Host-specific encapsidation of a defective RNA 3 of Cucumber mosaic virus

Viruses in the family Bromoviridae have a divided, single-stranded RNA genome, comprised of three, separately encapsidated, positive-sense RNAs (Roossinck et al., 2000). In addition, the capsid protein subgenomic RNA (designated RNA 4) is also encapsidated, often together with RNA 3. In general, RNAs 1 and 2 encode proteins involved in replication, while RNA 3 encodes proteins functioning in movement and encapsidation (Roossinck et al., 2000). However, in some cases, the proteins encoded by RNAs 1 and 2 also affect virus movement (Canto & Palukaitis, 2001; Carr et al., 1994; Ding et al., 1995; Gal-On et al., 1994; Traynor et al., 1991). In the case of some members of the genus Cucumovirus, particles may contain genomic RNAs, subgenomic RNAs, satellite RNAs and also defective (D) RNAs (reviewed by Palukaitis & García-Arenal, 2003). Satellite RNAs contain little or no sequence similarity to the helper virus genome, while D RNAs are derived from the helper virus genome.

D RNAs and defective interfering (DI) RNAs have been described for a number of plant viruses (see reviews by Graves et al., 1996; Simon & Nagy, 1996; White, 1996; White & Morris, 1999). DI and D RNAs usually contain in-frame deletions within one or more genes and in some cases may be associated with changes in pathology (Graves et al., 1996). In the case of the cucumovirus Cucumber mosaic virus (CMV), a D RNA derived from RNA 3 of CMV has been described (Graves & Roossinck, 1995). This D RNA, designated D RNA 3α, was found associated with a local lesion isolate of the Fny strain of CMV (Fny-CMV) and contained a 156 nt in-frame deletion within the 3a gene encoding the movement protein. In addition, a second D RNA 3 molecule (D RNA 3β), which contained an in-frame deletion of 309 nt within the Fny-CMV 3a gene, was also described, although it was not characterized further (Graves & Roossinck, 1995). By definition, such D RNAs could be maintained by the associated wild-type (WT) virus, as well as some other strains of CMV. However, it is not clear whether these D RNA 3s were originally present at subliminal levels in the population and subsequently were amplified to high levels, perhaps following further mutation, or whether they truly arose de novo.

During the passage of CMV derived from cDNA clones, two D RNA 3s were generated. These D RNA 3s were characterized and compared with those described previously, with regard to both the nature of the deletion and their biological properties. Host-specific maintenance of one D RNA 3 was observed and shown to be due to host-specific encapsidation.

Generation of virus inocula and passage in plants.
RNA transcripts were generated from cDNA clones of Fny-CMV RNAs 1 (pFny109) and 2 (pFny209) as well as various cloned sources of RNA 3 (see below), all as described previously by Zhang et al. (1994). RNA transcripts were inoculated into either tobacco (Nicotiana tabacum cv. Samsun NN) leaves or zucchini squash (Cucurbita pepo cv. Black Beauty) cotyledons.

Sap from a tobacco plant inoculated with transcripts representing Fny-CMV RNAs 1, 2 and 3 (Rizzo & Palukaitis, 1990) was inoculated into two tobacco plants and passaged at 2-weekly intervals for a total of 20 passages. Leaves from passages 220 were dried and stored at 4 °C. Dried leaves from passages 2, 3 and 20 were homogenized in 50 mM sodium phosphate, pH 7, and the slurry was used to inoculate two to four tobacco plants. Virus was purified from these infected plants 2 weeks post-inoculation and viral RNAs were extracted and stored at 20 °C (Palukaitis et al., 1992).

Various sources of CMV RNA 3 were used in different inoculations. In some experiments, RNA transcripts were derived from full-length cDNA clones of Fny-CMV RNA 3 (pFny309) or strain M CMV (M-CMV) RNA 3 (pMCMV3; Shintaku et al., 1992). In other experiments, RNA 3 transcripts generated from a cDNA clone of D RNA 3-1 (see below) were used.

Construction of D RNA 3-1.
All standard manipulations were done as described by Sambrook et al. (1989). cDNA was synthesized from gel-purified Fny-CMV D RNA 3, using avian myeloblastosis virus reverse transcriptase and a primer (5'-TGGTCTCCTTTTGGAG-3') complementary to the 16 nt at the 3'-end of Fny-CMV RNA 3. PCR was done using a primer corresponding to the 10 nt at the 5'-end of Fny-CMV RNA 3, preceded by a T7 promoter sequence and a 5'-terminal BamHI site (Owen et al., 1990) and a primer complementary to nt 12821299 (containing a SalI site) of Fny-CMV RNA 3. The PCR product was digested with BamHI and SalI and incubated in a three-piece ligation reaction. The other components of the ligation included the larger fragment of pUC18 cleaved with BamHI and PstI and the smaller fragment of pFny309 digested with SalI and PstI (containing the capsid protein gene and 3' non-coding region). Escherichia coli DH5α cells were transformed by the ligation reaction products. One such transformant contained a plasmid with an insert designated D RNA 3-1. The 3a gene of D RNA 3-1 was sequenced. Other cDNA clones from the same transformation event showed the same sequence. In addition, the non-coding regions as well as the genes encoding the capsid protein and 3a protein were sequenced directly from the gel-purified D RNA 3, to confirm that D RNA 3-1 contained a deletion typical of the D RNA 3 population and that only the 3a gene contained a deletion in the Fny-CMV D RNA 3.

Characterization of progeny viral RNAs.
Viral RNAs were extracted from virions and total plant RNAs were extracted from plants, both as described by Kaplan et al. (1995). The various RNAs were analysed by agarose gel electrophoresis and Northern blot hybridization as described previously (Kaplan et al., 1995). Blots were probed with ³²P-labelled transcripts complementary to the 3'-terminal 200 nt of Fny-CMV RNA 3 (Gal-On et al., 1994). WT RNA 3 and D RNA 3-1 were subjected to RT-PCR, using a primer complementary to the intergenic region (nt 11351148) for reverse transcription and the same primer plus a primer corresponding to the sequence to the first 10 nt at the 5' end of Fny-CMV RNA 3 (see above) for the PCR step. PCR products were analysed by agarose gel electrophoresis.

The D RNA designated [M] D RNA 3-2, detected after inoculation of tobacco plants with transcripts of Fny-CMV RNAs 1 and 2 plus M-CMV RNA 3 (Canto & Palukaitis, 1998), was amplified from total viral RNA containing this D RNA by RT-PCR. The primers used corresponded to the first 13 nt at the 5' end of Fny-CMV RNA 3 and sequences complementary to the 15 nt at the 3' end of Fny-CMV RNA 3, respectively. The nature of the deletion in [M] D RNA 3-2 was established by sequence analysis of the PCR product.

Preparation, infection and analysis of virus accumulation and encapsidation in zucchini squash protoplasts.
Protoplasts were prepared from leaves of zucchini squash plants and infected by electroporation using viral RNAs, all as described previously (Lee et al., 2001). Protoplasts were incubated for 24 or 48 h and then processed. This experiment was done four times. Total RNAs were extracted and purified from the infected protoplasts as described by Gal-On et al. (1994). Virions were isolated from infected protoplasts and RNAs were extracted from the isolated virions, both as described by Osman et al. (1998). Total plant RNAs and viral RNAs were analysed by Northern blot hybridization or by RT-PCR, both as indicated above, except that the CMV probe was labelled with digoxigenin and the blots were processed following the manufacturer's instructions (Roche Diagnostics).

Generation and sequence analysis of D RNA 3s
Fny-CMV derived from biologically active cDNA clones was passaged and propagated in tobacco. Virions purified from plants infected with an extract derived from one such passage were found to contain an additional RNA with an electrophoretic mobility between that of CMV RNAs 3 and 4 (Fig. 1, lane 2 vs lane 1). Upon further passage, this additional RNA persisted (Fig. 1, lane 3), but did not increase in accumulation at the level of encapsidation relative to the other CMV RNAs. Northern blot analyses indicated that this additional RNA was related in sequence to CMV RNA 3, but not to CMV RNA 1 or 2 (data not shown).

(64K): Fig. 1. Detection of D RNA in CMV RNA preparations. Virion RNAs extracted from infected tobacco plants were analysed by electrophoresis on a 1·6 % agarose gel and ethidium bromide staining. RNAs were from virus preparations at 2 weeks post-inoculation containing WT Fny-CMV (lane 1) and Fny-CMV containing D RNA 3-1 (lanes 2 and 3). Lane 3 contained viral RNA from a plant infected with an extract from the plant in lane 2. The positions of CMV RNAs 1 and 2, 3, 3-1 and 4 are shown on the left.

The cloning and sequencing of this additional RNA established that it was a D RNA 3 containing an in-frame deletion of 339 nt within the 3a gene, from nt 296 or 297 to nt 634 or 635. This D RNA 3 was designated D RNA 3-1. The same deletion was present in the original D RNA 3-1 as well as in the cDNA clones that were sequenced (data not presented), indicating that the cDNA clones were representative of the population of D RNA 3-1 molecules.

The lineage of the generation of D RNA 3-1 was analysed. The first virus preparation containing the progenitor of D RNA 3-1 was derived after passage in tobacco of a virus preparation that did not contain visible D RNA 3. This parental virus preparation was derived from passage 3a (Fig. 2). This was a separate lineage of a 20-passage experiment starting from passage 2, i.e. the virus from passage 3a was obtained from plants inoculated with an extract of dried leaves containing passage 2 of Fny-CMV derived from an inoculation involving RNA transcripts of the biologically active Fny-CMV cDNA clones. When dried leaves from the original passages 3 and 20 were used as inocula (Fig. 2), no D RNA 3 was detectable in virions of the successive passages (data not presented). Viral RNAs obtained from these virions (extracted from passages 4a and 21) were each inoculated into five tobacco plants at a concentration of 1 mg ml¹ to determine whether the appearance of D RNA 3 was stimulated by a high concentration of inoculum, but no D RNAs were visible in the progeny virions (data not shown). Thus, the D RNA 3 was most probably generated in subsequent passages of virus derived from the second-passage dried leaves (i.e. after passage 3a), rather than during the initial 20 passages in tobacco.

(17K): Fig. 2. Flow chart of passages of Fny-CMV in tobacco. RNA transcripts derived from biologically active cDNA clones of Fny-CMV RNAs 1, 2 and 3 were inoculated into two tobacco plants. The resulting virus then was passaged serially for 20 passages (descending arrows) through two tobacco plants at 2-weekly intervals, using sap from the previously infected plants as inoculum. Leaves from these infected plants then were dried and dried leaves from passages 2, 3 and 20 were used to inoculate a second set of tobacco plants (rightward arrows) generating passages 3a, 4a and 21. Virus was purified from these secondary passage plants and viral RNAs were used in further inoculations as described in the text.

A second D RNA 3 was derived from biologically active CMV cDNA clones. During passage of a pseudorecombinant virus constructed from biologically active transcripts of Fny-CMV RNAs 1 and 2 plus M-CMV RNA 3, a D RNA 3 of M-CMV RNA 3 was generated (see Fig. 9a in Canto & Palukaitis, 1998). As in the above experiments, this D RNA 3, designated [M] D RNA 3-2, was maintained upon passage at levels similar to the WT M-CMV RNA 3 (Canto & Palukaitis, 1998). Sequence analysis of a cDNA clone of [M] D RNA 3-2 showed that it contained an in-frame deletion of 411 nt within the 3a gene of M-CMV RNA 3, from nt 303 to 713.

These experiments demonstrated that D RNAs derived from CMV RNA 3 could appear spontaneously in CMV RNAs derived from cDNA clones, although the generation of D RNAs was rare. In addition, although the virions formed by the capsid proteins of Fny-CMV and M-CMV are known to differ in their stability (Ng et al., 2000), this did not affect the ability of these capsid proteins to encapsidate the respective D RNA 3s.

Effects of host on maintenance and encapsidation of D RNA 3-1
To assess potential host effects on the accumulation of the D RNA 3s, transcripts of the cDNA clone of D RNA 3-1 were mixed with Fny-CMV RNAs 1, 2 and 3, also derived from transcripts, and were passaged serially in tobacco plants. D RNA 3-1 was supported by the Fny-CMV RNAs in tobacco (Fig. 3a) and had no effect on the yield of virus or the symptoms induced by Fny-CMV (data not presented). By contrast, when sap or purified virus obtained from tobacco infected by Fny-CMV containing D RNA 3-1 was passaged through squash plants, D RNA 3-1 was not readily detectable (Fig. 3a and b). Analysis of the gel containing virion RNAs shown in Fig. 3(a) could not unambiguously establish whether D RNA 3-1 was absent from such virions or was simply maintained at a reduced level, due to the presence of a background of degraded genomic RNAs. Therefore, small cotyledons of squash plants were inoculated with Fny-CMV RNA, with and without D RNA 3-1, and virion RNAs were analysed at 2 weeks post-inoculation (Fig. 3b). In these preparations, which were made before the cotyledons began to deteriorate due to abscission, breakdown of the larger RNAs did not obscure the presence of stainable levels of D RNA 3-1. Nevertheless, no D RNA 3-1 could be detected in virions extracted from either the cotyledons or true leaves.

(52K): Fig. 3. Detection of D RNA in CMV RNA preparations. Virion RNAs extracted from infected tobacco or squash plants were analysed by electrophoresis in 1·6 % agarose gels and ethidium bromide staining. (a) Sap from tobacco infected with Fny-CMV RNA containing D RNA 3-1 was inoculated into tobacco or squash and passaged at 2-week intervals. Passages are designated 14. (b) Squash plants infected with Fny-CMV RNAs plus or minus D RNA 3-1 were used for virus preparations 2 weeks post-inoculation. The inoculated cotyledons (C) and systemically infected leaves (L) were extracted separately. The positions of CMV RNAs 1 and 2, 3, 3-1 and 4 are indicated on the left of the gels.

This analysis could not distinguish between replicated D RNA 3-1 either being turned over rapidly or not being stabilized by encapsidation. Thus, virion RNAs as well as total RNAs extracted from inoculated squash cotyledons and infected upper leaves were analysed by RT-PCR (Fig. 4), using primers flanking the 3a gene. In the case of the encapsidated RNAs, the DNA fragment corresponding to the partial 3a gene in D RNA 3-1 was not observed in squash cotyledons or upper leaves (Fig. 4a). By contrast, in total RNAs extracted from infected squash cotyledons and upper leaves, the fragment corresponding to the partial 3a gene of D RNA 3-1 was present only in the cotyledons and not in the upper leaves (Fig. 4b).

(41K): Fig. 4. Analysis of replication and encapsidation of CMV D RNA 3-1 in squash and tobacco plants. Squash and tobacco plants were inoculated with Fny-CMV RNA with and without D RNA 3-1. RNAs were extracted either from virions purified from the infected plants at 2 weeks post-inoculation (a) or directly from systemically infected leaves (L) or inoculated cotyledons (C) at 1 week post-inoculation (b). Tobacco samples were all from systemically infected leaves. The RT-PCR products generated using RNA 3-specific primers were analysed by electrophoresis in 1·6 % agarose gels and stained with ethidium bromide. Size markers (M) were loaded on each gel. From top to bottom, these were 1500, 1200, 1000, 900, 800, 700, 600 and 500 bp. The arrows indicate the positions of the RT-PCR products generated from WT RNA 3 and D RNA 3-1.

The much lower level of D RNA 3-1 present in infected squash cotyledons compared with infected tobacco leaves (Fig. 4b) indicated that squash was able to support accumulation of D RNA 3-1, but was not able to support the encapsidation and systemic movement of D RNA 3-1. Whether this difference in accumulation in the squash cotyledons was due to a lower level of replication or a similar level of replication to that occurring in tobacco but with poor accumulation because of a lack of stabilization by encapsidation could not be determined by this assay. To distinguish between these alternatives, protoplasts prepared from squash mesophyll cells were infected with Fny-CMV RNAs with and without D RNA 3-1, and virion RNAs as well as total RNAs were examined. Extracts of total RNAs revealed that a band corresponding to that of the D RNA 3-1 could be discerned in the total RNAs from various protoplast preparations infected with CMV RNAs plus D RNA 3-1 (Fig. 5a), although additional hybridizing bands due to rRNA shadowing also were present. However, D RNA 3-1 was not detectable in RNAs extracted from purified virions (Fig. 5b). Thus, while the D RNA 3-1 was replicated in protoplasts derived from squash mesophyll cells, the D RNA 3-1 was not stabilized by encapsidation.

(88K): Fig. 5. Detection of D RNA 3-1 in inoculated squash protoplasts. Squash protoplasts were inoculated with various CMV RNAs and incubated for 48 h. (a) Northern blot of total RNAs extracted from squash protoplasts infected with Fny-CMV RNAs with (lanes 36) or without (lane 2) D RNA 3-1. Lane 1 shows RNAs extracted from mock-inoculated protoplasts and lanes 7 and 8 show marker RNAs: CMV RNAs 14 and D RNA 3-1 transcript, respectively. The samples in lanes 36 were from four separate batches of infected protoplasts. (B) Northern blot of virion RNAs extracted from squash protoplasts infected with (lane 2) or without (lane 1) D RNA 3-1, compared with virion RNAs of CMV without D RNA 3-1 extracted from infected plants (lane 3). Lane 4 contains D RNA 3-1 transcript.

Whether [M] D RNA 3-2 could be maintained in squash was not examined, since in squash, the capsid protein of M-CMV (present in [M] D RNA 3-2) is unable to promote systemic infection and only promoted slow cell-to-cell movement (Wong et al., 1999). Two D RNA 3s found associated with CMV derived from cDNA clones were characterized. One of these was found associated with the Fny strain of CMV and one with a pseudorecombinant virus containing RNAs 1 and 2 of Fny-CMV and RNA 3 derived from the M strain of CMV, which shows 99 % sequence similarity to Fny-CMV RNA 3 (Owen et al., 1990). Thus, the same replicase gene sequences are present in both viruses, although the respective capsid proteins contained sequence differences that affected the stability of the virions formed from these capsid proteins (Ng et al., 2000). In both of these cases, the D RNA 3s were generated spontaneously in virus derived from passaged sap. However, the generation of natural D RNA 3s is a rare process, since we have not seen such D RNA 3s in the virion RNAs of over 30 isolates of CMV examined (unpublished data). Once established, D RNA 3-1 was detectable in all subsequent virus preparations generated from a single virus preparation that itself did not contain detectable levels of D RNA 3. However, D RNA 3 was not detected in separate virus preparations derived from other passages of the 20-passage source material (Fig. 2).

The D RNA 3s described here did not affect the yield of virus or the symptoms induced by the corresponding viruses (Fny-CMV induces a light green/dark green mosaic on tobacco, while the capsid protein of M-CMV induces a yellow, systemic chlorosis). This was also reported for the two D RNA 3s described previously, D RNA 3α and D RNA 3β, both found associated with WT Fny-CMV (Graves & Roossinck, 1995). Those D RNA 3s, differing in the nature of the deletion within the 3a gene from the D RNA 3s reported here (Fig. 6), were derived from a local lesion isolate and not from cDNA clones (Graves & Roossinck, 1995). Nevertheless all four natural D RNA 3s of CMV have been derived and maintained by the Fny-CMV replicase. Which factors lead to the spontaneous generation of D RNA 3s with Fny-CMV remains unknown.

(23K): Fig. 6. Comparison of deletions in various D RNA 3s. D RNA 3-1 and [M] D RNA 3-2 are described in the text. D RNAs 3α and 3β have been characterized by Graves & Roossinck (1995) and the various artificial deletions have been described previously (Kaplan & Palukaitis, 1998). Those D RNA 3s with a prefix [M] arose from M-CMV RNA 3, while the remainder arose from Fny-CMV RNA 3 (designated RNA 3 wt). The various D RNA 3s were generated either naturally (D RNA 3-1, [M] D RNA 3-2, D RNA 3α and D RNA 3β) or by passage of artificial D RNAs (ΔNheI, [M] ΔNheI-1, [M] ΔNheI-2, ΔE-H and ΔKpnI) in 3a-transgenic plants. The locations and boundaries of the in-frame deletions are indicated as well as the size of the deletions. The lines connecting rectangles represent the deletions. The numbers flanking the deletions represent the first and last nucleotides of the deleted regions, using the numbering system of Fny-CMV RNA 3. The open rectangle is the 3a ORF (nt 120956), while the 5' non-translated region (5' NTR) and intergenic region (IGR) are indicated by filled rectangles.

In addition to the natural D RNA 3s, several artificial D RNA 3s were generated from both Fny- and M-CMV RNA 3 containing deletions in the 3a gene (Kaplan & Palukaitis, 1998). In contrast to the natural D RNA 3s, the artificially generated D RNAs 3 of CMV, all containing in-frame deletions within the 3a gene (Fig. 6), were maintained by CMV RNAs 1 and 2 during complementation in the transgenic tobacco expressing the 3a gene, but could not be maintained in the presence of WT CMV RNA 3 (Kaplan & Palukaitis, 1998). Thus, the nature of the deletion must affect the accumulation properties of the corresponding D RNA 3. The reasons for these differences in properties are unknown, but again do not appear to relate to differences in the stability of the corresponding virions, since they were observed with virions formed from both Fny-CMV and the less stable M-CMV.

The differential accumulation of D RNA 3-1 in tobacco vs squash (Figs 1, 3 and 4) was due to an inability of the D RNA 3-1 to become encapsidated in squash cotyledons and to accumulate in upper leaves of squash plants (Figs 4b and 5). The failure to accumulate in upper leaves is probably due to an inability of D RNA 3-1 to move systemically in the absence of encapsidation rather than an inhibition of cell-to-cell movement, since the latter process does not require either encapsidation (Kaplan et al., 1998) or interaction between CMV RNAs and the capsid protein (Kim et al., 2004). It seems likely that virion formation is essential for systemic movement in squash, since it was shown that CMV capsid protein mutants that did not form virions still supported systemic movement in some Nicotiana species but not in squash (Kaplan et al., 1998). Moreover, the capsid protein of M-CMV did not promote systemic infection in squash (Wong et al., 1999). This might suggest that either there is a fundamental difference in how CMV moves in squash compared with tobacco or that movement in squash is more restrictive than in tobacco. The data presented here do not distinguish between these alternatives.

Why is there host-specific encapsidation of D RNA 3-1? Such host-specific encapsidation was also observed for a DI RNA 2 of the bromovirus Broad bean mottle virus (Romero et al., 1993). In that case, it was suggested that the lack of encapsidated DI RNA 2 may have been due to a lower stability of such virions in pea compared with broad bean, although why only virions containing the DI RNA would be less stable in one host than another was not explained (Romero et al., 1993). Moreover, CMV strains with considerable differences in virion stability (Ng et al., 2000) did not show strain-specific effects on encapsidation of D RNAs (Figs 1 and 3; Canto & Palukaitis, 1998). The assembly of most plant viruses is not well characterized, although generally assembly of RNA viruses with only a single type of capsid protein has been considered to be a process that does not require scaffolding proteins, but only interactions between the viral RNAs and the capsid proteins. Thus, in the absence of known host-specific effects on assembly or stability of virions, it may be that differences in the interactions of genomic, subgenomic and D/DI RNAs with host components in various plants may determine whether particular RNAs are available for encapsidation or can be mobilized to the site(s) of assembly. This in turn may then determine whether the RNAs are encapsidated or not. Studying the host-specific encapsidation of D RNAs may provide some enlightenment of this selection process and ultimately contribute to a better understanding of the CMV encapsidation process as such.

This work was supported in part by Grant DE-FG02-86ER13505 from the US Department of Energy, Grant 95-33120-1876 from the USDA NRICGP, support from a grant-in-aid from the Scottish Executive Rural Affairs Division to the SCRI (for P. P.) and funds from the National University of Singapore (to S.-M. W.).