P0 proteins of European beet-infecting poleroviruses display variable RNA silencing suppression activity

RSS activity analysis using a patch test in green fluorescent protein-positive (GFP⁺) transgenic N. benthamiana

To determine whether the P0 proteins of the poleroviruses BChV and BMYV could suppress GFP silencing in GFP-expressing transgenic N. benthamiana line 16c, binary vectors expressing P0 of BChV (P0^BC) or BMYV (P0^BM) were constructed, and those expressing P0 of TuYV (pBin-P0^Tu) and P14 of BNYVV (pBin-P14) were used as described previously (Pfeffer et al., 2002). Expression of GFP in the leaves and stems of N. benthamiana line 16c resulted in a green fluorescence that could be visualized under UV light. However, silencing of this GFP expression was visualized by red fluorescence after agro-infiltration of the 16c line with a binary vector pBin-GFP carrying ³⁵S-GFP and designed to transiently express a GFP transcript. The same red fluorescence appeared when pBin-GFP was co-agro-infiltrated with the empty vector (pBin-∅) (Fig. 1⇓, panel 1). The GFP silencing produced in line 16c by infiltration of pBin-GFP can be suppressed by co-infiltration of a pBin vector expressing an RSS protein (Voinnet et al., 1998).

As expected, when pBin-GFP was co-infiltrated with most pBin-P0^BM, the infiltrated patches displayed an intense green fluorescence at 4 days post-inoculation (p.i.), indicating that P0 of BMYV, as already shown for P0 of TuYV and P14 of BNYVV, exhibited an RSS activity (Fig. 1⇑, panels 2–4; Table 1⇓). Patches that were co-infiltrated with pBin-GFP/pBin-P0^BC did not inhibit GFP silencing and displayed the same phenotype as patches infiltrated with pBin-GFP/pBin-∅ (Fig. 1⇑, panel 6). In contrast, no or weak suppression of GFP silencing was detected for P0 of two BMYV isolates (BMYV-N32 and -26) (Fig. 1⇑, panel 5, Table 1⇓), even though GFP mRNA transcripts were detected in the corresponding plants (Fig. 2⇓). This weak GFP fluorescence probably corresponded to incomplete suppression by P0 of the two isolates.

The persistence of the intense GFP fluorescence was observed even after 21 days p.i. in patches that were co-infiltrated with either pBin-GFP/pBin-P0^Tu or pBin-GFP/pBin-P0^BM (data not shown). In contrast, in the case of BNYVV pBin-P14 agro-infiltration, the intense green fluorescence faded after 1 week under our experimental conditions. Indeed, the transcripts of P14 underwent RNA silencing in turn, as the accumulation of siRNA of these transcripts has been revealed (H. Guilley, unpublished results). This could be due to P14 instability in a non-viral context. This observation suggests that P14 certainly does not have the same operating mode as P0 of TuYV and BMYV. In the case of P0^BM-N32 and P0^BM-26, the weak fluorescence faded after 1 week, similar to BNYVV P14-infiltrated patches (data not shown).

Molecular hybridization using a GFP probe (Himber et al., 2003) of total RNA extracted from patches at 4 days p.i. confirmed the visual observations (Fig. 2a, b⇑). Northern blot analyses showed that GFP mRNA was moderately to highly abundant in the green fluorescent patches of leaves that were co-infiltrated with pBin-GFP/pBin-P0^Tu, pBin-GFP/pBin-P14 or pBin-GFP/pBin-P0^BM (Fig. 2a⇑, lanes 3, 4, 11, 12, 15 and 16). Conversely, the siRNA accumulation found in these patches was not intense (Fig. 2b⇑, lanes 3, 4, 11, 12, 15 and 16). Both fluorescence and RNA analyses of GFP expression in the infiltrated N. benthamiana leaves clearly identified most of the P0^BM as RSSs, as already shown for P0^Tu (Pfeffer et al., 2002). In contrast, for all P0^BC studied, as well as for two P0^BM (P0^BM-N32 and P0^BM-26), no mRNA GFP overaccumulation was observed (Fig. 2a⇑, lanes 5–10 and 13 and 14, respectively) and in correlation with this, siRNA molecules accumulated to high levels (Fig. 2b⇑, lanes 5–10 and 13, 14 respectively), indicating that these P0s, if they are stable under our experimental conditions, have an undetectable or weak RSS activity.

RSS activity analysis using BNYVV inoculation of C. quinoa

In order to confirm the results of the aforementioned agro-infiltrations, we used a viral vector complementation method developed recently by Guilley et al. (2009) and based on a BNYVV-derived recombinant vector to identify and characterize RSS activity of different heterologous viral proteins. One advantage of this assay lies in the fact that the expression of the heterologous potential RSS by the recombinant virus reproduces the natural conditions of a cytoplasmic viral infection. In this study, we used a variant of this technique as described below.

BNYVV RNA1, -2, -3 and -4 are necessary for the natural infection and transmission processes for sugar beet (Koenig et al., 1986; Bouzoubaa et al., 1991). However, when BNYVV is mechanically inoculated into experimental hosts such as C. quinoa, RNA1 and -2 are sufficient to induce and maintain infection (Quillet et al., 1989). At 6 days p.i., the inoculated plants displayed chlorotic local lesions of medium size (Fig. 3⇓, panel 1).

We used a BNYVV RNA3-based replicon (Jupin et al., 1992; Bleykasten-Grosshans et al., 1997; Guilley et al., 2009) as an amplification–translation vector to express P0 proteins of different poleroviruses. To assess the capacity of P0^BC and P0^BM to cross-complement the RSS activity of BNYVV P14, we used infectious transcripts comprising RNA1+2ΔP14 of BNYVV strain F2 (Quillet et al., 1989), an artificial strain deficient in P14 synthesis (Hehn et al., 1995). C. quinoa leaves inoculated with this strain developed small necrotic local lesions at 5 days p.i. (Fig. 3⇑, panel 2). These lesions were ten times smaller than those developed after inoculation by wild-type BNYVV RNA1+2 (Fig. 3⇑, panel 1). This phenotype of local lesions was consistent with a low RNA accumulation (Fig. 4⇓, lane 9). As a positive control and in order to show that it was possible to cross-complement the lack of P14, we used tomato bushy stunt virus (TBSV; genus Tombusvirus) P19 protein and TuYV P0 protein as heterologous RSSs (Guilley et al., 2009).

To check the expression ability of RNA3 replicon-based constructs harbouring polerovirus P0 ORFs, we co-inoculated wild-type RNA1+2 along with a replicon, either empty (rep0) or containing the ORF of BNYVV P14, TBSV P19 or P0 of various beet poleroviruses (TuYV-FL1, BChV-2a and BMYV-2ITB). All of these replicons showed the same accumulation as wild-type RNA1 and -2, as visualized on Northern blots (Fig. 4⇑, lanes 3–8). The accumulation rate of viral RNA was somewhat variable between experiments. Moreover, P0 sequence insertion did not interfere with RNA1 and -2 replication (Fig. 4⇑, lanes 6–8). This finding allowed us to use these replicons in heterocomplementation experiments.

Complementation tests were carried out with P0^BC and P0^BM RNA3 replicons, which were inoculated onto C. quinoa in the presence of helper RNA1+2ΔP14 transcripts. At 6 days p.i., chlorotic local lesions appeared on inoculated leaves in the case of P0^BM (Fig. 3⇑, panel 3, Table 1⇑), similar to those with wild-type phenotype obtained with RNA1+2 (Fig. 3⇑, panel 1) or with an inoculum containing RNA1+2ΔP14+repP14. However, no local lesions were observed in the case of the replicons harbouring the P0 ORF from all BChV isolates and two BMYV isolates (BMYV-N32 and -26) (Table 1⇑). This phenotype was similar to that displayed with RNA1+2ΔP14+rep0 (RNA3 empty replicon construct) (Fig. 3⇑, panel 6) and may result from an additional effect to RNA silencing, the defective interfering effect of the RNA replicon described previously (Hehn et al., 1994). The presence or lack of RNA accumulation of these replicons was confirmed by Northern blot analysis (Fig. 4⇑). P0^BM replicons allowed their autonomous RNA multiplication as well as RNA1+2ΔP14 accumulation (Fig. 4⇑, lanes 14 and 15). Amplification of viral RNA was the consequence of an effective heterocomplementation, as was the case with inoculum containing RNA1+2ΔP14+repP14 and RNA1+2ΔP14+repP19^TBSV where BNYVV P14 and TBSV P19 were expressed and complemented viral multiplication by inhibiting RNA silencing (Fig. 4⇑, lanes 11 and 12). In contrast, using P0^BC replicons, no accumulation of either their own RNA or RNA1+2ΔP14 of the helper virus was observed (Fig. 4⇑, lanes 18–21), similar to the experiment with RNA1+2ΔP14+rep0 where BNYVV RNA decreased drastically (Fig. 4⇑, lane 10). This was consistent with the absence of symptoms observed previously (Fig. 3⇑, panel 6). This result showed that all P0^BC proteins as well as P0^BM-26 and P0^BM-N32 (Fig. 4⇑, lanes 16 and 17) did not display any P14 heterocomplementation activity. Furthermore, Western blot analysis of BNYVV CP confirmed the presence or lack of virus multiplication (Fig. 5⇓). CP is produced when P14 is expressed but no CP can be detected when P14 is absent (Hehn et al., 1995). CP accumulation correlated with viral RNA multiplication, except for P0^Tu, for which the CP accumulation was weaker than for the other P0s (Fig. 5⇓, lane 10). This was probably due to its necrotic effect on C. quinoa leaves: the chlorotic local lesions were necrotic in the centre at 5 days p.i. (Fig. 3⇑, panel 4) and became totally necrotic at 7 days p.i. (data not shown).

Discussion

Analysis of different plant viruses indicates that the ability to suppress RNA silencing is a widespread property (Voinnet, 2005). More than 35 individual RSSs have been identified from almost all plant virus taxa (Li & Ding, 2006). These proteins are structurally and functionally very different. For TuYV (syn. BWYV-FL1), the RNA silencing suppression function has been assigned to the P0 protein (Pfeffer et al., 2002). ORF0 is the most variable genomic region within the genus Polerovirus. However, phylogenetic clustering based on the P0 sequence of CABYV, BMYV and BWYV-USA has confirmed their high identity for this part of the genome (Fig. 6⇓; Beuve et al., 2008) and has revealed the potential ability of P0 of these species to suppress RNA silencing.

The aim of this study was to investigate whether P0 of the closely related BMYV and BChV displayed a similar function and whether this ability varied among European isolates. Using two experimental systems based on either nuclear (agro-infiltration) or cytoplasmic (viral context) systems of P0 transcript expression, we demonstrated that the ability to suppress RNA silencing was variable, even among closely related members of the same virus genus. We showed that most P0s of BMYV displayed RSS activity in both of the heterologous systems used. Moreover, the suppression induced by P0 of BMYV was long-lasting in agro-infiltration experiments, similar to P0 of TuYV (Pfeffer et al., 2002) but in contrast to that induced by BNYVV P14, which decreased over a period of 6 days p.i. This observation suggests that P0 of BMYV and P14 of BNYVV may have different modes of action in their RSS mechanism. Conversely, we demonstrated that P0 of six BChV isolates had no detectable RSS activity.

The ability of P0^BC and P0^BM to suppress RNA silencing was investigated using the second system involving a viral context expression, i.e. C. quinoa inoculation experiments using BNYVV replicons as we described recently (Guilley et al., 2009). Thus, using TBSV P19 or P0^Tu, we confirmed that suppression activity analysis by this method is a reliable means of testing various proteins for their RSS ability. We showed that replicons carrying P0 of most BMYV isolates can complement BNYVV P14, but replicons carrying P0^BC, P0^BM-N32 or P0^BM-26 cannot. Therefore, these results are consistent with our aforementioned observations using agro-infiltration of the N. benthamiana 16c line.

In addition, we noted that, after C. quinoa inoculation, P0^Tu induced chlorotic local lesions, becoming necrotic in the centre at 5 days p.i. (Fig. 3⇑, panel 4) and totally necrotic at 7 days p.i. This protein is already known as a pathogenicity and symptom determinant (Pfeffer et al., 2002). In contrast, C. quinoa infected leaves with the P0^BM RNA3 replicon displayed chlorotic local lesions (Fig. 3⇑, panel 3), as well as the phenotype observed in the case of BNYVV infection (Fig. 3⇑, panel 1). Therefore, P0^BM protein was able to restore a BNYVV wild-type phenotype similar to that produced in the case of BNYVV P14.

It is noteworthy that the BChV host range appears to be narrower than that of BMYV, BWYV-USA and TuYV (Hauser et al., 2002; Beuve et al., 2008), suggesting that BChV has evolved separately and probably appeared by recombination between an unidentified polerovirus-like ancestor, which provided the 5′ part of the genome encoding P0 and the polymerase, and BMYV or TuYV, which provided ORF3, -4 and -5 (Hauser et al., 2002). Exchanging ORF0 sequences between BChV and BMYV in order to investigate their biological activity would be informative.

Among the six BMYV isolates, two isolates (BMYV-N32 and -26) did not exhibit RSS activity in either assay. The P0 sequences of these two isolates seem to be closest to the BChV P0s within the BMYV group (Fig. 6⇑). However, amino acid sequence analysis of the isolates BMYV-N32 and -26 did not reveal any changes, in comparison with the other isolates of BMYV, in their F-box-like domain (Table 2⇓).

Moreover, the amino acid sequence comparison with other poleroviruses revealed the presence of the short motif of the leucine-rich repeat sequence LP(LL)L/I (residues 61–65 of BChV), which matches the start of the F-box sequence in P0 of all BChV isolates (Table 2⇑). However, the amino acid composition of a C-terminal-proximal sequence (K/R)IYGEDGX₃FWR, which could be a so far undescribed domain of interaction for F-box proteins (Pazhouhandeh et al., 2006), showed 30, 60 and 90 % similarity, respectively, for isolates of BChV, BMYV and BWYV-USA towards CABYV and TuYV-FL1.

To conclude, these results provide the first reported evidence that P0 of BMYV is an RSS. In contrast, P0 of the closely related polerovirus BChV did not exhibit this function. Moreover, we demonstrated that the RSS ability can vary, even among isolates of the same virus species, as was the case for BMYV in this study. Such an extreme variation in RSS ability in closely related members of a single virus group has been shown for potexviruses (Voinnet et al., 2000; Senshu et al., 2009) and for some CMV strains (Zhang et al., 2006; Lewsey et al., 2007). As we tested P0 as an isolated protein for its potential RSS ability, we cannot exclude the existence of another viral RSS protein, as has been reported for citrus tristeza virus (genus Closterovirus), which possesses at least three RSSs, P20, P23 and CP (P25) (Lu et al., 2004). In the case of BChV, the RSS could correspond to another viral protein or to an interaction between P0 and another viral factor. It is also possible that P0 proteins that do not show suppression activity could be unstable under our experimental conditions. However, as no specific antibodies targeting these proteins are currently available, we are unable to draw conclusions about their stability in planta. Further studies of beet polerovirus P0s are required to understand their variability and mode of action as RSSs. Thus, it would be informative to investigate recognition of AGO1 by these P0s in transgenic A. thaliana plants, as has been performed for P0^Tu (Bortolamiol et al., 2007).