclear NbRAF2 in BrYV. A nuclear localization sequence (NLS; Wen et al., 1995) was fused to the N-terminus of NbRAF2-ΔTP-mCherry to generate NLS-NbRAF2-ΔTP-mCherry (see Supplementary Fig. S3). Confocal microscopy showed that NLS-NbRAF2-ΔTP-mCherry was exclusively redirected to the nucleus (Fig. 7A). Hence, we transiently expressed NLS-NbRAF2-ΔTP-mCherry and mCherry through agro-infiltration in N. benthamiana leaves. Then, at 1 dpi, we inoculated BrYV-A through agro-infiltration in the same leaves for 2 d. The accumulation of BrYV-A CP was examined with western blotting, which demonstrated that overexpressing nuclear NbRAF2 decreased the level of BrYV-A CP compared with mCherry (Fig. 7B). To determine whether nuclear AtRAF2 had the same role in resistance to BrYV-A, we constructed AtRAF2-ΔTP-mCherry, which was exclusively redirected to the nucleus ( Supplementary  Fig. S11). The results of western blotting demonstrated that overexpressing nuclear AtRAF2 also inhibited virus accumulation ( Supplementary Fig. S11). These results suggested that the increased nuclear accumulation of NbRAF2 or AtRAF2 enhances resistance to BrYV-A.

Discussion
We screened a polerovirus P0-interacting chloroplast/nucleus protein, NbRAF2, and found that it was localized not only to the cell periphery and nucleus but also to the chloroplasts and stromules. In addition, BiFC and Y2H assays showed that NbRAF2 could self-interact in chloroplasts and stromules, and the self-interaction required the 183-211 aa of its C-terminal. We obtained similar results for AtRAF2.
The C-terminus of the RAF2 protein is highly conserved in N. benthamiana, A. thaliana, and Z. mays, indicating that the C-terminus and self-interaction may play important roles in the functions of plant RAF2 proteins. NbRAF2 interacted with NbRbcL, and NbRAF2-silenced plants showed reduced accumulation of NbRbcL, demonstrating that NbRAF2 may play a role similar to ZmRAF2 and AtRAF2 in Rubisco assembly (Feiz et al., 2014;Fristedt et al., 2018).
Increasing evidence indicates that several chloroplast proteins are localized in both chloroplasts and nuclei. These proteins have pivotal roles in plastid-to-nucleus communication (retrograde signaling), which can regulate specific changes in nuclear gene expression, including photosynthesis-related and stress-responsive genes (Koussevitzky et al., 2007;Pogson et al., 2008;Krause et al., 2012). In response to retrograde signals, PTm, a chloroplast envelope-bound plant homeodomain PHD transcription factor, is cleaved. Its N-terminus is transported through the cytosol to the nucleus, and there it activates ABI4 transcription in a PHD-dependent manner to regulate Lhcb expression (Sun et al., 2011). Several proteins with dual localization in the plastids and the nucleus are involved in pathogen defense reactions (Desveaux et al., 2000(Desveaux et al., , 2005Caplan et al., 2008;Lai et al., 2011). For instance, after NRIP1 is released from the chloroplast to the cytoplasm and the nucleus during Tobacco mosaic virus infection, it is then recognized by the N immune receptor to activate defenses, and this is accompanied by the induction of the hypersensitive response (Caplan et al., 2008). The Why1 protein is released from the chloroplast and transported to the nucleus where it acts as a nuclear transcription factor to mediate the elicitor-induced expression of the PR-10a gene (Desveaux et al., 2000;Isemer et al., 2012).
In our study, NbRAF2 showed chloroplast/nucleus dual localization, but the dual localization mechanism and the function of NbRAF2 in the nucleus remain unknown. Interestingly, we observed that NbRAF2 was also localized to the stromules. The stromules, from which chloroplasts send out dynamic tubular extensions, aid in the transport of immune signals into the nucleus and other subcellular compartments during innate immunity (Caplan et al., 2015;Gu and Dong, 2015). Stromules can be induced during immune responses and are in close contact with nuclei (Kwok and Hanson, 2004;Holzinger et al., 2007;Caplan et al., 2008Caplan et al., , 2015. We hypothesized that chloroplast/nucleus NbRAF2 may be transported from the chloroplasts to the nucleus through the stromules and may have a role in transcriptional programming in the nucleus, similar to other chloroplast/ nuclear proteins. We observed that P0 BrA interacted and co-localized with NbRAF2 both in the nucleus and the cell periphery. The molecular mass of NbRAF2-mCherry corresponded to the predicted size of the mature NbRAF2-mCherry that was processed by the cleavage of the N-terminal chloroplast transit peptide (TP), and the nuclear isoform was the same size as the mature NbRAF2 protein, indicating that the nuclear NbRAF2 was released from chloroplasts and could interact with P0 BrA . In addition, the subcellular distribution of NbRAF2 was altered by P0 BrA . Confocal microscopy and the nucleocytoplasmic fractionation assay showed that P0 BrA -GFP inhibited the accumulation of nuclear NbRAF2. AtRAF2 is degraded by the E3 ligases AtSDIR1 and AIRP2 in the cytosol when responding to ABA and high-salt stress (Zhang et al., 2015a;Oh et al., 2017). However, the total NbRAF2 level was not affected when leaves were inoculated with P0 BrA , indicating that the reduction in the nuclear enrichment of NbRAF2 was not caused by the degradation triggered by the F-box protein P0 BrA . Interestingly, P0 BrA decreased the nuclear pool of NbRAF2-ΔCter, which localized to chloroplasts and nuclei, rather than the NbRAF2-ΔTP, which localized to nuclei but not chloroplasts. This indicated that the P0 BrA -mediated reduction of nuclear NbRAF2 requires the dual localization of NbRAF2 in chloroplasts and nuclei. We hypothesize that the chloroplast-to-nucleus translocation of NbRAF2 was regulated by chloroplast retrograde signals. Thus, P0 BrA interfered with the chloroplast retrograde signals to inhibit the chloroplast-to-nucleus translocation of NbRAF2. As a result, P0 BrA inhibited the accumulation of nuclear NbRAF2. It is unlikely that P0 BrA induced NbRAF2 export from the nucleus to the cytoplasm and/or chloroplast through interactions with NbRAF2 in the nucleus and cell periphery, because we observed no change in the nuclear pool of NbRAF2-ΔTP when leaves were inoculated with P0 BrA . However, the mechanism remains elusive and will be further investigated in the future.
The nucleus of a eukaryotic cell is essential for controlling immune responses. Immune signals are transmitted to the nucleus and reprogram gene expression, which shifts the cells into defense mode (Maleck et al., 2000;Tao, 2003;Dodds and Rathjen, 2010). Upon pathogen infection, the nuclear pool of immune receptors increases or immune receptors are targeted to the nucleus where they activate defense responses (Deslandes et al., 2002(Deslandes et al., , 2003Shen et al., 2007;Bernoux et al., 2008). Other essential signaling components in the nucleus also play important roles in defense responses. The nuclear pool of the nucleo-cytoplasmic immune regulator EDS1 increases after a pathogen challenge and is essential for resistance to biotrophic and hemi-biotrophic pathogens (García et al., 2010). Under unchallenged conditions, NPR1 is present in the cytosol as a stable oligomer (Mou et al., 2003); however, upon pathogen infection, monomeric forms of NPR1 are transferred to the nucleus to bind the TGA family of transcription factors, activating the expression of defense-related genes, such as PR1 and WRKY (Zhang et al., 1999;Mou et al., 2003;Wang et al., 2006). We found that the nuclear pool of NbRAF2 decreased during BrYV-A infection and could negatively regulate BrYV-A infection. Silencing NbRAF2 increased BrYV-A accumulation in the inoculated leaves and enhanced viral systemic infection. However, we cannot rule out the possibility that NbRAF2 could also be involved in BrYV-A movement. In contrast, overexpression of nuclear NbRAF2 inhibited BrYV-A accumulation in the inoculated leaves, indicating that nuclear NbRAF2 possesses antiviral activity against BrYV-A infection. Interestingly, overexpressing nuclear NbRAF2 enhanced the necrosis when inoculated with BrYV-A ( Supplementary Fig. S12), which is consistent with our hypothesis that the increased resistance to BrYV-A is probably caused by the enhanced defense signaling when overexpressing RAF2 in the nucleus. Future work will address these issues. In mammals, PCD/DCoH, a homolog of plant RAF2, acts as a transcriptional co-factor to bind to, and enhance the activity of, the HNF1 transcription factor in the nucleus (Suck and Ficner, 1996). There is a good possibility that NbRAF2 acts as a transcriptional co-factor in the nucleus to regulate defense response-related genes. Thus, P0 BrA may affect antiviral activity of NbRAF2 through two pathways. One is that the nuclear P0 protein directly interacts with nuclear NbRAF2 to interfere with its antiviral function; another is that P0 BrA inhibits the accumulation of the nuclear pool of NbRAF2 by sequestering it outside the nucleus in order to enhance the viral infection. Moreover, P0 proteins encoded by PLRV, Melon aphid-borne yellows virus, and Sugarcane yellow leaf virus interacted with AtRAF2 or NbRAF2 ( Supplementary Fig. S13), P0 PL decreased the nuclear pool of NbRAF2 ( Supplementary Fig. S14), and overexpression of nuclear AtRAF2 also inhibited BrYV-A accumulation. Thus, the host RAF2 protein might be a common target of polerovirus P0 proteins.
In light of our data, we hypothesize that, under normal conditions, NbRAF2 forms dimers and participates in Rubisco assembly in the chloroplast. Nuclear NbRAF2 functions as a transcriptional co-factor to regulate defense response-related genes. Upon pathogen infection, the inactive nuclear NbRAF2 is activated to initiate expression of defensive genes. Meanwhile, chloroplasts sense the changes and transport chloroplastic immune signals, including NbRAF2, to the nucleus through stromules or other pathways to enhance resistance to pathogens. However, the effector P0 protein interacts with NbRAF2, thereby interfering with the antiviral function of nuclear NbRAF2. Subsequently, P0 BrA inhibits NbRAF2 nuclear accumulation by interfering with the chloroplast retrograde signals to amplify the suppression process for the the benefit of the virus, which further facilitates pathogen infection.