Selective autophagic receptor NbNBR1 prevents NbRFP1-mediated UPS-dependent degradation of βC1 to promote geminivirus infection

Autophagy is an evolutionarily conserved, lysosomal/vacuolar degradation mechanism that targets cell organelles and macromolecules. Autophagy and autophagy-related genes have been studied for their antiviral and pro-viral roles in virus-infected plants. Here, we demonstrate the pro-viral role of a selective autophagic receptor NbNBR1 in geminivirus-infected Nicotiana benthamiana plants. The βC1 protein encoded by tomato yellow leaf curl China betasatellite (TYLCCNB) that is associated with tomato yellow leaf curl China virus (TYLCCNV) enhanced the expression level of NbNBR1. Then NbNBR1 interacted with βC1 to form cytoplasmic granules. Interaction of NbNBR1 with βC1 could prevent degradation of βC1 by the NbRFP1, an E3 ligase. Overexpression of NbNBR1 in N. benthamiana plants increased βC1 accumulation and promoted virus infection. In contrast, silencing or knocking out NbNBR1 expression in N. benthamiana suppressed βC1 accumulation and inhibited virus infection. A single amino acid substitution in βC1 (βC1K4A) abolished its interaction with NbNBR1, leading to a reduced level of βC1K4A. The TYLCCNV/TYLCCNBK4A mutant virus caused milder disease symptoms and accumulated much less viral genomic DNAs in the infected plants. Collectively, the results presented here show how a viral satellite-encoded protein hijacks host autophagic receptor NbNBR1 to form cytoplasmic granules to protect itself from NbRFP1-mediated degradation and facilitate viral infection.

Introduction Autophagy, originated from a Greek word describing self-eating [1], is an evolutionary conserved degradation pathway that targets macromolecules, organelles and/or pathogens [2]. Three main types of autophagy have been reported in eukaryotes: macroautophagy (thereafter referred to as autophagy), microautophagy and chaperone-mediated autophagy [3]. Autophagosomes are double membrane-bound vesicles formed through a series of steps, and are important components involved in the autophagy pathway. These membrane-bound vesicles fuse to host cell vacuoles (yeast and plants) or lysosomes (mammals) to deliver cargos for proteolytic degradation and monomer recycling. More than 40 autophagy-related proteins (ATGs) have been identified in plants and function in autophagy induction, phagophore nucleation, autophagosomes expansion, and vacuolar membrane fusion [2]. Autophagy was previously considered as bulk and non-selective. In recent years, more and more reports have shown that autophagy operates in selective ways via selective autophagic cargo adaptors [4]. For example, cargo adaptor p62/Sequestosome-1 (p62/SQSTM1) and neighbor of BRCA1 gene 1 (NBR1) recognize and bind ubiquitinated protein complexes, while BCL2/adenovirus E1B 19 kDa-interacting protein 3 (BNIP3) targets endoplasmic reticulum and mitochondria for autophagosome degradation [5,6]. In a different report, Beclin1 (ATG6) has been shown as a new selective plant autophagy receptor that can mediate autophagic degradation and inhibit the replication and infection of turnip mosaic virus (TuMV) through interaction with viral RNA-dependent RNA polymerase NIb [7].
Based on the microarray study, Laura et al. found that many genes involved in autophagy pathway were up-regulated in tomato after infection by tomato yellow leaf curl Sardinia virus [8]. Many reports have demonstrated that autophagy can be activated upon virus infection to play important roles in plant resistance to virus infection [9][10][11][12][13][14]. For example, autophagy can regulate plant hypersensitive cell death (PCD) response to restrict tobacco mosaic virus (TMV) infection [15]. ATG8f, also an autophagy-associated protein, has been shown to interact with cotton leaf curl Multan virus (CLCuMuV) βC1 protein to suppress virus infection [16]. Arabidopsis NBR1 (AtNBR1) has been shown to directly target cauliflower mosaic virus (CaMV) capsid protein (CP) for degradation, while CaMV safeguards its CP via sequestering it into viral inclusions [17]. In addition, the NBR1-mediated selective autophagy could suppress TuMV infection via targeting viral RNA silencing suppressor HC-Pro [18]. To prevent this degradation, TuMV utilizes its own proteins to interfere the NBR1-mediated selective autophagy [18].
During plant and virus arm races, virus has evolved different strategies to combat host defenses against viral replication and/or infections, including antagonizing autophagy directly or hijacking autophagic components to facilitate viral replication and infection. The γb protein of barley stripe mosaic virus (BSMV) has been shown to disturb the autophagy-mediated degradation through competing with ATG8 for ATG7 [19]. Bamboo mosaic virus (BaMV) infection induces the expressions of ATGs, and among these ATGs, ATG8f has been shown to associate with chloroplast-derived vesicles that are ideal sites for viral RNA replication and protection of viral RNAs from host silencing machinery [20]. TuMV 6K2 protein was reported to induce NBR1 expression, and TuMV could exploit the NBR1-ATG8f-mediated autophagy via an interaction between viral NIb and NBR1 to target viral replication complexes (VRCs) to tonoplasts for virus replication and virion formation [21].
Geminiviruses contain single-stranded circular DNA genomes and each genomic DNA encodes 4-6 classical viral proteins and many additional small viral proteins identified in a recent report [22]. Geminiviruses are transmitted by insect vectors and can infect many economically important crops, causing significant losses to agricultural industries worldwide [23]. The genus Begomovirus contains the largest member of species and is divided into the monopartite and bipartite begomoviruses, according to their genome components. Most monopartite begomoviruses are known to have betasatellites [24,25], which are circular single-stranded DNAs with about 1,350 nucleotides (nt). βC1 is a betasatellite-encoded protein and is a determinant of disease symptoms as well as a repressor of host defense [26]. For example, tomato yellow leaf curl China betasatellite (TYLCCNB)-encoded βC1 is reported to counteract plant defense responses including transcriptional gene silencing, post-transcriptional gene silencing, and mitogen-activated protein kinase-mediated plant immunity [27][28][29]. βC1 is also a key target of plant defense. Tomato (Solanum lycopersicum) SNF1-related kinase (SlSnRK1) has been shown to function defense against geminivirus infection through phosphorylation of viral βC1 [30]. Nicotiana tabacum E3 ligase, a RING-finger protein known as NtRFP1, has been shown to interact with and mediate βC1 degradation by the ubiquitin-26S proteasome system (UPS) [31].
In this study, we have identified a N. benthamiana selective autophagic receptor NbNBR1, a homologous of AtNBR1 in Arabidopsis and NtJoka2 in N. tabacum. TYLCCNB βC1 up-regulates the expression level of NbNBR1, and interacts with NbNBR1 to form cytoplasmic granules. NbNBR1 could prevent NbRFP1-mediated UPS-dependent degradation of βC1 to benefit geminivirus infection. Different from previous reports, our results indicate that a selective autophagy receptor can be exploited by a specific viral satellite-encoded protein. This new finding unveils a previously unknown mechanism modulating the roles of autophagy-related proteins in the arm race between plant and geminivirus.

NbNBR1 interacts with βC1 to form cytoplasmic granules
To investigate the role of autophagy in the infection of geminivirus associated with a betasatellite, we used βC1 as a bait to screen βC1-interacting autophagy-related proteins via Y2H assay. The result showed that the yeast cells transformed with AD-βC1+BD-NbNBR1 grew on the selective medium, while the yeast cells transformed with AD-βC1+BD or AD+BD-NbNBR1 did not (Fig 1A). To further validate this interaction, we performed bimolecular fluorescence complementation (BiFC) assay in the leaves of the RFP-H2B transgenic N. benthamiana plants. The N-terminal half or the C-terminal half of YFP was fused to the C-terminus of NbNBR1 or βC1 to generate YN-NbNBR1, YC-NbNBR1, YN-βC1 and YC-βC1, respectively. Co-expression YN-NbNBR1+YC-βC1 or YC-NbNBR1+YN-βC1 in the RFP-H2B transgenic N. benthamiana leaves resulted in the formation of yellow fluorescent cytoplasmic granules (green) by 48 hours post infiltration (hpi), indicating a positive interaction between these two proteins. We also used TuMV P3N-PIPO, a viral movement protein, as a negative control. When YN-P3N-PIPO+YC-βC1, YN-P3N-PIPO+YC-NbNBR1, YC-P3N-PIPO+YN-βC1 or YC-P3N-PIPO+YN-NbNBR1 were co-expressed in leaves, no yellow fluorescent granules were observed (Fig 1B). In order to know if such interaction also happens in other sub-cellular

PLOS PATHOGENS
NbNBR1 prevents degradation of geminivirus βC1 protein compartments, we analyzed the co-localization of some reported markers of sub-cellular compartments and the NbNBR1-βC1 interaction complex. As shown in S1A Fig, we found that the interaction complex of NbNBR1-βC1 did not localize in chloroplast, Golgi, peroxisome or endosome, but it co-localized with the endoplamic reticulum (ER) marker. βC1-GFP, Myc-NbNBR1, GFP, and Myc-GUS were then transiently co-expressed for co-immunoprecipitation (Co-IP) assay using GFP-Trap_MA magnetic agarose beads (ChromoTek). As shown in Fig  1C, Myc-NbNBR1 rather than Myc-GUS was specifically immunoprecipitated by βC1-GFP, but not by GFP using anti-GFP and anti-Myc antibodies, confirming the presence of the NbNBR1-βC1 complex in planta.
Previously, our laboratory has reported that βC1 accumulates primarily in the cell nucleus and cytoplasm [32]. To investigate whether the interaction between NbNBR1 and βC1 could change the subcellular localization pattern of βC1, we expressed βC1-YFP and NbNBR1-CFP, individually or together, in the RFP-H2B transgenic N. benthamiana leaves. At 48 hpi, βC1-YFP expressed alone was observed in both nucleus and cytoplasm, while NbNBR1-CFP expressed alone was observed as small granules in the cytoplasm. In the cells co-expressing βC1-YFP and NbNBR1-CFP, some βC1-YFP was found to co-localize with NbNBR1-CFP to form cytoplasmic granules (Fig 1D). The overlapping fluorescence signal and spectra of βC1-YFP and NbNBR1-CFP (Fig 1D and 1E) further confirmed that βC1 and NbNBR1 were co-localized in the cytoplasmic granules.  (Fig 2A). Western blot showed that the NbNBR1-silenced TYLCCNV/TYLCCNB-inoculated plants accumulated much less βC1 in the inoculated leaves at 3 dpi ( Fig 2B). qPCR analyses at 7 dpi and Southern blot analyses at 14 dpi of viral DNA accumulations in the systemic leaves of plants showed less viral DNAs accumulated in the NbNBR1-silenced plants than the non-silenced plants (Fig 2C and 2D).

Silencing and knocking out of NbNBR1 expression inhibits TYLCCNV/ TYLCCNB infection
To further confirm the function of NbNBR1 described above, we generated two NbNBR1knockout N. benthamiana lines using CRISPR/Cas9-based technology. The T1 NbNBR1-Cas9 Line3 and NbNBR1-Cas9 Line4 (referred to as NbNBR1-Cas9-L3 and NbNBR1-Cas9-L4, respectively) plants showed similar growth phenotypes as the Wt plants (S2C Fig). DNA sequencing results showed that the NbNBR1-Cas9-L3 line carried five altered and three deleted nts at the cleavage site, while the NbNBR1-Cas9-L4 line carried five deleted nts at the cleavage cite ( Fig 2E). After inoculation of these plants with TYLCCNV/TYLCCNB through agroinfiltration, milder leaf curling symptoms appeared on NbNBR1-knockout N. benthamiana plants at 7 dpi compared to those in the Wt plant ( Fig 2F). Western blot showed less βC1 accumulations in the virus-inoculated leaves of the NbNBR1-Cas9-L3 and NbNBR1-Cas9-L4 plants compared to those in the Wt plants at 3 dpi ( Fig 2G). qPCR analyses at 7 dpi and Southern blot analyses at 14 dpi of viral DNA accumulations in the systemic leaves of plants indicated in Fig 2F, showed less viral DNAs accumulated in the NbNBR1-knouck out plants than the Wt plants (Fig 2H and 2I). Furthermore, to investigate whether NbNBR1 could affect the βC1-mediated disease symptom formation and virus accumulation, we inoculated the

Transient overexpression of NbNBR1 increases βC1 accumulation to benefit TYLCCNV/TYLCCNB infection
We transiently co-expressed Myc-NbNBR1 and βC1-YFP in N. benthamiana leaves through agroinfiltration to analyze the effect of NbNBR1 on βC1 accumulation. Western blot and end point qPCR results showed that co-expression of Myc-NbNBR1 and βC1-YFP increased the accumulation of βC1-YFP protein, but not βC1-YFP mRNA (Figs 3A and S4A). Furthermore, to avoid the potential effect of YFP tag, we transiently co-expressed Myc-NbNBR1, Myc-GUS, GD-βC1, and GD, in N. benthamiana leaves through agroinfiltration, and then analyzed these leaves for the accumulation of GD-βC1 through Western blot assay using anti-Myc and anti-βC1 antibodies. The results showed that transient co-expression of Myc-NbNBR1 and GD-βC1 did increase the accumulation of GD-βC1 ( Fig 3B). When the concentration of Agrobacterium tumefaciens culture carrying Myc-NbNBR1 was increased from OD 600 = 0.2 to 0.4 and mixed with an A. tumefaciens culture carrying βC1-YFP or GD-βC1 prior to agroinfiltration, the accumulation levels of βC1-YFP and GD-βC1 were also significantly increased (Fig 3C and 3D). Besides, we also conducted BiFC assays to observe interactions between NbNBR1 and βC1 at different time points. As shown at S4B Fig, the cytoplasmic granules of NbNBR1-βC1 complex became bigger and more stable at

PLOS PATHOGENS
NbNBR1 prevents degradation of geminivirus βC1 protein 60 hpi compared to those at 48 hpi and at 36 hpi. Western blot verified that accumulation level of NbNBR1 or βC1 at 60 hpi increased as compared to that at 48 hpi or 36 hpi by using anti-HA antibodies (S4C and S4D Fig). We also used a quantification method to support our finding by size estimation of NbNBR1-βC1 complex at different time (S4E Fig). Above all, we confirmed that NbNBR1 enhanced βC1 accumulation.
In the case that NbNBR1 could enhance βC1 protein level, it may affect TYLCCNV/ TYLCCNB infection as feedback. We inoculated N. benthamiana leaves with TYLCCNV/ TYLCCNB and Myc-NbNBR1 or Myc-GUS (control) via agroinfiltration. The TYLCCNV/ TYLCCNB and Myc-NbNBR1-inoculated plants at 7 dpi showed accelerated and aggravated symptoms as compared to the plants inoculated with TYLCCNV/TYLCCNB and Myc-GUS ( Fig 4A). Western blot analyses of the infiltrated leaf samples at 3 dpi showed that transient overexpression of Myc-NbNBR1 increased the accumulations of the βC1 protein (Fig 4B). qPCR and Southern blot analyses of the viral accumulation from the systemic leaves at 7 dpi ( Fig 4C) and at 14 dpi ( Fig 4D) showed that NbNBR1 positively regulates viral DNA accumulations. Besides, different concentrations of A. tumefaciens cultures carrying Myc-NbNBR1 (OD 600 = 0, 0.2, 0.4) were also utilized to validate its effect on TYLCCNV/TYLCNNB infection. Consistently, the increased protein levels of NbNBR1 led to accelerated viral symptom development ( Fig 4E) with more βC1 accumulations at 3 dpi ( Fig 4F) and higher viral genomic DNA accumulations in systemic leaves (Fig 4G and 4H).

Transgenic overexpression of NbNBR1 facilitates TYLCCNV/TYLCCNB infection
Furthermore, we generated two NbNBR1-YFP-HA transgenic N. benthamiana T1 lines (NbNBR1-YFP-HA-L1 and NbNBR1-YFP-HA-L2) and inoculated them with TYLCCNV/ TYLCCNB through agroinfiltration. Growth phenotypes of the two transgenic lines were similar to that shown by the Wt plants ( Fig 5A). Yellow fluorescence could be observed under a confocal microscope from leaf samples of these two transgenic lines (Fig 5B). Western blot assay using anti-HA antibodies further confirmed the expression of NbNBR1-YFP-HA in these two transgenic lines (Fig 5C). After inoculation with TYLCCNV/TYLCCNB, much severe disease symptoms appeared on NbNBR1-transgene N. benthamiana plants at 7 dpi compared to those in the Wt plant (Fig 5D). At the same time, the NbNBR1-YFP-HA transgenic plants accumulated more βC1 protein levels than the Wt plants in the infiltrated leaves at 3 dpi ( Fig 5E). qPCR analyses at 7 dpi and Southern blot analyses at 14 dpi of viral DNA accumulations in the systemic leaves of plants indicated in Fig 5D, showed higher viral DNAs accumulated in the NbNBR1-YF-P-HA-overexpressing N. benthamiana plants than that in the Wt plants (Fig 5F and 5G). Meanwhile, we inoculated these two NbNBR1-YFP-HA transgenic N. benthamiana lines with PVX or PVX-βC1. Overexpression of NbNBR1 did not affect PVX infection (S5A and S5B Fig). However, at 5 dpi, NbNBR1-overexpressing lines inoculated with PVX-βC1 showed much severe disease symptoms together with increased βC1 protein level as compared to the Wt plants (S5C and S5D Fig). Additionally, we used qRT-PCR to check mRNA level of βC1 on TYLCCNV/ TYLCCNB infected NbNBR1-YFP-HA transgenic plants at 2 dpi, no marked change was found, confirming that NbNBR1 affected the level of βC1 protein rather than mRNA (S5E Fig). Above data suggest that NbNBR1 is a susceptible factor for TYLCCNV/TYLCCNB infection.

βC1 escapes the NbRFP1-mediated degradation in the presence of NbNBR1
Our previous study revealed NtRFP1 could bind and ubiquitinate βC1 to mediate their degradation [31]. Because NBR1 as a selective autophagic adaptor recognizes ubiquitinated protein complexes [5], we hypothesized that the ubiquitinated NbRFP1 might be a substrate of NbNBR1. To test this hypothesis, we investigated the interaction between NbNBR1 and NbRFP1 (a homolog of NtRFP1 in N. benthamiana) using Y2H, BiFC, and Co-IP assays. The result of Y2H assay showed that the yeast cells co-transformed with AD-NbRFP1+ BD-NbNBR1 or AD-NbNBR1+BD-NbRFP1 failed to grow on the selective medium supplemented with 10 mM 3-amino-1,2,4-triazole (3-AT), while yeast cells co-transformed with AD-βC1+BD-NbRFP1 did (Fig 6A). Similarly, no interaction was found between NbNBR1 and NbRFP1 in BiFC, and Co-IP assays, and no co-localization was observed between NbRFP1-YFP and NbNBR1-CFP in plant cell (Fig 6B-6D), indicating that NbNBR1 and NbRFP1 did not interact in vivo. In order to know if NbNBR1 affects the expression level of NbRFP1, we analyzed NbRFP1 mRNA expression level in NbNBR1-YFP-HA transgenic lines,

PLOS PATHOGENS
NbNBR1 prevents degradation of geminivirus βC1 protein NbNBR1-Cas9 transgenic lines and TRV-NbNBR1 inoculated lines. No change of NbRFP1 mRNA expression level in the above plants compared to control plants was found (Fig 6E and  6F). Besides, co-expression of Myc-NbNBR1 and NbRFP1-GFP in N. benthamiana was used to investigate NbRFP1 protein level by anti-GFP antibodies. As shown in Fig 6G, no specific variation of NbRFP1 was found either when co-expression with Myc-NbNBR1 or not ( Fig  6G). These results indicate that NbNBR1-mediated accumulation of βC1 is not because of altering the expression level of NbRFP1.
Based on these results, we thus speculated that NbNBR1 can compete with NbRFP1 for βC1. To test this, we firstly investigated the interaction among the three proteins through Y3H assay. We found that NbRFP1 interacted with βC1, and this interaction was not affected in the presence of MBP (Fig 7A). In contrast, the presence of NbNBR1-MBP inhibited the interaction between βC1 and NbRFP1. Next, we conducted confocal imaging to check whether NbNBR1 To further validate the relationship of NbNBR1, NbRFP1, and βC1 in plant, we co-expressed GD-βC1+Myc-NbNBR1+NbRFP1-GFP (Fig 7C, 1st lane), GD+Myc-NbNBR1+NbRFP1-GFP (Fig 7C, 2nd lane), GD-βC1+Myc-NbNBR1+GFP (Fig 7C, 3rd lane), GD+Myc-NbNBR1+GFP (Fig 7C, 4th lane), GD-βC1 +Myc-GUS+NbRFP1-GFP (Fig 7C, 5th lane), GD+Myc-GUS+NbRFP1-GFP (Fig 7C, 6th lane), and the corresponding protein accumulations were analyzed using anti-βC1, anti-Myc and anti-GFP antibodies. The result showed that much more βC1 accumulated in the extracts containing GD-βC1, Myc-NbNBR1, and NbRFP1-GFP than the extracts containing GD-βC1, Myc-NbNBR1 and GFP (Fig 7C, compare the 1st lane with the 5th lane in input Western blot using anti-βC1 antibodies), suggesting that the presence of NbNBR1 can protect βC1 from the NbRFP1-mediated degradation. In addition, less GD-βC1 was immunoprecipitated using GFP-Trap_MA magnetic agarose beads from the extracts containing GD-βC1+Myc-NbNBR1 +NbRFP1-GFP (Fig 7C, 1st lane) than the extracts containing GD-βC1+Myc-GUS+-NbRFP1-GFP (Fig 7C, 5th lane), supporting that NbNBR1 could interfere the interaction between NbRFP1 and βC1. Furthermore, we analyzed the ubiquitination level of the total protein after TYLCCNV/TYLCCNB infection in both NbNBR1-overexpression transgenic lines (NbNBR1-YFP-HA-L1/L2) and NbNBR1-Cas9 lines (NbNBR1-Cas9-L3/L4). As expected, overexpression of NbNBR1 reduced the ubiquitination level of total protein, while vice versa (S6A and S6B Fig). A single amino acid substitution in βC1 (βC1 K4A ) abolishes its interaction with NbNBR1 and attenuates TYLCCNV/TYLCCNB infection To further investigate the interaction among NbNBR1, βC1, and NbRFP1, we mutated βC1 through truncated mutation and site-directed mutagenesis. Firstly, we divided βC1 into 12 fragments, and found the first fragment (1-10 aa) of βC1 is required for its interaction with NbNBR1 (S7A and S7B Fig). Then we made a series of amino acid site mutations, and found that βC1 with the fourth lysine mutation no longer interacted with NbNBR1 in yeast cells ( Fig  8A). Under the confocal microscope, no YFP fluorescence was observed in the RFP-H2B transgenic N. benthamiana leaf cells when co-expressing YN-βC1 K4A and YC-NbNBR1 ( Fig  8B). Co-IP assay also showed that βC1 K4A and Myc-NbNBR1 were not immunoprecipitated together any more ( Fig 8C). However, this mutation did not interfere with the interaction between βC1and NbRFP1 (Fig 8B). It is noteworthy that co-expression of Myc-NbNBR1 +βC1-YFP in cells increased the intensity of βC1-YFP fluorescence, while the increased yellow fluorescence was not seen in the cells that co-expression of Myc-NbNBR1+βC1 K4A -YFP ( Fig  8D), suggesting that the interaction between NbNBR1 and βC1 stabilized βC1.
To investigate the role of βC1 and NbNBR1 interaction in βC1-induced symptoms, PVX-βC1 and PVX-βC1 K4A were inoculated individually to N. benthamiana plants. By 7 dpi, the PVX-βC1-inoculated plants showed strong mosaic and leaf curling symptoms (Fig 9A). Although the PVX-βC1 K4A -inoculated plants also showed strong mosaic symptoms, they did not show strong leaf curling symptoms and Western blot result showed PVX-βC1 K4A -inoculated plants accumulated much less βC1 as compared with the PVX-βC1-inoculated plants (Fig 9A and 9B). We then introduced the K4A mutation into the TYLCCNB (TYLCCNB K4A ) and inoculated TYLCCNV/TYLCCNB and TYLCCNV/TYLCCNB K4A , respectively, to N. benthamiana plants through agroinfiltration. By 30 dpi, the TYLCCNV/TYLCCNB K4A -inoculated plants displayed attenuated disease symptoms (Fig 9C). Western blot result showed that the TYLCCNV/TYLCCNB K4A -inoculated plants also accumulated much less βC1 than the TYLCCNV/TYLCCNB-inoculated plants at 3 dpi ( Fig 9D) and Southern blot assay at 30 dpi ( Fig 9E) showed that less viral genomic DNAs in Fig 9C when compared to the TYLCCNV/ TYLCCNB-inoculated plants. Furthermore, co-expression of Myc-NbNBR1 and βC1 K4A -HA in N. benthamiana leaves resulted in a much lower level of βC1 K4A than that in the leaves coexpression of Myc-NbNBR1 and βC1-HA (Fig 9F). Taken together, these results indicate that the interaction between βC1 and NbNBR1 is required for the βC1 protein accumulation and βC1-dependent viral pathogenicity.

PLOS PATHOGENS
NbNBR1 prevents degradation of geminivirus βC1 protein Rice gall dwarf virus has been shown to induce virion containing autophagosomes that can facilitate virus transmission through insect vectors [39].
NBR1 was initially reported as a selective autophagy receptor that has a typical conserved LC3 (microtubule-associated protein 1 light chain 3)-interacting region (LIR) and an ubiquitin (Ub)-binding domain (UBA) and can be recruited to ubiquitinate its targets. Arabidopsis AtNBR1 is involved in autophagy-associated degradation of vacuoles and contains two UBA domains [40]. N. tabacum NtJoka2 is a homolog of AtNBR1, and acts as a selective receptor to inhibit the colonization of potato blight pathogen, Phytophthora infestans [41][42]. Here we showed a geminivirus satellite-encoded protein could upregulate NbNBR1 expression when transgenic overexpression and transient overexpression of βC1 in N. benthamiana leaves (S8 Fig), and interact with NbNBR1 to form cytoplasmic granules. These cytoplasmic granules can prevent βC1 from being degraded by the NbRFP1-mediated UPS-dependent degradation system to facilitate geminivirus infection, which demonstrated a distinct function of NBR1. NBR1 as an autophagy adaptor can participate in the autophagic degradation of its substrates. However, co-expression of NbNBR1 and βC1 in our study failed to degrade βC1 but increased its  (Fig 3). Consistent with this result, transient or stable overexpression of NbNBR1 in N. benthamiana enhanced βC1 virulence, leading to stronger viral symptoms (Figs 4 and 5), and vice versa (Fig 2). In addition, overexpression of NbNBR1 in N. benthamiana also promoted the infection of another monopartite geminivirus: tomato yellow leaf curl virus Beijing isolate (TYLCV-BJ) (S9 and S10 Figs), and its V2 protein interacted with NbNBR1 and formed cytoplasmic granules as well (S11 Fig). These findings suggest that the NbNBR1-mediated pro-viral role may be a general phenomenon during geminivirus infections.
NtRFP1 can be ubiquitinated as a substrate of the UPS-dependent degradation machinery and mediate the degradation of βC1 [31], and it may be recognized by NBR1. However, we did not observe a direct interaction between NbNBR1 and NbRFP1 through Y2H, BiFC, co-localization and Co-IP assays (Fig 6). Y3H assay showed that NbNBR1 competed with NbRFP1 for βC1, which disrupted the interaction between NbRFP1 and βC1 (Fig 7A), indicating a different fate of βC1 when it is hijacked by RFP1 or NBR1 in plant. A mutant βC1 with a single amino acid substitution (βC1 K4A ) which failed to interact with NbNBR1 but it could still interact with NbRFP1, was found in this study (Fig 8). We found less βC1 accumulations and milder βC1-induced viral symptoms in PVX-βC1 K4A mutant in contrast to wild type PVX-βC1 -infected plants (Fig 9A and 9B). Similarly, the TYLCCNV/TYLCCNB K4A also displayed much weaker virulence and accumulated less viral genomic DNA compared to wild type TYLCCNV/TYLCCNB in infected plants (Fig 9C-9E), supporting the interaction between βC1 and NbNBR1 contributes to the stability and virulence of βC1.
Based on the above results, we proposed a working model for the interactions among NbNBR1, NbRFP1, and βC1 during geminivirus infection in plant (Fig 10). Geminiviruses are delivered into plant cells by insect vectors. Inside the infected cells, viral proteins are translated and become the targets of host surveillance. The NbRFP1-mediated UPS-dependent degradation system recognizes, interacts with, and then degrades the viral betasatellite-encoded βC1, leading to an inhibition of virus accumulation and disease symptom formation [26]. To counteract host defense, βC1 interacts with NbNBR1 to form cytoplasmic granules to prevent βC1 degradation by NbRFP1. The finding of NbNBR1, NbRFP1, and βC1 three-way-interactions expands our knowledge on the arm races between plant and geminiviruses. In addition, our finding also suggests that NbNBR1 is a susceptible host gene to geminivirus invasion, which might be applied into molecular breeding in other crops for restricting viral infection.

Plant materials and growth conditions
Wild type (Wt), RFP-H2B transgenic, βC1 transgenic, NbNBR1-YFP-HA transgenic, and NbNBR1-Cas9 mutant N. benthamiana seedlings were grown in pots inside an insect-free greenhouse maintained at 22˚C/18˚C (day/night), 16 h light/8 h dark photoperiod, and 60% relative humidity. These plants were used for assays after they grew to 4-5 leaves.

PLOS PATHOGENS
NbNBR1 prevents degradation of geminivirus βC1 protein transformed into Y2HGold strain cells. The transformed cell cultures were grown individually on the selective medium (SD) with -Leu-Trp for 96 h at 30˚C, and then on selective medium with SD/-Leu-Trp-His-Ade to determine the interactions between NbNBR1 and βC1 or βC1 K4A . The selective medium with SD/-Leu-Trp-His-Ade and 10 mM 3-aminotriazole (3-AT) was used to determine the interactions between NbRFP1 and NbNBR1. In this study, cell cultures co-transformed with AD-T7-T and BD-T7-53 (referred to as AD-T+BD-53 thereafter) or AD-T7-T and BD-T7-Lam (AD-T+BD-Lam) were used as the positive and negative controls, respectively. For Y3H assays different combinations of plasmids were transformed into AH109 strain cells using the Yeast Transformation II kit as instructed (ZYMO, USA). The transformed cell cultures were grown individually on the selective medium (SD) with -Leu-Trp for 96 h at 30˚C, and then on selective medium with SD/-Leu-Met-Trp (SD/-3), and SD/-His-Leu-Met-Trp (SD/-4) media, respectively.
For BiFC and subcellular localization assays, the excitation wavelength of CFP was set at 458 nm and the emission was captured at 470-500 nm. The excitation wavelength of YFP was set at 514 nm and the emission was captured at 565-585 nm. The excitation wavelength of GFP was set at 488 nm and the emission was captured at 510-550 nm, while the excitation wavelength of RFP was set at 543 nm and the emission was captured at 590-630 nm as described [7]. Epidermal cells of the assayed leaves were harvested at 48 and 72 hours post infiltration (hpi) and examined under a confocal laser scanning microscopy 980 (Zeiss, GRE).

DNA extraction and Southern blot analyses
Total DNA was extracted from leaf samples using the CTAB method. After denaturation and neutralization, total DNA was separated in agarose gels through electrophoresis and then transferred to Hybond N + nylon membranes (GE Healthcare, Pittsburgh, PA, USA). The membranes were hybridized at 55 o C with digoxin-labeled probes prepared using the DIG High Prime DNA Labeling and Detection Starter Kit (Roche, Mannheim, GRE). Agarose gels stained with GelStain (TransGen Biotech, Beijing, CHN) were used to show equal sample loadings.

Quantitative polymerase chain reaction (qPCR) and quantitative reverse transcription PCR (qRT-PCR)
Total genomic DNA (gDNA) and total RNA (T. RNA) were extracted, respectively, from the infiltrated and expanding young leaves using the CTAB method or Trizol reagent (Invitrogen, Carlsbad, CA, USA). For qPCR, 100 ng gDNA was used in each 10 μL reaction. For qRT-PCR, 1000 ng T. RNA was used in a 10 μL RT reaction made with a reverse transcription kit (TAKARA, JPN) followed by qPCR using the universal SYBR Green Master kit (TAKARA, JPN). The expression levels of 25S rRNA and F-box mRNA were used as the internal controls for qPCR and qRT-PCR assays, respectively. At the end of each run, melting curve was used to assess the specificity of amplification product (60 to 95˚C with heating rate at 0.5˚C for 10 s and measure fluorescence continuously).

Transient gene expression and Western blot assays
Transient gene expression assays were performed in N. benthamiana leaves through agroinfiltration as described above. Total protein was extracted from individual leaf samples homogenized in the extraction buffer and separated in gels through electrophoresis. After transferring to blotting membranes, protein bands were detected using anti-GFP (Roche, Mannheim, GRE) or anti-Myc (Genscript, Piscataway, NJ, USA) polyclonal antibodies, or anti-HA (Roche, Mannheim, GRE) or anti-βC1 (made in our laboratory) monoclonal antibodies.  Table. A detailed list of primers used in this study (5'-3'). (DOCX)