Mutations in the 5′ Nontranslated Region of Bovine Viral Diarrhea Virus Result in Altered Growth Characteristics

Abstract

The 5′ nontranslated region (NTR) of pestiviruses functions as an internal ribosome entry site (IRES) that mediates cap-independent translation of the viral polyprotein and probably contains additional cis-acting RNA signals involved in crucial processes of the viral life cycle. Computer modeling suggests that the 5′-terminal 75 nucleotides preceding the IRES element form two stable hairpins, Ia and Ib. Spontaneous and engineered mutations located in the genomic region comprising Ia and Ib were characterized by using infectious cDNA clones of bovine viral diarrhea virus. Spontaneous 5′ NTR mutations carrying between 9 and 26 A residues within the loop region of Ib had no detectable influence on specific infectivity and virus growth properties. After tissue culture passages, multiple insertions and deletions of A residues occurred rapidly. In contrast, an engineered mutant carrying 5 A residues within the Ib loop was genetically stable during 10 tissue culture passages. This virus was used as starting material to generate a number of additional mutants. The analyses show that (i) deletion of the entire Ib loop region resulted in almost complete loss of infectivity that was rapidly restored during passages in cell culture by insertions of variable numbers of A residues; (ii) mutations within the 5′-terminal 4 nucleotides of the genomic RNA severely impaired virus replication; passaging of the supernatants obtained after transfection resulted in the emergence of efficiently replicating mutants that had regained the conserved 5′-terminal sequence; (iii) provided the conserved sequence motif 5′-GUAU was retained at the 5′ end of the genomic RNA, substitutions and deletions of various parts of hairpin Ia or deletion of all of Ia and part of Ib were found to support replication, but to a lower degree than the parent virus. Restriction of specific infectivity and virus growth of the 5′ NTR mutants correlated with reduced amounts of accumulated viral RNAs.

Bovine viral diarrhea virus 1 (BVDV-1), BVDV-2, Classical swine fever virus, and Border disease virus are members of the genus Pestivirus, which together with the genera Flavivirus and Hepacivirus constitute the family Flaviviridae (2, 35). Pestiviruses cause economically important livestock diseases such as classical swine fever, bovine viral diarrhea, and mucosal disease. The pestivirus genome consists of a positive-stranded nonpolyadenylated RNA molecule of approximately 12.3 kb that contains one large open reading frame (ORF) flanked by 5′ and 3′ nontranslated regions (NTRs) (5, 12, 28, 36). In the virus-encoded polyprotein, the mature viral proteins are arranged in the following order (from N to C terminus): N^pro, C, E^rns, E1, E2, p7, NS2-3, (NS2), (NS3), NS4A, NS4B, NS5A, and NS5B; N^pro refers to an N-terminal autoprotease, and E^rns refers to a structural glycoprotein with RNase activity (see references 30 and 43 for reviews). The structural proteins are represented by the capsid protein C and the three envelope proteins E^rns, E1, and E2 (44). The remaining proteins are presumably nonstructural (NS). Nonstructural protein NS3 (NS2-3) possesses multiple enzymatic activities, namely, serine protease (41, 49, 50), nucleoside triphosphatase (39), and helicase activity (47). For NS5B, the predicted RNA-dependent RNA polymerase (RdRp) activity has been directly demonstrated (21, 53).

Pestivirus RNA replication occurs in the cytoplasm of infected cells. For a complete RNA replication cycle, the genomic RNA is first copied into minus-strand RNA. The minus-strand RNA in turn serves as the template for plus-strand synthesis. Taking into account our knowledge about other positive-strand RNA viruses, RNA replication of pestiviruses requires the RdRp (NS5B), RNA sequence elements, and probably additional virus-encoded and cellular proteins. Recently, sequence and structural elements essentially implicated in RNA replication have been identified in the 3′-terminal region of BVDV genomic RNA (52).

Similar to picornaviruses and hepatitis C viruses (HCV), the 5′ NTR of pestiviruses functions as an internal ribosome entry site (IRES) that mediates cap-independent translation of the viral polyprotein (23, 33, 34, 37, 46). Based on functional studies and comparative sequence analyses, including computer-predicted RNA secondary-structure models, it has been reported that the structural and functional organization of the pestivirus 5′ NTR is similar to that of HCV. The 5′ border of the BVDV IRES is at the 5′ end of stem-loop II near nucleotide (nt) 75, and the 3′ border extends into the 5′ region of the ORF (11). The region preceding the IRES element has been predicted to form two hairpins, termed Ia and Ib (9, 16). The high conservation of nucleotide sequences and structural features within this region among different pestiviruses (4, 16) suggests the presence of cis-acting elements required for viral replication, including signals for plus-strand RNA synthesis, regulation of translation versus replication, and possibly also for packaging of the viral genome.

In this study, several spontaneous and engineered mutations within the genomic region comprising the 5′-terminal hairpins Ia and Ib of the pestivirus BVDV were analyzed. The availability of an infectious BVDV cDNA clone allowed characterization of the mutants with regard to (i) specific infectivity of the RNA, (ii) virus growth properties, and (iii) accumulation of viral genomic RNA. The data presented contribute to our understanding of sequence and structural elements within the 5′ NTR important for efficient replication of pestiviruses.

MATERIALS AND METHODS

Cells and viruses.

MDBK cells were obtained from the American Type Culture Collection (Rockville, Md.). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% horse serum. Cells were tested regularly for the absence of pestiviruses by reverse transcription (RT)-PCR and immunofluorescence (3). The cytopathogenic BVDV-1 strain CP7 has been described previously (13, 29).

Oligonucleotides.

Oligonucleotides were purchased from MWG Biotech GmbH (Ebersberg, Germany). The antisense primers Ol 200R (corresponding to nt 235 to 252 of the CP7-5A sequence) and Ol 380R (corresponding to nt 370 to 390 of the CP7-5A sequence) have been described previously (4). The sense primers Ol B54 (5′-CTAGCAAAACTGGCCATT-3′), Ol B54A (5′-GTTGAGAGTTCTGCTTATG-3′), and Ol B56 (5′-ACGTCCACGGTTGGACTAG-3′) correspond to nt 11907 to 11924, nt 12064 to 12082, and nt 12256 to 12274 of the CP7-5A sequence, respectively. Other primers used for construction of CP7 5′ NTR mutants are described below.

RT-PCR, molecular cloning, and nucleotide sequencing.

RT of heat-denatured RNA and PCR were done as described previously (3). The cDNA fragments obtained after RT-PCR were separated by agarose gel electrophoresis and purified using the Qiaex DNA purification kit (Qiagen, Hilden, Germany). The respective cDNA fragments were cloned using the TA cloning kit (Invitrogen, De Schelp, The Netherlands). Nucleotide sequences were determined by cycle sequencing using the Thermo Sequenase kit (Amersham Buchler, Braunschweig, Germany) and dye (IRD 800)-labeled standard primers (MWG Biotech). Analysis of sequencing gels was carried out with the DNA sequencer Li-Cor 4000 L (MWG Biotech). All sequences were determined by sequencing both complementary strands. Computer analysis of sequence data was performed using HUSAR (DKFZ, Heidelberg, Germany) which provides the GCG software package (17).

Analysis of 5′ and 3′ sequences.

For determination of the 5′- and 3′-terminal sequences of BVDV CP7 wild-type and mutant viruses, an RNA ligation method was employed. RNA from infected cells was prepared by using either the RNeasy total RNA kit (Qiagen) or the RNA Extraction kit (Pharmacia Biotech) as recommended by the supplier. Two micrograms of RNA was ligated by using 20 U of T4 RNA ligase (New England Biolabs) in a reaction mixture containing 50 mM Tris-HCl (pH 7.8), 10 mM MgCl₂, 1 mM β-mercaptoethanol, 1 mM ATP, and 16 U of RNase inhibitor (RNaseOUT; Gibco-BRL) for 4 h at 37°C in a volume of 10 μl. After phenol-chloroform extraction and ethanol precipitation, the pellet was resuspended in 10 μl of diethylpyrocarbonate-treated H₂O. A 2.5-μl volume of this solution was used for RT-PCR with primers Ol 380R and Ol B54. A second, nested PCR was performed with primers Ol 200R and Ol B54A (CP7 wild type) or Ol 200R and Ol B56 (CP7 mutants). For BVDV CP7 and each of the mutant viruses, at least 10 cDNA clones were subjected to nucleotide sequence analysis.

Construction of 5′ NTR mutant plasmids.

Construction of BVDV CP7 full-length cDNA was based on plasmids pA/BVDV (29) and HHDI9, which lacks the genomic region encoding the structural proteins and NS2 (42). For generation of plasmid HHDI9, the 5′-terminal 21 bases and the 3′-terminal 33 bases were derived from the BVDV Osloss sequence; immediately upstream of the viral cDNA, an NheI site and an SP6 RNA polymerase promoter were integrated (42). An XhoI (nt 222 to 227 of the CP7-5A sequence)-ClaI (nt 11075 to 11080 of the CP7-5A sequence) fragment from pA/BVDV (29) was inserted in plasmid HHDI9 that had been digested with XhoI and ClaI, resulting in plasmid pCP7-Os.

For construction of full-length CP7 cDNA clones carrying the authentic 5′ terminus and various numbers of A residues downstream of position 44, the cDNA clones obtained after RNA ligation and RT-PCR were used as templates for PCR with Ol 200R and Ol CP7-SP6 (5′-TACGCTAGCATTTAGGTGACACTATAGTATACGAGGTTAGGCAAGTTC-3′; the underlined region corresponds to nt 1 to 22 of the CP7-5A sequence; an SP6 RNA polymerase promoter preceded by an NheI site is located directly upstream of the CP7-specific sequence). Finally, the NheI-XhoI fragment of pCP7-Os was replaced with CP7-specific NheI-XhoI fragments carrying 5, 9, 20, 26, or 54 A residues following position 44, resulting in full-length cDNA clones pCP7-5A, pCP7-9A, pCP7-20A, pCP7-26A, and pCP7-54A, respectively.

For construction of mutant CP7-T, which lacks the entire loop region of hairpin Ib, a cDNA fragment obtained after PCR with primers Ol 200R and Ol CP7-SP6T (5′-TACGCTAGCATTTAGGTGACACTATAGTATACGAGGTT AGGCAAGTTCTCGTATACATATTGGACACTCTTTTAGGCCTAGGGG ACAA-3′; the underlined regions encompass nt 1 to 44 and nt 53 to 69 of the CP7-5A sequence, respectively) using pCP7-5A as the template was cloned in pCR2.1 (Invitrogen). After digestion with NheI and XhoI, this fragment was used to replace the NheI-XhoI fragment of CP7-5A, resulting in the full-length cDNA clone pCP7-T.

All other full-length CP7 cDNA clones carrying mutations within the 5′ NTR described in this study were cloned in a similar way. Briefly, each mutation was first introduced into the 5′-terminal sequence of CP7 by PCR with a sense primer encompassing the respective mutated sequence (preceded by an NheI site and the SP6 RNA polymerase promoter) and the antisense primer Ol 200R using plasmid pCP7-5A as the template. As a second step, the NheI-XhoI fragments of the resulting clones were introduced into pCP7-5A digested with NheI and XhoI. For all full-length CP7 cDNA clones, the sequence of the entire 5′ NTR was verified by nucleotide sequencing using dye IRD 800-labeled primer Ol 380R. All 5′ NTR mutations are described in detail in the Results. Details of the cloning strategies, including the sequences of sense primers used for construction of the individual BVDV CP7 5′ NTR mutants, are available on request.

Computer-predicted RNA folding.

Modeling of the RNA secondary structures was performed with the programs RNAFOLD, MFOLD, and FOLDANALYZE (17). Computer-predicted models of RNA secondary structures were used to assess the locations and putative changes of the RNA secondary structure for each mutant.

In vitro synthesis of RNA.

Full-length cDNA clones of CP7-5A and the 5′ NTR mutants were digested to completion with SmaI, extracted with phenol-chloroform, and precipitated with ethanol. One microgram of linearized plasmid DNA was transcribed with 50 U of SP6 RNA polymerase (BioWhittaker Europe, Verviers, Belgium) in 20 μl using standard conditions. Reaction mixtures were incubated for 1 h at 37°C. For degradation of the template DNA, the reaction mixture was digested with 5 U of DNase I (Boehringer GmbH, Mannheim, Germany) for 1 h at 37°C, followed by extraction with phenol-chloroform and precipitation with ethanol. Photometric quantification of the transcribed RNAs was carried out in a GeneQuantII photometer (Pharmacia). The quality and the calculated amount of each RNA were controlled by ethidium bromide staining of samples after agarose gel electrophoresis. RNA transcripts used for transfection contained >60% full-length RNA.

Transfection of RNA.

MDBK cells were trypsinized and washed by three rounds of centrifugation (780 × g for 5 min) and resuspension in phosphate-buffered saline (PBS) without Ca²⁺ and Mg²⁺. For each transfection, the confluent cells from a 10-cm-diameter dish were resuspended in 0.4 ml of PBS without Ca²⁺ and Mg²⁺ and mixed with 2 μg of in vitro-transcribed RNA immediately before the pulse (950 μF and 180 V). For electroporation, a Gene Pulser II (Bio-Rad, Munich, Germany) was used. The electroporated cells were seeded on two six-well dishes and adjusted to 2 ml with medium containing 10% horse serum; 1/10 of the transfected cells were used for a plaque assay in order to determine the specific infectivity of the RNA and the plaque sizes of the recovered viruses (see below). At 24 h posttransfection (p.t.), the portion of positive cells was assessed by indirect immunofluorescence analysis (IFA) using monoclonal antibody 8.12.7 (directed against NS3), kindly provided by E. J. Dubovi (Cornell University, Ithaca, N.Y.). For comparative analyses, the transcription-transfection experiments using the whole set of 5′ NTR mutants were performed in parallel and repeated several times to exclude that a technical error might account for the observed differences; these analyses included IFA, determination of virus yields after transfection, and plaque assays (see below).

Plaque assay.

MDBK cells were transfected with 2 μg of each RNA, and 10-fold serial dilutions of transfected cells together with 2 × 10⁶ untreated MDBK cells were seeded into six-well dishes. After incubation at 37°C for 4 h, the attached cells were overlaid with semisolid medium containing 0.6% low-melting-point agarose (Gibco-BRL) and 5% horse serum. The mutant viruses produced small plaques, which were visualized by immunostaining. After 6 days of incubation at 37°C, the agarose overlays were removed, and the cells were washed with PBS and then fixed with acetone-methanol (1:1) for 1 h at −20°C. After incubation with a mixture of BVDV E2-specific monoclonal antibodies (1:10 in PBS–0.05% Tween 20; kindly provided by E. Weiland, Tübingen, Germany) for 2 h at room temperature, monolayers were washed twice with PBS–0.05% Tween 20 and then incubated with peroxidase-conjugated goat anti-mouse immunoglobulin (1:500 in PBS–0.05% Tween 20; Sigma-Aldrich Chemie GmbH, Steinheim, Germany). After 1 h, the monolayers were washed twice with PBS, and plaques were visualized by using the peroxidase substrate 3-amino-9-ethyl-carbazole (Sigma-Aldrich Chemie GmbH).

Determination of growth kinetics.

MDBK cells (10⁶) in a six-well dish were infected with transcript-derived virus at a multiplicity of infection (MOI) of 0.05. After adsorption for 1 h at room temperature, the cells were washed six times with PBS and then overlaid with medium containing 10% horse serum followed by incubation for 4 days. After the indicated time intervals, aliquots (200 μl) of the cell culture supernatant were removed and used for titration on MDBK cells. The virus yields were determined as the titer of 50% tissue culture infectious doses (TCID₅₀) per milliliter.

Analysis of viral RNA synthesis by Northern blot.

MDBK cells (10⁶) were infected with transcript-derived virus at an MOI of 0.05 and processed in parallel to cells used for determination of growth kinetics. After 2 and 3 days of incubation at 37°C, RNA was prepared using the RNeasy total RNA kit (Qiagen). Five micrograms of glyoxylated RNA (26) was separated in a phosphate-buffered 1.0% agarose gel containing 5.5% formaldehyde and transferred to Duralon-UV membranes (Stratagene, Heidelberg, Germany). Radioactive labeling of the probe, hybridization, and washing conditions were as described (1). A 2.5-kb NotI-NsiI fragment from the cDNA clone pA/BVDV was used as a probe (29). The viral genomic RNAs were detected by autoradiography, and the intensity of bands was determined with a phosphorimager (Fujik BAS 1000; Fuji).

Nucleotide sequence accession number.

The complete genomic sequence of BVDV-1 strain CP7, including the previously published complete coding sequence (30) together with the sequences of the 5′ and 3′ NTRs determined in this study, has been deposited in the EMBL and GenBank data libraries and assigned accession no. AF 220247; this sequence encompasses 5 A residues following position 44.

RESULTS

Determination of the 5′ and 3′ sequences of BVDV CP7.

The first infectious BVDV cDNA clone was based on the well-characterized cytopathogenic BVDV-1 strain 7 (CP7); its genomic sequence was almost completely known (29). Only the 5′-terminal sequence of the CP7 genomic RNA comprising the 5′-proximal 21 to 23 nt and the 3′-terminal 33 nt (compared to published sequences of other BVDV strains) had not been determined, and thus the sequence of the heterologous BVDV-1 strain NADL (12) was introduced into the CP7 cDNA. In order to establish an authentic infectious cDNA clone of BVDV CP7, we determined the 5′- and 3′-terminal sequences of this virus. After ligation of the viral genomic RNA, a nested RT-PCR assay (for details, see Materials and Methods) resulted in specific amplification of a cDNA fragment of the expected size, which was subsequently cloned in a bacterial vector. Ten cDNA clones were subjected to sequence analysis. With respect to the unknown 5′-terminal 21 nt, the sequence obtained was identical for all clones and differed at nine positions from the sequences of BVDV-1 strains NADL and Osloss (Fig. 1A). Similar to other pestivirus strains, two conserved complementary 9-mer motifs with the sequences GUAUACGAG (positions 1 to 9) and CUCGUAUAC (positions 22 to 30) are present in the 5′-terminal region of BVDV CP7. This part of the pestivirus genome has been predicted to form a highly conserved, stable stem-loop structure at the 5′ end of the viral RNA, termed Ia (16). With respect to the 3′-terminal 33 nt of the CP7 genome, six nucleotide differences were found compared to the 3′ terminus of BVDV NADL, while only two nucleotides differed from the sequence of BVDV strain Osloss, which has been used for recently described subgenomic cDNA constructs of CP7 (42) and all cDNA clones described here.

FIG. 1.

5′ sequences of pestiviruses, in particular BVDV-1 strain CP7. (A) Alignment of the 5′-terminal sequences of representative strains from four pestivirus species. For BVDV-1 CP7, the consensus sequence was determined from 10 cDNA clones. Other sequences were extracted from the GenBank/EMBL database (BVDV-1 Osloss [15], BVDV-1 NADL [12], classical swine fever virus [CSFV] Alfort-T [28], CSFV Brescia [31], border disease virus [BDV] X818 [4], BVDV-2 890 [36]). Conserved nucleotides are indicated with an asterisk. The region of the CP7 5′ sequence shown in panel B is marked by a line above the alignment; the region with the variable numbers of A residues is marked by arrows. (B) Nucleotide sequences of three CP7 5′ clones carrying either 9, 20 or 26 A residues following position 44 of the CP7 sequence. The numbers on top correspond to those of the CP7 sequence shown in panel A. (C) Nucleotide sequences of 10 CP7 clones. The clones shown in panel B are included. Note the variable number of A residues following position 44. (D) Predicted RNA secondary structure of the 5′ NTR of BVDV-1 CP7-9A. Modeling was performed with the computer programs RNAFOLD, MFOLD, and FOLDANALYZE (17). In addition, the proposed structure is based on comparative sequence analysis. Nt 362 to 389 were excluded from secondary-structure interactions because this region contributes to a putative pseudoknot structure. The BVDV 5′ NTR stem-loops are designated according to the nomenclature proposed by Brown et al. (9). The initiation codon AUG is indicated.

Naturally occurring 5′ NTR mutations.

Surprisingly, the 10 clones analyzed for determination of the 5′- and 3′-terminal sequences of BVDV CP7 exhibited marked variation in the number of A residues following nt 44 (Fig. 1B and C). While the previously reported CP7 full-length cDNA clone pA/BVDV (29) comprised 8 A residues at this position, 9 to 26 A residues were present in the cDNA clones obtained in this study. To our knowledge, this is the first description of such a variation within the 5′ NTR of a pestivirus. According to computer-predicted RNA secondary-structure models of the 5′ NTR of CP7, the region encompassing the different numbers of A residues forms the loop of hairpin Ib (Fig. 1D). The secondary structure of the BVDV CP7 5′ NTR shown in Fig. 1D is in good agreement with the secondary structures of the pestivirus 5′ NTR suggested by other groups (23, 33, 46). According to previous studies, the predicted 5′-terminal hairpins Ia and Ib do not represent parts of the IRES element (11).

Genetic stability of 5′ NTR variants.

Three sequences with the authentic 5′ terminus of the BVDV CP7 genome followed at position 44 by 9, 20, or 26 A residues were introduced into the infectious BVDV CP7 cDNA clone; the constructs were termed pCP7-9A, pCP7-20A, and pCP7-26A, respectively. After in vitro transcription of full-length genomic RNA and transfection of MDBK cells, infectious virus was recovered for all three constructs. To investigate the genetic stability of the 5′ NTR mutants, the viruses obtained (CP7-9A, CP7-20A, and CP7-26A) were repeatedly passaged in MDBK cells, and the lengths of the 5′ NTR fragments were determined after 10 passages by RT-PCR analysis using primers Ol 200R and Ol CP7-SP6. Interestingly, larger fragments were detected by RT-PCR for all three variants, suggesting the emergence of viral genomes with larger 5′ NTRs. For further characterization, the fragments were cloned and subjected to sequence analysis. For each of the variants, the sequences of 12 clones were determined. Nucleotide sequence analysis showed multiple insertions of A residues downstream of position 44. For CP7-9A, up to 21 A residues were found after 10 passages in MDBK cells, while passages of CP7-20A and CP7-26A led to viral genomes with up to 36 and 54 A residues at this position, respectively (Fig. 2). For CP7-20A, both multiple insertions and deletions of several A residues occurred. This demonstrates that each of the variants is genetically unstable.

FIG. 2.

Multiple insertions and deletions of A residues downstream of position 44 during propagation of BVDV 5′ NTR mutants in bovine cells. After transfection with engineered full-length RNAs, the recovered BVDV mutants CP7-9A, CP7-20A, CP7-26A, CP7-54A, and CP7-5A were repeatedly passaged in MDBK cells. After 10 passages, the 5′ NTR sequences and the numbers of A residues following position 44 were determined for each mutant by sequence analysis of 12 cDNA clones. Each value represents one clone, except for CP7-5A, for which all 12 clones gave the same value. The boxed numbers of A residues were introduced into the infectious CP7 full-length cDNA clone.

Identification of a genetically stable 5′ NTR variant and construction of a deletion mutant.

After tissue culture passages of CP7-9A, CP7-20A, and CP7-26A, insertions and deletions of A residues were observed. For further characterization of this phenomenon, a full-length cDNA clone carrying a stretch of 54 A residues (pCP7-54A) downstream of position 44 was constructed. Transfection of the RNA transcribed from pCP7-54A resulted in recovery of infectious virus, which was passaged 10 times in MDBK cells and then subjected to nucleotide sequence analysis of the 5′ NTR. Interestingly, deletions of A residues were detected for 11 of 12 clones investigated; for some clones, only 5 or 7 A residues were found after position 44 (Fig. 2). As a next step, we constructed the full-length cDNA clone pCP7-5A, which carries only 5 A residues following position 44. Transfection of RNA transcribed from pCP7-5A resulted in recovery of infectious virus (CP7-5A), which was subsequently passaged in MDBK cells. Remarkably, sequence analysis of the 5′ NTR of CP7-5A after 10 passages revealed that all 12 clones investigated carried 5 A residues following position 44 (Fig. 2). CP7-5A therefore represents a genetically stable variant of BVDV CP7, while multiple insertions and deletions of A residues occurred rapidly in the 5′ NTR of the other variants. Apart from the variable number of A residues, no other differences were observed in any of the 5′ NTR clones analyzed.

To find out whether the oligo(A) motif downstream of position 44 represents an essential element within the pestivirus 5′ NTR, the oligo(A) motif of pCP7-5A together with the trinucleotide sequence TAA was removed and replaced with 1 T residue. According to the predicted RNA secondary structure, the resulting construct (pCP7-T) lacks the entire loop of hairpin Ib.

Comparative analysis of hairpin Ib mutants.

For comparative analysis of the 5′ NTR mutants described above, the specific infectivities of the transcribed RNAs were determined. The RNAs derived from pCP7-5A, pCP7-9A, pCP7-20A, and pCP7-26A each yielded between 4.8 × 10⁵ and 6.0 × 10⁵ PFU/μg, while 8.0 × 10⁴ and 8.0 × 10² PFU/μg were obtained for the RNAs transcribed from pCP7-54A and pCP7-T, respectively (Fig. 3). Transcript-derived viral plaques of all variants were slightly smaller than BVDV CP7 plaques; the plaques generated by CP7-54A appeared smaller than those of the other virus variants (data not shown). In addition, an immunofluorescence (IF) assay using an anti-NS3 monoclonal antibody was performed 24 h p.t. For the variants carrying 5, 9, 20, and 26 A residues following position 44, the portion of IF-positive cells was between 50 and 70%, while only ca. 10% and <0.1% IF-positive cells were detected after transfection of RNAs derived from pCP7-54A and pCP7-T, respectively (Fig. 3). Furthermore, determination of the virus yields at 24, 48, and 72 h p.t. showed that the virus titers obtained for CP7-5A, CP7-9A, CP7-20A, and CP7-26A were very similar, while the titers reached by CP7-54A at 24 and 48 h p.t. were about 10-fold lower; the final titer obtained for CP7-54A at 72 h p.t. was comparable to that of the other variants with 5 to 26 A residues within the Ib loop (Fig. 4). For CP7-T, the virus yields determined at 24 and 48 h p.t. were about 1,000-fold below the titers obtained for the other variants.

FIG. 3.

IF analysis of MDBK cells at 24 h p.t. with engineered full-length RNAs from CP7-5A, CP7-9A, CP7-20A, CP7-26A, CP7-54A, and CP7-T. Magnification, ×100. The specific infectivities of the BVDV CP7 mutant RNAs are indicated.

FIG. 4.

Virus titers obtained for BVDV 5′ NTR mutants CP7-5A, CP7-9A, CP7-20A, CP7-26A, CP7-54A, and CP7-T at 24, 48, and 72 h p.t. The titers of released virus were determined on MDBK cells.

After two passages of CP7-T in MDBK cells, high titers of infectious virus (>1 × 10⁵/ml) were obtained, suggesting the emergence of a mutation(s). RT-PCR analysis of the 5′ NTR of the CP7-T-derived virus stock from the fifth tissue culture passage using primers Ol 200R and Ol CP7-SP6 led to detection of fragments larger than expected for CP7-T. After molecular cloning, the nucleotide sequences were determined for 10 cDNA clones. Sequence analysis revealed the presence of 22 to 48 A residues downstream of position 44, each followed by the trinucleotide sequence TAA, while no other changes occurred within the analyzed part of the 5′ NTR during propagation of CP7-T (data not shown). This suggests that the oligo(A) motif forming the loop of hairpin Ib is an important structural element for efficient infectivity of the BVDV genomic RNA.

Mutations within hairpin Ia.

For further characterization of the 5′-terminal region preceding the IRES element, we concentrated on the 5′-proximal stem-loop Ia. This hairpin includes a stem which is formed by base-pairing of nt 1 to 10 of the CP7 genome to the complementary sequence of nt 21 to 30, while the remaining 10 nt (11 to 20) form an apical loop. The first set of mutations within the 5′-terminal 30 nt of CP7-5A were designed to alter both the nucleotide sequence and predicted RNA secondary structure of hairpin Ia at various positions. The RNAs of mutants SL-1, SL-2, and SL-3 lack nt 2, 6 and 7, and 14 to 17, respectively. In addition, several mutants were constructed which contain nucleotide substitutions at positions 2 to 4 (SL-4), 5 to 7 (SL-5), 10 to 13 (SL-6), and 27 to 29 (SL-7). To assess how each individual mutation is predicted to alter the RNA secondary structure of hairpin Ia, the minimum free-energy structures of each mutant sequence were calculated. The computer-based RNA secondary-structure predictions of hairpin Ia of the individual mutations introduced into the infectious cDNA clone pCP7-5A are illustrated in Fig. 5A. Accordingly, different parts of the stem or the apical region forming the loop were deleted or substituted.

FIG. 5.

Hairpin Ia mutants. (A) Computer-predicted RNA secondary structures of BVDV hairpin Ia mutations and IF analysis of MDBK cells at 24 h p.t. with full-length RNAs from CP7-5A, SL-1, SL-2, SL-3, SL-4, SL-5, SL-6, SL-7, SL-8, and SL-9. Magnification, ×100. Modeling was performed with the computer programs RNAFOLD and MFOLD (17). Deleted (Δ) and substituted nucleotides within hairpin Ia are indicated. (B) Specific infectivities of the BVDV mutant RNAs in MDBK cells and virus titers obtained at 24, 48, and 72 h p.t.

The clones carrying these mutations were used for in vitro transcription of full-length genomic RNAs. After transfection of MDBK cells, the specific infectivities and virus yields were determined for each of the mutant RNAs and the RNA transcribed from pCP7-5A. The specific infectivities of SL-2, SL-3, SL-5, SL-6, and SL-7 (2.8 × 10⁴ to 6.4 × 10⁴ PFU/μg) were moderately reduced compared to that of CP7-5A (2.4 × 10⁵ PFU/μg) (Fig. 5B). In contrast, the specific infectivities of the RNAs from SL-1 (4.4 × 10¹ PFU/μg) and SL-4 (<10 PFU/μg) were near or below the limit of detection. These differences correlated well with the portions of IF-positive cells determined at 24 h p.t. and the virus titers obtained at different time points after transfection (Fig. 5). To reduce the possibility that a technical error accounted for the observed phenotypical differences, the transcription-transfection experiments using the whole set of mutant RNAs were done in parallel and repeated several times. The results of separate experiments were very similar.

Interestingly, the two mutations resulting in a severe loss of infectivity (SL-1 and SL-4) concern the 5′-terminal 4 bases, whereas all mutants carrying alterations downstream of nt 4 (SL-2, SL-3, SL-5, SL-6, and SL-7) showed only moderate reductions in infectivity and virus yield after transfection. It should be noted that the predicted RNA secondary structure of SL-7 is reminiscent of that of SL-4 (Fig. 5A). In contrast to SL-7, the replication competence of SL-4 was severely impaired (Fig. 5; see above). The results of our analysis suggest that the 5′-terminal sequence 5′-GUAU of the BVDV genomic RNA is essential for efficient replication of BVDV.

Deletion of hairpin Ia.

The results obtained prompted us to construct two additional mutants which lack hairpin Ia. SL-8 carries a deletion of nt 1 to 24, while SL-9 lacks the 5′-terminal 29 nt; for efficient in vitro synthesis of RNA, a G residue was fused to the 5′ end of SL-9. For SL-8, the 5′-terminal 6 nt (5′-GUAUAC) are identical to the 5′ terminus of CP7-5A. In contrast, the 5′ end of SL-9 RNA contains a different sequence. Transcription-transfection experiments indicated that the specific infectivity of the SL-8 RNA (5.2 × 10³ PFU/μg) was about 50-fold lower than that of CP7-5A, while the infectivity of SL-9 was near the limit of detection (6.4 × 10² PFU/μg). Similar differences were also observed by a comparison of the portions of IF-positive cells at 24 h p.t. and the virus titers determined after transfection (Fig. 5). Accordingly, deletion of the complete hairpin Ia (SL-9) resulted in a dramatic loss of infectivity, while an RNA lacking Ia but encompassing the conserved sequence motif 5′-GUAU at the 5′ end (SL-8) was capable of producing considerable amounts of infectious virus (7.3 × 10⁴/ml at 72 h p.t.). These results provide further evidence that the 5′ sequence motif GUAU plays an essential role in efficient replication of BVDV. While the presence of hairpin Ia is apparently not obligatory for replication competence, all mutations and deletions of Ia resulted in a reduction in specific infectivity and virus titers, suggesting an important role for this highly conserved element in efficient viral replication.

Analysis of secondary mutations within the 5′ NTR of hairpin Ia mutants.

To investigate whether secondary mutations had occurred within the 5′ NTR, each mutant was repeatedly passaged in MDBK cells. Using RNAs from the third tissue culture passage, the 5′- and 3′-terminal sequences were amplified by the above-described RNA ligation–RT-PCR method and cloned in a bacterial vector; at least 10 cDNA clones were characterized for each mutant. Sequence analysis indicated the absence of any secondary mutations within the analyzed genomic regions of mutants SL-2, SL-3, SL-5, SL-6, SL-7, and SL-8.

After two tissue culture passages of supernatants from cells transfected with the RNAs of SL-1, SL-4, and SL-9, titers of >1 × 10⁵ infectious virus per ml were obtained, suggesting the emergence of secondary mutations. Determination of the 5′-terminal sequences of the viruses obtained actually demonstrated the presence of mutations in the 5′ UTR. Sequence analysis indicated that in the mutant obtained during propagation of SL-1, a U residue was inserted at position 2 of the RNA; the resulting 5′-terminal sequence is identical to that of BVDV CP7. For the mutant that emerged during propagation of SL-4, a deletion of nt 2 to 31 was observed, resulting in removal of hairpin Ia. This deletion led to the emergence of the 5′-terminal sequence 5′-GUAU, which is identical to the 5′ end of the BVDV CP7 RNA. Interestingly, the 5′ NTR sequence of the mutant evolved from SL-9 was identical to the sequence of the SL-4-derived mutant virus. Emergence of this sequence can be explained by deletion of nt 2 and 3 from the SL-9 RNA. For further analysis, the deletion identified for the latter two mutants was introduced into pCP7-5A. After transfection of the in vitro-synthesized RNA, infectious virus was recovered and termed Δ2-31. The specific infectivity of Δ2-31 RNA (6.0 × 10³ PFU/μg) and the virus titers obtained after transfection were reduced compared to those of the parent virus CP7-5A (Fig. 6). After three passages of Δ2-31 in MDBK cells, secondary mutations were not found within the 5′-terminal region of the genome.

FIG. 6.

Deletion of hairpins Ia and Ib. (A) Schematic representation of the 5′ NTR of BVDV CP7-5A and deletion mutants Δ2-31, Δ5-57, and Δ5-73. (B) IF analysis of MDBK cells at 24 h p.t. Magnification, ×100. (C) Specific infectivities in MDBK cells and virus titers obtained at 24, 48, and 72 h p.t. (D) 5′-terminal sequences of CP7-5A, Δ5-73, and mutant viruses M1 and M2 evolved after passaging of supernatants from cells transfected with Δ5-73 RNA. For M1 and M2, the inserted nucleotides are highlighted.

Deletion of hairpins Ia and Ib.

Our analysis of SL8 and Δ2-31 showed that genomic RNAs lacking hairpin Ia or a substantial part of it are still capable of producing infectious virus as long as the motif GUAU is present at the 5′ end of the genome. In order to investigate the effects of larger deletions, two constructs were generated lacking nt 5 to 57 (Δ5-57) and nt 5 to 73 (Δ5-73). The specific infectivity of Δ5-57 was about 20-fold reduced compared with that of the parent virus CP7-5A, while the specific infectivity of Δ5-73 was near the limit of detection (Fig. 6). Determination of the 5′- and 3′-terminal sequences of Δ5-57 using RNA from the third tissue culture passage demonstrated the absence of any secondary mutations within the regions analyzed. After two tissue culture passages of the transfection supernatant of Δ5-73, higher titers of infectious virus (>10⁵/ml) were obtained. Analysis of the 5′-terminal sequences derived from 12 cDNA clones indicated the emergence of mutants with duplications of 2 nt (M1) or 3 nt (M2) near the 5′ terminus of the genome (Fig. 6D). These results suggest that hairpins Ia and Ib are not required for replication of pestiviruses provided that the sequence motif 5′-GUAU remains at the 5′ terminus of the viral RNA.

Growth properties of the mutants.

Progeny virus recovered from the in vitro-transcribed RNAs of CP7-5A, SL-2, SL-3, SL-5, SL-6, SL-7, SL-8, Δ2-31, and Δ5-57 was characterized by plaque assay on MDBK cells using dilutions of transfected cells (Fig. 7); RNAs of SL-1, SL-4, SL-9, and Δ5-73, which led rapidly to the emergence of secondary mutations, were not included. The transcript-derived parent virus CP7-5A formed plaques with an average size of 2.7 mm. Each of the mutants produced smaller plaques. SL-2, SL-5, and SL-7 produced plaques with an average size ranging from 1.2 to 1.4 mm, while plaques generated by SL-3, SL-6, SL-8, Δ2-31, and Δ5-57 had an average size of ≤0.8 mm (Fig. 7).

FIG. 7.

Plaques produced by BVDV 5′ NTR mutants at day 6 p.t. (A) Plaque size of CP7-5A, SL-2, SL-3, SL-5, SL-6, SL-7, SL-8, Δ2-31, and Δ5-57 in MDBK cells. The average plaque size of 20 randomly selected plaques is indicated. (B) As examples, the plaques generated by CP7-5A, SL-7, and SL-8 are shown.

For further characterization, the growth rate and yield of the mutant viruses were determined. The amount of infectious virus in the supernatants from cells transfected with the individual mutant RNAs was determined, and the supernatants were then used to infect MDBK cells at an MOI of 0.05; mutant SL-8 was not included in this analysis because the virus yield obtained was too low. After incubation at 37°C, virus released into the medium was counted over a 4-day period (Fig. 8A). The peak titer for CP7-5A was 2.2 × 10⁵ TCID₅₀/ml, achieved on day 3 postinfection. Mutants SL-2, SL-3, SL-5, SL-6, and SL-7 had about 10-fold lower peak titers, while the peak titers reached by mutants Δ2-31 and Δ5-57 were about 100-fold lower than that of CP7-5A (Fig. 8A). Establishment of growth curves at 33 and 40.5°C showed that neither CP7-5A nor any of the mutants exhibited a temperature-sensitive phenotype (data not shown). Our results demonstrate that the differences of growth kinetics correlate with the observed plaque sizes. Moreover, the growth restriction of the mutant viruses mirrors the reduction of specific infectivity detected for the individual RNAs (Fig. 7; see also Fig. 5 and 6).

FIG. 8.

(A) Growth curves of BVDV CP7-5A and mutant BVDV strains SL-2, SL-3, SL-5, SL-6, SL-7, Δ2-31, and Δ5-57 determined on MDBK cells infected at an MOI of 0.05. The titers of released virus were determined over a 4-day period. (B) Northern blot analysis of total RNA from MDBK cells infected with BVDV CP7-5A and the indicated mutant BVDV strains at an MOI of 0.05. The infected cells were processed in parallel to those used to determine the growth rates. RNAs were extracted at 48 h after infection. The blot was hybridized with a BVDV CP7-specific cDNA fragment. RNA ladder sizes are indicated on the left (in kilobases). The intensity of bands was determined with a phosphorimager. The relative amounts of viral genomic RNAs are indicated below the blot (percent of parent CP7-5A [100%] value). n.i., RNA from noninfected MDBK cells.

Analysis of viral RNA synthesis.

For analysis of viral RNA synthesis, cells were infected at an MOI of 0.05 and processed in parallel to cells used for the growth kinetics. Total cellular RNA was prepared 2 days postinfection and used for Northern blot analysis. Viral RNA was visualized by autoradiography, and the intensity of bands was determined with a phosphorimager (Fig. 8B). CP7-5A RNA was more abundant than that of any of the mutant viruses. The RNA of SL-7 was next most abundant, followed by RNA from SL-2, SL-3, SL-5, and SL-6. As expected from its replication kinetics, the RNAs of Δ2-31 and Δ5-57 were least abundant. Northern blot analysis of RNAs prepared 3 days postinfection gave very similar results. It can be concluded that the amounts of viral RNAs detected correlated with the virus titers.

DISCUSSION

The NTRs of positive-strand RNA viruses are thought to contain signals important for translation, transcription, replication, and probably also packaging of viral genomes (7–10, 22, 25, 27, 48). Translation initiation of pestiviruses occurs by a cap-independent mechanism by use of an IRES element which is located within the 5′ NTR and the 5′-terminal region of the ORF (11, 23, 33, 34, 37). Two hairpins, termed Ia and Ib, precede the IRES element. In this study, several spontaneous and engineered mutations located within Ia and Ib of the BVDV 5′ NTR were analyzed by use of infectious BVDV cDNA clones. The results of our analysis show that (i) the observed spontaneous 5′ NTR mutations carrying 9 to 26 A residues downstream of position 44 did not lead to detectable differences in infectivity and virus growth properties; (ii) deletion of the entire loop of Ib including the oligo(A) motif resulted in dramatic loss of infectivity; (iii) the conserved 5′-terminal sequence 5′-GUAU represents an essential element for efficient replication of pestiviruses; (iv) mutations altering the primary and predicted RNA secondary structure of hairpin Ia as well as deletion of Ia together with a substantial part of Ib resulted in recovery of infectious virus provided the sequence motif 5′-GUAU was present at the 5′ terminus of the viral genomic RNA. This indicates that hairpins Ia and Ib are not essential for replication of pestiviruses.

During propagation of BVDV CP7 in cell culture, we observed spontaneous insertions of A residues downstream of position 44 of the viral genome. Control experiments using in vitro-transcribed and DNase I-treated BVDV CP7 genomic RNAs with 9, 20, or 26 A residues revealed minor variations in the number of A residues (deletion of 1 to 4 A's in about 55% of cDNA clones and addition of 1 to 3 A's in 15% of cDNA clones), indicating that in vitro transcription, RT-PCR, and/or nucleotide sequencing can contribute to some extent to the observed variation. For most of the mutants obtained after 10 passages in MDBK cells, however, additions of 4 to 28 A's were detected (Fig. 2). The majority of the additions and deletions in the poly(A) tract are best explained by stuttering of the viral RdRp during replication. In contrast to the mutants with 9 to 54 A residues following position 44, the engineered mutant CP7-5A, carrying 5 A residues, remained genetically stable during 10 passages in MDBK cells. This efficiently replicating mutant was used in our study as the starting material to introduce mutations within the BVDV 5′ NTR and should be useful for future studies of pestiviruses that include reverse genetic approaches.

The 5′-terminal sequence motif GUAU is conserved among all pestiviruses analyzed so far, including BVDV-1, BVDV-2, border disease virus, and classical swine fever virus (4), suggesting a strong functional pressure. Our finding that replication of pestiviruses apparently requires this sequence at the 5′ end of the BVDV genomic RNA is compatible with recently reported data obtained by the analysis of 5′ HCV-BVDV chimeras (19). This study revealed that the HCV IRES can functionally replace that of BVDV as long as the 5′-terminal 3 or 4 bases of the BVDV genome are present at the 5′ terminus of the chimeric viruses. Taking into account the analysis of the 5′ HCV-BVDV chimeras together with our results, it appears reasonable to assume that the 3′ end of the minus-strand RNA plays an important role in initiation of plus-strand synthesis. Recently communicated experiments using a recombinant BVDV RdRp actually demonstrated that short RNA templates with the sequence AUAC-3′ at the 3′ end were capable of directing de novo synthesis of RNA (21). In addition, it has recently been suggested that a recombinant HCV RdRp is capable of de novo RNA synthesis (32). In contrast, all other studies on RdRps from members of the Flaviviridae, including HCV, flaviviruses, and pestiviruses, failed to detect de novo RNA synthesis, while primer-dependent synthesis of RNA and elongation of RNA templates by a copy-back mechanism were observed (6, 18, 24, 38, 40, 51). Future studies may include experiments with full-length minus-strand RNA templates and will shed more light on the properties of these RdRps and the function of the conserved sequence motif AUAC-3′ as well as other factors important for viral plus-strand RNA synthesis.

Our experiments were designed to characterize the growth properties of BVDV mutants derived from engineered full-length genomic RNAs in bovine cells. All engineered 5′ NTR mutants either carrying alterations of hairpin Ia or lacking Ia and part of Ib exhibited a growth-restricted phenotype characterized by considerably smaller plaques and lower growth rates than the parent virus CP7-5A. The mutations described here may influence RNA stability, RNA replication, translation of the viral polyprotein, and/or encapsidation of the viral genome. Northern blot analysis of cells infected with different 5′ NTR mutants using the same MOI revealed that the amount of accumulated viral genomic RNA correlated with both the specific infectivity of the transcribed RNAs and the growth rate of the mutants. After transfection or infection of cells, a negative effect on viral RNA replication may result in decreased synthesis of viral proteins and vice versa. Analysis of viral protein synthesis of the BVDV mutants estimated by immunoblot with an anti-NS3 monoclonal antibody actually demonstrated a correlation between the amount of viral genomic RNA and the amount of viral protein (data not shown). While it cannot be excluded that the region comprising hairpins Ia and Ib has an influence on translation of the viral polyprotein, it appears more likely that it contains cis-acting elements important for efficient viral RNA synthesis. This assumption is supported by the fact that the translation efficiency of the BVDV 5′ NTR lacking either Ia or Ib was not decreased compared to that of the entire 5′ NTR (11).

For flaviviruses and HCV, viral and cellular factors have been found to bind to the terminal regions of viral genomes and were proposed to play important roles in viral replication (7, 8, 10, 14, 20, 45). The high conservation of the pestivirus 5′-terminal sequence and its implication for viral replication described in this study strongly suggest an interaction of this RNA sequence element with viral and/or cellular proteins. Future studies of the molecular determinants for pestivirus replication will aim at the identification of cellular and viral proteins that interact with the 5′- and 3′-terminal sequences of pestivirus genomes. After identification of such proteins, it will be interesting to examine whether some of the mutants described here exhibit differences in binding.

For control of diseases caused by BVDV, vaccination of cattle with live or inactivated vaccines is commonly used. Thus far, an attenuated safe and efficacious live BVDV vaccine is not available. For other RNA viruses, it has been reported that mutations in the NTRs can lead to restricted growth together with a significant decrease in virulence (22, 25, 27). Some of the mutants described here may be useful as safe and efficient vaccines against pestivirus-induced diseases.