|Home | About | Journals | Submit | Contact Us | Français|
Point mutations were introduced into or near five conserved sequence motifs of the readthrough domain of the beet western yellows virus minor capsid protein P74. The mutant virus was tested for its ability to accumulate efficiently in agroinfected plants and to be transmitted by its aphid vector, Myzus persicae. The stability of the mutants in the agroinfected and aphid-infected plants was followed by sequence analysis of the progeny virus. Only the mutation Y201D was found to strongly inhibit virus accumulation in planta following agroinfection, but high accumulation levels were restored by reversion or pseudoreversion at this site. Four of the five mutants were poorly aphid transmissible, but in three cases successful transmission was restored by pseudoreversion or second-site mutations. The same second-site mutations in the nonconserved motif PVT(32-34) were shown to compensate for two distinct primary mutations (R24A and E59A/D60A), one on each side of the PVT sequence. In the latter case, a second-site mutation in the PVT motif restored the ability of the virus to move from the hemocoel through the accessory salivary gland following microinjection of mutant virus into the aphid hemocoel but did not permit virus movement across the epithelium separating the intestine from the hemocoel. Successful movement of the mutant virus across both barriers was accompanied by conversion of A59 to E or T, indicating that distinct features of the readthrough domain in this region operate at different stages of the transmission process.
Beet western yellows virus (BWYV) is a member of the Polerovirus genus of the Luteoviridae family (6). Luteoviruses are transmitted by aphids in a circulative nonpropagative manner and multiply in the phloem tissue of their hosts (22). The manner in which luteoviruses are acquired and transmitted by their specific vectors is complex. After uptake during feeding, the virus must first traverse the epithelial cell barrier separating the alimentary canal from the hemocoel, a process which involves receptor-mediated endocytosis and exocytosis (10, 11, 13). Once in the hemolymph, the virions diffuse until they encounter an accessory salivary gland (ASG). At this point, the virus must first penetrate the ASG basal lamina, which can act as a selective barrier to certain virus species (15, 25), and then undergo another round of receptor-mediated endocytosis and exocytosis to move into the salivary duct (13). Interactions between the virus capsid and aphid components at all of these sites may be important in conferring vector specificity (12, 14). Finally, in the hemocoel, binding of the virus to symbionin, a chaperonin secreted into the hemolymph by endosymbiotic bacteria of the aphid, is known to stabilize luteoviruses in a vector-nonspecific manner (9, 32, 33).
The genome of the poleroviruses consists of a monopartite plus-sense RNA of about 5.6 kb contained in isometric particles of 25-nm diameter (22, 23). Six major open reading frames (ORFs) are arranged on the genomic RNA into 5′-proximal (ORF0, ORF1, and ORF2) and 3′-proximal (ORF3, ORF4, and ORF5) gene clusters (see Fig. Fig.11 for a genetic map). The 3′-located genes are expressed from a subgenomic RNA (7, 29, 35). ORF3 encodes the major capsid protein of about 22 kDa. ORF4 is embedded in the capsid protein gene in another reading frame and encodes a 17- to 19-kDa protein. In potato leafroll virus, this protein has many of the properties expected for a viral movement protein (27, 29). BWYV P19, on the other hand, is not required for whole-plant infection of several hosts (37), suggesting that another BWYV protein or proteins may substitute for or act in parallel to P19 in supplying movement functions.
The 3′-proximal ORF 5 or readthrough domain (RTD) of BWYV, as in other poleroviruses, is translated as a 74-kDa fusion protein (known as P74 or RT protein) by episodic readthrough of the major coat protein termination codon (34). The RT protein is a minor component of the luteoviral capsid (1, 2, 20, 36). The RTD consists of several subdomains (22). A cytidine-rich sequence encoding a tract of 7 to 13 alternating proline residues in the RTD is located just downstream of the coat protein cistron-suppressible termination codon. This is followed by a region of about 200 amino acid residues with considerable sequence similarity throughout the Luteoviridae. The C-terminal half of the RTD is divergent. During virus purification, the RT protein of BWYV and other luteoviruses is cleaved to produce a C-terminally truncated form of about 53 kDa (1, 2, 36).
Mutagenesis of cloned full-length luteovirus cDNA has revealed that much of the C-terminal half of the RTD is dispensable for whole-plant infection and aphid transmission (2, 4). The conserved portion of the RTD, on the other hand, contains domains which are important for (i) aphid transmission of the virus (2, 4, 5), (ii) efficient accumulation of the virus in whole plants (4, 5, 24), and (iii) efficient suppression of major coat protein translation-termination (3, 4).
In this study we have produced a set of point mutants, mostly alanine substitutions, in the conserved part of the BWYV RTD in an attempt to better characterize determinants involved in aphid transmission. The mutants were assayed by infection of protoplasts with full-length transcripts and by agroinoculation of plants, followed by aphid transmission tests using either infected protoplast extracts, agroinfected plants, or purified virus as a virus source. The stability of the various mutations was assessed by sequence analysis of the mutant progeny in the agroinfected and aphid-infected plants. The results have allowed us to identify amino acids important for virus accumulation in planta and for transmission of BWYV by Myzus persicae Sulz. Our observations also reveal that successful aphid transmission of the RTD mutants is often accompanied by compensatory mutations elsewhere in the RTD, suggesting that the RTD possesses considerable structural redundancy.
All point mutants were created by PCR mutagenesis (16) using pBW0 as the template and oligonucleotides BW35 (nucleotides [nt] 4003 to 4022) and BW21 (complementary to nt 4829 to 4846) as external primers. The mutagenic primers were designed to substitute alanine residues at position 24 (mutant BW5.121; numbering refers to amino acids in the RTD), positions 59 and 60 (mutant BW5.123), positions 113 and 114 (BW5.125), and positions 225 and 226 (BW5.129). In mutant BW5.127, the amino acids K200 and Y201 were replaced by AD (Fig. (Fig.1).1). Full-length viral cDNAs containing the different mutations were reconstructed by cutting the PCR fragments bearing the mutations with BamHI and NcoI and substituting this fragment for the wild-type BamHI-NcoI (nt 4006 to 4822) fragment in pBW0. The 1.1-kb BamHI fragment (nt 2904 to 4006) was then introduced to create the final full-length mutant transcription vectors.
Protoplasts of Chenopodium quinoa were inoculated with viral RNA transcripts as described elsewhere (4). Mutant constructs for agroinfection (pBinBW5.121, pBinBW5.123, etc.) were made by replacing the SpeI-SalI fragment (extends from nt 1350 to 32 nt downstream of the insert 3′ terminus) of pBinBW0 by the SpeI-SalI fragment from the corresponding mutant transcription vector. The resulting plasmids were introduced into Agrobacterium tumefaciens LBA4404 for agroinfection (2, 18). Infected plants were identified by double-antibody sandwich enzyme-linked immunosorbent assay (DAS-ELISA) with a rabbit polyclonal antiserum raised against virus (4). Virus was purified as described by van den Heuvel et al. (31).
Transmission experiments used either detached leaves, crude extracts of infected protoplasts, or purified virus as an inoculum (4) and second- to fourth-instar M. persicae larvae as the vector. Membrane feeding of the larvae on crude protoplast extracts and on purified virus suspensions was performed as described elsewhere (4). For microinjection (4), the larvae were positioned with a micromanipulator and injected with 10 or 20 nl of purified virus (25 μg/ml), using 12- to 15-μm (outer diameter) glass capillaries attached to an Inject+ Matic microinjector (Gabay Instruments, Geneva, Switzerland).
BWYV RNA in total RNA extracts of infected plants and protoplasts was detected by Northern blotting using a 32P-labeled RNA probe complementary to the 3′-terminal 196 nt of the viral RNA (26). Viral structural proteins in total protein extracts of BWYV-infected protoplasts and plants and in purified virus were detected by Western blotting using antisera specific for BWYV coat protein and RT protein (26) following sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (2) except that urea was omitted from the gel loading buffer and immunodetection was performed with an enhanced chemiluminescence Western blotting kit (Amersham or Pierce).
The stability of the mutations in progeny viral RNA following agroinfection or aphid transmission was examined by amplifying a DNA fragment spanning the mutation site by reverse transcription followed by PCR (RT-PCR) as described elsewhere (2). Reverse transcription was primed with an oligonucleotide complementary to nt 5360 to 5376 (BW94), and a PCR product was synthesized with oligonucleotides BW35 and BW21 or BW35 and BW94. The PCR product was then cleaved with one of the following pairs of enzymes (numbers in parentheses refer to the position of each restriction site): BamHI(4006)/MscI(4544), MscI(4544)/NcoI(4822), BamHI(4006)/NcoI(4822), or BamHI(4006)/HindIII(5357), depending on the mutant to be analyzed. The appropriate fragments were cloned into a modified pBluescript cut with the same enzymes, and the insert sequences were determined for randomly selected clones with a 373 DNA sequencer (Applied Biosystems).
Sequence comparisons among members of the Luteoviridae reveal extensive amino acid sequence similarities within the N-terminal half of the RTD (22). Point mutations were generated in this region of the BWYV RTD at or near five positions that were conserved in all sequenced poleroviruses. In selecting the targets, preference was given to amino acids with charged side chains, as we reasoned that such residues are more likely to be exposed on the surface of the protein when it is folded into the native configuration. In the following discussion, amino acid substitutions within the RTD will be referred to by the following convention: XaZ, where X refers to the wild-type residue at position a of the RTD and Z refers to the substituted amino acid.
The positions of the different mutations in the RTD are shown in Fig. Fig.1.1. In mutant BW5.121, the strictly conserved residue R24, which is located just downstream of the alternating proline tract, was replaced by an alanine (R24A; codon CGT replaced by GCT). In BW5.123, the strictly conserved amino acids ED(59–60) were replaced by alanines (E59A/D60A; GAG.GAC replaced by GCG.GCC). In BW5.125, KD(113–114), amino acids which are not conserved in the other poleroviruses but which lie near the conserved motif G-IAY (residues 118 to 122), were replaced by alanines (K113A/D114A; AAG.GAC replaced by GCG.GCC). Mutant BW5.127 (K200A/Y201D; AAA.TAT replaced by GCA.GAT) targets residues KY(200–201), which are adjacent to and in the first position of the conserved motif YNY (residues 201 to 203). Finally, in BW5.129, the strictly conserved doublet DE(225–226), which marks the end of the conserved region of the RTD, was substituted by two alanines (D225A/E226A; GAT.GAA replaced by GCT.GCA).
When inoculated to C. quinoa protoplasts, viral transcripts containing the above point mutations directed synthesis of viral genomic and subgenomic RNA in amounts similar to those observed in protoplasts infected with wild-type BWYV transcript (data not shown). Proteins were extracted from the infected protoplasts and tested by Western blotting for the presence of the major coat protein (P22.5) and P74, using coat protein- and RTD-specific antibodies. The protoplasts infected with the RTD mutants were found to produce relative amounts of P22.5 and P74 similar to those observed in wild-type-infected protoplasts (Fig. (Fig.2),2), indicating that none of the point mutations had interfered with translation suppression of the P22.5 cistron's termination codon. The efficiency of packaging of the mutant viral RNAs into virions and their stability were not tested directly. However, it has been shown previously that deletion of the entire RTD does not interfere with packaging of mutant BW6.4 RNA into stable virions (26), making it unlikely that the point mutations in the RTD would have a dramatic effect.
Full-length viral cDNAs containing the various RTD mutations were moved into the binary vector pBin19 under control of the cauliflower mosaic virus 35S promoter, and the resulting constructs were introduced into A. tumefaciens. Nicotiana clevelandii plants were inoculated with recombinant agrobacteria harboring either the wild-type cDNA (BW0) or one of the five above-described mutants. The content of virus antigen was assayed by ELISA on tissue samples from upper noninoculated leaves of the agroinoculated plants (18 to 21 plants tested for each construct) at 5 weeks postinoculation (p.i.).
The average ELISA A405 values for the plants agroinfected with BW5.121 (0.92 ± 0.07), BW5.123 (0.95 ± 0.08), BW5.125 (0.95 ± 0.05), and BW5.129 (1.03 ± 0.07) were slightly lower than the average A405 value observed in a parallel experiment for BW0-infected plants (1.10 ± 0.07). BW5.127-infected plants, on the other hand, had a much lower ELISA titer (A405 = 0.55 ± 0.15). This is similar to the average A405 observed in parallel for plants agroinfected with mutant BW6.4 (0.53 ± 0.11), a mutant in which virtually the entire RTD is deleted (2) and which has been shown previously to accumulate about 10-fold less virus than the wild type (4). Measurements made 5 weeks later, however, revealed that the titer of viral antigen in the BW5.127-infected plants had risen significantly (0.93 ± 0.08), whereas the virus concentration in the BW6.4-infected plants remained low.
By 6 weeks p.i., most of the ELISA-positive plants infected with the RTD point mutants had developed typical interveinal yellowing symptoms. When total proteins were extracted from such plants and analyzed by Western blotting both P22.5 and P74 were readily detected (Fig. (Fig.3A).3A). In a previous study (4), short in-frame deletions in the conserved portion of the RTD were shown to interfere with incorporation of RT protein into viral particles. To determine if any of the RTD point mutants were similarly affected in RT protein packaging, we purified virus from the agroinfected plants and tested their protein contents by Western blotting. All mutant virions contained the C-terminally truncated form of RT protein of about 53 kDa (P53) in amounts comparable to that observed for the wild-type virus (Fig. (Fig.3B).3B).
The stability of the engineered RTD point mutations during virus multiplication in the agroinfected plants was investigated by RT-PCR. Total RNA was isolated from systemically infected leaves of two plants which had been agroinfected with each mutant, and a region of the genome encompassing the mutation was selectively amplified by RT-PCR. The amplified cDNA (nt 4006 to 4544 for BW5.121, BW5.123, and BW5.125; nt 4006 to 5327 for BW5.127; nt 4545 to 4822 for BW5.129) was then cloned, and inserts from randomly selected clones were sequenced.
All sequence alterations detected in the RT-PCR clones obtained from agroinfected N. clevelandii are summarized in Fig. Fig.4A.4A. In the cDNA clones, two types of nucleotide substitutions were observed: (i) those which changed an amino acid residue within or near the primary mutation and/or were present in several RT-PCR clones and (ii) those (either silent or resulting in an amino acid substitution) which were distant from the primary mutation site and/or were detected only one time. Mutations of the second type were observed in the RT-PCR progeny for all mutants except BW5.129. When averaged over the entire set of RT-PCR clones analyzed for the five mutants, these mutations were found to occur with a frequency of about one for every 1,800 nt sequenced, similar to the substitution frequency of one per 1,700 nt sequenced observed for RT-PCR clones derived from plants agroinfected with the wild-type construct BW0 (Fig. (Fig.4A).4A). Some of these background mutations no doubt reflect errors introduced during RT-PCR, while most if not all of the others probably represent neutral sequence variants which arise naturally in the virus population. For this reason we will limit the following discussion to mutations of the first type, whose appearance is more likely to have been favored by selective pressure exerted by the primary mutations.
For BW5.121, the original R24A mutation was maintained in each of the 12 RT-PCR clones characterized. In two clones, a second-site mutation, P32L or T34I, was present near the primary mutation site (Fig. (Fig.5A).5A). For brevity, we shall refer to the wild-type sequence PVT(32–34) to which these mutations map as the PVT patch.
The primary mutations E59A and D60A were conserved in all 10 RT-PCR clones analyzed, but four of them also contained a second-site mutation, P32L (Fig. (Fig.5A).5A). Interestingly, this is the same as one of the PVT patch substitutions observed in BW5.121, although in this case the PVT patch is 26 nt upstream of the primary mutation site at positions 59 and 60.
The primary mutation K113A/D114A was maintained in all 10 RT-PCR clones analyzed, and no significant second-site mutations were noted (Fig. (Fig.44A).
As shown above, plants agroinfected with BW5.127 initially accumulated low levels of virus, but near-wild-type levels appeared at later times. This observation suggests that the primary K200A/Y201D mutation had affected a motif in the RTD important for virus accumulation but that modified forms of the virus which overcome this defect appear later in infection. To test this hypothesis, viral RT-PCR clones were produced from RNA extracted from leaves taken either early (4 weeks) or later (7 weeks) following agroinfection. The sequence analysis revealed that both of the primary mutations (K200A and Y201D) were present in 20 of 21 RT-PCR clones from three different plants at 4 weeks p.i. (Fig. (Fig.6A).6A). In the single exception, D201 had been replaced by N in a clone derived from plant 9. We also detected the mutation S206L in two other clones from plant 9 at 4 weeks p.i. (Fig. (Fig.6A),6A), but the significance, if any, of this particular second-site substitution has not been further investigated.
In contrast, RT-PCR clones prepared from RNA extracted 7 weeks p.i. revealed numerous modifications at D201. Thus, of the 20 RT-PCR clones analyzed, D201 had reverted to the wild-type Y in 8 clones and had undergone a pseudoreversion to N in 9 (Fig. (Fig.6A).6A). The primary mutation D201 was retained in the remaining three clones. We conclude that the D201 substitution probably impedes efficient virus accumulation and that selection for revertants or pseudorevertants that overcome this defect is responsible for the enhanced virus accumulation at later times. The K200A substitution, on the other hand, is apparently neutral with respect to virus accumulation in planta.
All of the seven RT-PCR clones examined for BW5.129 retained the primary mutations D225A/E226A and no modifications were noted elsewhere in the region sequenced (Fig. (Fig.44A).
Extracts of transcript-infected protoplasts were used as a virus source in some transmission experiments. The aphids were allowed a 24-h acquisition access period (AAP) on the protoplast extract. The aphids were then transferred to healthy Montia perfoliata plants for a 4-day inoculation access period (IAP), and the plants were assayed for viral infection by ELISA 3 to 4 weeks later. In such experiments, only mutant BW5.125 was efficiently transmitted (Table (Table1).1). Mutant BW5.129 was weakly transmitted, infecting only 3 of 26 test plants following challenge with 30 aphids per test plant. No transmission events were recorded for BW5.121, BW5.123, or BW5.127 even though 100% transmission efficiencies were routinely observed in parallel tests with extracts of BW0-infected protoplasts (Table (Table1).1).
In other transmission experiments, young, fully expanded leaves of agroinfected N. clevelandii (4 to 6 weeks p.i.) or a solution of purified virus prepared from agroinfected plants was used as a virus source. Nonviruliferous M. persicae nymphs were allowed a 24-h AAP on the leaves or the purified virus solution. Then, 8 or 30 aphids were transferred to healthy M. perfoliata for a 4-day IAP, and the plants were assayed for infection as described above. The tests (Table (Table1)1) revealed that for both types of inoculum, mutants BW5.125 and BW5.129 were transmitted with efficiencies similar to that observed in parallel experiments with BW0. Mutants BW5.121 and BW5.127 were transmitted with intermediate efficiency. Transmission of BW5.123, on the other hand, occurred at a very low rate even though a high inoculation level (30 aphids/plant) was used; only 1 of 8 test plants became infected when BW5.123-agro-infected plants were used as the virus source, and only 2 of 34 plants became infected by aphids which had been given a high concentration (25 μg/ml) of purified virus (Table (Table11).
Experiments were carried out to determine if the lower transmissibility observed when the aphids were allowed to acquire BW5.121, BW5.123, and BW5.127 from leaves of agroinfected plants or by membrane feeding on purified virus was due to a defect in a step or steps of the mutant's circuit through the aphid subsequent to passage of the virions from the intestine into the hemocoel. To this end, 0.25 to 0.5 ng of purified wild-type or mutant virus was microinjected into the hemocoel of M. persicae nymphs. The nymphs were then placed on test plants (five aphids per plant) for a 4-day IAP, and infection was scored by ELISA 4 weeks later. BW5.121, BW5.123, and BW5.127 had transmission efficiencies similar to that of wild-type virus (Table (Table2),2), suggesting that the RTD mutations in question interfere with the initial step of the transmission circuit, i.e., movement of virions through the intestinal epithelial cell barrier.
An alternative explanation for the above observations is that the inefficient transmission of BW5.121, BW5.123, and BW5.127 following membrane feeding was not due to a defect in virus acquisition from the intestine but rather to lower stability of these mutant virions in the hemolymph, and that the high concentrations of virus introduced by microinjection masked this effect. Although it is difficult to strictly rule out this possibility at present, we regard it as unlikely because symbionin, which is believed to stabilize virions in the hemolymph (32, 33), has similar affinities for BW5.121, BW5.123, and BW5.127 as for wild-type virus in an in vitro binding assay (J. F. J. M. van den Heuvel, personal communication).
The foregoing experiments illustrate that except for BW5.125 (which is readily transmitted from both sources), transmission of the RTD mutants is more efficient following acquisition access to virus from agroinfected plants than it is when the virus is offered as an extract of transcript-infected protoplasts. A plausible explanation for this difference is that the mutant virus is defective in one or more steps of the transmission process but reversions or compensatory mutations which permit transmission have the time to appear during the 4- to 6-week duration of an agroinfection experiment. Furthermore, selective pressure for appearance of compensatory mutations in whole plants could also be exerted by a requirement for a functional RTD for efficient virus movement in planta.
To detect possible compensatory mutations, we sequenced progeny virus following successful aphid transmission of the RTD mutants when the inoculum was provided as purified virus from agroinfected N. clevelandii. The aphids were allowed to acquire the purified virus by membrane feeding or were rendered viruliferous by microinjection of the virus. The aphids were then given a 4-day IAP on M. perfoliata, and total RNA was extracted from infected leaves 4 weeks p.i. Viral cDNA sequences were amplified by RT-PCR, and regions encompassing each of the RTD mutations were cloned and sequenced as before.
Figure Figure4B4B summarizes of all sequence alterations detected in the RT-PCR clones obtained following aphid transmission. Again, we will consider only those nucleotide changes which provoke an amino acid substitution within or near the primary mutation site and which were observed in several RT-PCR clones. As observed for the progeny RT-PCR clones following agroinfection, such modifications (see below) were accompanied by background mutations scattered along the cloned cDNA. In the case of BW5.121, BW5.123, BW5.125, and BW5.127, these mutations occurred at frequencies similar to or below that observed among the RT-PCR clones obtained from plants infected with BW0 (data not shown). The frequency of accumulation of background mutations in BW5.129, however, was almost three times higher than for BW0, suggesting that for unknown reasons, the BW5.129 primary mutation promotes genetic instability. The sequence variants which we regard as significant for each mutant are discussed separately below.
The primary mutation (R24A) was strictly conserved in the aphid-infected plants, but proximal second-site amino acid substitutions were detected in all but one of the 14 RT-PCR clones examined (Fig. (Fig.5B).5B). These second-site mutations (P32L in seven clones; T34I in six clones) were the same PVT patch mutations encountered as minor components of the sequence population in the agroinfected plants (Fig. (Fig.5A).5A). We conclude that strong selective pressure for appearance of the PVT patch mutations is exerted during aphid transmission of the virus.
The foregoing observations led us to introduce one of the PVT patch mutations, P32L, directly into BW5.121 and into the wild-type virus by site-directed mutagenesis to produce BW5.121-P32L and BW0-P32L, respectively. Both mutants accumulated to near-wild-type levels in the systemic leaves of agroinoculated N. clevelandii and were readily transmitted to M. perfoliata by aphids that were allowed to feed on the agroinfected leaves (Table (Table3).3). The stability of the mutants in both the agroinfected plants and the target plants after aphid transmission was verified by sequence analysis of 10 or more RT-PCR clones of the progeny RNA. In every case, both the primary mutation (in BW5.121) and the P32L second-site modification were conserved, and no significant sequence variations were observed elsewhere in the region sequenced (data not shown). We conclude that the P32L second-site mutation favors aphid transmission in the BW5.121 background and does not interfere with virus accumulation or transmission in the wild-type background.
The situation with BW5.123 was more complex. Here, analysis of RT-PCR clones obtained from two plants inoculated with aphids which had been membrane-fed purified virus revealed that one of the primary mutations had undergone a modification: the first A in the mutant sequence AA(59–60) had been replaced by T in plant 2 or had reverted to the wild-type residue E59 in plant 1 (Fig. (Fig.5B).5B). To determine if the partial reversion EA(59–60) found in plant 1 restored full transmissibility of the virus, this plant was used as virus source for a second round of aphid transmission. Transmission to eight new M. perfoliata plants occurred with 100% efficiency with both 8 and 30 aphids per test plant. The progeny viral RNA was extracted from two of these plants, and nine RT-PCR clones were sequenced. EA(59–60) was present in all progeny clones, and no mutations elsewhere were noted (data not shown).
Successful transmission of BW5.123 following microinjection of aphids with virus was associated with a different set of compensatory mutations. RT-PCR clones obtained from two such plants retained the original E59A/D60A mutation but exhibited the same PVT patch second-site mutations (P32L or T34I) observed in the BW5.121 progeny (Fig. (Fig.5B).5B). The fact that these mutations were encountered only in plants infected by microinjected aphids indicates that at least in the BW5.123 background, the sequence signals permitting virus movement through the ASG are distinct from those which allow transit from the intestine into the hemocoel. To test this point directly, the double mutant BW5.123-P32L was constructed and agroinoculated to N. clevelandii. Virus accumulated to near-wild-type levels in the plants but could not be aphid transmitted to M. perfoliata (Table (Table3).3). Aphids rendered viruliferous with BW5.123-P32L by microinjection, on the other hand, were efficient virus transmitters (Table (Table3).3). We conclude that the A59E reversion and, probably, the A59T pseudoreversion permit movement of the mutant virus across both the intestine/hemocoel interface and through the ASG but that at least in the BW5.123 background, the PVT patch mutation operates only at the second barrier.
The primary mutation (K113A/D114A) was conserved in each of the 12 clones analyzed, and no second-site amino acid mutations were noted elsewhere in the region sequenced (Fig. (Fig.4B).4B). We conclude that the particular motif targeted in BW5.125 is not required for efficient aphid acquisition or transmission.
For mutant BW5.127 (K200A/Y201D), sequence analysis of progeny RT-PCR products following transmission revealed that the K200A primary mutation was conserved in 14 of the 18 clones. The four exceptions, with a V at position 200, all derived from the same plant (Fig. (Fig.6B).6B). We conclude that K200 is dispensable for aphid transmission. At position 201, on the other hand, an amino acid with a phenolic side chain (F or Y) was present in all but one of the progeny clones. The fact that only 1 of 18 RT-PCR clones examined after aphid transmission contained N201 even though this substitution was relatively abundant in BW5.127-agroinfected plants (Fig. (Fig.6A)6A) suggests that selective pressure against retention of the N201 variant is applied during transmission.
Following aphid transmission, pseudorevertants with an F at position 201 were abundant in the progeny even though this substitution was not detected among the RT-PCR clones obtained from plants agroinfected with BW5.127 (Fig. (Fig.6A).6A). To test the effect of F201 directly, the substitution was introduced into the BW5.127 background to yield BW5.127-D201F. The mutant multiplied to near-wild-type levels following agroinoculation and was efficiently aphid transmitted (Table (Table3).3). Sequence analysis of 16 RT-PCR clones obtained following aphid transmission revealed that the D201F mutation was conserved in every case. These observations suggest that the low level of the F201 variant observed following agroinfection with BW5.127 does not reflect unfitness but rather indicates that conversion of D201 to F requires two nucleotide transversions (GAU to UUU) while conversion of D201 to Y or to N requires only one transversion or transition, respectively. Indeed, the finding that the F201 pseudorevertant is rather efficiently selected during aphid transmission from a state of low abundance in the BW5.127-agroinfected plants suggests that at least in the BW5.127 background, it has a selective advantage over the true revertant, Y201.
It was shown above that BW5.129 was transmitted poorly when the virus was supplied to the aphids as an extract of transcript-infected protoplasts but that transmission was efficient when aphids were allowed to acquire virus from agroinfected leaves or by membrane feeding of purified virus (Table (Table1).1). We anticipated that the efficient transmission observed in the latter situation would be associated with reversion or compensatory mutations in the BW5.129 genome. However, when the progeny RNA following successful transmission was characterized by RT-PCR, the original D225A/E226A mutations were retained in each of the 11 clones examined (Fig. (Fig.4B).4B). Furthermore, although several second-site amino acid substitutions or silent mutations were observed in eight of the progeny RT-PCR clones, none were present more than once except for a silent mutation replacing T(4472) by C, which was detected twice (Fig. (Fig.4).4). Thus, although RTD second-site mutations were relatively frequent in the aphid-transmitted progeny of BW5.129, there was no obvious candidate for a compensatory mutation which could restore efficient transmissibility, as was observed after transmission of BW5.121 and BW5.123.
In this report we have shown that introduction of point mutations in the conserved portion of the BWYV RTD often leads to the appearance of second-site mutations or pseudoreversions which restore wild-type or near-wild-type function to the virus. This observation suggests that there is considerable structural redundancy in the RTD; that is, the overall folding pattern can tolerate modifications even when highly conserved residues are targeted. It is evident that analysis of such compensatory mutations may help identify motifs in the RTD which are implicated in virus accumulation in planta and in aphid transmission.
The appearance of compensatory mutations in our experiments is certainly related to the relatively high mutation rate characteristic of RNA viruses (8). Even though the inoculum used to agroinfect plants was derived from a cDNA clone and was hence perfectly uniform at the start, sequence variants which are neutral or only slightly deleterious can accumulate during the 4- to 6-week period of virus multiplication in plants. When point mutations are introduced intentionally into the RTD, several scenarios can be envisaged. (i) If the mutation is neutral, the mutant virus will be expected to persist and have properties similar to those of the wild type. This is apparently the case with BW5.125. (ii) If the mutation interferes with viral movement in planta, then the virus will accumulate to low levels following agroinoculation unless reversion or a compensatory mutation restores the movement function. Mutant BW5.127 belongs to this class, at least with respect to the Y201D substitution. (iii) If the mutation does not greatly affect virus accumulation in planta but interferes with aphid transmission of the virus, then transmission efficiency will be reduced and viral genomes carrying reversions or compensatory mutations will appear in the progeny after successful transmission events. BW5.121 and BW5.123 both belong to this category.
At present, BW5.129 cannot be classified. The mutant virus multiplied efficiently in planta following agroinfection. The primary mutations were strictly retained in the progeny virus in the agroinfected plants, and no notable second-site mutations were observed. The primary mutations were also retained following aphid transmission of virus from agroinfected plants, but the high transmission efficiency, compared to the inefficient transmission of virus from infected protoplast extracts, suggested that a compensatory mutation would be present in the progeny. However, although the progeny virus population contained a number of second-site amino acid substitutions and silent mutations in the RTD, there was no obvious pattern in their distribution pointing to a particular second-site mutation as being important. Possibly, there are several distinct second-site mutations in the RTD which can act in concert to compensate for the primary BW5.129 mutations. Alternatively, the compensatory second-site mutations may lie elsewhere in the genome. Further experiments will be necessary to distinguish between these possibilities.
One noteworthy feature of our experiments is that second-site mutations and pseudoreversions were frequently observed but that true reversions were uncommon, being detected only at Y201 for some of the BW5.127 progeny. The low frequency of true revertants is probably related to the fact that the amino acid substitutions introduced into the mutants always involved at least two nucleotide substitutions, generally transversions. The various second-site amino acid substitutions or pseudoreversions which restored function, on the other hand, generally involved a single nucleotide change, and this was almost always a transition. For example, the P32L and T34I second-site mutations in the PVT patch both arose by a T-for-C substitution in the codon. For a number of RNA viruses, it has been demonstrated that transition mutations occur at higher frequency than transversions (17, 19, 28). If the BWYV replicase displays a similar bias in transition/transversion frequency, transitions will be expected to predominate in the pool of potential second-site mutations.
Another unanticipated finding was the ability of the same second-site mutations in the PVT patch (P32I and T34L) to compensate for two distinct primary mutations (R24A in BW5.121 and E59A/D60A in BW5.123) at sites separated by 34 residues. Possibly, R24 and ED(59–60) are brought into proximity during folding of the wild-type RTD, and the PVT patch mutations permit subtle structural adjustments which restore function in the presence of either mutation. It is also possible that the PVT patch mutations act as global stabilizers, i.e., as mutations which augment the stability of the protein (21). If present as a second-site mutation, a global stabilizer may compensate for a primary mutation whose principal effect is to destabilize the protein. It appears unlikely, however, that the PVT patch mutations can function solely as global stabilizers, as it is difficult to imagine how a global stabilizer could act in a differential manner at different steps of the acquisition/transmission process, as observed for BW5.123-P32L. Evidently, although identification of second-site compensatory changes can provide preliminary information about structure-function relationships, knowledge of the three-dimensional structure of the RTD will be essential for an in-depth interpretation of the data.
Finally, the double mutant BW5.123-P32L, which is apparently unable to cross from the intestine into the hemocoel, could provide a useful probe for dissecting the acquisition process. Acquisition of BWYV involves receptor-mediated endocytosis of virus particles at the apical plasmalemma of intestinal epithelial cells, unidirectional intracellular transport through these cells in vesicles, and exocytosis into the hemocoel (10, 11). Electron microscopic observations of thin sections of intestinal epithelia of M. persicae which have been membrane fed with this mutant may tell us at what stage of this process the virus is blocked and so provide information about receptor localization.
We thank Philippe Hammann for help with nucleotide sequence analysis.