Expression of bovine viral diarrhoea virus glycoprotein E2 by bovine herpesvirus-1 from a synthetic ORF and incorporation of E2 into recombinant virions

▪ Cell culture and viruses.

BHV-1 strain Schönböken (BHV-1/Schö) was obtained from O. C. Straub (Tübingen, Germany) and propagated on Madin–Darby bovine kidney cell clone Bu100 (MDBK-Bu100; kindly provided by W. Lawrence and L. Bello, Philadelphia, USA). Cells were grown in Dulbecco’s minimum essential medium supplemented with 5% foetal calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin and 0·35 mg/ml l- glutamine. The gD-negative mutant BHV-1/80-221 was propagated on the cell line BU-Dorf, which constitutively expresses BHV-1 glycoprotein D (gD), as described previously (Schröder et al., 1997⇓ ).

▪ Sequencing of the BVDV strain C86 E2 ORF.

Whole-cell RNA was isolated from cells infected with BVDV strain C86, which is a component of the Intervet Bovilis BVD-MD vaccine (kindly provided by Intervet International, Boxmeer, Netherlands), by using the Pharmacia RNA extraction kit as recommended by the supplier. Reverse transcription and PCR amplification of the E2 ORF were done as described previously (Becher et al., 1997⇓ ) by using primers O1 35A (5′ AARTARTCTGTGACATAACT, antisense) and O1 W2 (5′ CGCGGATCCTGGTGGCCTTATGA, sense). PCR fragments were purified after agarose gel electrophoresis and cloned by using the TA cloning kit (Invitrogen) following the supplier’s instructions. Nucleotide sequences of both strands from three independent clones were determined by cycle sequencing with the Thermo Sequenase kit (Amersham).

▪ Construction and cloning of the BVDV E2_syn ORF.

Synthetic oligonucleotides (Gibco-BRL) were hybridized as described previously (Kühnle et al., 1998⇓ ) to generate the double-stranded DNA fragments shown in Table 1⇓. Cloning was performed by established methods (Sambrook et al., 1989⇓ ).

Fragment 1 was ligated via its cohesive ends into BglII/XhoI-cut pSP73, resulting in plasmid pNH1, which was in turn cleaved with StuI and XhoI to receive fragment 2. The resulting plasmid, pNH2, was cleaved with PstI and Xho I and ligated with fragment 3 to obtain pNH3. pNH3 was cleaved with HindIII and XhoI and ligated with fragment 4 to give pNH4. pNH4 was cleaved with SfuI and XhoI and received fragment 5, resulting in pNH5, which was in turn cleaved with Bsu36I and XhoI and ligated with fragment 6. The resulting plasmid, pNH6, was cleaved with EagI and Xho I and ligated with fragment 7 to give plasmid pNH7.

For the integration of fragments A to F, pSP73 was cleaved with BglII and XhoI. Fragment A was ligated into this DNA, resulting in plasmid pCA, which was in turn cleaved with BglII and BstXI and ligated with fragment B. The resulting plasmid, pCB, was cleaved with BglI and SacII and fragment C was integrated to give pCC, which was in turn cleaved with Bgl II and ApaI and ligated with fragment D to give pCD. pCD was cleaved with BglII and BsrGI and ligated with fragment E, resulting in pCE. pCE was cleaved with BglII and AccI and ligated with fragment F to obtain plasmid pCF.

Finally, plasmid pNH7 was cleaved with BamHI and Xho I and ligated with the purified BamHI–XhoI insert from plasmid pCF to yield plasmid pspE2_syn, containing the reconstructed BVDV E2 ORF. The construction was verified by sequencing after each cloning step.

▪ Construction of other plasmids.

To provide a start codon for initiation of translation and a signal peptide for transport and processing of the BVDV E2 glycoprotein, a BamHI–KpnI DNA fragment GATCCACCATGGCCCTGTTGGCTTGGGCGGTGATAACAATCTTGCTGTACCAGCCTGTAGCAGGGTAC (start codon in bold), encompassing the codons for the pestivirus E^rns signal sequence (Sig; MetAlaLeuLeuAlaTrpAlaValIleThrIleLeuLeuTyrGlnProValAlaGlyTyr; Rümenapf et al., 1993⇓ ) was cleaved from plasmid pgsCP7LE2 (P. Becher & H.-J. Thiel, unpublished) with BglII and KpnI. The isolated fragment was ligated into the BglII/ KpnI-cut plasmid pROMe (Kühnle et al., 1996⇓ ) to give plasmid pROMeSig. Plasmid pROMeSig was cut with KpnI and NotI and ligated with the BVDV E2_syn ORF, which had been cleaved from plasmid pspE2syn with KpnI and NotI. The resulting plasmid was named pROMeSigE2_syn and contained the reconstructed BVDV E2_syn ORF preceded by the pestivirus signal peptide-coding sequence.

In order to test the influence of an intron on the level of expression of the SigE2_syn gene in recombinant BHV-1- infected cells, the chimeric intron sequence contained within a 207 bp AflII fragment of plasmid pCI-neo (Promega) was integrated 8 bp upstream from the SigE2_syn ORF into AflII-cleaved pROMeSigE2_syn. This insertion position is 127 nt downstream of the 5′ cap site of the MCMV e1 promoter. The correct orientation of the intron sequence within the resulting plasmid pROMeSigE2_syn-intron was verified by sequencing.

For in vitro transcription and translation, the SigE2_syn ORF was isolated from plasmid pROMeSigE2_syn by cleavage with AflII and XhoI, made blunt-ended with Klenow polymerase and integrated into SmaI-cleaved pSP73. In the resulting plasmid, pspSigE2_syn, in vitro transcription of the SigE2_syn ORF is under the control of the phage T7 promoter.

▪ Construction of recombinants BHV-1/SigE2_syn and BHV-1/SigE2_syn-intron.

In order to integrate the E2_syn ORF into the genome of BHV-1, recombination plasmids pROMeSigE2_syn and pROMeSigE2 _syn-intron were co-transfected into MDBK cells with purified BHV-1/80-221 DNA. After complete lysis of the transfected cultures, progeny virus was titrated on MDBK cells. In BHV-1/80-221, the ORF encoding the essential gD is replaced by a lacZ expression cassette (Fig. 2⇓) and, therefore, only viruses that have acquired the gD ORF from the recombination plasmids should be able to replicate productively on non-complementing cells (Fehler et al. , 1992⇓ ; Kühnle et al. , 1996⇓ ). Virions from plaques that did not stain blue under a Bluo- Gal-containing agarose overlay (Fehler et al., 1992⇓ ) were plaque-purified and the recombinants BHV-1/SigE2_syn and BHV-1/SigE2_syn-intron were selected for further characterization. To verify integration of the expression cassettes into the genomes of the respective viruses, whole-cell DNA from MDBK cells infected with BHV-1/SigE2_syn and BHV-1/SigE2_syn -intron was prepared 20 h after infection, cleaved with HindIII, transferred to nitrocellulose membranes after electrophoresis in 0·6% agarose gels and hybridized to E2_syn -, BHV-1 gD-, BHV-1 gE- and lacZ-specific ³²P- labelled DNA fragments. No hybridization was found with the lacZ -specific probe and the sizes of the fragments detected by the other probes were as expected (data not shown).

▪ In vitro transcription and translation.

Plasmid pspSigE2_syn was linearized with BglII and cRNA was transcribed by T7 or SP6 RNA polymerase in the presence of the cap analogue m⁷GpppG according to the manufacturer’s protocol (Boehringer). In vitro translation of the cRNAs was performed in the presence of 60 μCi [³⁵S]methionine per reaction mixture in the absence or presence of canine microsomal membranes, as recommended by the supplier (Promega). Labelled proteins were visualized by fluorography after separation on SDS–10% polyacrylamide gels as described previously (Keil et al., 1985⇓ ).

▪ RNA isolation, Northern blot hybridization and primer-extension analysis.

Cytoplasmic RNA was isolated as described previously (Schmitt & Keil, 1996⇓ ). Glyoxal-treated RNA (5 μg) was separated in 1% formaldehyde gels, transferred to nitrocellulose filters and hybridized to ³²P-labelled DNA following established procedures (Keil et al., 1987⇓ ; Sambrook et al., 1989⇓ ). Primer-extension analysis was performed as described previously (Bü hler et al., 1990⇓ ). Elongated fragments were separated on 6% polyacrylamide sequencing gels containing urea and visualized by autoradiography. 5′ end-labelled HpaI fragments of pBR322 and 123 bp and 1 kbp ladders (Gibco-BRL) were used as size markers.

▪ Antibodies and sera.

BVDV-neutralizing, E2 glycoprotein-specific MAbs 1a16, 1a5, 1c17 and 1D5 (Weiland et al., 1989⇓ ) were kindly provided by E. Weiland (Tübingen, Germany) and used as mixtures of equal amounts of hybridoma culture supernatants (BVDV E2-specific MAb cocktail). The calf anti-BVDV serum, kindly provided by J. Patel (Intervet UK) was raised by repeated inoculation of BVDV strain C86. The polyclonal serum directed against BHV-1 gD and the gD-specific MAb 21/3/3 have been described elsewhere (Fehler et al., 1992⇓ ).

▪ Immunoprecipitation and deglycosylation reactions.

Immunoprecipitation of [³⁵S]methionine-labelled proteins from infected cells and purified virions was done as reported previously (Keil et al., 1985⇓ ; Fehler et al. , 1992⇓ ). Immunoprecipitated proteins were visualized by fluorography after separation on SDS–10% polyacrylamide gels as described previously (Keil et al., 1985⇓ ).

For deglycosylation, immunoprecipitated proteins were incubated overnight at 37 °C with 0·4 U N- glycosidase F (Boehringer), 1 mU neuraminidase (Boehringer) or 1 mU neuraminidase and 1·5 mU O-glycosidase (Boehringer) under conditions recommended by the supplier.

▪ Single-step growth curves.

MDBK-Bu100 cells were infected with BHV-1 at 10 p.f.u. per cell, incubated for 2 min with citrate buffer (40 mM citric acid, 10 mM KCl, 135 mM NaCl, pH 3·0) at 2 h post-infection (p.i.), in order to inactivate virions that had not penetrated into the cells, and washed twice with cell culture medium. At the times indicated, supernatants were collected and stored at −70 °C. Cells were washed with PBS, incubated with citrate buffer for 2 min to inactivate cell-associated extracellular virus, washed with PBS and harvested by low-speed centrifugation after trypsinization. Cell pellets were resuspended in 1 ml cell culture medium and stored at −70 °C. Cells and supernatants were thawed and sonicated for 20 s at 80 W in a Branson ultrasonic water-bath. Serial dilutions were titrated on MDBK-Bu100 cells and plaques were counted 3 days later after incubation under semi-solid medium containing methyl cellulose.

▪ Penetration kinetics.

Penetration kinetics were determined essentially as described previously (Fehler et al., 1992⇓ ) through low pH inactivation of extracellular virions at different times after a shift of infected cells from 4 to 37 °C. The plaque count of untreated control cultures was defined as 100% penetration.

Integration of the BVDV-E2_syn ORF into the genome of BHV-1/80-221

Attempts to express cDNA-encoded glycoproteins F and G of BRSV, which replicates in the cytoplasm of infected cells, by BHV-1 via the nucleus were not successful. Modification of the BRSV glycoprotein G ORF was necessary for transcription of stable mRNA in recombinant BHV-1- infected cells. It was assumed that splice signals or as yet unidentified sequence motifs resulted in the degradation of nuclear transcripts from the respective cDNA ORFs (Kühnle et al., 1998⇓ ).

Analysis of the cDNA sequence encoding the E2 glycoprotein of BVDV strain C86 revealed the presence of motifs identical or similar to splice-donor consensus sequences (Mount, 1982⇓ ) at positions 152–157, 419–424, 614–619 and 1061–1066 and a consensus sequence for polyadenylation (Birnstiel et al., 1985⇓ ; McLauchlan et al., 1985⇓ ) at positions 913–918 (Fig. 1⇓). Since it has been shown that the presence of a splice-donor consensus sequence within a pestivirus E2 glycoprotein-encoding cDNA led to splicing of transcripts after synthesis in the nucleus (Shiu et al., 1997⇓ ), we constructed a modified ORF encoding the BVDV strain C86 E2 glycoprotein (E2_syn ORF) for expression by BHV-1. If compatible with the cloning strategy, the codon usage of the E2_syn ORF was adapted to that of BHV-1 gD and the above-mentioned splice-donor and polyadenylation consensus sequences were altered. It should be noted that, in this study, instability or splicing of transcripts from a genomic RNA-derived cDNA fragment has not been analysed. In a previous report, cDNA that encompassed the ORF encoding the classical swine fever virus (CSFV) E2 glycoprotein was integrated into the genome of pseudorabies virus (van Zijl et al., 1991⇓ ). The resulting recombinants expressed the CSFV E2 glycoprotein in cell culture and pigs vaccinated with infectious virions were protected against disease caused by both pseudorabies virus and CSFV. The CSVF E2 ORF used in that study did not contain splice-donor consensus sequences, and therefore could result in nuclear transcripts that were not processed by splicing.

In order to allow proper intracellular transport and processing of the E2 glycoprotein, the sequence encoding the signal peptide preceding the pestivirus glycoprotein E^rns (Sig) was fused in-frame to the 5′ end of the E2_syn ORF (see Methods). To test whether RNA encompassing the resulting SigE2_syn ORF within plasmid pspSigE2_syn had the potential to direct synthesis of a glycoprotein, mRNA was transcribed in vitro and translated in a cell-free rabbit reticulocyte lysate with or without canine microsomal membranes. Translation products were analysed by SDS–10% PAGE under reducing (see Fig. 5a⇓) and non-reducing conditions (not shown). The apparent molecular mass of 43 kDa for the protein synthesized in the absence of membranes was in agreement with the calculated size of 44·7 kDa. In the presence of canine microsomal membranes, the in vitro translation products migrated at 54 kDa, indicating that the primary translation product encoded by the E2_syn ORF was translocated and modified by addition of N-linked carbohydrates. Electrophoresis of the in vitro-translated polypeptides under non-reducing conditions revealed the same migration behaviour as after reduction, indicating that in vitro synthesis did not result in dimerization, even after translocation and core glycosylation.

Attempts to immunoprecipitate the in vitro translation products by using a calf serum against BVDV strain C86 or a cocktail containing BVDV E2-specific MAbs (Weiland et al., 1989⇓ ) were not successful (see Fig. 5a⇓).

Transcription from SigE2_syn and SigE2_syn- intron genes in recombinant BHV-1-infected cells

In order to test for transcripts containing the recombinant BVDV E2 ORFs, cytoplasmic RNA isolated at 6 h p.i. from cells infected with BHV-1/SigE2_syn (Fig. 3⇓, lanes 1) or BHV-1/SigE2_syn- intron (Fig. 3⇑, lanes 2) was analysed by Northern blot hybridization. The probe representing the BVDV E2 ORF hybridized to RNAs of 1·8 and 1·9 kb after infection with BHV- 1/SigE2_syn and BHV-1/SigE2_syn-intron, respectively (Fig. 3a⇑, lanes 1 and 2). A BHV-1 gD-specific probe detected transcripts of 1·8 kb in both BHV-1/SigE2_syn- and BHV-1/SigE2_syn-intron- infected cells (Fig. 3a⇑, lanes 3 and 4). Taking into account the increase in size of mRNAs resulting from polyadenylation, the lengths of the RNAs were as expected. Comparison of the intensities of the signals between the SigE2_syn and SigE2_syn-intron RNAs and the signals obtained with the corresponding gD transcripts as internal standards indicated that the presence of the intron within the SigE2_syn-intron gene might have increased the steady-state level of the transcripts slightly. However, any effect was apparently less significant than the effects observed in other systems (Petitclerc et al., 1995⇓ ).

In order to demonstrate that the transcripts originating from the SigE2_syn-intron gene represented spliced RNAs, an oligonucleotide complementary to positions 103–144 (Fig. 1 ⇑) was hybridized to the RNA and elongated by reverse transcriptase. Extended DNA fragments of 339 and 409 nt, respectively, were detected with RNA from BHV-1/SigE2_syn-and BHV-1/SigE2_syn-intron-infected cells (Fig. 3b⇑, lanes 1 and 2). No specific extension product was found when RNA from non-infected cells was used (Fig. 3b⇑, lane 3). Provided that initiation of transcription downstream from the MCMV e1 promoter occurs at the same position in murine MCMV-infected cells (Bühler et al., 1990⇓ ) and bovine recombinant BHV-1-infected cells, the expected sizes of the elongated transcripts were 343 nt for BHV-1/SigE2_syn, 413 nt for spliced BHV-1/SigE2_syn-intron RNA and 546 nt for unspliced BHV-1/SigE2_syn -intron transcripts. Thus, the chimeric intron sequence was removed from the primary SigE2_syn-intron transcripts, which demonstrates that the splice signals provided by the chimeric intron were recognized. The four nucleotide difference between the expected and determined sizes of the extended molecules is probably due to compressions during electrophoresis. It is also possible that alternative 5′ cap sites for the MCMV e1 promoter directed transcription within the BHV-1 genome.

Identification and characterization of the BVDV E2 glycoprotein expressed by BHV-1

In order to identify the BVDV SigE2_syn ORF-encoded polypeptide, metabolically ³⁵S-labelled proteins from cells infected with BHV-1/SigE2_syn, BHV-1/SigE2_syn- intron and wild-type BHV-1 were immunoprecipitated by using the calf polyclonal anti-BVDV serum or BHV-1 gD-specific MAb 21/3/3 (Fig. 4 ⇓a). Comparable amounts of a protein that migrated in SDS–polyacrylamide gels with an apparent molecular mass of 54 kDa under reducing conditions and 94 kDa under non-reducing conditions were precipitated specifically from BHV- 1/SigE2_syn- and BHV-1/SigE2_syn-intron-infected cells (Fig. 4a⇑; lanes 2, 3, 5 and 6). These proteins were not precipitated from wild-type BHV-1-infected cells. BHV- 1 gD, which migrates with an apparent molecular mass of 72 kDa under both reducing (not shown) and non-reducing conditions, was precipitated by MAb 21/3/3 from all cell lysates (Fig. 4a⇑, lanes 7–9). The similar intensities of the bands generated by gD demonstrated that the cells were comparably infected. We conclude from these results that, in infected cells, the BHV-1- expressed BVDV E2 glycoprotein exhibited an apparent molecular mass of 54 kDa and formed a 94 kDa homodimer. This is in good agreement with the migration behaviour of the monomeric and homodimeric forms of the E2 glycoprotein found in BVDV-infected cells (Weiland et al. , 1990⇓ ). In addition, the results show that the presence of the intron in BHV-1/SigE2_syn-intron does not improve expression of the E2 glycoprotein significantly.

In order to investigate whether the E2 glycoprotein was associated with recombinant BHV-1 particles, proteins from purified, ³⁵ S-labelled BHV-1/SigE2_syn virions were precipitated by the calf anti-BVDV serum, the BVDV E2-specific MAb cocktail and gD-specific MAb 21/3/3. Immunoprecipitations of infected-cell proteins, labelled with ³⁵S under conditions that allow detection of the 63 kDa gD-precursor molecule (Fig. 4b⇑, lane 3), which is not incorporated into virions, were performed to control the purity of the virus preparation (Fig. 4b⇑, lanes 6 and 9). Interestingly, the glycoprotein E2 monomer and the dimeric form precipitated from virions by the calf serum and the MAb cocktail migrated slightly more slowly in SDS–polyacrylamide gels (Fig. 4b⇑, lanes 4, 5, 7 and 8) than the corresponding proteins from infected cells (Fig. 4b⇑, lanes 1 and 2; data for non-reducing conditions not shown) and exhibited apparent molecular masses of 57 and 101 kDa, respectively. In lane 8 of Fig. 4(b)⇑, diffuse bands are also visible at positions slightly below that of the glycoprotein E2 monomer. Although they were not analysed further, it is assumed that this signal was created by breakdown products of the dimeric form and did not represent an intact monomer. In conclusion, the E2 glycoprotein is associated with BHV-1/SigE2_syn virions as a dimer, where it exhibits a higher apparent molecular mass than that found for the E2 glycoprotein in infected cells. We assume that the modification(s) leading to the decreased mobility is linked to the release of the recombinant BHV-1 virions.

Cell-free translation of in vitro-transcribed SigE2_syn mRNA without (Fig. 5⇓a, lane 1) or with (Fig. 5a⇑, lane 2) canine pancreatic microsomal membranes indicated that the E2 glycoprotein contained N-linked carbohydrates. After synthesis in the presence of microsomal membranes, the protein with an apparent molecular mass of 54 kDa migrated in a comparable manner to the E2 glycoprotein monomer expressed in BHV- 1/SigE2_syn-infected cells. However, attempts to immunoprecipitate the in vitro-synthesis products with the calf anti-BVDV serum or the anti-E2 MAb cocktail were not successful (Fig. 5a⇑, lanes 3–6), suggesting that intracellular transport and/or dimerization of the E2 glycoprotein is crucial for proper processing. This interpretation was supported by pulse–chase experiments with BHV-1/SigE2_syn-infected cells (Fig. 5c⇑). In contrast to BHV-1 gD, for which conversion of the 63 kDa precursor molecules to the mature 72 kDa glycoprotein could be observed (Fig. 5c⇑, lanes 7–12), only increasing amounts of the presumably mature intracellular 54 kDa glycoprotein E2 form were detected by the calf anti-BVDV serum during the chase period (Fig. 5c⇑, lanes 1–6). Only the 94 kDa intracellular homodimer appeared after electrophoresis under non-reducing conditions (not shown). Conversion to the 57 or 101 kDa forms found in virions was not observed. The same result was obtained with the anti-E2 MAb cocktail. The observation that the calf anti-BVDV serum and the anti-E2 MAb cocktail did not react with the in vitro- synthesized, core-glycosylated E2 protein and the failure of these antibodies to precipitate E2 glycoprotein precursor molecules in pulse–chase experiments are in accordance with the conclusion that intracellular transport is required for proper folding and/or modification of the BHV-1-expressed E2 glycoprotein. Longer exposure of the gel shown in Fig. 5(c)⇑ gave no indication of intracellular formation of the 57 kDa glycoprotein E2 form found in BHV-1/SigE2_syn virions, supporting the assumption that this modification is connected with the release of recombinant BHV-1 virus particles.

The carbohydrate composition of the E2 glycoprotein was analysed by incubation of immunoprecipitated E2 with N-glycosidase F and neuraminidase together with O-glycosidase. The electrophoretic mobility of the 54 kDa E2 glycoprotein (Fig. 5b⇑, lane 1) appeared to be unaffected after incubation with neuraminidase together with O-glycosidase (Fig. 5b⇑, lane 3). Digestion with N-glycosidase F resulted in a reduction of the apparent molecular mass to 43 kDa, a size comparable to that of the in vitro-translation product shown in Fig. 5(a)⇑, lane 1. However, since the BHV-1-expressed E2 glycoprotein probably lacks the signal peptide, the removal of the N-glycans should have resulted in an apparent molecular mass of about 40 kDa. Whether this discrepancy is due to additional modifications, the presence of O-glycosidase- resistant O-linked carbohydrates (Kurilla et al., 1995⇓ ), dimerization or folding after synthesis in infected cells needs to be analysed. The origin of the distinct bands migrating above and below the E2 glycoprotein in each lane in Fig. 5(b) ⇑ is not clear. They may represent breakdown products of high molecular mass proteins that bind non-specifically to the Staphylococcus aureus cells used for immunoprecipitation (Keil et al., 1996⇓ ).

Effect of E2 expression on entry and release of BHV-1

The penetration behaviour of BHV-1/SigE2_syn was analysed to test whether the presence of the E2 glycoprotein influenced the entry of virions into the target cells. Fig. 6⇓ shows the result of a representative experiment. About 50% of infectious wild-type BHV-1 particles were protected from low pH-mediated inactivation after about 15 min, whereas entry of BHV-1/SigE2_syn virions required 60 min for 50% penetration and more than 2 h to enter the cells completely. Thus, association of the E2 glycoprotein with BHV-1 hindered the penetration process significantly, an effect that was not seen with recombinant virions containing BRSV glycoprotein G (Kühnle et al., 1998⇓ ).

Replication and release of BHV-1/SigE2_syn and wild-type BHV-1 in MDBK-Bu100 cells is illustrated in Fig. 7⇓, which shows single-step growth curves from a representative experiment. The generation of intra- and extracellular infectivity in BHV-1/SigE2_syn-infected cells was different to that found for wild-type BHV-1. In BHV-1/SigE2_syn -infected cells, release of virions appears to have been impaired, since infectivity in the supernatants only surpassed that found in infected cells after about 32 h, whereas a similar ratio was attained by wild-type BHV-1 before 20 h p.i. This result may indicate interference of the E2 glycoprotein with the egress of BHV-1. However, we have always observed a delay in release of recombinant BHV- 1 virions that express a heterologous glycoprotein (Kühnle et al., 1996⇓ , 1998⇓ ; G. M. Keil, unpublished results). The delay might therefore be related to the incorporation of de novo-synthesized glycoproteins into the cellular compartments involved in BHV-1 glycoprotein processing and viral envelope formation.

In conclusion, our results demonstrate that BHV-1 is a suitable vector for expression of the BVDV E2 membrane glycoprotein in a conformation that is at least similar to that of BVDV-expressed E2. They show that incorporation of the E2 glycoprotein into BHV-1 particles does not require herpesvirus-specific signal sequences. With regard to vector vaccine development, the presence of the E2 glycoprotein in recombinant BHV-1 virions raises the possibility of using BHV-1 expressing BVDV E2 as a live or inactivated vaccine.