Host GRXC6 restricts Tomato yellow leaf curl virus infection by inhibiting the nuclear export of the V2 protein

Geminiviruses cause serious symptoms and devastating losses in crop plants. With a circular, single-stranded DNA genome, geminiviruses multiply their genomic DNA in the nucleus, requiring the nuclear shuttling of viral proteins and viral genomic DNAs. Many host factors, acting as proviral or antiviral factors, play key roles in geminivirus infections. Here, we report the roles of a tomato glutaredoxin (GRX), SlGRXC6, in the infection of Tomato yellow leaf curl virus (TYLCV), a single-component geminivirus. The V2 protein of TYLCV specifically and preferentially interacts with SlGRXC6 among the 55-member tomato GRX family that are broadly involved in oxidative stress responses, plant development, and pathogen responses. We show that overexpressed SlGRXC6 increases the nuclear accumulation of V2 by inhibiting its nuclear export and, in turn, inhibits trafficking of the V1 protein and viral genomic DNA. Conversely, the silenced expression of SlGRXC6 leads to an enhanced susceptibility to TYLCV. SlGRXC6 is also involved in symptom development as we observed a positive correlation where overexpression of SlGRXC6 promotes while knockdown of SlGRXC6 expression inhibits plant growth. We further showed that SlGRXC6 works with SlNTRC80, a tomato NADPH-dependent thioredoxin reductase, to regulate plant growth. V2 didn’t interact with SlNTRC80 but competed with SlNTR80 for binding to SlGRXC6, suggesting that the V2-disrupted SlGRXC6-SlNTRC80 interaction is partially responsible for the virus-caused symptoms. These results suggest that SlGRXC6 functions as a host restriction factor that inhibits the nuclear trafficking of viral components and point out a new way to control TYLCV infection by targeting the V2-SlGRXC6 interaction.


Introduction
Geminiviruses are a group of plant viruses with a circular, single-stranded DNA genome that is encapsulated in twinned icosahedral particles [1][2][3]. Viral infection causes disease symptoms that include chlorosis and/or necrosis, leaf curling, and altered plant stature and morphology, leading to extensive agricultural losses worldwide [2,4,5].
Tomato yellow leaf curl virus (TYLCV) is a typical monopartite begomovirus in the family Geminiviridae that contains a single genome component with six open reading frames (ORFs). Multiple studies have shown that V2 is a suppressor of gene silencing at both the post-transcriptional stage (PTGS) [17,18] and the transcriptional level (TGS) [19,20]. V2 is also involved in the regulation of host defence responses [21] and viral movement [2,[22][23][24][25][26], playing important roles in viral spread and systemic infection [27,28]. Moreover, it was found that V2, when expressed from a PVX vector, caused severe leaf curling, stunting, vein yellowing and necrotic lesions on systemic leaves [13,19,29], suggesting that V2 may serve as an important symptom determinant in TYLCV pathogenicity. Nevertheless, it is unclear how and what host factors are involved in the V2-mediated functions during TYLCV infection.
Glutaredoxins (GRXs) are a group of low-molecular-weight thiol oxidoreductasea and are implicated in response to oxidative stress [30]. Based on the sequences of the active site, GRXs are classified into three distinct subgroups: the CPYC type, which contains C[P/G/S] [Y/F][C/S] motifs; the CGFS type, which has a strictly conserved CGFS active site; and the CC type, which contains the CC[M/L][C/S/G/A/I] active site [31]. While the CPYC-and CGFS-type GRXs are widely found in all eukaryotic and prokaryotic species, the CC-type GRXs are restricted to land plants [32]. In plants, GRXs are involved in plant development, signal transduction, and other biological processes [30,33,34]. It has recently been reported that GRXs play an important role in response to pathogens [34], but the mechanism is not well known.
Tomato (Solanum lycopersicum) glutaredoxin-C6 (SlGRXC6) belongs to the CC-type GRXs, which is the largest type in plants [31,35]. Early reports show that CC-type GRXs are involved in plant development [36] and pathogen responses through jasmonic acid (JA) or salicylic acid (SA) signalling pathways [34,37]. Among the five well-studied CC-type GRXs in Arabidopsis, three are related to host defence response: ROXY19 (GRX480), ROXY4 and GRXS13 [38][39][40]. The other two (ROXY1 and ROXY2) are involved in flower development and regulating floral organ primordium formation [36,[41][42][43]. The two CC-type GRXs in rice (OsROXY1 and OsROXY2) are also involved in defence responses to pathogens [44]. These collectively suggest that the CC-type GRXs play an important role in plant growth and host defence responses to pathogens.
It is well-known that there are cross-talks between the thioredoxin (TRX) and the glutathione systems [45]. TRXs and GRXs are critical for plant development and cell division. They also act as key signalling molecules in response to abiotic and biotic stresses [46][47][48][49]. The interrelation between GRX and NADPH-dependent thioredoxin reductase (NTR) in the regulation of plant growth has previously been investigated [50,51]. In Arabidopsis, the overexpression of NTR enhanced leaf growth and the overall height of plants, which was about 40% higher than that of wild-type (WT) plants [52]. Conversely, a lack of NTR results in stunted growth and decreased fertility [51,53,54]. On the other hand, the glutathione pathway plays a major role in compensating for NTR inactivation in the mutant transgenic plants [55]. In addition, plants lacking NTR show enhanced disease susceptibility to fungal and bacterial pathogens [56,57]. These results suggest that GRXs may work together with NTR in regulating plant growth and response to pathogen attacks; however, there is no direct evidence supporting this claim.
In this study, we demonstrate that SlGRXC6 is a host restriction factor of TYLCV infection. We identify and characterize a specific interaction between V2 and SlGRXC6, and show that overexpressed SlGRXC6 inhibits the nuclear export of V2 and, in turn, inhibits viral systemic infection in tomato plants. Conversely, knocking down the expression of SlGRXC6 promotes viral infection. We additionally demonstrate that expression levels of SlGRXC6 have a positive correlation with plant growth and that the SlGRXC6-mediated plant growth is related to SlNTRC80, which interacts with SlGRXC6. Using a competitive pulldown assay, we demonstrate that V2 competes with SlNTRC80 for a direct binding to SlGRXC6, suggesting that the disrupted SlGRXC6-SlNTRC80 interaction is likely accountable for the affected host growth during viral infection. These results show that SlGRXC6 is a host restriction factor that prohibits TYLCV infection and that regulating the V2-SlGRXC6 interaction could provide a new way to control TYLCV.

The V2 protein Interacts with SlGRXC6
It has been reported that the expression of TYLCV V2 protein induced severe symptom-like phenotype in plants [13,19,29]. We observed a similar phenotype in tomato transgenic plants expressing V2 of TYLCV (S1 Fig), indicating that V2 plays an important role in viral pathogenicity. To identify host target(s) involved in the V2-mediated pathogenesis in host plants, we performed a yeast two-hybrid (Y2H) screen of a tomato cDNA library using V2 as the bait. Two cDNAs encoding full-length SlGRXC6 (GenBank accession no. XM004251147) were identified, suggesting that SlGRXC6 could be an interacting partner of V2. To further confirm the SlGRXC6-V2 interaction, the full-length coding sequence of SlGRXC6 was amplified using tomato leaf tissue as a source and its interaction with V2 was confirmed in Y2H ( Fig 1A).
A bimolecular fluorescence complementation (BiFC) assay was performed to verify the interaction between V2 and SlGRXC6 in plant cells. SlGRXC6 and V2 were fused to the N-terminal and C-terminal fragments of yellow fluorescent protein (YFP), respectively. The corresponding constructs were co-delivered into N. benthamiana leaves by agroinfiltration and fluorescence was observed using a confocal microscope at 48 hours post-agroinfiltration (hpai). A positive interaction between nYFP-SlGRXC6 and cYFP-V2 was observed in both the and SlGRXC6. SlGRXC6 was fused with a GAL4 activation domain (AD-SlGRXC6) and V2 was fused to a GAL4-binding domain (BD-V2), respectively. Yeast cells expressing the indicated protein pairs were plated onto the selection medium (SD/-His/-Leu/-Trp/-Ade) with X-α-Gal to screen for positive interactions. Yeast cells coexpressing pGADT7-largeT (AD-largeT) and pGBKT7-p53 (BD-p53) or AD-largeT and pGBKT7-LaminC (BD-LaminC) served as a positive and negative control, respectively. (b) The V2-SlGRXC6 interaction was confirmed by using a BiFC assay in N. benthamiana cells. The V2-SlGRXC6 interaction led to a reconstituted fluorescence signal. DAPI stains DNA in the nucleus. Bars: 50 μm. Experiments were repeated three times and 30 cells were observed in each repeat. (c) A co-IP assay to test the interaction between V2 and SlGRXC6, SlGRXC6 T53A or SlGRXC6 C58A . N. benthamiana leaves were co-infiltrated with FLAG-V2 and SlGRXC6-YFP (Lane 1), FLAG-V2 and SlGRXC6 T53A -YFP (Lane 2), FLAG-V2 and SlGRXC6 C58A -YFP (Lane 3), or FLAG-V2 and YFP (Lane 4). Cell lysates were incubated with FLAG-Trap beads and proteins pulled down with beads were tested using the indicated antibodies. Samples before (Input) and after (IP) immunoprecipitation were analyzed by using anti-GFP or -FLAG antibody. ( cytoplasm and the nucleus as indicated by the presence of the reconstituted fluorescence in all cells (n = 30) (Fig 1B). No fluorescence signal was generated when nYFP-SlGRXC6 and cYFP or nYFP and cYFP-V2 were coexpressed (Fig 1B), reinforcing a specific interaction between V2 and SlGRXC6 in plant cells.
To provide more evidence of the interaction between V2 and SlGRXC6, a co-immunoprecipitation (co-IP) assay was performed with coexpressed FLAG-tagged V2 (FLAG-V2) and YFP or YFP-tagged SlGRXC6 (SlGRXC6-YFP) in N. benthamiana leaves. Total protein extracts were incubated with FLAG-Trap beads and the resulting precipitates were analysed by western blot assays using anti-FLAG or -YFP antibody. We found that SlGRXC6-YFP, but not YFP, co-precipitated with FLAG-V2 (Fig 1C), even though both YFP and SlGRXC6-YFP were well-expressed ( Fig 1C, top panel). These results indicate a specific association between V2 and SlGRXC6 in plant cells.
In tomato, there are 55 GRXs that are divided into four groups; the CC-type is the largest type with 35 members [35]. Amino acid sequence analysis revealed that SlGRXC6 is a close homolog of AtGRXC10 and AtGRXC6 and belongs to the CC-type GRXs with a conserved CCMC motif (S2A Fig). To test whether SlGRXC6 was the only one among the tomato GRXs that specifically interacted with V2, we selected multiple members to test their interactions with V2 (S2B Fig). Three members were from the CC-type GRXs: SlGRX25, SlGRX38, and SlGRX39. SlGRX38 and SlGRX39 have the same CCMC active motif as SlGRXC6, but SlGRX25 has a CCIS active motif. We also included two GRXs from the CGFS-type (SlGRX9 and SlGRX36), two from the CYPC-type (SlGRX27 and SlGRX43), and two from the GRLtype (SlGRXL1 and SlGRXL3) [35]. As shown in S2B Fig, all yeast cells expressing V2 and one of different tomato GRXs grew well on medium without selection, but most did not grow at all on the selection medium. Yeast cells expressing V2 and SlGRX43 or SlGRX39 grew much slower than those with V2 and SlGRXC6, showing that V2 preferentially interacts with SlGRXC6 among tomato GRX members.
After confirming the V2-SlGRXC6 interaction, we set out to identify the domain, motif, and amino acids in SlGRXC6 that were critical for the interaction. Based on a motif scan analysis (http://www.ebi.ac.uk/interpro/), there are two conserved domains in SlGRXC6. One domain is glutaredoxin PF00462, spanning amino acids 49 to 84 and the other is glutaredoxin PS51354, spanning amino acids 37 to 113 of SIGRXC6. We constructed seven SlGRXC6 deletion mutants as shown in Fig 1D, and determined that the shortest region interacting with V2 was SlGRXC6-M3 (aa 49-84) ( Fig 1D).
To more specifically identify the amino acids in the M3 fragment that are involved in the interaction, we selected five amino acids based on post-translational modifications. We constructed mutants with a single alanine substitution in T53, G56, C58, S69, and S72 to make constructs SlGRXC6 T53A , SlGRXC6 G56A , SlGRXC6 C58A , SlGRXC6 S69A , and SlGRXC6 S72A , respectively. Based on the Y2H analysis (Fig 1E), SlGRXC6 T53A and SlGRXC6 C58A failed to interact with V2, suggesting that T53 and C58 are possibly the key sites in SlGRXC6 for interaction with V2.
We then tested the interactions of SlGRXC6 T53A and SlGRXC6 C58A with V2 using the co-IP assay in plant cells. YFP-tagged WT SlGRXC6, SlGRXC6 T53A , SlGRXC6 C58A , or YFP were coexpressed with FLAG-V2, and the total lysate was subjected to the co-IP assay. Although WT or SlGRXC6 mutants accumulated to similar levels, only WT SlGRXC6-YFP, but not SlGRXC6 T53A -or SlGRXC6 C58A -YFP, was pulled down along with FLAG-V2 ( Fig 1C). This confirmed that the Ala substitution in T53 or C58 affected the ability of SlGRXC6 to interact with V2.
We next tested whether such mutations affect localization of SlGRXC6. When SlGRXC6 T53A -or SlGRXC6 C58A -YFP was expressed in H2B transgenic N. benthamiana, the fluorescence signal of SlGRXC6 T53A -or SlGRXC6 C58A -YFP was similarly observed in the cytoplasm and nucleus in all cells (n = 30) (Fig 1F), indicating that the T53A or C58A substitution had no effect on localization of SlGRXC6. Western blotting confirmed that SlGRXC6 T53A -or SlGRXC6 C58A -YFP accumulated to a level similar to SlGRXC6 (Fig 1F), indicating that both mutations did not affect the expression and stability of mutant proteins.

SlGRXC6 Regulates TYLCV Infection in Tomato Plants through Its Interaction with the V2 protein
To assess the biological significance of the V2-SlGRXC6 interaction in vivo, we overexpressed SlGRXC6 in tomato using a PVX vector [58]. The accumulated SlGRXC6 transcripts in plants overexpressing SlGRXC6 increased 3-fold more than that of PVX control plants at 8 days postagroinfiltration (dpai), based on qRT-PCR (Fig 2A). Unexpectedly, the plants overexpressing SlGRXC6 grew faster and taller than those with PVX only (Fig 2B). At 16 dpai, the average aboveground height of SlGRXC6-overexpressed plants was 28 cm, a 60% increase over that of PVX-inoculated plants ( Fig 2C).
PVX-and PVX-SlGRXC6-treated plants were then infected with TYLCV at 8 dpai. At 13 days post-inoculation (dpi), TYLCV-infected PVX-SlGRXC6-plants showed milder symptoms compared to that of PVX-treated plants ( Fig 2D). The aboveground height of SlGRXC6-overexpressed plants infected with TYLCV was much taller than that of WT plants (Fig 2E). At 11 days after TYLCV inoculation, 2 out of the 15 PVX-treated plants started showing symptoms, but all 15 PVX-SlGRXC6-treated plants remained symptomless ( Fig 2F). Importantly, only 7 out of a total of 15 inoculated plants developed mild symptoms compared to 12 of the TYLCVinfected PVX plants, which had strong symptoms at 16 dpi ( Fig 2F). Although all plants developed symptoms by 21 days, the TYLCV-infected PVX-SlGRXC6 plants showed weaker symptoms and in addition, at 23 dpi the accumulated TYLCV genomic DNA decreased by 57% compared to that of mock plants as revealed by qPCR ( Fig 2G). However, accumulated viral DNA decreased 16% in PVX-SlGRXC6 plants compared to that in PVX plants by 33 dpi ( Fig  2G). Given that the accumulated TYLCV genomic DNA in SlGRXC6-overexpressing plants was much lower at 23 dpi but similar at 33 dpi than those in mock plants, we conclude that overexpressed SlGRXC6 can mitigate but cannot totally block TYLCV infection.
We further tested TYLCV infection in SlGRXC6-silenced tomato plants. To knock down SlGRXC6 expression, we used a virus-induced gene silencing (VIGS) approach that was mediated by Tobacco rattle virus (TRV) [59]. We used the endogenous phytoene desaturase gene (PDS) as a reporter for monitoring the effectiveness and progress of TRV-induced gene silencing. The upper leaves of TRV-PDS-infiltrated plants turned white at 12 dpai and the accumulated SlPDS transcripts were about 30% of those in TRV-infected leaf tissues (S3 Fig), similar to that reported in tomato plants [59]. Following the same protocol, we next inoculated tomato plants with TRV-SlGRXC6. The expression of SlGRXC6 was downregulated in the SlGRXC6silenced lines to 50% of that in the TRV control lines at 12 dpai ( Fig 3A). In sharp contrast to the promoted plant growth in the SlGRXC6-overexpressing lines (Fig 2B), TRV-SlGRXC6-inoculated plants were shorter than TRV-inoculated plants (Fig 3B). At 16 dpai, the height of TRV-SlGRXC6 plants was 12 cm, significantly shorter than that of TRV-treated plants at 20 cm ( Fig 3C).
TRV and TRV-SlGRXC6 plants were then inoculated with TYLCV at 12 dpai. The TRV-SlGRXC6 plants developed more severe symptoms and were much shorter than those of TYLCV-infected TRV plants (Fig 3D and 3E

Host GRXC6 restricts TYLCV infection
To further verify the effect of V2-SlGRXC6 interaction on viral infection, we overexpressed SlGRXC6 T53A or SlGRXC6 C58A in tomato plants using the PVX vector. The aboveground plant height of SlGRXC6 T53A -and SlGRXC6 C58A -overexpressed plants were similar to that of mock plants ( Fig 4A). Moreover, the symptoms and viral DNA accumulated at similar levels in all TYLCV-inoculated plants (Fig 4B and 4C). Given the fact that both mutants accumulated at levels similar to WT (Fig 1F), these data showed that SlGRXC6 mutations had no effect on viral infection, suggesting that SlGRXC6 regulates TYLCV infection possibly through its interaction with V2 in tomato plants.

Host GRXC6 restricts TYLCV infection
To rule out the possibility that the inhibited TYLCV infection in TRV-SlGRXC6-treated plants was not due to the slowed plant growth, we included three other different tomato GRXs that were from the same CC-type group as SlGRXC6, SlGRX25, SlGRX38, and SlGRX39.

SlGRXC6 Inhibits the Nuclear Export of the V2 Protein
We previously reported that the fluorescence signal was observed in the cytoplasm and the perinuclear region in cells expressing V2-YFP [60]. However, the V2-SlGRXC6 complex was observed in the nucleus and cytoplasm (Fig 1B), suggesting that SlGRXC6 may retain V2 in the nucleus. We then tested the effect of SlGRXC6 on the localization of V2. When FLAGtagged SlGRXC6 (FLAG-SlGRXC6) and V2-YFP were coexpressed in N. benthamiana cells, strikingly, a very strong V2-YFP fluorescence signal was found in the nucleus in almost all the cells (n = 30) (Fig 5A), showing that SlGRXC6 can enhance nuclear accumulation of V2. We also coexpressed V2-RFP with SlGRXC6-YFP and found that V2-RFP was present in the nucleus and the cytoplasm (n = 20) (S5 Fig). These results suggest that SlGRXC6 inhibits the nuclear export of V2, increasing its accumulation in the nucleus.
To complement our visual observations with a biochemical assay, we performed a fractionation assay to separate total lysate into its nucleus and cytoplasm fractions [61]. As shown in Fig 5B, we only detected the cytoplasmic marker phosphoenolpyruvate carboxylase (PEPC) in the cytoplasmic fraction, and the nuclear marker H2B-RFP was only present in the nuclear fraction. Under such conditions, the amount of V2-YFP in the nuclear fraction significantly increased in the presence of SlGRXC6-FLAG, which is consistent with the results based on fluorescence microscopy (Figs 5A and S5). To provide numeric readings, we set the sum of the V2-YFP signal intensity in the cytoplasm and nucleus at 100%. In the presence of

Host GRXC6 restricts TYLCV infection
SlGRXC6-FLAG, we found 53% of V2-YFP in the nuclear fraction compared to 26% when V2-YFP was expressed alone (Fig 5B). This was in sharp contrast to H2B-RFP and PEPC whose localizations were not affected in the presence of SlGRX6-FLAG. According to our early results, V2 depends on exportin-α to exit from the nucleus [60]; thus, we concluded that SlGRXC6 is able to enrich V2 in the nucleus, possibly by preventing V2 from leaving the nucleus.

Host GRXC6 restricts TYLCV infection
We next tested whether the V2-SlGRXC6 interaction is required for SlGRXC6-mediated enrichment of V2 in the nucleus. When coexpressed with either SlGRXC6 T53A or SlGRXC6 C58A , V2-YFP was primarily found in the cytoplasm and the perinuclear region in most cells (n = 30), but not in the nucleus (Fig 5C), suggesting that T53 and C58 are critical for SlGRXC6 to re-localize V2. Because both mutants accumulated at a similar level to WT SlGRXC6 (Fig 1F), these results showed that SlGRXC6 inhibited the nuclear export of V2 through its interaction with V2.
We previously showed that V2 facilitates the nuclear export of TYLCV V1 protein [60]. As shown in Fig 5D and 5E, V1-YFP was found in the nucleus in all the cells (n = 30) when expressed alone but in only 17% cells (n = 30) in the presence of FLAG-V2. Given that SlGRXC6 enhanced the nuclear accumulation of V2 (Fig 5A and 5B), it is possible that SlGRXC6 may prevent the nuclear egress of V1 via the inhibited V2 export. Agreeing with this notion, V1-YFP was detected in the nucleus in 86% of cells (n = 30) when coexpressed with FLAG-V2 and SlGRXC6-FLAG (Fig 5D and 5E). We also tested whether SlGRXC6 executed its functions via its interaction with V2 by using mutants SlGRXC6 T53A and SlGRXC6 C58A . When co-expressed with FLAG-V2 and SlGRXC6 T53A -or SlGRXC6 C58A -FLAG, only 24-33% of cells had V1-YFP in the nucleus (Fig 5D and 5E), suggesting that the SlGRXC6-V2 interaction is necessary to inhibit the V2-mediated nuclear egress of V1.
Based on earlier reports that V1 is involved in binding and nuclear shuttling of viral genomic DNA [22,62], we hypothesized that SlGRXC6 may restrict TYLCV infection by inhibiting the nuclear export of the V1-viral genomic DNA complex. We measured accumulated TYLCV genomic DNA in the nucleus in TYLCV-inoculated SlGRXC6-overexpressed and -silenced tomato plants. We harvested emerging leaves from TYLCV-inoculated plants and extracted DNA from the total lysate or the nuclear fraction separately. As shown in S6 Fig, in the nuclear fraction we only detected proliferating cell nuclear antigen (PCNA) but not the cytochrome C oxidase subunit 1 (COX1, locus ATMG01360), which is the mitochondrial catalytic-core subunit of the respiratory chain complex and is encoded by mitochondrial DNA [63]. Under such conditions, viral DNA levels in the nuclear fraction and total lysates were detected by qPCR and the percentages of accumulated TYLCV genomic DNA in the nuclear fraction compared to that of total lysates were calculated. A significant difference in the nucleus-accumulated viral DNA was observed in SlGRXC6-overexpressed plants compared to that of control plants at 23 dpi: 14.4% of viral DNA accumulated in the nucleus in SlGRXC6-overexpressed plants compared to 4.85% in the control plants ( S7A Fig). These results showed that SlGRXC6 inhibited the viral DNA export from the nucleus during the period of high levels of DNA replication, likely by inhibiting V2 export. Conversely, as shown in S7B Fig, statistical analysis showed that the accumulated viral DNA in the nucleus was significantly reduced in SlGRXC6-silenced plants at 23 dpi. In addition, overexpression of SlGRXC6 T53A -or SlGRXC6 C58A failed to inhibit viral infection (Fig 4), showing that the SlGRXC6-V2 interaction is necessary for inhibiting viral infection. These results suggested that the V2-SlGRXC6 interaction is required for the inhibited nuclear egress of V2 and V1, as well as TYLCV infection.
We also tested whether SlGRXC6 restricted TYLCV by affecting gene expression of pathogenesis-related (PR) genes. As shown in S8 Fig, accumulated transcripts of PR1-a, GLUA and CHI3, three tomato PR genes [64], decreased by 42%-58% in TYLCV-inoculated, TRV-SlGRXC6 plants compared to those in the TYLCV-infected TRV plants (S8 Fig). Knocking down gene expression of SlGRX38 or SlGRX39 also decreased the accumulated PR gene transcripts similarly to that SlGRXC6 (S8 Fig), however, only the knockdown of SlGRXC6 inhibited TYLCV infection but not SlGRX38 and SlGRX39 (S4 Fig), suggesting that the enhanced nuclear egress of V2, but not the decreased expression of PR genes, is the major force contributing to the promoted TYLCV infection in the SlGRXC6-silenced tomato plants.

SlGRXC6 Mediates Plant Growth along with SlNTRC80
It is well-known that GRXs function together with TRX in plant development, stress responses, and host-mediated pathogen responses [45,47,48]. Given the fact that the levels of SlGRXC6 are related to plant growth (Figs 2B and 3B), we hypothesized that the stunting symptom induced by TYLCV infection may be related to SlGRXC6 directly, or indirectly via SlGRXC6-interacting TRX(s). Based on our analysis using the STRING program (https:// string-db.org/), 10 tomato TRXs were predicted to be associated with SlGRXC6 (S9A Fig). Among them, only SlTRX1-140 (Thioredoxin like 1-140) and SlNTRC80 (NADPH-dependent thioredoxin reductase C80) were found to interact with SlGRXC6 in the Y2H (Figs 6A  and S9B). SlGRXC6-SlNTRC80 and SlGRXC6-SlTRX1-140 interactions were further tested by the co-IP assay. However, only FLAG-SlNTRC80 was detected when SlGRXC6-YFP was pulled down using an anti-GFP polyclonal antibody (Fig 6B).

Host GRXC6 restricts TYLCV infection
Although it is well-known that NTRs play an important role in plant growth [51,54], the possible role of SlNTRC80 and its interaction with other GRXs is unknown. We knocked down the expression of SlNTRC80 using the TRV vector. A strong growth defect was observed in SlNTRC80-silenced plants, including yellowish young leaves and dwarf plants (Fig 6C, 6D and 6E). These results agreed with previous studies in Arabidopsis [51][52][53], suggesting that SlNTRC80's role in plant growth that might be conserved among multiple plant species.
We next examined the subcellular localization of SlNTRC80-YFP in H2B-RFP transgenic N. benthamiana cells. The fluorescence of SlNTRC80-YFP was localized in the cytoplasm and nucleus in all cells (n = 20) (S10 Fig). We next tested the effect of SlGRXC6 on the expression of SlNTRC80. An up-regulation of SlNTRC80 expression was found in tomato plants when SlGRXC6 was overexpressed using PVX as a vector (S11A Fig), and a decreased level of SlNTRC80 was found in TRV-SlGRXC6-infiltrated tomato plants (S11B Fig). The fact that the levels of SlGRXC6 correlated well with those of SlNTRC80 suggest that they may be coordinately involved in plant growth.
We also tested whether the SlGRXC6-SlNTRC80 interaction is necessary for growth regulation. We first tested whether SlGRXC6 T53A -and SlGRXC6 C58A -FLAG interacted with SlNTRC80-YFP using the co-IP assay. As shown in S12 Fig, only a very weak band of SlNTRC80-YFP was detected when FLAG-SlGRXC6 C58A and SlNTRC80 were coexpressed. No SlNTRC80-YFP was pulled down when they are co-expressed with SlGRXC6 T53A -FLAG, showing that SlGRXC6 T53A and SlGRXC6 C58A lost the ability to interact with SlNTRC80. In addition, SlGRXC6 T53A -or SlGRXC6 C58A -overexpressed tomato plants were not significantly different from the control plants (Fig 4A and 4B), suggesting that SlGRXC6 and SlNTRC80 may work together to regulate the growth of tomato plants.

V2 Competes with SlNTRC80 for Direct Binding to SlGRXC6
We next tested whether V2 interacted with SlNTRC80. As shown in Fig 6F, no interaction between SlNTRC80 and V2 was observed in Y2H. Given that SlGRXC6 functioned as a restriction factor (Figs 2 and 3) and TYLCV infection promoted the expression of SlGRXC6 (S13 Fig), we hypothesized that SlGRXC6 and SlNTRC80 may function as a complex in regulating plant growth but that SlGRXC6 may be recruited away or sequestered by V2 during TYLCV infection. To test the possibility that V2 may compete with SlNTRC80 for binding with SlGRXC6, we performed a competitive pulldown assay using glutathione S-transferase (GST)tagged SlGRXC6 (GST-SlGRXC6), maltose binding protein (MBP)-tagged SlNTRC80 (MBP-SlNTRC80), and His6-tagged V2 (His6-V2) that were purified from E. coli. Different amounts of His6-V2 or His6 were mixed with the same amount (2 μg) of GST-SlGRXC6 and incubated with GST-beads for 1 hour. Then, 2 μg MBP-SlNTRC80 were added and incubated with the beads. After extensive washing, GST-beads were collected and proteins that were pulled down with the beads were detected using anti-His6, -MBP, or -GST antibody. We found that the amount of MBP-SlNTRC80 pulled down by GST-SlGRXC6 was reduced when increasing amounts of V2 were added to the GST-SlGRXC6 and the MBP-SlNTRC80 mixture, but not by His6 (Fig 6G), suggesting that V2 has a higher affinity to SlGRXC6 than SlNTRC80 and possibly recruits SlGRXC6 away from SlNTRC80.

Discussion
Geminiviruses infect a broad spectrum of plants and induce a wide range of symptoms. Some early reports showed that V2 of geminiviruses is an important symptom determinant [6,12-14, 16,19,29]. TYLCV is a major tomato pathogen worldwide that causes extensive tomato losses [65]. We found that TYLCV V2 caused yellowing and curly leaves, as well as dwarfness

PLOS PATHOGENS
Host GRXC6 restricts TYLCV infection when expressed using a PVX vector in N. benthamiana [29] or expressed stably in transgenic tomato plants (S1 Fig), indicating that V2 plays a very important role in symptom development during TYLCV systemic infection. TYLCV V2 is a multifunctional protein and executes its functions along with many of its host partners. V2 inhibits TGS by interacting with host histone deacetylase 6 (HDA6), resulting in the decreased methylation of viral DNA [20]. V2 interacts with the cellular suppressor of gene silencing 3 (SGS3), which is required in the RNA silencing pathway, to suppress host RNA silencing [18]. V2 also interacts with CYP1, a tomato papain-like cysteine protease that is involved in plant defence against diverse pathogens [21]. To identify the host factors that might be involved in viral infection and symptom development along with V2, we identified SlGRXC6, out of 55 available GRXs in tomato, as a specific V2-interacting protein. We further demonstrated that SlGRXC6 possibly plays two roles in TYLCV infection: SlGRXC6 functions as a restriction factor of TYLCV that prevents V2 from moving out of the nucleus (Fig 5A and 5B), and in turn, the nuclear export of V1 (Fig 5D and  5E) and viral genomic DNA (S7 Fig). SlGRXC6 also contributes to symptom development (Figs 2 and 3) via its interaction with SlNTRC80 (Fig 6A and 6B), where V2 sequesters SlGRXC6 away from forming the SlGRXC6-SlNTRC80 complex (Fig 6G), and thus, inhibits plant growth (Fig 4). Our work, therefore, identified a new host partner of V2, and revealed the mechanisms whereby V2 functions as a pathogenicity determinant and can be targeted for antiviral defense.
For the role of GRXs in plant defense response, most reports focus on the enhanced generation of reactive oxygen species (ROS) that mediates plant defense gene activation and/or conjuncts with other plant signaling molecules, and thus, regulating pathogen infection, such as fungi and bacteria [47,[66][67][68]. However, little has been reported on the interaction between plant viruses and host GRXs. We demonstrated that V2-YFP was localized in the cytoplasm and perinuclear region, but was highly enriched in the nucleus in the presence of SlGRXC6-FLAG in N. benthamiana cells (Fig 5A), suggesting a possible role of SlGRXC6 in restricting V2 from exporting out of the nucleus. Agreeing with prior reports that V2 facilitates the nuclear export of V1 [60] and V1 is involved in the nuclear shuttling of viral genomic DNA [22,62], we showed that V1 was also enriched in the nucleus in the presence of SlGRXC6 and V2 (Fig 5D and 5E) and the accumulated viral genomic DNA in the nucleus increased in the presence of overexpressed SlGRXC6 but decreased when the SlGRXC6 expression was knocked down (S7 Fig). Our data also demonstrated that SlGRXC6 executed its functions via its interaction with V2 based on two mutations in SlGRXC6, T53A and C58A, that disrupted the V2-SlGRXC6 interaction (Fig 1C and 1E). When coexpressed with SlGRXC6 T53A -or SlGRXC6 C58A -FLAG, V2-YFP localized in the perinuclear region and the cytoplasm ( Fig 5C); when coexpressed with FLAG-V2 and SlGRXC6 T53A -or SlGRXC6 C58A -FLAG, only 24-33% of V1-YFP in the nucleus (Fig 5D and 5E), compared to 86% of cells with V1-YFP in the nucleus when co-expressed with FLAG-V2 and WT SlGRXC6, suggesting that the SlGRXC6-V2 interaction is necessary to inhibit the V2-mediated nuclear export of V1. In addition, we found that overexpression of SlGRXC6 T53A -or SlGRXC6 C58A failed to inhibit viral infection (Fig 4), showing that the SlGRXC6-V2 interaction is necessary for inhibiting viral infection. These results suggested that SlGRXC6 functions as a host restriction factor for TYLCV infection by regulating the viral nuclear export process: SlGRXC6 interacts and inhibits the nuclear export of V2, and thus, the V2-mediated nuclear export of V1 and viral genomic DNA, and therefore TYLCV systemic infection.
In plants, GRXs and TRXs play key roles in the maintenance of cellular redox homeostasis, development, and responses to biotic or abiotic stresses [34,36,55,69]. In addition, the glutathione pathway compensates for the reduced NTR activity, suggesting an interplay between the TRX and GRX pathways [55,69]. It was reported that in transgenic Arabidopsis, the overexpression of NTR enlarged plants, but the lack of NTR led to smaller plants [51][52][53]. However, it is unknown whether the functions of NTR in regulating plant growth is conserved among different plant species. Our data showed that tomato plants were taller than control plants in the presence of overexpressed SlGRXC6 (Fig 2B) but became shorter when SlGRXC6 was down-regulated (Fig 3B), suggesting that GRX(s) have a similar function in plant growth. In addition, plants with silenced SlNTRC80 were smaller than control plants (Fig 6C), indicating that NTR from different plant species share similar functions in regulating plant growth. Our data support the notion that SlGRXC6 and SlNTRC80 work coordinately in regulating plant growth because SlGRXC6 interacted with SlNTRC80 (Fig 6A and 6B) and an increased level of SlNTRC80 was associated with the overexpressed SlGRXC6 (S11 Fig). Further supporting this notion, the overexpression of SlGRXC6 T53A or SlGRXC6 C58A failed to stimulate plant growth (Fig 4) and this failed stimulation was most likely related to the disrupted SlGRXC6-SlNTRC80 interaction (S12 Fig).
Several reports have shown that GRXs play an important role in host defense response [34,37]. We found that SlGRXC6 interacted with both V2, to restrict viral infection and viruscaused disease symptoms, and SlNTRC80, to regulate plant growth. Our data suggested that V2 and SlNTRC80 might bind to the same site in SlGRXC6 because in an in vitro competitive pull-down assay (Fig 6G), an increasing amount of V2 competed off SlNTRC80, suggesting V2 may have a higher affinity to SlGRXC6 than to SlNTRC80. Furthermore, SlGRXC6 T53A and SlGRXC6 C58A are defective in interacting with V2 (Fig 1C and 1E) and also SlNTRC80 (S12 Fig). Both mutants accumulated to the same levels to WT, suggesting that both were wellexpressed and stable, even though we could not totally rule out that each mutation cause other defects in SlGRXC6.
Based on our work and results from others, we propose a working model underlying a possible mechanism by which SlGRXC6 restricts viral infection and contributes to symptom development (Fig 7). SlGRXC6 restricts TYLCV infection by preventing V2 from moving out of the nucleus, and thus, inhibiting V2-mediated nuclear export of V1 and V1-viral DNA complexes. TYLCV V2 is an important symptom determinant that appears to sequester SlGRXC6 from its association with SlNTRC80, causing dwarf symptoms in plant. Our work, therefore, extended and complemented the current understanding of V2's role as a symptom determinant and as a target for host defense, among other important roles in geminivirus infection. Given the fact that transient overexpression of SlGRXC6 promoted plant growth, inhibited viral infection, and delayed symptom development, our work also points out a potential antiviral strategy by overexpressing SlGRXC6 in tomato plants.

Plant materials and growth conditions
All agroinfiltration experiments were performed in wild-type (WT) or H2B-RFP (red fluorescent protein fused to the C terminus of histone 2B) [70] transgenic N. benthamiana. Plants were grown in a growth chamber with temperatures at 26˚C (16 h, light) and 22˚C (8 h, dark) for 4-6 weeks before being infiltrated with agrobacterium cultures. After infiltration, the plants were kept under the same growth conditions.
To make BiFC vectors, SlGRXC6 ORF was cloned into the BamHI site as a fusion with the N-terminal fragment of YFP and V2 ORF was cloned into the BamHI site as a fusion with the C-terminal fragment of YFP, resulting in nYFP-SlGRXC6 and cYFP-V2.
FLAG tagged V2 and SlGRXC6 ORFs were inserted into the BamHI site between the 35S promoter and the 35S terminator in the pCambia1300 binary vector to generate FLAG-V2 and SlGRXC6-FLAG for co-IP experiments.
To generate vectors overexpressing SlGRXC6, SlGRXC6 ORF (ClaI/SalI) was inserted into the PVX vector. To obtain a construct for virus-induced gene silencing (VIGS) assays, partial coding sequences of SlGRXC6 (encoding amino acids 13 to 113 of SlGRXC6, which is not conserved in the GRX family), or SlNTRC80 (amino acids 150 to 244, which is not conserved in

PLOS PATHOGENS
Host GRXC6 restricts TYLCV infection the TRX family) were cloned into the EcoRI-XhoI sites of a TRV vector [20]. Primers that were used in plasmid construction are shown in S1 Table. Agroinfection assays in N. benthamiana plants Target vectors were transformed into A. tumefaciens strain GV3101 by electroporation. Agrobacterial cultures were harvested when the optical density measured at a wavelength of 600 nm (OD 600 ) reached approximately 0.8-1.0, then collected by centrifugation, resuspended in induction buffer (10 mM MgSO 4 , 100 mM 2-(N-morpholino) ethanesulfonic acid [pH 5.7], 2 mM acetosyringone), and incubated for more than 2 hours at room temperature. The suspensions were then adjusted to OD 600 = 0.5 and infiltrated into leaves of 4-to 6-week-old WT or H2B-RFP transgenic N. benthamiana plants.

Co-Immunoprecipitation
The co-IP assay was performed as previously described [29]. Briefly, infiltrated N. benthamiana leaves were harvested at 40 hpai. Proteins were extracted and incubated with FLAG-conjugated beads (Sigma, USA) and resuspended in 2×SDS loading buffer. Target proteins were detected using a polyclonal anti-GFP antibody (GenScript, USA) or a monoclonal anti-FLAG (Sigma, USA) antibody.
Yeast two-hybrid screen RNA was extracted from TYLCV-infected tomato leaves at different developmental stages to construct a yeast two-hybrid library (Invitrogen, China). TYLCV V2 cloned into bait Gal4-BD vector pGBKT7 was transformed into the AH109 yeast strain and used to screen the tomato cDNA library. Mating between bait and prey yeast cells was performed on selective media (SD/-Leu/-Trp/-His) for 3-5 days at 30˚C. After the positive clones were cultured, they were transferred to SD/-Leu/-Trp/-His/-Ade with X-α-Gal. The positive clone plasmids were subsequently extracted using the Plasmid Miniprep kit (CoWin Biosciences, China). Following sequencing, a BLAST search was conducted using GenBank (National Center for Biotechnology Information) to determine the associated genes.
The Y2H system was used to examine interactions. SlNTRC80, SlGRXC6 and V2 were cloned into the activation domain (AD)-containing vector or the vector harboring the DNA binding domain (BD). Y2H assays were performed as described previously [60]. Briefly, we transformed constructs into yeast cells and grew them on synthetic defined medium at 30˚C for 72 h to test protein-protein interactions.

Nuclear-cytoplasmic fractionation assay
Nuclear-cytoplasmic fractionation assays were performed as described previously [60]. Briefly, we harvested infiltrated leaves at 40 hpai and separated the nuclear and cytoplasmic fractions. Target proteins were detected using the indicated antibody. Membranes were incubated in Supersignal West Femto substrate (Thermo Scientific) and the protein signals were detected using an Azure C400 ChemiDoc imager. The intensity of signals was quantified by using Ima-geQuant TL (GE healthcare). Signal areas were automatically identified and the pixel values were measured, and normalized to that of the background signal to obtain the signal intensity. The sum of the protein signal intensity in the cytoplasm and the nucleus was set at 100%. Values represent the average of three plants. Experiments were repeated three times.
PEPC protein and H2B-RFP were used as quality controls for fractionation assays as a cytoplasmic and a nuclear marker, respectively.

Agrobacterium-mediated TYLCV inoculation
TYLCV infectious clone was constructed as previously reported [29]. The agrobacteria culture was injected into tomato stems with a syringe. The inoculated plants were grown in an insectfree cabinet with supplementary lighting corresponding to a 16-h day length.

Quantitative PCR analysis
Total RNA was extracted from SlGRXC6-scilenced-, SlNTRC80-scilenced-, or SlGRXC6-overexpressed tomato leaves at different time points using Trizol Reagents (Life Technologies, USA). RNA samples were treated with DNase I and converted to cDNA following manufacturer's instructions (PrimeScript RT reagent Kit with gDNA Erase, Takara, Japan).
To quantify viral DNA levels by qPCR, total DNA was extracted from mock-or TYLCVinfiltrated tomato leaves at different time points and TYLCV V1 was detected to represent the viral genomic DNA. The qRCR reaction mixes consisted of 6 μl of SYBR Green supermix (Bio-Rad, USA), 0.10 μl of each primer (10 pmol) and 1.5 μl of DNA or cDNA (10 ng/μl) in a total volume of 12 μl. SlActin was used as an internal control for tomato. Each experiment was performed in triplicate and repeated three times. PCR reactions were done in an Applied Biosystems 7500 (ThermoFisher Scietific, USA) real-time PCR detection system. Data analysis was performed using Applied Biosystems 7500 software version 2.0.6.

Nuclear DNA isolation
Nuclear DNA isolation assays were performed as described previously [72] with minor modifications. Fresh leaves were harvested and mixed with homogenization buffer (1 M hexylene glycol, 10 mM Tris [pH 7.5], 10 mM MgCl 2 , 0.5% triton X-100, and 5 mM β-mercaptoethanol) on ice. The homogenate was transferred to a 50 mL tube and centrifuged at 1,800g for 20 min at 4˚C. The pellet was then suspended gently in 10 mL of nuclei wash buffer (0.5 M hexylene glycol, 10 mM Tris [pH 7.5], 10 mM MgCl 2 , 0.5% triton X-100, and 5 mM β-mercaptoethanol) and centrifuged for another 10 minutes at 4˚C. The pellet was resuspended with 10 mL of extraction buffer (0.35 M sorbitol, 0.1 M Tris [pH 7.5], 5 mM EDTA, 2% CTAB, 4 M NaCl, 0.4% SDS and 0.1% β-mercaptoethanol) with 10 μg RNase H to remove RNA. The mixture was incubated at 65˚C for 30 min and then extracted with chloroform-isoamytalcohol (24:1). After 10 min of centrifugation, the upper phase was mixed with isopropanol and incubate at -20˚C for 30 min to precipitate DNA.