Insertion of a Sequence Encoding Light Chain 3 of Microtubule-Associated Proteins 1A and 1B in a Pestivirus Genome: Connection with Virus Cytopathogenicity and Induction of Lethal Disease in Cattle

Abstract

Pestiviruses represent the first RNA viruses for which recombination with cellular protein-coding sequences has been reported. As a result of such recombinations cytopathogenic (cp) pestiviruses can develop from noncytopathogenic (noncp) viruses. In the case of bovine viral diarrhea virus (BVDV), the generation of cp mutants is linked to the induction of the lethal syndrome mucosal disease (MD) in cattle. The cp BVDV JaCP was isolated from an animal which had come down with MD. The genome of JaCP contains a novel kind of cellular insertion (LC3*) which is flanked by duplicated pestivirus sequences. Neither insertion nor duplication is present in the genome of the accompanying noncp virus JaNCP. As part of the viral polyprotein, the insertion in the JaCP genome is translated into a polypeptide almost identical to a fragment of light chain 3, a subunit of the microtubule-associated proteins 1A and 1B from the rat. Transient-expression studies revealed that the LC3* sequence is able to induce an additional cleavage of the viral polyprotein. The respective cleavage occurs directly downstream of the LC3*-encoded sequence and is not dependent on the NS3 serine protease. Insertion of LC3* into an infectious noncp pestivirus cDNA clone without duplicated viral sequences resulted in recovery of a defective cp virus able to replicate only in the presence of a noncp helper virus. In contrast, introduction of both insertion and duplication led to an autonomously replicating cp virus.

Economically important diseases of farm animals such as classical swine fever and bovine viral diarrhea are caused by pestiviruses, which together with flaviviruses and hepatitis C virus constitute the family Flaviviridae (for reviews see references 35 and 45). Pestiviruses have a positive-sense single-stranded RNA genome of about 12.5 kb, which consists of 5′ and 3′ noncoding regions flanking a long open reading frame. The genomic RNA is translated into a polyprotein of approximately 3,900 amino acids (aa). Processing by host and virus-encoded proteases leads to the mature virus proteins arranged in the polyprotein in the order NH₂–N^pro–C–E^rns–E1–E2–p7–NS2-3–NS4A–NS4B–NS5A–NS5B–COOH. C, E^rns, E1, and E2 are present in pestivirus virions, whereas the other polypeptides represent nonstructural proteins (4, 6, 11, 36, 40, 41, 46, 50).

The most severe clinical condition resulting from infection with the pestivirus bovine viral diarrhea virus (BVDV) is called mucosal disease (MD) (45). The etiology of MD has been elucidated only recently. Two biotypes of BVDV, cytopathogenic (cp) and noncytopathogenic (noncp) viruses, are required for induction of this lethal disease. In a first step, intrauterine infection with a noncp BVDV that leads to induction of specific immunotolerance and birth of persistently infected calves has to occur (45). Such animals frequently go down with MD early in life; the disease is either induced by superinfection with an antigenically closely related cp BVDV or by generation of a cp mutant of the persisting noncp virus (24, 25). This switch from a noncp to a cp phenotype is in most cases the result of RNA recombination (reviewed in reference 31). Recombination can occur between the noncp BVDV genome and RNAs of viral or cellular origin (26, 29–31, 34, 42–44). So far, two types of cellular sequences, which code for ubiquitin or a protein of unknown function, have been identified in pestivirus RNAs (2, 27, 29–31, 34, 43). We report here the identification of a third type of cellular insertion and present an analysis of its effects on polyprotein processing and virus phenotype.

MATERIALS AND METHODS

Cells and viruses.

MDBK cells and BVDV isolate NADL were obtained from the American Type Culture Collection (Rockville, Md.). BVDV JaCP and JaNCP have been isolated from the serum of a bullock named Jasper, which developed MD while housed in isolation (14). The cp virus, JaCP, was cloned twice by plaque purification, and the noncp virus, JaNCP, was obtained by two rounds of limiting dilution. MDBK cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS) and nonessential amino acids.

Infection of cells.

Since pestiviruses tend to be associated with the host cells, suspensions composed of cell culture supernatant and infected cell lysate were used for infection of culture cells. Material for infection was prepared by freezing and thawing cultures 48 h postinfection and was stored at −70°C. If not specified, a multiplicity of infection of ∼0.1 was used. Infection with noncp BVDV was detected by immunofluorescence with a bovine hyperimmune serum.

Northern (RNA) hybridization.

RNA preparation by the guanidine isothiocyanate method, glyoxylation of RNA, and electrophoresis through 1% agarose gels containing formaldehyde were carried out as described before (30). Following electrophoresis, the RNA was transferred to nylon membranes (Duralon; Stratagene, Heidelberg, Germany) according to standard procedures (37). Hybridization with radioactive probes labeled by nick translation (Nick Translation Kit; Amersham and Buchler, Braunschweig, Germany) at 54°C and posthybridization washes at the same temperature were carried out as described before (30). The cDNA insert of BVDV clone NCII.1 (29) was used as a probe.

cDNA synthesis, cloning, and nucleotide sequencing.

Establishment of cDNA libraries in lambda ZAPII (Stratagene) was done as described before (29). Briefly, total RNA of cells infected with JaCP or JaNCP, respectively, was used for cDNA synthesis primed with oligonucleotides Ol-BVDV13, Ol-BVDV14, and Ol-Pes9 (29). Second-strand synthesis and ligation of EcoRI adaptors was done with the You-Prime cDNA synthesis kit as suggested by the supplier (Pharmacia, Freiburg, Germany). Size selection of double-stranded cDNA for molecules larger than 2 kb was done by preparative agarose gel electrophoresis as previously described (29). Ligation of cDNA fragments with lambda ZAPII DNA, packaging with Gigapack III-Gold, and plating on Escherichia coli XL-Blue1 cells were done as recommended by the manufacturer (Stratagene). Screening was done with the same probe used for Northern hybridization. Subcloning of cDNA fragments into pBluescript plasmids by in vivo excision was performed as recommended by the supplier (Stratagene). Exonuclease III and S1 were used to establish deletion libraries of cDNA clones (16). Dideoxy sequencing (38) of double-stranded DNA templates was carried out by using the T7 sequencing kit (Pharmacia). Computer analysis of sequence data was performed with the Genetics Computer Group software (9).

From a variety of cDNA clones derived from RNA of cells infected with BVDV JaCP, pJaCP/A17 and pJaCP/A19 were chosen for sequencing. The insert of pJaCP/A17 has a length of about 1.6 kb; the 5′ and 3′ ends of the cDNA fragment correspond to positions 7096 and 5808, respectively, in the published sequence of noncp BVDV SD-1 (8). Clone pJaCP/A19 contains an insert of about 2.8 kb which starts at position 5815 and ends at position 5808 compared to BVDV SD-1. The sequence data have been obtained by sequencing both strands of the two clones and deposited at the EMBL/GenBank data libraries under accession no. U80885.

Clone pJaNCP35 was isolated from the library established with RNA of cells infected with BVDV JaNCP. The insert in this plasmid has a length of 4.6 kb and corresponds to positions 2051 to 6814 of the BVDV genome. From this cDNA fragment a region of only 0.3 kb, extending from position 5142 to 5439, was sequenced.

Construction of clones for transient expression and recovery of infectious virus.

Restriction, subcloning and other standard procedures were done essentially as described previously (37). Restriction and modifying enzymes were purchased from New England Biolabs (Schwalbach, Germany), Pharmacia, and Boehringer-Mannheim (Mannheim, Germany). Plasmid pACYC177 was obtained from NEB. To obtain pEx7, a 3.1-kb NcoI/SalI fragment of pA/BVDV (28) was cloned into pCITE-2A (Angewandte Gentechnologie Systeme GmbH, Heidelberg, Germany); the cDNA fragment does not contain the 27-nucleotide insertion responsible for NS2-3 cleavage and cytopathogenicity of BVDV CP7 (42). Plasmids pEx7/JaCP and pEx7/JaNCP were established by exchanging a 0.3-kb BamHI/HpaI fragment of pEx7 for equivalent fragments of 0.65 and 0.3 kb from cDNA clones pJaCP/A17 and pJaNCP35, respectively. The former fragment contains the LC3* insertion, whereas the latter is colinear with the BVDV CP7 sequence. The position of the BamHI site corresponds to residues 5142 to 5147 of the BVDV SD-1 genome, and the HpaI site is located at positions 5434 to 5439. Similarly, pEx7/JaCP and pEx7/JaNCP, each with a destroyed NS3 protease, were generated by insertion of the appropriate BamHI/HpaI fragment into a pEx7 homolog in which the triplet encoding the active-site serine of the NS3 protease had been exchanged for an alanine codon.

The full-length constructs pA/B-JaCP and pA/B-JaNCP were obtained by exchanging the NcoI/SalI fragment in pA/BVDV/Ins− for the corresponding fragments of pEx7/JaCP and pEx7/JaNCP, respectively. By starting with pA/B-JaCP, pA/BVDV/Ins− was reconstructed by deletion of a 0.45-kb NotI/SacI fragment together with a 1-kb SacI fragment and insertion of a 1-kb NotI/SacI fragment derived from pA/BVDV/Ins−.

To obtain full-length clones containing the duplication detected in the genome of BVDV JaCP, pEx7/JaCP was restricted with SacI and PstI. The latter enzyme cuts in the multiple-cloning site of the vector. Integration of a SacI/BamHI fragment from cDNA clone pJaCP/A19 together with the BamHI/PstI fragment from pEx7/JaCP or pEx7/JaNCP resulted in plasmids pEx7/JaCP/dp or pEx7/JaNCP/dp, respectively. To establish the full-length construct pA/B-JaCP/dp harboring the JaCP-specific duplication of viral sequences, the NcoI/SalI fragment of pEx7/JaCP/dp was used to replace the corresponding fragment in pA/BVDV/Ins−. Similarly, pA/B-JaNCP/dp was constructed by starting with pEx7/JaNCP/dp and pA/BVDV/Ins−.

Plasmid pEx7Δ was generated by cutting pEx7 with HpaI and XhoI, followed by end filling with Klenow polymerase and religation. The BVDV-derived cDNA fragment of the resulting construct corresponds to positions 4648 to 5436 of the BVDV SD-1 genome. The equivalent clones containing sequences derived from BVDV JaCP or JaNCP (pEx7/JaCPΔ or pEx7/JaNCPΔ, respectively) were constructed the same way, starting with plasmids pEx7/JaCP or pEx7/JaNCP, respectively.

Site-directed mutagenesis.

Mutagenesis according to the method of Kunkel et al. (21) was done with the Muta-Gene Phagemid in vitro-mutagenesis kit (Bio-Rad, Munich, Germany) essentially as recommended by the manufacturer, with the exception that single strands were produced with the filamentous phage VCSM13 (Stratagene). The presence of the desired mutations was verified by nucleotide sequencing. For mutagenesis, single-stranded DNA of pEx7 was produced and annealed with oligonucleotide Ol-C7/S-A and the mutagenized strand was completed by incubation with T4 DNA polymerase and T4 ligase in synthesis buffer using the materials supplied with the mutagenesis kit.

The sequence of the oligonucleotide used for mutagenesis of the active-site serine codon of the NS3 protease gene to an alanine codon, Ol-C7/S-A, was 5′-TGAAGGATGGGCGGGTCTACCCAT-3′.

Transient expression, metabolic labeling, immunoprecipitation, and SDS-PAGE.

Transient expression of transfected plasmids using vaccinia virus vTF7-3 (kindly provided by B. Moss [12]) was done as described before (44), except that labeling time was reduced to 5 h. BVDV-infected MDBK cells (1.5 × 10⁶ per 3.5-cm-diameter dish) were labeled for 8 h with 0.5 mCi/ml of [³⁵S]methionine-[³⁵S]cysteine (Promix; Amersham). Nonradioactive cysteine and methionine were absent from the labeling medium. Cell extracts were prepared under denaturing conditions (15). Extracts were incubated with 5 μl of undiluted serum. Precipitates were formed with cross-linked Staphylococcus aureus (18), analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and processed by fluorography using En³Hance (New England Nuclear, Boston, Mass.). For analyses of proteins with molecular masses of up to 60 kDa, Tricine gels prepared according to the method of Schägger and Jagow (39) were used, while larger proteins were analyzed by electrophoresis performed according to the method of Doucet and Trifaro (10).

The following antisera were used for detection of BVDV proteins. (i) Antiserum A3 was generated against a bacterial fusion protein encompassing sequences of CSFV Alfort/Tübingen and recognizes NS2-3 and NS3 (46). (ii) Antiserum pep6 was raised against a peptide corresponding to residues 1571 to 1586 of the BVDV CP7 polyprotein and reacts with NS2 and NS2-3 (42). (iii) Antiserum P1 was raised against a bacterial fusion protein containing sequences of CSFV Alfort/Tübingen and recognizes both NS4A and NS4B (30).

In vitro transcription and RNA transfection.

cDNA constructs (2 μg) were linearized with SmaI (full-length BVDV clones, partial cut for pA/B-JaCP/dp and pA/B-JaNCP/dp) or PvuII (pEx7Δ and equivalent constructs) and purified by phenol extraction and ethanol precipitation. Transcription with T7 RNA polymerase (NEB) was carried out in a total volume of 50 μl of transcription mix (40 mM Tris-HCl, pH 7.5; 6 mM MgCl₂; 2 mM spermidine; 10 mM NaCl; 0.5 mM [each] ATP, GTP, CTP, and UTP; 10 mM dithiothreitol; 100 μg of bovine serum albumin per ml) with 50 U of T7 RNA polymerase in the presence of 15 U of RNAguard (Pharmacia). After incubation at 37°C for 1 h, the reaction mixture was passed through a Sephadex G-50 spun column (37) and further purified by phenol extraction and ethanol precipitation.

Transfection was done with a suspension of 3 × 10⁶ MDBK cells and about 100 ng of in vitro-transcribed RNA bound to DEAE-dextran (Pharmacia). The RNA–DEAE-dextran complex was established by mixing RNA dissolved in 100 μl of HBSS (47) with 100 μl of DEAE-dextran (1 mg/ml in Hanks balanced salt solution [HBSS]) and incubating the mixture for 30 min on ice (32). Pelleted cells were washed once with DMEM without FCS, centrifuged, and then resuspended in the RNA–DEAE-dextran mixture. After 30 min of incubation at 37°C, 20 μl of dimethyl sulfoxide was added and the mixture was incubated for 2 min at room temperature. After the addition of 2 ml of HBSS, cells were pelleted and washed once with HBSS and once with medium without FCS. Cells were resuspended in DMEM with FCS and seeded in a 10.0-cm-diameter dish. At 48 to 72 h posttransfection, cells were apportioned and seeded as appropriate for subsequent analyses.

In vitro translation.

Translation of in vitro-transcribed RNA was done with nuclease-treated rabbit reticulocyte lysate (RRL) (Promega, Heidelberg, Germany). The reaction was carried out in the presence or absence of microsomal membranes (1.8 μl; Promega) with 0.5 μg of RNA in a total volume of 25 μl in the presence of [³⁵S]methionine (Amersham), according to the manufacturer’s suggestions. For analysis of the translation products, 10% of the reaction mix was separated in a 10% acrylamide gel (39).

N-terminal sequence analysis of radiolabeled NS3.

The NS3 protein used for radiosequencing was transiently expressed from pEx7/JaCP by using vaccinia virus vTF7-3 (see above). Labeling was carried out with 500 μCi of [³⁵S]cysteine and about 10⁶ cells in 0.5 μl of labeling medium. The protein was immunoprecipitated with antiserum A3, 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.

RESULTS

Virus isolation and genome analysis.

To investigate putative virus recombinants involved in the development of MD, a persistently infected bullock named Jasper was housed under isolated conditions from the age of 1 year until the clinical syndrome developed over 2 years later (14). The animal was euthanatized when in extremis, and cp as well as noncp BVDV was isolated from serum samples, biologically cloned, and further analyzed. In a Northern blot hybridization, the RNA of the noncp virus, JaNCP, comigrated with the genome of the BVDV reference strain, NADL, with a size of 12.5 kb (5). In contrast, the genomic RNA of JaCP was found to have a length of more than 14 kb (Fig. 1). This finding is reminiscent of other cp BVDV isolates, such as CP1 or III-C, which contain ubiquitin-coding cellular insertions that are flanked by large duplications of viral RNA (29, 31, 34). However, the genome of JaCP did not hybridize to a ubiquitin probe or to the second known cellular insert (cIns) (data not shown).

FIG. 1.

Northern blot with RNA from bovine kidney (MDBK) cells infected with the indicated viruses. An RNA ladder served as a size marker (in kilobases). Control, RNA from noninfected MDBK cells.

After cDNA cloning and sequencing, a large duplication of viral sequences and a nonviral insertion were identified in the genome of JaCP. The 5′ part of the determined sequence is colinear with a noncp BVDV genome from residue 5815 to 7517. The latter position (C₅ in Fig. 2A) is located in the NS4B gene. Residue C₅ is followed by a fragment derived from the NS2 gene, starting with position 5057 (F") and ending with position 5152 (position A). Downstream of position A is located a nonviral insertion which is followed by the viral sequence starting with position B (Fig. 2A). Position B corresponds to nucleotide 5153 of a BVDV genome without rearrangement. Taken together, it can be concluded that the JaCP genome contains a duplication of 2,460 nucleotides.

FIG. 2.

(A) Genome organization of BVDV JaCP (bottom) compared to that of a noncp BVDV (top) and BVDV isolates Osloss, CP1, and III-C, which contain ubiquitin-coding insertions. For CP1 and III-C, duplicated viral sequences were found in addition to the cellular insertion (29, 34). The viral genomes are indicated as bars. The regions coding for NS2-3 or NS2-3-derived polypeptides are shaded. Important genomic positions flanking insertions or representing the start or end of duplicated sequences are labeled with letters as in reference 31. The genome of BVDV CP1 contains a duplication of the sequence located between B and C. In the RNAs of BVDV III-C or JaCP, the regions flanked by F′ and C₄ or F" and C₅, respectively, are duplicated. In both cases the duplicated region located closer to the genomic 3′ end contains the cellular insertion. Note that A and B mark the equivalent genomic position in all five genomes. Hatched bar, ubiquitin-coding insertion (Ub); cross-hatched bar, insertion homologous to the rat LC3 sequence (LC3*). (B) Comparison of the sequence published for rat LC3 of MAPs 1A and 1B (23) with the amino acid sequence encoded by the nonviral insertion identified in the RNA of JaCP. For JaCP, the sequence encoded by the insertion is shown in capital letters while the flanking regions are given in lowercase letters. The translational stop of rat LC3 is symbolized by an asterisk.

The genome organization of BVDV JaCP as deduced from these data is very similar to that proposed for BVDV III-C. Both virus genomes contain a cellular insertion integrated into a 5′-terminally truncated version of the NS2-3 gene that has been duplicated in the course of the recombination. In contrast to BVDV JaCP and C-III, the genome of cp BVDV CP1 contains no duplicated NS2 sequences and the RNA of cp BVDV Osloss contains a ubiquitin-coding insertion but no duplication at all (Fig. 2A). Position B is not only conserved as an integration site in the genomes of BVDV JaCP and III-C but represents the crossing-over site for all viruses containing ubiquitin-coding insertions and also for most pestiviruses with duplicated and rearranged viral sequences (31). Position B is regarded as the 5′ end of the NS3 gene.

The nonviral insertion in the JaCP RNA has a length of 348 nucleotides and shows no homology to ubiquitin genes or cIns, the other cellular sequence known to be present in the genomes of cp pestiviruses. A search in data libraries revealed 84% identity between the insertion and a rat cDNA sequence coding for light chain 3 (LC3), a subunit of the neuronal microtubule-associated proteins (MAPs) 1A and 1B (23); the deduced amino acid sequences exhibit 98% identity. The JaCP insertion (LC3*) encompasses nucleotides 39 to 386 of the published LC3 sequence. LC3* has been integrated into the viral genome in the same reading frame as in the LC3 mRNA. Thus, a viral polyprotein is expressed that contains aa 5 to 120 of LC3 (Fig. 2B).

Protein analysis.

The genomes of most cp BVDV isolates exhibit individual rearrangements. In all cases, these rearrangements affect the genomic region coding for the nonstructural viral protein NS2-3, a chymotrypsin-like serine protease that also has helicase function (1, 13, 48, 49). The protease identified in NS2-3 is responsible for cleavage at its own carboxy terminus and the nonstructural protein sites 4A/4B, 4B/5A, and 5A/5B (41, 50). In consequence of the rearrangements, processing of cp BVDV polyproteins yields NS3, which is not present in cells infected with non-cp viruses. Either NS3 is generated by cleavage of an NS2-3 protein containing, e.g., a ubiquitin insertion, or it is expressed from duplicated sequences which again contain an insertion (29–31, 43, 44) (Fig. 2A, BVDV CP1). Also, in the case of BVDV JaCP, expression of NS3 was observed in addition to that of NS2-3, whereas in JaNCP-infected cells, only NS2-3 could be detected (data not shown). To analyze whether LC3* is responsible for the generation of NS3, this sequence was integrated at the appropriate site into pEx7, a plasmid that was originally designed for expression of proteins from BVDV CP7. pEx7 codes for aa 1422 to 2447 of the BVDV polyprotein (numbers are according to the sequence of BVDV SD1 [8]). The amino-terminally truncated NS2-3 expressed from this construct is not processed (Fig. 3). To integrate the LC3* insertion, a BamHI/HpaI fragment from pEx7 was exchanged for the corresponding fragment from cDNA clone pJaCP/A17. The resulting construct, pEx7/JaCP encodes a polypeptide that contains the cellular sequence and in addition exhibits four amino acid exchanges in the NS2-3 region. As a control, a third construct was assembled from pEx7 and cDNA clone pJaNCP35; the resulting plasmid, pEx7/JaNCP, lacks the cellular insert but contains the same mutations as pEx7/JaCP. Thus, the only difference between the polyproteins encoded by the last two of these constructs is the absence or presence of the LC3* insertion, respectively.

FIG. 3.

Immunoprecipitation of proteins transiently expressed from cDNA constructs pEx7, pEx7/JaNCP, and pEx7/JaCP. The constructs encompass the genomic region coding for the carboxy-terminal half of NS2 together with NS3, NS4A, and the amino-terminal half of NS4B (codons 1422 to 2447 of the open reading frame). The upper panels show the proteins precipitated with an antiserum directed against nonstructural protein NS3 (antiserum A3 [46]) analyzed on a 12% gel according to the method of Doucet and Trifaro (10) (A) and the polypeptides recognized by an antiserum directed against NS2 (antiserum pep6 [42]) separated on a 10% gel according to the method of Schägger and Jagow (39) (B). Numbers on the left side of each gel indicate the molecular masses (in kilodaltons) of marker proteins, whereas the letters and arrows on the right side mark the important bands. The panels on the bottom show schematic presentations of the different expression products. NS2*, aminoterminally truncated NS2; LC3*, insertion derived from the cellular LC3 of MAPs 1A and 1B control, precipitation of proteins extracted from cells infected with vaccinia virus vTF7-3 (12).

The proteins derived from the three constructs were analyzed by transient expression using vaccinia virus vTF7-3 (12) and precipitation with antisera directed against NS2 or NS3, respectively (42, 46). For pEx7 and pEx7/JaNCP, further processing of the amino-terminally truncated NS2-3 protein was not observed; only one specific band of ∼92 kDa precipitated with the antisera against NS2 and NS3 (bands B in Fig. 3). In contrast, the same protein containing the LC3* polypeptide was efficiently processed to yield NS3 (band C in Fig. 3A); only a rather small proportion of the expressed protein remained uncleaved (band A in Fig. 3A). Using the anti-pep6 serum that is directed against the assumed carboxy-terminal part of NS2 (42) the amino-terminal cleavage product of 32 kDa was detected in addition to the uncleaved product (bands D and A, respectively, in Fig. 3B). The reaction of this cleavage product with the anti-pep6 serum, its size, and the fact that the obtained NS3 protein comigrated with NS3 from cp BVDV isolates (not shown) strongly indicate that the 32-kDa protein contains at least the majority of the LC3*-encoded polypeptide.

For cp BVDV isolates containing ubiquitin-coding insertions, host ubiquitin-specific proteases were found to cleave at the carboxy terminus of the cellular sequence and thus liberate the amino terminus of NS3 (43). The genomes of the second type of cp strains contain duplicated N^pro genes; in these cases, the autoproteolytic activity of N^pro generates the amino terminus of NS3. However, for the other cp BVD viruses, the protease responsible for the cleavage at the 2-3 site is still not known. In order to test whether the NS3 serine protease is involved in NS2-3 processing, mutants of pEx7/JaCP and pEx7/JaNCP were generated that encode, at a position corresponding to aa 1752 of the polyprotein, an alanine instead of a serine; the latter amino acid represents the active-site residue of the NS3 protease (1, 13, 49). After expression of these constructs, release of NS4A (band F in Fig. 4) or the carboxy-terminally truncated NS4B encoded by the plasmids (band G in Fig. 4) was no longer observed. Instead, polypeptides of high molecular weight were precipitated with the respective antiserum for these two constructs. Thus, the NS3 protease activity was indeed destroyed by the mutation. Accordingly, only the full-length expression product of ∼115 kDa was detected after expression of pEx7/JaNCP harboring the protease mutation (band B in Fig. 4). However, the LC3*-dependent processing of NS2-3 still occurred, since for the mutated pEx7/JaCP both the amino-terminally truncated NS2/LC3* and a product of ∼97 kDa composed of NS3, NS4A, and the truncated NS4B could be detected in addition to the full-length expression product (bands C, D, and A, respectively, in Fig. 4). These experiments show that the NS3 serine protease itself is not involved in the observed NS2-3 cleavage.

FIG. 4.

Immunoprecipitation of proteins transiently expressed from cDNA constructs pEx7, pEx7/JaNCP, and pEx7/JaCP. In the latest two of these constructs, the codon coding for the active-site serine of the NS3 protease (position 1752 of the BVDV polyprotein) was exchanged for an alanine codon. The panel on the upper-left side shows the results of immunoprecipitations with the antiserum A3, directed against NS3 (46), analyzed on a 12% gel (10); the other two panels present 10% gels (39) on which the polypeptides recognized by the antiserum against NS2 (42) (middle) or the antiserum P1 (30) (right), directed against NS4A/4B, were separated. Numbers on the left side of each gel indicate the molecular masses (in kilodaltons) of marker proteins, whereas the letters and arrows on the right side mark the important bands. A schematic presentation of the different expression products is shown in a box below the gels. NS2*, amino-terminally truncated NS2; NS4B*, carboxy-terminally shortened NS4B; LC3*, insertion derived from the cellular LC3 of MAPs 1A and 1B. Control, precipitation of proteins extracted from cells infected with vaccinia virus vTF7-3 (12).

In vitro translation studies.

To obtain further information on the processing of NS2-3 in the presence of the LC3*-encoded sequence, in vitro translation studies were conducted. The majority of the NS3 coding region was removed from pEx7, pEx7/JaCP, and pEx7/JaNCP by deletion of the 3′-terminal 2.3 kb of the cDNA inserts, resulting in constructs pEx7Δ, pEx7/JaCPΔ, and pEx7/JaNCPΔ. Because of the carboxy-terminal truncation of the encoded proteins, the serine residue of the active center of the intrinsic serine protease was deleted from NS3. RNA was transcribed from these plasmids after linearization with PvuII and was translated in RRL either in the absence or presence of canine microsomal membranes. The translation products were analyzed by SDS-PAGE and subsequent fluorography. Proteins obtained by transient vaccinia virus expression served as controls. For pEx7Δ and pEx7/JaNCPΔ, only one major translation product (29 kDa), which comigrated with the polypeptide obtained after transient expression, was detected (bands B in Fig. 5). However, translation of the RNA transcribed from pEx7/JaCPΔ yielded a cleavage product of 31 kDa (band C in Fig. 5) in addition to the full-length product (band A in Fig. 5). The second cleavage product, representing the truncated NS3 protein of about 14 kDa, could not be detected, probably because truncated NS3 proteins tend to be unstable (unpublished observation). The detection of NS2-3 cleavage was not dependent on the presence of microsomal membranes as observed in the case of BVDV Oregon, a cp virus without recombination-induced genome alteration (20). In conclusion, the insertion of LC3* leads to expression of NS3 through mediating polyprotein cleavage by a not yet identified protease that either is contained in the RRL or represents an intrinsic activity of the expressed viral protein. Moreover, the experiment showed that major parts of NS3 can be deleted without deleterious effects on the cleavage.

FIG. 5.

Proteins obtained by in vitro translation of RNA transcribed from cDNA constructs pEx7Δ, pEx7/JaNCPΔ, and pEx7/JaCPΔ, separated by 10% SDS-PAGE (39). The translation was performed either in the absence (−) or presence (+) of canine microsomal membranes (MM). Proteins obtained by immunoprecipitation with the antiserum pep6 serum after T7-driven (vTF7-3) transient expression of the same constructs in eucaryotic cells served as controls. Numbers on the left side of the gel indicate the molecular masses (in kilodaltons) of marker proteins, whereas the letters and arrows on the right side mark the important bands. A schematic presentation of the different expression products is shown in the box below the gel. NS2*, amino-terminally truncated NS2; NS3*, carboxy-terminally shortened NS3; LC3*, insertion derived from the cellular LC3 of MAPs 1A and 1B.

N-terminal sequencing of NS3.

The results of the protein analyses indicated that the processing of the truncated NS2-3 expressed from pEx7/JaCP occurred very close to the carboxy-terminal end of the LC3* encoded insertion. This conclusion is based on the size of the precipitated proteins and the reactivity of the antiserum that is directed against aa 1571 to 1586 of the BVDV polyprotein (antiserum pep-6 [42]). To determine the cleavage site precisely, NS3 transiently expressed from pEx7/JaCP and labeled with [³⁵S]cysteine was precipitated with the antiserum A3 serum, run through a polyacrylamide gel, transferred to an Immobilon membrane, and subjected to 20 cycles of automated Edman degradation. The radioactivity released in each degradation step was counted and plotted, resulting in the curve shown in Fig. 6. Two peaks, at positions 5 and 14, were detected. Within the protein encoded by pEx7/JaCP, such a spacing of two cysteine residues is found only once, namely, at positions corresponding to residues 1594 and 1603 of the BVDV polyprotein (numbers refer to BVDV SD1 [8]). The LC3*-derived sequence is located in the pEx7/JaCP-encoded polypeptide as well as in the JaCP polyprotein between an arginine and a glycine corresponding to residues 1589 and 1590 of the BVDV polyprotein without insertion. Thus, cleavage of the polypeptide expressed from pEx7/JaCP occurs between the last residue of the LC3*-encoded sequence and the first amino acid of the viral sequence downstream of the cellular insertion. The amino terminus of the NS3 protein generated by this processing corresponds to glycine 1590 of the BVDV polyprotein.

FIG. 6.

N-terminal sequence of NS3 expressed from pEx7/JaCP. The proteins transiently expressed from the cDNA construct were labeled with [³⁵S]cysteine and purified by immunoprecipitation and SDS-PAGE. After transfer of the proteins to an Immobilon membrane, NS3 was isolated and automated Edman degradation was performed. The graph shows the counts per minute released per sequencing cycle. Above the graph, the amino acids determined by sequencing (in boldface type and underlined) and the flanking residues deduced from the pEx7/JaCP sequence are shown.

RNA transfection experiments.

Expression of NS3 by cp BVDV is correlated with induction of a cytopathic effect (CPE). In several cases, a direct connection between genome rearrangement, expression of NS3, and lysis of the infected cells could be demonstrated (28). To test whether the insertion of LC3* is able to convert a noncp BVDV into a cp virus, a full-length construct named pA/B-JaCP was generated by insertion of the NcoI/SalI insert of pEx7/JaCP into pA/BVDV/Ins− cut with the same enzymes (Fig. 7A). Construct pA/BVDV/Ins− represents a cDNA clone from which infectious noncp BVDV RNA can be transcribed (28). As a control, an equivalent construct without the LC3* insertion was established by using a fragment of pEx7/JaNCP; this plasmid was termed pA/B-JaNCP (Fig. 7A). RNA was transcribed from the two plasmids and used for transfection of MDBK cells. Transcripts derived from pA/BVDV/Ins− or pA/BVDV served as controls for non-cp and cp genomes, respectively. As described before (28), transfection of RNA derived from pA/BVDV resulted in lysis of the transfected cells, whereas a CPE was not observed in the case of pA/BVDV/Ins− (Fig. 7B). Upon transfection of RNA derived from pA/B-JaNCP and pA/B-JaCP, a CPE could not be detected (Fig. 7B, upper row). While this finding was expected in the case of pA/B-JaNCP, it was surprising for the RNA containing the LC3* insertion. Even more surprisingly, further analyses showed that in the latter case no infectious virus had been generated. In a Northern blot with total RNA isolated 72 h posttransfection, BVDV-specific signals could be detected for cells transfected with transcripts derived from pA/BVDV, pA/BVDV/Ins−, and pA/B7-JaNCP RNA but not for cells transfected with pA/B-JaCP-derived RNA (Fig. 7C). Similarly, immunofluorescence analyses allowed the detection of positive cells for the first three constructs but not for pA/B-JaCP (data not shown). These results indicated that transcripts derived from pA/B-JaCP were not infectious.

FIG. 7.

Results of transfection of MDBK cells with RNA transcribed from the full-length cDNA constructs pA/B-JaCP and pA/B-JaNCP. RNA derived from pA/BVDV and pA/BVDV/Ins− (28) served as controls for cp and noncp viruses, respectively. (A) Schematic presentation of the protein coding region of the cDNA constructs. pA/BVDV/Ins− was established from pA/BVDV by deletion of a 27-nucleotide insertion responsible for cytopathogenicity of the virus recovered from the latter construct (28). The regions derived from pJaCP/A17 or pJaNCP35 and inserted via BamHI and HpaI cuts are marked by a thick broken or a dotted line, respectively. For further information, see the legend to Fig. 2. Checkered bar, N^pro gene; dark gray bar, NS2 gene; light gray bar, NS3 gene; small white bar within the NS2 gene of pA/BVDV, 27-nucleotide insertion responsible for cytopathogenicity. (B) Crystal violet staining of cells that were fixed 72 h after transfection with the RNAs transcribed in vitro from the indicated cDNA constructs. Cells were washed once with phosphate-buffered saline, fixed for 10 min with 5% formaldehyde, washed extensively with water, and stained for 5 min with 1% (wt/vol) crystal violet (in 50% ethanol). The upper row shows the results of the transfection of noninfected MDBK cells, whereas the lower row shows dishes into which were seeded cells that had been infected 24 h before transfection, with non-cp BVDV NCP1 at a multiplicity of infection of 0.1. (C) Northern blot with RNA derived from cells at 72 h posttransfection hybridized with a BVDV-specific probe. Only the result for the cells which were not infected prior to transfection are shown. RNA ladder (in kilobases) is on the left side of the gel.

In order to verify that the latter finding was not due to a deleterious mutation in the cDNA construct, different experiments were conducted. After transient expression in the vTF7-3 vaccinia virus system, viral proteins were analyzed by immunoprecipitation, SDS-PAGE, and fluorography. The expression of polypeptides corresponding to all viral proteins could be demonstrated. No abnormality with regard to size or amount of the individual proteins was observed (data not shown).

As a second control, a plasmid equivalent to pA/BVDV/Ins− was reconstructed, starting with the defective clone pA/B-JaCP. The region of pA/B-JaCP containing the LC3* insertion was exchanged for the corresponding fragment from pA/BVDV/Ins−. After transfection of RNA transcribed from the reconstructed plasmid, infectious non-cp virus was recovered (not shown). The sequence of the fragments deleted from pA/B-JaCP did not exhibit differences with regard to the expected sequence. Thus, no obvious mistake present in pA/B-JaCP was responsible for the fact that RNA derived from this plasmid did not yield infectious virus.

Different cp pestivirus isolates were found to be composed of noncp helper viruses and cp-defective interfering particle (31). In these cases, cell lysis was observed upon transfection of viral RNA when the cells were infected with a noncp helper virus prior to introduction of the RNA. We therefore investigated whether RNA derived from pA/B-JaCP was also able to induce a CPE in the presence of a helper virus. Accordingly, the transfection experiments were repeated with target cells previously infected with a noncp BVDV. Whereas the RNA transcribed from pA/BVDV/Ins− or pA/B-JaNCP was again not able to induce a CPE, transfection of the RNA containing LC3* resulted in cell lysis (Fig. 7B, lower row). This result was not dependent on the helper virus strain, since cells infected with NCP7 or V(pA/BVDV/Ins−), the virus derived from pA/BVDV/Ins−, also developed CPE upon transfection with the RNA containing LC3* (data not shown). Taken together, these results indicate that the presence of LC3* in pA/B-JaCP leads to recovery of a defective cp virus.

In order to demonstrate that an autonomously replicating cp BVDV could be generated by integration of JaCP cDNA fragments into pA/BVDV/Ins−, a construct resembling the JaCP genome was assembled. A 3-kb NcoI/SalI fragment from pA/BVDV/Ins− was exchanged for the corresponding 5.5-kb fragment from a cDNA clone, pEx7/JaCP/dp, that was assembled from pEx7/JaCP and a SacI/BamHI fragment of cDNA clone pJaCP/A19 containing the duplication. The resulting construct was termed pA/B-JaCP/dp (Fig. 8A). As a control, a second construct, which contained the duplicated sequence but not the LC3* insertion, was established (Fig. 8A). For this clone the NcoI/SalI fragment from cDNA clone pEx7/JaNCP/dp was used instead of the corresponding fragment from pEx7/JaCP/dp. After transfection of in vitro-transcribed RNA, cell lysis was observed in the case of pA/B-JaCP/dp (Fig. 8B). It therefore can be concluded that RNA containing sequences from JaCP and CP7 can serve as the genome of an autonomously replicating cp BVDV. For RNA derived from pA/B-JaNCP/dp, cell lysis could not be observed (Fig. 8B). Moreover, immunofluorescence analysis and Northern hybridization indicated that no viable virus could be recovered (not shown and Fig. 8C, respectively). Experiments with a reconstructed plasmid similar to pA/BVDV/Ins−, which was obtained by deletion of a SacI fragment containing the duplication, resulted in recovery of noncp BVDV (not shown). It is therefore very likely that a BVDV RNA containing the JaCP duplication without the LC3* insertion is defective.

FIG. 8.

Results of the transfection of MDBK cells with RNA transcribed from the full-length cDNA constructs pA/B-JaCP/dp and pA/B-JaNCP/dp. RNA derived from pA/BVDV/Ins− served as a control. (A) Schematic presentation of the protein coding region of the cDNA constructs. The regions derived from pJaCP/A17 or pJaCP/A19 are marked by a thick broken line. Sequences derived from pJaNCP35 are marked by a dotted line. The positions of restriction sites important for the cloning procedures are indicated. For further information, see the legends to Fig. 2 and 7. (B) Crystal violet staining of cells that were fixed 72 h after transfection with the RNAs transcribed in vitro from the indicated cDNA constructs. (C) Northern blot with RNA derived from cells at 72 h posttransfection hybridized with a BVDV-specific probe. RNA ladder (in kilobases) is on the left side of the gel.

DISCUSSION

Molecular characterization of pestiviruses has revealed that the genomes of most cp viruses exhibit alterations which are not found in the RNAs of noncp isolates. Different rearrangements of viral sequences or insertions of cellular sequences, sometimes accompanied by large duplications of viral sequences, have been identified (2, 26, 27, 29–31, 34, 42–44). In the past, two types of cellular insertions, which code for either ubiquitin or part of a cellular protein of unknown function, termed cIns, were found (2, 26, 29, 31, 34). In this communication, we report on a novel kind of cellular insertion that shows no homology to the already known cellular inserts and is derived from an mRNA coding for LC3 of MAPs 1A and 1B. Cellular LC3 represents a small basic protein which is abundant only in neurons and which is able to bind microtubules as well as the MAPs 1A and 1B. The function of LC3 in normal cell life has not yet been elucidated (23). Interestingly, the polypeptide encoded by LC3* exhibits only four exchanges with respect to the LC3 sequence from rat. Since LC3* was most likely derived from a bovine mRNA, it can be concluded that the LC3 amino acid sequence has been highly conserved during evolution, indicating an important function of this protein.

The genome of JaCP represents the first BVDV RNA, for which the presence of an LC3 coding sequence has been identified. In contrast, insertions coding for ubiquitin have been found in the genomes of several independent virus isolates (for a review, see reference 31). With regard to the 3′ end, all nine ubiquitin-coding insertions identified so far are found at the same genomic position, which represents the 5′ end of the NS3 gene (position B in Fig. 2A). In the polyprotein, the presence of ubiquitin induces cleavage by a ubiquitin-specific cellular protease and thus is responsible for generation of the amino terminus of NS3. Taking into account our knowledge about different cp pestiviruses, it can be concluded that the presence and the location of the cellular sequences or other rearrangements in the genomes of cp pestiviruses are not random but should be regarded as a result of functional selection. The genome rearrangement leading to a cp BVDV has to induce processing at the amino terminus of NS3, thus allowing expression of this protein, which represents a specific feature of cp BVDV. In accordance with these considerations, LC3* is located in the JaCP genome at the conserved position B (Fig. 2A) and is able to mediate NS2-3 cleavage. It therefore can be hypothesized that the polypeptide encoded by LC3* either (i) exhibits autoproteolytic activity, (ii) serves as a signal for a cellular protease, or (iii) induces a defined conformation which allows NS2-3 cleavage via a usually cryptic mechanism. The N-terminal sequencing of the NS3 protein expressed from pEx7/JaCP showed that the LC3*-induced processing occurs just downstream of the cellular insertion. This is again reminiscent of cp BVDV-encoded proteins containing ubiquitin. It might turn out in the future that LC3*, like ubiquitin, serves as a target for a specific cellular protease. In any case, elucidation of the mechanism of LC3*-induced cleavage of NS2-3 should also shed light on the function of its cellular counterpart.

Processing of NS2-3 induced by the LC3*-encoded polypeptide results in an NS3 protein starting with a glycine corresponding to residue 1590 in a BVDV polyprotein without insertion. Cleavage of BVDV proteins containing ubiquitin most likely leads to NS3 proteins with the same amino terminus. Interestingly, at least three other types of cp BVDV express NS3 with the identical amino terminus, although the mechanisms responsible for generation of this protein apparently are different. In the case of BVDV Pe515CP, CP6, and DI9, the pestivirus N^pro protease is located upstream of NS3. This protease cleaves autocatalytically at its own carboxy terminus and thereby releases NS3 (30, 31, 44). The NS2 expressed by BVDV NADL contains the cellular clns insertion located 54 residues upstream of the position where ubiquitin or LC3* is found. The mechanism of NS2-3 processing has not been elucidated for this virus, but preliminary results obtained by N-terminal sequencing of NS3 indicated cleavage just upstream of a glycine corresponding to residue 1590 (50). Finally, NS2-3 of BVDV Oregon is cleaved because of point mutations within NS2, giving rise to NS3 with glycine 1590 as the amino terminus (20). In vitro, cleavage of NS2-3 of BVDV Oregon is dependent on microsomal membranes, whereas both the ubiquitin and the LC3*-induced processing can occur in the absence of membranes. As there is obviously no common mechanism responsible for this high conservation of the amino terminus of NS3, the function of this protein for either cytopathogenicity or viability of the viruses has to be dependent on the correct N terminus. Future analyses with infectious BVDV full-length clones and replicons will hopefully help to elucidate this interesting phenomenon.

According to a widely accepted hypothesis, RNA virus genomes recombine by template switching of the viral RNA-dependent RNA polymerase during replication (for reviews, see references 3, 17, 19, and 22). Direct generation of the JaCP genomic RNA requires three consecutive template switches, which is a very improbable event. Alternatively, the JaCP RNA could have been generated via two independent recombination reactions. In a first step, a recombination between viral and cellular RNA could yield a virus genome containing LC3*, integrated between NS2 and NS3, but no duplication. The resulting genomic RNA could be amplified by replication and then undergo recombination with the RNA of the original noncp BVDV to produce the JaCP genome containing LC3* and the duplication of viral sequences. Selection of the mutant generated in the second step would be favored if the first virus were somehow defective and the introduction of the duplication helped to overcome this problem. It has to be stressed in this context that the virus recovered after transfection of the RNA transcribed from pA/B-JaCP was defective and therefore dependent on a helper virus. The in vitro-generated RNA exactly mimics the product of the hypothetical first recombination, since it contains LC3* at the correct position of the NS2-3 gene but no duplication. Since viable viruses could be recovered from pA/B-JaCP/dp, the corresponding construct containing the duplication, and from pA/B-JaNCP, which lacks both LC3* and the duplication, the defect of pA/B-JaCP can hardly be due to the genetic background of our infectious clone. The majority of cp BVDV isolates contain in their genomes duplicated viral sequences encompassing the NS3 gene and sequences located further downstream in the genome. The 3′ ends of the duplications vary between position 7456 and 8788 of a regular genome (BVDV Pe515 CP and BVDV CP6, respectively [31]). The prevalence of rearranged genomes with duplications could be explained by the recombination mechanism. A recombination leading to a duplication allows flexibility with regard to the last template switch and should therefore be more likely than a recombination integrating an additional sequence between two neighboring nucleotides. However, the findings reported here indicate that at least in some cases the duplications might be functionally relevant. Further investigations are necessary to elucidate the role of the duplicated sequences with regard to viability and cytopathogenicity of the viruses.

In BVDV the ability to express NS3 is directly linked to the cp phenotype (7, 33). Since in the case of JaCP LC3* is apparently responsible for expression of NS3, the integration of the cellular sequence has to be regarded as causative for the CPE induced by this virus; this conclusion is strongly supported by the recovery of cp viruses upon transfection of the in vitro-transcribed RNA containing LC3*. JaCP has been isolated from an animal suffering from lethal MD (14). The generation of a cp virus in an animal persistently infected with a noncp virus is regarded as crucial for development of the disease. The homology found for corresponding parts of the sequences from noncp and cp BVDV isolated from one diseased animal proved that the cp virus represents a mutant of the persisting noncp virus (31). Since Jasper had been housed under isolated conditions a long time before he came down with MD (14) it is obvious that JaCP has also developed from JaNCP within the persistently infected animal. Thus, Jasper died most likely as a consequence of a recombination between the genome of JaNCP and an mRNA coding for MAP LC3. Molecular analysis of further samples taken just before the onset of MD and during the clinical disease will help to improve understanding of the processes leading to this interesting disease.