Point mutations in the potato leafroll virus major capsid protein alter virion stability and aphid transmission

The coat proteins (CPs) of many plant viruses are multifunctional, and are involved in a variety of interactions besides the formation of virus particles. These include virus replication, virus movement between plant cells, interactions with biological vectors that result in virus spread from infected to uninfected plants, and interactions with the host plant that result in susceptibility, disease symptoms or resistance (reviewed by Callaway et al., 2001). The CP plays an essential role in viral movement throughout an infected plant, either in systemic movement or in both cell-to-cell and systemic movement (reviewed by Waigmann et al., 2004). Some plant viruses have adapted their CPs for transmission by various vectors, including arthropods, nematodes and fungi (Gray & Rochon, 1999), and the CPs are important determinants of vector specificity.

Potato leafroll virus (PLRV) is a member of the genus Polerovirus within the family Luteoviridae (Mayo & D'Arcy, 1999). Some of the unique characteristics of viruses within this family include obligate transmission by aphids in a circulative, non-propagative manner and restriction of viral infection to the phloem (Harrison, 1999). None of these viruses are mechanically transmitted. Since the accumulation and spread of PLRV are limited mainly to the phloem cells, this experimental system allows the study of phloem-associated long-distance movement without the necessity for short-distance, cell-to-cell movement through the mesophyll and parenchyma tissues.

The PLRV genome consists of a 5.8 kb, monopartite, single-stranded, positive-sense RNA that is encapsidated in isometric particles. The PLRV genome contains eight open reading frames (ORFs) that are divided into two parts, separated by a non-coding region. Three 5'-proximal ORFs, which are expressed from the genomic RNA, encode the proteins involved in virus replication and suppression of gene silencing (Pfeffer et al., 2002). Five other ORFs are expressed by translation from two subgenomic RNAs (sgRNAs). Two structural proteins, the CP and the readthrough domain (RTD), and a 17 kDa movement protein (P17) (Miller et al., 1995), are encoded by sgRNA1. sgRNA2 encodes two 3'-proximal proteins for which the functions are not known (Ashoub et al., 1998). We have recently shown that P17 is required for virus movement in potato or Physalis floridana, but is not essential for virus movement in Nicotiana benthamiana or Nicotiana clevelandii (Lee et al., 2002).

Wild-type (WT) polerovirus capsids contain the major CP and a less abundant minor protein, the readthrough protein (RTP), that is a fusion of the CP and RTD, translated by suppression of the CP stop codon (Bahner et al., 1990). Assembled virions are required for systemic movement in plant hosts, but virions can be assembled from CP only, although these RTD-deficient particles are not aphid-transmissible (Brault et al., 2000, 1995; Bruyere et al., 1997) and do not accumulate as efficiently in systemically infected tissues of host plants (Brault et al., 1995; Bruyere et al., 1997). Some evidence suggests that the CP alone can mediate virus particle passage through the aphid gut membrane, but not through the aphid accessory salivary gland (Chay et al., 1996; Gildow et al., 2000; Reinbold et al., 2001). Other results suggest that pseudoparticles of PLRV, in which the capsid consists only of CP, are able to achieve the complete route of the virus within the vector (Gildow et al., 2000). However, RTP association with the virion is required for aphids to transmit the virus to new host plants. The RTD may function to regulate the efficiency of virus movement across the gut and salivary tissues, rather than be an absolute requirement (Reinbold et al., 2001). The RTD may also be involved in avoidance of the aphid immune response and may regulate the stability or accumulation of virus in plant tissues (van den Heuvel et al., 1999).

There are no crystallographic data available for virions of any members of the family Luteoviridae, but the general shape of the particle is icosahedral (Kojima et al., 1969). Interestingly, icosahedral RNA plant virus CPs generally do not have significant sequence similarities, but there is conservation of secondary and tertiary structure, especially within the shell (S) domain of the CP (Dolja & Koonin, 1991). There is also a conserved arginine-rich (R) domain at the N terminus of many of these CPs that presumably interacts with the viral RNA. The conserved structure of the S domain among small, icosahedral RNA plant viruses is an eight-stranded β-barrel that forms the jelly-roll structure (Harrison et al., 1978). Alignment of amino acid sequences of CPs of luteoviruses and poleroviruses have identified strongly conserved regions within the S domain (Mayo & Ziegler-Graff, 1996). Based on these conserved structures, several models of the polerovirus CP structure have been developed (Brault et al., 2003; Lee et al., 2005; Terradot et al., 2001), and some biological data have been generated to validate the predicted structures (Brault et al., 2003; Lee et al., 2005). It appears that the basic structural features are conserved among the CPs of different poleroviruses, but that biologically active domains and their structural relationships with each other may be organized differently (Lee et al., 2005).

We report mutational analyses of the PLRV CP and characterize some additional molecular determinants involved in particle assembly/stability, virus movement in different plant hosts and aphid transmission. These results are then assessed with regard to the existing structural model for PLRV CP.

Recombinant plasmids.
Plasmids pBNUP110 and pUC.PLRV were described previously; both contain a full-length cDNA copy of PLRV (Canadian isolate) between the cauliflower mosaic virus 35S RNA promoter and the corresponding transcription termination signal sequence (Franco-Lara et al., 1999; Nurkiyanova et al., 2000). The pUC.PLRV plasmid contains the pUC118 backbone and was used as an intermediate for ease of cloning (high copy number), and the pBNUP110 plasmid contains the pBIN19 backbone, which is the disarmed Ti plasmid used for agroinoculation (Frisch et al., 1995).

Generation of recombinant cDNA PLRV CP constructs.
Site-directed mutagenesis was accomplished by overlapping PCR and ligation with restriction enzymes. Overlapping PCR used pUC.PLRV as the template and the following primer pairs, (i) PLRV5'p3438 (corresponding to PLRV nt 34383461) and the reverse primer which is complementary to the PLRV sequence except for the introduced mutation, and (ii) PLRTH1R (complementary to PLRV nt 48694897) and the forward primer which corresponds to the PLRV sequence except for the introduced mutation. Primer sequences containing the various mutations are available upon request. A PmlIBstBI fragment corresponding to PLRV nt 35724636 was digested from the PCR product and used to replace the similar PmlIBstBI fragment of pUC.PLRV. After the region corresponding to the PLRV CP ORF (nt 34614248) was sequenced, the PmlIXhoI fragment (nt 35725482 in the PLRV RNA sequence) containing the altered CP sequence in pUC.PLRV was used to replace the PmlIXhoI fragment of pBNUP110. An additional mutant, ΔRTD (Liang et al., 2004), was used as a control in some experiments. The translation of this mutant results in the generation of a PLRV that produces WT CP, but no RTP.

Particle assembly/stability assay.
Agroinfiltrated leaf tissue was used for particle assembly/stability, RNase-sensitivity and structural protein expression analyses. Agrobacterium tumefaciens (strain LBA4404) carrying pBNUP110 (WT) or PLRV CP mutant constructs was grown in YEB (0.5 % Difco nutrient broth, 0.5 % peptone, 0.1 % yeast extract, 0.5 % sucrose, 2 mM MgSO₄, pH 7.2) broth with 50 µg kanamycin ml¹ for 48 h. The cultures were centrifuged and resuspended to 0.30.4 OD₆₀₀ in water and used for agroinfiltration into the expanded leaves of N. benthamiana at the nine to ten leaves stage. Five to six days post-infiltration, tissue was collected for the following analyses.

Particle assembly and stability were analysed by several methods including ELISA, virus purification and RNase-protection assays. Infiltrated tissue (0.10.2 g) was homogenized in 500 µl PBS (pH 7.2) and tested by double-antibody sandwich ELISA (DAS-ELISA) and triple-antibody sandwich ELISA (TAS-ELISA) (Lee et al., 2005). Unassembled capsid proteins are not detected by DAS-ELISA, but unassembled viral protein complexes are detected, albeit inefficiently, by TAS-ELISA using mAb SCR3 (Lee et al., 2005). Further examination of particle stability was assayed by virus purification and immunosorbent electron microscopy. Stable, assembled virus particles are defined as those that are able to survive the harsh conditions of purification and be recovered from sucrose gradients. A more direct measure of particle stability utilized an RNase-protection assay. Agroinfiltrated tissue (0.10.2 g) was frozen in liquid nitrogen, ground in 200 µl PIPES buffer (50 mM PIPES, pH 6.5; 0.1 % Tween), incubated at 37 °C for 30 min in endogenous nucleases. Total RNA was then extracted using an RNeasy Plant Mini kit (Qiagen). An untreated sample of the same agroinfiltrated tissue was ground after freezing in liquid nitrogen and total RNA was immediately extracted. The structural integrity of the viral RNA was determined by amplification of a 965 bp fragment encoding the CP and partial RTD genes. RT-PCR analysis used primer PLRV5'p3438 (corresponding to PLRV nt 34383461) and PLRV3'p4382 (complementary to PLRV nt 43824402) and a One Step RT-PCR kit (Invitrogen). The RT-PCR parameters were: reverse transcription at 53 °C for 45 min, followed by 95 °C for 2 min and 20 cycles of 95 °C for 30 s, 53 °C for 30 s and 72 °C for 1.5 min, followed by 72 °C for 10 min. Products were visualized on ethidium bromide-stained agarose gels.

Structural protein expression analysis.
Agroinfiltrated tissue (0.10.2 g, eight discs using a no. 5 cork borer) was disrupted in 250 µl 2x SDS/sample buffer (Sambrook et al., 1989) and heated to 95 °C for 5 min. Samples were then separated by electrophoresis in SDS-12.5 % polyacrylamide gels (Sambrook et al., 1989). Proteins were transferred to nitrocellulose membrane Nitrobind 0.2 µm (GE Osmonics) using a Trans-Blot Semi-Dry Cell (Bio-Rad), and blots were incubated with mAb SCR3 (1 : 2500), which recognizes a linear epitope on the N terminus of the CP and RTP (Torrance, 1992). Binding of the mAb SCR3 was detected with goat anti-mouse antibody conjugated to alkaline phosphatase (1 : 5000; Sigma-Aldrich) followed by addition of the chromogenic substrate 1-Step NBT/BCIP (Pierce Biotechnology).

Agrobacterium-mediated infection of plants.
A. tumefaciens (strain LBA4404) carrying pBNUP110 (WT) or PLRV CP mutant construct was grown in YEB broth with 50 µg kanamycin ml¹ for 48 h. The culture was centrifuged and resuspended in 1/10 volume of water and used for agroinfection of N. benthamiana and N. clevelandii plants at the five to six leaves stage (Lee et al., 2002; Nurkiyanova et al., 2000). Approximately 40 µl Agrobacterium suspension was injected using a Hamilton syringe into the midrib of each leaf (three in total). Inoculated plants were placed at 20 °C for 7 days prior to being placed in the greenhouse.

The accumulation of PLRV was assessed by detection of viral CP using DAS-ELISA (Agdia) (Lee et al., 2002). Newly emerged, uninoculated leaves from agroinfected plants were sampled after 3 weeks. To determine whether the mutations were retained by the virus, viral RNA was analysed by RT-PCR and direct sequencing of the PCR products. RT-PCR of the CP gene fragment was described above. The fragments amplified were sequenced using an automated 3700 DNA analyser with BigDye Terminator chemistry and AmpliTaq-FS DNA polymerase (Applied Biosystems).

Aphid transmission assays.
Systemically infected N. benthamiana or N. clevelandii plants were used as a virus source in aphid transmission tests. Non-viruliferous Myzus persicae nymphs were allowed to feed on detached leaves of N. benthamiana or N. clevelandii for 24 and 48 h, respectively. The aphids were transferred to healthy plants (10 aphids per plant) for 72 h inoculation access periods and then killed by fumigation. The accumulation of PLRV was assessed by detection of viral CP using DAS-ELISA (Agdia) 3 weeks post-inoculation. The progeny viruses from systemically infected plants were analysed by RT-PCR and direct sequencing of the PCR products.

Generation of PLRV structural models.
The procedure to generate the three-dimensional (3D) models for the PLRV CP is described in detail in an earlier publication (Lee et al., 2005). Briefly, predictions from different threading servers (Bujnicki et al., 2001; Kelley et al., 2000) were used to obtain a set of templates and initial sequence alignments that were used as input for MODELLER (Sali & Blundell, 1993), a program used in comparative protein modelling that generates 3D coordinates through satisfaction of spatial restraints. Generation of the final model was carried out by going through a repetitive process in which current 3D models were compared visually with the structures of the templates. This led to the identification of improperly aligned regions in the target sequence. New trial alignments were produced and the 3D models were updated until no obvious misalignment was detected.

Strategy for mutational analysis of PLRV CP
The 13 mutants described in this study were generated prior to the development of the various 3D polerovirus CP structural models. Targets for mutation were selected based on the alignments of CP sequences of poleroviruses and luteoviruses, as well as the CP predicted secondary structures (Mayo & Ziegler-Graff, 1996). Targets included amino acids that are highly conserved among known polerovirus CP sequences, mapped epitopes (Torrance, 1992) and amino acids predicted to be on the surface of the virus particle (Mayo & Ziegler-Graff, 1996) (Fig. 1; Table 1). The CP ORF partially overlaps ORF 4 that encodes the P17 movement protein; therefore many of the mutations in CP also resulted in amino acid changes in P17 (Fig. 1, Table 1). We have previously shown that deletion of the P17 movement protein did not result in any observable phenotype different from a WT infection in N. benthamiana or N. clevelandii (Lee et al., 2002).

(33K): Fig. 1. Genome organization of potato leafroll virus (PLRV) and linear amino acid sequence of the PLRV coat protein (CP) and the overlapping P17 movement protein (17K). The position and composition of the CP mutants are identified as shaded text. The altered amino acids are in bold type. If the mutations in the CP sequence also altered the amino acid sequence of P17, the WT amino acid is underlined and bold. The mutant name used in the text is given above each shaded area.

Table 1. Potato leafroll virus coat protein (CP) mutant constructs and their predicted structural attributes

PLRV structural protein expression
The accumulation of both CP and RTP varied among experiments and agroinfiltrated N. benthamiana tissue samples, but the levels of CP were not consistently different from WT virus for any of the CP mutants (Fig. 2a). However, detection, accumulation and size of the RTP differed among several mutants. Several distinct sizes of RTP were observed from tissue infiltrated with WT virus, and these protein species were consistent among experiments (Fig. 2a, lane WT). These appear to be proteolytic cleavage products from the C terminus of the RTP. The proteins were detected using the monoclonal antibody SCR3 that recognizes an epitope on the N terminus of the CP sequence. Purified preparations of poleroviruses and luteoviruses usually contain a truncated RTP of approximately 5055 kDa and the full-length RTP is often not observed. A similar set of proteolytic cleavage products seen in tissue infected with WT virus was consistently observed for mutants S-R, GG, D-P-K, HCK and TK (Fig. 2a). For several other mutants including FT, GP, LH, ELD and GNG, full-length RTP was rarely observed, and there was a consistent accumulation of an approximately 40 kDa protein (Fig. 2a). We believe this 40 kDa protein to be a CP dimer rather than a cleavage or degradation product of the RTP, similar to that described by Filichkin et al. (1994). The 40 kDa protein described here was detected by a monoclonal antibody that recognizes an epitope located at the N terminus of the CP. The protein was not recognized by an antibody made against a 20 aa sequence in the N-terminal region of the RTD. This RTD-specific antibody reacts weakly with full-length RTP and strongly against truncated RTP associated with purified virus, but not with CP (data not shown). We believe that the conformation of the full-length RTP is not optimal for binding of this RTD-specific antibody. Variable levels of full-length RTP were observed from tissue infiltrated with mutants SE and TIR and, in samples with reduced levels of full-length RTP, the levels of the 40 kDa protein increased. A similar phenomenon was reported for several other PLRV CP mutants (Lee et al., 2005). Interestingly, the original TK mutant (TK-1) was found to contain an additional nucleotide substitution that resulted in a K to R substitution at position 76. When the originally intended TK mutation (TK-2) was tested, it did not accumulate RTP, but there was an accumulation of the 40 kDa protein (Fig. 3).

(77K): Fig. 2. PLRV structural proteins and RNA detected in plant tissue infected with various CP mutants. (a) Western blot analysis of PLRV proteins in agroinfiltrated N. benthamiana tissue detected by a monoclonal antibody, SCR3, which detects an epitope on the N terminus of the CP. The position of the CP, full-length RTP and an approximately 40 kDa protein are identified. (b, c) RT-PCR amplification of a 965 bp product encoding the CP and partial RTP genes from RNA extracted from tissue agroinfiltrated with CP mutants following incubation (b) with endogenous RNases or (c) without exposure to RNase.

(86K): Fig. 3. Western blot analysis of PLRV proteins detected by a monoclonal antibody, SCR3, that detects an epitope on the N terminus of the CP. N. benthamiana tissue was agroinfiltrated with various PLRV CP mutants. The position of the CP, full-length RTP and an approximately 40 kDa protein are identified. MM, Molecular mass markers, values are in kDa.

cis- and trans-acting RNA sequences located in ORF5 have been shown to affect efficient translation of the RTP (Brown et al., 1996), but this is the first report that alterations in the CP sequence have an effect on the translation of the RTP. Virus particles will assemble in the absence of RTP (Filichkin et al., 1994; Reutenauer et al., 1993), but they are not transmissible by aphids and they are slower to accumulate in systemically infected plant tissues (Brault et al., 1995; Chay et al., 1996). The RTP-lacking virions may preferentially assemble from CP dimers, similarly to other icosahedral plant viruses (Fox et al., 1994). Specific amino acids or resulting tertiary structures in the CP may suppress dimer formation and allow RTP translation, and thereby promote the assembly of RTP-containing virions that are transmissible and better able to systemically infect plant hosts.

Particle assembly/stability
The assembly of virus particles was initially determined by detection of virus from agroinfiltrated N. benthamiana tissue using DAS-ELISA and TAS-ELISA (Lee et al., 2005). WT PLRV and five CP mutants (S-R, GG, D-P-K, HCK, TK-1) were all detected at similar absorbance levels by both ELISA formats (data not shown), suggesting that particles were being assembled. Absorbance values were not above background levels for any of the eight other CP mutants (FT, GP, L-H, SE, ELD, TK-2, GNG and TIR) in either DAS-ELISA or TAS-ELISA (data not shown), suggesting that no stable virions were formed. We attempted to purify several of the viruses from agroinfiltrated tissue. Typical virus peaks were observed from sucrose gradients for WT, ΔRTP (Fig. 4a) and D-P-K (not shown). Only a small peak was observed for TK-1 and no peak was observed for SEM (Fig. 4a). Samples from the peaks or from the sucrose gradient fractions where virus was expected to be eluted were observed using transmission electron microscopy. High concentrations of expected virus-like particles were observed for WT, ΔRT (Fig. 4b) and D-P-K (data not shown), only a few particles were observed for TK-1 (Fig. 4b) and no particles were observed for SEM (Fig. 4b). The ΔRT and D-P-K particles observed were indistinguishable from WT particles. Although the ELISA data suggested that virions of TK-1 were being assembled, the purification results suggest that the particles were not stable.

(136K): Fig. 4. (a) UV absorbance scans of PLRV preparations separated on 1040 % sucrose gradients. Arrows indicate the position or expected position of the purified virus peak. (b) Immunosorbent electron micrographs of purified virus particles trapped using an anti-PLRV polyclonal antibody.

An RNase-protection assay was used to examine further whether any of the CP mutations interfered with assembly of viral RNA into virions. Assembly of RNase-resistant PLRV virions resulted in amplification of PLRV genomic RNA, whereas poorly assembled or unstable virions resulted in degradation of viral RNA by plant endogenous RNases and little or no amplification of genomic RNA. WT PLRV and four CP mutants (S-R, GG, D-P-K and HCK) packaged viral RNA into virions and their RNAs were detected in the RNase-protection assay (Fig. 2b, c). The RNA of the correct TK mutant (TK-2) was not protected (data not shown), yet the RNA of the TK-1 mutant was partially protected (Fig. 2b, c). The RNAs from the other CP mutants (FT, GP, L-H, SE, ELD, GNG and TIR) were not protected in the RNase-sensitivity assay and thus did not appear to assemble properly or form stable particles. These data support the conclusions derived from the ELISA and purification data (Table 2).

Table 2. Detection of PLRV CP mutants and their ability to systemically infect multiple hosts following agroinoculation or aphid inoculation

Systemic infection
We have previously shown that the PLRV P17 movement protein is not essential for systemic infection of N. benthamiana or N. clevelandii, but is required for the systemic infection of potato or P. floridana (Lee et al., 2002). Thus, the ability of these PLRV CP mutants to infect host species systemically was assessed initially in two plant species, N. benthamiana and N. clevelandii, by agroinoculation.

In both N. benthamiana and N. clevelandii, WT PLRV and four CP mutants (S-R, GG, D-P-K and HCK) accumulated systemically (Table 2). Two mutants (S-R and GG) are mutated within the R domain of the PLRV CP, and two mutants (D-P-K and HCK) are mutated within α-helices. RT-PCR and subsequent direct sequencing of the PCR products confirmed that the progeny viruses that systemically infected these plants retained their mutated sequences and had not reverted back to WT PLRV. All of the other CP mutants (FT, GP, L-H, SE, ELD, TK-2, GNG and TIR) were unable to infect systemically either N. benthamiana or N. clevelandii (Table 2).

Interestingly, the results of agroinoculation on two other hosts, Solanum tuberosum and P. floridana, are consistent with the results found for systemic infection of N. benthamiana and N. clevelandii (Table 2). Therefore, the changes in the P17 movement protein that would occur due to the mutations in the CP did not have an effect on the systemic movement in these hosts that do require a functional P17 (Lee et al., 2002). These results also confirm previous results (Ziegler-Graff et al., 1996) that stable virions are required for systemic movement of poleroviruses in Nicotiana sp., and extend the findings to indicate that this is not a host-specific phenomenon.

Aphid transmission
The four PLRV CP mutants that assembled stable virions and systemically infected hosts were assayed for their ability to be transmitted by M. persicae from plants systemically infected by agroinoculation with the CP mutants. Non-viruliferous M. persicae nymphs were used in experiments to determine whether the mutants were affected in transmission from either N. clevelandii to N. clevelandii, N. clevelandii to P. floridana or P. floridana to P. floridana.

WT PLRV and four of the CP mutants (S-R, GG, D-P-K and HCK) were transmitted by aphids, although with varying efficiencies (Table 2). The progeny viruses were analysed by RT-PCR and subsequent direct sequencing of the PCR products, confirming that the progeny viruses that were transmitted by the aphids and systemically infected these plants retained their mutated sequences and did not contain any additional mutations. The GG and HCK mutants can be transmitted at efficiencies similar to WT. The lower transmission efficiencies when N. clevelandii is the virus source probably have more to do with the aphidplant interaction than with the aphidvirus interaction. The S-R and D-P-K mutants were not transmitted as efficiently as the WT virus. It is unknown whether the virus is not transported as efficiently across the midgut or accessory salivary gland or if the virion is less stable in the aphid.

PLRV CP modelling
Our model of the PLRV CP structure (Lee et al., 2005) can be used to explain the various biological data reported here. When the location of CP mutations on our model structure is compared with the ability of those mutants to assemble stable virions and systemically infect plant hosts, we find good correlation between lack of assembly and mutations located at or near subunit interfaces. Mutations located in loops and away from subunit interfaces have little measurable effect on biological characteristics. All of the mutants that do not assemble virions have mutations located within or bordering β-strands, except for mutant L-H which is within an α-helix. The CP mutant FT (F₇₄T₇₅), for example, is putatively located at the AB subunit interface. Mutations in these residues would be expected to disrupt potential hydrophobic subunit interactions with neighbouring residues such as F₇₂ and I₁₀₃, leading to an inability both to assemble stable particles and to systemically infect their plant hosts (see Fig. 5a). The leucine in the L-H (L₁₀₄H₁₀₈) mutant is predicted by the model to be located close to the AB subunit interface, whereas the histidine would be located close to the centre of the asymmetric trimer of subunits, the site of the biologically active acidic patch domain (Lee et al., 2005).

(54K): Fig. 5. Structures of the PLRV CP S domain predicted by the model of Lee et al. (2005) that identifies the relative positions of the amino acids modified in the various CP mutants. (a) Altered residues in the FT (red) and L-H (green) mutants that lie along the twofold axis, where two CP monomers come in close contact. F74 appears to form an aromatichydrophobic cluster involving F72, I103, Y107 and possibly L104, which is part of the L-H mutant. This cluster is likely to be relevant for the structural stability of the monomer. The H108 residue of the L-H mutant is not part of the twofold axis, but lies in close association with the acidic patch region (Lee et al., 2005) at the centre of the asymmetric trimer of subunits. The C monomers are in gold and the A and B monomers are in blue. (b) Residues involved in the GG (silver), SE (red), TK (green) and GNG (purple) mutants. The mutations in the GG mutant are located in the N-terminal arm of the CP monomer, and would be predicted to lie in close proximity to the arms of two other monomers at the threefold axis of symmetry. These are probably located internally in the virus particle and the mutations have no measurable effect on virion stability or transmission. Changes in the other residues identified in this panel all affect virion assembly or stability and are all likely to be involved in monomer binding along loops forming the threefold axis of symmetry. (c) Residues involved in the GP (silver), D-P-K (purple), HCK (green) and TIR (red) mutants. The mutations in the TIR mutant affect virion assembly or stability and appear to be located on a β-strand that faces toward the centre of the virus cavity. This region does not appear to be involved in monomer interactions. Mutations in the GP and HCK mutants are located on protruding loops and are not involved in monomer interactions. The residues of the D-P-K mutant are also located on an external loop and do not interact with other monomers, yet changes in one or more of these residues reduce aphid transmission efficiency. The K100 residue is at the interface of three monomers, but not the acidic patch described by Lee et al. (2005), and near a number of other positively charged residues (K76, K100, K105, K150, R164, K184, K188, R197, R210 and K208) that may be involved in binding other molecules, e.g. receptors, in the aphid.

The amino acids targeted in several of the mutants located throughout the C terminus of the CP are predicted by the model to be located in close proximity to the threefold axis of symmetry (Fig. 5b). The TK (T₁₅₄K₁₅₅) mutation may disrupt the contact between monomers at the threefold axis. A serine-rich region of the CP (residues 121130, SEASSTSSGS) is predicted by the model to fold into this same threefold axis. The SE (S₁₂₁E₁₂₂) mutation, which falls in this serine-rich region, abolishes the stable assembly of virus particles and systemic movement of virus in plants. The GNG (G₁₈₅/G₁₈₇) mutation would also be located near this threefold axis, and might be expected to disrupt virion stability. The GG mutation, which is located in the N-terminal R-domain, may also be proximal to the threefold axis region, but it does not affect particle stability, host range or vector transmission. This N-terminal domain is likely to be directed into the centre of the virion and interact with the viral RNA.

Two other mutants, GP (G₉₀P₉₁) and TIR (T₁₉₉/R₂₀₁), do not assemble stable virions, but the location of the mutations is not predicted to occur in regions of the folded CP monomer that would influence subunit interactions in any obvious way (Fig. 5c). The mutations in the GP mutant are predicted to be located in a hydrophobic loop near the threefold axis of symmetry. The mutations in the TIR mutant are predicted to be located on a β-strand that would face toward the interior of the virus particle. These mutations could affect the stabilization of the external loop (GP) or interactions of the CP with the viral RNA (TIR). Alternatively, these two mutations may point to areas of the CP where the model does not accurately predict the structure.

Two of the four mutants that were able to assemble stable particles and move systemically in plant hosts have mutations at the N-terminal domain of the CP (S-R and GG). The arginine-rich N-terminal domain would be expected to be internal and interacting with the viral RNA, but the dynamics of virus structure may allow this domain to assume a different conformation on some monomers (Johnson, 2003) and to be exposed to the outside environment, making it available for interactions with other host or vector components. The S-R (S₂₄/R₂₈) mutation reduced aphid transmission efficiency (Table 2). The other two mutants that were able to assemble stable virions, D-P-K (D₉₅/P₉₇/K₁₀₀) and HCK (C₁₃₉), have mutations predicted to be located on exposed loops. The aphid transmission efficiency of the D-P-K mutant was reduced relative to that of WT PLRV, while the aphid transmission efficiency of the HCK mutant was similar to that of WT (Table 2).

Most of the PLRV CP mutants described here that had mutations predicted by the model to be located within β-strands were unable to assemble stable particles, move systemically within the plant hosts or be transmitted by aphids. Those PLRV CP mutants that had mutations predicted to be on surface loops or outside β-strands were unaffected in their ability to assemble stable particles, move systemically within the plant hosts or be transmitted by aphids. Thus, the biological phenotypes of our various mutants can be explained for the most part by the models proposed, either by alignment of the luteovirus and polerovirus CP sequences (Mayo & Ziegler-Graff, 1996) or by homology modelling (Lee et al., 2005; Terradot et al., 2001); however, the biological phenotypes are not always consistent with the model proposed for beet western yellows virus (BWYV) (Brault et al., 2003). For example, the FT mutation, which lies in a β-strand in our model, would be similar to the S70A mutation described by Brault et al. (2003) for BWYV. In BWYV, the sequence is FS, not FT as in PLRV. This mutation in BWYV also lies in a β-strand on the BWYV CP model, but does not have any observable phenotype, unlike the PLRV FT mutant, which does not assemble virions or move systemically in plants. Another mutation in the BWYV CP was T83A, which is analogous to T88 in PLRV and is located just before the GP mutation reported here. Both mutations are predicted to lie in exposed loops in both our model and the BWYV model (Brault et al., 2003), yet while the GP mutation disrupts virion formation and systemic infection, the T83A mutation has no observable phenotype. Additional differences in the biological phenotypes of similar mutations in the acidic patch domain of PLRV and BWYV CP were described by Lee et al. (2005). Based on the two distinct models that have been proposed for the polerovirus CP S-domain (Brault et al., 2003; Lee et al., 2005; Terradot et al., 2001) and the differences in biological phenotypes of some similar mutations in the PLRV and BWYV CP, the structure and arrangement of biologically active CP domains in the assembled virion may differ among polerovirus species. Our model does fit the biological data for PLRV CP mutations to date and should facilitate the continued identification of important domains and amino acid residues that affect particle assembly, virus movement within hosts and aphid transmission.

We thank Dawn Smith and Tom Hammond for assistance with aphid transmission assays and greenhouse work. This research was partially supported by USDA CSREES NRI grant 96-01120 and a USDA, ARS Postdoctoral Research Associate Grant. Part of this work was carried out using the resources of the Computational Biology Service Unit supported by Cornell University and Microsoft Corporation.

Footnotes

^,†, Lawrence Lee^1,2 †These authors contributed equally to this work.

A supplementary table showing the primers used in this study is available with the online version of this paper.