Rational design of a classical swine fever C-strain vaccine virus that enables the differentiation between infected and vaccinated animals

Journal of Virological Methods 163 (2010) 175–185

Contents lists available at ScienceDirect

Journal of Virological Methods journal homepage: www.elsevier.com/locate/jviromet

Rational design of a classical swine fever C-strain vaccine virus that enables the differentiation between infected and vaccinated animals J. Kortekaas ∗ , R.P.M. Vloet, K. Weerdmeester, J. Ketelaar, M. van Eijk, W.L. Loeffen Virology Division, Central Veterinary Institute of Wageningen University Research Centre, P.O. Box 65, 8200 AB Lelystad, The Netherlands

a b s t r a c t Article history: Received 4 July 2009 Received in revised form 5 September 2009 Accepted 10 September 2009 Available online 19 September 2009 Keywords: Classical swine fever virus Vaccine DIVA Forced virus evolution Genetic stability

The C-strain of the classical swine fever virus (CSFV) is considered the gold standard vaccine for the control of CSF. This vaccine, however, does not enable the serological differentiation between infected and vaccinated animals (DIVA). Consequently, its use can impose severe trade restrictions. The immunodominant and evolutionarily conserved A-domain of the E2 structural glycoprotein is an important target in CSFV-specific ELISAs. With the ultimate aim to render the C-strain suitable as a DIVA vaccine, mutations were introduced that were expected to dampen the immunogenicity of the A-domain. In the first of two approaches, the feasibility of shielding the A-domain by N-linked glycans was evaluated, whereas in the second approach C-strain mutants were created with targeted deletions in the A-domain. Analysis of the antibody responses elicited in rabbits suggested that shielding of the A-domain by an N-linked glycan had a minor effect on the immune response against the A-domain, whereas a targeted deletion of only a single amino acid severely dampened this response. C-strain mutants with larger deletions were highly debilitated and incapable of sustained growth in vitro. By providing the viruses with the opportunity to increase their fitness by mutation, a mutant was rescued that found a way to compensate for the imposed fitness cost. Most of the identified mutations occurred in several independently evolved viruses, demonstrating parallel evolution. By virtue of this compensatory evolution, a well replicating and genetically stable C-strain mutant was produced that can be serologically differentiated from wildtype CSFV. The findings provide the molecular basis for the development of a novel, genetically stable, live attenuated CSF DIVA vaccine. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Classical swine fever virus (CSFV) is an enveloped, positive-strand RNA virus which, together with Bovine viral diarrhoea virus (BVDV) and Border disease virus (BDV), comprises the Pestivirus genus of the Flaviviridae family (Fauquet et al., 2005). Introduction of CSFV into herds of domesticated pigs can result in important economic losses (Terpstra and de Smit, 2000). Vaccination with a CSF virus that has been attenuated by repeated passage in rabbits and cell culture, the so-called “Chinese” or “C”-strain, generally results in swift and lifelong immunity against highly virulent CSFV after a single intramuscular inoculation. However, this vaccine does not enable the serological differentiation between infected and vaccinated animals (DIVA). This is a major disadvantage, since the inability to detect CSF-infected animals in a vaccinated population can impose severe trade restrictions. Although outbreaks of CSF are currently

∗ Corresponding author at: Virology Division, Central Veterinary Institute of Wageningen University Research Centre, Edelhertweg 15, 8219 PH Lelystad, The Netherlands. Tel.: +31 320 238052; fax: +31 320 238225. E-mail address: [email protected] (J. Kortekaas). 0166-0934/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jviromet.2009.09.012

controlled effectively by quarantine restrictions and slaughtering of suspected herds, there is an urgent need for the implementation of more humane and more economical intervention strategies to control future CSF outbreaks. Serological diagnosis of CSF can be performed by ELISAs that detect antibodies directed against either the structural glycoprotein ERNS or E2. Several candidate vaccines have been developed that fulfil the DIVA criterion when accompanied by the appropriate ELISA, varying from subunit vaccines (Depner et al., 2001; Hulst et al., 1993; Uttenthal et al., 2001) to live attenuated viruses (Reimann et al., 2004; van Gennip et al., 2000; van Zijl et al., 1991; Koenig et al., 2007; Wehrle et al., 2007; Holinka et al., 2009) and replicon-based vaccines (Maurer et al., 2005; van Gennip et al., 2002; Widjojoatmodjo et al., 2000) (for a recent review see Beer et al., 2007). The current commercially available CSF DIVA vaccine is based on baculovirus-produced E2 and is accompanied by a serological test that detects antibodies directed against ERNS (van Aarle, 2003). Although this vaccine can provide protection against CSF, it is far less efficacious than the C-strain vaccine (van Oirschot, 2003). More importantly, the ERNS ELISA that can be used as an accompanying DIVA test (i.e. the Chekit-CSF-Marker, IDEXX Laboratories), also detects other members of the pestivirus genus (i.e.

176

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

BVDV and BDV) and its use may therefore not be recommended in regions with a high seroprevalence of BVDV and/or BDV infection in swine (European Commission, 2003). This explains why E2 ELISAs are greatly preferred over ERNS ELISAs to accompany a future DIVA vaccine. The E2 protein contains two major antigenic domains, namely the B/C domain and the A-domain (van Rijn et al., 1993; Wensvoort, 1989). Domain A, which is located between amino acids 766 and 866 of the CSFV polyprotein, is divided into subdomains A1, A2 and A3 (Wensvoort, 1989). Despite the fact that the A1 domain is a dominant target for neutralizing antibodies, it has been conserved throughout evolution. In fact, its sequence conservation and immunodominance have rendered it the dominant target in E2 ELISAs. The A-domain contains a recently identified neutralizing B-cell epitope that comprises the amino acid sequence TAVSPTTLR (residues 829–837 of the CSFV polyprotein) (Lin et al., 2000). This epitope shares all characteristic features of the A1 domain, being immunodominant, evolutionarily conserved, specific for CSFV and a target for neutralizing antibodies. From this, it is hypothesized that the A1 domain and the TAVSPTTLR epitope correspond, at least partially. It was recently demonstrated that the substitution of amino acids from the TAVSPTTLR epitope of a virulent CSF virus with the corresponding amino acids from BVDV makes it possible to differentiate this virus serologically from wildtype CSFV (Holinka et al., 2009). Previous attempts to substitute these amino acids in the C-strain vaccine virus were unsuccessful, however (unpublished results). In the current work, alternative methods were used to mutate the TAVSPTTLR epitope of the C-strain vaccine virus. In the first of two approaches, C-strain mutants were created that contain newly introduced N-linked glycans, whereas in the second approach C-strain mutants were created with targeted deletions in the TAVSPTTLR epitope. Considering the error-prone replication of RNA virus genomes, the evolution of viable C-strain mutants upon passage of the viruses in vitro was given appropriate attention. Although a C-strain mutant that contains a newly introduced N-linked glycan anchored to the centre of the TAVSPTTLR epitope was produced successfully, deletion of conserved amino acids from this epitope seemed a more promising approach to dampen its immunogenicity. C-strain mutants with progressively larger deletions were, however, highly debilitated. The noncytopathic nature of CSFV enabled passage of persistently infected cells. By doing so, the mutant viruses were provided with the opportunity to increase their fitness by mutation. By exploiting this method of forced virus evolution, it was possible to produce a genetically stable C-strain virus that can be differentiated serologically from wildtype CSFV. 2. Materials and methods 2.1. Viruses and cells Swine kidney cells constitutively expressing T7 RNA polymerase (SK6.T7) (van Gennip et al., 1999) were grown in K1000 medium supplemented with glutamine (0.3 mg/ml, Invitrogen, Breda, The Netherlands), 5% fetal bovine serum and the antibiotics penicillin (100 U/ml, Invitrogen), streptomycin (100 U/ml, Invitrogen), amphotericin B (2.5 ␮g/ml, Invitrogen) and, when appropriate, with 10 mM l-histidinol dihydrochloride (Sigma, St. Louis, USA). Unless indicated otherwise, virus stocks were produced by passaging the virus three to four times on SK6.T7 cells, followed by two successive freeze/thaw cycles of the infected monolayers. The latter was performed to maximize release of the C-strain virus from infected cells. Virus stocks were titrated on SK6.T7 cells in log 10 dilutions and were determined as TCID50 /ml.

2.2. Construction of C-strain mutants Plasmid pPRK-flc34, which contains a DNA copy of the “Cedipest” CSFV C-strain under T7-promoter control, was used as a template to introduce mutations by site-directed mutagenesis. The previously published DNA copy of the C-strain virus, named pPRK-flc133 (Moormann et al., 1996) was found to lack a cytosine at the −10 position at the 3 -end of the genome. In pPRK-flc34, this error is corrected. Primers are described in Table 1. The name of the forward primer corresponds to the name of the constructed recombinant virus. Primer RV-r was used as a reverse primer for each construction. PCR amplification was performed using the Expand High-Fidelity PCR system (Roche, Almere, The Netherlands). The PCR products were cloned into pGEM-T Easy vectors according to the instructions of the manufacturer (Promega, Leiden, The Netherlands) and sequenced using an ABI PRISM 310 genetic analyzer (Applied Biosystems, Nieuwerkerk a/d IJssel, The Netherlands). The PCR products were released from pGEM-T plasmids by digestion with ApaLI and used to replace the corresponding genome fragment of plasmid pPRc129, a pOK12-derived plasmid containing cDNA encoding the 5 half of the C-strain virus. The pPRc129 plasmids were subsequently digested with NotI and SalI, and the resulting fragments were used to replace the corresponding segment of plasmid pPRK-flc34, resulting in plasmids pFlc-N1, pFlc-N2, pFlc-N3, pFlc-N4, pFlc-N5, pFlc-P, pFlc-PT, pFlc-SP, pFlc-SPT, pFlcVSP, pFlc-AVSP and pFlc-SPTTL. 2.3. Production of C-strain mutants Plasmids containing full-length cDNA clones were linearized with XbaI and transfected into SK6.T7 cells as described previously (van Gennip et al., 1999). Four days after transfection, expression of viral proteins was determined by immunoperoxidase monolayer assays (IPMAs) using mAb WB103, which is directed against the CSFV non-structural protein NS3 (Edwards et al., 1991). Cells from another well were treated with trypsin, transferred to a 25-cm2 tissue culture flask and grown for three to four days. When necessary, cells were passaged repeatedly to support virus growth. Monolayers were freeze-thawed, centrifuged to remove cell debris and subsequently stored at −70 ◦ C. The cleared lysates were used to prepare seedlots of the vaccine candidates by infecting fresh SK6.T7 cells followed by harvest four days later. Growth of the viruses was always performed on SK6.T7 cells. Although the viruses described here also replicated normally on SK6 cells, production yields were reproduced more accurately when using SK6.T7 cells. To study the growth kinetics of the rescued viruses, subconfluent monolayers in 25-cm2 tissue culture flasks were infected at a multiplicity of infection of 0.1. After 24, 48, 72, 96, 120, 144 and 168 h post infection, the virus titres in cell lysates were determined. The material was freeze-thawed twice, clarified by centrifugation at 2500 × g at 4 ◦ C and stored at −70 ◦ C. Virus titres (10 log TCID50 /ml) were determined on SK6.T7 cells. The E2 genes of the rescued viruses were sequenced. When appropriate, the genes encoding the capsid (C), ERNS and E1 proteins (i.e. the structural proteins) were also sequenced. To this end, viral RNA was isolated using the High Pure Total RNA isolation kit (Roche) and used for cDNA synthesis using the Superscript FirstStrand Synthesis system (Invitrogen) and a gene-specific primer. The cDNA was sequenced as described above. 2.4. Western blots Lysates of infected SK6.T7 cells were prepared from confluent monolayers grown in 25-cm2 tissue culture flasks. To this end, cells were lysed in 0.5 ml phosphate-buffered saline (PBS) containing 1% Nonidet P40 (VWR, Boxmeer, The Netherlands) and Complete

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

177

Table 1 Primers used to construct C-strain mutants. Primer

Sequencea

RV-N1 RV-N2 RV-N3 RV-N4 RV-N5 RV-P RV-PT RV-SP RV-SPT RV-VSP RV-AVSP RV-SPTTL RV-r

5 -GAGTGCACAGCAGTGAGCAATACAACTCTGAGAACAGAAG-3 5 -GAGTGCACAAATGTGAGCAATACAACTCTGAGAACAGAAGTGGTAAAGACCTTC-3 5 -GAGTGCACAGCAGTGAGCAATACAACTAATAGAACAGAAGTGGTAAAGACCTTCAGGAGA-3 5 -GAGTGCACAAATGTGAGCAATACAACTAATAGAACAGAAGTGGTAAAGACCTTC-3 5 -GAGTGCACAGCAGTGAGCAATAATACAACTCTGAGAACAGAAG-3 5 -GAGTGCACAGCAGTGAGCACAACTCTGAGAACAGAAGTGGTAAAGACC-3 5 -GAGTGCACAGCAGTGAGCACTCTGAGAACAGAAGTGGTAAAGACC-3 5 -GAGTGCACAGCAGTGACAACTCTGAGAACAGAAGTGGTAAAGACCTTC-3 5 -GAGTGCACAGCAGTGACTCTGAGAACAGAAGTGGTAAAGACCTTC-3 5 -GAGTGCACAGCAACAACTCTGAGAACAGAAGTGGTAAAGACC-3 5 -GAGTGCACAACAACTCTGAGAACAGAAGTGGTAAAGACC-3 5 -GAGTGCACAGCAGTGAGAACAGAAGTGGTAAAGACCTTCCAGGAGA-3 5 -CTCTGTCTTCACAGACGGTGCAC-3

a

ApaLI restriction sites (underlined) and nucleotide substitutions (bold) are indicated.

protease inhibitor cocktail (Roche). Cell debris was removed by centrifugation of 4 min at 10 000 × g at 4 ◦ C. Proteins were separated in 12% polyacrylamide gels (NuPAGE system, Invitrogen) and subsequently transferred to nitrocellulose paper (Protran, Schleicher and Schuell, VWR). After blocking with PBS containing 0.05% Tween-20 and 1% Protifar (Nutricia, Zoetermeer, The Netherlands), the blots were incubated with C-strain specific mAb C2, which is directed against the B/C domain of E2 (Bognár and Mészáros, 1963) and subsequently with peroxidase-conjugated rabbit anti-mouse immunoglobulins (DAKO, Heverlee, Belgium). Peroxidase activity was detected with the enhanced chemiluminescence system (ECL Plus, GE Healthcare, Diegem, Belgium) using a Storm 860 molecular imager (GE Healthcare). 2.5. Immunoperoxidase monolayer assay (IPMA) Monolayers were washed with D-PBS (Invitrogen), dried to the air, and frozen at −20 ◦ C. The monolayers were fixed with paraformaldehyde (4%, w/v in PBS) for 15 min and subsequently washed with PBS. Peroxidase-conjugated A-domain-specific mAbs b2, b3, b4, b7 (Wensvoort, 1989), c1, c4, c8, c11 (Bognár and Mészáros, 1963) and the mAb used in the PrioCHECK CSFV Ab 2.0 E2 ELISA were used in PBS containing 0.05% Tween-80 (PBS-T) and 5% horse serum. After incubation at 37 ◦ C for 1 h, the plates were washed three times with PBS-T after which activity of peroxidase was detected using 3-amino-9-ethyl-carbazole (Sigma) as the substrate. 2.6. Inoculation of rabbits New Zealand white rabbits of approximately 2 kg were housed in groups of two to four animals. Body temperatures were monitored daily, starting from three days before the inoculation until five days after. The normal body temperature of rabbits varies from 38.5 to 40.1 ◦ C. Accordingly, fever was defined as a body temperature above 40.1 ◦ C. Rabbits were inoculated via the marginal ear vein with 200 ␮l growth medium containing 103 TCID50 of virus. Every seven days, serum was collected. EDTA blood, to be used for virus isolation, was collected four days after the inoculation. These experiments were approved by the Ethics Committee for Animal Experiments of the Central Veterinary Institute of Wageningen UR. 2.7. Virus isolation Peripheral blood leukocytes (PBLs) were concentrated from the EDTA blood samples by ammonium chloride precipitation (0.83% NH4 Cl) as described (Terpstra and Wensvoort, 1988). The PBLs were resuspended in PBS and frozen at −70 ◦ C. The next day,

subconfluent SK6.T7 cell monolayers were incubated with the suspension of freeze-thawed PBLs for 1 h, after which the suspension was replaced by fresh growth medium, followed by an incubation period of four days. To produce sufficient virus for sequence analysis, the rescued viruses were passaged two to three times on SK6.T7 cells. 2.8. PEPSCAN analysis A complete set of overlapping 15-amino-acid-long peptides derived from the CSFV strain Brescia E2 protein, spanning amino acids 690–851 of the CSFV polyprotein, was synthesized in creditcard format miniPEPSCAN cards at PEPSCAN Presto (Lelystad, The Netherlands) and the binding of antibodies from sera to each peptide was tested as described (Slootstra et al., 1996). 2.9. Liquid-phase TAVSPTTLR ELISA Antibody responses directed against the TAVSPTTLR epitope were determined using a liquid-phase peptide ELISA (lp-ELISA) assay as described by van Eijk et al. (manuscript in preparation). The peptide that is used in this assay comprises the TAVSPTTLR peptide, preceded by a glycine-serine-glycine (GSG) linker. The N-terminus is linked to biotin; the C-terminus is amidated. The peptide was synthesized by standard procedures on a Symphony peptide synthesizer (PEPSCAN Presto). Briefly, 100 ␮l rabbit sera (diluted 1:50 in PBS containing 4% (v/v) horse serum and 0.05% (v/v) Tween-80) were incubated in the presence of 10 ng biotin–GSGTAVSPTTLR peptide for 1 h at 37 ◦ C and subsequently transferred to neutravidin-coated 96well plates. After incubation for 1 h at 37 ◦ C, the plates were washed with PBS containing 0.05% (v/v) Tween-80. Neutravidincaptured biotin–TAVSPTTLR–antibody complexes were detected using horseradish peroxidase-labelled goat anti-rabbit IgG (DAKO). After staining using a standard 3,3 ,5,5 -tetramethylbenzidine substrate solution, the OD was measured at 450 nm. As a negative control, each sample was also measured in the absence of peptide. 3. Results 3.1. Production and characterization of C-strain mutants with newly introduced potential N-linked glycosylation sites (PNGSs) Full-length cDNA constructs encoding mutant C-strain viruses with newly introduced PNGSs in the TAVSPTTLR epitope (Fig. 1) were constructed by genetic modification of pPRK-flc34, a cDNA clone of the “Cedipest” C-strain under T7-promoter control. The minimum requirement for N-linked glycosylation is the presence of the amino acid sequon asparagine (Asn, N)-X-threonine (Thr, T) or

178

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

Fig. 1. Introduced and adaptive mutations in E2. Comparison of the amino acids from positions 772 to 791 and 823 to 842 of the recombinant viruses under study. Introduced amino acid substitutions are indicated in bold, adaptive mutations are indicated in bold italics, deleted amino acids are indicated by . The TAVSPTTLR and LFDGTNP epitopes are boxed. Viability: +, virus was successfully passaged in vitro; †, virus was lost from the culture medium after a few passages; –, no virus was detected.

asparagine (Asn, N)-X-serine (Ser, S), where X can be any amino acid except proline (Pro, P) or aspartate (Asp, D) (Kornfeld and Kornfeld, 1985). The mutant virus to be produced from pFlc-N1 (i.e. vFlc-N1) contains a single newly introduced PNGS, whereas mutants to be produced from pFlc-N2, pFlc-N3, pFlc-N4 and pFlc-N5 (i.e. vFlc-N2, vFlc-N3, vFlc-N4 and vFlc-N5, respectively) contain multiple PNGSs (Fig. 1). In the virus to be produced from pFlc-N5 (i.e. vFlc-N5), the central Pro (P) residue of the TAVSPTTLR epitope is substituted by two Asn (N) residues, resulting in two overlapping PNGSs (i.e. NNTT). Four days after transfection of each individual full-length cDNA construct, the presence of infectious virus in the culture medium was demonstrated by infection of fresh SK6.T7 cells (data not shown). However, substantial differences in viral fitness were observed. Although foci of infection of vFlc-N1 were somewhat smaller compared to those of vFlc34 (Fig. 2C), multistep growth curves showed no significant difference in the fitness of these two viruses (Fig. 2A). Virus vFlc-N2 was clearly of lower fitness than vFlc34, yielding considerably smaller foci of infection (Fig. 2C) and lower final titres (Fig. 2A). Viruses from the supernatant of cells transfected with pFlc-N3 or pFlc-N4 could be passaged once or twice but were eventually lost. To provide the viruses with the opportunity to increase their fitness by mutation, cells containing these viruses were passaged repeatedly. However, the number of positively stained cells did not increase during these passages, suggesting that no fitness-compensating mutations were introduced. Interestingly, the virus produced from pFlc-N5 (Fig. 1) could be rescued after only a few passages of transfected cells, eventually yielding a titre of 104 TCID50 /ml. The E2 gene of this virus was sequenced and found to be unchanged. To produce vFlc-N5

in larger amounts, the virus was used to infect fresh SK6.T7 cells which resulted in titres similar to those normally obtained from vFlc34 (data not shown). However, consensus sequencing demonstrated that the virus had lost one of the two newly introduced Asn (N) residues and was, thus, essentially identical to vFlc-N1. Consequently, only viruses vFlc-N1 and vFlc-N2 were considered suitable for further study. To be suitable as a DIVA vaccine, the vaccine candidate must be incapable of inducing A-domain-specific antibodies in vivo. However, to get a first idea of whether or not the modifications introduced in the current candidates affected the antigenic structure of the A-domain, immunoperoxidase monolayer assays (IPMAs) were used to determine whether the recombinant viruses could be recognized by A-domain-specific monoclonal antibodies (mAbs). The A-domain-specific mAbs used were either raised against the CSFV strain Brescia (i.e. mAbs b2, b3, b4, b7) or the C-strain (i.e. c1, c4, c8 and c11). In these experiments, the mAb used in the PrioCHECK CSFV Ab 2.0 E2 ELISA (Prionics, Lelystad, The Netherlands) was also included. Since some of the A-domainspecific antibodies only weakly stained the wildtype C-strain, the ability of the antibodies to recognize the mutant C-strain viruses was studied by staining foci of infection, produced by growing the virus under methylcellulose overlay. Whereas foci of infection of the wildtype C-strain were clearly stained by all antibodies used, these experiments suggested that the Pro (P)833 → Asn (N) substitution, as present in virus vFlc-N1, was sufficient to prevent recognition of the A-domain by mAbs in vitro (data not shown). To study the evolution of vFlc-N1 upon growth in tissue culture, the virus was passaged thirty times. Infected monolayers were immunostained with an antibody directed against the non-

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

179

Fig. 2. In vitro growth characteristics of vFlc34 (wildtype C-strain) and C-strain mutants. (A) Multistep growth curves of viruses vFlc34 (♦), vFlc-N1 (), vFlc-N2 (). (B) Multistep growth curves of viruses vFlc34 (♦), vFlc-P (×) and vFlc-PTa1 (). SK6.T7 cells were infected with a multiplicity of infection of 0.1 and virus yield was determined at various time points post infection. (C) SK6.T7 monolayers were infected, covered with growth medium containing methylcellullose and incubated at 37 ◦ C for four days. Monolayers were fixed with 4% paraformaldehyde and immunostained with peroxidase-conjugated mAb WB103.

structural protein NS3, named WB103, and duplicate monolayers were stained with a mixture of A-domain-specific antibodies b3 and b4 to establish the presence of phenotypic revertants (i.e. viruses in which the A-domain could be stained with these antibodies). After only three passages, a small percentage of cells infected with vFlc-N1 were stained with the b3/b4 mixture, indicating that phenotypic reversion occurred (data not shown). Strikingly, the phenotypic revertant appeared to be maintained as a minority variant in the virus population. Even after repeated passage of the virus, only a few cells were stained with the b3/b4 mAb mixture. To identify the mutation that was responsible for the revertant phenotype, the virus population was enriched for the revertant virus to enable genome sequencing. By seeding dilutions of the virus in 96-well plates, populations that were enriched for the phenotypic revertants could be selected. In two individual experiments, the phenotypic revertant was shown to contain a Ser (S) residue at the position where the Asn (N) was introduced (i.e. position 833 of the CSFV polyprotein), resulting from a single transition in the Asn (N) codon (from AAU to AGU). Since this mutation results in a loss of the PNGS, it is not surprising that it was accompanied by a restored ability of mAbs to recognize the A-domain. In conclusion, the detection of the revertant subpopulation demonstrates that vFlc-N1 explored possibilities to increase its fitness by mutation. However, since the newly introduced AAU codon

was maintained by the master genotype, the virus is considered genetically stable at the population level. Analysis of vFlc-N2 revealed phenotypic variants with similar characteristics as those observed during experiments with vFlc-N1, but these were not further characterized. To determine if the newly introduced PNGSs in the TAVSPTTLR epitope of vFlc-N1 and vFlc-N2 resulted in the attachment of new carbohydrate moieties to E2, the relative electrophoretic mobilities of the modified E2 proteins were studied by polyacrylamide gel electrophoresis (PAGE) under reducing conditions followed by Western blotting. Western blots, containing separated proteins of cells infected with vFlc34, vFlc-N1 and vFlc-N2, demonstrated that the E2 protein of vFlc-N1 was of higher molecular weight than the corresponding protein of vFlc34 (Fig. 3). Oligomers containing E2 were also detected. However, considering that the proteins were analyzed under denaturing conditions, these oligomers are not considered to be physiological (Fig. 3). Treatment of the cell lysates with PNGase F, an enzyme that removes N-linked glycans from proteins, yielded E2 proteins of identical molecular weight. From this, it was concluded that the newly introduced PNGS in virus vFlc-N1 is indeed N-glycosylated. Western blots containing separated proteins of vFlc-N2 suggested that at least one of the two newly introduced glycosylation sequons is used as an anchor site for a carbohydrate moiety. These lysates also seemed to contain E2 proteins of even higher molecu-

180

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

Fig. 3. Western blot analysis of E2. Western blot of a denaturing PAGE gel containing lysates of SK6.T7 cells non-infected (mock), or infected with vFlc34, vFlc-N1 or vFlcN2, respectively, and the same samples after treatment with PNGase F. Proteins were detected by mAb C2 and peroxidase-conjugated rabbit anti-mouse immunoglobulins as the secondary antibody. The positions of oligomers (O), monomers (M) and PNGase F-treated monomers (M*) are indicated. The positions of molecular weight standard proteins are indicated to the left.

lar weight, which suggests the presence of glycan moieties at both newly introduced glycosylation sites (Fig. 3). 3.2. Production and characterization of C-strain mutants with targeted deletions in the TAVSPTTLR epitope Full-length cDNA constructs encoding C-strain viruses with targeted deletions in the TAVSPTTLR epitope (Fig. 1) were constructed as described above for glycosylation mutants. Deletion of the central Pro (P) residue of the TAVSPTTLR epitope yielded vFlc-P, which produced somewhat smaller foci of infection when compared to vFlc34 (Fig. 2C). vFlc-P grew in most experiments to 10-fold lower titres (Fig. 2B), although the highest titres obtained still exceeded 106 TCID50 /ml. Analysis of vFlc-P by IPMAs demonstrated that this C-strain mutant was not detected by A-domain-specific mAbs. Notably, after 20 passages, consensus sequencing demonstrated a transition mutation in the Ser (S)789 codon (from UCC to UUC), which resulted in the substitution of Ser (S) for phenylalanine (Phe, F). Intriguingly, Ser (S)789 is present in all C-strain viruses and related lapinized CSFV strains, whereas Phe (F)789 is completely conserved in virulent CSFV strains and even highly conserved among other members of the pestivirus genus (van Rijn et al., 1997). The resulting virus, named vFlcPa1 (Fig. 1), was not further characterized, however. In contrast to vFlc-P, viruses with deletions of more than one amino acid were highly debilitated. Similar to vFlc-N3 and vFlcN4, virus from the supernatant of cells transfected with plasmids encoding these viruses could be passaged a few times, but were subsequently lost. To provide the viruses with the opportunity to increase their fitness by mutation, cells transfected with these constructs were passaged repeatedly. Passage of cells transfected with pFlc-SP, pFlc-SPT, pFlc-VSP, pFlc-AVSP and pFlc-SPTTL did not result in an increase in virus production. During initial passages of cells transfected with pFlc-PT, the results were similar to those obtained with the other deletion constructs, yielding foci of infection that remained small in size (average of 10–20 cells). However, after a few additional passages, immune staining demonstrated a sudden improved growth of the virus, which suggested that the virus had introduced fitness-compensating mutations. The resulting virus was named vFlc-PTa1. To identify putative resuscitating mutations in vFlc-

PTa1, the consensus sequence of its E2 gene was determined. This revealed that the virus not only retained the introduced deletion, but also had introduced two mutations in the E2 gene. Conveniently, a clear double peak in the sequence chromatogram of one of the mutations suggested that a transition mutation resulting in a codon change from GAC to AAC was the first to occur (data not shown). This mutation resulted in the substitution of Asp (D)774 to Asn (N) and, interestingly, introduced a new PNGS in the A-domain of E2 (Figs. 1 and 4). The second change was a transversion mutation within the valine (Val, V)-831 codon of the TAVSPTTLR epitope (from GUG to GGG), which resulted in a Val (V) to glycine (Gly, G) substitution. It is interesting to note that position 774 might be located in the proximity of the TAVSPTTLR epitope in the native E2 structure. Asp(D)774 is part of a recently identified epitope comprising amino acids 772 LFDGTNP778 (Peng et al., 2008). Like the 829 TAVSPTTLR837 epitope, the 772 LFDGTNP778 epitope shares all three features that define the A1 domain, being CSFV-specific, evolutionarily conserved and a target for neutralizing antibodies. Considering these similarities and the experimental findings, it is hypothesized that the TAVSPTTLR and the LFDGTNP epitope co-localize in the E2 native structure, shaping the A1 subdomain (Fig. 4). Considering the possibility that adaptive mutations could also be present in genes encoding other structural proteins, the consensus sequences of the C, ERNS and E1 genes were also determined. Sequencing of the C gene and the ERNS gene revealed only a single silent mutation in the latter (U1549 → C). Interestingly, in the E1 gene, a transversion mutation at position 2275 (A2275 → U) was detected that resulted in the substitution of glutamic acid (Glu, E)634 for an aspartic acid (Asp, D) (Table 2). Considering that the E2 protein is known to assemble into disulphide-linked heterodimers Table 2 Sequence determinationa of the C, E1, ERNS and E2 genes of independently evolved viruses produced from pFlc-PTb . Virus

Mutation

Codon change

Amino acid change

Protein

PTa1

U1549 → C A2275 → U G2693 → A U2865 → G

GUU →GUC GAA → GAU GAC → AAC GUG → GGG

None E634 → D D774 → N V831 → G

ERNS E1 E2 E2

PTa2

U1549 → C C2695 → A

GUU → GUC GAC → GAA

None D774 → E

ERNS E2

PTa3

U1549 → C G2693 → U

GUU → GUC GAC → AAC

None D774 → N

ERNS E2

PTa4

U1549 → C

GUU → GUC

None

ERNS

PTa5

U1549 → C G1706 → A G2693 → U

GUU → GUC GCA → ACA GAC → AAC

None A445 → T D774 → N

ERNS ERNS E2

PTa6

U1549 → C G2693 → U

GUU → GUC GAC → AAC

None D774 → N

ERNS E2

PTa7

U1549 → C G2693 → U

GUU → GUC GAC → AAC

None D774 → N

ERNS E2

PTa8

U1549 → C U2865 → A

GUU → GUC GUG → GGG

None V831 → G

ERNS E2

PTa9

U1549 → C

GUU → GUC

None

ERNS

PTa10

U1549 → C G2693 → U

GUU → GUC GAC → AAC

None D774 → N

ERNS E2

PTa11

U1549 → C G2693 → U

GUU → GUC GAC → AAC

None D774 → N

ERNS E2

a Adaptive mutations found in the genes encoding structural proteins. Transfected cells were passaged repeatedly. After an increase in virus production was noted, RNA was isolated and used for reverse transcriptase PCR followed by sequencing. Identified mutations are indicated in bold. b The in vitro growth characteristics of virus vFlc-PTa1 are depicted in Fig. 2. The growth characteristics of the other viruses were not determined.

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

181

Fig. 4. Proposed antigenic structure of the ectodomain of the CSFV C-strain E2 protein (modified from van Rijn et al., 1994). Antigenic domains B/C and A are indicated. Positions of PNGSs referred to in the text are indicated by arrows. In these mutants the indicated amino acid is substituted for an Asn (N) residue. Authentic N-linked glycans are drawn with solid lines, newly introduced (putative) N-linked glycans are drawn in dotted lines. Positions in E2 that are predicted to co-localize in the native E2 structure are shaded. B-cell epitopes 829 TAVSPTTLR837 and 772 LFDGTNP778 are indicated in bold.

with the E1 protein (Thiel et al., 1991), it is conceivable that this mutation contributed to the fitness recovery of vFlc-PTa1. The in vitro growth characteristics of vFlc-PTa1, relative to vFlc34 and vFlc-P, are visualized by multistep growth curves (Fig. 2B). Although the vFlc-PTa1 virus grows somewhat slower than vFlc34, comparable final titres are obtained. 3.3. Experimental evolution of viruses produced from plasmid pFlc-PT The results obtained from sequence analysis of vFlc-PTa1 suggested that the silent mutation in ERNS , the amino acid substitution in E1 and the two substitutions in E2 were all involved in the fitness recovery of this mutant. To test this hypothesis, ten independent transfections with pFlc-PT were performed and the transfected cells were passaged repeatedly while the presence of virus was monitored by IPMAs using mAb WB103. The number of positive cells clearly increased after only two to three passages, suggesting that the viruses had introduced fitness-compensating mutations. After eleven passages, the genomes of the viruses were analyzed as described for vFlc-PTa1 (Table 2). The mutation that resulted in the Asp (D)774 → Asn (N) substitution was found in six out of ten evolved viruses, demonstrating parallel evolution. However, in one of the viruses, a mutation was found that results in an Asp (D)774 → Glu (E) substitution while in the three remaining viruses no mutation was found in this codon. The mutation responsible for the Val (V)831 → Gly (G) substitution in vFlc-PTa1 was detected in one virus (i.e. vFlc-PTa8, Table 2). Interestingly, the silent mutation in the ERNS gene at position 1549 was found in all viruses. This observation suggests that natural selection also operated at the RNA level and that additional adaptive mutations could also have occurred in regions of the genome that were not analyzed in this experiment. However, it was striking to find that three of four mutations detected in the genome of vFlc-PTa1 were again introduced during evolution of the viruses in the current experiment. Of note, a mutation at a fifth position in virus vFlc-PTa5 was detected. Besides the silent mutation in the ERNS gene and the Asp (D)774 → Asn (N) substitution in E2, this virus had introduced a transition mutation in the Ala (A)454 codon (from GCA to ACA) of the ERNS gene, which resulted in the substitution of Ala (A)454 for Thr

(T) (Table 2). Interestingly, in analogy with the previously described Ser (S)789 to Phe (F) substitution that was selected for during the evolution of vFlc-P, the amino acid that altered in vFlc-PTa5, in this case Ala (A)445 , is changed into an amino acid that is conserved among CSFV field strains, in this case Thr (T)445 . Although adaptive mutations in the structural genes were anticipated, the recurrent detection of a silent adaptive mutation was surprising. This finding implies that mutations in regions of the genome that were not analyzed in the current work could also have contributed to the fitness recovery of vFlc-PTa1. 3.4. Analysis of the antibody responses against vFlc-P and vFlc-N1 The inability of A-domain-specific antibodies to recognize vFlcP and vFlc-N1 in vitro suggested that the antigenic structure of the A-domain was successfully modified. However, to be suitable as a DIVA vaccine, the antibody response induced in vivo must be sufficiently dampened to enable the serological differentiation between infected and vaccinated animals. Although the DIVA properties of the vaccine candidates will ultimately be tested in pigs, in the current work rabbits were preferred for analysis of the humoral immune response for two main reasons. First, it was important to determine whether the vaccine viruses under study are capable of a productive infection in vivo. In pigs, inoculation of C-strain viruses does not induce any clinical symptoms, whereas inoculation of rabbits induces a temporal febrile illness. Hence, using rabbits enables the confirmation of productive infection and, in addition, allows the study of potential differences in fitness of the viruses in vivo (de Smit et al., 2000). Secondly, the C-strain virus can be isolated from the blood of infected rabbits, a procedure that is often unsuccessful when using pigs. The isolation of the vaccine virus after replication in vivo allows for the verification of stable maintenance of the introduced genetic modifications. Groups of four rabbits were inoculated with vFlc-P or vFlcN1. Control animals were inoculated with either vFlc34 or culture medium. During the acclimatization period, the average body temperatures of the rabbits were normal (39.2 ◦ C, SD ±0.28, n = 68). Fever was defined as a body temperature above 40.1 ◦ C. Fever was first noted in the groups inoculated with vFlc34 (40.7 ◦ C, SD ±0.48,

182

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

Fig. 5. Analysis of the antibody responses against vFlc-N1 and vFlc-P by ELISA. Antibody responses induced by vFlc-34 (), vFlc-N1 (), or vFlc-P () as determined by the Chekit ERNS ELISA (A) and the PrioCHECK CSFV Ab 2.0 E2 ELISA (B), respectively. Values depicted are averages (n = 4; ±SD).

n = 4) and vFlc-N1 (40.5 ◦ C, SD ±0.32, n = 4), both at two days after inoculation. An elevated body temperature in rabbits inoculated with vFlc-P (40.0 ◦ C, SD ±0.13, n = 4) was noted at three days after inoculation. Virus was isolated from peripheral blood lymphocytes (PBLs) of three of four rabbits inoculated with vFlc34, three of four rabbits inoculated with vFlc-N1 and two of four rabbits inoculated with vFlc-P. Consensus sequencing demonstrated that the E2 genes of these viruses were not altered by the passage in rabbits (data not shown). To determine whether the C-strain mutants enable differentiation between infected and vaccinated animals, the rabbit antisera were analyzed by the PrioCHECK CSFV Ab 2.0 E2 ELISA (Prionics, Lelystad, The Netherlands). This ELISA specifically detects antibodies against the A-domain of E2. As a reference control for effective immunization, the Chekit CSF ERNS ELISA (IDEXX laboratories, Hoofddorp, The Netherlands) was used. The ERNS responses induced by vFlc34, vFlc-N1 and vFlc-P were similar (Fig. 5A). Comparison of the A-domain-specific E2 responses induced by vFlc34 and vFlc-N1, in the PrioCHECK CSFV Ab 2.0 E2 ELISA, demonstrated that the modification present in vFlc-N1 had a minor effect on this response (Fig. 5B). In contrast, comparison of the E2 responses of vFlc-P with that of vFlc34 demonstrated that deletion of Pro (P)833 results in a significant dampening of the antibody response against the A-domain (Fig. 5B). Considering the minor effect of the newly introduced N-linked glycosylation site of vFlc-N1 on the A-domain-specific antibody response, it was decided not to study the antigenic properties of vFlc-N2 in rabbits any further, but to focus on the deletion mutants instead.

(40.3 ◦ C), whereas two of the rabbits inoculated with vFlc-PTa1 displayed fever (i.e. 40.4–40.6 ◦ C) at five to seven days post inoculation. At four days after inoculation, virus was isolated from PBLs of both animals inoculated with vFlc34. No virus could be isolated from PBLs of the three rabbits inoculated with vFlc-PTa1. Sera derived from both rabbits inoculated with vFlc34 gave results in the ERNS and E2 ELISA that were comparable to those obtained in the first experiment (animals 1.1 and 1.2; Fig. 6). That is, the blocking percentages determined by the E2 ELISA were higher than those detected by the ERNS ELISA. In contrast, analysis of sera from rabbit 2.1, which was inoculated with vFlc-PTa1, demonstrated that the blocking percentages in the E2 ELISA were lower than those detected in the ERNS ELISA (Fig. 6). This result suggests a specific dampening of the A-domain-specific antibody response elicited by vFlc-PTa1, similar to the results described earlier for vFlc-P (Fig. 5). Unexpectedly, however, analysis of sera from animals 2.2 and 2.3 did not confirm this result. These animals developed high levels of A-domain-specific E2 antibodies (Fig. 6B). To analyze the antibody responses more thoroughly, PEPSCAN analysis was performed using previously constructed 15-aminoacid-long overlapping peptides derived from the E2 protein of CSFV strain Brescia. Results showed that the sera obtained at day 36 from animals inoculated with the wildtype C-strain recognized two epitopes of CSFV strain Brescia (Fig. 7A, upper panels). As expected, the first of these epitopes was the 829 TAVSPTTLR837 epitope in which the residues 831 VSPTTLR837 appeared most critical (Fig. 7B, left panel). The second epitope comprised amino acids 754 YLASLHKDAPT764 . Interestingly, this epitope, which is described here for the first time, is located outside the previously defined Adomain (i.e. amino acids 766–866; Fig. 4). None of the sera obtained from rabbits inoculated with vFlc-PTa1 recognized either of the abovementioned epitopes. Although the lack of recognition of the 829 TAVSPTTLR837 epitope (Fig. 7B, right panel) can be explained

3.5. Analysis of the antibody response against vFlc-PTa1 In the second animal trial, the antibody response against vFlcPTa1 was compared with that induced against vFlc34. The average body temperature of the rabbits prior to inoculation was normal (39.2 ◦ C, SD ±0.37, n = 28). One of two rabbits inoculated with vFlc34 experienced fever at two days after inoculation

Fig. 6. Analysis of the antibody response against vFlc-PTa1 by ELISA. Antibody response induced by vFlc34 (rabbits 1.1 and 1.2) and vFlc-PTa1 (rabbits 2.1, 2.2 and 2.3) as determined by the Chekit ERNS ELISA (A) and the PrioCHECK CSFV Ab 2.0 E2 ELISA (B).

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

183

Fig. 7. Analysis of rabbit antisera by PEPSCAN analysis. Rabbit antisera obtained at 36 days post inoculation were analyzed by PEPSCAN analysis. (A) The reactivity of each rabbit antiserum with 162 peptides (x-axis) is shown. (B) Comparison of the reactivity of antibodies from rabbit antisera raised against vFlc34 (left panel) and vFlc-PTa1 (right panel) with the TAVSPTTLR epitope. The numbers on the x-axis correspond to the position of the peptide N-termini in the CSFV polypeptide. The y-axis depicts the ODccd values.

by the mutations introduced in vFlc-PTa1, the lack of recognition of the 754 YLASLHKDAPT764 epitope was unexpected. This result suggests that the 754 YLASLHKDAPT764 epitope in some way interacts with the 829 TAVSPTTLR837 epitope and that these epitopes are located in close proximity in the native E2 structure (Fig. 4). Although PEPSCAN analysis did not reveal the epitopes of the Adomain that are recognized by the antisera of rabbits 2.2 and 2.3, it did demonstrate that neither of these sera recognize the TAVSPTTLR epitope. This result suggests that vFlc-PTa1 could fulfil the DIVA criterion when accompanied by a TAVSPTTLR-based peptide ELISA. To test this hypothesis, an experimental TAVSPTTLR-based liquid-phase peptide ELISA was used (Van Eijk et al., manuscript in preparation). Indeed, sera raised against the wildtype C-strain reacted strongly in this ELISA whereas antisera raised against vFlcPTa1 did not show any response (Fig. 8). 4. Discussion In the past decade, several approaches have been proposed for the development of CSF DIVA vaccines. With respect to protective efficacy, chimeric pestiviruses seem to be among the most promising (for a review see Beer et al., 2007). The first chimeric CSF DIVA vaccine candidates were produced by exchanging the E2

Fig. 8. Comparison of the reactivity of antibodies from rabbit antisera raised against vFlc34 (rabbits 1.1 and 1.2) or vFlc-PTa1 (rabbits 2.1, 2.2 and 2.3) with the TAVSPTTLR epitope as determined by the liquid-phase TAVSPTTLR ELISA.

184

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185

ectodomain or the complete ERNS protein of the C-strain virus for the corresponding sequences of BVDV (van Gennip et al., 2000; de Smit et al., 2001). As an alternative, the feasibility of using BVDV as the backbone was later demonstrated by replacing the BVDV E2 protein for the E2 protein of CSFV (Koenig et al., 2007). Vaccination of swine with these chimeric pestiviruses generally results in complete protection against lethal challenge after a single inoculation and enables the serological differentiation between infected and vaccinated animals using either E2 or ERNS ELISAs. Although the chimeric pestiviruses clearly hold great promise as DIVA vaccines, there are also some concerns about their use. Firstly, replicating chimeric viruses might reveal unanticipated and undesired features such as changing tissue or species tropism. Secondly, the presence of BVDV or BDV proteins in the chimeric vaccines could lead to vaccination failure in animals that have been previously infected with BVDV or BDV. Maximum safety of a chimeric pestivirus would be obtained, at least in theory, using a CSF vaccine strain with a high safety record, and preferably exchanging a single antigenic domain or even a single epitope, that would still enable the differentiation between infected and vaccinated animals. Unfortunately, attempts to produce a CSF vaccine virus in which only the A-domain was exchanged for the corresponding region of BDV were not successful (Wehrle et al., 2007). A promising step forward was made by the successful substitution of conserved amino acids of the TAVSPTTLR epitope of the virulent Brescia strain with amino acids from BVDV (Risatti et al., 2006). Progressive mutations in the TAVSPTTLR epitope resulted in attenuation of this normally highly pathogenic CSFV strain. In a subsequent paper, is was demonstrated that one of these Brescia mutants protected against a lethal challenge with the wildtype virus and enabled the differentiation between infected and vaccinated animals using a TAVSPTTLR-based peptide ELISA (Holinka et al., 2009). Unfortunately, attempts to introduce these mutations into the C-strain vaccine virus were so far unsuccessful (unpublished results). In the current work, the TAVSPTTLR epitope of the C-strain vaccine virus was modified without introducing gene sequences from other pestiviruses. In the first of two approaches, C-strain mutants were created that contain newly introduced N-linked glycans, whereas in the second approach C-strain mutants were created with targeted deletions in the TAVSPTTLR epitope. Although a Cstrain mutant was successfully produced that stably maintains a newly introduced N-linked glycan anchored to the centre of the TAVSPTTLR epitope (i.e. vFlc-N1), deletion of amino acids from the TAVSPTTLR epitope seemed a more promising approach to dampen the immunogenicity of this epitope. Plasmids encoding C-strain mutants with deletions of two to four amino acids from the TAVSPTTLR epitope produced infectious viruses, but these viruses were highly debilitated and incapable of sustained growth. The noncytopathic nature of CSFV, however, enabled passage of the cells persistently infected with these viruses. Using this method, the mutant viruses were provided with the opportunity to increase their fitness by mutation. A C-strain mutant virus that lacks Pro833 and Thr834 from the TAVSPTTLR epitope was the only mutant virus that found a way to compensate for the imposed fitness cost by introducing adaptive mutations. Most of the identified mutations occurred in several independently evolved viruses. This demonstration of parallel evolution confirmed that the adaptive mutations emanated from positive Darwinian selection and that they are, thus, unlikely to revert to the wildtype sequence. It will be interesting to study the effects of these substitutions on the (antigenic) structure of E2 in future experiments. To the authors’ knowledge, this is the first report that describes the successful production of a viable CSFV mutant that maintains a deletion in a structural protein. The method of forced virus evolution that was employed in this study is believed to be a valuable

tool for the study and manipulation of CSFV, but also of other RNA viruses. Interestingly, forced virus evolution is now being used as a systematic research tool for the study of human immunodeficiency virus biology (Berkhout and Das, 2009) and has recently provided new insights into hepatitis C virus biology (Phan et al., 2009). Previous studies with mAb-resistant mutants performed by van Rijn et al. (1994), suggested that Pro (P)833 and Thr (T)834 of the TAVSPTTLR epitope are very important for the integrity of conserved epitopes of the A-domain. The successful production of a thriving C-strain mutant (i.e. vFlc-PTa1) that lacks both these amino acids and contains an additional mutation in the TAVSPTTLR epitope as well as a stable mutation in the LFDGTNP epitope, suggested that the antigenic structure of the A-domain was changed dramatically. It was, therefore, expected that the immune response induced by vFlc-PTa1 would be distinguishable from that induced by wildtype CSFV using the PrioCHECK CSFV Ab 2.0 E2 ELISA. It was surprising to find that vFlc-PTa1 is actually quite potent in inducing A-domain-specific antibodies. In fact, vFlc-PTa1 was more potent in inducing A-domain-specific antibodies than its predecessor, the vFlc-P virus (compare Figs. 6B with 5B). It was hypothesized that the significant disruption of the otherwise immunodominant TAVSPTTLR epitope resulted in a refocusing of the antibody response towards epitopes of the A-domain that are normally subdominant. In an attempt to identify these epitopes, antisera were analyzed by PEPSCAN analysis. Although this analysis did not reveal any possible newly recognized epitopes, it did demonstrate that none of the antisera recognized peptides containing the TAVSPTTLR epitope. In accordance with this finding, this study demonstrates that serological differentiation between wildtype CSFV and the vFlc-PTa1 virus is feasible using an experimental liquid-phase TAVSPTTLR peptide-based ELISA. In summary, a molecular basis is provided for the development of a C-strain-based DIVA vaccine and an experimental accompanying ELISA is described that targets the conserved and CSFV-specific TAVSPTTLR epitope. Experiments are in progress to evaluate the protective efficacy and DIVA property of vFlc-PTa1 in swine. Acknowledgements The authors thank Rob Moormann, Piet van Rijn, René van Gennip and Wim van der Poel for useful discussions and Franz Daus and Sjaak Quak for technical assistance. They also thank the animal technicians for performing the animal trials. Furthermore, thanks are due to Jerry Slootstra and Nard Langendijk (PEPSCAN Presto, Lelystad, The Netherlands) for conducting PEPSCAN analysis and peptide synthesis, respectively. This work was commissioned and financed by the Dutch Ministry of Agriculture, Nature and Food Quality, project code BO-08-010-1.4. References Beer, M., Reimann, I., Hoffmann, B., Depner, K., 2007. Novel marker vaccines against classical swine fever. Vaccine 25, 5665–5670. Berkhout, B., Das, A.T., 2009. Virus evolution as a tool to study HIV-1 biology. Methods Mol. Biol. 485, 436–451. Bognár, K., Mészáros, J., 1963. Experiences with lapinized hog cholera virus strain of decreased virulence. Acta Vet. Acad. Sci. Hung. 13, 429–438. Depner, K.R., Bouma, A., Koenen, F., Klinkenberg, D., Lange, E., de Smit, H., Vanderhallen, H., 2001. Classical swine fever (CSF) marker vaccine. Trial II. Challenge study in pregnant sows. Vet. Microbiol. 83, 107–120. de Smit, A.J., Bouma, A., van Gennip, H.G.P., de Kluijver, E.P., Moormann, R.J.M., 2001. Chimeric (marker) C-strain viruses induce clinical protection against virulent classical swine fever virus (CSFV) and reduce transmission of CSFV between vaccinated pigs. Vaccine 19, 1467–1476. de Smit, A.J., van Gennip, H.G., Miedema, G.K., van Rijn, P.A., Terpstra, C., Moormann, R.J., 2000. Recombinant classical swine fever (CSF) viruses derived from the Chinese vaccine strain (C-strain) of CSF virus retain their avirulent and immunogenic characteristics. Vaccine 18, 2351–2358.

J. Kortekaas et al. / Journal of Virological Methods 163 (2010) 175–185 Edwards, S., Moennig, V., Wensvoort, G., 1991. The development of an international reference panel of monoclonal antibodies for the differentiation of hog cholera virus from other pestiviruses. Vet. Microbiol. 29, 101–108. European Commission, 2003. Report on the evaluation of a new classical swine fever discriminatory test (2003/265/EC). SANCO/10809/2003. European Commission, Directorate-General for Health and Constumer Protection, Brussels. Fauquet, C.M., Mayo, M.A., Maniloff, J., Desselberger, U., Ball, L.A., 2005. Virus Taxonomy: Eighth Report of the International Committee on Taxonomy of Viruses. Academic Press, an imprint of Elsevier. Holinka, L.G., Fernandez-Sainz, I., O’Donnell, V., Prarat, M.V., Gladue, D.P., Lu, Z., Risatti, G.R., Borca, M.V., 2009. Development of a live attenuated antigenic marker classical swine fever vaccine. Virology 384, 106–113. Hulst, M.M., Westra, D.F., Wensvoort, G., Moormann, R.J., 1993. Glycoprotein E1 of hog cholera virus expressed in insect cells protects swine from hog cholera. J. Virol. 67, 5435–5442. Koenig, P., Lange, E., Reimann, I., Beer, M., 2007. CP7 E2alf: a safe and efficient marker vaccine strain for oral immunisation of wild boar against classical swine fever virus (CSFV). Vaccine 25, 3391–3399. Kornfeld, R., Kornfeld, S., 1985. Assembly of asparagine-linked oligosaccharides. Annu. Rev. Biochem. 54, 631–664. Lin, M., Lin, F., Mallory, M., Clavijo, A., 2000. Deletions of structural glycoprotein E2 of classical swine fever virus strain alfort/187 resolve a linear epitope of monoclonal antibody WH303 and the minimal N-terminal domain essential for binding immunoglobulin G antibodies of a pig hyperimmune serum. J. Virol. 74, 11619–11625. Maurer, R., Stettler, P., Ruggli, N., Hofmann, M.A., Tratschin, J.-D., 2005. Oronasal vaccination with classical swine fever virus (CSFV) replicon particles with either partial or complete deletion of the E2 gene induces partial protection against lethal challenge with highly virulent CSFV. Vaccine 23, 3318–3328. Moormann, R.J.M., van Gennip, H.G.P., Miedema, G.K.W., Hulst, M.M., van Rijn, P.A., 1996. Infectious RNA transcribed from an engineered full-length cDNA template of the genome of a pestivirus. J. Virol. 70, 763–770. Peng, W.P., Hou, Q., Xia, Z.H., Chen, D., Li, N., Sun, Y., Qiu, H.J., 2008. Identification of a conserved linear B-cell epitope at the N-terminus of the E2 glycoprotein of Classical swine fever virus by phage-displayed random peptide library. Virus Res. 135, 267–272. Phan, T., Beran, R.K., Peters, C., Lorenz, I.C., Lindenbach, B.D., 2009. Hepatitis C virus NS2 protein contributes to virus particle assembly via opposing epistatic interactions with the E1-E2 glycoprotein and NS3-NS4A enzyme complexes. J. Virol. 83, 8379–8395. Reimann, I., Depner, K., Trapp, S., Beer, M., 2004. An avirulent chimeric Pestivirus with altered cell tropism protects pigs against lethal infection with classical swine fever virus. Virology 322, 143–157. Risatti, G.R., Holinka, L.G., Carrillo, C., Kutish, G.F., Lu, Z., Tulman, E.R., Sainz, I.F., Borca, M.V., 2006. Identification of a novel virulence determinant within the E2 structural glycoprotein of classical swine fever virus. Virology 355, 94–101. Slootstra, J.W., Puijk, W.C., Ligtvoet, G.J., Langeveld, J.P., Meloen, R.H., 1996. Structural aspects of antibody-antigen interaction revealed through small random peptide libraries. Mol. Divers. 1, 87–96.

185

Terpstra, C., de Smit, A.J., 2000. The 1997/1998 epizootic of swine fever in the Netherlands: control strategies under a non-vaccination regimen. Vet. Microbiol. 77, 3–15. Terpstra, C., Wensvoort, G., 1988. The protective value of vaccine-induced neutralising antibody titres in swine fever. Vet. Microbiol. 16, 123–128. Thiel, H.J., Stark, R., Weiland, E., Rumenapf, T., Meyers, G., 1991. Hog cholera virus: molecular composition of virions from a pestivirus. J. Virol. 65, 4705–4712. Uttenthal, A., Le Potier, M.F., Romero, L., De Mia, G.M., Floegel-Niesmann, G., 2001. Classical swine fever (CSF) marker vaccine. Trial I. Challenge studies in weaner pigs. Vet. Microbiol. 83, 85–106. van Aarle, P., 2003. Suitability of an E2 subunit vaccine of classical swine fever in combination with the E(rns)-marker-test for eradication through vaccination. Dev. Biol. (Basel) 114, 193–200. van Gennip, H.G., Bouma, A., van Rijn, P.A., Widjojoatmodjo, M.N., Moormann, R.J., 2002. Experimental non-transmissible marker vaccines for classical swine fever (CSF) by trans-complementation of E(rns) or E2 of CSFV. Vaccine 20, 1544– 1556. van Gennip, H.G., van Rijn, P.A., Widjojoatmodjo, M.N., de Smit, A.J., Moormann, R.J., 2000. Chimeric classical swine fever viruses containing envelope protein E(RNS) or E2 of bovine viral diarrhoea virus protect pigs against challenge with CSFV and induce a distinguishable antibody response. Vaccine 19, 447–459. van Gennip, H.G., van Rijn, P.A., Widjojoatmodjo, M.N., Moormann, R.J., 1999. Recovery of infectious classical swine fever virus (CSFV) from full-length genomic cDNA clones by a swine kidney cell line expressing bacteriophage T7 RNA polymerase. J. Virol. Methods 78, 117–128. van Oirschot, J.T., 2003. Vaccinology of classical swine fever: from lab to field. Vet. Microbiol. 96, 367–384. van Rijn, P.A., van Gennip, H.G., de Meijer, E.J., Moormann, R.J., 1993. Epitope mapping of envelope glycoprotein E1 of hog cholera virus strain Brescia. J. Gen. Virol. 74, 2053–2060. van Rijn, P.A., van Gennip, H.G.P., Leendertse, C.H., Bruschke, C.J.M., Paton, D.J., Moormann, R.J.M., van Oirschot, J.T., 1997. Subdivision of the pestivirus genus based on envelope glycoprotein E2. Virology 237, 337–348. van Rijn, P.A., Miedema, G.K., Wensvoort, G., van Gennip, H.G., Moormann, R.J., 1994. Antigenic structure of envelope glycoprotein E1 of hog cholera virus. J. Virol. 68, 3934–3942. van Zijl, M., Wensvoort, G., de Kluyver, E., Hulst, M., van der Gulden, H., Gielkens, A., Berns, A., Moormann, R., 1991. Live attenuated pseudorabies virus expressing envelope glycoprotein E1 of hog cholera virus protects swine against both pseudorabies and hog cholera. J. Virol. 65, 2761–2765. Wehrle, F., Renzullo, S., Faust, A., Beer, M., Kaden, V., Hofmann, M.A., 2007. Chimeric pestiviruses: candidates for live-attenuated classical swine fever marker vaccines. J. Gen. Virol. 88, 2247–2258. Wensvoort, G., 1989. Topographical and functional mapping of epitopes on hog cholera virus with monoclonal antibodies. J. Gen. Virol. 70, 2865–2876. Widjojoatmodjo, M.N., van Gennip, H.G.P., Bouma, A., van Rijn, P.A., Moormann, R.J.M., 2000. Classical swine fever virus E(rns) deletion mutants: trans-complementation and potential use as nontransmissible, modified, liveattenuated marker vaccines. J. Virol. 74, 2973–2980.