Reconstitution of an RNA Virus Replicase in Artificial Giant Unilamellar Vesicles Supports Full Replication and Provides Protection for the Double-Stranded RNA Replication Intermediate

Understanding the mechanism of replication of positive-strand RNA viruses, which are major pathogens of plants, animals, and humans, can lead to new targets for antiviral interventions. These viruses subvert intracellular membranes for virus replication and coopt numerous host proteins, whose functions during virus replication are not yet completely defined. To dissect the roles of various host factors in Tomato bushy stunt virus (TBSV) replication, we have developed an artificial giant unilamellar vesicle (GUV)-based replication assay. The GUV-based in vitro approach recapitulates critical steps of the TBSV replication process. GUV-based reconstitution of the TBSV replicase revealed the need for a complex mixture of phospholipids, especially phosphatidylserine and phosphatidylethanolamine, in TBSV replication. The GUV-based approach will be useful to dissect the functions of essential coopted cellular factors.

ABSTRACT

Positive-strand RNA [(+)RNA] viruses are important pathogens of humans, animals, and plants and replicate inside host cells by coopting numerous host factors and subcellular membranes. To gain insights into the assembly of viral replicase complexes (VRCs) and dissect the roles of various lipids and coopted host factors, we have reconstituted Tomato bushy stunt virus (TBSV) replicase using artificial giant unilamellar vesicles (GUVs). We demonstrate that reconstitution of VRCs on GUVs with endoplasmic reticulum (ER)-like phospholipid composition results in a complete cycle of replication and asymmetrical RNA synthesis, which is a hallmark of (+)RNA viruses. TBSV VRCs assembled on GUVs provide significant protection of the double-stranded RNA (dsRNA) replication intermediate against the dsRNA-specific RNase III. The lipid compositions of GUVs have pronounced effects on in vitro TBSV replication, including (−) and (+)RNA synthesis. The GUV-based assay has led to the discovery of the critical role of phosphatidylserine in TBSV replication and a novel role for phosphatidylethanolamine in asymmetrical (+)RNA synthesis. The GUV-based assay also showed stimulatory effects by phosphatidylinositol-3-phosphate [PI(3)P] and ergosterol on TBSV replication. We demonstrate that eEF1A and Hsp70 coopted replicase assembly factors, Vps34 phosphatidylinositol 3-kinase (PI3K) and the membrane-bending ESCRT factors, are required for reconstitution of the active TBSV VRCs in GUVs, further supporting that the novel GUV-based in vitro approach recapitulates critical steps and involves essential coopted cellular factors of the TBSV replication process. Taken together, this novel GUV assay will be highly suitable to dissect the functions of viral and cellular factors in TBSV replication.

IMPORTANCE Understanding the mechanism of replication of positive-strand RNA viruses, which are major pathogens of plants, animals, and humans, can lead to new targets for antiviral interventions. These viruses subvert intracellular membranes for virus replication and coopt numerous host proteins, whose functions during virus replication are not yet completely defined. To dissect the roles of various host factors in Tomato bushy stunt virus (TBSV) replication, we have developed an artificial giant unilamellar vesicle (GUV)-based replication assay. The GUV-based in vitro approach recapitulates critical steps of the TBSV replication process. GUV-based reconstitution of the TBSV replicase revealed the need for a complex mixture of phospholipids, especially phosphatidylserine and phosphatidylethanolamine, in TBSV replication. The GUV-based approach will be useful to dissect the functions of essential coopted cellular factors.

INTRODUCTION

Positive-strand RNA [(+)RNA] viruses are intracellular pathogens and greatly depend on numerous host components to support their replication. In the complex intracellular environment, viruses must be able to coopt proviral host factors due to the rather limited coding capacity of their small RNA genomes. (+)RNA virus replication takes place on subcellular membrane surfaces in viral replication compartments (1–3). Therefore, viruses have to interact with different lipids, such as phospholipids and sterols. They also have to deform membranes and, thus, depend on the biophysical features of membranes, including fluidity and curvature (4, 5). The hijacked cellular membranes protect the viral RNA from recognition by the host cytosolic nucleic acid sensors or from destruction by the cellular innate immune system (6, 7). Moreover, the subverted membranes allow the sequestration of viral and coopted host proteins with the viral (+)RNA, leading to high local concentrations and efficient macromolecular assembly needed for formation of the viral replicase complex (VRC) or virion assembly. RNA viruses manipulate lipid composition of intracellular membranes to support efficient replication (5, 8–15). Understanding the roles of various lipids and coopted host factors in RNA virus replication is important to control pathogenic RNA viruses.

Among the various lipids, the highly abundant phospholipids and sterols are especially targeted by RNA viruses (1, 14, 16–18). Lipidomics analyses of cells infected with Dengue virus and hepatitis C virus (HCV) (9) or genetic studies with brome mosaic virus and enteroviruses revealed virus-induced lipid biosynthesis, which results in altered global lipid profiles of host cells and enrichment of various lipids in the viral replication compartments (15, 19–21). The minor regulatory phosphatidylinositol-4-phosphate (PI4P) is enriched at replication sites of enterovirus and HCV due to the recruitment of cellular lipid kinases (8, 22). Overall, the membranous microenvironment is enriched for various phospholipids, PI4P, and sterols to facilitate (+)RNA virus replication (6, 14–16, 23–26). However, our knowledge of the roles of various phospholipids and other lipids in RNA virus replication is currently incomplete. This is despite the significance of the interaction between cellular lipids and viral components as one of the possible targets for antiviral approaches against a great number of viruses.

Research based on tombusviruses, such as tomato bushy stunt virus (TBSV), which are small RNA viruses of plants that can replicate in a yeast surrogate host (27, 28), has revealed a major role for global phospholipid and sterol biosynthesis (29–32). Moreover, the critical role of phosphatidylethanolamine (PE) in the formation of VRCs and, together with phosphatidylcholine (PC), in the activation of the virus-coded p92 RdRp has been shown (32–35). Sterols also greatly affect viral replicase functions due to the direct binding of sterols to the p33 replication protein, which is the master regulator of VRC assembly and viral (+)RNA recruitment into VRCs (33, 36).

Despite the development of novel in vitro assays to study TBSV replication based on yeast cell extracts (CFE), mitochondrial or endoplasmic reticulum (ER) preparations (37, 38), artificial vesicles (liposomes) (32) with purified recombinant tombusvirus p33 and p92^pol replication proteins, and TBSV positive-strand replicon RNA [(+)repRNA], our understanding of TBSV replication is far from complete, partially due to the limitations of these assays. For example, the CFE-based reconstitution of the TBSV VRC is a powerful approach, but it is difficult to manipulate the lipid compositions of membranes in yeast CFEs. On the other hand, the liposome-based replicase complexes could not protect the viral RNAs from ribonucleases during or at the end of the replication assay (32). Altogether, the liposome-based membranes were able to fulfill some major functions but also failed to provide some other functions required for TBSV replicase assembly and viral RNA synthesis. This is likely due to the rather small sizes of liposomes (∼100 nm) and their highly curved membranes. In contrast, TBSV replication takes place in the much larger aggregated peroxisomal membrane environment in infected cells (39, 40). Therefore, the membrane deformations induced by the TBSV replication proteins and coopted host factors in cells (∼40- to 70-nm vesicle-like structures, depending on the length of the viral RNA) (41) may not be possible with the small liposomes. Therefore, we decided to develop a novel replication assay based on giant unilamellar vesicles (GUVs), which are large (more than 1 μm in diameter, frequently as large as 20 to 50 μm) vesicles with minimal membrane curvature. In this paper, we show evidence that the novel TBSV VRC reconstitution assay based on GUVs with ER-like phospholipid composition successfully recapitulated many critical steps in the replication process. These include the dependence of TBSV replication on complex lipid membranes and coopted host factors. The GUV-based VRC reconstitution has led to the discovery of the critical role of phosphatidylserine (PS) in TBSV replication and a novel role for phosphatidylethanolamine (PE) in asymmetrical (+)RNA synthesis. Overall, the major advantage of this GUV-based system is the ease of use of various cellular lipids to construct the GUV membranes and test their suitability for the reconstitution of the TBSV replicase in vitro.

RESULTS

Authentic replication of TBSV is supported by artificial GUV vesicles with ER-like phospholipid composition.

To support the in vitro assembly of the TBSV replicase, we used GUVs with various phospholipid compositions in combination with purified recombinant TBSV p33 and p92^pol replication proteins, TBSV positive-strand replicon RNA [(+)repRNA], and the soluble fraction of a yeast cell extract (SF-CFE) (Fig. 1A and Table 1). The idea is that the artificially made GUVs would allow the selection of a desirable lipid composition of membranes, whereas the soluble fraction of the CFEs provides the protein factors needed for TBSV replication. We found that GUVs with ER-like phospholipid composition (42) supported efficient in vitro replication of TBSV repRNA at a level comparable to that of complete CFEs, which contained both soluble and membranous fractions of yeasts (Fig. 1B). TBSV RNA replication supported by the GUVs included both negative- and positive-strand synthesis in an asymmetric manner. Accordingly, the TBSV replicase in the GUV-based assay supported an ∼1:4 ratio of dsRNA intermediate versus positive-strand single-stranded RNA [(+)ssRNA] progeny, whereas the corresponding ratio was 1:11 for the CFE-based assay (Fig. 1B, lanes 1 and 4). Nevertheless, we conclude that the above-described GUV-based TBSV replicase reconstitution assay results in a complete asymmetrical cycle of viral RNA replication on the input (+)RNA template.

FIG 1.

In vitro reconstitution of the TBSV replicase in GUVs with ER-like phospholipid composition. (A) Scheme of the in vitro GUV-based TBSV replicase reconstitution assay. Purified recombinant TBSV p33 and p92^pol replication proteins in combination with the TBSV (+)repRNA were added to GUVs with ER-like lipid composition (Table 1). The S40 fraction of yeast CFE (SF-CFE) was also added to each sample to provide soluble host factors required for TBSV VRC assembly. (B) Nondenaturing PAGE analysis of the ³²P-labeled repRNA products obtained is shown. The full-length (+)repRNA progeny and the dsRNA replication intermediate are pointed at by arrowheads. The level of ssRNA produced in the GUV-based replication assay was chosen as 100% (lane 4). Lane 5 demonstrates the dsRNA nature of the TBSV repRNA product, which was denatured at 85°C treatment. Standard deviation was calculated. Note that the SF-CFE preparation alone supports a basal level of TBSV RNA synthesis when purified recombinant TBSV p33 and p92^pol replication proteins in combination with the TBSV (+)repRNA are added (lane 2). This is likely due to the presence of a small amount of membrane contamination in SF-CFE preparations. The control sample in lane 1 contained a mixture of the membrane fraction (ME) and SF-CFE obtained from wt yeast. (C) Asymmetrical RNA synthesis by TBSV VRCs assembled in GUVs with ER-like lipid composition. The amounts of ³²P-labeled TBSV positive-stranded repRNA products and dsRNA replication intermediate produced by the reconstituted TBSV VRCs are measured during the time course as shown. (Bottom) The graph shows the accumulation of dsRNA and the new (+)repRNA products produced in GUVs with ER-like lipid composition but with longer incubation times than those in the top panel. (D) The TBSV replicase assembled in the GUVs with ER-like lipid composition protects the dsRNA replication intermediate from dsRNA-specific RNase III added at the end of the assay. Note that the protection of the TBSV dsRNA replication intermediate from dsRNA-specific RNase III by the reconstituted VRCs in GUVs with ER-like lipid composition or supplemented with 1% PI(3)P depends on the lipid bilayer. This conclusion is based on the lack of protection after disruption of GUVs by Triton X-100 treatment, as shown in lanes 1 and 7. Each sample contained SF-CFE from wt yeast. The nondenaturing PAGE analysis of the replicase products is as described for panel B. Each experiment was repeated at least three times.

TABLE 1.

In vitro TBSV RNA replication in GUV-based replicase reconstitution assays^a

No.	Lipid components in GUVs	% repRNA	SD
61	55% PC, 17% PE, 18% PI, 10% PS	105	7
63	60% PC, 40% PE (or any other PC+PE combination)	<1
74	Mitochondrion-like (33% PC, 36% PE, 4% PI, 2% PG, 2% PS, 18% CA) + 0.5% NBD-PE	8	2
76	Plastid-like (36% PC, 39% PE, 8% PI, 8% PG, 5% PS) + 0.5% NBD-PE	37	6
77	45% PC, 30% PE, 25% PS	144	31
81	35% PC, 35% PE, 15% PI, 15% PS	169	22
113	ER-like (46% PC, 30% PE, 13% PI, 5% PG, 3% PS, 3% CA) + 2% NBD-PE	100
114	65% PC, 21% PI, 12% PS + 2% NBD-PE	7	2
115	48% PC, 30% PE, 20% PI + 2% NBD-PE	11	2
116	40% PC, 25% PE, 15% PI, 20% CA + 2% NBD-PE	2	1
117	40% PC, 25% PE, 15% PI, 20% PG + 2% NBD-PE	13	2
118	40% PC, 25% PE, 15% PI, 20% PS + 2% NBD-PE	108	11

^aGUVs were prepared using the shown lipids, followed by programming the preparations with purified recombinant p33 and p92^pol replication proteins, (+)repRNA transcripts, and SF-CFE prepared from wt yeast. The total ³²P-labeled TBSV repRNA accumulation was measured by PAGE. Standard deviations were calculated from three or more repeats.

Time course experiments to measure the dynamics of TBSV replicase assembly in GUVs with ER-like phospholipid composition revealed that 45 to 60 min was needed for the initial replicase assembly, leading to the start of dsRNA production, whereas the maximum level of dsRNA production was reached at the 120-min time point (Fig. 1C). The dynamics of TBSV replicase assembly in GUVs is somewhat slower than that observed with yeast CFEs, which could assemble the TBSV replicase in 30 to 45 min (7, 37). The slower dynamics in TBSV replicase assembly based on GUV is likely due to the lack of a group of host membrane proteins, which are present in CFEs but missing from GUVs, and they could facilitate TBSV VRC assembly, possibly serving as VRC assembly platforms or assembly hubs (43).

Importantly, the GUV preparations with ER-like phospholipid composition were able to support the assembly of TBSV replicases that led to significant protection of the dsRNA replication intermediate against the dsRNA-specific RNase III enzyme (Fig. 1D, lane 5 versus 6). The level of dsRNA protection was comparable to that achieved with complete CFE preparations from wild-type (wt) yeast (7, 37, 44). Triton X-100 treatment in combination with RNase III enzyme completely destroyed the dsRNA replication intermediate in the GUV-based assay (Fig. 1D, lanes 1 and 7). This observation strongly supports that the RNase protection of the TBSV dsRNA is provided by the VRCs assembled in the phospholipid membrane in GUVs, similar to the findings with yeast CFEs (37). Based on these findings, we propose that GUV preparations with ER-like phospholipid composition could support the in vitro assembly of TBSV VRCs, which produce both dsRNA replication intermediates and new (+)RNA progeny in an asymmetric manner, and the dsRNA becomes protected by the membranous VRCs from ribonucleases.

Critical roles of phosphatidylethanolamine and phosphatidylserine in artificial GUV vesicles in supporting TBSV replication.

To characterize the minimal requirement for phospholipid composition in GUVs, which is necessary to reconstitute the active TBSV replicase in vitro, we omitted particular phospholipids and/or changed the ratio among various phospholipids compared with the ER-like phospholipid composition, as described in Table 1. Unlike with liposomes/small artificial vesicles (32), we found that GUVs with the simple combination of only phosphatidylethanolamine (PE) and phosphatidylcholine (PC) phospholipids did not support TBSV replication in vitro (Table 1, number 63), suggesting that a more complex combination of phospholipids is needed during replicase reconstitution in GUVs. Accordingly, the minimal combinations of phospholipids to support TBSV replication in vitro were the following: 35% PC, 35% PE, 15% phosphatidylinositol (PI), and 15% phosphatidylserine (PS) (Fig. 2A, lane 2, and Table 1, number 81) and 45% PC, 30% PE, and 25% PS (Table 1, number 77), showing that GUV membranes with high PE, PC, and PS contents are needed for TBSV replication. In the absence of PS (Fig. 2A, lane 5, and Table 1, number 115) or in the presence of a small amount of PE (Fig. 2B, and Table 1, number 114), the GUVs supported TBSV replication only at an ∼10% level compared with the GUVs with ER-like membrane composition (Table 1, number 113). Thus, the GUV-based reconstitution of the TBSV VRC led to the discovery of the critical role of PS and confirmed the requirement of PE and PC in TBSV replication.

FIG 2.

Effect of lipid compositions of GUVs on the activity of the TBSV VRCs. (A) Defining the minimal phospholipid requirement of GUVs to support TBSV replication. Decreased VRC activity is detected in GUV vesicles lacking PS phospholipid (lane 5). The GUVs contained a mixture of phospholipids, as described in Table 1, namely, GUV no. 113 in lane 1, no. 61 (lane 2), no. 117 (lane 3), no. 116 (lane 4), no. 115 (lane 5), and no GUV (lane 6). Nondenaturing PAGE analysis of the replicase products is as described for Fig. 1B. Each sample contained SF-CFE from wt yeast. Each experiment was repeated at least three times. (B) Reverse asymmetrical TBSV replication in GUVs containing a minor amount of PE. The lipid composition of CIS-GUV (no. 114) is described in Table 1. The nondenaturing PAGE analysis of the replicase products is as described for Fig. 1B. Triangles show that the GUV preparations were used in increasing amounts. Each experiment was repeated three times.

Interestingly, GUVs with mitochondrion-like phospholipid compositions supported the assembly of TBSV VRC much less efficiently than the GUVs with the ER-like composition (Table 1, number 74 versus number 113). We observed that GUVs with high cardiolipin (CA) contents supported TBSV replication to a reduced extent (Fig. 2A, lane 4, and Table 1, number 116 and number 74), suggesting that the presence of high CA content in GUVs is inhibitory to TBSV replication, as observed previously with liposomes or in an RdRp activation assay (32, 35).

A surprising observation from these experiments was that GUVs with PC, PI, PS, and low PE content (here called CIS-GUV) supported dsRNA production, but the newly made (+)RNA progeny produced was only slightly above background levels (Fig. 2B, lane 4, and Table 1, number 114). Thus, the TBSV replicase assembled in CIS-GUVs had a reverse asymmetrical replication process by producing an ∼5-fold excess amount of dsRNA replication intermediate over the (+)ssRNA progeny. This is in contrast to the TBSV replicase assembled in GUVs with ER-like lipid composition, which produces excess amounts of the (+)ssRNA progeny (Fig. 1B and 2B). This finding suggests that the lipid composition of the membrane, especially the PE content, hijacked by TBSV for VRC assembly has a pronounced effect on the outcome of TBSV replication, including negative- and positive-strand RNA synthesis. Taken together, we found a new, critical role for PS in TBSV replication and that low PE content in GUVs results in a TBSV replicase supporting mostly (−)RNA synthesis and only limited positive-strand synthesis. We also confirmed the inhibitory effect of CA in the assembly of the TBSV replicase. Overall, a mixture of PE, PS, and PC or more complex phospholipid compositions are required in GUVs to support the full activity of the reconstituted TBSV replicase.

PI(3)P phosphoinositide and ergosterol facilitate TBSV replication in GUV vesicles.

The major advantage of using GUVs over yeast CFEs is the easy manipulation of lipid compositions of membranes. For example, to test the role of a minor signaling lipid, phosphatidylinositol-3-phosphate [PI(3)P], in TBSV replication, we incorporated 1% PI(3)P phosphoinositide into GUVs with the ER-like lipid composition. This led to ∼2- to 3-fold enhanced replication of TBSV in vitro (Fig. 1D, lane 3 versus lane 5). Both (+)RNA progeny and dsRNA replication intermediate levels were enhanced in the presence of PI(3)P, suggesting that this minor lipid plays a proviral role during the replicase assembly process. The GUV preparations with ER-like phospholipid plus 1% PI(3)P composition provided significant protection of the dsRNA replication intermediate against the dsRNA-specific RNase III enzyme (Fig. 1D, lanes 2 versus 3). Thus, this approach confirmed the role of PI(3)P in TBSV replicase assembly, as shown previously using yeast CFEs, yeast cells, and plants (45).

We also used the SF-CFE of vps34Δ yeast, which lacks the only PI(3)P kinase in yeast (46). Interestingly, the combination of GUV with ER-like lipid composition and the SF-CFE of vps34Δ yeast supported only ∼50% of the TBSV replicase activity of SF-CFE derived from wt yeast; thus, it contains Vps34p protein (Fig. 3A, lanes 1 and 2 versus 5 and 6). Supplementing the GUV with 1% PI(3)P in the presence of SF-CFE of vps34Δ yeast led to an almost wt level of TBSV replication in vitro (Fig. 3A, lanes 3 and 4), confirming the proviral roles of PI(3)P and Vps34 PI3K.

FIG 3.

Proviral effect of PI(3)P and Vps34 PI3K on TBSV replication based on GUV assays. (A) The GUV-based assay with ER-like lipid composition was supplemented with 1% PI(3)P, whereas the comparable SF-CFEs were prepared from wt or vps34Δ yeasts as shown. The nondenaturing PAGE shows the accumulation levels of ³²P-labeled ssRNA progeny and the dsRNA replication intermediate. (B) Blocking TBSV replication with bacterial effectors from L. pneumonia. The GUV-based assay with ER-like lipid composition was supplemented with 1% PI(3)P as shown, whereas the SF-CFE was prepared from wt yeast. Note that the RavZ effector binds to PI(3)P and highly curved membranes, whereas LegC8 glucosylates and inactivates eEF1A translation elongation factor. The GST-tagged bacterial effectors were expressed and purified from E. coli and then added to the GUV assays at the beginning of the assay. Each experiment was repeated.

Bacterial effectors can be used as subcellular probes (47–49), and here we selected RavZ of Legionella pneumophila, which binds to PI(3)P and highly curved membranes (50, 51). RavZ was shown to inhibit TBSV replication in yeast (49). The addition of purified recombinant RavZ to the GUV assay with 1% PI(3)P and the ER-like lipid composition inhibited TBSV replication by over 10-fold, decreasing both (+)repRNA and dsRNA replication intermediate levels (Fig. 3B). On the contrary, RavZ was not inhibitory when the GUV’s lipid composition was ER-like but had a small amount of PI(3)P (Fig. 3B, lane 5 versus 4). These experiments demonstrated that PI(3)P is important for TBSV replication in vitro, and the bacterial effector RavZ is only an effective inhibitor of TBSV replication when PI(3)P is present in larger amounts.

Other important lipids in the cellular membranes required for TBSV replication are sterols (31, 33, 52). To test if sterol could also promote TBSV replication in vitro, we added 5% ergosterol to the GUV with ER-like lipid composition. The GUV preparations containing sterol supported an enhanced level of TBSV replication in vitro by 70% (Fig. 4A). Both (+)RNA progeny and dsRNA replication intermediate levels were enhanced in the presence of ergosterol, suggesting that this lipid plays a proviral role during the replicase assembly process. Combined incorporation of 1% PI(3)P and 5% ergosterol into the GUVs with ER-like lipid composition led to a more than ∼2-fold increase in TBSV replication in vitro (Fig. 4B). Overall, these experiments with various lipid compositions in GUVs to reconstitute the TBSV replicase revealed that GUVs with complex lipid composition can support TBSV replication in vitro and could provide dsRNase protective environments through supporting correct VRC assembly, just like the membranes in living cells.

FIG 4.

Proviral effect of ergosterol on TBSV replication based on GUV assays. (A) The GUV-based assay with ER-like lipid composition was supplemented with 5% ergosterol (lanes 1 to 8). Each sample contained the same amount of SF-CFE prepared from wt yeast. Nondenaturing PAGE shows the accumulation levels of ³²P-labeled ssRNA progeny and the dsRNA replication intermediate. The time course shows when the assay was stopped for analysis. (B) The GUV-based assay with ER-like lipid composition was supplemented with 5% ergosterol and 1% PI(3)P (lanes 1 to 8). Each sample contained the same amount of SF-CFE prepared from wt yeast. Further details are provided for panel A. Each experiment was repeated.

The roles of coopted replicase assembly factors in GUV-based replicase reconstitution in vitro.

To test if the in vitro reconstitution of the TBSV replicase in GUVs requires coopted replicase assembly factors, first we targeted the eukaryotic eEF1A translation elongation protein (Tef1/2 in yeast), which has proviral functions in VRC assembly, p33 stabilization, and minus-strand synthesis (53–56). To inhibit eEF1A activity, we utilized the LegC8 effector of L. pneumophila, which glucosylates and inactivates eEF1A (57). LegC8 was shown to inhibit TBSV replication in yeast (49). The addition of the purified recombinant LegC8 to the GUV assay with the ER-like lipid composition inhibited TBSV replication by over 2-fold, decreasing both (+)repRNA and dsRNA replication intermediate levels (Fig. 3B, lane 8 versus 7). The inhibitory effect by LegC8 on in vitro TBSV replication was independent of the presence of high or low levels of PI(3)P in the GUVs (Fig. 3B, lanes 1 and 8). We interpret these data to mean that inactivation of the coopted eEF1A by LegC8 effector inhibits TBSV replication via blocking VRC assembly and minus-strand synthesis in the GUV-based assay.

The second coopted replicase assembly factor that we targeted was the cellular Hsp70 protein chaperone. Hsp70 (SSA1 and SSA2 in yeast) is coopted by TBSV to activate the p92 RdRp and for facilitating the membrane insertion of the p33 and p92 replication proteins (37, 58, 59). To this end, we applied the allosteric inhibitor MKT-077 (60) of Hsp70 to block the chaperone activity in vitro. We found that the activity of the reconstituted TBSV replicase in the GUV-based replication assay was almost completely lost in the presence of MKT-077 inhibitor (Fig. 5A, lane 5). The production of both dsRNA replication intermediate and the new (+)repRNA progeny was inhibited, suggesting that Hsp70 is needed for the reconstitution of the TBSV replicase in the GUV-based assay.

FIG 5.

Essential role of the coopted Hsp70 in TBSV replication based on GUV assays. (A) The GUV-based assay with ER-like lipid composition was performed with the SF-CFE prepared from wt yeast. The allosteric inhibitor MKT-077 of Hsp70 or DMSO (as a negative control) was added to the assay as shown. Nondenaturing PAGE shows the accumulation levels of ssRNA progeny and the dsRNA replication intermediate. (B) Purified recombinant yeast Ssa1 (Hsp70) was added to the GUV assay with ER-like or CIS (PC, PI, and PS plus a small amount of PE; Table 1, no. 114) lipid compositions. The positive control contained SF-CFE prepared from wt yeast (lane 1). Further details are described in the legend to Fig. 1. Each experiment was repeated.

To test if Hsp70 by itself is sufficient to assemble the TBSV replicase in the GUV-based assay, we provided recombinant purified Ssa1 (the yeast Hsp70) to our standard GUV-based replication assay. However, unlike in the presence of a soluble fraction of yeast CFE, we did not detect the production of a dsRNA replication intermediate in the GUV-based replicase reconstitution assay (Fig. 5B). These data suggest that Ssa1 Hsp70 by itself is not sufficient to support TBSV replicase assembly in the GUV-based assay. The purified Ssa1 preparation is active in a simple p92^pol activation assay in vitro (35, 61), indicating that the purified Ssa1 is a functional chaperone. Altogether, we have demonstrated that eEF1A and Hsp70 coopted replicase assembly factors are required for the reconstitution of the active TBSV replicase in GUVs, further supporting that our GUV-based in vitro approach recapitulates critical steps and involves essential coopted cellular factors of the TBSV replication process.

The roles of coopted membrane-bending ESCRT proteins in GUV-based replicase reconstitution in vitro.

To further demonstrate the authentic nature of TBSV replication in the artificial GUV preparations, we chose a group of ESCRT proteins (endosomal sorting complex required for transport), which are coopted by TBSV to support the formation of spherules (vesicle-like invaginations) that harbor the TBSV VRCs (40, 41). The ESCRT proteins are involved in membrane deformations and herding cargo proteins together in the selected portion of the membrane (62). To test the roles of selected proviral ESCRT proteins, we prepared SF-CFEs from yeasts lacking an ESCRT component. Interestingly, the lack of either Snf7 (CHMP4A in mammals) or Vps20 (CHMP6 in mammals) ESCRT-III factor inhibited the production of the dsRNA replication intermediate by ∼50-fold in the GUV-based assay (Fig. 6, lanes 11 and 12). Another observation is that the reverse asymmetric replication supported by CIS-GUV was also inhibited by the lack of ESCRT-III factors in the replicase reconstitution assay (Fig. 6, lanes 3 and 4 versus 1). Thus, similar to our previous findings in yeast and plants (40, 41), ESCRT-III components involved in membrane deformations via creating negative membrane curvature are critical for TBSV replication in the GUV-based assays.

FIG 6.

Critical roles for ESCRT-III and ESCRT-I proteins in TBSV replication based on GUV assays. (A) The GUV-based assays with ER-like lipid composition were performed with comparable SF-CFEs, which were prepared from wt, bro1Δ vps23Δ (double deletion), snf7Δ, or vps20Δ yeasts as shown. Nondenaturing PAGE shows the accumulation levels of ssRNA progeny and the dsRNA replication intermediate. We also tested TBSV replication in GUV-based assays with CIS (PC, PI, and PS plus a small amount of PE; Table 1, no. 114) lipid composition. Further details are described in the legend to Fig. 1. Each experiment was repeated three times. (B and C) In vitro translation mixture producing p33 and p92^pol replication proteins was used to program the GUVs with ER-like lipid composition in the absence of SF-CFE. The control experiment in panel C included the in vitro translation mixture producing p33 and p92 replication proteins in the presence of yeast CFE (lanes 3 and 4). Nondenaturing PAGE shows the accumulation levels of ssRNA progeny and the dsRNA replication intermediate. Each experiment was repeated three times.

We also tested the roles of the Vps23 ESCRT-I component (Tsg101 in mammals) and Bro1 accessory ESCRT protein (ALIX in mammals), which have somewhat redundant functions in the recruitment of other ESCRT factors (63). The GUV-based assay, which contained the soluble fraction of CFEs from yeast lacking both Vps23 and Bro1 (41), supported the production of dsRNA replication intermediates 5-fold less efficiently than the corresponding preparation from wt yeast (Fig. 6, lane 10). The reverse asymmetric replication supported by CIS-GUV was also inhibited by the lack of Vps23 and Bro1 host factors in the replicase reconstitution assay (Fig. 6, lane 2 versus 1). Taken together, similar to the yeast- and plant-based replication assays, ESCRT-I and Bro1 accessory ESCRT factors are important for the reconstitution of the TBSV replicase in our GUV assays.

TBSV p33 and p92 replication proteins induce the aggregation of GUVs in vitro.

Since the TBSV p33 replication protein is known to cause the aggregation of peroxisomes into large replication organelles in yeast and plant cells (39, 64, 65), we analyzed the structure of GUVs with confocal laser microscopy. The GUVs were programmed with TBSV (+)repRNA transcripts and tombusvirus p33 and p92^pol replication proteins, which were produced via in vitro translation. Under these conditions, the GUVs supported TBSV replication almost as efficiently as that observed with yeast CFEs and purified recombinant viral replication proteins (Fig. 6B and C).

To facilitate the visualization of GUVs, we included the fluorescent 2% NBD-PE [1,2-dioleoyl-sn-glycero-3 phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl)] during GUV preparations (Fig. 7A). Interestingly, we frequently observed the aggregation and deformation of GUVs with ER-like lipid plus 5% sterol composition in the presence of in vitro-translated p33 and p92^pol and (+)repRNA and SF-CFE (Fig. 7B and C), which were not observed with comparable GUV preparations incubated in the absence of the viral components. Another observation is the uneven distribution pattern of NBD-PE between the individual aggregated GUVs in the presence of recombinant p33 and p92^pol and (+)repRNA (Fig. 7B and C), suggesting enrichment/sequestration of NBD-PE in those areas. We also observed similar GUV aggregation and deformation when purified recombinant p33 and p92^pol replication proteins were added to GUVs (in the presence of SF-CFE) but not in their absence (Fig. 7B). These observations suggest that p33 and p92^pol replication proteins could induce the aggregation and deformation of individual GUVs. The ability to aggregate numerous peroxisomes in combination with membrane deformation is a documented property of TBSV replication proteins in yeast and plant cells (Fig. 7D) (39, 52).

FIG 7.

Aggregation and deformation of GUVs in the presence of p33 and p92^pol replication proteins. (A) Control confocal laser microscopy images of GUVs with purified MBP in the absence of p33 and p92^pol replication proteins. NBD-PE (2%) was added to the GUVs with ER-like lipid composition or with 1% PI(3)P supplement. Differential interference contrast (DIC) images also show the GUVs. Scale bars represent 10 μm. (B) Aggregation and deformation of GUVs in the presence of p33 and p92^pol replication proteins are visualized through confocal laser microscopy. P33 and p92^pol replication proteins were expressed and purified from E. coli, or they were produced via in vitro translation without purification as shown. NBD-PE (2%) was added to the GUVs with ER-like lipid composition. Note that we also added SF-CFEs prepared from wt yeast to the assay. Scale bars represent 10 μm. (C) Confocal laser microscopy images of GUVs containing ER-like lipid composition plus 5% ergosterol supplement. The negative-control experiments show images of GUVs on the left in the presence of an in vitro translation mixture not producing p33 and p92^pol replication proteins. Further details are provided in the legend to panel B. Scale bars represent 10 μm. (D) A confocal laser microscopy image of aggregated peroxisomes in an epidermal cell of the plant Nicotiana benthamiana infected with TBSV (39, 52). The cell expresses p33-BFP and RFP-SKL peroxisomal matrix marker. Scale bars represent 5 μm.

DISCUSSION

Replication of (+)RNA viruses inside the infected cells is a rather complex process, with the participation of many factors, such as viral replication proteins, including the virus-coded RdRp, the viral (+)RNA and dsRNA replication intermediate, numerous coopted host proteins, subcellular membranes, and various metabolites (1–3, 23, 24). The functions of these components during replication are incompletely understood. In addition, the individual VRCs are assembled into large membranous viral replication organelles (VROs), which can be as large as the host nucleus, which is the case with TBSV in plant cells (41). The critical roles of subcellular membranes and lipids make it especially challenging for biochemical and structural studies. This is important, because (+)RNA viruses replicate in various subcellular compartments that contain unique compositions of lipids (14, 15, 23). Therefore, it could be very beneficial to develop in vitro systems in which the roles of lipids and membranes could be studied in a more regulated or defined environment. Accordingly, in this work, we present the development of an in vitro replicase reconstitution assay with TBSV based on GUVs with various lipid compositions. GUVs were also used to reconstitute the budding process of HIV-1 with the help of hijacked ESCRT proteins (66).

We found that many of the critical features of TBSV replication could be recapitulated with the GUV-based in vitro assays. (i) For example, the in vitro assembled TBSV replicase in artificial GUVs can support a complete cycle of viral RNA replication and asymmetrical RNA synthesis, which is a hallmark of positive-strand RNA viruses replicating in infected cells. (ii) Importantly, TBSV replicase could form a protective microenvironment for the dsRNA replication intermediate with the help of GUVs, which is also the case in infected cells and in the yeast CFE-based assays (7, 37, 44), but this was not the case with artificial liposomes/vesicles (32). (iii) The requirement for PE in TBSV replication was also documented with GUVs. We found that GUVs with a minimum of three to four phospholipids (PE, PC, PS, and PI) were needed to reconstitute the functional TBSV replicases. (iv) We also found that the presence of ergosterol (similar to cholesterol) and PI(3)P phosphoinositide in the GUV’s membrane stimulated TBSV replication in vitro, similar to their proviral roles in yeasts or plants (31, 33, 45). (v) The GUV-based TBSV replicase reconstitution assay confirmed the previously documented negative effect of cardiolipin on TBSV replication (32). Moreover, we observed that GUVs with unusual lipid composition (PC, PI, PS, and only a small amount of PE) supported reverse asymmetrical replication, where the dsRNA replication intermediate was produced in excess amounts compared with the (+)RNA progeny. This observation highlights the essential role of PE in TBSV positive-strand RNA synthesis. In addition, a novel critical role for PS in stimulation of TBSV VRC assembly was discovered using GUVs. Thus, similar to cellular membranes with complex lipid composition that can support viral replication in vivo, TBSV replication was greatly dependent on the lipid composition of GUVs.

We also found that a complex set of coopted host proteins was needed to reconstitute the functional TBSV VRCs using GUVs. These host proteins included the VRC assembly factors, such as Hsp70 and eEF1A and membrane-deforming ESCRT-III and ESCRT-I/BroI factors, which have defined proviral roles in yeasts or plants (37, 40, 41, 54). The stimulatory role of Vps34 PI3K (45) has also been documented using GUVs with incorporated PI. Based on these advances, the GUV-based TBSV replicase reconstitution assay will likely be suitable to dissect the functions of lipids, membranes, and coopted host proteins and their complexes in TBSV replication.

Although the previously developed liposome/artificial vesicle-based TBSV replicase reconstitution assay was useful to identify and characterize several functions of lipids, mainly revealing the critical role of PE, in TBSV replication, the liposome-based assay did not recapitulated many, more complex functions/features of the viral replication process (32). These included the lack of protection of the dsRNA replication intermediate by the assembled replicase, and TBSV replicases assembled with liposomes with simple lipid compositions, which do not exist in cells, could also support TBSV RNA synthesis. The liposomes are small (∼100 nm), and TBSV replicates in much larger aggregated peroxisomes with highly deformed membranes (39, 40, 52). Our new GUV-based TBSV replicase reconstitution assay seems to solve these limitations due to its much larger size and the need for more complex lipid compositions to support in vitro TBSV replication in a nuclease-protected microenvironment.

However, the GUV-based TBSV replicase reconstitution assay in the current format lacks coopted membrane-localized host proteins, such as the syntaxin 18-like Ufe1p SNARE protein, Sac1 PI(4)P phospatase, and VAP protein Scs2p, which are important for the biogenesis of the TBSV replication compartment in the more complex cellular environment (43, 52, 67). However, the incorporation of these coopted cellular proteins into GUV membranes will likely be achievable in future works to further advance this powerful assay.

Conclusions.

We have developed an in vitro GUV-based TBSV replicase reconstitution assay, which recapitulates many of the known features of TBSV replication in infected cells. This unique assay will be highly suitable to dissect the various roles of viral replication factors, numerous coopted proviral host proteins, antiviral host proteins, subcellular membranes, lipids, and various metabolites during TBSV replication. The development of similar replicase reconstitution assays for other (+)RNA viruses with medical or agricultural significance would be beneficial, including high-throughput drug screening and structural studies.

MATERIALS AND METHODS

Yeast strains and plasmids used.

Saccharomyces cerevisiae wt haploid strain BY4741 and Δvps34::kanMX4, Δsnf7::kanMX4, and Δvps20::kanMX4 single-gene-deletion yeast knockout strains were obtained from Open Biosystems (Huntsville, AL, USA). The Δbro1::kanMX4 Δvps23::hphNT1 yeast strain was described previously (41).

Plasmids pAG416GAL-LegC8 and pAG416GAL-RavZ (kindly provided by Jun-ichi Inaba) (49) were used as templates in PCRs with primers 7793 (CCAGGGATCCATGAGCGAACAATATTGGCGTTTC) and 7794 (CCAGCTCGAGTTATTTTGATAATCGTGCTTGCTCACTAGGC) for LegC8 and 7795 (CCAGGGATCCATGAAAGGCAAGTTAACAGGTAAAGAC) and 7796 (CCAGCTCGAGCTATTTTACCTTAATGCCACCATGCAATG) for RavZ. The obtained PCR products were digested with BamHI and XhoI and were ligated into pGEX-His vector, digested with BamHI and XhoI, to obtain pGEX-His-LegC8 and pGEX-His-RavZ, respectively.

Protein purification from E. coli.

Expression and purification of the recombinant MBP-tagged TBSV p33 and p92 replication proteins from E. coli CodonPlus cells were performed as published earlier, with modifications (68). Purification of glutathione S-transferase (GST)-tagged proteins was performed using a similar protocol. Briefly, the expression plasmids (pGEX-His, pGEX-His-LegC8, and pGEX-His-RavZ) were transformed into E. coli strain BL21(DE3) CodonPlus. Protein expression was induced using isopropyl β-d-thiogalactopyranoside (IPTG) for 7 h at 23°C. The cells were harvested by centrifugation at 5,000 rpm at 4°C for 5 min to remove the culture medium prior to suspension in MBP column buffer containing 30 mM HEPES-KOH, pH 7.4, 25 mM NaCl, 1 mM EDTA, 10 mM β-mercaptoethanol, followed by sonication. Affinity columns containing glutathione resin (NEB) were used to purify GST-tagged recombinant proteins from cleared sonicated E. coli extracts. After washing the resin 3 times with the column buffer, the proteins were eluted with column buffer containing 10 mM glutathione. Eluted proteins were aliquoted for storage at −80°C. The concentration of the purified recombinant proteins was measured by Bio-Rad protein assay. Protein fractions used for the replication assays were at least 95% pure, as determined by SDS-PAGE. Two or 6 pmol purified GST-tagged proteins or GST protein was added to the CFE or GUV CFE assays.

For the purification of FLAG-tagged yeast Ssa1, the procedure described in reference 35 was used. Briefly, wt FLAG-tagged Ssa1p was expressed in yeast from plasmid pEsc-His/Cup-FLAG/ssa1. Purification was performed using anti-FLAG M2 resin, and the bound Ssa1p protein was eluted with TG buffer containing 100 μg/ml FLAG. One microgram of purified protein was added to the GUV- and CFE-based assays. TG buffer, containing 100 μg/ml FLAG, was used as a negative control.

Preparation of GUVs.

Solutions of phospholipid (1 mg/ml; with 18:1 fatty acid chains) mixtures in chloroform, which were prepared from 25 or 10 mg/ml stock solutions (Avanti, USA) in chloroform and stored at –20°C under nitrogen gas, were used for GUV preparation. A Vesicle Prep Pro (Nanion, Germany) workstation was used for the preparation of GUVs by electroformation. Typically, 0.2 ml of a 1-mg/ml solution of phospholipids in chloroform was placed on the conductive side of an ITO slide. Following the evaporation of chloroform, an 18-mm O-ring was placed around the lipid film, and 0.25 ml of 0.25 M sorbitol solution in MilliQ water was added over the lipid film. Following the covering of the first slide by the second slide (conductive side facing down), the sandwich was placed in the Vesicle Prep Pro chamber. The following adjusted protocol parameters were used: Freq, 005.0; Ampl, 03.00; Temp, 050.0; Rise time, 05:00; Main time, 999:00; Fall time, 05:00. At the end of the protocol, sorbitol solution containing GUVs was pipetted into an Eppendorf tube and stored at 4°C for 1 to 2 weeks.

Reconstitution of the TBSV replicase in the GUV-based assay.

A procedure described earlier (44) was used to prepare whole-cell-free yeast (BY4741, Δvps34, Δsnf7, Δvps20, and Δbro1Δvps23 yeast strains [41]) extracts capable of supporting TBSV replication in vitro. For the CFE-based assay, the procedure described earlier was used (37, 44). For the GUV-based assays, the procedure was modified to remove the membranous fraction of CFE that resulted in SF-CFE. To remove the membranes from the whole CFE, it was centrifuged for 20 min at 42,000 × g at 4°C for the separation of the soluble (SF; supernatant) and membrane (pellet) fractions. The SF-CFE was added to a reaction mixture containing all the components [0.25 μg DI-72 (+)repRNA transcripts, 200 ng E. coli-expressed and affinity-purified recombinant MBP-p33, 200 ng recombinant MBP-p92^pol, 30 mM HEPES-KOH, pH 7.4, 150 mM potassium acetate, 5 mM magnesium acetate, 0.13 M sorbitol, 0.4 μl actinomycin D (5 mg/ml), 2 μl of 150 mM creatine phosphate, 0.2 μl of 10 mg/ml creatine kinase, 0.2 μl of RNase inhibitor, 0.2 μl of 1 M dithiothreitol (DTT), 2 μl of rNTP mixture (10 mM ATP, CTP, and GTP and 0.25 mM UTP)] and aliquoted GUVs. The assays were performed at 25°C for 2 h or 2.5 h. The ³²P-labeled repRNA products synthesized in the replication assay were loaded without heat treatment onto the 5% polyacrylamide gel (PAGE) containing 8 M urea and separated by electrophoresis in 0.5× Tris-borate-EDTA (TBE) buffer (7).

To test the protection of TBSV dsRNA replication intermediate products, treatment with dsRNA-specific RNase RNase III (NEB) was performed as described previously (7), with some modifications. Briefly, 0.2 U of RNase III was added to the reaction mixture at the end of the TBSV replication assay (Fig. 1A), but 15 min prior to the extraction of RNA samples with phenol-chloroform- and isopropanol-based precipitation. The treatment of reaction mixture with 0.1% Triton X-100 (Amersham, USA) was used to solubilize the GUVs and remove the protection of dsRNA.

MKT-077 (60), an inhibitor of Hsp70 protein, was dissolved in DMSO and added to the GUV-based assay at 100 μM final concentration (35). DMSO was added to the control reaction mixtures.

The DNA sequences corresponding to the p33 and p92 mRNAs were PCR amplified using primers 2144 (GTAATACGACTCACTATAGGGAAGCTATACCAAGCATACAATC) and 2145 (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTAAAACCTAAGAGTCAC) from plasmids pGBKCNV33 and pGADCNV92, respectively (69). In addition to the p33 and p92 sequences, the PCR products contained the 5′ and 3′ noncoding regions from the yeast expression vectors, a T7 promoter at the very 5′ end and a poly(A) tail at the very 3′ end. T7 transcripts were produced according to the manufacturer’s recommendations (TaKaRa) and purified by phenol-chloroform extractions and 2 subsequent isopropanol-ammonium acetate precipitation steps. In vitro translation reactions with the wheat germ extract were done according to the manufacturer’s recommendations (Promega), except that the amino acid mix contained all the amino acids. In vitro replication assays were carried out as described above, except replacing the E. coli-expressed and affinity-purified TBSV proteins with the in vitro translation reaction mixtures containing 5 μl in vitro translation mix, which produced either one or both of the viral replication proteins, as indicated in the figures. Note that the obtained replication proteins were not purified and the wheat germ extract provided the host factors for tombusvirus RNA replication in this assay.

Confocal laser microscopy.

To visualize the structure of GUVs in the presence of viral components, we used a fluorescent PE [18:1 NBD-PE (1,2-dioleoyl-sn-glycero-3 phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl))] (Avanti Polar Lipids, Inc.), which was dissolved in DMSO at 8 mM concentration (70) and stored at –20°C. Following the preparation of GUVs containing NBD-PE, an unlabeled (in the absence of ³²P-UTP) GUV CFE reaction was performed. After 2 h of incubation, the confocal microscopic analysis using a 488-nm laser (green fluorescent protein channel) in an Olympus FV1000 confocal laser scanning microscope was conducted.