Comparison of Two Aquatic Alphaviruses, Salmon Pancreas Disease Virus and Sleeping Disease Virus, by Using Genome Sequence Analysis, Monoclonal Reactivity, and Cross-Infection

Abstract

Cell culture isolates of salmon pancreas disease virus (SPDV) of farmed Atlantic salmon and sleeping disease virus (SDV) of rainbow trout were compared. Excluding the poly(A) tracts, the genomic nucleotide sequences of SPDV and SDV RNAs include 11,919 and 11,900 nucleotides, respectively. Phylogenetic analysis places SPDV and SDV between the New World viruses of Venezuelan equine encephalitis virus and Eastern equine encephalitis virus and the Old World viruses of Aura virus and Sindbis virus. When compared to each other, SPDV and SDV show 91.1% nucleotide sequence identity over their complete genomes, with 95 and 93.6% amino acid identities over their nonstructural and structural proteins, respectively. Notable differences between the two viruses include a 24-nucleotide insertion in the C terminus of nsP3 protein of SPDV and amino acid sequence variation at the C termini of the capsid and E1 proteins. Experimental infections of Atlantic salmon and rainbow trout with SPDV and SDV confirmed that the disease lesions induced by SPDV and SDV were similar in nature. Although infections with SPDV and SDV produced similar levels of histopathology in rainbow trout, SDV induced significantly less severe lesions in salmon than did SPDV. Virus neutralization tests performed with sera from experimentally infected salmon indicated that SPDV and SDV belonged to the same serotype; however, antigenic variation was detected among SDV and geographically different SPDV isolates by using monoclonal antibodies. Although SPDV and SDV exhibit minor biological differences, we conclude on the basis of the close genetic similarity that SPDV and SDV are closely related isolates of the same virus species for which the name Salmonid alphavirus is proposed.

Pancreas disease of farmed Atlantic salmon, which has been reported to occur throughout Western Europe and in North America, is associated with runting and production losses (17, 29, 39). Sleeping disease syndrome, so called because freshwater-reared rainbow trout affected with the disease can be found lying on their sides at the bottom of the tank, has been reported in France (4). Viral etiologies were suspected in each case. Demonstrations that both diseases are associated with similar histological lesions of the pancreas, heart, and muscle of their respective species suggested that they were caused by related viruses, and this view was confirmed when immune cross-protection was demonstrated in rainbow trout (5). Viruses known as salmon pancreas disease virus (SPDV) and sleeping disease virus (SDV) were subsequently isolated in Chinook salmon embryo (CHSE-214) cells (31) and rainbow trout gonad (RTG-2) cells (6), respectively, and were later shown to be the causal agents in disease reproduction experiments (6, 27).

The ability to propagate SPDV and SDV in cell culture facilitated their biochemical and molecular characterizations (53-55). In separate investigations, nucleotide sequence analyses of cDNA clones encompassing parts of their viral RNA genomes indicated that both SPDV and SDV exhibited nucleotide and amino acid sequence homologies to members of the genus Alphavirus of the family Togaviridae. Alphaviruses, as exemplified by Sindbis virus (SINV) or Semliki Forest virus, are arthropod-borne and contain a positive-sense, single-stranded genome of ca. 12 kb. The four nonstructural proteins (nsP1 to nsP4) involved in virus replication are encoded by the 5′-terminal two-thirds of the genome, whereas the structural proteins are encoded by the 3′-terminal one-third of the genome (48). Unpublished and more recently published comparisons of previously reported sequences of SPDV and SDV genomic regions encoding the structural proteins indicated that these viruses share high levels of homology (40).

The primary aim of this study was to more fully investigate the relationship between SPDV and SDV. In the first part of the study, their complete genomic nucleotide sequences were determined and compared. In the second part, isolates of SPDV and SDV collected from geographically different locations within Europe were compared in terms of their reactivity with monoclonal antibodies (MAbs) raised to SPDV and SDV (51). Lastly, contemporaneous experimental infections were used to compare the pathogenic potential of cell culture isolates of SPDV and SDV in both Atlantic salmon and rainbow trout. A secondary aim of this study was to determine how SPDV and SDV relate to other alphaviruses. This was achieved by comparing their complete genomic sequences with those of other previously sequenced alphaviruses.

MATERIALS AND METHODS

Viruses, cells, and virus growth.

The F93-125, F97-12, and N2P6 SPDVs were isolated in Chinook salmon embryo (CHSE-214) cells from disease-affected salmon, farmed in Ireland, Scotland, and Norway, respectively (8, 31, 43). The F93-125 isolate is identified as the reference isolate of SPDV on the grounds that it was the first reported isolate (31), that it has been the subject of biochemical characterization (54), and that its genomic sequence is now reported in this study. The N3P12 virus was similarly isolated in CHSE-214 cells from rainbow trout farmed in Norway that were showing clinical signs and histopathology characteristic of salmon pancreas disease. On this basis, the N3P12 virus was regarded by Christie et al. (8) as an SPDV isolate. These four isolates of SPDV were propagated in CHSE-214 cells at 15°C as described previously (31). The P42P isolate of SPDV was isolated in CHSE-214 cells from Atlantic salmon farmed in Scotland (26). This virus was subsequently passaged in CHSE-214 cells grown at either 15 or 10°C. The S49P isolate of SDV was initially isolated in CHSE-214 cells from trout farmed in France with sleeping disease (6). This virus was subsequently passaged in RTG-2 cells or CHSE-214 cells, which were maintained at 10°C.

SPDV was reisolated from experimentally infected fish by subjecting tissue homogenates to two passages in CHSE-214 cells incubated at 15°C for 7 days. Reisolation of SDV from experimentally infected fish was carried out by using a similar procedure except that incubation of infected cells was performed at 10°C for 14 days.

Experimental infections.

Groups of mixed-sex Atlantic salmon (Salmo salar L.) (20 ± 5 g) and rainbow trout (Oncorhyncus mykiss) (15 ± 5 g) were experimentally infected by intraperitoneal inoculation with the F93-125 isolate of SPDV or isolate S49P of SDV by using the experimental design shown in Table 1. The experimental population was acclimatized 1 week before the experiment was initiated. The fish populations tested negative for infectious pancreatic necrosis virus and for antibodies to infectious pancreatic necrosis virus and SPDV prior to injection. Water (0% salinity) in the experimental tanks was maintained at 15°C with a pH of 6.6 to 6.9 and an oxygen content of 10.5 to 11.5 mg/liter. Ten fish were removed from each test group for sampling at 1, 2, 3, and 6 weeks postinfection (p.i.), when the experiment ended. Blood samples for antibody analysis were taken preinfection and 6 weeks p.i. Kidney tissue samples for virus reisolation were taken 1 week p.i., and samples of pancreas, heart, and skeletal muscle tissue for histopathological examination were collected 2, 3, and 6 weeks p.i.

TABLE 1.

Experimental design^a

Test group	Tank no.	Fish species	Challenge inoculum	Challenge dose (TCID50)/fishc	No. of fish
1	1	Trout	SPDV	6.3 × 107	40
2	1	Salmon	SPDV	6.3 × 107	40
3	2	Trout	SDV	2 × 107	40
4	2	Salmon	SDV	2 × 107	40
5	3	Trout	EMEMb	-	40
6	3	Salmon	EMEM	-	40

^aThe table outlines the six test groups used in this study.

^bEMEM, Eagle minimal essential medium.

^cTCID₅₀, 50% tissue culture infective dose. -, No virus.

Histopathology.

Samples of pyloric ceca with pancreas, heart, and lateral muscle (red and white) were taken for histopathological examinations. The samples were fixed in 10% formaldehyde in buffered saline (pH 7.0) and examined by microscopy for histopathological lesions. A scoring system was used to evaluate the severity of virus-induced lesions (0 = none, 1 = minimal, 2 = mild, 3 = moderate, and 4 = severe). Scores of 2 or higher were considered to be consistent with virus infection. Statistical analysis of the results was performed by nonparametric statistics (Kruskal-Wallis one-way analysis of variance).

Virus neutralization assay.

Blood samples, collected at 6 weeks p.i. from each experimental group, were tested for the presence of antibodies against SPDV by using a virus neutralization test involving a fixed virus concentration and various serum dilutions (31). Briefly, an equal volume of SPDV (100 50% tissue culture infective doses per 0.1 ml) was added to 0.1 ml of twofold dilutions (starting dilution of 1:10; a dilution of 1:2 was toxic after inoculation into CHSE-214 cells) of antiserum and incubated at 15°C for 1 h. The mixtures were then inoculated into CHSE-214 cells in 24-well plates at 0.1 ml per well and allowed to absorb for 1 h at 15°C; then, 1 ml of minimal essential medium was added, followed by incubation at 15°C for 14 days. A titer of ≥20 was taken as a positive result. A similar method was used to detect virus-neutralizing antibodies against SDV except that cultures were incubated at 10°C for 14 days before the test was read.

Mouse MAb production.

The panel of seven SPDV-derived MAbs was produced as previously described (51, 54). The three SDV-derived MAbs were produced by intraperitoneal injection of concentrated supernatants from SDV (S49P isolate)-infected RTG-2 cells into BALB/c mice. Immunization was carried out every 2 weeks for a 6-week period; then, 3 days after the last boost, immunized mouse spleen lymphocytes were fused with SP20 myeloma cells, and hybridomas were selected in hypoxanthine-aminopterine-thymidine medium. Positive SDV-derived secreting hybridomas were selected by using an indirect immunofluorescence (IIF) assay on SDV-infected and mock-infected RTG-2 cells.

IIF and IAP assays.

IIF assay was performed by using the SPDV-derived mouse MAbs as previously described (51). Virus-infected and uninfected CHSE-214 cells that had been grown on multiwell slides and fixed with acetone were used with SPDV isolates F93-125, F97-12, N2P6, and N3P12. To test the reactivities of the P42P SPDV and the S49P SDV isolates with the panel of MAbs, virus-infected and uninfected CHSE-214 cells that had been grown in 24-well Costar plates and fixed with ethanol and acetone were used in an indirect alkaline phosphatase (IAP) assay. In this assay, the presence of reactive MAb was detected by incubating with alkaline phosphatase conjugated goat anti-mouse immunoglobulin, followed by incubation with the enzyme substrate. In both the IIF and the IAP assays, the optimal test dilutions were determined for each SPDV-derived MAb by using cultures infected with the F93-125 SPDV isolate and for each SDV-derived MAb by using cultures infected with the S49P SDV isolate.

cDNA cloning, including 5′ RACE for SPDV and SDV.

SPDV and SDV cDNA libraries were prepared with RNAs that had been extracted from purified SPDV (F93-125 isolate) and SDV (S49P) as previously described (55, 53). Clones mapping to specific regions along the length of the SPDV and SDV genomes from the 5′ to the 3′ termini were identified on the basis of nucleotide homologies shared with previously sequenced alphaviruses. Sequence information from these clones was used to design primers, which were then used in reverse transcription-PCR assays to amplify overlapping fragments representing the complete SPDV and SDV genomes. For both SPDV and SDV, 5′ RACE (rapid amplification of cDNA ends) analysis was carried out by using the Roche 5′/3′ Race Kit (Roche Diagnostics, Ltd.) according to the manufacturer's instructions.

Sequencing.

Cycle sequencing was performed by using the ABI PRISM dye terminator ready reaction kit on purified plasmid DNA according to the manufacturer's protocol (Perkin-Elmer Cetus). Electrophoresis was carried out by an ABI 310 analyzer (Perkin-Elmer Cetus). Electropherograms were interpreted by using Sequence Navigator software (Perkin-Elmer Cetus). Sequencing was carried out by a primer walking strategy, with constructs being sequenced on both strands more than twice. Sequences were analyzed by using the MacDNASIS sequence analysis software (Hitachi), and similarity searches were performed with the BLAST program (1) by using the nonredundant nucleic acid sequence database at the National Center for Biotechnology Information, Bethesda, Md. Sequence alignments were performed by using the LALIGN (30) and CLUSTALW (1) programs. Phylogenetic analysis was carried out by using the Phylip package (Department of Genetics, University of Washington). Sequences were aligned by using CLUSTAL, and data sets were statistically analyzed by using the ESEQBOOT algorithm, EPROTDIST, and ENEIGHBOR. Phylogenetic trees were calculated by using the ECONSENSE algorithm. The maximum-likelihood branch lengths of the unrooted consensus tree were computed by using the Dayhoff model for substitution and rate heterogeneity as implemented in PUZZLE (49) and drawn by using Treeview software (37).

Nucleotide sequence accession numbers.

The complete nucleotide sequences of both SPDV (isolate F93-125) and SDV (S49P) are available from GenBank under accession numbers AJ316244 (SPDV) and AJ316246 (SDV).

RESULTS

Nucleotide sequences of SPDV and SDV RNA.

The genome organizations of the two viruses are similar to those of other alphaviruses in that a 5′ nontranslated region 27 nucleotides (nt) long is followed by a continuous open reading frame (ORF) encoding the nonstructural proteins. This ORF is 7,803 nt in SPDV and 7,779 nt in SDV. In both SPDV and SDV, a nontranslated region of 38 nt occurs after this long ORF, before the start of the second ORF encoding the structural proteins. The second ORF is 3,960 nt in SPDV and 3,966 nt in SDV. The 3′ nontranslated region is 91 nt for SPDV and 90 nt for SDV. Overall, therefore, excluding the poly(A) tracts at the 3′ termini, the total lengths of the SPDV and SDV genomes are 11,919 and 11,900 nt, respectively.

Structural proteins of SPDV and SDV.

SPDV and SDV show 93.6% amino acid identity over their structural polyprotein regions, while the identities shared by both SPDV and SDV with other alphaviruses were in the range of 31.3 to 33.7% (Table 2). When the individual structural proteins of SPDV and SDV were compared, the amino acid identities were 88.3% for the capsid protein, 94.4% for E3, 94.3% for E2, 92.6% for the 6K component, and 96.1% for E1. The amino acid differences generally occurred singly; however, two regions with relatively high numbers of amino acid differences were seen at the C termini of the capsid and E1 proteins. The capsid protein genes of SPDV and SDV exhibited 90.6% nucleotide identity; however, comparison of corresponding regions of the capsid protein genes of SPDV (8,477 to 8,577 nt) and SDV (8,554 to 8,557 nt) showed differences in 20 nt positions, leading to 22 amino acid changes in the C terminus region. The E1 glycoprotein genes of SPDV and SDV showed 94.1% nucleotide sequence identity, with amino acid differences in the C termini being attributed to three nucleotide deletions in SPDV when SPDV (11,742 to 11,769 nt) and SDV (11,721 to 11,751 nt) were compared.

TABLE 2.

Pairwise comparison of SPDV and SDV protein-coding regions

Virus protein	Virus protein (% amino acid identity)a
	SPDV	SDV	SINV	SFV	RRV	BFV	ONNV	AURAV	VEEV	EEEV	WEEV
SPDV		95.0	43.1	42.0	42.6	43.3	42.9	41.8	42.9	41.7	41.9
SDV	93.6		42.9	42.1	42.8	43.6	43.3	41.9	42.6	41.8	42.2
SINV	32.5	32.6		59.7	58.7	57.0	58.5	68.5	56.4	55.9	57.2
SFV	33.3	32.5	46.3		73.0	62.8	68.8	59.1	58.4	59.3	59.4
RRV	32.2	31.8	46.8	73.4		62.0	68.1	58.0	58.8	58.9	58.9
BFV	33.6	33.5	45.6	55.6	53.9		60.5	57.1	57.2	57.6	57.2
ONNV	32.5	31.9	43.4	60.6	59.3	51.8		57.8	58.4	58.7	58.5
AURAV	33.5	33.1	61.0	45.0	45.0	44.8	43.6		55.6	55.5	56.4
VEEV	31.3	31.3	46.4	45.2	45.7	43.2	45.2	45.9		68.1	67.8
EEEV	33.7	33.1	48.8	48.1	46.7	47.0	46.9	47.3	56.5		80.1
WEEV	31.9	32.0	67.8	44.9	44.6	44.2	42.7	56.0	50.3	56.4

^aThe upper right half of the table represents nonstructural proteins; the bottom left half of the table represents structural proteins. Abbreviation are as defined in the legend to Fig. 2.

The structural protein regions of SPDV and SDV have been comprehensively described (52, 54). However, the present comparative investigation has revealed a number of differences from those reported earlier. Thus, it is now reported that the SPDV structural genes are encoded by an ORF of 3,960 nt, versus 3,852 nt as previously described, thereby giving rise to a polyprotein of 1,320 amino acids as opposed to 1,286 amino acids (Fig. 1). This difference concerns the size of the 6K protein, which was previously reported to be 32 amino acids. Based on the recent sequencing of additional cDNA clones, it is now accepted that the previously reported SPDV sequence contained a 108-nt deletion in the 6K gene region. Therefore, the 6K protein of SPDV is now sized at 68 amino acids and as such resembles that of SDV. During the present investigation, a number of other sequence variations of the SPDV 6K coding sequence were identified. These included other deletions in the 6K gene region, which resulted in changes in the ORF at a location 3′ to the E2 gene and which led to early termination of the E2 ORF. Besides the deletion of 108 nt from nt 10272 to 10380 in the SPDV genome, another sequence deletion of 225 nt was of particular note. This occurred between nt 10251 to 10476 in the SPDV genome and resulted in a deletion of 65 amino acids, largely from the 6K protein but also including the first 10 amino acids of the contiguous E1 glycoprotein with no change in the ORF. Additional cDNA sequencing showed that the SDV structural protein genes are encoded from an ORF of 3,966 nt and not of 3,972 nt as previously reported. Thus, a polyprotein comprising 1,322 amino acids and not 1,324 amino acids is encoded by SDV (Fig. 1). As a consequence, the capsid protein gene is now recognized as comprising a 283-amino-acid protein, as opposed to the 285-amino-acid protein previously reported (53).

FIG. 1.

Salmon pancreas disease virus and sleeping disease virus genome organization. The number of amino acids is shown for each protein.

Nonstructural proteins of SPDV and SDV.

The nonstructural polyproteins of SPDV and SDV are predicted to be 2,601 and 2,593 amino acids, respectively. Cleavage sites in these polyproteins were deduced from amino acid sequence homologies with other alphaviruses and showed that SPDV and SDV contain generally larger individual nonstructural and structural proteins compared to other alphaviruses (Fig. 1). SPDV and SDV exhibited 95% amino acid identity over their nonstructural proteins; however, compared to other alphaviruses they showed only 41.7 to 43.6% identity. This is considerably lower than the percent amino acid identities (55.5 to 80.1%) that the other alphaviruses share with one another in their nonstructural protein regions (Table 2). When the individual nonstructural proteins of SPDV and SDV were compared, the amino acid sequence identities were 95.2% for nsP1, 96.6% for nsP2, 90.0% for nsP3, and 97.2% for nsP4. It is evident that while the nsP4 proteins of SPDV and SDV share the greatest homology, their nsP3 proteins are the most divergent. Like other alphaviruses the nsP3 of SPDV and SDV is conserved at its N terminus and is less conserved at its C terminus (46). Thus, the N-terminal 338-amino-acid regions of the SPDV and SDV nsP3 proteins show 97.3% identity, while their C-terminal regions are only 79.5% identical. Although this difference in the nsP3 C termini of the two viruses is primarily due to the presence of a 24-nt insertion in the SPDV nsP3 gene, a number of substitutions and a single deletion also exist. As a result, the C terminus of the SPDV nsP3 is 233 amino acids, while that of SDV nsP3 is 226 amino acids long.

Comparison of SPDV and SDV with other alphaviruses: nonstructural proteins.

Amino acid and nucleotide sequence analysis of the nonstructural protein gene region showed that SPDV and SDV appear to be equally closely related to the previously sequenced alphaviruses. Phylogenetic analysis based on the nonstructural polyprotein sequences of SPDV and SDV and 12 other alphaviruses showed that SPDV and SDV form a separate group between the New World viruses of Venezuelan equine encephalitis virus (VEEV) and Eastern equine encephalitis virus and the Old World viruses of Aura and SINV (Fig. 2).

FIG. 2.

Phylogenetic analysis of SPDV and SDV. The unrooted tree was based on the entire alphavirus nonstructural polyprotein sequences nsP1 to nsP4. The percent bootstrap support values for neighbor joining are given for each fork. Branch lengths of the rectangular cladogram are informative and drawn to scale. GenBank accession numbers of the sequences are as follows: SPDV, AJ316244; SDV, AJ316246; Aura virus (AURAV), AF126284; Semliki Forest virus (SFV), X04129; SINV, J02363; ONNV, M20303; Igbo Ora, AF079457; Sagiyama virus, AB032553; Ross River virus (RRV), M20162; Ockelbo virus, M69205; Eastern equine encephalitis virus (EEEV), X63135; VEEV, L04653; Western equine encephalitis virus (WEEV), AF214040; Barmah Forest virus (BFV), U73745. (Igbo Ora is a strain of ONNV, Sagiyama virus is a strain of Ross River virus, and Ockelbo virus is a strain of SINV [52].)

The nsP4 RNA-dependent RNA polymerase is the most highly conserved alphavirus nonstructural protein, and the levels of amino acid identity (range, 53.9 to 56.5%) shared by the nsP4 of SPDV and SDV with those of other alphaviruses are higher than when other nonstructural proteins were compared (results not shown). The nsP4 proteins of alphaviruses are also the most conserved in length, and this applies to both SPDV and SDV, which are both 609 amino acids long; all other sequenced alphaviruses have nsP4 proteins of 607 to 614 amino acids. Like other alphavirus nsP4 proteins, the aquatic alphaviruses contain the conserved GDD motif found at residues 466 to 468.

The nsP3 proteins of SPDV and SDV share the least homologies with those of other alphaviruses. Closer examination indicates that the N terminus of the nsP3 of SPDV and SDV exhibit 37.3 to 41.7% amino acid identities with those of other alphaviruses, whereas identities of between 53.2 and 80% were estimated when the nonaquatic alphaviruses were compared in this region. Like other alphaviruses, the C terminus of the nsP3 of SPDV and SDV is serine rich and shows little sequence identity to other alphavirus nsP3s (23). The amino acid identities shared by all alphaviruses, including SPDV and SDV, over the C-terminal nsP3 regions were comparatively low, ranging from 15.1 to 22.7%. The function of this protein during virus replication is unknown; however, it has been suggested to have a role in minus-strand and subgenomic RNA synthesis (20).

Motifs common to alphavirus nonstructural proteins and thought to have important roles in virus growth and replication can be clearly identified in both SPDV and SDV. The conserved motifs thought to make up the catalytic site of the nsP1 methyltransferase can be found in both SPDV and SDV. Motif I (containing the conserved histidine residue at 169 to 171 nt), motif II (AspXXArg [where X is any amino acid at 334 to 345 nt]), and motif IV (a tyrosine residue at 817 to 819 nt) are all found in the nsP1 of both SPDV and SDV (2, 18, 28, 44). The SPDV and SDV nsP2 proteins contain the GXXGXGKT motif (2,278 to 2,301 nt [SPDV], 2,276 to 2,298 nt [SDV]), responsible for NTP binding, and the residues Cys⁴⁸² and His⁵⁵² within the cysteine proteinase domain (9, 13, 14, 41, 42, 47, 50).

There is no opal termination codon between the nsP3 and nsP4 proteins of the SPDV and SDV isolates sequenced in this study, and this is also the case with a number of alphaviruses, including Semliki Forest virus and O'nyong-nyong virus (ONNV) (46). However, when the translational sequences around the nsP3-nsP4 junction region are aligned, a glutamine as opposed to an arginine residue replaces the termination codon. A more recent isolate of the ONNV has been shown to contain this termination signal (19). Therefore, until further sequence analysis on other isolates of SPDV and SDV has been performed, the lack of a termination codon may not be a true representation of these two viruses.

Comparison of SPDV and SDV with other alphaviruses: nucleotide sequence elements.

Alphaviruses have four well-characterized, conserved nucleotide sequence elements. Three of these—CS1, CS3, and CS4—occur in the nontranslated regions at the 5′ end of the genomic RNA, the junction region between the nonstructural and structural protein genes and the 3′ nontranslated region, respectively. CS2 is a 51-nt sequence found within the nsP1 gene. All of these sequence elements have important roles during virus replication (36, 48). CS1 in alphaviruses can form a double stem-loop structure (21, 36, 45), which is believed to play a role in plus-strand RNA synthesis (32). SPDV and SDV show no similarity to other alphaviruses in this region and have shorter 5′ nontranslated regions than previously reported alphaviruses (27 nt); however, they do contain a stem-loop structure in their 5′ nontranslated region, with the predicted translational start codon being found within the stem for both viruses.

SPDV and SDV contain the 51-nt sequence (CS2), found in nsP1, which is proposed to have a role in minus-strand synthesis (33). The SPDV and SDV CS2 motifs are at 150 to 200 nt, respectively, with the element showing 94.1% identity between the two viruses. RNA secondary structure analysis shows that the 5′ region of the SPDV/SDV CS2 motif is capable of forming a relatively stable stem-loop (Δ−9.5 kcal), while the 3′ half could give rise to a less-stable stem-loop (Δ−4.0 kcal). Therefore, this region may be capable of forming two stem-loops, a result similar to that found at this region in the nsP1 proteins of other alphaviruses (33, 45).

The nucleotide sequence of the 26S RNA junction region (CS3) has been previously described for SPDV and SDV (53, 55) and has a role as a promoter for the transcription of the subgenomic 26S RNA (22, 35). SPDV and SDV show 95.8% identity in this 24-nt sequence, differing at only 1 nt, which is otherwise conserved in other alphaviruses.

The 3′ nontranslated regions for SPDV and SDV are (i) 91 and 90 nt long, respectively, (ii) are two of the shortest described for alphaviruses (38), (iii) share 94.6% nucleotide identity, (iv) contain no repeat sequence elements, and (v) have similar RNA secondary structures (results not shown). The 3′-terminal 19-nt region, conserved among alphaviruses (22, 35), shows three differences between SPDV and SDV.

Cross-reactivities of SPDV and SDV isolates with MAbs.

An earlier study, in which the reactivities of four different SPDV isolates with SPDV-derived MAbs were reported, has been extended to include investigation of the cross-reactivities of additional SPDV and SDV isolates with panels of SPDV- and SDV-derived MAbs. IIF or IAP detection performed with fixed virus-infected and uninfected CHSE-214 cells were used to determine reactivities of the SPDV and SDV isolates (Table 3).

TABLE 3.

Reactivities of SPDV- and SDV-derived MAbs with SPDV and SDV isolates

Fusion/MAb	Virus (strain) to which MAb was raised	Reactivitya as determined by:
		IIF assay				IAP assay
		Irish F93-125 salmon SPDV	Norwegian N2P6 salmon SPDV	Norwegian N3P12 trout SPDV	Scottish F97-12 salmon SPDV	Scottish P42P salmon SPDV	French S49P trout SDV
F180/2D9	SPDV (F93-125)	+	+	+	+	+	+
F180/5D3	SPDV (F93-125)	+	+	+	+	+	+
F197/4H1	SPDV (F93-125)	+	+	+	+	+	+
F197/7A2	SPDV (F93-125)	+	+	+	+	+	+
F197/7B2	SPDV (F93-125)	+	+	+	+	+	+
F197/5A5	SPDV (F93-125)	+	+	+	+	+	+
F197/5D1	SPDV (F93-125)	+	+	+	+	+	+
I16	SDV (S49P)	−	−	−	−	−	+
K16	SDV (S49P)	−	−	−	−	+	+
L2	SDV (S49P)	−	−	−	−	−	+

^aThe reactivities of the five SPDV isolates and one SDV isolate with the panel of SPDV- and SDV-derived MAbs were compared. A “+” denotes a positive reaction with that MAb; a “−” denotes a negative reaction with that MAb.

Whereas all seven SPDV-derived MAbs reacted with all SPDV or SDV isolates tested, only one (K16) of the three SDV-derived MAbs showed reactivity with any of SPDV isolates. The P42P isolate of SPDV, with which the MAb K16 was reactive, was isolated in CHSE-214 cells from Atlantic salmon farmed in Scotland with salmon pancreas disease.

Experimental infections with SPDV and SDV.

The experimental design in which groups of Atlantic salmon and rainbow trout were inoculated with cell culture isolates of SPDV and SDV is outlined in Table 1. No mortality or gross pathology was detected during the 6 weeks p.i. in any of the six test groups. Virus was reisolated from kidney samples taken from all four virus-infected test groups at 1 week p.i. (Table 4). A more severe cytopathic effect, characterized by heavy necrosis but no vacuolization, was seen in CHSE-214 cultures that were inoculated with samples from SDV-infected fish, compared to cultures that were inoculated with samples from SPDV-infected fish. This finding indicated that SPDV and SDV isolates used in this study induce different cytopathic effects.

TABLE 4.

Detection of virus and neutralizing antibodies in experimentally infected fish

Test group	Challenge inoculum	Fish species	No. of fish/total no.a:
			From which virus could be isolated at 1 wk p.i.	With neutralizing antibody against SPDV at 6 wk p.i.	With neutralizing antibody against SDV at 6 wk p.i.
1	SPDV	Trout	5/5	7/10	4/5
2	SPDV	Salmon	5/5	10/10	NT
3	SDV	Trout	5/5	7/10	5/5
4	SDV	Salmon	5/5	5/10	NT
5	No virus	Trout	0/5	0/10	0/5
6	No virus	Salmon	0/5	0/10	NT

^aVirus reisolation was carried out by using kidney homogenates taken from the test groups and used to inoculate CHSE-214 cells. NT, fish were not tested due to insufficient serum.

Antibodies with neutralizing activity against SPDV were detected in all of the infected groups at 6 weeks p.i. (Table 4). Due to the scarcity of serum in samples collected from salmon, the presence of neutralizing antibodies against SDV were investigated in trout only and were detected in fish infected with SPDV and in fish infected with SDV. These results indicate that there is high serological cross-reactivity between the two viruses.

When the virus-infected groups were considered, the highest histopathology scores were seen at 2 weeks p.i. in pancreas and heart tissues and at 6 weeks p.i. in lateral muscle tissues (Table 5). Histopathology scores at these time points were used for statistical analysis of group comparison (Table 6). Significant differences between the SPDV-infected and uninfected groups (groups 1 and 5 and groups 2 and 6; Table 6) were seen in the heart, pancreas, and skeletal tissues for both fish species. Significant differences were observed between the SDV-infected and uninfected trout groups (groups 3 and 5; Table 6) in all three tissues; however, comparison of the SDV-infected and uninfected salmon groups (groups 4 and 6; Table 6) showed that a significant histopathological difference was detected with the pancreas tissue only and not with the heart or skeletal muscle tissue. Comparison of groups 1 and 3 (Table 6) indicated that no significant differences were detected between SPDV and SDV regarding their ability to cause histopathological effects in trout. However, by comparing the results from groups 2 and 4, it was evident that SPDV differed from SDV in its ability to induce histopathology in the skeletal muscle (6 weeks p.i.) of experimentally infected salmon.

TABLE 5.

Histopathological scores at 2 to 6 weeks p.i.^a

Test group	Mean score (frequency)b at:
	2 wk p.i.			3 wk p.i.			6 wk p.i.
	Heart	Pancreas	Lateral muscle	Heart	Pancreas	Lateral muscle	Heart	Pancreas	Lateral muscle
1 (trout-SPDV)	2.2 (9/10)	2.2 (8/10)	2.7 (8/10)	1.7 (7/10)	2.3 (7/10)	3 (9/10)	0.6 (0/10)	0.8 (0/10)	3.4 (9/10)
2 (salmon-SPDV)	2.2 (6/10)	2.5 (7/10)	0.5 (1/10)	2.2 (8/10)	1.9 (6/10)	1.3 (3/10)	0.7 (1/10)	0.8 (0/10)	1.6 (5/10)
3 (trout-SDV)	3.2 (10/10)	2 (6/10)	2.9 (9/10)	2.3 (2/10)	2.1 (8/10)	3.4 (10/10)	1.3 (4/10)	0.7 (0/10)	3.9 (10/10)
4 (salmon-SDV)	1 (3/10)	1.4 (4/10)	0.2 (1/10)	1.2 (3/10)	0.9 (2/10)	0.8 (0/10)	0 (0/10)	0.2 (0/10)	0.2 (0/10)
5 (trout-control)	0.6 (0/10)	0.1 (0/10)	0 (0/10)	0.2 (0/10)	0.2 (0/10)	0.1 (0/10)	0.4 (0/10)	0.4 (0/10)	0.2 (0/10)
6 (salmon-control)	0.1 (0/10)	0 (0/10)	0 (0/10)	0.1 (0/10)	0.1 (0/10)	0.0 (0/10)	0.1 (0/10)	0 (0/10)	0 (0/10)

^aHistopathological examination of the six test groups was done to look for characteristic lesions in heart, pancreas, and lateral muscle tissues associated with salmon and trout affected with salmon pancreas disease or sleeping disease.

^bScores are as defined in Materials and Methods. The frequency is number of affected animals/total number of animals tested.

TABLE 6.

Statistical (Kruskal-Wallis) analysis of histopathology results^a

Groups compared	Test group	Control group	P
			Pancreas at 2 wk p.i.	Heart at 2 wk p.i.	Muscle at 6 wk p.i.
1 and 5	SPDV-infected trout	Uninfected trout	<0.01	<0.05	<0.01
2 and 6	SPDV-infected salmon	Uninfected salmon	<0.01	<0.01	<0.05
3 and 5	SDV-infected trout	Uninfected trout	<0.01	<0.01	<0.01
4 and 6	SDV-infected salmon	Uninfected salmon	<0.05	NS	NS
1 and 3	SPDV-infected trout	SDV-infected trout	NS	NS	NS
2 and 4	SPDV-infected salmon	SDV-infected salmon	NS	NS	<0.05

^aThe histopathological scores for the six test groups from Table 5 were used for statistical analysis to allow test group comparison. NS, no significant difference.

DISCUSSION

In this study, we report the complete nucleotide sequences of SPDV and SDV, which, on the basis of sequence analyses of the structural protein genes, were previously and independently characterized as alphaviruses. Analyses of the genomic sequences presented in this collaborative paper indicate that these viruses possess genome organizations that are very similar to those of previously sequenced, arthropod-borne, mammalian alphaviruses. SPDV and SDV are very similar at both the nucleotide level and the amino acid sequence level, sharing 91.1% nucleotide sequence identity over their complete genomes and 94.5% amino acid identity over their complete protein-coding regions. Compared to other alphaviruses, SPDV and SDV share 41.7 to 43.6% identity in their nonstructural polyprotein and 31.3 to 33.7% identity in their structural polyprotein. These values are considerably less than the amino acid identities that the mammalian alphaviruses share with one another for the nonstructural polyproteins (55.5 to 80.1%) and for the structural polyproteins (43.2 to 73.4%) (Table 2). Likewise, both SPDV and SDV show little similarity to the recently characterized marine alphavirus isolated from the louse Lepidophthirus macrorhini collected from the southern elephant seal (SES) (25). For example, the E2 proteins of SPDV and SDV share 25.9% amino acid identity to that of SES virus, and this value is similar to the identities that SPDV and SDV share with other alphaviruses when their E2 proteins are compared. Although the aquatic alphaviruses show no particularly close relationship to any of the sequenced alphaviruses, they contain most or all of the many alphavirus conserved sequence elements believed to play an important role in virus replication and virus life cycle.

SPDV and SDV meet the species definition criteria of the International Committee on Taxonomy of Viruses, differing by up to 49.5% at the nucleotide sequence level and 64.5% at the amino acid sequence level when their E1 protein genes are compared to other alphaviruses (52). These results support the view that SPDV and SDV are closely related isolates of the same virus species, which we believe should be classified as a new member of the genus Alphavirus of the family Togaviridae, and for which we propose the name Salmonid alphavirus.

Although earlier disease investigations showed similarities in the histopathologies associated with naturally occurring salmon pancreas disease and sleeping disease in their respective salmonid fish species (5, 27), this is the first study reporting the results of contemporaneous experiments in which Atlantic salmon and rainbow trout were each infected with cell culture isolates of SPDV and SDV. Importantly, these cross-infection experiments indicated that, while the histopathological lesions induced by SPDV and SDV in salmonid fish were similar in nature, differences existed between the two viruses regarding the severity of the histopathology induced. Although the effects of infection of trout with SPDV and SDV were not statistically different, it was noted that the SDV induced less severe lesions in the skeletal muscle of salmon than SPDV (Table 6). It is difficult to predict whether this is a consistent feature of all SPDV and SDV isolates. Villoing et al. (53) reported that wild-type SDV, present as a tissue homogenate from naturally affected fish, could be detected at higher levels and for longer times after experimental infection of trout than an SDV isolate that had been grown in cell culture. Thus, there is some uncertainty as to how much influence the cell passage histories of the SPDV and SDV isolates, used in this investigation, have on the severity of the disease induced in different salmonid species. Demonstration that SDV isolates are incapable of inducing severe lesions in salmon would have epidemiological implications and may offer up the prospect of live attenuated vaccines, should such be considered acceptable.

The results of antibody titration assays for SPDV and SDV indicated that there was full cross-neutralization and confirmed that these isolates belonged to the same serotype. Unfortunately, due to the poor immunogenicity of SPDV and SDV accompanied with difficulties in producing sufficient amounts of highly purified virus for immunization, there are no satisfactory polyclonal antisera available with which to compare these isolates. However, the results of the MAb reactivity investigation indicated that antigenic differences exist between SDV, as exemplified by isolate S49P, and the SPDV isolates. Although all seven MAbs raised against SPDV reacted with all five SPDV isolates and the single SDV isolate under investigation, two of the three MAbs raised against SDV failed to react with any of the five SPDV isolates tested. The remaining MAb (K16), raised to SDV, reacted with the N42P isolate of SPDV only. Without additional research, it is uncertain whether this antigenic variation has arisen due to differences in the procedures used for cell culture propagation. It is noteworthy that the S49P isolate of SDV was grown in RTG-2 cells at 10°C prior to either mouse immunization and MAb reactivity testing, whereas the F93-125, F97-12, N2P6, and N3P12 isolates of SPDV were grown at 15°C. With viral RNA possibly existing as a quasispecies (15), it is probable that many sequence-distinct virus variants are present in clinical samples and that minor populations of these variants will be selected when the virus is isolated and propagated in cell culture. Thus, it is possible that genetically and antigenically different viruses will be selected from the mixed populations present in clinical samples by isolation and growth in different cell types maintained at different temperatures.

The major difference between the sequenced SPDV and SDV isolates is the occurrence of a 24-nt deletion found in the nsP3-coding region of SDV. Other minor sequence differences exist between SPDV and SDV, notably at the C termini of the capsid and E1 proteins. It has been suggested that the nsP3 sequence diversity of VEEV isolates may be related to adaptation of the virus to different hosts (34). However, until more isolates of SPDV and SDV are sequenced in these regions, we cannot be certain if these are characteristic differences that allow SPDV to be distinguished from SDV. Analysis of the nucleotide sequence of viruses present in clinical specimens may help to determine whether SPDV and SDV are genuinely different or whether differences in the methods used for cell culture isolation are important. This work has provided evidence that cell culture-derived virus pools comprise sequence-distinct viruses, including variants that contain deletions. Thus, additional sequence analysis performed within this investigation has revealed that deletions can occur within the genomic region encoding the 6K and E1 proteins, such that we now recognize that the SPDV sequence originally published (55) represented that of a virus variant that contained a deletion of 108 nt in the 6K gene region. The corrected genome sequence, specifying a 6K protein of 68 amino acids and not of 32 amino acids, resembles that of SDV. The frequency of such deletions within other regions of salmonid alphavirus genomes is unknown. At present it is impossible to attribute genetic differences to the variation observed in biological properties such as disease-causing capability and MAb reactivity, but this may be facilitated by the development of full-length infectious clones. Infectious clones would also be useful for investigating the viability of viruses that possess 6K gene deletions and that were shown to occur in the SPDV populations. Mutations in the 6K gene affect virus assembly and budding (10-12, 16, 48) and a Semliki Forest virus mutant with deletion of the 6K protein has been shown to produce low levels of infectious virus particles (24). The construction of an infectious SPDV clone and characterization of these 6K deletion mutants may prove useful in further understanding the role this SPDV 6K protein has in virus replication in fish cells.

Since most alphaviruses are transmitted by arthropods (7), it has been speculated that the sea lice species Lepeophtheirus salmonis may have a role to play in the transmission of salmonid alphaviruses (55). The isolation of a different alphavirus, SES virus, from the elephant seal louse, Lepidophthirus macrorhini, has fueled such speculation. Although it has been demonstrated in cohabitation experiments that direct fish-to-fish transmission occurs (3), the possibility that sea lice are also involved cannot be dismissed. It will be of interest to determine whether salmonid alphaviruses can be detected in sea lice and, if so, whether there is evidence of virus replication.