VIROLOGY
190, 763-772
Epitope
(1992)
Mapping
of the gp53 Envelope D. J. PATON,*,’
*Central
Veterinary
Protein
J. P. LOWlNGS,*
AND
Laboratory! Weybridge, Surrey, KT15 3NB, University of Surrey, Guildford, Surrey, Received
April
13, 1992; accepted
of Bovine Viral Diarrhea
Virus
A. D. T. BARRETTt U.K.; and tSchool GU2 5X 17, U.K.
of Biological
Sciences,
July I, 1992
Epitopes recognized by nine monoclonal antibodies (mAbs) on the envelope protein, gp53, of two strains of bovine viral diarrhoea virus (NADL and Oregon C24V) were mapped by competitive binding assays and by the characterization and sequence analyses of mAb neutralization escape mutants. This defined an antigenic domain on gp53 that was shared by many BVDV strains, while other less conserved epitopes were possibly distinct. Sequencing of escape mutant viruses revealed that a cluster of three amino acids in the N-terminal half of gp53 were involved in the main antigenic domain shared by both NADL and Oregon C24V viruses, while an amino acid 31 residues further toward the N-terminus was involved in a second site present only on the NADL strain. Since other amino acids defining these epitopes were located at distant positions within the gp53 protein, it is likely that both domains on gp53 consist of composite, conformation-dependent epitopes.
INTRODUCTION
(Collett et al., 198813) as has the El gene of HCV (Stark eta/., 1990; Moormann eta/., 1990). The N-terminus of gp53/El is indicated by a putative signalase cleavage site in the nucleotide sequence and the C-terminus by a region of hydrophobic amino acids that represent the membrane-anchoring site (van Zijl et a/., 1991). The deduced amino acid sequence of gp53 is about 370 residues long. Antigenic variation exists between different strains of BVDV (Heuschele, 1975; Edwards et al., 1988) and both epitope conservation studies with mAbs (Paton et al., 1991 b) and genomic sequence comparisons (Moormann et al., 1990) point to gp53 as a site that contributes to such variation. Knowledge of the organization of gp53 epitopes will help in the understanding of antigenic variation, of antibody-mediated neutralization of virus infectivity, and ultimately in the generation of appropriate vaccines. Using mAb competition studies, three to four antigenie domains have been described for BVDV gp53 (Moennig et al., 1989; Xue et a/., 1990) while Wensvoort (1989) defined four such domains for El of HCV. Van Rijn et al. (1992), extending the work of Wensvoort (1989) have indicated that most El epitopes are situated in the N-terminal half of the protein, although their precise locations were not demonstrated. In this paper we report on the mapping of two antigenic domains on the gp53 of BVDV, showing that the arrangement of epitopes differs for the two virus strains studied. We provide evidence for the location of the domains and demonstrate that they are probably defined by discontinuous sequences of amino acids.
Bovine viral diarrhea virus (BVDV) is an important pathogen of cattle worldwide. Along with the closely related viruses of ovine border disease and hog cholera (HCV), it makes up the pestivirus genus recently reclassified within the Flaviviridae (Francki eta/., 1991). BVDV is enveloped and contains a single-stranded, positive-sense RNA molecule of approximately 12.5 kb (Renard et a/., 1985). The RNA has a single open reading frame that encodes a number of structural and nonstructural polypeptides that are cotranslationally and post-translationally cleaved from polyprotein precursors (Collett et al., 1988a,b). Three putative, structural glycoproteins have been ascribed to BVDV (Thiel et al., 1991) the largest of which, gp53, has an apparent molecular weight of between 51 and 58 kDa, depending in part on glycosylation differences (Donis and Dubovi, 1987). It is an immunodominant protein to which neutralizing monoclonal antibodies (mAbs) are directed (Donis et a/., 1988). The equivalent, major envelope protein of HCV, El, is known to occur as both a homodimer and as a heterodimer complexed with a smaller glycoprotein, gp33 (Wensvoort et al., 1990; Thiel et a/., 1991). Vaccination with Aujeszky’s disease virus expressing only El of HCV has been shown to protect against HCV challenge (van Zijl et a/., 1991). The approximate position of the gp53 gene of BVDV has been mapped toward the 5’ end of the genome
’ To whom
reprint
requests
should
be addressed. 763
0042-6822192
$5.00
764
PATON.
MATERIALS Viruses
AND
LOWINGS.
METHODS
and mAbs
Viruses were grown in calf testis (CT) cells at passage 4-9 or in a bovine turbinate cell line (BT) as described by Paton et al. (199 1 a). The Oregon C24V virus (Gillespie et al., 1960) was cloned by three cycles of limiting dilution and then passaged once more prior to use (a total of 67 in vitro passages), while the NADL strain (Gutekunst and Malmquist, 1963) had been twice plaque purified and then passaged seven times before use (a total of 12 in vitro passages). Anti-BVDV mAbs used in this study were raised against either NADL or Oregon C24V viruses (Table 1). Production of the mAbs and their previous characterization has been reported by Edwards et a/. (1988) and Paton et al. (1991 b). Some of these data are summarized in Table 1, which shows the neutralization titer of each mAb for BVDV strains NADL and Oregon C24V and the relative conservation of the different epitopes among 92 different strains of BVDV. Only mAb WB214n showed significant neutralization of both strains. MAb binding
studies
Recognition of viruses by mAbs was assessed by peroxidase-linked assay (PLA; Holm Jensen, 1981) using infected and mock-infected BT or CT cells grown in 96-well plates (Nunc). After fixing the cells with 20% acetone in PBS, pH 7.6, containing 200 mg/liter bovine serum albumin, predetermined optimal dilutions of mAbs were added and incubated for 15 min at 37”. Bound mAbs were detected with a rabbit anti-mouse peroxidase conjugate and the substrate 3-amino-gethylcarbazole. Staining of infected cells was assessed microscopically with reference to the mock-infected wells. Competition
studies
Qualitative results of competitive binding between these mAbs have been reported previously (Paton et a/., 1991 b). The method, a modified PLA, was altered by use of a soluble color substrate TMB (0.1 mg/ml tetramethyl benzidine in phosphate-citrate buffer, pH 5, with 0.0075% hydrogen peroxide) to permit quantitative assessment of competition. BT cells were prepared in microtiter plates as for the PLA. Unlabeled mAbs were in the form of murine ascitic fluids precipitated with ammonium sulfate and redissolved at 2.5 mg/ml in phosphate-buffered saline, pH 7.6, containing 50% glycerol. Serial dilutions of these mAbs were added to the fixed cell monolayers as blockers, immediately before adding biotinylated competitor mAbs. The
AND
BARREll
biotinylated mAbs, which had been previously titrated in the assay system, were used at twice the minimum concentration for plateau binding. Binding of biotinylated mAbs was detected by a biotin streptavidin-peroxidase complex (Amersham, U.K.) and TMB stopped with 2 M sulfuric acid. Color development was recorded by a spectrophotometer at a wavelength of 450 nm and net optical densities were calculated from the difference between the values for infected and mockinfected wells. The percentage inhibition of binding was calculated from the ratio of the net optical density when biotinylated mAb competed with either an irrelevant unlabeled mAb or unlabeled anti-gp53 mAbs. MAb neutralization
escape
mutants
Viruses that escaped neutralization by mAbs were selected by incubating sequential, 1O-fold virus dilutions with equal volumes of mAb at concentrations of between 0.12 and 0.20 mg/ml. After 1 hr at 37”, cell suspension was added and incubation continued for a further 3 days. Supernatant fluids were retained, while monolayers were fixed and immunostained using a pan-pestivirus-reactive anti-p80 mAb as previously described (Paton et a/., 1991 b). Supernatants from cultures containing only plaques of stained cells, at virus dilutions at which 100% infection would have been expected in the absence of mAb, were passaged twice without antibody to produce stocks of potential escape mutant viruses. These mutants received one further passage, either for characterization with mAbs or for extraction of RNA for molecular cloning. Each stock was titrated in 1 O-fold steps and stained with the selecting mAb and an anti-p80 mAb. Only those stocks which were recognized by the anti-p80 mAb and not the selecting mAb were retained as escape mutant viruses. Escape mutants were titrated in lo-fold steps with and without the presence of mAbs (final dilution 0.05 mg/ml) to test whether they resisted neutralization. They were also examined for recognition by the full panel of gp53 mAbs to assess epitopic changes. All escape mutant viruses were named after the mAb/ virus combination selected from (e.g., Al 214n was selected from NADL with WB214n, while C3 214~ was selected from Oregon C24V with the same mAb) and all were derived from separate wells. Nucleotide
sequencing
Because of uncertainty over the exact position of gp53, all BVDV nucleotide and amino acid numbering is from the start of the genome and the beginning of the open reading frame, respectively, (as described by Collett et al., 1988a, for BVDV strain NADL). Gp53 was sequenced for both parental strains of virus and for eight escape mutants (four each of NADL and Oregon
EPITOPES
ON
THE
C24V). Total cellular RNA was extracted from BVDV-infected or mock-infected CT cells and virus-specific cDNA was synthesized using moloney murine leukaemia virus reverse transcriptase and amplified by the polymerase chain reaction (PCR) using Taq DNA polymerase, as described by Roehe and Woodward (1991). Oligonucleotide primers for cDNA synthesis and amplification were derived from published BVDV sequence data (Collett et al., 1988a) and were synthesized on an Applied Biosystems 381A DNA synthesizer. Single fragments of Oregon C24V and variant viruses were amplified with the PCR primer combination: forward 2336-2357Ireverse 3486-3467. NADL and variant viruses were amplified as two fragments with two pairs of primers: forward 2336-2356/reverse 2874-2855 and forward 2855-2874lreverse 34853466. Each primer began with an additional 9-l 0 nucleotides of nonviral sequence, encoding a specific endonuclease restriction site. PCR-derived cDNA was purified by low-melting-point agarose gel electrophoresis, extracted with a silica matrix (Prepagene, Bio-Rad), cut by restriction endonucleases to give cohesive ends, and ligated into similarly digested, calf alkaline-phosphatase-treated phagemid vectors (pBluescript II KS+, Stratagene) according to the supplier’s instructions. The fscherichia co/i strain XL1 -Blue (Stratagene) was transformed with the phagemid construct using the CaCI, method (Sambrook et al., 1989) and singlestranded, insert-specific DNA was isolated using the helper phage VCSM 13 and the Stratagene protocol (based on the method of Vieira and Messing, 1987). The single-stranded DNA was sequenced using the Sequenase (United States Biochemical) protocol and oligonucleotide primers derived from Bluescript or BVDV sequence data. The locations of the primers for both PCR and sequencing are marked in Fig. 1. All genes were sequenced in both directions using two to six clones per virus; increased numbers of clones being sequenced across regions where differences between viruses were detected. The region amplified and sequenced extended from nucleotide 2336 to 3486, beginning upstream of the putative signalase cleavage site between bases 2461 and 2462 and ending just before the start of the putative membrane-anchoring region (3491) probably about 91 nucleotides short of the C-terminus of gp53. Sequence data was analyzed using the GCG software package (Devereux et al., 1984) via the Daresbury Laboratory of the Science and Engineering Research Council. RESULTS MAb binding
studies
At optimal mAb dilutions, background staining was minimal. No significant differences in mAb binding
gp53
PROTEIN
OF
BVDV
765
were detected whether BT or CT cells were used. Recognition of the NADL and Oregon C24V strains of BVDV by the mAb panel is shown in Table 1. Competition
studies
The results of competition experiments for NADL and Oregon C24V viruses are shown in Table 2. WB166n and WB170n which were not successfully biotinylated were only assessed as blockers. All of the mAbs, except WB165c, blocked the binding of at least one other biotinylated mAb, while each biotinylated mAb, except for WBl58n, was blocked by at least one other unbiotinylated mAb. Reciprocal competition of >45% between mAbs is shown in Fig. 2. One-way blocking data concerning WBl66n and WB170n was also included in this figure as the possibility of reciprocal block could not be excluded for these mAbs. If reciprocal blocking is taken as an indication of spatial proximity, it can be seen from Fig. 2 that, for both viruses, many of the epitopes appeared to cluster within a single domain and only that of WBl58n was clearly distinct. The four epitopes present on both viruses (those of WB166n, WB214n, WB215c, and WB163c) showed similar, but not identical interrelationships whichever virus was used as antigen. A single mAb combination showed evidence of binding synergism in that WB162c increased the binding of biotinylated WB165c by 54%. MAb neutralization
escape
mutants
Mutant viruses, which escaped neutralization by their selecting mAb, were obtained for all neutralizing mAb/virus combinations except for WB163c/Oregon C24V and WB170n/NADL. The log,0 neutralization of parental virus by the selecting mAbs was between 2.5 and 3.75. All mutants showed at least a two log,0 reduction in neutralization, in the presence of the inducing mAb, compared to that seen for the parent virus with the same mAb. Some mutants showed a similar order of neutralization resistance with additional mAbs. Table 3 summarizes the pattern of mAb neutralization resistance of the mutants and also shows their loss of binding epitopes. Where binding epitopes were lost, multiple losses were considered suggestive of epitopic overlap. On this basis, NADL epitopes were split into three groups: (1) WBI 66n and WB215c, (2) WB214n,and(3)WB158n.TheWB170nandWB163c epitopes seemed distinct, in that they were not lost from any of the NADL mutants, but their relationship to one another was not established. For Oregon C24V virus, the pattern of epitope loss suggested a domain consisting of the epitopes of WB214n, WB166n, WB162c, WBl15c, and WB215c. The WB163c and
PATON,
766
LOWINGS,
AND NADL
Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL
BARRETT
C24v
AGATTTAACACGCATTTGG .f AACGCTGCCACGACTACTGCATTCCTAGTATGCCTTGTTAGATGGTCAG
2351
AACGCTGCAACAACTACTGCTTTTTTAGTATGCCTTGTTAAGATAGTCAG
C24~
GGGCCAGAT;;GTACAGGGCATCCTATGGCTACTACTGAT~CAGGAGTG~
2407
GGGCCAGATGGTACAGGGCATTCTGTGGCTACTATTGATACAGGGGTAC
C24V
AGGGGCACCTAGACTGCAAACCTGAATACTCATATGCCATAGCCAAGAGT
2451
AAGGGCACTTGGATTGCAAACCTGAATTCTCGTATGCCATAGCAAAGGAC
C24V
GATAGAATT;;GCCTACAAG;;AGCTGAAGA&TTACTACThTTTGGAAGGi
2507
GAAAGAATTGGTCAACTGGGGGCTGAAGGCCTTACCACCACTTGGAAGGA
C24V
TTACTCACATGGAATGACACTGGAAGACACAATGGTCATAGCATGGTGCA
2551 C24V
ATACTCGCCTGGAATGAAGCTGGAAGACACAATGGTCATTGCTTGGTGCG f AAGATGGTAAGTTAACATATTATGCAAGGTGCACTAGGGA
2601
AAGATGGGAAGTTTATGTACCTCCAAAGATGCACGAGAGAAACCAGATAT
C24V
CTTGCAATTiTGCATTCAAiAGCCTTACCiACCAGTGTG'iTATTCAAAAi
2657
CTCGCAATCTTGCATACA~GAGCCTTG~~~~~~~~~~~~~~~~~~A~A
IIIIIIII
II
IIIlIIII
II
llllllllllllllIlIIIII
I IIIIII
II
IIIIIIII
II
II
I IIIIIIII
IIIII
lllll
lIlIIIIII
IIIIIIIIIIIII
I II
IIIIIII
IIIIIIIIIIllIIIIIll
II
I II
IIIII
2331
IIIIIII
IIIIII
II
IIIIII
III
IIIIIIIIII
II
NADL Oregon NADL
2707
ACTCTTTGATGGGCGAAAGCAAGAGGATGTAGTCGAAATGAACGAC~CT
C24V
TTGAATTTGGACTCTGCCCATGCGATGCCAAACCCATAGTAAGAGGGACT
2151
TTGAATTTGGACTCTGCCCATGTGATGCCAAACCCA'FAGGAAG
II
IIII
I
IIIIIIIIl
lllllllllllllllllllIII
IIIIII
2606
2656
2706
.
ACTTTTCGAGGGGCAAGGGCAAGAGGACACAGTCGAAATGGATGACAACT
II
2556
IIIlIIIIIIIIIIIIIIIIIII
C24V
III
IIIlIIIII
II
2456
2506
llIIlIIIIIII
r
Oregon
I
Ill
II
IIIII
2406
II
IIIIIIIIIII
IIIIlIIIIIIIIIIIIlII
I
IIIII
IIIIIIIIII
III
2356
IIIIlIIIII
I IIIIIII
2756
IIIllllllllllllllllllllll
2806
f Oregon NADL Oregon NADL Oregon NADL Oregon NADL
C24V
TACAATACAACACTGCTAAACGGACCAGCCTTCCAGATG;
2807
TTCAATACAACGCTGCTGAACGGACCGGCCTTCCAGATGGTATGCCCCAJ
C24V
AGGTTGGACAGGGACTGTGAGCTGTATGTTAGCTAATAGAGACACCCTA~
2851 C24V
AGGATGGACAGGGACTGTAAGCTGTACGTCATTCAATATGGACACCTTAG f/r . ACACAGCAGTAGTACGGACGTATAGAAGGTCCAGACCAGACCATTCCCTTATAGG
2907
CCACAnCTGTGGTeCGGhCATATAGAAGGTCTAAACCATTCCCTCATAGG
C24V
CAAGGCTG;iTTACCCAAAiGACCCTGGG;;GAGGATCTC+ATGACTGTAi
2957
CAAGGCTGTATCACCCAAAAGAATCTGGGGGAGGATCTCCATAACTGCAT
I IlIIIIIII
III
lllll
IIIIIlII
IIIIIIIIIIIlII
III1
IIIIIlIIIII
I II
lIlllllllllllllllllllll
IIIIIII
IIIlIIII
IIIIlIIIII
II
I
IIIIIIlIIII
IIIIIIIIIIIIIII
IIII
2856
IIIIII
I IIIIIIIlII
III
IIIII
II
IIII
2906
2956
II
3006'
FIG. 1. Gp53 nucleotide sequence of Oregon C24V aligned to that of CVL-NADL and showing oligonucleotide (2337-3486) alongside NADL sequence of Collett eta/. (1988a) is from 5’start of genome. Underlined, PCR primer sequencing primer locations; f, forward primer; r. reverse primer
WB165c epitopes were not lost from any of the mutants, but again their relationship to one another could not be established. Nucleotide sequence of the gp53 gene of BVDV strains Oregon C24V and NADL Since PCR products inevitably have 3’ and 5’ ends that are exactly homologous to the primers used in their synthesis (and not necessarily the DNA being amplified), the reliable sequence obtained was: Oregon C24V and variants, nucleotides 2357-3466 and NADL and variants, nucleotides 2357-2854 plus 2875-3466. Sequencing revealed unique nucleotide
primer locations. locations; broken
Numbering underlined,
changes in a minority of individual clones from the same virus. It was considered likely that these had resulted from reverse transcriptase or Taq-polymeraseinduced copying errors and they were therefore discounted. The gp53 NADL sequence determined in this study differed from that obtained by Collett et al. (1988a) for the same strain by four nucleotide changes, two of which encoded amino acid substitutions: leutine to phenylalanine and valine to glutamic acid at residues 745 and 771, respectively (Fig. 3). The nucleotide sequence for gp53 of Oregon C24V is shown in Fig 1 aligned to that of CVL-NADL. Figure 3 shows the amino acid sequence for part of the gp53 of NADL aligned to the deduced sequence of Oregon C24V and
EPITOPES
Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL
ON THE
gp53
PROTEIN
OF
BVDV
C24V
TCTTGGAGGAAACTGGACTTGTGTAACTGGGGACCAACTACAATACACAG
3007
CCTTGGAGGAAATTGGACTTGTGTGCCTGGAGACCAACTACTATACAAAG
C24V
GAGGCTCTG~TGAATCTTG>GGTGT'iGTTTTAAAT&XAAAAAG~
3057
GGGGCTCTATTGAATCTTGCAAGTGGTGTGGCTATCAATTTAAAGAGAGT
C24V
GAGGGACTA&ACACTACC&ATTGGCAA'iTGTAGGTTGiAGAACGAGAi
3107
C24V
GAGGGACTACCACAC~@X~A~~~~~A+X'~TAAATTGGAGAACGAGAC . f/r . TGGCTACAGATTCGTGGATGGTACCTCTTGCAACAGAGAAGGTGTGGCCA
3157
TGGTTACAGGCTAGTAGACAGTACCTCTTGCAATAGAGAAGGTGTGGCCA
C24V
TAGTGCCACAAGGACTGGTAAAGTGTAAGATAGGAGACACAATTGTACAG
3207
TAGTACCACAAGGGACATTAAAGTGCAAGATAGGAAAAACAACTGTACAG
C24V
GTCATAGCTCTTGACACCAAACTTGGGCCTATGCCTTGCAAGCCATATGA
3257
GTCATAGCTATGGATACCAAACTCGGG~~~~~~~~~~~~~~~~~~~~~~A
C24V
f GATCATACCiAGCGAGGGGCCTGTGGAAAiGACGGCGTGiACCTTCAACi
3307
AATCATATCAAGTGAGGGGCCTGTAGAAAAGACAGCGTGTACTTTCAACT
C24V
ACACAAGGACATTGAAAAATAAATATTTTGAGCCCAGAGACAGTTACTTC
3357
ACACTAAGACATTAAAAAATAAGTATTTTGAGCCCAGAGACAGCTACTTT
C24V
CAACAATATATGCTAAAAGGAGAGTATCAATACTGGTTTGACCTGGAGGT
3407
CAGCAATACATGCTAAAAGGAGAGTATCAATACTGGTTTGACCTGGAGGT
C24V
CACTGACCAT
3457
GACTGACCATCACCGGGATTACTTCGCTGA r
IIIIIlIIIII
IIIlIIIIIIl
I IIIIII
IIII
IIIIIlIIII
IIIII
III1
IIIIlIIlIIIlI
IIIIlIII
IIIIIII
IIIIIIIII
III1
I IIIIII
II lllll
IIIIIIIlII
I III1
IIIIIIIIIIIIlIII
IIIIIIIIIII
IIIIIIII
IIIIIIII
IIIIIII
II IIIIIII
IIIIIIIIIlIIIIIIIIII
IIIII
IIIlIIIIIIIIIIIlIlIlIIIIIIIIIIIIIIIIlIIII
IIIIIIIII
3056
3106
3156
3206
IIIIIlII
IIIII
II
II I III
Illlllllllllllll
IIIIIIIII
I II IIIIIIlI
IIIIII
IIII
III
I II II
IIIII
I I IIII
IIIIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII III
767
3256
3306
3356
3406
3456
3486
FIG. 1 -ConGnued
to published sequences for other sequence begins at the putative due 693 of the protein-coding nome. Comparison of the Oregon quence from bases 2357-3466,
strains of BVDV. signalase site at sequence of the C24V nucleotide with published
The resigesedata
for the homologous regions of other (NADL: Collett et a/., 1988a; Osloss: 1987, and Singer, nucleotides 2357-31 1987) showed nucleotide homologies and 86%, compared to 60% homology
TABLE BINDING AND NEUTRALIZATION
OF NADL
AND OREGON C24V
1
VIRUSES BY MAW AND EPITOPE CONSERVATION Neutralizing
BVDV MAb
strains recognized (out of 92)
WB166n” WB2 14n WBl58n WB170n WB115c” WB162c WB 163~ WB165c WB2 15~ Note. BVDV strains recognized: replicates of 50 TCID,, of BVDV. a n/c: mAb raised against BVDV
84 82 3 1 35 71 69 9 75 recognition NADUOregon
in PLA (Paton C24V.
BVDV strains Renard et al,, 12; Dale eta/., of between 74 for comparison
AMONG
OTHER STRAINS OF BVDV
titer
MAb
NADL
C24V
NADL
26,900 45,300 3,365 42
190, 763-772
Epitope
(1992)
Mapping
of the gp53 Envelope D. J. PATON,*,’
*Central
Veterinary
Protein
J. P. LOWlNGS,*
AND
Laboratory! Weybridge, Surrey, KT15 3NB, University of Surrey, Guildford, Surrey, Received
April
13, 1992; accepted
of Bovine Viral Diarrhea
Virus
A. D. T. BARRETTt U.K.; and tSchool GU2 5X 17, U.K.
of Biological
Sciences,
July I, 1992
Epitopes recognized by nine monoclonal antibodies (mAbs) on the envelope protein, gp53, of two strains of bovine viral diarrhoea virus (NADL and Oregon C24V) were mapped by competitive binding assays and by the characterization and sequence analyses of mAb neutralization escape mutants. This defined an antigenic domain on gp53 that was shared by many BVDV strains, while other less conserved epitopes were possibly distinct. Sequencing of escape mutant viruses revealed that a cluster of three amino acids in the N-terminal half of gp53 were involved in the main antigenic domain shared by both NADL and Oregon C24V viruses, while an amino acid 31 residues further toward the N-terminus was involved in a second site present only on the NADL strain. Since other amino acids defining these epitopes were located at distant positions within the gp53 protein, it is likely that both domains on gp53 consist of composite, conformation-dependent epitopes.
INTRODUCTION
(Collett et al., 198813) as has the El gene of HCV (Stark eta/., 1990; Moormann eta/., 1990). The N-terminus of gp53/El is indicated by a putative signalase cleavage site in the nucleotide sequence and the C-terminus by a region of hydrophobic amino acids that represent the membrane-anchoring site (van Zijl et a/., 1991). The deduced amino acid sequence of gp53 is about 370 residues long. Antigenic variation exists between different strains of BVDV (Heuschele, 1975; Edwards et al., 1988) and both epitope conservation studies with mAbs (Paton et al., 1991 b) and genomic sequence comparisons (Moormann et al., 1990) point to gp53 as a site that contributes to such variation. Knowledge of the organization of gp53 epitopes will help in the understanding of antigenic variation, of antibody-mediated neutralization of virus infectivity, and ultimately in the generation of appropriate vaccines. Using mAb competition studies, three to four antigenie domains have been described for BVDV gp53 (Moennig et al., 1989; Xue et a/., 1990) while Wensvoort (1989) defined four such domains for El of HCV. Van Rijn et al. (1992), extending the work of Wensvoort (1989) have indicated that most El epitopes are situated in the N-terminal half of the protein, although their precise locations were not demonstrated. In this paper we report on the mapping of two antigenic domains on the gp53 of BVDV, showing that the arrangement of epitopes differs for the two virus strains studied. We provide evidence for the location of the domains and demonstrate that they are probably defined by discontinuous sequences of amino acids.
Bovine viral diarrhea virus (BVDV) is an important pathogen of cattle worldwide. Along with the closely related viruses of ovine border disease and hog cholera (HCV), it makes up the pestivirus genus recently reclassified within the Flaviviridae (Francki eta/., 1991). BVDV is enveloped and contains a single-stranded, positive-sense RNA molecule of approximately 12.5 kb (Renard et a/., 1985). The RNA has a single open reading frame that encodes a number of structural and nonstructural polypeptides that are cotranslationally and post-translationally cleaved from polyprotein precursors (Collett et al., 1988a,b). Three putative, structural glycoproteins have been ascribed to BVDV (Thiel et al., 1991) the largest of which, gp53, has an apparent molecular weight of between 51 and 58 kDa, depending in part on glycosylation differences (Donis and Dubovi, 1987). It is an immunodominant protein to which neutralizing monoclonal antibodies (mAbs) are directed (Donis et a/., 1988). The equivalent, major envelope protein of HCV, El, is known to occur as both a homodimer and as a heterodimer complexed with a smaller glycoprotein, gp33 (Wensvoort et al., 1990; Thiel et a/., 1991). Vaccination with Aujeszky’s disease virus expressing only El of HCV has been shown to protect against HCV challenge (van Zijl et a/., 1991). The approximate position of the gp53 gene of BVDV has been mapped toward the 5’ end of the genome
’ To whom
reprint
requests
should
be addressed. 763
0042-6822192
$5.00
764
PATON.
MATERIALS Viruses
AND
LOWINGS.
METHODS
and mAbs
Viruses were grown in calf testis (CT) cells at passage 4-9 or in a bovine turbinate cell line (BT) as described by Paton et al. (199 1 a). The Oregon C24V virus (Gillespie et al., 1960) was cloned by three cycles of limiting dilution and then passaged once more prior to use (a total of 67 in vitro passages), while the NADL strain (Gutekunst and Malmquist, 1963) had been twice plaque purified and then passaged seven times before use (a total of 12 in vitro passages). Anti-BVDV mAbs used in this study were raised against either NADL or Oregon C24V viruses (Table 1). Production of the mAbs and their previous characterization has been reported by Edwards et a/. (1988) and Paton et al. (1991 b). Some of these data are summarized in Table 1, which shows the neutralization titer of each mAb for BVDV strains NADL and Oregon C24V and the relative conservation of the different epitopes among 92 different strains of BVDV. Only mAb WB214n showed significant neutralization of both strains. MAb binding
studies
Recognition of viruses by mAbs was assessed by peroxidase-linked assay (PLA; Holm Jensen, 1981) using infected and mock-infected BT or CT cells grown in 96-well plates (Nunc). After fixing the cells with 20% acetone in PBS, pH 7.6, containing 200 mg/liter bovine serum albumin, predetermined optimal dilutions of mAbs were added and incubated for 15 min at 37”. Bound mAbs were detected with a rabbit anti-mouse peroxidase conjugate and the substrate 3-amino-gethylcarbazole. Staining of infected cells was assessed microscopically with reference to the mock-infected wells. Competition
studies
Qualitative results of competitive binding between these mAbs have been reported previously (Paton et a/., 1991 b). The method, a modified PLA, was altered by use of a soluble color substrate TMB (0.1 mg/ml tetramethyl benzidine in phosphate-citrate buffer, pH 5, with 0.0075% hydrogen peroxide) to permit quantitative assessment of competition. BT cells were prepared in microtiter plates as for the PLA. Unlabeled mAbs were in the form of murine ascitic fluids precipitated with ammonium sulfate and redissolved at 2.5 mg/ml in phosphate-buffered saline, pH 7.6, containing 50% glycerol. Serial dilutions of these mAbs were added to the fixed cell monolayers as blockers, immediately before adding biotinylated competitor mAbs. The
AND
BARREll
biotinylated mAbs, which had been previously titrated in the assay system, were used at twice the minimum concentration for plateau binding. Binding of biotinylated mAbs was detected by a biotin streptavidin-peroxidase complex (Amersham, U.K.) and TMB stopped with 2 M sulfuric acid. Color development was recorded by a spectrophotometer at a wavelength of 450 nm and net optical densities were calculated from the difference between the values for infected and mockinfected wells. The percentage inhibition of binding was calculated from the ratio of the net optical density when biotinylated mAb competed with either an irrelevant unlabeled mAb or unlabeled anti-gp53 mAbs. MAb neutralization
escape
mutants
Viruses that escaped neutralization by mAbs were selected by incubating sequential, 1O-fold virus dilutions with equal volumes of mAb at concentrations of between 0.12 and 0.20 mg/ml. After 1 hr at 37”, cell suspension was added and incubation continued for a further 3 days. Supernatant fluids were retained, while monolayers were fixed and immunostained using a pan-pestivirus-reactive anti-p80 mAb as previously described (Paton et a/., 1991 b). Supernatants from cultures containing only plaques of stained cells, at virus dilutions at which 100% infection would have been expected in the absence of mAb, were passaged twice without antibody to produce stocks of potential escape mutant viruses. These mutants received one further passage, either for characterization with mAbs or for extraction of RNA for molecular cloning. Each stock was titrated in 1 O-fold steps and stained with the selecting mAb and an anti-p80 mAb. Only those stocks which were recognized by the anti-p80 mAb and not the selecting mAb were retained as escape mutant viruses. Escape mutants were titrated in lo-fold steps with and without the presence of mAbs (final dilution 0.05 mg/ml) to test whether they resisted neutralization. They were also examined for recognition by the full panel of gp53 mAbs to assess epitopic changes. All escape mutant viruses were named after the mAb/ virus combination selected from (e.g., Al 214n was selected from NADL with WB214n, while C3 214~ was selected from Oregon C24V with the same mAb) and all were derived from separate wells. Nucleotide
sequencing
Because of uncertainty over the exact position of gp53, all BVDV nucleotide and amino acid numbering is from the start of the genome and the beginning of the open reading frame, respectively, (as described by Collett et al., 1988a, for BVDV strain NADL). Gp53 was sequenced for both parental strains of virus and for eight escape mutants (four each of NADL and Oregon
EPITOPES
ON
THE
C24V). Total cellular RNA was extracted from BVDV-infected or mock-infected CT cells and virus-specific cDNA was synthesized using moloney murine leukaemia virus reverse transcriptase and amplified by the polymerase chain reaction (PCR) using Taq DNA polymerase, as described by Roehe and Woodward (1991). Oligonucleotide primers for cDNA synthesis and amplification were derived from published BVDV sequence data (Collett et al., 1988a) and were synthesized on an Applied Biosystems 381A DNA synthesizer. Single fragments of Oregon C24V and variant viruses were amplified with the PCR primer combination: forward 2336-2357Ireverse 3486-3467. NADL and variant viruses were amplified as two fragments with two pairs of primers: forward 2336-2356/reverse 2874-2855 and forward 2855-2874lreverse 34853466. Each primer began with an additional 9-l 0 nucleotides of nonviral sequence, encoding a specific endonuclease restriction site. PCR-derived cDNA was purified by low-melting-point agarose gel electrophoresis, extracted with a silica matrix (Prepagene, Bio-Rad), cut by restriction endonucleases to give cohesive ends, and ligated into similarly digested, calf alkaline-phosphatase-treated phagemid vectors (pBluescript II KS+, Stratagene) according to the supplier’s instructions. The fscherichia co/i strain XL1 -Blue (Stratagene) was transformed with the phagemid construct using the CaCI, method (Sambrook et al., 1989) and singlestranded, insert-specific DNA was isolated using the helper phage VCSM 13 and the Stratagene protocol (based on the method of Vieira and Messing, 1987). The single-stranded DNA was sequenced using the Sequenase (United States Biochemical) protocol and oligonucleotide primers derived from Bluescript or BVDV sequence data. The locations of the primers for both PCR and sequencing are marked in Fig. 1. All genes were sequenced in both directions using two to six clones per virus; increased numbers of clones being sequenced across regions where differences between viruses were detected. The region amplified and sequenced extended from nucleotide 2336 to 3486, beginning upstream of the putative signalase cleavage site between bases 2461 and 2462 and ending just before the start of the putative membrane-anchoring region (3491) probably about 91 nucleotides short of the C-terminus of gp53. Sequence data was analyzed using the GCG software package (Devereux et al., 1984) via the Daresbury Laboratory of the Science and Engineering Research Council. RESULTS MAb binding
studies
At optimal mAb dilutions, background staining was minimal. No significant differences in mAb binding
gp53
PROTEIN
OF
BVDV
765
were detected whether BT or CT cells were used. Recognition of the NADL and Oregon C24V strains of BVDV by the mAb panel is shown in Table 1. Competition
studies
The results of competition experiments for NADL and Oregon C24V viruses are shown in Table 2. WB166n and WB170n which were not successfully biotinylated were only assessed as blockers. All of the mAbs, except WB165c, blocked the binding of at least one other biotinylated mAb, while each biotinylated mAb, except for WBl58n, was blocked by at least one other unbiotinylated mAb. Reciprocal competition of >45% between mAbs is shown in Fig. 2. One-way blocking data concerning WBl66n and WB170n was also included in this figure as the possibility of reciprocal block could not be excluded for these mAbs. If reciprocal blocking is taken as an indication of spatial proximity, it can be seen from Fig. 2 that, for both viruses, many of the epitopes appeared to cluster within a single domain and only that of WBl58n was clearly distinct. The four epitopes present on both viruses (those of WB166n, WB214n, WB215c, and WB163c) showed similar, but not identical interrelationships whichever virus was used as antigen. A single mAb combination showed evidence of binding synergism in that WB162c increased the binding of biotinylated WB165c by 54%. MAb neutralization
escape
mutants
Mutant viruses, which escaped neutralization by their selecting mAb, were obtained for all neutralizing mAb/virus combinations except for WB163c/Oregon C24V and WB170n/NADL. The log,0 neutralization of parental virus by the selecting mAbs was between 2.5 and 3.75. All mutants showed at least a two log,0 reduction in neutralization, in the presence of the inducing mAb, compared to that seen for the parent virus with the same mAb. Some mutants showed a similar order of neutralization resistance with additional mAbs. Table 3 summarizes the pattern of mAb neutralization resistance of the mutants and also shows their loss of binding epitopes. Where binding epitopes were lost, multiple losses were considered suggestive of epitopic overlap. On this basis, NADL epitopes were split into three groups: (1) WBI 66n and WB215c, (2) WB214n,and(3)WB158n.TheWB170nandWB163c epitopes seemed distinct, in that they were not lost from any of the NADL mutants, but their relationship to one another was not established. For Oregon C24V virus, the pattern of epitope loss suggested a domain consisting of the epitopes of WB214n, WB166n, WB162c, WBl15c, and WB215c. The WB163c and
PATON,
766
LOWINGS,
AND NADL
Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL
BARRETT
C24v
AGATTTAACACGCATTTGG .f AACGCTGCCACGACTACTGCATTCCTAGTATGCCTTGTTAGATGGTCAG
2351
AACGCTGCAACAACTACTGCTTTTTTAGTATGCCTTGTTAAGATAGTCAG
C24~
GGGCCAGAT;;GTACAGGGCATCCTATGGCTACTACTGAT~CAGGAGTG~
2407
GGGCCAGATGGTACAGGGCATTCTGTGGCTACTATTGATACAGGGGTAC
C24V
AGGGGCACCTAGACTGCAAACCTGAATACTCATATGCCATAGCCAAGAGT
2451
AAGGGCACTTGGATTGCAAACCTGAATTCTCGTATGCCATAGCAAAGGAC
C24V
GATAGAATT;;GCCTACAAG;;AGCTGAAGA&TTACTACThTTTGGAAGGi
2507
GAAAGAATTGGTCAACTGGGGGCTGAAGGCCTTACCACCACTTGGAAGGA
C24V
TTACTCACATGGAATGACACTGGAAGACACAATGGTCATAGCATGGTGCA
2551 C24V
ATACTCGCCTGGAATGAAGCTGGAAGACACAATGGTCATTGCTTGGTGCG f AAGATGGTAAGTTAACATATTATGCAAGGTGCACTAGGGA
2601
AAGATGGGAAGTTTATGTACCTCCAAAGATGCACGAGAGAAACCAGATAT
C24V
CTTGCAATTiTGCATTCAAiAGCCTTACCiACCAGTGTG'iTATTCAAAAi
2657
CTCGCAATCTTGCATACA~GAGCCTTG~~~~~~~~~~~~~~~~~~A~A
IIIIIIII
II
IIIlIIII
II
llllllllllllllIlIIIII
I IIIIII
II
IIIIIIII
II
II
I IIIIIIII
IIIII
lllll
lIlIIIIII
IIIIIIIIIIIII
I II
IIIIIII
IIIIIIIIIIllIIIIIll
II
I II
IIIII
2331
IIIIIII
IIIIII
II
IIIIII
III
IIIIIIIIII
II
NADL Oregon NADL
2707
ACTCTTTGATGGGCGAAAGCAAGAGGATGTAGTCGAAATGAACGAC~CT
C24V
TTGAATTTGGACTCTGCCCATGCGATGCCAAACCCATAGTAAGAGGGACT
2151
TTGAATTTGGACTCTGCCCATGTGATGCCAAACCCA'FAGGAAG
II
IIII
I
IIIIIIIIl
lllllllllllllllllllIII
IIIIII
2606
2656
2706
.
ACTTTTCGAGGGGCAAGGGCAAGAGGACACAGTCGAAATGGATGACAACT
II
2556
IIIlIIIIIIIIIIIIIIIIIII
C24V
III
IIIlIIIII
II
2456
2506
llIIlIIIIIII
r
Oregon
I
Ill
II
IIIII
2406
II
IIIIIIIIIII
IIIIlIIIIIIIIIIIIlII
I
IIIII
IIIIIIIIII
III
2356
IIIIlIIIII
I IIIIIII
2756
IIIllllllllllllllllllllll
2806
f Oregon NADL Oregon NADL Oregon NADL Oregon NADL
C24V
TACAATACAACACTGCTAAACGGACCAGCCTTCCAGATG;
2807
TTCAATACAACGCTGCTGAACGGACCGGCCTTCCAGATGGTATGCCCCAJ
C24V
AGGTTGGACAGGGACTGTGAGCTGTATGTTAGCTAATAGAGACACCCTA~
2851 C24V
AGGATGGACAGGGACTGTAAGCTGTACGTCATTCAATATGGACACCTTAG f/r . ACACAGCAGTAGTACGGACGTATAGAAGGTCCAGACCAGACCATTCCCTTATAGG
2907
CCACAnCTGTGGTeCGGhCATATAGAAGGTCTAAACCATTCCCTCATAGG
C24V
CAAGGCTG;iTTACCCAAAiGACCCTGGG;;GAGGATCTC+ATGACTGTAi
2957
CAAGGCTGTATCACCCAAAAGAATCTGGGGGAGGATCTCCATAACTGCAT
I IlIIIIIII
III
lllll
IIIIIlII
IIIIIIIIIIIlII
III1
IIIIIlIIIII
I II
lIlllllllllllllllllllll
IIIIIII
IIIlIIII
IIIIlIIIII
II
I
IIIIIIlIIII
IIIIIIIIIIIIIII
IIII
2856
IIIIII
I IIIIIIIlII
III
IIIII
II
IIII
2906
2956
II
3006'
FIG. 1. Gp53 nucleotide sequence of Oregon C24V aligned to that of CVL-NADL and showing oligonucleotide (2337-3486) alongside NADL sequence of Collett eta/. (1988a) is from 5’start of genome. Underlined, PCR primer sequencing primer locations; f, forward primer; r. reverse primer
WB165c epitopes were not lost from any of the mutants, but again their relationship to one another could not be established. Nucleotide sequence of the gp53 gene of BVDV strains Oregon C24V and NADL Since PCR products inevitably have 3’ and 5’ ends that are exactly homologous to the primers used in their synthesis (and not necessarily the DNA being amplified), the reliable sequence obtained was: Oregon C24V and variants, nucleotides 2357-3466 and NADL and variants, nucleotides 2357-2854 plus 2875-3466. Sequencing revealed unique nucleotide
primer locations. locations; broken
Numbering underlined,
changes in a minority of individual clones from the same virus. It was considered likely that these had resulted from reverse transcriptase or Taq-polymeraseinduced copying errors and they were therefore discounted. The gp53 NADL sequence determined in this study differed from that obtained by Collett et al. (1988a) for the same strain by four nucleotide changes, two of which encoded amino acid substitutions: leutine to phenylalanine and valine to glutamic acid at residues 745 and 771, respectively (Fig. 3). The nucleotide sequence for gp53 of Oregon C24V is shown in Fig 1 aligned to that of CVL-NADL. Figure 3 shows the amino acid sequence for part of the gp53 of NADL aligned to the deduced sequence of Oregon C24V and
EPITOPES
Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL Oregon NADL
ON THE
gp53
PROTEIN
OF
BVDV
C24V
TCTTGGAGGAAACTGGACTTGTGTAACTGGGGACCAACTACAATACACAG
3007
CCTTGGAGGAAATTGGACTTGTGTGCCTGGAGACCAACTACTATACAAAG
C24V
GAGGCTCTG~TGAATCTTG>GGTGT'iGTTTTAAAT&XAAAAAG~
3057
GGGGCTCTATTGAATCTTGCAAGTGGTGTGGCTATCAATTTAAAGAGAGT
C24V
GAGGGACTA&ACACTACC&ATTGGCAA'iTGTAGGTTGiAGAACGAGAi
3107
C24V
GAGGGACTACCACAC~@X~A~~~~~A+X'~TAAATTGGAGAACGAGAC . f/r . TGGCTACAGATTCGTGGATGGTACCTCTTGCAACAGAGAAGGTGTGGCCA
3157
TGGTTACAGGCTAGTAGACAGTACCTCTTGCAATAGAGAAGGTGTGGCCA
C24V
TAGTGCCACAAGGACTGGTAAAGTGTAAGATAGGAGACACAATTGTACAG
3207
TAGTACCACAAGGGACATTAAAGTGCAAGATAGGAAAAACAACTGTACAG
C24V
GTCATAGCTCTTGACACCAAACTTGGGCCTATGCCTTGCAAGCCATATGA
3257
GTCATAGCTATGGATACCAAACTCGGG~~~~~~~~~~~~~~~~~~~~~~A
C24V
f GATCATACCiAGCGAGGGGCCTGTGGAAAiGACGGCGTGiACCTTCAACi
3307
AATCATATCAAGTGAGGGGCCTGTAGAAAAGACAGCGTGTACTTTCAACT
C24V
ACACAAGGACATTGAAAAATAAATATTTTGAGCCCAGAGACAGTTACTTC
3357
ACACTAAGACATTAAAAAATAAGTATTTTGAGCCCAGAGACAGCTACTTT
C24V
CAACAATATATGCTAAAAGGAGAGTATCAATACTGGTTTGACCTGGAGGT
3407
CAGCAATACATGCTAAAAGGAGAGTATCAATACTGGTTTGACCTGGAGGT
C24V
CACTGACCAT
3457
GACTGACCATCACCGGGATTACTTCGCTGA r
IIIIIlIIIII
IIIlIIIIIIl
I IIIIII
IIII
IIIIIlIIII
IIIII
III1
IIIIlIIlIIIlI
IIIIlIII
IIIIIII
IIIIIIIII
III1
I IIIIII
II lllll
IIIIIIIlII
I III1
IIIIIIIIIIIIlIII
IIIIIIIIIII
IIIIIIII
IIIIIIII
IIIIIII
II IIIIIII
IIIIIIIIIlIIIIIIIIII
IIIII
IIIlIIIIIIIIIIIlIlIlIIIIIIIIIIIIIIIIlIIII
IIIIIIIII
3056
3106
3156
3206
IIIIIlII
IIIII
II
II I III
Illlllllllllllll
IIIIIIIII
I II IIIIIIlI
IIIIII
IIII
III
I II II
IIIII
I I IIII
IIIIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII III
767
3256
3306
3356
3406
3456
3486
FIG. 1 -ConGnued
to published sequences for other sequence begins at the putative due 693 of the protein-coding nome. Comparison of the Oregon quence from bases 2357-3466,
strains of BVDV. signalase site at sequence of the C24V nucleotide with published
The resigesedata
for the homologous regions of other (NADL: Collett et a/., 1988a; Osloss: 1987, and Singer, nucleotides 2357-31 1987) showed nucleotide homologies and 86%, compared to 60% homology
TABLE BINDING AND NEUTRALIZATION
OF NADL
AND OREGON C24V
1
VIRUSES BY MAW AND EPITOPE CONSERVATION Neutralizing
BVDV MAb
strains recognized (out of 92)
WB166n” WB2 14n WBl58n WB170n WB115c” WB162c WB 163~ WB165c WB2 15~ Note. BVDV strains recognized: replicates of 50 TCID,, of BVDV. a n/c: mAb raised against BVDV
84 82 3 1 35 71 69 9 75 recognition NADUOregon
in PLA (Paton C24V.
BVDV strains Renard et al,, 12; Dale eta/., of between 74 for comparison
AMONG
OTHER STRAINS OF BVDV
titer
MAb
NADL
C24V
NADL
26,900 45,300 3,365 42
