Structure and origin of a novel class of defective interfering particle of vesicular stomatitis virus

The genome structure and terminal sequences of a 'copyback' defective interfering (DI) particle ST1, and a novel complexly rearranged 'snapback' DI particle ST2 of vesicular stomatitis virus have been determined. The ST1 DI genome RNA possesses 54 base long inverted complementary termini, the 5' end of which is homologous to the standard virus genome 5' end. Following this region of inverted complementarity the DI RNA 5' end continues to be homologous to standard virus RNA 5' sequences, whereas the 3' end diverges into sequences within the virus L gene internal sequences. ST2 DI genome RNA does not contain colinear covalently linked plus and minus sense RNA copies of the standard infectious virus RNA 5' terminus as predicted from the prototype snapback DI structure, but instead appears to be a hairpin copy of the ST1 DI RNA genome. This is the first evidence suggesting that DI particles may be generated from RNA templates other than the standard virus RNA. Generation models and the implications of these findings for RNA virus evolution are discussed.     

Continuing coevolution of virus and defective interfering particles and of viral genome sequences during undiluted passages: virus mutants exhibiting nearly complete resistance to formerly dominant defective interfering particles

We quantitatively analyzed the interference interactions between defective interfering (DI) particles and mutants of cloned vesicular stomatitis virus passaged undiluted hundreds of times in BHK-21 cells. DI particles which predominated at different times in these serial passages always interfered most strongly (and very efficiently) with virus isolated a number of passages before the isolation of the DI particles. Virus isolated at the same passage level as the predominant DI particles usually exhibited severalfold resistance to these DI particles. Virus mutants (Sdi- mutants) isolated during subsequent passages always showed increasing resistance to these DI particles, followed by decreasing resistance as new DI particles arose to predominate and exert their own selective pressures on the virus mutant population. It appears that such coevolution of virus and DI particle populations proceeds indefinitely through multiple cycles of selection of virus mutants resistant to a certain DI particle (or DI particle class), followed by mutants resistant to a newly predominant DI particle, etc. At the peak of resistance, virus mutants were isolated which were essentially completely resistant to a particular DI particle; i.e., they were several hundred thousand-fold resistant, and they formed plaques of normal size and numbers in the presence of extremely high multiplicities of the DI particle. However, they were sensitive to interference by other DI particles. Recurring population interactions of this kind can promote rapid virus evolution. Complete sequencing of the N (nucleocapsid) and NS (polymerase associated) genes of numerous Sdi- mutants collected at passage intervals showed very few changes in the NS protein, but the N gene gradually accumulated a series of stable nucleotide and amino acid substitutions, some of which correlated with extensive changes in the Sdi- phenotype. Likewise, the 5' termini (and their complementary plus-strand 3' termini) continued to accumulate extensive base substitutions which were strikingly confined to the first 47 nucleotides. We also observed addition and deletion mutations in noncoding regions of the viral genome at a level suggesting that they probably occur at a high frequency throughout the genome, but usually with lethal or debilitating consequences when they occur in coding regions.     

Structure and origin of a snapback defective interfering particle RNA of vesicular stomatitis virus

The nucleotide sequence of the region which covalently links the complementary strands of the "snapback" RNA of vesicular stomatitis virus, DI011, is (Formula: see text). Both strands of the defective interfering (DI) particle RNA were complementary for their full length and were covalently linked by a single phosphate group. Because the strands were exactly the same length and complementary, template strand and daughter strand nucleocapsids generated during replication of DI 011 were undistinguishable on the basis of sequence, a property not shared by other types of DI particle RNAs. Treatment of the RNA with RNase T1 in high-ionic-strength solutions cleaved the RNA only between positions 1 and 1'. These results and the availability of the guanosine residue in position 1' to kethoxal, a reagent that specifically derivatizes guanosines of single-stranded RNA, suggest that steric constraints keep a small portion of the "turnaround" region in an open configuration. The sequence of the turnaround region was not related in any obvious way to the sequences at the 3' and 5' termini and limited the number of possible models for the origin of this type of DI particle RNA. Two models for the genesis of DI 011 RNA are discussed. We favor one in which the progenitor DI 011 RNA was generated by replication across a nascent replication fork.     

A specific internal RNA polymerase recognition site of VSV RNA is involved in the generation of DI particles.

The 3′ terminal sequences of four different DI particle RNAs ranging in size from 10S to 30S have been determined directly using rapid RNA sequencing methods or deduced, in the case of the fourth DI RNA, from the complementary sequence of a small RNA transcribed from this part of the genome (Schubert et al., 1978). One DI particle (DI 011) contains covalently linked genomic and antigenomic RNA. The 5′ end of this RNA is identical to that of VSV RNA, as determined by annealing for at least 1 kb, as well as to the other DI particle RNAs used in this study. The 3′ ends of the other three DI particle RNAs are exact copies of the common 5′ terminal sequence for 48 nucleotides in two cases and 45 nucleotides in the third. Beyond these complementary regions the sequences are different for each DI RNA. The fact that these regions differ in length by only three nucleotides, despite the wide differences in the overall size of the DI particle RNAs, indicates that if these DIs were formed by the copy-back mechanisms similar to those proposed by Leppert, Kort and Kolakofsky (1977) and Huang (1977), a specific recognition site for the RNA polymerase must be involved in copying the 5′ terminus. We determined the 5′ terminal sequence from position 43–48 at the end of the complementary region and found it to be 5′-GGUCUU-3′. This hexamer is also part of other highly conserved terminal RNA polymerase initiation sites (Keene et al., 1978; Keene, Schubert and Lazzarini, 1979) and may be a specific internal RNA polymerase recognition site. We conclude that this sequence is one of the elements involved in the genesis of DI particle chromosomes containing short complementary sequences at their termini. The ability of the polymerase to resume synthesis at or near a specific recognition site is discussed.     

Terminal sequences of vesicular stomatitis virus RNA are both complementary and conserved.

The nucleotide sequences at the 5' and 3' termini of RNA isolated from the New Jersey serotype of vesicular stomatitis virus [vsV(NJ)] and two of its defective interfering (DI) particles have been determined. The sequence differs from that previously demonstrated for the RNA from the Indiana serotype of VSV at only 1 of the first 17 positions from the 3' terminus and at only 2 of the first 17 positions from the 5' terminus. The 5'-terminal sequence of VSV(NJ) RNA is the complement of the 3'-terminal sequence, and duplexes which are 20 bases long and contain the 3' and 5' termini have been isolated from this RNA. The RNAs isolated from DI particles of VSV(NJ) have the same base sequences as do the RNAs from the parental virus. These results are in sharp contrast to those obtained with the Indiana serotype of VSV and its DI particles, in which the 3'-terminal sequences differ in 3 positions within the first 17. However, with both serotypes, the 3'-terminal sequence of the DI RNA is the complement of the 5'-terminal sequence of the RNA from the infectious virus. These findings suggest that the 3' and 5' RNA termini are highly conserved in both serotypes and that the 3' terminus of DI RNA is ultimately derived by copying the 5' end of the VSV genome, as recently proposed (D. Kolakofsky, M. Leppert, and L. Kort, in B. W. J. Mahy and R. D. Barry, ed., Negative-Strand Virus and the Host Cell, 1977; M. Leppert, L. Kort, and D. Kolakofsky, Cell 12:539-552, 1977; A. S. Huang, Bacteriol. Rev. 41:811-8218 1977).     

Defective interfering viral particles in acute dengue infections

While much of the genetic variation in RNA viruses arises because of the error-prone nature of their RNA-dependent RNA polymerases, much larger changes may occur as a result of recombination. An extreme example of genetic change is found in defective interfering (DI) viral particles, where large sections of the genome of a parental virus have been deleted and the residual sub-genome fragment is replicated by complementation by co-infecting functional viruses. While most reports of DI particles have referred to studies in vitro, there is some evidence for the presence of DI particles in chronic viral infections in vivo. In this study, short fragments of dengue virus (DENV) RNA containing only key regulatory elements at the 3' and 5' ends of the genome were recovered from the sera of patients infected with any of the four DENV serotypes. Identical RNA fragments were detected in the supernatant from cultures of Aedes mosquito cells that were infected by the addition of sera from dengue patients, suggesting that the sub-genomic RNA might be transmitted between human and mosquito hosts in defective interfering (DI) viral particles. In vitro transcribed sub-genomic RNA corresponding to that detected in vivo could be packaged in virus like particles in the presence of wild type virus and transmitted for at least three passages in cell culture. DENV preparations enriched for these putative DI particles reduced the yield of wild type dengue virus following co-infections of C6-36 cells. This is the first report of DI particles in an acute arboviral infection in nature. The internal genomic deletions described here are the most extensive defects observed in DENV and may be part of a much broader disease attenuating process that is mediated by defective viruses.     

Comparison of ribonucleotide sequences from the genome of vesicular stomatitis virus and two of its defective interfering particles.

RNA genomes from standard vesicular stomatitis virus and two defective interfering (DI) particles dI 0.33 (DI-T) and DI 0.52, were purified and digested with RNase T1. The resulting oligonucleotides were labeled at the 5' end with [32P]ATP and separated by two-dimensional electrophoresis in polyacrylamide gels. All of the major oligonucleotides containing 20 or more nucleotides were sequenced. Those oligonucleotides that were thought to be in common by their migration on polyacrylamide gels actually did have identical sequences. Those oligonucleotides thought to be unique to the DI RNAs either differed by only one nucleotide from oligonucleotides of the standard RNA or contained new sequences which were complementary to known sequences at the 5' end. These data indicate that RNAs from DI particles are not simple deletions but contain point mutations and additional complementary sequences.     

The complete sequence of a unique RNA species synthesized by a DI particle of VSV.

The 2S RNA synthesized in vitro by the RNA polymerase of a defective interfering (DI) particle of vesicular stomatitis virus was labeled at its 3' terminus with 32P-cytidine 3', 5' bisphosphate and RNA ligase. Analysis of the labeled RNA showed that it was a family of RNAs of different length but all sharing the same 5' terminal sequence. The largest labeled RNA was purified by gel electrophoresis, and the sequence of 41 of its 46 nucleotides was determined by rapid RNA sequencing methods. The assignment of the remaining 5 nucleotides was made on the basis of an analysis of one of the smaller RNAs and published data. A new approach in RNA sequencing based on the identification of 3' terminal nucleotides of rna fragments originally present in the DI product or generated during the ligation reaction confirmed most of the sequence. The complete sequence of this 46 nucleotide long plus-sense RNA is: ppACGAAGACCACAAAACCAGAUAAAAAA UAAAAACCACAAGAGGGUC-OH. This RNA anneals to the RNA of the DI particle from which it was synthesized, indicating that its synthesis is template-specified. At least the first 17 and possibly all of the nucleotides are also complementary to sequences at the 3' end of two other VSV DI particles which were derived independently and whose genomes differ significantly in length. These data suggest a common 3' terminal sequence among all VSV DI particles which contain part of the Lgene region of the parental genome.     

Interference among defective interfering particles of vesicular stomatitis virus

Three defective interfering (DI) particles of vesicular stomatitis virus (VSV), all derived from the same parental standard San Juan strain (Indiana serotype), were used in various combinations to infect cells together with the parental virus. The replication of their RNA genomes in the presence of other competing genomes was described by the hierarchical sequence: DI 0.52 particles greater than DI 0.45 particles less than or equal to DI-T particles greater than standard VSV. The advantage of one DI particle over another was not due simply to multiplicity effects nor to the irreversible occupation of limited cellular sites. Interference, however, did correlate with a change in the ratio of plus and minus RNA templates that accumulated intracellularly and with the presence of new sequences at the 3' end of the DI genomes. DI 0.52 particles contained significantly more nucleotides at the 3' end that were complementary to those at the 5' end of its RNA than did DI-T or DI 0.45 particles. The first 45 nucleotides at the 3' ends of all of the DI RNAs were identical. VSV and its DI particles can be separated into three classes, depending on their terminal RNA sequences. These sequences suggest two mechanisms, one based on the affinity of polymerase binding and the other on the affinity of N-protein binding, that may account for interference by DI particles against standard VSV and among DI particles themselves.     

Within-host spatial dynamics of viruses and defective interfering particles

Defective-interfering (DI) viruses arise spontaneously by deletion mutations. The shortened genomes of the DI particles cannot replicate unless they coinfect a cell with a wild-type virus. Upon coinfection, the DI genome replicates more quickly and outcompetes the wild type. The coinfected cell produces mostly DI viruses. At the population level, the abundances of DI and wild-type viruses fluctuate dramatically under some conditions. In other cases, the DI viruses appear to mediate persistent infections with relatively low levels of host cell death. This moderation of viral damage has led some to suggest DI particles as therapeutic agents. Previous mathematical models have shown that either fluctuation or persistence can occur for plausible parameter values. I develop new mathematical models for the population dynamics of DI and wild-type viruses. My work extends the theory by developing specific predictions that can be tested in the laboratory. These predictions, if borne out by experiment, will explain the key processes that control the diversity of observed outcomes. The most interesting prediction concerns the rate at which killed host cells are replaced. A low rate of replacement causes powerful epidemics followed by a crash in viral abundance. As the rate of replacement increases, the frequency of oscillations increases in DI and wild-type viral abundances, but the severity (amplitude) of the fluctuations declines. At higher replacement rates for host cells, nearly all cells become infected by DI particles and a low level of fluctuating, wild-type viremia persists.     

Structure of the rat cytomegalovirus genome termini.

The lytic replication cycle of herpesviruses can be divided into the following three steps: (i) circularization, in which, after infection, the termini of the linear double-stranded viral genome are fused; (ii) replication, in which the circular DNA serves as template for DNA replication, which generates large DNA concatemers; and (iii) maturation, in which the concatemeric viral DNA is processed into unit-length genomes, which are packaged into capsids. Sequences at the termini of the linear virion DNA are thought to play a key role in both genome circularization and maturation. To investigate the mechanism of these processes in the replication of rat cytomegalovirus (RCMV), we cloned, sequenced, and characterized the genomic termini of this betaherpesvirus. Both RCMV genomic termini were found to contain a single copy of a direct terminal repeat (TR). The TR sequence is 504 bp in length, has a high GC content (76%), and is not repeated at internal sites within the RCMV genome. The TR comprises several small internal direct repeats as well as two sequences which are homologous to herpesvirus pac-1 and pac-2 sites, respectively. The organization of the RCMV TR is unique among cytomegaloviruses with respect to the position of the pac sequences: pac-1 is located near the left end of the TR, whereas pac-2 is present near the right end. Both RCMV DNA termini carry an extension of a single nucleotide at the 3' end. Since these nucleotides are complementary, circularization of the viral genome is likely to occur via a simple ligation reaction.     

5'-proximal hot spot for an inducible positive-to-negative-strand template switch by coronavirus RNA-dependent RNA polymerase

Coronaviruses have a positive-strand RNA genome and replicate through the use of a 3' nested set of subgenomic mRNAs each possessing a leader (65 to 90 nucleotides [nt] in length, depending on the viral species) identical to and derived from the genomic leader. One widely supported model for leader acquisition states that a template switch takes place during the generation of negative-strand antileader-containing templates used subsequently for subgenomic mRNA synthesis. In this process, the switch is largely driven by canonical heptameric donor sequences at intergenic sites on the genome that match an acceptor sequence at the 3' end of the genomic leader. With experimentally placed 22-nt-long donor sequences within a bovine coronavirus defective interfering (DI) RNA we have shown that matching sites occurring anywhere within a 65-nt-wide 5'-proximal genomic acceptor hot spot (nt 33 through 97) can be used for production of templates for subgenomic mRNA synthesis from the DI RNA. Here we report that with the same experimental approach, template switches can be induced in trans from an internal site in the DI RNA to the negative-strand antigenome of the helper virus. For these, a 3'-proximal 89-nt acceptor hot spot on the viral antigenome (nt 35 through 123), largely complementary to that described above, was found. Molecules resulting from these switches were not templates for subgenomic mRNA synthesis but, rather, ambisense chimeras potentially exceeding the viral genome in length. The results suggest the existence of a coronavirus 5'-proximal partially double-stranded template switch-facilitating structure of discrete width that contains both the viral genome and antigenome.     

In vivo resolution of circular plasmids containing concatemer junction fragments from minute virus of mice DNA and their subsequent replication as linear molecules.

During replication of their linear, single-stranded DNA genomes, parvoviruses generate a series of concatemeric duplex intermediates. We have cloned, into Escherichia coli plasmids, junction fragments from these palindromic concatemers of minute virus of mice DNA spanning both the right end-to-right end (viral 5' to 5') and left end-to-left end (viral 3' to 3') fusions. When mouse cells were transfected with these circular plasmids and superinfected with minute virus of mice, the viral junctions were resolved and the plasmids replicated as linear chromosomes with vector DNA in their centers and viral DNA at their termini. Resolution did not occur when the concatemer joint was replaced by a different palindromic sequence or when the transfected cells were not superinfected, indicating the presence of latent origins of replication which could only be activated by a viral trans-acting factor(s). Moreover, the products of resolution and replication from the two termini were characteristically different. Analysis of individual terminal fragments showed that viral 5' (right-end) sequences were resolved predominantly into "extended" structures with covalently associated copies of the virally encoded NS-1 polypeptide, while bridges derived from the 3' (left) end resolved into both NS-1-associated extended termini and lower-molecular-weight "turn-around" forms in which the two DNA strands were covalently continuous. This pattern of resolution exactly coincides with that seen at the two termini of replicative-form intermediates in normal virus infections. These results demonstrate that the bridge structures are authentic substrates for resolution and indicate that the frequency with which extended versus turn-around forms of each terminus are generated is an intrinsic property of the telomere.