The E2 Glycoprotein of Classical Swine Fever Virus Is a Virulence Determinant in Swine

Abstract

To identify genetic determinants of classical swine fever virus (CSFV) virulence and host range, chimeras of the highly pathogenic Brescia strain and the attenuated vaccine strain CS were constructed and evaluated for viral virulence in swine. Upon initial screening, only chimeras 138.8v and 337.14v, the only chimeras containing the E2 glycoprotein of CS, were attenuated in swine despite exhibiting unaltered growth characteristics in primary porcine macrophage cell cultures. Additional viral chimeras were constructed to confirm the role of E2 in virulence. Chimeric virus 319.1v, which contained only the CS E2 glycoprotein in the Brescia background, was markedly attenuated in pigs, exhibiting significantly decreased virus replication in tonsils, a transient viremia, limited generalization of infection, and decreased virus shedding. Chimeras encoding all Brescia structural proteins in a CS genetic background remained attenuated, indicating that additional mutations outside the structural region are important for CS vaccine virus attenuation. These results demonstrate that CS E2 alone is sufficient for attenuating Brescia, indicating a significant role for the CSFV E2 glycoprotein in swine virulence.

Classical swine fever (CSF) is a highly contagious and often fatal hemorrhagic disease of swine and is caused by Classical swine fever virus (CSFV), a member of the genus Pestivirus of the family Flaviviridae (5). Intermittent CSF outbreaks in Europe and other parts of the world result in significant economic losses. While infection with highly virulent CSFV strains can cause acute CSF characterized by high morbidity and mortality, isolates of moderate to low virulence induce a prolonged, chronic disease (35).

CSFV is a small enveloped virus with a single-stranded, 12.5-kb RNA genome of positive polarity. The CSFV genome contains a single open reading frame encoding an approximately 4,000-amino-acid polyprotein which through cellular and viral protease-mediated co- and posttranslational processing gives rise to the following 11 to 12 final cleavage products:NH₂-Npro-C-E^rns-E1-E2-p7-NS2-NS3-NS4A-NS4B-NS5A-NS5B-COOH (23). Protein C and the glycoproteins E^rns, E1, and E2 represent structural components of the virion (28). E1 and E2 are anchored to the envelope by their carboxy termini, with E^rns loosely associated with the envelope. E^rns and E2 are present as homodimers linked by disulfide bridges on the surfaces of CSFV virions, whereas E2 is also found dimerized with E1 (28, 37, 38). The genetic basis of CSFV virulence and host range remains poorly understood (35).

Development of infectious CSFV cDNA clones has enabled genetic approaches for defining mechanisms of viral replication and pathogenesis. Infectious clones (IC) of the attenuated CSFV C-strain and the pathogenic Alfort/Tübingen and Eystrup strains have been constructed, enabling identification of E^rns and N^pro as virulence factors in swine and of the role of different E^rns mutations in virus attenuation (17, 19). Additionally, attempts to mutate and inactivate E^rns in the attenuated C-strain resulted in a virus atypically cytopathic for cell cultures (10, 39). Thus, CSFV IC hold great promise for identifying viral proteins or protein domains functioning in viral virulence and host range and for rationally engineering marker live attenuated CSF vaccines (17, 19, 21, 30, 31).

Here we have used chimeras of the highly pathogenic Brescia strain and the attenuated vaccine strain CS to identify genetic determinants of CSFV virulence and host range. Results demonstrate that the CSFV E2 glycoprotein is a major determinant of virulence in swine.

MATERIALS AND METHODS

Viruses and cells.

Swine kidney cells (SK6) (27) free of bovine viral diarrhea virus (BVDV) were used throughout this study. SK6 cells were cultured in Dulbecco's minimal essential medium (Gibco, Grand Island, N.Y.) with 10% fetal calf serum (Atlas Biologicals, Fort Collins, Colo.). CSFV strain Brescia (obtained from the Animal and Plant Health Inspection Service, Plum Island Animal Disease Center) was propagated in SK6 cells and used for IC construction. The CS vaccine strain is derived from the LK VNIIVIM parental vaccine strain by serial passages and cloning (40). Titrations of CSFV from clinical samples were conducted by using SK6 cells in 96-well plates (Costar, Cambridge, Mass.), stained via immunoperoxidase assay 4 days postinoculation by using the CSFV monoclonal antibody 303 (4) and the Vectastain ABC kit (Vector Laboratories, Burlingame, Calif.) (24), calculated by using the method of Reed and Muench (22), and expressed as 50% tissue culture infective dose (TCID₅₀)/milliliter. As performed, test sensitivity was ≥1.8 log₁₀/TCID₅₀/ml. Primary swine macrophage cell cultures were prepared as described by Zsak et al. (41).

Construction of CSFV Brescia (pBIC), CS (pCSIC), and chimeric cDNA IC.

Total cellular RNA was Trizol extracted (Gibco) from SK6 cells infected with peripheral blood obtained from pigs infected with the CSFV strain Brescia or vaccine strain CS. cDNAs spanning the entire genome of CSFV Brescia and CS were obtained by reverse transcription (RT) as follows: the reaction mixture (1 μl of Moloney murine leukemia virus RT buffer, 4 mM deoxynucleosidetriphosphates, 10 pmol of specific reverse primer R12292, R8538, or R4567 [Table 1], 2 μl of total RNA, and 5.3 μl of water) was incubated at 70°C for 5 min, cooled at room temperature for 20 min, and further incubated for 1 h at 37°C after addition of 2 U of Moloney murine leukemia virus reverse transcriptase (Stratagene, Cedar Creek, Tex.). The entire viral genome of both strains was PCR amplified in three overlapping fragments, using specific primers as described below (Table 1 and Fig. 1A).

TABLE 1.

Oligonucleotides used for construction of pBIC, pCSIC, and chimeric p 138, p319.1, and p312.1 infectious clones^a

Primer	Sequence	Function
MCSFBrescia	5′TCGAGCCATGGATGGCCGCTACGGCCATATGCATATACGCGTGACTAGTGACGT3′	Multiple cloning
MCSRBrescia	5′CACTAGTCACGCGTATATGCATATGGCCTAGCGGCCATCCATGGC	Brescia
R4567	5′CTCCTGCTATTTCATCTA3′	RT 5′ end
Brescia 5′	5′ACTAGTACGCGTTAATACGACTCACTATAGTATACGAGGTTAGTTCA3′	PCR 5′ end
CS5′	5′GTCAGTCGACTTAATACGACTCACTATAGTATACGAGGTTAGTTCA3′	PCR 5′ end
R3668	5′ATCCATTTCTTTATAGGC3′	PCR 5′ end
R8538	5′CCCTATCCTATCATCCGT3′	RT middle
R8127	5′GTGTTCTCTTCTGCT CAC3′	PCR middle
R12292	5′GGGCCGTTAGAAATTACCTTA3′	RT 3′ end
Brescia 3′	5′ATTGCCATGGCCCGGGCCGTTAGAAATTACCTTA3′	PCR 3′ end
F8017	5′AAGCGATGGTTTGCTAGG3′	PCR 3′ end
F2379	CCTCATCTGCTTGATAAAAG	p138.8
R6636	CCTTCTCTGGGCTTGTTC	p138.8
FSacII	5′GAACAACTCGCCGCGGGTCTACAGTTAGG3′	SacII site
RSacII	5′CCTAACTGTAGACCCGCGGCGAGTTGTTC3′	SacII site
MCS+	5′CCGGCCGCGTCGACATAGATCTATGGCCGTAGCGGCC3′	Multiple cloning
MCS−	5′TCGAGGCCGCTACGGCCATAGATCTATGTCGACGCGGCCGGACGT3′	site for CS
F2008	5′GATTCTCCATGAGATGGG3′	PCR middle
CS3′	5′ATTGCTCGAGCCCGGGCCGTTAGGAAATTACCTTA3′	PCR 3′ end
F8210	5′GATGAATTAGTCAAGGAG3′	PCR 3′ end
F3237	5′ACACAACTGTCAAGGTGC3′	PCR middle
R3126	5′TCTGTAACCTGTCTCATT3′	PCR 5′ end

^aRestriction sites are shown in bold italics and underlined. 5′ to 3′, for MCSF Brescia, MluI, NsiI, SfiI, and NcoI; for Brescia, 5′, MluI; and for Brescia, 3′, overlapping NcoI and SrfI; bold letters indicate the T7 promoter. RT, reverse transcription. 5′ end, middle, and 3′ end are amplicons shown in Fig. 1.

FIG. 1.

(A) Construction of CSFV infectious cDNA clones pBIC and pCSIC, using amplicon fragments 5′, M, and 3′. Corresponding restriction sites and nucleotide numbers are depicted (pBIC/pCSIC). (B) Growth curve of CSFV Brescia and CS and viruses derived from ICs BIC and CSIC. Primary macrophage cell cultures were infected (MOI = 0.1) and harvested at the indicated time points. Samples were titrated on SK6 cells with virus detected by immunohistochemistry as described in Material and Methods. Data are means and standard errors from three independent experiments.

A 4.3-kb region comprising the 3′ end of the CSFV genome was amplified by seminested PCR, using the primer combination F8017-Bresia3′ (Brescia virus) or F8017-CS3′ (CS virus), and then reamplified by using the primer combination F8210-Bresia3′ or CS3′. The amplified DNA fragments were excised from a 1% agarose gel and cloned, using SfiI/NcoI sites (Brescia virus) or SfiI/XhoI sites (CS virus), into pACYC177GR and pACYC177AZ (see below) to obtain plasmids pBrescia 3′end or pCS 3′end, respectively.

A central 4.8-kb genomic region was amplified by using the primer combination F3237-R8127 (Brescia virus) or F2008-R8127 (CS virus). The amplified fragments were purified as described above and cloned into pCR4Blunt, using a Topo cloning kit (Invitrogen, San Diego, Calif.) to create the plasmid pTOPOBmiddle (Brescia virus) or pTOPOCSmiddle (CS virus). pTOPOBmiddle was then NsiI-SfiI digested, and the fragment was cloned into pACYC177GR to create the plasmid pBresciamiddle. Similarly, pTOPOCSmiddle was BglII-SfiI digested, and the fragment was cloned into pACYC177AZ to generate the plasmid pCSmiddle.

A 3.1-kb region comprising the 5′ end of the CSFV genome was amplified by using primers Brescia 5′ and R3126 (Brescia virus) or CS 5′ and R3668 (CS virus). Amplified products were cloned in pCR4Blunt as described above to generate plasmid vectors pTOPOB5′end and pTOPOCS5′end. pTOPOB5′end was MluI-NsiI digested and cloned into pBresciamiddle to generate the vector pBresciamiddle-5′end. pBrescia 3′end was SfiI-SacII digested and cloned into pBresciamiddle-5′end to generate a full-length plasmid, pBIC (Fig. 1A). Similarly, pTOPOCS5′end was SalI-BglII digested and cloned into pCSmiddle to generate the vector pCSmiddle-5′end. pCS3′end was SfiI-XhoI digested and cloned into pCSmiddle-5′end to generate a full-length plasmid, pCSIC (Fig. 1A).

The low-copy-number plasmids used here, pACYC177GR and pACYC177AZ, are pACYC177 derivatives (New England Biolabs, Beverly, Mass.), obtained by removing a 1.2-kb AatII-XhoI fragment from pACYC177 and replacing it with a synthetic oligonucleotide containing unique restriction sites (Table 1).

Chimeric cDNA ICs p189.9, p182.1, p138.8, p167.1, p113.1, and p337.14 were constructed by replacing, respectively, the restriction fragments ClaI-SrfI, ClaI-NaeI, NaeI-BamHI, BamHI-SfiI, SfiI-SrfI, and ClaI-SacII in pBIC with the homologous fragment from the CS strain (Fig. 2). Chimeric cDNA ICs p137.1 and p343.8 were constructed by replacing, respectively, the restriction fragments NaeI-BamHI and ClaI-SacII in pCS with the homologous fragment from pBIC (Fig. 2).

FIG. 2.

Construction of CSVF chimeric viruses. Corresponding restriction sites and nucleotide numbers are depicted.

Chimeric cDNA ICs p319.1 and p312.1 were constructed as follows. The NaeI-BamHI fragments from pBIC and pCSIC were obtained by PCR by using the primers F2379 and R6636 and cloned by using a TA Topo kit (Invitrogen) to create the plasmids pTOPOBreN/B and pTOPOCSN/B, respectively. These plasmids were used as targets for site-directed mutagenesis, using the QuikChange XL kit (Stratagene) and the specific oligonucleotide primers FSacII and RsacII, complementary to each strand of the target. Following mutagenesis, the reaction was treated with 10 U of the endonuclease DpnI to eliminate methylated target DNA, and extended products were transformed into XL10 Gold ultracompetent cells (Stratagene). DNA obtained from the transformants was purified (QIAGEN plasmid kits, Valencia, Calif.) and sequenced to verify the presence of the desired mutation. Extension resulted in creation of a SacII site and introduction of a silent mutation at position 3540 to yield the following plasmids: pTOPOBreN/BSacII and pTOPOCSN/BSacII. NaeI-SacII fragments of each plasmid containing the genomic region E2/p7-NS2-3 were reciprocally inter changed, creating the chimeric plasmids pTOPOBreCS and pTOPOCSBre (Brescia/CS and CS/Brescia, respectively). Finally, the NaeI-BamHI fragments from pTOPOBreCS and pTOPOCSBre were cloned in pBIC to create cDNA ICs p319.1 and p312.1, respectively (Fig. 2).

Rescue of CSFV BICv, CSICv, and chimeras.

Full-length genomic ICs were linearized with SrfI and in vitro transcribed by using the T7 Megascript system (Ambion, Austin, Tex.). RNA products were precipitated with LiCl and transfected into SK6 cells by electroporation at 500 V, 720 Ω, 100 W with a BTX 630 electroporator (BTX, San Diego, Calif.). Cells were plated in 12-well plates and 25-cm² flasks and incubated for 4 days at 37°C and 5% CO₂. Virus was detected by immunoperoxidase staining as described above. Stocks of rescued viruses were stored at −70°C.

DNA sequencing and analysis.

Full-length ICs, in vitro rescued viruses, and viruses recovered from infected animals were completely sequenced with CSFV-specific primers by the dideoxynucleotide chain termination method (25). Sequencing reactions were prepared with the Dye Terminator cycle sequencing kit (Applied Biosystems, Foster City, Calif.). Reaction products were sequenced on a PRISM 3700 automated DNA sequencer (Applied Biosystems). Sequence data were assembled with the Phrap software program (http://www.phrap.org), with confirmatory assemblies performed by using CAP3 (7). The final DNA consensus sequence represented on average eightfold redundancy at each base position.

Nucleotide and amino acid comparisons and alignments were done by using BLAST (bl2seq) (1) and Clustal (29). Protein characterization and secondary structure predictions (Chou-Fasman and Garnier-Osguthorpe-Robson) were done by using the MacVector (Accelrys, Inc.) and GCG (3) software packages, SAPS (2), memsat (14), and psipredict (13).

Animal infections.

Initial screening of chimeric viruses 189.9v, 182.1v, 138.8v, 167.1v, 113.1v, 337.14v, 137.1v, and 343.8v to determine a virulence phenotype was conducted by intranasal inoculation of two 10- to 12-week-old pigs with 10⁵ TCID₅₀ of each virus. The presence of clinical symptoms and temperature were observed daily. More-detailed pathogenesis studies with attenuated chimeric viruses 138.8v, 319.1v, and 312.1v were conducted with similar pigs (n = 30) randomly allocated into five groups of six animals each. Pigs were intranasally inoculated with 10⁵ TCID₅₀ of BICv, CSICv, or chimeric viruses. Clinical signs (anorexia, depression, fever, purple skin discoloration, staggering gait, diarrhea, and cough) were observed daily throughout the experiment and scored as previously described (20). One pig per group was necropsied at 6 and 8 days postinfection (dpi), at which time blood, serum, nasal swabs, and tonsil scraping samples were obtained. Tissue samples (tonsil, mandibular lymph node, and spleen) were snap-frozen in liquid nitrogen for virus titration or fixed in 10% neutral buffered formalin for histopathological studies.

Total and differential white-blood-cell counts.

Blood was collected from the anterior vena cava in tubes containing EDTA (Vacutainer). Total white-blood-cell, lymphocyte, and platelet counts were obtained with a Beckman Coulter ACT (Beckman Coulter, Fullerton, Calif.).

Immunohistochemistry.

Tissue samples were fixed in 10% neutral buffered formalin, embedded in paraffin, and sectioned. Tissue sections were allowed to adhere to Superfrost/plus slides (Fisher Scientific, Pittsburg, Pa.), heated for 20 min at 65°C, and then deparaffinized with xylene. Sections were rehydrated through a graded alcohol series and washed with phosphate-buffered saline (PBS) (pH 7.4). Immunohistochemistry was performed as described by Sur et al. (26). Briefly, 4-μm sections were first treated with 3% hydrogen peroxide in PBS for 20 min, followed by washes in PBS and digestion with 0.05% Protease XIV (Sigma Chemical Co., St. Louis, Mo.) for 2 min at 37°C. After several washes with PBS, sections were incubated in blocking solution (5% normal goat serum in PBS) for 30 min at room temperature and then incubated for 2 h at 4°C with CSFV-specific monoclonal antibody directed against the E2 glycoprotein (4) diluted 1:500 in PBS. After PBS washes, slides were incubated with alkaline phosphatase-conjugated, goat antimouse antibody for 20 min at room temperature.

Nucleotide sequence accession numbers.

Brescia and CS genomic sequences were deposited in GenBank under accession numbers AY578687 and AY578688, respectively.

RESULTS

Construction of CSFV Brescia (pBIC) and CS (pCSIC) infectious clones.

Infectious RNA was in vitro transcribed from a full-length copy of CSFV strain Brescia or CS (Fig. 1A). cDNAs amplified by RT-PCR (see Materials and Methods) using specific primers (Table 1) were cloned into the low-copy-number plasmids pACYC177GR and pACYC177AZ to obtain pBIC and pCSIC, full-length DNA copies of the Brescia and CS viral genomes, respectively (Fig. 1A). RNA transcripts were obtained from these plasmids and used to transfect SK6 cells. Virus was rescued from transfected cells at day 4 posttransfection. Nucleotide sequences of the rescued virus BICv/CSICv genomes were identical to those of parental viruses Brescia and CS, respectively.

In vitro and in vivo analysis of CSFV BICv and CSICv.

Growth kinetics of BICv and CSICv were compared with their parental viruses, Brescia and CS, in a multistep growth curve. Primary porcine macrophage cell cultures were infected at a multiplicity of infection (MOI) of 0.1 TCID₅₀ per cell. Virus was adsorbed for 1 h (time zero), and samples were collected at times postinfection through 72 h. Replication kinetics and virus yields between the parental Brescia and its IC-derived virus, BICv, were indistinguishable (Fig. 1B). However, titers of CSICv were approximately 10-fold lower than those of parental CS virus (Fig. 1B). To assess pig virulence phenotypes, three pigs were inoculated intranasally with 10⁵ TCID₅₀ of parental viruses (Brescia and CS) or IC-derived viruses, BICv and CSICv. No significant differences in time to onset of clinical disease, mortality rate (which was 100%), time to death, or viremia titers were observed for Brescia- and BICv-infected animals (data not shown). Animals inoculated with CS or CSICv were free of CSFV-related clinical signs during the experimental period (data not shown). These data indicate that the IC-generated viruses, BICv and CSICv, have parental phenotypes in vitro and in vivo.

CSFV E2 is a virulence determinant.

To identify CSFV virulence and host range determinants, we constructed chimeric viruses in which genomic regions of the pathogenic Brescia strain were replaced with homologous regions from the CS vaccine strain (Fig. 2). Chimeric viruses 189.9v, 182.1v, 138.8v, 167.1v, 113.1v, and 337.14v were constructed as described in Material and Methods and found to contain nucleotide sequences identical to those of their corresponding cDNA IC (data not shown). Chimeras were screened for virulence in pigs by intranasal inoculation of 10⁵ TCID₅₀/animal. Only two chimeras, 138.8v and 337.14v, exhibited the attenuated phenotype of CSICv, while the other four chimeras exhibited the virulence phenotype of parental BICv (Table 2). The CS region common to both 138.8v and 337.14v encoded the E2 gene, suggesting that a significant virulence determinant absent in CS E2 was responsible for attenuation of these chimeras.

TABLE 2.

Swine survival, clinical scores, and fever response following infection with BICv, CSICv, and chimeric viruses

Virus	No. of survivors/ total no.	Mean time to death (days ± SD)	Fever
			No. of days to onset (± SD)	Duration, no. of days (± SD)
BICv	0/10	9.1 (3.1)	4.4 (0.8)	4.6 (0.8)
CSICv	4/4
189.9v	0/4	8.75 (2.6)	3 (0)	5.2 (3.2)
182.1v	0/2	11.5 (6.4)	4 (0.7)	5.5 (4.9)
138.8v	2/2		8.5 (0.5)	1 (0)
167.1v	0/2	12 (4.65)	6.5 (0.7)	9 (2.8)
113.1v	0/2	8 (0)	4 (0)	4 (0)
337.14	2/2
137.1v	2/2
343.8v	2/2		8a	1

^aOnly one animal showed fever for 1 day.

To confirm the role of E2 in swine virulence and further map the region responsible for 138.8v attenuation, two additional chimeric viruses, 319.1v and 312.1v, were constructed to contain CS regions encoding E2 only or p7 and the amino-terminal 894-amino-acid residues of NS2-3, respectively (Fig. 2). Chimeric viruses were rescued, and their nucleotide sequences were found to be identical to those of their corresponding cDNA IC (data not shown).

Growth characteristics of 138.8v, 319.1v, and 312.1v relative to those of BICv and CSICv were examined in a multistep growth curve by infecting primary swine macrophage cell cultures (MOI = 0.1) and determining titers at various times postinfection (Fig. 3A). While growth characteristics of BICv, 138.8v, 312.1v, 319.1v, and 337.14v were similar, CSICv growth was reduced by approximately 1,000-fold. Notably, the plaque size of 138.8v and 319.1v was reduced by approximately 50 to 60% relative to that of BICv and 312.1v. The plaque size of CSICv was reduced by 90% relative to that of BICv (Fig. 3B).

FIG. 3.

(A) Growth characteristics of CSFV BICv, CSICv, and chimeric viruses 138.8v, 312.1v, and 319.1v in swine macrophage cell cultures. Primary swine macrophage cell cultures were infected (MOI = 0.1) with CSFV BICv, CSICv, 138.8v, 319.1v, 312.1v, 337.14v, 343.8v, or 137.1v. At times postinfection, samples were collected and titrated for virus yield. Data are means and standard errors from three independent experiments. (B) Plaque formation of CSFV BICv, CSICv, and chimeric viruses on SK6 cell cultures. Cell cultures were infected with 50 to 100 TCID₅₀, overlaid with 0.5% agarose, and incubated at 37°C for 3 days. Plates were fixed with 50% (vol/vol) ethanol-acetone and stained by immunohistochemistry as described in Materials and Methods.

Chimeric viruses 319.1v and 312.1v were evaluated for swine virulence relative to BICv, CSICv, and 138.8v. Parental viruses exhibited characteristic virulence phenotypes (Table 3). Animals infected with chimera 138.8v or 319.1v survived infection and, with the exception of a transient fever (1 to 2 days), remained clinically normal. All animals infected with 312.1v presented with clinical signs of CSF starting at 3 to 7 dpi, with kinetics and severity of disease indistinguishable from those observed for animals infected with BICv (Table 3). White-blood-cell and platelet numbers dropped drastically by 6 dpi in BICv- or 312.1v-infected animals and remained low until death, while a transient and much less dramatic effect was observed for 138.8v- or 319.1v-infected animals (Fig. 4). CSICv infection induced no significant hematological changes.

TABLE 3.

Swine survival, clinical scores, and fever response following infection with BICv, CSICv, or chimeric viruses

Virus	No. of survivors/total no.	Mean time to death days (± SD)	Clinical scores, range (5-10 dpi)	Fever
				No. of days to onset (± SD)	Duration, no. of days (± SD)	Max temp, °F (± SD)
BICv	0/6	12.2 (3.1)	11-20	7 (0)	5.2 (0.5)	106.5 (1)
CSICv	4/4		1-1			103.4 (0.3)
138.8v	6/6		1-1	9.5 (1.1)	1 (0.5)	103.4 (0.2)
319.1v	6/6		1-1.5	9.3 (1.2)	1.25 (1.7)	103.7 (0.7)
312.1v	0/6	10.6 (2.1)	10-20	5.0 (0.0)	8 (0.7)	106.8 (0.5)

FIG. 4.

Cell counts of peripheral white blood cells (A) and platelets (B) for pigs infected with BICv, CSICv, or 319.1v, 138.8v, or 312.1v. Counts were determined as described in Material and Methods. Each point represents the mean and standard errors for four or more animals.

Viremia in 138.8v- or 319.1v-infected animals was transient (8 to 10 dpi) and significantly reduced by 10⁵ to 10⁶ log₁₀ from that with BICv or 312.1v infection but was higher than that observed for CSICv-infected animals where no viremia was detected (Fig. 5C). Similar titration patterns were obtained for nasal swab and tonsil scraping samples (Fig. 5A and B).

FIG. 5.

Titers of virus from (A) nasal swabs, (B) tonsil scrapings, or (C) blood from pigs infected with BICv, CSICv, or chimeric viruses 319.1v, 138.8v, or 312.1v. Each point represents the mean and standard errors for four or more animals.

Nucleotide sequences of viruses recovered from infected animal tonsil scraping samples (6 dpi) were identical to those of 138.8v, 319.1v, and 312.1v stock viruses used for inoculation (data not shown).

For a more-detailed comparison of the pathogenesis induced by 138.8v, 319.1v, and 312.1v, randomly selected animals were euthanized at 6 and 8 dpi (one animal/time point/group), and tissue samples (tonsil, submandibular lymph node, and spleen) were collected for virus titration and histopathological and immunohistochemical analysis. Titers of virus in tissue are shown in Table 4. Replication of CSICv, 138.8v, and 319.1v was transient in tonsils, with tissue titers of virus significantly reduced (approximately 10 to 10⁶ log₁₀ depending on virus and time postinfection) relative to those of BICv and 312.1v. CSICv and 138.8v were not detected in regional draining mandibular lymph node or spleen tissues, and 319.1v was detected in these tissues only transiently (6 dpi) and at significantly lower titers (10² log₁₀) than those observed for BICv and 312.1v.

TABLE 4.

Titers of virus in tissues after infection with BICv, CSICv, or chimeric viruses

Virus	No. of dpi	Log10/TCID50/g in:
		Tonsils	Mandibular lymph node	Spleen
BICv	6	4.97	5.13	5.97
	8	5.47	5.97	5.80
CSICv	6	Nega	Neg	Neg
	8	2.13	Neg	Neg
319.1v	6	3.63	3.47	3.63
	8	3.13	Neg	Neg
138.8v	6	2.80	Neg	Neg
	8	Neg	Neg	Neg
312.1v	6	5.80	5.97	5.97
	8	6.30	7.30	7.13

^aNeg, less than 1.8 TCID₅₀ (log₁₀).

Histological lesions in pigs inoculated with BICv were already present at 6 dpi (Fig. 6). This included tonsillar, splenic, and submandibular lymphoid necrosis and cell depletion along with a moderate to marked (6 and 8 dpi, respectively) histiocytic hyperplasia and reticuloendothelial cell proliferation. Additionally, the spleen was characterized by moderate to severe lymphoid follicular depletion, necrosis, and a mild to moderate histiocytosis. Animals infected with viruses CSICv and 319.1v showed histopathological changes in tonsils and the regional submandibular lymph node similar to those described for the BICv-infected pigs, but the degree of severity was markedly reduced: there was minimal to mild lymphoid necrosis and depletion and a mild histiocytosis. Notably, the pathology already had started to resolve, with evidence of lymphoid hyperplasia, at 6 dpi. In addition, spleens of animals infected with CSICv or 319.1v showed near-negligible changes, with minimal lympholysis and depletion (Fig. 6).

FIG. 6.

Detection of CSFV antigens by immunohistochemistry at 6 dpi in lymphoid organs of pigs infected with BICv or 319.1v. Note (1a) the abundance of CSFV antigen in tonsillar crypt epithelium (Δ = crypt), lymphoid follicles (*), and parafollicular regions (arrow) of the BICv-infected animal. Both in the follicular (1b) and parafollicular (1c) regions, immunoreactive cells are morphologically characterized as dendritic cells. In the 319.1v-infected animal (2a), there is minimal CSFV antigen in tonsillar crypt epithelium (Δ = crypt) and rare CSFV antigen in lymphoid follicles (* and 2b) and parafollicular (2c) region. In spleen of BICv-infected animal (1d), note lymphoid depletion with poorly defined periarteriolar lymphoid sheaths (large arrows) and abundant CSFV antigen in periarteriolar regions (small arrow) and follicles with associated lymphoid depletion (1e). In the 319.1v-infected animal, there are prominent periarteriolar lymphoid sheaths (2d and e: arrows) with follicles, with minimal to no expression of CSFV antigen in periarteriolar regions and follicles. Lymphoid follicles are well populated with lymphoid proliferation (2e). Bar in (1a), (2a), (1d), and (2d) = 160 μm; bar in (1b), (1c), (2b), (2c), (1e), and (2e) = 30 μm.

The presence of CSFV antigen in tissues of infected animals was consistent with the degree and distribution of lesions and titers of virus in tissues. In BICv-infected pigs, there was antigen in superficial and cryptic tonsillar epithelial cells and tonsillar endothelium by 6 dpi. At 6 and 8 dpi, antigen was observed in tonsillar macrophages and lymphoid follicular and parafollicular dendritic cells with moderate to marked immunoreactivity (Fig. 6). Similar immunoreactivity was observed in the spleen, where macrophages and endothelium were affected. Conversely, animals infected with CSICv or 319.1v showed CSFV antigen concentrated in the tonsil, with a significant relative reduction in immunoreactivity in the regional submandibular lymph node and spleen compared to the BICv-infected animals.

These data indicate that glycoprotein E2 markedly affects pig virulence. 319.1v infection is characterized by mild and transient clinical disease, decreased viral replication in tonsils, limited generalization of infection, and a significant reduction in virus shedding.

Brescia structural protein E2 is not sufficient to restore CS virus virulence.

To test the ability of Brescia virus E2 to restore a virulent phenotype to the attenuated CS vaccine strain, two chimeric viruses which contained the genomic region encoding all Brescia virus structural proteins (343.8v) or the Brescia E2/p7 protein and the amino-terminal 894 amino acid residues of the NS2-3 protein (137.1v) inserted into the CS genetic background were constructed and evaluated for virulence in swine (Fig. 2; Table 2). Both chimeric viruses, although demonstrating normal growth in swine macrophages, demonstrated an attenuated phenotype in swine, indicating that these Brescia genomic regions alone are not sufficient to restore a virulent phenotype to CS virus.

DISCUSSION

Here we have shown that the CSFV E2 glycoprotein markedly affects viral virulence in swine. The replacement of the E2 gene in the highly pathogenic strain Brescia with the E2 gene from the vaccine strain CS resulted in significant in vivo attenuation. Unlike the acute fatal disease induced by Brescia and its IC BICv, 319.1v infection was subclinical, characterized by decreased viral replication in tonsils, limited generalization of infection, and reduced virus shedding. The 319.1v attenuation phenotype was independent of its ability to replicate in primary macrophage cell cultures. It is notable that no other CS genome region was able to attenuate the virulent Brescia phenotype, indicating the significance of CS E2 in attenuation.

The genetic basis and the molecular mechanisms underlying pestivirus virulence are largely unknown. In the case of CSFV, there are two reports implicating a specific viral protein or genomic region in virulence. A single- or double-codon mutation abrogating E^rns RNase activity of the CSFV Alfort/Tübingen strain attenuated the virus for pigs (19). Notably, a similar result was obtained by mutating the same E^rns domain of BVDV (18). More recently, N^pro has also been shown to be involved in CSFV virulence, since deletion of N^pro in the virulent Eystrup strain abrogated its virulence for swine (17). Here a third CSFV virulence determinant associated with the E2 glycoprotein has been identified. Although E2 is a highly immunogenic CSFV glycoprotein, inducing protective immune responses in swine (8, 15, 36, 37), its role in virus virulence has not been previously described.

Specific features responsible for the functional difference of CS E2 are unknown. Comparative sequence analysis of Brescia and CS E2 regions revealed no insertions or deletions and 6% nucleotide divergence over 1,119 nucleotides, including 40 synonymous and 22 nonsynonymous substitutions. The 22 predicted amino acid substitutions in CS relative to Brescia are distributed along E2, with only two adjacent positions (238 and 239) affected. Substitutions are generally conservative or neutral, as only six substitutions (T56R, T197M, P200L, S238L, R290S, and L370P) are relatively nonconservative (≤0.3 in a normalized PAM250 substitution matrix [3]). Highly conservative I137V and I342V substitutions were present in a previously predicted fusion peptide region (6) and near the strongly predicted carboxyl-terminal transmembrane domain (position 342 to 366) of CS E2, respectively. Few substitutions were found in regions consistently altered in predicted secondary structure, but they included K45R, a residue previously affecting domain C antigenicity when experimentally mutated in Brescia (34), G99E, and notably T197M and P200L, which locally affect hydrophobicity and predicted B-sheet formation in CS E2. Also notable is L370P, a mutation which resulted in loss of a predicted carboxyl-terminal alpha-helical structure in CS E2 and in the reversed transmembrane topology predicted for CS E2. How these specific changes contribute to the attenuated phenotype observed for CSICv E2 remains to be determined.

Attenuation of 319.1v in pigs could conceivably involve some aspect of virus attachment and/or efficiency of entry into critical target cells in vivo. Given that macrophages are targeted in vivo, it is notable that 319.1v exhibited normal growth in macrophage cell cultures in vitro. E^rns, E1, and E2 are structural glycoproteins in the CSFV virion envelope, and available evidence indicates that the E2 gene is an essential gene (32). Anchored to the envelope, E2 appears as both homo- and heterodimers linked by disulfide bridges (28, 37, 38) and, along with E^rns, is known to be critical for virus reception (9). Notably, pestiviruses containing chimeric E2 proteins have an altered host range. Partial replacement of the amino terminus of CSFV E2 with the homologous sequence from BVDV resulted in a virus that, although differing from BVDV in its capacity to infect fetal bovine epithelial cells, resembled BVDV with a 10-fold decrease in its ability to replicate in SK6 cells (31). Additionally, a chimeric BVDV containing the complete E2 gene of border disease virus (BDV), similarly to BDV, is unable to replicate in MDBK cells, a cell line permissive for BVDV (16).

Consistent with a possible direct or indirect effect for altered E2 on virus attachment and/or spread, 319.1v exhibited a small-plaque phenotype in SK6 cells compared with parental BICv. Plaque size was reduced approximately 50%. Interestingly, CSICv also has a reduced plaque size compared with BICv, although in this case reduction was 90%. Thus, additional genetic determinants in the CSICv genome are likely involved in the small-plaque phenotype. A correlation between plaque size in vitro and pathogenesis in vivo has yet to be established. Hulst et al. (11, 12) serially passaged CSFV Brescia in SK6 cells to generate heparan sulfate binding-dependent small-plaque variants that contained mutations in the E^rns protein; however, these varients had an unaltered virulence phenotype in pigs (11, 12).

Despite the data here implicating E2 in swine virulence, the presence of additional CSFV virulence determinants outside the E2 region is indicated. Notably, when inserted into the CS genetic background, all Brescia structural proteins, including E2, were unable to restore a virulent phenotype to CS. This result suggests that CS contains additional mutations affecting virulence in untranslated and nonstructural protein-encoding regions. Given the history of CS virus, a vaccine virus safely used for more than 30 years (40), it is not surprising that CS may contain additional attenuating mutations which affect multiple proteins. In fact, CS demonstrated 531 nucleotide differences with Brescia across the genome, resulting in 105 amino acid substitutions affecting both 5′ and 3′ untranslated regions and all viral proteins except NS4A (data not shown). Additional support for other contributing virulence determinants is suggested by the somewhat more attenuated phenotype for 138.8v (E2/p7/NS2-3 substitution) than for 319.1v (E2 substitution only) observed as reduced blood and tissue titers and the absence of generalized infection (Table 4; Fig. 5). Most significantly, any additional CS mutations affecting virulence were unable to attenuate BICv in the context of the chimeras generated in this study, further highlighting the significance of E2 in swine virulence.

In summary, a novel CSFV virulence determinant associated with the E2 glycoprotein has been identified. An improved understanding of the genetic basis of CSFV virulence and host range will permit rational design of live attenuated CSF vaccines of enhanced safety, efficacy, and utility.

ADDENDUM

Since completion of the work presented here, additional studies have identified the CSFV E2 protein as contributing to the virulence of strain Brescia; however, these E2 mutants also required mutations in the E^rns protein to yield an attenuated phenotype (33).