INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bovine viral diarrhea virus (BVDV), classical swine fever virus (CSFV), and border disease virus (BDV) comprise the genus Pestivirus within the family Flaviviridae. This family also includes the genus Flavivirus and the hepatitis C-like viruses (HCV) (59). The pestivirus genome consists of a positive-stranded nonpolyadenylated RNA molecule that typically is 12.3 kb long (3, 5, 10, 11, 30, 37, 43, 44). The genomic RNA encodes a polyprotein of approximately 4,000 amino acids (aa), which encompasses all viral proteins arranged in the order NH₂-N^pro- C-E^rns-E1-E2-p7-NS2-NS3-NS4A-NS4B-NS5A-NS5B-COOH (7, 14, 50, 51, 56). C, E^rns, E1, and E2 represent structural components of the virion, whereas the remaining proteins are nonstructural (NS). The release of the pestivirus proteins occurs co- and posttranslationally by host cell- and virus-derived proteases (45, 51, 61, 62). The first cleavage event in pestivirus protein biogenesis is due to the autoproteolytic activity of N^pro, which cleaves in cis at the N^pro/C junction (51). Processing at the C/E^rns, E1/E2, and E2/p7 sites is probably mediated by host signal peptidases (14, 45). The proteases responsible for cleavage at the E^rns/E1 and p7/NS2 sites have not been defined; however, cleavage at the latter site may also be mediated by cellular signal peptidase (14, 55). The release of the nonstructural proteins located downstream of NS3 is mediated by a serine protease residing in NS3 (62). The N termini of NS4A, NS4B, NS5A, and NS5B have recently been determined (55, 63). According to these studies, the NS3 protease cleaves between leucine and serine or leucine and alanine. For processing at the 4B/5A and 5A/5B sites, the NS4A protein is required as a cofactor for the NS3 protease (63).

According to their effects during growth in tissue culture cells, BVDV can be divided into cytopathogenic (cp) and noncytopathogenic (noncp) viruses (57). Both biotypes play an important role in the pathogenesis of fatal mucosal disease (MD). As a prerequisite for MD, infection of pregnant cows with a noncp BVDV has to occur at an early stage of gestation. This results in the birth of persistently infected calves, which are immunotolerant to the respective noncp BVDV (1, 36, 57). Interestingly, a "pair" of antigenetically closely related cp and noncp BVDV can be isolated from each animal with MD (27, 28, 60). The molecular characterization of several BVDV pairs demonstrated that cp BVDV strains develop from noncp BVDV by RNA recombination. The genomes of cp BVDV strains contain different alterations like host cell-derived insertions, sometimes combined with large duplications, and genome rearrangements, including duplications or deletions of viral sequences (see reference 34 for a review). These genome alterations all affect the NS2-3 coding region of the viral genome. As a consequence, cp BVDV strains express NS3, which is regarded as molecular marker for cp viruses. NS3 is not found in cells infected with noncp BVDV; instead, noncp viruses express NS2-3, a protein also present in cells infected with cp BVDV (6, 8, 39, 40).

For several cp BVDV isolates, partial genome analyses indicated that the NS2-3 coding region of the viral RNA might not contain recombination-induced alterations (9, 18, 29, 38, 41). Some of these viruses were not obtained during recent outbreaks of MD but have been propagated for a long time in different laboratories. The history of such strains is often not known in detail, and in most cases a noncp counterpart is not available. The reason for the cytopathogenicity of these viruses remained obscure. Since the published sequences encompass only rather small parts of the genomes, the presence of recombination-induced genome alterations in other regions of the genome is not excluded. In this paper, we describe detailed analyses for one of these viruses, namely, BVDV Oregon. These investigations allowed us to determine the molecular basis for the expression of NS3.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells and viruses. MDBK cells were obtained from the American Type Culture Collection (Rockville, Md.). BHK-21 cells (BSR clone) were kindly provided by J. Cox (Federal Research Centre for Virus Diseases of Animals, Tübingen, Germany). Cells were grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum and nonessential amino acids. The cp BVDV strain Oregon was kindly provided by B. Liess (University of Hanover, Hanover, Germany). The T7 vaccinia virus (vTF7-3) (15) was generously provided by B. Moss (Laboratory of Viral Diseases, National Institute of Allergy and Infectious Diseases, Bethesda, Md.). BVDV infection of cells was assayed by immunofluorescence with monoclonal antibodies (58).

Infection of cells. Since pestiviruses are mainly cell associated, lysates of infected cells were used for reinfection of culture cells. Lysates were prepared by freezing and thawing cells 48 h postinfection and stored at 70°C.

cDNA cloning and nucleotide sequencing. Synthesis of cDNA, cloning, and library screening were done essentially as described previously (32). For the first library, cDNA synthesis was primed with oligonucleotides BVDV13 and BVDV14 (32). The probe for screening was the insert of BVDV cDNA clone NCII.1 (32). For a second library, cDNA synthesis was primed with BVDV35a (54) and BVDV11. The screening was done with a CP7-derived cDNA fragment corresponding to nucleotides 845 to 2738 (35). Exonuclease III and nuclease S1 were used to establish deletion libraries of cDNA clones (19). Sequencing of double-stranded DNA was carried out with the T7 polymerase sequencing kit (Pharmacia, Freiburg, Germany) (47). Sequence analysis and sequence alignments were done with Genetics Computer Group software (12). The sequence of oligonucleotide BVDV11 is 5' TCR AAC CAR TAY TGR TAY TC-3'.

RT-PCR and PCR. Total RNA from MDBK cells infected with BVDV Oregon was used as starting material for reverse transcription-PCR (RT-PCR). RT-PCR was performed as described previously (35). The primers used to amplify nucleotides 1007 to 1566 were A39 and B78; the primers used to amplify nucleotides 3337 to 4085 were C78 and D2 (Table 1). PCR was carried out with Taq polymerase (Appligene, Heidelberg, Germany). The reaction volume was 50 µl and contained 25 mM Tris-HCl (pH 8.3), 75 mM KCl, 2.5 mM MgCl₂, 250 µM deoxynucleotide triphosphates, and 30 pmol of each primer. Amplification was carried out for 30 cycles (30 s at 94°C, 30 s at 54°C, and 60 s at 72°C). The PCR products were separated by agarose gel electrophoresis; for purification of DNA fragments, a QIAEX II gel extraction kit (Qiagen, Hilden, Germany) was used.

Cloning of expression constructs. All eukaryotic expression constructs were based on pCITE-2 obtained from AGS GmbH, Heidelberg, Germany. Restriction, subcloning, and other standard procedures were done essentially as described previously (46). Restriction and modifying enzymes were obtained from New England BioLabs (Schwalbach, Germany), Pharmacia, and Boehringer Mannheim GmbH (Mannheim, Germany). To create blunt ends, the Klenow fragment of DNA polymerase I (Klenow) was used. Dephosphorylation was carried out with calf intestinal phosphatase.

Site-directed mutagenesis. All mutants were generated by the method of Kunkel et al. (24), using the Muta-Gene Phagemid in vitro mutagenesis kit (Bio-Rad, Munich, Germany) essentially as described by the manufacturer, except that single strands were produced with the filamentous phage VCSM13 (Stratagene). Oligonucleotide primers specifying single or double base changes were used in the mutagenesis reactions. All of the subcloned fragments used for mutagenesis were sequenced to verify the presence of the desired mutation(s) and the absence of second-site mutations.

Transient expression with the T7 vaccinia virus system. BSR cells (5 × 10⁵ cells in a 3.5-cm-diameter dish) were infected with the recombinant T7 vaccinia virus vTF7-3 (15) at a multiplicity of infection of 5 in medium without fetal calf serum. After incubation for 1 h at 37°C, the cells were washed once with serum-free medium and transfected with 10 µg of plasmid DNA (mammalian transfection kit; Stratagene). After 4 h at 37°C, the cells were washed twice with medium containing no methionine or cysteine (label medium) and incubated in this medium for 1 h at 37°C. The cells were labeled in 0.5 ml of label medium containing 0.25 mCi of [³⁵S]methionine-[³⁵S]cysteine ([³⁵S]Trans-Label; ICN, Eschwege, Germany) for 4 to 5 h, washed twice with phosphate-buffered saline, and stored at 70°C.

Radioimmunoprecipitation and SDS-PAGE. Extracts of MDBK cells (1.5 × 10⁶ cells in a 3.5-cm-diameter dish) infected with BVDV Oregon and labeled for 7 h with 0.25 mCi [³⁵S]methionine-[³⁵S]cysteine ([³⁵S]Trans-Label; ICN) or of BSR cell infected with vTF7-3 and transfected with the respective plasmid were prepared under denaturing conditions (2% sodium dodecyl sulfate [SDS]). After dilution to a final concentration of 0.2% SDS, aliquots of cell extracts were incubated with 5 µl of undiluted rabbit serum. For the formation of precipitates, cross-linked Staphylococcus aureus was used (22). SDS-polyacrylamide gel electrophoresis (PAGE) of proteins with molecular masses up to 60 kDa was carried out on Tricine gels by the method of Schägger and Jagow (48); for the separation of larger proteins, SDS gels as described by Doucet and Trifaro (13) were used. The gels were processed for fluorography by using En³Hance (New England Nuclear, Boston, Mass.). The following antisera were used for the detection of BVDV proteins: antiserum against NS2 (anti-Pep6, generated against a peptide corresponding to residues 1571 to 1586 of the BVDV CP7 polyprotein [numbers refer to BVDV SD-1]) (54), antiserum against NS3 (anti-A3, raised against a bacterial fusion protein encompassing sequences of CSFV Alfort Tübingen) (56), and antiserum against NS4 (anti-P1, generated against a bacterial fusion protein containing sequences of CSFV Alfort Tübingen) (33).

Evaluation of NS2-3 cleavage efficiencies. NS2-3 cleavage efficiencies were quantified after precipitation of NS2-3 and NS3 with anti-NS3 (56) and SDS-PAGE analysis (see above). The gels were exposed to a Fujifilm imaging plate (Raytest, Straubenhardt, Germany) and analyzed with a Fujifilm BAS-1500 phosphorimager (Raytest). Computer-aided determination of the intensities of the respective signals was carried out with TINA 2.0 software (Raytest). For evaluation of cleavage efficiencies, the number of methionines and cysteines within NS2-3 or NS3 was determined. After measurement of the radioactivity of the NS2-3 protein, the percentage of the counts resulting from the NS3 moiety was calculated. This value together with the counts determined for the cleaved NS3 protein was defined as total NS3 (100%). The calculated percentage of cleaved NS3 with respect to total NS3 is given as the percent cleavage efficiency.

N-terminal sequence analysis of radiolabeled NS3. The NS3 protein used for radiosequencing was generated by transient expression of pO1 in the T7 vaccinia virus system (see above). Labeling of about 5 × 10⁵ cells was performed with 1 mCi of [³⁵S]cysteine (ICN) in 0.5 ml of label medium lacking cysteine. Material from about 10⁶ cells was used for isolation of radiolabeled NS3. The NS3 protein was immunoprecipitated (see above), separated by SDS-PAGE, and transferred to an Immobilon polyvinylidene difluoride membrane (Millipore, Eschborn, Germany). The protein was localized by autoradiography and subjected to automated Edman degradation.

Nucleotide sequence accession number. The sequence data for BVDV Oregon were deposited at the GenBank/EMBL data library (accession no. AF041040).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Genome analysis. The genomes of several cp BVDV isolates contain large duplications, deletions, and/or cellular sequences coding for ubiquitin or a fragment of another cellular protein of unknown function, termed cIns (34). Northern blot analyses demonstrated that the genome of the cp BVDV strain Oregon does not contain a large duplication, and there was no indication of the presence of a cp defective interfering particle. Moreover, it could be shown by hybridization with specific probes that the genome contains neither ubiquitin-encoding nor cIns sequences (data not shown). Partial cloning of the BVDV Oregon genome revealed the absence of recombination-induced alterations in the genomic region coding for NS2-3 (38).

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the cDNA clones derived from the BVDV Oregon genome. The upper part shows a scheme of the BVDV genome together with the encoded polyprotein. The structural proteins are shown as shaded boxes, and the nonstructural proteins are shown as open boxes. Below, the cDNA clones are shown. The position and size of the clones obtained after cDNA library screening (A) and RT-PCR (B) are given. The indicated numbers represent the first and last nucleotides of each cDNA clone and refer to the sequence of BVDV SD-1 (11).

Protein expression studies. The cytopathogenicity of BVDV is correlated with the expression of the nonstructural protein NS3 (6, 8, 39, 40). To demonstrate that BVDV Oregon expresses NS3, immunoprecipitation was carried out with extracts of MDBK cells, which had been infected with this BVDV strain and labeled with [³⁵S]methionine-[³⁵S]cysteine. For precipitation, an antiserum directed against NS3 was used (anti-A3, 56). SDS-PAGE analysis revealed that in addition to NS2-3, NS3 was present (see Fig. 3B, lane 2); the latter protein comigrated with NS3 of other cp BVDV strains (data not shown).

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Schematic drawing of the constructs used for transient expression in eucaryotic cells. The diagram at the top represents the BVDV polyprotein, with the structural proteins shown as shaded boxes and the nonstructural proteins shown as open boxes. Below, the expression constructs are presented as lines and drawn to scale to indicate the region of the Oregon polyprotein expressed. The names of the constructs are given on the left. The numbers on the right indicate the amino acids of the BVDV Oregon polyprotein expressed by each construct. Numbers refer to the sequence of BVDV Oregon. In pO1/S-A, inactivation of the NS3 proteinase by a serine-to-alanine change at position 1752 is indicated.

View larger version (32K): [in this window] [in a new window]   FIG. 3.   SDS-PAGE analysis of immunoprecipitates obtained after transient expression of pO1 in BSR cells. For production of authentic BVDV proteins, MDBK cells were infected with BVDV Oregon. Metabolic labeling was performed with [35S]methionine-[35S]cysteine. Nontransfected BSR cells infected with T7 vaccinia virus (vTF7-3) were used as a control. BVDV-specific proteins are indicated on the right. Numbers on the left refer to the molecular masses (in kilodaltons [K]) of marker proteins. +, presence; , absence. (A) The anti-NS2 serum, which is directed against the carboxy-terminal part of NS2, was used to precipitate NS2 (54). A rabbit preimmune serum (NS) served as a control. (B) Precipitation was carried out with an anti-NS3 serum (56), which recognizes NS2-3 and NS3, or with a rabbit preimmune serum (NS).

N-terminal sequencing of NS3. To determine the cleavage site between NS2 and NS3, N-terminal sequencing of NS3 was performed. Since processing of BVDV polyproteins seems to take place authentically after expression in the T7 vaccinia virus system and since the quantity of NS3 obtained after transient expression was larger than after BVDV infection, NS3 transiently expressed from pO1 was used as a source for protein sequencing. NS3 protein labeled with [³⁵S]cysteine was precipitated with anti-NS3 serum, transferred to an Immobilon membrane, and subjected to 16 cycles of Edman degradation. The radioactivity released in each degradation step was measured and resulted in detection of peaks at steps 5 and 14 (Fig. 4). Since the NS2-specific serum anti-Pep6 is directed against aa 1571 to 1586 (54), the NS2-3 cleavage site must be located downstream of this sequence. Only two cysteine residues, located at positions 1594 and 1603 of the polyprotein, fit with the spacing of the two peaks. Thus, glycine 1590 represents the N-terminal residue of NS3.

View larger version (14K): [in this window] [in a new window]   FIG. 4.   N-terminal sequencing of NS3 from BVDV Oregon. NS3 was generated by transient expression of pO1 in cells metabolically labeled with [35S]cysteine. The graph shows the distribution of radioactivity in counts per minute released during automated Edman degradation after subtraction of background radiation. At the top of the diagram, the amino acid sequence of BVDV Oregon beginning with glycine 1590 is aligned with the degradation steps. The labeled residues are shown in boldface type.

Inactivation of the NS3 protease. NS2-3 represents a chymotrypsin-like serine protease that generates its own C terminus and releases the viral nonstructural proteins downstream of NS2-3 from the precursor molecule (2, 16, 62). The catalytic center of this protease is located in the N-terminal one-third of NS3. Mutation of serine residue 1752 of the proposed catalytic triad to alanine leads to inactivation of the NS3 protease (62). It has been proposed that the NS3 protease is also involved in cleaving NS2-3 into NS2 and NS3 (62). To study the role of this enzyme for the generation of NS3 in the case of BVDV Oregon, the codon for serine 1752 was mutated to a codon for alanine, resulting in plasmid pO1/S-A (Fig. 2). In contrast to the expression of pO1, for which NS4A and the truncated NS4B (NS4B*) could be detected (Fig. 5A, lane 4), expression of pO1/S-A yielded neither NS4A nor NS4B* (lane 5). This indicates that the NS3 protease has been inactivated by the mutation. For pO1/S-A, two fusion proteins of about 125 and 155 kDa could be detected, which both precipitated with the anti-NS3 serum as well as with the anti-NS4 serum (Fig. 5B, lanes 3 and 4). The protein of about 155 kDa most probably represents NS2-3 fused with NS4A and NS4B*. The second protein, of about 125 kDa, could consist of NS3 and NS4A/NS4B*. Evidence for this hypothesis was obtained after precipitation with the serum directed against NS2. Despite the inactivation of the NS3 protease in the protein encoded by pO1/S-A, NS2 was generated upon transient expression (Fig. 5A, lane 3). These results demonstrate that the NS3 protease is not involved in the cleavage of NS2-3.

View larger version (24K): [in this window] [in a new window]   FIG. 5.   SDS-PAGE analysis after transient expression of pO1 and pO1/S-A in BSR cells metabolically labeled with [35S]methionine-[35S]cysteine. BVDV-specific proteins are indicated on the right. The truncated NS4B protein is marked with an asterisk. For further details, see the legend to Fig. 3. (A) Precipitation was carried out with the anti-NS2 or anti-NS4 serum. (B) Sera directed against NS3 or NS4 were used for precipitation. NS2-3/4A/4B* and NS3/4A/4B* represent fusion proteins composed of the respective proteins.

Expression of truncated NS2-3 proteins. The fact that the NS3 protease is not involved in NS2-3 cleavage did not rule out the possibility that part of NS3 is required for NS2-3 cleavage, as was observed for HCV (17, 20). To narrow down the region necessary for generation of NS2 a series of plasmids encoding C-terminally truncated NS2-3 proteins was established; the constructs were named pO2, pO3 and pO4 (Fig. 2). Transient expression and precipitation with the anti-NS2 serum showed that NS2 is produced after transfection of all three truncated constructs (Fig. 6A, lanes 2 to 4). Since the truncated NS3 proteins expressed from pO2, pO3, and pO4 did not contain the sequence recognized by our serum directed against NS3, the encoded proteins were fused with six histidine residues. After precipitation with a specific monoclonal antibody recognizing the His tag, no truncated NS2-3/NS3 proteins were detected, indicating that these products were unstable (data not shown). Accordingly, truncated NS2-3 proteins were also not detected with the anti-NS2 serum (Fig. 6A, lanes 2 to 4). Nevertheless, precipitation of NS2 was observed with this serum, which indicates that processing takes place at the original NS2/NS3 cleavage site even when only the N-terminal 66 aa of NS3 is present.

View larger version (56K): [in this window] [in a new window]   View larger version (46K): [in this window] [in a new window]   View larger version (36K): [in this window] [in a new window]   FIG. 6.   SDS-PAGE of immunoprecipitates after transient expression of the indicated plasmids in BSR cells metabolically labeled with [35S]methionine-[35S]cysteine. For details, see the legend to Fig. 3. (A) Precipitation was carried out with the anti-NS2 serum or a rabbit preimmune serum (NS). (B) The anti-NS2 serum or a preimmune serum (NS) were used for precipitation. NS3# represents an aberrant cleavage product migrating slightly slower than NS3. NS3# is detectable after precipitation with the anti-NS2 serum as well as after precipitation with the anti-NS3 serum (see Fig. 6C). The arrows in lanes 4, 5, and 6 mark the N-terminally truncated NS2-3 proteins. (C) Precipitation with a serum directed against NS3 or a rabbit preimmune serum (NS). See also the legend to panel B.

Expression of chimeric NS2-3 proteins. The results described above indicate that the information leading to the processing of NS2-3 is located within NS2 of BVDV Oregon. To verify this hypothesis, chimeric constructs were established with cDNA fragments derived from BVDV Oregon and BVDV CP7. Fragments obtained from the latter virus were changed by deletion of a 27-nucleotide insertion normally present in the NS2 gene; the resulting sequence was termed CP7Ins. It has been shown previously that for BVDV CP7 the 27-nucleotide insertion is responsible for NS2-3 cleavage (54) and cytopathogenicity (35). Thus, NS2-3 translated from the CP7Ins sequence is no longer processed into NS2 and NS3 (54). The chimeric constructs are all based on pO1 or a construct containing the equivalent cDNA fragment of CP7Ins (pN1; Fig. 7). In the first step, a cDNA fragment was exchanged between pO1 and pN1, which contains the NS2-encoding sequence. This exchange was performed after introduction of appropriate restriction sites by silent mutagenesis and led to pN1-1, which contains the NS2 gene from BVDV Oregon in the context of the CP7Ins sequence, and pO1-1, with the reciprocal arrangement (Fig. 7).

View larger version (32K): [in this window] [in a new window]   FIG. 7.   Schematic drawing of the constructs obtained after exchanges of cDNA fragments or single codons in the NS2 coding region. All constructs encode proteins that start within the hypothesized C-terminal part of p7 and end within NS4B. Open boxes indicate sequences derived from BVDV Oregon, whereas shaded boxes represent the CP7Ins sequence. The CP7Ins sequence is derived from BVDV CP7 by deletion of the 27-nucleotide insertion within the NS2 gene (54). Chimeric constructs based on plasmid pO1 are shown on the left; chimeric constructs based on pN1 are shown on the right. In each case, the name of the expression construct is given on the left, whereas numbers on the right indicate the positions of the amino acids encoded by the exchanged cDNA fragment or of the exchanged codons, respectively. The abbreviation INS represents the 27-nucleotide insertion specific for BVDV CP7. NS4B*, truncated NS4B.

View larger version (40K): [in this window] [in a new window]   FIG. 8.   Results of transient expression of constructs with heterologous cDNA fragments or point mutations. (A) The upper part shows the SDS-PAGE analysis after transient expression of the chimeric plasmids based on pO1. For details, see the legend to Fig. 3. On top, precipitation was carried out with the anti-NS3 serum. Below, the serum directed against NS2 was used for precipitation. Below the gels, the quantification of the NS2-3 cleavage efficiencies is represented schematically. The quantification was carried out with a phosphorimager after precipitation of NS2-3 and NS3 and SDS-PAGE analysis. For evaluation of cleavage efficiencies, the number of methionine and cysteine residues within each NS2-3 or NS3 protein was determined. After measurement of the radioactivity of the NS2-3 protein, the percentage of the counts resulting from the NS3 moiety was calculated. This value, together with the counts determined for the cleaved NS3 protein, was defined as total NS3 (100%). Cleavage efficiency is equivalent to the percentage of cleaved NS3 with respect to total NS3. Cleavage efficiencies are given as the average of three independent experiments. Values in parentheses are not due to NS3, since in these cases no NS2 could be detected. The exposure time was different for the gels showing NS2-3/NS3 on the one hand and NS2 on the other hand. (B) Results obtained after transient expression of the chimeric plasmids based on pN1. For further details, see the legend to panel A.

Contribution of single amino acid exchanges to NS2-3 cleavage. The expression of different chimeras showed that the C-terminal region of NS2 of BVDV Oregon is important for NS2-3 cleavage. The presence of codons 1467 to 1581 of the BVDV Oregon sequence in pN1-3 is sufficient to induce significant processing (Fig. 8B, lane 5). Therefore, an alignment of aa 1463 to 1596 was performed for several pestiviruses to look for striking differences in this part of the NS2 sequence of BVDV Oregon (Fig. 9). Four amino acids differing between the BVDV Oregon sequence and the other pestivirus proteins were selected to analyze the influence of individual residues on cleavage efficiency. These investigations were carried out with constructs based on pO1. Four constructs were made, each containing one mutation leading to replacement of the residue present in the BVDV Oregon sequence by the corresponding amino acid from BVDV CP7; I 1508 was replaced by K (pO1/I-K), T 1514 was replaced by M (pO1/T-M), L 1515 was replaced by T (pO1/L-T), and S 1555 was replaced by F (pO1/S-F) (Fig. 7). Analyses of the cleavage efficiencies of these four constructs revealed that the S-to-F exchange at position 1555 had the largest influence, reducing the cleavage efficiency from 88 to 44% (Fig. 8A). The other three exchanges had only minor effects (Fig. 8A). The importance of the serine residue at position 1555 could also be seen after expression of pN1-3/S-F, which resulted from pN1-3 by mutation of the S codon to an F codon (Fig. 7); the protein harboring this mutation was cleaved much less efficiently (9%) than the protein encoded by pN1-3 (42%) (Fig. 8B).

View larger version (23K): [in this window] [in a new window]   FIG. 9.   Comparison of sequences encompassing the C-terminal one-third of NS2 from different pestiviruses. Sequences of several cp BVDV strains (Oregon, CP7, NADL, and Osloss), a noncp BVDV strain (SD-1), and a CSFV strain (Alfort-Tübingen) were aligned. Only differences from the BVDV Oregon sequence are specified. Numbers indicate the corresponding position in the BVDV Oregon sequence. The positions of the cIns insertion of BVDV NADL and the ubiquitin (UBI) insertion of BVDV Osloss are shown by vertical lines between adjacent amino acids. Arrows mark the positions of amino acids that have been exchanged between BVDV Oregon and BVDV CP7. The pestivirus sequences are from the following sources: CP7 (35), SD-1 (11), NADL (5), Osloss (10), Alfort-Tübingen (30).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Molecular analyses of different BVDV strains revealed that cp viruses develop from noncp BVDV strains by RNA recombination. Cellular insertions (sometimes accompanied by duplications of viral sequences), duplications and rearrangements of viral sequences, and deletions, have been found in the genomes of cp BVDV strains which are not present in the genomes of their noncp counterparts (34). It has been shown in several cases that these genome alterations are responsible for the expression of NS3 (33, 34, 52, 54), the marker protein of cp BVDV (6, 8, 39, 40). As shown in this report, the genome of the cp BVDV strain Oregon does not possess changes due to RNA recombination. It was therefore of great interest to determine the molecular basis for NS2-3 cleavage. Since NS2-3 processing was shown to take place after transient expression of parts of the BVDV Oregon polyprotein, the region of the genome necessary for this cleavage could be narrowed by expressing a series of consecutively shortened cDNA fragments. While C-terminal truncations of NS2-3 did not influence cleavage at the NS2/3 site, N-terminal truncations blocked cleavage at the original site, and led to an aberrant processing upstream of the NS2/3 cleavage site. This aberrant processing could be a consequence of misfolding of the shortened proteins, and it is not clear whether it is executed by the same protease that is responsible for the authentic processing. These experiments showed that in BVDV Oregon the information necessary for authentic processing of NS2-3 resides within the region of the polyprotein encompassing NS2 and the first 66 aa of NS3. Final proof that the main determinants for NS2-3 cleavage are localized within NS2 could be achieved by construction of chimeras with sequences of BVDV Oregon and the recombinant noncp BVDV CP7Ins. By using this approach, the genomic region necessary for NS2-3 cleavage could be narrowed to the 3' one-third of the NS2 gene. However, while the transfer of the complete NS2 gene of BVDV Oregon led to a cleavage efficiency comparable to the one obtained for the BVDV Oregon construct, exchanges of only parts of the NS2 gene reduced cleavage efficiency significantly. These data indicate that the whole NS2 protein is important for efficient NS2-3 cleavage. Because of the absence of recombination-induced genome alterations, point mutations within the NS2 gene have to be responsible for NS2-3 processing. The complete NS2 protein of BVDV Oregon displays nearly 100 aa exchanges with respect to the CP7Ins sequence. The C-terminal one-third of this protein, which in chimeric constructs is sufficient for the induction of significant processing, harbors 19 of these exchanges. Introduction of single amino acid exchanges within the C-terminal region of NS2 revealed that the serine at position 1555 plays an important role. After this amino acid was changed to phenylalanine, cleavage efficiency was reduced to 50% of the wild-type level. Introduction of the reciprocal exchange into the BVDV CP7Ins sequence was able to induce processing of an NS2-3 that was not cleaved before. Interestingly, cp BVDV Singer, for which, like cp BVDV Oregon, no genome alteration due to RNA recombination could be identified within the NS2 gene, does not possess a serine at position 1555 of its polyprotein but instead contains leucine (38), a large hydrophobic amino acid that is much more similar to phenylalanine than to serine. However, an amino acid alignment revealed some striking differences with respect to other BVDV sequences including BVDV Oregon (38). Thus, for BVDV Singer, cleavage of NS2-3 could also be due to strain-specific point mutations. Perhaps the amino acid exchanges identified in the NS2 proteins of BVDV Oregon and Singer induce a specific conformation of NS2 which is important for efficient cleavage. Such a conformation could result from different combinations of amino acid exchanges in the NS2 region. It seems rather unlikely that in these viruses a proteolytic activity or a protease cleavage site has been generated de novo by accumulation of point mutations. More probably, a cryptic enzymatic activity or cleavage site is present in NS2-3 of pestiviruses, which is inactive or not accessible in noncp BVDV. In the case of BVDV Oregon or other cp isolates like Singer, a set of point mutations could change the conformation of NS2-3 in such a way that the cryptic cleavage mechanism becomes active. Remarkably, CSFV and some isolates of border disease virus (BDV) also express NS3 without exhibiting striking genome alterations (3, 56). It is likely that in all these cases a similar mechanism accounts for cleavage of NS2-3.

Induction of a conformation allowing NS2-3 cleavage could also represent the mechanism by which the insertions present in the NS2 proteins of BVDV CP7 and NADL induce processing of NS2-3. In the case of BVDV CP7, a duplicated sequence of 27 nucleotides was identified, which is inserted in the NS2 gene in another reading frame and thus codes for an additional 9 aa not present elsewhere in the polyprotein (54). Analyses based on transient expression of polyprotein fragments and an infectious BVDV clone revealed that this insertion is responsible for NS2-3 cleavage as well as the cytopathogenicity of the virus (35, 54). For BVDV NADL, a host sequence termed cIns was found (31); the function of the cellular counterpart is not known. Even though definite proof is still missing, the cIns insertion most probably causes the cleavage of NS2-3 and cytopathogenicity of BVDV NADL. Both the cIns insertion and the additional 9 aa in the CP7 polyprotein are located upstream of the NS2-3 cleavage site and therefore cannot serve as a direct target for a protease, unlike ubiquitin (52). The polypeptides encoded by the insertions do not contain known protease motifs. Thus, induction of a cleavable conformation represents an attractive explanation for the function of these insertions with regard to NS2-3 processing. It therefore could turn out in the future that NS2-3 processing in pestiviruses is achieved by two principal mechanisms. One possibility would be introduction of a new protease and/or protease cleavage site at the N terminus of NS3 by RNA recombination. The second possibility would rely on the generation of a cleavable conformation as a consequence of either a recombination upstream of the NS2-3 cleavage site or introduction of a specific set of point mutations.

For several cp BVDV strains, the mechanism leading to the expression of NS3 has been elucidated in detail. For cp viruses with genomes containing cellular insertions coding for ubiquitin, release of NS3 occurs via cellular ubiquitin C-terminal hydrolases (52). In the polyproteins encoded by other cp BVDV genomes, N^pro is located just upstream of NS3, which is due either to duplication and rearrangement or to deletion of sequences; in these cases, processing at the N terminus of NS3 is mediated by the autocatalytic activity of N^pro (33, 53). In contrast, the protease responsible for NS2-3 cleavage has not yet been identified for the BVDV strains NADL, CP7, and Oregon; also, in the CSFV and BDV isolates showing NS2-3 processing, the responsible protease is unknown. For members of the genus Flavivirus, NS2-3 cleavage is carried out by the NS3 protease, with NS2B representing an essential cofactor (see reference 42 for a review). Such a mechanism can be excluded for BVDV Oregon since inactivation of the NS3 protease does not inhibit processing at the 2/3 site. Analogous results were also obtained for BVDV CP7 and NADL (54, 63).

The polyprotein of HCV contains a so-called NS2-3 protease, which is different from the NS3 serine protease and mediates processing at the 2/3 site (42). As shown by truncation experiments, this enzymatic activity encompasses nearly the whole NS2 region and the N-terminal one-third of the NS3 protein (17, 20). In the case of BVDV Oregon, cleavage at the 2/3 site still occurred when only the N-terminal 66 aa of NS3 was left. Moreover, the HCV NS2-3 protease is active after expression in Escherichia coli and after translation in an animal cell-free system in the absence of microsomal membranes (17, 20). Preliminary data indicate that the NS2-3 protein of BVDV Oregon is not cleaved after expression in bacteria and processing of NS2-3 is observed only in the presence of microsomal membranes during in vitro translation. These data do not necessarily mean that cleavage occurs by a cellular protease associated with membranes. Perhaps membranes are necessary for correct folding of NS2-3. The N-terminal part of NS2 is highly hydrophobic and thus presumably is associated with membranes. Moreover, preliminary results suggest cleavage at the p7/NS2 site by signal peptidase, indicating that at least the N terminus of NS2 is located within the lumen of the endoplasmatic reticulum (14, 55). Correct folding of NS2 either may play a crucial role in activating a cryptic viral protease or may be a prerequisite for making the protein accessible for cleavage by a cellular protease, which could be associated with the membranes or could be located in the cytoplasm. Taken together, our results reveal significant differences between BVDV Oregon and HCV with regard to processing of NS2-3. Further studies are necessary to identify the protease responsible for cleavage of NS2-3 from BVDV Oregon and other pestiviruses exhibiting so far unknown mechanisms of NS2-3 processing.

BVDV Oregon represents the first pestivirus for which the sequence at the N terminus of NS3 is published. This information was eagerly awaited since indirect evidence indicated a remarkable conservation of this position. This indirect evidence is based on the genome structures of a variety of cp BVDV strains and the mechanisms leading to NS3 expression. Until now, nine cp BVDV strains have been isolated that contain ubiquitin-encoding insertions (34). In all these cases, the 3' end of the cellular insertions is located exactly upstream of the position corresponding to codon 1590 of a BVDV genome without insertion. The known mechanism of NS2-3 processing in these cases strongly suggests that the N terminus of NS3 corresponds to glycine 1590 of the polyprotein. Similarly, the NS3 proteins expressed by viruses with N^pro coding sequences located directly upstream of NS3 should start with glycine 1590 (34). Since all these viruses were generated by recombination, a specific feature of the sequences around positions 5152 and 5153 of the BVDV genome could be responsible for this conservation. However, the finding that NS3 expressed by BVDV Oregon via a mechanism that is not based on recombination starts with glycine 1590 clearly shows that this terminus is functionally important. Remarkably, preliminary analysis of NADL NS3 indicated that glycine 1680 was the N-terminal residue (63); the respective position in the NADL polyprotein is equivalent to glycine 1590 in the SD-1 sequence. Thus, with one known exception (25), the N terminus of NS3 is highly conserved between different cp BVDV strains and may play an important role in the cytopathogenicity and/or viability of these viruses.

To date, about 40 cp pestiviruses have been analyzed at the genome level. Among these, the vast majority has been generated by RNA recombination. For only 14 strains have recombination-induced genome alterations not been demonstrated (9, 18, 29, 34, 38, 41). However, in most of these cases, the reported analyses were based on error-prone PCR assays (9, 18, 38). The choice of the primers and, to a lesser extent, the limitations of gel electrophoretic detection of small insertions could have led to wrong conclusions during these investigations. Thus, the results reported here prove for the first time that in addition to RNA recombination, a second mechanism can lead to a cp pestivirus. The first argument for this conclusion is the failure to detect recombination-induced alterations after sequencing nearly the entire genome of cp BVDV Oregon. Elucidation of the general mechanism resulting in the expression of NS3, the marker protein of cp BVDV, serves as a second strong argument for our theory. The newly demonstrated mechanism is based on a specific combination of point mutations. The analysis of the cp BVDV Singer sequence indicates the existence of different sets of such mutations, which can be regarded as a certain degree of flexibility. In the light of these data, it is astonishing that the generation of cp viruses by point mutations apparently occurs less often than by RNA recombination. According to the generally accepted model, the latter mechanism is based on template switching of the viral RNA polymerase during genome replication (4, 21, 23, 26). Generation of an autonomously replicating cp virus requires at least two switches, which have to be precise with regard to target sequence position, orientation of RNA strand, and reading frame. This process obviously is more complicated than the introduction of point mutations which is known to occur at high frequency in RNA viruses (49). It also has to be kept in mind that in most cases the intermediate resulting from the first template switch during RNA recombination will not be viable or at least will not be able to replicate autonomously. Thus, the two template switches probably have to occur in immediate succession or at least in a short time. In contrast, it can be assumed that mutations could be introduced at more or less arbitrary time intervals, with every new mutant able to replicate and generate viable progeny virus. The fact that generation of cp BVDV by accumulation of point mutations does not occur more often indicates that a barrier might exist in the generation of cp BVDV strains like Oregon, which must be equivalent to or even higher than that due to the complicated RNA recombination mechanism. Further investigations, including analysis of the correlation between specific point mutations and cytopathogenicity or viability of different mutant viruses, are necessary to answer this interesting question. To perform such studies, an infectious cDNA clone of BVDV Oregon is needed.

Our data on the molecular biology of BVDV Oregon add an interesting new chapter to the fascinating story of cytopathogenicity in pestiviruses. Since the generation of a cp virus mutant during persistent infection of calves with noncp BVDV can lead to lethal MD, introduction of point mutations could have major effects on the pathogenesis of this disease. This could be of special interest with regard to persistent infections of humans with HCV, which exhibits significant similarity to pestiviruses.