A Conserved Capsid Protein Surface Domain of Cucumber Mosaic Virus Is Essential for Efficient Aphid Vector Transmission

Abstract

A prominent feature on the surfaces of virions of Cucumber mosaic virus (CMV) is a negatively charged loop structure (the βH-βI loop). Six of 8 amino acids in this capsid protein loop are highly conserved among strains of CMV and other cucumoviruses. Five of these amino acids were individually changed to alanine or lysine (an amino acid of opposite charge) to create nine mutants (the D191A, D191K, D192A, D192K, L194A, E195A, E195K, D197A, and D197K mutants). Transcripts of cDNA clones were infectious when they were mechanically inoculated onto tobacco, giving rise to symptoms of a mottle-mosaic typical of the wild-type virus (the D191A, D191K, D192A, E195A, E195K, and D197A mutants), a systemic necrosis (the D192K mutant), or an atypical chlorosis with necrotic flecking (the L194A mutant). The mutants formed virions and accumulated to wild-type levels, but eight of the nine mutants were defective in aphid vector transmission. The aspartate-to-lysine mutation at position 197 interfered with infection; the only recovered progeny (the D197K∗ mutant) harbored a second-site mutation (denoted by the asterisk) of alanine to glutamate at position 193, a proximal site in the βH-βI loop. Since the disruption of charged amino acid residues in the βH-βI loop reduces or eliminates vector transmissibility without grossly affecting infectivity or virion formation, we hypothesize that this sequence or structure has been conserved to facilitate aphid vector transmission.

The vectors of plant viruses include insects, nematodes, mites, and fungi, and the majority of viruses have arthropod vectors (26). In cases where plant virus transmission has been studied in detail, the capsid protein has been shown to be the primary determinant for transmission by vectors (13, 14, 37, 40). The interactions between virus capsid proteins and their vectors are either direct or mediated by an accessory or helper protein (22, 38). Transmission involves the binding and retention of virions to sites within the vector, although little is known about the ligands to which viruses bind. Genetic studies have revealed that specific viral capsid protein domains and amino acids are determinative for transmissibility and vector specificity. Examples can be found for aphid-transmitted viruses in the genera Potyvirus (1, 2, 11), Luteovirus (53), Polerovirus (3), Enamovirus (7), and Cucumovirus (35, 36, 44); for the nematode-transmitted viruses in the genus Tobravirus (25); and for fungus-transmitted viruses in the genera Tombusvirus (18, 40) and Benyvirus (52). An emerging theme for a number of these viruses is that a minor capsid protein readthrough protein plays an important role in transmission; this protein is generated by a mechanism involving readthrough of the major capsid protein stop codon (reviewed in reference 4).

Cucumber mosaic virus (CMV) is an aphid-transmitted virus, the transmission of which has been studied in some detail (33). It is a plus-sense RNA virus with three single-stranded genomic RNAs (32). RNAs 1 and 2 each encode a replication-related protein (15). CMV RNA 2 also encodes an 11- to 13-kDa protein (2b) that influences virulence and functions as a suppressor of plant host-mediated gene silencing (8, 9). RNA 3 encodes two proteins, the 3a protein involved in the cell-to-cell movement of the virus and the 24.1-kDa capsid protein. The capsid protein is multifunctional; in addition to having a role in encapsidation, it affects virus movement in plants (19, 48), transmission, symptom expression, and host range (43).

CMV virions have T=3 icosahedral symmetry and consist of 180 copies of the single, virus-encoded capsid protein. The 3.2-Å-resolution X-ray crystallographic structure of CMV has recently been described (46). This structure revealed a number of remarkable differences between the A subunits (those at the fivefold axis of symmetry) and the B and C subunits (those at the threefold axis of symmetry). The most striking difference was a cluster of six amphipathic helices present in the B and C subunits and oriented down into the virion core at the threefold axis; these helices were not present in the A subunits at the fivefold axis (46). The lateral βF-βG loop structure faces the threefold and fivefold axes of symmetry, but its structure differs in the respective hexameric and pentameric capsomeres. A similar difference in structure was observed for the βE-βF loop, which makes contact among the A subunits about the fivefold axis and forms the contact surface between the B and C subunits at the threefold axis. In contrast to the βE-βF and βF-βG loops, all of the three prominent surface loop structures, βB-βC, βD-βE, and βH-βI, have nearly identical structures in the A, B, and C subunits.

The transmission of CMV in nature is dependent upon aphid vectors, and virions are transmitted in a nonpersistent or stylet-borne manner (10). The CMV capsid protein has been shown to be the primary determinant for vector transmission (5, 12); no additional helper component is known to be required for the transmission of virions (27). Molecular and genetic analyses of strains that are defective in aphid transmission have revealed that different regions of the viral capsid protein variably affect transmission and that the efficiency of transmission of CMV varies with the species of aphid vector. Amino acids that affect virus transmission by the aphid Aphis gossypii have been mapped to two capsid protein positions, 129, located in the βE-βF loop, and 162, located in the βF-βG loop (36). Three additional capsid protein amino acid positions affect transmission by the aphid Myzus persicae (35). Surprisingly, most of these amino acids that affect transmission are buried in the folded polypeptide or between subunits in assembled virions. In examining what capsid protein domains are clearly exposed on the virus, it was noted that the amino acid sequence of one of three prominent surface loops was conserved among cucumoviruses. This report describes the amino acid conservation of the βH-βI loop and how mutations in this CMV surface domain affect vector transmission.

MATERIALS AND METHODS

Construction of virus mutants.

PCR-based site-directed mutagenesis was used to introduce mutations into the capsid protein gene of a CMV-Fny RNA 3 cDNA clone, pFny309 (39). The strategy for the generation of mutated PCR fragments was based on the procedure of Higuchi et al. (16). The high-fidelity, thermostable polymerase Pfu (Stratagene) was used in accordance with the manufacturer's instructions. Recombinant DNA techniques were performed by standard procedures (42). The nomenclature and labeling of putative capsid protein domains were as described for the atomic structure of the virus (46). For introducing mutations into the βH-βI loop, PCR-generated fragments encoding single amino acid mutations were digested with HindIII and PstI. This DNA was ligated into pFny309 that had previously been digested with HindIII and PstI. The 5′-to-3′ nucleotide sequences of the oligonucleotide primers used in this study, shown with the flanking nucleotide numbers corresponding to the numbering of the sequence of CMV-Fny RNA 3 (accession number D10538 [31]) and with the mutated base positions in bold, are as follows: for the D191A mutant, 1818-TATTCAAAAGCGGATGCGCTCGAGACG-1844; for the D191K mutant, 1818-TATTCAAAAAAGGATGCGCTCGAGACG-1844; for the D192A mutant, 1817-GTATTCAAAAGACGCGGCGCTCGAGACG-1844; for the D192K mutant, 1817-GTATTCAAAAGACAAGGCGCTCGAGACG-1844; for the L194A mutant, 1827-GACGATGCGGCCGAGACGGACGAG-1850; for the E195A mutant, 1827-GACGATGCGCTCGCCACGGACGAG-1850; for the E195K mutant, 1827-GACGATGCGCTCAAGACGGACGAG-1850; for the D197A mutant, 1833-GCGCTCGAGACGGCGGAGCTAGTACTT-1859; and for the D197K mutant, 1833-GCGCTCGAGACGAAGGAGCTAGTACTT-1859.

These primers were used in combination with a 3′-end primer with the sequence 5′-GCATGCCTGCAGTGGTCTCCT-3′, corresponding to the 9 terminal nucleotides of RNA 3 (in italics; positions 2216 to 2208) with an introduced PstI restriction enzyme site. The restriction enzyme sites in this primer and in other primers are underlined. In order to produce a second PCR-mutated fragment required for the Higuchi procedure (16), primers complementary to each of the above primers were synthesized and used in combination with a primer overlapping the HindIII site in the cDNA for CMV RNA 3, 1554-GATAAGAAGCTTGTTTCGCGC-1574. The mutants were all engineered to contain a single capsid protein amino acid change; the mutation terminology specifies the wild-type amino acid (in the single-letter code) followed by the amino acid position number and then the mutant amino acid. For the D197K∗ mutant, the asterisk indicates the presence of an additional second-site mutation (see Results). The coat protein genes in all of the resulting clones were sequenced to confirm both the presence of the introduced mutations and the absence of any spurious nucleotide changes. The ribbon model depicting the positions of the βH-βI loop and the mutations was created with the program Molview (45; http://www.danforthcenter.org/smith/molview.htm).

In vitro transcription of viral RNAs, the inoculation of transcripts, and aphid transmissions.

Capped, in vitro transcripts from the full-length cDNA clones pFny109 (RNA 1), pFny209 (RNA 2), and pFny309 (RNA 3) (39) were synthesized with a MEGAscript T7 transcription kit (Ambion). Transcripts were inoculated onto tobacco (Nicotiana tabacum cv. Xanthi NN) and Chenopodium quinoa, and the mutant viruses were subsequently maintained in tobacco. Ten to 21 days after inoculation, virus transmission by the aphid Aphis gossypii was assayed as previously described (35). N. tabacum was used as both the source and the target plant, with 10 aphids transferred onto each target plant. Approximately five target plants were used in each experiment. Young tobacco plants (leaf stage, four to six leaves) were employed as target plants, and under these conditions, infection by the both the wild-type and mutant viruses gave rise to clearly visible symptoms. Plants were rated for virus (transmission) 14 to 21 days following transfer of the aphids to the target plants. In the case of the D197K mutant, asymptomatic plants were assayed by an enzyme-linked immunosorbent assay (34).

Virus purification, electron microscopy, and cDNA cloning of RNA 3.

Viruses were purified from infected tobacco by the methods of Lot et al. (24), and the concentrations of the viruses were determined spectrophotometrically (29). Purified virions or those in crude sap were negatively stained with 2% uranyl acetate and examined on a Philips EM-200 transmission electron microscope. For analyses of mutant viruses derived from transcript-inoculated plants, the coat protein genes from single cDNA clones were sequenced to determine whether the introduced mutation had been retained and whether any additional nucleotide changes were present. In the case of the D197K∗ mutant, two cDNA clones from independent reverse transcription-PCRs were sequenced. cDNA corresponding to CMV RNA 3 was synthesized with a SuperScript preamplification system for a First Strand cDNA synthesis kit (GibcoBRL) used in accordance with the manufacturer's instructions. Template RNAs were genomic CMV RNAs extracted from purified virions (34). The 3′-end primer described above was used for first-strand cDNA synthesis. Full-length RNA 3 cDNAs were amplified by PCR by using Pfu DNA polymerase (Stratagene) with a 5.5-min extension reaction. The sequence of the second primer was 5′-CCGGATCCTAATACGACTCACTATAGGTAATCTTACCACTGT-3′; this contains a T7 RNA polymerase promoter sequence, a BamHI site (underlined), and CMV RNA 3 nucleotide positions 1 to 16 (in italics), as previously described (39). Full-length PCR products were digested with BamHI and PstI and ligated into the corresponding polylinker site of the vector pUC19.

Agarose gel electrophoresis of CMV particles.

Approximately 10 μg of purified virus was prepared in 40 μl of borate buffer (5 mM sodium borate, 0.5 mM EDTA [pH 9.0]). Samples were electrophoresed at 4°C in a 1% agarose gel in a Tris-acetate buffer (40 mM Tris-HCl, 20 mM sodium acetate, 1 mM EDTA [pH 7.5]) for 2 h at 3.0 V/cm and then for 5 h at 2.4 V/cm. Virus and nucleic acids were stained in 0.5 mg of ethidium bromide per ml for 15 min, rinsed with distilled water, and photographed. Gels were treated with 40% (vol/vol) methanol and 10% glacial acetic acid for 30 min and dried in a gel dryer under vacuum for 1 h without heating and then for 30 min at 80°C. The dried gels were stained with Coomassie blue for 1 to 2 h, destained, and then dried.

RESULTS

A CMV capsid protein surface domain is highly conserved.

Preliminary structural analyses of CMV using cowpea chlorotic mottle virus (CCMV) as a model resulted in an alignment of capsid protein sequences for all representative members of the Bromoviridae (54). A determination of the atomic structure of CMV confirmed that the βH-βI loop of the capsid protein was on the surfaces of virions and included 8 amino acids (46) (Fig. 1). In reviewing the capsid protein gene sequences of members of the genus Cucumovirus, we observed that 6 of these amino acids are nearly invariant among the three subgroups of CMV (41), the three subgroups of Peanut stunt virus (55), and Tomato aspermy virus (Table 1). This degree of amino acid conservation in a surface loop structure is uncommon among members of an entire genus and suggests an essential function. We hypothesized that the βH-βI loop domain plays a role in vector transmission.

FIG. 1.

Ribbon model of the CMV capsid protein and sites of mutagenesis. (A) The CMV capsid protein is presented as a ribbon highlighting the location of the βH-βI loop structure in the center and the bounding βB-βC and βD-βE loops. The carboxy and amino termini of the protein are indicated. The orientation is such that the surface of the virus would be at the top of the figure and the N-terminal alpha helix is directed downward toward the center of the virion. The positions of amino acids in the βH-βI loop that were mutagenized as part of this study are depicted as black spheres. (B) The amino acid sequence of the βH-βI loop, corresponding to positions 191 to 198 of the CMV-Fny capsid protein, is shown. The 5 amino acids that were mutagenized are underlined, and the amino acids to which they were changed are shown below the sequence.

TABLE 1.

Comparison of the amino acid sequences in the capsid protein βH-βI loop domains of cucumoviruses

Virus-strain (subgroup)a	βH-βI loop amino acid sequenceb	Reference
CMV-Fny (IA)	D-D-A-L-E-T-D-E	31
CMV-K (IB)	D-D-A-L-E-T-D-E	41
CMV-Q (II)	D-D-K-L-E-K-D-E	6
PSV-J (I)	D-D-T-L-E-D-D-E	20
PSV-W (II)	D-D-V-L-Q-A-D-E	17
PSV-Mi (III)	D-D-V-L-E-A-D-E	55
TAV-C	D-D-V-L-E-A-D-E	30

^aPSV, Peanut stunt virus; TAV, Tomato aspermy virus.

^bThe βH-βI loop is that observed in the structure of CMV strain Fny by Smith et al. (46). The corresponding domain in other cucumoviruses was identified in the alignment by Wikoff et al. (54). Amino acids conserved in the three viruses are shown in bold.

Capsid protein mutants with altered βH-βI loop amino acids form virions and systemically infect tobacco.

Five of the 8 capsid protein amino acids that form the βH-βI loop were targeted for mutation. A total of nine constructs were prepared, with single encoded amino acid changes introduced into the capsid protein gene (Fig. 1). Transcripts for the nine mutants were infectious when they were mechanically inoculated onto C. quinoa and/or N. tabacum (tobacco). The parental Fny and mutant CMVs induced the formation of local lesions on C. quinoa, with symptoms progressing from chlorotic to necrotic; no systemic infections were observed on this host.

Mechanical inoculations of transcripts for all nine mutants gave rise to systemic infections in tobacco. Seven of the nine mutants induced mottling and mosaic symptoms comparable to those induced by the wild-type CMV-Fny; these symptoms were usually visible 6 to 10 days postinoculation (Table 2). Two of the mutants gave rise to atypical symptoms. Inoculation of tobacco with the D192K mutant resulted in the formation of necrotic lesions on the inoculated leaves. These lesions were not limiting, and the necrosis spread down the petiole and into the stem. In younger tobacco plants (those with four to six fully expanded leaves), the systemic necrosis often resulted in the death of the host. Older plants were able to limit and outgrow the spread of necrosis in the stem and were not killed. In these older plants, upper leaves continued to become systemically infected and necrotic, but new growth outpaced the necrosis. Infection of tobacco by the L194A mutant gave rise to an atypical mottle-mosaic with chlorotic bright-yellow patches; this was followed by the appearance on mottled leaves of a necrotic flecking.

TABLE 2.

Symptoms and rates of aphid transmission in tobacco plants infected with CMV βH-βI loop mutants^a

Strain	Symptom(s) on tobaccob		Aphid transmissionc [no. of plants infected/total (%)]
	Inoculated leaves	Systemically infected leaves
D191A mutant	Chlorotic LL	Mottle-mosaic	28/64 (44)
D191K mutant	Chlorotic LL	Mottle-mosaic	0/20 (0)
D192A mutant	Chlorotic LL	Mottle-mosaic	0/30 (0)
D192K mutant	Necrosis	Mottle-mosaic, necrosis	8/27 (30)
L194A mutant	Chlorotic LL	Yellow mosaic, necrotic flecks	3/31 (10)
E195A mutant	Chlorotic LL	Mottle-mosaic	4/47 (9)
E195K mutant	Chlorotic LL	Mottle-mosaic	1/35 (3)
D197A mutant	Chlorotic LL	Mottle-mosaic	5/58 (9)
D197K∗d mutant	Chlorotic LL	Mottle-mosaic	20/20 (100)
Fny (wild type)	Chlorotic LL	Mottle-mosaic	85/87 (98)

^aTobacco leaves were mechanically inoculated with in vitro-transcribed RNAs or crude sap extracts from infected plants.

^bSymptoms were evaluated 7 to 21 days postinoculation. LL, local lesions.

^cTransmission was tested with 10 Aphis gossypii aphids per tobacco plant.

^dThe D197K mutant appeared to be inviable, as no systemic infection was observed. The asterisk in D197K∗ indicates the presence of a second-site mutation, A193E, as described in the text; this was the only recovered progeny.

Virions of the nine mutants accumulated in tobacco to levels similar to that of the wild-type virus, as determined by purification. Yields for the wild-type and mutant viruses from tobacco ranged from 0.5 to 1.3 mg of virus per g (fresh weight); this type of variation in yield is representative of that observed in multiple purifications of the wild-type virus CMV-Fny. The integrity and properties of virions were evaluated by agarose gel electrophoresis. Virions of six of the nine mutants (the D191A, D192A, L194A, E195A, D197A, and D197K∗ mutants) migrated in the gel; three of the mutants (the D191K, D192K, and E195K mutants) remained in the well (Fig. 2). Eight of the nine introduced mutations were expected to reduce the net negative charge on virions. Consistent with this prediction, virions of the D191A, D192A, E195A, and D197A mutants were retarded in their migration toward the anode relative to that of the wild-type virus. The L194A mutant migrated faster than the wild type, and the D197K∗ mutant migrated to the same position as the wild-type virus. All of the mutants were also evaluated by electron microscopy; the relative sizes and appearances of negatively stained virions were indistinguishable from those of the wild-type virus (data not shown).

FIG. 2.

Agarose gel electrophoresis of CMV mutants. (A) Results for a 1% agarose-Tris-acetate-EDTA gel at 4°C in which virions were subjected to electrophoresis followed by staining with ethidium bromide; (B) same gel after being dried and stained with Coomassie blue. The virus identifications at the bottom of the figure apply to corresponding lanes of both panels.

Mutations in the βH-βI loop dramatically affect transmission by an aphid vector.

The wild-type virus CMV-Fny is efficiently transmitted by the aphid A. gossypii. In 18 experiments with 10 aphids per target plant, 98% of the plants (85 of 87) became infected (Table 2). Five target plants were used per experiment, although in some cases individual plants were lost due to other diseases. In experiments with the βH-βI loop mutants, eight of the nine mutants were defective in aphid transmission (Table 2). Two of the mutants (the D191K and D192A mutants) were not transmitted by A. gossypii. Six of the mutants (the D191A, D192K, L194A, E195A, E195K, and D197A mutants) exhibited markedly reduced rates of transmission.

A compensatory second-site amino acid change restores systemic infection.

The sequences of the capsid protein genes of all of the mutants were determined both prior to their inoculation onto plants and again after they had been introduced onto and replicated in tobacco. This allowed confirmation of amino acid sequences that were associated with the measured aphid transmissibilities. For all of the mutants except the D197K mutant, the introduced single amino acid changes were retained in a background of an otherwise wild-type capsid protein sequence. In the case of the D197K mutant, only one of four inoculation experiments with independent transcription reaction products gave rise to a systemic infection; in the other three experiments, no local or systemic infections were observed and synthesis of the capsid protein could not be detected by immunoassay. Positive-control experiments with the same RNA 1 and 2 transcripts but with a wild-type RNA 3 transcript resulted in infection. Sequencing of the coat protein gene from two independent cDNA clones of the virus derived from the one successful D197K mutant inoculation experiment revealed that the engineered amino acid change was retained but that a second-site mutation encoding a change of Ala to Glu at position 193 had been selected. This mutation was both observed in the initial screening of the virus population and confirmed subsequently in plants infected by aphid transmission. This second-site mutant is referred to as the D197K∗ mutant and is the only mutant that was unaffected in its aphid transmission (Table 2). It appears that a D197K mutation is debilitating and renders the virus unable to systemically infect tobacco. The fortuitous, compensatory second-site change of A193E restored infectivity, with wild-type symptoms and full aphid transmissibility.

DISCUSSION

The conservation of an amino acid sequence on the surface of a virus suggests an essential role beyond that of encapsidation. Results presented in this study are consistent with our hypothesis that the conserved CMV βH-βI loop structure plays a role in aphid vector transmission. While this loop forms a conspicuous, negatively charged electrostatic field on the surfaces of virions, the two other prominent surface domains, the βB-βC and βD-βE loops, make minimal charge contributions (46). In studies of spontaneous mutants defective in aphid transmission, amino acids in two other loop structures (βE-βF and βF-βG) were shown to impact transmission; these loops do not appear to alter the surface charge on virions but do affect virion stability (29, 46). It is useful to compare and contrast what is observed in CMV with what is seen in the genetically related CCMV. Although the capsid structure of CCMV is quite similar to that of CMV, CCMV is transmitted by beetles and not by aphids. As the capsid protein is a primary determinant for vector transmission, comparative analyses may provide insights into the underlying structural basis of vector specificity. In CCMV, the βH-βI loop structure is the same length but has fewer acidic residues (2 versus 5). Analogous to the arrangement in CMV, the two other prominent surface loop structures in CCMV do not contribute significantly to the surface charge density (47). Our working hypothesis is that a unique structure and charge density present on the surface of CMV facilitate the reversible binding of virions in the aphid vector.

The presentation of a putative vector binding domain on the surface of a virion must be fully integrated and compatible with other capsid protein functions such as RNA binding, encapsidation, cell-to-cell and systemic movement within the host, and any essential interactions with plant factors (4). Although six of the βH-βI loop mutations did not give rise to detectable phenotypic differences beyond those of their transmission defects, two of the mutants induced altered symptoms on tobacco and a third mutation interfered with replication and/or movement. Of particular interest is the D192K mutation, which induces necrosis. The induction of a similar type of necrosis has been reported for CMV capsid protein mutants with amino acid substitutions at position 129 (49). The necrotic phenotype is dependent on the specific amino acid replacement, as some substitutions at position 129 result in symptoms of chlorosis or a wild-type mottle-mosaic (44, 49). It is not known whether virions, the free capsid protein, or assembly intermediates are the entities responsible for the induction of necrosis or chlorosis. Interestingly, position 129 mutations also affect the aphid transmission phenotype, reducing or eliminating vector transmission (35, 36). With regard to the structure of CMV, amino acid position 129 and the βE-βF loop on which it is located face a depression surrounding the pentameric and hexameric capsomeres. The βE-βF loop is at or proximal to the surfaces of virions, and this region would be accessible to factors in the plant host or the aphid vector. In contrast, although amino acid position 192 is clearly on the surfaces of virions and accessible to factors in the host or vector, its effects on symptoms or transmission may be indirect. Position 192 is implicated in the binding of a metal cation (46), and mutations that affect metal binding may influence the structure or presentation of the βH-βI loop or affect subunit-subunit interactions. Capsid protein mutations that interfere with aphid transmission have also been shown to reduce the stability of virions (29). The stabilities of virions of the βH-βI loop mutants are comparable to that of the wild-type virus (J. C. Ng, C. Josefsson, and K. L. Perry, unpublished results). Therefore, we believe that the defect in aphid transmission is most likely due to an alteration of virion binding (or release) from sites in the aphid vector.

A second mutant with an altered symptom phenotype was the L194A mutant. The induced yellow chlorosis was similar to that described for CMV strains M (28) and Y (51), but it differed in the subsequent appearance of necrotic flecks within the chlorotic, mottled areas. In a subgroup II strain of CMV, a capsid protein change of K193N or K193S also radically altered symptoms (50); these changes are of interest because they are in the βH-βI loop, although the amino acid at position 193 is not conserved among CMVs. All three CMV RNAs encode symptom determinants (8, 21, 23, 32, 56), but the mechanisms by which viral proteins induce symptom expression are not understood. It is known that expression of the CMV capsid protein alone in a viral vector does not induce symptoms characteristic of the CMV infection (50).

A D197K mutation appeared to be lethal, as multiple attempts to infect plants with transcripts of the D197K mutant failed. It is not known whether this mutation affects replication, movement, or both. The only recoverable mutant with the D197K change was a virus (the D197K∗ mutant) with a compensatory second-site mutation located 4 amino acid residues away in the same loop structure. Compensatory second-site mutations were also described by Suzuki et al. (49) in their study of an engineered mutant (the Phe-129 mutant); this mutant induced necrotic local lesions but did not move systemically. Remarkably, in that study both the introduced and second-site mutations were in the same loop structure (βE-βF); this situation is analogous to the results described above for the D197K∗ mutant. The results from both of these studies suggest that the perturbation of a structural motif can be compensated for by a second, proximal amino acid change that restores a structure or charge density.

Consistent with a predicted alteration of a surface loop in the capsid protein and a reduction in the net negative charge on virions, the migration of mutant virions in agarose gels was retarded. It should be noted that the three D-to-A mutants (the D191A, D192A, and D197A mutants) did not migrate to the same position in the gel, suggesting that there may be conformational differences in the capsid protein of each mutant. A similar inference can be made to account for the faster migration of the L194A mutant relative to that of the wild-type virus. Three of the mutants with introduced lysine residues (the D191K, D192K, and E195K mutants) did not migrate in the gel; for each of these mutants, material stained by both ethidium bromide and Coomassie blue was observed in the well. This may have been due to aggregation; alternatively, the mutations may have conferred a net neutral charge upon virions. Wild-type-like virions of these lysine mutants were purified and observed by negative staining, indicating that a perturbation in virion formation was not responsible for the defect in transmission. The one feature that all of the mutants had in common was a predicted alteration in their surface charge, and this property appears to underlie the defect in vector transmission.

One issue that remains to be addressed is the extent to which amino acid residues in other surface loop structures can be modified and the extent to which they affect viability, symptomatology, and aphid vector transmission. Although it would not be surprising to find that the mutation of any surface residue gives rise to a phenotypic change, altering conserved amino acids (such as those found in the βH-βI loop) would be expected to have the most profound effects. From the present study, it is clear that disrupting amino acid residues in a conserved surface loop structure reduces or eliminates vector transmissibility without grossly affecting virion formation. The availability of these transmission-defective mutants will facilitate studies aimed at identifying ligands in the aphid vector and understanding molecular mechanisms underlying the vector transmission of a plant virus.