SGM Journals
Research Article

Spontaneous excision of BAC vector sequences from bacmid-derived baculovirus expression vectors upon passage in insect cells

Gorben P. Pijlman, Jessica E. van Schijndel, Just M. Vlak

  • Wageningen University, Laboratory of Virology, Binnenhaven 11, 6709 PD, Wageningen, The Netherlands
  • Correspondence
    Just Vlak
    just.vlak{at}wur.nl

    • Journal of General Virology 2003; 84(10):2669 · https://doi.org/10.1099/vir.0.19438-0

    View at publisher PubMed

    Abstract

    Repeated baculovirus infections in cultured insect cells lead to the generation of defective interfering viruses (DIs), which accumulate at the expense of the intact helper virus and compromise heterologous protein expression. In particular, Autographa californica multicapsid nucleopolyhedovirus (AcMNPV) DIs are enriched in an origin of viral DNA replication (ori) not associated with the homologous regions (hrs). This non-hr ori is located within the coding sequence of the non-essential p94 gene. We investigated the effect of a deletion of the AcMNPV non-hr ori on the heterologous protein expression levels following serial passage in Sf21 insect cells. Using homologous ET recombination in E. coli, deletions within the p94 gene were made in a bacterial artificial chromosome (BAC) containing the entire AcMNPV genome (bacmid). All bacmids were equipped with an expression cassette containing the green fluorescent protein gene and a gene encoding the classical swine fever virus E2 glycoprotein (CSFV-E2). For the parental (intact) bacmid only, a strong accumulation of DIs with reiterated non-hr oris was observed. This was not observed for the mutants, indicating that removal of the non-hr ori enhanced the genetic stability of the viral genome upon passaging. However, for all passaged viruses it was found that the entire BAC vector including the expression cassette was spontaneously deleted from the viral genome, leading to a rapid decrease in GFP and CSFV-E2 production. The rationale for the (intrinsic) genetic instability of the BAC vector in insect cells and the implications with respect to large-scale production of proteins with bacmid-derived baculoviruses are discussed.

    • ↵†Present address: Sir Albert Sakzewski Virus Research, Royal Children's Hospital Centre, Herston Rd, Herston, 4029 QLD, Australian

    INTRODUCTION

    The baculovirus–insect cell expression system is versatile and widely used for the high-level production of heterologous (eukaryotic) proteins (Possee, 1997). The proteins are often properly folded and post-translationally modified to obtain similar biological activities to their authentic counterparts (Vialard et al., 1995). Advances have been made over the last decade to make more convenient and speed up the process of generating baculovirus recombinants, which were initially made via homologous recombination between viral and transfer vector DNA in insect cells (Kitts, 1996). A system based on site-specific transposon-mediated insertion of foreign genes into an infectious baculovirus genome, propagated as a bacterial artificial chromosome (BAC) in Escherichia coli (bacmid), has reduced the time taken to obtain pure recombinants from 1–2 months to 1–2 weeks (Luckow et al., 1993). However, for both classical and bacmid recombinants, the rapid generation and accumulation of defective interfering viruses (DIs) upon passage in infected cells is still a major obstacle for the efficient large-scale production of virus and heterologous proteins in cell culture systems (Kool et al., 1991; Wickham et al., 1991). This phenomenon is known as the passage effect and causes a sharp drop in protein production upon serial passage of baculoviruses in cultured insect cells (reviewed by Krell, 1996).

    Many reports have shown that genomic deletions and/or insertions of foreign DNA into the viral genome readily occur upon baculovirus infection in cell culture. For example, DIs with deletions of approximately 43 % of the Autographa californica multicapsid nucleopolyhedovirus (AcMNPV) genome (d43 DIs) are rapidly generated (Pijlman et al., 2001) and subsequently accumulate in cell culture (Kool et al., 1991). Furthermore, DIs with reiterations of small viral sequences become abundant in later stages of passaging (Kool et al., 1993; Lee & Krell, 1992). These cis-acting sequences were subsequently identified as putative origins of viral DNA replication (oris) by transient replication assays. Ori activity in baculoviruses is associated with the homologous repeated regions (hrs) (Lu et al., 1997), which are scattered throughout the viral genome and can also act as transcriptional enhancers (Friesen, 1997). The presence of hrs is a common feature of baculoviruses, but they are also found in other large circular DNA viruses such as nimaviruses and ascoviruses (Van Hulten et al., 2001; Bigot et al., 2000), implying an important role for these interspersed repetitive sequences in viral DNA replication.

    In a detailed study on DI formation following serial passage of AcMNPV in Spodoptera frugiperda (Sf21) insect cell culture (Lee & Krell, 1992, 1994), a specific 2·8 kb AcMNPV sequence predominated in later passages. This fragment contained a viral sequence located on the HindIII-K restriction fragment of AcMNPV. In an independent study (Kool et al., 1994), it was demonstrated that this AcMNPV HindIII-K fragment exhibited a strong ori activity in transient replication assays. The HindIII-K sequence was designated non-hr ori because it did not contain hr sequences. The AcMNPV non-hr ori is located within the open reading frame (ORF) encoding the p94 gene, which is an early gene of unknown function (Friesen & Miller, 1987) and has probably co-evolved with the adjacent apoptosis-inhibiting gene, p35 (Clem et al., 1994). A related baculovirus, Bombyx mori nucleopolyhedrovirus (BmNPV), lacks a p94-homologous ORF, but has retained 151 bp of the p94 gene containing the essential non-hr ori regions II and III as identified by Kool et al. (1994). This thus suggests that the non-hr ori is somehow involved in baculovirus replication. Non-hr oris are identified in many other baculoviruses and share structural similarities rather than sequence homology (Heldens et al., 1997; Pearson et al., 1993; Huang & Levin, 1999; Luque et al., 2001; Hu et al., 1998; Jehle, 2002). In Spodoptera exigua multicapsid nucleopolyhedrovirus (SeMNPV), the non-hr ori was shown to be non-essential for virus replication in vitro and in vivo. Deletion of the non-hr ori even led to enhanced genome stability by preventing DIs from becoming predominant upon passage (Pijlman et al., 2002).

    Since it has been clearly shown that the non-hr origin of DNA replication of AcMNPV accumulates in DIs upon serial passaging, the question is whether deletion of the non-hr ori can prevent the accumulation of DIs and therefore suppress the passage effect. A bacmid-mutagenesis approach was used to make site-specific deletions in the p94 coding sequence. On analysis of the AcMNPV mutants by serial undiluted passage in Sf21 insect cells, we observed that the viruses became more stable, but that the bacmid insertion containing the foreign genes was specifically lost.

    METHODS

    Cells and virus.

    Spodoptera frugiperda (Sf-AE-21) cells (Vaughn et al., 1977) were maintained at 27 °C in Grace's supplemented insect medium (Gibco-BRL) with 10 % foetal calf serum (FCS; Gibco-BRL). Routine cell culture maintenance was performed according to established procedures (Summers & Smith, 1987; King & Possee, 1992). Isolation and transfection of bacmid DNA to Sf21 cells was carried out according to the Bac-to-Bac baculovirus expression system manual (Gibco-BRL). The bacmid-derived budded virus (BV) inoculum was defined as passage 1 (P1) and was used to initiate undiluted serial passaging in Sf21 cells at an initial m.o.i. of 100 TCID50 units per cell. Serial undiluted passaging was carried out as previously described (Pijlman et al., 2001). Infectious BV titres were determined using the endpoint dilution assay (Vlak, 1979).

    Deletion mutagenesis by ET recombination in E. coli.

    For deletion mutagenesis of the p94 coding sequence of the AcMNPV bacmid, 74–76 bp primers were designed with 50 bp 5′ ends flanking the deletion target region on the AcMNPV genome (Table 1⇓). The 3′ ends of the primers annealed to the chloramphenicol gene of pBeloBAC11 (Shizuya et al., 1992) from nt 735 to 1671. PCR on pBeloBAC11 was performed using the Expand long-template PCR system (Roche) according to the manufacturer's protocol, giving a product of 1048 bp. The PCR product was purified using the High pure PCR purification kit (Roche), digested with DpnI to eliminate residual pBeloBAC11 template, phenol/chloroform extracted and ethanol precipitated. For ET recombination, 70 ml LB medium was inoculated with 0·7 ml of an overnight culture of E. coli DH10β containing the AcMNPV bacmid and homologous recombination helper plasmid pBAD-αβγ (Muyrers et al., 1999). At an OD600 of 0·1–0·15, ET protein expression from pBAD-αβγ was induced by the addition of 0·7 ml 10 % l-arabinose. The cells were harvested at an OD600 of 0·3–0·4 and made electrocompetent by three washes with ice-cold 10 % glycerol. The cells were transformed with ∼0·5 μg purified PCR product in 2 mm electroporation cuvettes (Eurogentec) using a Biorad Gene Pulser (2·3 kV, 25 μF, 200 Ω). The cells were resuspended in 1 ml LB medium and incubated for 1 h at 37 °C and subsequently spread on agar plates containing kanamycin and chloramphenicol. Colonies were picked and screened by restriction analysis and PCR.

    Table 1.

    Oligonucleotides used for site-specific genomic deletions by ET recombination

    Donor plasmid construction for green fluorescent protein (GFP) and classical swine fever virus E2 glycoprotein (CSFV-E2) expression.

    To introduce the GFP and CSFV-E2 genes in the (mutant) AcMNPV bacmids, a donor plasmid pFBD-GFPE2 was constructed. First, the red-shifted GFP gene (Davis & Vierstra, 1998) was cloned as an EcoRI/HindIII fragment into the pFastBac-DUAL vector (Invitrogen) under the polyhedrin promoter to generate pFDB-GFP. To create a translational start codon for CSFV-E2, pSeMO7 (Dai et al., 2000) was digested with BamHI and XbaI and a linker was inserted. This linker was made with oligonucleotides DZ195 (5′-GATCATCGATTATGGATCCT-3′) and DZ196 (5′-CTAGAGGATCCATAATCGAT-3′) and is BamHI compatible, contains a ClaI site (italics) followed by a ATG–BamHI fusion (underlined) and an XbaI site. The CSFV-E2 gene was then cloned as a BamHI fragment from pAcE2 (Van Oers et al., 2001) behind the ATG of the inserted linker. The CSFV-E2 gene was subcloned as a ClaI fragment into pBluescript and then cloned downstream of the p10 promoter as a XhoI–XbaI fragment into pFBD-GFP digested with XhoI and NheI, generating pFBD-GFPE2. Before transposition, E. coli DH10β containing the recombinant AcMNPV bacmids were transformed with transposition helper plasmid pMON7124 (Luckow et al., 1993). The protocol from the Bac-to-Bac manual (Gibco-BRL) was followed to transpose the GFP and CSFV-E2 genes from pFBD-GFPE2 into the attTn7 transposon integration site of the AcMNPV bacmids.

    Viral DNA isolation, Southern hybridization, molecular cloning and sequencing.

    Intracellular viral DNA was isolated as previously described (Summers & Smith, 1987). Digested viral DNA was run overnight in ethidium bromide-stained 0·6 % agarose gels and Southern blotting was performed by standard capillary upward blotting (Sambrook et al., 1989) using Hybond-N (Amersham Pharmacia) filters. As a DNA size marker, λ DNA digested with EcoRI/HindIII/BamHI was used. Random-primed DNA probes for Southern hybridization were made using the DIG non-radioactive nucleic acid labelling and detection system (Roche). For the non-hr ori probe, a PCR product (1036 bp) of the AcMNPV non-hr ori was made with primers DZ123 (5′-TGCGGCCAGGTTTTGTAGAATG-3′; nt 114056–114077; Ayres et al., 1994) and DZ124 (5′-GCATGGAACGCGTTTGTCAC-3′; nt 115072–115091), purified using the High pure PCR purification kit (Roche) and DIG-labelled overnight. For the BAC vector probe, BAC-Bsu36I (Pijlman et al., 2002) was DIG-labelled overnight. Hybridization and colorimetric detection with NBT/BCIP (Roche) were performed according to the manufacturer's recommendations. Hypermolar viral HindIII bands were cut from the gel, purified with the Matrix gel extraction system (Marlingen) and cloned into pBluescript by electrotransformation of E. coli DH5α using standard methods (Sambrook et al., 1989). Automatic sequencing was performed at Baseclear, The Netherlands. Sequence analyses were performed using blast (Altschul et al., 1997). In silico reassembling of bacmid sequences and computational predictions of restriction digests were done using the Lasergene dnastar package.

    SDS-PAGE and immunodetection.

    Protein samples were analysed in 12 % SDS-PAGE gels as described in Sambrook et al. (1989). Protein masses were determined using the low molecular mass protein marker (Amersham Pharmacia Biotech). Semi-dry blotting was performed onto an Immobilon-P membrane (Millipore) using a Tris/glycine buffer (25 mM Tris base, 192 mM glycine, 10 %, v/v, methanol, pH 8·3). Immobilon-P membranes were blocked in 2 % low-fat milk powder (Campina, The Netherlands) in TBS (0·2 M NaCl, 50 mM Tris/HCl, pH 7·4). The marker was visualized on the membrane by Ponceau-red staining (Sambrook et al., 1989). Immunodetection of CSFV-E2 was performed by incubation with a monoclonal antibody (mAb A18, Intervet International B.V.) diluted 1 : 10 000 in TBS with 1 % low-fat milk powder for 1 h at room temperature. Subsequently, anti-mouse antibody conjugated with horseradish peroxidase (Amersham) was used at a concentration of 1 : 10 000 and detection was performed with an Enhanced Chemiluminescent-light (ECL) Detection Kit (Amersham).

    RESULTS

    Construction and serial passage of recombinant viruses

    To investigate whether a deletion of the AcMNPV non-hr ori could prevent the rapid accumulation of DIs and thereby improve the expression of heterologous genes in infected insect cells upon prolonged passaging, three AcMNPV mutants were constructed. Substitutions with a chloramphenicol resistance gene (CmR) were made in the p94 coding sequence by ET recombination of an AcMNPV bacmid in E. coli (Fig. 1⇓A). In AcΔCp94, the essential domains of the non-hr ori (according to Kool et al., 1994) located within the C terminus of p94 were deleted, leaving the N terminus of p94 intact. As a control for a putative effect of a p94 disruption, two other mutants were constructed. AcΔNp94 lacked the N terminus of p94 and thus retained the functional domains of the non-hr ori, whereas AcΔp94 lacked the whole p94 gene including the non-hr ori.

    Figure image not available in archive
    Fig. 1.

    (A) Schematic overview of (mutant) AcMNPV bacmids. Nucleotide positions according to the complete AcMNPV genome sequence (Ayres et al., 1994) are indicated on top. The genomic organization including endonuclease restriction sites of a region containing p94 and 35K genes and hr5 is depicted for the intact AcMNPV bacmid AcGFPE2. Three mutants, designated AcΔCp94, AcΔNp94, AcΔp94, were constructed by homologous ET recombination in E. coli in which CmR replaced the p94 C terminus, the p94 N terminus and the entire p94 gene, respectively. All viruses were equipped with an expression cassette containing the GFP and CSFV-E2 genes under the control of the AcMNPV polyhedrin and p10 promoters, respectively. (B) Infectious budded virus titres upon serial passage in Sf21 cells. TCID50 values ml−1 were based on GFP expression. ▪, AcGFPE2; ▵, AcΔCp94; ○, AcΔNp94; ⧫, AcΔp94. (C) Immunodetection intracellular CSFV-E2 levels upon passage in Sf21 cells. Molecular mass markers (kDa) are indicated on the left.

    The parental and the three mutant AcMNPV bacmids were equipped with an expression cassette containing GFP under the control of the polyhedrin promoter to visualize infection and a gene encoding the classical swine fever virus E2 glycoprotein (CSFV-E2) under the control of the p10 promoter to measure protein production. Recombinant bacmid DNA was transfected into insect cells, BV titres were determined and serial undiluted passage was carried out with an initial m.o.i. of 100. Virus titres (monitored by GFP expression) were determined during serial passage (Fig. 1B⇑). As expected, the GFP-based titres for AcGFPE2 dropped rapidly from almost 109 at P0 to less than 107 at P10 and to almost 106 TCID50 ml−1 at P20. Similarly, GFP expression for the AcΔNp94 mutant (containing the non-hr ori) also dropped to lower levels upon passaging, although less rapidly than for AcGFPE2. In sharp contrast to what was expected, namely that a non-hr ori deletion would prevent the formation of DIs and therefore result in enhanced protein expression, a similar rapid decrease in GFP expression was observed for the non-hr ori deletion mutants AcΔCp94 and AcΔp94. Coinciding with this overall drop in GFP-based virus titres, a major decrease in the production of CSFV-E2 following virus passage was observed by Western blot analysis (Fig. 1C⇑) for AcGFPE2, AcΔCp94 and AcΔp94 and to a lesser extent for AcΔNp94. In summary, this experiment showed that the presence or absence of the non-hr ori was not correlated with the level of recombinant protein expression upon serial passage.

    Analysis of genomic alterations upon serial passage

    To investigate the putative formation of DIs and to find the molecular basis for the unexpected rapid decrease in protein production upon serial passage of the parental virus and the three mutants, intracellular viral DNA of P1, P10 and P20 was subjected to digestion with HindIII (Fig. 2⇓). For all viruses, two bands of 10·9 and 9·2 kb were submolar in P20 with respect to the other fragments. The genetic content of these bands was investigated in more detail (see Fig. 4⇓). Typically, novel fragments of 3·2, 1·8 and 1·6 kb appeared in P20 of AcGFPE2 (Fig. 2A⇓), but not in the DNA preparations of the other (mutant) viruses (Fig. 2B⇓–D). These fragments accumulated relative to the genomic HindIII fragments of the parental virus, suggesting that these sequences were part of DI molecules. According to the hypothesis, we investigated whether these novel fragments contained the non-hr ori. Therefore, the viral DNA was digested and transferred to a membrane and Southern hybridized with a DIG-labelled non-hr ori probe (Fig. 2⇓). The original non-hr ori-containing HindIII-K fragment in AcGFPE2 of 2971 bp hybridized to the non-hr ori probe (Fig. 2A⇓). Moreover, the results showed that the novel 3·2, 1·8 and 1·6 kb bands in AcGFPE2 hybridized strongly with the non-hr ori probe (Fig. 2A⇓), indicating that the accumulated sequences indeed contained the non-hr ori. As expected, the non-hr ori probe did not hybridize to AcΔCp94 (Fig. 2B⇓) and AcΔp94 (Fig. 2D⇓), but did hybridize to the fragment with the CmR insertion in AcΔNp94 (Fig. 2C⇓). Thus, deletion of the non-hr ori (in AcΔCp94 and AcΔp94) or a non-hr ori flanking sequence (in AcΔNp94) prevented the generation and accumulation of DIs, but apparently did not prevent the decrease in recombinant protein production.

    Figure image not available in archive
    Fig. 2.

    Restriction analysis of intracellular viral DNA upon serial passage in Sf21 cells and Southern detection with a non-hr ori probe. Viral DNA isolated from Sf21 cells infected with AcGFPE2, AcΔCp94, AcΔNp94 and AcΔp94 at P1, P10 and P20 was subjected to digestion with HindIII. Restriction fragments containing the non-hr ori and/or a CmR insertion are indicated with an asterisk. The two submolar fragments of 10·9 and 9·2 kb are indicated with an arrow on the right. Novel, hypermolar fragments of 3·2, 1·8 and 1·6 kb hybridizing to the non-hr ori probe are indicated with arrows on the right of the AcGFPE2 Southern blot.

    Sequence and replicative form of hypermolar non-hr ori fragments

    Subsequently it was investigated whether the novel 3·2, 1·8 and 1·6 kb HindIII fragments hybridizing to the non-hr ori originated from ordinary deletions in the 2971 bp HindIII-K fragment, or, alternatively, were derived from non-hr ori mini-circles or concatenated non-hr ori repeats within DI molecules, as previously found with SeMNPV (Pijlman et al., 2002). Viral DNA from AcGFPE2 P20 was digested with BamHI, HindIII and XhoI and Southern hybridized with the same non-hr ori probe as described above (Fig. 3⇓A). The 3·2 kb fragment appeared in the BamHI, HindIII and XhoI digests, whereas the 1·8 and 1·6 kb fragments appeared in the HindIII and XhoI digests only (Fig. 3A⇓). This indicated that the novel non-hr ori containing fragments must appear in a circular form or as tandem repeats in larger concatemers. The HindIII fragments of 3·2, 1·8 and 1·6 kb were subsequently cloned and sequenced (Fig. 3B⇓). As expected from the Southern blots, all fragments contained the five essential domains (I–V) of the non-hr ori sequence between the HindIII and XhoI sites. The sequence results also demonstrated the presence of a junction site between sequences from either side of the non-hr ori (Fig. 3B⇓, asterisks). The sequence overlaps at the junction sites may indicate the imprints of the recombination steps involved in the formation of these circular or concatenated molecules.

    Figure image not available in archive
    Fig. 3.

    Genetic organization of novel hypermolar non-hr ori fragments. (A) Intracellular viral DNA of AcGFPE2 P20 was digested with BamHI, HindIII and XhoI and Southern detected with a DIG-labelled non-hr ori probe. (B) Schematic representation of the (HindIII-linearized) sequences of the hypermolar non-hr ori fragments of 3·2, 1·8 and 1·6 kb, compared with the genomic organization of the parental virus. Nucleotide positions according to the complete AcMNPV genome sequence (Ayres et al., 1994) are indicated on top, as well as endonuclease restriction sites and genes encoding alk-exo and p94. The five essential domains of the non-hr ori (after Kool et al., 1994) are represented by a white bar below the p94 gene. The sequence to the left of the HindIII site is depicted as a grey bar, whereas the sequence to the right of the XhoI site is depicted as a black bar. Corresponding sequences in the hypermolar fragments have the same colours. Nucleotide positions and sequence overlaps are indicated at the junction sites (asterisks).

    Specific deletion of BAC vector sequences upon serial passage

    To find an explanation for the overall decrease in protein production, the genetic content of the 10·9 and 9·2 kb HindIII fragments, which became submolar upon serial passage of all viruses (Fig. 2A⇑–D), was determined. The fragments were mapped on computer-predicted digests of in silico reassembled bacmid sequences. Both 10·9 and 9·2 kb HindIII fragments were located on the BAC vector insertion, which contains a kanamycin resistance gene, a LacZ-mini-attTn7 cassette and a bacterial mini-F replicon (Luckow et al., 1993). More specifically, the 10·9 kb HindIII fragment was located entirely on the mini-F replicon and the 9·2 kb HindIII fragment on the kanamycin resistance gene (Fig. 4⇓B). To map the deletions of the BAC vector sequences in more detail, a PstI digest was performed and a DIG-labelled probe specific for the BAC vector (Fig. 4B⇓) was used specifically to distinguish the BAC-derived fragments from co-migrating genomic fragments (Fig. 4A⇓). The results showed that all PstI fragments in the BAC vector from the kanamycin resistance gene (KanR) to the end of the mini-F replicon were deleted in AcGFPE2, AcΔCp94 and AcΔp94 upon passage and were submolar in AcΔNp94 (Fig. 4A⇓). In AcΔp94, only the PstI fragment of 1240 nt, which is localized in the kanamycin resistance gene, was not fully deleted. This indicated that the moment of deletion and the actual size of the BAC vector deletions may differ between separate experiments and that there is probably not a well-defined deletion mechanism.

    Figure image not available in archive
    Fig. 4.

    Restriction mapping of specific deletions in the BAC vector. (A) Southern detection of a PstI digest with a probe specific for the BAC vector. Sizes of BAC vector-specific restriction fragments are indicated on the left and correspond to the sizes in (B). (B) Schematic representation of the organization of the bacterial expression cassette containing the GFP and CSFV-E2 genes under the control of AcMNPV polyhedrin and p10 promoters, respectively. Sizes of HindIII, PstI, XhoI and KpnI restriction fragments (nt) are indicated. (C) Restriction analysis of intracellular viral DNA with XhoI and KpnI. Genomic XhoI and KpnI fragments are indicated on the right with letters according to the complete genome sequence of AcMNPV (Ayres et al., 1994). The increased size of the XhoI-J fragment due to the CmR insertion is indicated with an asterisk (J*). In AcΔp94 the XhoI-J fragment is not present, because of a disruption of the XhoI site (see also Fig. 1). Sizes of BAC vector-specific restriction fragments are indicated on the left and correspond to the sizes in (B).

    To investigate whether, in addition to a deletion of the BAC vector parts, the expression cassette containing GFP and CSFV-E2 genes was also deleted, restriction digests with XhoI and KpnI were carried out (Fig. 4C⇑). Again, the results showed that all XhoI and KpnI fragments derived from the BAC vector as well as from the GFP and CSFV-E2 genes (sizes of fragments on the left in Fig. 4C⇑ correspond to sizes on the physical map in Fig. 4B⇑) were deleted upon passage in AcGFPE2, AcΔCp94 and AcΔp94 and were submolar in AcΔNp94 (Fig. 4C⇑). In contrast to the deletions in the BAC vector and the expression cassette, the genomic XhoI-J, -K, -L and KpnI-E fragments of viral origin (named after digests of the complete AcMNPV genome sequence; Ayres et al., 1994) remained at nearly equimolar levels in the three mutants. In AcGFPE2, however, all fragments became submolar as a consequence of predominating non-hr ori concatemers with 3·2, 1·8 and 1·6 kb units (arrows, Fig. 4C⇑). The results clearly indicated that, irrespective of the presence or absence of the non-hr ori, sequences from the BAC vector and the expression cassette were systematically deleted and that sequences of viral origin were retained upon passage in insect cells.

    DISCUSSION

    In this paper we investigated whether a deletion of the non-hr ori from the AcMNPV genome could prevent the accumulation of DIs. Three mutants were constructed with deletions in the p94 coding sequence, of which the C terminus comprises the non-hr ori. Similar to SeMNPV (Pijlman et al., 2002), the results showed that the AcMNPV non-hr ori is dispensable for in vitro replication and confirmed that p94 is not essential for AcMNPV replication in cultured insect cells (Friesen & Nissen, 1990). Also, a p94 gene is not present in the closely related BmNPV, which is a baculovirus with a very high sequence homology to AcMNPV (Kamita et al., 1993). However, BmNPV has retained (essential) parts of the non-hr ori, implying that the non-hr ori may be important for virus replication (Kool et al., 1994).

    Passaging of the parental AcMNPV bacmid (AcGFPE2) resulted in the predominance of DI molecules containing reiterated non-hr oris (Figs 2 and 3⇑⇑). Sequence overlaps on the junction sites of the concatenated non-hr ori molecules suggested that they were generated by homologous recombination during viral DNA replication (Pijlman et al., 2002). The generation of non-hr ori DIs was also expected to occur for AcΔNp94, because this mutant only lacks the p94 N-terminal part but still contains the five non-hr ori subdomains, which are required for optimal ori activity (Kool et al., 1994; Fig. 1A⇑). However, this was not the case. It may be that auxiliary sequences located in the adjacent N-terminal part of p94 are necessary either for the excision of non-hr oris from the genome, or are responsible for a higher ori activity, thereby giving the DIs a stronger replicative advantage. The latter hypothesis is supported by data from transient replication assays by Kool et al. (1994), who showed that plasmids containing the entire sequence from HindIII to EcoRV (see Fig. 1A⇑) had a slightly greater replication ability than the smaller HindIII–XhoI region, which is retained in the AcΔNp94 mutant. Thus, the AcMNPV non-hr ori is not essential for virus replication in vitro, and deletion of the non-hr ori and/or flanking auxiliary sequences prevent the accumulation of DIs enriched in non-hr oris upon serial passage in insect cells.

    In spite of the fact that the formation of non-hr ori DIs did not occur in the three recombinants, BV titres (based on GFP expression) and CSFV-E2 production decreased rapidly upon passaging of all viruses (Fig. 1B, C⇑). We demonstrated that the non-essential BAC vector including the expression cassette was spontaneously deleted from the viral genome upon passage in insect cells (Fig. 4⇑). This cassette comprises a bacterial mini-F replicon (or BAC vector), two antibiotic resistance genes and two foreign genes (CSFV-E2 and GFP) under the control of baculovirus p10 and polyhedrin promoters. Spontaneous baculovirus mutants carrying this deletion quickly became predominant upon passage, which explained the drop in foreign protein production levels.

    Instability of mini-F plasmids, which are also known as bacterial artificial chromosomes (BACs), in eukaryotic cells has been reported to occur in several other cases. An infectious clone of the pseudorabies virus was maintained as a stable BAC in E. coli, but reconstitution of the virus led to the spontaneous deletion of the BAC vector insertion upon transfection of mammalian cells (Smith & Enquist, 1999). Approximately 5–6 kb of flanking viral sequence was deleted along with the BAC vector sequence. In contrast, when the BAC vector was inserted at a different locus, the virus was stable, suggesting that the location of BAC vector insertion might also be important (Smith & Enquist, 2000). Wagner et al. (1999) showed that during construction of a mouse cytomegalovirus (CMV) BAC, overlength genomes were not stable in mammalian cells. To overcome this problem, they designed duplicated viral sequences flanking the BAC vector insertion, allowing spontaneous excision by homologous recombination. Adler et al. (2001) further showed that excision of BAC vector sequences (by Cre-lox recombination) from cloned MHV-68 genomes was critical for reconstitution of wild-type properties. Similarly, insertion of the BAC vector in CMV requires deletion of non-essential genes, because CMVs only tolerate 5 kb of additional sequence in their genomes (Brune et al., 2000).

    Most likely, these properties of herpesvirus BACs are the result of physical limitations of the virus capsid, which can only package a genome of a defined (maximum) size. Maximum packaging capacity is also observed for other DNA viruses. An overlength of only 5 % leads to unstable adenovirus and Epstein–Barr virus genomes (Bett et al., 1993; Bloss & Sugden, 1994). For baculoviruses, a maximum packaging capacity may also exist, although the rod-shaped baculovirus nucleocapsids are presumed to be more flexible with respect to DNA content as they contain genomes of up to almost 180 kb (e.g. Xestia c-nigrum granulovirus; Hayakawa et al., 1999) and allow inserts of up to 25 kb (Roosien et al., 1986). Still, the occurrence of spontaneous (major) deletions in baculoviruses is a general phenomenon. In SeMNPV, deletions of up to 25 kb of non-essential sequences are routinely observed upon infection of cultured insect cells (Heldens et al., 1996; Dai et al., 2000; Pijlman et al., 2002) and are located in the largest region between two adjacent hrs (SeMNPV hr1 and hr2; IJkel et al., 1999). In AcMNPV, deletions are frequently found in the EGT/DA26 locus (Kumar & Miller, 1987), which is located in the middle of AcMNPV inter-hr region hr1–2 (Ayres et al., 1994). Since hrs are believed to be involved as ori in viral DNA replication, we hypothesize that the occurrence of genomic deletions is more likely in regions with a low ori density. This may explain why deletions in the BAC vector sequence are likely to occur. Yet the BAC vector itself may also display a certain intrinsic genetic instability. Alternatively, the heterologous gene may confer a certain level of toxicity to the infected cells, thereby creating an added selection pressure against intact bacmids. However, toxicity in insect cells has never been observed with CSFV-E2, which is a commercialized baculovirus expression product used as the major constituent of a marker vaccine against classical swine fever (Van Rijn et al., 1999). In addition, bacmids equipped with an expression cassette not containing CSFV-E2 also showed specific loss of BAC vector sequences (G. P. Pijlman, unpublished results).

    In this paper we have shown that reconstitution in insect cells of infectious baculovirus from a bacmid is accompanied by genetic instability of BAC vector sequences. Recently we have obtained similar results with SeMNPV bacmid-derived viruses in cultured insect cells (G. P. Pijlman, unpublished results). Once the instability is removed by spontaneous deletion of the (non-essential) BAC vector during viral DNA replication, a more stable virus is generated, which predominates subsequent passages. The present observations may constitute a major concern for the utilization of bacmid-derived baculoviruses for the large-scale production of heterologous proteins, especially in insect-cell bioreactors involving many virus passages. Although the generation of mutant baculoviruses by the classical method is more time-consuming than the generation of a recombinant bacmid, it may yield a virus with greater stability. An improvement for the bacmid strategy would be to introduce the heterologous gene(s) at a more stable locus remote from the BAC vector insertion. Alternatively, a bacmid could be developed in which the BAC vector is deliberately excised (using Cre-lox recombination) upon replication in insect cells, while leaving the introduced heterologous gene(s) intact.

    Acknowledgments

    The authors would like to thank Dr Paul D. Friesen for the donation of classical p94 recombinants, on which we based the deletions in our mutant bacmids. Dr Douwe Zuidema, Dr Dirk E. Martens and Professor Dr Rob W. Goldbach are acknowledged for their continued interest and helpful discussions. This research was supported by the Technology Foundation STW (grant no. 790-44-730), applied science division of NWO and the technology programme of the Dutch Ministry of Economic Affairs.

    References

    1. Adler, H., Messerle, M. & Koszinowski, U. H. (2001). Virus reconstituted from infectious bacterial artificial chromosome (BAC)-cloned murine gammaherpesvirus 68 acquires wild-type properties in vivo only after excision of BAC vector sequences. J Virol 75, 5692–5696.
    2. Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25, 3389–3402.
    3. Ayres, M. D., Howard, S. C., Kuzio, J., Lopez Ferber, M. & Possee, R. D. (1994). The complete DNA sequence of Autographa californica nuclear polyhedrosis virus. Virology 202, 586–605.
    4. Bett, A. J., Prevec, L. & Graham, F. L. (1993). Packaging capacity and stability of human adenovirus type 5 vectors. J Virol 67, 5911–5921.
    5. Bigot, Y., Stasiak, K., Rouleux-Bonnin, F. & Federici, B. A. (2000). Characterization of repetitive DNA regions and methylated DNA in ascovirus genomes. J Gen Virol 81, 3073–3082.
    6. Bloss, T. A. & Sugden, B. (1994). Optimal lengths for DNAs encapsidated by Epstein–Barr virus. J Virol 68, 8217–8222.
    7. Brune, W., Messerle, M. & Koszinowski, U. H. (2000). Forward with BACs – new tools for herpesvirus genomics. Trends Genet 16, 254–259.
    8. Clem, R. J., Robson, M. & Miller, L. K. (1994). Influence of infection route on the infectivity of baculovirus mutants lacking the apoptosis-inhibiting gene p35 and the adjacent gene p94. J Virol 68, 6759–6762.
    9. Dai, X. J., Hajós, J. P., Joosten, N. N., van Oers, M. M., IJkel, W. F. J., Zuidema, D., Pang, Y. & Vlak, J. M. (2000). Isolation of a Spodoptera exigua baculovirus recombinant with a 10·6 kbp genome deletion that retains biological activity. J Gen Virol 81, 2545–2554.
    10. Davis, S. J. & Vierstra, R. D. (1998). Soluble, highly fluorescent variants of green fluorescent protein (GFP) for use in higher plants. Plant Mol Biol 36, 521–528.
    11. Friesen, P. D. (1997). Regulation of baculovirus early gene expression. In The Baculoviruses, pp. 141–170. Edited by L. K. Miller. New York: Plenum Press.
    12. Friesen, P. D. & Miller, L. K. (1987). Divergent transcription of early 35- and 94-kilodalton protein genes encoded by the HindIII K genome fragment of the baculovirus Autographa californica nuclear polyhedrosis virus. J Virol 61, 2264–2272.
    13. Friesen, P. D. & Nissen, M. S. (1990). Gene organization and transcription of TED, a lepidopteran retrotransposon integrated within the baculovirus genome. Mol Cell Biol 10, 3067–3077.
    14. Hayakawa, T., Ko, R., Okano, K., Seong, S. I., Goto, C. & Maeda, S. (1999). Sequence analysis of the Xestia c-nigrum granulovirus genome. Virology 262, 277–297.
    15. Heldens, J. G., van Strien, E. A., Feldmann, A. M., Kulcsar, P., Muñoz, D., Leisy, D. J., Zuidema, D., Goldbach, R. W. & Vlak, J. M. (1996). Spodoptera exigua multicapsid nucleopolyhedrovirus deletion mutants generated in cell culture lack virulence in vivo. J Gen Virol 77, 3127–3134.
    16. Heldens, J. G., Broer, R., Zuidema, D., Goldbach, R. W. & Vlak, J. M. (1997). Identification and functional analysis of a non-hr origin of DNA replication in the genome of Spodoptera exigua multicapsid nucleopolyhedrovirus. J Gen Virol 78, 1497–1506.
    17. Hu, Z. H., Arif, B. M., Sun, J. S., Chen, X. W., Zuidema, D., Goldbach, R. W. & Vlak, J. M. (1998). Genetic organization of the HindIII-I region of the single-nucleocapsid nucleopolyhedrovirus of Buzura suppressaria. Virus Res 55, 71–82.
    18. Huang, J. & Levin, D. B. (1999). Identification and functional analysis of a putative non-hr origin of DNA replication from the Spodoptera littoralis type B multinucleocapsid nucleopolyhedrovirus. J Gen Virol 80, 2263–2274.
    19. IJkel, W. F., van Strien, E. A., Heldens, J. G., Broer, R., Zuidema, D., Goldbach, R. W. & Vlak, J. M. (1999). Sequence and organization of the Spodoptera exigua multicapsid nucleopolyhedrovirus genome. J Gen Virol 80, 3289–3304.
    20. Jehle, J. A. (2002). The expansion of a hypervariable, non-hr ori-like region in the genome of Cryptophlebia leucotreta granulovirus provides in vivo evidence for the utilization of baculovirus non-hr oris during replication. J Gen Virol 83, 2025–2034.
    21. Kamita, S. G., Majima, K. & Maeda, S. (1993). Identification and characterization of the p35 gene of Bombyx mori nuclear polyhedrosis virus that prevents virus-induced apoptosis. J Virol 67, 455–463.
    22. King, L. A. & Possee, R. D. (1992). The baculovirus expression system. Dordrecht: Kluwer.
    23. Kitts, P. A. (1996). Construction of baculovirus recombinants. Cytotechnology 20, 111–123.
    24. Kool, M., Voncken, J. W., van Lier, F. L., Tramper, J. & Vlak, J. M. (1991). Detection and analysis of Autographa californica nuclear polyhedrosis virus mutants with defective interfering properties. Virology 183, 739–746.
    25. Kool, M., van den Berg, P. M., Tramper, J., Goldbach, R. W. & Vlak, J. M. (1993). Location of two putative origins of DNA replication of Autographa californica nuclear polyhedrosis virus. Virology 192, 94–101.
    26. Kool, M., Goldbach, R. W. & Vlak, J. M. (1994). A putative non-hr origin of DNA replication in the HindIII-K fragment of Autographa californica multiple nucleocapsid nuclear polyhedrosis virus. J Gen Virol 75, 3345–3352.
    27. Krell, P. J. (1996). Passage effect of virus infection in insect cells. Cytotechnology 20, 125–137.
    28. Kumar, S. & Miller, L. K. (1987). Effects of serial passage of Autographa californica nuclear polyhedrosis virus in cell culture. Virus Res 7, 335–349.
    29. Lee, H. Y. & Krell, P. J. (1992). Generation and analysis of defective genomes of Autographa californica nuclear polyhedrosis virus. J Virol 66, 4339–4347.
    30. Lee, H. & Krell, P. J. (1994). Reiterated DNA fragments in defective genomes of Autographa californica nuclear polyhedrosis virus are competent for AcMNPV-dependent DNA replication. Virology 202, 418–429.
    31. Lu, A., Krell, P. J., Vlak, J. M. & Rohrmann, G. F. (1997). Baculovirus DNA replication. In The Baculoviruses, pp. 171–216. Edited by L. K. Miller. New York: Plenum Press.
    32. Luckow, V. A., Lee, S. C., Barry, G. F. & Olins, P. O. (1993). Efficient generation of infectious recombinant baculoviruses by site-specific transposon-mediated insertion of foreign genes into a baculovirus genome propagated in Escherichia coli. J Virol 67, 4566–4579.
    33. Luque, T., Finch, R., Crook, N., O'Reilly, D. R. & Winstanley, D. (2001). The complete sequence of the Cydia pomonella granulovirus genome. J Gen Virol 82, 2531–2547.
    34. Muyrers, J. P., Zhang, Y., Testa, G. & Stewart, A. F. (1999). Rapid modification of bacterial artificial chromosomes by ET-recombination. Nucleic Acids Res 27, 1555–1557.
    35. Pearson, M. N., Bjornson, R. M., Ahrens, C. & Rohrmann, G. F. (1993). Identification and characterization of a putative origin of DNA replication in the genome of a baculovirus pathogenic for Orgyia pseudotsugata. Virology 197, 715–725.
    36. Pijlman, G. P., van den Born, E., Martens, D. E. & Vlak, J. M. (2001). Autographa californica baculoviruses with large genomic deletions are rapidly generated in infected insect cells. Virology 283, 132–138.
    37. Pijlman, G. P., Dortmans, J. C. F. M., Vermeesch, A. M. G., Yang, K., Martens, D. E., Goldbach, R. W. & Vlak, J. M. (2002). Pivotal role of the non-hr origin of DNA replication in the genesis of defective interfering baculoviruses. J Virol 76, 5605–5611.
    38. Possee, R. D. (1997). Baculoviruses as expression vectors. Curr Opin Biotechnol 8, 569–572.
    39. Roosien, J., Usmany, M., Klinge-Roode, E. C., Meijerink, P. H. S. & Vlak, J. M. (1986). Heterologous recombination between baculoviruses. In Fundamental and Applied Aspects of Invertebrate Pathology, pp. 389–392. Edited by R. A. Samson, J. M. Vlak & D. Peters. ICIP86 Foundation.
    40. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory.
    41. Shizuya, H., Birren, B., Kim, U. J., Mancino, V., Slepak, T., Tachiiri, Y. & Simon, M. (1992). Cloning and stable maintenance of 300-kilobase-pair fragments of human DNA in Escherichia coli using an F-factor-based vector. Proc Natl Acad Sci U S A 89, 8794–8797.
    42. Smith, G. A. & Enquist, L. W. (1999). Construction and transposon mutagenesis in Escherichia coli of a full-length infectious clone of pseudorabies virus, an alphaherpesvirus. J Virol 73, 6405–6414.
    43. Smith, G. A. & Enquist, L. W. (2000). A self-recombining bacterial artificial chromosome and its application for analysis of herpesvirus pathogenesis. Proc Natl Acad Sci U S A 97, 4873–4878.
    44. Summers, M. D. & Smith, G. E. (1987). A Manual of Methods for Baculovirus Vectors and Insect Cell Culture Procedures. Texas Agricultural Experiment Station Bulletin No. 1555.
    45. Van Hulten, M. C. W., Witteveldt, J., Peters, S., Kloosterboer, N., Tarchini, R., Fiers, M., Sandbrink, H., Lankhorst, R. K. & Vlak, J. M. (2001). The white spot syndrome virus DNA genome sequence. Virology 286, 7–22.
    46. Van Oers, M. M., Thomas, A. A. M., Moormann, R. J. M. & Vlak, J. M. (2001). Secretory pathway limits the enhanced expression of classical swine fever virus E2 glycoprotein in insect cells. J Biotechnol 86, 31–38.
    47. Van Rijn, P. A., van Gennip, H. G. & Moormann, R. J. (1999). An experimental marker vaccine and accompanying serological diagnostic test both based on envelope glycoprotein E2 of classical swine fever virus (CSFV). Vaccine 17, 433–440.
    48. Vaughn, J. L., Goodwin, R. H., Tompkins, G. J. & McCawley, P. (1977). The establishment of two cell lines from the insect Spodoptera frugiperda (Lepidoptera; Noctuidae). In Vitro 13, 213–217.
    49. Vialard, J. E., Arif, B. M. & Richardson, C. D. (1995). Introduction to the molecular biology of baculoviruses. Methods Mol Biol 39, 1–24.
    50. Vlak, J. M. (1979). The proteins of nonoccluded Autographa californica nuclear polyhedrosis virus produced in an established cell line of Spodoptera frugiperda. J Invertebr 34, 110–118.
    51. Wagner, M., Jonjic, S., Koszinowski, U. H. & Messerle, M. (1999). Systematic excision of vector sequences from the BAC-cloned herpesvirus genome during virus reconstitution. J Virol 73, 7056–7060.
    52. Wickham, T. J., Davis, T., Granados, R. R., Hammer, D. A., Shuler, M. L. & Wood, H. A. (1991). Baculovirus defective interfering particles are responsible for variations in recombinant protein production as a function of multiplicity of infection. Biotechnol Lett 13, 483–488.