Volume 2, Issue 2 p. 59-63
Open Access

Cucumber mosaic virus, a model for RNA virus evolution

Marilyn J. Roossinck

Corresponding Author

Marilyn J. Roossinck

Plant Biology Division, The Samuel Roberts Noble Foundation, Ardmore, OK 73402, USA

: E-mail:[email protected]Search for more papers by this author
First published: 07 July 2008
Citations: 84


Taxonomic relationships: Cucumber mosaic virus (CMV) is the type member of the Cucumovirus genus, in the family Bromoviridae. Additional members of the genus are Peanut stunt virus (PSV) and Tomato aspermy virus (TAV). The RNAs 3 of all members of the genus can be exchanged and still yield a viable virus, while the RNAs 1 and 2 can only be exchanged within a species.

Physical properties: The virus particles are about 29 nm in diameter, and are composed of 180 subunits (T = 3 icosahedral symmetry). The particles sediment with an s value of approximately 98. The virions contain 18% RNA, and are highly labile, relying on RNA–protein interactions for their integrity. The three genomic RNAs, designated RNA 1 (3.3 kb in length), RNA 2 (3.0 kb) and RNA 3 (2.2 kb) are packaged in individual particles; a subgenomic RNA, RNA 4 (1.0 kb), is packaged with the genomic RNA 3, making all the particles roughly equivalent in composition. In some strains an additional subgenomic RNA, RNA 4A is also encapsidated at low levels. The genomic RNAs are single stranded, plus sense RNAs with 5′ cap structures, and 3′ conserved regions that can be folded into tRNA-like structures.

Satellite RNAs: CMV can harbour molecular parasites known as satellite RNAs (satRNAs) that can dramatically alter the symptom phenotype induced by the virus. The CMV satRNAs do not encode any proteins but rely on the RNA for their biological activity.

Hosts: CMV infects over 1000 species of hosts, including members of 85 plant families, making it the broadest host range virus known. The virus is transmitted from host to host by aphid vectors, in a nonpersistent manner.

Useful web sites: http://mmtsb.scripps.edu/viper/1f15.html (structure); http://www.ncbi.nlm.nih.gov/ICTVdb/ICTVdB/10040001.htm (general information)


Cucumber mosaic, first described in 1916 (Doolittle, 1916), was one of the earliest plant diseases attributed to a virus (Jagger, 1916). Reports of the disease soon came from elsewhere in the USA, and later from Europe and Africa (Price, 1934) and other parts of the world. In the early days the tools for determining the presence of specific viruses were limited, and as many as 40 different plant diseases were later shown to be caused by Cucumber mosaic virus (CMV) (Kaper and Waterworth, 1981). A number of extensive reviews have been published on CMV which detail the biology of the virus (Edwardson and Christie, 1991; Kaper and Waterworth, 1981; Palukaitis et al., 1992; Roossinck, 1999). A large number of CMV strains have been described, and the sequence databases contain about 60 different coat protein sequences, as well as 15 complete viral genome sequences. The species includes three subgroups, IA, IB and II, with as much as 25% nucleotide sequence divergence between them (Roossinck et al., 1999). Thus, CMV has proved itself as a highly adaptable virus, with an unusual capacity for evolutionary change, making it both a menace to agriculture worldwide, and an ideal model for studying RNA virus evolution.

CMV encodes five proteins, distributed on three genomic RNAs (Fig. 1). RNA 1 is the only monocistronic RNA, encoding the 1a protein that is required for viral replication and contains methyl transferase and helicase motifs [for reviews see (Kadaré and Haenni, 1997; Rozanov et al., 1992)]. RNA 2 encodes the 2a protein, the viral polymerase (Ishihama and Barbier, 1994; O’Reilly and Kao, 1998), and the 2b protein, expressed from a low abundance subgenomic RNA, RNA 4A (Ding et al., 1994). The 2b open reading frame (ORF) is overprinted on the carboxy terminal portion of the 2a ORF. This ORF is not found in most other members of Bromoviridae genera, although an ORF in a similar position is found in Tobacco streak virus, a member of the Ilarvirus genus. The 2b protein of subgroup II CMV strains was shown to inhibit host post-transcriptional gene silencing (PTGS) (Béclin et al., 1998; Brigneti et al., 1998). RNA 3 encodes the movement protein (MP) expressed from the 5′ ORF, and the coat protein (CP) expressed from the subgenomic RNA 4. Both are required for virus movement (Canto et al., 1997).

Details are in the caption following the image

Genome organization of CMV. The nucleotide and amino acid numbers are for the Fny strain. RNAs 1, 2, and 3 are genomic, and required for infection. RNAs 4 and 4 A are subgenomic. RNA 4 is packaged in virions of all strains; RNA4A is only packaged in subgroup II strains. The satRNA is a molecular parasite sometimes associated with the virus. It is packaged in the virions of the virus.

The satellite RNAs (satRNAs) of CMV are small linear RNAs that do not carry any apparent coding capacity [reviewed in (García-Arenal and Palukaitis, 1999; Roossinck et al., 1992)]. SatRNAs can have a dramatic effect on the symptoms induced by CMV, ranging from attenuation (Yoshida et al., 1985) to severe exacerbation (Gonsalves et al., 1982; Kaper and Waterworth, 1977). Recently the CMV D satRNA was shown to induce programmed cell death in tomato, leading to systemic necrosis and the eventual death of the plant (Xu and Roossinck, 2000). The origin of satRNAs is still unknown. By definition they do not share any significant sequence identity with the helper virus. Database searches for similar sequences have not resulted in a clear relationship with any plant sequences (unpublished data). While satRNAs frequently occur in greenhouse studies of CMV when they are not intentionally included in the inoculum, they have not been generated during extensive passage of virus from infectious transcripts from cDNA clones (P. Palukaitis, personal communication; our unpublished results). Hence they may exist in subliminal levels in natural isolates, and if they are ‘spontaneously’ generated by mutation and selection of host or viral sequences during plant–virus interactions, it is not a common event.

CMV is transmitted by at least 75 species of aphids in a nonpersistent manner (Palukaitis et al., 1992). Although coat protein mutants have been identified that render the virus no longer transmissible by aphids (Perry et al., 1994), there is little vector specificity. CMV can also be transmitted by the parasitic plant dodder (Cuscuta spp.), in which it replicates (Francki et al., 1979), and through seeds. In experimental work, the virus is most frequently transmitted mechanically, and sap, purified virions, and viral RNA are all infectious via mechanical transmission.


RNA viruses can evolve rapidly due to the error-prone nature of their polymerases (Domingo and Holland, 1994) and a very short generation time. A single replicating population of an RNA virus can maintain a high level of variation, and these populations are known as quasispecies [for reviews see (Domingo et al., 1995; Eigen et al., 1988; Holland et al., 1992)]. However, not all viruses have the same rates of evolution, and even closely related viruses may differ dramatically in their mutation frequencies (Schneider and Roossinck, 2000).

Plant viruses are useful and convenient as models for understanding many general principles of virus evolution and disease. The ability to propagate plants that are essentially genetically identical makes plants an extremely attractive model for many studies that cannot be done in animal systems. Growth, maintenance and harvesting of plant hosts is relatively inexpensive and does not raise any ethical concerns, as compared to animal hosts. The infectivity of unencapsidated RNA enables the use of cDNA clones capable of producing infectious transcripts, making controlled genetic studies in an intact host organism very simple.

CMV is an extremely successful parasite. It has spread around the world, and evolved to infect over 1000 species of plant hosts in tropical, subtropical and temperate climates (Edwardson and Christie, 1991). Its evolutionary elasticity makes it an excellent model system for studying RNA virus evolution. In addition, the CMV satRNAs make useful reporters for experimental evolution studies. They are very easy to clone and sequence, and population data is easy to obtain.


A number of studies have analysed the variation in consensus sequences (i.e. the average sequence of a quasispecies population) of CMV isolates, in some cases from field isolates collected from a specific region (Crescenzi et al., 1993; Fraile et al., 1997; Rodríguez-Alvarado et al., 1995), or from strains from around the world (Owen and Palukaitis, 1988). These studies have demonstrated the highly variable nature of CMV.

Phylogeny estimations for the Cucumovirus genus revealed that the trees for each RNA were noncongruent, indicating that the evolutionary histories of the RNAs were independent of each other (White et al., 1995). This strongly supports a model of reassortment as a mechanism for strain generation and speciation. The characterization of a naturally occurring CMV-PSV interspecific reassortant further supports this model (White et al., 1995). Phylogeny estimations within CMV can reveal other interesting features of the evolutionary past. For example, an analysis of 53 CMV coat protein sequences revealed radial evolution within three subgroups (Roossinck et al., 1999). This pattern is not seen with some other open reading frames (unpublished data), and may reflect differing constraints on divergence. In addition, one of the subgroups (IB), was restricted to Asia, whereas the other two subgroups (IA and II) displayed a worldwide distribution, suggesting that CMV has had at least two widespread dispersals (Roossinck et al., 1999). The levels of nucleotide sequence similarity between the subgroups is shown in Table 1.

Table 1. Nucleotide sequence similarity (% identity) between CMV subgroups*.
IA 92 78
IB 78
IA 92 74
IB 73
IA 94 78
IB 73
  • * Sequence comparisons were made between Fny-CMV (subgroup IA), Nt9-CMV (subgroup IB), and LS-CMV (subgroup II).

Recombination is a mechanism both for the generation of variation and new strains, and for the rescue of deleterious mutations occurring in disparate parts of a viral RNA segment. It is a potentially important component of RNA virus evolution, and has been studied extensively in other Bromoviridae[(Bujarski and Nagy, 1996; Nagy and Bujarski, 1996, 1997; Nagy and Simon, 1997; Nagy et al., 1999; Bruyere et al., 2000), and references therein]. CMV is capable of generating defective RNAs (Graves and Roossinck, 1995), and therefore undergoing RNA-RNA recombination, but the mechanism has not been characterized. The 5′ nontranslated regions of CMV RNA 3 appear to have undergone recombination events to generate the three extant subgroups (Roossinck et al., 1999), but other studies on the importance of recombination to CMV evolution have not been carried out. In a field survey of CMV isolates in Spain, evidence of reassortment and recombination events was rare (Fraile et al., 1997).

Recently CMV was used as a model for experimental evolution studies. In a comparison of three closely related Sindbis-like plant viruses, CMV, Tobacco mosaic virus and Cowpea chlorotic mottle virus, CMV had the highest mutation frequency in progeny from cDNA clones of the viruses in a common host, Nicotiana benthamiana (Schneider and Roossinck, 2000). CMV also infects at least fivefold more host species than the other viruses in the study, correlating host range with population sequence variation.


The CMV satRNA has been used for a number of studies of within-population variation (Aranda et al., 1993; Grieco et al., 1997; Kurath and Palukaitis, 1989). However, these studies either compared consensus sequences between isolates in a given field or region (Aranda et al., 1993; Grieco et al., 1997), or examined a very limited number of members of a quasispecies (Kurath and Palukaitis, 1989). A thorough characterization of a single naturally occurring satRNA quasispecies has not been done.

The establishment of sequence variation after infection with transcript generated from a cDNA clone of the D4 satRNA showed that variation occurred reproducibly in the same region (Kurath and Palukaitis, 1990), but was dependent on the helper virus (Palukaitis and Roossinck, 1995). In addition, minor changes in the initial sequence could effect the generation of variation (Palukaitis and Roossinck, 1995), suggesting that RNA structure might play a role in the establishment of hypervariation. Other lines of evidence suggest that satRNA variation is constrained by structure (Fraile and García-Arenal, 1991), and a careful structural analysis of the D4 satRNA in planta (Rodríguez-Alvarado and Roossinck, 1997) shows that the hypervariable region lies in a nonbase-paired portion of the satRNA.

The CMV helper virus plays an important role in selection of specific satRNA variants from mixed populations (Moriones et al., 1991). In one study using two satRNA strains, WL and D, one helper virus (LS-CMV) preferentially amplified the WL satRNA, while another helper virus (TAV) preferentially supported D satRNA, even though both satRNAs replicated with high efficiency with either helper virus in single infections (Roossinck and Palukaitis, 1995). In another study a single nucleotide variation was spontaneously generated in a B5 satRNA transcript from a cDNA clone, and strongly selected-for by two different helper viruses (Fny- and LS-CMV). The new variant resulted in a change in phenotype, from attenuating to pathogenic, in conjunction with LS-CMV, but not with Fny-CMV. Hence selection in this case was for an unknown factor, and was unrelated to virulence, or the ability to cause disease (Palukaitis and Roossinck, 1996).


The above discussion was intended to convince the reader that CMV is an ideal virus for studying RNA virus evolution. There are many open questions in virus evolution that provide fertile areas for research. One important question in the evolution of plant viruses is the role of genetic bottlenecks during transmission. The natural transmission of CMV occurs by seed and by aphid vector. The highly variable nature of CMV could result in a dramatic loss in fitness if the bottlenecks are too narrow, a phenomenon known as Muller’s ratchet (Chao et al., 1992; Chao, 1990). Muller’s ratchet has been demonstrated in mammalian viruses through the use of artificial narrow bottlenecks (Duarte et al., 1992; Escarmís et al., 1996), but has not been demonstrated for a plant virus. Other aspects about the population biology of plant viruses are also unknown. For example, it is not known what the size of a generation is, i.e. when a viral RNA enters a cell, how many ‘offspring’ are produced in a single complete round of replication (plus-strand to minus-strand to progeny plus-strand). How many virions accumulate in a cell, and how many move to establish a systemic infection (another potential bottleneck in the life-cycle of a virus)? Finally, the underlying mechanisms that are responsible for sequence variation, such as polymerase fidelity, are not known for any plant virus. Clearly CMV can provide ample studies for anyone interested in the understanding of virus evolution.


The ability to evolve rapidly has allowed CMV to infect a very broad range of hosts. This adaptability has implications for the eventual control of viruses, and it is likely that without an understanding of evolutionary mechanisms, long-term methods for virus resistance will be difficult to achieve. In addition, highly adaptable viruses are likely to emerge into new host species. There is evidence that the most severe virus diseases result when a virus has recently ‘jumped species’ (reviewed in Villarreal et al., 2000). The ability to prevent or predict viral emergence would be a great benefit to agriculture and human health. Finally, RNA virus evolution mirrors the evolution of other life, but occurs rapidly enough to observe in as little as 10 days. Hence many mechanisms of molecular evolution will likely be elucidated by studying RNA viruses such as CMV.