Co-opting the fermentation pathway for tombusvirus replication: Compartmentalization of cellular metabolic pathways for rapid ATP generation

Summary

The viral replication proteins of plus-stranded RNA viruses orchestrate the biogenesis of the large viral replication compartments, including the numerous viral replicase complexes, which represent the sites of viral RNA replication. The formation and operation of these virus-driven structures require subversion of numerous cellular proteins, membrane deformation, membrane proliferation, changes in lipid composition of the hijacked cellular membranes and intensive viral RNA synthesis. These virus-driven processes require plentiful ATP and molecular building blocks produced at the sites of replication or delivered there. To obtain the necessary resources from the infected cells, tomato bushy stunt virus (TBSV) rewires cellular metabolic pathways by co-opting aerobic glycolytic enzymes to produce ATP molecules within the replication compartment and enhance virus production. However, aerobic glycolysis requires the replenishing of the NAD⁺ pool. In this paper, we demonstrate the efficient recruitment of pyruvate decarboxylase (Pdc1) and alcohol dehydrogenase (Adh1) fermentation enzymes into the viral replication compartment. Depletion of Pdc1 in combination with deletion of the homologous PDC5 in yeast or knockdown of Pdc1 and Adh1 in plants reduced the efficiency of tombusvirus replication. Complementation approach revealed that the enzymatically functional Pdc1 is required to support tombusvirus replication. Measurements with an ATP biosensor revealed that both Pdc1 and Adh1 enzymes are required for efficient generation of ATP within the viral replication compartment. In vitro reconstitution experiments with the viral replicase show the pro-viral function of Pdc1 during the assembly of the viral replicase and the activation of the viral p92 RdRp, both of which require the co-opted ATP-driven Hsp70 protein chaperone. We propose that compartmentalization of the co-opted fermentation pathway in the tombusviral replication compartment benefits the virus by allowing for the rapid production of ATP locally, including replenishing of the regulatory NAD⁺ pool by the fermentation pathway. The compartmentalized production of NAD⁺ and ATP facilitates their efficient use by the co-opted ATP-dependent host factors to support robust tombusvirus replication. We propose that compartmentalization of the fermentation pathway gives an evolutionary advantage for tombusviruses to replicate rapidly to speed ahead of antiviral responses of the hosts and to outcompete other pathogenic viruses. We also show the dependence of turnip crinkle virus, bamboo mosaic virus, tobacco mosaic virus and the insect-infecting Flock House virus on the fermentation pathway, suggesting that a broad range of viruses might induce this pathway to support rapid replication.

Keywords:

Yeast – Protein interactions – Viral replication – Leaves – Glycolysis – Reverse transcriptase-polymerase chain reaction – Fluorescence resonance energy transfer – Fermentation

Introduction

Similar to other positive-strand RNA viruses, the plant-infecting tombusviruses cause major structural rearrangements and metabolic changes in infected cells. The changes include the subversion of pro-viral host factors to support their replication and the induction of lipid synthesis, lipid transfer, membrane proliferation and alteration of vesicular trafficking. The major outcome of all these virus-driven processes is the biogenesis of the unique and extensive viral replication compartments and the formation of numerous viral replicase complexes (VRCs) on subverted subcellular membrane surfaces [1–7]. All these cellular changes serve several purposes, including supporting robust viral RNA replication, and protection of the viral RNA, including the dsRNA replication intermediate, from recognition by the cellular innate immune system or from elimination by the host RNAi machinery, which is also called post transcriptional gene silencing in plants [8–11]. Also, sequestration of viral and co-opted host proteins together with the viral (+)RNA into the replication compartment results in high local concentrations and efficient macromolecular assembly needed for the optimal formation of VRCs [6,12–19]. Our increasing knowledge of the roles of various lipids/membranes and co-opted host factors in RNA virus replication will be useful to control RNA viruses.

Tombusviruses, such as tomato bushy stunt virus (TBSV) and carnation Italian ringspot virus (CIRV), are small (+)RNA viruses, which can replicate in the surrogate model host yeast (Saccharomyces cerevisiae) [20–22]. Intensive genome-wide and proteome-wide research with the TBSV–yeast system has led to a catalog of host factors co-opted for viral RNA replication [4,22,23]. Induction of global phospholipid biosynthesis and redistribution of sterol and alteration of vesicular trafficking has been revealed to play major roles in the formation of VRCs and the activation of the viral-coded p92 RdRp [24–28]. All these subcellular changes are guided by the p33 replication protein, which is the master regulator of VRC assembly and viral (+)RNA recruitment into the VRCs [24,29]. Additional characteristic alterations caused by TBSV include the subversion of the actin network, the induction of subcellular membrane proliferation, peroxisome aggregation and the formation of membrane contact sites to support efficient virus replication [22,30,31].

Most of these viral-induced processes require ATP-based energy and the production of new metabolites in infected cells. Accordingly, we have previously discovered that several glycolytic enzymes, such as glyceraldehyde-3-phosphate dehydrogenase (GAPDH, Tdh2/3 in yeast), phosphoglycerate kinase (Pgk1) and pyruvate kinase (PK, Cdc19 in yeast) are recruited into the viral replication compartment [32–35] and Eno2 phosphopyruvate hydratase binds to p92^pol replication protein [27]. These findings suggest that tombusviruses co-opt the aerobic glycolytic pathway, which leads to the production of plentiful ATP within the viral replication compartment [32,33]. Our studies also revealed that the locally produced ATP is used up to fuel the co-opted Hsp70 proteins and DEAD-box helicases, and possibly the ESCRT-associated Vps4 AAA ATPase to promote viral replication within the viral replication compartment [32,33].

Sustaining aerobic glycolysis pathway, however, requires the replenishing of NAD⁺, which is a critical regulatory compound in glycolysis [36,37]. Because a previous proteomic-based screen indicated that p92^pol replication protein binds to Pdc1 pyruvate decarboxylase fermentation protein [27], in this work we studied the role of the fermentation pathway in tombusvirus replication.

Aerobic glycolysis occurs in many fast-growing microbes, and in cancer cells, some embryonic cells and immune cells, such as fibroblasts and lymphocytes [36–38]. Aerobic glycolysis is hijacked by the malaria parasite [36]. It is a process that regulates the balance between fast ATP production and biosynthesis of ribonucleotides, lipids and several amino acids. However, aerobic glycolysis requires the replenishing of NAD⁺ compound, which is produced by fermentation in eukaryotic cells. NAD⁺ is converted to NADH during glycolysis by GAPDH, which is required for many biosynthetic processes [36–38].

Our surprising discovery is that tombusviruses co-opt the host Pdc1 and the alcohol dehydrogenase (Adh1) fermentation enzymes by re-localizing them from the cytosol into the large viral replication compartment through direct interaction with the tombusvirus replication proteins. The subversion of the fermentation enzymes is critical to support VRC assembly, the activation of p92 RdRp protein and viral RNA synthesis. Most important, however, is the role of the co-opted fermentation enzymes in the maintenance of ATP synthesis within the viral replication compartment. Altogether, we discovered that tombusviruses compartmentalize entire cellular metabolic pathways to promote intensive viral replication within the viral replication compartment at the expense of the infected host cells. Based on these and previous findings, we propose that compartmentalization of the fermentation pathway gives an evolutionary advantage for tombusviruses to replicate quickly to speed ahead of antiviral responses of the hosts.

Results

Pdc1 fermentation enzyme is required for tombusvirus replication

Pdc1 was previously identified in a co-purification assay with the TBSV p92^pol replication protein [27]. To test the relevance of PDC1 in tombusvirus replication, first we created a double mutant, which allowed for the depletion of Pdc1p in the absence of Pdc5p paralog in yeast (GAL::PDC1 pdc5Δ). Then, we used this double mutant yeast to launch TBSV replication via co-expressing the p33 and p92^pol replication proteins and the replicon repRNA. Northern blot analysis revealed a ~10-fold decrease in TBSV repRNA accumulation when Pdc1p expression was suppressed versus induced (Fig 1, lanes 13–15 versus 16–18). These data suggest that Pdc1p and Pdc5p are critical for TBSV replication. Expression of Pdc1p from a plasmid in pdc1Δ yeast increased TBSV repRNA accumulation by ~2-fold (Fig 1B). We observed similar ~2-fold enhanced replication of the closely-related CIRV, which replicates on the outer membranes of mitochondria (Fig 1C). Depletion of Pdc1p in double mutant yeast (GAL::PDC1 pdc5Δ) also inhibited the replication of the unrelated Flock House virus (FHV), which is an insect-infecting RNA virus (S1 Fig, lanes 17–20 versus 1–24). FHV replicates on the outer membranes of mitochondria [39]. These data suggest that Pdc1p has a pro-viral role in different subcellular microenvironments.

To test if the canonical enzymatic function of Pdc1p is needed for TBSV replication, we expressed Pdc1^S455F mutant, which has a reduced pyruvate decarboxylase activity [40], in pdc1Δ yeast replicating TBSV repRNA. Unlike the WT Pdc1p, Pdc1^S455F mutant was unable to enhance the replication of TBSV repRNA (Fig 1D, lanes 7–9 versus 4–6). Therefore, we suggest that the canonical role of Pdc1p in the fermentation pathway is required for efficient TBSV replication in yeast.

To obtain additional evidence for the pro-viral role of Pdc1p, we replaced the original promoter of PDC2 gene with the regulatable GAL1 promoter in the haploid yeast chromosome. Pdc2p is a transcription factor regulating the transcription of both PDC1 and PDC5 genes in yeast [41]. Depletion of Pdc2p in the above yeast (GAL1::HA-PDC2) resulted in ~3-fold reduction in TBSV repRNA accumulation (S2 Fig, lanes 13–16 versus 17–24). Therefore, these findings confirm the pro-viral role of Pdc1/5p in TBSV replication in yeast.

Pdc1 protein interacts with the tombusvirus replication proteins

To test if Pdc1p could interact with the tombusvirus replication proteins, we used the membrane yeast two-hybrid assay (MYTH), which is based on the split-ubiquitin strategy [42]. We found that the yeast Pdc1p and the homologous Arabidopsis AtPdc1 proteins interacted with the TBSV p33 replication protein (Fig 2A).

We then purified the TBSV replicase from yeast membrane fraction through detergent-solubilization and Flag-affinity purification. Interestingly, the yeast Pdc1p was co-purified with the Flag-p33/Flag-p92 replication proteins (Fig 2B). We also found that the enzymatically inactive Pdc1^S455F mutant was co-purified with the TBSV replicase from yeast (Fig 2C, lane 3). The mitochondrial CIRV Flag-p36/Flag-p95 showed a comparable co-purification profile with the WT Pdc1p and Pdc1^S455F mutant to that observed with the TBSV replication proteins (Fig 2C). The homologous AtPdc1 was also co-purified with either TBSV Flag-p33/Flag-p92 or the CIRV Flag-p36/Flag-p95 replication proteins from the mebrane-fraction of yeast (Fig 2D and 2E). Similar co-purification experiments with the Flag-p33 replication protein from detergent-solubilized fraction of Nicotiana benthamiana also confirmed the interaction of the replication proteins with AtPdc1 protein (Fig 2F). These data suggest that the interaction between Pdc1/AtPdc1 and p33 replication protein occurs in both yeast and plant cells.

To confirm direct interactions between the TBSV p33 and Pdc1p, we applied a pull-down assay with MBP-tagged Pdc1p or MBP-AtPdc1 and GST-His₆-tagged p33C (the C-terminal, soluble portion) proteins from E. coli (Fig 2G). Both MBP-Pdc1p and MBP-AtPdc1 captured the GST-His₆-p33C protein on the maltose-column, indicating direct interaction between these host and viral proteins. This conclusion was confirmed using the TBSV MBP-p33C or the CIRV MBP-p36C proteins, which captured the GST-His₆-AtPdc1 in the second pull-down assay (Fig 2H). In the pull-down assay, we used truncated TBSV p33 and CIRV p36 replication proteins missing their membrane-binding regions to aid their solubility in E. coli (Fig 2G and 2H). Altogether, these data suggest that the direct interactions between the replication proteins of TBSV and CIRV and Pdc1p/AtPdc1 host proteins occur within the viral protein C-terminal domain facing the cytosolic compartment.

To examine if Pdc1p was co-opted as a permanent or temporary component of the tombusvirus replicase, first, we stopped the formation of new tombusvirus replicase complexes by blocking ribosomal translation via adding cycloheximide to the yeast growth media [43]. Second, we performed Flag-affinity-purification of the tombusvirus replicase from the membrane fraction of yeast at various time-points. Interestingly, the amount of the co-purified Pdc1p was decreased by ~50% in the purified replicase preparations at the 2.5 h time point (Fig 2I, lanes 3–4 versus 2). The reduction of Pdc1p amount suggests that Pdc1p is likely released from the replicase. The release of Pdc1p likely occurs before the final assembly of the viral replicase, which ultimately forms a rather closed vesicle-like structure during replication [31,43,44]. Based on this observation, we suggest that the function of Pdc1p is temporary with the replication proteins, which likely takes place during the early steps of tombusvirus replication.

The Adh1 family of fermentation enzymes is co-opted for tombusvirus replication

The fermentation pathway consists of two different sets of enzymes, pyruvate decarboxylase (Pdc1/5 in yeast) and alcohol dehydrogenase (Adh1-5 in yeast). The end result of the pathway is ethanol, however, the critical product is NAD⁺ from NADH [36,38]. Importantly, NAD⁺ is required to replenish the glycolytic pathway via providing the regulatory compound of Glyceraldehyde-3-phosphate dehydrogenase (GAPDH, coded by TDH2/3 in yeast). Because the subversion of the catalytically active Pdc1 enzyme is needed to support TBSV replication (see above), we tested if the NAD⁺ producing Adh family members are also co-opted by tombusviruses.

To test if Adh1-5p could interact with the tombusvirus replication proteins, we used the MYTH assay [42], which revealed that the five members of the yeast Adh family as well as the homologous Arabidopsis AtAdh1 protein interacted with the TBSV p33 replication protein (Fig 3A). To confirm this unexpected finding, we purified the TBSV replicase from yeast membrane fraction through detergent-solubilization and Flag-affinity purification. We found that the yeast Adh1p, Adh2p and Adh3p were all co-purified with the Flag-p33/p92 replication proteins (Fig 3B). Adh1p was also co-purified with the mitochondrial CIRV Flag-p36/p95 replication proteins (Fig 3C), suggesting that different tombusviruses recruit Adh proteins into the membrane fraction of yeast. Similar co-purification experiments with the TBSV Flag-p33 or the CIRV Flag-p36 replication proteins from detergent-solubilized membranous fraction of yeast confirmed the interactions of the TBSV and CIRV replication proteins with AtAdh1 protein (Fig 3D). These data suggest that the interactions between Adh1/AtAdh1 and the tombusviral replication proteins occur in yeast cells.

To confirm direct interactions between TBSV p33 and Adh1p, we applied a pull-down assay using the TBSV MBP-p33C or the CIRV MBP-p36C proteins, which captured the GST-His₆-Adh1p in the pull-down assay (Fig 3E). Similar pull-down experiment also confirmed the direct interaction between GST-His₆-AtAdh1 and the viral replication proteins (Fig 3F). As above, we used the truncated TBSV p33 and CIRV p36 replication proteins missing their membrane-binding regions to increase their solubility in E. coli in the pull-down assay (Fig 3E–3F). Altogether, these data suggest that the direct interactions between the replication proteins of TBSV and CIRV and the Adh1 host protein occur within the viral protein domain facing the cytosol.

Both Pdc1 and Adh1 have pro-viral functions in plants

The homologous PDC1 and ADH1 genes are present in plants, but they are expressed at a very low level in plant cells under normal growth conditions [45,46]. Therefore, tombusviruses likely need to induce the expression of Pdc1 and Adh1 mRNAs in order to exploit Pdc1 and Adh1 for pro-viral functions during plant infections. Indeed, RT-PCR analysis of Pdc1 mRNA levels in TBSV-infected versus mock-treated N. benthamiana leaves revealed robust up-regulation of Pdc1 mRNA level in the TBSV inoculated leaves (Fig 4A) as well as the leaves expressing only the p33 replication protein (Fig 4C, lanes 1–3 versus 4–6). We observed a comparable up-regulation of Pdc1 mRNA level in CIRV infected N. benthamiana leaves or the leaves only expressing the CIRV p36 replication protein (Fig 4B and 4C). Based on these observations, we propose that TBSV and CIRV replication induces a high level of Pdc1 expression.

Similarly, RT-PCR analysis of Adh1 mRNA level in TBSV-infected versus mock-treated N. benthamiana leaves revealed an up-regulation of Adh1 mRNA level in the TBSV and CNV inoculated leaves (Fig 4D–4F) and the systemically-infected leaves (Fig 4E). Based on these observations, we propose that TBSV and CNV replication induces a high level of Adh1 expression in plant leaves.

To study if tombusviruses depend on the Pdc1 function in plants, we knocked-down Pdc1 expression via virus-induced gene-silencing (VIGS) in N. benthamiana plants. Knockdown of Pdc1 in N. benthamiana resulted in a ~3-fold reduction of TBSV RNAs in the inoculated leaves (Fig 5A). Knockdown of Pdc1 level did not cause an obvious phenotype in N. benthamiana (Fig 5A). To test the effect of Pdc1 depletion on virus accumulation in the absence of cell-to-cell spread, we also tested TBSV replication in Pdc1 knockdown protoplasts. Interestingly, TBSV RNA accumulation was reduced by ~7-fold in Pdc1 knockdown protoplasts in comparison with control protoplasts (Fig 5B), suggesting that Pdc1 affects the viral replication process.

Similar experiments with CIRV in Pdc1 knockdown N. benthamiana plants also revealed a ~3-fold reduced level of tombusvirus accumulation (Fig 5C). These data confirmed the pro-viral role of Pdc1 in supporting tombusvirus replication in plants. To test if the pro-viral function of Pdc1 is also exploited by a more distantly-related carmovirus, we measured the accumulation of turnip crinkle virus (TCV) in Pdc1 knockdown N. benthamiana plants. The accumulation of TCV RNAs decreased by ~3-fold in Pdc1 knockdown plants (Fig 5D). It seems that tombusviruses and a carmovirus can exploit Pdc1 functions to support viral replication.

Knockdown of Adh1 in N. benthamiana resulted in a ~2-fold reduction of TBSV RNAs in the inoculated leaves (Fig 6A). Knockdown of Adh1 in N. benthamiana also reduced the accumulation of the peroxisomal-replicating CNV and the mitochondrial-replicating CIRV by ~2-fold (Fig 6B and 6C). These data confirmed the pro-viral role of Adh1 in supporting tombusvirus replication in plants.

Both Pdc1 and Adh1 proteins are recruited into the tombusvirus replication compartment in plants

To determine if Pdc1 is recruited by TBSV into the extensive viral replication compartment, we co-expressed the BFP-tagged TBSV p33 replication protein and the RFP-tagged AtPdc1 with the GFP-SKL peroxisome matrix protein in N. benthamiana leaves, followed by confocal imaging. These experiments revealed a high level of co-localization of the TBSV p33 replication protein and the RFP-AtPdc1 within the replication compartments consisting of aggregated peroxisomes, even in the absence of TBSV replication (Fig 7A). We observed a similar re-distribution of the RFP-AtPdc1 in the presence of CIRV p36-BFP within the replication compartments consisting of aggregated mitochondria (Fig 7B). Therefore, we suggest that the TBSV p33 and the CIRV p36 replication proteins alone are enough to recruit Pdc1 to the replication compartment to a similar extent as the actively replicating TBSV or CIRV (Fig 7A and 7B). In the absence of viral components, AtPdc1 is localized in the cytosol (Fig 7C). Based on these experiments, we propose that Pdc1 is efficiently recruited by the tombusvirus replication proteins to the extensive tombusvirus replication compartments in plants.

Similar co-localization experiments revealed a high level of co-localization of the TBSV p33-RFP replication protein and the BFP-AtAdh1 within the replication compartments consisting of aggregated peroxisomes in the absence or presence of TBSV replication (Fig 7D). We also observed re-distribution of the BFP-AtAdh1 in the presence of CIRV p36-RFP within the replication compartments consisting of aggregated mitochondria (Fig 7E). Based on these observations, we suggest that the TBSV p33 and the CIRV p36 replication proteins alone are capable of recruiting AtAdh1 to the replication compartment (Fig 7). In the absence of viral components, AtAdh1 is localized in the cytosol (Fig 7F). These experiments support a model that Adh1 is efficiently recruited by the tombusvirus replication proteins to the extensive tombusvirus replication compartments in plants.

To demonstrate whether AtPdc1 is recruited into the TBSV replication compartment, which actively replicates the viral RNAs, we utilized a modified repRNA carrying an ssRNA sensor [47]. This sensor consists of six repeats of a hairpin RNA from MS2 bacteriophage, which is specifically recognized by the MS2 coat protein (MS2-CP) [48]. Co-expression of the TBSV p33-BFP with the GFP-tagged AtPdc1 and the RFP-tagged MS2-CP revealed the re-localization of AtPdc1 to the active TBSV replication compartment containing the new (+)repRNA product (Fig 8A and S3A Fig) or the (-)repRNA, which is part of the replication intermediate (Fig 8C). In the control experiments, in the presence of only the TBSV repRNA and p33-BFP (no replication due to the absence of p92^pol replication protein), AtPdc1 was still localized in the viral replication compartment with p33, whereas RFP-MS2-CP was located in the nucleus (Fig 8B and 8D and S3B Fig). Therefore, we conclude that Pdc1 is present at the sites of tombusvirus replication and Pdc1 likely plays a role in the formation of the tombusvirus replication compartments.

Similar experiments with AtAdh1 revealed the re-localization of AtAdh1 to the active TBSV replication compartment containing the new (+)repRNA product (Fig 8E) or the (-)repRNA, which is part of the replication intermediate (Fig 8G and S3C Fig). In the control experiments, when only the TBSV repRNA and p33-BFP were expressed without the p92^pol replication protein, then AtAdh1 was still re-localized into the viral replication compartment with p33, but RFP-MS2-CP was located in the nucleus (Fig 8F and 8H and S3D Fig). Therefore, we conclude that AtAdh1, similar to AtPdc1, is present at the sites of active tombusvirus replication.

To provide additional evidence that the AtPdc1 is recruited into the viral replication compartments through interacting with the TBSV p33 or CIRV p36 replication proteins, we have conducted bimolecular fluorescence complementation (BiFC) experiments in N. benthamiana leaves. The BiFC experiments revealed robust interactions between AtPdc1 and either the TBSV p33/p92^pol or the CIRV p36 replication proteins within the replication compartment (Fig 9A, see also S4 Fig for the negative control BiFC experiments).

Similar BiFC experiments in N. benthamiana leaves revealed interactions between AtAdh1 and the TBSV p33/p92^pol or the CIRV p36 replication proteins within the replication compartment (Fig 9B). These data confirmed the replication protein-driven re-localization of AtPdc1 and AtAdh1 into the viral replication compartment.

Pdc1 is required for efficient tombusvirus replication in vitro

To obtain direct evidence of the role of Pdc1 in TBSV replication, we used an in vitro replicase reconstitution assay based on a cell-free extract (CFE) from GAL::PDC1 pdc5Δ yeast strain with depleted Pdc1p level. Programming the CFE with the (+)repRNA and purified replication proteins led to ~3-to-4-fold reduced replication, including the production of both dsRNA replication intermediate and the (+)repRNA progeny when compared with CFE prepared from the same yeast strain with induced Pdc1p expression (Fig 10A). Based on these data, we suggest that Pdc1p is required for robust replication and likely during the replicase assembly step since all TBSV repRNA products were reduced when Pdc1p was depleted.

We also performed another approach to test the efficiency of replicase assembly in yeast, which is based on the purification of the tombusvirus replicase from yeast, followed by in vitro RdRp assay with added template RNA. The purified replicase prepared from GAL::PDC1 pdc5Δ yeast strain with depleted Pdc1p level had only ~50% activity of that obtained from the same yeast strain with induced Pdc1p expression on both (-) and (+)RNA templates (Fig 10B). Because the replicase has to pre-assemble in yeast in this approach, the reduced activity of replicase with depleted Pdc1p is likely due to defect in the replicase assembly.

Interestingly, the replicase assembly involves the activation of the p92 RdRp, which depends on several viral- and host factors, most notably the Hsp70 protein chaperone [27]. We have tested the p92 RdRp activation in a simplified in vitro assay, based on a purified N-terminally-truncated p92 RdRp and the soluble fraction of yeast CFEs, which should provide the needed host components [27,28]. We observed a ~40% reduction in p92 RdRp activation when the CFE was derived from the double-mutant GAL::PDC1 pdc5Δ yeast strain with depleted Pdc1p level versus the CFE obtained from the same yeast strain with induced Pdc1p expression (Fig 10C). This reduction might indicate a low-level activity for the ATP-dependent Hsp70 in CFE, which could be due to reduced ATP production by glycolysis in yeast with a depleted Pdc1p level (see Discussion). Overall, all in vitro assays suggest the direct involvement of Pdc1 in tombusvirus replication, which is likely due to reduced ATP production by glycolysis.

Robust generation of ATP by glycolysis in the tombusvirus replication compartment is dependent on the co-opted Pdc1 and Adh1 in yeast and plants

Because several pro-viral co-opted host proteins, such as Hsp70, the ESCRT-associated Vps4 AAA ATPase and DEAD-box helicases, require plentiful ATP within the replication compartment to fuel robust viral replication [43,49–53], it is possible that the co-opted Pdc1 and Adh1 are needed within the replication compartment to rapidly supply the NAD⁺ substrate. NAD⁺ is critical to replenish the glycolytic pathway, which is dependent on reducing NAD⁺ to NADH via the co-opted GAPDH (Tdh2/2p in yeast, GAPC in plants) [36,38]. The facts that both fermentation enzymes are recruited to the sites of virus replication and the catalytic activity of Pdc1 is required for its pro-viral function (Fig 1D), also support this hypothesis.

To estimate the ATP level within the tombusviral replication compartment, we used a FRET-based biosensor [54], which was previously adapted to estimate ATP levels [32,33]. Briefly, ATeam- p92^pol can measure ATP level due to the conformational change in the enhanced ε subunit of the bacterial F₀F₁-ATP synthase upon ATP binding [32,33]. The ε subunit bound to ATP draws the CFP and YFP fluorescent tags in close vicinity, increasing the FRET signal in confocal laser microscopy (Fig 11A). On the contrary, the ε subunit in the ATP-free form is present in an extended conformation, which places CFP and YFP tags in a distal position, thus reducing the FRET signal (Fig 11A) [54]. We found previously [32,33] that the ATeam-tagged p92^pol is a fully functional RdRp, which localizes to the viral replication compartment representing aggregated peroxisomes. Since these experiments are best performed in the presence of glucose in yeast media [33], we used pdc1Δ yeast strain expressing Pdc1p from a plasmid. We found that the ATP level was ~4-fold higher in pdc1Δ yeast strain expressing WT Pdc1p than in the control lacking PDC1 or expressing Pdc1^S455F mutant (Fig 11B). Expression of the low-sensitive variant of ATeam (ATeam^RK-p92) [33,54] in pdc1Δ yeast strain expressing Pdc1p showed low FRET values, confirming that the FRET data is derived from the ATP biosensor in this assay. Overall, the obtained data support the model that Pdc1p is recruited into the tombusvirus replication compartment in yeast to facilitate the generation of ATP locally for viral RNA synthesis.

To confirm that tombusviruses co-opt Pdc1 into the viral replication compartment to support efficient ATP generation in plants, we expressed ATeam-p33 replication protein in N. benthamiana leaves, which were either silenced for Pdc1 expression or not (Fig 12A). The obtained data showed up to a ~4-fold reduction in ATP production within the viral replication compartment in the Pdc1 knockdown plants versus the control plants (Fig 12B). Similar experiments with N. benthamiana infected with TBSV showed a ~3-fold reduction in ATP level within the viral replication compartment in the knockdown plants versus the control plants (Fig 12C). Intensive TBSV replication likely uses up some of the produced ATP in the latter experiments as we observed previously [33]. Applying the same approach showed that similar pictures on ATP production within the viral replication compartment in the Pdc1-knockdown plants versus the control plants exist during the peroxisomal CNV (Fig 12D) and the mitochondrial CIRV (Fig 13) infections of N. benthamiana leaves. However, we did observe a range in ATP production [between ~3-fold (Fig 13A) and ~2-fold (S5 Fig)] in the absence of CIRV replication within the viral replication compartment in the Pdc1-knockdown plants versus the control plants in different experiments. It is possible that glucose concentrations and/or the efficiency of fermentation within the viral replication compartments in leaves are influenced by several physiological processes in plants. Nevertheless, the emerging picture is that subversion of Pdc1 into the viral replication compartment is required to support efficient ATP generation locally.

Because Pdc1 works together with Adh1 in the fermentation pathway, we were curious if tombusviruses also co-opt Adh1 into the viral replication compartment to support efficient ATP generation locally in plants. Therefore, we expressed the ATeam-p33 replication protein in N. benthamiana leaves, which were either silenced for Adh1 expression or not. We observed a ~2-fold reduction in ATP production within the viral replication compartment in the Adh1 knockdown plants versus the control plants (Fig 14A). Similar experiments with N. benthamiana infected with TBSV or CIRV also showed a ~2-fold reduction in ATP-level within the viral replication compartment in the knockdown plants versus the control plants (Fig 14). Based on these findings, we propose that subversion of both Pdc1 and Adh1 fermentation enzymes by tombusviruses facilitates the glycolytic process to produce plentiful ATP locally within the replication compartment in N. benthamiana leaves.

Dependence of bamboo mosaic virus and tobacco mosaic virus replication on the fermentation pathway in plants

To investigate if additional plant viruses also depend on the fermentation pathway for robust replication, we chose bamboo mosaic virus (BaMV), a potexvirus, and tobacco mosaic virus (TMV), a tobamovirus, which are unrelated to TBSV.

We found that BaMV and TMV replication led to the efficient induction of both Pdc1 mRNA and Adh1 mRNA expression in the inoculated as well as the systemically-infected N. benthamiana leaves (Fig 15A and 15B and S6A and S6B Fig). VIGS-based silencing of Pdc1 level in N. benthamiana leaves resulted in a ~60% reduction in the accumulation of both BaMV and TMV gRNAs (Fig 15C and S6C Fig). Similarly, knocking down Adh1 level in N. benthamiana leaves reduced the accumulation of BaMV and TMV gRNAs by 70% and 60%, respectively (Fig 15C and S6C Fig). To test if BaMV can recruit the fermentation proteins directly through protein-protein interactions, we used a BiFC approach. Co-expression of either the capping enzyme domain or the helicase domain of the BaMV replicase with Pdc1 in N. benthamiana leaves resulted in punctate and cytosolic signals, respectively (Fig 15D). In contrast, co-expression of the RdRp domain of the BaMV replicase with Pdc1 did not produce signals, suggesting the lack of interaction. Interestingly, we also observed interaction between the capping enzyme domain or the helicase domain, but not the RdRp domain of the BaMV replicase with Adh1 in N. benthamiana leaves (Fig 15E). Therefore, it is possible that BaMV also exploits the fermentation pathway via interaction between the viral replicase and the fermentation enzymes. Based on these observations, we suggest that similar to tombusviruses, other unrelated and rapidly replicating plant viruses also depend on the fermentation pathway in plants. Further experiments will be needed on the mechanistic details on the role of the fermentation pathway in the replication of BaMV and TMV.

Discussion

Compartmentalization of the co-opted fermentation pathway in the tombusviral replication compartment to support tombusvirus replication

The viral replication proteins orchestrate the biogenesis of the large tombusviral replication compartment, including the numerous spherules/VRCs, which represent the sites of viral RNA replication [22,31,55]. The formation and operation of these virus-driven structures require subversion of numerous cellular proteins, membrane deformation, membrane proliferation, changes in lipid composition of the hijacked cellular membranes and intensive viral RNA synthesis. To obtain the necessary resources from the infected cells, tombusviruses have to rewire cellular pathways to fuel the biogenesis of the replication compartment. These robust processes require plentiful ATPs and molecular building blocks produced at the sites of replication or delivered there. The emerging picture with tombusviruses is that by co-opting the aerobic glycolysis, the ATP molecules are produced and utilized within the replication compartment [32,33]. However, the aerobic glycolysis requires the replenishing of the NAD⁺ pool, which is used by the glycolytic GAPDH to produce NADH. NAD⁺ is efficiently generated by the fermentation pathway, which also utilizes pyruvate, the end product of the glycolytic pathway [36,38]. Accordingly, in the current work we show the efficient recruitment of Pdc1 and Adh1 fermentation enzymes into the viral replication compartment. Depletion of Pdc1 combined with deletion of the homologous PDC5 in yeast or knockdown of Pdc1 and Adh1 in plants reduced the efficiency of tombusvirus replication. A complementation approach revealed that the enzymatically functional Pdc1p is required to support tombusvirus replication. We provide evidence that both Pdc1 and Adh1 enzymes are required for efficient generation of ATP within the replication compartment based on the measurements with an ATP biosensor inside the viral replication compartment (Figs 11–14). Moreover, in vitro works show the pro-viral function of Pdc1 during the assembly of the viral replicase and the activation of the p92 RdRp, both of which require the co-opted ATP-driven Hsp70 protein chaperone.

Is the co-opted fermentation pathway only required for facilitating ATP production within the tombusviral replication compartment? Albeit not studied in this work, it is very likely that the co-opted aerobic glycolysis in combination with the subverted fermentation pathway also provide plentiful metabolic precursors, which could be utilized by the cell to make molecular building blocks, such as ribonucleotides, lipids and amino acids [36,37]. These newly made molecular building blocks are likely exploited by tombusviruses to build the viral replication compartment and support intensive viral RNA synthesis. Accordingly, high glucose concentration stimulates TBSV replication in yeast, whereas blocking the aerobic glycolysis with 2DG compound strongly inhibited TBSV accumulation in yeast and plants [56].

Why are the relatively inefficient aerobic glycolytic and fermentation pathways co-opted by tombusviruses? These metabolic pathways are present in the cytosol, thus easily accessible for subversion by the cytosolic tombusviruses. Moreover, the ATP generation by the aerobic glycolytic and fermentation pathways is fast if plentiful glucose is present in the cells. Plants produce plentiful glucose based on chloroplasts, thus glucose is not expected to be rate limiting for the aerobic glycolytic and fermentation pathways in the infected plant cells. Moreover, these metabolic pathways do not require free oxygen, which could be an advantage for tombusviruses that also replicate efficiently in plant roots. Moreover, tombusviruses require the synthesis of new phospholipids and ribonucleotides [26,57]. The nexus point of the metabolic pathways, which is pyruvate, the end-product of glycolysis, has to be re-routed into the fast fermentation pathway. This then leads to the rapid regeneration of NAD⁺ to replenish the glycolytic pathway. NAD⁺ is also necessary for the biosynthesis of nucleotides and amino acids, and the fermentation pathway supports fast glucose flux through glycolysis. Thus, the rapid regeneration of NAD⁺ allows fast incorporation of glucose into metabolites [36–38]. Altogether, by providing plentiful precursor compounds in the cytosol, the aerobic glycolytic and fermentation pathways are far more efficient to facilitate the production of molecular building blocks than the oxidative phosphorylation pathway [36–38]. Then, the generated new metabolites can be exploited by tombusviruses to build extensive replication.

Why is compartmentalization of the aerobic glycolytic and fermentation pathways in the replication compartment advantageous for tombusviruses? The combined subversion of the aerobic glycolytic and fermentation pathways allows for the rapid production of ATP locally, including replenishing of the regulatory NAD⁺ pool by the fermentation pathway. Then, the locally produced ATP could be used efficiently by the co-opted ATP-dependent host factors, such as the Hsp70 protein chaperone, the ESCRT-associated Vps4 AAA ATPase and the pro-viral DEAD-box helicases [32,33]. These co-opted host factors are required for pro-viral processes, including VRC assembly, the activation of p92 RdRp, and the utilization of both ssRNA templates and dsRNA replication intermediates for viral RNA synthesis [22,32,33]. By producing the ATP locally within the replication compartment, tombusviruses do not need to compete with cellular processes for the common ATP pool and all the molecular processes could be accelerated by the high local concentration of ATP within the replication compartment. It is also possible that the feedback regulation of these metabolic processes by the cell is less efficient when compartmentalized in the viral replication compartment. Overall, there is an evolutionary pressure for tombusviruses to replicate fast and speed ahead of antiviral responses of the hosts and to outcompete other pathogenic viruses. Therefore, there are numerous advantages for tombusviruses to subvert the cellular aerobic glycolytic and fermentation pathways to support the infection process.

Aerobic glycolysis is induced during cancer and other diseases as well, including type 2 diabetes, amyloid-based brain diseases, wound repair and oncogenic virus infections [58–61]. Switching to the aerobic glycolytic metabolism can also occur with healthy cells, for example, during Endothelial cell differentiation, monocytes-based trained immunity, in rapidly dividing cells during embryogenesis, during T cell differentiation and motor adaptation learning in human brain [37,58,62,63]. The fetal heart primarily produces ATP via glycolytic metabolism [38]. All these cells/tissues utilize aerobic glycolysis as a metabolic compromise to provide ATP and produce enough new metabolic compounds to perform their functions.

In summary, we show evidence that TBSV exploit the fermentation pathway to support rapid virus replication. The dependence on the fermentation pathway is also shown for several other related and unrelated plant viruses. These viruses induce the fermentation pathway, thus indicating that a broad range of viruses takes advantage of the rapid cytosolic generation of ATP and numerous metabolic precursors. It will also be interesting to learn if other (+)RNA viruses exploit the aerobic glycolytic and fermentation pathways for their replication. Because all plant, animal and human (+)RNA viruses require the biogenesis of the membranous viral replication compartment/organelle, thus they likely use plenty ATP and they might depend on the production of new metabolic precursors, it is possible that hijacking the aerobic glycolytic and fermentation pathways occurs in other viruses as well. This could open up new common antiviral strategies targeting the fermentation pathway.

Materials and methods

Plant materials, yeast strain and plasmids

Wild type N. benthamiana plants were potted in soil and placed in growth room at 25°C under a 16-h-light/8-h-dark cycle. S. cerevisiae strain BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) was purchased from Open Biosystems. Yeast strains pdc1Δ was from the YKO library (Openbiosystems). To create pdc5Δ yeast strain, the hygromycin resistance gene hphNTI was PCR-amplified from vector pFA6a–hphNT1 (Euroscarf) [64] with primers #7504 and primers #7505 and the PCR product was transformed into BY4741. To generate GAL1::PDC1 pdc5Δ and GAL1::HA-PDC2 yeast strains, the transformants with GAL1 promoter along with the nourseothricin resistance gene were PCR-amplified from plasmid pYM-N23 with primers #7508 and #7509 or from pYM-N24 with #7475 and #7476 and then transformed into pdc5Δ and BY4741 yeast strains, respectively. Yeast strain NMY51 was obtained from Dualsystems. Plasmids and their constructions are listed in S1 and S2 Tables and primers used are described in S3 Table.

Analysis of virus replication in yeast

To determine the effect of Pdc1 on the replication of TBSV in yeast, BY4741, pdc5Δ and GAL1::PDC1 pdc5Δ strains were transformed with HpGBK-CUP1-Flagp33, LpGAD-CUP1-Flag92 and UpCM189-Tet-DI72. TBSV replication was induced by growing cells at 23°C in SC-ULH⁻ (synthetic complete medium without uracil, leucine and histidine) medium supplemented with 2% galactose or 2% raffinose for 16 h. Then, yeast cultures were resuspended in SC-ULH⁻ medium supplemented with 50 μM CuSO₄ and 2% galactose or 2% raffinose, and grown for 24 h at 23°C.

To complement TBSV or CIRV replication with Pdc1 in pdc1Δ yeast strain, plasmids HpGBK-CUP1-Hisp33/Gal-DI72 and LpGAD-CUP1-Hisp92 or HpESC-CUP1-Flagp36/Gal-DI72 and LpESC-CUP1-Flagp95, respectively, were co-transformed with UpCM189-Tet-empty or UpCM189-Tet-HisPdc1 or UpCM189-Tet-Pdc1 into yeast strain. To test if the enzymatic function of Pdc1 is required for TBSV replication, plasmids HpGBK-CUP1-Hisp33/Gal-DI72, LpGAD-CUP1-Hisp92 and UpCM189-Tet-Pdc1^S455F were co-transformed into pdc1Δ yeast strain. Transformed yeast cells were pre-grown in 2 ml SC-ULH⁻ medium supplemented with 2% galactose and 100 μM BCS for 16 h at 23°C. Then, yeast cultures were resuspended in SC-ULH⁻ medium supplemented with 2% galactose and 50 μM CuSO₄ and grown for 24 h at 23°C.

Co-purification assay

To understand the dynamics of Pdc1 association with the viral replicase, transformed yeast cells were pre-grown in SC-ULH⁻ medium supplemented with 2% glucose and 100 μM BCS at 29°C for 16 h. Then yeast cultures were centrifuged and the pellets were resuspended in SC-ULH⁻ medium supplemented with 2% galactose and 100 μM BCS and grown at 23°C for 24 h, followed by culturing yeast cells in SC-ULH⁻ medium supplemented with 2% galactose and 50 μM CuSO₄ at 23°C for 6 h. Next, the yeast cells were shifted to SC-ULH⁻ medium supplemented with 2% glucose and cycloheximide (100 μg/ml) and samples were taken at 0, 1 h and 2.5 h time points. Yeast cultures were treated with formaldehyde and glycine and performed Flag-immunoaffinity purification as described below.

Co-purification assay from plants was performed by slight modifications of a previously described method [65]. Briefly, N. benthamiana leaves were co-infiltrated with agrobacterium carrying pGD-HA-AtPdc1, pGD-GFP-HA, pGD-T33-Flag, pGD-p19 and pGD-empty. Then, samples were harvested at 2.5 days post agroinfiltration and ground in cooled mortar in PPEB buffer (10% [v/v] glycerol, 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, 10 mM DTT, 0.5% [v/v] Triton X-100 and protease inhibitor cocktail). The supernatant was incubated with anti-FLAG M2 affinity agarose (Sigma-Aldrich) in Bio-spin chromatography columns (Bio-rad) for 2 h at 4°C on a rotator, followed by washing with the CP buffer (10% [v/v] glycerol, 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, 1mM DTT and 0.1% [v/v] Triton X-100). Elutions of the purified proteins were as described in the co-purification assay in yeast [66].

Knockdown of NbPdc1 and NbAdh1 in N. benthamiana plants by VIGS

The VIGS-based knockdown of host genes in N. benthamiana was performed as described previously [65]. To generate VIGS constructs TRV2-NbPdc1 and TRV2-NbAdh1, cDNA fragments were PCR-amplified with primers #5847/#5848 and #7911/#7912 from N. benthamiana cDNA preparations and inserted into the plasmid pTRV2 [67]. At 12 days after VIGS treatment of N. benthamiana (pTRV1 together with pTRV2-NbPdc1 or pTRV2-NbAdh1 or pTRV2-cGFP or TRV-LUC), the levels of N. benthamiana NbPdc1 and NbAdh1 mRNAs were determined by semi-quantitative RT-PCR. Then, the silenced leaves were either sap inoculated with TBSV, CIRV or TCV inocula or agroinfiltrated with pGD-CNV^20KSTOP or pGD-CIRV, to launch virus replication. At different time points, samples from the inoculated and systemically-infected leaves were collected, followed by total RNA extraction and northern blot analysis as described previously [65]. In case of BaMV, the VIGS-silenced leaves were sap-inoculated with BaMV inocula to launch viral replication. Then, the inoculated leaves were collected at 2.5 dpi, followed by total RNA extraction and quantitative real-time RT-PCR analysis as described previously [68]. The N. benthamiana EF1α gene was used as an internal control to normalize the level of viral gene expression.

The VIGS-silenced NbPdc1 and NbAdh1 leaves of N. benthamiana were agroinfiltrated with pJL-36 vector carrying TMV cDNA and plant samples were collected 2 days after infection from the inoculated leaves. The TMV RNA levels were measured by northern blot analysis.

Plant protoplasts preparation and viral RNA transfection

Protoplasts preparation from plant leaves was performed by some modifications of a previously described method [69]. Briefly, the NbPdc1-silenced or the mock-treated leaves were harvested at 12 days post VIGS silencing. Then, the leaves were sliced into 0.5–1 mm strips, digested with an enzyme solution containing 1.2% [w/v] Cellulase, 0.16% [w/v] Macerozyme, 0.12% [w/v] BSA and 0.5 M mannitol. To improve the isolation of protoplasts, leaf strips were vacuum infiltrated for 20 min in the dark using Vacufuge Plus (Eppendorf) and further digested in the dark for at least 3 h at room temperature. The protoplasts preparations were passed through a sieve set (Scienceware Mini-Sieve Microsieve Set from Fisher cat# 14-306A) and collected by centrifugation at 900 rpm for 2 min, followed by washing once with the W5 solution (154 mM NaCl, 125 mM CaCl₂, 5 mM KCl, 2 mM MES pH5.7) and re-suspending in the W5 solution. Then, 0.6 M sucrose was layered under the W5 solution with protoplasts and centrifuged at 900 rpm for 3 min. Protoplasts were transferred from the interface between the W5 solution and 0.6 M sucrose layers in the same amount of the W5 solution, followed by washing once with the W5 solution and re-suspending at 2 x10⁵ ml^–1 in the MMG solution (4 mM MES pH5.7, 0.4 M mannitol and 15 mM MgCl₂). For RNA transfection, protoplasts were incubated with the PEG-calcium transfection solution containing 40% PEG 4000, 0.2 M mannitol and 100 mM CaCl₂ and either viral RNA transcripts or total RNA extracts obtained from virus-infected plants at room temperature for up to 15 min. The transfection mixtures were diluted with the W5 solution and centrifuged at 100 g for 2 min at room temperature and incubated in the WI solution (4 mM MES, pH 5.7, 0.5 M mannitol and 20 mM KCl). The protoplasts were harvested at 16 h or 24 h post-transfection and subjected to RNA extraction and northern blot analysis as mentioned above.

Visualization and measurement of ATP levels in yeast and plants

To visualize ATP production within the TBSV replication compartments in yeast, the previously adapted ATeam-based biosensor LpGAD-ADH-ATeam^YEMK-p92 (high sensitivity) and LpGAD-ADH-ATeam^RK-p92 (low sensitivity) were utilized [33,54]. pdc1Δ yeast strain was co-transformed with HpGBK-CUP1-Hisp33 and UpCM189-Tet-HisPdc1 or UpCM189-Tet-HisPdc1^S455F or UpCM189-Tet. The transformed yeast cells were pre-grown in 2 ml SC-ULH⁻ medium supplemented with 2% glucose and 100 μM BCS for 16 h at 23°C. Then, the yeast cultures were re-suspended in SC-ULH⁻ medium supplemented with 2% glucose and 50 μM CuSO₄ and grown for 3 h at 23°C. Then samples were collected for confocal laser microscopy analysis. FRET values (YFP/CFP ratio) were obtained based on the quantification of CFP and Venus images using ImageJ software and calculation using Microsoft Excel as described [33].

To measure the ATP level within the tombusvirus replication compartments in the NbPdc1- or the NbAdh1-silenced N. benthamiana leaves, the previously adapted ATeam-based biosensor pGD-p33-ATeam^YEMK or pGD-p36-ATeam^YEMK [33] were transformed into agrobacterium strain C58C1. In case of TBSV, the silenced or the control leaves of N. benthamiana (at 12 days after VIGS treatment using pTRV1 in combination with pTRV2-NbPdc1 or pTRV2-NbAdh1 or pTRV2-cGFP control) were co-agroinfiltrated with plasmids pGD-p33-ATeam^YEMK with or without pGD-p92 and pGD-DI72. In case of CNV, the silenced or the control leaves were co-agroinfiltrated with plasmids pGD-p33-ATeam^YEMK with or without pGD-CNV-^20KSTOP. In case of CIRV, the silenced or control leaves were co-agroinfiltrated with plasmids pGD-p36-ATeam^YEMK with or without pGD-p95 and pGD-DI72. Then samples were harvested at 1.5 day after agroinfiltration for the confocal microscopy analysis. FRET values (YFP/CFP ratio) were obtained based on the quantification of CFP and Venus images using ImageJ software and calculation using Microsoft Excel [33].

Yeast cell free extract (CFE)-based in vitro replication assay

CFEs from BY4741, pdc5Δ and GAL1::PDC1 pdc5Δ yeast strains were prepared as described earlier [53,70]. These yeast stains were pre-grown in YPD or YPG media at 29°C for 16 h. Then, the yeast cultures were diluted (to 0.4 OD₆₀₀) with fresh YPD or YPG media and grown at 29°C for 5 h, followed by 37°C treatment for 30 min. The individual CFE preparations were made following the published protocol [53,70] and adjusted to contain comparable amounts of total proteins. The in vitro CFE assay was performed in 20 μl total volume containing 2 μl of adjusted CFE, 0.5 μg DI-72 (+)RNA transcripts, 0.5 μg affinity-purified MBP-p33, 0.5 μg affinity-purified MBP-p92 (both recombinant proteins were obtained from E. coli) [71,72], 30 mM HEPES-KOH, pH 7.4, 150 mM potassium acetate, 5 mM magnesium acetate, 0.13 M sorbitol, 0.2 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 10 mM ATP, CTP, and GTP and 0.1 mM UTP and 0.2 μl of ³²P-UTP. Reaction mixtures were incubated for 3 h at 25°C, followed by phenol/chloroform extraction and isopropanol/ammonium acetate (10:1) precipitation. The ³²P-UTP-labeled RNA products were analyzed in 5% acrylamide/8 M urea gels [53,70]. Additional methods used are described in S1 text.

Supporting information

S1 Fig [a]
FHV replication depends on the expression of Pdc1/5 fermentation enzymes in yeast.

S2 Fig [a]
Pdc2 transcription factor, which regulates the expression of fermentation enzymes, is an essential host factor for tombusvirus replication in yeast.

S3 Fig [pdf]
Additional experiments to show the recruitment of Pdc1 and Adh1 to the sites of tombusviral replication in . .

S4 Fig [a]
Negative control experiments for the BiFC studies.

S5 Fig [pdf]
The co-opted cellular Pdc1 fermentation enzyme affects ATP accumulation locally within the CIRV replication compartment in . .

S6 Fig [a]
Dependence of TMV replication on Pdc1 and Adh1 proteins in . .

S1 Text [docx]
Experimental procedures.

S1 Table [docx]
Plasmids constructed in this study.

S2 Table [docx]
Plasmids described in previous studies.

S3 Table [docx]
Primers used in this study.