|Home | About | Journals | Submit | Contact Us | Français|
Membranous structures derived from various organelles are important for replication of plus-stranded RNA viruses. Although the important roles of co-opted host proteins in RNA virus replication have been appreciated for a decade, the equally important functions of cellular lipids in virus replication have been gaining full attention only recently. Previous work with Tomato bushy stunt tombusvirus (TBSV) in model host yeast has revealed essential roles for phosphatidylethanolamine and sterols in viral replication. To further our understanding of the role of sterols in tombusvirus replication, in this work we showed that the TBSV p33 and p92 replication proteins could bind to sterols in vitro. The sterol binding by p33 is supported by cholesterol recognition/interaction amino acid consensus (CRAC) and CARC-like sequences within the two transmembrane domains of p33. Mutagenesis of the critical Y amino acids within the CRAC and CARC sequences blocked TBSV replication in yeast and plant cells. We also showed the enrichment of sterols in the detergent-resistant membrane (DRM) fractions obtained from yeast and plant cells replicating TBSV. The DRMs could support viral RNA synthesis on both the endogenous and exogenous templates. A lipidomic approach showed the lack of enhancement of sterol levels in yeast and plant cells replicating TBSV. The data support the notion that the TBSV replication proteins are associated with sterol-rich detergent-resistant membranes in yeast and plant cells. Together, the results obtained in this study and the previously published results support the local enrichment of sterols around the viral replication proteins that is critical for TBSV replication.
IMPORTANCE One intriguing aspect of viral infections is their dependence on efficient subcellular assembly platforms serving replication, virion assembly, or virus egress via budding out of infected cells. These assembly platforms might involve sterol-rich membrane microdomains, which are heterogeneous and highly dynamic nanoscale structures usurped by various viruses. Here, we demonstrate that TBSV p33 and p92 replication proteins can bind to sterol in vitro. Mutagenesis analysis of p33 within the CRAC and CARC sequences involved in sterol binding shows the important connection between the abilities of p33 to bind to sterol and to support TBSV replication in yeast and plant cells. Together, the results further strengthen the model that cellular sterols are essential as proviral lipids during viral replication.
Subcellular membranes are critical for replication of plus-stranded (+)RNA viruses, which takes place in membrane-bound viral replicase complexes (VRCs) in the cytoplasm of infected cells. The replication proteins of (+)RNA viruses have to interact with various lipids present in intracellular membranes, including the endoplasmic reticulum (ER), mitochondria, peroxisomes, and endosomal membranes, to aid the replication process. Overall, the recruited membranes are thought to facilitate virus replication by providing platforms to assemble the VRCs, which concentrate viral and host components and protect the viral RNA and proteins from nucleases and proteases. The major activity of the VRCs is viral RNA synthesis leading to the generation of abundant (+)RNA progeny (1,–4).
Although the important roles of co-opted host proteins in RNA virus replication have been appreciated for a decade (1, 4), the equally important functions of cellular lipids in virus replication have been gaining full attention only recently (5,–7). Indeed, the replication proteins of several (+)RNA viruses have been shown to bind to various lipids, recruiting those lipids or cellular enzymes involved in lipid synthesis, or modification, to the site of replication (3, 6). Remodeling of intracellular membranes within viral replication compartments also depends on replication protein-lipid interactions and recruitment of host proteins affecting membrane curvature, as shown for recruitment of ESCRT and reticulon proteins to the replication compartments (8,–11).
Tomato bushy stunt virus (TBSV) is a small (+)RNA virus used to study virus replication, recombination, and virus-host interactions (12). Genome-wide screens and global proteomics approaches in model host yeast (Saccharomyces cerevisiae) have led to the identification of ~500 host genes/proteins with putative roles in TBSV replication or recombination (12, 13). In addition to the usurped host proteins, TBSV replication also depends on two viral replication proteins (14). The abundant auxiliary p33 replication protein, which is an RNA chaperone, is the master regulator of TBSV replication. The essential p33 interacts not only with the viral RNAs and the viral p92pol replication protein but also with numerous host proteins (12, 13, 15, 16). Tombusvirus p33 is an integral membrane protein that binds to various phospholipids (17). Among these, phosphatidylethanolamine (PE) is enriched at the sites of TBSV replication (17), which takes place on the cytosolic surface of peroxisomal membranes (18). PE is required for TBSV replication in an artificial vesicle (liposome)-based in vitro assay (17). Moreover, the PE level is increased during TBSV replication in yeast and plant cells, and a high PE level in yeast via modulation of phospholipid biogenesis genes also leads to enhanced TBSV replication (17, 19).
However, in addition to phospholipids, sterols are also critical for TBSV replication based on genetic approaches that lead to reduced sterol levels or applying sterol synthesis inhibitors (20, 21). We have also shown efficient redistribution of sterols during TBSV replication from the plasma membrane to internal membranes, including the site of TBSV replication. Co-opted cellular oxysterol-binding protein (OSBP)-related proteins (ORPs; called Osh proteins in yeast) and VAMP/synaptobrevin-associated protein (VAP) facilitate the localized enrichment of sterols at the replication sites via the formation of membrane contact sites (MCSs) in yeast and plant cells (20). The incorporation of sterols into PE-based liposomes also enhanced TBSV replication in vitro (20). Therefore, sterols seem to play essential roles in TBSV replication by affecting several facets of the replication process, including the stability of the viral replication proteins, VRC assembly, and the efficiency of viral replication (20, 21).
One intriguing aspect of viral infections is their dependence on efficient assembly platforms serving replication, virion assembly, or virus egress via budding out of infected cells (22). It is proposed for several viruses that these processes depend on lipid rafts, which are specialized domains in membranes enriched in proteins, sterols, and sphingolipids (23). These lipid rafts are resistant to detergents; thus, they are also called detergent-resistant membranes (DRMs). Another model is based on the existence of membrane microdomains, which are heterogeneous and highly dynamic nanoscale structures usurped by various viruses (24, 25). Membrane microdomains are likely formed during TBSV replication, because the viral replication compartment contains numerous stable vesicle-like structures (membrane invaginations with a narrow neck structure, called spherules) enriched in sterols and PE, numerous co-opted host proteins, and large amounts of replication proteins (5, 18, 20).
We investigated the role of sterols in tombusvirus replication, and here we show that viral p33 and p92 are sterol-binding proteins in vitro. We define cholesterol recognition/interaction amino acid consensus (CRAC) and CARC-like sequences (26) within the two transmembrane (TM) domains of p33 involved in sterol binding. Mutagenesis analysis of p33 shows the critical role of sterol binding in the ability of p33 to support TBSV replication in yeast and plant cells. We also show the enrichment of sterols in the detergent-resistant membrane fraction obtained from TBSV-replicating yeast and plant cells, which can support viral RNA synthesis on both the endogenous and exogenous templates. Our results further strengthen our model on the essential role of cellular sterols as proviral lipids during viral replication.
Because sterols affect so many facets of TBSV replication (5, 20, 21), we analyzed if p33 could directly bind to sterol. An in vitro sterol binding assay revealed the binding of the in vitro translated 35S-labeled p33 to sterol (Fig. 1A). Since the p92pol replication protein contains the same sequence as p33 in its N-terminal region (due to the overlapping-expression strategy [Fig. 1D]), we also tested sterol binding by 35S-labeled p92pol, which was comparable (Fig. 1A). To compare the sterol binding by p33 to that of a known sterol-binding protein, we performed the in vitro sterol-binding assay with the in vitro-translated 35S-labeled yeast Osh4p, which is a sterol transfer protein (27). Comparable amounts of Osh4p and p33 showed similar efficiencies in sterol binding in vitro (Fig. 1B and andC).C). Thus, these in vitro experiments suggest that the TBSV replication proteins could interact with sterols.
To examine if the cytosolic C-terminal segment of p33 is involved in sterol binding, we compared the N-terminally truncated p33 (called p33C [Fig. 1D]) with the full-length p33, which includes the two transmembrane (TM) regions. The sterol binding assay revealed that p33C missing the N-terminal sequence (including the TMs) showed poor binding to sterol compared with that of the full-length p33 (Fig. 1D to toF).F). These data indicate that the N-terminal portion of p33 (and likely p92) is critical for sterol binding.
Many sterol-binding proteins contain CRAC or CARC sterol-binding motifs (26, 28). Interestingly, the N-terminal part of TBSV p33 contains three CRAC- and one CARC-like sterol-binding motif within or in the vicinity of the TM domains (Fig. 2A). These sterol-binding motifs could be important to determine the topology of p33 in membranes, as predicted based on their locations (Fig. 2B), thus suggesting that these sequences might be important for p33 functions. Accordingly, these CRAC- or CARC-like motifs are conserved in the closely related Cucumber necrosis virus (CNV) p33 (peroxisomal replication) and Carnation Italian ringspot virus (CIRV) p36 (mitochondrial replication) replication proteins (Fig. 2A).
To test the importance of the predicted CRAC motifs in p33 in sterol binding, we chose the CRAC motif within the TM2 domain (which includes Y151 [Fig. 2A]). p33-121C, which lacks the N-terminal 120 amino acids (aa) but includes the single Y151 CRAC motif (Fig. 2A), bound to sterol (Fig. 2C). This CRAC motif is absent in p33C (which includes 168 to 296 aa ), which shows poor sterol binding (Fig. 1D). Sterol binding by the CRAC sequence is known to depend on a hydrophobic surface structure/shape within the protein, which can be most efficiently inhibited by altering the conserved Y to a proline to change the surface structure, which still remains hydrophobic (26, 28). We found that p33-121C carrying the Y151P mutation inhibited sterol binding ~5-fold, while p33-121C carrying the Y151A mutation reduced sterol binding ~3-fold, suggesting that the Y151 CRAC motif contributes to sterol binding by p33 replication protein.
To test whether the four CRAC or CARC motifs in p33 play redundant roles in sterol binding, we have introduced mutations to p33 by altering the most critical and conserved Y amino acids within CRAC or CARC sequences (Fig. 2B). p33 carrying the combined mutations in each of the four sterol-binding motifs (Fig. 2D) showed almost as poor sterol-binding capacity as p33C did in vitro (Fig. 1B). However, individual restoration of mutations in each position out of four CRAC or CARC motif mutants restored the sterol binding to 65 to 88% that of wild-type (wt) p33 (Fig. 2D), suggesting that each of these motifs contributes to the ability of p33 to bind to sterol. Thus, it seems that any of the four CRAC or CARC motifs in p33 could bind sterol to, or near, the wt level. Overall, the in vitro results suggest that these CRAC or CARC motifs play a role in sterol binding by p33.
Testing the ability of p33 mutants to facilitate TBSV replication revealed a low level of TBSV replicon RNA (repRNA) replication supported by two CRAC (i.e., Y42P and Y151P) and one CARC (Y83P) p33 mutant in yeast cells (Fig. 3A). The Y151P mutant, which shows poor sterol binding in vitro (Fig. 2C), did not support TBSV repRNA replication (Fig. 3A, lanes 13 to 15), suggesting that this site is critical for p33 function. In contrast, the CRAC mutant (Y100P) in the lumen-facing sequence in p33 had no detrimental effect on TBSV repRNA replication (Fig. 3A). Note that due to the separate expression strategy, only p33, and not p92 (which was wt in these experiments), carried the above-mentioned mutations in yeast. Together, these data suggest that three of the predicted sterol-binding sites in p33 seem to be essential for TBSV replication. These essential CRAC or CARC sequences are conserved in three tombusviruses (Fig. 2A), while the lumen-facing CRAC sequence is present only in TBSV and CNV p33 and not in the CIRV p36 replication protein.
The enrichment of sterols at TBSV replication sites via MCSs and ORPs is critical for p33 stability (20). Therefore, sterol binding by p33 might be important to stabilize p33. This was tested using the p33 Y42,83,100,151P mutant, which shows poor sterol binding in vitro (Fig. 2D). Interestingly, the Y42,83,100,151P mutant showed much reduced stability in comparison with that of the stable wt p33 in yeast (Fig. 3B). The p33 Y83,151A mutant also had a shortened half-life (Fig. 3B), which might be due to, at least in part, to the reduced sterol binding or sterol enrichment around p33 when CRAC and CARC sequences are mutated.
To confirm that the sterol-binding sites in the TBSV replication proteins are also important for the replication of the full-length TBSV genomic RNA (gRNA) in plants, we introduced these mutations into the infectious transcripts, followed by testing the accumulation of gRNA in Nicotiana benthamiana protoplasts. Note that due to the overlapping expression strategy, both p33 and p92pol RdRp (produced via a stop codon readthrough mechanism ) carry the given mutation in these experiments. Northern blot analysis revealed that mutations in the three predicted cytosol-facing sterol-binding sites in p33/p92 blocked TBSV replication, while the mutation in the lumen-facing CRAC sequence had no effect (Fig. 3C). To test if the above-mentioned mutations affect translation of p33 or p92, we performed an in vitro translation assay in the presence of [35S]methionine in wheat germ extract (WGE), programmed with TBSV gRNA. Each mutant replication protein was expressed in a cap-independent translation assay in WGE (Fig. 3D).
In addition, we found that TBSV expressing p33/p92 with the combined mutations in each sterol-binding motif did not accumulate at a detectable level in protoplasts (Fig. 3E, lanes 22 to 24). Also, separate Y-to-A mutations within three motifs (two in CRAC [i.e., Y42A and Y151A] and one in CARC [Y83A]) had detrimental effects, leading to a 40 to 80% reduction in viral gRNA accumulation in protoplasts (Fig. 3E). As expected, the Y-to-F mutations, which preserve the hydrophobic nature of these motifs, which are also present in some CRAC or CARC-like motifs (26), had lesser or no effects on TBSV gRNA accumulation (Fig. 3E).
We further tested Y83A and Y151A, which are parts of the TM domains (Fig. 2A), as a double mutant in a protoplast-based replication assay. Interestingly, the double Y83,151A mutant was nonviable (Fig. 3F, lanes 5 and 6), while the single mutants were replication competent (at 30 to 60% of the level of the wt [Fig. 3E, lanes 7 to 9 and 13 to 15]). These data suggest that at least one of the two motifs at either site must be functional in sterol binding to support TBSV replication. Together, the plant protoplast experiments support the critical role of p33/p92 sterol-binding motifs in TBSV replication.
The ability of p33 and p92 replication proteins to bind sterols might help their localization to sterol-rich membrane microdomains. Since the sterol-rich microdomains are frequently resistant to complete solubilization by detergents, such as Triton X-100, we treated the isolated membrane fraction of yeast cells replicating TBSV repRNA with Triton X-100 at 4°C. Interestingly, a large portion of both p33 and p92 replication proteins remained in the detergent-resistant membrane (DRM) fraction (Fig. 4A, lane 3), while six different membrane proteins of yeast were mostly solubilized. In addition, the DRM fraction sedimented farther in the sucrose gradients (Fig. 4B versus C) than did the membrane fraction without detergent treatment, suggesting a higher density for this DRM fraction. Moreover, discontinuous Nycodenz gradient centrifugation revealed that the DRM fraction containing p33/p92 showed high RdRp activity on added TBSV RI/RIII (−)repRNA (Fig. 4D to toF).F). The pattern of in vitro RdRp products from the DRM fraction carrying p33/p92 suggested that the RdRp mostly synthesized complementary-strand cRNA products comparable to that obtained with the purified tombusvirus replicase preparations from either plants or yeast (30, 31).
Additional attempts with combination of high-salt and detergent treatments also failed to completely solubilize p33 and p92 from the DRM fraction (Fig. 4G). Comparable experiments with the membrane fraction from plants infected with the full-length TBSV gRNA demonstrated the abundant presence of p33 and p92 replication proteins (the native proteins lacking affinity tags) in the DRM fraction (Fig. 5A, lane 3). Interestingly, the DRM fraction from TBSV-infected plants contained highly active tombusvirus replicase, which supported the synthesis of genomic RNA and the two subgenomic RNAs in vitro (Fig. 5B, lanes 1 and 2). The DRM fraction from mock-infected plants did not show RdRp activity (Fig. 5B, lanes 3 and 4). The RNA templates in the DRM fraction containing p33/p92 were derived from plants and were resistant to micrococcal nuclease treatment (lanes 9 and 10), suggesting the presence of protein or lipid-based protection from RNA degradation in the DRM fraction. Moreover, the DRM fraction containing p33/p92 could be programmed with an external template [i.e., DI-72(−) repRNA] (Fig. 5B, lanes 5 and 6). Surprisingly, the abundant external template competed very efficiently with the endogenous templates for the RdRp, suggesting that the RdRp in the DRM fraction is accessible to suitable RNA templates.
To test if the sterol-binding CRAC and CARC motifs are important for association of p33 with the DRM fraction, we performed experiments with the Y42,83,100,151P mutant, which shows ~3-fold-reduced sterol binding in vitro (Fig. 2D). We observed a shift toward a higher portion of the p33 Y42,83,100,151P mutant in the Triton-solubilized fraction than in the insoluble fraction (by ~20% [Fig. 4H]), suggesting that CRAC and CARC motifs affect the association of p33 with the DRM fraction. The reduced presence of the p33 Y42,83,100,151P mutant in the Triton-insoluble fraction is likely due to the residual sterol-binding activity of this mutant and the oligomerization of p33 (32), which could trap many p33 molecules via protein-protein interaction in this fraction.
We compared the lipid composition of DRMs (with or without p33/p92) with the total membranes to identify the changes in lipid composition. The DRMs derived from yeast lacking or expressing TBSV components were highly enriched in sterols in comparison with phospholipids (Fig. 6A versus versusB).B). The total sterol content, including ergosterol and lanosterol, increased only slightly in yeast replicating TBSV repRNA, in contrast with the large (40%) increase in total phospholipids in yeast with TBSV (Fig. 6A versus versusB).B). Similar results were obtained from plant samples, based on mock-inoculated or TBSV-infected plants, showing enrichment of sterols in DRMs and barely detectable amounts of phospholipids (Fig. 6C and andD).D). Also, TBSV infection led to an increased level of phospholipids, but not sterols, in comparison with mock-inoculated plants (Fig. 6C and andD).D). These data support the idea that the TBSV replication proteins are associated with sterol-rich detergent-resistant membranes in yeast and plant cells (Fig. 6E).
Sterols represent major components of cellular membranes in eukaryotes. Viruses frequently exploit sterol-rich microdomains that contain a unique set of proteins, called raft proteins (25, 33). Tombusviruses, similar to many other (+)RNA viruses, depend on sterols and phospholipids, especially PE, for their replication (19,–21, 34, 35). Sterols affect VRC assembly and tombusvirus p33 and p92 stability (20, 21), indicating that the replication proteins may interact with sterols. Accordingly, in this work we have shown direct sterol binding by p33 and p92 (Fig. 1), likely through three CRAC and CARC-like sequences, which are conserved in several tombusviruses. Individual and combined mutagenesis of the critical Y amino acid in the three CRAC- or CARC-like sequences revealed their essential roles for TBSV replication in yeast and plants (Fig. 2 and and33).
The presence of multiple sterol-binding sites in p33 has made determining the roles of individual CRAC- and CARC-like sequences a challenge. Nevertheless, the presence of a single CRAC sequence including Y151 in an N-terminally truncated p33 allowed the recombinant protein to bind sterol in vitro (Fig. 2C). Mutagenesis (Y151P) of this site greatly inhibited sterol binding and made p33 nonfunctional in supporting TBSV replication in yeast (Fig. 3A). When the less debilitating Y-to-A mutation, which inhibits p33 sterol binding to a lesser extent than Y151P (Fig. 2C), was used, then simultaneous mutations of the two cytosol-facing CRAC- and CARC-like sequences (Y83,151A [Fig. 2A and and3F])3F]) were needed to render p33 nonfunctional in viral replication in plant protoplasts. The basic amino acids (K/R) within a CRAC/CARC motif are proposed to be positioned at the polar-apolar interface of transmembrane domains (TM); thus, they facilitate correct placement/topology of a transmembrane protein in a membrane bilayer (26). Our findings suggest that the sterol binding by these sequences in p33 in combination with the proximal positively charged amino acids (Fig. 2A) serves as a flattened head of a nail that likely controls the proper topology of the membrane-interacting TM sequences in the membrane.
While separate mutagenesis of each of the CRAC- or CARC-like sequences had a detrimental effect on TBSV replication, we observed major decrease in sterol binding by p33 only when all CRAC- or CARC-like sequences were mutagenized (Fig. 2C). The difference in yeast versus in vitro results is likely due to a more complex role of p33-sterol interaction in TBSV replication and the more complex cellular environment. First, we propose that the ability of p33 to bind to sterol is important for local enrichment of sterols in the replication compartments to stabilize the spherule structures harboring the tombusviral VRCs. Second, binding to sterols within all three CRAC- or CARC-like sequences likely affects the structure or topography of the replication proteins embedded in the membranes, thus facilitating the proper oligomerization of replication proteins, which is needed for VRC formation. Additional high-resolution structural data will be needed to address the structures of sterol-bound replication proteins.
Similar to the hepatitis C viral replicase (36), the TBSV replicase was partly associated with DRM fractions obtained from yeast and plants (Fig. 4 and and5).5). The DRM fractions not only contained p33 and p92 replication proteins (Fig. 4 and and5)5) and many host proteins (not shown) but also were enriched for sterols (Fig. 6E) and poor in phospholipid content (Fig. 6E). The association of p33/p92 with sterol-enriched DRMs suggests that the tombusvirus replicase is likely associated with lipid microdomains in subcellular membranes. The observation that the added excess amount of (−)RNA template competed very efficiently with the endogenous templates for the RdRp in the DRM fraction suggests that the RdRp in the DRM fraction is not present within membranous structures (such as semiclosed vesicles or spherules), which could not take up exogenous template RNAs after their formations (37, 38). Thus, the DRM fraction does not represent the complete, membrane-bound tombusvirus replicase based on characterization of the in vitro activity (Fig. 5B). Indeed, in addition to the competition between added and endogenous templates, the DRM fraction synthesized many different products on the (−)RNA template. These activities of the DRM fractions are similar to those of the detergent-solubilized and affinity-purified tombusvirus replicase preparations (31), suggesting that the high sterol content in DRMs might not affect tombusvirus RdRp activity. Indeed, unlike the DRM fractions or the purified replicase preparations, tombusvirus replicases assembled in yeast cell extracts (containing subcellular membranes, such as the ER) or via using artificial PE or PE/sterol vesicles (liposomes) could only be programmed with (+)RNA templates and produced both double-stranded RNA (dsRNA) replication intermediates and excess new (+)RNA progeny (17, 39). Moreover, after the in vitro VRC assembly step, the replicase preparations could not take up (+)RNA or (−)RNA from the solutions (17, 37, 40). Therefore, based on these similarities and differences among the various replicase preparations, we suggest that the DRM fractions from yeast or plants do not have the complete VRCs but most likely contain only the catalytic core of the replicase. This is in line with the finding that phospholipids, which are required for complete TBSV replication in vitro and in vivo, are barely present in DRM fraction (Fig. 6E).
In spite of sterol enrichment in DRM fractions containing p33/p92 (Fig. 6E), the presented lipidomic analysis showed the absence of sterol synthesis induction in tombusvirus-infected cells (Fig. 6A and andC).C). This phenomenon is different from phospholipid synthesis, which is induced by TBSV both in yeast and in plant cells (17, 34). In contrast, it seems that TBSV enriches sterols at the sites of replication via inducing rapid redistribution of sterols to the peroxisomes at the expense of the plasma membrane (20, 41). Cellular ORPs and VAPs likely facilitate sterol redistribution via MCS formation induced by viral replication proteins and facilitated by the actin network (20, 41).
What could be the function of sterols at the replication sites? TBSV and other tombusviruses form hundreds of spherules per cell in boundary membranes of peroxisomes (or mitochondria in case of CIRV) that are frequently present in close proximity to one another (18). The virus-induced robust spherule formation requires large subcellular membrane surfaces, which might require sterol-rich microdomains to facilitate the efficient sequestration of viral replication proteins and co-opted host factors, while other host components should be excluded from spherules. Enrichment of sterols in the spherules might also lead to more stable spherules that are predicted to exist for several hours to support robust viral RNA synthesis and replication. Accordingly, sterols allow tighter packing of phospholipids in membranes, which might be important for stabilizing spherules via limiting spontaneous lipid diffusion. Interaction between sterols and the numerous p33 and p92 replication proteins present in each spherule might affect the topology, structure, and oligomerization state of these proteins embedded in membranes.
TBSV p33 and its truncated form (termed p33C) were used in an in vitro cholesterol binding assay as described previously (42). Briefly, different amounts of plant-derived cholesterol (Avanti Polar Lipids, Inc.; catalog number 700100P) dissolved in chloroform were dotted onto the surfaces of polyvinylidene fluoride (PVDF) membranes. The PVDF membrane strips were blocked by 3% bovine serum albumin (BSA) in Tris-buffered saline (TBS; pH 7.5) overnight, followed by incubation with an equal amount of [35S]methionine-labeled p33 or p33C translated in a wheat germ (WGE) in vitro translation system (43) in TBS (pH 7.5) for 3 h at room temperature. As a negative control, no mRNA transcripts were added into the translation system. The PVDF strips were washed with TBS (pH 7.5) three times and analyzed by phosphorimaging (Typhoon; GE).
Saccharomyces cerevisiae strain BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) was obtained from Open Biosystems. Single amino acid mutations in p33 were carried out by site-directed mutagenesis in the original TBSV genome by using plasmid pT100, resulting in pT100-p33-Y42P, pT100-p33-Y83P, pT100-p33-Y100P, and pT100-p33-Y151P. Sequences of p33 single amino acid mutants were amplified by primer pair 788 (GAGGGATCCGAGACCATCAAGAGAATG)/810 (GGAGCTCGAGCTATTTGACACCCAGGGAC) from plasmids described above, digested with BamHI and XhoI, and inserted into BamHI and XhoI-digested pESC-T33/DI72 (40), resulting in pESC-T33-Y42P/DI72, pESC-T33-Y83P/DI72, pESC-T33-Y100P/DI72, and pESC-T33-Y151P/DI72.
For measuring the effects of TBSV p33 single amino acid mutations on TBSV replicon RNA (repRNA) replication in yeast, yeast strain BY4741 was transformed with pYES-T92 (40) together with pESC-T33/DI72, pESC-T33-Y42P/DI72, pESC-T33-Y83P/DI72, pESC-T100P-Y42P/DI72, or pESC-T33-Y151P/DI72. Induction of TBSV repRNA replication, total RNA extraction, and RNA analysis by Northern blotting were described previously (44).
To test the effect of single amino acid mutants of TBSV p33 (together with p92 due to the overlapping expression strategy) on TBSV genomic RNA (gRNA) replication in plant cells, DNA templates for in vitro transcription of TBSV gRNA using T7 RNA polymerase were PCR amplified with primer pair 359 (GTAATACGACTCACTATAGGAAATTCTCCAGGATTTC)/157 (GGGCTGCATTTCTGCAATGTTCC) from plasmids pT100, pT100-p33-Y42P, pT100-p33-Y83P, pT100-p33-Y100P, and pT100-p33-Y151P. TBSV gRNAs were transcribed in vitro by T7 RNA polymerase using above-listed DNA templates. wt and mutated TBSV gRNAs were electroporated into isolated Nicotiana benthamiana protoplasts as described previously (45).
For TBSV repRNA replication studies in yeast, yeast strain BY4741 was transformed with pESC-T33/DI72 and pYES-T92 (40) plasmids. Transformed yeast cells were pregrown in synthetic complete medium lacking appropriate amino acids and containing 2% glucose overnight and then in synthetic complete medium lacking appropriate amino acids and containing 2% galactose for 24 h. Yeast cells were harvested by centrifugation at 4,000 rpm for 4 min at 4°C, washed once with ice-cold water and once with ice-cold buffer A (30 mM HEPES-KOH [pH 7.4], 100 mM potassium acetate, 2 mM magnesium acetate), and homogenized with glass beads in buffer A containing 1 mM dithiothreitol (DTT) and complete mini-protease inhibitor cocktail (Roche Applied Science) for 5 min. The homogenate was centrifuged at 500 × g for 5 min to remove cell debris, nuclei, and glass beads, and then the supernatant was withdrawn. The total membrane fractions were recovered by centrifugation of the above-mentioned supernatant fraction at 20,000 × g for 15 min at 4°C. The total membrane fractions were resuspended in buffer A plus 500 mM potassium acetate containing 1% Triton X-100 (Pd and Sd fractions) or without Triton X-100 (Pb and Sb fractions), 1 mM DTT, and complete mini-protease inhibitor cocktail (Roche Applied Science) and kept on ice for 1 h. The membranes were centrifuged at 20,000 × g for 15 min at 4°C. Then both the supernatant and pellet were collected. The detergent-resistant membrane fraction (Pd fraction) was fully resuspended in buffer A plus 500 mM potassium acetate containing 1% Triton X-100, 1 mM DTT, and complete mini-protease inhibitor cocktail (Roche Applied Science) and then used for lipid analysis.
To perform sucrose/Nycodenz gradient analysis of total membrane fractions or DRM fractions from yeast, 100 μl of total membrane fractions resuspended in buffer A plus 500 mM potassium acetate or DRM fractions resuspended in buffer A plus 500 mM potassium acetate containing 1% Triton X-100 were applied to 700 μl of a continuous 10% to 70% (wt/vol) sucrose gradient or 800 μl of a Nycodenz discontinuous gradient (300 μl of 30% [wt/vol] plus 400 μl of 50% plus 400 μl of 80% Nycodenz solution) and centrifuged at 100,000 × g and 4°C for 2 h using a desktop ultracentrifuge (Optima MAX-XP ultracentrifuge system [Beckman Coulter] with a TLS-55 rotor). Fractions with equal volumes were collected after centrifugation, diluted 10 times with buffer A plus 500 mM potassium acetate, and centrifuged at 20,000 × g for 30 min. The pellets containing membranes were resuspended and subjected to protein analysis and RdRp assay.
To further test the effects of detergents and salts on p33/p92 protein solubility from the yeast Pd fraction, we centrifuged the yeast Pd fraction at 20,000 × g and 4°C for 15 min. The supernatants were removed, and the following buffers were used to resuspend the pellets: buffer 1, buffer A plus 1 M lithium chloride plus 4% Triton X-100; buffer 2, buffer A plus 1 M potassium chloride plus 5% Triton X-100 plus 5% SB3-10; buffer 3, buffer A plus 1 M potassium chloride plus 5% Triton X-100 plus 5% NDSB-256 (nondetergent sulfobetaine 256); and buffer 4, buffer A plus 5% Triton X-100 plus 1% sodium deoxycholate. The mixes were kept on ice for 1 h and then centrifuged at 20,000 × g and 4°C for 15 min. The pellet fraction from each treatment was suspended in sodium dodecyl sulfate (SDS) loading buffer for protein analysis. The total proteins from supernatant fractions were subjected to trichloroacetic acid (TCA) precipitation and then used for protein analysis.
Three-week-old N. benthamiana plants were inoculated with TBSV gRNA transcripts or mock inoculated with buffer. Upper leaves showing TBSV symptoms were harvested at 6 days postinoculation (dpi). Leaves were ground in liquid nitrogen and resuspended in buffer B (50 mM HEPES-KOH [pH 7.4], 500 mM potassium acetate, 15 mM magnesium acetate, 0.2 M sorbitol). The homogenates were passed through 4 layers of cheesecloth to remove debris. The resulting homogenates were centrifuged at 20,000 × g and 4°C for 15 min. Then the pellet was resuspended in buffer B containing 1% Triton X-100 (Pd and Ps fractions) or without Triton X-100 (Pb and Sb fractions), 1 mM DTT, and complete mini-protease inhibitor cocktail (Roche Applied Science) and kept on ice for 1 h. The membranes were centrifuged at 20,000 × g and 4°C for 15 min. Then both the supernatant and pellet fractions were collected. The detergent-resistant membrane (DRM) fraction (Pd) was fully resuspended in buffer A containing 1% Triton X-100, 1 mM DTT, and complete mini-protease inhibitor cocktail (Roche Applied Science) and used for lipid analysis, protein analysis, and the viral RdRp assay.
Each DRM fraction or total membrane fraction from 1.0-g plant samples or 1.0-g yeast samples was mixed with 5 ml of solution containing 94% (vol/vol) ethanol and 6% (vol/vol) 5.88 N KOH solution (33% [wt/vol] KOH solution), 0.25 ml of 20% (wt/vol) ascorbic acid, and 10 μg of 5α-cholestane (as an internal standard) in glass tubes with Teflon liners. Samples were vortexed and incubated at 50°C for 1 h and then cooled on ice for 10 min. To each sample, 7.5 ml of water and 7.5 ml of hexane were added and mixed. Each tube was kept still to allow the organic phase to separate from the aqueous phase. The organic phase (upper phase) was collected. An additional 7.5 ml of hexane was added to repeat the extraction step another 2 times. The organic phases from 3 repeated extraction steps were combined and dried under nitrogen gas. Before gas chromatography-mass spectrometry (GC-MS) analysis, 100 μl of pyridine and 100 μl of N-trimethylsilyl-N-methyl trifluoroacetamide (MSTFA) plus 1% trimethylchlorosilane (TMCS) were added to each dried sample for derivatization, and 1-μl samples were withdrawn for GC-MS analysis on a GC-MS system (Agilent Technologies; model 6890N).
Phospholipid analysis of plant or yeast total membrane fractions or DRM fractions was performed as described previously (17).
To test the RdRp activity of DRM fractions derived from either plant or yeast cells replicating TBSV, the DRM fraction (5 μl) was incubated in a 25°C water bath for 3 h in 50 μl of RdRp reaction buffer containing 50 mM Tris-Cl (pH 8.0); 10 mM MgCl2; 10 mM DTT; 1 mM ATP, CTP, and GTP; 0.5 μl of [32P]UTP; 0.1 μl of RNase inhibitor; 0.05 mg/ml of actinomycin D; and 0.5 μg of viral RNA transcript. The reaction was stopped by adding 110 μl of stop buffer (1% SDS and 0.05 M EDTA, pH 8.0). RNA was extracted by phenol-chloroform, precipitated by isopropanol-ammonium acetate, and washed with 70% ethanol. RNA samples were separated by electrophoresis in a denaturing gel (5% polyacrylamide gel containing 8 M urea) and analyzed by phosphorimaging (Typhoon; GE).
We thank Judit Pogany for critical reading of the manuscript and for very helpful suggestions. We thank Melissa Molho for assisting with one of the experiments, Herman B. Scholthof (Texas A&M) for the anti-p33 primary antibody, and Zuodong Jiang and Joseph Chappell (University of Kentucky) for assisting in sterol profiling. Sss1 antibody was provided by Randy Schekman at UC Berkeley, while Tim23p and Tom40p antibodies were provided by Jan Brix at Gramsch Laboratories, Schwabhausen, Germany.
This work was supported by the NIH-NIAID (5R21AI109529-02) and the University of Kentucky.