Host-Specific Cell-to-Cell and Long-Distance Movements of Cucumber Mosaic Virus Are Facilitated by the Movement Protein of Groundnut Rosette Virus

Transcription

1 Virology 260, (1999) Article ID viro , available online at on Host-Specific Cell-to-Cell and Long-Distance Movements of Cucumber Mosaic Virus Are Facilitated by the Movement Protein of Groundnut Rosette Virus Eugene V. Ryabov, Ian M. Roberts, Peter Palukaitis, 1 and Michael Taliansky Virology Department, Scottish Crop Research Institute, Invergowrie, Dundee, DD2 5DA, United Kingdom Received January 27, 1999; returned to author for revision February 22, 1999; accepted May 12, 1999 The cucumovirus, cucumber mosaic virus (CMV), requires both the 3a movement protein (MP) and the capsid protein (CP) for cell-to-cell movement. Replacement of the MP of CMV with the MP of the umbravirus, groundnut rosette virus (GRV), which does not encode a CP, resulted in a hybrid virus, CMV(ORF4), which could move cell to cell in Nicotiana tabacum and long distance in N. benthamiana. After replacement of the CMV CP in CMV(ORF4) with the gene encoding the green fluorescent protein (GFP), the hybrid virus, CMV(ORF4.GFP), expressing both the GRV MP and the GFP, could move cell to cell but not systemically in either Nicotiana species. Immunoelectron microscopic analysis of cells infected by the hybrid viruses showed different cellular barriers in the vasculature preventing long-distance movement of CMV(ORF4) in N. tabacum and CMV(ORF4.GFP) in N. benthamiana. Thus the GRV MP, which shows limited sequence similarity to the CMV MP, was able to support CP-independent cell-to-cell movement of the hybrid virus, but CP was still required for long-distance movement and entry of particular vascular cells required functions encoded by different proteins Academic Press INTRODUCTION 1 To whom reprint requests should be addressed. Fax ppaluk@scri.sari.ac.uk. Plant viruses require one to four viral-encoded proteins to facilitate cell-to-cell (local) or leaf-to-leaf (systemic or long-distance) movement (reviewed by Maule, 1991), and movement occurs via several different mechanisms (reviewed by Carrington et al., 1996). Even among viruses that have movement proteins (MPs) showing sequence similarities, there are differences in requirements for the number of viral-encoded components needed for movement. For example, some viruses, such as tobacco mosaic virus (TMV), require the capsid protein (CP) for long-distance but not cell-to-cell movement (Siegel et al., 1962), whereas other viruses, such as brome mosaic virus (BMV), require CP for cell-to-cell movement (Schmitz and Rao, 1996). By contrast, the two bromoviruses BMV and cowpea chlorotic mottle virus (CCMV) differ in their requirement of CP for cell-to-cell movement (Rao, 1997), although CCMV still requires the CP for long-distance movement (Schneider et al., 1997). A few RNA viruses, such as hordeiviruses (Petty and Jackson, 1990), tombusviruses (reviewed by Russo et al., 1994), and umbraviruses (Reddy et al., 1985), do not require CP for long-distance movement, although in the absence of CP, the efficiency of long-distance movement is sometimes reduced. Thus there appears to be numerous subtle interactions between host and viral gene products to facilitate movement within plants, and different viruses encode different proteins to exploit these subtle interactions. The umbravirus, groundnut rosette virus (GRV), does not encode its own CP (Murant et al., 1995; Taliansky et al., 1996) but nevertheless accumulates and spreads very efficiently within infected plants (Reddy et al., 1985). Although umbraviruses depend on the assistance of luteovirus for aphid transmission, the presence or absence of the luteovirus and its CP does not affect systemic infection by umbraviruses (Reddy et al., 1985). GRV contains a monopartite, single-stranded RNA genome with four open reading frames (ORFs) (Taliansky et al., 1996), two of which encode proteins involved in virus movement. The 27-kDa ORF3 gene product was able to facilitate the long-distance movement of TMV, in place of the TMV CP, in a common host (Ryabov et al., 1999), whereas the overlapping 28-kDa ORF4 gene product was able to facilitate limited cell-to-cell movement of potato virus X (PVX), in place of the PVX triple gene block (TGB)-encoded MPs (Ryabov et al., 1998). The GRV ORF4 gene product shows limited sequence similarity to the 30-kDa, 3a MP of cucumber mosaic virus (CMV) (Taliansky et al., 1996). However, GRV and CMV are taxonomically unrelated. CMV contains a tripartite, single-stranded RNA genome encoding five genes (Palukaitis et al., 1992; Ding et al., 1994) with the 3a MP and the CP both encoded on CMV RNA 3 (Fig. 1). To gain a better understanding of the role of various proteins in virus movement, hybrid viruses were generated in which the CMV 3a gene was replaced by GRV ORF4 in the presence or absence of the CMV CP gene. These hybrid viruses were used to determine (1) whether the GRV ORF4 gene product could potentiate /99 $30.00 Copyright 1999 by Academic Press All rights of reproduction in any form reserved. 98

2 CAPSID PROTEIN DEPENDENCY OF CMV MOVEMENT 99 FIG. 1. Schematic representation of cdna constructs of cucumoviral RNA containing WT and foreign genes. Unmodified cdna constructs (pfny109, pfny209, and pfny309; Rizzo and Palukaitis, 1990) were used for the synthesis of Fny-CMV RNAs 1, 2, and 3, respectively. pf.orf4 was used for the synthesis of Fny-CMV RNA 3, in which the gene for the 3a movement protein was replaced by GRV ORF4 (F.ORF4 RNA 3). pfl.orf4.gfp was generated as a hybrid CMV construct consisting of the 5 half corresponding to Fny-CMV RNA 3 containing GRV ORF4 in place of the 3a movement protein gene and the 3 half corresponding to LS-CMV RNA 3 containing the GFP gene inserted in place of the CP gene (FL.ORF4.GFP RNA 3). the movement of CMV, (2) to what extent such complementation occurred, (3) whether movement of the hybrid virus was restricted to common hosts of CMV and GRV, and (4) whether GRV ORF4-assisted movement was CMV CP dependent. RESULTS GRV MP facilitates cell-to-cell movement of CMV To determine whether the GRV MP could functionally replace the CMV MP, a hybrid RNA 3 was constructed, and accumulation of CMV containing this hybrid RNA 3 was assessed in protoplasts and whole plants. Specifically, the construct pf.orf4 was generated using a fulllength cdna clone of Fny-CMV RNA 3 (Fig. 1) in which the GRV ORF4 replaced the gene encoding the CMV 3a MP. Transcript generated in vitro from pf.orf4 and coinoculated with Fny-CMV RNAs 1 and 2 replicated in Nicotiana benthamiana (Fig. 2A, lane 3) and N. tabacum (Fig. 2B, lane 3) protoplasts to give CMV(ORF4). It may be noted that the ratio between levels of accumulation of particular RNA species in this case was different from that of the corresponding wild-type (WT) Fny-CMV [i.e., although RNAs 1 and 2 of the hybrid CMV accumulated to higher levels compared with the WT CMV, the levels of accumulation of the hybrid RNA 3 and subgenomic RNA 4 were significantly lower (Fig. 2, lane 3 versus lane 1)]. This may have been due to poor replication and/or encapsidation of the hybrid RNA 3. Despite these differences, the hybrid virus CMV(ORF4) containing F.ORF4 as RNA 3 (Fig. 1) was able to infect N. benthamiana systemically, producing symptoms in noninoculated leaves at the same time as for WT Fny-CMV [7 10 days postinoculation (p.i.)], although the symptoms induced by CMV(ORF4) were noticeably milder than those induced by the WT CMV (data not shown). Northern blot hybridization analysis, using radiolabeled probe complementary to the 3 -terminal noncoding sequences of CMV RNA 3, revealed the presence of all four CMV RNAs in FIG. 2. Accumulation of viral RNA in N. benthamiana (A) or N. tabacum (B) protoplasts 48 h after inoculation with transcripts from clones of WT CMV RNAs and hybrid viral RNAs. Northern blot hybridization analyses were of total nucleic acids extracted from protoplasts after electrophoresis in 1.5% agarose gels, with a probe specific to the 3 noncoding region of each CMV RNA. The inocula were CMV RNAs 1 plus 2 and transcripts derived from one of the following RNA 3s: (1) WT RNA 3, (2) FL.ORF4.GFP RNA 3, or (3) F.ORF4 RNA 3. Lane 4 contains the RNAs extracted from mock inoculated protoplasts. Equal amounts of total extracted RNA were analyzed, and the blots were autoradiographed for 8 h. The mobilities of CMV RNAs 1, 2, and 3, as well as the subgenomic RNA 4, are indicated.

3 100 RYABOV ET AL. FIG. 3. Accumulation of viral RNA in N. benthamiana (A) and N. tabacum (B) plants 8 days p.i. with transcripts from full-length clones of WT and hybrid CMV RNAs. RNAs representing the following viruses were used for inoculations: (1) Fny-CMV and (2) CMV(ORF4). Northern blot hybridization analyses were of total nucleic acids extracted from inoculated (i) and noninoculated (n) leaves after electrophoresis in 1.5% agarose gels with a probe specific to the 3 noncoding region of each CMV RNA. Equal amounts of total extracted RNA were analyzed, and the blots were autoradiographed for 8 h. The electrophoretic mobilities of CMV RNA 1 to RNA 4 are indicated. both inoculated and noninoculated leaves of N. benthamiana plants infected by CMV(ORF4) (Fig. 3A, lanes 2i and 2n), confirming the ability of CMV(ORF4) to move long distance. Other radiolabeled probes specific to GRV ORF4, CMV CP, or 3a MP genes were used to confirm the original hybrid nature of the virus (Fig. 4). Moreover, sequence analysis of the viral progeny of CMV(ORF4) did not detect nucleotide changes in the ORF4 (data not shown). Thus the GRV ORF4 MP is able to facilitate movement of the hybrid CMV. In N. tabacum, CMV(ORF4) also moved in inoculated leaves (Fig. 3B, lane 2i) but was unable to move long distance (Fig. 3B, lane 2n), in contrast to either WT CMV (Fig. 3B, lane 1n) or CMV(ORF4) in N. benthamiana (Fig. 3A, lane 2n). In the inoculated leaves of N. tabacum, RNAs 3 and 4 of CMV(ORF4) accumulated to much lower levels than did RNAs 3 and 4 of WT CMV (Fig. 3B, lane 2i versus lane 1i), similar to the situation observed previously in N. tabacum protoplasts (Fig. 2B, lane 3 versus lane 1). Immunoelectron microscopy was used to examine the boundary to long-distance movement of CMV(ORF4) in N. tabacum versus the absence of such a boundary in N. benthamiana. Similar patterns of immunogold labeling (IGL) were detected in samples collected from the inoculated leaves of N. benthamiana plants infected with WT CMV or CMV(ORF4). Significant IGL was observed in mesophyll and bundle sheath cells surrounding minor veins, as well as in some cells of phloem tissues, particularly in phloem parenchyma cells (Table 1; Figs. 5B and 5C). As both these viruses moved long distance efficiently in N. benthamiana, we assume that they can move from phloem parenchyma cells into and through the sieve elements. However, IGL was not detected in sieve elements, probably because of a low concentration of rapidly moving virus and/or the limited sensitivity of the detection method used. Parallel IGL studies were performed on inoculated leaves of N. tabacum plants infected with WT CMV and CMV(ORF4). In both these cases, CMV CP was readily detected in mesophyll, bundle sheath, and phloem-associated phloem parenchyma and companion cells, although the levels of CP accumulation in phloem cells were lower (Table 1; Figs. 5D, 5E, and 5F). However, only WT CMV could move long distance in these plants, apparently crossing boundaries between phloem paren- FIG. 4. Characterization of RNA 3 of viral RNAs accumulating in N. benthamiana plants infected with WT and hybrid CMV. Plants were inoculated with RNAs of (1) Fny-CMV, (2) CMV(ORF4.GFP), or (3) CMV(ORF4), and the inoculated (i) and upper, noninoculated (n) leaves were sampled. Total RNAs extracted from these leaves were analyzed by Northern blot hybridization, after electrophoresis in 1.5% agarose gels. The radiolabeled probes were specific to (A) GRV ORF4, (B) the CMV CP gene, and (C) the CMV 3a gene. The electrophoretic mobilities of CMV RNAs 3 and 4 are indicated.

4 CAPSID PROTEIN DEPENDENCY OF CMV MOVEMENT 101 TABLE 1 Analysis of the Cell Types in Nicotiana benthamiana and Nicotiana tabacum Infected by CMV Expressing WT and Foreign MP N. benthamiana N. tabacum Inoculum M a BS PP C M BS PP C CMV b,c (18/18) (12/12) (5/6) (0/9) (8/8) (12/12) (5/6) (8/9) CMV(ORF4) c (15/15) (12/12) (6/6) (0/9) (14/14) (12/12) (6/6) (8/9) CMV(ORF4.GFP) d (21/21) (13/16) (1/8) (1/12) (0/16) (0/12) (0/6) (0/9) a Cell types in minor veins examined for the presence of a virus: M, mesophyll cells; BS, bundle sheath cells; PP, phloem parenchyma cells; C, companion cells. b Data based on transmission electron microscopy immunogold cytochemical detection. Immunolocalization of virus infection is represented by, indicating the presence of virus within the specified cell type. The number of signs indicates the average relative levels of immunogold labeling. Cell types in which virus antigens were not detected or were detected at extremely low levels (only marginally above the label detected in control preimmune studies) by immunogold labeling are indicated as and, respectively. The number of cells infected over the number of cells analyzed is indicated in brackets. c Immunolocalization using primary antiserum against CMV particles and gold-labeled IgG as the secondary antiserum. d Immunolocalization using primary antiserum against the GFP and gold-labeled IgG as the secondary antiserum. chyma/companion cells and the sieve elements. CMV(ORF4) appeared to be as capable as WT CMV of reaching phloem cells in the inoculated leaves of N. tabacum, but in contrast to WT CMV, CMV(ORF4) could not move through the vascular system. This indicates that the block in the long-distance movement of CMV(ORF4) was associated with a boundary after phloem parenchyma and companion cells in the movement pathway (i.e., with entry into or exit from the sieve elements). Virus particles were isolated and analyzed to determine whether the difference in long-distance movement was associated with differences in levels of virus and/or the composition of viral RNAs encapsidated into virions by WT CMV versus CMV(ORF4). Electron microscopy showed that the CMV(ORF4) particles were essentially similar and did not differ either in appearance from WT CMV particles or in their behavior in immunoelectron microscopy tests (data not shown). However, the yield of CMV(ORF4) particles (20 g/g of leaf tissue) was considerably lower than that of WT CMV (150 g/g of leaf tissue). Moreover, extraction and Northern blot hybridization analysis of RNA from CMV(ORF4) particles isolated from inoculated leaves of N. tabacum plants showed an RNA profile that was rather different from the typical CMV profile (Fig. 6, lane 4 versus lane 2). Although WT CMV preparations contained predominantly RNAs 3 and 4, the levels of these RNA species in the hybrid virus preparations were considerably reduced compared with the levels of RNAs 1 and 2 (Fig. 6, lane 2 versus lane 4). Analysis of virus preparations isolated from N. benthamiana plants infected with WT CMV and CMV(ORF4) showed results similar to those above. Levels of CMV(ORF4) (15 g/g leaf tissue) were considerably lower than those of WT CMV (200 g/g leaf tissue). The proportional content of RNAs 3 and 4 in virus preparations isolated from CMV(ORF4)-infected N. benthamiana plants compared with RNAs 1 and 2 was even less than in CMV(ORF4) preparations isolated from N. tabacum (Fig. 6, lane 4 versus lane 3). Thus extremely poor accumulation of virus particles, particularly containing RNAs 3 and 4, did not prevent CMV(ORF4) from longdistance movement in N. benthamiana but may be one reason for the lack of long-distance spread in N. tabacum. GRV MP facilitates cell-to-cell movement of CMV in the absence of CP, but CP is required for long-distance movement To determine whether the GRV MP could promote the cell-to-cell movement of CMV in the absence of CMV CP, the construct pfl.orf4.gfp was generated, in which the MP and CP genes of a chimeric CMV RNA 3, containing sequences derived from the Fny- and LS-CMV strains, were replaced by the genes encoding the GRV MP and the green fluorescent protein (GFP), respectively (Fig. 1). The replication and movement of in vitro synthesized transcript from pfl.orf4.gfp in either N. benthamiana or N. tabacum plants were assessed by coinoculation with Fny-CMV RNAs 1 and 2 to produce CMV(ORF4.GFP). In N. benthamiana plants, CMV(ORF4.GFP) caused the development of green fluorescent foci in inoculated leaves, which were clearly visible under long-wavelength UV light, starting on the third day p.i. (as in Fig. 7A). This CP-independent cellto-cell movement was not due to stabilization of the CMV RNAs by the ORF4 gene product in the absence of the CMV

5 102 RYABOV ET AL. FIG. 5. Distribution of WT CMV and hybrid viruses in various vascular cell types. (A) A section of a class V vein from N. tabacum used to identify the various cell types examined for the presence of virus. The bar, 7.5 m. (B H) Sections of a class V vein from plants infected with either CMV(ORF4) (B F) or CMV(ORF4.GFP) (G and H) in the inoculated leaves of either N. benthamiana (B, C, G, and H) or N. tabacum (D F) at 14 days p.i. Immunogold labeling of sections B H was done using polyclonal antibodies raised against either CMV CP (B F) or the GFP (G and H) and goat anti-rabbit secondary antibody conjugated to 15-nm gold particles. Bars in B H, 150 nm. BS, bundle sheath; CC, companion cell; CW, cell wall; PP, phloem parenchyma; SE, sieve element; X, xylem; XP, xylem parenchyma. CP because the CMV RNAs were barely detectable in N. benthamiana plants (Fig. 4A, lane 2i), N. benthamiana protoplasts (Fig. 2A, lane 2 versus lane 1), or N. tabacum protoplasts (Fig. 2B, lane 2 versus lane 1) infected by the RNAs of CMV(ORF4.GFP). These latter observations are in agreement with previous reports concerning CMV CP-defi-

6 CAPSID PROTEIN DEPENDENCY OF CMV MOVEMENT 103 FIG. 6. RNA composition of WT and hybrid CMV particles isolated from N. benthamiana (1 and 3) or N. tabacum (2 and 4). Northern blot hybridization analyses were of RNA samples isolated from WT CMV particles (1 and 2) or from CMV(ORF4) particles (3 and 4) with a probe specific to the 3 -noncoding region of each CMV RNA. One-twentieth of the samples obtained from 5 g of leaf material was applied for electrophoresis in 1.5% agarose gels. The electrophoretic mobilities of CMV RNAs 1 4 and exposure times for autoradiography are indicated, at the left and below the blots, respectively. cient mutants (Suzuki et al., 1991; Boccard and Baulcombe, 1993) showing that the absence of CP resulted in considerable reduction of CMV RNA accumulation. This effect may be enhanced in the case of RNAs 3 and 4, possibly due to reduced efficiency of replication of the hybrid RNA 3. Even though the levels of CMV(ORF4.GFP) RNA 3 accumulation were very low in protoplasts, infection by RNAs of this hybrid virus caused the appearance of intense green fluorescence in protoplasts, detected using confocal laser scanning microscopy (data not shown). Similarly, the extent of cell-to-cell movement of CMV(ORF4.GFP) in inoculated leaves seemed to be very efficient (Fig. 7A). Because it was not possible to generate WT CMV expressing the GFP gene, it was impossible to compare the rates of cell-to-cell spread of the CMV(ORF4.GFP) and WT CMV. However, the growth of the green fluorescent foci formed by CMV(ORF4.GFP) was similar to the pattern seen after infection by other plant viruses expressing the GFP gene, such as PVX and TMV (data not shown). Moreover, as was the case with PVX expressing GFP (Ryabov et al., 1998), CMV(ORF4.GFP) accumulated and spread efficiently into palisade and spongy mesophyll cells underlying infected areas of epidermal cells. More than 90% of mesophyll cells were infected by 7 days p.i. (data not shown). Thus in contrast to the CMV 3a MP, which requires CP for virus movement, the MP of GRV acts independently of a CP even when trafficking CMV RNA from cell to cell. Nevertheless, despite this efficient cell-to-cell movement, CMV(ORF4.GFP) did not induce green fluorescence in noninoculated leaves, suggesting the absence of longdistance movement by the hybrid virus. This suggestion was confirmed by confocal laser scanning microscopy of noninoculated leaves, as well as by back inoculation tests, Northern blot hybridization analysis, and RT-PCR, using total RNAs extracted from noninoculated leaves (Fig. 4A, lane 2n, and data not shown). To determine the cellular boundary responsible for the confinement of CMV(ORF4.GFP) to the inoculated leaves, tissue samples were collected from the inoculated leaves of N. benthamiana plants 14 days p.i. and were analyzed by immunoelectron microscopy. Using antibodies against GFP, heavy IGL patterns were detected in mesophyll and bundle sheath cells surrounding minor veins (Table 1; Figs. 5G and 5H). In contrast, in the phloem tissues containing phloem parenchyma and companion cells, immunogold label was either absent or present at levels that were only marginally above the level of label detected in studies using preimmune serum (Table 1; Figs. 5G and 5H). These results show that the bundle sheath phloem interface appears to be a boundary for the systemic movement of CMV(ORF4.GFP). In control experiments with WT CMV and using antibodies against the CP, significant levels of immunogold label were obtained not only in mesophyll and bundle sheath cells but also in some phloem cells, particularly in the phloem parenchyma cells (Table 1). N. benthamiana is a permissive host common to CMV and GRV. To determine whether the function of the GRV FIG. 7. The development of fluorescence in a N. benthamiana plant (A) and cells of N. tabacum (B and C) infected with a GFP-expressing derivatives of CMV. (A) A plant infected with CMV(ORF4.GFP) photographed under long-wavelength UV light at 5 days p.i., showing fluorescent foci in inoculated (In) leaves. (B) Confocal image of a single epidermal cell infected with CMV(GFP) (4 days p.i.) obtained using CMV containing RNA 3 from pf:gfp/cp (Canto et al., 1997) as a control. (C) Confocal image of a multiple cell infection site of CMV(ORF4.GFP) at 4 days p.i.

7 104 RYABOV ET AL. MP require specific adaptations to host plant factors, CMV(ORF4.GFP) was inoculated to N. tabacum L. cv. Samsun and Cucumis sativus L., which are systemic hosts for CMV but cannot be infected with GRV. In inoculated leaves of N. tabacum, this hybrid virus caused the development of green fluorescent foci visible under both long-wavelength UV light (not shown) and the confocal laser scanning microscope (Fig. 7C), as it did on N. benthamiana (Fig. 7A). Moreover, in N. tabacum, CMV(ORF4.GFP) spread efficiently into palisade and spongy mesophyll cells underlying infected areas of epidermal cells but did not move long distance (data not shown), similar to the situation in N. benthamiana (Fig. 7A). Attempts to identify the cellular boundary restricting infection of CMV(ORF4.GFP) to the inoculated leaves on N. tabacum were unsuccessful because immunoelectron microscopy did not detect any significant immunolabeling using antibodies against GFP, either in phloem parenchyma and companion cells or in mesophyll and bundle sheath cells surrounding minor veins (Table 1). Thus the level of accumulation of GFP, and presumably the level of accumulation of CMV(ORF4.GFP), was much lower in N. tabacum than in N. benthamiana. In contrast to infection in N. benthamiana and N. tabacum, infection by CMV(ORF4.GFP) in C. sativus was restricted to single inoculated epidermal cells (data not shown). These results suggest that the GRV ORF4-encoded protein may function as a cell-to-cell MP in tobacco but not in cucumber plants. DISCUSSION The observations reported here have different consequences for our understanding of movement by GRV and CMV. In the case of GRV, these results confirm previous suggestions that the GRV ORF4-encoded protein is a GRV cell-to-cell MP (Ryabov et al., 1998). This MP can function in a host common to CMV and GRV (N. benthamiana), as well as in another host of CMV, which is not a host for GRV (tobacco), but not in a second host for CMV that is not a host for GRV (cucumber). Thus it is clear that the reason for the resistance of tobacco to GRV is not due to the failure to move cell to cell. The data also show that the GRV ORF4-encoded protein may function in nonvascular tissues of inoculated leaves but cannot overcome the bundle sheath phloem interface without some other helper function. This function may be provided by the ORF3 gene product, which has the ability to support the long-distance movement of a heterologous viral RNA (TMV), in the absence of the GRV ORF4 gene product, but together with the TMV MP (Ryabov et al., 1999). Because ORFs 3 and 4 of GRV overlap each other, similar to the MP and CP genes of luteoviruses (Taliansky et al., 1996), the GRV ORF3-encoded protein may be either a primitive coat protein that does not form capsids or a degenerate CP, no longer able to form capsids and not needed for cell-to-cell movement in epidermal, mesophyll, or bundle sheath cells but functioning in concert with the ORF4-encoded MP to facilitate movement into and through the vasculature. It is noteworthy that the efficiency of cell-to-cell transport of CMV RNA by the GRV MP is rather high and comparable with those of WT viruses like PVX or TMV. In contrast to the results presented here with CMV RNA, the GRV MP trafficked RNA of another unrelated virus, PVX with a deleted TGB, poorly and only to a limited extent, and also only in epidermal and neighboring mesophyll cells of N. benthamiana (Ryabov et al., 1998). Thus some properties of viral RNA itself, such as specific sequences, secondary structure, or size, may be important for trafficking by GRV and perhaps also by other viral MPs. It also cannot be completely ruled out that levels of the GRV MP or the timing of expression of GRV ORF4, from the hybrid PVX genome described by Ryabov et al. (1998), was not well coordinated with movement events. However, this suggestion does not seem very likely because the GRV MP efficiently facilitated the cell-to-cell movement of a PVX RNA that contained the TGB but lacked the PVX CP gene (Ryabov et al., 1998). It is more likely that the GRV MP is essential but insufficient for efficient virus movement and requires additional virusencoded factors as has been suggested earlier for the TMV 30-kDa MP (Taliansky et al., 1992). For CMV, such a factor may be one of the replication-associated proteins. With PVX coding for multiple MPs, one or more of the TGB-encoded proteins might complement functions of the GRV MP, to allow PVX RNA to move efficiently from cell to cell. In the case of CMV, the observations presented here indicate that the CMV MP has sequences that enable it to move cell to cell in a broader range of species than the GRV MP. Moreover, the data indicate that the CMV CP has a function beyond protection of viral RNAs because neither a high steady-state level of virus accumulation (Fig. 3; Canto and Palukaitis, 1998) nor the formation of virus particles (Kaplan et al., 1998) is required for CMV cell-to-cell movement. This CP function is required by the CMV MP for cell-to-cell movement but not for the GRV MP (Fig. 7A). The ability of the CMV MP to traffic itself and RNA through plasmodesmata in the absence of CP in microinjection experiments (Ding et al., 1995) appears in contradiction to the data from infectivity experiments. This may be explained by either of the two following models for the role of the CP in CMV movement: (1) the CP may be needed to inhibit some host response elicited by virus replication that prevents movement. This either may not be induced in microinjected cells or can be overcome by the massive amount of MP injected. (2) The CP may be needed to either directly or indirectly provide some chaperone-like function needed for virus movement (through activation of host factors), such as to conformationally

8 CAPSID PROTEIN DEPENDENCY OF CMV MOVEMENT 105 change the MP. Such a function has been proposed for the cell-to-cell trafficking by the CMV MP (Kragler et al., 1998). Again, the microinjected MP, which has been subjected to denaturation and renaturation, may contain a subpopulation with the appropriate conformation needed to facilitate movement in the absence of the function expressed by the CP. This model would also explain why the CMV MP did not function in place of the BMV MP in a BMV CMV MP hybrid, whereas a variant of this hybrid, in which a deletion and frameshift in the CMV 3a gene resulted in a MP with a deletion of the C-terminal 33 amino acids, did function for movement (Nagano et al., 1997). That is, the BMV CP may not be able to facilitate the effects required for the CMV MP to function, but a deletion variant of the MP may already contain a subpopulation of MP with the conformation needed for trafficking. This may also explain why hybrid viruses, in which the CPs of BMV and CMV were exchanged for each other, were unable to move in common hosts (Osman et al., 1998). Again, this may be due to the inability of the respective CPs to provide the function required for the heterologous MPs. The ability of the GRV MP to facilitate the cell-to-cell movement of CMV(ORF4.GFP) may be because the GRV MP either is not affected by some host response that prevents movement, or does not require the chaperone-like activity. It is interesting that the same barrier for long-distance movement of CMV(ORF4.GFP) identified here in N. benthamiana has been identified in cucumber for a reassorted cucumovirus containing RNAs 1 and 2 of Fny- CMV and RNA 3 of tomato aspermy virus (Thompson and García-Arenal, 1998), as well as in a transgenic line of tobacco expressing replicase-mediated resistance to CMV (Wintermantel et al., 1997) and in a line of soybean showing resistance to CCMV (Goodrick et al., 1991). By contrast, CMV(ORF4), a hybrid CMV containing the GRV MP and the CMV CP, was able to cross the bundle sheath phloem interface and spread long-distance in N. benthamiana. Thus the CMV CP, possibly in a concert with GRV MP, is able to modify interconnections between bundle sheath cells and phloem cells, allowing the longdistance movement of the hybrid virus in this plant species but not in N. tabacum. Experiments with CMV(ORF4) showed that this hybrid virus was able to enter phloem parenchyma and companion cells of vascular tissues in inoculated leaves of N. tabacum, demonstrating that the presence of CP allowed the hybrid virus to pass beyond the bundle sheath phloem barrier. However, further stages of the long-distance movement process did not occur. The lack of long-distance spread by CMV(ORF4) may be due to the inability to either enter or exit the sieve elements in N. tabacum. The process of the entering into sieve elements might involve compatible functional interactions between the viral CP and MP and some hypothetical host phloem factors. In that case, it may be suggested that GRV ORF4 MP is incompatible with the N. tabacum factor, although it can function in N. benthamiana. As a result, CMV(ORF4) either might stop at the boundary between phloem parenchyma companion cells and the sieve elements or enter into, move through, but not exit from the sieve elements. This boundary is similar to that described for tobacco etch virus (TEV), both for a mutation in the central region of the helper component-proteinase of TEV, which prevented the virus from moving long distance, by blocking either entry into or exit from the sieve elements (Cronin et al., 1995), and in a line of tobacco resistant to systemic infection by WT TEV (Schaad and Carrington, 1996). Another reason for the lack of systemic infection may be related to the fact that CMV most probably moves through sieve elements in the form of virions (Blackman et al., 1998; Kaplan et al., 1998). The low levels of CMV(ORF4) particle accumulation would not in itself explain why long-distance movement occurred in N. benthamiana but not in N. tabacum because transgenic N. tabacum expressing CMV RNA 1 and inoculated with CMV RNAs 2 and 3 were able to support the longdistance movement of CMV RNAs 2 4, even though the levels of accumulated virus were times less than what occurs during infection by WT CMV (Canto and Palukaitis, 1998). However, in the latter case, the ratio of accumulating viral RNAs 2 4 was very similar, whereas in this study, the ratios of RNAs 1 plus 2 to RNAs 3 plus 4 were quite abnormal, with much lower levels of RNAs 3 and 4 accumulating. A similar situation was observed in plants inoculated with CMV containing two RNA 3s: one with the MP and the other with the CP replaced by GFP. In that case, cell-to-cell movement occurred but not systemic infection, and although the level of total CMV accumulation was lower than that for WT CMV in leaves, the levels of RNAs 3 and 4 were much lower than the levels of RNAs 1 and 2 (Canto et al., 1997). This observation is also similar to one made in cowpea plants resistant to the long-distance movement of CCMV: CCMV accumulated in the inoculated leaves of a resistant line 12- to 20-fold less than in a susceptible line, and resistance was manifested in lower levels of encapsidated RNA 3 (Wyatt and Kuhn, 1979), even though long-distance movement was controlled genetically by sequences in CCMV RNA 1 (Wyatt and Kuhn, 1980). Thus low levels of virus particles, which contain abnormal ratios of RNAs 1 or 2 to RNA 3, may have an effect on the ability of the virus to establish a systemic infection, and different hosts may have different thresholds for this abnormal ratio. MATERIALS AND METHODS Plasmids and generation of chimeric cdna constructs and mutants Plasmids pfny109, pfny209, and pfny309, containing full-length cdna copies of Fny-CMV genomic RNAs 1, 2,

9 106 RYABOV ET AL. and 3, respectively, have been described previously (Rizzo and Palukaitis, 1990) (Fig. 1). The GRV cdna clone grmp2 (Taliansky et al., 1996) was used as a template for PCR amplification of GRV ORF4. Plasmids pf:gfp/cp and pl:3a/gfp (Canto et al., 1997), containing sequences corresponding to modified Fny-CMV RNA 3 and LS-CMV RNA 3, were used for generation of CMV RNA 3 constructs containing GRV ORF4 and the GFP gene (as shown in Fig. 1). Standard DNA manipulation techniques (Sambrook et al., 1989) were used for the generation of these constructs. pf.orf4 (Fig. 1). A cdna fragment of the 5 -terminal noncoding region of Fny-CMV RNA 3 fused with GRV ORF4 was amplified by overlap extension PCR (Higuchi et al., 1988) using pfny309 and grmp2 as templates and flanking oligonucleotides. The latter were (1) 5 -AA- CAGCTATGACCATG-3, identical to the region preceding the SmaI site located in the multiple cloning region of pfny309 upstream the CMV-specific sequences, as the forward primer; (2) 5 -GGCCACTAGTTACGTCGCTTT- GCG-3, with an SpeI site preceding 14 nucleotides (nt) complementary to those of the 3 end of the GRV ORF4, as the reverse primer; and two (3 and 4) complementary mutagenic oligonucleotides. Primer 3 was 5 -GTTAGATT- TCCCGAGGCATGTCTTCGCAAGTGGC-3, with the 5 half identical to terminal nt of the Fny-CMV RNA 3 5 noncoding region and the 3 half identical to the first 17 nt of GRV ORF4. The resulting fragment, digested with SmaI and SpeI restriction endonucleases, was used to replace the SmaI SpeI fragment of pf:gfp/cp (Canto et al., 1997) to give pf.orf4. pfl.orf4.gfp. A cdna fragment of the plasmid pl: 3a/GFP (Canto et al., 1997) containing the intergenic and 3 -terminal noncoding sequences of LS-CMV RNA 3 and the GFP gene inserted between them was amplified from this plasmid using oligonucleotides 5 (5 -GGCCACTAGT- GTTTTGTTACGTTGTAC-3, with an SpeI site preceding 17 nt identical to those of the 5 end of the intergenic region of LS-CMV RNA 3) as the forward primer and 6 (5 -CTAAAACGACGGCCAGT-3, complementary to the sequence located downstream of the SphI site in the multiple cloning region of pl:3a/gfp) as the reverse primer. The resulting fragment, digested with SpeI and SphI restriction endonucleases, was used to replace the SpeI SphI fragment of pf.orf4 to generate pfl.orf4.gfp (Fig. 1). In vitro transcription and inoculation of plants and protoplasts Viral RNAs were transcribed from full-length cdna clones pfny109, pfny209, and pfny309, representing Fny-CMV RNAs 1, 2, and 3, respectively (Rizzo and Palukaitis, 1990), as well as from chimeric cdna clones generated in this work. The plasmids were linearized and the RNA transcripts were synthesized using T7 RNA polymerase as described previously (Zhang et al., 1994). Transcript mixtures were concentrated by ethanol precipitation before use in mechanical inoculations on aluminum oxide-dusted leaves of N. tabacum cv. Samsun NN, N. benthamiana, or C. sativus. Mesophyll protoplasts were prepared from leaves of N. benthamiana and N. tabacum plants as described by Power and Chapman (1985), and 10 6 protoplasts were electroporated with RNA transcripts, as described by Gal-On et al. (1994). Samples were collected for analysis of the viral RNAs after 48 h of incubation. Detection of GFP fluorescence in plants Plants were illuminated with long-wavelength UV light and photographed as described previously (Baulcombe et al., 1995; Oparka et al., 1995). GFP fluorescence in plant tissues was viewed with a Bio-Rad (Hercules, CA) MRC 1000 confocal laser scanning microscope, using methods that were described previously (Baulcombe et al., 1995; Oparka et al., 1995). Analysis of RNA and virus particles Total RNA was isolated from 0.2 g of leaf tissue and 10 6 protoplasts by the method of Blok et al. (1994). For Northern blot analysis, one-tenth of each total RNA preparation was denatured with formaldehyde and formamide. Electrophoresis was in 1.5% agarose gels, as outlined by Sambrook et al. (1989). RNA was transferred to Hybond N membrane by the capillary method with 20 SSC (3 M sodium chloride and 0.3 M sodium citrate, ph 7.0) and immobilized by UV cross-linking. Hybridization was done as described by Sambrook et al. (1989) with [ 32 P] RNA probes complementary to 3 -terminal noncoding region of each CMV RNA (Gal-On et al., 1994), washed and autoradiographed, all as described by Sambrook et al. (1989). cdna probes corresponding to GRV ORF4 (nt of GRV RNA), the CMV 3a gene (nt of Fny-CMV RNA 3), and the CMV CP gene (nt of Fny-CMV RNA 3) were also used. The probes were labeled with [ - 32 P]dATP using a random primer DNA labeling kit (Boehringer Mannheim, Mannheim, Germany). The standard CMV isolation procedure (Palukaitis et al., 1992) was used to purify virus particles. RNA was extracted from purified virions with phenol SDS and recovered by ethanol precipitation. Tissue sampling, fixation, and embedding N. benthamiana and N. tabacum plants infected with the various WT and hybrid CMVs were used for immunolocalization analysis. Two leaf pieces from each of three different leaves were sampled for each virus isolate. With samples on which symptoms were obvious, the samples included all areas around the lesion; with samples expressing the GFP, the lesions were detected first

10 CAPSID PROTEIN DEPENDENCY OF CMV MOVEMENT 107 under UV irradiation and then sampled in a similar manner. Pieces of leaf were processed essentially as described by Roberts (1994). The individual pieces of leaf were vacuum infiltrated in 3 5ml of 5% glutaraldehyde in piperazine-n,n -bis(2-ethanesulfonic acid) buffer, ph 8.0 (Salema and Brandao, 1973) and allowed to fix for h at room temperature. After fixation, the samples were washed, dehydrated, and embedded in Araldite resin (Ciba-Geigy). Ultrathin sections (silver gold interference colors) were cut on a Reichert Ultracut E ultramicrotome using tungsten-coated glass knives (Roberts, 1975), heat-stretched (Roberts, 1970), and mounted on pyroxylin-filmed nickel grids (Roberts, 1994). IGL IGL was optimized and performed as described previously (Roberts, 1994). Sections were pretreated with Decon 75 before incubation with either CMV-CP antiserum (diluted 1:100; Ding et al., 1995) or GFP antibodies (diluted 1:2000; Molecular Probes, Eugene, OR), which had been cross-absorbed with matching healthy protein (Fasseas et al., 1989) from N. benthamiana or N. tabacum, respectively. The sections were then labeled with goat anti-rabbit IgG/gold (GAR-G, 15 nm; Amersham International). After IGL, the sections were post-stained with uranyl acetate followed by lead citrate before examination on a Philips CM 10 electron microscope. Mapping of infected leaf cells Templates of line drawings of the classic class V veins and the surrounding cells in transverse sections through a leaf were scored for every vein found in each section examined. The lowest score value ( ) represents an IGL level similar to that found in control sections; the highest value ( ) represents a dense, uniform distribution of gold throughout the cell cytoplasm of infected cells. The reproducibility of label density on the same cells in serial sections was high (Roberts, 1994), indicating an accurate relative estimate of virus antigen distribution. ACKNOWLEDGMENTS This work was supported by a grant-in-aid from the Scottish Office Agriculture, Environment and Fisheries Department (SOAEFD) and by an EMBO Fellowship (to E.V.R.). REFERENCES Baulcombe, D.C., Chapman, S., and Santa Cruz, S. (1995). Jellyfish green fluorescent protein as a reporter for virus infections. Plant J. 7, Blackman, L. M., Boevink, P., Santa Cruz, S., Palukaitis, P., and Oparka, K. J. (1998). The movement protein of cucumber mosaic virus traffics into sieve elements in minor veins of Nicotiana clevelandii. Plant Cell 10, Blok, V. C., Ziegler, A., Robinson, D. J., and Murant, A. F. (1994). Sequences of 10 variants of the satellite-like RNA-3 of groundnut rosette virus. Virology 202, Boccard, F., and Baulcombe, D. C. (1993). Mutational analysis of cisacting sequences and gene function in RNA3 of cucumber mosaic virus. Virology 193, Canto, T., and Palukaitis, P. (1998). Transgenically expressed cucumber mosaic virus RNA 1 simultaneously complements replication of cucumber mosaic virus RNAs 2 and 3 and confers resistance to systemic infection. Virology 250, Canto, T., Prior, D. A. M., Hellwald, K.-H., Oparka, K. J., and Palukaitis, P. (1997). Characterization of cucumber mosaic virus. IV. Movement protein and coat protein are both essential for cell-to-cell movement of cucumber mosaic virus. Virology 237, Carrington, J. C., Kasschau, K. D., Mahajan, S. K., and Schaad, M. C. (1996). Cell-to-cell and long-distance transport of viruses in plants. Plant Cell 8, Cronin, S., Verchot, J., Haldeman-Cahill, R., Schaad, M. C., and Carrington, J. C. (1995). Long-distance movement factor: A transport function of the potyvirus helper component proteinase. Plant Cell 7, Ding, B., Li, Q., Nguyen, L., Palukaitis, P., and Lucas, W. J. (1995). Cucumber mosaic virus 3a protein potentiates cell-to-cell trafficking of CMV RNA in tobacco plants. Virology 207, Ding, S.-W., Anderson, B., Haase, H., and Symons, R. H. (1994). New overlapping gene encoded by the cucumber mosaic virus genome. Virology 198, Fasseas, C., Roberts, I. M., and Murant, A. F. (1989). Immunogold localisation of parsnip yellow fleck virus particle antigen in thin sections of plant cells. J. Gen. Virol. 70, Gal-On, A., Kaplan, I. B., Roossink, M. J., and Palukaitis, P. (1994). The kinetics of infection of zucchini squash by cucumber mosaic virus indicate a function of RNA 1 in virus movement. Virology 205, Goodrick, B. J., Kuhn, C. W., and Hussey, R. S. (1991). Restricted systemic movement of cowpea chlorotic mottle virus in soybean with nonnecrotic resistance. Phytopathology 81, Higuchi, R., Krummel, B., and Saiki, R. K. (1988). A general method of in vitro preparation and specific mutagenesis of DNA fragments: Study of protein and DNA interactions. Nucleic Acids Res. 16, Kaplan, I. B., Zhang, L., and Palukaitis, P. (1998). Characterization of cucumber mosaic virus. V. Cell-to-cell movement requires capsid protein but not virions. Virology 246, Kragler, F., Monzer, J., Shash, K., Xoconostle-Cázares, B., and Lucas, W. J. (1998). Cell-to-cell transport of proteins: requirement for unfolding and characterization of binding to a putative plasmodesmal receptor. Plant J. 15, Maule, A. J. (1991). Virus movement in infected plants. Crit. Rev. Plant Sci. 9, Murant, A. F., Robinson, D. J., and Gibbs, M. J. (1995). Genus Umbravirus. In Virus Taxonomy-Classification and Nomenclature of Viruses. Sixth Report of the International Committee on the Taxonomy of Viruses (F. A. Murphy, C. M. Fauquet, D. H. L. Bishop, S. A. Gabrial, A. W. Jarvis, G. P. Martelli, M. A. Mayo, and M. D. Summers, Eds.), pp , Springer-Verlag, Vienna. Nagano, H., Okuno, T., Mise, K., and Furusawa, I. (1997). Deletion of the C-terminal 33 amino acids of cucumber mosaic virus movement protein enables a chimeric brome mosaic virus to move cell-to-cell. J. Virol. 71, Oparka, K. J., Roberts, A. G., Prior, D. A. M., Chapman, S., Baulcombe, D. C., and Santa Cruz, S. (1995). Imaging the green fluorescent protein in plants: Viruses carry the torch. Protoplasma 189, Osman, F., Choi, Y. G., Grantham, G. L., and Rao, A. L. N. (1998). Molecular studies on bromovirus capsid protein. V. Evidence for the specificity of brome mosaic virus encapsidation using RNA3 chimera of brome mosaic and cucumber mosaic viruses expressing heterologous coat proteins. Virology 251,

11 108 RYABOV ET AL. Palukaitis, P., Roossinck, M. J., Dietzgen, R. G., and Francki, R. I. B. (1992). Cucumber mosaic virus. Adv. Virus Res. 41, Power, J. B., and Chapman, J. V. (1985). Isolation, culture and genetic manipulation of plant protoplasts. In Plant Cell Culture (R. A. Dixon, Ed.), pp , ICR Press, Oxford. Petty, I. T. D., and Jackson, A. O. (1990). Mutational analysis of barley stripe mosaic virus RNA. Virology 179, Rao, A. L. N. (1997). Molecular studies on bromovirus capsid protein. III. Analysis of cell-to-cell movement competence of coat protein defective variants of cowpea chlorotic mottle virus. Virology 232, Reddy, D. V. R., Murant, A. F., Duncan, G. H., Ansa, O. A., Demski, J. W., and Kuhn, C. W. (1985). Viruses associated with chlorotic rosette and green rosette diseases of groundnut in Nigeria. Ann. Appl. Biol. 107, Rizzo, T. M., and Palukaitis, P. (1990). Construction of full-length cdna clones of cucumber mosaic virus RNAs 1, 2 and 3: Generation of infectious RNA transcripts. Mol. Gen. Genet. 222, Roberts, I. M. (1970). Reduction of compression artefacts in ultrathin sections by the application of heat. J. Microsc. 92, Roberts, I. M. (1975). Tungsten coating: A method of improving glass microtome knives for cutting ultrathin sections. J. Microsc. 103, Roberts, I. M. (1994). Factors affecting the immunogold labelling of plant virus antigens in thin sections. J. Virol. Methods 50, Russo, M., Burgyan, J., and Martelli, G. P. (1994). Molecular biology of tombusviridae. Adv. Virus Res. 44, Ryabov, E. V., Oparka, K. J., Santa Cruz, S., Robinson, D. J., and Taliansky, M. E. (1998). Intracellular location of two groundnut rosette umbravirus proteins delivered by PVX and TMV vectors. Virology 242, Ryabov, E. V., Robinson, D. J., and Taliansky, M. E. (1999). A plant virus-encoded protein facilitates long-distance movement of heterologous viral RNA. Proc. Natl. Acad. Sci. USA 96, Salema, R., and Brandao, I. (1973). The use of PIPES buffer in the fixation of plant cells for electron microscopy. J. Submicrosc. Cytol. 5, Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual, 2nd Ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. Schaad, M. C., and Carrington, J. C. (1996). Suppression of longdistance movement of tobacco etch virus in a non susceptible host. J. Virol. 70, Schmitz, I., and Rao, A. L. N. (1996). Molecular studies on bromovirus capsid protein. I. Characterization of cell-to-cell movement-defective RNA3 variants of brome mosaic virus. Virology 226, Schneider, W. L., Greene, A. E., and Allison, R. F. (1997). The carboxyterminal two-thirds of the cowpea chlorotic mottle bromovirus capsid protein is incapable of virion formation yet supports systemic movement. J. Virol. 71, Siegel, A., Zaitlin, M., and Seghal, O. P. (1962). The isolation of defective tobacco mosaic virus strains. Proc. Natl. Acad. Sci. USA 48, Suzuki, M. S., Kuwata, S., Kataoka, J., Masuta, C., Nitta, N., and Takanami, T. (1991). Functional analysis of deletion mutants of cucumber mosaic virus RNA3 using an in vitro transcription system. Virology 183, Taliansky, M. E., Malyshenko, S. I., Kaplan, I. B., Kondakova, O. A., and Atabekov, J. G. (1992). Production of the TMV 30 K protein in transgenic plants is essential but insufficient for complementing foreign virus transport: A need for full-length TMV genome or some other TMV-coded products. J. Gen. Virol. 73, Taliansky, M. E., Robinson, D. J., and Murant, A. F. (1996). Complete nucleotide sequence and organization of the RNA genome of groundnut rosette umbravirus. J. Gen. Virol. 77, Thompson, J. R., and García-Arenal, F. (1998). The bundle sheathphloem interface of Cucumis sativus is a boundary to systemic infection by tomato aspermy virus. Mol. Plant-Microbe Interact. 11, Wintermantel, W. M., Banerjee, N., Olivier, J. C., Paolillo, D. J., and Zaitlin, M. (1997). Cucumber mosaic virus is restricted from entering minor veins in transgenic tobacco exhibiting replicase-mediated resistance to cucumber mosaic virus. Virology 231, Wyatt, S. D., and Kuhn, C. W. (1979). Replication and properties of cowpea chlorotic mottle virus in resistant cowpeas. Phytopathology 69, Wyatt, S. D., and Kuhn, C. W. (1980). Derivation of a new strain of cowpea chlorotic mottle virus from resistant cowpeas. J. Gen. Virol. 49, Zhang, L., Hanada, K., and Palukaitis, P. (1994). Mapping local and systemic symptom determinants of cucumber mosaic cucumovirus in tobacco. J. Gen. Virol. 75,