Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Cytosine methylation is an epigenetic mark that promotes gene silencing and plays an important role in genome defence against transposons and invading DNA viruses. Previous data showed that the largest family of single-stranded DNA viruses, Geminiviridae, prevents methylation-mediated transcriptional gene silencing (TGS) by interfering with the proper functioning of the plant methylation cycle.
Here, we describe a novel counter-defence strategy used by geminiviruses, which reduces the expression of the plant maintenance DNA methyltransferases, METHYLTRANSFERASE 1 (MET1) and CHROMOMETHYLASE 3 (CMT3), in both locally and systemically infected tissues.
We demonstrated that the virus-mediated repression of these two maintenance DNA methyltransferases is widespread among geminivirus species. Additionally, we identified Rep (Replication associated protein) as the geminiviral protein responsible for the repression of MET1 and CMT3, and another viral protein, C4, as an ancillary player in MET1 down-regulation. The presence of Rep suppressed TGS of an Arabidopsis thaliana transgene and of host loci whose expression was strongly controlled by CG methylation. Bisulfite sequencing analyses showed that the expression of Rep caused a substantial reduction in the levels of DNA methylation at CG sites.
Our findings suggest that Rep, the only viral protein essential for replication, displays TGS suppressor activity through a mechanism distinct from that thus far described for geminiviruses.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Geminiviruses constitute a large family of plant viruses with circular, single-stranded (ss) DNA genomes that can infect a broad range of plants (Rojas et al., 2005; Seal et al., 2006). The Geminiviridae family is divided into four genera, Begomovirus, Curtovirus, Topocuvirus and Mastrevirus, based on their genome organization, host range and insect vectors (Fauquet et al., 2008). The largest genus is Begomovirus, which is comprised of species with monopartite or bipartite genomes and has a coding capacity of six to seven genes. Begomovirus replication depends almost entirely on host factors. The viral genomes encode only two proteins that play a role in viral replication: Rep (Replication associated protein, also named C1, AL1 or AC1), which is essential for this process (Elmer et al., 1988), and C3 (also named AL3, AC3 or REn (Replication enhancer protein)), which enhances the accumulation of viral DNA (Sunter et al., 1990). Geminiviruses replicate in the nucleus of an infected cell by a rolling circle replication (RCR) that involves double-stranded (ds) DNA replicative form (RF) intermediates, although some recombination-dependent replication (RDR) also occurs (Jeske, 2009). The dsDNA RF molecules are templates for replication and transcription and become associated with cellular histone proteins to form viral minichromosomes (Pilartz & Jeske, 1992, 2003), which could carry marks associated with repressive or active chromatin (methylation of histone H3 at lysine 9 and DNA methylation, or acetylation of histone H3, respectively; Raja et al., 2008).
RNA silencing has emerged as an important antiviral mechanism in plants (Ruiz-Ferrer & Voinnet, 2009; Ding, 2010; Llave, 2010). Upon infection, plants can process viral RNA into small interfering RNAs (siRNAs) and use those siRNAs to direct specific antiviral silencing activity. Several lines of evidence, especially from studies on siRNAs and viral pathogenesis, have led to the current view of cytoplasmic RNA silencing, which is referred to as post-transcriptional gene silencing (PTGS), as a ubiquitous defence against RNA viruses and transcripts produced by DNA viruses such as geminiviruses. As a counter-defence, geminiviruses encode proteins (i.e. C2, C4 and V2) that are capable of suppressing the PTGS pathway at different steps such as siRNA amplification, siRNA-effector complex loading or the expression of gene silencing-related pathway proteins (for reviews see Diaz-Pendon & Ding, 2008; Raja et al., 2010; Burgyan & Havelda, 2011). In addition, viral DNA and chromatin are subject to repressive siRNA-directed methylation that leads to transcriptional gene silencing (TGS). The idea that the epigenetic arms of RNA silencing are deployed against geminiviruses is based on numerous observations: (1) geminivirus DNA is methylated during infection, and hypermethylation is associated with host recovery (Hagen et al., 2008; Raja et al., 2008; Rodriguez-Negrete et al., 2009; Paprotka et al., 2011); (2) in vitro methylation of geminivirus DNA greatly reduces viral replication in tobacco (Nicotiana tabacum) protoplasts, mostly by inhibiting viral transcription (Brough et al., 1992; Ermak et al., 1993); (3) transgenes driven by geminivirus promoters could be hypermethylated and transcriptionally silenced in plants inoculated with cognate but not heterologous geminiviruses (Seemanpillai et al., 2003; Bian et al., 2006); and (4) direct genetic evidence for chromatin methylation as an epigenetic defence against geminiviruses has been obtained using Arabidopsis thaliana mutants (Raja et al., 2008). These observations suggest that sequence-specific signals that are capable of eliciting TGS are generated during infection. It is expected that geminiviruses also encode suppressors of TGS, and to date, only one viral genome-encoded protein, C2, has been shown to revert TGS and reduce plant DNA methylation by interfering with the proper functioning of the methyl cycle (Buchmann et al., 2009; Zhang et al., 2011). Two mechanisms have been proposed to underlie the function of C2 as a suppressor of TGS: (1) inhibition of adenosine kinase (ADK; Wang et al., 2003) and (2) attenuation of proteasome-mediated degradation of S-adenosyl-methionine decarboxilase 1 (SAMDC1; Zhang et al., 2011). Additionally, the begomovirus-betasatellite βC1 protein has been shown to also interfere with the proper functioning of the methyl cycle by inhibiting the activity of S-adenosyl homocysteine hydrolase (SAHH; Yang et al., 2011).
In mammals, DNA methylation occurs almost exclusively in a symmetric CG context, while in plants, this modification occurs in all sequence contexts: CG, CNG, and CHH (where H is A, T or C). De novo DNA methylation is catalysed by DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2), and once methylation is established, it is maintained by three distinct enzymes: METHYLTRANSFERASE 1 (MET1), CHROMOMETHYLASE 3 (CMT3, a plant-specific DNA methyltransferase) and DRM2 itself, which maintain CG, CNG and asymmetric CHH methylation, respectively (Law & Jacobsen, 2010). DNA methylation seems to be a stable epigenetic mark in plants because methylation patterns are maintained in a multigenerational manner. However, changes in the methylation levels of the plant genome are observed, especially during development. Reduction in methylation can occur passively, by replication in the absence of functional methylation maintenance pathways, or actively, by removal of methylated cytosines. Active demethylation is achieved in plants by DNA glycosylase activity and probably in combination with the base excision repair (BER) pathway (Law & Jacobsen, 2010). DEMETER (DME) and REPRESSOR OF SILENCING 1 (ROS1) are the founding members of a family of DNA glycosylases in A. thaliana that also includes DEMETER-LIKE 2 (DML2) and DEMETER-LIKE 3 (DML3), and mutations in these genes cause increased DNA methylation in all sequence contexts at specific genomic loci.
Raja et al. (2008) demonstrated that A. thaliana mutants in cytosine methyltransferases (drm1 drm2 double mutant and cmt3 and met1 single mutants) are hypersusceptible to infection by geminiviruses, and that viral methylation is significantly reduced in some of these mutants. In this study, we present evidence that plant DNA viruses, like animal DNA viruses (de Souza et al., 2010), alter the levels of expression of the host DNA methyltransferases. Specifically, geminiviruses impair DNA methylation by reducing the transcript levels of the plant DNA methyltransferases responsible for the maintenance of symmetric methylation, MET1 and CMT3. We also present data indicating that Rep is the main viral protein involved in the repression of the maintenance DNA methyltransferases and leads to TGS reversion of a transgene and of host loci. This novel Rep-dependent mechanism would act in concert with the previously described inhibition of the methyl cycle that is exerted by C2. Taken together, the results obtained in this work indicate that this family of viruses have evolved multiple, independent strategies to overcome this host defence response. Moreover, our data unveil a novel function for the geminivirus Rep protein as a TGS suppressor that modifies the plant epigenome and demonstrate that this pathogenicity factor plays a role in the infection process that goes beyond DNA replication.
Materials and Methods
Plasmids and cloning
Cloning details are provided in Supporting Information Methods S1 and Table S1, which lists all the oligonucleotides used; Table S2 indicates the primer pairs used; and Table S3 summarizes the engineering of the plasmids used in this work.
Plant material and growth and treatment conditions
Wild-type Arabidopsis thaliana (L.) Heynh used in this study was the Columbia (Col-0) ecotype. The transgenic L5 line and the efr-1 (EF-Tu RECEPTOR) mutant have been described elsewhere (Morel et al., 2000; Zipfel et al., 2006). Plant growth and treatment conditions are described in Methods S1.
Infection and transient expression assays
Geminiviral inoculations of A. thaliana or Nicotiana benthamiana L. were performed as previously described (Lozano-Duran et al., 2011a; Caracuel et al., 2012). For local infection and transitory expression assays in N. benthamiana, fully expanded leaves of 6-wk-old plants were agroinfiltrated with Agrobacterium tumefaciens clones carrying the corresponding binary vector. For local infection and transitory expression assays in A. thaliana, four fully expanded rosette leaves of 5-wk-old efr-1 mutant plants (Zipfel et al., 2006) were agroinfiltrated. As a control, plants were mock-inoculated with an A. tumefaciens culture harbouring the empty binary vector pGreen-0229 (Hellens et al., 2000) or pBINX1 (Sanchez-Duran et al., 2011). To analyse local infections or transient expression, samples were collected 3–4 d after agroinfiltration. For each experiment three (N. benthamiana) or 12 leaves (A. thaliana) from three independent plants were pooled. Viral infectious clones are described in Table S4.
Arabidopsis thaliana transformation
Transformation of A. thaliana wild-type or L5 (Elmayan et al., 2005) plants was performed using the floral dip method (Clough & Bent, 1998). Procedures are detailed in Methods S1.
Analysis of nucleic acids and proteins
Nucleic acid and protein expression analyses are described in Methods S1.
One to two micrograms of total DNA extracted from 12 to 15 A. thaliana seedlings was used for bisulfite modification with the MethylEasy Xceed Rapid DNA Bisulfite Modification kit (Human Genetic Signatures, Sydney, Australia). Bisulfite-modified DNA was purified and resuspended in 30 μl of elution buffer according to the manufacturer's instructions. Conversion controls provided with the kit were included in all experiments.
Nested PCR was performed to amplify FWA (FLOWERING WAGENINGEN) or CACAT-Like fragments. Specific primers were designed and used in the first (FWA: AtFWAB1-F/AtFWAB1-R; CACAT-Like: AtCACAB1-F/AtCACAB1-R) and the second PCR reactions (FWA: AtFWAB2-F/AtFWAB2-R; CACAT-Like: AtCACAB2-F/AtCACAB2-R). Amplified products were cloned into the pGEM-T Easy vector. Cloned DNA fragments were sequenced and analysed using the Kismeth free-access software (http://katahdin.mssm.edu; Gruntman et al., 2008) to quantify cytosine methylation levels (12 independent clones for each condition).
Histochemical staining of GUS activity
GUS staining was performed according to the protocol previously described in Ranjan et al. (2012) with minor modifications. Plant tissues were immersed in histochemical GUS staining buffer (100 mM NaPO4, 0.5 mM K3[Fe(CN)6, 0.5 mM K4[Fe(CN)6], 10 mM EDTA and 1 mg ml−1 5-bromo-4-chloro-3-indolyl-β-d-glucuronide (X-gluc), pH 7), vacuum-infiltrated under pressure for 2 min three times, and then incubated at 37°C for 8 h. Samples were then washed several times with absolute ethanol until complete tissue clarification.
Geminiviruses reduce transcript levels of plant maintenance DNA methyltransferases in Nicotiana benthamiana
In recent years, several studies have shown that animal DNA viruses alter the expression of the maintenance and de novo DNA methyltransferases as a mechanism to harness host-mediated methylation to their own advantage (de Souza et al., 2010). To determine if plant DNA viruses employ similar strategies, we analysed the expression of the N. benthamiana DNA methyltransferases and demethylases during geminivirus infection. As a first step, we amplified, cloned and sequenced cDNA fragments of c. 500 bp using primers designed for NbMET1 (GenBank sequence accession number FJ222441) or for conserved regions of the CMT3 and DRM2 homologues in A. thaliana, N. tabacum and Solanum lycopersicum (Tables S1, S2). Reverse transcription PCR (RT-PCR) of total RNA from N. benthamiana leaves yielded single cDNA fragments of NbMET1, NbCMT3 and NbDRM2 that showed a high degree of homology to their counterpart genes (Fig. S1A). Although the ROS demethylase is encoded by a single gene in A. thaliana (AtROS1), several homologues have been identified in the genomes of Solanaceae plants: four in N. tabacum (NtROS1, NtROS2a, NtROS2b and NtROS3; Choi & Sano, 2007) and two in tomato (S. lycopersicum) (SlROS1-Solyc09 g009080 and SlROS2b-Solyc10 g083630). Using specific primers for each N. tabacum ROS gene, we were able to isolate two partial cDNA clones that were homologous to NtROS1/SlROS1 and NtROS3/SlROS2b. This finding suggested that two ROS homologues, which we named NbROS1 and NbROS2, also exist in N. benthamiana (Fig. S1B,C).
In a first approach, we determined whether the transcript levels of the DNA methyltransferases and demethylases were altered in a local infection. Nicotiana benthamiana leaves were agroinfected with monopartite (Tomato yellow leaf curl Sardinia virus (TYLCSV) or Tomato yellow leaf curl virus, mild strain (TYLCV-Mld)) or bipartite (African cassava mosaic virus (ACMV) or Tomato golden mosaic virus (TGMV)) begomoviruses. Total RNA was extracted from leaf patches and used to measure the mRNA expression levels of the DNA methyltransferases NbMET1, NbCMT3 and NbDRM2 and the demethylases NbROS1 and NbROS2 by reverse transcription quantitative real-time PCR (RT-qPCR). As a control for the wild-type level of expression, plants were agroinfiltrated with the empty vector (mock). As shown in Fig. 1(a), the expression levels of the maintenance DNA methyltransferases NbMET1 and NbCMT3 were clearly reduced in leaves that were infected with any of the begomoviruses compared with the control sample, whereas the expression of the de novo methyltransferase NbDRM2 was only reduced after ACMV agroinfection. Interestingly, a reduction in the mRNA levels was also observed for NbROS1 and NbROS2 (Fig. 1a). Southern blot analysis and RT-PCR were performed to detect the accumulation of viral DNA forms and the expression of the respective viral Rep transcript to confirm that the geminiviruses were replicating in all assayed samples (Fig. S2A,B). These results indicate that begomoviruses down-regulate the mRNA expression levels of the plant maintenance DNA methyltransferases NbMET1 and NbCMT3; however, with the exception of ACMV, these viruses did not alter the expression of the de novo DNA methyltransferase NbDRM2.
To determine whether the expression of plant DNA methyltransferases is also altered during a systemic infection, we infected transgenic 2IRGFP N. benthamiana plants with TYLCSV (Morilla et al., 2006; Lozano-Duran et al., 2011b). These plants overexpress Rep-dependent green fluorescent protein (GFP), which allows the localization of the tissues where the virus is replicating at a certain time. Because geminiviruses replicate in a reduced subset of plant cells, the use of these transgenic plants improves sensitivity for detecting changes in the expression of the DNA methylation machinery that occur during a systemic geminiviral infection. Plants were inoculated with TYLCSV, and GFP-expressing tissues were collected by manual dissection. As a control, equivalent leaf tissues at a similar stage of development were collected from mock-infected plants. Total RNA was extracted at 35 d post-inoculation (dpi), and the transcript levels of NbMET1, NbCMT3, NbDRM2, NbROS1 and NbROS2 were measured by RT-qPCR. The transcript levels of NbMET1 and NbCMT3 in systemically infected tissue were clearly reduced and represented ≈50% of those in the mock-inoculated tissues (Fig. 1b). Down-regulation of the DNA demethylase NbROS1 was also detected, but unexpectedly, the transcript levels of NbROS2 were only slightly reduced, which differed from what occurred during the local infection (Fig. 1b). No significant changes in the levels of NbDRM2 were detected. Expression of Rep was verified by RT-qPCR (Fig. S2C). These findings confirm that systemic geminivirus infection down-regulates the two plant genes that are responsible for maintaining symmetric DNA methylation, which represents 94% of the methylated cytosines in A. thaliana (Cokus et al., 2008).
Down-regulation of NbMET1 and NbCMT3 depends on Rep but not on the previously known geminivirus silencing suppressors
To identify the viral protein involved in the down-regulation of maintenance DNA methyltransferase genes upon TYLCSV infection, we constructed different versions of TYLCSV that harboured mutations in the three known geminiviral silencing suppressors, C2, C4 and V2. Nicotiana benthamiana leaves were agroinfiltrated with TYLCSV wild-type or C2 (C2T2C; Lozano-Duran et al., 2011a), C4 (C4T2C) or V2 (V2T2G) null mutants, and transcript levels were measured by RT-qPCR. Viruses with mutations in C2 or V2 down-regulated NbMET1 and NbCMT3 to the same level as the wild-type virus, which indicates, that neither of these proteins is necessary for repressing the DNA methyltransferases (Fig. 2a). The C4 mutant repressed NbCMT3 to the same extent as the wild-type virus, but its repression of NbMET1 was significantly weaker (64% of the basal expression for the C4 mutant compared with 39% of the basal expression for wild-type TYLCSV; Student's t-test confirmed that this difference is statistically significant at P <0.01; Fig. 2a). This result indicates that NbMET1 repression is partially dependent on C4, but another viral protein is probably involved. To ensure that variations in NbMET1 or NbCMT3 expression are not attributable to differences in the ability of the mutants to replicate, we determined the level of viral DNA and Rep transcripts in the infiltrated tissues, and no significant differences were detected between the wild-type and the mutant viruses (Fig. S3A,B). To further investigate these results, we overexpressed the viral proteins in N. benthamiana leaves. As shown in Fig. 2(b), agroinfiltration with vectors expressing C2, C4 or V2 did not result in a reduction in the accumulation of NbMET1 or NbCMT3 transcripts compared with mock-infiltrated leaf patches. Expression of the viral transcripts was confirmed by RT-PCR (Fig. S3C). In summary, our findings show that none of the known TYLCSV silencing suppressors is involved in the down-regulation of NbCMT3. Only C4 seems to play an ancillary role in suppressing NbMET1 because its presence is not sufficient to alter the expression of this DNA methyltransferase, but it is required to reach the level of repression of NbMET1 that is detected after infection with the wild-type virus.
Replication initiation proteins that exhibit functional similarities to geminiviral Rep, such as polyomavirus large T antigen (TAg) and adenovirus E1a activate MET1 homologue (DNMT1, DNA methyltransferase 1) through the pRB (retinoblastoma)/E2F pathway (McCabe et al., 2006; Jung et al., 2007). Interestingly, the activation of MET1-dependent imprinted genes in plants is also controlled, among other mechanisms, by transcriptional repression of AtMET1 through a retinoblastoma-dependent pathway (Jullien et al., 2008). As the interaction between the geminiviral Rep and plant RBR homologues has been reported to be essential for full infection (Hanley-Bowdoin et al., 2004), we wondered if begomoviral Rep was involved in down-regulating MET1. Unfortunately, we could not use a Rep null mutant for infection experiments because Rep is absolutely required for viral replication. Therefore, we overexpressed TYLCSV Rep in N. benthamiana leaves. Because the Rep open reading frame (ORF) encompasses the C4 ORF, we generated two different constructs: one that expressed both proteins from the same vector (35S:Rep+C4) and another that expressed only Rep (35S:Rep−C4). In the second construct, a mutation was made that introduced a premature stop codon that truncated the C4 protein after nine amino acids but did not alter the amino acid sequence of Rep (C2437G). mRNA expression levels of NbMET1 and NbCMT3 were measured by RT-qPCR in the agroinfiltrated tissues. The results showed that overexpression of Rep decreased the transcript levels of NbMET1 and NbCMT3, which indicated that this viral protein is responsible for down-regulating the maintenance DNA methyltransferases (Fig. 2b). The Rep-mediated repression of NbCMT3 seems to be independent of C4 because the expression of Rep−C4 or Rep+C4 led to similar levels of repression (c. 55–60%). However, although the down-regulation of NbMET1 is also dependent on Rep, it is enhanced by the presence of C4 (Student's t-test confirmed that this difference is statistically significant at P <0.05; Fig. 2b). RT-qPCR analyses of Rep+C4, Rep−C4 and C4 transcript levels were performed as controls and, although some differences in the expression of the viral constructs were detected, these differences cannot explain the results obtained for the DNA methyltransferase genes (Fig. S3D).
To complete the analysis of the effect of viral proteins on the expression of host DNA methyltransferases and demethylases, we measured the expression of NbDRM2, NbROS1 and NbROS2 in leaves agroinfiltrated with the constructs that expressed Rep−C4, Rep+C4, C2, C4 or V2. No effect on NbDRM2 transcript levels was observed after the expression of any of the viral proteins; however, NbROS1 levels were significantly reduced when Rep and C4 were present (Rep+C4), and this correlates with the strongest repression for NbMET1 (Fig. S3E).
To gain insight into the mechanism of Rep-induced MET1 repression, we investigated whether the interaction between Rep and RBR is required for this effect. Mutations in TGMV Rep that alter its interaction with RBR have been previously described (Kong et al., 2000). Two mutated versions of Rep that were impaired in the interaction with maize (Zea mays) and A. thaliana RBR1 (RS-R125 and KEE146) and a mutant that displayed a higher binding activity to both RBR1 proteins (Q-HN165) were used for this analysis. Wild-type or mutated versions of the complete ORFs of TGMV Rep were cloned into binary plasmids, under the control of the 35S promoter, and were used to agroinfiltrate N. benthamiana leaves. RT-qPCR analysis of infiltrated tissues showed that all mutated Rep proteins suppressed NbMET1 (and NbCMT3) expression like the wild-type Rep protein (no statistically significant differences were detected between the mutated versions of Rep and the wild-type protein; Fig. 2c). These results were reinforced by the finding that the expression of a mutated version of TYLCSV Rep (RS-R124), which was equivalent to the RS-R125 mutation for TGMV Rep, also repressed NbMET1 to the same extent as the wild-type protein (Fig. 2d). Although differences in expression of the different versions of Rep, from TGMV or TYLCSV, were detected by RT-qPCR, those variations did not correlate with the results obtained for NbMET1 and NbCMT3 (Fig. S4). Taken together, these results suggest that the Rep–RBR1 interaction does not mediate the down-regulation of NbMET1 or NbCMT3.
Geminivirus infection or expression of TYLCSV Rep down-regulates the expression of the MET1 and CMT3 genes in A. thaliana
To further investigate the consequences of repressing the maintenance DNA methyltransferases, we decided to take advantage of the molecular and genetic tools that are available for A. thaliana. As a first step, we determined if geminivirus local infection also alters the expression of the DNA methyltransferases and demethylases in this model organism. Because TYLCSV cannot infect A. thaliana, we used two geminiviruses that can: TYLCV-Mld (a TYLCSV-related begomovirus) and Beet curly top virus (BCTV; a curtovirus). Transcript levels of the DNA methyltransferases (AtDRM2, AtMET1 and AtCMT3) and a DNA demethylase (AtROS1) were measured by RT-qPCR after agroinfiltration of an A. thaliana efr-1 mutant, which supports higher levels of transient transformation with A. tumefaciens (Zipfel et al., 2006), with the geminivirus species. As in N. benthamiana, we detected a significant reduction in AtMET1, AtCMT3 and AtROS1 expression after infection with both viruses; however, expression of AtDRM2 was not altered (Fig. 3a). We confirmed by RT-PCR that Rep transcripts from both viruses were being produced (Fig. S5A). These results suggest that the repression of the plant DNA methylation machinery after geminivirus infection occurs in different plant species.
Next, we tested if the expression of Rep and C4 of TYLCSV down-regulates AtMET1 and AtCMT3 in A. thaliana as it does in N. benthamiana. We agroinfiltrated the efr-1 mutant with the binary constructs that express Rep and C4 (35S:Rep+C4), Rep (35S:Rep−C4) or C4 (35S:C4), and measured the transcript levels of AtMET1 and AtCMT3 by RT-qPCR. The results confirmed that the expression of TYLCSV Rep is sufficient to down-regulate AtMET1 and AtCMT3, but C4 alone did not alter their transcript levels. As previously observed in N. benthamiana, C4 expression enhances the Rep-mediated repression of AtMET1 but not of AtCMT3 (Fig. 3b; Student's t-test confirmed that this difference is statistically significant at P <0.05). Although differences in the expression of the three viral constructs were detected by RT-qPCR (Fig. S5B), the variations cannot explain the results that were obtained for AtMET1 and AtCMT3. These data confirm that Rep also mediates the repression of these two maintenance DNA methyltransferases in A. thaliana.
To determine whether Rep/C4-mediated repression of AtMET1 and AtCMT3 leads to a reduction in plant DNA methylation and induces the expression of transcriptionally silenced loci, we constructed transgenic lines that expressed the viral ORFs under the control of a β-estradiol-inducible promoter (Zuo et al., 2000), as a decrease of AtMET1 or AtCMT3 expression will cause a loss of DNA methylation only in cells that are actively replicating. Two independent lines, RC7 and RC10, which showed low transgene expression in the absence of β-estradiol and reached similar levels of Rep transcript and Rep protein after β-estradiol treatment, were selected (Fig. S5C,D). Total RNA was extracted from seedlings that were either treated or not treated with β-estradiol for 24 or 72 h and used to measure AtMET1 and AtCMT3 transcript levels by RT-qPCR. As a control, we used RNA extracted from transgenic plants that had been transformed with the empty vector (Ev). The reduction of AtMET1 and AtCMT3 transcript levels in the transgenic lines (RC7 and RC10) was significant compared with the control line (Ev) at both time-points (Fig. 4a). As expected, the AtROS1 mRNA levels were reduced, but the AtDRM2 transcript levels were not consistently affected by the expression of Rep (Fig. S5E).
Expression of Rep reverses endogenous epigenetic gene silencing by reducing cytosine methylation in A. thaliana
To investigate whether Rep-mediated repression of AtMET1 and AtCMT3 also leads to changes in the expression of A. thaliana genes that were regulated by DNA methylation, we measured the expression of transcriptionally gene-silenced loci (TGS loci) in the same samples used to detect the level of the DNA methyltransferases in Fig. 4(a). Primers were designed to detect the level of transcripts from loci whose expression is specifically detected on a met1 mutant background (i.e. the AtFWA gene and the AtCACTA-like DNA transposon) and loci that are specifically expressed on mutant backgrounds that primarily impact non-CG methylation (i.e. the retrotransposons AtSN1 (SHORT INTERSPERSED ELEMENT1) and AtTa3; Zhang et al., 2006). RT-qPCR analysis showed that the expression of AtFWA and AtCACTA-like was increased in both transgenic lines after β-estradiol treatment, while the level of AtSN1 transcripts was not altered at 24 or 72 h after Rep induction (Fig. 4b). Although induction of AtFWA and AtCACTA-like was detected in both transgenic lines, the increase was higher in the RC10 line (3–4-fold rise for AtFWA and 4–5-fold for AtCACTA-like compared with the control line), which correlates with a stronger repression of AtMET1 and AtCMT3 observed in this line (Fig. 4a). No transcripts from the retrotransposon AtTa3 could be detected in any of the lines under any condition. Because we could detect AtTa3 transcripts in a ddm1 (DECREASED DNA METHYLATION1) mutant (Fig. S6), we conclude that the presence of Rep is not sufficient to alleviate the silencing of this locus. These results confirm that Rep-mediated repression of AtMET1 is capable of reversing the TGS of loci that are controlled predominantly by CG methylation and pinpoint Rep as a novel geminiviral TGS suppressor.
The increase in the expression of AtFWA and AtCACTA-like in the RC10 line was already detected after 24 h of hormone treatment. To determine whether the expression of the AtFWA gene and AtCACTA-like correlated with a reduction in the DNA methyltransferase activity in planta, we analysed the methylation status of both loci by bisulfite sequencing. DNA obtained from transgenic RC10 seedlings grown in the presence of β-estradiol was treated with bisulfite reagent to convert unmethylated cytosines to uracil. PCR fragments of the AtFWA promoter region (region B1; (Kinoshita et al., 2007)) and the AtCACTA-like upstream region (coordinates 1668390–1671115 from AT2G04760) were amplified, cloned and fully sequenced. As controls, equivalent PCR fragments from β-estradiol-treated Ev plants were also analysed. The results from 12 independent clones for each locus were compiled and are shown in Fig. 4(c). At the AtFWA locus, the expression of Rep resulted in a 18% decrease in the number of cytosine residues that were methylated at CG sites (74% in the RC10 line compared with 91% in the Ev line), but no significant changes were detected at non-CG methylated sites. Similar results were obtained for the AtCACTA-like transposon, which showed a 18% reduction in cytosine methylation in the CG context (75% in the RC10 line compared with 91% in the Ev line) and no significant changes at non-CG methylated sites. Cytosine methylation profiles for both loci are presented in Fig. S7. Similar levels of hypomethylation at non-CG sites were obtained for AtSN1 loci (28% for CHG and 13% for CHH) when the previously known geminiviral TGS suppressor, C2, was expressed in A. thaliana plants (Buchmann et al., 2009). Thus, our data indicate that Rep-mediated repression of AtMET1 leads to the hypomethylation of CG sites in A. thaliana.
The A. thaliana L5 transgenic line contains a transcriptionally silenced and methylated GUS transgene (Morel et al., 2000). All currently tested TGS mutants, including met1 and cmt3, reactivated expression of L5-GUS (Elmayan et al., 2005), which confirmed that the GUS transgene is transcriptionally silenced and indicates that the A. thaliana L5 line is a useful tool to further confirm the role of Rep in suppressing TGS. Because geminivirus-mediated suppression of the transcriptionally repressed GUS transgene has not been reported, we first infected L5 plants with BCTV or TYLCV-Mld and monitored GUS expression in leaves that were stained at different times post-infection. A larger number of blue foci were apparent at 28 dpi in leaves from systemically infected L5 plants compared with mock-inoculated plants, which indicated that geminivirus infection activates a transcriptionally silenced transgene in A. thaliana (Fig. 5a). Total DNA was extracted from the histochemically stained tissue, and a PCR was performed to confirm the presence of viral DNA (Fig. 5a).
Next, we generated transgenic L5 plants that expressed the Rep and C4 ORFs under the control of a β-estradiol-inducible promoter. A transgenic line, named RCL1, which showed low transgene expression in the absence of β-estradiol and reached high levels of Rep transcript after hormone treatment, was selected. Total RNA was extracted from seedlings that were either treated or not treated with β-estradiol for 24 or 72 h and used to measure Rep (Fig. S8), AtMET1 and AtCMT3 transcript levels by RT-qPCR. As in the previous transgenic lines (RC7 and RC10; Fig. 4a), the reduction in the transcript levels of AtMET1 and AtCMT3 was detected at 24 and 72 h (Fig. 5b). GUS expression was detected by histochemical staining. Differences between the treated and nontreated transgenic plants started being apparent after 72 h and were clearly noticeable 5–7 d after the hormone induction (Fig. 5c). GUS expression was stronger in roots, which are in direct contact with the medium that contains the hormone.
Together, these experiments demonstrate that geminivirus systemic infection or expression of Rep alone induces the expression of plant TGS loci by interfering with their levels of DNA methylation and suggest a novel function for Rep as a TGS suppressor.
The addition of methyl groups to cytosine residues in DNA is a common epigenetic mark that is associated with inactivation of gene transcription. In plants, the genomic DNA methylation status is controlled by the activity of methyltransferases and demethylases and the constant supply of methyl group donors. Previous studies have shown that geminiviruses interfere with the proper functioning of the cellular methyl cycle (Buchmann et al., 2009; Yang et al., 2011; Zhang et al., 2011). We now provide evidence that geminiviruses use an alternative mechanism to interfere with the host DNA methylation machinery during the infection, by reducing the transcript levels of key enzymes that maintain this modification. Local infection with the begomovirus TYLCSV, TYLCV-Mld, ACMV or TGMV reduced the expression levels of the plant maintenance DNA methyltransferases MET1 and CMT3 in N. benthamiana. This repression was also produced during a local infection by BCTV or TYLCV-Mld in A. thaliana, which indicates that transcriptional repression of DNA methyltransferases is a mechanism that is widely used by geminiviruses and is conserved between different hosts. Interestingly, no changes in the transcript levels of de novo DNA methyltransferase DRM2 were detected for any of the viruses that were assayed except for ACMV, which suggests that geminiviruses suppress the maintenance rather than the establishment of DNA methylation. Although begomovirus local infection also reduced the accumulation of N. benthamiana demethylase (NbROS1 and NbROS2) transcripts, the fact that, in A. thaliana, the maintenance of CG methylation by AtMET1 is required for the expression of AtROS1 (Mathieu et al., 2007) suggests that reducing NbROS1 and NbROS2 expression would be a consequence of the viral-induced change in the level of NbMET1 transcripts. This scenario was also observed in systemically infected tissues. The data obtained from systemically infecting 2IRGFP plants with TYLCSV verify that the levels of NbMET1 and NbCMT3 transcripts are reduced in tissues where the virus is actively replicating, confirming the biological relevance of this finding. In a previous study, suppression of these genes was also apparent in microarray data obtained from A. thaliana plants infected with another begomovirus, Cabbage leaf curl virus (Ascencio-Ibanez et al., 2008).
Although three geminivirus-encoded proteins, C2, C4 and V2, suppress PTGS, only C2 interferes with TGS (Raja et al., 2010). Using two different approaches, infection with null mutants and transient expression of the proteins, we have demonstrated that neither C2 nor V2 is implicated in the reduction of the transcript levels of the DNA methyltransferases NbMET1 and NbCMT3. However, although infection with a C4 mutant produced repression of NbCMT3 similar to that produced by the wild-type virus, the reduction in NbMET1 transcript levels was consistently weaker, which suggested that C4 was partially responsible for NbMET1 repression. The fact that transient expression of C4 did not significantly alter the transcript levels of any of the DNA methyltransferases confirms that C4 is not involved in CMT3 repression and suggests that it functions as an enhancer of MET1 repression. The results obtained by transiently expressing Rep in N. benthamiana and A. thaliana leaves points to Rep as the viral protein responsible for this viral-induced repression of the DNA methyltransferases and confirms the ancillary role of C4 in down-regulating MET1 but not CMT3. Although our data clearly demonstrate that the presence of TYLCSV Rep reduces the expression of the DNA methyltransferases MET1 and CMT3, the impossibility of using a Rep null mutant virus prevents us from completely ruling out the possibility that a viral factor(s) other than C4 could have an auxiliary role in the down-regulation of these plant genes.
The previous reports on transcriptional regulation of the mammalian MET1 homologue (DNMT1) by DNA viruses through the pRB/E2F pathway (McCabe et al., 2006; Jung et al., 2007) and the results published by Jullien et al. (2008), which showed that the transcription of AtMET1 is regulated by a protein complex that includes AtRBR1, led us to consider the possibility that Rep-mediated MET1 repression is dependent on the Rep–RBR interaction. However, the data that were obtained by overexpressing mutated versions of TGMV and TYLCSV Rep, which impaired or increased Rep–RBR binding, suggest that this interaction is not necessary for repressing the plant DNA methyltransferases. Rep-mediated repression of its own transcription by binding to specific sequences of its promoter has been reported (Shung & Sunter, 2007). Although we have not found Rep-binding motives in AtMET1 or AtCMT3 promoters, we cannot rule out completely the possibility that Rep binding could play a role in the repression of these plant genes.
Plant recovery from geminiviral infection requires that the methylation machinery hypermethylate the viral genome, which leads to a reduction in viral replication and disease symptoms. Because Rep is essential for viral replication, we could not directly demonstrate that the viral genome is hypermethylated in the absence of Rep. However, the fact that several methylation-deficient A. thaliana mutants, including met1 and cmt3, are more susceptible to infection by geminiviruses (Raja et al., 2008) clearly indicates that the maintenance of methylation function is a defence mechanism against geminiviruses and supports the idea that Rep-mediated suppression of MET1 benefits viral propagation. Our work shows that geminiviruses encode at least two suppressors (Rep and C2) that inhibit TGS at different levels and perhaps at different stages during infection of the plant. Early in infection, the virus requires the expression of Rep to replicate, and the expression of C2 is enhanced when Rep represses its own transcription and activates an internal C2 promoter (Shung & Sunter, 2007). Geminivirus replication occurs via a combination of RDR and RCR (Jeske, 2009). The RDR mechanism relies on homologous recombination, and the two strands are synthesized de novo. By contrast, RCR is a semiconservative process in which viral ssDNA is complemented by second-strand synthesis to produce dsDNA RF intermediates. These RF intermediates should be hemimethylated due to the pre-existing methylated and the de novo unmethylated strand. Because the actual model for geminivirus replication considers that RCR may precede RDR, evading the maintenance of viral DNA methylation of the RF intermediates at the early stages of infection will increase the number of ssDNA viral particles that are present in the plant. Successive suppression of maintenance DNA methyltransferases and methyl group donors by Rep and C2, respectively, would efficiently repress the methylation pathway and allow the virus to escape its inhibitory effect.
The apparent complementary suppressive activity of Rep and C2 during a viral infection is also evident in the DNA context in which these proteins alter methylation. They both hypomethylate plant TGS loci to similar levels (c. 15–20%) but they differ in the type of DNA methylation that is affected. Bisulfite sequence analysis of TGS loci in plants that were infected with C2 mutants or overexpressed this viral protein indicates that C2 mainly suppresses non-CG methylation (Buchmann et al., 2009). By contrast, transgenic expression of Rep reduces CG but not non-CG DNA methylation, and, consequently, Rep expression induces the expression of loci that are mainly controlled by CG-methylation (AtFWA and the AtCACTA-like DNA transposon) but not the expression of loci that are controlled only or mostly by non-CG methylation, such as the retrotransposon AtTa3 or AtSN1 (Zhang et al., 2006). Although a met1 mutant lacks virtually all CG methylation and a substantial amount of non-CG methylation (Tariq et al., 2003; Zhang et al., 2006), we could not detect a consistent reduction in the levels of non-CG methylation because the decrease in AtMET1 transcript levels after Rep expression was less severe than in a met1 null mutant. As expected, the down-regulation of AtCMT3 after Rep expression seems to have no impact on genomic CNG methylation levels because it has previously been shown that AtDRM1/2 and AtCMT3 genes act in a partially redundant manner to control non-CG methylation (Cao & Jacobsen, 2002). Furthermore, a drm1 drm2 cmt3 triple mutant is required to eliminate the vast majority of this methylation (Zhang et al., 2006).
Experiments using A. thaliana L5 transgenic plants revealed that transgenic expression of TYLCSV Rep and BCTV or TYLCV-Mld systemic infection can revert TGS of a transgene, which confirms that geminiviruses suppress epigenetic silencing in a biologically significant context and that this reversal probably depends on Rep. Because the construct that was used to express TYLCSV Rep, in all of the transgenic lines that were generated in this work, also contains the C4 ORF, we cannot rule out the possibility that the level of TGS suppression that was observed in the transgenic plants is partially attributable to the presence of C4. Experiments are in progress to address this question.
Finally, in addition to a direct beneficial effect for viral multiplication, a reduction in DNA methylation could benefit viral infection through changes in the plant epigenome. Studies in plants infected with bacteria and Tobacco mosaic virus have shown that pathogen infection leads to hypomethylation of the host genome, which can influence the expression of defence genes or promote an increase in recombination (Alvarez et al., 2010; Dowen et al., 2012). This increase in recombination affects the activity or integrity of genes that are clustered in regions that are rich in transposable elements and repetitive sequences, such as the NBS-LRR (NUCLEOTIDE BINDING SITE LEUCINE REACH REPEAT) genes.
In summary, we conclude that the geminivirus Rep protein can suppress TGS by reducing CG methylation, and this process is accomplished by reducing the expression of the maintenance methyltransferases MET1 and CMT3 in different plant species. These data reinforce the importance of TGS suppression for plant DNA viruses that, through the combined and complementary actions of at least two viral proteins, Rep and C2, reduce DNA methylation in all sequence contexts. Furthermore, our results confirm that transcriptional modulation of DNA methyltransferases is a universal strategy of eukaryotic DNA viruses, although with opposite outcomes in animals and plants; animal viruses activate the expression of MET1 and geminiviruses repress it. These different strategies would reflect the distinct roles that CG methylation plays in the regulation of mammalian and plant epigenomes. While in animals CG methylation largely contributes to the regulation of gene expression (it is mostly found near gene promoters in dense clusters known as CpG islands), in plants it plays an important role in the maintenance of the stability of plant genome by TGS of transposons and other repetitive DNA elements. The biological importance of the hypomethylation of the plant genome after geminivirus infection remains to be elucidated.
We thank Mayte Duarte, Silvia Hernández and Esther Sánchez for technical assistance and Miguel Sánchez-Durán for generating the anti-Rep antibody. We also thank Dr Linda Hanley-Bowdoin for sharing materials and Dr Hervé Vaucheret for providing the A. thaliana L5 line. This research was supported by grants from the Spanish Ministerio de Educación, Ciencia y Tecnología (MEC, AGL2010-22287-C02-02), FEDER and Junta de Andalucía (P07-CVI-02605). R.L-D. was awarded a predoctoral fellowship from the Spanish MEC.