Analysis of Geminivirus AL2 and L2 Proteins Reveals a Novel AL2 Silencing Suppressor Activity

ABSTRACT

Both posttranscriptional and transcriptional gene silencing (PTGS and TGS, respectively) participate in defense against the DNA-containing geminiviruses. As a countermeasure, members of the genus Begomovirus (e.g., Cabbage leaf curl virus) encode an AL2 protein that is both a transcriptional activator and a silencing suppressor. The related L2 protein of Beet curly top virus (genus Curtovirus) lacks transcription activation activity. Previous studies showed that both AL2 and L2 suppress silencing by a mechanism that correlates with adenosine kinase (ADK) inhibition, while AL2 in addition activates transcription of cellular genes that negatively regulate silencing pathways. The goal of this study was to clarify the general means by which these viral proteins inhibit various aspects of silencing. We confirmed that AL2 inhibits systemic silencing spread by a mechanism that requires transcription activation activity. Surprisingly, we also found that reversal of PTGS and TGS by ADK inactivation depended on whether experiments were conducted in vegetative or reproductive Nicotiana benthamiana plants (i.e., before or after the vegetative-to-reproductive transition). While AL2 was able to reverse silencing in both vegetative and reproductive plants, L2 and ADK inhibition were effective only in vegetative plants. This suggests that silencing maintenance mechanisms can change during development or in response to stress. Remarkably, we also observed that AL2 lacking its transcription activation domain could reverse TGS in reproductive plants, revealing a third, previously unsuspected AL2 suppression mechanism that depends on neither ADK inactivation nor transcription activation.

IMPORTANCE RNA silencing in plants is a multivalent antiviral defense, and viruses respond by elaborating multiple and sometimes multifunctional proteins that inhibit various aspects of silencing. The studies described here add an additional layer of complexity to this interplay. By examining geminivirus AL2 and L2 suppressor activities, we show that L2 is unable to suppress silencing in Nicotiana benthamiana plants that have undergone the vegetative-to-reproductive transition. As L2 was previously shown to be effective in mature Arabidopsis plants, these results illustrate that silencing mechanisms can change during development or in response to stress in ways that may be species specific. The AL2 and L2 proteins are known to share a suppression mechanism that correlates with the ability of both proteins to inhibit ADK, while AL2 in addition can inhibit silencing by transcriptionally activating cellular genes. Here, we also provide evidence for a third AL2 suppression mechanism that depends on neither transcription activation nor ADK inactivation. In addition to revealing the remarkable versatility of AL2, this work highlights the utility of viral suppressors as probes for the analysis of silencing pathways.

INTRODUCTION

Viruses belonging to the Geminiviridae package small, circular single-stranded DNA (ssDNA) genomes in unique double icosahedral particles. The ssDNA genome, which may be monopartite or bipartite, is replicated by rolling circle- and recombination-mediated mechanisms using double-stranded DNA (dsDNA) intermediates that associate with cellular histones to form minichromosomes. Replication and transcription from viral templates are accomplished by host polymerases and accessory proteins. Viral gene products are typically multifunctional proteins that direct the host machinery to viral templates, initiate specific steps in replication and/or transcription, enable virus spread within and between hosts, provide a cellular environment favorable to replication, and suppress host defenses (1–3).

RNA silencing refers to a set of mechanistically related and evolutionarily conserved processes, including posttranscriptional gene silencing (PTGS) and transcriptional gene silencing (TGS). In plants, PTGS typically leads to translational inhibition and/or cytoplasmic degradation of mRNAs (4). TGS is an epigenetic, nuclear phenomenon associated with repressive histone modifications and RNA-directed DNA methylation (RdDM) (5). Antiviral roles for PTGS and TGS are well established, and their importance is underscored by the fact that virtually all plant viruses encode proteins that act as silencing suppressors (2, 6–9). These viral counterdefensive proteins are pathogenicity factors that often employ multiple mechanisms to block different aspects of silencing. Known mechanisms include binding short or long dsRNA, binding to and inhibiting silencing pathway components, and interfering with cellular pathways that regulate or support silencing. Some suppressors inhibit the establishment of silencing, whereas others reverse silencing, suggesting that they target maintenance pathways. A remarkable feature that greatly enhances the defensive value of silencing is its ability to spread cell-to-cell and systemically throughout the plant (10), and some viral suppressors antagonize this process as well.

The best-studied geminivirus silencing suppressor is the AL2 protein encoded by members of the genus Begomovirus, which includes Tomato golden mosaic virus (TGMV) and Cabbage leaf curl virus (CaLCuV). The 15-kDa AL2 (also known as AC2, C2, or TrAP, transcriptional activator protein) was initially characterized as a transcription factor required for expression of viral late genes (11, 12). The related L2 protein (also known as C2) found in the Curtovirus genus typified by Beet curly top virus (BCTV), is not required for late viral gene expression and lacks transcription activation activity (13, 14). However, AL2 and L2 share pathogenicity functions (15–17). AL2 can suppress PTGS by multiple mechanisms. One involves transactivation of host genes that may encode negative regulators of RNA silencing, including WEL1 (Werner exonuclease-like 1) and rgsCaM (regulator of gene silencing-calmodulin-like protein) (18, 19). This mechanism is referred to as transcription-dependent suppression (20). A transcription-independent PTGS suppression mechanism correlates with the ability of AL2 and L2 to interact with and inactivate adenosine kinase (ADK), a methyl cycle-associated enzyme required for the efficient production of S-adenosyl methionine (SAM), an essential methyltransferase cofactor (17, 21, 22).

Methylation-mediated TGS also acts as a defense against geminiviruses (23). AL2 can suppress methylation and TGS by both transcription-dependent and -independent means, while L2 again is limited to the latter mechanism. Transcription-independent reversal of TGS by AL2 and BCTV L2 also correlates with methyl cycle interference by ADK inhibition (24). The importance of the methyl cycle for defense against geminiviruses is further highlighted by the observation that Beet severe curly top virus (BSCTV) C2 protein inhibits proteosome-mediated degradation of SAM decarboxylase (SAMDC), which stabilizes a competitive inhibitor of SAM (25). In addition, the βC1 protein of geminivirus satellite DNA β interferes with SAM synthesis by inhibiting S-adenosyl homocysteine hydrolase (SAHH) (26).

Despite advances in our understanding of the suppression mechanisms employed by AL2 and L2, much remains to be learned about these proteins and the specific aspects of PTGS and TGS they target. The goal of the present study was to determine whether, and in general terms how, AL2 and L2 inhibit systemic spread of silencing or cause the reversal of PTGS and TGS. In the process, evidence for the existence of an additional transcription-independent AL2 suppression mechanism was obtained, and developmental changes in the role of ADK in the Nicotiana benthamiana methyl cycle were revealed.

MATERIALS AND METHODS

Growth conditions and suppression of systemic silencing spread.

N. benthamiana plants used in this study were maintained in a growth room at 24°C with a 12-h light-dark cycle. Details specific for PTGS and TGS reversal assays are provided below.

An assay to assess suppression of systemic silencing has been described previously (27). Briefly, lower leaves of transgenic N. benthamiana line 16c plants, containing an active 35S-green fluorescent protein (35S-GFP) transgene (28), were coinfiltrated with Agrobacterium tumefaciens cells harboring plasmids designed to express double-stranded GFP RNA (dsGFP, silencing trigger) and control or geminivirus proteins from the 35S promoter. β-Glucuronidase (GUS) was a negative control, and p19 was a positive control. All expression plasmids used in these experiments have been described previously (14, 22, 24). Plants were screened for GFP expression with a handheld long-wave UV lamp (Blak-Ray model B 100 YP) and photographed with a Nikon D40 camera equipped with 52-mm UV lens and a yellow filter.

PTGS reversal assays.

Seeds of N. benthamiana line 16c were sown in 6-cm pots, and at 2 weeks postgermination, seedlings were agroinfiltrated with dsGFP to trigger systemic silencing. Two weeks later plants were screened under UV light for GFP expression. Those not exhibiting systemic silencing were removed from the study. The remaining plants were transferred to fresh 6-cm pots for reproductive growth (reproductive plants) or to 16.5-cm pots to prolong vegetative growth (vegetative plants). Plants were monitored for an additional 2 weeks for complete silencing, indicated by the entire plant appearing red under UV light. Plants were then agroinfiltrated with recombinant potato virus X (PVX) vectors expressing control or geminivirus proteins (24, 29). Leaf sections were collected 4 weeks postinoculation from symptomatic, systemically infected leaves and analyzed for GFP expression using a Nikon PCM 2000 confocal laser scanning microscope as described previously (30).

TGS reversal assays.

Two-week old seedlings of N. benthamiana 16-TGS, a line derived from 16c that contains a transcriptionally silenced 35S-GFP transgene (24, 31), were divided into vegetative and reproductive populations as described above. At 3 weeks of age, plants were agroinoculated with geminiviruses, PVX vectors, or gene-silencing vectors derived from tobacco rattle virus (TRV) (24, 32). GFP expression indicating TGS reversal was visualized under UV light and photographed as previously described.

Nucleic acid extraction and analysis.

To minimize plant-to-plant variation, in all cases sample extracts were prepared from pooled leaves from three to four plants.

RNA extraction and Northern blotting procedures have been described previously (22). Northern blots contained 4 μg of RNA per lane, and gels were stained with ethidium bromide to visualize the 28S rRNA loading control. Blots for GFP expression analysis utilized an antisense GFP riboprobe synthesized in vitro with [α-³²P]UTP (3,000 Ci/mmol; Perkin-Elmer) using a T7 MAXIscript kit (Ambion). Blots analyzed for TRV or PVX viral titers were probed with a mixture of antisense 5′-end-labeled oligonucleotides prepared using [γ-³²P]ATP (3,000 Ci/mmol; Perkin-Elmer) and T4 polynucleotide kinase (Fermentas).

DNA extraction and Southern blotting techniques were previously described (23). DNA extracts from geminivirus-infected plants were restricted overnight with ScaI (for BCTV) or XcmI (for CaLCuV) prior to 1% gel electrophoresis to linearize circular viral DNA. Membranes were cross-linked and probed with oligonucleotides specific for either the 18S rRNA gene (loading control) or geminivirus DNA. Mixtures of five antisense oligonucleotides were 5′ end labeled using [γ-³²P]ATP (3,000 Ci/mmol; Perkin-Elmer) and T4 polynucleotide kinase (Fermentas). Oligonucleotides used for probing gel blots are available on request.

ADK assays.

ADK activity was determined as described previously (17), using total protein extracts from individual mock-inoculated plants or from plants infected with geminiviruses or TRV vectors. Extracts (250 ng) were incubated with adenosine and [γ-³²P]ATP (3,000 Ci/mmol; Perkin-Elmer) at 30°C for 20 min, and products were fractionated by thin-layer chromatography (TLC) on polyethyleneimine-cellulose plates (Sigma) in 1 M glacial acetic acid. Radiolabeled AMP was quantitated using a phosphorimager (Molecular Imager FX; Bio-Rad).

Methylation-sensitive extension assay.

Global genome methylation was evaluated using a cytosine extension assay (33). Briefly, 1 μg of total N. benthamiana DNA was digested overnight with a 10-fold excess of the methylation-sensitive restriction enzyme MspI (C/CGG; New England BioLabs), leaving an overhang at nonmethylated sites. A second DNA aliquot from mock samples was incubated without enzyme to provide a background control. Single-nucleotide extension reactions using Taq DNA polymerase and [α-³²P]dCTP were then performed as described previously (24).

RESULTS

AL2, but not L2, prevents systemic spread of silencing in a transcription-dependent manner.

Using agroinfiltration assays in Nicotiana benthamiana, we previously showed that TGMV AL2 and a truncated AL2 protein consisting of residues 1 to 100 (AL2_1–100, lacking the transcription activation domain), the BCTV L2 protein, a double-stranded RNA corresponding to ADK (dsADK), and an inhibitory adenosine analogue are able to suppress the establishment of local silencing directed against the coding region of a transiently expressed green fluorescent protein (GFP) mRNA (22). Thus, the transcription-independent mechanism shared by AL2 and L2, which correlates with ADK inactivation by these proteins, is sufficient to suppress local PTGS.

An assay using N. benthamiana line 16c plants, which contain a GFP transgene driven by the constitutive 35S promoter (35S-GFP) (28), was adopted to determine whether the viral proteins could also interfere with systemic spread of PTGS. Lower leaves of line 16c plants were coagroinfiltrated with a construct expressing inverted repeat RNA corresponding to the GFP coding region (dsGFP) to induce systemic PTGS and with test or control constructs. The p19 protein of Cymbidium ringspot virus is a strong suppressor of systemic silencing and served as a positive control (27). β-Glucuronidase (GUS) was a negative control. Test constructs expressed AL2, AL2_1–114, AL2 with a C33A substitution (AL2-C33A), or L2. AL2_1–114 lacks the minimal transcription activation domain and cannot activate transcription (34). AL2-C33A is defective in self-interaction and has reduced transcription activation activity (14). Both mutant proteins retain the ability to interact with ADK. The AL2, AL2_1–114, AL2-C33A, and L2 constructs used in these experiments were previously shown to inhibit the establishment of PTGS (14, 22).

When lower leaves of N. benthamiana 16c plants were coinfiltrated with plasmids expressing dsGFP (silencing trigger) and GUS (negative control), systemic silencing of the GFP transgene was first visible under UV light after ∼10 days on the crown leaves as dark red areas against the salmon-colored GFP-expressing tissue, which then spread downward to the remainder of the plant (Fig. 1). Dark red areas due to chlorophyll autofluorescence were apparent in more than 80% of the plants tested, while on average about 20% of the plants escaped silencing. In contrast, when a plasmid expressing the p19 positive control was coagroinfiltrated with dsGFP, none of the test plants showed evidence of systemic silencing. Despite its relatively weak activity in local PTGS assays (22), AL2 proved nearly as effective as p19 at inhibiting systemic silencing, with greater than 90% nonsilenced plants in three independent experiments. This is consistent with a previous study, which used a different assay to show that AC2/AL2 protein from Mungbean yellow mosaic virus interferes with silencing spread (18). In contrast, AL2_1–114 and L2, which lack transcription activation activity, did not prevent systemic silencing and were essentially equivalent to the GUS negative control. These observations indicate that the transcription-independent suppression mechanism is insufficient to block systemic silencing in N. benthamiana, and the intermediate suppression activity displayed by AL2-C33A bolsters the argument that transcription activation is required (Fig. 1). We concluded that AL2, but not L2, inhibits systemic spread of silencing by a transcription-dependent mechanism.

FIG 1.

AL2, but not L2, blocks systemic silencing. N. benthamiana line 16c plants containing a 35S-GFP transgene were coagroinfiltrated on lower leaves with a construct expressing dsRNA corresponding to GFP (dsGFP) to trigger silencing and with control or test constructs, as indicated. The graph depicts the percentage of plants that failed to show systemic silencing following administration of dsGFP. Numbers above the bars indicate the number of plants of the total tested (in two to three independent experiments with 4 to 12 plants per treatment) that showed no evidence of systemic GFP silencing after 4 weeks. Error bars represent the range between individual experiments. The bottom panel illustrates the progression of systemic silencing, from nonsilenced to a fully silenced leaf. Silencing is first evident on upper leaves (shown) and later spreads to lower leaves.

Host developmental stage can influence whether transcription-dependent or -independent mechanisms are able to reverse PTGS.

Only a relatively few silencing suppressors, including HC-Pro and AL2, have been shown to reverse established PTGS (35–39). In preliminary experiments to compare AL2 and L2 activities, we serendipitously observed that the ability of L2 to reverse silencing depended on whether the N. benthamiana test plants had produced bolts (floral shoots) that signal the onset of flowering. This was surprising as studies in Arabidopsis indicated that AL2 and L2 suppression activities are not sensitive to developmental stage in this host (24). Thus, AL2 and L2 PTGS reversal activities were studied in N. benthamiana line 16c plants both before and after the vegetative-to-reproductive transition. To do this, plants were grown in small (6-cm) pots, and lower leaves were infiltrated 2 weeks postgermination with the dsGFP construct to provoke systemic PTGS. Plants that failed to show systemic silencing were eliminated from the study. Two to 3 weeks later, the silenced plants were divided into two groups. Reproductive plants were transferred to fresh small pots (6 cm) and quickly flowered. Vegetative plants were moved into larger (16.5-cm) pots and resumed vegetative growth. Thus, the silenced test plants were the same age but differed with respect to the vegetative-to-reproductive transition at the time of analysis. Plants in both groups were then inoculated with potato virus X (PVX) vectors expressing AL2, AL2_1–114, or L2 (24). PVX (empty vector) served as a negative control, and PVX expressing HC-Pro (PVX::HC-Pro) served as a positive control. Four weeks postinoculation, plants were observed under UV light, and samples were collected for GFP mRNA analysis. Samples consisted of systemically infected, symptomatic leaf and stem tissue pooled from three to four plants.

Before describing the results of these experiments, a few words concerning the relationship between early flowering and stress are in order. We recognize that continued growth in small pots constitutes a stress that triggers flowering and that stress has a multitude of effects on plant physiology. However, determining which of these effects or combinations of effects impact the silencing phenomena under study is beyond the scope of this communication. As the appearance of floral bolts coincides with the changes in silencing phenotypes we observed, we have chosen to use the terms “vegetative” and “reproductive” plants as a matter of convenience.

GFP fluorescence indicating silencing reversal was not reliably detected by visual inspection of whole plants, even with the PVX::HC-Pro positive control. Thus, we examined water-mounted tissue segments from systemically infected leaves using a confocal microscope, and under these conditions yellow to green GFP-expressing cells could easily be seen against a background of red cells in which the GFP transgene remained silenced (Fig. 2A). A total of three experiments were carried out with vegetative and reproductive plants, and one additional experiment was performed with only reproductive plants. Each experiment included four plants per treatment. Leaves showing obvious PVX symptoms and three to four leaves below the crown were harvested, and one or two tissue segments were taken from each. A segment was considered positive for reversal if a field of view from the section showed GFP expression.

FIG 2.

The ability of L2 to reverse established PTGS depends on host developmental status. (A) Systemically silenced N. benthamiana line 16c plants were inoculated with a PVX vector or PVX expressing the indicated viral protein. Leaf sections were observed at 4 weeks postinoculation under a confocal microscope at a magnification of ×20. Representative sections are shown. Deep red indicates chlorophyll autofluorescence in the absence of GFP, and yellow to green fluorescence indicates cells expressing GFP. The images are representative of at least three independent experiments with four plants per treatment. Four to eight tissue sections were examined per plant. The top panel shows sections from nonsilenced and silenced 16c plants for comparison and from a silenced control plant inoculated with PVX, which showed expression only in guard cells. The lower panels show sections collected from silenced vegetative or reproductive plants. (B) GFP mRNA analysis. Shown are representative Northern blots of RNA (three independent experiments) from vegetative and reproductive 16c plants, inoculated as indicated. The ³²P-labeled probe was specific for GFP mRNA. The 18S rRNA loading controls were visualized by staining the gel with ethidium bromide.

Curiously, the empty PVX vector reproducibly elicited silencing reversal in stomatal guard cells of both reproductive and vegetative plants, suggesting that PVX affects silencing maintenance in this cell type. However, infection of vegetative plants with PVX::HC-Pro, PVX::AL2, PVX::AL2_1–114, and PVX::L2 resulted in more extensive reversal of GFP silencing that included epidermal and mesophyll cells (Fig. 2A). PVX::HC-Pro elicited yellow-green GFP fluorescence in isolated patches of cells, whereas PVX vectors expressing the geminivirus proteins induced GFP expression in a more evenly dispersed pattern. Silencing reversal was apparent in >75% to 100% of the tissue segments examined in all experiments with all recombinant viruses, with the exception of one experiment in which PVX::L2 was positive in ∼50% of the segments. Northern blot analysis of GFP mRNA levels confirmed these observations (Fig. 2B).

In reproductive plants, PVX::HC-Pro and PVX::AL2 elicited GFP expression that was similar to that of vegetative plants (Fig. 2A). GFP expression was observed in >75% to 100% of the tissue segments examined in each of four experiments. However, GFP expression outside guard cells was never observed with PVX::AL2_1–114; it was observed in only one of four experiments with PVX::L2-inoculated plants, and in this case ∼50% of sections were positive. These observations were again confirmed by Northern blotting (Fig. 2B). We concluded that AL2 is capable of reversing silencing directed against the GFP coding region in both vegetative and reproductive N. benthamiana plants, whereas L2 and transcriptionally inactive AL2 can reverse silencing only in leaves of vegetative plants.

Again, it is worth noting that the loss of AL2_1–114 and L2 suppression activity corresponded with the onset of flowering since even plants in the vegetative group failed to show PTGS reversal with these proteins if the PVX vectors expressing them were administered after bolts appeared (data not shown).

It was concluded that in vegetative N. benthamiana plants, the transcription-independent mechanism is sufficient to reverse PTGS, while suppression in reproductive plants appears to require transcription activation. This, in turn, suggests that silencing-maintenance mechanisms in N. benthamiana can change during development or in response to stress.

Host developmental stage can also influence whether geminiviruses can reverse TGS.

In an earlier study, we created an N. benthamiana line (16-TGS, derived from line 16c) containing a transcriptionally silenced 35S-GFP transgene. Because TGS is heritable, 16-TGS seedlings emerge with GFP fully silenced and appear red under UV light. We previously used this line to show that CaLCuV, BCTV, and the viral AL2 and L2 proteins can reverse established TGS (24). These experiments were carried out using vegetative plants, and here we repeated them to compare suppression activities in reproductive plants. A BCTV L2⁻ mutant virus, which contains a stop codon that truncates the L2 protein after 72 (of 173) amino acids (L2-2 mutation), was used as a negative control (13). BCTV L2⁻ replicates and generates systemic symptoms that are similar to, albeit somewhat milder than, wild-type BCTV although plants eventually recover from infection with this mutant. AL2-deficient begomoviruses (including CaLCuV and TGMV) are not systemically infectious because AL2 is required to activate the expression of the BR1 movement protein (11).

For these experiments, 2-week-old seedlings of N. benthamiana line 16-TGS were screened for GFP silencing under UV light, and only plants showing uniform red chlorophyll autofluorescence were used. Seedlings were then divided to generate vegetative and reproductive populations as described above. Three independent experiments were conducted with eight vegetative and reproductive plants each. At 3 to 4 weeks of age, plants were infected with CaLCuV, BCTV, or BCTV L2⁻. A handheld UV light was used to screen for TGS reversal at 3 weeks postinfection, and plants were considered positive when GFP expression was observed in symptomatic tissue. Tissue was harvested for RNA, DNA, and ADK activity analysis, with samples consisting of symptomatic leaf and stem tissue pooled from three to four plants.

Consistent with previous results, more than 80% of vegetative plants infected with wild-type BCTV or CaLCuV exhibited GFP expression indicative of TGS reversal, whereas fewer than 20% of BCTV L2⁻-infected plants showed visual evidence of GFP expression (Fig. 3A) (2, 24). However, fewer than 20% of BCTV-infected reproductive plants showed visual evidence of GFP expression even though disease symptoms were comparable to those observed in vegetative plants. Consistent with these observations, GFP mRNA levels were relatively high in leaves from vegetative plants infected with BCTV, whereas levels in reproductive plants were relatively low and similar to those of plants infected with BCTV L2⁻ (Fig. 3B). In contrast, robust GFP expression was observed in both vegetative and reproductive plants infected with CaLCuV, indicating that this virus is able to reverse TGS at both developmental stages (Fig. 3A and B). Similar results were obtained in two experiments with TGMV (data not shown).

FIG 3.

Host developmental stage can determine whether geminiviruses reverse TGS. (A) Vegetative and reproductive transgenic N. benthamiana plants containing a transcriptionally silenced GFP transgene (line 16-TGS) were infected with the indicated geminiviruses and photographed under UV light at 3 weeks postinoculation. The dark red resulting from chlorophyll autofluorescence in the absence of GFP is apparent in plants infected with BCTV L2⁻ mutant virus. GFP expression and TGS reversal are indicated by yellow to green fluorescence in veins and mesophyll of vegetative plants infected with the wild-type viruses. CaLCuV continues to reverse TGS in reproductive plants, while BCTV cannot. Results shown are representative of three independent experiments with eight plants per treatment. (B) Northern blot analysis of GFP mRNA from line 16-TGS plants, inoculated as indicated. V and R indicate samples from vegetative and reproductive plants, respectively. The blot shown is representative of results observed in three independent experiments. The ³²P-labeled probe was specific for GFP mRNA, and 28S rRNA loading controls were visualized by staining the gel with ethidium bromide. (C) Southern blot analysis shows that viral DNA levels are similar in vegetative and reproductive plants. The blot shown is representative of three individual experiments. Oligonucleotide probes were specific for CaLCuV, BCTV, or the 18S rRNA gene (rDNA; loading control). Samples were cleaved with ScaI (BCTV) or XcmI (CaLCuV) to linearize circular viral dsDNA. These enzymes produce 18S rRNA gene fragments of different sizes. (D) BCTV inhibits ADK activity in vegetative and reproductive plants in an L2-dependent manner. Total protein extracts were obtained from mock-, BCTV-, and BCTV L2⁻-infected plants and analyzed for ADK activity using thin-layer chromatography (TLC). The graph represents the averaged results of three individual experiments ± standard errors. Radioactively labeled AMP resulting from phosphorylation of adenosine was quantitated, and the average level in mock extracts was set to 1.0. The right panel shows a representative TLC plate.

To rule out the possibility that differences in virus replication might account for differences in TGS reversal observed with BCTV, viral DNA levels were examined. DNA extracts were obtained from infected vegetative and reproductive plants, and viral dsDNA was linearized with ScaI (BCTV) or XcmI (CaLCuV). Southern blots were probed with oligonucleotides specific for the 18S rRNA gene (loading control) and the BCTV or CaLCuV intergenic region. Not surprisingly, viral DNA levels were lowest in plants infected with BCTV L2⁻ (Fig. 3C). However, for each virus, DNA levels in vegetative and reproductive tissues were similar, and thus the inability of BCTV to suppress silencing in reproductive tissue cannot be attributed to impaired replication.

We previously showed that ADK inhibition in infected plants is L2 dependent (17) and considered the possibility that L2 expression levels might be reduced in reproductive plants infected with wild-type BCTV. To indirectly test this, ADK activity levels were measured in protein extracts obtained from vegetative and reproductive plants infected with BCTV or BCTV L2⁻. The results of these experiments confirmed that ADK inhibition is L2 dependent and showed that wild-type BCTV is capable of suppressing ADK activity in both vegetative and reproductive plants (Fig. 3D). Thus, BCTV replication and L2 expression levels appear to be similar in both tissues. Taken together, these experiments suggest that, similar to PTGS, distinct mechanisms of TGS maintenance may be active during different stages of development.

A novel AL2 TGS suppression mechanism is revealed in reproductive N. benthamiana plants.

The experiments described in the previous section indicate that begomoviruses and curtoviruses differ in their abilities to suppress TGS in reproductive tissue. To confirm that this reflects the activities of their respective AL2 and L2 proteins, these were individually expressed from PVX vectors in line 16-TGS plants. We previously used this system to show that expression of AL2, AL2_1–114, AL2-C33A, or L2 reverses TGS in vegetative N. benthamiana plants (24), and here we extended the experiments to include reproductive plants. Experiments were performed in triplicate with eight vegetative and reproductive plants in each experiment. Interestingly, microscopy showed that the PVX vector control did not elicit GFP expression in guard cells of transcriptionally silenced 16-TGS plants (data not shown), as was the case in 16c plants systemically silenced by PTGS (Fig. 2).

In support of results obtained with the geminiviruses, PVX::AL2 reversed silencing in greater than 80% of both vegetative and reproductive 16-TGS plants, while TGS reversal by PVX::L2 was limited to vegetative plants (Fig. 4A and B). Again, it is important to note that the loss of L2 activity corresponded with the onset of flowering since plants in the vegetative group also failed to respond to L2 once floral bolts appeared (data not shown).

FIG 4.

A novel AL2-mediated, transcription-independent TGS suppression mechanism is uncovered in reproductive 16-TGS plants. (A) Vegetative and reproductive transgenic N. benthamiana plants containing a transcriptionally silenced GFP transgene (line 16-TGS) were infected with the PVX vector or PVX expressing the indicated viral protein and photographed under UV light 2 to 3 weeks postinoculation. The images shown are representative of three independent experiments with eight plants per treatment. Note that in addition to PVX::AL2, PVX::AL2_1–114 and PVX::AL2-C33A (but not PVX::L2) reversed TGS in reproductive tissue. These results indicate a suppression mechanism that is independent of transcription activation and is not shared by L2. (B) Northern blot analysis of GFP mRNA from line 16-TGS plants, inoculated as indicated, supports the visual data. The blot shown is representative of results observed in three independent experiments. The ³²P-labeled probe was specific for GFP mRNA, and 28S rRNA loading controls were visualized by ethidium bromide staining. Note that a relatively weak signal was observed with the PVX vector control in some cases. (C) Northern blot analysis shows that PVX titers are similar in vegetative and reproductive plants. The blot shown is representative of three independent experiments. The probe was specific for PVX RNA.

Surprisingly, however, the transcription activation-deficient AL2_1–114 and activation-impaired AL2-C33A proteins also reversed silencing in both vegetative and reproductive plants (Fig. 4A and B). PVX vector accumulation was similar in both types of plants regardless of the geminivirus protein expressed, suggesting comparable protein expression levels (Fig. 4C). We concluded that in vegetative plants, TGS reversal can occur in a transcription-independent manner by a mechanism that is shared by AL2 and L2 (ADK inhibition). But only AL2 and its transcription-defective derivatives can reverse TGS in reproductive N. benthamiana plants, in this case by a distinct mechanism that is not shared by L2. Thus, a third mechanism of AL2-mediated silencing suppression, which does not correlate with either ADK inhibition or transcription activation, was uncovered in this study.

TGS in reproductive N. benthamiana plants does not require ADK activity.

We showed, using a tobacco rattle virus (TRV)-based virus-induced gene silencing (VIGS) vector, that knockdown of the methyl cycle enzymes SAHH and ADK phenocopies geminivirus TGS reversal of a GFP transgene in vegetative plants (24). However, studies described in the previous section indicate that BCTV and PVX::L2 cannot reverse TGS in reproductive tissue even though the L2 protein produced by the virus continues to efficiently inhibit ADK activity after N. benthamiana plants have begun to flower (Fig. 3D). Therefore, TGS may become ADK independent in reproductive tissue. To explore this possibility, the knockdown experiments were repeated in reproductive plants.

Experiments using line 16-TGS were carried out as described above, except that plants were inoculated with TRV empty vector (negative control) or TRV containing fragments of the ADK or SAHH coding region. A fragment of methyltransferase 1 (MET1) was a positive control. MET1 is required for TGS maintenance while SAHH is an essential methyl cycle enzyme, and ADK enhances methyl cycle activity. A handheld UV light was used to monitor GFP expression, and tissue was harvested for RNA and ADK activity analysis. At least eight plants from both the vegetative and reproductive populations were examined in three independent experiments.

Fewer than 10% of the vegetative and reproductive plants infected with TRV empty vector showed any evidence of GFP expression. As observed previously, TRV::ADK, TRV::SAHH, and TRV::MET1 infection resulted in TGS reversal in nearly all of the vegetative plants (24). In contrast, only TRV::SAHH and TRV::MET1 treatments showed yellow-green fluorescence in nearly all of the reproductive plants (Fig. 5A). These observations indicate that SAHH activity and the methyl cycle are necessary to support TGS in both vegetative and reproductive plants. However, the requirement for ADK is conditional as ADK knockdown is not sufficient to reverse TGS in reproductive plants. Analysis of GFP mRNA accumulation by Northern blotting hybridization supported visual observations (Fig. 5B). Additionally, oligonucleotide probes complementary to the TRV coat protein gene showed that, for the most part, vector replication levels were similar in vegetative and reproductive tissues although TRV::ADK levels were somewhat lower than those of TRV::SAHH and TRV::MET1 (Fig. 5C).

FIG 5.

TGS is independent of ADK activity in reproductive N. benthamiana plants. (A) Vegetative and reproductive transgenic N. benthamiana plants containing a transcriptionally silenced GFP transgene (line 16-TGS) were infected with the TRV vector (control), TRV::ADK, TRV::SAHH, or TRV::MET1 to knock down expression of the cognate mRNA. Plants were photographed under UV light at 3 weeks postinoculation. Yellow to green fluorescence indicates TGS reversal. Images are representative of three independent experiments with eight plants per treatment. (B) Northern blot analysis of GFP mRNA from line 16-TGS plants, inoculated as indicated. V and R indicate samples from vegetative or reproductive plants, respectively. The blot shown is representative of three independent experiments. The ³²P-labeled probe was specific for GFP mRNA, and 28S rRNA controls were visualized by ethidium bromide staining. (C) Northern blot analysis with a TRV-specific probe shows similar TRV vector (pTV00) titers in vegetative and reproductive plants. The blot is representative of three independent experiments. (D) TRV::ADK inhibits ADK activity in vegetative and reproductive plants. Protein extracts were obtained from mock-, TRV-, and TRV::ADK-infected plants and analyzed for ADK activity using TLC. The graph illustrates averaged results from three individual experiments performed in duplicate ± standard errors. Labeled AMP resulting from phosphorylation of adenosine was quantitated, and the average level in mock extracts was set to 1.0. The right panel shows a representative TLC plate.

Finally, to ensure that the loss of TGS reversal in reproductive tissue was not due to insufficient ADK knockdown, an ADK activity assay was employed. ADK assays were conducted with the total protein extracts obtained from mock-inoculated and TRV- and TRV::ADK-infected vegetative and reproductive plants, and activity was normalized to that of the mock samples. We found that ADK activity levels were similar in extracts from mock-inoculated vegetative and reproductive plants and that activity was stimulated by TRV infection (Fig. 5D). Increased ADK activity was previously observed following infection of N. benthamiana plants with PVX and cucumber mosaic virus (CMV) and, to a lesser extent, with BCTV L2⁻ (Fig. 3D) (17). Nevertheless, TRV::ADK reduced ADK activity to between 30% and 60% of levels seen in mock-inoculated reproductive and vegetative plants, respectively (Fig. 5D). These results allow us to conclude that the TRV::ADK construct was effective in both vegetative and reproductive plants and that (in contrast to Arabidopsis) ADK activity is no longer needed to support silencing maintenance in reproductive N. benthamiana plants.

ADK inhibition inhibits global cytosine methylation in vegetative but not reproductive plants.

Because TGS is linked to DNA methylation, a methylation-sensitive cytosine extension assay was used to interrogate the status of N. benthamiana genomic DNA following geminivirus infection and ADK or SAHH knockdown (24, 26). These experiments rely on the inability of MspI to cleave its target site (C/CGG) when the external cytosine is methylated. Following digestion of total genomic DNA with MspI, a single-nucleotide extension assay was performed using [³²P]dCTP and Taq DNA polymerase. Under these conditions, nucleotide incorporation depends on the number of sites cleaved, which negatively correlates with methylation at CNG sites.

DNA was isolated from mock-, BCTV-, BCTV L2⁻-, and CaLCuV-infected vegetative and reproductive N. benthamiana plants and also from plants inoculated with TRV, TRV::ADK, and TRV::SAHH VIGS vectors. At least three individual plants were sampled, and duplicate extension reactions were performed with each sample.

In vegetative plants, BCTV L2⁻ infection had no significant impact on cellular DNA methylation levels compared to mock-inoculated plants, whereas increased incorporation (indicating decreased methylation) was observed in extracts from plants infected with BCTV (>1.5-fold) and CaLCuV (>2-fold), as expected. However, only CaLCuV infection resulted in increased cytosine incorporation in samples from reproductive plants (>3-fold) (Fig. 6A). Again, this contrasts with previous studies in Arabidopsis, where expression of AL2 or L2 in reproductive plants resulted in increased cytosine incorporation (24).

FIG 6.

Reducing ADK expression does not affect global cytosine methylation in reproductive N. benthamiana plants. (A) The histograms illustrate relative incorporation of cytosine observed in methylation-sensitive extension assays. DNA was obtained from vegetative or reproductive N. benthamiana plants infected with geminiviruses, as indicated. (B) The same experiment as in panel A, except that plants were infected with the indicated TRV VIGS vectors. BCTV L2⁻ and TRV served as negative controls. In all cases, DNA was digested to completion with MspI and incubated with [³²P]dCTP and Taq DNA polymerase to allow single-nucleotide extension. Increased incorporation reflects enhanced MspI cleavage due to reduced methylation. At least three individual plants were sampled per treatment, and duplicate extension reactions were performed with each sample. Averaged values (± standard errors) were normalized to mock controls. Asterisks indicate significant differences at the 95% (*) or 99% (**) confidence level, as determined by Student's t test.

Cytosine extension assays also showed no apparent difference between mock samples and samples from plants inoculated with the TRV vector, while increases in incorporation were noted in vegetative plants infected with TRV::ADK (>1.2-fold) and TRV::SAHH (>1.5-fold). However, only TRV::SAHH treatment was able to increase incorporation in reproductive plants (>1.5-fold) (Fig. 6B). These results provide further evidence that suppression mechanisms involving ADK inhibition are effective in vegetative, but not reproductive, N. benthamiana plants and suggest that the role of ADK in sustaining the methyl cycle is developmental stage dependent in this species.

DISCUSSION

Previous studies demonstrated that AL2 could suppress PTGS establishment and spread by a mechanism that depends on the presence of its transcription activation domain and involves upregulation of cellular genes (transcription-dependent suppression), while both AL2 and L2 could also suppress PTGS establishment and TGS by a mechanism that correlates with inactivation of ADK and consequent inhibition of methyl cycle activity (transcription-independent suppression) (18, 22, 24). The studies described here confirm and extend our knowledge of AL2 and L2 activities and suggest a conditional role for ADK in the N. benthamiana methyl cycle.

First, we confirmed that AL2 effectively prevents systemic spread of silencing while a mutant protein lacking the activation domain cannot. The intermediate suppression observed here with AL2-C33A, which has reduced transcription activation activity, verifies that this type of suppression requires transcription activation and not merely the presence of the AL2 activation domain, which could (for example) also interact with factors not involved in transcription. Further study is needed to determine whether AL2 inhibits the generation, spread, or perception of the silencing signal. Additionally, we found that the BCTV L2 protein failed to impede systemic silencing, suggesting that spread does not require methyl cycle activity. This observation may also suggest that, unlike begomoviruses, curtoviruses are unable to effectively inhibit systemic silencing. Alternatively, as systemic spread of silencing is generally considered to be an important component of antiviral defense, it is possible that another BCTV protein(s) is able to inhibit this aspect of silencing.

AL2 was previously shown to reverse PTGS (38), and here we demonstrated that AL2, AL2 lacking the transcription activation domain, and L2 could reverse silencing in vegetative plants, indicating that a transcription-independent mechanism is sufficient to interfere with silencing maintenance. However, only AL2 could reverse PTGS in N. benthamiana plants after the onset of flowering, indicating that a transcription-dependent activity is required and also suggesting that maintenance mechanisms change with the vegetative-to-reproductive transition.

These ideas were subsequently tested in a more extensive set of experiments using vegetative and reproductive N. benthamiana 16-TGS plants containing a transcriptionally silenced GFP transgene. Again, while CaLCuV, BCTV, and their AL2 or L2 proteins reversed TGS in vegetative plants (24), only CaLCuV and its AL2 protein could do this in reproductive plants. However, in contrast to PTGS reversal, AL2_1–114 lacking the transcription activation domain also effectively reversed silencing in reproductive plants, thereby identifying a third, previously unrecognized suppression activity that is not shared by L2 and does not depend on transcription activation. That AL2_1–114 is able to reverse TGS but not PTGS in reproductive plants is surprising and suggests AL2 might interact with or otherwise adversely affect the activity of a methylation pathway component.

Since the shared, transcription-independent mechanism of TGS reversal by AL2 and L2 correlates with ADK inactivation, it was important to carefully evaluate the role of methyl cycle inhibition in relation to silencing suppression in vegetative and reproductive 16-TGS plants. SAHH is an essential methyl cycle enzyme, while ADK promotes flux through the pathway. We found that although BCTV infection and ADK knockdown (using TRV::ADK) significantly reduced ADK activity levels in both vegetative and reproductive plants, TGS reversal was observed only in vegetative plants. Thus, TGS and ADK inhibition are uncoupled in reproductive N. benthamiana plants. Methyl cycle activity nevertheless remains essential to support and maintain TGS as SAHH knockdown (using TRV::SAHH) effectively reversed silencing in both vegetative and reproductive plants. These findings were supported by methylation-sensitive cytosine extension assays, which showed that BCTV and TRV::ADK caused global reductions in cytosine methylation in genomic DNA from vegetative but not reproductive plants, whereas CaLCuV and TRV::SAHH infection significantly decreased methylation regardless of developmental status. Thus, the methyl cycle appears to function without ADK in reproductive N. benthamiana plants, suggesting that an alternative enzyme is recruited to remove adenosine and maintain methyl cycle activity after the vegetative-to-reproductive transition in this species.

Most methyltransferase enzymes, including cytosine and histone methyltransferases, utilize SAM as a methyl group donor. Methyl group transfer generates S-adenosyl homocysteine (SAH), a potent methyltransferase inhibitor. SAHH reversibly hydrolyzes SAH to homocysteine and adenosine, and removal of adenosine is necessary to shift the equilibrium in favor of hydrolysis (40, 41). In plant species including Arabidopsis thaliana, Spinacia oleracea, and Beta vulgaris, ADK is primarily responsible for adenosine salvage and methyl cycle maintenance at all stages of development and also in inflorescence tissues (21, 42–44). This is consistent with earlier work which showed that L2 expression and ADK inhibition cause TGS reversal in both vegetative and reproductive Arabidopsis plants (24).

Because the methyl cycle appears to be ADK independent in reproductive N. benthamiana, adenosine must be salvaged by an alternative pathway for the cycle to continue to function. Although three enzymes, adenosine deaminase (ADA), adenosine nucleosidase (AN), and ADK, can potentially remove adenosine, only ADK and AN activities have been detected in plant extracts (41, 45, 46). While ADK is believed to carry out adenosine recycling in most plants, elevated levels of AN activity in some species, including Camellia sinensis (tea) and Avicenna marina (mangrove shrub), suggest that this enzyme can also contribute to adenosine salvage (47, 48). Thus, it is reasonable to suspect that AN could play a similar role in reproductive N. benthamiana. Since we found ADK activity levels to be similar in vegetative and reproductive plants, there appears to be no a priori reason for a switch from ADK to another enzyme. Thus, we speculate that AN activity might be elevated in reproductive N. benthamiana, providing an alternative and redundant means of supporting the methyl cycle.

To summarize, AL2 is a remarkably versatile silencing suppressor that employs both transcription-dependent and -independent mechanisms to block the establishment of silencing as well as its subsequent systemic spread and to reverse established PTGS and TGS. BCTV L2 appears limited to a transcription-independent suppression mechanism that correlates with ADK inhibition, which it shares with AL2. This mechanism cannot inhibit silencing spread but is sufficient to block the establishment of PTGS and to reverse PTGS and TGS in most plants. However, in some species, including reproductive N. benthamiana, this suppression mechanism is not effective, and an alternative enzyme is apparently recruited to replace or supplement ADK activity and support the methyl cycle. This unique situation allowed us to uncover a novel AL2 suppression activity: TGS reversal mediated by transcription activation-defective AL2_1–114 and AL2-C33A in reproductive N. benthamiana plants, where L2 is inactive, revealed a mechanism that depends on neither transcription activation nor ADK inactivation.

In addition to regulating the expression of endogenous genes and invasive DNAs, the several aspects of RNA silencing (i.e., PTGS, TGS, and systemic spread) act as a multivalent antiviral defense system to which viruses have responded by elaborating silencing suppressors. That silencing pathways and/or support systems could change during development or in response to stress, as illustrated here, adds an additional layer of complexity to host defense and viral counterdefense. The varied and potentially changeable nature of silencing no doubt explains why some viruses encode multiple suppressors and why some of them act by multiple mechanisms. Our findings also encourage caution: as N. benthamiana is widely used to study viral silencing suppressors, investigators should consider evaluating their activities at different stages of development, under stress and nonstress conditions, and in natural hosts where possible.

In conclusion, these studies have clarified differences between AL2 and L2 silencing-suppression activities, identified a new transcription-independent mechanism of AL2 suppression, and provided evidence that the importance of ADK in supporting the methyl cycle can change during development in at least some plant species. The stage is now set for confirming and determining the molecular mechanisms of AL2-mediated suppression. Much has been and will be learned about host silencing and methylation pathways as we further unravel the mechanisms by which they are suppressed by geminivirus proteins.