Reovirus reverse genetics: Incorporation of the CAT gene into the reovirus genome.

We have modified the infectious reovirus RNA system so as to generate a reovirus reverse genetics system. The system consists of (i) the plus strands of nine wild-type reovirus genome segments; (ii) transcripts of the genetically modified cDNA form of the tenth genome segment; and (iii) a cell line transformed so as to express the protein normally encoded by the tenth genome segment. In the work described here, we have generated a serotype 3 reovirus into the S2 double-stranded RNA genome segment of which the CAT gene has been cloned. The virus is stable, replicates in cells that have been transformed (so as to express the S2 gene product, protein final sigma 2), and expresses high levels of CAT activity. This technology can be extended to members of the orbivirus and rotavirus genera. This technology provides a powerful system for basic studies of double-stranded RNA virus replication; a nonpathogenic viral vector that replicates to high titers and could be used for clinical applications; and a system for providing nonselectable viral variants (the result of mutations, insertions, and deletions) that could be valuable for the construction of viral vaccine strains against human and animal pathogens.

Construction and characterization of a reovirus double temperature-sensitive mutant.

The infectious reovirus RNA system was used to construct a mutant with two temperature-sensitive (ts) lesions in genome segments M2 and S2, respectively. The double mutant is about 300 times more ts than either of its parents, which are about 1,500 and 170 times more ts than their wild-type parent reovirus ST3 strain Dearing. At 39 degrees C the double mutant is essentially unable to multiply. In spite of its striking temperature sensitivity, the double mutant elicits the formation of significant amounts of neutralizing antibodies in newborn mice. Possible mechanisms responsible for this are discussed, as is the significance of this double ts mutant in relation to current searches for safe and efficient vaccine strains.

Identification of signals required for the insertion of heterologous genome segments into the reovirus genome.

In cells simultaneously infected with any two of the three reovirus serotypes ST1, ST2, and ST3, up to 15% of the yields are intertypic reassortants that contain all possible combinations of parental genome segments. We have now found that not all genome segments in reassortants are wild type. In reassortants that possess more ST1 than ST3 genome segments, all ST1 genome segments appear to be wild type, but the incoming ST3 genome segments possess mutations that make them more similar to the ST1 genome segments that they replace. In reassortants resulting from crosses of the more distantly related ST3 and ST2 viruses that possess a majority of ST3 genome segments, all incoming ST2 genome segments are wild type, but the ST3 S4 genome segment possesses two mutations, G74 to A and G624 to A, that function as acceptance signals. Recognition of these signals has far-reaching implications for the construction of reoviruses with novel properties and functions.

Reovirus exists in the form of 13 particle species that differ in their content of protein sigma 1.

When electrophoresed in 0.7% agarose gels, populations of reovirus particles can be resolved into 13 well-defined bands that possess from 0 to 12 projection/spike-associated trimers of protein sigma 1. This state of affairs is not an artifact of purification, of the techniques used to demonstrate it, of aggregation, or of virus particle instability. Complexes of monoclonal antibody against protein sigma 1 with virus particles that possess only 1, or, to a lesser extent, 2 sigma 1 trimers are less stable (that is, more readily dissociated by sonication) than complexes of antibody and virus particles that possess 3 or more sigma 1 trimers. The specific infectivity of virus particles that possess 3 or more sigma 1 trimers is essentially the same; virus particles that possess only 2 sigma 1 trimers are about two-thirds as infectious; and particles that possess only 1 sigma 1 trimer still possess very significant infectivity (about one-third of maximum). Reovirus particles that possess no sigma 1 trimers (about 1 in 30) are essentially noninfectious. The reason reovirus particles do not possess a full complement of sigma 1 trimers is presumably the fact that only very small amounts of protein sigma 1 are synthesized in infected cells; and since possession of 3 such trimers is sufficient for maximal infectivity, and since the average number of sigma 1 trimers per reovirus particle is 7.1, there is presumably no selection for variants that synthesize larger amounts of sigma 1. On the contrary, such variants may well be at a selective disadvantage.

Generation of reovirus core-like particles in cells infected with hybrid vaccinia viruses that express genome segments L1, L2, L3, and S2.

When mouse L fibroblasts are infected with various combinations of recombinant vaccinia viruses possessing thymidine kinase (TK) genes with inserted reovirus genes that encode core components, particles are formed that closely resemble reovirus cores. In cells infected with vaccinia viruses that express reovirus proteins lambda 1 and sigma 2, particles are formed that are very similar to reovirus core shells; if, in addition, the cells are also infected with vaccinia virus that expresses protein lambda 2, particles are formed that also possess the characteristic icosahedrally located projections/spikes that are present on reovirus cores. If, in either case, the cells are also infected with vaccinia virus that expresses the reovirus RNA polymerase, protein lambda 3, the resultant particles are morphologically identical with those formed in its absence, but also contain protein lambda 3.

Translation of reovirus RNA species m1 can initiate at either of the first two in-frame initiation codons.

The m1 species of reovirus RNA, which encodes the minor protein component mu 2, possesses two initiation codons, one "strong" according to Kozak rules and preceded by 13 residues (IC1), the other "weak" and located 49 codons downstream of the first (IC2). In reovirus-infected cells only IC2 is used, but initiation from IC1 can be activated, and efficiency of initiation from either initiation codon modulated over a wide range, by coupling unrelated sequences to either or both ends of m1 RNA. For example, when the M1 genome segment is cloned into the thymidine kinase gene of vaccinia virus in such a way that various "irrelevant" stretches of nucleotides comprising restriction endonuclease cleavage sites or promoter remnants are coupled to the 5' end of m1 RNA, translation of the resultant transcripts is also initiated at IC2, with frequencies controlled by the nature of the attached sequences. However, in rabbit reticulocyte lysates these same transcripts are translated from IC1 as well as from IC2, and transcripts in which m1 RNA is preceded by long sequences of encephalomyocarditis virus RNA (from the T7 polymerase-controlled pTM1 vector) are translated exclusively from IC1. By contrast, m1 RNA itself is translated only from IC2. It appears that the most important factor that controls the extent to which translation is initiated from IC1 and IC2 is their "availability," which is likely to be a function of the extent to which the regions on either side of them interact with each other (and also, to a lesser extent, with the 3' untranslated region) either directly or via interaction with host cell proteins. The effects described here are of considerable potential significance when genetic material is rearranged as a result of translocations, insertions, deletions, and amplifications--that is, when sequences that are normally separated are brought into apposition.

Reovirus protein lambda 3 is a poly(C)-dependent poly(G) polymerase.

Reovirus protein lambda 3 has been isolated from cells infected with two recombinant vaccinia viruses into the TK gene of which the reovirus serotype3 strain Dearing L1 genome segment under the control of the bacteriophage T7 RNA polymerase promoter, or the T7 polymerase gene itself, had been cloned. Highly purified protein lambda 3 does not transcribe double-stranded reovirus RNA into single-stranded RNA, or plus-stranded reovirus RNA into minus-stranded RNA, but it does transcribe poly(C) into poly(G). It prefers Mn2+ to Mg2+. A polymer consisting of poly(C) linked linearly to poly(U) provided template activity only for its poly(C) moiety. Protein lambda 3 forms complexes with protein lambda 1, as well as with protein lambda 2, and with both lambda 1 and lambda 2, which are sufficiently stable to be precipitated by monospecific antisera. None of these complexes are capable of transcribing either ds- or ssRNA.

Reovirus genome segment assortment into progeny genomes studied by the use of monoclonal antibodies directed against reovirus proteins.

Using a panel of monoclonal antibodies (MABs) against reovirus proteins, we have identified proteins that associate with reovirus messenger RNA molecules prior to the generation of progeny double-stranded (ds) genome segments and proteins that are components of the structures within which progeny ds genome segments are generated. The following conclusions can be drawn from the results obtained. (1) Three proteins rapidly become associated with mRNA molecules to form single-stranded RNA-containing complexes (ssRCCs): the nonstructural protein microNS, the nonstructural protein sigma NS, and protein sigma 3. (2) Analysis of populations of ssRCCs in density gradients and by sequential exposure to various MABs indicates that some ssRCCs contain only microNS, others microNS and sigma NS or sigma 3, and others all three proteins. Each ssRCC contains one RNA molecule and, depending on the size of the RNA, 10-30 protein molecules. (3) The relative proportions of the individual RNA species in the ssRCC populations reflect the composition of the total mRNA population present in infected cells (which differs substantially from equimolarity). (4) RCCs that contain dsRNA, which become detectable as early as 4 hr after infection, contain not only microNS, sigma NS, and sigma 3, but also lambda 2. (5) The relative proportions of the 10 genome segments in dsRCCs are equimolar. This suggests that genome segment assortment into progeny genomes is linked to the transcription of plus strands into minus strands.

Isolation and enzymatic characterization of protein lambda 2, the reovirus guanylyltransferase.

Protein lambda 2 of reovirus serotype 3 has been purified to homogeneity from extracts of cells infected with hybrid vaccinia virus strain WR into whose TK gene of the reovirus L2 genome segment under the control of the CPV ATI protein gene promoter had been inserted. Protein lambda 2 is formed in large amounts (final purification factor about 180) as a monomer that shows no tendency to pentamerize into the reovirus core projections/spikes. Isolated protein lambda 2 is reversibly guanylylated by GTP (that is, it carries out the GTP-PPi exchange reaction) and can transfer the -GMP moiety to GTP to form GppppG, to GDP to form GpppG, and to 5'-pp-terminated RNA to form GpppG- caps. These studies confirm previous studies on reovirus cores that indicated that protein lambda 2 is the reovirus guanylyltransferase. Protein lambda 2 possesses neither nucleoside nor RNA triphosphatase activities, nor methyltransferase activities; thus it is the reovirus capping enzyme, but provides neither the required 5'-ppG-terminated substrate nor does it methylate the cap structure. These must be functions of lambda 2 pentamers or of other individual or complexed components of reovirus cores.

Reovirus RNA is infectious.

Conditions under which reovirus RNA is infectious have been worked out. In brief, single-stranded (plus-stranded, ss) and/or double-stranded (ds) RNA of reovirus serotype 3 (ST3 virus) is lipofected into L929 mouse fibroblasts together with a rabbit reticulocyte lysate in which ss or melted dsRNA has been translated. After 8 hr the cells are then infected with a helper virus, ST2 reovirus. Virus yields are harvested 24 or 48 hr later. Under these conditions virus that forms plaques by 5 days is produced, all of which is ST3 virus; ST2 virus forms plaques only after 12 days. No reassortants are present among the progeny. The virus yields are about 0.2 PFU/cell; immunofluorescence assays show that this progeny is derived from about 4% of the cells. Double-stranded RNA is 20 times as infectious as ssRNA; ds and ssRNA together yield 10 times as much infectious virus as dsRNA alone, the reason being that dsRNA greatly increases the infectiousness of ssRNA. All species of both ss and dsRNA are required for the operation of this additive effect. The primed rabbit reticulocyte lysate is not essential, but increases virus yields by 100-fold. Its activity is proportional to the time for which translation has proceeded; however, this activity is not due solely to newly synthesized proteins because destruction of the RNA following translation abolishes activity which cannot be restored by simple addition of more RNA. Translation of all species of RNA is essential. Whereas no reassortants are formed when ss and dsRNA of different genotypes are lipofected together, mixtures of dsRNAs of different genotypes do yield reassortants. The same is true for such mixtures of ssRNA. These findings will permit the introduction of new or altered genome segments into the reovirus genome. They open the way to the identification of encapsidation and assortment signals on reovirus genome segments, the characterization of functional domains on reovirus proteins, and the development of reovirus as an expression vector.

Reovirus protein sigma 1 translated in vitro, as well as truncated derivatives of it that lack up to two-thirds of its C-terminal portion, exists as two major tetrameric molecular species that differ in electrophoretic mobility.

Reovirus protein sigma 1 is the cell attachment protein that modulates tissue tropism and the nature of the antiviral immune response. This protein is present in reovirus particles in the form of 12 tetramers that are associated with the projections or spikes. We have analyzed a series of deletion mutants of protein sigma 1 in order to localize its oligomerization domain and found that progressive deletion from the C-terminus fails to affect ability to oligomerize, even when the deletion extends into the N-terminal heptapeptide repeat region. It was also found that native tetrameric protein sigma 1 synthesized in vitro, as well as its truncated derivatives, exists in two forms that differ in electrophoretic mobility. Possible reasons for this are discussed.

Incorporation of the reovirus M cell attachment protein into small unilamellar vesicles: incorporation efficiency and binding capability to L929 cells in vitro.

Neutral phosphatidylcholine/cholesterol (10 : 3) liposomes with the reovirus M cell attachment protein sigma 1 were made by means of detergent dialysis method. An immunological evaluation method revealed an incorporation efficiency (pg protein/microgram lipid) of 314. Binding studies with mouse fibroblasts (L929 cells) yielded a 10 fold improvement of uptake of coated liposomes compared to uncoated liposomes. Competition studies with reovirus serotype 3 demonstrated that coated liposomes were capable of binding to the reovirus receptor. In vitro incubation of rat Peyer's patches with either coated or uncoated liposomes resulted in a 10-20 fold higher uptake of the coated liposomes. These results suggest that selective adherence of a carrier system to M cells may facilitate delivery of carrier contents to mucosal underlying lymphoid tissue, thereby enhancing immune response.

Identification of conserved domains in the cell attachment proteins of the three serotypes of reovirus.

Sequence analysis of reovirus serotype 1 (ST1) and 2 (ST2) S1 genome segment cDNAs identified several differences from previously reported versions of their sequences. The sequences reported here comprise 1463 and 1440 base pairs, respectively; for comparison, the ST3 S1 genome segment is 1416 nucleotides long. The serotype 1 and 2 sigma 1 proteins are predicted to contain 470 and 462 amino acids, respectively; the ST3 sigma 1 protein is 455 amino acids long. As previously observed, the ST1 and ST2 sigma 1 proteins are much more closely related to each other than to that of ST3 (about 48 and 25% similarity, respectively, using a computer program that finds about 14% similarity among unrelated proteins). The sequences of the three S1 genome segments have diverged very extensively in all three codon positions, in some cases almost to the extent of randomness. Despite this, not only function but also shape and configuration have been retained (since the three sigma 1 proteins can be incorporated efficiently into completely heterologous capsids). Seventy-nine amino acid residues are conserved among all three serotypes, many of them clustered into five regions in which one-third or more of the residues are triply conserved. These regions may represent functionally conserved domains involved in oligomerization, cell attachment, and hemagglutination.