Kinetics and functional studies on interaction between the replicase proteins of Tomato Bushy Stunt Virus: requirement of p33:p92 interaction for replicase assembly

The assembly of the functional replicase complex via protein:protein and RNA:protein interactions among the viral-coded proteins, host factors and the viral RNA on cellular membranes is a key step in the replication process of plus-stranded RNA viruses. In this work, we have characterized essential interactions between p33:p33 and p33:p92 replication proteins of Tomato bushy stunt virus (TBSV), a tombusvirus with a non-segmented, plus-stranded RNA genome. Surface plasmon resonance (SPR) measurements with purified recombinant p33 and p92 demonstrate that p33 interacts with p92 in vitro and that the interaction requires the S1 subdomain, whereas the S2 subdomain plays lesser function. Kinetic SPR analyses showed that binding of S1 subdomain to the C-terminal half of p33 takes place with moderate binding affinity in the nanomolar range whereas S2 subdomain binds to p33 with micromolar affinity. Using mutated p33 and p92 proteins, we identified critical amino acid residues within the p33:p92 interaction domain that play essential role in replication and the assembly of the tombusviral replicase. In addition, we show that interaction takes place between replication proteins of TBSV and the closely related Cucumber necrosis virus but not between TBSV and the more distantly related Turnip crinkle virus, suggesting that selective protein interactions might prevent the assembly of chimeric replicases carrying replication proteins from different viruses during mixed infections.     

Yeast as a model host to study replication and recombination of defective interfering RNA of Tomato bushy stunt virus

Defective interfering (DI) RNA associated with Tomato bushy stunt virus (TBSV), which is a plus-strand RNA virus, requires p33 and p92 proteins of TBSV or the related Cucumber necrosis virus (CNV), for replication in plants. To test if DI RNA can replicate in a model host, we coexpressed TBSV DI RNA and p33/p92 of CNV in yeast. We show evidence for replication of DI RNA in yeast, including (i) dependence on p33 and p92 for DI replication; (ii) presence of active CNV RNA-dependent RNA polymerase in isolated membrane-containing preparations; (iii) increasing amount of DI RNA(+) over time; (iv) accumulation of (-)stranded DI RNA; (v) presence of correct 5' and 3' ends in DI RNA; (vi) inhibition of replication by mutations in the replication enhancer; and (vii) evolution of DI RNA over time, as shown by sequence heterogeneity. We also produced evidence supporting the occurrence of DI RNA recombinants in yeast. In summary, development of yeast as a host for replication of TBSV DI RNA will facilitate studies on the roles of viral and host proteins in replication/recombination.     

Phosphorylation of the p33 replication protein of Cucumber necrosis tombusvirus adjacent to the RNA binding site affects viral RNA replication

Replication of the nonsegmented, plus-stranded RNA genome of Cucumber necrosis tombusvirus (CNV) requires two essential overlapping viral-coded replication proteins, the p33 replication co-factor and the p92 RNA-dependent RNA polymerase. In this paper, we demonstrate that p33 is phosphorylated in vivo and in vitro by a membrane-bound plant kinase. Phosphorylation of p33 was also demonstrated in vitro by using purified protein kinase C. The related p28 replication protein of Turnip crinkle virus was also found to be phosphorylated in vivo, suggesting that posttranslational modification of replication proteins is a general feature among members of the large Tombusviridae family. Based on in vitro studies with purified recombinant p33, we show evidence for phosphorylation of threonine and serine residues adjacent to the essential RNA-binding site in p33. Phosphorylation-mimicking aspartic acid mutations rendered p33 nonfunctional in plant protoplasts and in yeast, a model host. Comparable mutations within the prereadthrough portion of p92 did not abolish replication. The nonphosphorylation-mimicking alanine mutants of CNV were able to replicate in plant protoplasts and in yeast, albeit with reduced efficiency when compared to the wild type. These alanine mutants also showed altered subgenomic RNA synthesis and a reduction in the ratio between plus- and minus-strand RNAs produced during CNV infection. These findings suggest that phosphorylation of threonine/serine residues adjacent to the essential RNA-binding site in the auxiliary p33 protein likely plays a role in viral RNA replication and subgenomic RNA synthesis during tombusvirus infections.     

Inhibition of in vitro RNA binding and replicase activity by phosphorylation of the p33 replication protein of Cucumber necrosis tombusvirus

Tombusviruses, which are small plus-strand RNA viruses of plants, require the viral-coded p33 replication co-factor for template selection and recruitment into replication in infected cells. As presented in the accompanying paper [Shapka, N., Stork, J., Nagy, P.D., 2005. Phosphorylation of the p33 replication protein of Cucumber necrosis tombusvirus adjacent to the RNA binding site affects viral RNA replication. J. Virol. 343, 65-78.], p33 can be phosphorylated in vitro at serine and threonine residues adjacent to its arginine-proline-rich RNA binding site. To test the effect of phosphorylation on p33 function, in this paper, we used phosphorylation-mimicking aspartic acid mutants of Cucumber necrosis virus (CNV) p33 and in-vitro-phosphorylated p33 in gel mobility shift experiments. We found that phosphorylation inhibited the ability of p33 to bind to the viral RNA. In contrast, the nonphosphorylation-mimicking alanine mutants of p33 bound to viral RNA as efficiently as the nonphosphorylated wild type p33 did. In vitro assays with purified CNV replicase preparations revealed that phosphorylation-mimicking mutants of p33 did not support the assembly of functional CNV replicase complexes in yeast, a model host. Based on these results, we propose that the primary function of reversible phosphorylation of p33 is to regulate the RNA binding capacity of p33, which could affect the assembly of new viral replicase complexes, recruitment of the viral RNA template into replication and/or release of viral RNA from replication. Thus, phosphorylation of p33 might help in switching the role of the viral RNA from replication to other processes, such as viral RNA encapsidation and cell-to-cell movement in infected hosts.     

The overlapping RNA-binding domains of p33 and p92 replicase proteins are essential for tombusvirus replication

Two of the five viral-coded proteins of tombusviruses, which are small, nonsegmented, plus-stranded RNA viruses of plants, are required for replication in infected cells. These replicase proteins, namely, p33 and p92, of cucumber necrosis virus are expressed directly from the genomic RNA via a readthrough mechanism. Their overlapping domains contain an arginine/proline-rich RNA-binding motif (termed RPR, which has the sequence RPRRRP). Site-directed mutagenesis of p33 expressed in Escherichia coli, followed by a gel shift assay, defined two of the four arginines as required for efficient RNA binding in vitro. In vivo testing of 19 RPR motif mutants revealed that the RPR motif, and therefore the ability to bind RNA, is important for the replication of tombusviruses and their associated defective interfering (DI) RNAs. Mutation within the RPR motif also affected the ratio of subgenomic versus genomic RNAs in infected cells. To test whether the RPR motif is essential for the function of either p33 or p92 in replication, we used a two-component system developed by, J. Virol. 5845-5851), in which p92 was expressed from the genomic RNA of a tombusvirus, while p33 was expressed from a DI RNA. The protoplast experiments with the two-component system revealed that the RPR motif is essential for the replication function of both proteins. Interestingly, mutations within the RPR motif of p33 and p92 had different effects on RNA replication, suggesting different roles for the RNA-binding motifs of these proteins in tombusvirus replication.     

The host Pex19p plays a role in peroxisomal localization of tombusvirus replication proteins

Replication of Tomato bushy stunt virus (TBSV) RNA takes place on the cytosolic membrane surface of peroxisomes in plants and in yeast, a model host. To identify the host proteins involved in assisting the peroxisomal localization of the tombusvirus p33 replication protein, we tested if p33 could bind directly to yeast proteins involved in peroxisomal transport in vitro. This work has led to the demonstration of Pex19p-p33 interaction via pull-down and co-purification experiments. Pex19p was also detected in the tombusvirus replicase after protein cross-linking, suggesting that Pex19p transiently binds to the replicase as could be expected from a transporter. To validate the importance of Pex19p-p33 interaction in TBSV replication in yeast, we re-targeted Pex19p to the mitochondria, which resulted in the re-distribution of a large fraction of p33 to the mitochondria. The expression of the mitochondrial-targeted Pex19p inhibited TBSV RNA accumulation by 2-4-fold in vivo and reduced the in vitro activity of the tombusvirus replicase by 80%. These data support the model that Pex19p is a cellular transporter for localization of p33 replication protein to the host peroxisomal membranes.     

Ubiquitination of tombusvirus p33 replication protein plays a role in virus replication and binding to the host Vps23p ESCRT protein

Post-translational modifications of viral replication proteins could be widespread phenomena during the replication of plus-stranded RNA viruses. In this article, we identify two lysines in the tombusvirus p33 replication co-factor involved in ubiquitination and show that the same lysines are also important for the p33 to interact with the host Vps23p ESCRT-I factor. We find that the interaction of p33 with Vps23p is also affected by a “late-domain”-like sequence in p33. The combined mutations of the two lysines and the late-domain-like sequences in p33 reduced replication of a replicon RNA of Tomato bushy stunt virus in yeast model host, in plant protoplasts, and plant leaves, suggesting that p33-Vps23p ESCRT protein interaction affects tombusvirus replication. Using ubiquitin-mimicking p33 chimeras, we demonstrate that high level of p33 ubiquitination is inhibitory for TBSV replication. These findings argue that optimal level of p33 ubiquitination plays a regulatory role during tombusvirus infections.     

Host transcription factor Rpb11p affects tombusvirus replication and recombination via regulating the accumulation of viral replication proteins

Previous genome-wide screens identified over 100 host genes whose deletion/down-regulation affected tombusvirus replication and 32 host genes that affected tombusvirus RNA recombination in yeast, a model host for replication of Tomato bushy stunt virus (TBSV). Down-regulation of several of the identified host genes affected the accumulation levels of p33 and p92(pol) replication proteins, raising the possibility that these host factors could be involved in the regulation of the amount of viral replication proteins and, thus, they are indirectly involved in TBSV replication and recombination. To test this model, we developed a tightly regulated expression system for recombinant p33 and p92(pol) replication proteins in yeast. We demonstrate that high accumulation level of p33 facilitated efficient viral RNA replication, while the effect of p33 level on RNA recombination was less pronounced. On the other hand, high level of p92(pol) accumulation promoted TBSV RNA recombination more efficiently than RNA replication. As predicted, Rpb11p, which is part of the polII complex, affected the accumulation levels of p33 and p92(pol) as well as altered RNA replication and recombination. An in vitro assay with the tombusvirus replicase further supported that Rpb11p affects TBSV replication and recombination only indirectly, via regulating p33 and p92(pol) levels. In contrast, the mechanism by which Rpt4p endopeptidase/ATPase and Mps1p threonine/tyrosine kinase affect TBSV recombination is different from that proposed for Rpb11p. We propose a model that the concentration (molecular crowding) of replication proteins within the viral replicase is a factor affecting viral replication and recombination.     

Interaction between the replicase proteins of Tomato bushy stunt virus in vitro and in vivo

Tomato bushy stunt virus (TBSV), a tombusvirus with a non-segmented, plus-stranded RNA genome, codes for p33 and p92 replicase proteins. The sequence of p33 overlaps with the N-terminal domain of p92, which also contains the signature motifs of RNA-dependent RNA polymerases (RdRps) in its non-overlapping C-terminal portion. In this research, we demonstrate in vitro interactions between p33:p33 and p33:p92 using surface plasmon resonance analysis with purified recombinant p33 and p92. The sequence in p33 involved in the above protein-protein interactions was mapped to the C-terminal region, which also contains an RNA-binding site. Using the yeast two-hybrid assay, we confirmed that two short regions within p33 could promote p33:p33 and p33:p92 interactions in vivo. Mutations in either p33 or p92 within the short regions involved in p33:p33 and p33:p92 interactions decreased the replication of a TBSV defective interfering RNA in yeast, a model host, supporting the significance of these protein interactions in tombusvirus replication.     

Translation elongation factor 1A is a component of the tombusvirus replicase complex and affects the stability of the p33 replication co-factor

Host RNA-binding proteins are likely to play multiple, integral roles during replication of plus-strand RNA viruses. To identify host proteins that bind to viral RNAs, we took a global approach based on the yeast proteome microarray, which contains 4080 purified yeast proteins. The biotin-labeled RNA probes included two distantly related RNA viruses, namely Tomato bushy stunt virus (TBSV) and Brome mosaic virus (BMV). Altogether, we have identified 57 yeast proteins that bound to TBSV RNA and/or BMV RNA. Among the identified host proteins, eleven bound to TBSV RNA and seven bound to BMV RNA with high selectivity, whereas the remaining 39 host proteins bound to both viral RNAs. The interaction between the TBSV replicon RNA and five of the identified host proteins was confirmed via gel-mobility shift and co-purification experiments from yeast. Over-expression of the host proteins in yeast, a model host for TBSV, revealed 4 host proteins that enhanced TBSV replication as well as 14 proteins that inhibited replication. Detailed analysis of one of the identified yeast proteins binding to TBSV RNA, namely translation elongation factor eEF1A, revealed that it is present in the highly purified tombusvirus replicase complex. We also demonstrate binding of eEF1A to the p33 replication protein and a known cis-acting element at the 3′ end of TBSV RNA. Using a functional mutant of eEF1A, we provide evidence on the involvement of eEF1A in TBSV replication.     

Exploiting alternative subcellular location for replication: tombusvirus replication switches to the endoplasmic reticulum in the absence of peroxisomes

Plus-strand RNA virus replication takes place on distinct membranous surfaces in infected cells via the assembly of viral replicase complexes involving multiple viral and host proteins. One group of tombusviruses, such as Tomato bushy stunt virus (TBSV), replicate on the surfaces of peroxisomal membranes in plant and yeast cells. Surprisingly, previous genome-wide screen performed in yeast demonstrated that a TBSV replicon RNA replicated as efficiently in yeast defective in peroxisome biogenesis as in the wt yeast (Panavas et al., Proc Natl Acad Sci U S A, 2005). To further test how the lack of peroxisomes could affect tombusvirus replication, we used yeast cells missing either PEX3 or PEX19 genes, which are absolutely essential for peroxisome biogenesis. Confocal microscopy-based approach revealed that the wild-type tombusvirus p33 replication protein accumulated in the endoplasmic reticulum (ER) in pex3Delta or pex19Delta yeast, suggesting that tombusvirus replication could take place on the surface of ER membrane. The activities of the isolated tombusvirus replicase preparations from wt, pex3Delta or pex19Delta yeasts were comparable, demonstrating that the assembly of the replicase was as efficient in the ER as in the authentic subcellular environments. The generation/accumulation of tombusvirus recombinants was similar in wt, pex3Delta and pex19Delta yeasts, suggesting that the rate of mistakes occurring during tombusvirus replication is comparable in the presence or absence of peroxisomes. Overall, this work demonstrates that a tombusvirus, relying on the wt replication proteins, can efficiently replicate on an alternative intracellular membrane. This suggests that RNA viruses might have remarkable flexibility for using various host membranes for their replication.     

Analysis of the subcellular targeting of the smaller replicase protein of Pelargonium flower break virus

Replication of all positive RNA viruses occurs in association with intracellular membranes. In many cases, the mechanism of membrane targeting is unknown and there appears to be no correlation between virus phylogeny and the membrane systems recruited for replication. Pelargonium flower break virus (PFBV, genus Carmovirus, family Tombusviridae) encodes two proteins, p27 and its read-through product p86 (the viral RNA dependent-RNA polymerase), that are essential for replication. Recent reports with other members of the family Tombusviridae have shown that the smaller replicase protein is targeted to specific intracellular membranes and it is assumed to determine the subcellular localization of the replication complex. Using in vivo expression of green fluorescent protein (GFP) fusions in plant and yeast cells, we show here that PFBV p27 localizes in mitochondria. The same localization pattern was found for p86 that contains the p27 sequence at its N-terminus. Cellular fractionation of p27GFP-expressing cells confirmed the confocal microscopy observations and biochemical treatments suggested a tight association of the protein to membranes. Analysis of deletion mutants allowed identification of two regions required for targeting of p27 to mitochondria. These regions mapped toward the N- and C-terminus of the protein, respectively, and could function independently though with distinct efficiency. In an attempt to search for putative cellular factors involved in p27 localization, the subcellular distribution of the protein was checked in a selected series of knockout yeast strains and the outcome of this approach is discussed.     

Identification of domains in p27 auxiliary replicase protein essential for its association with the endoplasmic reticulum membranes in Red clover necrotic mosaic virus

Positive-strand RNA viruses require host intracellular membranes for replicating their genomic RNAs. In this study, we determined the domains and critical amino acids in p27 of Red clover necrotic mosaic virus (RCNMV) required for its association with and targeting of ER membranes in Nicotiana benthamiana plants using a C-terminally GFP-fused and biologically functional p27. Confocal microscopy and membrane-flotation assays using an Agrobacterium-mediated expression system showed that a stretch of 20 amino acids in the N-terminal region of p27 is essential for the association of p27 with membranes. We identified the amino acids in this domain required for the association of p27 with membranes using alanine-scanning mutagenesis. We also found that this domain contains amino acids not critical for the membrane association but required for the formation of viral RNA replication complexes and negative-strand RNA synthesis. Our results extend our understanding of the multifunctional role of p27 in RCNMV replication.     

RNA chaperone activity of the tombusviral p33 replication protein facilitates initiation of RNA synthesis by the viral RdRp in vitro

Small plus-stranded RNA viruses do not code for RNA helicases that would facilitate the proper folding of viral RNAs during replication. Instead, these viruses might use RNA chaperones as shown here for the essential p33 replication protein of Tomato bushy stunt virus (TBSV). In vitro experiments demonstrate that the purified recombinant p33 promotes strand separation of a DNA/RNA duplex. In addition, p33 renders dsRNA templates sensitive to single-strand specific S1 nuclease, suggesting that p33 can destabilize highly structured RNAs. We also demonstrate that the RNA chaperone activity of p33 facilitates self-cleavage by a ribozyme in vitro. In addition, purified p33 facilitates in vitro RNA synthesis on double-stranded (ds)RNA templates up to 5-fold by a viral RNA-dependent RNA polymerase. We propose that the RNA chaperone activity of p33 facilitates the initiation of plus-strand synthesis as well as affects RNA recombination. Altogether, the TBSV RNA chaperone might perform similar biological functions to the helicases of other RNA viruses with much larger coding capacity.     

Nucleolin/Nsr1p binds to the 3′ noncoding region of the tombusvirus RNA and inhibits replication

Previous genome-wide screens identified > 100 host genes affecting tombusvirus replication using yeast model host. One of those factors was Nsr1p (nucleolin), which is an abundant RNA-binding shuttle protein involved in rRNA maturation and ribosome assembly. We find that overexpression of Nsr1p in yeast or in Nicotiana benthamiana inhibited the accumulation of tombusvirus RNA by ∼10-fold. Regulated overexpression of Nsr1p revealed that Nsr1p should be present at the beginning of viral replication for efficient inhibition, suggesting that Nsr1p inhibits an early step in the replication process. In vitro experiments revealed that Nsr1p binds preferably to the 3′ UTR in the viral RNA. The purified recombinant Nsr1p inhibited the in vitro replication of the viral RNA in a yeast cell-free assay when preincubated with the viral RNA before the assay. These data support the model that Nsr1p/nucleolin inhibits tombusvirus replication by interfering with the recruitment of the viral RNA for replication.     

Replication of Tomato bushy stunt virus RNA in a plant in vitro system

An ideal system to investigate individual determinants of the replication process of (+)-strand RNA viruses is a cell-free extract that supports viral protein and RNA synthesis in a synchronized manner. Here, we applied a translation/replication system based on cytoplasmic extracts of Nicotiana tabacum cells to Tomato bushy stunt virus (TBSV) RNA. In vitro translated TBSV proteins p33 and p92 form viral replicase, which, in the same reaction, accomplishes the entire replication cycle on exogenous TBSV DI or full-length RNA. Tests of mutant TBSV RNAs confirmed the template specificity of the in vitro replication reaction. Complementation experiments ascertained the significance of an earlier identified TBSV host factor. Interestingly, formation of the viral replicase occurs also in the absence of concurrent protein synthesis demonstrating that translation and RNA replication are not functionally linked in this system. Our studies with cell-free extracts of a plant host thus confirmed earlier findings and enabled novel insights into the TBSV RNA replication process.     

Expression of the Arabidopsis Xrn4p 5'-3' exoribonuclease facilitates degradation of tombusvirus RNA and promotes rapid emergence of viral variants in plants

Rapid RNA virus evolution is a major problem due to the devastating diseases caused by human, animal and plant-pathogenic RNA viruses. A previous genome-wide screen for host factors affecting recombination in Tomato bushy stunt tombusvirus (TBSV), a small monopartite plant virus, identified Xrn1p 5'-3' exoribonuclease of yeast, a model host, whose absence led to increased appearance of recombinants [Serviene, E., Shapka, N., Cheng, C.P., Panavas, T., Phuangrat, B., Baker, J., Nagy, P.D., (2005). Genome-wide screen identifies host genes affecting viral RNA recombination. Proc. Natl. Acad. Sci. U. S. A. 102 (30), 10545-10550]. In this paper, we tested if over-expression of Xrn1p in yeast or expression of the analogous Xrn4p cytoplasmic 5'-3' exoribonuclease, which has similar function in RNA degradation in Arabidopsis as Xrn1p in yeast, in Nicotiana benthamiana could affect the accumulation of tombusvirus RNA. We show that over-expression of Xrn1p led to almost complete degradation of TBSV RNA replicons in yeast, suggesting that Xrn1p is involved in TBSV degradation. Infection of N. benthamiana expressing AtXrn4p with Cucumber necrosis tombusvirus (CNV) led to enhanced viral RNA degradation, suggesting that the yeast and the plant cytoplasmic 5'-3' exoribonuclease play similar roles. We also observed rapid emergence of novel CNV genomic RNA variants formed via deletions of 5' terminal sequences in N. benthamiana expressing AtXrn4p. Three of the newly emerging 5' truncated CNV variants were infectious in N. benthamiana protoplasts, whereas one CNV variant caused novel symptoms and moved systemically in N. benthamiana plants. Altogether, this paper establishes that a single plant gene can contribute to the emergence of novel viral variants.     

West Nile virus genome cyclization and RNA replication require two pairs of long-distance RNA interactions

West Nile virus (WNV) genome cyclization and replication require two pairs of long-distance RNA interactions. Besides the previously reported 5'CS/3'CSI (conserved sequence) interaction, a 5'UAR/3'UAR (upstream AUG region) interaction also contributes to genome cyclization and replication. WNVs containing mutant 5'UARs capable of forming the 5'/3' viral RNA interaction were replicative. In contrast, WNV containing a 5'UAR mutation that abolished the 5'/3' viral RNA interaction was non-replicative; however, the replication defect could be rescued by a single-nucleotide adaptation that restored the 5'/3' RNA interaction. The 5'UAR/3'UAR interaction is critical for RNA synthesis, but not for viral translation. Antisense oligomers targeting the 5'UAR/3'UAR interaction effectively inhibited WNV replication. Phylogenic analysis showed that the 3'UAR could alternate between pairing with the 5'UAR or with the 3' end of the flaviviral genome. Therefore, the 5'UAR/3'UAR pairing may release the 3' end of viral genome (as a template) during the initiation of minus-strand RNA synthesis.