Both Structural and Non-Structural Forms of the Readthrough Protein of Cucurbit aphid-borne yellows virus Are Essential for Efficient Systemic Infection of Plants

Cucurbit aphid-borne yellows virus (CABYV) is a polerovirus (Luteoviridae family) with a capsid composed of the major coat protein and a minor component referred to as the readthrough protein (RT). Two forms of the RT were reported: a full-length protein of 74 kDa detected in infected plants and a truncated form of 55 kDa (RT*) incorporated into virions. Both forms were detected in CABYV-infected plants. To clarify the specific roles of each protein in the viral cycle, we generated by deletion a polerovirus mutant able to synthesize only the RT* which is incorporated into the particle. This mutant was unable to move systemically from inoculated leaves inferring that the C-terminal half of the RT is required for efficient long-distance transport of CABYV. Among a collection of CABYV mutants bearing point mutations in the central domain of the RT, we obtained a mutant impaired in the correct processing of the RT which does not produce the RT*. This mutant accumulated very poorly in upper non-inoculated leaves, suggesting that the RT* has a functional role in long-distance movement of CABYV. Taken together, these results infer that both RT proteins are required for an efficient CABYV movement.


Introduction
Cucurbit aphid-borne yellows virus (CABYV) is a member of the Polerovirus genus in the Luteoviridae family [1]. Poleroviruses are strictly transmitted by aphids in a circulative and non-propagative manner [2]. In plants, polerovirus infection is limited to phloem tissues: cell-to-cell movement occurs between vascular parenchyma cells, companion cells and the enucleated sieve elements whereas long-distance movement follows the sieve elements. Because polerovirus particles were detected in plasmodesmata connecting phloem cells [3][4][5], the virus is thought to move in the form of virions and recent data confirmed that particles of Turnip yellows virus (TuYV, Polerovirus genus) are essential for long-distance movement [6]. Polerovirus have isometric particles of about 25 nm of diameter containing two structural proteins: the major coat protein (CP) encoded by ORF3 and a minor component, referred to as the readthrough (RT) protein, which is synthesized after ribosomes bypass the ORF3 stop codon during translation. The RT protein is incorporated into the capsid by its CP moiety, with the RT domain protruding from the virion surface [7,8]. While the full-length RT protein of 74 kDa is readily detected in infected plants, a C-terminal truncated form of the protein of about 55 kDa, hereafter referred to as RT*, is only easily detected when incorporated into virions [7,[9][10][11]. The RT* protein was observed in crude extracts of protoplasts infected with Barley yellow dwarf virus (BYDV, Luteovirus genus, Luteoviridae family) suggesting that the cleavage of the RT protein is not due to fortuitous degradation but reflects a conserved, presumably biologically significant processing event [9,11]. The identification of viral proteins, or viral domains, involved in polerovirus movement has been the subject of numerous studies. Poleroviruses encode a movement protein (P4) that, in spite of having cellular and biochemical characteristics of viral movement proteins [12][13][14][15][16], cannot support virus movement outside phloem cells. Interestingly, this protein was shown to be host-dependent, suggesting the existence of a P4-independent transport of poleroviruses in some plants [17,18]. The CP which is essential for particle formation is also required for virus movement [17]. Conversely, the RT* protein was found dispensable for BYDV and TuYV transport, although virions devoid of RT* were greatly impaired in their ability to invade plants [7,19,20]. The number of infection foci along sieve elements was also reduced [5]. Apart from the RT*, the non-structural C-terminal domain of the RT protein was also shown to be involved in virus accumulation in plants using TuYV deletion mutants [21]. Moreover, plant infection with RTengineered mutants of TuYV and Potato leafroll virus (PLRV, Polerovirus genus) unable to encapsidate the RT* suggested that the complete RT protein cannot achieve in trans the transport function carried by the incorporated RT* [21,22]. The complexity of the polerovirus RT protein function in virus movement was recently further illustrated by the fact that the PLRV RT protein could act in trans on virions to retain them in the phloem where they are available for aphid acquisition [20]. In addition to its complex role in virus movement and phloem limitation, the RT protein is a key factor in aphid transmission since it intervenes in virus transport across the aphid gut cells, specifies intestinal tropism and interacts with aphid endosymbionts [7,11,[20][21][22][23][24][25][26].
A way to differentiate the involvement of each RT protein form (full-length and truncated forms) in polerovirus transport throughout the plant is to generate a polerovirus mutant able to synthesize only the RT* protein in a form that retains its capability to be incorporated into virions. Up to now, deletions in the RT protein C-terminal moiety generated mutants unable to synthesize or to anchor the RT protein into the particles [21,22]. Internal sequences present at distance in the RT open reading frame were shown to be required for the readthrough translation mechanism and for the RT* packaging into virions [21,27].
CABYV is a polerovirus infecting cucurbits [28] and was used as a model to decipher the role of the two RT proteins (full-length and truncated versions) in viral long-distance transport. By analyzing the movement of a set of CABYV mutants modified either by deletion or point mutations in the RT sequence, we showed that both RT proteins are essential for CABYV longdistance movement. Based on these results, we propose a hypothetical model where the free full-length RT protein would act in trans on virions bearing the truncated RT protein (RT*).

Generation of CABYV mutants
All point mutants were created by PCR mutagenesis using pCA-WT as template [29]. The mutagenic primers were designed to substitute successively a series of two to three amino acids for a set of two or three alanine residues around the putative cleavage site of the RT protein (Fig. 1A). The generated PCR products bordered by the EagI and SalI restriction sites were introduced back into the pCA-WT plasmid digested by the same enzymes giving nine constructs referred to as pCA-PCS (Putative Cleavage Site). Another mutant (PCS3+) was obtained that contains a tryptophan in addition to three alanine substitutions. Finally, vectors for agroinfection were constructed by replacing the AgeI-SalI fragment of pBin35SCA-WT with the mutated AgeI-SalI fragment from the pCA-PCS constructs. All PCS constructs presented in Fig. 1A were obtained following this procedure except for PCS4 which was produced using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies, France) to introduce directly the mutations into pBin35SCA-WT [29] following manufacturer's instructions. In order to obtain CA-BYV-RTDC ter (Fig. 1B), two unique restriction sites were first introduced into pCA-WT. A NheI site was placed right after the CP stop codon (nt 4104) and a MluI site (nt 5481) 25 nucleotides upstream of the RT protein stop codon (nucleotide positions refer to NC 003688 accession). The resulting plasmid is referred to as pCA-NM3 (Fig. S1). To produce CABYV-RTDC ter , pCA-WT was used as a template to amplify the N-terminal domain of ORF5 from nt 4098 to 4890 with primers containing NheI and MluI restriction sites at either end. The resulting PCR product, cleaved with NheI and MluI, was introduced into pCA-NM3 cut with the same enzymes. Vectors for agroinfection corresponding to CABYV-NM3 and CABYV-RTDC ter were made as described above for the PCS mutants. All the PCR-amplified fragments were sequenced to verify the absence of additional mutations. Binary plasmids bearing the CABYV mutated sequences were introduced into Agrobacterium tumefaciens strain C58C1 for agro-infection [29].

Plant inoculation, virus detection by ELISA and virus purification
A. tumefaciens harboring the mutant constructs were grown to an optical density of 0.5 at 600 nm and agroinfiltrated to Montia perfoliata or Cucumis sativus [29]. Agroinfiltrated plants were analyzed 2 to 4 weeks post-inoculation by double-antibody sandwich (DAS) enzyme-linked immunosorbent assay (ELISA) [30] with a rabbit polyclonal antiserum raised against CABYV (SEDIAG, France). Virions were purified from inoculated or systemic leaves as previously described [31].
Western blotting and analysis of viral RNA progeny Viral proteins present in total protein extracts of infected plants, in phloem exudates collected from infected cucumbers [32] or in purified virions, were detected by western blot using a CABYV polyclonal antiserum [28], a CABYV-RT-C ter specific antiserum [24] or a polyclonal antiserum raised against a synthetic peptide corresponding to aa 155 to 177 of the CABYV-CP. Detection of viral proteins was performed with the Lumi-Light Western Blotting Substrate (Roche, France). The stability of the mutations Figure 1. Schematic representation of CABYV mutants. A) Genetic organization of CABYV RNA with the position of the encoded structural proteins (CP and RT*) and the full-length RT protein (RT) (arrows). The restriction sites AgeI, EagI and SalI used to obtain the PCS mutants are shown. The wild-type amino acid sequence (amino acid 430 to 470) of the region on the CABYV RT protein targeted by alanine scanning is shown below with the corresponding changes in the ten PCS mutants. The underlined R residue represents the last C-terminal amino acid identified by mass spectrometry on the CABYV-RT* protein [34]. The C-terminal amino acid on RT* is tentatively positioned at amino acid 461 (grey-shaded serine). B) Genetic map of CABYV-RTDC ter deletion mutant. NheI and MluI restriction sites originally introduced in CABYV-NM3 to obtain CABYV-RTDC ter mutant are positioned. doi:10.1371/journal.pone.0093448.g001 in the viral progeny following agroinfection was examined by RT-PCR in two infected plants per construct after extraction of total RNA using RNeasy Plant Mini kit (Qiagen, France). Reverse transcription was primed using an oligonucleotide complementary to CABYV nt 5004 to 4987, and a PCR fragment encompassing the mutation was synthesized with an additional oligonucleotide corresponding to nt 4652 to 4670. The PCR products were further sequenced. To analyze CABYV-NM3 viral progeny, reverse transcription was primed with two oligonucleotides, one downstream the NheI restriction site from nt 4800 to 4780, and the second downstream the MluI restriction site from nt 5670 to 5653. PCR products covering the CABYV sequence from nt 3822 to 4800 and from 4891 to 5670 were sequenced. To analyze the presence of CABYV-RTDC ter in systemic plant tissue, the oligonucleotide corresponding to CABYV nt 5514 to 5501 was used for cDNA synthesis and a PCR fragment encompassing the deletion introduced in the RT sequence was amplified using an oligonucleotide complementary to nt 4261 to 4278. To detect CABYV-NM3 and CABYV-RTDC ter in aphids, the reverse transcription was primed with the oligonucleotide corresponding to nt 3889 to 3869 on total RNA extracted from aphids following the animal tissue protocol of the RNeasy Plant Mini kit (Qiagen, France). A PCR fragment of 386 bp was amplified using the additional oligonucleotide corresponding to nt 3503 to 3520.

Mass spectrometry
Mass spectrometry was conducted at the proteomic facility hosted at the Institute of Molecular and Cellular Biology of Strasbourg (France). Proteins were extracted from polyacrylamide gels and samples were prepared as described by Hamrita et al. [33]. MALDI-TOF-MS and Nano LC-MS-MS analyses were performed following the procedure given by Bencharki et al. [32].

Results
The CABYV RT* protein is present in phloem exudate collected from infected cucumbers The complete RT protein of poleroviruses (74 kDa theoretical MW) was previously detected in total protein extracts prepared from infected plants [21][22][23][24]. Conversely, the truncated version of the protein (RT*) that represents the only form incorporated into virions [7,10] was so far only reported in BYDV-infected protoplasts [9,11]. The RT* protein lacks about 22 kDa of the C-terminal domain of the RT protein. Absence of detection of the RT* protein in polerovirus-infected plants may result from a lack of sensitivity of the antibodies used for its detection, or from a rapid degradation of the protein in its unassembled form. We took advantage of the ability of CABYV to infect cucumber plants, from which phloem exudate can be easily collected, to investigate the presence of the RT* protein in sieve elements. Although a CPspecific antiserum can potentially recognize both RT proteins, detection of the complete RT by western blot was very weak (see Fig. 2B). We therefore used a CABYV-polyclonal antiserum to detect the RT* protein and a CABYV-RT-C ter antiserum which recognizes unambiguously the RT protein. While the complete CABYV-RT of apparent MW of 95 kDa was present in infected C. sativus and M. perfoliata plants, no RT* could be detected in total plant extracts (Fig. 3A). Conversely, a band of about 55 kDa was specifically observed in phloem exudate collected from infected cucumber plants (Fig. 3B) and could presumably correspond to the RT* protein. The discrepancy between these results may arise from virion enrichment during phloem exudate collection thus enabling the detection of the incorporated RT* protein. Contrary to the complete RT that was easily and reproducibly found in phloem exudate sampled from infected plants, the 55 kDa product was not always detected, neither at the same RT*/CP ratio in phloem exudate collected from infected C. sativus. Similarly, virions were not consistently observed by transmission electron microscopy in phloem exudate sampled from infected C. sativus (Fig. 3C). This suggests that virions are unevenly distributed in sieve tubes which may explain the difficulty to detect reproducibly the incorporated RT* in phloem exudate extracts. On the other hand, free full-length RT could be released into sieve elements where the protein may migrate independently of virions. Distribution and local accumulation of virions in sieve tubes may not reflect those of the free complete RT protein.

Systemic movement of CABYV mutants bearing mutations in the putative cleavage site of the RT protein
To address the biological functions of CABYV RT-derived proteins in the virus life cycle, we aimed at engineering virus mutants able to generate either the complete RT protein or the RT* protein. The former one could be obtained by modifying the RT protein processing. Therefore point mutations were introduced around the putative cleavage site in the RT protein sequence located approximately in its middle. The arginine at position 440 in CABYV RT is the last C-terminal amino acid detected by mass spectrometry on RT* protein [34] suggesting that the RT protein cleavage must occur at any close position downstream of it. Starting from this position, downstream amino acid pairs or triplets were sequentially changed into two or three alanine residues along a 24 amino acid stretch, leading to ten CABYV mutants, referred to thereafter as PCS 1-8 mutant for ''Potential Cleavage Site'' mutant ( Fig. 1A). Full-length viral cDNAs, containing the mutated RT sequences, were transferred into a binary vector and further transformed into A. tumefaciens for agroinoculation of M. perfoliata.
Four weeks post-inoculation, the ability of the mutants to move systemically was tested by ELISA on newly developed noninoculated leaves. This assay detected more specifically intact virions than free structural proteins since disassembled CP subunits, obtained by addition of SDS to whole virions, gave a significantly lower absorbance than whole virions (Fig. S2). It should be pointed out that CABYV accumulation in infected plants, measured by ELISA, is often subject to important variations between plants as revealed by the high standard deviation values observed with the wild-type virus (Table 1). This may reflect heterogeneous partitioning of CABYV in infected plants, as already mentioned in the previous section. All ten mutants were able to invade non-inoculated tissue (Table 1).
Two point mutants, PCS1 and PCS3+, showed a significant reduction of infectivity (the percentage of plants infected with PCS1 and PCS3+ is statistically different from wild-type-infected plants, p-values = 0, Exact Fisher Test) and accumulated strikingly less than the wild-type virus in systemic leaves (Table 1). Furthermore, no CP or RT proteins could be detected in noninoculated leaves by western blot (Fig. 2A). However, the virus titre within the infiltrated leaves was close to that observed with the wild-type virus as determined by ELISA (p-values calculated with Kruskal Wallis test: 0.677 for PCS1 and 0.173 for PCS3+), and by western blot (nearly similar CP accumulation in the different protein samples) indicating that virus multiplication of both mutants was almost unaffected (Table 1; Fig. 2B). All other PCS mutants accumulated in plants up to levels close to that of the wildtype virus ( Table 1).
As viruses with RNA genomes are subject to high mutation rates during replication, we investigated the stability of the point mutations introduced in the PCS mutants. Total RNA was extracted from systemic leaves of two infected plants and the cDNA encompassing the mutation site (from nt 4652 to 5004) was amplified by RT-PCR and then sequenced. All modifications introduced into the RT sequence were conserved in the progeny of the corresponding PCS mutants. In the case of PCS1, the progeny showed a synonymous mutation at nucleotide 4767 in one of the two plants analyzed. A non-synonymous mutation also appeared at position 4663 in PCS5 progeny in one of the two plants analyzed. This latter modification, that changed the cysteine residue at position 386 on the RT sequence into a serine (66 amino acids upstream the first mutation introduced in PCS5), had no obvious effect on the infectivity of PCS5 since the virus mutant was readily detected in non-inoculated leaves of M. perfoliata.

Systemic movement of a C-terminally truncated RT CABYV mutant
To design a virus mutant able to synthesize solely a truncated RT protein similar to that incorporated into viral particles, the serine at position 461 was tentatively designated as the last amino acid of the RT* protein (Fig. 1A). Therefore, the downstream 205 amino acids in the RT sequence were deleted from CABYV-NM3, a full-length clone containing two unique restriction sites, NheI and MluI, respectively downstream and upstream of the ORF3 and ORF5 stop codons ( Fig. 1B and Fig. S1). Note that due to the cloning strategy, this mutant referred to as CABYV-RTDC ter displays altered amino acids: two following the ORF3 stop codon and three at the C-terminus of the protein (Fig. S1). M. perfoliata plants were inoculated with this mutant and the corresponding virus controls (CABYV-WT and CABYV-NM3) but CABYV-RTDC ter was never detected by ELISA in young non-infiltrated leaves (Table 1), nor by Northern blot (data not shown). Although CABYV-NM3 that generated CABYV-RTDC ter accumulated in M. perfoliata less than the wild-type virus, the percentage of infected plants was high (87%), suggesting that the amino acid modifications introduced into CABYV-NM3 (Fig. S1) and present in CABYV-RTDC ter , did not affect virus infectivity but rather reduced the virus titre ( Table 1). The CABYV-RTDC ter deficiency in long distance movement was also noticed in two additional hosts of CABYV, A. thaliana and C. sativus, where no accumulation was detected by ELISA in non-inoculated leaves (Table S1). However, CABYV-RTDC ter was readily detectable by ELISA in agroinfiltrated leaves of M. perfoliata (Table 1), and the CP accumulated at a similar level compared to CABYV-NM3 as observed by western blot (Fig. 2B). In order to test whether CABYV-RTDC ter loading into sieve tubes was impaired, the presence of the virus mutant in sieve tubes was investigated by feeding non-viruliferous aphids for 48 h on agroinoculated C. sativus systemic leaves. As CABYV-RTDC ter particles contain both CP and RT* proteins required for efficient virus acquisition into the hemolymph [25], longstanding feeding of aphids in sieve tubes may concentrate virions in the insect's body and facilitate virus detection. Virus detection in aphids could therefore confirm the presence of virus particles in sieve tubes. CABYV-NM3 was detected by RT-PCR in a batch of 30 aphids fed on systemic leaves of plants agroinfected with this virus, but no signal was found in the RNA samples prepared from two batches of 30 aphids fed on the upper leaves of plants agroinoculated with CABYV-RTDC ter (Fig. S3). These results suggest that the virus was not, or hardly, present in sieve tubes presumably because the upstream step, i.e. the loading from nucleated cells into sieve elements, is compromised in the absence of the complete RT protein.

Accumulation of the mutated RT proteins in planta
The effect of the mutations introduced in the RT sequence on its accumulation was investigated by western blot analysis on protein extracts prepared from plants agroinoculated with the different RT mutants. Both CP and full-length RT (95 kDa apparent MW) proteins were detected in systemically infected leaves of plants inoculated with CABYV-NM3 and the wild-type virus ( Fig. 2A) and in plants inoculated with the mutants PCS2, PCS2-3, PCS3, PCS4, PCS5, PCS6, PCS7 and PCS8 (Fig. S4). In contrast, neither of the CP or RT proteins could be detected in systemic leaves of plants inoculated with PCS1, PCS3+ or CABYV-RTDC ter ( Fig. 2A) which correlates with low or absence (in the case of CABYV-RTDC ter ) of virus accumulation in systemic tissue (Table 1). However, when the protein content was investigated in infiltrated leaves, the major CP was readily detected for all three mutants (Fig. 2B) and the complete RT (apparent MW 95 kDa) could be observed in leaves infiltrated with PCS3+ mutant only after a longer exposure of the blot (Fig. 3B,  right panel). This suggests that the PCS3+ RT protein accumulated at a lower level compared to wild-type protein. Surprisingly, no 95 kDa RT protein was detected in PCS1-inoculated leaves (Fig. 2B), even after an overexposure of the blot (not shown) suggesting that the mutations introduced in the RT sequence might have affected either the readthrough translation mechanism or the stability of the mutated RT protein. However, a minor band of about 35 kDa was reproducibly present in the protein extracts from plants infected by PCS1, but the identity of this protein has not been further analyzed (Fig. 2B). Similarly, no RT protein was observed in C. quinoa protoplasts infected with PCS1 transcripts synthesized in vitro (data not shown). The C-terminally deleted RT encoded by CABYV-RTDC ter was also detected in agroinoculated leaves and migrated with an apparent MW of ca. 57 kDa, slightly above the theoretical 52 kDa MW (Fig. 2B). Taken together, these results show that most of the mutations introduced in the potential cleavage site of the RT (8 mutants out of 10) did not affect RT protein synthesis, nor the viral infectious cycle. Conversely, both PCS1 and CABYV-RTDC ter mutants that produced no RT protein or only a C-terminal deleted version respectively were impaired in their ability to move systemically in infected plants. Finally, PCS3+ that expressed lower levels of the full-length RT protein also exhibited an inefficient long-distance movement.

RT protein cleavage and incorporation of the RT* protein into virions
The ability of the mutated RT proteins to be correctly processed and incorporated into viral particles was further addressed by analyzing protein composition of purified virus particles prepared from infected plants. We assumed that if the RT* protein was properly processed by the mutant virus it would be packaged into virions. The purified mutant viruses were prepared from systemic leaves of M. perfoliata infected individually with the eight PCS mutants that performed like the wild-type virus or from M. perfoliata agroinoculated leaves for PCS1, PCS3+, and CABYV-RTDC ter whose accumulation in systemic leaves was not detectable or too low to allow virion purification. The major CP (ca. 24 kDa) was detected in all virus preparations (Fig. 2C lower panel and Fig. S5). The RT* protein (apparent MW 55 kDa) was present in the particles of all mutants except PCS1 ( Fig. 2C and Fig. S5). A longer blot exposure did not further allow RT* detection (Fig. S5 right panel) what was predictable since no fulllength RT protein was detected in plants infected with this mutant ( Fig. 2A). In microscopy, PCS1 particles without the minor capsid component exhibited an apparent structure similar to wild-type virions (Fig. 4). This confirms that the RT* protein is dispensable for virion formation and does not alter the global particle shape, as previously reported for other members of the Luteoviridae family [7,9,35]. An additional smaller band of ca. 30 kDa was reproducibly detected in the purified particles of PCS3+ using the antibodies directed against CABYV virions (Fig. 2C and Fig.  S5). In order to further characterize the 30 kDa viral protein present in PCS3+ particles, a band of the same molecular weight was excised from a gel loaded in parallel with proteins extracted from PCS3+ virions. Mass spectrometry analysis identified peptides covering 46.7% of the CP sequence and matching the CP sequence up to the stop codon at position 200 (Fig. S6). This suggests that the 30 kDa protein present in PC3+ virions comprises the entire CP sequence. However, based on its apparent MW (ca. 30 kDa), this CP-fusion protein is expected to contain additional amino acids probably those that extend beyond the CP stop codon (CP is 24 kDa). Although a peptide exhibiting a mass of 1155.55 Da that could match to the positions 256-264 in the RT sequence was detected by mass spectrometry (Fig. S6), it could not be confirmed by Nano LC-MS-MS, because of too low amounts of peptide. Failure to detect any peptide downstream the CP stop codon may originate from the proline-rich sequence downstream the CP stop codon which is known to impede trypsin cleavage [36]. Additional enzymes (chymotrypsin and Asp-N) were used to digest the protein but mass spectrometry analysis of the released products did not provide more information on the protein. Therefore we cannot conclude whether the 30 kDa protein present in PCS3+ purified particles is a CP posttranslational modification or a shorter version of the RT*. Whatever the nature of this additional product anchored into PCS3+ particles, the overall virion shape assessed by transmission electron microscopy was indistinguishable from wild-type CABYV particles (Fig. 4).
A high MW protein (ca. 90 kDa) was also present in all purified virus samples, sometimes visible only after longer exposure of the blot. (Fig. 2C and Fig. S5). This protein, referred to as P90, is recognized by virus-targeted antibodies and was previously identified as a plant glycosyl hydrolase protein by mass spectrometry [34]. Identification of this protein was further confirmed by its strong interaction with anti-P90 specific antibodies (Fig. S7A). However, to definitively discard the presence of an unprocessed RT protein in the purified PCS3+ mutant virions, the blot was incubated with antibodies directed against the C-terminal part of the RT protein. No protein of MW 95 kDa (apparent MW of the full-length RT protein) was detected in purified virus extracts confirming that no unprocessed RT protein is incorporated into virus particles of PCS3+ (Fig. S7B).
In summary, the absence of 55 kDa RT* protein in purified PCS3+ virus preparation strongly supports the fact that the correct processing of the RT protein produced in infected plants is affected by the mutation introduced in PCS3+.

Discussion
In addition to its major role in virus transmission, the minor capsid component of poleroviruses, the RT protein, was suggested to control virus loading into the phloem [5]. Two forms of the RT are produced in plants: a 74 kDa full-length protein and a 55 kDa C-terminal truncated version (RT*), the latter being the only detectable species incorporated into virions [7,[9][10][11]. In this paper, we showed for the first time that both forms of the CABYV RT protein can be detected in phloem exudate collected from infected cucumber plants. To address the function of each RT protein species in polerovirus life cycle, and in particular in long-distance movement, we engineered CABYV mutants producing either only the full-length RT or the truncated form. In addition, a series of CABYV mutants affected in the potential cleavage site of the RT protein (amino acids 440-463 in the RT sequence) was generated. Our results showed that the mutations introduced in eight out of the ten alanine substitutions virus mutants (PCS2, PCS2-3, PCS3, PCS4, PCS5, PCS6, PCS7 and PCS8, see Fig. 1) did not affect synthesis, processing of the RT and RT* incorporation into viral particles. In addition, none of these mutants were strongly affected in their ability to move systemically. In the case of PCS1 mutant, no RT was detected in infected plant extracts, and we suspect that the five nucleotide change introduced in the PCS1 RT sequence (Fig. S8) may have affected a sequence required for the translational readthrough mechanism [1,21,27]. An alternative hypothesis for the absence of RT protein in PCS1-infected plants could be protein instability due to the mutation.
A particularly interesting observation was made with PCS3+ mutant as it produces a full-length mutated RT protein that could be detected in infected plants in lower amounts but was likely improperly processed. The RT* could never be detected in PCS3+ particles, although another CP-containing product of about 30 kDa was reproducibly assembled into virions. This 30 kDa protein could represent an RT-processed product released after cleavage at an incorrect site exposed on the PCS3+ mutated RT protein. It is noteworthy that the mutation introduced in PCS3+ spans the alanine substitutions introduced in PCS2, PCS2-3 and PCS3 (amino acids 443-447). Taken individually, these substitutions did not modify the RT processing, suggesting that the effect observed with PSC3+ may be related to the combination of the individual mutations or to the introduction of a bulky aromatic tryptophan at position 444 (Fig. 1). This latter modification is likely to impose a strong structural constraint to the RT protein conformation of PCS3+ that may mask or change the cleavage site of the protein. These considerations led us to propose a cleavage model controlled by the structural context around the cleavage site of the RT protein rather than the sequence per se. Implication of the local secondary structure of the cleavage site, and not just the sequence per se, has been suggested to play a role in the processing of the polyproteins of Sesbania mosaic virus (Sobemovirus genus) and of the P3-6K1 protein of Plum pox virus (Potyviridae family) [37,38].
The PCS3+ mutant inefficiency to move to systemic tissues shows that the full-length RT protein produced by the PCS3+ mutant cannot complement the absence of RT* incorporation. However, as the RT protein produced by PCS3+ contains a 4 amino acid change, we cannot completely discard the hypothesis that the mutation per se is deleterious for its function in virus longdistance movement. Previous data showed that the PLRV RT protein was able, but only to some extent, to act in trans on the movement of viral particles without RT* [20], while no such in trans action of the full-length RT was reported for TuYV [21]. Finally, the very poor infectivity of PCS1 mutant, whose particles are only composed of the CP, supports the hypothesis that some inefficient virus movement, independent of the RT protein, can take place for viruses devoid of RT* [5]. This result correlates with the poor ability of CABYV mutants, in which the entire ORF5 sequence was deleted, to invade systemic leaves (our unpublished results).
A second valuable CABYV mutant we generated is CABYV-RTDC ter which produces a C-terminally deleted RT that is incorporated into virions ( Fig. 2B and Fig. 2C). This mutant represents to our knowledge the first example of an engineered member of the Luteoviridae family able to synthesize solely an RTtruncated protein resembling the RT* and to package it into particles. Although CABYV-RTDC ter produces virions displaying physical and biological properties similar to the wild-type virus (shape, size, virions protein composition; see Fig. 3C and Fig. 4), this mutant was unable to reach upper leaves. By using aphids as virus ''concentrators'' of phloem exudate and by assessing the presence of the virus in the aphids by RT-PCR, we were unable to find CABYV-RTDC ter mutant in the sieve elements, suggesting that it was incapable to invade this plant compartment. In absence of the full-length RT, the RT* packaged into particles seems therefore unable to fulfill the entire transport function required for systemic transport of CABYV. These data strongly support a model whereby long-distance movement requires both the incorporated RT* and the entire RT. The absolute requirement of the C-terminal domain of the CABYV RT protein for longdistance movement reminds the case of Pea enation mosaic virus-1 (PEMV-1, genus Enamovirus, Luteoviridae family) that lacks the homologous C-terminal part of the RT protein and is unable to move systemically in the absence of the helper umbravirus PEMV-2 [39,40].
To explain the requirement of both RT proteins for CABYV systemic movement, direct or indirect interactions between these two proteins can be proposed. Evidence for interactions between PLRV RT* proteins has recently been provided by Chavez et al. [41]. Based on the assumption that such interactions could also take place for CABYV, we propose a model setting the basis for its systemic transport that could eventually be extended to all members of the Luteoviridae family (Fig. 5). As already reported for PLRV [41], the C-terminal part of the RT protein of CABYV is predicted to be highly disordered using the disorder prediction algorithms IUPred [42] and PONDR [43] (Fig. S9). We can therefore assume that the RT protein may adopt different conformations in plant cells and that only a fraction of the RT population is cleaved and incorporated into particles. Incomplete cleavage of the RT would leave a pool of free RT proteins able to act in trans on virions. The free RT proteins could bridge the gap between viral particles and some cellular elements mediating virus transport into sieve tubes (e.g. plasmodesmata or microtubules) (Fig. 5). Only viral particles able to incorporate the RT* will be efficiently loaded into the sieve tubes from where they can be acquired by aphids. How the free RT protein also enters the sieve tubes and whether it is thereafter acquired by aphids and involved in aphid transmission, as suggested for BYDV [11], remains an open question. In parallel to this efficient systemic transport of wild-type CABYV virions in the sieve tubes, virus particles lacking the RT* might also follow an alternative route independent of the RT that remains to be explored.
In order to support the model presented in Fig. 5, it would be interesting to complement the deficiency in long-distance transport of CABYV mutants by inoculating transgenic plants expressing either the full-length RT, or the C-terminal domain of the protein.
Several other additional aspects of polerovirus transport will also need to be addressed in the future like the identification of the protease (viral or cellular) responsible for RT protein cleavage and regulation of the processing. Indeed, this process is crucial to release a defined ratio of RT* versus RT proteins, two essential viral determinants governing major steps in polerovirus propagation, i.e. movement within plants via the phloem and between plants via vector transmission.