Tomato Mosaic Virus Replication Protein Suppresses Virus-Targeted Posttranscriptional Gene Silencing

Abstract

Posttranscriptional gene silencing (PTGS), a homology-dependent RNA degradation system, has a role in defending against virus infection in plants, but plant viruses encode a suppressor to combat PTGS. Using transgenic tobacco in which the expression of green fluorescent protein (GFP) is posttranscriptionally silenced, we investigated a tomato mosaic virus (ToMV)-encoded PTGS suppressor. Infection with wild-type ToMV (L strain) interrupted GFP silencing in tobacco, coincident with visible symptoms, whereas some attenuated strains of ToMV (L₁₁ and L₁₁A strains) failed to suppress GFP silencing. Analyses of recombinant viruses containing the L and L₁₁A strains revealed that a single base change in the replicase gene, which causes an amino acid substitution, is responsible for the symptomless and suppressor-defective phenotypes of the attenuated strains. An agroinfiltration assay indicated that the 130K replication protein acts as a PTGS suppressor. Small interfering RNAs (siRNAs) of 21 to 25 nucleotides accumulated during ToMV infection, suggesting that the major target of the ToMV-encoded suppressor is downstream from the production of siRNAs in the PTGS pathway. Analysis with GFP-tagged recombinant viruses revealed that the suppressor inhibits the establishment of the ToMV-targeted PTGS system in the inoculated leaves but does not detectably suppress the activity of the preexisting, sequence-specific PTGS machinery there. Taken together, these results indicate that it is likely that the ToMV-encoded suppressor, the 130K replication protein, blocks the utilization of silencing-associated small RNAs, so that a homology-dependent RNA degradation machinery is not newly formed.

RNA silencing, including posttranscriptional gene silencing (PTGS) in plants, is a homology-dependent RNA degradation system occurring in the cytoplasm (41) and characterized by the presence of 21- to 25-nucleotide (nt) small interfering RNAs (siRNAs) (11). Biochemical and genetic analyses have shown that the core mechanisms of PTGS are shared among eukaryotes (30, 42, 46). PTGS is induced by double-stranded RNAs (dsRNAs) or structured single-stranded RNAs (ssRNAs), which are processed into siRNAs by RNase III-like enzymes, such as Dicer in Drosophila. A sort of siRNA is incorporated into the sequence-specific degradation complex (RNA-induced silencing complex [RISC]) and thereby guides the RISC to destroy target RNA that contains a sequence complementary to the siRNA in the RISC. In some organisms, including plants, a host RNA-dependent RNA polymerase (RdRp) is also involved in PTGS (5, 24), presumably converting the target ssRNA into dsRNA, which is the substrate used by Dicer homologs to generate secondary siRNAs. In plants, PTGS generates mobile signals with sequence-specific information that spread through plasmodesmata and phloem to distant organs, establishing systemic silencing (23).

Plant viruses are strong inducers and targets of PTGS; therefore, defending plants against viruses is thought to be a role of PTGS (18, 45). Consequently, many plant viruses have evolved to encode a suppressor of PTGS to combat this resistance mechanism (41, 44). The suppressor proteins identified so far are structurally and functionally diverse. Potyvirus helper component proteinase (HC-Pro), one of the first identified suppressors (1, 4, 13), probably inhibits the maintenance step of PTGS at or upstream from the production of siRNA (16, 17) and/or downstream from the production of siRNA (14, 32). Cucumovirus 2b, which was also one of the first suppressors identified (4), suppresses systemic silencing, possibly after the generation of the silencing signal (8). Potato virus X (PVX) p25 prevents systemic silencing (43), presumably through the inhibition of the class of long (∼25-nt) siRNAs (10). Recently, tombusvirus p19 was shown to bind siRNAs in vitro, and so it may prevent the spread of mobile silencing signals (35). The coat protein (CP) of turnip crinkle virus probably functions at an earlier step of PTGS, upstream from siRNA production (32).

Tomato mosaic virus (ToMV) is a positive-sense ssRNA virus belonging to the Tobamovirus genus. The ToMV genome encodes at least four proteins: two replicase components (130K and 180K proteins), a movement protein (MP), and CP (22). The 130K and 180K proteins are translated from the genomic RNA, and MP and CP are translated from the respective subgenomic mRNAs. The 180K protein is synthesized by read-through of the leaky termination codon of the 130K protein gene. Like Tobacco mosaic virus (TMV, the type member of the Tobamovirus genus), ToMV infects tobacco plants systemically, causing mosaic symptoms, which are characterized by intermingled light and dark green regions (20).

Previously, Voinnet et al. (44) showed that TMV infection reverses green fluorescent protein (GFP) silencing systemically in leaves of Nicotiana benthamiana and suggested, based on the suppression pattern, that TMV may counteract PTGS by preventing the spread of systemic signals as well as by a further mechanism. However, wild-type TMV and ToMV cause severe systemic necrosis in N. benthamiana approximately 1 week after infection, a fact which imposes a limitation on the investigation of these tobamovirus suppressor functions in the normal infection process. In this work, we investigated the activity of the ToMV-encoded suppressor by using transgenic tobacco in which GFP is expressed constitutively or in which the expression of GFP is silenced posttranscriptionally. We found that infection with ToMV interrupted GFP silencing, in parallel with the visible symptoms, through the action of a suppressor. An agroinfiltration assay indicated that the 130K replication protein functions as a suppressor of PTGS. We also found that the suppressor probably exerts its effects at a specific step after siRNA production to prevent the de novo formation of the RISC in ToMV-infected leaves.

MATERIALS AND METHODS

Transgenic plants.

N. tabacum cv. Samsun and N. benthamiana were transformed with Agrobacterium tumefaciens LBA4404 harboring plasmid pBK-erG3 (see below) by the leaf disk method essentially as described previously (7). The tobacco and N. benthamiana lines constitutively expressing an endoplasmic reticulum (ER)-localized, highly fluorescent GFP variant (erG3GFP) (38) were G3Sm1 and G3Nb3, respectively, which carry the transgene homozygously at a single locus. For the GFP-silenced tobacco line (G3Sm2), T2 plants and a few T3 lines showing a strong phenotype were used; these had various combinations of multiple copies of the transgene that differed in transgene number and in the locus of integration. Because the same results were obtained with different T2 and T3 plants, a T3 line (G3Sm2-1) carrying three copies of the transgene homozygously and showing a strong phenotype was used in the later stages of the experiments.

Inoculation with viruses and infectious transcripts.

The ToMV strains used were wild-type strain L (27) and L-derived attenuated strains L₁₁ and L₁₁A (25). Infectious ToMV and PVX transcripts were derived from MluI-cut pTL-series plasmids and SpeI-cut pTX-series plasmids (see below), respectively, with T7 RNA polymerase (Takara Bio). As the standard method, ToMV and its derivatives (including in vitro transcripts) were suspended in 10 mM sodium phosphate (pH 7.0) and inoculated onto the single fifth true leaf of Samsun plants when it was 7.5 to 9.0 cm long. Unless otherwise specified, the inocula contained infectious viruses at concentrations that caused approximately 100 to 150 local necrotic lesions on Xanthi nc tobacco leaves. Inoculated plants were grown at 26 or 28°C (16 h of light). Recombinant PVX was first amplified in N. benthamiana, and homogenates of the inoculated leaves were used to inoculate tobacco leaves.

Construction of plasmids.

pTLW3 carries the full-length ToMV cDNA plus a linker-derived 25-bp sequence (between the MluI and the EcoRI sites) of pLFW3 (21) downstream of a T7 promoter between the PstI and EcoRI sites of pUC18. The 2.93-kb BstXI-AccIII fragment (nt 832 to 3759) of pLFA1 (21) was replaced with the corresponding fragment of pTLW3 to create pTLJA1. The 2.47-kb NheI-KpnI fragment (nt 1926 to 4394) was exchanged between pTLW3 and pTLJA1 to yield pTLJ and pTLA. The 1.41-kb KpnI-BstEII fragment (nt 4395 to 5799) of pTLW3 was replaced with the corresponding fragment (from the KpnI site to the end of the G3GFP open reading frame [ORF]) of piL.G3 (38) to yield pTL.G3. The 0.59-kb fragment (from the filled-in BstEII site to the MluI site) of pTL.G3 was replaced with an NcoI (filled-in)-MluI fragment of pTLW3 to create pTLBN.G3. The KpnI-MluI fragment of pTLBN.G3 containing the G3GFP ORF was introduced into pTLJ to yield pTLBN.G3(J). pTX.erG3 was constructed by replacing the ApaI-XhoI fragment of pP2C2S (3) with the corresponding fragment of piX.erG3 (37). pTX.GUS was constructed by inserting the SmaI-SacI fragment of pBI121 (Clontech), encoding β-glucuronidase (GUS), into the SalI site of pP2C2S after blunting. Note that the erG3GFP gene in pTX.erG3 and pBK-erG3 (see below) encodes the signal peptide at the N terminus and the ER retention signal at the C terminus, whereas the G3GFP gene in pTLBN.G3 and pTLBN.G3(J) encodes cytosolic GFP.

The XbaI-SacI fragment of pBI-erG3, which encodes erG3GFP (38), was inserted between the corresponding sites of pBI121 (Clontech) to create pBK-erG3. The coding region for HC-Pro of potato virus Y (PVY; Japanese O strain) was amplified by reverse transcription (RT)-PCR with primers 5′-GCTCTAGAAACATGGCCTCAAATGCTGAGAATTTTTGGAAGGGTCTG-3′ and 5′-CGGGATCCTTAACCAACTCTATAGTGTTTTATATCAGACTC-3′, cut with XbaI and BamHI (shown in italic type in the primers), and cloned into pMD1 (a pBI121-derived vector with a modified cloning site; supplied by B. Baker) to create pMD-YHCP. The Met and Ala codons and a termination codon were included in the primers. pBI-L130NRT contains the 5′ noncoding region and the 130K protein ORF of pTLW3. An additional termination codon was positioned just after the leaky termination codon of the 130K protein ORF. pBI-L130FS was created by filling in the XbaI site located in the 130K protein ORF of pBI-L130NRT (nt 1002 of the ToMV genome). pBI-L130J contains the pTLJ-derived 130K protein ORF.

GFP observation (imaging) and GUS staining.

GFP signals were observed under an epifluorescence stereomicroscope (MZ-FLIII; Leica) equipped with a filter set (GFP Plant; Leica) and a charge-coupled device camera (DXM1200; Nikon). Plants were excited with a halogen lamp (Intralux 6000-1; Volpi AG) through a no. 24902 filter (440 to 480 nm) and then photographed with a digital camera (Camedia E-20; Olympus) equipped with a cut filter (520 nm, SC52; Fuji Film). This procedure resulted in a low level of red autofluorescence. Alternatively, plants were irradiated with the mercury lamp of the epifluorescence stereomicroscope through the GFP filter (450 to 490 nm) and then photographed with a charge-coupled device camera (DC500; Leica) equipped with a zoom lens (70 to 180 mm, Zoom-Micro Nikkor; Nikon) and a band-pass filter (500 to 550 nm; Omega Optical, Inc.) to reduce red autofluorescence. GUS staining was performed as described previously (19).

RNA analysis.

Leaves, excluding major veins, were homogenized in liquid nitrogen. Total RNAs were extracted with TRIzol reagent (Invitrogen) according to the manufacturer's instructions. For Northern blot analysis of GFP mRNA, total RNAs were glyoxylated, separated on a 1% agarose gel, and then transferred to a Hybond-N membrane (Amersham Biosciences). Prehybridization and hybridization were performed as described previously (11) with a DNA probe derived from the erG3GFP ORF (the EcoRI-ClaI fragment).

For small-RNA analysis, high-molecular-weight RNAs were removed from total RNAs (isolated from 0.5 to 2 g of leaf tissues) by centrifugation in the presence of 5% polyethylene glycol and 0.5 M NaCl (11). Low-molecular-weight RNAs were recovered by ethanol precipitation, suspended in formamide, and separated on a 15% polyacrylamide-7 M urea gel. The amounts of low-molecular-weight RNAs loaded were adjusted on the basis of the ethidium bromide-stained bands, which contained the major species of tRNAs, 5S rRNAs, and 5.8S rRNA, migrating near the ∼400-bp DNA marker on a 1% agarose gel (17).

After blotting onto Hybond-NX (Amersham Biosciences), hybridization was performed at 40°C overnight with riboprobes as described previously (17). Riboprobes corresponded to the erG3GFP ORF or the 3′-terminal 0.8-kb region of the ToMV genome and were prepared with T7 or T3 RNA polymerase from the appropriate subclones, followed by partial alkaline hydrolysis into 50- to 100-nt (on average) fragments. Size markers were prepared by filling in the SpeI-cut end of pBluescript in the presence of [α-³²P]dCTP and cutting with SacII (20-nt oligonucleotide) or EcoRI (25-nt oligonucleotide). Synthetic RNAs with the same sequences as the above 20- and 25-nt oligonucleotides were also used in some experiments. The oligonucleotide DNA markers migrated ∼1 base faster than the corresponding RNA markers. Radioactive bands were detected with a Typhoon 8600 imager (Amersham Biosciences).

Specific amplification of transgene-derived GFP mRNA and GFP-tagged viral genomes by RT-PCR was performed as follows. Central portions (2 mm in diameter) of GFP lesions induced by TLBN.G3 or TLBN.G3(J) were punched out under an epifluorescence stereomicroscope and subjected to total RNA extraction as described above. Disks of the same diameter were punched from areas outside the GFP lesions on the same inoculated leaves or from mock-inoculated leaves.

cDNAs were synthesized with Ready-to-Go RT-PCR beads (Amersham Biosciences) and with oligo(dT)_12-16 or a 20-nt oligonucleotide complementary to the 3′ end of ToMV RNA as a primer. PCR primers used to amplify the erG3GFP mRNA sequence were 5′-ATGAAGACTAATCTTTTTCTCTTTCTCATC-3′, corresponding to the N terminus of erG3GFP (part of the signal peptide), and 5′-TTAAAGCTCATGTTTGTATAGTTCATC-3′, corresponding to the C terminus (including the ER retention signal). Primers used to amplify actin mRNA were 5′-ATGGCAGACGGTGAGGATATTCA-3′ and 5′-GCCTTTGCAATCCACATCTGTTG-3′. Primers used to specifically detect TLBN.G3 and TLBN.G3(J) were an oligonucleotide corresponding to nt 5271 to 5300 of ToMV and 5′-TGTGGGAGTTGAAGTTGTATTCCAA-3′, which corresponds to an internal portion of the G3GFP ORF. Amplification was carried out with ExTaq DNA polymerase (Takara Bio) for 22 cycles (for viral RNA or GFP mRNA) or 35 cycles (for actin mRNA) of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min. Each PCR (50 μl) contained cDNAs derived from total RNAs equivalent to 0.1 GFP lesion.

Protein analysis.

Leaf tissues were homogenized in 100 mM Tris-HCl (pH 6.8)-4% sodium dodecyl sulfate (SDS)-12% 2-mercaptoethanol. To detect the 130K protein, protease inhibitor CompleteMini (Roche-Boehringer) was included. After debris was removed, supernatants were boiled and mixed with an equal volume of 20% glycerol containing 0.02% bromophenol blue. To detect CP, proteins were separated by SDS- 15% polyacrylamide gel electrophoresis (PAGE) and stained with Coomassie brilliant blue. To immunodetect GFP or the ToMV 130K protein, proteins were separated by SDS-12.2% PAGE or SDS-7.5% PAGE, respectively, and transferred to a polyvinylidene difluoride membrane (Millipore). The membrane was treated with anti-GFP antibodies (Clontech) or anti-130K protein antibodies (9) as primary antibodies and horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin G antibody (Amersham Biosciences) as the secondary antibody, and immunoreactive bands were visualized by enhanced chemiluminescence (Amersham Biosciences).

Agroinfiltration.

A. tumefaciens C58C1/pGV2260 harboring the appropriate plasmid was cultured, set to an optical density at 600 nm of 1.0, and used to infiltrate small leaves of ∼3-week-old N. benthamiana plants essentially as described previously (6). The infiltration mixture contained two parts Agrobacterium carrying an empty vector, pOCA18 (28), or a plasmid encoding a suppressor and one part Agrobacterium carrying pBK-erG3, which induces local silencing. Infiltrated plants were kept at 22°C for 1 day to ensure transformation and then kept at the same or different temperatures for testing of the temperature dependency of silencing induction (36).

RESULTS

GFP-silenced tobacco lines.

We transformed tobacco with a gfp gene encoding an ER-localized, highly fluorescent GFP variant (erG3GFP), excitable by blue light but not by UV light (38). Of the transgenic lines, one line (G3Sm2) underwent silencing of GFP expression (Fig. 1A). A decrease in GFP fluorescence was first recognized along the major veins of the first and second true leaves ∼2 weeks after germination, after which the silenced regions spread. The GFP-silenced patterns in the aerial parts of the plants were categorized into two types. In the weakly silenced plants, GFP fluorescence remained in the interveinal regions even when the leaves were fully expanded. In the strongly silenced plants, which were used in this work, GFP fluorescence became undetectable under a fluorescence stereomicroscope by 4 or 5 weeks after germination, except in the cotyledons and a tiny region at the growing point. As a reference, we selected a line (G3Sm1) constitutively expressing GFP in all aerial parts of the plant (Fig. 1A).

FIG. 1.

Characteristics of GFP-silenced tobacco line. (A) GFP fluorescence in G3Sm1 and G3Sm2. Images were obtained at 2, 3, and 6 weeks (w) after germination. Bars, 2 mm. (B) Northern analysis of GFP mRNA in mature leaves of NT tobacco, G3Sm1, and G3Sm2. The lower panel shows ethidium bromide (EtBr) staining of the same samples. (C) Detection of GFP in mature leaves by Western blotting. (D) Presence of siRNAs containing the gfp sequence in G3Sm2. The lower panel shows EtBr staining (5.8S and 5S rRNAs) of the gel. Positions of 20- and 25-nt DNA markers are indicated on the left.

Both the G3Sm1 and the G3Sm2 lines grew normally. G3Sm2 was resistant to infection with GFP-tagged recombinant ToMV and PVX and susceptible to GUS-tagged viruses, whereas G3Sm1 was susceptible to both types of recombinants (see Fig. 5 and 6 for related data). GFP mRNA was detected at low levels in the silenced leaves of G3Sm2 but at much lower levels than were detected in G3Sm1 (Fig. 1B). GFP was not detected in the silenced leaves of G3Sm2 by Western blotting (Fig. 1C) or was detected only faintly after a long exposure (data not shown). Two classes of small RNAs (21- to 23-nt siRNA and ∼25-nt siRNA) (10) with the gfp sequence were detected in the GFP-silenced leaves of G3Sm2, whereas correspondingly sized small RNAs were not found in the leaves of G3Sm1 or nontransgenic (NT) tobacco plants (Fig. 1D). From these observations, we concluded that the expression of GFP was posttranscriptionally silenced in G3Sm2.

FIG. 5.

Suppression of PTGS in L0. (A) Schematic representation of GFP-tagged ToMV. Bent arrows indicate the start positions of subgenomic RNAs. (B) GFP lesions formed on tobacco (NT and G3Sm1). Arrows indicate the central portions of lesions, where the expression of the resident GFP gene was silenced. Inoculated plants were kept at 28°C, and images were obtained at 6 dpi. (C) Decrease in GFP mRNA levels in the central portions of TLBN.G3(J)-induced GFP lesions formed on G3Sm1. RNAs were extracted from the central portions of GFP lesions (lanes 3, 5, 7, and 9) or from areas outside the lesions (lanes 4, 6, 8, and 10) on leaves inoculated with TLBN.G3 (lanes 3 to 6) or with TLBN.G3(J) (lanes 7 to 10) or from mock-inoculated leaves (lanes 1, 2, 11, and 12). RT-PCR was performed as described in Materials and Methods. Paired lanes indicated at the top were derived from the same leaves. For lanes 11 and 12, PCRs were performed as for lanes 1 and 2 but with cDNAs that had been synthesized in the absence of primers.

FIG. 6.

Effect of ToMV infection on the preestablished PTGS system. NT tobacco or G3Sm2 was mock inoculated or inoculated with ToMV. At 4 dpi, when ToMV had spread over the inoculated leaves, a mixture of PVX.GUS and PVX.erG3 was challenge inoculated. The expression of GFP and GUS was examined 3 days later. Leaves were photographed under white-light irradiation (WL, top panels) and under blue-light irradiation (GFP, bottom panels). The GUS staining patterns of the same regions are shown (middle panels). Bar, 1 cm for all panels.

Symptom-associated suppression of GFP silencing during ToMV infection.

G3Sm2 was susceptible to wild-type virus infections, displaying symptoms indistinguishable from those developing in NT tobacco. Figure 2A shows ToMV-infected G3Sm2. Interestingly, strong green fluorescence was observed in leaves with mosaic symptoms (mosaic leaves) under blue-light irradiation. When the fluorescent regions were examined under a high magnification with a fluorescence microscope, the ER network was visible and was not excited by UV light, indicating that the green fluorescence was derived from erG3GFP. Such green fluorescence was not detected in the mosaic leaves of NT tobacco. The expression of GFP in the mosaic leaves of ToMV-infected G3Sm2 was confirmed by Western and Northern blot analyses (Fig. 3C and 4A). Therefore, the GFP fluorescence must have resulted from the suppression of GFP silencing, reflecting the activity of a ToMV-encoded suppressor.

FIG. 2.

Symptom-associated suppression of GFP silencing in G3Sm2 infected with ToMV. (A) GFP fluorescence pattern on mosaic leaves at 11 dpi. (B) Close-up view at 12 dpi of a mosaic leaf, in which GFP fluorescence is visible in yellow tissues. (C) Restoration of GFP expression in L3 around veins with vein-clearing symptoms at 6 dpi. Paired views in panels A and B were taken under white-light irradiation (left panels) and blue-light irradiation (right panels). Bars, 2 mm.

FIG. 3.

Mapping the mutation responsible for the suppressor-defective phenotype of ToMV L₁₁A. (A) Schematic representation of the genomic structures of ToMV and its derivatives. Nucleotides differing from wild-type ToMV (ToMV-L and TLW3) are indicated by open triangles (silent substitutions) and closed triangles (substitutions causing coding changes). The activity that interrupts GFP silencing in the systemically infected leaves of G3Sm2 is indicated at the right: +, symptom-associated suppression of GFP silencing; −, symptomless infection without suppression of GFP silencing. (B) Symptoms and suppression of GFP silencing in L5 of G3Sm2 infected with TLW3, TLJ, or TLA. Leaves were photographed under white-light irradiation (upper panels) and blue-light irradiation (lower panels). Dark green tissues are marked by asterisks. Bar, 4 mm. (C) No increase in GFP levels in TLJ- or TLJA1-infected leaves. Proteins extracted from 1 mg of leaf tissue were separated by SDS-PAGE, followed by Coomassie brilliant blue staining to detect CP (upper panel) or by Western blot-ting to detect GFP (lower panel). Lanes 1 to 3, healthy leaves (the fifth true leaves); lanes 4 to 10, upper leaves (L5) of G3Sm2 inoculated with buffer (lane 4) or the in vitro transcripts indicated above the panel (lanes 5 to 10). D and Y, tiny dark green islands and yellow tissues, respectively. (D) Transient restoration of GFP silencing in TLJ-infected G3Sm2. TLJ was inoculated onto a small leaf (L0), and a typical GFP patch appearing on L1 was photographed at 5 dpi (left panel). At 10 dpi, most GFP patches had disappeared, or the intensity of GFP fluorescence had decreased (middle panel). The right panel shows L1 of TLW3-inoculated G3Sm2 at 10 dpi. Bar, 4 mm.

FIG. 4.

RNA analysis of ToMV-infected tobacco. (A) Detection of genomic RNA and GFP mRNA in systemically infected leaves. Plants (NT tobacco, G3Sm1, or G3Sm2) were mock inoculated (lanes m) or inoculated with TLW3 (lanes L) or TLJ (lanes J) and then grown at 28°C. RNAs were extracted from L6 and L7 (excluding class I and II veins) at 12 dpi. The same amounts of total RNAs were loaded in the lanes. Genomic RNA was visualized by ethidium bromide (EtBr) staining (upper panel), and GFP mRNA was visualized by Northern blot analysis (lower panel). (B) Detection of small RNAs derived from the total RNAs shown in panel A. The gel was stained with EtBr prior to blotting (bottom panel). The amounts of RNAs loaded were adjusted based on EtBr staining of 5.8S and 5S rRNAs. The same membrane was reprobed to detect ToMV- and gfp-specific RNAs, indicated at the right. (+) and (−), small RNAs with plus (sense) and minus (antisense) polarities, respectively. Positions of 20- and 25-nt RNA markers are indicated by closed and open arrowheads, respectively. (C) Profiles of accumulation of genomic RNA and GFP mRNA in L0 of G3Sm2. Total RNAs were extracted at the indicated dpi and analyzed as described for panel A. The amounts of RNAs loaded were adjusted based on rRNA levels; accordingly, the RNA amounts loaded in lanes 7 and 8 were 1.3- and 1.7-fold higher, respectively, than those loaded in the other lanes. (D) Detection of small RNAs in L0. Small RNAs in each lane were derived from the total RNAs in the corresponding lane in panel C. The amounts of RNAs loaded were adjusted based on EtBr staining of 5.8S and 5S rRNAs. The gel was processed as described for panel B.

To understand the relationship between the suppression of GFP silencing and the development of visible symptoms, we inoculated ToMV onto the single fifth true leaf when it was between 7.5 and 9 cm in length, according to the standard inoculation method used by Atkinson and Matthews (2) for a detailed description of the development of tobacco mosaic disease. The leaves were numbered sequentially from the inoculated leaf (L0) to the upper leaves.

GFP fluorescence was not restored to leaves L0 to L2. The GFP signal was first observed on L3 and L4 at 3 or 4 days postinoculation (dpi), along with class I to IV veins (Fig. 2C). GFP expression had been silenced in these leaves at the time of inoculation. In class III and IV veins, GFP fluorescence was associated with visible vein clearing. Mosaic symptoms developed in L4 and the upper leaves, where GFP fluorescence was detected, coinciding well with yellow tissues (Fig. 2A and B). No green fluorescence was detected in the tiny dark green islands that were surrounded by yellow tissues. However, large dark green tissues often contained irregularly shaped, light yellow regions that exhibited weak green fluorescence. Once GFP silencing was broken, GFP fluorescence remained until the leaves were apparently senescent.

A replicase gene mutation results in a symptomless phenotype and defective suppressor activity.

That the suppression of GFP silencing or the activity of the ToMV-encoded suppressor was associated with mosaic symptoms suggested the possibility that some symptomless mutants may be suppressor defective. On examining some attenuated strains, we found that ToMV L₁₁ and L₁₁A represent this type of virus, spreading systemically without causing symptoms or restoring GFP fluorescence in G3Sm2 (Fig. 3A). The L₁₁A genome differs from the wild-type L genome at 10 positions, including three coding changes (21, 25). L₁₁, the parental attenuated strain of L₁₁A, has the first mutation at nt 1117 (25), suggesting that this is the causative mutation. However, because the L₁₁ genome has been sequenced only partially, it was possible that another mutation in the L₁₁ genome was responsible for the defect in suppressor function. Therefore, we prepared several recombinant viruses to identify with certainty the mutation that causes defective suppressor activity.

TLJA1 (in vitro transcript from pTLJA1), which differs from TLW3 (the wild-type construct equivalent to L) at four positions, including all three coding changes (Fig. 3A), displayed the same phenotype as L₁₁A in every characteristic examined; e.g., it caused no systemic symptoms on Samsun tobacco, formed normal local lesions on Xanthi nc tobacco with the N gene and, importantly, failed to suppress GFP silencing in G3Sm2. TLJ contains one of the four substitutions of TLJA1, and TLA contains the other three (Fig. 3A). TLJ caused no mosaic symptoms and did not restore GFP fluorescence in G3Sm2 (Fig. 3B). On the other hand, TLA caused mild mosaic symptoms, and GFP fluorescence was detected in systemically infected leaves of TLA-infected G3Sm2, coinciding with the visible symptoms, although its intensity was weaker than that seen on L- or TLW3-infected mosaic leaves (Fig. 3B).

To confirm that TLJ and TLJA1 caused systemic infection but did not restore GFP expression in G3Sm2, leaf extracts were subjected to protein analysis (Fig. 3C). Yellow tissues contain a high level of virus, whereas tiny dark green islands are virus free or contain a low level of virus (2). Consistent with this description, high levels of CP were detected in yellow tissues of TLW3- or TLA-induced mosaic leaves, where GFP was also expressed at high levels (Fig. 3C, lanes 6 and 9). In dark green islands, CP was often detected at low levels and, correspondingly, GFP levels were slightly increased (Fig. 3C, lanes 5 and 8). In the nonsymptomatic leaves of TLJ- or TLJA1-infected G3Sm2, no increase in GFP levels was observed, even though CP accumulated (Fig. 3C, lanes 7 and 10). It is therefore likely that the mutation at nt 1117 impairs the activity of a factor that functions in the active suppression process.

Although TLJ and TLJA1 caused no systemic symptoms in leaves and did not restore GFP fluorescence in the corresponding leaves of G3Sm2, faint GFP fluorescence was detected around regions showing very weak vein clearing in L3 and L4, but only at an early stage of infection (data not shown), as was the case for L₁₁ and L₁₁A. Interestingly, this GFP expression did not persist and became fully silenced within a few days. Such transient restoration of GFP expression was more pronounced when these viruses were inoculated onto small leaves, ∼3 cm or less in length (Fig. 3D). Infection with either suppressor-active or suppressor-defective constructs restored GFP fluorescence on these small L0 leaves as well as on a few upper leaves, accompanied by slight chlorotic or vein-clearing symptoms. A few days later, these GFP signals in TLJ- or TLJA1-inoculated plants weakened (Fig. 3D) and then disappeared. No GFP fluorescence was detected in newly emerging leaves, as was the case after infection with the standard inoculation method. In TLW3-infected plants, GFP fluorescence did not disappear once it was restored. Therefore, the mutation at nt 1117 did not completely inactivate ToMV-encoded suppressor activity.

Production of ToMV-specific small RNAs in infected leaves.

To better understand how the ToMV-encoded suppressor functions, we analyzed ToMV- and gfp-derived RNAs in upper leaves of mock-, TLW3-, or TLJ-inoculated plants. Total RNAs for the TLW3 genome were detected at high levels, severalfold higher than for the TLJ genome (Fig. 4A, upper panel), as expected from a previous report for L₁₁A (26). GFP mRNA was detected at similar levels in mock-, TLW3-, or TLJ-inoculated G3Sm1 plants (Fig. 4A, lower panel, lanes 4 to 6). In G3Sm2 leaves, TLW3 infection increased the expression of GFP mRNA (Fig. 4A, lane 8), reflecting the suppression of GFP silencing in yellow tissues (Fig. 2 and 3). GFP mRNA was detected at low steady-state levels in mock- and TLJ-infected G3Sm2 leaves, which were nonsymptomatic and displayed no GFP fluorescence (Fig. 4A, lanes 7 and 9).

Low-molecular-weight RNAs were enriched from total RNAs and then subjected to small-RNA analysis (Fig. 4B). ToMV-derived small RNAs were detected infection specifically in either TLW3- or TLJ-infected leaves. No clear difference was found in the overall distribution patterns of TLW3- and TLJ-derived small RNAs. The minus-strand small RNAs migrated at 21 to 23 nt and at ∼25 nt, like gfp siRNAs in G3Sm2. The ∼25-nt ToMV siRNA band was often faint (in three independent series of experiments), compared with the corresponding band of gfp siRNAs. The plus-strand small RNAs of ToMV migrated at 21 to 24 nt and also contained abundant longer species (>25 nt). gfp siRNAs were detected only in G3Sm2 (Fig. 4B, lanes 7 to 9) and were not generated in G3Sm1 during infection (Fig. 4B, lanes 4 to 6).

These observations indicate that ToMV-specific siRNAs (21- to 25-nt species) were produced after infection. However, because it was unclear when the leaves were infected systemically and how the viruses were distributed there (in particular, in symptomless leaves of TLJ-infected plants), the timing and distribution of ToMV siRNAs in the upper leaves were not resolved. Therefore, we examined siRNA production in L0 of TLW3- and TLJ-inoculated G3Sm2 over time (Fig. 4C and D). TLW3 accumulated vigorously (Fig. 4C, lanes 5 to 8). Because TLW3 levels were very high at 6 and 8 dpi, the total RNAs loaded were adjusted on the basis of the amounts of rRNAs stained (Fig. 4C and D). The accumulation of TLJ reached a nearly maximum level at approximately 4 dpi (Fig. 4C, lanes 9 to 12), consistent with previous reports for L₁₁A (15, 26). GFP mRNA levels did not change significantly upon infection, consistent with the observation that no GFP fluorescence was restored in L0.

TLW3-derived small RNAs were detected at 2 dpi, and their levels increased rapidly as viruses accumulated (Fig. 4D, lanes 5 to 8). Therefore, virus-derived small RNAs were generated in infected tissues from an earlier stage of infection, suggesting that the siRNA production step was not the major target of the ToMV-encoded suppressor. TLJ-derived small RNAs accumulated to levels similar to those of TLW3-derived small RNAs (Fig. 4D, lanes 9 to 12). However, the ratio of siRNA to genomic RNA appeared to be higher in TLJ-infected leaves than in TLW3-infected leaves after infection proceeded (e.g., >4 dpi). gfp siRNA levels were not detectably influenced by infection (Fig. 4D, lanes 5 to 12).

ToMV-encoded suppressor inhibits de novo formation of the PTGS system in L0.

ToMV small RNAs accumulated in L0, whereas GFP silencing was not interrupted in L0. A simple explanation is that the ToMV-encoded suppressor inhibits de novo formation of the sequence-specific RNA degradation machinery. However, in G3Sm2 plants showing the weak silencing phenotype, GFP expression was not silenced in the interveinal regions even when the leaves became senescent. Therefore, it was still possible that the PTGS system was not efficiently established de novo in L0 that we used under standard inoculation conditions. Therefore, we examined directly whether the cells in L0 were capable of establishing the ToMV-targeted PTGS machinery de novo and whether the ToMV-encoded suppressor inhibited its establishment.

For this purpose, we prepared two constructs carrying the GFP ORF between the duplicated CP subgenomic promoters, TLBN.G3 and TLBN.G3(J) (Fig. 5A). TLBN.G3 carries the wild-type replicase gene, and TLBN.G3(J) has a mutation responsible for the suppressor-defective phenotype. If PTGS is activated upon ToMV infection in L0 and if the replicase gene is involved in the suppression of PTGS there, GFP signals from TLBN.G3(J) will weaken even in L0 in due course. Unlike wild-type ToMV, which causes chlorosis on L0, these GFP-tagged constructs did not cause visible symptoms. Blue-light irradiation revealed infection sites, hereafter referred to as GFP lesions. As shown in Fig. 5B, TLBN.G3(J) formed ring-shaped GFP lesions on both NT tobacco and G3Sm1, whereas TLBN.G3 formed round GFP lesions (Fig. 5B). In the center of the TLBN.G3(J)-induced lesions, a decrease in GFP fluorescence was observed from 3 or 4 dpi, after which the central region enlarged. Importantly, in G3Sm1, GFP fluorescence emitted from the central portion of each lesion became less intense than the fluorescence emitted from areas outside the lesion (Fig. 5B), suggesting that transgene-derived GFP expression was silenced simultaneously.

We used RT-PCR analysis to confirm this finding at the RNA level. Approximately the same amounts of total RNAs were subjected to cDNA synthesis, and then the transgene-derived GFP mRNA sequence and the GFP-tagged recombinant ToMV sequence were specifically amplified by PCR (Fig. 5C). The virus-derived PCR band was amplified only from tissues inside GFP lesions (Fig. 5C, lanes 3, 5, 7, and 9). The GFP mRNA-derived band was amplified to a lesser degree from the central portions of TLBN.G3(J)-induced lesions (Fig. 5C, lanes 7 and 9) than from areas outside the lesions (lanes 8 and 10), whereas the actin mRNA-derived band, used as a reference, was amplified similarly from both regions (lanes 7 to 10). These results indicate that the decrease in GFP fluorescence in the central portions of the TLBN.G3(J)-induced lesions was actually associated with a decrease in the level of transgene-derived GFP mRNA. The GFP mRNA-derived band was amplified similarly from tissues inside and outside the TLBN.G3-induced lesions and from mock-inoculated leaves (Fig. 5C, lanes 1 to 6). These observations indicate that the cells in L0 were competent to establish PTGS directed against the ToMV genome but that the ToMV-encoded suppressor encoded by TLBN.G3 blocked the completion of this PTGS system. Furthermore, it is clear that the replicase gene is involved in this suppression process.

ToMV infection did not inactivate preestablished PTGS directed against the gfp sequence in L0 of G3Sm2.

As described above and shown in Fig. 4C, no restoration of GFP fluorescence and no increase in the GFP mRNA level was observed in L0 of ToMV-infected G3Sm2. These observations suggest that ToMV infection did not affect preestablished gfp-specific PTGS activity in L0. To confirm this idea, we examined whether ToMV infection influenced the activity of gfp-targeted PTGS in G3Sm2, as expressed by the resistance of the plants to GFP-tagged viruses.

G3Sm2 was first inoculated with ToMV at a high concentration (∼500 lesion-forming units) or mock inoculated. At 4 dpi, the same leaves were challenge inoculated with a mixture of PVX.erG3 and PVX.GUS (recombinant PVX expressing erG3GFP and GUS, respectively). As shown in Fig. 6, PVX.GUS infected both ToMV- and mock-inoculated G3Sm2 leaves similarly, whereas PVX.erG3 did not. Both viruses infected either ToMV- or mock-inoculated NT tobacco. Therefore, infection with ToMV did not affect (or only negligibly affected) preexisting gfp-specific PTGS activity in inoculated leaves.

The 130K replication protein is a PTGS suppressor.

A replicase gene mutation responsible for the suppressor-defective phenotype of ToMV mutants causes an amino acid change, implying that the 130K replication protein itself is a suppressor. To clarify this point, we used an agroinfiltration assay with N. benthamiana (Fig. 7). This assay is based on the observation that when Agrobacterium harboring a gfp gene under the control of the cauliflower mosaic virus 35S promoter (35S/GFP) infiltrated into N. benthamiana leaves, GFP expression is shut off by local PTGS within a few days after infiltration, whereas local PTGS is inhibited when a suppressor-encoding gene is cointroduced (41).

FIG. 7.

Suppression of local GFP silencing by ToMV 130K replication protein. GFP-transgenic N. benthamiana (G3Nb3) was agroinfiltrated with the 35S/GFP gene and a gene for the ToMV 130K replication protein (130K), its frameshift mutant [130K(FS)], the TLJ 130K protein [130K(J)], or PVY HC-Pro. Vector, cointroduction of a vector. (A) GFP fluorescence visualized under blue-light irradiation. The infiltrated plant was kept at 22°C for 5 days. Arrows indicate the silencing of resident gene-derived GFP expression around the infiltrated zone. (B) Expression of the 130K and 130K(J) proteins in infiltrated zones. Proteins were extracted from infiltrated plants kept at 22°C for 2 days. (C) GFP fluorescence visualized as in panel A. The plant was kept at 22°C for 1 day to establish transformation and at 28°C for 3 days. (D) Detection of GFP in infiltrated zones of plants treated as in panel C. (E) Northern analysis of GFP mRNA in infiltrated zones. The plants were kept at 22°C for 1 day and at 28°C for 2 days. The lower panel shows ethidium bromide (EtBr) staining of the same samples. (F) Analysis of siRNAs containing the gfp sequence in infiltrated zones. Small RNAs in each lane were derived from total RNAs in the corresponding lane of panel E. The gel was stained with EtBr prior to blotting (bottom panel). Positions of 20- and 25-nt RNA markers are indicated by closed and open arrowheads, respectively.

When the 35S/GFP gene was agroinfiltrated into G3Nb3, a transgenic N. benthamiana line constitutively expressing GFP, GFP fluorescence in the infiltrated zone decreased from ∼2 days after infiltration. The concurrent silencing of GFP expression from the resident gene was first observed in marginal regions, where GFP levels were apparently lower than in noninfiltrated regions (Fig. 7A). When the ToMV 130K protein or the well-known suppressor, HC-Pro of PVY (4), was coexpressed, GFP silencing was inhibited and, consequently, a higher level of fluorescence was observed in the infiltrated zone (Fig. 7A). GFP expression was not prolonged when a frameshift mutant of the 130K protein [130K (FS)] was expressed with the 35S/GFP gene (Fig. 7A). The expression of the 130K protein in the infiltrated zone was confirmed by Western blotting (Fig. 7B, lane 4). No enhanced expression of the resident GFP gene occurred when the 130K protein was expressed alone (data not shown). Therefore, the ToMV 130K protein acted as a suppressor of PTGS in the absence of replicating virus.

The TLJ-encoded 130K protein, designated 130K(J), also exhibited suppressor activity under the assay conditions shown in Fig. 7A, although this activity was weaker than that of the wild-type 130K protein (data not shown). However, it is possible that the effect of the mutation is temperature dependent, because we found that the central holes of TLBN.G3(J)-induced GFP lesions on tobacco were larger at 28°C than at lower temperatures. Furthermore, it was recently shown that PTGS is more strongly induced at higher temperatures in N. benthamiana (36). Therefore, we examined the activity of the 130K(J) protein at 28°C.

As shown in Fig. 7C, the 130K(J) protein still suppressed local GFP silencing; however, the resulting GFP fluorescence declined earlier than when the wild-type 130K protein was expressed. Western blotting showed that the intensity of green fluorescence reflected the accumulation of GFP in the infiltrated zone (Fig. 7D), confirming that the suppressor activity of the 130K(J) protein is weaker than that of the wild-type 130K protein. Furthermore, the amounts of immunoreactive GFP (Fig. 7D) were in good agreement with the amounts of GFP mRNA in the corresponding samples extracted a day before (Fig. 7E). Small RNAs with the gfp sequence accumulated in the infiltrated zone, irrespective of whether the 130K protein was expressed (Fig. 7F), supporting the idea that the ToMV-encoded suppressor inhibits the PTGS pathway at a step after the generation of siRNAs in infected cells.

The suppressor activity exhibited by the wild-type 130K protein was weaker than the activity of HC-Pro (Fig. 7C to E). Incomplete inhibition of silencing by the wild-type 130K protein in N. benthamiana was also confirmed by the observation that TLBN.G3 induced ring-shaped GFP lesions on both NT and GFP-transgenic N. benthamiana leaves, although the central holes were smaller than those of TLBN.G3(J)-induced lesions (data not shown).

DISCUSSION

In this work, we investigated ToMV-encoded suppressor activity by using GFP-constitutive (G3Sm1) and GFP-silenced (G3Sm2) transgenic tobacco lines. Wild-type ToMV (L and TLW3) infection resulted in the suppression of GFP silencing in G3Sm2 together with visible systemic symptoms in the leaves, whereas no suppression of GFP silencing was observed in the nonsymptomatic upper leaves of G3Sm2 infected with the L-derived attenuated strains, L₁₁ or L₁₁A. Mapping the causative mutation that results in the suppressor-defective phenotype of the attenuated strains showed that the replicase gene is involved in the suppression of PTGS (Fig. 3). An agroinfiltration assay indicated that the 130K replication protein itself is able to suppress local silencing in the absence of other ToMV-encoded proteins (Fig. 7).

PTGS is known to be associated with the accumulation of siRNAs of 21 to 25 nt (10, 11). As in animals, some sorts of siRNAs (the class of short siRNAs) are thought to be incorporated into the RISC to guide the complex to the target in plants (11, 42). In ToMV-infected leaves, this class of siRNAs were found with both polarities, and their levels increased as the virus accumulated (Fig. 4), suggesting that ToMV-directed PTGS was initiated in the infected cells. These findings also suggest that the major target of the ToMV-encoded suppressor is downstream from the production of siRNAs. Analyses with GFP-tagged ToMV constructs demonstrated that the suppressor of wild-type ToMV prevents the establishment of the ToMV-directed PTGS system (Fig. 5). Furthermore, resistance to GFP-tagged PVX in the GFP-silenced leaves of G3Sm2 was not broken by ToMV infection (Fig. 6), indicating that ToMV infection did not detectably inhibit the activity of the preexisting sequence-specific PTGS machinery, namely, the RISC. Taken all, it is likely that the ToMV suppressor acts after the production of siRNAs and inhibits their utilization, so that the RISC machinery is not newly formed.

The replicase mutation impaired the activity that prevents de novo formation of the virus-targeting RISC (Fig. 5). However, agroinfiltration experiments indicated that the mutation in the replication protein weakened the suppressor activity but did not completely abolish it (Fig. 7). Incomplete inhibition of the PTGS pathway would explain, at least partly, why TLJ and the other attenuated strains with the same replicase mutation can infect systemically but accumulate to lower levels than wild-type ToMV. The accumulation of L₁₁ and L₁₁A in inoculated leaves ceases a few days after inoculation, when the rate of L₁₁ and L₁₁A RNA synthesis is down-regulated (15, 26). These findings are consistent with our observation that TLBN.G3(J) formed ring-shaped GFP lesions (Fig. 5), in which a decrease in GFP fluorescence was apparent at 3 or 4 dpi.

Suppression of PTGS depends on both the activity of the host PTGS machinery and the activity of a given suppressor. It has recently been shown that PTGS is more strongly induced at higher temperatures in N. benthamiana (36) and, therefore, suppression of PTGS by a suppressor is more difficult at higher temperatures. This would explain the observation that the central holes of TLBN.G3(J)-induced GFP lesions on tobacco were larger at higher temperatures (Fig. 5B). However, it is also possible that the mutation itself has a temperature-sensitive nature. In relation to this, it should be noted that TLJ and other attenuated strains showed the same symptomless and suppressor-defective phenotypes on the systemically infected leaves of tobacco lines at all temperatures examined.

It is generally accepted that siRNAs are generated from the precursor dsRNA by the action of Dicer or its homologs (39, 41, 42). In transgene-mediated PTGS, dsRNA may be formed by the annealing of sense and antisense transcripts or directly derived from an inverted repeat-containing transcript (40). dsRNA could also be formed from structured ssRNAs through the amplification system in which host-encoded RdRp is involved (5, 24). The ToMV dsRNA, or the precursor of siRNAs, may be generated by virus-encoded RdRp during replication or by the host amplification system, which may recognize the abundantly synthesized (sub)genomic RNAs in the infected cells as aberrant ssRNAs. The minus-strand ToMV siRNAs, which migrated like the sense and antisense gfp siRNAs detected in G3Sm2 (Fig. 4B and D), probably reflect the Dicer-like activity in the infected cells. On the other hand, the plus-strand small RNAs contained longer species (>25 nt) in addition to 21- to 25-nt siRNAs. These >25-nt RNAs must be derived from single-stranded (sub)genomic RNAs, because no corresponding minus-strand RNAs accumulated. These plus-strand, >25-nt RNA fragments might be exclusively degradation intermediates. However, it is possible that some might be used in the amplification cycle of PTGS (e.g., as primers or templates for dsRNA synthesis).

Irrespective of how ToMV dsRNA and siRNAs are generated in infected cells, the dsRNA must be produced after replication proceeds. During infection by ToMV, the 130K and 180K replication proteins are translated directly from the genomic RNA. Of these, the 180K protein contains the RdRp domain and is, therefore, absolutely required for the minus-strand synthesis (12). The accumulation of the 130K protein is more than 10-fold greater than that of the 180K protein and accordingly, the 130K protein thus produced would have a good chance of inactivating the host PTGS pathway, before the replication machinery is set up and subsequently synthesizes a large amount of progeny RNAs. Therefore, it is a reasonable strategy for the 130K protein to act as the suppressor to prevent the establishment of PTGS that targets its own genome.

GFP silencing of G3Sm2 was differently suppressed by ToMV infection depending on the positions of the leaves. Under standard inoculation conditions, GFP silencing was not disrupted in L0 (Fig. 4C), where the preexisting RISC activity was not, or negligibly, affected by infection (Fig. 6). This observation suggests that the RISC in mature leaves may have rather a long half-life. In contrast, GFP silencing was interrupted in the vein-clearing regions and also in the small inoculated leaves (Fig. 2C and 3D). In these parts of the plant, GFP expression was apparently silenced at the time of inoculation. This indicates that ToMV can block the silencing system at work in some tissues (e.g., sink leaves) and suggests that ToMV may suppress PTGS by weakening preexisting RISC activity. This suppression might result from an active process in which a virus-encoded or -induced factor is involved, or from a passive process that results from the disturbance of the intracellular milieu caused by the abrupt invasion and subsequent replication of the virus. Another possibility is that in young tissues, the RISC might have a short half-life or that siRNA in the RISC might be rapidly exchanged. Whichever the case, the ability to inhibit the PTGS system, albeit transiently as seen in TLJ-infected plants (Fig. 3D), would facilitate the establishment of infection, because the invading virus would have time to produce sufficient replication protein to suppress the PTGS directed against its own genome.

Mosaic symptoms develop only on leaves that were small or invisible at the time of inoculation (20). Suppression of GFP silencing in the yellow tissues of the mosaic leaves may be explained by the inhibition of siRNA utilization; infection at an earlier stage of leaf development would result in a decrease in the number of active RISC, because it would be diluted during proliferation. It is also possible that the PTGS system is concurrently inhibited in mosaic leaves by an as-yet-unidentified mechanism, as seen in the vein-clearing regions.

ToMV L₁₁A has been used successfully as an attenuated strain to protect tomato plants from infection by severe strains (29). Our finding that the replicase mutation in L₁₁A results in a suppressor-defective phenotype suggests that the cross-protection observed in L₁₁A-infected plants is primarily PTGS-mediated; i.e., PTGS directed against the L₁₁A genome prevents infection of any virus with sufficient homology to L₁₁A. The attenuated phenotype of L₁₁ and L₁₁A may also be attributable to their defective suppressor activity, because viral suppressors often determine pathogenicity (31, 33, 34, 44). However, the replicase gene mutation responsible for the suppressor-defective phenotype did not completely abolish suppressor activity (Fig. 7C). Furthermore, we have found that some symptomless strains are suppressor-active insofar as they were tested with G3Sm2 (data not shown). Therefore, we cannot exclude the possibility that the symptomless and suppressor-defective phenotypes of L₁₁ and L₁₁A derive from a single mutation but are mechanistically distinct.