David Rowlands collection

Probing the molecular interactions within the foot-and-mouth disease virus (FMDV) RNA replication complex has been restricted in part by the lack of suitable reagents. Random insertional mutagenesis has proven an excellent method to reveal domains of proteins essential for virus replication as well as locations that can tolerate small genetic insertions. Such insertion sites can subsequently be adapted by the incorporation of commonly used epitope tags, facilitating their detection with commercially available reagents. In this study, we used random transposon-mediated mutagenesis to produce a library of 15 nt insertions in the FMDV non-structural polyprotein. Using a replicon-based assay, we isolated multiple replication-competent as well as replication-defective insertions. We adapted the replication-competent insertion sites for the successful incorporation of epitope tags within FMDV non-structural proteins for use in a variety of downstream assays. Additionally, we showed that replication of some of the replication-defective insertion mutants could be rescued by co-transfection of a ‘helper’ replicon, demonstrating a novel use of random mutagenesis to identify intergenomic trans-complementation. Both the epitope tags and replication-defective insertions identified here will be valuable tools for probing interactions within picornavirus replication complexes.

Infectious hepatitis C virus (HCV) particle production in the genotype 2a JFH-1-based cell culture system involves non-structural proteins in addition to canonical virion components. NS2 has been proposed to act as a protein adaptor, co-ordinating the early stages of virion assembly. However, other studies have identified late-acting roles for this protein, making its precise involvement in infectious particle production unclear. Using a robust, bipartite trans-encapsidation system based upon baculovirus expression of HCV structural proteins, we have generated HCV-like particles (HCV-LP) in the absence of NS2 with overt similarity to wild-type virions. HCV-LP could transduce naive cells with trans-encapsidated subgenomic replicon RNAs and shared similar biochemical and biophysical properties with JFH-1 HCV. Both genotype 1b and JFH-1 intracellular HCV-LP were produced in the absence of NS2, whereas restoring NS2 to the JFH-1 system dramatically enhanced secreted infectivity, consistent with a late-acting role. Our system recapitulated authentic HCV particle assembly via trans-complementation of bicistronic, NS2-deleted, chimeric HCV, which is otherwise deficient in particle production. This closely resembled replicon-mediated NS2 trans-complementation, confirming that baculovirus expression of HCV proteins did not unduly affect particle production. Furthermore, this suggests that separation of structural protein expression from replicating HCV RNAs that are destined to be packaged alleviates an early stage requirement for NS2 during particle formation. This highlights our current lack of understanding of how NS2 mediates assembly, yet comparison of full-length and bipartite systems may provide further insight into this process.

A characteristic of many positive-strand RNA viruses is that, whilst replication of the viral genome is dependent on the expression of the majority of non-structural proteins in cis, virus particle formation can occur when most or all of the structural proteins are co-expressed in trans. Making use of a recently identified hepatitis C virus (HCV) isolate (JFH1) that can be propagated in tissue culture, this study sought to establish whether this is also the case for hepaciviruses. Stable cell lines containing one of two bicistronic replicons derived from the JFH1 isolate were generated that expressed non-structural proteins NS3–5B or NS2–5B. Release and transmission of these replicons to naïve Huh7 cells could then be demonstrated when baculovirus transduction was used to express the HCV proteins absent from the subgenomic replicons. Transmission could be blocked by a neutralizing antibody targeted at the E2 envelope protein, consistent with this phenomenon occurring via trans-encapsidation of replicon RNA into virus-like particles. Transmission was also dependent on expression of NS2, which was most effective at promoting virus particle formation when expressed in cis on the replicon RNA compared with in trans via baculovirus delivery. Density gradient analysis of the particles revealed the presence of a broad infectious peak between 1.06 and 1.11 g ml−1, comparable to that seen when propagating full-length virus in tissue culture. In summary, the trans-encapsidation system described offers a complementary and safer approach to study HCV particle formation and transmission in tissue culture.

Hyperphosphorylation of NS5A is thought to play a key role in controlling hepatitis C virus (HCV) RNA replication. Using a tetracycline-regulable baculovirus delivery system to introduce non-culture-adapted HCV replicons into HepG2 cells, we found that a point mutation in the active site of the viral polymerase, NS5B, led to an increase in NS5A hyperphosphorylation. Although replicon transcripts lacking elements downstream of NS5A also had altered NS5A hyperphosphorylation, this did not explain the changes resulting from polymerase inactivation. Instead, two additional findings may be related to the link between polymerase activity and NS5A hyperphosphorylation. Firstly, we found that disabling polymerase activity, either by targeted mutation of the polymerase active site or by use of a synthetic inhibitor, stimulated translation from the replicon transcript. Secondly, when the rate of translation of non-structural proteins from replicon transcripts was reduced by use of a defective encephalomyocarditis virus internal ribosome entry site, there was a substantial decrease in NS5A hyperphosphorylation, but this was not observed when non-structural protein expression was reduced by simply lowering replicon transcript levels using tetracycline. Therefore, one possibility is that the point mutation within the active site of NS5B causes an increase in NS5A hyperphosphorylation because of an increase in translation from each viral transcript. These findings represent the first demonstration that NS5A hyperphosphorylation can be modulated without use of kinase inhibitors or mutations within non-structural proteins and, as such, provide an insight into a possible means by which HCV replication is controlled during a natural infection.

We have developed a baculovirus delivery system that enables tetracycline-regulated expression of polII-derived hepatitis C virus (HCV) transcripts in hepatocyte-derived cell lines ( McCormick et al., 2002 ). As part of a study to determine whether such transcripts are replication competent, the transcription start site of the tetracycline-regulable promoter was mapped and three baculovirus transfer vectors containing a neo R-expressing culture adapted replicon cDNA were generated. These vectors either had the first nucleotide of the 5′UTR positioned −2 (mkI) and +1 (mkII) with respect to the transcription start site, or included a hammerhead ribozyme at the 5′ end of the transcript (5′HH) that cleaves between the ribozyme–5′UTR boundary. Transfection of all of the culture-adapted replicon constructs into Huh7 cells resulted in the formation of more neomycin-resistant colonies than seen with a polymerase knock-out replicon construct, although this was less pronounced in the mkI group. Furthermore, both the positive- and negative-strands of the replicon could be detected in all neomycin-resistant polyclonal cell lines except for those derived from transfection of the polymerase knock-out construct. Transduction of Huh7 cells with recombinant baculoviruses carrying the same expression cassettes improved replicon delivery, but the relative efficiency of the constructs remained the same. The baculovirus vectors were also used to introduce the replicon transcript into HepG2 cells. Expression of the culture-adapted but not the polymerase knock-out construct induced transcription of the β-interferon gene, a response that may contribute to this cell line being unable to maintain the replicon over long-term culture.

Baculovirus vectors have been used as efficient delivery vehicles for constitutive gene expression in a variety of mammalian cells. We have further developed the system to allow for regulable expression by placing the gene of interest under the control of an inducible promoter, and complementing it with a second baculovirus vector providing the control elements necessary for promoter activity. We have used this system to express (a) the lacZ gene, (b) a ‘minigenome’ derived from hepatitis C virus (HCV) and carrying lacZ or (c) the full-length HCV viral genome, in human hepatocyte cell lines in an inducible fashion. Control systems that rely on either the absence of tetracycline or presence of ponasterone to induce gene expression were tested. Expression of lacZ was controlled by ponasterone, but β-galactosidase activity was limited to 10–20% of cells. In contrast, the tetracycline-controlled expression system gave a low basal activity and was highly inducible in almost 100% of cells. Inducible expression was also obtained in almost 100% of cells infected with baculoviruses in which an HCV minigenome was placed downstream of the tetracycline-inducible promoter and upstream of either a hammerhead or hepatitis δ virus ribozyme. Northern blot analysis was consistent with accurate cleavage of the minigenome transcript by the hepatitis δ virus ribozyme. Finally, regulable transcript production and viral polypeptide processing could be demonstrated in HepG2 cells infected with baculoviruses bearing the full-length HCV genome. This system thus provides a novel tool for the analysis of HCV replication and host–cell interactions.

The non-structural protein NS3 of hepatitis C virus has been expressed in bacteria as a polyhistidine fusion protein which can be produced in a soluble form and easily purified by affinity chromatography. Using an in vitro transcription and translation system we have been able to demonstrate that this protein can proteolytically process substrate molecules derived from the non-structural region of the polyprotein. Using this assay system we have been able to optimize basic biochemical characteristics of the purified enzyme. Parallel experiments show that the full-length NS3 protein also possesses ATPase activity, indicating the bifunctional nature of the protein. In contrast, purified NS3 in which the predicted catalytic serine has been mutated loses protease but retains ATPase activity.

The protease activity of the hepatitis C virus (HCV) NS3 protein has been investigated using transient expression methods in mammalian cells, as well as in vitro transcription/translation systems. We confirmed that expression of the NS3−5 polyprotein in rabbit reticulocyte lysates results in efficient cis processing at the NS3/NS4 junction. However, processing at the other predicted sites of NS3-mediated cleavage varied markedly in efficiency, the site most susceptible being that between NS5A and NS5B. Time-course analysis of the proteolytic processing of the HCV non-structural precursor showed that the cis cleavage between NS3 and NS4 occurred extremely rapidly. However, efficient cleavage at this position was dependent on the prior removal of the NS2 protein. Furthermore, the presence of uncleaved NS2 sequences on the enzyme severely impeded NS3-mediated proteolysis at downstream sites in the polyprotein. This suggests therefore that efficient cleavage at the NS2/NS3 junction is a pivotal event in HCV replication. During the course of this study a proteolytically inactive mutant of NS3 was characterized carrying a previously unreported amino acid substitution near the proposed active site of the enzyme. Molecular modelling suggested that the amino acid present at this position may influence the conformation of the active site of the enzyme. Recently a number of reports have described a second protease activity, located in the NS2/NS3 region, which is responsible for cleavage at the NS2/NS3 junction. We have identified an isolate of HCV, obtained from a U.K. patient, which has a virtually inactive NS2/NS3 protease. The possible implications of this observation are discussed.

A total of 38 neutralization escape mutant viruses have been selected from a cloned stock of human rhinovirus serotype 2 (HRV-2), using either of two monoclonal antibodies (MAbs) which recognize overlapping epitopes as judged by competition binding. The mutant viruses were analysed for their sensitivity to a panel of antiviral MAbs by antibody binding and virus neutralization assays. The position and nature of the selected mutations was determined by sequencing of the virus RNAs, and the location of the substituted amino acids on the three-dimensional structure of the virus predicted from the co-ordinates determined for the closely related HRV-1A. Escape from neutralization could be attributed to single amino acid substitutions in all but one case, which had a deletion of four amino acids. In all cases in which the same mutation was found more than once, these mutations were transitions. The ratio of transition to transversion mutations was about 5:1 overall or about 1·7:1 if only unique substitutions are considered. Each antibody selected for a discrete cluster of mutations and the area of these clusters was considerably less than that determined to be in contact with antibodies from X-ray crystallographic analyses of antibody/protein complexes. One mutation did not occur within the cluster of others selected with the same antibody. This substitution occurred at the base of a small loop and may cause conformational changes at the virus surface.

Guinea-pigs were challenged with homologous or heterologous strains of foot-and-mouth disease virus (FMDV) following vaccination with baby hamster kidney (BHK) monolayer cell-adapted or BHK suspension cell-adapted strains of FMDV serotype A22 Iraq 24/64. The protection afforded by these vaccines was analysed as a function of antigen dose and the in vitro serum virus neutralization titres achieved. The results show that the level of neutralizing antibody induced that afforded 50% protection was similar for both vaccines in homologous or heterologous challenge situations. However, although the dose of antigen required to achieve this titre against homologous virus was similar for the two vaccines, approximately 20-fold more of the suspension cell-adapted virus was required to elicit a protective titre against heterologous challenge compared to the dose of monolayer cell-adapted virus required. A synthetic peptide representing the amino acid sequence 135 to 167 of VP1, which is identical in the A22 Iraq 24/64 variant viruses, was shown to induce protection against both homologous and heterologous virus challenge.

We have investigated the interactions of purified radiolabelled hepatitis A virus (HAV) with a variety of continuous cell lines. Virus labelled either in vitro with radiolabelled iodine or in vivo with radiolabelled uridine bound to cells with similar efficiency. Attachment to BS-C-1 cells was calcium ion-dependent and this correlated with infectivity assay results. The cell tropism of HAV attachment was examined using cell suspensions and confluent cell monolayers at both 4 °C and 37 °C. The maximum level of attachment was observed at 4 °C with cells in suspension, but was severely inhibited by 2% foetal calf serum; these results again correlated with infectivity assays. The components of serum which inhibit attachment have been characterized by gel filtration chromatography, sucrose density gradient analysis, immunoprecipitation and Western blotting. The data show that such components are of high M r and that the serum glycoprotein, α2-macroglobulin, can partly mimic the inhibitory effect of whole serum.

We recently reported the enhanced immunogenicity of a peptide epitope when it was presented as a fusion protein with hepatitis B core antigen. In those experiments the fusion protein was expressed in vaccinia virus. We have now refined the system so that large amounts of highly immunogenic particles can be produced using a simple bacterial expression system. We describe the expression of three different viral epitopes as chimeric particles that induce good antibody responses to each epitope after one dose of low amounts of antigen. Finally we demonstrate that the immunogenicity is a reflection of both T helper cell sites within the core protein and also the particulate nature of the immunogens.

Eleven neutralizing monoclonal antibodies (MAbs) were produced to the O1BFS 1860/67 strain of foot-and-mouth disease virus (FMDV), and were characterized for their ability to bind viral and subviral antigens in different ELISA tests and to neutralize heterologous type O isolates. Neutralization escape variants of the homologous virus, isolated under pressure from five of these MAbs, were used in cross-neutralization tests with all of the 11 antibodies. These studies identified three functionally independent, conformational, neutralizing sites. The most conformationally dependent site bound antibody which neutralized a range of type O virus isolates. A second site was less dependent on conformation and was recognized by antibody that was strain-specific. The least conformational site bound MAbs which showed limited cross-neutralization of other type O strains. This latter site appeared to be immunodominant and contained several overlapping epitopes which showed some differences in their specificities. Isoelectrofocusing and sequencing studies of the variants strongly suggested that polypeptide VP2 contributes to the immunodominant site.

The amino acid sequence RGD (arginine-glycine-aspartic acid) is highly conserved in the VP1 protein of foot-and-mouth disease virus (FMDV), despite being situated in the immunodominant hypervariable region between amino acids 135 and 160. RGD-containing proteins are known to be important in promoting cell attachment in several different systems, and we report here that synthetic peptides containing this sequence are able to inhibit attachment of the virus to baby hamster kidney (BHK) cells. Inhibition was dose-dependent and could be reversed on removal of the peptide. A synthetic peptide corresponding to a portion of the same hypervariable region but not containing the RGD sequence did not inhibit virus attachment under the same conditions. Antibody against the RGD region of VP1 blocked attachment of the virus to BHK cells, and neutralizing monoclonal antibodies, which neutralize virus by preventing cell attachment, were blocked by RGD-containing peptides from binding virus in an ELISA test. Cleavage of the C-terminal region of virus VP1 in situ with proteolytic enzymes reduced cell attachment, and antiserum against a peptide corresponding to this region was also able to inhibit attachment of virus to BHK cells. These results indicate that the amino acid sequence RGD at positions 145 to 147 and amino acids from the C-terminal region of VP1 (positions 203 to 213) contribute to the cell attachment site on FMDV for BHK cells.

Foot-and-mouth disease virus (FMDV) A22 Iraq 24/64 adapted to grow in BHK monolayer cells induced antibodies which neutralized many isolates belonging to the A serotype. Plaque-purified virus isolated from this stock also induced broadly reactive antibodies, showing that this property is not due to the combined response to a mixture of variants in the original stock virus. However, viruses obtained by passage in suspension BHK cells of either the monolayer cell-adapted virus or a virus cloned from this stock resulted in the selection of virus which induced antibodies with highly specific neutralizing activity. In addition to their antigenic properties the monolayer and suspension cell-adapted viruses could be distinguished by plaque morphology, tendency to aggregate and ability to attach to BHK cells. Monoclonal antibodies (MAbs) induced with the plaque-purified monolayer-adapted virus had neutralizing activity almost as broad as polyclonal serum, showing that this property can be represented by a single epitope on the virus. These neutralizing MAbs recognize a trypsin-sensitive epitope on the virus. Surprisingly, sequence analysis of the structural protein-coding regions of the genomic RNAs of monolayer and suspension cell-adapted viruses showed no amino acid differences in VP1, the protein known to contain the major neutralization epitope in FMDV and to be the only protein susceptible to cleavage by trypsin in the virus particle. Although three coding differences were found in the capsid protein these were all located in VP2.

Translation of foot-and-mouth disease virus RNA for extended periods in rabbit reticulocyte lysates results in the appearance of a previously undescribed protein. A protein with similar properties can also be detected in BHK cells at late times after virus infection. Specific immunoprecipitation has shown that this protein (Lb′) is closely related to the smaller of the two leader proteins, Lb, although it migrates with an apparently higher M r in SDS-polyacrylamide gels. The conversion of Lb to Lb′ can be mimicked by treatment with carboxypeptidase B. It is suggested that C-terminal trimming of Lb to produce Lb′ results in an increase in negative charge and is responsible for its slower migration in SDS-PAGE.

The surface structure of foot-and-mouth disease virus (FMDV) and the interaction of the individual capsid proteins with the virus RNA have been examined using modification reagents. By measuring the extent of modification of the lysine residues of intact and disrupted virus particles and the 12S protein subunit with Bolton & Hunter reagent it was found that 54 % of the residues of VP 1, 15 % of the residues of VP2 and 37 % of the residues of VP3, equivalent to five, two and four lysine residues respectively, are on the surface of the intact virus particle. Polypeptide VP4 was not modified in intact virus particles, indicating that it has no lysine residues on the surface of the virus. Modification with sodium metabisulphite, which causes a specific transamination reaction between cytidylic acid residues in ssRNA and closely associated basic amino acids, cross-linked all four structural proteins to the virus RNA. Both fragments of VP1, produced by treatment of the virus particle with trypsin, are also cross-linked to the RNA. These observations have been combined with the evidence that the immunogenic activity of VP1 may be contained in two discontinuous sites, at amino acids 141 to 160 and 200 to 213, in proposing a model for the arrangement of this polypeptide in the virus particle.

Typically, the translation of eukaryotic mRNAs into protein is initiated at a single site. However, we have recently shown that not one but two primary products, P20a and P16, are translated from the 5′ end of the coding region of the genome of foot-and-mouth disease virus (FMDV). In this paper we show by partial protease digestion of these proteins that they differ only at their N termini, thus confirming the presence of two initiation sites for translation of FMDV RNA. Sequence analysis of two subtypes of the virus (A10 and A12) confirms the presence of two initiator AUG codons in the expected position on the genome. By correlation with protein synthesis data from these subtypes it appears that the relative use of each initiation site is dependent on its surrounding nucleotide sequence. In addition, the ratio of the two proteins when synthesized in vitro differs markedly from that when they are synthesized in vivo, suggesting the presence of a control mechanism for synthesis of P20a in vivo which may be absent in vitro. We also show that the cleavage site between these two proteins and the structural protein precursor, P88, is located closer to the N terminus of the polyprotein than has previously been reported.

Antibody produced against preparations of VP1, one of the four structural polypeptides of foot-and-mouth disease virus, neutralized the virus and reacted with both full and empty particles in radioimmunoassays (RIA). Antiserum against VP2 reacted with artificial empty particles of the virus but not with full particles. In contrast, none of the individual polypeptides of poliovirus produced antisera which neutralized the virus nor reacted with it in RIA. However, antisera produced with VP1 and VP2 reacted with artificial empty particles in RIA.

Foot-and-mouth disease virus (FMDV) attached to pig kidney cells at 0 °C and could only be recovered in a form with a sedimentation coefficient and buoyant density lower than that of the native virus. Incubation of the virus-cell complex at 37 °C caused disruption of about 80% of the particles into a 12S protein sub-unit that had the same polypeptide composition as that produced by reducing the pH of the virus below pH 7. The remaining 20% had the same polypeptide and RNA composition as the native virus but it had a lower sedimentation coefficient, buoyant density and specific infectivity. These lower values are probably due to the association of the virus with cell membrane components. The 12S subunits were shown to be located inside the cell, indicating that disruption of the virus had occurred within the cell. The results are discussed in relation to the different cell mediated alteration of other picornaviruses.

Chymotrypsin cleaves only one of the four major polypeptides of foot-and-mouth disease virus (FMDV serotype O) in situ. This polypeptide (VP1, mol. wt. 29 × 103) was first cleaved into fragments of mol. wt. 20 and 9 × 103 and further cleavage could be prevented by the addition of a large excess of bovine serum albumin. The infectivity of the virus particles at this stage was the same as that of the intact virus although the rate of attachment to BHK 21 cells was slower and the immunogenic activity was reduced. If hydrolysis was allowed to continue, VP1 was cleaved into fragments with mol. wt. r8 and < 9 × 103, similar to those obtained with trypsin, and the virus particles then had a greatly reduced infectivity and a lower immunogenicity. Treatment of strains from five other serotypes of the virus with the two enzymes cleaved only VP1 in each instance and there was a corresponding loss of infectivity. The results are discussed in relation to the location and biological activity of the virus polypeptides.

In addition to the major infective component, which bands at a density of 1.34 g/ml in caesium chloride (‘light component’), a component with a density of 1.44 g/ml (‘heavy component’) has been found in harvests of poliovirus (type 1), Coxsackie B5 virus, a bovine enterovirus (VG-5-27) and swine vesicular disease virus (SVDV). With SVDV about 98% of the infectivity equilibrated at 1.34 g/ml but approx. 2% was present as a peak at 1.44 g/ml. The morphology of the two forms was similar but the heavy component had a smaller diameter (28 nm) than the light component (30 nm). No inter-conversion of the two forms was observed on re-cycling in fresh caesium chloride gradients and the two components had the same proportions of RNA and protein and the same polypeptide composition. Each component gave a similar proportion of the light and heavy forms on replication, but the light component had a specific infectivity about fourfold higher than that of the heavy component and was also much more efficient in eliciting the formation of neutralizing antibodies in guinea pigs. Although these results suggest that the two particles are alternative stable configurations of the virus, iodination failed to reveal any differences in the extent or pattern of labelling of the polypeptides in the two forms.

Treatment of foot-and-mouth disease virus with 4% glutaraldehyde increases the diam. of the particles by 25% and makes them permeable to phosphotungstic acid so that they appear empty. The treated particles also resemble naturally-occurring empty particles in their low sedimentation coefficient (about 75S) but, in contrast to empty particles, they have a normal content of RNA and a higher than normal buoyant density in caesium chloride. The RNA can be removed from fixed particles by ribonuclease. Two models are suggested which account for these alterations in the structure of the virus particles. These results show that fixation with glutaraldehyde, far from maintaining the structural integrity of the virus particles, leads to considerable alterations in the arrangement of the RNA and protein subunits.

Several of the physico-chemical properties of representative members of the Picornaviridae family have been examined. On the basis of their buoyant density in caesium chloride, stability at pH 3 to 7 and the base composition of the virus RNA, a division of this family of viruses into six subgroups is suggested.

Further evidence has been obtained which confirms that foot-and-mouth disease virus contains several structural proteins. By electrophoresis in urea-polyacrylamide gels, virus of type O gave six distinct bands. In sodium dodecyl sulphate-polyacrylamide gels four proteins with molecular weights of 34, 30, 26 and 13.5 × 103 were clearIy demonstrated. When virus preparations were labelled with a single amino acid, in both sodium dodecyl sulphate-polyacrylamide and urea-polyacrylamide gel electrophoresis, the fastest migrating protein contained no arginine and only traces of cysteine. This protein also stained differently from the other bands with Coomassie Blue and was absent from the 12s protein subunit prepared by mild acid (pH 6.5) disruption of the virus. This protein was separated from the 12s subunit by sucrose gradient centrifugation and by ion exchange chromatography on AmberIite IRC-50.

Unfractionated harvests of foot-and-mouth disease virus grown in baby hamster kidney cells fixed complement with both heterotypic and homotypic antisera but the freshly prepared intact virus (25 nm. component) from these harvests fixed complement only with the homotypic antiserum. Storage at 4° or heating at 37° released an antigen from the 25 nm. component which fixed complement with heterotypic serum. This antigen could also be prepared by mixing the 25 nm. component with baby hamster kidney cells but it was obtained in greatest yield by disrupting with guanidine. It had a sedimentation coefficient of 14S in sucrose gradients. Serum from hyperimmunized infected guinea pigs which had been absorbed with excess homotypic 25 nm. component fixed complement with the disrupted virus but not with intact virus. The disrupted virus also reacted with heterotypic antiserum produced by inoculation of guinea pigs with inactivated 25 nm. component, providing further evidence that the antigen is a structural component of the virus particle.