Search

Tomato black ring virus (TBRV, genus Nepovirus) infects a wide range of economically important plants such as tomato, potato, tobacco and cucumber. Here, a successful construction of infectious full-length cDNA clones of the TBRV genomic RNAs (RNA1 and RNA2) is reported for the first time. The engineered constructs consisting of PCR-amplified DNAs were cloned into binary vector pJL89 immediately downstream of a double cauliflower mosaic virus (CaMV) 35S promoter, and upstream of the hepatitis delta virus (HDV) ribozyme and nopaline synthase terminator (NOS). The symptoms induced on plants agroinoculated with both constructs were indistinguishable from those caused by the wild-type virus. The infectivity of obtained clones was verified by reinoculation to Nicotiana tabacum cv. Xanthi, Chenopodium quinoa and Cucumis sativus. The presence of viral particles and RNA was confirmed by electron microscopy and reverse transcription polymerase chain reaction, respectively. Constructed full-length infectious cDNA clones will serve as an excellent tool to study virus-host-vector interactions., (Copyright © 2017 Elsevier B.V. All rights reserved.)

Nepoviruses supplied seminal landmarks to the historical trail of plant virology. Among the first agriculturally relevant viruses recognized in the late 1920s and among the first plant viruses officially classified in the early 1970s, nepoviruses also comprise the first species for which a soil-borne ectoparasitic nematode vector was identified. Early research on nepoviruses shed light on the genome structure and expression, biological properties of the two genomic RNAs, and mode of transmission. In recent years, research on nepoviruses enjoyed an extraordinary renaissance. This resurgence provided new insights into the molecular interface between viruses and their plant hosts, and between viruses and dagger nematode vectors to advance our understanding of some of the major steps of the infectious cycle. Here we examine these recent findings, highlight ongoing work, and offer some perspectives for future research., (© 2017 Elsevier Inc. All rights reserved.)

The NTB-VPg polyprotein from tomato ringspot virus is an integral membrane replication protein associated with endoplasmic reticulum membranes. A signal peptidase (SPase) cleavage was previously detected in the C-terminal region of NTB-VPg downstream of a 14 amino acid (aa)-long hydrophobic region (termed TM2). However, the exact location of the cleavage site was not determined. Using in vitro translation assays, we show that the SPase cleavage site is conserved in the NTB-VPg protein from various ToRSV isolates, although the rate of cleavage varies from one isolate to another. Systematic site-directed mutagenesis of the NTB-VPg SPase cleavage sites of two ToRSV isolates allowed the identification of sequences that affect cleavage efficiency. We also present evidence that SPase cleavage in the ToRSV-Rasp2 isolate occurs within a GAAGG sequence likely after the AAG (GAAG/G). Mutation of a downstream MAAV sequence to AAAV resulted in SPase cleavage at both the natural GAAG/G and the mutated AAA/V sequences. Given that there is a distance of seven aa between the two cleavage sites, this indicates that there is flexibility in the positioning of the cleavage sites relative to the inner surface of the membrane and the SPase active site. SPase cleavage sites are typically located 3-7 aa downstream of the hydrophobic region. However, the NTB-VPg GAAG/G cleavage site is located 17 aa downstream of the TM2 hydrophobic region, highlighting unusual features of the NTB-VPg SPase cleavage site. A putative 11 aa-long amphipathic helix was identified immediately downstream of the TM2 region and five aa upstream of the GAAG/G cleavage site. Based on these results, we present an updated topology model in which the hydrophobic and amphipathic domains form a long tilted helix or a bent helix in the membrane lipid bilayer, with the downstream cleavage site(s) oriented parallel to the membrane inner surface., Competing Interests: The authors have declared that no competing interests exist.

We report here the development of tobacco ringspot virus (TRSV)-based vectors for the transient expression of foreign genes and for the analysis of endogenous gene function in plants using virus-induced gene silencing. The jellyfish green fluorescent protein (GFP) gene was inserted between the TRSV movement protein (MP) and coat protein (CP) regions, resulting in high in-frame expression of the RNA2-encoded viral polyprotein. GFP was released from the polyprotein via an N-terminal homologous MP-CP cleavage site and a C-terminal foot-and-mouth disease virus (FMDV) 2 A catalytic peptide in Nicotiana benthamiana. The VIGS target gene was introduced in the sense and antisense orientations into a SnaBI site, which was created by mutating the sequence following the CP stop codon. VIGS of phytoene desaturase (PDS) in N. benthamiana, Arabidopsis ecotype Col-0, cucurbits and legumes led to obvious photo-bleaching phenotypes. A significant reduction in PDS mRNA levels in silenced plants was confirmed by semi-quantitative RT-PCR., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Arabidopsis thaliana leucine-rich repeat-containing (NLR) proteins RPS4 and RRS1, known as dual resistance proteins, confer resistance to multiple pathogen isolates, such as the bacterial pathogens Pseudomonas syringae and Ralstonia solanacearum and the fungal pathogen Colletotrichum higginsianum. RPS4 is a typical Toll/interleukin 1 Receptor (TIR)-type NLR, whereas RRS1 is an atypical TIR-NLR that contains a leucine zipper (LZ) motif and a C-terminal WRKY domain. RPS4 and RRS1 are localised near each other in a head-to-head orientation. In this study, direct mutagenesis of the C-terminal LZ motif in RRS1 caused an autoimmune response and stunting in the mutant. Co-immunoprecipitation analysis indicated that full-length RPS4 and RRS1 are physically associated with one another. Furthermore, virus-induced gene silencing experiments showed that hypersensitive-like cell death triggered by RPS4/LZ motif-mutated RRS1 depends on EDS1. In conclusion, we suggest that the RRS1-LZ motif is crucial for the regulation of the RPS4/RRS1 complex.

The N-terminal domains of the RNA 2-encoded 2A(HP) proteins of the arabis mosaic (ArMV) and grapevine fanleaf (GFLV) nepoviruses were shown to be highly variable and a hotspot for intra- and inter-species recombination events. Chimeric ArMV-NW clones in which the N-terminal domain of 2A(HP) or the entire 2A(HP) of GFLV isolates replaced the corresponding domains of ArMV retained their infectivity, showing that the 2A(HP) proteins of ArMV-NW and GFLV are exchangeable. ArMN-NW clones with deletions of the N-terminal, core, or C-terminal domains of the ArMV-NW 2A(HP) were infectious in Chenopodium quinoa although viral RNA (especially RNA 2) accumulated at reduced levels. In contrast, deletion of the entire 2A(HP) protein or of the C-terminal two thirds of the protein abolished infectivity of the ArMV-NW clones. These results suggest that multiple functional domains are distributed throughout the 2A(HP) protein and are essential for the accumulation of viral RNA 2., (Copyright © 2014. Published by Elsevier Inc.)

Most Arabidopsis ecotypes display tolerance to the Tobacco ringspot virus (TRSV), but a subset of Arabidopsis ecotypes, including Estland (Est), develop lethal systemic necrosis (LSN), which differs from the localized hypersensitive responses (HRs) or systemic acquired resistance (SAR) characteristic of incompatible reactions. Neither viral replication nor the systemic movement of TRSV was restricted in tolerant or sensitive Arabidopsis ecotypes; therefore, the LSN phenotype shown in the sensitive ecotypes might not be due to viral accumulation. In the present study, we identified the Est TTR1 gene (tolerance to Tobacco ringspot virus 1) encoding a TIR-NBS-LRR protein that controls the ecotype-dependent tolerant/sensitive phenotypes by a map-based cloning method. The tolerant Col-0 ecotype Arabidopsis transformed with the sensitive Est TTR1 allele developed an LSN phenotype upon TRSV infection, suggesting that the Est TTR1 allele is dominant over the tolerant ttr1 allele of Col-0. Multiple sequence alignments of 10 tolerant ecotypes from those of eight sensitive ecotypes showed that 10 LRR amino acid polymorphisms were consistently distributed across the TTR1/ttr1 alleles. Site-directed mutagenesis of these amino acids in the LRR region revealed that two sites, L956S and K1124Q, completely abolished the LSN phenotype. VIGS study revealed that TTR1 is dependent on SGT1, rather than EDS1. The LSN phenotype by TTR1 was shown to be transferred to Nicotiana benthamiana, demonstrating functional conservation of TTR1 across plant families, which are involved in SGT-dependent defense responses, rather than EDS1-dependent signaling pathways.

Many animal and plant viruses rely on vectors for their transmission from host to host. Grapevine fanleaf virus (GFLV), a picorna-like virus from plants, is transmitted specifically by the ectoparasitic nematode Xiphinema index. The icosahedral capsid of GFLV, which consists of 60 identical coat protein subunits (CP), carries the determinants of this specificity. Here, we provide novel insight into GFLV transmission by nematodes through a comparative structural and functional analysis of two GFLV variants. We isolated a mutant GFLV strain (GFLV-TD) poorly transmissible by nematodes, and showed that the transmission defect is due to a glycine to aspartate mutation at position 297 (Gly297Asp) in the CP. We next determined the crystal structures of the wild-type GFLV strain F13 at 3.0 Å and of GFLV-TD at 2.7 Å resolution. The Gly297Asp mutation mapped to an exposed loop at the outer surface of the capsid and did not affect the conformation of the assembled capsid, nor of individual CP molecules. The loop is part of a positively charged pocket that includes a previously identified determinant of transmission. We propose that this pocket is a ligand-binding site with essential function in GFLV transmission by X. index. Our data suggest that perturbation of the electrostatic landscape of this pocket affects the interaction of the virion with specific receptors of the nematode's feeding apparatus, and thereby severely diminishes its transmission efficiency. These data provide a first structural insight into the interactions between a plant virus and a nematode vector.

One of the mechanisms of functioning for viral cap-independent translational enhancers (CITEs), located in 3' non-translated regions (NTRs), is 3' NTR-5' leader long-distance base pairing. Previously, we have demonstrated that the RNA2 3' NTR of Blackcurrant reversion nepovirus (BRV) contains a CITE, which must base pair with the 5' NTR to facilitate translation. Here we compared translation strategies employed by BRV RNA1 and RNA2, by using mutagenesis of the BRV NTRs in firefly luciferase reporter mRNA, in plant protoplasts. Translation mechanisms, based on 3' CITEs, 5' NTR-3' NTR base pairing and poly(A) tail-stimulation, were found conserved between RNA1 and RNA2. The 40S ribosomal subunit entry at the RNA1 leader occurred, at least partly, via an internal ribosomal entry site (IRES). Two RNA1 leader segments complementary to plant 18S rRNA enhanced translation. A model for BRV RNAs translation, involving IRES-dependent 40S subunit recruitment and long-distance 5' NTR-3' NTR base pairing, is discussed.

Replication of Tomato ringspot virus (ToRSV) occurs in association with endoplasmic reticulum (ER)-derived membranes. We have previously shown that the putative nucleotide triphosphate-binding protein (NTB) of ToRSV is an ER-targeted protein and that an intermediate polyprotein containing the domains for NTB and for the genome-linked viral protein (VPg) is associated with the replication complex. We now report the detection of a 95-kDa polyprotein that contains the domains for the RNA-dependent RNA polymerase (Pol), the proteinase (Pro) and the VPg. This polyprotein appears to be a truncated version of the full-length 111-kDa VPg-Pro-Pol polyprotein and was termed VPg-Pro-Pol'. A subpopulation of VPg-Pro-Pol' was peripherally associated with ER-derived membranes active in viral replication. However, the VPg, Pro and Pol domains did not target to membranes in the absence of viral infection. We propose a model in which VPg-Pro-Pol' is brought to the site of replication through interaction with a viral membrane protein.

The putative NTP-binding protein (NTB) of Tomato ringspot nepovirus (ToRSV) contains a hydrophobic region at its C terminus consisting of two adjacent stretches of hydrophobic amino acids separated by a few amino acids. In infected plants, the NTB-VPg polyprotein (containing the domain for the genome-linked protein) is associated with endoplasmic reticulum-derived membranes that are active in ToRSV replication. Recent results from proteinase K protection assays suggested a luminal location for the VPg domain in infected plants, providing support for the presence of a transmembrane domain at the C terminus of NTB. In this study, we have shown that NTB-VPg associates with canine microsomal membranes in the absence of other viral proteins in vitro and adopts a topology similar to that observed in vivo in that the VPg is present in the lumen. Truncated proteins containing 60 amino acids at the C terminus of NTB and the entire VPg exhibited a similar topology, confirming that this region of the protein contains a functional transmembrane domain. Deletion of portions of the C-terminal hydrophobic region of NTB by mutagenesis and introduction of glycosylation sites to map the luminal regions of the protein revealed that only the first stretch of hydrophobic amino acids traverses the membrane, while the second stretch of hydrophobic amino acids is located in the lumen. Our results provide additional evidence supporting the hypothesis that the NTB-VPg polyprotein acts as a membrane-anchor for the replication complex.

The nepovirus Grapevine fanleaf virus (GFLV) is specifically transmitted by the nematode Xiphinema index. To identify the RNA2-encoded proteins involved in X. index-mediated spread of GFLV, chimeric RNA2 constructs were engineered by replacing the 2A, 2B(MP), and/or 2C(CP) sequences of GFLV with their counterparts in Arabis mosaic virus (ArMV), a closely related nepovirus which is transmitted by Xiphinema diversicaudatum but not by X. index. Among the recombinant viruses obtained from transcripts of GFLV RNA1 and chimeric RNA2, only those which contained the 2C(CP) gene (504 aa) and 2B(MP) contiguous 9 C-terminal residues of GFLV were transmitted by X. index as efficiently as natural and synthetic wild-type GFLV, regardless of the origin of the 2A and 2B(MP) genes. As expected, ArMV was not transmitted probably because it is not retained by X. index. These results indicate that the determinants responsible for the specific spread of GFLV by X. index are located within the 513 C-terminal residues of the polyprotein encoded by RNA2., (Copyright 2001 Elsevier Science.)

Tomato ringspot nepovirus RNA-1-encoded polyprotein (P1) contains the domains for the putative NTP-binding protein, VPg, 3C-like protease and a putative RNA-dependent RNA polymerase in its C-terminal region. The N-terminal region of P1, with a coding capacity for a protein (or a precursor) of 67 kDa, has not been characterized. Using partial cDNA clones, it is shown that the 3C-like protease can process the N-terminal region of P1 at a novel cleavage site in vitro, allowing the release of two proteins, X1 (located at the N terminus of P1) and X2 (located immediately upstream of the NTB domain). P1 precursors in which the protease was inactive or absent were not cleaved by exogenously added protease, suggesting that P1 processing was predominantly in cis. Results from site-directed mutagenesis of putative cleavage sites suggest that dipeptides Q(423)/G and Q(620)/G are the X1-X2 and X2-NTB cleavage sites, respectively. The putative X1 protein contains a previously identified alanine-rich sequence which is present in nepoviruses but not in the related comoviruses. The putative X2 protein contains a region with similarity to the comovirus 32 kDa protease co-factor (the only mature protein released from the N terminus of comovirus P1 polyproteins) and to the corresponding region of other nepovirus P1 polyproteins. These results raise the possibility that the presence of two distinct protein domains in the N-terminal part of the P1 polyprotein may be a common feature of nepoviruses.

Tomato ringspot nepovirus (TomRSV) RNA-1 encodes a putative NTP-binding protein (NTB), a putative viral genome-linked protein (VPg), a putative RNA-dependent RNA polymerase (Pol) and a serine-like protease (Pro), which have been suggested to be involved in viral RNA replication. Proteolytic processing of protease precursors containing these proteins was studied in Escherichia coli and in vitro. The TomRSV protease could cleave the precursor proteins and release the predicted mature proteins or intermediate precursors. Although processing was detected at all three predicted cleavage sites (NTB-VPg, VPg-Pro and Pro-Pol), processing at the VPg-Pro cleavage site was inefficient, resulting in accumulation of the VPg-Pro intermediate precursor in E. coli and in vitro. In addition, the presence of the VPg sequence in the precursor resulted in increased cleavage at the Pro-Pol cleavage site in E. coli and in vitro. Direct N-terminal sequencing of the genomic RNA-linked VPg, of the mature protease purified from E. coli extracts and of radiolabelled mature polymerase purified from in vitro translation products revealed the sequences of the NTB-VPg, VPg-Pro and Pro-Pol dipeptide cleavage sites to be Q/S, Q/G and Q/S, respectively. In vitro processing at the NTB-VPg and Pro-Pol cleavage sites was not detected upon mutation or deletion of the conserved glutamine at the -1 position of the cleavage site. These results are discussed in light of the cleavage site specificity of the TomRSV protease.

Recent studies of the hairpin ribozyme have revealed a distinct catalytic mechanism for this small RNA motif. Inner-sphere coordinated metal ions are not required, as the inert metal ion complex cobalt hexammine promotes catalysis. Detailed kinetic analyses have defined rates of individual steps in the catalytic cycle. Functional group modification, NMR studies of subdomains and cross-linking experiments, in combination with computer modeling, have led to a proposal for domain interactions in the substrate-ribozyme complex.

Each of the two genomic RNAs of tobacco ringspot nepovirus is known to have a 5'-linked protein, the VPg. We report a simplified analysis of the covalent VPg-RNA connection that allowed us to identify the 5' nucleotide residue of each genomic RNA and its phosphodiester link to a specific serine residue of the VPg, without resorting to in vivo labeling with 32P, in vitro radioiodination, or separation of the two genomic RNAs. Unfractionated genomic RNA was incubated with an oligodeoxyribonucleotide specific for the 5' region of either RNA 1 or RNA 2 and ribonuclease H. Reaction products were 3'-end-labeled and were fractionated by gel electrophoresis. The most highly labeled product derived from each genomic RNA was identified as a VPg-oligoribonucleotide (VPg-5'-oligo) by its sensitivity to proteinase. In a presumed beta-elimination reaction that apparently was more rapid than phosphodiester cleavage, incubation in alkaline sodium bicarbonate released a rapidly migrating product, 5'-oligo. Phosphatase-treated 5'-oligo accepted 5'-label in a polynucleotide kinase-catalyzed reaction, and uridylate was identified as the 5' terminal residue for both RNA 1 and RNA 2. Results from Edman degradation of the VPg suggest that the VPg is linked at serine 5 to the 5' uridylate of each genomic RNA.

The capsid protein gene of tobacco ringspot virus (TobRV), which had been modified to contain an amino-terminal methionine codon, was ligated into a baculovirus transfer vector downstream from the polyhedrin promoter. The resulting plasmid was cotransfected with linearized baculovirus DNA into insect cells. Recombinant baculovirus expressed high levels of the TobRV capsid protein that assembled to form virus-like particles that were similar in size and shape to authentic TobRV capsids. These virus-like particles did not encapsidate any RNA, including the capsid protein mRNA. The capsid protein mRNA is a truncated RNA 2, which may lack a putative encapsidation signal. To determine whether an intact packaging substrate could be encapsidated by the TobRV capsid protein, another recombinant baculovirus, concomitantly expressing both capsid protein and TobRV satellite RNA, was constructed. Surprisingly, the vast majority of the satellite RNA molecules expressed from this recombinant baculovirus were ligated in the insect cells to form circular RNA molecules. Like circular forms of satellite RNA generated in planta, these circular satellite molecules remained unencapsidated by the TobRV capsid protein. Computer-generated three-dimensional reconstruction using electron cryomicrographs of the empty virus-like particles allowed the first structural analyses of any nepovirus capsid. This 22-A resolution reconstruction resembled capsids of other members of the picornavirus superfamily. These data support the hypothesis that the nepovirus capsid is structurally analogous to those of the como- and picornaviruses.

Tomato ringspot nepovirus (TomRSV) produces a 45 kDa movement protein and a 58 kDa coat protein in infected plants. Accumulation of the movement protein in relation to that of the coat protein was studied in infected protoplasts using a monoclonal antibody against the movement protein and polyclonal antibodies against the coat protein. Unlike most other viral movement proteins, the TomRSV movement protein was present at late stages of infection. Pulse-chase labelling experiments revealed that the release of the movement protein from the precursor polyprotein was coordinated with that of the coat protein. However, the movement protein was less stable than the coat protein in the extractable fraction of the protoplasts. The expression pattern of the TomRSV movement protein is discussed in the light of the proposed mechanism of cell-to-cell movement of virus-like particles through tubular structures composed of the movement protein.

The putative 38 kDa movement protein (P38) gene, located on the RNA2 of grapevine fanleaf nepovirus (GFLV), was cloned in Escherichia coli and expressed as a fusion protein fused to six histidines (6HisP38). This protein was purified and used to produce a specific polyclonal antiserum from which immunoglobulins were isolated by immunoaffinity against the recombinant 6HisP38. Western immunoblot analyses on GFLV RNA2 in vitro maturation products showed that the antibodies were specific for P38 protein. This protein was detected as early as 18 h post-inoculation in GFLV-infected Chenopodium quinoa protoplasts and accumulated to very high levels. Tissue-prints and time course experiments on infected C. quinoa plants confirmed that P38 is present at a high level late in infection and is a final maturation product of the GFLV RNA2 polyprotein in vivo. P94 and P66 intermediates of maturation and polyprotein P2 were also detected in vivo but in very low concentrations. No significant difference was observed in the relative amounts of P66 and P94 detected in vivo, contrary to what occurs in vitro. Subcellular fractionation studies showed that P38, although mainly cytosolic, is also found in association with cell wall and membranes. Thought to be the GFLV movement protein, P38 would thus behave in an 'atypical' manner.

Arabis mosaic virus (ArMV) is a nepovirus that is serologically distantly related to grapevine fanleaf virus (GFLV). Both ArMV and GFLV induce grapevine degeneration disease. Several ArMV isolates, unlike isolates of GFLV, produce upon in vitro translation of RNA2 a polyprotein (P2) that forms a double band in polyacrylamide-SDS gels. Cloning of full-length copies of RNA2 of an ArMV isolate from grapevine (ArMV-S) revealed that this isolate contained two RNA2s of different length, called RNA2-U and RNA2-L. The two species were not readily separated by electrophoresis of the virion RNA under denaturing gel electrophoresis conditions but could be distinguished by analysis of primer extension and in vitro translation products. The size difference of the two RNA2s is due mostly if not exclusively to differences in their coding regions. The 124 kDa RNA2-U-encoded polyprotein P2' and the 119 kDa RNA2-L-encoded polyprotein P2", which co-migrate, respectively, with the upper and lower polyprotein bands produced by RNA2 of ArMV-S, were more than 95% identical except in their N-terminal domains. In vitro maturation experiments and sequence comparisons indicate that the N-terminal products of P2' and P2" have a molecular mass of 31 kDa and 26 kDa. The genomic organization proposed is similar to that of GFLV RNA2.

A non-isotopic tissue-print hybridization technique was developed to study long-distance plant virus movement. By using digoxigenin-labeled RNA probes the distribution pattern of the viral RNA was observed in leaf, stem and petiole tissues. In leaf tissue viral RNA was confined preferentially to symptoms and veins, and in stem and petiole sections, the hybridization signal was observed in vascular tissue. Both chemiluminescent and colorigenic detection methods were used. The colorigenic method, though less sensitive, is advantageous in that it gives some anatomical information on the signal distribution. This non-isotopic tissue-print hybridization technique can provide considerable information about the spatial and temporal virus expression with regard to its symptoms.

Translation of tomato black ring virus (TBRV) RNA-1 in a rabbit reticulocyte lysate leads to the synthesis of a 250K polyprotein which cleaves itself into smaller proteins of 50, 60, 120, and 190K. Polypeptides synthesized from synthetic transcripts corresponding to different regions of TBRV RNA-1 are processed only when they encode the 23K protein delimited earlier by sequence homology with the cowpea mosaic virus 24K protease. The proteolytic activity of this protein is completely lost by mutating residues C170 (to I) or L188 (to H), residues which align with conserved residues of the viral serine-like proteases. The 120K protein is generated by cleavage of the dipeptide K/A localized in front of the VPg but is not further cleaved in vitro at the K/S site (at the C terminus of the VPg) or between the protease and polymerase domains. However, both the protein VPgProPol (120K) and the protein ProPol (117K) produced in vitro from synthetic transcripts can cleave in trans the RNA-2-encoded 150K polyprotein, but they cannot cleave in trans polypeptides containing a cleavage site expressed from RNA-1 transcripts in which the protease cistron is absent or modified.

The 359 nucleotide residue (nt) satellite RNA of tobacco ringspot virus increases detectably only in association with replicating tobacco ringspot virus and becomes encapsidated in the virus coat protein. Results from previous reports are consistent with participation of rolling circle transcription in replication of the satellite RNA: (i) both the more abundant plus polarity satellite RNA, s(+)RNA and the complementary s(-)RNA occur in multimeric forms that self-cleave to release the unit length, 359 nt satellite RNA, (ii) circles of both s(+)RNA and s(-)RNA are present in extracts of infected tissue, and (iii) s(-)RNA, but not s(+)RNA, spontaneously and efficiently circularizes in vitro. Our analyses of RNA in tissue extracts suggest that s(+)RNA of all forms is about 100-fold more abundant than s(-)RNA. Nucleic acids were purified rapidly to minimize interconversion of linear and circular forms. For s(+)RNA and for s(-)RNA, the circular and the linear forms were detected in about equal amounts in tissue extracts. The linear s(-)RNA from tissue extracts was found to have the same 5'-terminal sequence as previously was found for s(-)RNA self-cleaved from in vitro transcripts. Like s(-)RNA synthesized in vitro, the circular and linear s(-)RNA from tissue extracts spontaneously and readily interconverted during incubation in vitro. In contrast, the bulk of the circular and linear forms of s(+)RNA were stable. The very limited interconversion of s(+)RNA forms in vitro suggests that circulation in vivo is enzymically catalyzed. Encapsidated satellite RNA was found to be composed of linear, unit length and multimeric forms, including previously undocumented s(-)RNA present in approximately the same relative abundance compared to s(+)RNA as was observed for RNA from tissue extracts. Circles were not detected in encapsidated RNA. We interpret our results in the context of a rolling circle model for satellite RNA replication.