JOURNAL OF VIROLOGY, June 1995, p. 3683–3689 Vol. 69, No. 6 0022-538X/95/$04.00ϩ0 Copyright ᭧ 1995, American Society for Microbiology

Cytopathogenicity of Classical Swine Fever Caused by Defective Interfering Particles

GREGOR MEYERS* AND HEINZ-JU¨ RGEN THIEL† Federal Research Centre for Virus Diseases of Animals, 72001 Tu¨bingen, Germany

Received 22 December 1994/Accepted 10 March 1995

For three independent cytopathogenic isolates of classical swine fever virus, defective RNAs were found in infected cells in addition to full-length viral genomes. These RNAs represent the genomes of typical defective interfering (DI) particles because of strict dependence on a complementing helper virus and interference with the replication of the helper virus. Analysis of the DI genomes revealed internal deletions of 4,764 nucleotides encompassing the complete structural protein-coding region of the virus and two flanking nonstructural genes. Plaque isolation and RNA transfection experiments showed that the DI particles are responsible for the cytopathic effect caused by these classical swine fever virus isolates.

Classical swine fever virus (CSFV), the causative agent of an and nonessential amino acids. Cells and virus stocks were tested regularly for the economically important disease of swine, is a small enveloped absence of mycoplasma contamination. Infection of cells. Since tend to be associated with the host cells, virus (23). The CSFV genome is represented by a single- lysates of infected cells were used for reinfection of culture cells. Lysates were stranded RNA of about 12.3 kb which is of positive polarity prepared by freezing and thawing cells 48 h postinfection and were stored at and comprises one long open reading frame (ORF). Transla- Ϫ70ЊC. If not indicated differently in the text, a multiplicity of infection (MOI) tion of the genomic RNA leads to a hypothetical polyprotein of about 0.5 was used for infections. which is co- and posttranslationally processed by both virus- Northern (RNA) hybridization. RNA preparation, gel electrophoresis, radio- active labeling of the probe, hybridization, and posthybridization washes were and host cell-encoded proteases (18, 24, 30, 31, 40, 41). done as described before (33). A 2.2-kb SalI fragment from CSFV Alfort cDNA CSFV is a member of the family , genus Pestivi- clone 4.5 (18) was used as a probe. rus; the latter also includes virus (BVDV) cDNA cloning and nucleotide sequencing. Establishment of cDNA libraries in and border disease virus (BDV) of (39). For BVDV, the lambda ZAPII (Stratagene, Heidelberg, Germany), library screening (same existence of both cytopathogenic (cp) and noncytopathogenic probe as for Northern hybridization), and nucleotide sequence determination were done as described before (21). Sequence analysis was done with the Ge- (noncp) isolates is well known. Either biotype can be isolated netics Computer Group software (10). Oligonucleotides for priming of the first from field cases of bovine viral diarrhea (2, 23). The coexist- strand were BVD13, BVD14, PES9 (21), and BVD24, which is complementary to ence of a cp and a noncp virus in one animal has been dem- nucleotides 7757 to 7773 of the published CSFV Alfort sequence (18). The onstrated for cattle which came down with mucosal disease sequence of BVD24 is CCYTCYTGYTGNGTYTC. Positions of the 5Ј and 3Ј (MD) (5, 6, 17, 23). Molecular analyses revealed that com- ends of the obtained cDNA clones with respect to the CSFV Alfort sequence are as follows: Alfort/M K1, 12 to 6516; ATCC K5, 32 to 7746; ATCC K9, 320 to pared with noncp BVDV, the genomes of cp isolates show 7280; ATCC K16, 99 to 7746; ATCC K13, 3419 to 5925; Steiermark K3, 31 to rearrangements due to recombination events. Integration of 7758; Steiermark K9, 31 to 7767; and Steiermark K11, 30 to 7773. host cellular sequences with or without duplication of viral Plaque isolation. PK15 cells in a 6.0-cm-diameter dish were infected with the sequences, complex duplication and rearrangement of viral respective CSFV isolate and overlaid with 1.8% methylcellulose. After about 36 h, plaque formation was observed. Material from individual plaques was isolated sequences, and internal deletions of considerable parts of the with Pasteur pipettes and diluted in 500 ␮l of medium. These suspensions were genome were identified (19–22, 27, 32, 33). used for infection of PK15 cells seeded in 24-well plates; 250 ␮l of plaque In the case of CSFV, the majority of isolates do not exhibit suspension was used for each infection. Superinfection with CSFV Alfort was cytopathogenicity in tissue culture (23). There are a few older done 1 h later. After 48 h, cells from wells without detectable cytopathic effect publications on identification of cp CSFV (9, 11, 16, 37). Many (CPE) were passaged into 3.5-cm-diameter dishes. RNA transfection. Transfection was done with a suspension of 3 ϫ 106 PK15 publications actually state that CSFV is generally considered to cells and 10 ␮g of RNA bound to DEAE-dextran (Pharmacia, Freiburg, Ger- be noncytopathogenic. We report here data on molecular char- many). Total RNA from PK15 cells infected with the respective CSFV isolate acterization of three independent cp CSFV isolates. was used for transfection. The RNA/DEAE-dextran complex was established by mixing RNA dissolved in 100 ␮l of Hanks balanced salt solution (HBSS) with 100 ␮l of DEAE-dextran (1 mg/ml in HBSS) and incubation for 30 min on ice (38). MATERIALS AND METHODS 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 Cells and . PK15 cells and CSFV ATCC VR-531 were obtained from 37ЊC, 20 ␮l of dimethyl sulfoxide was added and the mixture incubated for 2 min the American Type Culture Collection (Rockville, Md.). CSFV Alfort (1) (ob- at room temperature. After addition of 2 ml of HBSS, cells were pelleted and tained from B. Liess [Veterinary School, Hannover, Germany]) was reisolated washed once with HBSS and once with medium without FCS. Cells were resus- from organs of an experimentally infected moribund animal (29). A cp variant of pended in DMEM with FCS and split into culture dishes. For each transfection CSFV Alfort (named CSFV Alfort/M) was isolated after about 230 passages in experiment, cells were seeded in two 3.5-cm-diameter dishes (5% of the cells for pig lymphoma cell line 38A1D (kindly provided by W. Scha¨fer, Max-Planck- each dish) and one or two 10.0-cm-diameter dishes (40% of the cells per dish). Institut fu¨r Virusforschung, Tu¨bingen, Germany). CSFV Steiermark was kindly At 48 to 72 h posttransfection, one of the small dishes was used for immunoflu- provided by R. Ahl and A. Kosmidou, Federal Research Centre for Virus orescence analysis, and the other was used for crystal violet staining. The cells Diseases of Animals, Tu¨bingen, Germany. Cells were grown in Dulbecco mod- from a 10.0-cm-diameter dish were lysed for RNA preparation. ified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS) Immunofluorescence assay. Immunofluorescence with a mixture of anti-CSFV monoclonal antibodies a18 and 24/16 was done as described previously (35, 36). Since the antibodies are directed against CSFV structural proteins, and massive * Corresponding author. Mailing address: Federal Research Centre antigen production is necessary for positive results, only CSFV helper viruses but for Diseases of Animals, P.O. Box 1149, 72001 Tu¨bingen, Germany. not CSFV defective interfering (DI) particles are monitored with this assay. † Present address: Institute of Virology, University of Giessen, Crystal violet staining of cells. Cells were washed once with phosphate-buff- D-35392 Giessen, Germany. ered saline, fixed for 10 min with 5% formaldehyde, washed with water, and

3683 3684 MEYERS AND THIEL J. VIROL. stained for 5 min with 1% (wt/vol) crystal violet in 50% ethanol. After extensive washing with water, cells were dried. Radioimmunoprecipitation and SDS-PAGE. CSFV-infected PK15 cells (106) were labeled for 14 h with 0.5 mCi of [35S]cysteine/methionine (Promix; Amer- sham, Braunschweig, Germany) per ml. Labeling medium contained no cysteine and 5% of the normal methionine content. Cell extracts were prepared under denaturing conditions (12). Extracts were incubated with 5 ␮l of undiluted serum. Precipitates were formed with cross-linked Staphylococcus aureus (15), analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- PAGE), and processed for fluorography using En3Hance (New England Nuclear, Boston, Mass.). Nomenclature. The designation of pestiviral nonstructural proteins used in this report has been suggested by a member of the ICTV Flaviviridae Study Group but has not been confirmed yet. CSFV has formerly been termed hog cholera virus. Nucleotide sequence accession number. The sequence data reported have been deposited with the EMBL/GenBank data libraries under accession numbers U21328 (CSFV ATCC), U21329 (CSFV Steiermark), and U21330 (CSFV Al- fort/M).

RESULTS Characteristics of virus isolates. CSFV isolates usually do not exhibit CPE on tissue culture cells. This is also true for a CSFV isolate derived from CSFV Alfort after one passage in a pig and reisolation from the moribund animal in 1987 (29). In the following years, the virus was characterized extensively both at the genome and protein levels. During this time, CSFV Alfort was passaged about 230 times with a standard MOI of about 0.02 without detection of changes with respect to tissue FIG. 1. Northern blot analysis of RNA from PK15 cells infected with CSFV culture behavior, genome characteristics, or protein pattern. isolates Alfort, Alfort/M, ATCC, and Steiermark, respectively. The MOI based However, during one experiment in the course of which in- on immunofluorescence was about 0.5 for all viruses. Hybridization was done fected cells were serially passaged, CPE was observed. We with a CSFV Alfort-derived probe. Exposure times were 2 h for lane 1 and 24 h were not able to determine at which stage of this experiment for the other three lanes. Numbers on the left indicate the sizes of an RNA ladder in kilobases. For CSFV Alfort/M, a second band comigrating with the the cp virus was generated, but subsequent control experi- genomic RNA of CSFV Alfort was detectable after longer exposure times (see ments with virus from older passages did not result in CPE. also Fig. 2). Additional bands visible in the smear of degraded CSFV RNA below The isolated cp virus was termed CSFV Alfort/M. CSFV Al- the 8-kb band in lane 2 represent gel artifacts resulting from the presence of large fort/M is able to provoke complete lysis of infected tissue amounts of rRNA in the corresponding region of the gel. culture cells in less than 3 days postinfection, while in parallel experiments, the original noncp virus has apparently no effect on the cells (data not shown). tion interferes with replication of the helper virus. Accordingly, CSFV ATCC VR-531, termed CSFV ATCC below, has been the ratio between DI particle and helper virus varies in the described as a cp isolate obtained from a persistently infected course of infection and is heavily influenced by the MOI used young pig (11). CSFV Steiermark was isolated during an for the initial inoculation (3, 13, 14, 28). To assess the influence outbreak of classical swine fever in Austria in 1992. The cyto- of MOI on the appearance of the 8-kb RNA, infections with pathogenicity of this virus was observed during the first pas- serial dilutions of virus stocks were carried out. The infected sages, indicating that CSFV Steiermark was already cytopatho- cultures were harvested after 24 to 72 h. Total cellular RNA genic when isolated from the infected animal. was prepared and analyzed by Northern blotting. Figure 2 RNA analysis. In the BVDV system, generation of cp viruses shows the result for CSFV Alfort/M. When a high MOI was is a result of genome rearrangements which in some cases lead used for inoculation, only the 8-kb RNA was identified (Fig. 2, to changes in genome size easily detectable by Northern blot dilutions of 1:2 and 1:20). Longer exposure times resulted in analyses (21, 22, 27, 33). Characterization of the cp CSFV detection of faint bands corresponding to genomic CSFV RNA isolates was therefore first performed by Northern hybridiza- (data not shown). Dilution of the original virus stock 1:200 tion with a CSFV-specific probe. RNA from cells infected with prior to infection led to about equal amounts of 12.3- and 8-kb the original noncp CSFV Alfort served as a control (Fig. 1, RNAs; infection with an even lower MOI reduced the amount lane 1). Surprisingly, in all lanes with RNA of cells infected of the smaller RNA species to or below the background level with cp CSFV, a specific band of about 8 kb was detected (Fig. (Fig. 2, dilutions of 1:200, 1:2,000, and 1:20,000). The proper- 1, lanes 2 to 4). In addition, a band comigrating with the ties of the 8-kb RNA determined by the dilution experiment 12.3-kb genomic RNA of CSFV Alfort (Fig. 1, lane 1) was mirror the above-mentioned characteristics of DI genomes, visible for CSFV ATCC and CSFV Steiermark (Fig. 1, lanes 3 namely, interference with helper replication at a high MOI and and 4). loss of DI particles by separation from the helper function at a Characterization of the 8-kb RNA. The result of the North- low MOI. Equivalent results were obtained for the other two ern blot analysis is reminiscent of BVDV CP9, for which also cp CSFV isolates. For all three isolates, infection and serial two RNAs of about 12.5 and 8 kb were detected in infected passage with a low MOI resulted in virus stocks no longer cells. While the larger RNA represented a full-length pestivi- generating the 8-kb RNAs in infected cells (data not shown). rus genome, the 8-kb molecule was found to be the genome of Accordingly, all three cp CSFV isolates represent mixtures a DI particle (33). composed of DI particles and nondefective helper viruses. DI particles are characterized by two typical properties. On Identification of the cytopathogenic agent. For BVDV iso- the one hand, they are dependent on a helper virus to com- late CP9, it was shown that the DI particle represented the plement their genomic defects; on the other hand, DI replica- cytopathogenic agent (33). It was therefore possible that the VOL. 69, 1995 CYTOPATHOGENIC CSFV DEFECTIVE INTERFERING PARTICLES 3685

results strongly suggest that development of CPE is dependent on the presence of a helper virus and thus is linked to repli- cation of the DI particles. The failure to detect CPE without addition of exogenous helper virus is probably due to low amounts of helper virus in the plaque material resulting from the interference with the DI particles. Cells from the 12 wells which had not been superinfected with noncp helper virus and had not shown CPE were passaged once. After 48 h, CPE could be detected in 6 of the 12 dishes (Table 1). Probably the initially isolated material of these six plaques contained enough helper virus to allow rescue of DI particles after one passage. This explanation implies that the DI particles have the ability to survive within an infected cell for a rather long time without helper virus. In the second series of experiments, noninfected PK15 cells or CSFV Alfort-infected PK15 cells were transfected with total RNA from cells infected with CSFV Alfort, Alfort/M, ATCC, or Steiermark. To study CPE development, the transfected cells were seeded in culture dishes and incubated at 37ЊC for 4 to 6 days. CPE was observed only when transfection of RNA derived from cells infected with cp CSFV was preceded by infection with noncp CSFV Alfort (Fig. 3A). Accordingly, re- covery of cp virus after transfection is observed only in the FIG. 2. Northern blot analysis of total RNA from PK15 cells infected with presence of a helper virus in the target cells. different MOIs of CSFV Alfort/M. The dilution of the virus stock used for each For further analysis of the transfection experiment, RNA infection is indicated. A 1:2 dilution of the stock results in an MOI of about 0.25 was prepared and analyzed by Northern hybridization. When (titration based on immunofluorescence assay). Sizes of an RNA ladder are RNA from cells infected with noncp CSFV Alfort was used for indicated in kilobases on the left. See the legend to Fig. 1 for information concerning additional bands visible in the smear below the main bands. transfection of noninfected target cells, genomic CSFV RNA could easily be demonstrated in the transfected cells (Fig. 3B). In contrast, no viral or DI genome was detectable when RNA from cells infected with one of the cp CSFV strains was used in detected CSFV DI particles are responsible for the CPE of the the equivalent experiment (Fig. 3B). Since RNA samples with cp CSFV isolates. A first hint supporting this hypothesis was high DI/helper virus genome ratios (equivalent to Fig. 1, lane obtained by the dilution experiments reported above. Infec- 2) were used, this result is probably due to the lack of sufficient tions with a low MOI resulting in elimination of the DI RNA helper virus to support DI replication. In accompanying im- also led to loss of cytopathogenicity. Cloning of autonomously munofluorescence analyses, only a few positive cells were de- replicating viruses from the cp CSFV stocks by endpoint dilu- tected in the samples. Passaging of the transfected cells re- tion rescued only noncp viruses. sulted in the recovery of viral genomic RNA but without Further evidence for the assumption that the DI particles development of CPE (data not shown). represent the cytopathogenic agents of the cp CSFV isolates When RNA from cells infected with cp CSFV was trans- was sought by experiments again relying on the helper depen- fected into cells harboring a noncp helper virus, high amounts dency of a DI particle. For the first set of experiments, 12 of genomic DI RNA could be detected on the Northern blot individual CSFV Alfort/M plaques were isolated. The material (Fig. 3C). The transfection experiments demonstrate that for from each plaque was used for infection of two dishes with all cp CSFV isolates, development of CPE requires a helper PK15 cells. One of these dishes was superinfected with noncp virus and correlates with the presence of the respective DI CSFV Alfort 1 h after addition of the plaque material. During genomes. Therefore, CSFV Alfort/M, ATCC, and Steiermark the following days, the cultures were monitored for CPE de- each represent mixtures of a noncp helper virus and a cp DI. velopment. Within 48 h, complete lysis of cells was detectable Genome structure of CSFV DI particles. The cp BVDV DI9 in all wells infected with noncp CSFV in addition to the plaque has a genome of the internal deletion type. The complete material. In contrast, plaques were not detectable in wells not structural protein-coding region and the 5Ј part of the NS2-3 superinfected with helper virus (Table 1). CPE could not be (p125) gene have been removed with respect to a full-length detected in control experiments in which cells infected with BVDV genome (33). Northern blot analysis had shown that noncp CSFV were superinfected with the same virus. These the CSFV DI genomes are smaller than the BVDV CP9 DI genome (data not shown). For further analysis, cDNA cloning and sequencing were performed. RNA with a high content of TABLE 1. Inoculation of PK15 cells with material isolated from DI genomes served as template for first-strand cDNA synthe- individual CSFV Alfort/M plaques with and without sis. Northern hybridization experiments had suggested that the superinfection with noncp CSFV Alfort deletions of the cp CSFV DI particles were located in the 5Ј half of the genome (data not shown). Therefore, specific oli- Superinfection Induction of CPE with CSFV Alfort gonucleotides complementary to the genome around positions 6000, 7000, 8000, and 9000 were used as cDNA prim- ϩ ϩϩϩϩϩϩers. Clones with CSFV-specific inserts were further analyzed ϩϩϩϩϩϩ aa a aby sequencing of the 5Ј and 3Ј ends. From comparison of the ϪϪϪϪϪ ϪϪterminal sequences with the sequence of a full-length CSFV ϪϪa ϪϪ Ϫa Ϫ genome, putative DI particle-specific cDNA clones were iden- aCPE was detected after one passage. tified. DI particle-derived clones should meet the criterion that 3686 MEYERS AND THIEL J. VIROL.

fragments are colinear with the full-length genome until posi- tion 366, which represents the last nucleotide of the translation initiation codon of the long ORF. The following region is homologous to a part of the CSFV genome starting at position 5131 (Fig. 4B). Accordingly, all three DI genomes have been generated by an internal deletion removing the Npro gene, the region coding for the structural proteins C, E0, E1, and E2, and part of the NS2-3 gene (Fig. 4C). The location of the deletion’s 3Ј end is exactly the same as that identified for the BVDV CP9 DI genome (33). The identities and locations of the deletions found in the genomes of the CSFV DI particles were confirmed by sequenc- ing the respective parts of additional cDNA clones as well as by direct sequencing of PCR products using specific primers (data not shown). For a region of about 300 nucleotides, no se- quence variation was detected among clones from a single isolate. The CSFV DI particles are derived from three different origins. Since for all three the same genome structure was found, the possibility of one DI particle contaminating the other virus stocks had to be excluded. Accordingly, sequence comparison studies were conducted. Comparison of 1,739 nucleotides showed only four base differences between the CSFV Alfort genome (18) and the same region of its DI par- ticle. Finding more than 99% sequence identity strongly sug- gests that the defective genome has arisen from the CSFV Alfort RNA. The extent of similarity between CSFV isolates Alfort/M and ATCC, Alfort/M and Steiermark, or ATCC and Steiermark is 89, 92, or 90%, respectively. These results can be taken as proof that the three DI genomes are different from each other. Protein analysis. In the BVDV system, the cytopathic phe- notype is correlated with expression of NS3 (p80), which rep- resents the carboxy-terminal region of the nonstructural pro- tein NS2-3 (p125) (7, 8, 25, 26). While NS2-3 is found in cells infected with either cp or noncp viruses, NS3 can be detected only in cells infected with cp BVDV. Generation of NS3 in cp BVDV-infected cells was found to be the result of recombina- tional rearrangements in cp BVDV genomes (22, 32). This is also true for the BVDV isolate CP9, in which the DI particle represents the cytopathogenic agent and expresses NS3. In the case of the CP9 DI particle, the amino terminus of NS3 is FIG. 3. Results of transfection experiments. Total RNA from cells infected generated by the proteolytic activity of the preceding protease pro with different CSFV isolates was transfected into PK15 cells or CSFV Alfort- N (33). infected PK15 cells. (A) Crystal violet staining of tissue culture dishes into which For CSFV, the situation is more complex than that described cells had been seeded after transfection. Two sets of experiments are shown, one above for BVDV. Both NS2-3 and NS3 are expressed from with noninfected target cells (top row) and one with CSFV Alfort-infected target cells (bottom row). The slightly cloudy staining of the dishes in the top row and noncp CSFV genomes (34). To analyze whether differences in the leftmost dish in the bottom row is not due to virus-induced CPE but results NS3/NS2-3 expression could be detected between cells in- from an inhomogeneous cell layer due to very high cell density. (B and C) fected with noncp CSFV and cp CSFV, immunoprecipitation Northern blot analysis of total RNA isolated from PK15 cells (B) or CSFV experiments were carried out. As expected, the anti-A3 serum, Alfort-infected PK15 cells (C) 3 days after transfection with RNA of the viruses indicated at the top. Sizes of an RNA ladder are indicated in kilobases on the which is directed against the carboxy-terminal region of NS2-3, left. precipitated both NS2-3 and NS3 from cells infected with CSFV Alfort (Fig. 5). In contrast, NS2-3 is not detected for CSFV Alfort M, while NS3 is visible as a strong band. After much longer exposure times, a faint NS2-3 band was observed their actual insert size is about 5 kb below the theoretical size. (data not shown). In the lane with CSFV ATCC, proteins Clones Alfort/M K1, ATCC K5, K9, and K16, and Steiermark NS2-3 and NS3 are detectable; however, the ratio between the K3, K9, and K15 all met this criterion (Fig. 4A). Since RNAs two bands again is dramatically different from that for noncp with a high DI/helper virus genome ratio were used for cDNA CSFV Alfort. Similar results were obtained in equivalent ex- cloning, only few clones derived from the helper genome were periments with CSFV Steiermark (data not shown). isolated. One of these clones is ATCC K13 (Fig. 4A). Clones Alfort/M K1, ATCC K16, and Steiermark K3 were chosen for DISCUSSION sequence determination. Interestingly, the genome structure determined by comparison of the analyzed sequences with the One interesting property of pestiviruses is the existence of CSFV Alfort sequence was exactly the same for all three DI two biotypes consisting of cp and noncp viruses. Most de- genomes. Starting in the 5Ј noncoding region, the three cDNA scribed cp pestiviruses are BVDV strains. Both cp and noncp VOL. 69, 1995 CYTOPATHOGENIC CSFV DEFECTIVE INTERFERING PARTICLES 3687

FIG. 4. (A) Schematic presentation of the cDNA clones isolated from libraries derived from RNA of cells infected with CSFV Alfort/M, ATCC, and Steiermark. The top bar with numbers represents a size scale in kilobases. Below the size scale, a CSFV genome is indicated by bars. The thin bars represent the 5Ј and 3Ј noncoding regions, while the thick bar indicates the long ORF. The different cDNA clones are indicated below the genome scheme by bars. The thick bars represent the initially determined terminal sequences and their positions with respect to the genome. The thin bar shows the region of the viral genome theoretically spanned by the cDNA fragment (according to the location of the termini). The actual insert size of the cDNA clones is given in brackets following the name of the clone. For exact positions of 5Ј and 3Ј ends, see Materials and Methods. (B) Alignment of part of the sequences determined for the CSFV DI genomes with that of CSFV Alfort. The position of each fragment of the published CSFV Alfort sequence is indicated. Homologous sequence stretches are marked by a box. Within the box, nucleotides identical with the residues of the 5Ј nontranslated region are shown as Ϫ, while those identical with the NS3 gene are indicated by ϩ. (C) Deduced structure of the three CSFV DI genomes compared with that of CSFV Alfort (top bar) and BVDV CP9 DI genomes (bottom bar). The deletions identified in the different DI genomes are indicated by stippled lines between the bars representing the different regions of the genomes. The genes coding for the proteins Npro (viral autoprotease, nonstructural protein), C (capsid protein), E0, E1, and E2 (glycoproteins), and nonstructural protein NS2-3 are indicated.

BVDV strains can be isolated from field cases (2, 23). To the cp CSFV isolates CSFV Alfort/M, CSFV ATCC, and identify the molecular basis for cytopathogenicity, the genomes CSFV Steiermark. All three isolates were found to contain of several noncp and cp BVDV strains were compared. These defective viruses in addition to standard CSFV. The defective studies allowed the conclusion that genomes of cp variants of variants exhibit characteristics of internal deletion-type DI par- noncp BVDV can be generated by recombination resulting in ticles. Interestingly, plaque isolation and RNA transfection extensive rearrangements (19–22, 27, 32, 33). experiments showed that the CSFV DI particles represent the The situation is quite different for the other two pestiviruses, cytopathogenic agents of these cp CSFV isolates. As described CSFV and BDV. For both species, the overwhelming number for BVDV, the generation of cp CSFV is the result of a re- of isolates are noncp, and only few publications report the combination. The CSFV isolates containing DI particles re- identification of cp variants of CSFV, without, however, pre- semble the cp BVDV isolate CP9, which is also composed of a senting details about the molecular biology of these viruses (9, cp DI genome and a noncp helper virus (33). 11, 16, 37). We report here the molecular characterization of The sequence comparison studies showed that the three 3688 MEYERS AND THIEL J. VIROL.

FIG. 5. SDS-PAGE (12% gel) analysis of immunoprecipitates after meta- bolic labeling of PK15 cells infected with the viruses indicated on top. The antisera used for precipitation are noted below the gel (NS, preimmune serum; A3, anti-A3 serum, directed against NS3). NS2-3 (p125) and NS3 (p80) are marked with arrows.

analyzed CSFV DI genomes have been generated in individual FIG. 6. Comparison of 3Ј recombination positions in different cp pestivirus polyproteins. The polyproteins are shown as bars below the names of the virus recombination reactions starting with different parental virus isolates. M, G, P, A, V, amino acids in one-letter code; NS2, nonstructural genomes. Surprisingly, all three analyzed CSFV DI genomes protein (amino-terminal region of NS2-3); Npro, viral autoprotease, first protein have exactly the same structure. The deletion starts with the encoded by the viral ORF; Npro*, Npro encoded by a duplicated gene inserted far second codon of the long ORF and extends to position 5130 downstream of the original context; ubi, host cell-derived ubiquitin sequence. (codon 1589). Thus, the CSFV DI genomes lack the genes encoding Npro, the four structural proteins, and NS2, the ami- no-terminal region of the nonstructural protein NS2-3 (p125). certain concentration of NS3 may be required for exhibition of In the case of the CP9 DI genome, the deletion starts 501 cytopathogenicity which is not achieved after infection with nucleotides downstream of that of the CSFV DI genome. Ac- noncp CSFV. Alternatively, NS3 expressed by noncp CSFV or cordingly, the deletion does not affect the Npro gene but starts BDV could be different from NS3 of cp viruses. As indicated by with the first codon of the capsid protein gene (33). However, the conservation of the 3Ј recombination position identified the 3Ј recombination position is exactly the same as that of the now for cp pestiviruses from cattle and pigs, the amino termi- CSFV DI genomes. The same genomic position was also iden- nus of NS3 might be critical for cytopathogenicity. It is not tified as 3Ј recombination point in cp BVDV genomes with known at which position NS2-3 from noncp CSFV or noncp cellular ubiquitin-coding insertions or duplication and rear- BDV is cleaved. This important question will have to be an- rangement of viral sequences (Fig. 6) (19–22, 27, 32). Sequence swered by future experiments. Other points to address concern motifs which would mark this site as a recombination hot spot the nature of the protease cleaving NS2-3 of noncp CSFV have not been found. We therefore hypothesize that the ability strains and an explanation for the fact that the analogous to exhibit a cytopathogenic phenotype is connected with re- cleavage is not detected for noncp BVDV. combination at this specific site. Outbreak of lethal MD in cattle is dependent on the pres- In the BVDV system, the cytopathogenic phenotype is cor- ence of noncp and cp BVDV within one animal. It is believed related with expression of NS3 (p80), which is colinear with the that generation of a cp variant of a persisting noncp BVDV is carboxy-terminal part of NS2-3. While noncp BVDV strains causative for an outbreak of MD (5, 6, 17). The high degree of express only NS2-3, both NS2-3 and NS3 are found in cells sequence homology found for the genomes of members of such infected with cp BVDV. Therefore, NS3 is regarded as marker virus pairs proves that the respective cp viruses represent mu- protein for cp BVDV strains (8, 25, 26). For several cp BVDV tants of the persisting noncp BVDV (21, 22, 33). The patho- isolates, expression of p80 was linked with the respective ge- genesis of MD is unique, and a similar disease is not known for nome rearrangements resulting from recombination (22, 32). swine. Experimental data concerning the pathogenicity of cp Accordingly, a continuous line of evidence connects recombi- CSFV, in particular in direct comparison with the correspond- nation, cytopathogenicity, and expression of NS3. NS3 repre- ing helper virus, is currently not available. Since generation of sents the prime candidate for a cytopathogenic agent in this cp CSFV Steiermark occurred most likely in the field during an system. The situation becomes more complex when all three outbreak of classical swine fever, the data are eagerly awaited. pestiviruses are considered. Expression of NS3 is common for all CSFV and BDV strains analyzed so far regardless whether ACKNOWLEDGMENTS these viruses are cytopathogenic (4, 34). As shown above for CSFV Alfort and Alfort/M, the ratio between NS2-3 and NS3 We thank Silke Mauritz, Petra Ulrich, and Karin Pietschmann for is dramatically different when extracts from cells infected with excellent technical assistance. We are grateful to Reinhard Ahl and noncp and cp CSFV are compared. Thus, at least for CSFV, a Alexandra Kosmidou for providing CSFV Steiermark. VOL. 69, 1995 CYTOPATHOGENIC CSFV DEFECTIVE INTERFERING PARTICLES 3689

This study was supported jointly by the Bundesminister fu¨r For- 21. Meyers, G., N. Tautz, E. J. Dubovi, and H.-J. Thiel. 1991. Viral cytopatho- schung und Technologie and Intervet International BV (project genicity correlated with integration of ubiquitin-coding sequences. Virology 0319028A) and by grant Th 298/2-2 from the Deutsche Forschungsge- 180:602–616. meinschaft. 22. Meyers, G., N. Tautz, R. Stark, J. Brownlie, E. J. Dubovi, M. S. Collett, and H.-J. Thiel. 1992. Rearrangement of viral sequences in cytopathogenic pes- tiviruses. Virology 191:368–386. REFERENCES 23. Moennig, V., and P. G. W. Plagemann. 1992. The pestiviruses. Adv. Virus 1. Aynaud, J. M. 1976. Characteristics of a new live virus vaccine against swine Res. 41:53–98. fever prepared in tissue culture at low temperature. The Thiverval strain. 24. Moormann, R. J. M., P. A. M. Warmerdam, B. V. D. Meer, W. M. M. Comm. Eur. Commun. publication no. Eur 5486, 93-96. Schaaper, G. Wensvoort, and M. M. Hulst. 1990. Molecular cloning and 2. Baker, J. C. 1987. Bovine viral diarrhea virus: a review. J. Am. Vet. Med. nucleotide sequence of hog cholera virus strain Brescia and mapping of the Assoc. 190:1449–1458. genomic region encoding envelope protein E1. Virology 177:812–815. 3. Barrett, A. D. T., and N. J. Dimmock. 1986. Defective interfering viruses and 25. Pocock, D. H., C. J. Howard, M. C. Clarke, and J. Brownlie. 1987. Variation infections of animals. Curr. Top. Microbiol. 128:55–84. in the intracellular polypeptide profiles from different isolates of bovine viral 4. Becher, P., A. D. Shannon, N. Tautz, and H.-J. Thiel. 1994. Molecular diarrhea virus. Arch. Virol. 94:43–53. characterization of border disease virus, a pestivirus from sheep. Virology 26. Purchio, A. F., R. Larson, and M. S. Collett. 1984. Characterization of bovine 198:542–551. viral diarrhea virus proteins. J. Virol. 50:666–669. 5. Bolin, S. R., A. W. McClurkin, R. C. Cutlip, and M. F. Coria. 1985. Severe 27. Qi, F., J. F. Ridpath, T. Lewis, S. R. Bolin, and E. S. Berry. 1992. Analysis clinical disease induced in cattle persistently infected with noncytopathic of the bovine viral diarrhea virus genome for possible cellular insertions. bovine viral diarrhea virus by superinfection with cytopathic bovine viral Virology 189:285–292. diarrhea virus. Am. J. Vet. Res. 46:573–576. 28. Roux, L., A. E. Simon, and J. J. Holland. 1991. Effects of defective interfer- 6. Brownlie, J., M. C. Clarke, and C. J. Howard. 1984. Experimental production ing viruses on virus replication and pathogenesis in vitro and in vivo. Adv. of fatal mucosal disease in cattle. Vet. Rec. 114:535–536. Virus Res. 40:181–211. 7. Collett, M. S., R. Larson, S. K. Belzer, and E. Retzel. 1988. Proteins encoded 29. Ru¨menapf, T., G. Meyers, R. Stark, and H.-J. Thiel. 1989. Hog cholera by bovine viral diarrhea virus: the genomic organization of a pestivirus. virus—characterization of specific antiserum and identification of cDNA Virology 165:200–208. clones. Virology 171:18–27. 8. Corapi, W. V., R. O. Donis, and E. J. Dubovi. 1988. Monoclonal antibody 30. Ru¨menapf, T., G. Unger, J. H. Strauss, and H.-J. Thiel. 1993. Processing of analyses of cytopathic and noncytopathic viruses from fatal bovine viral the envelope glycoproteins of pestiviruses. J. Virol. 67:3288–3294. diarrhea virus infections. J. Virol. 62:2823–2827. 31. Stark, R., G. Meyers, T. Ru¨menapf, and H.-J. Thiel. 1993. Processing of the 9. De Castro, M. P. 1973. In Vitro 9:8–16. pestivirus polyprotein: cleavage site between autoprotease and nucleocapsid 10. Devereux, J., P. Haeberli, and O. Smithies. 1984. A comprehensive set of protein of classical swine fever virus. J. Virol. 67:7088–7095. sequence analysis programs for the VAX. Nucleic Acids Res. 12:387–395. 32. Tautz, N., G. Meyers, and H.-J. Thiel. 1993. Processing of poly-ubiquitin in 11. Gillespie, J. H., B. E. Sheffy, and J. A. Baker. 1960. Propagation of hog the polyprotein of an RNA virus. Virology 197:74–85. cholera virus in tissue culture. Proc. Soc. Exp. Biol. Med. 105:679–681. 33. Tautz, N., H.-J. Thiel, E. J. Dubovi, and G. Meyers. 1994. Pathogenesis of 12. Harlow, E., and D. Lane. 1988. Antibodies, a laboratory manual, p. 460. Cold mucosal disease: a cytopathogenic pestivirus generated by an internal dele- Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. tion. J. Virol. 68:3289–3297. 13. Holland, J. J. 1991. Defective viral genomes, p. 151–165. In B. N. Fields and 34. Thiel, H.-J., R. Stark, E. Weiland, T. Ru¨menapf, and G. Meyers. 1991. Hog D. M. Knipe (ed.), Fundamental virology, 2nd ed. Raven Press, Ltd., New cholera virus: molecular composition of virions from a pestivirus. J. Virol. York. 65:4705–4712. 14. Huang, A. S., and D. Baltimore. 1970. Defective viral particles and viral 35. Weiland, E., R. Ahl, R. Stark, F. Weiland, and H.-J. Thiel. 1992. A second disease processes. Nature (London) 226:325–327. envelope glycoprotein mediates neutralization of a pestivirus, hog cholera 15. Kessler, S. W. 1981. Use of protein A-bearing staphylococci for the immu- virus. J. Virol. 66:3677–3682. noprecipitation and isolation of antigens from cells. Methods Enzymol. 73: 36. Weiland, E., R. Stark, B. Haas, T. Ru¨menapf, G. Meyers, and H.-J. Thiel. 442–459. 1990. Pestivirus glycoprotein which induces neutralizing antibodies forms 16. Laude, H. 1978. Virus de la peste porcine classique: isolement d’une souche part of a disulfide-linked heterodimer. J. Virol. 64:3563–3569. cytolytique a partir de cellules IB-RS2. Ann. Microbiol. (Inst. Pasteur) 129A: 37. Van Bekkum, J. G., and S. J. Barteling. 1970. Plaque production by hog 553–561. cholera virus. Arch. Gesamte Virusforsch. 32:185–200. 17. McClurkin, A. W., M. F. Coria, and S. R. Bolin. 1985. Isolation of cytopathic 38. Van der Werf, S., J. Bradley, E. Wimmer, F. W. Studier, and J. J. Dunn. and noncytopathic bovine viral diarrhea virus from the spleen of cattle 1986. Synthesis of infectious poliovirus RNA by purified T7 RNA poly- acutely and chronically affected with bovine viral diarrhea. J. Am. Vet. Med. merase. Proc. Natl. Acad. Sci. USA 83:2330–2334. Assoc. 186:568–569. 39. Wengler, G. 1991. Family Flaviviridae, p. 223–233. In R. I. B. Francki, C. M. 18. Meyers, G., T. Ru¨menapf, and H.-J. Thiel. 1989. Molecular cloning and Fauquet, D. L. Knudson, and F. Brown (ed.), Classification and nomencla- nucleotide sequence of the genome of hog cholera virus. Virology 171:555– ture of viruses. Fifth report of the International Committee on Taxonomy of 567. Viruses. Springer-Verlag, Berlin. 19. Meyers, G., T. Ru¨menapf, and H.-J. Thiel. 1989. Ubiquitin in a togavirus. 40. Wiskerchen, M. A., S. K. Belzer, and M. S. Collett. 1991. Pestivirus gene Nature (London) 341:491. expression: the first protein of the bovine viral diarrhea virus large open 20. Meyers, G., T. Ru¨menapf, and H.-J. Thiel. 1990. Insertion of ubiquitin- reading frame, p20, possesses proteolytic activity. J. Virol. 65:4508–4514. coding sequence identified in the RNA genome of a togavirus, p. 25–29. In 41. Wiskerchen, M. A., and M. S. Collett. 1991. Pestivirus gene expression: M. A. Brinton and F. X. Heinz (ed.), New aspects of positive-strand RNA protein p80 of bovine viral diarrhea virus is a proteinase involved in polypro- viruses. American Society for Microbiology, Washington D.C. tein processing. Virology 184:341–350.