Large-Scale Synonymous Substitutions in Cucumber Mosaic Virus RNA 3 Facilitate Amino Acid Mutations in the Coat Protein

Recently, it has been known that synonymous substitutions in RNA virus genes affect viral pathogenicity and competitive fitness by alteration of global or local RNA secondary structure of the viral genome. We confirmed that large-scale synonymous substitutions in the CP gene of CMV resulted in decreased viral RNA titer. Importantly, when viral evolution was stimulated by serial-passage inoculation, viral RNA titer was rapidly restored, concurrent with a few amino acid changes in the CP. This novel finding indicates that the deleterious effects of large-scale nucleic acid mutations on viral RNA secondary structure are readily tolerated by structural changes in the CP, demonstrating a novel part of the adaptive evolution of an RNA viral genome. In addition, our experimental system for serial inoculation of large-scale synonymous mutants could uncover a role for new amino acid residues in the viral protein that have not been observed in the wild-type virus strains.

ABSTRACT

The effect of large-scale synonymous substitutions in a small icosahedral, single-stranded RNA viral genome on virulence, viral titer, and protein evolution were analyzed. The coat protein (CP) gene of the Fny stain of cucumber mosaic virus (CMV) was modified. We created four CP mutants in which all the codons of nine amino acids in the 5′ or 3′ half of the CP gene were replaced by either the most frequently or the least frequently used synonymous codons in monocot plants. When the dicot host (Nicotiana benthamiana) was inoculated with these four CP mutants, viral RNA titers in uninoculated symptomatic leaves decreased, while all mutants eventually showed mosaic symptoms similar to those for the wild type. The codon adaptation index of these four CP mutants against dicot genes was similar to those of the wild-type CP gene, indicating that the reduction of viral RNA titer was due to deleterious changes of the secondary structure of RNAs 3 and 4. When two 5′ mutants were serially passaged in N. benthamiana, viral RNA titers were rapidly restored but competitive fitness remained decreased. Although no nucleic acid changes were observed in the passaged wild-type CMV, one to three amino acid changes were observed in the synonymously mutated CP of each passaged virus, which were involved in recovery of viral RNA titer of 5′ mutants. Thus, we demonstrated that deleterious effects of the large-scale synonymous substitutions in the RNA viral genome facilitated the rapid amino acid mutation(s) in the CP to restore the viral RNA titer.

IMPORTANCE Recently, it has been known that synonymous substitutions in RNA virus genes affect viral pathogenicity and competitive fitness by alteration of global or local RNA secondary structure of the viral genome. We confirmed that large-scale synonymous substitutions in the CP gene of CMV resulted in decreased viral RNA titer. Importantly, when viral evolution was stimulated by serial-passage inoculation, viral RNA titer was rapidly restored, concurrent with a few amino acid changes in the CP. This novel finding indicates that the deleterious effects of large-scale nucleic acid mutations on viral RNA secondary structure are readily tolerated by structural changes in the CP, demonstrating a novel part of the adaptive evolution of an RNA viral genome. In addition, our experimental system for serial inoculation of large-scale synonymous mutants could uncover a role for new amino acid residues in the viral protein that have not been observed in the wild-type virus strains.

INTRODUCTION

Understanding how RNA virus genes evolve is an important issue because they can easily adapt to new environments or new hosts. Important properties of RNA viruses are the error-prone replication and short generation time of their genomes and the genome recombination between related viruses, leading to rapid evolution of viral pathogenicity. Cucumber mosaic virus (CMV) (genus Cucumovirus; family Bromoviridae), a single-stranded positive-sense RNA virus, has been used extensively as a model system for study of evolution of plant RNA viruses (1). CMV infects over 1,200 species in over 100 families of both monocot and dicot plants and causes variable disease symptoms. CMV encodes at least five proteins: the 1a and 2a viral RNA replication proteins, the 2b viral suppressor for RNA silencing, the 3a movement protein, and the coat protein (CP) (reviewed in reference 2).

In the evolution of viral pathogenicity, nonsynonymous substitutions with amino acid mutations in viral genes are the most important, because amino acid mutation may dramatically change the structure and function of viral proteins and directly affects interactions with viral factors or host factors. Nonsynonymous substitution has a major role in the evolution of virulence in certain hosts and environments, so a lot of research has focused on the effect of nonsynonymous substitutions in CMV genes on viral pathogenicity (3). On the other hand, synonymous substitutions without amino acid mutation (also called silent substitutions) do not usually result in changes pathogenicity. However, in the case of RNA viruses, there are two possible effects of synonymous mutations in viral genes: alteration of the local and global RNA secondary structure of the viral genome (4–8) and alteration of the codon usage bias of viral genes (9–11).

In animal viruses, codon usage deoptimization of viral virulence genes by large-scale synonymous substitutions resulted in an attenuated phenotype that has been considered for use in vaccine development (reviewed in reference 12). Large-scale synonymous substitution of viral gene also resulted in a decrease of viral titer, competitive fitness, and robustness against genomic mutations (9, 13). Thus, synonymous substitution may play an important role in virulence, adaptation, and evolution of RNA viruses.

Adams and Antoniw (14) reported that the codon usage bias of most plant RNA viruses is not adapted to that of the host plant, although some plant viruses have adapted codon usage to the bias of their hosts (15–17). Cardinale et al. (18) reported that the codon usage of plant viruses was influenced by the pressure from genome architecture or secondary structure but was not influenced by the host codon usage bias, at least for luteoviruses and potyviruses (single-stranded RNA [ssRNA]) and geminiviruses (single-stranded DNA [ssDNA]). Tubiana et al. (6) indicated that the global secondary RNA structure of brome mosaic virus RNA 2 becomes compact for efficient packaging in small icosahedral viral particles. Thus, the local and global RNA secondary structure of plant ssRNA viral genomes is considered to be a stronger negative selection pressure than codon usage bias for synonymous substitutions. However, the effects of large-scale synonymous substitution in plant RNA virus genomes on virulence, viral titer, and evolution are not known.

As a model experimental system to analyze the adaptive evolution of plant ssRNA viruses, we used CMV mutants in which large-scale synonymous substitutions without codon usage bias change were introduced into the CP gene, to understand how these mutated RNAs evolved during host adaptation. We show that the large-scale synonymous substitutions caused decreased viral RNA titer, but viral RNA titer of CMV mutants rapidly recovered by one to two amino acid changes in the CP during serial-passage inoculation.

RESULTS

Virulence and viral titer of large-scale synonymously mutated CMV.

To investigate the effect of large-scale synonymous substitutions in the CP gene of the Fny stain of CMV on virulence and viral titer, we created a CP gene mutant in which all the codons of 18 amino acids were replaced by the least frequently used synonymous codons in monocots (19). However, this mutant did not infect the dicot host Nicotiana benthamiana (data not shown). Next we created two CP gene mutants in which all the codons of nine amino acids (A, D, E, L, P, Q, R, S, and Y) were replaced by either the most frequently (Mmaj) or the least frequently (Mmin) used synonymous codons in monocots. These mutants did not show any symptoms and were barely detectable in uninoculated upper leaves of N. benthamiana (see below). Next, we created four further chimeric mutants (Mmaj/wt, Mmin/wt, wt/Mmaj, and wt/Mmin) in which the 5′ or 3′ half of the CP gene was modified (Fig. 1A). Importantly, the codon adaptation index, a measurement of the relative adaptiveness of the codon usage of a gene toward those of the target organism genes (20), of four chimeric mutants against dicots (arabidopsis) was similar to those of wild-type Fny (Table 1).

FIG 1.

Effect of large-scale synonymous substitutions in the coat protein (CP) gene on virulence and viral titer of CMV. (A) Diagrams of the four chimeric large-scale synonymous mutants of CMV Fny RNA 3. (B) Symptoms of Nicotiana benthamiana inoculated with Fny or the four chimeric mutants. The photographs were taken at 10 days postinoculation. (C) Viral RNA titer in uninoculated upper leaves of N. benthamiana inoculated with Fny or four chimeric mutants. Three N. benthamiana individuals were used for each mutant. The mean values of normalized intensity of the CMV RNA 4 band (RNA 4/28S rRNA) obtained in experiments with triplicate samples are indicated under the CMV RNA panel. The data are representative of those from three independent experiments.

TABLE 1.

Characters of large-scale synonymously substituted CP genes used in this study

Virus	Codon adaptation index		Effective no. of codonsa	GC content (%)	No. of mutated nucleotides	dN/dS ratiob
	Arabidopsis	Rice
Fny	0.714	0.73	61.0	50.1
Mmaj	0.615	0.876	29.8	59.7	123	0
Mmin	0.737	0.607	29.8	41.9	149	0
wt/Mmaj	0.675	0.781	46.2	54.3	52	0
Mmaj/wt	0.650	0.818	44.0	55.4	71	0
wt/Mmin	0.748	0.653	45.1	44.0	79	0
Mmin/wt	0.703	0.678	48.1	47.9	70	0

^aThe effective number of codons is a general measure of codon usage bias from equal codon usage in a gene.

^bRatio of nonsynonymous to synonymous substitutions.

When N. benthamiana was inoculated with each chimeric mutant, Fny and the 5′ mutants (Mmaj/wt and Mmin/wt) yielded systemic symptom at 4 days postinoculation (dpi), while symptom expression with the 3′ mutants (wt/Mmaj and wt/Mmin) was delayed to 7 to 10 dpi. All four chimeric mutants eventually showed mosaic symptoms similar to those for wild-type Fny (Fig. 1B). However, in the symptomatic upper mosaic leaves at 10 dpi, the titer of viral RNAs, especially RNAs 3 and 4, was decreased in both the 5′ and 3′ mutants (Fig. 1C). The amounts of viral RNA 4 for the 5′ and 3′ mutants were one-third and one-eighth that of Fny, respectively. The titer of viral RNAs 1 and 2 also was decreased in wt/Mmaj, wt/Mmin, and Mmaj/wt. These results showed that the viral RNA titer was reduced by the large-scale synonymous substitutions in the CP gene without codon usage bias change. It is also noteworthy that the viral RNA titers of wt/Mmaj and wt/Mmin were different among the replicated plants, showing a larger variation of fitness of large-scale synonymous mutants.

Viral RNA titer of 5′ mutants recovered by serial-passage inoculation.

To analyze how the viral genomes of large-scale synonymous mutants adapt in the context of strong selective pressure, 5′ mutants (Mmaj/wt and Mmin/wt) were serial-passage inoculated into N. benthamiana. Three independent passage series for each mutant and Fny were done (Fig. 2A). Viral RNA titer was examined at the 5th, 10th, and 15th passages, and the consensus CP sequences were examined at the 1st, 5th, 10th, and 15th passages by direct sequencing of the reverse transcription-PCR (RT-PCR) product of the CP gene.

FIG 2.

Rapid evolution of the large-scale synonymously substituted CMVs by serial-passage inoculation. (A) Schematic diagram of passage inoculation experiment. (B) Representative symptom of passage-inoculated revertant virus. The photographs were taken at 5 to 10 days postinoculation. (C) Viral RNA titer in uninoculated (systemically infected) upper leaves of N. benthamiana leaves at the 5th, 10th, and 15th passages. The values of normalized intensity of the CMV RNA 4 band (RNA 4/28S rRNA) are indicated under the CMV RNA panel.

Mmaj/wt-Rep1 and Mmin/wt-Rep2 showed a severe wilt phenotype at the 2nd passage. Viral RNA titers of these viruses were restored at the 5th passage (Fig. 2B and C). An amino acid change from Ser to Thr at position 84 (S84T) in the CPs of these viruses was observed at the 1st passage, although the mutated nucleotide was different between Mmaj/wt-Rep1 and Mmin/wt-Rep2 (Fig. 3). Further nucleic acid mutations did not occur until the 15th passage. Symptom severity of Mmaj/wt-Rep2, Mmin/wt-Rep1, and Mmin/wt-Rep3 did not change during the 15th passage (Fig. 2B). The viral RNA titers of Mmaj/wt-Rep2 and Mmin/wt-Rep1 were restored at the 5th passage, but the viral RNA titer of Mmin/wt-Rep3 remained low (Fig. 2C). Only one to three amino acid changes and some synonymous mutations in different sites were observed in these viruses at the 5th to 15th passages: amino acid changes occurred both in the 5′-mutated half and in the 3′ wild-type half (Fig. 3). Mmaj/wt-Rep3 showed an asymptomatic phenotype at the 7th passage (Fig. 2B). The viral RNA titer of Mmaj/wt-Rep3 was decreased to 1/100 that of Fny at the 10th passage (Fig. 2C). An amino acid change from Pro to Ser at position 17 in the CP was observed at the 10th passage. In addition, three nucleotide deletions in the 2b gene (nucleotides [nt] 174 to 176), which resulted in a newly introduced stop codon at amino acid (aa) 57 of the 2b protein, were observed in the 2b gene of Mmaj/wt-Rep3, indicating that the asymptomatic phenotype and low viral RNA titer of Mmaj/wt-Rep3 were probably due to a truncated 2b protein. Changes in symptom severity, viral RNA titer, and CP nucleic acid sequences were not observed in the three passage series of Fny (Fig. 2).

FIG 3.

Nucleic acid mutations in the synonymously mutated chimeric CP genes by serial-passage inoculation. Bold type indicates the nonsynonymous mutation with amino acid change (in parentheses). The circled number indicates the passage number in which the mutation was found. The secondary structure of the CP in the amino acid mutation region is shown at the bottom (23).

Recovery of viral RNA titer of serially passaged large-scale synonymous mutants is due to the amino acid change(s) in the CP.

Because viral sequences other than CP and 2b were not determined, we confirmed that the recovery of the viral RNA titer after serial passage inoculation was due to the amino acid change(s) in the CP gene. The CP gene of pG1143A was replaced by the CP gene from the 10th passage of each passage series of the 5′ mutants (Fig. 4A). Mmaj/wt-S84T (Thr at position 84) and Mmin/wt-S84T induced severe wilt, while other viruses showed mosaic symptoms similar to those for Fny (Fig. 4B). Viral RNA titers of Mmaj/wt-S84T and -R82K/L134I and Mmin/wt-G146R, -S84T, and -L114F/A193T were similar to those of Fny, while the viral RNA titer of Mmaj/wt-P17S remained low (Fig. 4C). Hence, recovery of viral RNA titer of serially passaged 5′ mutants is due to the nucleic acid change in the CP, except for Mmaj/wt-P17S.

FIG 4.

Amino acid mutation in the CP restores the viral RNA titer of 5′ mutants (Mmaj/wt and Mmin/wt). (A) Diagrams of the chimeric CP mutants used in this study. (B) Symptoms of N. benthamiana plants inoculated with 5′ mutants with revertant CP. The photographs were taken at 5 to 10 days postinoculation. (C) Viral RNA titers in uninoculated upper leaves of N. benthamiana plants infected with Fny or 5′ mutants with each six revertant CP. Two N. benthamiana individuals were used for each mutant. The mean values of normalized intensity of the CMV RNA 4 band (RNA 4/28S rRNA) obtained in experiments with duplicate samples are indicated under the CMV RNA panel. The data are representative of those from two independent experiments.

Three of the amino acid changes (84T, 114F, and 146R) were not found in any CMV strains deposited in the NCBI database. Each amino acid change was introduced into the CP genes of Mmaj and Mmin (Fig. 5A). The viral RNA titers of both Mmaj-S84T and Mmin-S84T were restored, while the viral RNA titers of Mmaj, Mmin, Mmaj-L114F, Mmin-L114F, Mmaj-G146R, and Mmin-G146R were not detected or very low (Fig. 5B). Thus, only the S84T mutation in the CP resulted in recovery of the viral RNA titers of Mmaj and Mmin in which large-scale synonymous substitutions were introduced in the full-length CP gene.

FIG 5.

An S84T mutation in the CP restores the viral RNA titer of Mmaj and Mmin, in which large-scale synonymous substitutions were introduced in the full-length CP gene. (A) Diagrams of the CP mutants used in this study. (B) Viral RNA titer in uninoculated upper leaves of N. benthamiana plants infected with Mmaj and Mmin with each amino acid mutation (S84T, L114F, or G146R) in the CP. Two N. benthamiana individuals were used for each mutant. The mean values of normalized intensity of the CMV RNA 4 band (RNA 4/28S rRNA) obtained in experiments with duplicate samples are indicated under the CMV RNA panel. The data are representative of those from two independent experiments.

RNA composition within the particles.

To clarify why the virus RNA titers of the large-scale synonymous mutants decreased and why the virus RNA titer was restored by amino acid mutation of the CP, RNA composition within the particles was examined. In addition to Fny and two 5′ mutants (Mmaj/wt and Mmin/wt), Mmaj/wt-S84T and Mmin/wt-S84T were used, since the S84T mutation occurred in both Mmaj/wt and Mmin/wt (Fig. 3) and the viral RNA titers of S84T mutants were higher than for Fny (Fig. 4). After partial purification of viral particles from infected N. benthamiana, relative levels of individual viral RNAs were examined (Fig. 6). The levels of RNAs 3 and 4 in the particles of the Mmaj/wt and Mmin/wt were similar to those of Fny, Mmaj/wt-S84T, and Mmin/wt-S84T. This result indicated that the large-scale synonymous substitutions in the 5′ half of the CP gene did not affect the RNA composition within the particles. Interestingly, the amount of RNA 1 in the particles of Mmaj/wt-S84T and Mmin/wt-S84T remarkably increased, showing that S84T mutation in the CP changed the RNA composition within the particles.

FIG 6.

RNA composition within the particles of Fny, 5′ mutants (Mmaj/wt and Mmin/wt), and 5′ mutants with amino acid mutation in the CP (Mmaj/wt-84T and Mmin/wt-84T). (A) Diagrams of the CP mutants used in this study. (B) The viral particles were partially purified from the N. benthamiana leaves infected with each virus, and the amount of viral RNAs was analyzed by Northern blot analysis. The data are representative of those from two independent experiments. The values under the panel indicate proportions of the CMV RNA 3 band (RNA 3/RNA 2).

Competitive fitness of S84T mutants of 5′ mutants.

The competitive fitness of Mmaj/wt-S84T and Mmin/wt-S84T was analyzed. An Fny-S84T mutant, in which Ser at position 84 in the Fny was replaced with Thr, was used in a competition with the other mutants (Fig. 7A and B). When 10 N. benthamiana plants were mix inoculated with Fny-S84T and Mmaj/wt-S84T or Mmin/wt-S84T, Fny-S84T became dominant in all inoculated plants (Fig. 7C and D), showing that the competitive fitness of Mmaj/wt-S84T and Mmin/wt-S84T in N. benthamiana was not restored.

FIG 7.

Competitive fitness of Mmaj/wt-S84T and Mmin/wt-S84T. (A) Diagrams of the CP mutants used in this study. (B) Viral RNA levels in uninoculated upper leaves of N. benthamiana plants infected with Fny and Fny-84T mutants. Three N. benthamiana individuals were used for each mutant. The mean values of normalized intensity of the CMV RNA 4 band (RNA 4/28S rRNA) obtained in an experiment with triplicate samples are indicated under the CMV RNA panel. The data are representative of those from three independent experiments. (C and D) Restriction fragment length polymorphism (RFLP) analysis for competitive-fitness analysis between wild-type Fny-S84T and Mmaj/wt-S84T (C) or Mmin/wt-S84T (D). Ten N. benthamiana plants (#1 to #10) were inoculated with a mixture of transcripts from the Fny-84T and each mutant. Total nucleic acid was extracted from inoculated leaves at 2 dpi and uninoculated upper leaves at 7 dpi, and CP fragments were amplified by the RT-PCR. RT-PCR products were digested by each virus-specific restriction enzyme (NruI [N] for Fny-S84T, PflFI [P] for Mmaj/wt-S84T, and SmaI [S] for Mmin/wt-S84T). In the control experiments, individual infections with Fny-84T or each mutant were used. A mixture containing equal amount of transcribed RNA from the Fny-84T and each mutant was used as a control. Note that the mutant-specific digested DNA bands as shown in the controls (white arrowheads) were not detected in the mixed-inoculation samples.

DISCUSSION

In general, there are two possible deleterious effects of large-scale synonymous substitutions in viral genes: altering the viral RNA secondary structure or changing the codon usage bias of the coding gene. Importantly, the codon adaptation index of four chimeric mutants against dicots was similar to that of the wild-type CP gene (Table 1), indicating that the reduction of the viral RNA titers of four chimeric mutants was due to altering the RNA secondary structure of CMV RNAs 3 and 4 rather than the changing the codon usage bias of the CP gene. In animal viruses, Le Nouën et al. analyzed phenotypic reversion of attenuated human respiratory syncytial virus, in which the codon usage of the L polymerase gene was deoptimized, by serial passage using a strong selection pressure of high temperature (21). The authors reported that the attenuated phenotype of deoptimized virus recovered through a single prominent amino acid mutation in the M2-1 gene (21). Similarly, we showed that the viral RNA titers of 5′ mutants (Mmaj/wt and Mmin/wt) were rapidly restored by one to two amino acid mutations in the CP during serial-passage inoculation (Fig. 4). Serial-passage inoculation selects viruses with strong propagative ability in the context of strong selective pressure, so mutants in which viral titers have been recovered became dominant during passage inoculation. It is unlikely that viral RNA secondary structure reverted to its original state by the few nucleic acids mutations in the CP gene. It is reasonable to interpret that the recovery of viral RNA titer was due to the amino acid mutation of the CP. CMV CP is the sole protein associated with the virus particles and is also required for cell-to-cell movement (2). Therefore, deleterious effects of large-scale synonymous substitutions in RNA 3 that were restored by the amino acid mutation of the CP are most likely in the integrity of the viral particles and/or cell-to-cell movement.

In small icosahedral single-stranded viruses, the viral RNA genome folds into a compact secondary/tertiary structure to drive packaging (5). Tubiana et al. (6) showed that synonymous mutations of brome mosaic virus RNA 2 reduced genome compactness compared to that of wild-type RNA 2. Thus, the physical features of viral RNA are optimized for assembly of their capsid and the physical nature of capsids of small icosahedral RNA viruses has negative selective pressure for viral RNA mutation. In addition, the integrity of CMV particles depends on the CP-viral RNA interactions, and CMV particles are easily dissociated by disruption of the electrostatic CP-RNA interactions (22). Although the large-scale synonymous substitutions in the 5′ half of the CP gene do not affect the relative RNA composition within the particles (Fig. 6), alteration of global RNA secondary structure of CMV RNAs 3 and 4 by large-scale synonymous substitutions might result in inefficient particle assembly or disintegration of viral particles by perturbation of the CP-RNA interactions, leading to decreased viral RNA titer. By stimulating the viral evolution by serial-passage inoculation, the viral RNA titers of 5′ mutants (Mmaj/wt and Mmin/wt) were restored by one to two amino acid mutations in the CP (Fig. 3). Interestingly, the mutated amino acids, except for at residue 17, surround the 6-fold axis of symmetry in the T3 structure of the virus particle (data not shown). Amino acids 82, 114, 134, 146, 193, and 194 are located in the loop structure, while residue 84 is the first amino acid of β strand C in the CP (Fig. 3) (23, 24). The loops in the CMV CP are more flexible than those of the tomato aspermy virus (TAV) CP (24, 25), and amino acid residues in these loops of the CMV CP are involved in aphid vector transmission or symptom determination (3, 26). The amino acid changes in the flexible loops could influence the interaction between capsid subunits or between CP and viral RNA, leading to effective assemble or stability of the CMV particle with large-scale synonymously substituted viral RNA. A similar possibility can also be considered for residue 84 because this amino acid is adjacent to the flexible loop. Detailed observations of particle structure by using cryo-transmission electron microscopy (cryo-TEM) are necessary to verify this.

During the cell-to-cell movement of CMV, the movement protein (MP) forms ribonucleoprotein complexes (RNP) with viral RNAs supported by the CP. The MP directly binds to viral RNAs without obvious sequence specificity, which is important for RNP structure and cell-to-cell movement. Andreev et al. (27) and Kim et al. (28) showed that C-terminally truncated MP changed the RNP structure and eliminated the requirement for the CP in CMV cell-to-cell movement. They suggested that the MP conformation was altered by the CP to increase its ability to bind to viral RNAs and form appropriate RNP structures. From these, it is suggested that the large-scale synonymous substitutions affect interaction between RNA 3 and the MP, leading to an inappropriate RNP structure and inefficient cell-to-cell movement. The CP with amino acid mutation may change the MP conformation that is suitable for interaction with large-scale synonymously substituted RNA 3, making an appropriate RNP structure for cell-to-cell movement in the large-scale synonymous mutant.

Most of the mutated amino acids are not found in the CMV strains deposited in the NCBI database. There is a possibility that these amino acid changes have negative effects on viral multiplication of the wild-type CMV. However, introduction of the S84T mutation in the wild-type CP did not change the viral titer (Fig. 7), suggesting that the S84T mutation was not eliminated by negative selection for reduced levels of the viral RNA. Since the S84T mutation resulted in a severe wilt phenotype (Fig. 4 and 5), it seems likely that the S84T mutant could not survive in a natural environment. It is also possible that amino acid changes in the CP affect aphid transmissibility (26).

In this study, we found that deleterious effects of large-scale synonymous substitutions facilitated amino acid changes in the CP. Because we used artificial mutants, the role of this phenomenon in viral evolution in nature is unclear. Recombination between related viruses is one powerful mechanism of RNA virus evolution, and 17% of the CMV strains isolated from plants in Spain were recombinants (29). The authors suggest that genetic exchange by recombination seems to have a fitness cost, and many hybrid genotypes might disappear from the population. One reason may be that recombination between different viruses results in a deoptimized RNA secondary structure. To resolve this problem, amino acid mutation in CP may be generated rapidly to tolerate the deoptimized RNA secondary structure during recombination events in nature. Finally, indirect effects of large-scale synonymous substitutions should be considered. The large-scale synonymous substitutions seem to have resulted in deleterious changes in RNA structure, leading to lower fitness of the mutants. During evolution, low-fitness mutants restored fitness by fixing beneficial mutations among various mutations generated by the error-prone replication of viral RNA. Fixation of beneficial mutations may appear prominently in low-fitness mutants but not in wild-type virus with high fitness, which could explain why the beneficial S84T mutation did not occur in the wild-type Fny during the passage inoculation or in any other isolates of CMV. In this case, fixation of beneficial mutations may be a direct effect of low fitness, while the effect of the large-scale synonymous substitutions on fixing the beneficial mutations may be an indirect effect of lowering the fitness by deleterious changes in RNA structure.

MATERIALS AND METHODS

Virus and construction of CMV RNA 3 mutants.

Clones of Fny CMV capable of producing infectious transcribed RNA have been described previously (30, 31). Plasmid pG1143A is a mutant derived from pFny309; residue A at nt 1143 in the intercistronic region (ICR) was changed to G, resulting in a new restriction enzyme ApaI site.

Two codon usage-modified CP genes in which all the codons of nine amino acids (A, D, E, L, P, Q, R, S, and Y) were replaced by either the most frequently (Mmaj) or the least frequently (Mmin) used synonymous codons in monocots (19) were designed. The de novo DNA fragments containing the ApaI site in the ICR, the modified CP gene, and the PstI site in the 3′ end of RNA 3 were artificially synthesized by Genscript (Piscataway, NJ). The ApaI-PstI fragment of pG1143A was replaced with each artificial DNA fragment to create pFny309-Mmaj and pFny309-Mmin, respectively.

The four chimeric CP mutants with modifications in the 5′ or 3′ half of the CP gene were created using the common HindIII sites in the CP gene and the pUC18 vector. These chimeras were constructed by exchanging the HindIII fragments, creating wt/Mmaj (pFny309-wt/Mmaj2), Mmaj/wt (pFny309-Mmaj/wt), wt/Mmin (pFny309-wt/Mmin2), and Mmin/wt (pFny309-Mmin/wt).

The six CP mutants that have the CP gene of each serially passaged 5′ mutant at the 10th passage were constructed using the ApaI and XhoI (nt 1836 in the CP gene) sites. ApaI/XhoI-digested RT-PCR fragments of each mutated CP gene were introduced in the ApaI/XhoI site of pG1143A, creating pFny309-jwRep1CP (Mmaj/wt-S84T), pFny309-jwRep2CP (Mmaj/wt-R82K/L134I), pFny309-jwRep3CP (Mmaj/wt-P17S), pFny309-iwRep1CP(Mmin/wt-G146R), pFny309-iwRep2CP (Mmin/wt-S84T), and pFny309-iwRep3CP(Mmin/wt-L114F/A193T).

A mutated Mmaj clone in which Ser at position 84 in the Mmaj CP was replaced with Thr was constructed using the PflFI (nt 1539 in the CP gene) and PstI sites (in the 3′ end of RNA 3). The PflFI/PstI-digested fragments from pFny309-Mmaj were introduced in the PflFI/PstI sites of pFny309-jwRep1CP. Similarly, a mutated Mmin clone in which Ser at position 84 in the Mmin CP was replaced with Thr was constructed using the HindIII sites. The HindIII-digested fragments from pFny309-Mmin were introduced in the HindIII sites of pFny309-iwRep2CP. The resulting mutated RNA 3 clones were named pFny309-MmajS84T and pFny309-MminS84T.

To create mutated Mmaj and Mmin clones in which Leu at position 114 or Gly at position 146 was replaced with Phe or Arg, respectively, site-directed mutagenesis was conducted by inverse PCR with a high-fidelity KOD Plus Neo DNA polymerase (Toyobo, Osaka, Japan) using pFny309-Mmaj or pFny309-Mmin as a template and mutagenesis primers MmajL114F-F (5′-ATC CGT TCC CGA AAT TTG A-3′), MmajL114F-R (5′-TAA CCC TAA TCT GAA TCC TG-3′), MmajG146R-F (5′-ATG TTC GCC GAC AGG GCC TC-3′), MmajG146R-R (5′-GGC GGA GAT GGC GGC AAC-3′), MminL114F-F (5′-CGA GTT AAT CCC TTT CCC AA-3′), MminL114F-R (5′-AAT TTG AAT TCG ACT AAC AAG C-3′), MminG146R-F (5′-CGC AGA TCG AGC AAG TCC-3′), and MminG146R-R (5′-AAC ATT GCA CTG ATT GCT G-3′). The PCR product was digested by DpnI, phosphorylated, self-ligated, and then transfected into Escherichia coli DH5α. The resulting mutated RNA 3 clones were named pFny309-MmajL114F, pFny309-MmajG146R, pFny309-MminL114F, and pFny309-MminG146R. Similarly, a mutated CP gene in which Ser at position 84 in wild-type CP was replaced with Thr was constructed by inverse PCR with primers FNY3-1515-fw (5′-GTA ATA AGT CCC ACG GTC TA-3′) and FNY3-T1506A-rv (5′-GGT AAA AGG TTG TTA CTA CCT G-3′).

Codon adaptation index calculation.

The codon usage of arabidopsis (Arabidopsis thaliana) and that of rice (Oryza sativa) were obtained from the Codon Usage Database (http://www.kazusa.or.jp/codon/). The codon adaptation index against arabidopsis and rice, effective number of codons, and GC content of the CP genes were calculated by the “CAI calculation” program in the CAIcal server (http://genomes.urv.es/CAIcal/).

Transcription, inoculation of infectious RNA, and maintenance of inoculated plants.

Infectious RNA were synthesized in vitro by combining each RNA 3 clone with pFny86.1 and pFny209 (30, 31). After linearization by PstI digestion, infectious RNA swere synthesized using the SP6 (for RNA 1) or T7 (for RNA 2 and RNA 3) RiboMAX large-scale RNA production system (Promega, Madison, WI). Two larger leaves of five- or six-leaf-stage N. benthamiana plants were mechanically inoculated with transcribed RNA in 50 mM Na₂HPO₄ (pH 9.2), and the inoculated plants were maintained in a growth chamber with a 16-h light/8-h dark cycle at 25°C.

Extraction of nucleic acids from infected leaves.

Circa 50 mg of leaf tissues (six leaf discs of 0.8 cm each) was ground with a mortar and pestle and mixed with 500 µl of extraction buffer (0.1 M NaCl, 0.01 M Tris-HCl [pH 8.0], 0.1 mM EDTA, and 1% SDS). The mixture was extracted with an equal volume of phenol-chloroform (1:1 [vol/vol]) twice and precipitated with 0.3 M sodium acetate and ethanol. After washing with 70% ethanol, the nucleic acids were resuspended in 50 μl of diethyl phosphoramidate (DEPC)-treated Milli-Q water.

Small-scale viral particle preparation.

Small-scale viral particle preparation was carried out by the clarified virus concentrate method (32), with some modifications. Approximately 100 mg of symptomatic upper leaf tissues (12 leaf discs of 0.8 cm each) was ground with a mortar and pestle in 500 μl of citrate buffer (0.5 M sodium citrate and 5 mM EDTA [pH 6.5]) containing 0.5% thioglycolic acid. After removal of cellular debris and nuclei by centrifugation at 10,000 rpm for 10 min, 500 μl of chloroform was mixed with the supernatant by vortexing, and the mixture was centrifuged at 10,000 rpm for 10 min. Viral particles in the supernatant were precipitated with 8% polyethylene glycol (PEG; molecular weight [MW], 6,000) and 0.125 M NaCl. Final pellets were resuspended in 300 μl of borate buffer (5 mM sodium borate and 0.5 mM EDTA [pH 9.0]) and centrifuged at 10,000 rpm for 10 min. The supernatant was used for RNA extraction. An equal volume of viral RNA extraction buffer (200 mM Tris [pH 8.0], 1 M NaCl, 1% SDS, 2 mM EDTA) was added, and then viral RNA was purified by phenol-chloroform extraction and ethanol precipitation (33).

Northern blot analysis for CMV RNA.

For the detection of CMV RNAs, 1 to 3 μg of total nucleic acids was loaded onto a 1.5% denaturing agarose gel and transferred onto a Biodyne Plus membrane (Pall, East Hills, NY). After linearization using HindIII, a digoxigenin (DIG)-labeled RNA probe was transcribed in vitro by T7 RNA polymerase from pFny2-3′ UTR, which contains the 430 bp for the 3′ end of CMV Fny RNA 2, which is shared by all four viral RNAs. The membrane was hybridized with DIG-labeled RNA probes in DIG Easy Hyb (Roche Diagnostics, Mannheim, Germany). The detection of DIG signals using CDP-star was performed according to the manufacturer's instructions (Roche Diagnostics). The intensity of CMV RNA 4 and 28S rRNA bands was determined with ImageJ software (National Institutes of Health, Bethesda, MD).

Serial-passage inoculation.

The wild-type Fny and two 5′ mutants (Mmaj/wt and Mmin/wt) were serially passaged in N. benthamiana 15 times by mechanical inoculation. The transcribed RNAs from clones for each virus were used to mechanically inoculate three N. benthamiana plants; these three plants represented the first passage in three independent series for each viruses (Fig. 2A). Six leaf discs sampled as described above from systemically infected leaves of the 1st N. benthamiana plant were ground in 300 μl of 50 mM Na₂HPO₄ (pH 7.0), and the sap extract was used to inoculate three fresh N. benthamiana plants. From the 2nd passage, we sampled two discs from each plant, for a total of six discs from three plants for the 2nd passage. Mixed sap extract from uninoculated leaves of all three plants per lineage was used to inoculate three new plants for the 3rd passage. This procedure was repeated for 15 passages.

At the 1st, 5th, 10th, and 15th passages, we sampled six leaf discs from three N. benthamiana plants (two discs from each plant) per series for analysis. Viral RNA titer was analyzed by Northern blot analysis, described above. The consensus CP gene sequences at the 1st, 5th, 10th, and 15th passages were also determined by direct sequencing of RT-PCR products. For the direct sequencing, CMV CP genes were amplified by the PrimeScript One Step RT-PCR kit version 2 (TaKaRa, Otsu, Japan) in a 15-μl reaction mixture containing 7.5 μl of 2× buffer, 0.6 μl of enzyme mix, 1.5 μl of primer mix (10 μM each IRM2-F and T1969A-R), 1 μl of RNA template solution, and 4.4 μl of DEPC-treated Milli-Q water. CMV 2b genes were also amplified by the PrimeScript One Step RT-PCR kit version 2 (TaKaRa) with CMV-2a-fw (5′-CAG AGT TGA GCG TGT AAA TTC C-3′) and CMV3′UTR-rv (5′-CAG AAC TGC CAA CTC AGC TCC-3′). After PEG precipitation, PCR products were Sanger sequenced with appropriate primers. Assembly and analysis of sequences were done by DNADynamo software (Blue Tractor Software, North Wales, UK).

Competitive-fitness analysis.

Ten N. benthamiana plants were inoculated with a mixture of RNA 1, RNA 2, and an equal amount of transcribed RNAs from pFny309-CPS84T (Fny-S84T) and pFny309-jwRep1CP (Mmaj/wt-S84T) or pFny309-iwRep2CP (Mmin/wt-S84T). Inoculated leaves were excised and sampled at 2 dpi. Uninoculated upper leaves were sampled at 7 dpi. Total nucleic acids were extracted from sampled tissues and used as a template for RT-PCR. The primer T2061R (5′-GCT GGC GTG GAA TTC TCC ACG AC-3′), which binds to nt 2049 to 2072 in the 3′ untranslated region (UTR) of CMV RNA 3, was used in the RT. The primers IRM2-F (5′-CTC CCT GTT GGG CCC CTT AC-3′) and T1969A-R (5′-CAG TTT ATA GCA GAT CTG CCA AC-3′), which amplify nt 1132 to 1983 from the intercistronic region to the 3′ UTR, were used for the PCR. RT was performed using ReverTra Ace (Toyobo) according to the manufacturer's instructions. PCR was conducted with KAPA taq Extra DNA polymerase (KAPA Biosystems, Wilmington, MA) in a 15-μl reaction mixture containing 3 μl of 5× buffer, 0.05 μl of KAPA taq enzyme, 0.45 μl of 10 mM deoxynucleoside triphosphates (dNTPs), 1.2 μl of MgCl₂ solution, 1 μl of each primer (10 μM), 2 μl of five-times-diluted RT product, and 6.5 μl of distilled Milli-Q water. The PCR was performed with a T100 thermal cycler (Bio-Rad) using an initial denaturation at 96°C for 5 min, followed by 35 cycles at 96°C for 30 s, 56°C for 1 min, and 72°C for 1 min and a final extension at 72°C for 10 min. The PCR product (5 μl) was digested with each virus-specific restriction enzyme shown in Fig. 1A and then analyzed by electrophoresis in a 2.5% agarose gel. The presence or absence of each wild-type Fny or mutants in a sample was determined by the presence or absence of its diagnostic bands. A mixture of equal amounts of transcribed RNA of wild-type Fny and each mutant was used as a control template for RT-PCR.