In a previous study we demonstrated that vesicular stomatitis virus (VSV) can be used as a vector to express a soluble protein in mammalian cells. Here we have generated VSV recombinants that express four different membrane proteins: the cellular CD4 protein, a CD4-G hybrid protein containing the ectodomain of CD4 and the transmembrane and cytoplasmic tail of the VSV glycoprotein (G) We reported recently that VSV can be used as a genetically stable, high-level expression vector (2). Here we have investigated VSVs ability to express cellular and viral membrane proteins and we have also examined the incorporation of these proteins, into the vital membrane. Incorporation of foreign viral spike glycoproteins into the envelope of recombinant VSV particles might allow development of a new generation of killed virus vaccines or retargeting of virus particles to specific cell types. Also, expression of other membrane proteins from the VSV genome and their subsequent incorporation into the virions could yield large quantities of the expressed protein in highly purified form for structural or functional studies. Because of these potential applications, we are especially interested in understanding the factors governing membrane protein incorporation into VSV particles. By generating VSV recombinants expressing foreign envelope proteins, we are able to achieve high-level expression of foreign membrane proteins at the same time that the VSV proteins are being made. This system allows us to examine the structural requirements for foreign protein incorporation into VSV particles.Previous studies have shown that the envelope proteins of several other viruses can be incorporated into the envelope of VSV particles when VSV and a second virus are propagated in the same cells. This phenomenon of pseudotype formation is well known (reviewed in ref.3), but the extent of foreign envelope protein incorporation into VSV has not been examined in detail. Also, the cellular CD4 protein expressed from a vaccinia vector was found to be incorporated into VSV particles at a low level of about 60 molecules per virion, and no preference was seen for a chimeric CD4 carrying the transmembrane and cytoplasmic domains of VSV G protein (4). In contrast, a study of HIV envelope protein incorporation into VSV particles lacking G protein indicated a requirement for the cytoplasmic tail of G on the HIV envelope protein (5). These studies complemented earlier studies showing a lower efficiency of incorporation of G proteins with truncated tails into the VSV G ts mutant lacking the G protein (6).Here, using recombinant VSVs expressing very high levels of foreign membrane proteins, we have been able to quantitate both the expression levels of the foreign proteins and their incorporation into VSV particles. We conclude that there is significant extra space in the VSV envelope that can, in many cases, accommodate large amounts of foreign membrane proteins. MATERIALS AND METHODSPlasmid Construction. A plasmid expressing the positive strand RNA complem...

Since the development of a system for generating vesicular stomatitis virus (VSV) from plasmid DNAs, our laboratory has reported the expression of several different glycoproteins from recombinant VSVs. In one of these studies, high-level expression of an influenza virus hemagglutinin (HA) from a recombinant VSV-HA and efficient incorporation of the HA protein into the virions was reported (E. Kretzschmar, L. Buonocore, M. J. Schnell, and J. K. Rose, J. Virol. 71:5982-5989, 1997). We report here that VSV-HA is an effective intranasal vaccine vector that raises high levels of neutralizing antibody to influenza virus and completely protects mice from bronchial pneumonia caused by challenge with a lethal dose of influenza A virus. Additionally, these recombinant VSVs are less pathogenic than wild-type VSV (serotype Indiana). This vector-associated pathogenicity was subsequently eliminated through introduction of specific attenuating deletions. These live attenuated recombinant VSVs have great potential as vaccine vectors.

The matrix protein of influenza virus is a major structural component of the virion which is generally believed to bridge between the membrane envelope and the ribonucleocapsid core. To investigate the interaction of M1 with cellular membranes in the absence of other influenza proteins, we examined its distribution by subcellular fractionation after expression in HeLa cells. Approximately 81 to 88% of M1 protein, expressed without other viral proteins, was soluble, whereas the remaining 12 to 19% was tightly associated with membranes. Conditions known to release peripherally associated membrane proteins did not detach M1 proteins from isolated membranes, suggesting that the fraction of M1 bound to membranes behaves as an integral protein. Coexpression of M1 with hemagglutinin or neuraminidase did not alter the extent of membrane association of M1 protein, indicating that there is no strong interaction between M1 and the cytoplasmic tails of the viral glycoproteins. Additional attempts were made to identify membrane binding domains in M1 protein. Mutants constructed with mutations in the four hydrophobic regions thought to be responsible for membrane association still exhibited the same levels of membrane association as that observed with wild-type matrix protein. Therefore, specific hydrophobic domains are apparently not required for membrane binding.

The hemagglutinin (HA) of influenza virus is a type I transmembrane glycoprotein which is acylated with long-chain fatty acids. In this study we have used oligonucleotide-directed mutagenesis of cloned cDNA and a simian virus 40 expression system to determine the fatty acid binding site in HA and to examine possible functions of covalently linked fatty acids. The results show that the HA is acylated through thioester linkages at three highly conserved cysteine residues located in the cytoplasmic domain and at the carboxy-terminal end of the transmembrane region, whereas a cysteine located in the middle of the membrane-spanning domain is not acylated. Mutants lacking fatty acids at individual or all three attachment sites acquire endoglycosidase H-resistant oligosaccharide side chains, are cleaved into HA1 and HA2 subunits, and are transported to the plasma membrane at rates similar to that of wild-type HA. All mutants are membrane bound and not secreted into the medium. These results exclude transport signal and membrane-anchoring functions of covalently linked fatty acids for this integral membrane glycoprotein. Furthermore, lack of acylation has no obvious influence on the biological activities of HA: cells expressing fatty acid-free HA bind to and, after brief exposure to mildly acidic pH, fuse with erythrocytes; the HA-induced polykaryon formation is not impaired, either. Other possible functions of covalently linked fatty acids in integral membrane glycoproteins which cannot be examined in conventional cDNA expression systems are discussed.

Spodoptera frugiperda (Sf9)-cells differ markedly in their protein glycosylation capacities from vertebrate cells in that they are not able to generate complex type oligosaccharide side chains. In order to improve the oligosaccharide processing properties of these cells we have used baculovirus vectors for expression of human (beta 1,2-N-acetylglucosaminyltransferase I (hGNT-I), the enzyme catalysing the crucial step in the pathway leading to complex type N-glycans in vertebrate cells. One vector (Bac/GNT) was designed to express unmodified GNT-I protein, the second vector (Bac/tagGNT) to express GNT-I protein with a tag epitope fused to its N-terminus. In Sf9-cells infected with Bac/tagGNT-virus a protein of about 50 kDa representing hGNT-I was detected with an antiserum directed against the tag epitope. HGNT-I activity was increased at least threefold in lysates of infected cells when N-acetylglucosamine (GlcNAc)-free ovalbumine was used as substrate. To monitor hGNT-I activity in intact Sf9-cells, the glycosylation of coexpressed fowl plague virus hemagglutinin (HA) was investigated employing a galactosylation assay and chromatographic analysis of isolated HA N-glycans. Coexpression of hGNT-I resulted in an at least fourfold increase of HA carrying terminal GlcNAc-residues. The only structure detectable in this fraction was GlcNAcMan3GlcNAc2. These results show that hGNT-I is functionally active in Sf9-cells and that the N-glycans of proteins expressed in the baculovirus/insect cell system are elongated by coexpression of glycosyltransferases of vertebrate origin. Complete complex type oligosaccharide side chains were not observed when hGNT-I was overexpressed, thus supporting the concept that Sf9-cells do not contain glycosyltransferases acting after hGNT-I.