- Volume 78, Issue 11, 1997

In a step towards a tetravalent dengue virus subunit vaccine which is economical to produce, highly immunogenic and stable, a hybrid dengue virus envelope (E) protein molecule has been constructed. It consists of 36 amino acids from the membrane protein, the N-terminal 288 amino acids of the dengue-2 virus E protein plus amino acids 289–424 of the dengue-3 virus E protein. It has been engineered for secretory expression by fusion to a mellitin secretory signal sequence and truncation of the hydrophobic transmembrane segment. Using the baculovirus expression system and serum- free conditions, more than 95 % of recombinant dengue-2 virus-dengue-3 virus hybrid E protein (rD2D3E) was secreted into the cell culture supernatant in a stable form with multiple features indicative of preserved conformation. The hybrid molecule reacted with a panel of dengue virus- and flavivirus-specific MAbs which recognize linear or conformational epitopes on dengue virions. Human dengue virus-specific antisera also reacted with the protein. The hybrid rD2D3E protein was able to inhibit the in vitro binding of dengue-2 and dengue- 3 viruses to human myelomonocytic cells, suggesting that the receptor-binding epitope(s) was preserved. Adjuvant-free immunization with the hybrid protein induced an antibody response to both dengue-2 and dengue-3 virus in outbred mice, comparable in strength to that of individual rD2E and rD3E proteins. Notably, these antibody responses were primarily of the IgG2a and IgG2b isotype. A strong dengue virus cross-reactive T cell response was also induced in the mice, suggesting that dengue virus hybrid E proteins could form the basis of an efficacious multivalent dengue virus vaccine.

The interaction between the host major histocompatibility complex (MHC) and the genotype of the hepatitis C virus (HCV) was analysed using syntheticfull-length non-structural (NS) 4Aproteins, residues 1658–1712, of genotypes 1b, 2b, 3a, 4a and 5a. Human and murine antibodies specific for the five NS4A genotypes analysed focused on residues 1688–1707. In immunized B10 H-2 con- genic mice, the H-2d, H-2f and H-2s haplotypes were good responders to NS4A, irrespective of the viral genotype. In contrast, the H-2k haplotype was a low or non-responder to all NS4A genotypes, except for genotype 2b. Also, H-2f- and H-2s-restricted NS4A genotype 1b-specific T-cells focused on residues 1670–1679 and 1683–1692, respectively, whereas H-2k-restricted NS4A genotype 2b-specific T-cells focused on the carboxy terminus. Interestingly, H-2f-restricted genotype 1b-specific T-cells did not cross-react with T-cell site analogues of seven other genotypes, whereas the H-2s-restric- ted, genotype 1b-specific T-cells cross-reacted with genotypes 1a, 4a and 5a. Thus the combination of viral genotype and host MHC profoundly influences the ability to mount an HCV NS4A-specific immune response.

Peripheral blood mononuclear cells (PBMCs) from 27 hepatitis C virus (HCV)-infected patients were analysed for the presence of HCV negative strand RNA with strand-specific Tth-based RT-PCR. No negative strand RNA was detected in any sample, and positive strand HCV sequences amplified from PBMCs were identical to those found in serum. These findings suggest that HCV does not replicate in PBMCs, and the presence of HCV sequences at this site is compatible with passive virus adsorption and/or contamination by circulating virus.

The synthesis of long cDNA molecules encoding the complete genome of RNA viruses has recently been demonstrated; this major improvement has numerous practical applications such as construction of infectious cDNA clones or study of sequence variability at the level of a single RNA molecule. Using hepatitis C virus (HCV) as a model, we established an RT-PCR technique for amplification of cDNA fragments with a length of about 5 kb. The RT reaction was carried out with a Moloney murine leukaemia virus reverse transcriptase lacking detectable RNase H activity. For PCR reactions an enzyme mix containing Taq and Pwo DNA polymerases was used. Hot start and addition of 5% DMSO were also important to efficiently achieve long PCR products.

About 106 HCV genome equivalents/ml in serum were needed in order to amplify the HCV genome in only two cDNA fragments covering about 98% of the complete genome. Analysis of the HCV quasispecies is also possible by this method as shown by sequencing of the hypervariable region 1 (HVR1) after cloning of cDNAs. The integrity of the long cDNA clones was proven by (1) restriction analyses, (2) partial sequencing and (3) expression of respective gene products. In vitro transcribed cDNAs were translated in rabbit reticulocyte lysate. Structural and nonstructural HCV proteins were identified by immunoprecipitation using patient serum. These results suggest that the two cDNA clones encode a complete and functional open reading frame of HCV.

A defective hepatitis C virus (HCV) genome in the ascitic fluid of a patient with hepatocellular carcinoma was cloned and sequenced up to the 3′ poly(U) stretch. When compared with the published Taiwanese HCV sequence, this defective genome contained deletions of single nucleotides at eight sites, double nucleotides at two sites, triple nucleotides at four sites, quadruple nucleotides at one site and replacement of a short stretch of sequence at one site. For comparison, the corresponding regions containing these mutations were also cloned from a serum sample from this patient. Except for deletions of two triple nucleotides in the hypervariable region, the reading frames of all serum-derived clones were intact. The defective HCV genome encoded a truncated core protein with 90 amino acid residues (the last 20 amino acid residues came from a different reading frame), whereas the serum-derived genome encoded a full- length core protein. When expressed in Huh-7 cells, these two proteins were localized to the nucleus and cytoplasm, respectively. Using specific primer-sets, ascites- and serum-derived genomes were each detected alone in ascitic fluid and serum samples, respectively, whereas both sequences were present in ascitic mononuclear cells. The defective sequence thus constituted the major virus population in the ascitic fluid whereas a putative helper genome coexisted with it inside the ascitic mononuclear cells. This sequence is possibly a defective and interfering genome.

A transfusion-transmissible agent provisionally designated hepatitis G virus (HGV) was recently identified. In this study, we examined the variability of the HGV genome by analysing sequences in the putative envelope region from 72 isolates obtained from diverse geographical sources. The 1561 nucleotide sequence of the E1/E2/NS2a region of HGV was determined from 12 isolates, and compared with three published sequences. The most variability was observed in 400 nucleotides at the N terminus of E2. We next analysed this 400 nucleotide envelope variable region (EV) from an additional 60 HGV isolates. This sequence varied considerably among the 75 isolates, with overall identity ranging from 79·3% to 99·5% at the nucleotide level, and from 83·5% to 100% at the amino acid level. However, hypervariable regions were not identified. Phylogenetic analyses indicated that the 75 HGV isolates belong to a single genotype. A single-tier distribution of evolutionary distances was observed among the 15 E1/E2/NS2a sequences and the 75 EV sequences. In contrast, 11 isolates of HCV were analysed and showed a threetiered distribution, representing genotypes, subtypes, and isolates. The 75 isolates of HGV fell into four clusters on the phylogenetic tree. Tight geographical clustering was observed among the HGV isolates from Japan and Korea.

Pure preparations of envelope glycoproteins Erns and E2 of classical swine fever virus (CSFV) synthesized in insect cells were used to study infection of porcine and bovine cells with the pestiviruses CSFV and bovine viral diarrhoea virus (BVDV). Almost 100% inhibition of infection of porcine kidney cells with CSFV was produced by 100 μg/ml Erns. After removal of the virus no Erns was needed in the overlay medium (growth medium) to maintain this level of inhibition. In contrast, 100% inhibition of infection of porcine kidney cells with CSFV by 10 ^g/ml E2 was only achieved when E2 was added to the overlay medium. When E2 was omitted, a maximum of 50% inhibition was achieved. This indicated that after the virus and E2 were removed from the cells, infection still occurred, by virus particles which were still bound to the cell surface. Treatment with 100 μg/ml Erns released these particles from the cell surface. Furthermore, Erns bound irreversibly to the surface of cells susceptible or unsusceptible to pestivirus infection and cell-to-cell spread of CSFV was completely inhibited by E2 but not by Erns. These results demonstrated that Erns and E2 interacted with different cell surface receptors. Inhibition of BVDV infection of porcine and bovine cells by CSFV E2 suggested that CSFV E2 and BVDV E2 share an identical receptor. BVDV strain 5250 isolated from pigs was efficiently inhibited by CSFV Erns, whereas several BVDV strains isolated from cattle were not, suggesting that the conformation of Erns plays a role in host tropism.

The translation products of the human coronavirus (HCV) 229E open reading frames 1a and 1b, the polyproteins 1a and 1ab, are processed by virus- encoded proteinases. One of the key enzymes in this process is a chymotrypsin-like enzyme, the 3C- like proteinase. In this study we have identified an ORF 1b-encoded, 41 kDa processing product in HCV 229E-infected cells by using a monoclonal antibody with defined specificity. We show that this polypeptide is released from polyprotein 1ab by 3C-like proteinase-mediated cleavage of the peptide bonds Gln-6110/Gly-6111 and Gln-6458/Ser- 6459. Also, we have investigated the subcellular localization of the 41 kDa processing product. Immunofluorescence microscopy revealed a punctate, perinuclear distribution of the 41 kDa polypeptide in infected cells and an identical subcellular localization was observed for three additional pp1ab-derived polypeptides. In contrast, the virus nucleocapsid protein showed a homogeneous cytoplasmic localization.

Aminopeptidase N (APN) is the major cell surface receptor for group 1 coronaviruses. In this study, we have isolated and characterized a feline APN cDNA and shown that the transfection of human embryonic kidney cells with this cDNA renders them susceptible to infection with the feline coronavirus feline infectious peritonitis virus, the human coronavirus (HCV) 229E and the porcine coronavirus porcine transmissible gastroenteritis virus. By using chimeric APN genes, assembled from porcine and feline sequences, we have shown that, analogously to the human APN protein, a region within the amino-terminal part of the feline APN protein (encompassing amino acids 132–295) is essential for its HCV 229E receptor function. Furthermore, by comparing the relevant feline, human and porcine APN sequences, we were able to identify a hypervariable stretch of eight amino acids that are more closely related in the feline and human APN proteins than in the porcine APN molecule. Using PCR- directed mutagenesis, we converted this stretch of amino acids within the porcine APN molecule to the corresponding residues of the human APN molecule. These changes were sufficient to convert porcine APN into a functional receptor for HCV 229E and thus define a small number of residues that are critically important for the HCV 229E receptor function of human APN.

The capsid protein of a French isolate of the European brown hare syndrome virus (EBHSV) was expressed in the baculovirus system. The recombinant EBHSV (rEBHSV) capsid protein was able to self-assemble into virus-like particles (VLPs). The VLPs were indistinguishable from the infectious EBHSV and displayed morphological characteristics similar to those we have described for the VLPs resulting from the expression of the capsid protein of rabbit haemorrhagic disease virus (RHDV), a closely related calicivirus. Cross-protection experiments showed that vaccination with rEBHSV particles did not protect rabbits against an RHDV challenge. A set of monoclonal antibodies (MAbs) was raised against rEBHSV capsid protein and used together with anti-RHDV and anti-EBHSV MAbs produced against native viruses to study the antigenic relationships between the two calici- viruses. All six anti-EBHSV MAbs delineated discontinuous epitopes; two of them reacted with specific surface epitopes and the remaining four reacted with internal epitopes which were also present in rRHDV. All anti-RHDV MAbs were monospecific; three reacted with surface linear epitope(s), two reacted with internal linear epitope(s) and one with a surface conformational epitope. On the basis of all these results, a classification of RHDV and EBHSV as two serotypes of a single serogroup is proposed.

A genetic engineering approach was made to generate a recombinant non-segmented negative- strand RNA virus, Sendai virus (SeV) of the family Paramyxoviridae, that expresses firefly luciferase. The DNA construct containing the entire open reading frame (ORF) of the luciferase gene followed by the SeV transcription stop and restart signals connected with the conserved intergenic three nucleotides was inserted immediately before the ORF of the viral 3′-proximal nucleocapsid (N) protein gene in a full-length SeV cDNA copy. After intracellular expression of full-length antigenomic transcripts from the engineered cDNA and of the viral nucleocapsid protein and RNA polymerase from the respective plasmids, a recombinant SeV expressing luciferase activity at a high level was recovered, although the tendency of this particular reporter gene product to aggregate in cells made it difficult to estimate the maximum level of expression. The increase in genome length brought about by inserting 1728 nucleotides into the 15 384 nucleotide parental SeV was associated with reduced plaque size, slightly slower replication kinetics and a severalfold decrease in yield of the virus. The inserted luciferase gene was stably maintained after numerous rounds of replication by serial passages in chick embryos. These results indicate the potential utility of SeV as a novel expression vector.

Two influenza viruses of differing virulence, clone 7a (virulent for humans and ferrets) and A/Fiji (attenuated for both species), induced differential levels of cytotoxicity (as measured by release of lactate dehydrogenase from cells) and apoptosis (as determined by altered nuclear morphology or flow cytometry) in MDCK and U-937 cells. Clone 7a induced more apoptosis and cytotoxicity than A/Fiji, despite the fact that its infectious virus yields were similar to or less than those for A/Fiji. This indicates a greater capacity of clone 7a to induce the two phenomena. Nevertheless the replication process was clearly essential, since UV-irradiated virus induced little or no apoptosis, and apoptosis occurred as a late event post-infection. The possible relevance of these findings to destruction of host cells by influenza virus in vivo is discussed.

We investigated the effect of rabies virus infection on the actin cytoskeleton using various techniques. Confocal microscopic examination of rabies virus- infected neuroblastoma cells at late stages of infection revealed a dramatic decrease in F-actin staining. The results of a fluorimetric assay with pyrenylated actin indicated that purified rabies virus nucleocapsid has no direct action on the kinetics of actin polymerization and only a weak effect on the final extent of polymerization. Videomicroscopy experiments with purified components showed that rabies virus nucleocapsid inhibits the actin-bundling effect induced by dephospho-syn- apsin I, a neuron-specific protein which is known to exert a control on the actin-based cytoskeleton. Thus, the observed decrease in F-actin staining in infected cells might be ascribed to an indirect action of rabies nucleocapsid on the effects of actin- binding proteins such as synapsin I.

To evaluate the genetic diversity of viral haemorrhagic septicaemia virus (VHSV), the sequence of the glycoprotein genes (G) of 11 North American and European isolates were determined. Comparison with the G protein of representative members of the family Rhabdoviridae suggested that VHSV was a different virus species from infectious haemorrhagic necrosis virus (IHNV) and Hirame rhabdovirus (HIRRV). At a higher taxonomic level, VHSV, IHNV and HIRRV formed a group which was genetically closest to the genus Lyssavirus. Compared with each other, the G genes of VHSV displayed a dissimilar overall genetic diversity which correlated with differences in geographical origin. The multiple sequence alignment of the complete G protein, showed that the divergent positions were not uniformly distributed along the sequence. A central region (amino acid position 245–300) accumulated substitutions and appeared to be highly variable. The genetic heterogeneity within a single isolate was high, with an apparent internal mutation frequency of 1·2 × 10 3 per nucleotide site, attesting the quasispecies nature of the viral population. The phylogeny separated VHSV strains according to the major geographical area of isolation: genotype I for continental Europe, genotype II for the British Isles, and genotype III for North America. Isolates from continental Europe exhibited the highest genetic variability, with sub-groups correlated partially with the serological classification. Neither neutralizing polyclonal sera, nor monoclonal antibodies, were able to discriminate between the genotypes. The overall structure of the phylogenetic tree suggests that VHSV genetic diversity and evolution fit within the model of random change and positive selection operating on quasispecies.

The nucleotide sequence of the small (S) RNA segment of Akabane (AKA) bunyavirus was determined. The segment is 858 nucleotides long and contains two overlapping open reading frames (ORFs), which encode the nucleocapsid (N) and nonstructural (NSs) proteins, consistent with other bunyaviruses. Comparisons with the Aino virus S RNA sequence indicated that there is 73·5% identity in nucleotide sequence. However, the sequence identity of the 5′ non-coding region of the genomic RNA between these two viruses is only 55%. The N ORFs from 20 Japanese and 2 Australian isolates of AKA virus were sequenced and subjected to phylogenetic analysis. This suggested that AKA virus has evolved in multiple lineages. Twenty-three isolates were grouped into three major clusters, and the cluster which includes recent isolates was subdivided into two branches. Thus, phylogenetic analysis of the AKA virus N protein gene gives a greater insight into bunyavirus evolution.

Eighteen strains of Rift Valley fever (RVF) virus collected over a period of 38 years and isolated from diverse localities in Africa and from various hosts (human, animal and arthropod) were investigated by RT-PCR followed by sequencing of the NSS protein coding region. This region was chosen to analyse variability because, in contrast to the N protein, the NSS protein differs in various phlebo- viruses and there exists an RVF virus (clone 13) in which 70% of the NSS ORF is deleted, suggesting that this sequence is under a weak selective pressure. Sequence data indicated that percentage di-vergence among isolates ranged from 0 to 9·6% at the nucleotide level and from 0 to 9·5% at the amino acid level. Phylogenetic analysis based on the NSS gene revealed two major lineages: Egyptian and sub-Saharan. This led to the establishment of the relatedness between strains and insights into the NSS protein, the function of which is still undetermined. Alignment of the deduced amino acid sequences indicated that the cysteine residues are conserved, as are several motifs representing potential phosphorylation sites.

To identify potential selection pressures which lead to RNA sequence conservation, we examined the occurrence rates of dinucleotides in 64 singlestranded RNA virus genomes. These viruses may offer a particular insight into these pressures since their RNA-dependent RNA polymerases lack proofreading capability. This potentiates introduction of mutations into their genomes, yet unidentified selection processes conserve the genomes to a large degree. We report a strong inverse correlation between the C G content and the occurrence of the CpG dinucleotide (r = 0·71) in the RNA virus genomes, in contrast to earlier reports (Karlin et al., 1994, Journal of Virology 68, 2889–2897). We also detected significant suppression of UpA, correlating inversely with genomic U A content. These sup-pressions are coupled with over-representation of the complementary pair of dinucleotides, CpA and UpG. In addition, we highlight the fact that odds ratios for dinucleotides are not independent variables, a situation apparently not widely appreciated in the literature. This led us to view the over-representation of CpA and UpG as a consequential outcome of UpA and CpG suppression in the virus genomes. Potential factors influencing these disturbances are discussed. In addition, higher than random incidence was observed for ‘out-of-frame’ stop codons in the viral RNA genomes, with some preferences for individual codons being exhibited by certain virus groups. The UAG codon appeared more common in the M1 frame, the UGA in the 1 frame.

In this study we have analysed variability in the V1 and V2 regions of human immunodeficiency virus type 1 (HIV-1) proviral sequences amplified from lymphoid tissue, brain and other non-lymphoid tissue collected at autopsy from three HIV-1-infected individuals with giant cell encephalitis. We found no evidence for any tissue-specific grouping of variants in the V1/V2 regions, in contrast to previous comparisons of sequences in the V3 region, but consistent with the existence of evolu- tionarily distinct lineages previously observed in these study subjects by sequence comparisons of the p17gag gene. Examination of inferred amino acid sequences from V1 and V2 revealed no correlations between tissue origin with overall charge, length or number of glycosylation sites. Length polymorphism analysis is a rapid method to compare whole populations of HIV-1 variants within a sample, and provides information on the length and diversity of the V1 and V2 hypervariable regions. Based upon a comparison of 42 individuals with CD4 counts ranging from 802 to < 1 at time of death, we found no evidence for changes in the length of V2 with development of AIDS. Using the number of length variants in the V1 and V2 hypervariable region as a marker of the overall degree of variability within HIV populations, we found no evidence for an increase or a decrease in diversity between those with and without AIDS defining illness.

It has previously been shown that trypsinized triplelayered particles of rotavirus induce destabilization of liposomes and membrane vesicles in the absence of Ca2 , a condition which leads to solubilization of the outer capsid proteins of the virus. In this work, we have studied the relationship between outer capsid solubilization and permeabilization of membrane vesicles, monitoring particle and vesicle size simultaneously by changes in light scattering. Per- meabilization of intact cells induced by solubilized outer capsid proteins was monitored by following the rate of entry of ethidium bromide into the cells. Solubilized outer capsid proteins separated from double-layered particles induced vesicle permeabi- lization. Solubilization of the outer capsid preceded and was required for vesicle or cell permeabiliza- tion. Membrane damage induced by rotaviral outer proteins was not repaired upon addition of 1 mM Ca2 to the medium. Rotavirus infection and cell permeabilization were correlated in six different cell lines tested. This phenomenon might be related to the mechanism of virus entry into the cell. We propose a new model for rotavirus internalization based on the permeabilizing ability of outer capsid proteins and the cycling of trapped calcium in the endosomal compartment.