Search

This Invited Review is a kind of scientific autobiography based on the presentation at the Symposium "Viruses: Discovering Big in Small" held in honor of the author's 90th birthday (Moscow, March 2019)., (Copyright © 2019 Elsevier Inc. All rights reserved.)

Replication of picornavirus genomes is accomplished by the virally encoded RNA-dependent RNA polymerase (RdRP). Although the primary structure of this enzyme exhibits a high level of conservation, there are several significant differences among different picornavirus genera. In particular, a comparative alignment indicates that the C-terminal sequences of cardiovirus RdRP (known also as 3D(pol)), are 1-amino-acid residue (arginine or tryptophan) longer than that of the enterovirus or rhinovirus enzymes. Here, it is shown that alterations of the last codon of the RdRP-encoding sequence of mengovirus RNA leading to deletion of the C-terminal Trp460 or its replacement by Ala or Phe dramatically impaired viral RNA replication and, in the former case, resulted in a quasi-infectious phenotype (i.e., the mutant RNA might generate a low yield of pseudorevertants acquiring a Tyr residue in place of the deleted Trp460). The replacement of Trp460 by His or Tyr did not appreciably alter the viral growth potential. Homology modeling of three-dimensional structure of mengovirus RdRP suggested that Trp460 may be involved in interaction between the thumb and palm domains of the enzyme. Specifically, Trp460 of the thumb may form a hydrogen bond with Thr219 and hydrophobically interact with Val216 of the palm. The proposed interactions were consistent with the results of in vivo SELEX experiment, which demonstrated that infectious virus could contain Ser or Thr at position 219 and hydrophobic Val, Leu, Ile, as well as Arg (whose side chain has a nonpolar part) at position 216. A similar thumb-palm domain interaction may be a general feature of several RdRPs and its possible functional significance is discussed.

The Sabin oral poliovaccine (OPV) is extremely efficacious and safe, despite its inherent genetic instability. While reversion to nearly wild-type phenotype regularly occurs soon after the onset of OPV reproduction in the gastro-intestinal tract of vaccine recipients or their contacts, this is usually not a big problem, provided the vaccine is used either for mass vaccination or in populations with a relatively high level of anti-polio immunity. However, if these conditions are not met, the vaccine viruses are likely to be converted into highly transmissible agents with a nearly wild-type level of neurovirulence. Moreover, OPV viruses may persist and evolve even in adequately immunized populations. The current strategy for the "endgame" of poliovirus eradication envisions cessation of OPV usage shortly after the last isolation of a wild poliovirus. If implemented, this strategy would result in rapid growth of non-immune human populations at the time when OPV derivatives would very likely be persisting. Therefore, the planned cessation of OPV vaccination is associated with a very high, and in the author's opinion, unacceptable risk of polio outbreaks caused by OPV derivatives. The only currently available tool to curb such outbreaks is OPV, which should have been used at a global scale. Safe discontinuation of OPV vaccination will be possible only after an efficient new vaccine or an anti-poliovirus drug is available. To achieve this goal, stimulation of poliovirus research and elimination of organizational and financial obstacles preventing it are needed.

In several cell types, poliovirus activates the apoptotic program, implementation of which is suppressed by viral antiapoptotic functions. In such cells, productive infection leads to a necrotic cytopathic effect (CPE), while abortive reproduction, associated with inadequate viral antiapoptotic functions, results in apoptosis. Here, we describe two other types of cell response to poliovirus infection. Murine L20B cells expressing human poliovirus receptor responded to the infection by both CPE and apoptosis concurrently. Interruption of productive infection decreased rather than increased the proportion of apoptotic cells. Productive infection was accompanied by the early efflux of cytochrome c from the mitochondria in a proportion of cells and by activation of DEVD-specific caspases. Inactivation of caspase-9 resulted in a marked, but incomplete, prevention of the apoptotic response of these cells to viral infection. Thus, the poliovirus-triggered apoptotic program in L20B cells was not completely suppressed by the viral antiapoptotic functions. In contrast, human rhabdomyosarcoma RD cells did not develop appreciable apoptosis during productive or abortive infection, exhibiting inefficient efflux of cytochrome c from mitochondria and no marked activation of DEVD-specific caspases. The cells were also refractory to several nonviral apoptosis inducers. Nevertheless, typical caspase-dependent signs of apoptosis in a proportion of RD cells were observed after cessation of viral reproduction. Such "late" apoptosis was also observed in productively infected HeLa cells. In addition, a tiny proportion of all studied cells were TUNEL positive even in the presence of a caspase inhibitor. Degradation of DNA in such cells appeared to be a postmortem phenomenon. Biological relevance of variable host responses to viral infection is discussed.

Human prothymosin alpha is a proliferation-related nuclear protein undergoing caspase-mediated fragmentation in apoptotic cells. We show here that caspase-3 is the principal executor of prothymosin alpha fragmentation in vivo. In apoptotic HeLa cells as well as in vitro, caspase-3 cleaves prothymosin alpha at one major carboxy terminal (DDVD(99)) and several suboptimal sites. Prothymosin alpha cleavage at two amino-terminal sites (AAVD(6) and NGRD(31)) contributes significantly to the final pattern of prothymosin alpha fragmentation in vitro and could be detected to occur in apoptotic cells. The major caspase cleavage at D(99) disrupts the nuclear localization signal of prothymosin alpha, which leads to a profound alteration in subcellular localization of the truncated protein. By using a set of anti-prothymosin alpha monoclonal antibodies, we were able to observe nuclear escape and cell surface exposure of endogenous prothymosin alpha in apoptotic, but not in normal, cells. We demonstrate also that ectopic production of human prothymosin alpha and its mutants with nuclear or nuclear-cytoplasmic localization confers increased resistance of HeLa cells toward the tumor necrosis factor-induced apoptosis.

A HeLa cell line expressing the green fluorescent protein fused to the SV40 T-antigen nuclear localization signal (EGFP-NLS) was established. Fluorescence in these cells was confined to the nuclei. After poliovirus infection, cytoplasmic fluorescence in a proportion of cells could be detected by 1 h postinfection (p.i.) and in virtually all of the fluorescent cells by 2 h p.i. The relocation could be prevented by cycloheximide but not by inhibition of poliovirus replication by guanidine. HCl. Nuclear exit of a protein composed of three copies of GFP fused to the NLS also occurred upon poliovirus infection. A similar redistribution of EGFP-NLS took place upon infection with coxsakievirus B3 and, to a lesser extent, with vesicular stomatitis virus. The EGFP-NLS efflux was not due to the loss of NLS. Thus, some positive-strand and negative-strand RNA viruses trigger a rapid nonspecific relocation of nuclear proteins., (Copyright 2000 Academic Press.)

The death of poliovirus-infected cells may occur in two forms: canonical cytopathic effect (CPE) (on productive infections) or apoptosis (when the viral reproduction is hindered by certain drugs or some other restrictive conditions). Morphological manifestations of the CPE and apoptosis, being distinct, share some traits (e.g., chromatin condensation and nuclear deformation). It was shown here that a permeable caspase inhibitor, benzyloxycarbonyl-Val-Ala-Asp-(OMe) fluoromethyl ketone (zVAD.fmk), prevented the development of the poliovirus-induced apoptosis on abortive infection. The apoptotic pathway could be dissected by an inhibitor of chymotrypsin-like serine proteases, N-tosyl-l-phenylalanine chloromethyl ketone (TPCK), which prevented the cleavage of DNA to oligonucleosome-sized pieces and nuclear fragmentation but did not suppress cellular shrinkage, cytoplasmic blebbing, and partial chromatin condensation. These results demonstrate that caspase activation is involved in the execution phase of the viral apoptosis and suggest that a nuclear subset of the apoptotic program is under a separate control, involving a TPCK-sensitive event. Neither zVAD.fmk nor TPCK, at the concentrations affecting the apoptotic response, exerted appreciable influence on the virus growth or cellular pathological changes on productive infection, indicating that the pathways leading to the poliovirus-evoked CPE and apoptosis are different., (Copyright 1998 Academic Press.)

Poliovirus mutants with extended (> 150-nt) deletions in the 5'-untranslated region between the internal ribosome entry site and the initiator codon have been selected previously (Pilipenko et al., Cell 68, 119-131, 1992; Gmyl et al., J. Virol. 67, 6309-6316, 1993). These deletions were transferred into the genome of a mouse-pathogenic poliovirus strain and found to be strongly attenuating. The deletions can be considered as covering three structural elements, a stem-loop (domain E) with a conserved cryptic AUG and two spacers, upstream and downstream of it. In an attempt to identify putative essential determinants of neurovirulence in these individual structural elements, appropriate mutants were engineered. The results demonstrated that neither of the above elements is essential for neurovirulence. The results strongly suggested that the presence of a cryptic AUG in the oligopyrimidine/AUG tandem followed, at a sufficient distance, by the initiator codon was necessary to ensure the neurovirulent phenotype of our constructs. On the other hand, the attenuated phenotype appeared to correlate with the occurrence of the initiator AUG as a moiety of the oligopyrimidine/AUG tandem. Possible mechanisms underlying these effects are discussed. Identification of the cryptic AUG as an essential determinant for neurovirulence provides a rational basis for the design of genetically stable attenuated poliovirus variants.

Field trials and the first mass immunization campaign using Oral Poliomyelitis Vaccine (OPV) from the strains derived by Albert Sabin were carried out in the former U.S.S.R. in 1959 under the guidance of Mikhail Chumakov. The results obtained during these trials contributed greatly to the general acceptance of OPV as a safe and efficient preparation. Mechanisms responsible for the attenuated phenotype of the Sabin strains have long been studied in Russia. For the first time, recombination analysis was used for mapping genetic determinants of attenuation, which were found to reside largely in the 5' half of the Sabin strains RNA. It has been established that the phenotype effect of a subset of the major attenuating mutations is due to the impairment of the translation initiation.

Tick-borne encephalitis virus (TBEV) RNA was translated in extracts from Krebs-2 cells and in rabbit reticulocyte lysates. In the former system, two polypeptides, p53 and p13, corresponding to envelope (E) and core (C) proteins of the virion, respectively, were synthesized preferentially. In contrast, the major product in reticulocyte lysates was represented by a heterogeneous set of high-molecular-weight polypeptides which did not appear to include p53 or p13. The reticulocyte lysates, however, acquired the ability to produce structural proteins (p53 at least) after addition of purified membranes isolated from the rough endoplasmic reticulum of Krebs-2 cells. On the other hand, the ability of Krebs-2 extracts to generate identifiable viral structural proteins was lost after degradation of membranes by the nonionic detergent Triton X-100. These findings strongly suggest that membrane-dependent processing of protein precursors is involved in the formation of TBEV structural proteins. Evidence has been obtained that only nascent precursor polypeptides can be processed efficiently into structural proteins in the membrane-dependent reaction.

Replication complexes (RC) containing endogenous viral RNA templates and the viral RNA polymerase were isolated from encephalomyocarditis (EMC) virus-infected Krebs II cells by two different procedures one of which included removal of the bulk of the associated proteins by treatment with 2 M LiCl. Replacement of one or all the four nucleoside triphosphate (NTP) substrates by corresponding nucleoside diphosphates (NDPs) did not eliminate the ability of the RC to synthesize viral RNA products. Moreover, the complexes were shown to accept as substrates even nucleoside monophosphates (NMPs), provided at least one NTP was present in the system. These results suggested that nucleotide kinases, NMP kinase and NDP kinase, were associated with the RC, and this suggestion was confirmed directly. The RC could be resolved, by sucrose gradient centrifugation, into distinct components that exhibited marked preference for either NDPs or NTPs as substrates for RNA synthesis. It is suggested that, in the NDP-preferring components, the NTP molecules generated by the built-in nucleotide kinase system are channeled directly to the vicinity of the replication fork.

Poliovirus strains of type 1 and type 3 carrying genetically mapped ts mutations and differring in growth response to guanidine have been used to infect HeLa cells. With four heterotypic pairs of the mutants, recombinants with the crossover points between the loci coding for the antigenic properties, on the one hand, and for the sensitivity to guanidine, on the other, have been obtained. The recombinants have been identified on the basis of their phenotypic properties and, in particular, of the pattern of inheritance of unselected markers. One recombinant has been characterized by fingerprinting virus-specific polypeptides. It has been found that the capsid proteins (VP2, VP3, and VP1) of this recombinant originate from the type 3 parent, whereas the nonstructural polypeptides (X, 2, and 4) are inherited from the type 1 parent. Implications of the poliovirus intertypic recombination are discussed.

Recombinants with a centrally located crossover point were selected from crosses between poliovirus type 1 strains and intertypic (type 3/type 1) recombinants. Two such recombinants were characterized in some detail. In one of them (v1/a1-6), the 5' half of the genome was derived from a virulent type 1 strain, while the 3' half came from an attenuated type 1 strain. The genome of the other recombinant (a1/v1-7) had the reverse organization, with the 5' and 3' halves being derived from the type 1 attenuated and virulent strains, respectively. As deduced from the RNase T1 oligonucleotide maps, the a1/v1-7 genome also had a relatively short centrally located insert of the poliovirus type 3 origin. Both recombinants exhibited ts phenotypes. The RNA phenotypes of the recombinants corresponded to that of the parent donating the 3' half of the genome, v1/a1-6 and a1/v1-7 expressing RNA- and RNA +/- characters, respectively. Despite being a ts RNA- virus, v1/a1-6 proved to be neurovirulent when injected intracerebrally into Cercopithecus aethiops monkeys, although it exhibited a somewhat diminished level of pathogenicity as compared to its virulent type 1 parent. Recombinant a1/v1-7 behaved as an attenuated strain. These data supported our previous conclusion drawn from the experiments with intertypic poliovirus recombinants that the attenuated phenotype of poliovirus depends largely on the structure of the 5' half of its genome, although mutations of the 3' half may alleviate the virulence of the virus to a degree.

An attempt was made to map, in a general way, the region of the poliovirus genome that is responsible for the neurovirulent and attenuated phenotypes of different virus strains. A set of four recombinants was investigated, one described previously (E. A. Tolskaya, L. I. Romanova, M. S. Kolesnikova, and V. I. Agol, 1983, Virology 124, 121-132) and three obtained in the present work with the following genetic structure: a 5' end-adjacent segment of the genome derived from either a virulent strain (452/62 3D), or from an attenuated strain (Leon-2) of poliovirus type 3, the remaining RNA sequences being derived from either a virulent strain (Mgr), or an attenuated strain (LSc-gr3) of poliovirus type 1. The crossover points in the recombinant genomes were centrally located, somewhere between the gene(s) that determines antigenic specificity of the virus and the locus that determines resistance of virus multiplication to low doses of guanidine. The recombinant nature of the newly selected clones was definitively established by mapping RNase T1 oligonucleotides of their genome. The recombinants were characterized with respect to their ability to produce infectious progeny and synthesize viral RNA at an elevated temperature. Neurovirulence of the recombinants was assayed by intracerebral inoculation of monkeys. Irrespective of the origin of the 3' end-adjacent segment of the genome, the recombinants that inherited the 5' end-adjacent segment from the neurovirulent parent were neurovirulent, whereas the recombinants with the 5' end-adjacent segment derived from the attenuated parent were not. The results suggest that the major determinants of neurovirulence of these recombinants (and by inference, of their parental viruses) reside in the 5' end-adjacent segment of poliovirus genome, known to code for capsid proteins.

Upon translation of poliovirus RNA in reticulocyte lysates, initiation occurs largely "incorrectly," that is, at sites in the middle of the viral genome rather than at the beginning of the polyprotein reading frame; the anomaly appears to be due to an initiation factor deficiency. Here, a fraction which stimulated initiation at the correct site, provisionally called "initiation correcting factor" (ICF), was partially purified from Krebs-2 cells. The ICF activity appeared to copurify with a complex of initiation factors eIF-2 and eIF-2B. The ability of ICF to stimulate, in reticulocyte lysates, the correct initiation of translation on the RNAs from neurovirulent and attenuated type 1 and type 3 poliovirus strains was investigated. Like crude initiation factor preparations, ICF appeared to be relatively less active with the RNAs from attenuated strains, the difference being especially pronounced for the type 3 strains. For the latter strains, the data suggested an important role of the nucleotide at, and perhaps around, position 472 in determining a response to the addition of ICF. It is proposed that interaction of a specific segment of the viral RNA with one or more of initiation factors plays an important part in the mechanism of translation of the picornavirus genomes, poliovirus attenuation, and, possibly, pathogenesis of poliomyelitis.

A model of secondary structure common for the central part (ca. 400 nucleotides) of the 5'-untranslated regions (5'-UTR) of all the so far sequenced genomes of polioviruses, coxsackieviruses, and rhinoviruses was derived on the basis of evolutionary and thermodynamic considerations. According to the model, this part of the genome comprises three domains, which appear to be involved, at least in the poliovirus genome, in the control of viral neurovirulence and in vitro translation. Some salient features of this model were supported by investigating RNAs of five poliovirus and one coxsackievirus strains with respect to their accessibility to modifications with dimethyl sulfate and sensitivity to single-strand- and double-strand-specific nucleases. In contrast to the previous suggestion, no major changes in the conformation of the Sabin vaccine poliovirus type 3 5'-UTR due to the transition in position 472 were observed. The biological relevance of the conserved primary and secondary structure elements in the picornaviral 5'-UTRs is discussed.

In mRNA-dependent extracts of Krebs-2 cells, RNAs from attenuated strains of poliovirus type 1 and type 3 exhibited diminished template activity as compared to RNAs from the respective virulent counterparts. This defect appeared to be due to the impaired initiation of viral polyprotein synthesis as evidenced by a relatively low level of accumulation of polypeptide 1a (which corresponds to an NH2-terminal region of the polyprotein) in samples programmed with RNAs from attenuated strains. In reticulocyte lysates, where poliovirus RNA is translated predominantly from abnormal (internal) sites [Dorner et al. (1984) J. Virol. 50, 507-514], this difference in the overall template activity of the attenuated and virulent poliovirus genomes was less pronounced, but the correct initiation (as judged by polypeptide 1a accumulation) was again more efficient on RNAs from virulent strains. It is suggested that template deficiency is a factor contributing to the attenuated phenotype of poliovirus strains studied. A possible involvement of nucleotide sequences located far upstream from the initiator codon in the control of translation of poliovirus genome is briefly discussed.

The nucleotide sequence of crossover sites in the genome of four intertypic (type 3/type 1) poliovirus recombinants has been determined. The approximate boundaries of the crossover regions were first estimated by RNase T1 oligonucleotide mapping of the recombinant genomes; then appropriate regions were sequenced by the chain termination method using oligonucleotide-primed reverse transcription of the recombinant RNAs. The crossover sites (defined as the contiguous sequences shared by the recombinant and both parental genomes flanked, in the recombinant genome, by heterotypic RNA segments) are 5, 5, 7, and 11 nucleotides long, respectively. The recombination was precise and was not accompanied by any other genetic alterations. The recombination sites were found to be located within genome segments having a potential to form secondary structure elements. Based on this observation, a model of recombination between picornaviral RNA genomes has been proposed. The essence of this model consists in bringing together homologous regions of two recombining RNA genomes via formation of intermolecular duplexes, detachment of the nascent 3' end of the newly synthesized complementary RNA from a "parting" site on the first template and its subsequent "jumping" to the identical (or closely related) "anchoring" site on the other template. Features of this model are discussed in some detail.

A series of intertypic (type 3/type 1) poliovirus recombinants was obtained whose crossover sites were expected to be located in the middle of the viral genome, between the loci encoding type-specific antigenic properties, on the 5' side, and an altered sensitivity to guanidine, on the 3' side. The primary structures of the crossover regions in the genomes of these recombinants were determined by the primer extension method. The length of the crossover sites (the uninterrupted sequences shared by the recombinant and both parental genomes that are flanked, in the recombinant RNAs, by two heterotypic segments) varied between 2 and 32 nucleotides, but the majority of the sites were 5 nucleotides long or shorter. The crossover sites were nonrandomly distributed over the presumably available genome region: only a single such site was found within the gene for polypeptide 2A, whereas an apparent clustering of the crossover sites was encountered in other genomic segments. When the crossover sites were superimposed on a model of the secondary structure of the relevant region of the viral RNA molecule, a pattern consistent with the previously proposed mechanism of poliovirus recombination (L.I. Romanova, V.M. Blinov, E.A. Tolskaya, E.G. Viktorova, M.S. Kolesnikova, E.I. Guseva, and V.I. Agol (1986) Virology 155, 202-213) was observed. It is suggested that the nonrandom distribution of the crossover sites in the genomes of intertypic poliovirus recombinants was due to two factors: the existence of preferred sites for recombination, and selection against recombinants with a lowered level of viability.

When the osmotic pressure of the maintenance medium is decreased, poliovirus reproduction is inhibited. Poliovirus strains may vary in their sensitivity to the effect of hypotonic solutions. Mutants have been selected (designated as osm mutants) the reproduction of which is characterized by particularly high resistance to media with reduced osmotic pressure. Initial stages of the virus-cell interaction proceed at similar rates in physiologic and hypotonic solutions. Reactions sensitive to hypotonic solutions take place in the second half of the latent period and during the stage of virus maturation. The multiplication of viruses in moderately hypotonic solutions exhibits a marked dependence upon the temperature and the presence of cystine in the medium, even though these factors have relatively small influence upon the reproduction of the viruses in isotonic solutions. Despite complete cessation of the production of infectious virus in certain hypotonic media, there is a considerable synthesis of virus-induced RNA. At least some of the RNA produced under these conditions is infectious. With a further decrease of the osmotic pressure of the medium the synthesis of the viral RNA is inhibited. Some possible explanations of the observed facts are briefly discussed.