The endosomal sorting nexin-BAR proteins are required for tombusvirus replication in yeast and plant cells

To test the putative role of the yeast PI(3)P-binding proteins in TBSV replication, we analyzed the accumulation of TBSV replicon (rep)RNA in yeast lacking one of fourteen known PI(3)P-effectors (S1 Table). We found that deletion of VPS5 in haploid yeast showed the largest inhibitory effect on the accumulation of TBSV among the genes tested (S1 Table). The absence of Vps5p SNX-BAR protein, which is a PI(3)P-binding endosomal protein, resulted in ~5-fold inhibition of TBSV repRNA accumulation in vps5Δ yeast (Fig 1A, lanes 16–18 versus 4–6), confirming that Vps5p SNX-BAR is critical for TBSV replication in yeast. The closely-related CIRV, which, unlike the peroxisome-associated TBSV, replicates on the boundary membranes of mitochondria, was inhibited by ~6-fold in vps5Δ yeast (S1A Fig, lanes 16–18). Thus, Vps5p is required for tombusvirus replication in different subcellular environments.

Fig 1

Pro-viral role of the endosomal SNX-BAR proteins in tombusvirus replication in yeast and plants.

(A) Top image: Northern blot analysis shows decreased TBSV (+)repRNA accumulation in vps5Δ yeast strain. Vps5p and its deletion mutants were expressed from the constitutive TEF1 promoter from a plasmid. The accumulation level of repRNA was normalized based on 18S rRNA levels (second panel). The accumulation of His₆-p33, His₆-p92 and Vps5-Flag is measured by western blotting and anti-His or anti-Flag antibodies. (B) VIGS-based knockdown (KD) of both Snx1 and Snx2b mRNA levels inhibits the accumulation of TBSV RNA in N. benthamiana. Top panel: The accumulation of TBSV gRNA and sgRNAs was measured using northern blot analysis of total RNA samples obtained from N. benthamiana leaves at 2 dpi. The upper, systemically-silenced leaves were inoculated with TBSV virions on the 12th day after VIGS. The control experiments included the TRV2-cGFP vector. Second panel: ethidium-bromide stained gels show ribosomal RNA level. Middle and bottom panels show the semi-quantitative RT-PCR and quantitative real-time PCR analyses of both Snx1 and Snx2b mRNA levels, whereas tubulin mRNA was used as a control in the VIGS plants. The Y-axis shows the relative levels of the Snx1 and Snx2b mRNAs with the control treatment (in blue) representing 100%. (C) Lack of phenotype in Snx1 and Snx2b knockdown N. benthamiana. The enlarged images show the delayed development of symptoms caused by TBSV infection in the silenced plants. The picture was taken 5 dpi. (D) VIGS-based knockdown of both Snx1 and Snx2b mRNA levels inhibits the accumulation of CIRV RNA in N. benthamiana. Samples for RNA extractions were taken 2.5 days post inoculation from the inoculated leaves. See further details in panel B. Each experiment was performed three times.

Complementation through expression of the Flag-tagged Vps5p from a plasmid in vps5Δ yeast restored efficient replication of both TBSV and CIRV (Fig 1A, lanes 19–21 and S1A Fig). Over-expression of Vps5p in wt yeast did not increase TBSV or CIRV repRNA accumulation (Fig 1A, lanes 7–9 and S1A Fig). Expression of Vps5p lacking either the BAR domain or the PX domain could not complement the replication of the tombusvirus RNAs in vps5Δ yeast (Fig 1A, lanes 13–15 and 22–24, and S1A Fig), suggesting that tombusviruses need both domains of Vps5p for supporting viral replication. Since Vps5p forms either a homodimer or heterodimer with Vps17 [29], we tested TBSV replication in vps17Δ yeast, which supported ~3-fold less TBSV repRNA accumulation (S1B Fig, lanes 7–9). Thus, both Vps5p and Vps17p SNX-BAR proteins are important for tombusvirus replication in yeast.

To explore if tombusviruses depend on the Vps5p orthologs in plants, we knocked down both Snx1 and Snx2b levels via virus-induced gene-silencing (VIGS) in Nicotiana benthamiana plants. The sequence of Snx2a is very similar to Snx2b in Arabidopsis (S2, S3 and S4 Tables). Therefore, VIGS of Snx1/2b likely leads to silencing of both Snx2a (only incomplete sequence is available in N. benthamiana) and Snx2b in N. benthamiana plants. Knock-down of the SNX-BAR proteins led to ~4-fold reduced accumulation of TBSV genomic (g)RNA and CIRV gRNA, respectively (Fig 1B and 1D, lanes 4–6). Knocking down Snx1 and Snx2b levels ameliorated the necrotic symptoms caused by TBSV infection in N. benthamiana (Fig 1C). Similarly, protoplasts (separate cell wall-free plant cells) supported ~75% less replication of TBSV and CIRV gRNA when obtained from Snx1/Snx2b-silenced versus control N. benthamiana plants (S1C and S1D Fig). The replication of the more-distantly related turnip crinkle virus (TCV) was also reduced by ~3-fold in protoplasts obtained from Snx1/Snx2b-silenced versus control N. benthamiana plants (S1E Fig), suggesting that several plant viruses in the Tombusviridae family could take advantage of the SNX-BAR proteins during their replication. However, the pro-viral effects of the co-opted SNX-BAR proteins on tombusvirus replication seem to be specific, because deletion of VPS5 in yeast did not have an adverse effect on the accumulation of the unrelated insect viruses, namely Nodamura virus (NoV) and Flock house virus (FHV) (S1G and S1H Fig).

In addition, we analyzed Snx1 and Snx2b mRNA levels in TBSV and CIRV-infected versus mock-treated N. benthamiana leaves. RT-PCR analysis showed the up-regulation of both Snx1 and Snx2b mRNA levels in tombusvirus-infected leaves (S1F Fig), suggesting that tombusvirus replication induces the expression of SNX1 and SNX2B genes in plants.

Tombusvirus replication proteins interact with and recruit the endosomal SNX-BAR proteins into VROs in yeast and plant cells

To test if Vps5p interacts with the viral replication proteins, we performed co-purification experiments with Flag-tagged viral replication proteins from yeast membranes. The HA-tagged Vps5p was expressed in yeast from its native promoter and natural chromosomal location. Western blot analysis of the purified functional replicase complex identified the co-purified Vps5p (Fig 2A, lanes 2 and 4). Similarly, we were able to co-purify the myc-tagged TBSV p33 replication protein with either the Flag-tagged Snx1 or Snx2b from membrane fraction of plant cells (Fig 2B). Pull-down experiments with purified proteins from E. coli showed the direct binding of the p33 replication protein with the yeast Vps5p and the plant Snx1 and Snx2b in vitro (Fig 2C).

Fig 2

SNX-BAR proteins are present in the tombusviral replicase complex.

(A) Co-purification of the yeast Vps5p with the TBSV VRCs. Top panel: Western blot analysis of co-purified 3xHA-tagged Vps5p (expressed from the native promoter in the original chromosomal location) with Flag-affinity purified p33 from membrane fraction of yeast. Vps5p was detected with anti-HA antibody. The negative control was His₆-tagged p33 purified from yeast extracts using a Flag-affinity column (lanes 1 and 3). The yeast samples were either cross-linked with formaldehyde or not cross-linked. Second panel: Western blot of purified Flag-p33 detected with anti-Flag antibody. Bottom panels: Western blots of Vps5-3xHA, Flag-p33 and His₆-p33 proteins in the total yeast extracts using anti-HA, anti-Flag and anti-His antibodies, respectively. (B) Co-purification of the myc-tagged p33 replication protein with the Flag-tagged Snx1 or Snx2b from detergent-solubilized membrane fraction of N. benthamiana cells. The proteins were co-expressed from plasmids based on agroinfiltration. The plants were also infected with TBSV. (C) A pull-down assay to test direct binding between the p33 replication protein and the shown SNX-BAR proteins expressed in E. coli. The MBP-tagged p33 was immobilized on beads, followed by addition of the affinity-purified GST-tagged SNX-BAR proteins. MBP was used as a negative control. Western blot analysis with the shown antibodies was used to detect the bound or input proteins. (D) Vps5p SNX-BAR is a permanent component of the TBSV replicase complex. Cycloheximide was used to block the formation of new VRCs in wt yeast via inhibition of cellular translation. Top panel: Western blot analysis shows the co-purified Vps5-3xHA with the viral replicase using Flag-based purification from the membrane fraction at the shown time points. Second panel: Western blot analysis of the purified Flag-p33 with anti-Flag antibody. Bottom panels: Western blot analysis of Vps5-3xHA, His₆-p33 and Flag-p33 in the total yeast lysates with anti-HA, anti-His and anti-Flag antibodies, respectively. See further details in panel A. Each experiment was repeated three times.

To gain insights into the subcellular site of interaction between Snx1, Snx2a and Snx2b with the replication proteins, we used bimolecular fluorescence complementation (BiFC) assay. This assay demonstrated that the p33 and p92^pol replication proteins interact with Snx1, Snx2a and Snx2b within the TBSV-induced aggregated peroxisomes, which represent the sites of TBSV replication (Figs 3A and S2A and S2B). The BiFC assay also showed that the CIRV p36 replication protein interacts with Snx1, Snx2a and Snx2b within the CIRV-induced aggregated mitochondria, which represent the sites of CIRV replication (S2C Fig). Confocal microscopy revealed that both Snx1 and Snx2b are recruited separately (S3A and S3B Fig versus S3C Fig), or simultaneously (Fig 3B) to the large TBSV VROs in plant cells and the yeast Vps5p in yeast cells (S3D and S3E Fig). Super-resolution microscopy showed the co-localization of Vps5 with the p33 and p92 replication proteins in yeast cells (Fig 3C). Based on these findings, we propose that TBSV co-opts the SNX-BAR proteins into the VROs through direct interaction with the viral replication proteins in yeast and plant cells.

Fig 3

PI(3)P phosphoinositide is required for recruitment of the endosomal SNX-BAR proteins into the viral replicase.

(A) BiFC assay to detect interaction between p33 replication protein and the SNX-BAR proteins in planta. TBSV p33-cYFP and nYFP-Snx2b, nYFP-Snx1 or nYFP-Snx2a proteins were co-expressed from the 35S promoter after co-agroinfiltration into N. benthamiana leaves. Note that the plants were infected with TBSV to induce VROs in cells. Co-localization of RFP-SKL (peroxisomal luminar marker) with the BiFC signal (see merged image) demonstrates that the interaction between p33 and SNX-BAR proteins occurs in VROs. Scale bars represent 5 μm (top two panels) and 10 μm (bottom panel). (B) Robust co-localization of TBSV p33-BFP with the GFP-tagged Snx1 and RFP-tagged Snx2b in N. benthamiana cells is detected by confocal laser microscopy. (C) Super-resolution laser microscopic images of yeast cells. The yeast cell replicating TBSV repRNA was imaged based on anti-TBSV p33 antibody and anti-Flag antibody for Vps5-Flag. The bars represent 1 μm. Arrows point at the co-localized Vps5p and p33 replication protein. The images were obtained by a Nikon N-STORM Super Resolution Microscope and image processing was performed using NIS-element software. Each experiment was repeated three times. (D) Co-localization of TBSV p33-BFP with the GFP-tagged Snx1 or GFP-tagged Snx2b and with PI(3)P in N. benthamiana protoplasts is detected by confocal microscopy. PI(3)P was detected with anti-PI(3)P antibody. Scale bars represent 20 μm. (E) BiFC assay shows the reduced level of interaction between p33 and the Snx2b protein in N. benthamiana treated with Vps34 PI3K inhibitor, Wortmannin or DMSO as a negative control. TBSV p33-cYFP and nYFP-Snx2b proteins were co-expressed from the 35S promoter after co-agroinfiltration into N. benthamiana leaves. RFP-SKL was expressed as a peroxisomal marker to identify VROs. Scale bars represent 10 μm. On the right, we show the quantitative evaluation of the BiFC signal with the Y-axis showing the relative BiFC signal levels with the control DMSO treatment (in red) representing 100%. The BiFC signals were quantified via Image J. Each experiment was repeated three times.

Next, we tested if Vps5p is a permanent component of the viral replicase complex by purification of the viral replicase from yeast cells at various time points. New VRC formation was halted via blocking yeast translation through cycloheximide. Interestingly, the co-purified Vps5p level has not changed in comparison with Flag-p33 through 2.5 hours (Fig 2D). This finding indicates that Vps5p is a permanent resident in the membrane-bound viral replicase complex. Similar to Vps5p, we found that other co-opted cellular proteins, such as Hsp70s (Ssa1/2 in yeast), Vps4 ESCRT protein, and the DDX3-like Ded1 DEAD-box helicase, are also permanent residents in VRCs [21,38,39]. This is in contrast with other co-opted host factors, such as Vps34 PI3K, the glycolytic Pgk1, or Pex19, which are not present in the assembled VRCs [21,25,40].

PI(3)P is required for the recruitment of the SNX-BAR proteins into VROs in yeast and plant cells

Confocal microscopy of plant protoplasts infected with TBSV revealed the efficient co-localization of PI(3)P (detected with anti-PI(3)P antibody or by using a RFP-2xFYVE biosensor, which specifically binds to PI(3)P in cells) [26,41], and Snx1 and Snx2b with the p33 replication protein at the replication sites (Figs 3D, S4A and S4B).

To test if PI(3)P affects the recruitment of SNX-BAR proteins into VROs, we performed BiFC assay with plant cells treated with Wortmannin, an inhibitor of Vps34 PI3K. The recruitment of both Snx1 (S5A–S5C Fig) and Snx2b (Figs 3E and S5C) into VROs was inhibited up to ~4-to-5 times by the reduced production of PI(3)P in plant cells treated with the PI3K inhibitor. Since PI(3)P is required for the endosomal localization of SNX-BAR proteins [42,43], and the TBSV p33 replication protein drives the recruitment of Rab5-positive endosomes into the TBSV VROs [24], the Wortmannin treatment likely inhibits the co-targeting of p33 and SNX-BAR proteins to the endosomes prior to final targeting to the VROs.

A mutagenesis approach was used to test if PI(3)P binding by SNX-BAR is important for TBSV replication. The conserved YR amino acids in the PI(3)P binding region (PX-domain) of the yeast Vps5p were mutated to AA to block PI(3)P binding [44]. Expression of Vps5^Y321A,R360A mutant in vps5Δ yeast poorly complemented TBSV replication (Fig 4A, lanes 16–18 versus 10–12 and S3D and S3E Fig), whereas Vps5^Y321A,R360A expression in WT yeast led to ~3-fold inhibition of TBSV replication (Fig 4A, lanes 7–9 versus 1–3), suggesting that Vps5^Y321A,R360A functions as a dominant negative mutant, likely as a component of a SNX-BAR dimer. Similar mutagenesis of the PI(3)P binding site in the plant Snx1 (Snx1^rry) inhibited the recruitment into VROs by the TBSV p33 and the CIRV p36 replication proteins in a BiFC assay (Figs 4B and S5D). In addition, phosphorylation mimicking mutant of Snx1 [45], termed Snx1^E, which has low binding affinity to PI(3)P, also interfered with the p33-driven recruitment of Snx1 into VROs, whereas nonphosphorylation mimicking mutant Snx1^A, which has high binding affinity to PI(3)P, was recruited by TBSV similar to the case with wt Snx1 (Fig 4C and 4D). Pull-down experiments with purified recombinant proteins demonstrated that all the above Snx1 mutants still bound to p33 replication protein in vitro (S5E Fig). Therefore, the enrichment of PI(3)P is critical for TBSV to co-opt SNX-BAR proteins into VROs.

Fig 4

Mutations in the endosomal SNX-BAR proteins inhibit TBSV replication and their recruitment into VROs.

(A) Vps5p mutant (Vps5^yr-AA) deficient in PI(3)P-binding does not efficiently complement TBSV replication in vps5Δ yeast. Top panels: Northern blot analysis shows reduced TBSV (+)repRNA accumulation in wt yeast strain expressing Vps5^yr-AA, and poor complementation of TBSV replication in vps5Δ yeast expressing Vps5^yr-AA. Bottom images: Western blot analysis of the level of His₆-p33 with anti-His antibody and Vps5-Flag with anti-Flag antibody, respectively. (B) Decreased interaction between TBSV p33-cYFP replication protein and the nYFP-Snx1^rry mutant protein, which is deficient in PI(3)P-binding, was detected by BiFC in N. benthamiana infected with TBSV. Scale bars represent 5 μm. Bottom image: The Western blots show the accumulation levels of the proteins used in these assays. (C-D) Interaction between TBSV p33-cYFP and the nYFP-Snx1 mutant proteins was detected by BiFC in N. benthamiana infcted with TBSV. Snx1^A is a non-phosphorylation mimicking mutant with high affinity for PI(3)P, whereas Snx1^E is a phosphorylation mimicking mutant with low affinity for PI(3)P. Scale bars represent 5, 5 and 10 μm. In panel D, we show the quantitative evaluation of the BiFC signals seen in panels B and C, with the Y-axis showing the relative BiFC signal levels with the wt Snx1 (in red) representing 100%. The BiFC signals were quantified via Image J. Each experiment was repeated three times. (E) Lack of recruitment of Vps5p into the p33-decorated VROs in yeast missing a retromer component. Vps5p was visualized with anti-Flag antibody. Scale bars represent 5 μm.

Since SNX-BAR proteins bind to PI(3)P and sense positive-membrane curvature, we wondered if it is possible to substitute the SNX-BAR proteins in the VRCs. We used the C-terminal domain of RavZ effector protein from Legionella as a fusion partner to p33 replication protein. RavZ^CT is known to bind to PI(3)P and senses positive-membrane curvature [46], similar to SNX-BAR proteins. Expression of RavZ^CT-p33, however, strongly inhibited TBSV replication in yeast (S5F Fig). Deletion of the PI(3)P-binding module in RavZ^CT-P-p33 resulted in fully functional replication protein (S5F and S5G Fig). These data suggest that the co-opted SNX-BAR protein—PI(3)P interaction is required for TBSV replication and this interaction cannot be replaced by p33—PI(3)P interaction in the form of RavZ^CT-p33.

Next, we tested how Vps5p endosomal protein is delivered to the VROs in wt yeast. Vps5p functions together with the retromer complex, which consists of three conserved proteins, Vps26p, Vps29p and Vps35p. Vps5p and the retromer induce the formation of tubular transport carriers from the endosomes to recycle cargos to the Golgi and ER or to the plasma membrane [29,47]. An analysis of Vps5p distribution revealed that the p33-decorated VROs lacked Vps5p in vps35Δ, vps26Δ or vps29Δ yeasts in contrast with the wt yeast (Fig 4E). We also tested the recruitment of the Snx2b SNX-BAR protein, the plant homolog of the yeast Vps5, into VROs in N. benthamiana cells. We found co-localization of Snx2b and the TBSV p33 replication protein with Vps26, Vps29, and Vps35 retromer complex proteins in the large TBSV VROs (S6A Fig). Snx2b and the components of the retromer complex are also co-localized in the absence of TBSV infection in N. benthamiana, but they are not localized in large punctate structures (S6B Fig). Taken together, it seems that the retromer is involved in facilitating the p33-driven recruitment of the endosomal SNX-BAR membrane-shaping proteins into the tombusvirus VROs. The functional roles of the retromer in TBSV replication will be presented elsewhere.

PI(3)P and the SNX-BAR proteins coordinate the protection of the viral RNAs from ribonucleases in vitro

Although previous works have shown that the ESCRT machinery is co-opted by TBSV to form spherules [21,22], the ESCRT factors might not be enough to render the spherules stable enough for an extended period to provide full protection for the dsRNA replication intermediate. Because local enrichment of PI(3)P results in positive membrane curvature and the SNX-BAR proteins sense, bind and reshape membranes into positive curvature (tube-like forms) [48], these host components are the best candidates to stabilize VRCs, including the neck structure of the tombusvirus VRCs (see model below). The narrow neck structure is proposed to serve as a gate of the VRCs, preventing ribonucleases to enter the VRCs, while allowing ribonucleotides entry and (+)RNA export out of VRCs during replication [4].

To examine if the membranous VRCs in vps5Δ yeast indeed provide reduced protection to the viral RNAs against nucleases, we used several cell-free extract (CFE)-based replicase reconstitution assays, which can efficiently probe the exposure of the replicating viral RNAs to ribonucleases [49,50]. In assay #1, we expressed p33 and p92^pol replication proteins without the TBSV repRNA to allow the replicase pre-assembly step in vps5Δ yeast and in WT yeast as a control (Fig 5A) [50]. Then, we programmed the obtained CFEs with the TBSV (+)repRNA, followed by addition of the micrococcal nuclease (MNase) at the 20 min time point and inactivation of MNase 15 min latter. At the end of the assay, we measured the level of the TBSV dsRNA replication intermediate [produced by minus-strand synthesis on the (+)RNA template], which is always present in the VRCs [51]. The TBSV dsRNA was more sensitive to MNase when the CFEs were prepared from either vps5Δ or vps5Δvps17Δ yeast strains in comparison with the CFE from WT yeast (Fig 5A).

Fig 5

Vps5 SNX-BAR protein and PI(3)P are required for TBSV replication and for the protection of the viral dsRNA in replicase reconstitution assays.

(A) Reduced dsRNA and (+)repRNA production and increased nuclease sensitivity by the tombusvirus replicase assembled in vitro in CFEs prepared from vps5Δ or vps5Δvps17Δ yeast strains. The yeast strains expressed the p33 and p92^pol, allowing the pre-assembly of the replicase complex in yeast. (+)repRNA was used to program the CFEs, which were treated with MNase 20 min latter, followed by the inactivation of MNase with EGTA after 15 min. Non-denaturing PAGE analysis shows the ³²P-labeled TBSV repRNA products, including the (+)repRNA progeny and the dsRNA replication intermediate, made by the reconstituted replicases. We also show the relative protection level of the dsRNA replication intermediate from the MNase treatment by adjusting the “no treatment” to 100% (see numbers in the bottom row). (B) In vitro reconstitution of the TBSV replicase using purified recombinant p33 and p92^pol and in vitro transcribed TBSV DI-72 (+)repRNA. The CFEs were prepared from the shown yeast strains. The MNase treatments, which lasted for 15 min, were done at three different time points as shown. Non-denaturing PAGE analysis was done as in panel A. Note that we loaded only half of the samples from the wt CFE-based assay. (C) The scheme of the novel GUV-based replicase reconstitution assay. (D) Both Vps5p SNX-BAR protein and PI(3)P are required to protect the TBSV dsRNA in a replicase reconstitution assay. GUVs were prepared with ER-like phospholipid composition (see also panel C). The soluble fraction of CFEs prepared from the shown yeast strains together with purified recombinant p33 and p92^pol and TBSV (+)repRNA were added to GUVs to reconstitute the replicase. The dsRNA-specific RNase III was added at the end of the reaction to destroy the unprotected TBSV dsRNA. Non-denaturing PAGE analysis was done as in panel A. (E) The Vps34 PI3K and Ymr1 PI(3)P phosphatase affect TBSV replication and the nuclease sensitivity of the viral dsRNA in vitro. The CFEs were prepared from wt, vps34Δ or ymr1Δ yeast strains to reconstitute the TBSV replicase as shown in panel B above. The MNase treatment and the RNA product analysis were as described in panel B. The panel on the right shows the comparable levels of yeast proteins in the obtained CFEs in an SDS-PAGE analysis.

Additional assay to test the level of protection provided by the VRCs included in vitro replicase reconstitution with purified recombinant viral proteins and (+)repRNA transcripts as schematically shown in Fig 5B. This CFE-based assay supports a single full cycle of RNA replication, including both (-) and (+)RNA synthesis [49]. The MNase was added at different time points (as shown) for 15 min to destroy the unprotected viral RNAs, followed by MNase inactivation with EGTA and TBSV repRNA replication on the protected TBSV dsRNAs up to 3 hours (Fig 5B). The in vitro assembled VRC based on the CFE prepared from vps5Δvps17Δ yeast provided 3-times less protection at the 60 min time point in comparison with the CFE from WT yeast. In the third assay, we isolated the VRCs replicating TBSV repRNA from vps5Δ and WT yeasts, followed by in vitro replication assay in the presence or absence of MNase [52]. The activity of VRC preparation obtained from vps5Δ yeast was inhibited by the MNase by 2-fold more efficiently than the corresponding preparation from WT yeast (S7A and S7B Fig). Altogether, the results from three separate in vitro replication assays with VRCs lacking Vps5p showed that ribonucleases could access the TBSV dsRNA replication intermediate much more efficiently, likely due to incomplete or unstable VRCs. All these data support the model that the co-opted SNX-BAR protein—PI(3)P interaction helps stabilizing the VRCs and protection of the viral dsRNA during TBSV replication.

To obtain additional evidence for the protective role of Vps5p SNX-BAR protein in TBSV replication, we used a replicase reconstitution assay based on giant unilamellar vesicles (GUVs) [53]. The artificially made GUVs allow the selection of the particular lipid composition of membranes, whereas the soluble fraction of the CFE provides the host factors needed for TBSV replication (Fig 5C). Interestingly, in the absence of Vps5p, the reconstituted TBSV replicase could not efficiently protect the TBSV dsRNA replication intermediate from the dsRNA-specific RNase III (Fig 5D, lanes 4 and 6 versus 3 and 5). In contrast, GUVs in combination with the soluble fraction of the WT yeast CFE provided almost complete protection of TBSV dsRNA replication intermediate against the dsRNA-specific RNase III (Fig 5D, lanes 1–2). These experiments also illustrate the power of using GUV-based replicase reconstitution to study the effects of particular lipids in viral replication.

Similar in vitro replicase reconstitution using CFE preparations obtained from vps34Δ yeast, lacking PI3K, provided lesser protection of dsRNA against MNase than WT CFE did (Fig 5E). In contrast, in vitro replicase reconstitution using CFE preparations obtained from ymr1Δ yeast, lacking PI(3)P phosphatase, increased dsRNA protection against MNase in comparison with the WT CFE (Fig 5E). Moreover, WT CFE preparations programmed with purified recombinant viral proteins and (+)repRNA transcripts in the presence of purified recombinant Ymr1 PI(3)P phosphatase to reduce PI(3)P level in membranes by converting PI(3)P to PI [54] showed less protection of the dsRNA replication intermediate against MNase treatment (S7C and S7D Fig). Altogether, all these in vitro data obtained by probing the exposure of the viral dsRNA during viral replication strongly support the protective role of SNX-BAR protein and PI(3)P phosphoinositide to the TBSV dsRNA replication intermediate within VRCs.

Co-opting the SNX-BAR proteins and enrichment of PI(3)P are needed to build RNAi-insensitive tombusvirus replication complexes

To further test if Vps5p SNX-BAR protein is needed to protect the viral RNA in yeast, we have used the reconstituted RNAi machinery from S. castellii with the two-component DCR1 and AGO1 genes [55]. We measured TBSV accumulation when RNAi activity was induced [22] in WT as well as vps5Δ yeasts. Activation of RNAi in vps5Δ yeast led to ~3-fold less TBSV repRNA accumulation, suggesting poor protection of the TBSV RNA in comparison with vps5Δ yeast with suppressed RNAi machinery (Fig 6A, lanes 10–12 versus lanes 7–9). Induction of the RNAi machinery in WT yeast had a minor effect on TBSV accumulation (Fig 6A, lanes 4–6). Based on these data, we suggest that Vps5p SNX-BAR protein is involved in protecting the viral dsRNA replication intermediate within VRCs.

Fig 6

Depletion of SNX-BAR proteins or PI(3)P renders tombusvirus RNA sensitive to RNAi-based degradation in yeast and plants.

(A) Co-expression of S. castellii Ago1 and Dcr1 proteins in vps5Δ yeast reduces TBSV repRNA accumulation to a larger extent than in wt yeast (BY4741). Replication of the TBSV repRNA was measured by northern blotting 16 h after initiation of TBSV replication. The accumulation level of repRNA was normalized based on the ribosomal (r)RNA. Each sample is obtained from different yeast colonies. Yeast strain not expressing RNAi components is chosen as 100%. Each experiment was repeated three times. (B) Inhibition of Vps34 PI3K with chemical inhibitors reduces TBSV (+)repRNA and (-)repRNA accumulation. Dcr1 was expressed in all yeast in these experiments. (C) Northern blot detection of vsiRNA(-) and vsiRNA(+) in the combined samples presented in panel B. The ³²P-labeled TBSV DI-72 (+)RNA and (-)RNA, respectively, were used as probes. (D-E) VIGS-based knock-down of both Snx1 and Snx2b mRNA levels inhibits the accumulation of CNV^20Kstop RNA in N. benthamiana. The accumulation of CNV^20Kstop gRNA and sgRNAs was measured using northern blot analysis of total RNA samples obtained from N. benthamiana inoculated (3 dpi) (D) or systemically-infected (E) leaves (6 dpi). Note that CNV^20Kstop used for infection does not express p20 silencing suppressor protein in these samples, thus allowing the full effect of the RNAi response. See further details in Fig 1B.

To test the role of PI(3)P phosphoinositide in the protection of the TBSV dsRNA replication intermediate, we measured vsiRNA abundance in yeast treated with Vps34p PI3K inhibitors in comparison with the DMSO-treated wt yeast expressing Dcr1 (Fig 6B). Interestingly, the abundance of vsiRNA(-) and vsiRNA(+), the products of Dcr1, did not change in yeast treated with PI3K inhibitors versus DMSO (Fig 6C lanes 2–3 versus 1). This observation is in contrast with the ~4-to-6-fold reduction of the target viral RNA level in yeasts treated with PI3K inhibitors versus DMSO (Fig 6B lanes 4–9 versus 1–3). Altogether, the data suggest that depletion of PI(3)P in yeast makes the TBSV VRCs highly accessible to Dcr1, resulting in efficient production of vsiRNAs.

To test if Snx1 and Snx2b proteins are needed to protect the viral RNA in plants, we knocked down Snx1 and Snx2b levels via VIGS and used CNV^20Kstop missing the p20 silencing suppressor to allow for efficient RNAi activity during infection. CNV^20Kstop accumulation was greatly suppressed in the inoculated leaves and CNV^20Kstop was close to undetectable in systemic leaves of N. benthamiana (Fig 6D and 6E). Knock down of the Snx1/Snx2b levels in N. benthamiana had a lesser inhibitory effect on other tombusviruses expressing the p19 suppressor of gene silencing (Fig 1B and 1D). Based on these data, we propose that tombusviruses require SNX-BAR proteins to protect their replicating RNAs against RNAi response in plants.

The SNX-BAR proteins bind to the viral RNA in tombusvirus replication complexes

We hypothesized that SNX-BAR proteins would likely contact the viral (+)RNA, while the RNA passes through the tight neck structure during exit from the spherule/VRC. Therefore, we tested if SNX-BAR proteins could bind to the viral RNA. In the first test, we used ³²P-labeled (+)repRNA and (-)repRNA of TBSV and purified yeast Vps5p, Snx1 (S7E and S7F Fig) and Snx2b (Fig 7A and 7B) in a gel mobility shift assay. All three SNX-BAR proteins bound to the (+)RNA and (-)RNA in vitro, indicating that these host proteins are indeed RNA-binding proteins.

Fig 7

The co-opted SNX-BAR proteins bind to the TBSV (+)RNA.

(A-B) In vitro RNA binding assay with purified Snx2b using ³²P-labeled ssRNA templates. The assay contained the (-) or (+) DI-72 repRNA (~0.1 pmol) plus increasing amounts (1, 3, 6, 15, and 30 pMol) of purified recombinant MBP-tagged Snx2b or MBP as a negative control. RNA gel shift analysis was conducted on nondenaturing 5% PAGE that separated the free or Snx2b-bound ssRNAs. (C) Co-purification of the TBSV (+)gRNA with Snx1 or Snx2b from N. benthamiana replicating TBSV. The membrane fraction of plants expressing Flag-tagged Snx1 or Snx2b was solubilized with Triton-X100, followed by Flag-affinity purification. The agarose gel image shows the RT-PCR analysis of the co-purified (+)gRNA (top panel) or (-) gRNA (middle panel). Bottom panel: Western blot analysis of the Flag-affinity purified proteins from the samples shown above using anti-Flag antibody. (D) Scheme of the trimolecular fluorescence complementation (TriFC)-based Snx1-repRNA interaction. The repRNA [repRNA(+)MS2hp] carries the MS2 hairpin sequence that is specifically recognized by MS2-CP-cYFP. Binding of nYFP-Snx1 to the repRNA and simultaneous binding of MS2-CP-cYFP to the MS2-hairpins present in the repRNA is expected to generate TriFC signals as shown schematically. (E) TriFC analysis of Snx1 binding to the repRNA(+)MS2hp or repRNA(-)MS2hp (the MS2 hairpin folds on the minus-stranded repRNA). Bottom panel: repRNA lacking the MS2 hairpin sequence is used as a negative control. (F) Co-purification of Vps5p SNX-BAR protein with the help of repRNA(+)MS2hp serving as an RNA bridge. Purified recombinant GST-tagged MS2-CP was immobilized on the GST resin. Then, the detergent-solubilized yeast lysates containing Flag-p33, Flag-p92 and Vps5-3xHA and the shown repRNA derivatives were applied to the column. Please find the scheme for this assay in S7G Fig. Western blot analysis was used to document the captured Vps5-3xHA and Flag-p33, respectively. (G) A model showing the proposed roles of PI(3)P and the SNX-BAR proteins in stabilizing the neck structure of a spherule harboring the TBSV VRC. Enrichment of PI(3)P by the co-opted Vps34 PI3K in coordination with binding of the SNX-BAR dimer reshapes the membrane area into positive curvature. The body of the spherule is formed with the help of the co-opted ESCRT proteins. The viral (+)RNA is likely contacted by the SNX-BAR proteins during the exit of the viral RNA out of the spherule. (H) A model showing the increased ribonuclease sensitivity of the VRC if the neck structure is not stabilized with the contribution of PI(3)P and the SNX-BAR proteins.

To demonstrate if similar interaction could occur in plant cells, we cross-linked plant samples replicating the TBSV gRNA, followed by purification of Flag-tagged Snx1 and/or Snx2b. RT-PCR analysis of the RNAs isolated from purified Snx1, Snx2b and Snx1/Snx2b preparations with strand-specific primers showed the presence of the co-purified TBSV (+)gRNA, but not the (-)RNA (Fig 7C). In the second test, we used a three-component TriFC-based assay (Fig 7D). The RNA component was a modified repRNA carrying an ssRNA sensor consisting of six repeats of a hairpin RNA (MS2hp) from MS2 bacteriophage, which is specifically recognized by the MS2 coat protein (MS2-CP) (Fig 7D) [56]. Co-expression of the repRNA(+)MS2hp and MS2-CP-cYFP and nYFP-Snx1 with TBSV p33 and p92^pol replication proteins leads to replication of the repRNA. If nYFP-Snx1 could bind to the repRNA, whereas MS2-CP-cYFP could bind to the MS2-hairpins present in the repRNA, then we expect TriFC signal as shown schematically in Fig 7D. Interestingly, we found many plant cells (93%) with the TriFC signals, where p33-BFP marked the sites of TBSV replication (Fig 7E, top panel). In contrast, it was difficult to find TriFC signals when the MS2 hairpins formed on the (-)repRNA, although a few cells (only 9%) had weak TriFC signals as shown in Fig 7E. In the control experiments, the TBSV repRNA replaced repRNA(+)MS2hp and there was no TriFC signal within the plant cell compartments marked by p33-BFP (Fig 7E, bottom panel). Therefore, we conclude that repRNA(+)MS2hp was required as an “RNA bridge” to allow the proximal positioning of MS2-CP-cYFP and nYFP-Snx1 within the membranous TBSV VROs marked by p33-BFP. Therefore, we suggest that either Snx1 is located in the proximity of the TBSV (+)repRNA or Snx1 binds to the viral (+)RNA during viral replication.

To confirm that the yeast Vps5p also interacts with the viral RNA in vivo, we co-expressed repRNA(+)MS2hp with p33 and p92^pol replication proteins (to support replication) and HA-tagged Vps5p in yeast. This was followed by capturing repRNA(+)MS2hp from yeast with the GST-tagged MS2-CP (S7G Fig). Interestingly, we detected both Vps5p and the p33 replication protein in the eluate from the GST resin (Fig 7F, lane 2). When we expressed repRNA(-)MS2hp under the above conditions, then the eluate from the GST resin with the bound GST-MS2-CP contained only p33, albeit to a lesser amount than that obtained with repRNA(+)MS2hp (Fig 7F lanes 3 versus 2). In the control experiments, expression of the replication proteins with the TBSV repRNA, lacking the (+)MS2hp sequence, did not result in the co-purification of Vps5p and the p33 replication protein with the GST-MS2-CP (Fig 7F, lane 1). Based on these results, we suggest that Vps5p binds to the viral (+)RNA within the VRCs, either directly or in complex with the p33 replication protein, during viral replication in yeast.

Yeast stains

Parental yeast strain BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0), and single knock out strains vps5Δ, vps17Δ, vps34Δ, vps26Δ, vps29Δ, vps35Δ and ymr1Δ were purchased from Open Biosystems. SC1 (MATa his3Δ1 leu2Δ trp1Δ289 uraΔ52) yeast strain was purchased from Invitrogen. vps5Δvps17Δ double deletion strain and yeast strain (BY4741: Vps5-3xHA) with chromosomal HA tagging of Vps5 was created by using yeast toolbox plasmids [64].

Plant and yeast expression plasmids

Plasmids and primers used in this study are described in S5 Table.

Analysis of viral replication in yeast and plant

To identify the function of yeast Vps5 and Vps17 SNX-BAR proteins in the replication of TBSV and CIRV, yeast strains BY4741, vps5Δ and vps17Δ were transformed with plasmids pESC-His-p33/DI72, pYES-His-p92 and pRS315-cFlag for TBSV replication, pESC-Strep-p36/DI72, pYES-Strep-p95 and pRS315-cFlag for CIRV replication. The transformed yeast cells were pre-grown in synthetic complete medium lacking uracil, leucine and histidine (ULH^-) supplemented with 2% glucose at 29°C for overnight, then tombusviral repRNA replication was induced by transferring the yeast to synthetic complete medium (ULH^-) supplemented with 2% galactose at 23°C for 24 h for TBSV or 30 h for CIRV. Yeast total RNA and total protein were isolated and analyzed by Northern blotting and Western blotting, respectively [14,25]. The functions of yeast Vps5 mutants in tombusviral replication were tested by using the same methods.

To measure the effect of plant SNX-BAR proteins on virus replication, NbSnx1 and NbSnx2b gene expression were silenced using tobacco rattle virus (TRV)-mediated virus-induced gene silencing (VIGS) method in N. benthamiana [65]. The NbSnx1 and NbSnx2b fragments were ligated together into the pTRV2 vector. The upper NbSnx1/2b-silenced leaves were inoculated with TBSV, CIRV or CNV^20KStop sap 12 day post agroinfiltration. The control plants were treated the same way, except using TRV-cGFP (The C terminus of GFP ORF was inserted into pTRV2 vector to prepare this control plasmid). Plant leaf discs from inoculated leaves and systemic leaves were collected for total RNA extraction and viral RNA detection by Northern blotting [66].

Plant protoplasts were isolated from NbSnx1/2b-silenced or cGFP-control N. benthamiana leaves [67]. About 5 x 10⁵ protoplasts were transformed with in vitro transcribed full-length TBSV, CIRV and TCV genomic RNAs, incubated in 35 x 10 mm petri dishes in dark at room temperature for 24 h. Total RNAs were extracted from these protoplasts, viral RNA accumulation was detected by Northern blotting [68].

Confocal microscopy analysis of plant and yeast cells

To examine the subcellular localization of plant SNX-BAR proteins upon virus infection, N. benthamiana leaves were co-infiltrated with agrobacterium carrying plasmids pGD-AtSnx1-GFP (0.3 OD₆₀₀) or pGD-AtSnx2b-GFP (0.3 OD₆₀₀), pGD-RFP-SKL (0.2 OD₆₀₀) together with pGD-p33-BFP (0.3 OD₆₀₀) and p19 (0.1 OD₆₀₀). Alternatively, pGD-p33-BFP, pGD-AtSnx1-GFP and pGD-AtSnx2b-RFP were co-agroinfiltrated into N. benthamiana. The agroinfiltrated leaves were inoculated with TBSV sap at 14 h post agroinfiltration. After 2 dpi, the agroinfiltrated leaves were subjected to confocal microscopy (FV1200 confocal laser scanning microscope, Olympus) using 405 nm laser for BFP, 488 nm laser for GFP and 559 nm for RFP. Images were captured successively and merged using the FLUOVIEW software [25].

To examine the co-localization of plant sorting nexins with the retromer complex upon virus infection, N. benthamiana leaves were co-infiltrated with agrobacterium carrying plasmids pGD-AtSnx2b-RFP (0.3 OD₆₀₀), pGD-p33-BFP (0.3 OD₆₀₀) and p19 (0.1 OD₆₀₀) together with pGD-AtVps26-GFP(0.3 OD₆₀₀) or pGD-AtVps29-GFP(0.3 OD₆₀₀) or pGD-AtVps35-GFP(0.3 OD₆₀₀). The agroinfiltrated leaves were subjected to confocal microscopy as above.

To analyze the subcellular localization of Vps5 in yeast cells upon tombusviral replication, pYes-His-p92, pEsc-GFP-p33/DI72 and pRS315-Vps5-Flag plasmids were transformed into BY4741 yeast. The transformed yeast cells were pre-grown in synthetic complete medium (ULH^-) supplemented with 2% glucose at 29°C overnight. Tombusviral repRNA replication was induced by changing the yeasts to synthetic complete medium (ULH^-) supplemented with 2% galactose for 21 h at 23°C. Yeast cells were collected to isolate spheroplasts for immunofluorescence [25]. Vps5-Flag was visualized with anti-Flag mouse antibody (Sigma-Aldrich, Cat#F1804), and Alexa 568 secondary antibody (Thermo Fisher Scientific, Cat#A11031). Confocal microscopy analysis using 488 nm laser for GFP and 559 nm for Alexa 568 fluorescent dye in an Olympus FV1200 confocal laser scanning microscope.

To analyze the subcellular localization of Vps5p in yeast retromer deletion stains upon tombusviral replication, pYes-His-p92, pEsc-GFP-p33/DI72 and pRS315-Vps5-Flag were transformed into vps26Δ, vps29Δ and vps35Δ yeast. Immunofluorescence analysis was conducted as above.

Yeast strain BY4741 was transformed with pESC-His-p33/DI72, pYES-His-p92 and pRS315-Vps5-Flag plasmids. Co-localization between p33 and Vps5 was analyzed by super-resolution microscopy (N-STORM Super Resolution Microscopy, Nikon). Anti-p33 monoclonal mouse antibody and anti-Flag rabbit antibody (Sigma-Aldrich, Cat#F7435) were used to visualize p33 and Vps5, respectively [69].

To identify the interaction between SNX-BAR proteins and tombusviral replication proteins in vivo, bimolecular fluorescence complementation (BiFC) assay was performed in N. benthamiana [24]. The plasmids pGD-p33-cYFP or pGD-p36-cYFP or pGD-p92-cYFP, pGD-nYFP-GST, pGD-RFP-SKL and pGD-CoxIV-RFP were transformed to Agrobacterium strain C58C1, as well as pGD-nYFP-AtSnx1, pGD-nYFP-AtSnx2b and SNX-BAR protein mutants. The obtained Agrobacterium transformants were co-infiltrated (0.3 OD₆₀₀, each) into the leaves of four weeks-old N. benthamiana plants. Agroinfiltrated leaves were inoculated with TBSV or CIRV 14 h after agroinfiltration. Plant samples were subjected to confocal laser microscopy at 48 h post virus inoculation.

To observe the distribution and association between cellular PI(3)P and SNX-BAR proteins upon tombusvirus replication in plant cells, pGD-AtSnx1-GFP (0.3 OD₆₀₀) or pGD-AtSnx2b-GFP (0.3 OD₆₀₀), together with pGD-p33-BFP (0.3 OD₆₀₀) were agroinfiltrated into N. benthamiana leaves. The agroinfiltrated leaves were inoculated with TBSV sap at 12 h post agroinfiltration. Protoplasts were isolated from the infiltrated leaves with enzyme solution containing 1.5% (wt/vol) Cellulase R10 (Yakult Pharmaceutical Ind. Co., Ltd., Japan) and 0.4% (wt/vol) Macerozyme R10 (Yakult Pharmaceutical Ind. Co., Ltd., Japan) 48 h post virus inoculation for immunofluorescence detection of PI(3)P [25]. The permeabilized cells were incubated with purified anti-PI(3)P mouse antibody (Echelon Biosciences Inc. Cat#Z-P003), and after washing steps, incubated with anti-mouse secondary antibody conjugated with Alexa Fluor 568 (Thermo Fisher Scientific, Cat#A11031). The images were captured with Olympus FV1200 confocal laser scanning microscope. RFP-2xFYVE was used as a PI(3)P biosensor to visualize PI(3)P distribution upon virus replication in plant protoplast system [25].

To investigate the role of PI(3)P in the recruitment of plant SNX-BAR proteins into viral replication compartments, the plasmids pGD-nYFP-AtSnx1, pGD-nYFP-AtSnx2b, pGD-p33-cYFP and pGD-RFP-SKL were separately transformed into Agrobacterium strain C58C1. The obtained Agrobacterium transformants were co-infiltrated (OD₆₀₀ 0.3 for each) into the leaves of four week-old N. benthamiana plants. Agroinfiltrated leaves were treated with 66 μM Wortmannin (an Vps34 PI3K inhibitor) 15 h after agroinfiltration. Plant samples were subjected to confocal laser microscopy at 31 h post inhibitor treatment.

Protein-protein interaction between SNX-BAR proteins and viral replication protein p33 by co-purification and pull down assays

To test the interaction between p33 and SNX-BAR proteins in plant, TBSV p33-myc was co-expressed with AtSnx1-Flag or AtSnx2b-Flag or GFP-Flag in plant by agroinfiltration (0.3 OD₆₀₀, each), followed by TBSV sap inoculation 12 h post agroinfiltration. Membrane fraction was isolated from the leaves 2 dpi and solubilized with 1% Triton X-100 followed by purification of AtSnx1-Flag or AtSnx2b-Flag with anti-Flag M2 agarose (Sigma-Aldrich, Cat#A2220). The co-purified p33-myc protein was detected by anti-myc antibody [25].

To dissect whether Vps5 SNX-BAR protein is a permanent component of the tombusvirus replicase complex, pEsc-URA-Vps5-3xHA was transformed into yeasts with pGAD-Cup1-Flag-p92 and pGBK-Cup1-Flag-p33-Gal1-DI72. pGAD-Cup1-His-p92 and pGBK-Cup1-His-p33-Gal1-DI72 were transformed into yeast as the negative control. The transformed yeast cells were pre-grown in synthetic complete medium lacking uracil, leucine and histidine (ULH^-) supplemented with 2% glucose and 100 μM BCS at 29°C overnight, followed by transferring the yeasts to synthetic complete medium (ULH^-) supplemented with 2% galactose and 100 μM BCS at 29°C for 24 h. Tombusviral repRNA replication was induced by changing the media to synthetic complete medium (ULH^-) supplemented with 2% galactose and 50 μM CuSO₄ for 3 h at 23°C, 100 μg/ml cycloheximide was added to the culture to inhibit protein translation [21]. Yeast samples were collected at 0 h, 1 h and 2.5 h after addition of cycloheximide. Cross-linking of yeast cells was conducted by suspending yeast pellets with 1x PBS buffer containing 1% formaldehyde and incubation for 1 h on ice. Then, glycine (to 0.1 M) was added to quench the extra formaldehyde and the yeasts were washed and collected by centrifugation. Yeast cells were broken with glass beads, membrane fraction was solubilized with 1% Triton X-100 followed by purification of Flag-p33 with anti-Flag M2 agarose (Sigma-Aldrich, Cat#A2220). The co-purified protein was detected with anti-HA antibody.

We also performed the co-purification assay using yeast expressing Vps5-3xHA from its natural promoter and the original chromosomal location in wt yeast. Yeast expressing Vps5-3xHA was transformed with pGAD-Cup1-Flag-p92 and pGBK-Cup1-Flag-p33-Gal1-DI72. pGAD-Cup1-His-p92 and pGBK-Cup1-His-p33-Gal1-DI72 were transformed into yeast as the negative control. The procedure of yeast culture and purification was the same as above.

To investigate whether the association between p33 and SNX-BAR proteins is direct or indirect, MBP-p33, GST-Vps5, GST-AtSnx1 and GST-AtSnx2b were obtained from E.coli (BL21 DE3 Codon Plus cells) [70]. First, we incubated 2 μg MBP-p33 with amylose resin (New England Biolabs, Cat#E8021L) at 4°C for 2 h. The same amount of MBP protein was used as the control. The unbound proteins were removed by repeated washing, followed by adding 1 μg GST-Vps5, GST-AtSnx1 and GST-AtSnx2b, respectively, onto amylose resin and rotating at 4°C for 4 h. The bound proteins were eluted with maltose elution buffer, and the co-purified proteins were detected by Western blotting [71].

Analysis of viral replication in yeast expressing Ago1 and Dcr1

To measure if Vps5 contributes to the protection of the viral RNA, we co-expressed Dcr1 and Ago1 proteins from S. castellii as an intracellular RNAi probe in yeast as described [22]. Details can be found in S1 Text.

In vitro cell-free extract-based replicase reconstitution assays

CFEs were prepared from untransformed BY4741, vps5Δ, vps5Δvps17Δ, vps34Δ and ymr1Δ yeast strains as described previously [15,22,49]. Reaction mixture for the in vitro replication contained 2 μl of CFE, 0.15 μg DI-72 (+)RNA, 400 ng affinity-purified MBP-p33, 400 ng affinity-purified MBP-p92^pol in 20 μl total volume. 0.1 U/μl micrococcal nuclease (MNase, Amersham Biosciences, Cat#E70196Y) was added to the reactions at different time points as shown in Figures, followed by incubation for 15 min at 25°C, then, 2.5 mM EGTA was added to the samples to inactivate the MNase. The CFE reactions were further incubated for a total of 3 h at 25°C.

The second in vitro replicase reconstitution assay utilized the CFE from wild-type yeast was pre-treated with purified Ymr1 PI(3)P phosphatase at 25°C for 50 min [25], then 0.15 μg DI-72 (+)RNA, 400 ng affinity-purified MBP-p33, 400 ng affinity-purified MBP-p92^pol were added to the reaction mixture in 20 μl total volume, 0.1 U/μl MNase was added to the reactions at different time points as shown in Figures, the reaction mixtures were incubated for 15 min at 25°C, then, 2.5 mM EGTA was added to the samples to inactivate the MNase. The CFE reactions were further incubated for a total of 3 h at 25°C.

The third in vitro replicase reconstitution assay utilized BY4741, vps5Δ and vps5Δvps17Δ yeast strains, which were transformed with pGAD-CUP1-His-p92 and pGBK-CUP1-His-p33. Expression of p33 and p92^pol was induced by adding 50 μM CuSO₄ for 40 min to the medium. CFEs were separately prepared from the above yeasts. The CFE-based reaction mixtures were programmed with 0.5 μg DI-72 (+)RNA transcripts as described previously [22,49]. 0.1 U/μl MNase was added to the reactions at 20 min post incubation. The reaction mixtures were incubated for 15 min at 25°C, then, 2.5 mM EGTA was added to the samples to inactivate the MNase. The CFE-based replication mixtures were incubated at 25°C for a total of 3 h.

The fourth in vitro assay was based on yeast strains BY4741 and vps5Δ, which were transformed with pGAD-CUP1-His-p92 and pGBK-CUP1-His-p33-GAL1-DI72. CFEs were separately prepared from the above yeasts. MNase treatment was performed as describe above. The CFE-based replication mixtures were incubated at 25°C for a total of 3 h.

The total RNAs were isolated from the CFE-based reactions, the ³²P-labeled RNA products were separated in 5% semi-denaturing polyacrylamide gel containing 8 M urea [22,49].

Protein-RNA interaction between SNX-BAR proteins and viral RNA in plant and yeast cells

To investigate the interaction between Vps5 SNX-BAR protein and viral RNA within viral replicase complex in yeast, MS2-tagged RNA affinity purification assay was performed [72]. pGAD-Cup1-Flagp92, pGBK-Cup1-Flagp33, pESC-URA-Vps5-3xHA with pESC-TRP-repRNA or pESC-TRP-repRNA(+)MS2hp or pESC-TRP-repRNA(-)MS2hp were transformed into yeast SC1 strain. The transformed yeasts were pre-grown in synthetic complete medium (ULHT^-) supplemented with 2% glucose and 100 μM BCS at 29°C overnight. This was followed by transferring the yeasts to synthetic complete medium (ULHT^-) supplemented with 2% galactose and 100 μM BCS at 29°C for 24 h. Tombusviral repRNA replication was induced by changing the media to synthetic complete medium (ULHT^-) supplemented with 2% galactose and 50 μM CuSO₄ for 12 h at 23°C. The tombusviral repRNA replication in yeasts was confirmed by Northern blotting. To do the RNA-based purification, yeast cells were broken with glass beads, membrane fraction was solubilized with 1% Triton X-100 followed by incubation with GST resin coated with GST-MS2-CP (purified from E. coli), and then, the GST resins were washed five times. The co-purified proteins were detected by anti-HA and anti-Flag antibody for Vps5-3xHA and Flag-p33, respectively.

To test if SNX-BAR proteins associate with viral RNAs within the viral replicase complex in plants, trimolecular fluorescence complementation (TriFC) was conducted in N. benthamiana [73,74]. MS2-CP binding RNA hairpin sequence (MS2hp) was fused with viral repRNA to get the pGD-35S-repRNA(+)MS2hp and pGD-35S-repRNA(-)MS2hp constructs [75]. MS2-CP ORF was inserted into pGD-cYFP to make pGD-MS2-CP-cYFP plasmid. All the plasmids were transformed into Agrobacteria C58C1 strain. pGD-MS2-CP-cYFP (0.1 OD₆₀₀), pGD-nYFP-Snx1 (0.1 OD₆₀₀), pGD-p33-BFP (0.3 OD₆₀₀), pGD-RFP-SKL (0.2 OD₆₀₀), pGD-p33 (0.3 OD₆₀₀), pGD-p92 (0.3 OD₆₀₀) and pGD-repRNA (0.3 OD₆₀₀) or pGD-35S-repRNA(+)MS2hp (0.3 OD₆₀₀) or pGD-35S-repRNA(-)MS2hp (0.3 OD₆₀₀) were co-agroinfiltrated into N. benthamiana. The plant leaves were studied by confocal microscopy 60 h post agroinfiltration, using confocal laser scanning microscope.

An RNA-based immuno-precipitation method was used to confirm the interaction between SNX-BAR proteins and viral RNA within the viral replicase complex in planta [76,77]. Agrobacterium harboring pGD-AtSnx1-Flag (0.3 OD₆₀₀) or pGD-AtSnx2b-Flag (0.3 OD₆₀₀) were infiltrated into in N. benthamiana leaves, and in parallel, pGD-AtSnx1-Flag (0.3 OD₆₀₀) and pGD-AtSnx2b-Flag (0.3 OD₆₀₀) were co-agroinfiltrated into plant leaves. pGD empty vector served as the negative control. TBSV sap inoculation was done 12 h post agroinfiltration. The leaf strips from the agroinfiltrated leaves (2 dpi) were cross-linked with 1% formaldehyde at 23°C for 1 h, then formaldehyde was quenched by adding 0.1 M glycine. Afterwards, the plant leaf strips were washed in 1xPBS buffer. The membrane fraction was isolated and solubilized with 1% Triton X-100. Flag-based purification of AtSnx1-Flag or AtSnx2b-Flag was performed with anti-Flag M2 agarose. The purified protein was detected with Western blot using anti-Flag antibody. Total RNAs were extracted from the purified protein-RNA complexes by phenol/chloroform, and then, cDNAs were synthesized for the detection of the co-purified viral genomic RNAs by RT-PCR. The PCR reactions were conducted with the same amount of samples and using the same number of PCR cycles for detecting the viral (+)gRNA and (-)gRNA, respectively [21].