Conserved structural domains in the 5'-untranslated region of picornaviral genomes: an analysis of the segment controlling translation and neurovirulence

Correlation of RNA secondary structure and attenuation of Sabin vaccine strains of poliovirus in tissue culture

Part of the 5' noncoding regions of all three Sabin vaccine strains of poliovirus contains determinants of attenuation that are shown here to influence the ability of these strains to grow at elevated temperatures in BGM cells. The predicted RNA secondary structure of this region (nt 464-542 in P3/Sabin) suggests that both phenotypes are due to perturbation of base-paired stems. Ts phenotypes of site-directed mutants with defined changes in this region correlated well with predicted secondary structure stabilities. Reversal of base-pair orientation had little effect whereas stem disruption led to marked increases in temperature sensitivity. Phenotypic revertants of such viruses displayed mutations on either side of the stem. Mutations destabilizing stems led to intermediate phenotypes. These results provided evidence for the biological significance of the predicted RNA secondary structure.

Conservation of RNA-protein interactions among picornaviruses

Picornavirus genomes encode unique 5' noncoding regions (5' NCRs) which are approximately 600 to 1,300 nucleotides in length, contain multiple upstream AUG codons, and display the ability to form extensive secondary structures. A number of recent reports have shown that picornavirus 5' NCRs are able to facilitate cap-independent internal initiation of translation. This mechanism of translation occurs in the absence of viral gene products, suggesting that the host cell contains the necessary components for the cap-independent internal initiation of translation of picornavirus RNAs as well as cellular mRNAs. In an attempt to identify some of the perhaps novel cellular proteins involved in this newly discovered mechanism of translation, we utilized RNA mobility shifts assays to identify and characterize interactions that occur between the 5'NCR of poliovirus type 1 (PV1) and cellular proteins. In this report, we describe two separate interactions between RNA structures from the 5' NCR of PV1 and proteins present in extracts from HeLa cells as well as other cell types. We describe the interaction between nucleotides 186 to 220 (stem-loop D) and a cellular protein(s) present in HeLa cell extracts. Mutational analysis of this stem-loop structure suggests that maintenance of a base-paired structure in the lower stem is necessary to present the sequences which directly interact with the protein(s). We also describe the interaction between nucleotides 220 to 460 (stem-loop E) and a cellular protein present in HeLa cell extracts. This RNA binding activity fractionates to a specific ammonium sulfate fraction (A cut) of a ribosomal salt wash. Mutational analysis of the stem-loop E structure suggests that the preservation of an extensive RNA structure is necessary for a strong interaction with the cellular protein(s), although smaller RNAs derived from this region of the 5' NCR can interact to lesser extents. Finally, we show that both of these RNA-protein interactions are conserved among the closely related enteroviruses PV1 and coxsackievirus type B3, human rhinovirus type 14, and the more distantly related cardiovirus Theiler's murine encephalomyelitis virus, suggesting that such RNA-protein interactions serve basic functions which are conserved and utilized by each of these picornaviruses.

Influence of Theiler's murine encephalomyelitis virus 5' untranslated region on translation and neurovirulence

The DA strain of Theiler's murine encephalomyelitis virus (TMEV), a picornavirus, causes a persistent, restricted infection and demyelinating disease in mice. In contrast, the GDVII strain causes an acute, fatal, neuronal disease and is highly neurovirulent. To investigate the role of the TMEV 5' untranslated region (UTR) in translational efficiency and the TMEV subgroup differences, we tested the translational efficiency of transcripts in vitro derived from plasmids containing DA, GDVII, or DA/GDVII chimeric 5' UTRs preceding a reporter gene or the rest of the TMEV genome. We demonstrated that GDVII RNA translates more efficiently in rabbit reticulocyte lysate than DA RNA and that this enhanced translation is mediated by multiple domains in the GDVII 5' UTR as well as a region of the genome outside of the 5' UTR. We also identified a region within DA nucleotides 14 to 395 which inhibits translation of DA RNA and could contribute to the persistent, restricted DA central nervous system infection; the predicted secondary structure of the 5' UTR demonstrates a remarkable stem-loop structure within this region that is relatively unique among picornaviruses. Data from experiments involving DA/GDVII chimeric 5' UTR full-length infectious cDNA clones suggested that sequences in the 5' UTR can affect the neurovirulence phenotype but that translational efficiency is necessary but not sufficient for neurovirulence. These studies emphasize the multigenic nature of neurovirulence and the importance of translation in the regulation of picornaviral gene expression.

Initiation of translation of human rhinovirus RNA: mapping the internal ribosome entry site

In order to map the 3' boundary of the segments needed for translation initiation at the correct site on human rhinovirus 2, deletions were made from the 3' end of the viral 5'-untranslated region. These truncated viral segments were placed immediately upstream of a reporter gene, a derivative of the influenza virus NS cDNA, either as monocistronic constructs or as dicistronic constructs in which the upstream cistron was the Xenopus laevis cyclin B2 cDNA. In vitro transcripts of these clones were translated in the rabbit reticulocyte lysate system, with or without supplementation with crude HeLa cell initiation factors, or in a HeLa cell-free system. When the full-length viral 5'-untranslated region was present, the HeLa cell factors strongly stimulated the synthesis of the NS-related polypeptides, especially in the case of the dicistronic mRNAs. Deletions from the 3' end extending up to nt 562 had little effect on translation efficiency or the response to HeLa cell factors, but more extensive deletions resulted in the complete loss of response to these factors, an almost total inhibition of NS synthesis from dicistronic mRNAs, and a partial inhibition in the case of the monocistronic mRNAs. In the case of a deletion extending to nt 554, insertion of a 15 nucleotide residue linker failed to restore efficient translation initiation. We conclude that the essential sequences for internal initiation extend to a point located between nt 554 and 562, and that the ribosome entry site, defined as the most 5'-proximal point where the ribosome can bind in an initiation-competent manner, must lie within 6 residues on either side of nt 562, and certainly not further downstream than nt 568.

Specific interactions of HeLa cell proteins with proposed translation domains of the poliovirus 5' noncoding region

To determine which sequences or structures in the poliovirus 5' noncoding region (5'NCR) are involved in binding proteins used for internal ribosome binding and protein synthesis initiation, translation competition assays were performed in rabbit reticulocyte lysates in the presence and absence of HeLa cell extract. The results revealed two functional domains in the poliovirus 5'NCR. One, requiring nucleotides (nts) 457 to 626, binds proteins that are required for translation of all mRNAs and that are present in both reticulocyte lysates and HeLa cell extracts. Another, contained within nts 286 to 456, interacts with proteins that are specific for poliovirus translation and are present in HeLa cells but not in significant amounts in rabbit reticulocyte lysates. In order to detect HeLa cell proteins that interact stably with the 5'NCR of poliovirus, UV cross-linking was used. At least four major protein-RNA complexes were identified, three of which were shown by RNA competition analysis to bind specifically to defined domains within the 5'NCR. Protein A (54 kDa) cross-linked to RNA sequences and/or structures located between nts 457 and 626; proteins B (48 kDa) and C (38 kDa) bound to nts 286 to 456.

Towards identification of cis-acting elements involved in the replication of enterovirus and rhinovirus RNAs: a proposal for the existence of tRNA-like terminal structures

On the basis of a comparative analysis of published sequences, models for the secondary structure of the 3'-terminal [poly(A)-preceding] untranslated region of the entero- and rhinovirus RNAs were worked out. The models for all these viruses share a common core element, but there are an extra enterovirus-specific element and still an additional element characteristic of a subset of enterovirus RNAs. The two latter models were verified for poliovirus and coxsackievirus B genomes by testing with single-strand and double-strand specific enzymatic and chemical probes. A tRNA-like tertiary structure model for the 3'-terminal folding of enterovirus RNAs was proposed. A similar folding was proposed for the 3' termini of the negative RNA strands as well as for the 5' termini of the positive strand of all entero- and rhinovirus RNAs. Implications of these data for template recognition during negative and positive RNA strands synthesis and for the evolution of the picornavirus genomes are discussed.

Cardioviral internal ribosomal entry site is functional in a genetically engineered dicistronic poliovirus

High mutation rates have driven RNA viruses to shorten their genomes to the minimum possible size. Mammalian (+)-strand RNA viruses and retroviruses have responded by reducing the number of cis-acting regulatory elements, a constraint that has led to the emergence of the polyprotein. Poliovirus is a (+)-stranded picornavirus whose polyprotein, encoded by an open reading frame spanning most of the viral RNA, is processed by virus-encoded proteinases. Despite their genetic austerity, picornaviruses have retained long 5' untranslated regions, which harbour cis-acting elements that promote initiation of translation independently of the uncapped 5' end of the viral messenger RNA. These elements are termed 'internal ribosomal entry sites' and are formed from highly structured RNA segments of at least 400 nucleotides. How these elements function is not known, but special RNA-binding proteins may be involved. The ribosome or its 40S subunit probably binds at or near a YnXmAUG motif (where Y is a pyrimidine and X is a purine) at the 3' border of the internal ribosomal entry site, which either provides the initiating codon or enables the ribosome to translocate to one downstream (E.W. et al., submitted). Initiation from most eukaryotic messenger RNAs usually occurs by ribosomal recognition of the 5' and subsequent scanning to the AUG codon. Here we describe a genetic strategy for the dissection of polyproteins which proves that an internal ribosomal entry site element can initiate translation independently of the 5' end.

Intracellular modifications induced by poliovirus reduce the requirement for structural motifs in the 5' noncoding region of the genome involved in internal initiation of protein synthesis

A series of genetic deletions based partly on two RNA secondary structure models (M. A. Skinner, V. R. Racaniello, G. Dunn, J. Cooper, P. D. Minor, and J. W. Almond, J. Mol. Biol. 207:379-392, 1989; E. V. Pilipenko, V. M. Blinov, L. I. Romanova, A. N. Sinyakov, S. V. Maslova, and V. I. Agol, Virology 168:201-209, 1989) was made in the cDNA encoding the 5' noncoding region (5' NCR) of the poliovirus genome in order to study the sequences that direct the internal entry of ribosomes. The modified cDNAs were placed between two open reading frames in a single transcriptional unit and used to transfect cells in culture. Internal entry of ribosomes was detected by measuring translation from the second open reading frame in the bicistronic mRNA. When assayed alone, a large proportion of the poliovirus 5' NCR superstructure including several well-defined stem-loops was required for ribosome entry and efficient translation. However, in cells cotransfected with a complete infectious poliovirus cDNA, the requirement for the stem-loops in this large superstructure was reduced. The results suggest that virus infection modifies the cellular translational machinery, so that shortened forms of the 5' NCR are sufficient for cap-independent translation, and that the internal entry of ribosomes occurs by two distinct modes during the virus replication cycle.

Prokaryotic-like cis elements in the cap-independent internal initiation of translation on picornavirus RNA

Initiation of translation on picornavirus RNAs is accomplished through internal binding of ribosomes to a complex cis-acting element. Here we show that efficient function of this element involves two appropriately spaced smaller elements: UUUCC and an AUG. This conclusion emerged from analysis of the genome structures of spontaneous revertants of mutant polioviruses with extended insertions between the UUUCC and AUG motifs. It was confirmed by the results obtained with specially designed constructs. A similarity to the prokaryotic translation initiation mechanism, which involves the Shine-Dalgarno sequence, is emphasized, but in the picornavirus system the position of the UUUCC must be strictly fixed relative to upstream cis-acting elements, and the AUG may not necessarily serve as an initiation codon.

Interaction of a cellular 57-kilodalton protein with the internal translation initiation site of foot-and-mouth disease virus

A cellular 57-kDa protein (p57) that binds specifically to the internal translation initiation site in the 5' untranslated region of foot-and-mouth disease virus RNA was detected in cell extracts of different mammalian species by UV cross-linking. The protein binds to two distinct sites of the translation control region which have as the only common sequence a UUUC motif. The first binding site consists of a conserved hairpin structure, whereas the second binding site contains an essential pyrimidine-rich region without obvious secondary structure. Competition experiments indicate that the complexes with the two binding sites were formed by a single p57 species. The protein binds also to the 5' untranslated region of other picornaviruses. Results from footprint analyses with foot-and-mouth disease RNA suggest the participation of additional cellular factors in the translation initiation complex.

In vitro mutational analysis of cis-acting RNA translational elements within the poliovirus type 2 5' untranslated region

Initiation of translation on poliovirus RNA occurs by internal binding of ribosomes to a region within the 5' untranslated region (UTR) of the mRNA. This region has been previously roughly mapped between nucleotides 140 and 631 of the 5' UTR and termed the ribosome landing pad. To identify cis-acting elements in the 5' UTR of poliovirus type 2 (Lansing strain) RNA that confer cap-independent internal initiation, we determined the in vitro translational efficiencies of a series of deletion and point mutations within the 5' UTR of the mRNA. The results demonstrate that the 3' border of the core poliovirus ribosome landing pad is located between nucleotides 556 and 585, whereas a region extending between nucleotides 585 and 612 confers enhanced translation. We studied two cis-acting elements within this region of the 5' UTR: a pyrimidine stretch which is critical for translation and an AUG (number 7 from the 5' end) that is located approximately 20 nucleotides downstream from the pyrimidine stretch and augments translation. We also show that the stem-loop structure which contains this AUG is not required for translation.

The 5' nontranslated region of hepatitis A virus RNA: secondary structure and elements required for translation in vitro

Although the lengthy 5' nontranslated regions (5'NTRs) of other picornaviral RNAs form highly ordered structures with important functions in viral translation, little is known about the 5'NTR of hepatitis A virus (HAV). We determined the nearly complete 5'NTR nucleotide sequences of two genetically divergent HAV strains (PA21 and CF53) and included these data in a comparative phylogenetic analysis of the HAV 5'NTR. We identified covariant nucleotide substitutions predictive of conserved secondary structures and used this information to develop a model of the 5'NTR secondary structure, which was further refined by thermodynamic predictions and nuclease digestion experiments. According to this model, the 5'NTR comprises six major structural domains. Domains I and II (bases 1 to 95) contain a 5'-terminal hairpin and two stem-loops followed by a single-stranded and highly variable pyrimidine-rich tract (bases 96 to 154). The remainder of the 5'NTR (domains III to VI, bases 155 to 734) contains several complex stem-loops, one of which may form a pseudoknot, and terminates in a highly conserved region containing an oligopyrimidine tract preceding the putative start codon by 13 bases. To determine which structural elements might function as an internal ribosome entry site, RNA transcripts representing the HAV 5'NTR with progressive 5' deletions were translated in rabbit reticulocyte lysates. The translation product was truncated, unprocessed P1 polyprotein. Removal of the 5'-terminal 354 bases of the 5'NTR had little effect on translation. However, deletion to base 447 slightly decreased translation, while deletion to base 533 almost completely abolished it. These data indicate that sequences 3' of base 355 play an important role in the translation mechanism utilized by genomic-length HAV RNA. Significantly, this region shares several conserved structural features with the internal ribosome entry site element of murine encephalomyocarditis virus.

Structural and functional analysis of the ribosome landing pad of poliovirus type 2: in vivo translation studies

The naturally uncapped genomic and mRNAs of poliovirus initiate translation by an internal ribosome-binding mechanism. The mRNA 5' untranslated region (UTR) of poliovirus is approximately 750 nucleotides in length and has seven to eight (depending on the serotype) AUG codons upstream of the initiator AUG. The sequence required for internal ribosome binding has been termed the ribosome landing pad (RLP). To better understand the mechanisms of internal initiation, we have determined the boundaries and critical elements of the RLP of poliovirus type 2 (Lansing strain) in vivo. By using deletion analysis, we demonstrate the existence of a core RLP in the poliovirus mRNA 5' UTR whose boundaries are between nucleotides 134 and 155 at the 5' end and nucleotides 556 and 585 at the 3' end. Sequences flanking the core RLP affect translational activity. The importance of several stem-loop structures in the RLP for internal initiation has been determined. Mutation of the phylogenetically conserved loop sequences in the proximal stem-loop structure of the RLP (stem-loop structure III; nucleotides 127 to 165) abolished internal translation. However, deletion of the second stem-loop in the RLP (stem-loop structure IV; nucleotides 189 to 223) reduced internal translation by only 50%. Internal deletions encompassing nucleotides 240 to 300, 350 to 380, or 450 to 480, predicted to disrupt stem-loop structure V and possibly VI, also abrogated internal initiation. Small point mutations within a short polypyrimidine sequence, highly conserved among all picornaviruses, abolished translation. A conservation of distance between the conserved polypyrimidine tract and a downstream AUG could play an important role in the mechanism of internal initiation.

Coupled mutations in the 5'-untranslated region of the Sabin poliovirus strains during in vivo passages: structural and functional implications

All entero- and rhinovirus RNAs sequenced thus far possess A and U residues at positions corresponding to nucleotides 480 and 525, respectively, of poliovirus type 1. These two nucleotides have been proposed previously to form a base pair. The single exception to this rule appears to be the Sabin type 1 strain, which has a G480. Isolates of the Sabin 1 virus from healthy vaccinees were shown to have either a reversion to A480 or a second-site mutation U525----C, both restoring a potential for efficient base pairing. In vitro translation experiments demonstrated that poliovirus type 1 RNAs with either A480-U525 or G480-C525 are more efficient in promoting translation initiation as compared with the Sabin 1 RNA (G480-U525). The Sabin 2 strain has a U and an A at position 398 and 481, respectively, while its predecessor, strain P712, is shown to have C398 and G481. All the derivatives of the Sabin 2 isolated from vaccine-associated paralytic poliomyelitis cases shown reversion to G481, and most of them reverted also to C398. It is proposed that bases at positions 398 and 481 may be involved in a tertiary interaction. The in vitro template activity of the Sabin type 2 RNA (A481) is significantly lower than that of the isolate RNAs with G481, thus confirming the relation between attenuation and translation efficiency demonstrated previously for the type 1 and type 3 Sabin strains. The C----U change at position 398 exerted only a minor effect on the RNA template activity.

Nucleotide sequences important for translation initiation of enterovirus RNA

An infectious cDNA clone was constructed from the genome of coxsackievirus B1 strain. A number of RNA transcripts that have mutations in the 5' noncoding region were synthesized in vitro from the modified cDNA clones and examined for their abilities to act as mRNAs in a cell-free translation system prepared from HeLa S3 cells. RNAs that lack nucleotide sequences at positions 568 to 726 and 565 to 726 were found to be less efficient and inactive mRNAs, respectively. To understand the biological significance of this region of RNA, small deletions and point mutations were introduced in the nucleotide sequence between positions 538 and 601. Except for a nucleotide substitution at 592 (U----C) within the 7-base conserved sequence, mutations introduced in the sequence downstream of position 568 did not affect much, if any, of the ability of RNA to act as mRNA. Except for a point mutation at 558 (C----U), mutations upstream of position 567 appeared to inactivate the mRNA. In the upstream region, a sequence consisting of 21 nucleotides at positions 546 to 566 is perfectly conserved in the 5' noncoding regions of enterovirus and rhinovirus genomes. These results suggest that the 7-base conserved sequence functions to maintain the efficiency of translation initiation and that the nucleotide sequence upstream of position 567, including the 21-base conserved sequence, plays essential roles in translation initiation. A deletion mutant whose genome lacks the nucleotide sequence at positions 568 to 726 showed a small-plaque phenotype and less virulence against suckling mice than the wild-type virus. Thus, reduction of the efficiency of translation initiation may result in the construction of enteroviruses with the lower-virulence phenotype.

Poliovirus translation initiation: differential effects of directed and selected mutations in the 5' noncoding region of viral RNAs

We have analyzed the translational defects of a number of mutations in the 5' noncoding region of poliovirus type 1 RNA. These mutations fall into three categories: (1) two mutations which resulted in temperature sensitive (ts) viruses, (2) the second-site mutations responsible for the reversion of the two ts viruses, and (3) mutations which were lethal to virus production. RNAs containing either of the ts mutations translated in vitro at levels significantly lower than wild-type levels. RNAs containing the respective second-site reversions had corrected these translational defects to levels corresponding to their viral growth potentials. Unlike in vitro translation of wild-type poliovirus RNA, translation of the RNAs which gave rise to ts mutant viruses was not stimulated by the addition of an S10 fraction from an uninfected HeLa cell extract to a rabbit reticulocyte lysate (RRL). In vitro translation of the mutant RNAs (corresponding to the ts viruses) in a RRL was stimulated by factors present in a ribosomal salt wash (RSW) from a HeLa extract, although the levels of stimulation were only half those seen for wild-type. These results suggest that the stimulatory factors present in the RSW have a decreased affinity for the mutant RNA templates but can, to some extent interact, with such RNAs if provided in high enough concentration. The in vitro translation of RNAs containing either of the lethal mutations was not stimulated by factors present in the S10 or the RSW. Taken together, our data suggest a correlation between the ability of a genetically altered RNA to respond to translation stimulatory factors in vitro and the ability of that mutation to be recovered in infectious virus. In addition, we have identified the in vivo-selected reversion of translational defects for two different ts viruses.

Quantitation of relative fitness and great adaptability of clonal populations of RNA viruses

We describe a sensitive, internally controlled method for comparing the genetic adaptability and relative fitness of virus populations in constant or changing host environments. Certain monoclonal antibody-resistant mutants of vesicular stomatitis virus can compete equally during serial passages in mixtures with the parental wild-type clone from which they were derived. These genetically marked "surrogate wild-type" neutral mutants, when mixed with wild-type virus, allow reliable measurement of changes in virus fitness and of virus adaptation to different host environments. Quantitative fitness vector plots demonstrate graphically that even clones of an RNA virus are composed of complex variant populations (quasispecies). Variants of greater fitness (competitive replication ability) were selected within very few passages of virus clones in new host cells or animals. Even clones which were well adapted to BHK21 cells gained further fitness during repeated passages in BHK21 cells.

Identification of a consistent pattern of mutations in neurovirulent variants derived from the sabin vaccine strain of poliovirus type 2

Complete nucleotide sequencing of the RNAs of two unrelated neurovirulent isolates of Sabin-related poliovirus type 2 revealed that two nucleotides and one amino acid (amino acid 143 in the major capsid protein VP1) consistently departed from the sequences of the nonneurovirulent poliovirus type 2 712 and Sabin vaccine strains. This pattern of mutation appeared to be a feature common to all neurovirulent variants of poliovirus type 2.

The 5'-terminal nucleotides of hepatitis A virus RNA, but not poliovirus RNA, are required for infectivity

A series of plasmids containing hepatitis A virus (HAV) cDNA was constructed such that positive-strand HAV RNA could be transcribed with T7 RNA polymerase. The plasmids differed in the number of 5'-terminal nucleotides representing the junctions between vectors and HAV sequences that were present in the transcripts. When these transcripts were used to transfect cultured BS-C-1 cells, it was found that only those transcripts that contained all of the 5'-terminal HAV nucleotides, in addition to one or more nucleotides from the vector, were capable of initiating an infectious cycle leading to production of progeny virus. Transcripts that contained one 5'-terminal nucleotide from the vector sequence but were missing two uridylate residues corresponding to the first two nucleotides of HAV sequences, or were missing U and C residues corresponding to nucleotides 2 and 3 of the HAV sequence, were not infectious. A similar plasmid containing poliovirus cDNA was engineered to produce transcripts similarly lacking the first two uridylate residues of the poliovirus RNA sequence. These transcripts were infectious.

The 5' noncoding region of the type 2 poliovirus vaccine strain contains determinants of attenuation and temperature sensitivity

Intratypic recombinants of P2/Sabin and P2/117, a neurovirulent vaccine revertant, have been generated in vitro using infectious cDNA clones and used to demonstrate that strong determinants of the attenuation and temperature-sensitive phenotypes of P2/Sabin reside in the 5' 492 nucleotides. In this region of the genome the viruses differ only at nucleotides 437 and 481. The ts phenotype associated with the 5' noncoding region is expressed at different temperatures in different cell lines, suggesting an involvement of cellular factors which may be species specific. Suppression of both the ts and attenuation phenotypes correlates with an A-G mutation at nucleotide 481, although other changes are also involved.

An RNA hairpin at the extreme 5' end of the poliovirus RNA genome modulates viral translation in human cells

Several mutations were introduced into an infectious poliovirus cDNA clone by inserting different oligodeoxynucleotide linkers into preexisting DNA restriction endonuclease sites in the viral cDNA. Ten mutated DNAs were constructed whose lesions mapped in the 5' noncoding region or in the capsid coding region of the viral genome. Eight of these mutated cDNAs did not give rise to infectious virus upon transfection into human cells, one yielded virus with a wild-type phenotype, and one gave rise to a viral mutant with a small-plaque phenotype. This last mutant, designated 1-5NC-S21, bears a 6-nucleotide insertion in the loop of a stable RNA hairpin at the very 5' end of the viral genome. Detailed analysis of the biological properties of 1-5NC-S21 showed that the primary defect in mutant-infected cells is a fivefold decrease in translation relative to wild-type-infected cells. Transfection into HeLa cells of in vitro-synthesized RNA molecules bearing either the 5' noncoding region of 1-5NC-S21 or wild-type poliovirus upstream of a luciferase reporter gene showed that the mutated RNA hairpin was responsible for the observed decrease in viral translation in mutant-infected cells and conferred this defect to heterologous RNAs. These findings indicate that an RNA hairpin located at the extreme 5' end of the viral RNA and highly conserved among enteroviruses and rhinoviruses profoundly affects the translation efficiency of poliovirus RNA in infected cells.

Use of type 1/type 2 chimeric polioviruses to study determinants of poliovirus type 1 neurovirulence in a mouse model

We previously described the characteristics of a type 1/type 2 (PV-1/PV-2) chimeric poliovirus, v510, which contains the six amino acids specific for PV-2 in the B-C loop of VP1. This virus was found to be mouse-adapted, as PV-2 and in contrast with PV-1. Determinants of host range were studied in detail and are reported here. PV-1/PV-2 chimeras containing partial PV-1----PV-2 substitutions in the B-C loop of VP1 were obtained by making use of a mutagenesis cartridge on PV-1 cDNA. Analysis of mouse neurovirulence of these chimeras, when correlated with the three-dimensional structure of the v510 capsid, revealed that PV-2 residues important for mouse tropism are those which determine the particular conformation of the B-C loop of VP1 in v510. The mutation of the adenine residue at position 480 of the 5' noncoding region into a guanine residue has been shown to be an important determinant of PV-1 attenuation in monkeys. We show that introduction of this mutation in the v510 genome results in a virus which is partially attenuated for mice. This suggests that analysis of genomic determinants important for PV-1 neurovirulence could be carried out in a mouse model by making use of a mouse-adapted PV-1/PV-2 chimera.

Genetic variability within the 5' nontranslated region of hepatitis A virus RNA. Implications for secondary structure and function

The RNA genome of hepatitis A virus (HAV) contains a lengthy and relatively well conserved 5' nontranslated region (5'NTR). In other picornaviruses, the 5'NTR has been shown to have important functions related to the initiation of viral translation and replication of viral RNA, functions which are critically dependent on both primary and secondary RNA structure. We have utilized a phylogenetic approach to construct a model of the secondary structure of the HAV 5'NTR. By comparing the nucleotide sequences of genetically divergent simian and human HAV strains, we identified a series of covariant nucleotide substitutions which are predictive of conserved, double-stranded helical structures within the 5'NTR, and which thus permitted improved thermodynamic modeling of the secondary structure. The model was further refined based on the observed sites of cleavage of synthetic RNA by single- and double-strand specific RNAses. The results of these studies suggest that the 5'NTR of HAV has a general organization similar to that of other picornaviruses, and shares certain structural features and perhaps specific functions with the 5'NTRs of the cardioviruses and aphthoviruses.

The 5'-untranslated region of picornaviral genomes

Picornaviruses are small naked icosahedral viruses with a single-stranded RNA genome of positive polarity. According to current taxonomy, the family includes four genera: Enterouirus (polioviruses, coxsackieviruses, echoviruses, and other enteroviruses), Rhinovirus, Curdiouirus [encephalomyocarditis virus (EMCV), mengovirus, Theiler's murine encephalomyelitis virus (TMEV)], and Aphthouirus [foot-and-mouth disease viruses (FMDV)]. There are also some, as yet, unclassified picornaviruses [e.g., hepatitis A virus (HAW] that should certainly be assessed as a separate genus. Studies on the molecular biology of picornaviruses might be divided into two periods: those before and after the first sequencing of the poliovirus genome. The 5'-untranslated region (5-UTR) of the viral genome was one of the unexpected problems. This segment proved to be immensely long: about 750 nucleotides or ∼10% of the genome length. There were also other unusual features (e.g., multiple AUG triplets preceding the single open reading frame (ORF) that encodes the viral polyprotein). This chapter shows that the picornaviral 5-UTRs are not only involved in such essential events as the synthesis of viral proteins and RNAs that could be expected to some extent, although some of the underlying mechanisms appeared to be quite a surprise, but also may determine diverse biological phenotypes from the plaque size or thermosensitivity of reproduction to attenuation of neurovirulence. Furthermore, a close inspection of the 5-UTR structure unravels certain hidden facets of the evolution of the picornaviral genome. Finally, the conclusions drawn from the experiments with the picornaviral5-UTRs provide important clues for understanding the functional capabilities of the eukaryotic ribosomes.

The novel mechanism of initiation of picornavirus RNA translation

The initiation of translation of picornaviral RNAs takes place by an unusual mechanism whereby ribosomes bind directly to an internal site rather than scan the RNA from the 5'-end. This internal entry mechanism requires a 450-nucleotide segment of the picornavirus 5'-untranslated region. The ribosome binds initially to a site at the 3'-end of this segment, and then may scan the RNA to reach the authentic initiation site. This novel mechanism may be of relevance to the translation of some cellular mRNAs.

Mapping of mutations associated with neurovirulence in monkeys infected with Sabin 1 poliovirus revertants selected at high temperature

Poliovirus type 1 neurovirulence is difficult to analyze because of the 56 mutations which differentiate the neurovirulent Mahoney strain from the attenuated Sabin strain. We have isolated four neurovirulent mutants which differ from the temperature-sensitive parental Sabin 1 strain by only a few mutations, using selection for temperature resistance: mutant S(1)37C1 was isolated at 37.5 degrees C, S(1)38C5 was isolated at 38.5 degrees C, and S(1)39C6 and S(1)39C10 were isolated at 39.5 degrees C. All four mutants had a positive reproductive capacity at supraoptimal temperature (Rct+ phenotype). Mutant S(1)37C1 induced paralysis in two of four cynomolgus monkeys, and the three other mutants induced paralysis in four of four monkeys. The lesion score increased from the S(1)37C1 mutant to the S(1)39 mutants. To map the mutations associated with thermoresistance and neurovirulence, we sequenced all regions in which the Sabin 1 genome differs from the Mahoney genome. The S(1)37C1 mutant had one mutation in the 5' noncoding region and another in the 3' noncoding region. Mutant S(1)38C5 had these mutations plus another mutation in the 3D polymerase gene. The S(1)39 mutants had three additional mutations in the capsid protein region. The mutations were located at positions at which the Sabin 1 and Mahoney genomes differ, except for the mutation in the 5' noncoding region. The noncoding-region mutations apparently confer a low degree of neurovirulence. The 3D polymerase mutation, which distinguishes S(1)38C5 and S(1)39 mutants from S(1)37C1, is probably responsible for the high neurovirulence of S(1)38C5 and S(1)39 mutants. The capsid region mutations may contribute to the neurovirulence of the S(1)39 mutants, which was the highest among the mutants.

Functional analysis of the internal translation initiation site of foot-and-mouth disease virus

Mutagenesis of the large untranslated sequence at the 5' end of the genome of foot-and-mouth disease virus revealed that a region of approximately 450 nucleotides preceding the open reading frame of the viral polyprotein is involved in the regulation of translation initiation at two internal start sites. Variations in two domains of this region reduced the translation efficiency up to 10-fold, whereas an intermediate segment seemed to be less essential. A pyrimidine-rich sequence preceding the start codon was most sensitive in that conversion of single pyrimidine residues to purines decreased the translation efficiency strongly. The data are in agreement with a recently proposed general structural model for the internal ribosome entry site of the cardiovirusaphthovirus subgroup of picornaviruses (E. V. Pilipenko, V. M. Blinov, B. K. Chernov, T. M. Dmitrieva, and V. I. Agol, Nucleic Acids Res. 17:5701-5711, 1989). They suggest, however, that this model represents only a core structure for the internal entry of ribosomes and that foot-and-mouth disease virus and other members of the picornaviruses need additional regulatory RNA elements for efficient translation initiation.

Cytoplasmic expression system based on constitutive synthesis of bacteriophage T7 RNA polymerase in mammalian cells

A mouse cell line that constitutively synthesizes the bacteriophage T7 RNA polymerase was constructed. Fluorescence microscopy indicated that the T7 RNA polymerase was present in the cytoplasmic compartment. The system provided, therefore, a unique opportunity to study structural elements of mRNA that affect stability and translation. The in vivo activity of the bacteriophage polymerase was demonstrated by transfection of a plasmid containing the chloramphenicol acetyltransferase (CAT) gene flanked by T7 promoter and termination signals. Synthesis of CAT was dependent on the presence of a cDNA copy of the untranslated region of encephalomyocarditis virus (ECMV) RNA downstream of the T7 promoter, consistent with the absence of RNA-capping activity in the cytoplasm. CAT expression from a plasmid, pT7EMCAT, containing the T7 and EMCV regulatory elements was detected within 4 hr after transfection and increased during the next 20 hr, exceeding that obtained by transfection of a plasmid with the CAT gene attached to a retrovirus promoter and enhancer. Nevertheless, the presumably cap-independent transient expression of CAT from pT7EMCAT was increased more than 500-fold when the transfected cells also were infected with wild-type vaccinia virus. A protocol for high-level expression involved the infection of the T7 RNA polymerase cell line with a single recombinant vaccinia virus containing the target gene regulated by a T7 promoter and EMCV untranslated region.

Cap-independent translation of encephalomyocarditis virus RNA: structural elements of the internal ribosomal entry site and involvement of a cellular 57-kD RNA-binding protein

Translation of encephalomyocarditis virus (EMCV) mRNA occurs by ribosomal internal entry into the 5'-nontranslated region (5' NTR) rather than by ribosomal scanning. The internal ribosomal entry site (IRES) in the EMCV 5' NTR was determined by in vitro translation with RNAs that were generated by in vitro transcription of EMCV cDNAs containing serial deletions from either the 5' or 3' end of the EMCV 5' NTR. Regions downstream of nucleotide 403 and upstream of nucleotide 811 of EMCV were required for efficient translation. Site-directed mutagenesis revealed that a stem-loop structure (400 nucleotides upstream of the initiation codon) was essential for IRES function. We discovered a 57-kD cellular protein whose specific interaction with this stem-loop appears to be prerequisite for IRES function. A A pyrimidine-rich stretch proximal to the initiation codon was also crucial for efficient translation of EMCV mRNA. We propose that ribosomes bind directly to the initiating AUG without scanning.

An initiation signal in the 5' untranslated leader sequence of encephalomyocarditis virus RNA

A 593 nucleotide fragment of the 5' leader of encephalomyocarditis virus RNA (EMCV-RNA) was linked to the SP6 promoter and inserted upstream of the reporter gene chloramphenicol acetyltransferase (CAT). The presence of the 5'-UTR of EMCV-RNA in the RNA transcripts, made in vitro with the SP6 polymerase, resulted in a strong translational enhancement when tested in the micrococcal nuclease-treated reticulocyte lysate. The transcripts were equally active with or without a 5' methylated capstructure as expected, since EMCV-RNA is one of the mRNAs capable of internal initiation. We searched for a signal in the 5' leader that allows the 43S preinitiation complex to bind internally and localized a hairpin containing a unique nucleotide sequence, CUUUA, present in a domain conserved among cardio- and aphtoviral RNAs. Replacing this sequence into AGCU resulted in a 50% loss of translational activity. A second mutation involving a U-G change in the stem of that hairpin resulted in an almost complete loss of initiation.

Common and specific sequences in picornaviruses

Enteroviruses and rhinoviruses were studied by nucleic acid hybridization using oligonucleotide probes. The reactivity of these sequences varied from serotype specificity to wide reactivity with nearly all human picornaviruses tested. Widely reactive sequences were from the 5' non-coding region while specific oligonucleotides were found from the coding region. Sequences that were common to both enteroviruses and human rhinoviruses were identified and one of the oligonucleotides reacted with a majority enterovirus serotypes.

The complete nucleotide sequence of a pathogenic swine vesicular disease virus

The nucleotide sequence of a swine vesicular disease virus (SVDV) strain that is pathogenic for pigs has been determined and compared with that of a non-pathogenic strain of SVDV, as well as a number of other enteroviruses. It shows only 98 base changes in comparison with a non-pathogenic strain of SVDV (Inoue et al., 1989, J. Gen. Virol. 70, 919-934). Fourteen of these nucleotide differences between the pathogenic and the non-pathogenic SVDV strains occur in the 5' non-coding region which, by analogy with the other picornaviruses, has been implicated in the efficiency with which the RNA is employed as mRNA. Additional differences found throughout the coding regions are largely conservative in nature. A number of residues are discussed as candidates for determinants of pathogenicity. This sequence has been submitted to the PIR database and has accession number A30061.

Gross rearrangements within the 5'-untranslated region of the picornaviral genomes

An analysis of reported nucleotide sequences revealed several cases of gross rearrangements in the 5'-untranslated region (5-UTR) of picornaviral genomes. A large (greater than 100 nt) duplication was discovered in a downstream region of poliovirus 5-UTR involved in the translational control. Properties of the poliovirus mutants with large deletions [Kuge and Nomoto (1987) J. Virol. 61, 1478-1487] show that a single copy of the appropriate repeating unit is compatible with a wild type phenotype of the virus. In contrast to poliovirus and another enterovirus genomes, human rhinovirus RNAs contain only a single copy of this repeating unit. Another similarly large repeat was found in an upstream segment of the bovine enterovirus 5-UTR. A comparison of the primary and secondary structures of cardio- and aphthovirus 5-UTRs demonstrated the existence of a large (ca. 250 nucleotides) insertion/deletion in a region preceding the poly(C) tract. The two latter rearrangements appear to involve elements of the viral genome replication machinery. Possible origin as well as evolutionary and functional implications of these structural peculiarities are discussed.

Translation deficiency of the Sabin type 3 poliovirus genome: association with an attenuating mutation C472----U

Previous studies have shown that the genome of Sabin type 3 poliovaccine strain (P3/Leon 12a1b) possesses a diminished translation efficiency as compared to genomes of closely related neurovirulent strains, the neurovirulent progenitor (P3/Leon/37), or a revertant (P3/119/70) of the vaccine (Y.V. Svitkin, S.V. Maslova, and V.I. Agol, 1985, Virology 147, 243-252). Here we attempted to evaluate the contribution of each mutation in the genome of the vaccine to this translation deficiency. Recombinants between P3/Leon 12a1b and P3/Leon/37 or P3/119/70 were constructed in vitro and their RNAs were translated in a cell-free system derived from Krebs-2 cells. The results show that of 10 nucleotide differences between the genomes of P3/Leon 12a1b and P3/Leon/37 9 have minor or no effect on translation and that the only mutation of significance is C472----U which is known to reduce the neurovirulence of the virus. Reversion from uridine to cytosine at position 472 in type 3 poliovaccine upon replication in the human gut resulted in an increase of both translation efficiency of polio RNAs and neurovirulence of corresponding strains. The data provide evidence for a common nucleotide sequence regulatory element for protein synthesis of the virus and its neurovirulence. In vitro translation assays may therefore prove to be useful for detection of attenuating mutations in the 5' noncoding region of poliovirus genome. The apparent involvement of the translation mechanism in the expression of neurovirulent or attenuated phenotype of poliovirus is briefly discussed.

Attenuation of Mengo virus through genetic engineering of the 5' noncoding poly(C) tract

The murine cardioviruses, such as the Mengo and encephalomyocarditis viruses, and the bovine aphthoviruses, such as foot-and-mouth disease virus, are distinguished among positive-strand RNA viruses by the presence of long homopolymeric poly(C) tracts within their 5' noncoding sequences. Although the specific lengths (60-350 bases) and sequence discontinuities (for example, uridine residues) that sometimes disrupt the homopolymer have served to characterize natural viral isolates, the biological function of the poly(C) region has never been clear. We now report that complementary DNA-mediated truncation of the Mengo virus poly(C) tract dramatically attenuates the pathogenicity of the virus in mice. Animals injected with viruses with short tracts not only survived inoculation of up to 50 micrograms live virus (10(11) plaque-forming units) but consistently produced high titres of neutralizing antibodies, which conferred long-term immunogenic protection from (normally) lethal virus challenge. We propose that analogous synthetic strains of foot and mouth disease virus could serve as the basis for new attenuated vaccines.

Construction of less neurovirulent polioviruses by introducing deletions into the 5' noncoding sequence of the genome

Viral attenuation may be due to lowered efficiency of certain steps essential for viral multiplication. For the construction of less neurovirulent strains of poliovirus in vitro, we introduced deletions into the 5' noncoding sequence (742 nucleotides long) of the genomes of the Mahoney and Sabin 1 strains of poliovirus type 1 by using infectious cDNA clones of the virus strains. Plaque sizes shown by deletion mutants were used as a marker for rate of viral proliferation. Deletion mutants of both the strains thus constructed lacked a genome region of nucleotide positions 564 to 726. The sizes of plaques displayed by these deletion mutants were smaller than those by the respective parental viruses, although a phenotype referring to reproductive capacity at different temperatures (rct) of viruses was not affected by introduction of the deletion. Monkey neurovirulence tests were performed on the deletion mutants. The results clearly indicated that the deletion mutants had much less neurovirulence than with the corresponding parent viruses. Production of infectious particles and virus-specific protein synthesis in cells infected with the deletion mutants started later than in those infected with the parental viruses. The rate at which cytopathic effect progressed was also slower in cells infected with the mutants. Phenotypic stability of the deletion mutant for small-plaque phenotype and temperature sensitivity was investigated after passaging the mutant at an elevated temperature of 37.5 degrees C. Our data strongly suggested that the less neurovirulent phenotype introduced by the deletion is very stable during passaging of the virus.

Cell proteins bind to multiple sites within the 5' untranslated region of poliovirus RNA

The 5' noncoding region of poliovirus RNA contains sequences necessary for translation and replication. These functions are probably carried out by recognition of poliovirus RNA by cellular and/or viral proteins. Using a mobility-shift electrophoresis assay and 1,10-phenanthroline/Cu+ footprinting, we demonstrate specific binding of cytoplasmic factors with a sequence from nucleotides 510-629 within the 5' untranslated region (UTR). Complex formation was also observed with a second sequence (nucleotides 97-182) within the 5' UTR. These two regions of the 5' UTR appear to be recognized by distinct cell factors as determined by competition analysis and the effects of ionic strength on complex formation. However, both complexes contain eukaryotic initiation factor 2 alpha, as revealed by their reaction with specific antibody.

Nucleotide sequence of a neurovirulent variant of the type 2 oral poliovirus vaccine

Infectious cDNAs of the Sabin type 2 poliovirus vaccine virus and a vaccine-derived neurovirulent type 2 strain (P2/117) have been cloned in Escherichia coli. Nucleotide sequence analysis revealed that P2/117 differs from the vaccine strain by just 23 point mutations. Three occur in the 5' noncoding region. The remainder result in a total of 5 coding changes located in VP1, VP4, 2B, and 3D. The likely role of these mutations in the evolution to neurovirulence is discussed.

Poliovirus translation: a paradigm for a novel initiation mechanism

All eukaryotic cellular mRNAs, and most viral mRNAs, are blocked at their 5' ends with a cap structure (m7GpppX, where X is any nucleotide). Poliovirus, along with a small number of other animal and plant viral mRNAs, does not contain a 5' cap structure. Since the cap structure functions to facilitate ribosome binding to mRNA, translation of polio-virus must proceed by a cap-independent mechanism. Consistent with this, recent studies have shown that ribosomes can bind to an internal region within the long 5' noncoding sequence of poliovirus RNA. Possible mechanisms for cap-independent translation are discussed. Cap-independent translation of poliovirus RNA is of major importance to the mechanism of shut-off of host protein synthesis after infection. Moreover, it is likely to play a role in determining poliovirus neurovirulence and attenuation.

Distinct modes of poliovirus polyprotein initiation in vitro

RNAs of poliovirus type 1 and type 3 were translated in extracts from Krebs-2 cells after annealing with oligodeoxyribonucleotides complementary to different sites in the 5'-untranslated region (5'-UTR). Due to a high level of endogenous RNase H activity in the extracts, such RNAs appeared to be efficiently 5'-truncated prior to translation. The observed levels of initiation on differently truncated templates suggested that a region in the middle of the poliovirus 5'-UTR is essential for the cap-independent initiation of viral polyprotein synthesis. The data reported here, in conjunction with the results from other laboratories, permitted to relate the essential cis-acting control elements to the 5'-UTR secondary structure domains defined previously (E.V. Pilipenko et al., Virology 168, 201-209). However, the removal of these domains activated another mode of polyprotein initiation, which appeared to require another set of translation initiation factors.

Conservation of the secondary structure elements of the 5'-untranslated region of cardio- and aphthovirus RNAs

An analysis of published nucleotide sequences of the 5'-untranslated region (5'-UTR) of 7 cardioviruses and 3 aphthoviruses has allowed us to derive a consensus secondary structure model that differs from that previously proposed for the 5'-UTR of entero- and rhinoviruses, though all these viruses belong to the same family, Picornaviridae. The theoretical model derived here was experimentally supported by investigating the accessibility of encephalomyocarditis virus RNA to modifications with dimethyl sulfate and its susceptibility to S1 and cobra venom nucleases. The possible involvement of the 5"-UTR secondary structure domains in the translational control is briefly discussed.

Initiation of protein synthesis by internal entry of ribosomes into the 5' nontranslated region of encephalomyocarditis virus RNA in vivo

Expression vectors that yield mono-, di-, and tricistronic mRNAs upon transfection of COS-1 cells were used to assess the influence of the 5' nontranslated regions (5'NTRs) on translation of reporter genes. A segment of the 5'NTR of encephalomyocarditis virus (EMCV) allowed translation of an adjacent downstream reporter gene (CAT) regardless of its position in the mRNAs. A deletion in the EMCV 5'NTR abolishes this effect. Poliovirus infection completely inhibits translation of the first cistron of a dicistronic mRNA that is preceded by the capped globin 5'NTR, whereas the second cistron preceded by the EMCV 5'NTR is still translated. We conclude that the EMCV 5'NTR contains an internal ribosomal entry site that allows cap-independent initiation of translation. mRNA containing the adenovirus tripartite leader is also resistant to inhibition of translation by poliovirus.