The C4 protein encoded by tomato leaf curl Yunnan virus reverses transcriptional gene silencing by interacting with NbDRM2 and impairing its DNA-binding ability

In plants, cytosine DNA methylation is an efficient defense mechanism against geminiviruses, since methylation of the viral genome results in transcriptional gene silencing (TGS). As a counter-defense mechanism, geminiviruses encode viral proteins to suppress viral DNA methylation and TGS. However, the molecular mechanisms by which viral proteins contribute to TGS suppression remain incompletely understood. In this study, we found that the C4 protein encoded by tomato leaf curl Yunnan virus (TLCYnV) suppresses methylation of the viral genome through interacting with and impairing the DNA-binding ability of NbDRM2, a pivotal DNA methyltransferase in the methyl cycle. We show that NbDRM2 catalyzes the addition of methyl groups on specific cytosine sites of the viral genome, hence playing an important role in anti-viral defense. Underscoring the relevance of the C4-mediated suppression of NbDRM2 activity, plants infected by TLCYnV producing C4(S43A), a point mutant version of C4 unable to interact with NbDRM2, display milder symptoms and lower virus accumulation, concomitant with enhanced viral DNA methylation, than plants infected by wild-type TLCYnV. Expression of TLCYnV C4, but not of the NbDRM2-interaction compromised C4(S43A) mutant, in 16c-TGS Nicotiana benthamiana plants results in the recovery of GFP, a proxy for suppression of TGS. This study provides new insights into the molecular mechanisms by which geminiviruses suppress TGS, and uncovers a new viral strategy based on the inactivation of the methyltransferase NbDRM2.

machinery to replicate the viral DNA genome, DRM-related viral DNA methylation establishment and maintenance may exert important functions in TGS of the viral genome.
TGS plays an important role in the host defense against geminiviruses, since methylation of the viral genome impairs the transcription of viral genes [24,36]. As a counter-defense strategy, geminiviruses encode proteins that serve as suppressors of TGS, hence promoting viral replication and spread. Suppressors of TGS encoded by geminiviruses exert their functions through divergent mechanisms. Well-known geminiviral TGS suppressors are the AC2 or AL2 protein encoded by some bipartite begomoviruses, such as cabbage leaf curl virus (CaLCuV) and TGMV, and the C2 or L2 protein encoded by some curtoviruses, such as beet curly top virus (BCTV) and beet severe curly top virus (BSCTV). Both AC2/AL2 and C2/L2 suppress TGS by interfering with the methyl cycle through inactivation of adenosine kinase (ADK) [37][38][39]. Additionally, BSCTV C2 attenuates 26S proteasomal degradation of S-adenosyl-methionine decarboxylase (SAMDC1) [40]. TGMV-encoded AC2 was shown to interact with the H3K9 histone methyltransferase SUVH4/KYP to suppress its activity to attenuate TGS [41]. The βC1 protein encoded by the TYLCCNV-associated betasatellite blocks the methyl cycle through the interaction with the key enzyme S-adenosyl homocysteine hydrolase (SAHH) [42]. Furthermore, the replication-associated proteins (Rep, also known as C1, AC1, or AL1) of several geminiviruses were also reported to act as TGS suppressors, since they reduce the expression of the maintenance methyltransferases MET1 and CMT3 [36]. The V2 protein encoded by cotton leaf curl Multan virus (CLCuMuV) acts as a TGS suppressor through its interaction with AGO4 [43]. Finally, the C4 or AC4 protein encoded by some geminiviruses has also been described to act as TGS suppressor [21,44]. The C4 protein from CLCuMuV interacts with S-adenosyl methionine synthetase (SAMS) and inhibits its enzymatic activity [44]; whether this is the mechanism underpinning the TGS suppression mediated by other geminivirus-encoded C4 proteins remains to be determined.
TLCYnV is a recombinant virus with TYLCCNV as major parent and pepper yellow leaf curl China virus (PepYLCCNV) as the donor of the C4 gene and a partial intergenic region (IR). The C4 protein encoded by TLCYnV is not only a pathogenicity determinant but also suppresses methylation-mediated TGS [21][22][23]. In this study, we report that TLCYnV C4 interacts with NbDRM2, a major plant DNA methyltransferase. TLCYnV C4 impairs the DNA-binding ability of NbDRM2 through direct protein-protein interaction. Silencing NbDRM2 suppresses methylation of the TLCYnV genome and enhances the viral infection in Nicotiana benthamiana plants. Our results provide additional evidence of the requirement of a functional methyl cycle for TGS-based anti-geminiviral defense in plants, and demonstrate that TLCYnV C4 suppresses TGS through a novel mechanism that involves the inhibition of the DNA-binding ability of NbDRM2.

TLCYnV C4 interacts with NbDRM2
To elucidate the molecular mechanism by which TLCYnV C4 suppresses TGS, we screened a N. benthamiana cDNA library by using yeast two-hybrid (Y2H) to identify C4-interacting proteins. Among the C4-interacting factors, one cDNA encoding a DNA methyltransferase was recovered. Interestingly, this protein is a novel interactor of geminivirus C4. We named this gene NbDRM2, since its sequence shares high similarity with that of NtDRM2 [45]. We validated the interaction between TLCYnV C4 and NbDRM2 in Y2H and bimolecular fluorescence complementation (BiFC) assays (Fig 1A and 1B). NbDRM2 has orthologues in solanaceous plant species (NaDRM2, NtDRM2, StDRM2, and CaDRM2) (S1A Fig). A computational analysis by using SMART (SMART, http://smart.embl-reidelbery.de) predicted NbDRM2 to harbor a conserved catalytic DNA-cytosine methyltransferase domain (483-597 aa) (S1B Fig), which shares high similarity with that of orthologues (S1C Fig). To identify whether the TLCYnV C4/NbDRM2 interaction identified above has biological relevance, we silenced NbDRM2 in N. benthamiana by using virus-induced gene silencing (VIGS) with a tobacco rattle virus (TRV)-based vector. Real-time quantitative reverse transcription PCR (qRT-PCR) analysis revealed that the transcripts of NbDRM2 were significantly reduced (over 50%) in NbDRM2-silenced N. benthamiana plants when compared with those in N. benthamiana plants inoculated with the control (TRV-GFP) (S2 Fig). Interestingly, NbDRM2-silenced N. benthamiana plants showed a late-flowering phenotype compared with mock (TRV-GFP) N. benthamiana plants, and phenocopied 35S::TLCYnV C4 transgenic N. benthamiana plants (Fig 1C and 1D). These results suggest that TLCYnV C4 might interfere with the physiological functions of NbDRM2 through protein-protein interaction.

NbDRM2-mediated methylation of the viral genome plays an important role in the plant defense against TLCYnV
To elucidate whether NbDRM2 has biological relevance in the virus infection, we inoculated NbDRM2-silenced (TRV-NbDRM2) and mock (TRV-GFP) N. benthamiana plants with TLCYnV. The symptoms of the viral infection in NbDRM2-silenced plants were more severe than those of mock plants, and the mean latent period appeared reduced (Fig 2A and S3 Fig). As expected, Southern blot analysis showed that TLCYnV accumulation is higher in NbDRM2-silenced N. benthamiana plants than in mock (TRV-GFP) plants ( Fig 2B). To test whether NbDRM2 plays an important role in the methylation of the TLCYnV genome, bisulfite sequencing was performed to detect the methylation of the viral intergenic region (IR) at

PLOS PATHOGENS
TLCYnV C4 reverses TGS by interacting with NbDRM2 high resolution. The TLCYnV IR contains 18 CG, 6 CHG, and 56 CHH sites. The bisulfite sequencing results indicate that the methylation level of the TLCYnV IR is significantly lower in NbDRM2-silenced N. benthamiana plants than in mock plants (Fig 2C and 2D). To confirm the contribution of NbDRM2 to the methylation of the TLCYnV genome, methylation-sensitive PCR was performed to examine the methylation status of the viral DNA in TLCYnVinfected mock and NbDRM2-silenced N. benthamiana plants. HpaI, and MspI are methylation-sensitive endonucleases whose cleavage activities are blocked by methylation of cytosine in their target sites; McrBc is a methylation-dependent endonuclease that preferentially digests methylated DNA. Total genomic DNA was extracted and digested with those three enzymes; the PCR product was significantly decreased in HpaII-and MspI-digested samples from TLCYnV-infected NbDRM2-silenced (TRV-NbDRM2) N. benthamiana plants, compared to samples from TLCYnV-infected mock (TRV-GFP) plants, while the PCR products were significantly increased in McrBc-digested samples (Fig 2E). These results suggest that NbDRM2 plays a critical role in the methylation of the TLCYnV genome.

The interaction between TLCYnV C4 and NbDRM2 alters the nuclear distribution pattern of NbDRM2
To investigate the subcellular localization of NbDRM2, a construct to express NbDRM2-GFP (where the GFP protein is fused to the C-terminus of NbDRM2) from the cauliflower mosaic virus (CaMV) 35S promoter was infiltrated into H2B-RFP transgenic N. benthamiana plants. Confocal imaging showed that NbDRM2-GFP localizes in the nucleus and forms speckles ( Fig  3A). This specific nuclear localization of NbDRM2-GFP raised the possibility that NbDRM2 might bind chromosomal DNA. To further examine the localization of NbDRM2-GFP in the nucleus, we isolated intact nuclei and observed them under the confocal microscope after staining them with propidium iodide (PI). Images show that NbDRM2-GFP co-localized with the PI signal, especially in the observed speckles (S4A Fig); GFP, on the contrary, distributes evenly in the nucleus and does not form speckles (S4A Fig). These results suggest that NbDRM2 indeed binds to the plant chromosomal DNA. To investigate the influence of TLCYnV C4 on the nuclear distribution pattern of NbDRM2, we co-expressed NbDRM2-GFP with CFP or TLCYnV C4-CFP in N. benthamiana. GFP and CFP fluorescence was observed by confocal microscopy at 48 hours post-infiltration (hpi). Strikingly, the nuclear NbDRM2-GFP-containing speckles disappeared in the presence of TLCYnV C4 (Fig 3B). We also isolated intact nuclei from wild type (WT) or 35S::TLCYnV C4 transgenic N. benthamiana plant leaves transiently expressing NbDRM2-GFP, and observed the nuclei after PI staining, confirming that the speckles formed by NbDRM2-GFP disappeared in the presence of C4 (S4B Fig).  To test whether it is the interaction between TLCYnV C4 and NbDRM2 that alters the nuclear distribution pattern of NbDRM2, we aimed to identify mutations rendering C4 unable to bind NbDRM2. For this purpose, we used C4 mutants constructed in previous studies [22,23,46] or replaced the amino acids harboring hydroxyl groups (such as Ser or Thr) with alanine and conducted Y2H assays with NbDRM2. Our results identified Ser43 in TLCYnV C4 as required for the TLCYnV C4/NbDRM2 interaction ( Fig 3C and  S5 Fig). To further confirm that Ser43 of TLCYnV C4 is critical for TLCYnV C4/NbDRM2 interaction, we transiently co-expressed Flag-NbDRM2 with GFP, TLCYnV C4-GFP, or TLCYnV C4(S43A)-GFP in leaves of N. benthamiana plants and performed co-immunoprecipitation (Co-IP) assays. The Co-IP results show that the C4(S43A) mutant loses the ability to interact with NbDRM2 (S6 Fig). These results suggest that Ser43 is indeed a key site for the TLCYnV C4/NbDRM2 interaction. Interestingly, the TLCYnV C4(S43A) mutant could not

PLOS PATHOGENS
TLCYnV C4 reverses TGS by interacting with NbDRM2 alter the nuclear distribution pattern of NbDRM2 in epidermal cells of N. benthamiana plants, in sharp contrast to WT TLCYnV C4 (Fig 3D), indicating that the interaction between TLCYnV C4 and NbDRM2 is required for C4 to alter the nuclear distribution pattern of NbDRM2. To further confirm the subcellular localization of the TLCYnV C4/NbDRM2 interaction, we performed BiFC assays in the epidermal cells of H2B-RFP transgenic N. benthamiana plants. Micrographs show that the TLCYnV C4/NbDRM2 complex localizes in the nucleus and does not form speckles. These results suggest that TLCYnV C4 interferes with the DNA-binding ability of NbDRM2 to inhibit the formation of NbDRM2/DNA complex ( Fig  3E). Previous studies reported that the TLCYnV C4/NbSKη interaction is critical for C4 to

PLOS PATHOGENS
TLCYnV C4 reverses TGS by interacting with NbDRM2 cause symptom-like developmental abnormalities [22,23]. We evaluated the ability of the NbDRM2-interaction-compromised C4 mutant version to cause these symptom-like abnormalities by using a cucumber mosaic virus (CMV)-based expression vector. Mutation in Ser43 did not affect the ability of C4 to cause developmental abnormalities, ruling out a role of the TLCYnV C4/NbDRM2 in this process (S7 Fig). In agreement with this, and as expected, Ser43 of TLCYnV C4 was not a key site for the TLCYnV C4/NbSKη interaction (S8 Fig). Taken together, these results indicate that the interaction between TLCYnV C4 and NbDRM2 influences the nuclear distribution pattern of NbDRM2 but does not affect the symptom determinant activity of TLCYnV C4.

TLCYnV C4 impacts the DNA-binding ability of NbDRM2
The altered nuclear distribution pattern of NbDRM2 caused by the TLCYnV C4/NbDRM2 association raised the possibility that TLCYnV C4 impacts the NbDRM2 DNA-binding ability through direct interaction. To test this hypothesis, chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR) was performed. For this purpose, we transiently co-expressed Flag-NbDRM2 with CFP, TLCYnV C4-CFP, and C4(S43A)-CFP in systemic leaves of TLCYnVinfected N. benthamiana plants, then detected the amount of TLCYnV genomic DNA bound by NbDRM2. Interestingly, we found that the amount of TLCYnV associated to NbDRM2 was significantly decreased in the presence of TLCYnV C4, but not of the NbDRM2-interactioncompromised C4 mutant (Fig 4A and S9 Fig). To further validate that TLCYnV C4 influences the DNA-binding ability of NbDRM2, we also examined the distribution of NbDRM2 in different nuclear fractions. The transcription machinery known to be tightly associated with chromatin appears in the insoluble fraction during nuclei preparation [47]. We found that NbDRM2 was mostly detected in the insoluble fraction, with a smaller amount of protein detectable in the soluble fraction ( Fig 4B). However, the amount of NbDRM2 in the insoluble fraction diminished in the presence of TLCYnV C4, with most NbDRM2 now appearing in the soluble fraction. Consistent with the histone H3 chromatin-bound nature, most of histone H3 was detected in the insoluble fraction ( Fig 4B). Next, we performed a competitive

PLOS PATHOGENS
TLCYnV C4 reverses TGS by interacting with NbDRM2 electrophoretic mobility shift assay (EMSA) to identify the effect of the TLCYnV C4/NbDRM2 interaction on the DNA-binding ability of NbDRM2 in vitro. With this aim, we expressed GST-NbDRM2, GST-TLCYnV C4, GST-TLCYnVC4(S43A), and GST alone in Escherichia coli and purified them, and amplified the TLCYnV IR DNA using Alexa Fluor 680-labeled oligonucleotides. The mixtures containing GST-NbDRM2 and A860-labeled TLCYnV IR were combined with different volumes of GST Tag, GST-C4, or GST-C4(S43A). As shown in Fig  4C, more free TLCYnV IR could be detected in the mixture containing GST-TLCYnV C4 ( Fig  4C). As expected, GST Tag and GST-TLCYnV C4(S43A) did not influence the amount of NbDRM2-bound TLCYnV IR DNA. These results suggest that the interaction between TLCYnV C4 and NbDRM2 indeed interferes with the DNA-binding ability of NbDRM2 and inhibits the formation of the NbDRM2/DNA complex.

TLCYnV C4 decreases the methylation level of the viral genome through interacting with NbDRM2
To identify whether the interaction between TLCYnV C4 and NbDRM2 has biological relevance, we constructed TLCYnV infectious clones expressing WT C4, the NbSKη-interaction compromised C4 mutant [C4(T35A)], or the NbDRM2-interaction compromised C4 mutant [C4(S43A)]. Consistent with our previous results, TLCYnV expressing C4(T35A) did not infect N. benthamiana plants systemically [22], while TLCYnV expressing the C4(S43A) induced milder symptoms compared to WT TLCYnV (Fig 5A), which correlated with lower viral DNA accumulation (Fig 5B). We then performed methylation-sensitive PCR to detect the methylation level of TLCYnV and TLCYnV harboring the C4(S43A) mutant. As expected, the methylation level of the TLCYnV mutant was significantly higher than that of WT TLCYnV (Fig 5C). These results were confirmed by bisulfite sequencing (Fig 5D and 5E), demonstrating that TLCYnV C4 decreases methylation of the viral genome through the TLCYnV C4/ NbDRM2 interaction. To assess the influence of this interaction on the methylation of endogenous plant DNA, we expressed TLCYnV C4, C4(T35A) and C4(S43A) mutants in N. tabacum by using a CMV-based expression vector. Consistent with previous results, expression of TLCYnV C4 (T35A) did not induce severe viral symptoms (S10A and S10B Fig). NtGRS1.3, a highly methylated repetitive DNA sequence in N. tabacum [48], was selected as a proxy to detect the influence of the TLCYnV C4/DRM2 interaction on endogenous gene methylation. We extracted total DNA and detected the methylation level of NtGRS1.3 by using a methylation-sensitive restriction enzyme (MspI). Southern blot analysis showed that more unmethylated NtGRS1.3 could be detected in the presence of TLCYnV C4 and C4(T35A), but not of C4 (S43A) (S10C Fig). To further confirm the above results, we performed methylation-sensitive PCR to detect the methylation level of NtGRS1.3; the results show that the TLCYnV C4/DRM2 interaction indeed decreases the methylation level of NtGRS1.3 in N. tabacum plants (S10D Fig). Taken together, these results suggest that TLCYnV C4 decreases the methylation level of both viral and host plant genomes through interacting with NbDRM2.

The interaction between TLCYnV C4 and NbDRM2 contributes to suppress TGS
Given that TLCYnV C4 inhibits NbDRM2 function through protein-protein interaction, we assumed that it might suppress the NbDRM2-mediated TGS. We expressed TLCYnV C4 and C4 mutants in 16c-TGS transgenic N. benthamiana plants, in which the GFP transgene driven by the CaMV 35S promoter is transcriptionally silenced, by using a potato virus X (PVX)based expression vector. 16c-TGS plants infected by PVX-TLCYnV C4 or PVX-TLCYnV C4 (S43A) showed abnormal development, including foliar distortion and longer internodes and

TLCYnV C4 reverses TGS by interacting with NbDRM2
NbDRM2 (Fig 6C and S11 Fig). When we inoculated 16c-TGS transgenic N. benthamiana plants with TLCYnV or the TLCYnV-C4(S43A) mutant, we found that infection by TLCYnV restores GFP expression at 12 dpi, but the TLCYnV mutant fails to do so (Fig 6D and 6E). In order to confirm an involvement of NbDRM2 in TGS of the GFP transgene in 16c-TGS plants, we silenced NbDRM2 in these plants by using a TRV-based vector, qRT-PCR analysis shows that the transcripts of NbDRM2 are significantly reduced (over 50%) in NbDRM2-silenced 16c-TGS transgenic N. benthamiana plants when compared with those in 16c-TGS transgenic N. benthamiana plants inoculated with the control (TRV-GUS) (S12 Fig). PVX-TLCYnV C4 was then inoculated in mock (TRV-GUS) and NbDRM2-silenced (TRV-NbDRM2) 16c-TGS transgenic N. benthamiana plants. As shown in Fig 6F and 6G, GFP expression was more strongly restored in NbDRM2-silenced 16c-TGS plants. These results suggest that the TLCYnV C4/NbDRM2 interaction plays a critical role in TGS suppression.

Discussion
Geminiviruses replicate their genome via dsDNA intermediates in the nucleus, which could be the target of methylation-mediated TGS [49]. In this study, we report that the C4 protein encoded by TLCYnV interacts with NbDRM2, a DNA methyltransferase required for the establishment and maintenance of DNA methylation, to inhibit its physiological function and break the TGS-mediated defense.
How does the interaction between TLCYnV C4 and NbDRM2 lead to suppressed TGS? Our results show that NbDRM2 localizes in the nucleus and forms speckles, which correlates with its DNA-binding capacity. However, TLCYnV C4 can interfere with the DNA-binding

PLOS PATHOGENS
TLCYnV C4 reverses TGS by interacting with NbDRM2 ability of NbDRM2, inhibiting the formation of DNA-NbDRM2 complexes and retaining NbDRM2 in the nucleoplasm. Inhibition of NbDRM2 DNA binding by TLCYnV C4 interferes with the NbDRM2-mediated DNA methylation. This model is supported by several pieces of evidence: (1) The nuclear speckles potentially containing DNA-bound NbDRM2 disappear in the presence of TLCYnV C4; (2) A higher proportion of soluble NbDRM2 can be detected in the presence of TLCYnV C4; (3) The TLCYnV infectious clone expressing the NbDRM2-interaction compromised C4 mutant has weaker pathogenicity and higher methylation levels of the viral genome; (4) TLCYnV C4, but not the mutant deficient in NbDRM2 interaction, decreases the methylation level of a DRM2 target locus. Taken together, these results strongly suggest that TLCYnV C4 impacts NbDRM2-mediated TGS through the interaction with this protein.
Although DRM2 plays an important role in maintenance of DNA methylation, only partial GFP could be restored in the systemic leaves of NbDRM2-silenced 16c-TGS transgenic N. benthamiana plants (Fig 6D). Strikingly, GFP expression was more strongly restored in NbDRM2-silenced 16c-TGS plants infected with PVX-TLCYnV C4 (Fig 6D). These results could be explained by at least two different possibilities: (1) Other proteins that function redundantly with NbDRM2 might exist and interact with TLCYnV C4; (2) The reduction of NbDRM2 in the silenced plants might be insufficient to restore GFP expression in 16c-TGS plants. Further efforts are necessary to explore the role of the methylation marks deposited by NbDRM2 in transcriptionally silenced GFP of 16c-TGS transgenic N. benthamiana plants.
Previous studies reported that TLCYnV C4 is the viral symptom determinant and induces symptoms through the TLCYnV C4/NbSKη interaction [21][22][23]. Interestingly, a residue (S43) essential for the TLCYnV C4/NbDRM2 interaction, does not affect the interaction between TLCYnV C4 and NbSKη. When we expressed the NbDRM2 interaction-compromised C4 mutant [C4(S43A)] in N. benthamiana by using a CMV-based expression vector and, the C4 mutant [C4(S43A)] produces developmental abnormalities similarly to WT C4 (S7 Fig); on the other hand, the TLCYnV C4 mutant [C4(T35A)] deficient in NbSKη interaction can suppress TGS in 16c-TGS transgenic N. benthamiana plants. Therefore, our results support the idea that the multifunctional C4 protein exerts its symptom determinant activity and its TGSsuppression ability by independent mechanisms, and that these two processes can be uncoupled.
Ser43 of TLCYnV C4 is critical for the interaction between TLCYnV C4 and NbDRM2. A C4 protein mutated in this residue [C4(S43A)] does not impact the DNA-binding ability of NbDRM2, as opposed to the wild-type C4 (Figs 4C and 6C). However, C4(S43A) could not completely inhibit the DNA-binding ability of NbDRM2, especially the viral DNA-binding ability (Fig 4A). These results suggest that another amino acid(s) might be involved in the TLCYnV C4/NbDRM2 interaction. Further efforts to investigate new site(s) important for the TLCYnV C4/NbDRM2 interaction will provide a more complete understanding of the molecular mechanism by which TLCYnV C4 impacts the DNA-binding ability of NbDRM2, and in turn TGS.
In this study, we show that TLCYnV C4 interferes with the DNA-binding ability of NbDRM2 to inhibit the NbDRM2-mediated viral genomic DNA methylation. We also found that the TLCYnV C4/NbDRM2 interaction decreased the methylation of one DRM2-targeted endogenous gene (NtGRS1.3) in N. tabacum (S10 Fig). This finding suggests that the interaction between TLCYnV C4 and DRM2 not only impacts TGS-mediated defense against geminiviruses, but has a general effect on DRM2 targets in plants; this idea is consistent with the effect of C4 on the nuclear distribution of NbDRM2 in the absence of the virus. Whether C4, through its interaction with NbDRM2, impacts genome-wide plant DNA methylation remains to be determined.

TLCYnV C4 reverses TGS by interacting with NbDRM2
Based on previous research and the work presented here, we propose a model to explain how TLCYnV C4 suppresses TGS. In the absence of TLCYnV C4, NbDRM2 binds to the viral DNA to silence gene expression; the presence of C4 negatively impacts the DNA-binding ability of NbDRM2 and inhibits the formation of the NbDRM2/DNA complex, consequently blocking the NbDRM2-mediated gene silencing (Fig 7). Strikingly, multiple geminiviral proteins encoded by different species have evolved to target the methyl cycle, in turn suppressing anti-viral gene silencing. For example, AC2/AL2 and C2/L2 proteins inactivate ADK and stabilize SAM decarboxylase (SAMDC), which causes decarboxylated SAM (dcSAM) levels to rise [37][38][39]40]; TYLCCNV βC1 protein inhibits SAHH [42]; the C4 protein encoded by CLCu-MuV interacts with and inhibits SAMS [44] (Fig 7). The convergent targeting of this pathway by geminiviruses underscores its importance in anti-geminiviral defence, and highlights its potentiality for the design of strategies to engineer resistance to the devastating crop diseases caused by these viruses. Its product, S-adenosyl homocysteine (SAH), inhibits transmethylation by competing with SAM for methyltransferases (MTases). SAH is converted to homocysteine (Hyc) and adenosine by S-adenosyl homocysteine hydrolase (SAHH). The phosphorylation of adenosine by adenosine kinase (ADK) is critical because the SAHH-catalyzed reaction is reversible and the equilibrium lies in the direction of SAM synthesis. By removing adenosine, ADK promotes flux through the cycle and SAM production, and minimizes competitive inhibition of the methyltransferase reaction by SAH. Geminivirus AC2/AL2 and C2/L2 proteins inactivate ADK and have also been shown to stabilize SAM decarboxylase (SAMDC), which causes decarboxylated SAM (dcSAM) levels to rise. The TYLCCNV betaC1 protein directly antagonizes the methyl cycle by inhibiting SAHH. The C4 protein encoded by CLCuMuV interacts with S-adenosyl methionine synthesase (SAMS) to inhibit SAMS activity for TGS suppression. As described in this work, TLCYnV C4 reverses TGS through interacting with and impacting the DNA-binding ability of NbDRM2. https://doi.org/10.1371/journal.ppat.1008829.g007

Plant material and growth conditions
Transgenic N. benthamiana plants expressing the nuclear marker H2B-RFP (full-length red fluorescent protein fused to the C-terminus of histone 2B) were kindly provided by Dr. Michael M. Goodin (University of Kentucky. KY, USA) [50]. Transgenic N. benthamiana plants expressing TLCYnV C4 were described previously [21]. N. benthamiana plants were grown inside a growth chamber at 26˚C, under a 16-h light/8-h dark photoperiod.

Co-IP assays
Co-IP assays were conducted essentially as described previously [22], with minor modifications. N. benthamiana leaves (0.5 g) transiently transformed to express the proteins of interest were harvested at 2 dpi and ground in 1 mL IP buffer (50mM Tris-HCl, 150mM NaCl, 10mM MgCl 2 , 5mM DTT, and Triton X-100 0.1%), centrifuged at 8,000 g at 4˚C for 15 min, and the soluble proteins were immunoprecipitated with 20 μL anti-Flag M2 magnetic beads at 4˚C for 2 h. Following three consecutive washes with IP buffer with 10 min incubation each at 4˚C, the protein complexes were eluted in 200 μL of elution buffer (200 μg/mL 3×Flag peptide, 50 mM Tris-HCl, 150 mM NaCl, 10 mM MgCl 2 , 5 mM DTT) then centrifuged at 1,600 g for 5 min. The supernatant and crude extracts were subjected to SDS-PAGE/western blot analysis.

Agroinfection assays in N. benthamiana
Agroinfection assays were conducted essentially as described previously [22,23]. The constructs whose backbone is pgR106 were transformed into A. tumefaciens GV3101 by electroporation; others were introduced into A. tumefaciens C58C1 by electroporation. The transformed bacteria cultures were grown individually until approximately OD 600 = 0.5~0.8. The cultures were collected and re-suspended using an induction buffer [10mM MgCl 2 , 100mM MES (pH 5.7), 2mM acetosyringone] for 3 h at room temperature. The suspensions were adjusted to OD 600 = 0.5. For co-expression experiments, the individual cultures were adjusted to OD 600 = 0.4 and equal volumes were mixed before leaf infiltration. The suspensions were infiltrated into leaves of 4-to 6-week old N. benthamiana leaves using 1-ml needleless syringes.

Isolation of total nuclear, nucleoplasmic, and chromatin-bound protein
Isolation of chromatin-bound protein assays were performed as described previously [51][52][53]. Leaves co-expressing NbDRM2 with TLCYnV C4 or C4 mutants were harvested and crosslinked (1% formaldehyde), followed by nucleus isolation using Honda buffer. Three volumes of 1% SDS (10 mM Tris, pH 7.5, 2 mM EDTA, and 1× proteinase inhibitor) were added to the nucleus pellet, which was vortexed for 1 min, followed by centrifugation at 14,000 g for 10 min. The supernatant containing soluble nucleoplasmic proteins was collected. The remaining pellet was resuspended in 3 volumes of 1% SDS and sonicated, followed by centrifugation at 14,000 g for 10 min. The supernatant containing chromatin-bound proteins was collected. Total nuclear protein was isolated by adding 3 volumes of 1% SDS directly to the nucleus pellet, followed by boiling at 95˚C for 10 min. Equal volumes of each fraction were mixed with loading buffer, boiled, gel-separated, and subjected to SDS-PAGE analysis.

Methylation-sensitive PCR
Methylation-sensitive PCR assays were performed as described previously [49]. Systemic leaves of N. benthamiana plants infected by TLCYnV at 14 dpi were harvested and prepared for total genomic DNA isolation. Genomic DNA was isolated by using DNeasy Plant Minikit (Qiagen). Then, 100 ng of genomic DNA was digested for 1 h at 37 with 2U of HpaII, MspI, or McrBc in 20 μl reaction mixtures. The enzymes were heat inactivated, and 2 μl of the cleaved DNA was loading into PCR mixtures containing primers for full-length TLCYnV. PCR products of undigested genomic DNA served as control.

Bisulfite sequencing assays
Bisulfite sequencing assays were performed essentially as described previously [42][43][44]. Genomic DNA was extracted from plant leaves using DNeasy Plant Minikit (Qiagen). To improve the efficiency of bisulfite treatment, DNA (1 mg) was digested with BamHI, which acts outside of interest to decrease the size of DNA, followed by overnight treatment with proteinase K. Bisulfite modification was carried out using the EZ DNA Methylation Gold Kit (Zymo Research). Bisulfite-modified DNA was purified using a Zymo-Spin IC column and dissolved in 20 μl of Elution Buffer according to the manufacturer's instructions. PCR polymerase and products were cloned into a pLB vector. Individual clones were sequenced. Primers were designed against templates are listed in S1 Table. PLOS PATHOGENS TLCYnV C4 reverses TGS by interacting with NbDRM2

ChIP quantitative-PCR
ChIP quantitative PCR was preformed essentially as described previously [54][55][56]. Chromatin was prepared from cross-linked material using extraction buffer, sonicated, diluted, and subjected to immunoprecipitation using anti-Flag M2 magnetic beads. The immunoprecipitation was performed at 4˚C for 3 h, after which the beads were washed four times with ChIP dilution buffer and twice with 1 × TE buffer (10 min each). Immunoprecipitated DNA was eluted after reverse cross-linking by boiling at 95˚C for 10 min, followed by treatment with proteinase K for 1 h at 55˚C. qPCR data was normalized to 1% of the input.

EMSA assays
EMSA assays were performed essentially as described previously [57,58]. The yeast strain Gold co-transformed with the indicated plasmids was subjected to 10-fold series dilution, and grown on a SD/-Leu/-Trp/-His medium. (TIF) S6 Fig. Ser43 of TLCYnV C4 is a key site vital for its interaction with NbDRM2. Leaves coexpressing Flag-NbDRM2 with GFP, TLCYnV C4-, or TLCYnV C4(S43A)-GFP were harvested at 2 dpi for co-immunoprecipitation (Co-IP) assays. Immunoblot analysis was conducted with antibodies specific to detect the indicated proteins.