Modification of Small RNAs Associated with Suppression of RNA Silencing by Tobamovirus Replicase Protein

Abstract

Plant viruses act as triggers and targets of RNA silencing and have evolved proteins to suppress this plant defense response during infection. Although Tobacco mosaic tobamovirus (TMV) triggers the production of virus-specific small interfering RNAs (siRNAs), this does not lead to efficient silencing of TMV nor is a TMV-green fluorescent protein (GFP) hybrid able to induce silencing of a GFP-transgene in Nicotiana benthamiana, indicating that a TMV silencing suppressor is active and acts downstream of siRNA production. On the other hand, TMV-GFP is unable to spread into cells in which GFP silencing is established, suggesting that the viral silencing suppressor cannot revert silencing that is already established. Although previous evidence indicates that the tobamovirus silencing suppressing activity resides in the viral 126-kDa small replicase subunit, the mechanism of silencing suppression by this virus family is not known. Here, we connect the silencing suppressing activity of this protein with our previous finding that Oilseed rape mosaic tobamovirus infection leads to interference with HEN1-mediated methylation of siRNA and micro-RNA (miRNA). We demonstrate that TMV infection similarly leads to interference with HEN1-mediated methylation of small RNAs and that this interference and the formation of virus-induced disease symptoms are linked to the silencing suppressor activity of the 126-kDa protein. Moreover, we show that also Turnip crinkle virus interferes with the methylation of siRNA but, in contrast to tobamoviruses, not with the methylation of miRNA.

RNA silencing is a posttranscriptional, RNA-guided, gene regulatory mechanism that operates through RNA-mediated sequence-specific interactions in the cytoplasm of eukaryotes, including plants (5, 47, 57).

RNA silencing is generally induced by double-stranded RNA (dsRNA), which can originate from various sources, such as transgenes, viral replication intermediates, or experimentally introduced dsRNA sequences. Central to the silencing process are dicers or “dicer-like” enzymes that cleave dsRNA into small double-stranded fragments, called small interfering RNAs (siRNAs). Single-stranded siRNAs are then incorporated into multicomponent RNA-induced silencing complexes (RISC), which contain an “argonaute” (AGO) family protein (in plants usually AGO1) (3) and inactivate homologous RNA through endonucleolytic cleavage. In addition to siRNAs, which are usually derived from foreign elements such as transgenes and viruses, other small RNA (sRNA) species are encoded by specific noncoding RNA genes. Among these, micro-RNAs (miRNAs) have predominant roles during plant development (28) and are processed from miRNA precursors encoded by miRNA genes. Similarly to siRNAs, miRNAs are incorporated into AGO-containing RISC complexes to guide the recognition of target RNAs. In plants, miRNA-RISC complexes usually cause target RNA cleavage, whereas in most mammalian cases miRNA-RISC inhibits translation of target mRNA (37). Plant siRNAs and miRNAs (commonly referred to as sRNAs) are predominantly seen as 21- and 24-nucleotide (nt) bands (1, 13, 53). Double-stranded sRNAs have 2-base-long 3′ overhangs and are phosphorylated at the 5′ end. Moreover, they are methylated at the 2′-OH group of the 3′-ribose by the activity of HEN1, which thus contributes to the stability and availability of sRNAs (1, 12, 54, 56).

A remarkable feature of RNA silencing in plants, as well as in Caenorhabditis elegans (44) and planaria (33), is its ability to act beyond the cells in which it is initiated (48). The signal must be a nucleic acid because it mediates a nucleotide sequence-specific effect. Although non-cell-autonomous silencing signaling is correlated with the production of siRNAs (10, 16), the signal might also include other RNA species (for a review, see reference 31). The non-cell-autonomous spread of silencing signals may provide an effective means in the defense of plants against spreading viruses. In fact, if viral RNA-derived signals spread and the silencing condition is established ahead of a viral infection, viral RNAs are degraded in newly infected cells prior to replication (49).

Viruses counteract silencing by evasion, e.g., by minimizing production and exposure of dsRNA, as well as by suppression, i.e., by interfering with the silencing pathway through the expression of silencing suppressors (46). More than 20 RNAi suppressors have been identified in plant viruses (42, 46). These suppressors are commonly involved in the enhancement of viral pathogenicity and the accumulation of viruses. They are usually multifunctional proteins, serving also other functions during the viral life cycle. They have nothing obvious in common, with the exception that most of them have single-stranded RNA (ssRNA) and/or dsRNA binding activity (23, 29, 42). Moreover, many of them, either in viral infections or as a transgene, cause plant abnormalities and developmental defects, which are likely to be caused by interference of these proteins not only with the siRNA pathway but also with the miRNA pathway (51). Silencing suppression can either lead to a decrease (27) or an increase in siRNAs (7), indicating that in the first case siRNA production and in the second case siRNA usage is inhibited. In principle, these proteins could interfere with any step in the silencing pathway. For example, by binding to ssRNA, suppressors could inhibit dsRNA formation; by binding to large dsRNA, they could inhibit its degradation by dicers (38); and by binding to siRNA duplexes, they could inhibit the formation of RISC. The latter case was exemplified by the isolation and detailed characterization of the tombusvirus p19 complex with double-stranded siRNA (24, 43). This complex is very specific with respect to the size of the siRNAs, preferring 21-nt siRNA duplexes. Similar complexes involving potyviral Hc-Pro or Beet yellows virus p21 are also specific for the presence of the 3′ 2-nt overhangs (23). Silencing suppressors could also act by interactions with host proteins that are important components of the silencing machinery. HcPro recruits calmodulin-related protein (rgsCaM) from host plants, and this protein, if overexpressed, mimics suppression by HcPro (2). More recently, the Cucumber mosaic virus 2b has been reported to interact with AGO1 leading to inhibition of its slicer activity (58), and the polerovirus P0 protein appears to act as an F-box protein that targets an essential component of the silencing machinery for degradation (36). Silencing suppressors may also act through manipulation of the host transcriptome. Here, one example is the geminivirus AC2 protein, which strongly enhances the transcription of several host genes, at least one of which, WEL1, acts as a silencing suppressor itself (45). Finally, some suppressors, such as HcPro (12) and an up to now unknown factor of Oilseed rape mosaic virus (ORMV) (1), inhibit not only the usage but also the HEN1-mediated methylation of sRNAs. Whether this is due to inhibition of the enzyme or due to masking of the substrate in the sRNA-suppressor complex is not yet known.

TMV infection leads to the accumulation of TMV-specific siRNAs, indicating that the plant recognizes TMV as a target for silencing (32). That TMV nevertheless is able to propagate in infected plants is likely due to the expression of a silencing suppressing activity, which resides in the 126-kDa (126k) replicase protein of the virus (9). A role of this protein in silencing suppression is also indicated by studies using the corresponding small (130k) replicase subunit of the TMV-related Tomato mosaic virus (ToMV) (22).

The 126k protein is one of the two replicase subunits translated from the TMV genomic RNA. Whereas this protein is produced upon terminating translation at an amber codon, the larger (183k) subunit results from read-through of this codon.

Here we show that the lack of methylation of small RNA species, previously shown for ORMV (1), and the suppressing activity of the 126k protein, previously shown for TMV (9) and ToMV (22), are linked processes, thus implying an active role for the 126k protein in silencing suppression through either inhibition of HEN1-mediated sRNA methylation or sRNA demethylation.

MATERIALS AND METHODS

Plant material.

Wild-type Nicotiana benthamiana plants and the transgenic N. benthamiana line 16c (40) were grown from seeds and maintained in ca. 70% relative humidity at 23°C during a 16-h photoperiod. Three- to five-week-old plants were used for either infiltration assays or inoculation experiments.

N. tabacum cv. Xanthi nn and N. tabacum cv. Xanthi NN plants were grown under the same conditions as N. benthamiana. Three-week-old plants were used for inoculation assays.

Arabidopsis thaliana (Col-0) plants were grown from seeds in soil in a growth chamber (Sanyo) at 20°C with a 12-h photoperiod.

DNA constructs.

The TMV silencing suppressor mutant was created by introducing an A-to-G nucleotide exchange mutation at position 1114 of the TMV genome using the QuikChange site-directed mutagenesis kit (Stratagene) and specific forward (5′-GACTCTTGCAATGTACAACAGCGAGAGAATCCTC-3′) and reverse (5′-GAGGATTCTCTCGCTGTTGTACATTGCAAGAGTC-3′) primers. This mutation causes replacement of the cytosine residue at position 349 of the 126k protein with a tyrosine residue corresponding to the reported silencing suppressor mutation in ToMV (22).

TMV-green fluorescent protein (GFP) is identical to TMVΔC-GFP (17) and expresses GFP instead of coat protein from the coat protein subgenomic promoter.

Silencing of GFP in 16c plants was induced by infiltration of leaves with Agrobacterium tumefaciens strain GV3101 transformed with p35S:GFP, a pBIN construct that expresses GFP under the control of the Cauliflower mosaic virus 35S promoter.

For transient expression of 126k protein by agroinfiltration, the 126k open reading frame was PCR amplified from TMV-encoding plasmid pU3-12/4 (19) and cloned into plasmid pG35Somega, a derivative of binary vector pGREENII0029 (18), into which an expression cassette under the control of the Cauliflower mosaic virus 35S promoter and the TMV omega leader (a translation enhancer) had been inserted. The resulting construct, pG126kD, was also used for site-directed mutagenesis as described above, yielding plasmid pG126kDmut. pBin-HcPro, a pBin61 derivative harboring the Hc-Pro coding region of Potato virus Y, was obtained from D. C. Baulcombe.

Virus inoculation and agroinfiltration.

N. benthamiana and N. tabacum plants were mechanically inoculated (in the presence of carborundum) with TMV constructs by using infectious transcripts prepared by in vitro reactions (MEGAscript T7 kit; Ambion). Arabidopsis and N. tabacum plants were mechanically inoculated with ORMV by using either isolated virions or tissue extract isolated from infected N. benthamiana plants. Arabidopsis plants were inoculated with Turnip crinkle virus (TCV) by agroinfiltration using bacteria harboring TCV-encoding plasmid (8) (kindly provided by O. Voinnet).

For agroinfiltration we used the method of Voinnet (50). Briefly, bacteria were grown in 50 ml of LB medium containing 50 μl/ml kanamycin and 20 μM acetosyringone at 28°C for 24 to 36 h. Cells were harvested by centrifugation and resuspended in 10 mM MgCl₂, 10 mM MES (morpholineethanesulfonic acid), and 100 μM acetosyringone to reach an optical density at 595 nm of 0.5. Before infiltration the cells were incubated for 3 h to overnight at room temperature. For silencing suppressor activity assays, cells containing the silencing inducer construct p35S:GFP were mixed in a ratio of 1:1 with cells harboring the test constructs. The final Agrobacterium concentration was kept constant at an optical density at 595 nm of 0.5 throughout the experiments. The mixtures were coinfiltrated into leaves of 3- to 4-week-old plants. Noncoding plasmid pG35Somega was used as a negative control in all agroinfiltration experiments.

Northern blot analysis and β-elimination assay.

Total RNA was extracted from 1 g of plant tissue ground in liquid nitrogen, using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Electrophoresis and detection of small RNA species, as well as the treatment of RNA samples for β elimination, were performed as described previously (1). The blot hybridization was performed using, as a probe, one or several short DNA oligonucleotides (Table 1) end labeled with ³²P by polynucleotide kinase.

TABLE 1.

Probes for RNA blot hybridization

Probea	Sequence (5′-3′)
miR157 as	GTG CTC TCT ATC TTC TGT CAA
miR160 as	TGG CAT ACA GGG AGC CAG GCA
miR166 as	GGG GAA TGA AGC CTG GTC CGA
miR172 as	ATG CAG CAT CAT CAA GAT TCT
tmv1 s	AGT TCG TGT TCT TGT CAT CAG CGT GGG
tmv2 s	TTA TAA TCG GAG CTC TTT CGA GAG CTC
ormv2 s	TAA CTA AAA GTG AGA GGT TCG AAT CCT
ormv3 s	ATC ACC TGT TAA CGT ACG CGT GGC GTA
ormv4 s	TTA GAT GAG GCC GTT GCC GAG GTC CAT
tcv1 s	ACG ACC TCG CTT CCC TCT ACA ACA TAG AGG
tcv2 s	CTT ATC GGT CCT GGG AGA AGC CGC AGC AGG
tcv3 s	GAT GGG GAT CAA GCT CTC TCC TGT GGC GCA

^aas, antisense; s, sense.

Separation and hybridization of long RNA was carried out under standard conditions (41). Probes for the detection of long RNA molecules were created by PCR and subsequent random labeling with ³²P using the Rediprime II DNA Labeling System (Amersham). The radioactive signal was detected by exposure to a phosphor screen using a Molecular Imager (Bio-Rad). Signal intensities were quantified by using Quantity One 1-D analysis software (version 4.4.1; Bio-Rad).

Preparation of protein extracts and Western blot analysis.

Tobacco leaves were ground into a fine powder under liquid nitrogen and homogenized in 1 ml/g of 1× phosphate-buffered saline (PBS) buffer containing one tablet of protease inhibitor cocktail (Roche)/10 ml. Cell debris was removed by centrifugation at 15,000 × g at 4°C for 30 min. The supernatant was collected and analyzed by Western blotting with an affinity-purified rabbit antibody raised against a synthetic peptide spanning amino acids 151 to 165 of TMV replicase and ECL Western blot reagent (Amersham) for signal detection.

RESULTS

Spread of TMV-GFP in GFP silenced and nonsilenced tissues of GFP transgenic plants.

To investigate the activity of the TMV 126k protein as silencing suppressor, we inoculated leaves of GFP-transgenic plants with a derivative of TMV expressing GFP in place of its coat protein (TMVΔC-GFP; here referred to as TMV-GFP) (17) (Fig. 1). The virus produced highly fluorescent and enlarged infection sites (Fig. 1A and B), showing that plant-encoded GFP expression interferes neither with TMV-GFP replication and spread nor with virus-encoded GFP expression in the inoculated leaf. In contrast, in fully silenced GFP-transgenic plants no infection sites could be observed (Fig. 1C), indicating that in cells with established GFP silencing the silencing suppressor is unable to prevent the degradation of the GFP-containing viral RNA genome. When similar fully GFP-silenced leaves were inoculated with TMV-Luc expressing luciferase (Fig. 1D) or TMV-Gus expressing β-glucuronidase (not shown), infection sites were obtained. This indicates that the lack of infection by TMV-GFP is mediated through silencing caused by specific recognition of the GFP-sequences in the viral genome and is not due to a general inability of TMV to infect the silenced leaves. Next, we chose leaves for infection that were just in the process of spreading GFP silencing, i.e., leaves that carried both green and red fluorescent areas (Fig. 1E). Here, viral infection sites were restricted to GFP-expressing cells, confirming that the GFP-expressing virus is recognized as a target for degradation and unable to infect cells undergoing established GFP silencing. Moreover, in such leaves (Fig. 1E and F) the spread of infection halted just at the border between nonsilenced and GFP-silenced cells over an extended period. Thus, GFP-containing virus can replicate and move cell to cell in GFP-expressing tissue up to the border of GFP-silenced cells. This finding illustrates that TMV-GFP is unable to spread from nonsilenced cells into GFP-silenced cells and thus suggests that the viral silencing suppressor can inhibit de novo silencing but not revert already-established silencing.

FIG. 1.

TMV does not suppress a preestablished silencing system in N. benthamiana. (A and B) Spread of TMV-GFP in GFP-transgenic line 16c. The virus spreads freely throughout the GFP-expressing leaf (A, 8 dpi; B, 20 dpi). (C) TMV-GFP does not infect fully GFP-silenced leaves. (D) Fully GFP-silenced leaf infected with TMV-luc. (E and F) The spread of TMV-GFP in a leaf undergoing the spread of GFP silencing (E, 4 dpi; F, 15 dpi). Infection is restricted to nonsilenced areas of the leaf, indicating that TMV-GFP can propagate in GFP-expressing cells but is a target for silencing in GFP-silenced cells. Magnified images of leaf areas indicated in panels E and F are shown. Because the virus was unable to spread into cells undergoing GFP silencing, infection sites developed aberrant shapes.

A mutation in the 126k protein reduces the activity of the viral silencing suppressor.

A G1117A point mutation in ToMV RNA affects the viral suppressor activity by causing a C349Y exchange in the 130k and 183k replicase subunits (22). The ToMV 130k protein is homologous to the TMV 126k protein, and the cognate amino acid position in the 126k protein can be easily identified (Fig. 2A). To confirm previous evidence for a silencing suppressing role of the 126k protein (9) and to directly test this activity, we engineered the C349Y exchange mutation into the 126k protein of TMV-GFP. As shown in Fig. 2B and C, parental TMV-GFP virus, here referred to as TMV-126k^wt-GFP, causes the formation of infection sites with homogenously distributed fluorescence in infected tobacco. In contrast, the virus carrying the C349Y mutation, to which we refer as TMV-126k^m-GFP, produces infection sites that appear as fluorescent rings with a dark center, thus indicating the silencing of the virus in cells behind the infection front. This observation indicates that the C349Y mutation (126k^m, Fig. 2A) interferes with the silencing suppressing activity of TMV-GFP. Next, we compared the silencing suppressor ability of the mutant 126k protein (126k^m) with that of wild-type 126k protein (126k^wt) in a transient assay. GFP-expressing N. benthamiana plants (line 16c) were co-agro-infiltrated with a silencing inducing construct (35S:GFP) and constructs either expressing the potyvirus Hc-Pro suppressor, the 126k^m protein, the 126k^wt protein, or none of them (empty vector). As shown in Fig. 2D, the expression of the different agrobacterial constructs led to distinct GFP expression levels at 5 days postinfiltration (dpi). Whereas the tissue patch expressing empty vector showed low GFP expression surrounded by a red rim of fully silenced cells, the Hc-Pro-expressing tissue exhibited very high GFP fluorescence confirming the known silencing suppressing activity of Hc-Pro (27). Tissue expressing 126k^wt also showed high levels of GFP fluorescence, and the red rim was visible. In contrast, tissue expressing 126k^m had a phenotype similar to that of the empty-vector control. These observations indicate that the 126k protein of TMV suppresses silencing, although to a lesser extent than Hc-Pro, and that this activity is affected by the C349Y mutation in 126k^m.

FIG. 2.

An amino acid exchange mutation in the 126k/183k replicase protein interferes with silencing suppressing activity. (A) Position of the C349Y exchange mutation in ToMV and TMV. (B) Infection sites (7 dpi) of TMV-126k^wt-GFP (TMV-GFP) appear in the form of green fluorescent disks. (C) Infection sites (7 dpi) of TMV-126k^m-GFP (TMV-GFP carrying the C349Y mutation) appear in the form of green fluorescent rings, thus indicating silencing of the virus in cells behind the infection front. (D) Transient agroinfiltration assay in GFP-transgenic line 16c to determine silencing suppressor activity of the TMV 126k protein. GFP-expressing construct (p35S:GFP) was agroinfiltrated together with test construct encoding the TMV 126k protein (126k^wt), or the 126k protein carrying the C349Y mutation (126k^m), or the PVY Hc-Pro or with empty vector. The results shown were obtained at 5 dpi. Consistent with the strong ability of potyviral Hc-Pro to suppress silencing, the coexpression of this protein with the GFP construct results in strong expression of GFP. Coexpression of the GFP construct with empty control vector results in silencing of GFP. Expression of the TMV 126k protein (126k^wt) allows some expression of GFP, indicating the ability of the protein to suppress the silencing of GFP expression. In contrast, expression of 126k^m results in only very little GFP expression, indicating that the mutation reduces the silencing suppressing activity of 126k protein.

The silencing suppressor activity of the 126k protein is correlated with increased virus induced symptoms in tobacco.

Next, we determined the effects of the 126k^m mutation in nonhybrid, wild-type TMV. Unlike TMV-126k^wt-GFP, TMV spreads rapidly in N. benthamiana, leading to plant death within 1 to 2 weeks after showing the first symptoms. To overcome the limitations of this system and to be able to study the effects of the suppressor mutation in the wild-type viral background over an extended period of time, we performed experiments in N. tabacum cv. Xanthi nn. These plants show a variety of pronounced symptoms, and their growth and development are strongly hampered, but they usually survive the infection and are able to set seeds. Figure 3 gives an overview of the symptoms observed in plants infected with wild-type TMV and TMV expressing the mutated 126k protein (TMV-126k^m). TMV infection led to severely stunted plants, a dark green appearance, and green mosaic leaves with an irregular shape (Fig. 3A and B). Systemic leaves displayed all patterns of deformation, ranging from an altered length-to-width ratio (Fig. 3E) over lancet-shaped leaves with distorted vein patterns (Fig. 3F) to extremely reduced leaf blades (Fig. 3G). In contrast, infection with TMV-126k^m caused only minor symptoms, i.e., the growth was slightly inhibited and the leaves had an almost normal appearance, except that they appeared lighter green compared to the mock-infected controls (Fig. 3C). The mutated virus could spread systemically throughout the plant, as was indicated by the mosaic pattern of systemic leaves (Fig. 3H and I). Apparently, there is a correlation between the ability of the virus to suppress RNA silencing and the severity of disease symptoms, which suggests that the TMV silencing suppression mechanism interferes with the miRNA pathway. To test this hypothesis, we checked the abundance and methylation status of miRNAs in infected plants.

FIG. 3.

Phenotypes of tobacco plants and leaves after infection. (A and B) TMV infection led to severely stunted plants, a dark green appearance, and green mosaic leaves with irregular shape. (E to F) Systemic leaves displayed all patterns of deformation, ranging from an altered length/width ratio (E) over lancet-shaped leaves with distorted vein patterns (F) to extremely reduced leaf blades (G). (C and D) In contrast, infection with TMV-126k^m caused only minor symptoms, i.e., the growth was slightly inhibited (C) and the leaves had an almost-normal appearance, except that they appeared lighter green compared to the mock-infected controls (D). (H and I) The mutated virus could spread systemically throughout the plant, as was indicated by the mosaic pattern of systemic leaves.

Silencing suppressing activity of 126k protein is correlated with enhanced accumulation of unmethylated sRNA.

Previously, it was shown that miRNAs and siRNAs isolated from Arabidopsis or N. benthamiana plants infected with the TMV-related ORMV are sensitive to β elimination, indicating that the infection interferes with HEN1-mediated methylation of these small RNA species (1, 4). Interference with the methylation of miRNAs had also been observed in transgenic plants expressing other viral silencing suppressors, such as p21 of Beet yellows virus, p19 of Tomato bushy stunt virus, or P1/Hc-Pro of Turnip mosaic virus (55), indicating that interference with HEN1-mediated methylation is a manifestation of silencing suppression by these proteins. Thus, in addition to the stabilization of siRNA/siRNA* and miRNA/miRNA* duplexes (6, 11), changing the quality of sRNAs through interference with methylation may be a mechanism by which silencing suppressors interfere with the miRNA pathway (6, 11, 21) and cause developmental defects in infected plants (46).

In order to test whether ORMV and TMV infection in tobacco interferes with miRNA methylation and whether the ability of TMV to suppress silencing is linked to this activity, we analyzed sRNA fractions isolated from infected plants at 14 dpi by β elimination and Northern hybridization using probes against miR166, miR160, miR157, and viral RNA. β Elimination removes the 3′ nucleotide of unmethylated sRNAs, leading to products that migrate faster during gel electrophoresis. Unlike in noninfected plants, in which sRNAs are methylated and therefore insensitive to β elimination, we observed in ORMV- or TMV-infected N. tabacum faster-migrating bands for each of the tested miRNA, as well as for viral siRNA (Fig. 4). These were similar as in infected A. thaliana and N. benthamiana (Fig. 4), confirming earlier results (1). In all cases, these faster-migrating bands were absent from untreated samples. These observations show that ORMV infection interferes with sRNA methylation in all three plants tested and that TMV infection interferes with sRNA methylation in tobacco. Finally, we compared extracts from TMV-infected plants with those from TMV-126k^m-infected plants and found that the fraction of unmethylated sRNAs within the 21 nt sRNA population was reduced by ca. 50% in the latter case. Based on these results we conclude that the decreased ability of TMV-126k^m to suppress silencing and the reduced occurrence of viral disease symptoms in plants infected with this virus are correlated with virus-induced interference with miRNA methylation. This indicates that the ability of the 126k protein to suppress silencing, as well as the formation of virus-induced disease symptoms, may both involve an interference with HEN1-mediated methylation of sRNAs.

FIG. 4.

sRNA analysis before (−) and after (+) β elimination (BE). ORMV infection causes the production of nonmethylated sRNA (viral siRNAs and indicated miRNAs) in Arabidopsis (A.t.), N. benthamiana (N.b.), and N. tabacum (N.t). TMV gives rise to the production of nonmethylated sRNA in N. tabacum like ORMV. The fraction (%) of β-elimination-sensitive nonmethylated miRNA within the total specific miRNA population is given for the TMV-infected samples. Compared to the fraction of nonmethylated miRNA observed in plants infected with wild-type TMV (column a), the fraction of nonmethylated miRNAs is reduced by ca. 50% in plants infected with TMV-126k^m (column b). In uninfected control tobacco plants, the tested miRNAs are fully methylated.

Interestingly, N. tabacum plants infected with either ORMV or TMV showed an overall increase in the level of miRNAs compared to noninfected plants or plants infected with TMV-126k^m (Fig. 4). This overaccumulation of miRNAs might indicate a second function of the 126k protein, e.g., sequestering of sRNAs.

The accumulation of unmethylated sRNAs is correlated with the time course of 126k protein expression.

To test whether indeed the 126k protein causes interference with sRNA methylation in infected plants, a time course experiment was performed, in which a potential correlation between the accumulation of 126k protein and interference with sRNA methylation was investigated. After inoculation of plants with either wild-type TMV (W) or TMV-126k^m (M) or after mock infection (C) at 8, 15, 30, and 40 dpi, sRNA and protein fractions were isolated and analyzed by Northern hybridization before and after β elimination and by Western blotting with antibodies against replicase. The results of a representative experiment are shown in Fig. 5. The patterns of viral siRNAs and miR160 bands before and after β elimination showed that virus-induced nonmethylated siRNAs and miRNAs accumulated at 8, 15, and 30 dpi, but strongly declined at 40 dpi. Furthermore, at 8 and 15 dpi the fraction of nonmethylated siRNA and miRNA within the sRNA population was lower in extracts derived from mutant TMV-infected compared to extracts derived from wild-type TMV-infected plants. Interestingly, this difference was not seen at the later time points. The relative accumulation of unmethylated sRNAs and the transient effects of the 126k^m mutation were correlated with the transient accumulation of 126k protein, which decreased at 30 dpi and was absent at 40 dpi. The level of long viral RNA was decreased at 15 dpi in plants infected with TMV-126k^m but not in plants infected with wild-type TMV, which is consistent with the reduced ability of TMV-126k^m to interfere with antiviral silencing. At 30 dpi, the level of RNA from both TMV-126k^m and wild-type TMV was reduced. Later, at 40 dpi, higher levels of both viral genomes were again detectable, probably due to protection of the viral genomes by encapsidation. Importantly, the occurrence of increased amounts of unmethylated versus methylated sRNAs is correlated with the level of 126k protein rather than with the level of viral RNA. This observation confirms that the virus-induced and silencing suppression-correlated interference with sRNA methylation resides in the 126k protein of the virus.

FIG. 5.

The level of nonmethylated sRNAs in TMV-infected plants correlates with the time course of infection and 126k expression. Comparison of RNA (A, B, C, D, and E) and protein (F and G) extracts derived from N. tabacum plants inoculated with wild-type TMV (W) or TMV-126k^m (M) or mock infected (C) and then harvested at 8, 15, 30, or 40 dpi. Viral siRNA (CP siRNA), shown before (−) (A) and after (+) (B) β elimination (BE), accumulates until about 30 dpi and shows a lower level at 40 dpi, suggesting protection of the virus genome by encapsidation. The virus-induced accumulation of nonmethylated siRNA (B) and miRNA (miR160) (C) is restricted to early time points (until 30 dpi) and correlated with the level of 126k/183k protein (F), as detected by immunoblotting. The level of virus-induced nonmethylated sRNA is reduced in extract derived from plants infected with TMV-126k^m (M); the effects of the suppressor mutation are more pronounced on miR160 (C) than on viral siRNA (B) and more pronounced during earlier time points (8 and 15 dpi) compared to later time points (30 dpi) (C). miR160 is fully methylated in uninfected control plants (lanes C). The level of TMV RNA (D) is reduced at 30 dpi, a finding consistent with high levels of CP siRNA (A), and increased at 40 dpi (D), a finding consistent with reduced levels of siRNAs (A) and protection of the viral genome by encapsidation. (E) Mitochondrial RPL2 mRNA is shown as RNA loading control; (F) immunoblot with antibody against TMV replicase; (G) bands in Coomassie blue-stained protein gel shown as loading control; (H) quantified fraction (%) of nonmethylated CP siRNA and miR160 within the total respective sRNA population in extracts derived from plants infected with wild-type TMV (W) and mutant TMV (TMV-126k^m, M), respectively. Also, the fraction (%) of nonmethylated sRNA accumulation seen in tissues infected by the mutant virus (M) compared to nonmethylated sRNA accumulation seen in tissues infected by the wild-type virus (W) (set to 100%) is shown.

Transient expression of the 126k replicase leads to accumulation of unmethylated sRNA.

To test whether the interference with sRNA methylation indeed reflects an activity of the 126k protein, N. tabacum tissues were agroinfiltrated with the construct expressing the 126k protein and analyzed by Northern blot hybridization using probes against miR160 and miR157 (Fig. 6). Although the 126k protein could be expressed at only very low levels, its presence led to the accumulation of nonmethylated miRNA molecules similar to what occurs in infected cells. This observation demonstrates the ability of the 126k protein to cause the accumulation of nonmethylated sRNAs.

FIG. 6.

Accumulation of nonmethylated miRNA in agroinfiltrated N. tabacum leaves expressing 126k protein. (A) Compared to TMV-infected leaves (TMV), the agroinfiltrated, 126k-transfected leaves (126k) express only very low levels of detectable 126k protein. Nevertheless, RNA samples treated (+) or not treated (−) for β elimination (BE) reveal that, similar to the infected leaves (TMV), 126k-transfected leaves also accumulate periodate-sensitive, nonmethylated miRNAs (arrow), whereas mock-infected or mock-transfected leaves (C) only accumulate periodate-insensitive and, therefore, fully methylated miRNA (see panel B).

TCV infection causes accumulation of unmethylated siRNAs but not of unmethylated miRNAs.

Both ORMV and TMV infections lead to the accumulation of β-elimination-sensitive, i.e., nonmethylated miRNAs. However, this is not the case for every RNA virus. Figure 7 shows examples of experiments demonstrating that miRNAs in plants infected with TCV are β elimination insensitive. For the experiment shown in Fig. 7A, sRNAs from Arabidopsis plants infected with either TCV, ORMV, or not at all were isolated at 14, 21, and 30 dpi and hybridized with probes against miR166, miR172, viral siRNA, and viral genomic RNA. Before gel electrophoresis and hybridization, the samples were treated for β elimination, where indicated. As shown in Fig. 7A, miRNAs from the uninfected control and the TCV-infected but not the ORMV-infected plants appeared as single bands, showing that they are β elimination insensitive, i.e., methylated. The additional miRNA-specific bands originating from the ORMV-infected plants appeared as a ladder with bands both smaller and larger than normal miRNA, indicating miRNA shortening and polyuridylation, respectively (26). Obviously, in contrast to ORMV or TMV, TCV infection does not lead to the accumulation of unmethylated miRNAs. As in the case of ORMV, however, a significant fraction of TCV siRNAs was sensitive to β elimination, since the distinct siRNA bands seen in the untreated sample were converted to a complex hybridization pattern in treated samples. The experiment shown in Fig. 7B confirms these conclusions. In this experiment, we inoculated plants of two Arabidopsis ecotypes, Col-0 and Ler, with TCV, and sRNAs isolated at 16 dpi were hybridized with probes against miR165 and viral siRNA. Again, we observed that miRNA was fully insensitive to β elimination and thus methylated, whereas a significant fraction of the TCV siRNAs was sensitive to this treatment and thus unmethylated. Moreover, the different effects of TCV infection on miRNA and viral siRNA methylation were the same for the two ecotypes tested.

FIG. 7.

In Arabidopsis, both ORMV and TCV infection lead to the accumulation of nonmethylated viral siRNA, but only ORMV infection leads also to the accumulation of nonmethylated miRNA. A. Analysis of miR166 (a), miR172 (b), TCV siRNA (with homology to the CP gene (c), ORMV siRNA (with homology to the CP gene) (d), and ORMV genomic RNA (e) in A. thaliana Col-0 extracts obtained at 14, 21, or 30 dpi from plants infected with either TCV or ORMV (OR) or from uninfected control plants (C). The sRNA extracts have been treated (+) or not treated (−) for β elimination (BE). The miRNAs are fully methylated in uninfected control plants (C), as seen by the stability of the miRNA and the absence of miRNA derivatives of higher (due to polyuridinylation) or lower molecular weight in the BE samples (a and b). Such miRNA derivatives are present in samples from ORMV-infected plants but not in samples from TCV-infected samples (a and b), indicating the accumulation of nonmethylated miRNA in ORMV-infected plants but not in TCV-infected plants. Viral siRNA is sensitive to BE, irrespective of whether the siRNA originates from TCV (c) or ORMV (d). B. Duplicate experiment. sRNAs extracted at 16 dpi from A. thaliana Col-0 and Ler ecotypes infected with TCV and either treated (+) or not treated (−) for β elimination (BE) were hybridized with probes against miR165 (a) and TCV siRNA (b). Whereas miRNA is fully methylated (a), a considerable fraction of the viral siRNA is nonmethylated (b).

DISCUSSION

We addressed here the potential mechanisms by which the TMV 126k protein could suppress silencing. We found that TMV-GFP can propagate in GFP-expressing tissues but not in tissues in which GFP is silenced, indicating that the TMV-encoded suppressor may inhibit the establishment of silencing de novo but does not interfere with already-established silencing. The accumulation of TMV-specific siRNAs in infected tissues indicates that the target of the TMV suppressor likely acts downstream of siRNA production, probably by blocking siRNA usage. To investigate how the virus could block siRNA usage, we made use of previous findings indicating that the silencing suppressing activity of tobamoviruses resides in the small replicase subunit (9, 22). Indeed, a specific C349Y mutation in the 130k protein of ToMV was reported to reduce the ability of the protein to suppress silencing (22). We show here that also in TMV the corresponding replicase mutation results in reduced suppressor activity. We then compared the silencing suppressing abilities of wild-type and mutant 126k protein with effects at the level of sRNAs and the formation of disease symptoms. The mutation caused a decrease of the 126k suppressor activity correlating with a strong reduction in the formation of virus-induced disease symptoms. This suggests that the suppressor may act through a mechanism that also affects miRNA properties. The results of our analysis comparing small RNA species derived from plants infected with wild-type TMV with those infected with TMV carrying the C349Y mutation indeed show that the silencing suppressing activity of the 126k protein correlates with the accumulation of unmethylated versions of both virus siRNAs and host miRNAs. Transient-expression experiments demonstrate that the expression of 126k protein alone is sufficient to cause accumulation of unmethylated sRNAs, thus providing strong evidence that indeed an activity of the 126k protein leads to accumulation of nonmethylated sRNAs during infection.

The occurrence of unmethylated sRNAs was also observed upon tobamovirus ORMV infection of either Arabidopsis (1, 4) or tobacco (as shown here) and thus appears to represent a general phenomenon associated with tobamoviruses. Based on our observations with the C349Y mutation in TMV, which allowed us to correlate the occurrence of nonmethylated sRNAs with silencing suppression by 126k replicase, we propose that tobamovirus suppressors in general act through altering the quality of sRNAs.

Interference with the methylation of miRNAs has also been reported for silencing suppressors derived from viruses of other families, such as p21 of Beet yellows virus, p19 of Tomato bushy stunt virus, and P1/Hc-Pro of Turnip mosaic virus (55). However, the evidence in these cases was derived from transgenic plants expressing the proteins. In contrast to these previous studies, our study supports the reduced sRNA methylation as an effect of the suppressing activity of the 126k protein during infection. Moreover, we show that an effect of viral suppressors on miRNA methylation cannot be generalized, since, as we found here, infections with the carmovirus TCV led only to the lack of siRNA methylation but not of miRNA methylation. Although further studies are necessary, this finding suggests that some suppressors could have the ability to distinguish siRNA and miRNA duplexes, for example, by the absence or presence of bulges or by their intracellular localization.

We do not know whether the 126k protein interferes with HEN1-mediated sRNA methylation (54) or whether it causes demethylation of already-methylated molecules. Osman and Buck (35) mapped an RNA-binding domain to amino acids 314 to 423 of the replicase protein of ToMV, and the 126k protein of TMV was shown to have methyltransferase activity, albeit with the cap as a target and S-adenosylmethionine as the methyl donor (30). It is possible that this activity could also use methylated sRNAs as methyl donors, thereby leading to their demethylation. However, whether the methyltransferase activity of the 126k protein is indeed required for the accumulation of unmethylated sRNA in infected cells and whether unmethylated sRNA molecules are directly involved or are rather a consequence of sRNA binding by the suppressor remains to be demonstrated. The latter hypothesis is more likely, since the 126k protein is not only inhibiting miRNA methylation but also increasing the concentration of miRNA, which could be explained by protection through sequestration and binding.

The methyltransferase domain carried by the 126k protein is also present on the longer 183k protein, which in addition to the methyltransferase and helicase domains has a RNA-dependent RNA polymerase domain (14, 39). Both proteins interact with each other (15) and are found in replication complexes isolated from infected plants (34, 52). Interestingly, while the 183k protein alone is sufficient for replication in protoplasts, replication efficiency is strongly increased if both versions are expressed (20, 25). Furthermore, although a mutation of the amber codon, which regulates termination and read-through of the 126k protein, to a tyrosin codon is viable, a pseudorevertant restored the production of the 126k protein through reversion to an ochre codon (20). These findings indicate that the 126k protein has important virulence functions that go beyond the functions already contributed by the 183k protein. It remains to be tested whether the suppressor activity could rely on the 126k protein only or whether the 183k protein also could contribute to the effects on sRNA methylation described here.