Salmon Gill Poxvirus, the Deepest Representative of the Chordopoxvirinae

ABSTRACT

Poxviruses are large DNA viruses of vertebrates and insects causing disease in many animal species, including reptiles, birds, and mammals. Although poxvirus-like particles were detected in diseased farmed koi carp, ayu, and Atlantic salmon, their genetic relationships to poxviruses were not established. Here, we provide the first genome sequence of a fish poxvirus, which was isolated from farmed Atlantic salmon. In the present study, we used quantitative PCR and immunohistochemistry to determine aspects of salmon gill poxvirus disease, which are described here. The gill was the main target organ where immature and mature poxvirus particles were detected. The particles were detected in detaching, apoptotic respiratory epithelial cells preceding clinical disease in the form of lethargy, respiratory distress, and mortality. In moribund salmon, blocking of gas exchange would likely be caused by the adherence of respiratory lamellae and epithelial proliferation obstructing respiratory surfaces. The virus was not found in healthy salmon or in control fish with gill disease without apoptotic cells, although transmission remains to be demonstrated. PCR of archival tissue confirmed virus infection in 14 cases with gill apoptosis in Norway starting from 1995. Phylogenomic analyses showed that the fish poxvirus is the deepest available branch of chordopoxviruses. The virus genome encompasses most key chordopoxvirus genes that are required for genome replication and expression, although the gene order is substantially different from that in other chordopoxviruses. Nevertheless, many highly conserved chordopoxvirus genes involved in viral membrane biogenesis or virus-host interactions are missing. Instead, the salmon poxvirus carries numerous genes encoding unknown proteins, many of which have low sequence complexity and contain simple repeats suggestive of intrinsic disorder or distinct protein structures.

IMPORTANCE Aquaculture is an increasingly important global source of high-quality food. To sustain the growth in aquaculture, disease control in fish farming is essential. Moreover, the spread of disease from farmed fish to wildlife is a concern. Serious poxviral diseases are emerging in aquaculture, but very little is known about the viruses and the diseases that they cause. There is a possibility that viruses with enhanced virulence may spread to new species, as has occurred with the myxoma poxvirus in rabbits. Provision of the first fish poxvirus genome sequence and specific diagnostics for the salmon gill poxvirus in Atlantic salmon may help curb this disease and provide comparative knowledge. Furthermore, because salmon gill poxvirus represents the deepest branch of chordopoxvirus so far discovered, the genome analysis provided substantial insight into the evolution of different functional modules in this important group of viruses.

INTRODUCTION

Poxviruses are large, complex viruses with linear, double-stranded DNA (dsDNA) genomes that replicate entirely in the cytoplasm and infect insects (Entomopoxvirinae) and vertebrates (Chordopoxvirinae). About half of the approximately 100 conserved genes of chordopoxviruses are also found in entomopoxviruses (1). The known chordopoxviruses have been divided into 10 genera and 1 unassigned genus by the International Committee on Taxonomy of Viruses. Vertebrate hosts include reptiles, birds, and mammals. There are no reports on the occurrence of poxviruses in wild fish populations. However, poxvirus-like particles have been found by transmission electron microscopy (TEM) in gills sampled during serious mortality of farmed koi carp (Cyprinius carpio L.) (2), ayu (Plecoglossus altivelis Temminck & Schlegel) (3), and Atlantic salmon (Salmo salar L., referred to here as salmon) (4). In addition, a poxvirus-like sequence has been reported from koi carp (5). Diagnostic use of a PCR assay based on this sequence suggests that the virus has spread through Europe and that the common grass carp (Ctenopharyngodon idella Valencienes) is also susceptible (6–8). There appear to be two disease manifestations: carp edema in very small fry in winter (2) and koi sleepy disease in larger juveniles in summer (9). In carp edema, the entire fish is swollen and the fish swim close to the surface, akin to fish suffering from hypoxia. In koi sleepy disease, the fish lie on the bottom in a lethargic state but swim away when touched. The gills are always affected, and skin and eye lesions may also occur. Both manifestations are strongly alleviated by immersing the fish in 0.5% saline, but as infectivity is retained, the treatment could be merely symptomatic (9, 10). The ayu suffers a severe proliferative gill disease with large basophilic inclusions that correspond to poxvirus-like particles on TEM. In other cases where poxvirus-like particles have been found in fish, no obvious inclusion bodies have been found using light microscopy (3).

Poxvirus infection in Atlantic salmon was suspected in the 1990s in cases of acute, high-mortality events in freshwater farms with juvenile fish (O. B. Dale and A. Kvellestad, unpublished data). Later, TEM images from diseased Atlantic salmon showed poxvirus-like particles that appeared to be distinct from those in carp, although both have a single lateral body like that found in entomopoxviruses instead of two lateral bodies like those found in other chordopoxviruses (4). Based on material from fish with TEM findings similar to those described above, we present the first complete poxvirus genome from a fish and compare it with that of other chordopoxviruses. Diagnostic tools derived from the sequence have allowed us to analyze the gill disease associated with salmon gill poxvirus (SGPV) in terms of pathology and the location of infection.

MATERIALS AND METHODS

Sample material.

Samples were collected from three different Norwegian salmon farms in which the fish had suspected SGPV-related disease (gill apoptosis) (Table 1) and at the following clinical stages: premortality (n = 20; samples were taken 1 to 3 days before mortality was observed), mortality (n = 60; samples were taken from tanks in which mortality occurred and lethargic fish crowded on the bottom), and postmortality (n = 10; samples were taken from tanks in which mortality was observed a week prior to sampling). The average weight was 27 g (range, 10 to 40 g).

TABLE 1.

Overview of material from Norwegian salmon farms

Fish	No. of fish with gill apoptosis			No. of fish/no. of farms
				Archival cases, 1995–2006	Controls (no gill apoptosis)
	Farm A	Farm B	Farm C		Diseased fish	Healthy fish
Archival cases and controls				39/14	48/8	3/1
Diseased fish
Premortality	20
Mortalitya	30	25	5
Postmortality	10

^aSampling was performed on 5 dead and 25 moribund fish. Except for the 5 dead fish, all fish used in this study were sampled while still alive.

Archived, formalin-fixed, paraffin-embedded (FFPE) gill tissue was identified from 14 cases with records of gill disease and apoptotic gill epithelial cells. These cases were geographically spread in both fresh- and seawater sites in Norway (Table 1). Included were 12 fish from the first known outbreak of so-called amoebic gill disease in Norway (11). A separate TEM study also demonstrated poxvirus-like particles in those 12 fish (4). In addition, samples from 48 fish with other gill diseases (without gill epithelial apoptosis) and 3 healthy fish were included as controls (Table 1; see also Table 3).

TABLE 3.

Overview of gill lesions in the controls (no gill apoptosis)

Farm	No. of fish	Gill histopathology and agents visible on light microscopy
D	9	Moderate adherences of lamellae and parasitic flagellates (Ichthyobodo spp.)
E	4	Moderate detachment of lamellar epithelial cells
F	4	Moderately thickened lamellae due to epithelial hypertrophy
G	4	Severe epithelial proliferation
H	4	Moderate lifting of epithelial cells and fungal infection
I	3	Moderate epithelial hypertrophy
J	4	Severe epithelial proliferation
K	10	Focal detachment and necrosis of lamellar epithelial cells and bacteria colonizing the apical surface of epithelium
L	6	Moderate hypertrophy and necrosis of epithelial cells; mucous cell proliferation

Tissue sampling for histology, TEM, and PCR.

All fish were anesthetized and autopsied, and gill tissues were fixed in neutral phosphate-buffered 10% formalin for histology and in RNAlater (Qiagen Inc., Valencia, CA, USA) for quantitative PCR (qPCR). Additional organs sampled for histology were heart, liver, intestine, spleen, kidney, muscle, and skin. Additional organs sampled for PCR were spleen, kidney, and skin from five fish in the premortality stage and five dead fish from farm A (Table 1). Formalin-fixed gill tissue from one fish in the premortality stage and one fish in the mortality stage was prepared for TEM as described previously (12).

In situ staining methods.

Paraffin-embedded and hematoxylin and eosin (H&E)-stained sections were made for histology. For a subset of the samples, a terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) in situ cell death detection kit AP (Roche, Basel, Switzerland) was used to confirm apoptosis. Prussian blue staining was used to verify hemosiderosis in the spleen and kidney. Staining for osmoregulatory chloride cells and proliferating cell nuclear antigen (PCNA) was performed as described previously (13, 14).

Immunohistochemistry (IHC) for SGPV.

Sections from gills were dewaxed, rehydrated, treated to demask antigen, and blocked with 5% bovine serum albumin in Tris-buffered saline to prevent nonspecific binding. Sections were incubated at 4°C overnight with a rabbit antibody (Pacific Immunology, Ramona, CA, USA) generated against the synthesized peptide GVNVDVKEFMQKFESNLSN-Cys (which is part of the L1 protein, a transmembrane protein expressed on the surface of the intracellular mature virion). Visualization was performed using an EnVision kit (Dako, Glostrup, Denmark) with horseradish peroxidase and 3-amino-9-ethylcarbazole as the chromogen.

DNA isolation and qPCR detection.

DNA was isolated from various tissues using a QIAcube system and a QIAamp DNA minikit according to the manufacturer's recommendations (Qiagen Nordic, Oslo, Norway). For archive material, a QIAamp DNA FFPE tissue kit was used (Qiagen Inc., Valencia, CA, USA). A qPCR assay based on the SGPV genomic sequence was designed for the molecular detection of virus DNA. The target locus was the homolog of the vaccinia virus (VACV) D13L open reading frame (ORF), which has been suggested to be a unique feature of poxviruses (15). The assay comprised forward primer ATCCAAAATACGGAACATAAGCAAT, reverse primer CAACGACAAGGAGATCAACGC, and the minor groove binding (MGB) probe CTCAGAAACTTCAAAGGA labeled with 6-carboxyfluorescein and a minor groove binding nonfluorescence quencher (MGBNFQ). The assay was run using a Platinum quantitative PCR SuperMix-uracil DNA glycosylase (UDG) kit (Life Technologies AS, Oslo, Norway) and the following PCR parameters: 50°C for 2 min (for UDG incubation), 95°C for 15 min (for UDG inactivation), and 50 cycles of 94°C for 15 s, 55°C for 30 s, and 72°C for 15 s. Reactions with threshold cycle (C_T) values above 40 were repeated for confirmation of the results. All results with C_T values above 45 were considered negative. In this study, there were no C_T values between 35 and 45.

When sections were cut from archival FFPE tissue, healthy fish gill tissue samples were interspersed between the samples from different disease outbreaks to control for carryover contamination. Isolation of DNA from FFPE tissue sections was done with a QIAamp DNA FFPE tissue kit according to the manufacturer's instructions (Qiagen Inc., Valencia, CA, USA).

RNA and DNA sequencing.

On the basis of the findings of TEM analyses, an Atlantic salmon gill tissue specimen containing poxvirus-like particles was selected for high-throughput sequencing. Total RNA was isolated from gill tissue fixed in RNAlater (Qiagen Norge, Oslo, Norway) using an RNeasy kit (Qiagen) and treated with Turbo DNA-free DNase (Life Technologies AS, Oslo, Norway) according to the manufacturer's recommendations. After DNase inactivation, 50 ng of total RNA was reverse transcribed and amplified using a QuantiTect whole-transcriptome kit (Qiagen). An initial round of total RNA pyrosequencing was done using a Roche 454 GS-FLX system and Titanium chemistry (454 Life Sciences, a Roche Company, Branford, CT, USA). All reads (521,710) from all reading frames and both strands were translated into protein sequences, and searches for sequence similarity to all poxvirus sequences available in GenBank were performed using the tblastx program (16). Two reads with weak similarity to known poxvirus sequences were identified, and one of these reads was used to design a qPCR assay (forward primer, ATCCAAAATACGGAACATAAGCAAT; reverse primer CAACGACAAGGAGATCAACGC; MGB probe, CTCAGAAACTTCAAAGGA; all sequences are written 5′ to 3′). Using the assay, a gill tissue sample with a high viral DNA content was selected for a subsequent round of sequencing. Total DNA was prepared using a DNeasy kit (Qiagen) and sequenced directly using a paired-end strategy and an Illumina HiSeq 2500 system (Illumina, Inc., San Diego, CA, USA). Reads were assembled de novo using a Velvet sequence assembler (17). Primers for PCR-based gap closing were designed using the software Primer Express (version 2.0.0; Applied Biosystems, Life Technologies Corporation, Carlsbad, CA, USA), and PCR was performed using a HotStarTaq master mix kit (Qiagen). Amplification products were sequenced directly using Sanger sequencing.

The sequencing with the Illumina system gave a total of 169,083,705 pairs of 101-bp reads. Using the Velvet sequence assembler, a total of 68,968 high-confidence contigs could be generated (coverage, >10 times; length, >100 bp). A contig containing the two reads originally identified as being poxvirus-like was found, and pairs of reads where one partner mapped uniquely to one contig and the other mapped to a different contig were extracted using the poxvirus-like contig as a starting point. Using this information, 24 of the contigs produced by the Velvet sequence assembler could be arranged into a tentative scaffold of the genome. PCR was successful across all gaps, but for a small number of loci, the exact number of low-complexity repeats could not be established using Sanger sequencing due to length and base compositional bias. Instead, the approximate lengths of repeat regions were determined using a 2100 Bioanalyzer and a DNA 1000 kit (Agilent Technologies, Santa Clara, CA, USA) to analyze the gap PCR products. Only 5 of the original 521,710 reads from the 454 sequencing data set could be mapped back to the final version of the virus genome. The five reads ranged in length from 52 to 528 bases, and the longest read was identical to the one that was used to design the PCR assay.

Genome annotation.

The SGPV genome was translated by GeneMarkS software (http://exon.biology.gatech.edu/) (18); long (>80-nucleotide) intergenic regions were checked for the presence of ORFs, and ORFs ranging from 50 to 100 codons were annotated to be predicted protein-coding genes if they showed significant sequence similarity to other proteins or to a conserved domain in the National Center for Biotechnology Information Conserved Domains Database (19) or contained predicted transmembrane helices and/or a signal peptide. Transmembrane helices were predicted using the TMHMM server (http://www.cbs.dtu.dk/services/TMHMM/) (20), and signal peptides were predicted using the SignalP (version 4.1) server (http://www.cbs.dtu.dk/services/SignalP/) (21). Tandem direct repeats were detected using the Tandem Repeats Finder program (22).

Protein sequence analysis and phylogenetic trees.

For detection of protein sequence similarity, the nonredundant protein sequence database at the National Center for Biotechnology Information (NIH, Bethesda, MD) was searched using the PSI-BLAST program (23). Predicted proteins of SGPV were assigned to clusters of nucleocytoplasmic virus orthologous genes (NCVOGs) using the PSI-COGNITOR program as previously described (24, 25). For phylogenetic analysis, protein sequences were aligned using the MUSCLE program (26) (http://www.ncbi.nlm.nih.gov/pubmed/15034147), and columns containing a large fraction of gaps (greater than 30%) and columns with low information content (27) were removed from the alignment. The alignment was used to construct an initial maximum likelihood (ML) phylogenetic tree with the FastTree program (http://www.ncbi.nlm.nih.gov/pubmed/20224823) with default parameters (28). The initial tree and the alignment were fed to the ProtTest program (29) to select the best substitution matrix. For each protein family, the best matrix found by ProtTest was used to construct the final ML tree with the TreeFinder program (30).

For the construction of the phylogenetic tree of poxviruses, multiple-sequence alignments of the sequences of 13 core genes present in all Poxviridae and African swine fever virus (ASFV) were employed (see Fig. S1 in the supplemental material). These genes belong to the following NCVOGs: NCVOG0022, major capsid protein; NCVOG0023, a D5-like helicase-primase; NCVOG0031, unclassified DEAD/SNF2-like helicases; NCVOG0038, DNA polymerase elongation subunit family B; NCVOG0076, DNA or RNA helicases of superfamily II; NCVOG0249, packaging ATPase; NCVOG0261, poxvirus early transcription factor (VETF), large subunit; NCVOG0262, poxvirus late transcription factor VLTF-3-like; NCVOG0267, RNA helicase DExH-NPH-II; NCVOG0271, DNA-directed RNA polymerase subunit beta; NCVOG0274, DNA-directed RNA polymerase subunit alpha; NCVOG1117, mRNA capping enzyme; NCVOG1164, A1L late transcription factor VLTF-2.

Reconstruction of gene content evolution.

The tree reconstructed from the concatenated alignment of 13 conserved proteins and the pattern of the presence-absence of SGPV proteins in the current version of the NCVOGs (24) were used to infer the gene loss and gene gain events and to obtain an ML reconstruction of the ancestral gene sets using COUNT software (31), as previously described (25).

Genome synteny analysis.

Genome synteny was visualized either with the Artemis genome comparison tool (32) or as dot plots of orthologous gene hits ordered by their positions in the genome (33). The synteny distance between viral genomes was calculated as previously described (33), with minor modifications, and a synteny-based neighbor-joining tree of the Poxviridae was constructed using the Neighbor program in the Phylip package (34).

Nucleotide sequence accession numbers.

The complete sequence of the SGPV genome was deposited in GenBank under accession number KT159937.

RESULTS

Evidence for poxvirus infection in farmed salmon.

The RNA isolated from salmon gill tissue containing poxvirus-like particles included sequences encoding putative proteins with significant similarity to those of poxviruses. A PCR probe was made using one such sequence in order to identify tissue with a high viral DNA content for direct paired-end sequencing. De novo assembly was performed, and gaps were filled in by PCR to generate a unique genome of 241,565 bp, excluding the termini, which were presumed to be covalently closed hairpins, as in other poxviruses. The relationship of the SGPV genome to the genomes of other poxviruses is detailed below.

PCR and peptide antibody probes were constructed from SGPV homologs of the highly conserved vaccinia virus D13L gene and L1R virion membrane protein, respectively. Poxvirus DNA was detected by PCR in the gills from all fish sampled from the three outbreak farms, with a trend of increasing C_T values over the disease course being detected (Table 2). In fish removed from tanks 1 to 3 days before death occurred (premortality stage), no lesions were found on autopsy, but most fish had no food in the gut, indicating appetite loss. On histopathology, changes were found only in the gills. Already at this stage, before clinical disease, apoptosis of lamellar epithelial cells was consistently found (Fig. 1b and c). Also, a general, moderate hypertrophy of this simple squamous epithelium was present, but no major blocking of the respiratory surfaces was found (Fig. 1b). A sparse fusion of lamellae due to epithelial proliferation and a moderate increase in the number of chloride cells were found in a few fish. All gills but one were IHC positive (Table 2). Only apoptotic epithelial cells either in the cytoplasm or in budding processes stained positive for poxvirus antigen (Fig. 1d and e). PCR of spleen and kidney tissue gave no C_T value for four fish, while one fish had a C_T value of 33.8 for spleen tissue and a C_T value of 34.4 for kidney tissue. All skin tissue PCRs were positive, with the median C_T value being 29.7 (range, 23.2 to 32.6).

TABLE 2.

Overview of results

Clinical stage	No. of fish	Median (range) CT value for poxvirus in gills by qPCR	% of fish with:
			IHC of gills	Hemophagocytosis
Premortality	220	18.1 (15.8–22.4)	95	0
Mortality	660	20.5 (15.7–28.9)	91.4a	66.7
Postmortality	110	24.7 (18.9–30.7)	20	20

^aTwo dead fish were not suited for IHC because of autolysis.

FIG 1.

Normal tissues and pathology in SGPV-infected Atlantic salmon. (a) A normal gill with thin lamellae (arrows) ensures efficient gas exchange. Chloride cells are present in normal numbers and at the normal location (arrowheads). (b and c) Detaching apoptotic cells with central clearing of chromatin (arrows) in the nuclear seen by H&E staining (b) and confirmed by red TUNEL staining (c). (d and e) IHC staining of poxvirus (brown) as cytoplasmic granules (d) and apical budding processes from apoptotic gill epithelial cells (e). (f) H&E staining of collapsed, adherent (arrows) thin lamellae losing apoptotic epithelial cells, creating an atelectasis-like condition hindering gas exchange. (g) H&E staining of proliferating (the arrow indicates metaphase), pale, foamy epithelial cells occluding the normally water-filled interlamellar space for gas exchange. Chloride cells are displaced and degenerated (arrowhead). (h) The lesion in panel g stained by IHC for PCNA showing brown nuclei, including proliferating cells in metaphase (arrow). (i and j) The lesion in panel g stained by IHC for chloride cells (red) that are displaced and enlarged (i) compared to the chloride cells in a normal gill (j). (k) TEM showing virus particles consistent with poxvirus in size and shape. Note the presence of crescents (CR), immature virions (IVs), and mature virions (MVs). (l) H&E staining of prominent hemophagocytosis (arrows) in the hematopoietic interrenal tissue. Methods included H&E staining (a, b, f, g, l); IHC staining for TUNEL (c), salmon gill poxvirus (d, e), PCNA (h), and chloride cells (i, j); and TEM (k).

In tanks in which fish were lethargic and there was some mortality (mortality stage), the main autopsy findings were swollen and slightly pale gills. Internal organs were also often pale, some spleens were enlarged, and no feed was found in the gut. Histopathology showed gill apoptosis in all fish at this stage, as described in the premortality stage (Fig. 1b to e). In addition, more severe gill changes obstructing the respiratory area were present in two different ways. First, in the phase with the severe detachment of apoptotic epithelial cells, the widespread adherence of the thin gill lamellae closed the water-filled space for gas exchange in an atelectasis-like manner (Fig. 1f). Second, the water-filled space between lamellae was solidified by proliferating epithelial cells (Fig. 1g), as demonstrated by PCNA staining (Fig. 1h). The proliferation also disrupted the tissue organization of the chloride cells (Fig. 1i), and apoptosis of chloride cells was also found. Histopathological lesions were also present in the spleen, kidney, and liver. A pronounced hemophagocytosis by scavenger endothelial cells and macrophages was found in the hematopoietic tissue of the spleen and kidney (Fig. 1l). Tissues with hemophagocytosis stained positive for Prussian blue Fe(III), demonstrating hemosiderosis (Table 2). Degenerative liver changes were variable but consistently present in dead fish. On IHC, over 90% of the gills were positive (Table 2), and labeling appeared as it did in the early stages (Fig. 1d and e). Furthermore, TEM demonstrated poxvirus-like particles in apoptotic cells (Fig. 1k). Crescents, spherical immature virions, and mature virions were seen in the cytoplasm, and these were also present in the extracellular space. Spleen and kidney were PCR negative in 3/5 fish (C_T value range, 32.6 to 35.8). All skin samples were PCR positive, with the median C_T value being 27.1 (range, 22.1 to 31.1).

In tanks in which mortality was observed a week prior to sampling (postmortality stage), most fish had no lesions on autopsy, except one fish had pale gills and four fish had enlarged spleens. Only a minor proliferation of gill epithelial cells, a very few apoptotic cells, and no IHC-positive cells were detected in all except two fish (Table 2). These two fish had pathology similar to that at the premortality stage, showing prevalent apoptosis, IHC-positive cells, and markedly lower C_T values (18.9 and 19.5) than the other fish. We also observed hemophagocytosis in the spleen and kidney at this stage, although to a much lower degree and in fewer fish than in the mortality stage (Table 2). All samples from the control cases with no signs of gill epithelial apoptosis were PCR negative, but a wide range of other gill pathologies as well as evidence for bacterial, fungal, and parasitic infections were present in the unhealthy fish (Table 3).

From each of the 14 archived, formalin-fixed, paraffin-embedded case series, at least one positive gill tissue sample was found by PCR. The 39 diseased fish had a median C_T value of 25.9 (range, 20.1 to 36.2). The interspersed control tissues had either high C_T values or no C_T value, indicating low or no cross contamination. All 12 samples from the so-called amoebic gill disease case were positive for poxvirus DNA by qPCR.

Genome analysis and evolutionary relationships of SGPV.

The SGPV genome consists of 241,565 bp (excluding the terminal hairpins) with a 37.5% GC content. The genome contains inverted terminal repeats of 5,679 bp each, similar to other poxviruses. Each of the inverted repeats, in turn, encompasses arrays of direct repeats. However, the tandem direct repeat arrays of SGPV, located at the very ends of the available genomic sequence, consist of only two 89-bp repeat units with 90% identity matches (each of these units consists of two 45-bp repeats with 88% identity). Thus, these direct repeat arrays are much smaller than those detected in other chordopoxviruses, although the possibility that they extend beyond the sequenced portion of the genome cannot be ruled out. Indeed, the highly conserved concatemer resolution sequence that is located between the repeat array and the apex of the terminal hairpin in other poxvirus genomes was not detected at the ends of the available SGPV sequence. The SGPV genome encompasses 206 unique predicted protein-coding genes (4 of these are contained within the terminal repeats and, accordingly, are present in the genome in two copies each; for details on gene prediction, see Materials and Methods). Comparison of the protein sequences encoded by these predicted genes to the sequences in the nonredundant protein sequence database at the National Center for Biotechnology Information (NIH, Bethesda, MD) using PSI-BLAST identified homologs with significant sequence similarity (E values, <10⁻⁴) for only 60 genes (for several additional predicted proteins, hits with apparently significant E values were identified as originating from regions of low sequence complexity and, accordingly, were dismissed as spurious). In addition to the standard database search, the predicted SGPV protein sequences were compared to the sequences of the NCVOGs (25), clusters of orthologous genes of nucleocytoplasmic large DNA viruses (NCLDVs), using a sequence profile search (see Materials and Methods). This comparison resulted in the assignment of 68 SGPV genes to NCVOGs, including 6 genes that showed no significant similarity to other proteins in BLAST searches. Additionally, a search for conserved domains led to a functional prediction in yet another protein (SGPV102). In total, specific, sequence conservation-based annotations were obtained through these procedures for 71 (34%) SGPV genes (Table 4 and Fig. 2). Among the genes without detectable homologs, 23 contained predicted transmembrane segments and/or a signal peptide, whereas 111 genes (55%) remained completely uncharacterized. Among the predicted products of these uncharacterized genes, many primarily consisted of low-complexity sequences and/or contained simple amino acid repeats (Table 4). These proteins are likely to be structurally disordered or assume unusual tertiary structures.

TABLE 4.

Predicted genes of SGPV^a

SGPV gene	Genome coordinates (protein lengthb)	NCVOG no.	Representation among NCLDVs	VACV gene name	Best hit (GI|Eval|% identity|aln_len|organism)	Predicted TM and SP	Functional annotation, comments, or inferred origin
001	1248–298 (317)						Hypothetical protein
002	2205–1288 (306)						Hypothetical protein
003	3227–2241 (329)						Hypothetical protein; low sequence complexity
004	4788–3361 (476)						Hypothetical protein
005	5884–4934 (317)						Hypothetical protein
006	7314–6361 (318)						Hypothetical protein
007	7770–7438 (111)						Hypothetical protein
008	8702–7830 (291)						Hypothetical protein
009	9383–9054 (110)					1 TM (C)	Hypothetical type I membrane protein, heptad repeats
010	9911–9546 (122)						Hypothetical protein; low sequence complexity
011	11295–10021 (425)						Hypothetical protein; low sequence complexity
012	12365–11373 (331)						Hypothetical protein
013	13629–12421 (403)						Hypothetical protein
014	15099–13681 (473)						Hypothetical protein
015	15912–15166 (249)						Hypothetical protein
016	16318–16007 (104)						Hypothetical protein
017	17728–16409 (440)						Hypothetical protein
018	18401–17922 (160)						Hypothetical protein
019	18870–18373 (166)						Hypothetical protein
020	19147–18863 (95)						Hypothetical protein
021	19468–19196 (91)						Hypothetical protein
022	19721–19533 (63)					1 TM (M)	Hypothetical membrane protein
023	20272–19751 (174)						Hypothetical protein
024	21336–20332 (335)						Hypothetical protein
025	21953–21345 (203)						Hypothetical protein
026	23067–21967 (367)						Hypothetical protein
027	24336–23167 (390)						Hypothetical protein
028	25255–24395 (287)						Hypothetical protein
029	25986–25294 (231)						Hypothetical protein
030	26353–26033 (107)						Hypothetical protein
031	26850–26389 (154)					1 TM (C)	Hypothetical type I membrane protein
032	27837–26851 (329)	0017	Phy, Mimi, Ent		401825817|3.E−11|28|180|Encephalitozoon hellem ATCC 50504		N-Myristoyl transferase; probable independent acquisition in different viruses; CACQ
033	28446–27847 (200)						Hypothetical protein
034	28512–30752 (747)	0330	Most NCLDVs, all families except Asf		660515722|3.E-07|26|242|Armadillidium vulgare iridescent virus		Divergent RING finger protein, potential E3 subunit of ubiquitin ligase; uncharacterized N-terminal domain upstream of RING domain; RING proteins in different NCLDVs likely have different origins; this SGPV protein is most similar to homologs from Iri and Phy; CACQ
035	31420–30749 (224)						Hypothetical protein
036	32306–31434 (291)						Hypothetical protein
037	34172–32325 (616)						Hypothetical protein
038	34466–34263 (68)					SP	Hypothetical protein
039	34829–34503 (109)						Hypothetical protein
040	35364–34792 (191)	0202	Pox, Iri		617520525|5.E-06|30|123|Poecilia formosa	1 TM (C), SP	Ig domain type I membrane protein; not closely related to Ig domain-containing proteins of other NCLDVs; CACQ
041	35957–35583 (125)						Hypothetical protein
042	36110–36592 (161)					1 TM	Hypothetical protein
043	36828–37679 (284)	0284	Some representatives of most NCLDVs except Asf		511086842|1.E-59|39|276|Entamoeba histolytica		Ser/Thr protein kinase; probable eukaryotic origin; putative ribosomal protein S6K, mTOR pathway component; not closely related to any other NCLDV kinase, likely independent origin; CACQ
044	38096–37665 (144)					1 TM (M)	Hypothetical membrane protein
045	39115–38135 (327)	1068	Scattered distribution in all NCLDV families	F2L	254568556|8.E-14|33|141|Komagataella pastoris GS115	1 TM (N)	Trimeric dUTPase highly similar to homologs from Phy but not poxviruses; contains uncharacterized N-terminal domain with a predicted TM; ANC
046	39674–39102 (191						Hypothetical protein
047	40162–39689 (158)						Hypothetical protein
048	40352–40882 (177)						Hypothetical protein
049	41988–40891 (366)						Hypothetical protein
050	42947–42033 (305)						Hypothetical protein
051	45667–42950 (906)				502875360|8.E-4|24|403|Planctomyces limnophilus		Metalloendopeptidase of the M60-like family; UAQ
052	46422–45664 (253)					1 TM (C)	Hypothetical type I membrane protein
053	46580–46422 (53)					1 TM (C)	Hypothetical type I membrane protein
054	47190–46636 (185)					1 TM (N)	Hypothetical type II membrane protein
055	48431–47325 (369)						Hypothetical protein
056	49116–48292 (275)						Hypothetical protein
057	50255–49101 (385)						Hypothetical protein
058	52022–50265 (586)						Hypothetical protein
059	52040–52516 (159)	1122	All Pox, some Iri, Mimi	J5		1 TM	Myristylated membrane protein, entry-fusion complex subunit; ANC; TAAATG
060	53257–52439 (273)						Hypothetical protein
061	53859–53299 (187)	0258	All Chor	J4R	13876678|1E−08|31|186|lumpy skin disease virus		DNA-dependent RNA polymerase subunit Rpo22; CPOX
062	54821–53886 (312)	1152	All Pox, Pith, some Mimi	J3R	41057489|3E−42|36|276|bovine papular stomatitis virus		Poly(A) polymerase small subunit, cap O-methyltransferase; ANC
063	55426–54788 (213)						Hypothetical protein
064	55824–55390 (145)					1 TM (M)	Hypothetical membrane protein
065	56600–55842 (253)	1063	All Pox	L4R			DNA-binding virion core protein VP8; POX; TAAATG
066	56631–57599 (323)	1168	All Pox	L3L	659488262|5.E-08|24|308|penguinpox virus		Virion protein required for early transcription; POX
067	59023–57596 (476)	0295	Most NCLDV families except Asf and Pan	F10L	544837|6.E-26|28|396|variola virus VAR, India-1967, peptide, 439 amino acids		Protein kinase involved in early stages of virion morphogenesis; ANC
068	59043–59897 (285)	0249	All NCLDVs	A32L	12085104|1.E-17|28|264|Yaba-like disease virus		DNA packaging ATPase; ANC; TAAATG
069	59901–60317 (139)						Hypothetical protein
070	62029–60323 (569)	1165	All Pox	E1L	9631476|2.E-14|23|381|Melanoplus sanguinipes entomopoxvirus		Poxvirus poly(A) polymerase catalytic subunit POX
071	62691–62071 (207)	0272	All Pox, most other NCLDVs	E4L	38229198|5.E-18|29|174|Yaba monkey tumor virus		Transcription factor S-II (TFIIS); ANC; TAAATG
072	63448–62681 (256)						Hypothetical protein
073	64802–63516 (429)						Hypothetical protein
074	64862–66538 (559)	1173	All Pox	E6R	40556061|3E-07|19|547|canarypox virus		Virion protein required for the formation of mature virions; POX; TAAATG
075	66539–67645 (369)						Hypothetical protein; low sequence complexity
076	70831–67634 (1,066)	0038	All NCLDVs	E9L	659488229|5E−142|32|1027|penguinpox virus		DNA polymerase ANC
077	71147–70848 (100)	0052	All NLCDVs	E10R	40556058|3E−24|48|92|canarypox virus	1 TM (false positive)	Disulfide (thiol) oxidoreductase (Erv1/Alr family) involved in disulfide bond formation during virion morphogenesis; ANC; TAAATG
078	71159–71788 (210)						Hypothetical protein
079	72152–71760 (131)						Hypothetical protein
080	73310–72168 (381)	1160	All Chor	I1L	5830616|4E−13|27|281|variola minor virus		DNA-binding virion core protein; CPOX
081	74930–73677 (418)						Hypothetical protein
082	76048–74933 (372)	1171	All Chor	I6L			Telomere-binding protein involved in viral DNA encapsidation; CPOX; TAAATG
083	77179–76049 (377)						Hypothetical protein
084	77295–78926 (544)						Hypothetical protein
085	79246–78857 (130)						Hypothetical protein; low sequence complexity
086	81234–79657 (526)						Hypothetical protein; low sequence complexity
087	81913–81221 (231)						Hypothetical protein; low sequence complexity
088	81912–82319 (136)					4 TM	Protein consists of hydrophobic decamer repeats; TM prediction could be spurious
089	86171–82275 (1,299)	0190					Hypothetical protein
090	88408–86198 (737)	0031	Nearly all NCLDVs	D6R	345107280|1E−156|41|657|Yoka poxvirus		SNF2-like helicase involved in early transcription; ANC
091	90927–88453 (825)	0023	All NCLDVs	D5R	571798002|5E−93|28|768|squirrelpox virus		Primase-helicase; ANC
092	91684–90920 (255)	0211	All Chor	F9L	9634782|8E−06|33|123|fowlpox virus	1 TM (C)	Myristylated IMV envelope protein; CPOX; TAAATG
093	92207–91647 (187)						Hypothetical protein
094	92262–92876 (205)	1067	Mimi		494264790|3E−09|28|178|Marinobacter algicola		Deoxynucleotide monophosphate kinase shared with Mimi, probable bacterial origin; CACQ
095	96496–93050 (1,149)	0037	Phy, Mimi, Mar, CrPV (multiple paralogs) but not other Pox		5121|2E−53|25|933| Schizosaccharomyces pombe		DNA topoisomerase II; ANC
096	96825–96496 (110)					1 TM (M)	Hypothetical membrane protein
097	97628–96828 (267)	0211	Most NCLDVs, all Pox	L1R	12085043|2E−29|31|225|Yaba-like disease virus	1 TM (C)	Myristylated IMV envelope protein; ANC; TAAATG
098	99427–97661 (589)	0022	All NCLDVs except Pan	D13L	345107288|8E−50|28|570|Yoka poxvirus		Major capsid protein (involved in morphogenesis but not incorporated into virions in poxviruses); ANC
099	99845–99456 (130)	1164	All NCLDV	A1L	289183841|1E−12|29|123|pseudocowpox virus		Late transcription factor VLTF-2; ANC
100	100665–99853 (271)	0262	All NCLDVs except Pith	A2L	571798015|7E−8|38|195|squirrelpox virus		Late transcription factor VLTF; ANC
101	103192–100967 (742)	1162	All Pox, Mimi	A3L	115531788|1E−49|24|697|Nile crocodilepox virus		Poxvirus P4B major core protein; POX
102	103950–103237 (238)		MCV, some Mimi, Phy				J domain-containing protein, putative cochaperonin; distantly related to J domains of other NCLDVs; CACQ
103	103956–104498 (181)	1377	All Chor	A5R	40556180|3E−14|33|172|canarypox virus		DNA-dependent RNA polymerase subunit Rpo19; CPOX
104	105738–104503 (412)	1179	All Chor	A6L			Virion core protein required for membrane biogenesis and formation of mature virions; CPOX; TAAATG
105	107964–105751 (738)	0261	All Pox, scattered in other NCLDVs	A7L	659488305|2E−104|32|734|penguinpox virus		VETF, large subunit; ANC
106	107945–109252 (436)	1176	All Chor	A8R	40556183|2E−08|24|248|canarypox virus		Poxvirus intermediate transcription factor VITF-3 subunit; CPOX; TAAATG
107	109518–109261 (86)					2 TM	Hypothetical membrane protein
108	113053–109535 (1,173)	0257	All Pox	A10L	157939724|4E−13|20|561|tanapox virus		Virion core protein P4; POX; TAAATG
109	113084–113923 (280)						Hypothetical protein
110	113964–114296 (111)					1 TM (N)	Hypothetical type II membrane protein
111	114326–114619 (98)						Hypothetical protein; low sequence complexity
112	114935–114600 (112)						Hypothetical protein containing serine-rich repeats
113	115188–115556 (123)						Hypothetical protein
114	115768–115556 (71)					1 TM (N)	Hypothetical type II membrane protein
115	117013–115769 (415)	1045	Some Iri and Mimi		339906034|2E−07|30|145|Wiseana iridescent virus		5′-3′ exoribonuclease of the XRN family; NCLDV proteins appear to be monophyletic; ANC
116	117383–117045 (113)					2 TM	Hypothetical protein
117	117721–117401 (107)						Hypothetical protein
118	118830–117736 (365)	1122	All Pox, Mimi, some Iri	A16L	41057529|2E−15|29|204|bovine papular stomatitis virus	1 TM	Myristylated protein, entry-fusion complex subunit; ANC; TAAATG
119	119936–118848 (363)					2 TM	Hypothetical membrane protein
120	119988–121418 (477)	0076	All Pox, in many other NCLDVs	A18R	115531805|1E−54|29|424|Nile crocodilepox virus		DNA helicase of superfamily 2, transcript release factor; ANC
121	121419–122474 (352)	2643	Some Mimi		504603808|3E−15|30|151|Ornithobacterium rhinotracheale		Apurinic-apyrimidinic endonuclease of the exonuclease III family; probable bacterial origin; CACQ
122	122812–122465 (116)	1370	All Pox	A21L	506498863|2E−06|24|111|Choristoneura rosaceana entomopoxvirus L.	1 TM (C)	Type I membrane protein, entry-fusion complex subunit; POX; TAAATG
123	122842–125133 (764)	0035	CrPV, Ent, some Iri, Mimi				NAD+-dependent DNA ligase; poorly conserved sequence but contains intact catalytic residues and shows the closest sequence similarity to NAD+-dependent ligases of Ent; ANC
124	125105–125602 (166)	0278	All Pox, majority of other NCLDVs	A22R	659488557|6E−16|32|149|pigeonpox virus		RuvC family Holliday junction resolvase; ANC
125	125599–126816 (406)	0263	All Pox	A23R	9634858|3E−25|27|395|fowlpox virus		Intermediate transcription factor; POX
126	126817–130305 (1,163)	0271	All NCLDVs except some Phy	A24R	225194776|0|47|1169|skunkpox virus		DNA-directed RNA polymerase subunit beta; TAAATG
127	130720–130310 (137)	1418	All Pox	A28L	51317191|3E−17|33|128|Diachasmimorpha longicaudata entomopoxvirus	1 TM (N)	Type I membrane protein, entry-fusion complex subunit beta; ANC; TAAATG
128	131699–130725 (325)	0260	All Pox	A29L	148912996|9E−08|27 |181|goatpox virus Pellor		DNA-directed RNA polymerase, 35-kDa subunit; POX
129	131870–132817 (316)						Hypothetical protein
130	132821–133525 (235)						Hypothetical protein
131	133536–135035 (500)						Hypothetical protein
132	135013–135495 (161)						Hypothetical protein
133	135919–135470 (150)						Hypothetical protein
134	136606–135941 (222)	1115	All Pox, scattered in other NCLDVs	D4R	9634732|1E−15|28|216|fowlpox virus		UDG; ANC
135	136671–138380 (570)						Hypothetical protein; low sequence complexity
136	138373–139212 (280)	0259	All Pox	D7R	9629029|4E−17|30|145|molluscum contagiosum virus subtype 1		DNA-directed RNA polymerase, 18-kDa subunit; POX
137	139235–139879 (215)	0236	All Pox, most other NCLDVs	D10R	9629031|1E−15|29 |161|molluscum contagiosum virus subtype 1		Nudix hydrolase, decapping enzyme; ANC
138	141785–139887 (633)	0027	All Pox, some Mimi	D11L	115531782|2E−174|43|635|Nile crocodilepox virus		Superfamily 2 helicase D11; POX; TAAATG
139	141949–142902 (318)	0330	All NCLDVs except Asco and Pith		658035022|2E−06|31|75|Malus domestica		RING finger-containing E3 ubiquitin ligase; probably independent acquisition in different NCLDV families; CACQ
140	143889–142951 (313)	1169	All Pox	D12L	9629033|3E−31|30|289|molluscum contagiosum virus subtype 1		Poxvirus mRNA capping enzyme, small subunit; POX; TAAATG
141	144893–143889 (335)	1122	All Pox, some Mimi, Iri	G9R	9634797|7E−06|36|78|fowlpox virus	1 TM	Myristylated protein, entry-fusion complex subunit; ANC; TAAATG
142	145769–144894 (292)	1369	All Chor	G8R	41057481|1E−06|26|171|bovine papular stomatitis virus		Protein containing a derived PCNA domain; VLTF-1; CPOX; TAAATG
143	145819–147291 (491)						Hypothetical protein
144	147884–147288 (199)	1182	All Pox	G6R			Predicted hydrolase or acyltransferase of the NlpC/P60 superfamily; weak sequence similarity to orthologs in other poxviruses; POX; TAAATG
145	148111–147914 (66)	1368	All Chor, one Ent, Asf	G5.5R	289183806|2E−04|24|65|pseudocowpox virus		RNA polymerase, subunit 10 (a very small protein, possibly missed during genome annotation of other viruses); POX
146	149892–148072 (607)	1060	All Pox, scattered in other NCLDVs	G5R	539191060|6E−13|36|176|myxoma virus		Flap endonuclease required for poxvirus genome replication; ANC
147	149931–150485 (185)				505137967|1E−05|41|59|Methanomethylovorans hollandica		Thioredoxin; no close homologs in other viruses; UAQ
148	150507–150884 (126)					1 TM (M)	Hypothetical membrane protein
149	150881–152773 (631)	1170	All Pox	G1L	115531736|6E−35|31|233|Nile crocodilepox virus		Metalloprotease essential for virion morphogenesis; POX; TAAATG
150	154796–152760 (679)	0267	All Pox, Asf, Mimi	I8R	41057099|5E−121|37|597|orf virus		RNA helicase of superfamily 2 implicated in early transcription termination; ANC; TAAATG
151	154823–156076 (418)	1161	All Pox, most other NCLDVs	I7L	115531734|7E−15|21|429|		Virion core cysteine protease involved in virion protein maturation; ANC; TAAATG
152	156073–156567 (165)						Hypothetical protein
153	156623–157354 (244)					1 TM (C), SP	Hypothetical protein
154	157464–164144 (2,227)	0269	All Chor; disrupted in some, including VACVs	(B22R VARV)	422933904|3E−120|29|1049|cyprinid herpesvirus 2	1 TM (C), SP	Giant type I membrane protein with homologs also in cyprinid herpesviruses, suggestive of gene transfer from SGPV to the herpesviruses (see the phylogenetic tree in Fig. 8); implicated in T cell inactivation; paralog of SGPV159 and SGPV162; CPOX
155	164257–168030 (1,258)					1 TM (C), SP	Hypothetical type I membrane protein
156	168031–169008 (326)						Hypothetical protein
157	168995–169900 (302)						Hypothetical protein
158	170583–169939 (215)						Hypothetical protein
159	170638–173652 (1,005)	0269	All Chor; disrupted in some, including VACV	0	9634792|5E−11|24|462|fowlpox virus	1 TM (C), SP	Giant type I membrane protein with homologs also in cyprinid herpesviruses, suggestive of gene transfer from SGPV to the herpesviruses (see the phylogenetic tree in Fig. 8); implicated in T cell inactivation; paralog of SGPV154 and SGPV162; CPOX
160	173910–173665 (82)					1 TM (N)	Hypothetical type II membrane protein containing pentapeptide repeats
161	173870–181351 (2,494)					SP	Hypothetical secreted protein
162	181528–185433 (1,302)	0269	All Chor; disrupted in some, including VACVs	0	9628967|5E−25|25|413|molluscum contagiosum virus subtype 1	1 TM (C), SP	Giant type I membrane protein with homologs also in cyprinid herpesviruses, suggestive of gene transfer from SGPV to the herpesviruses (see the phylogenetic tree in Fig. 8); implicated in T cell inactivation; paralog of SGPV154 and SGPV159; CPOX
163	185473–186558 (362)						Hypothetical protein
164	186693–188648 (652)					SP	Hypothetical secreted protein
165	188749–192687 (1,313)	0274	All NCLDVs except for some Phy	J6R	115531763|0|41|1311|Nile crocodilepox virus		DNA-directed RNA polymerase subunit alpha; ANC
166	193271–192684 (196)					1 TM (C)	Hypothetical type I membrane protein
167	193287–194597 (437)					SP	Hypothetical secreted protein, pentapeptide repeats
168	195155–194586 (190)	0253	All Pox	H2R	594019595|2E−35|40|151|avipoxvirus OKr-2014	1 TM (N)	Type II membrane protein, fusion-entry complex subunit; POX; TAAATG
169	197626–195161 (822)	1163	All Pox	H4L	6969751|3E−67|30|583|vaccinia virus Tian Tan		Pox_Rap94, RNA polymerase-associated transcription specificity factor, Rap94; POX; TAAATG
170	197724–198404 (227)						Hypothetical protein
171	198405–199343 (313)	0036	All Pox, Mimi	H6R	345107272|5E−60|40|310|Yoka poxvirus		DNA topoisomerase IB; ANC; TAAATG
172	199715–199329 (129)					SP	Hypothetical secreted protein
173	199747–202368 (874)	1451	All NCDLVs except Asco and Pan	D1R	225194732|4E−110|33|867|volepox virus		mRNA capping enzyme large subunit; ANC; TAAATG
174	204943–202382 (854)						Hypothetical protein
175	205237–204956 (94)						Hypothetical protein
176	205654–205238 (139)						Hypothetical protein
177	205659–207647 (663)						Hypothetical protein
178	207690–209033 (448)						Hypothetical protein
179	209178–209951 (258)						Hypothetical protein
180	210027–211280 (418)						Hypothetical protein
181	211532–213193 (554)						Hypothetical protein
182	213211–213954 (248)						Hypothetical protein
183	213947–214258 (104)						Hypothetical protein
184	214236–214847 (204)						Hypothetical protein
185	215300–214851 (150)						Hypothetical protein
186	215396–216664 (423)				167525479|6E−18|28|228|Monosiga brevicollis MX1		DNA or RNA methyltransferase; UAQ
187	216775–217242 (156)				209734208|9E−29|46|127|Salmo salar		Macrodomain, most similar to O-acetyl-ADP-ribose deacetylase; UAQ
188	217294–218286 (331)						Hypothetical protein
189	218360–219514 (385)						Hypothetical protein
190	219572–220492 (307)						Hypothetical protein; low sequence complexity; partly consists of tetrapeptide repeats
191	220576–221535 (320)						Hypothetical protein
192	221579–222580 (334)						Hypothetical protein
193	222676–223716 (347)						Hypothetical protein
194	224007–224258 (84)						Hypothetical protein; hydrophobic; 12-mer repeats
195	224390–225718 (443)						Hypothetical protein; low sequence complexity
196	226126–226542 (139)						Hypothetical protein
197	226596–227135 (180)						Hypothetical protein; cysteine rich; low sequence complexity
198	227202–228812 (537)						Hypothetical protein; low sequence complexity
199	228872–229897 (342)						Hypothetical protein
200	229951–230268 (106)						Hypothetical protein
201	230293–230985 (231)						Hypothetical protein
202	231098–231802 (235)						Hypothetical protein
203	232049–233533 (495)						Hypothetical protein
204	233892–233491 (134)					4 TM	Hypothetical protein; hydrophobic, consists mostly of hexapeptide repeats; TM prediction might be false positive
205	234526–235545 (340)						Hypothetical protein
206	235633–236631 (333)						Hypothetical protein
207	236777–238204 (476)						Inverted terminal repeat; identical to SGPV001 gene
208	238338–239324 (329)						Inverted terminal repeat; identical to SGPV002 gene
209	239360–240277 (306)						Inverted terminal repeat; identical to SGPV003 gene
210	240317–241267 (317)						Inverted terminal repeat; identical to SGPV004 gene

^aIn the first column, “SGPV” is omitted from the gene identifiers for brevity; in the last column “SGPV” is included; GI, GenInfo Identifier sequence identification number); aln_len, the length of pairwise protein alignment produced by BLASTP searches; SP, (predicted) signal peptide; TM, (predicted) transmembrane helix (C, M, and N denote the C-terminal, middle, and N-terminal location of the predicted transmembrane helix in the protein, respectively); the percent identity and alignment length are taken directly from BLASTP searches. IMV stands for intracellular mature virions; VARV stands for variola virus. The inferred origin of genes is indicated as follows: ANC, ancestral to NCLDV; POX, ancestral to poxviruses; CPOX, ancestral to chordopoxviruses; CACQ, convergent acquisition (with other NCLDVs); UAQ, unique acquisition. The transcription start element TAAAT is shown for those SGPV genes that have orthologs from other chordopoxviruses (the sequence TAAATG includes the translation start codon of the respective gene). Abbreviations for groups of viruses: Asco, Ascoviridae; Asf, Asfarviridae; Chor; Chordopoxvirinae; CrPV, crocodile poxvirus; Ent, Entomopoxvirinae; Iri, Iridoviridae; Mar, Marseilleviridae; MCV, molluscum contagiosum virus; Mimi, Mimiviridae; Pan, Pandoravirus; Phy, Phycodnaviridae; Pith, Pithovirus; Pox, poxviruses.

^bProtein lengths are in numbers of amino acids.

FIG 2.

Distribution of SGPV genes by tiers of inferred origin. The number of genes in each tier and the percentage of the total are indicated. NCLDV, genes inferred to have been present in the common ancestor of all NCLDVs; poxvirus, genes that originated in the common ancestor of the poxviruses; chordopoxvirus, genes that originated in the common ancestor of chordopoxviruses; TM/SP, transmembrane helix/signal peptide.

Among the predicted gene products of SGPV, homologs in other chordopoxviruses were detected for 59 proteins (Table 4 and Fig. 2). Among these conserved chordopoxvirus proteins, 32 belong to the previously inferred ancestral NCLDV gene set (24, 35), 17 are represented in all poxviruses (including entomopoxviruses), and 10 are specific for chordopoxviruses (Table 4 and Fig. 2). Eight genes have homologs in other NCLDVs but most likely were acquired independently (convergently), as suggested by sequence similarity and phylogenetic analysis, and only 4 genes appear to represent unique genes (with respect to the NCLDVs) captured from cellular organisms (Table 4 and Fig. 2; see Discussion).

The conserved gene set includes most of the essential genes involved in virus DNA replication and expression as well as the morphogenesis and structure of the virion core and the capsid (see the discussion of some notable exceptions in “Shared and distinct gene functions between SGPV and other chordopoxviruses and unexpected evolutionary patterns among SGPV genes” below). All conserved genes of SGPV showed the highest sequence similarity to the orthologs from chordopoxviruses, with only 3 exceptions, where the highest similarity (albeit by a small margin) was observed with entomopoxvirus orthologs (Table 4). These observations imply that the conserved SGPV genes share an evolutionary history, at least within the poxviruses. Accordingly, we used concatenated multiple-sequence alignments of the sequences of 13 highly conserved genes from this ancestral gene set to construct a maximum likelihood (ML) phylogenetic tree in which the root was placed between ASFV and the poxviruses, given that ASFV and poxviruses are sister groups in the overall NCLDV phylogeny (25). In the resulting tree, SGPV was placed at the root of the chordopoxvirus branch with unequivocal bootstrap support (Fig. 3). Thus, the phylogeny of chordopoxviruses generally follows the phylogeny of their hosts.

FIG 3.

Phylogenetic tree of poxviruses. The tree was constructed from a multiple-sequence alignment of 13 proteins that are conserved in all poxviruses and ASFV (NCVOG0022, major capsid protein; NCVOG0023, D5-like helicase-primase; NCVOG0031, unclassified DEAD/SNF2-like helicases; NCVOG0038, DNA polymerase elongation subunit family B; NCVOG0076, DNA or RNA helicases of superfamily II; NCVOG0249, packaging ATPase; NCVOG0261, poxvirus early transcription factor [VETF], large subunit; NCVOG0262, poxvirus late transcription factor VLTF-3-like; NCVOG0267, RNA helicase DExH-NPH-II; NCVOG0271, DNA-directed RNA polymerase subunit beta; NCVOG0274, DNA-directed RNA polymerase subunit alpha; NCVOG1117, mRNA capping enzyme; NCVOG1164, A1L transcription factor VLTF-2). The root position was forced between the two families. Numbers at internal nodes indicate bootstrap support (on a scale of from 0 to 1). Figure S1 in the supplemental material contains the alignments used to generate the tree.

In order to gain further insight into the evolution of the gene complement of SGPV, we performed an ML reconstruction of the ancestral gene sets using the poxvirus phylogenetic tree (Fig. 3) as a guide. The inferred ancestral gene sets showed an unexpected pattern (Fig. 4): 58 genes were mapped to the common ancestor of all poxviruses, and 62 genes were mapped to the common ancestor of chordopoxviruses. Thus, taken in their entirety, chordopoxviruses possess almost the same conserved gene set as the entire family Poxviridae, with very few additional conserved genes appearing after the divergence from the common ancestor with entomopoxviruses. In contrast, 38 additional genes were mapped to the common ancestor of the chordopoxviruses infecting tetrapods; i.e., these genes were gained along the tree branch between SGPV and crocodile poxvirus (CrPV). Thus, the reconstruction reveals a dramatic difference in the conserved gene repertoires between the common ancestor of all chordopoxviruses and the tetrapod poxvirus ancestor (Fig. 4). This difference likely reflects a major biological transition, the possible nature of which is discussed in “Shared and distinct gene functions between SGPV and other chordopoxviruses and unexpected evolutionary patterns among SGPV genes” below.

FIG 4.

Reconstruction of the evolution of the gene repertoire of the NCLDVs. The numbers at internal branches (shown only for the ASFV-Poxviridae branch and for the root) indicate the maximum likelihood estimates of the number of genes mapped to the respective ancestral form. The numbers after the virus names indicate the number of annotated genes. The NCLDV families used as outgroups are shown by triangles. The NCLDV tree topology is from reference 25.

The tetrapod chordopoxviruses, except for avipoxviruses, are characterized by a distinct genome architecture whereby the central portion of the genome shows a nearly perfect conservation of gene synteny and the terminal regions are highly divergent and often contain unique genes (36, 37), as depicted in the dot plots of Fig. 5. In contrast, genome-wide comparison of the gene orders between SGPV and other chordopoxviruses shows the extensive decay of synteny in SGPV and the complete disappearance of synteny between chordopoxviruses and entomopoxviruses (Fig. 5). Examination of the genomic dot plots (Fig. 5) and a genome architecture alignment (Fig. 6) between SGPV and other chordopoxviruses reveals several conserved gene blocks in the central part of the genome that are separated by strings of nonhomologous genes of variable length, along with at least two inversions of conserved genomic segments. To assess the evolution of the poxvirus genome architecture in more quantitative terms, we calculated the matrix of genome rearrangement distances and used it to construct an evolutionary tree of genome architectures (Fig. 7). This tree shows that the decay of synteny roughly follows the evolution of gene sequences (compare the trees in Fig. 7 and 3), but the rate of disruption of the ancestral gene order is nonuniform, with the major change mapping to the branch between SGPV and the rest of the chordopoxviruses.

FIG 5.

Dot plot comparison of poxvirus gene orders. Each dot corresponds to a pair of orthologous genes. The horizontal axis shows the SGPV genes, and the vertical axis shows the GenInfo Identifier sequence identification numbers for genes of the respective viruses.

FIG 6.

Alignment of the genome architectures of SGPV and VACV. The alignment was generated using the Artemis tool and the table of gene orthology derived from the NCVOG assignments obtained in this work. The orthologous genes are connected by red lines, and the names of the respective vaccinia virus genes are indicated. nt, nucleotides.

FIG 7.

Synteny-based evolutionary tree of poxviruses. The root between chordopoxviruses and entomopoxviruses was forced. The tree was constructed using the neighbor-joining method, and the distances between the genome architectures of the respective viruses that were estimated as described previously (33) are shown in the table underneath the tree; a unit distance means that the fraction of orthologous gene pairs that belong to synteny blocks is equal to e−1. Amsmo, Amsacta moorei entomopoxvirus; Melsa, Melanoplus sanguinipes entomopoxvirus; Vacco, vaccinia virus; Deevi, deerpox virus; Psevi, pseudocowpox virus; Canvi, canarypox virus; Crovi, crocodilepox virus; Squvi, squirrelpox virus; Molco, molluscum contagiosum virus; Yabvi, Yaba-like disease virus.

Shared and distinct gene functions between SGPV and other chordopoxviruses and unexpected evolutionary patterns among SGPV genes.

Here we discuss the predicted functions and some unusual aspects of evolution of the SGPV genes in the order of the tiers of ancestry, i.e., the point of gene origin for (acquisition by) this virus (Fig. 2). The ancestral NCLDV genes retained by SGPV encode the principal functions required for genome replication and expression, with no genes having been lost since the common ancestor of all poxviruses. However, two genes merit special mention in the context of poxvirus evolution, namely, the genes for DNA topoisomerase II (Topo II; SGPV095) and NAD-dependent DNA ligase (SGPV123). These (putative) ancestral NCLDV genes are uncharacteristic of chordopoxviruses, being present only in CrPV and SGPV, whose genomes encode both Topo II (which has multiple paralogs in CrPV) and Topo IB, which is conserved in the rest of the chordopoxviruses. The evolution of topoisomerases in NCLDVs appears to have been quite complex, involving both differential gene loss and the apparent independent acquisition of homologous genes (38). Thus, SGPV and CrPV could represent either the ancestral state with two distinct topoisomerase genes or an intermediate state after the Topo IB gene had been acquired but the Topo II gene had not been lost as it was in chordopoxviruses and entomopoxviruses independently.

The NAD-dependent ligase also appears to be an ancestral NCLDV gene but was replaced by the distinct, ATP-dependent ligase in several groups of viruses, including most of the chordopoxviruses, after the divergence from the common ancestor with CrPV (39). The finding that SGPV encodes an NAD-dependent ligase but not an ATP-dependent ligase is compatible with this scenario. The predicted NAD-dependent ligase of SGPV shows an unexpectedly low sequence similarity to homologs from other NCLDVs (Table 4), suggestive of some peculiarity in the DNA replication process of this virus.

The next evolutionary tier of the SGPV genes, those that are conserved in all poxviruses (Table 4) (1, 24), includes components of the transcription apparatus, such as several RNA polymerase subunits and the poly(A) polymerase catalytic subunit; several components of the virion core and proteins involved in virion morphogenesis, such as the metalloprotease G1; and six subunits of the fusion-entry complex (40, 41) (homologs of three paralogous subunits of this complex, A16, G9, and J5, are also detectable in mimiviruses and iridoviruses, suggesting that some form of this complex might be ancestral in NCLDVs). Of note is the presence in SGPV of a highly diverged ortholog of the G6 protein, a predicted amidase or acyltransferase that is thought to be important for the virus-host interaction but whose specific function remains obscure (42).

The genes that are conserved in chordopoxviruses, to the exclusion of entomopoxviruses, follow the same general functional themes, including RNA polymerase subunits, such as A5 and J4; the late-stage transcription factor G8 containing a highly diverged PCNA domain; and proteins involved in core morphogenesis, e.g., the telomere-binding protein I6 and the protein A6 required for virus membrane biogenesis. Particularly notable in this group are three paralogous genes (SGPV154, SGPV159, SGPV162), located near the right end of the genome, that encode homologs of variola virus B22R, a giant type I membrane protein implicated in the virus-induced shutdown of the host adaptive immunity, specifically, inhibition of T lymphocytes (43). Of these three paralogous genes, SGPV154 is similar in length to homologs from other poxviruses, whereas SGPV159 and SGPV162 are considerably shorter and, thus, have apparently been truncated during the evolution of the SGPV lineage. However, the conservation of the predicted signal peptide and the C-terminal transmembrane helix in all three proteins (Table 4) suggests that they remain functional. The proliferation of this gene in SGPV, which parallels its independent triplication in CrPV (44), implies an important role of this route of counterdefense. Interestingly, homologs of this gene were also detected in cyprinid herpesviruses, suggesting transfer of this gene, involved in virus-host interaction from SGPV (or its relative), to unrelated viruses within the same host (Fig. 8). Moreover, the cluster of SGPV genes that encompasses the three B22R paralogs also contains two other proteins with a similar, very large size (SGPV155 and SGPV161; Table 4) that are predicted to be, respectively, membrane associated and secreted. The sequences of these proteins show no similarity to the B22R sequence, but the proteins might perform roles similar to the role performed by B22R via a distinct mechanism.

FIG 8.

Phylogenetic tree of the viral B22R-like genes. The numbers on the left show bootstrap values as percentages. The bar shows the scale as the estimated number of amino acid substitutions per site. For the cyprinid herpesviruses, the GI numbers are indicated on the right. The three paralogs from SGPV are shown in red. The chordopoxvirus sequences are collapsed and shown as a triangle. Figure S2 in the supplemental material contains the alignments used for construction of the tree.

The late-stage genes of chordopoxviruses, as well as most intermediate and some early genes, contain a distinct sequence element within which transcription starts. This element has the sequence TAAAT, where the second T usually corresponds to the second nucleotide of the translation initiation ATG codon of the respective gene (45–47). In the process of transcription initiation, the complement of this element serves as the template for the formation of the 5′-terminal poly(A) sequence that is present in many chordopoxvirus transcripts and is produced by RNA polymerase slippage (48, 49). This TAAAT element is conserved in nearly all SGPV homologs of the respective chordopoxvirus genes (Table 4), suggesting that the main features of transcription initiation are shared by all chordopoxviruses.

Eight genes of SGPV have homologs in other NCLDVs and, thus, are formally assigned to NCVOGs, but as indicated by sequence similarity searches and phylogenetic analysis results (Table 4), they have probably been independently acquired by different viruses, which implies that they play important roles in virus-host interactions. Two of these genes (SGPV034 and SGPV139) encode RING finger proteins that could function as either specialized E3 subunits of ubiquitin ligases or inhibitors of ubiquitin pathways. RING finger-containing E3 proteins are encoded by many NCLDVs, including some of the orthopoxviruses, in which they are essential for pathogenicity (50–52). However, viral RING finger domains, including those encoded by the SGPV genome, show limited sequence similarity to each other and have probably been acquired independently. This independent acquisition was likely driven by the selection for virus interaction with the host ubiquitin networks. A similar trend of likely independent acquisition by diverse viruses is apparent for the DnaJ (J) domain, which was detected in the SGPV102 protein sequence. The J domain is present in mimiviruses, some phycodnaviruses, a single chordopoxvirus (molluscum contagiosum virus [36]), as well as polyomaviruses. The polyomavirus J domain has been shown to function as a cochaperonin, enhancing the activity of the Hsc70 chaperone in the infected cells (53). A similar role in viral protein folding could be played by SGPV102.

Of special interest is the SGPV043 protein, a predicted serine/threonine protein kinase that is highly similar to the eukaryotic ribosomal protein S6 kinase (S6K), with which it shows up to 60% amino acid sequence identity. S6K is a component of the mTOR pathway and, more specifically, of the TORC1 complex, an environmental sensor that promotes anabolic pathways and inhibits catabolic pathways (54). Thus, this gene, which seems to have been convergently captured by SGPV and several other NCLDVs, could act as a regulator of the global metabolic state of virus-infected cells.

Only four SGPV genes appear to be unique acquisitions from cellular organisms, as they have no homologs in other NCLDVs. These are the metalloendopeptidase SGPV051, the thioredoxin SGPV147, the predicted DNA or RNA methyltransferase SGPV186, and the macrodomain-containing protein SGPV187, a putative O-acetyl-ADP-ribose deacetylase. Each of these proteins showed a high level of divergence from cellular homologs, presumably due to the high rate of evolution upon transfer to the viral genome, precluding a convincing inference of origin by phylogenetic analysis (not shown). The presence of the macrodomain is of special interest. Previously, this domain has been detected in several groups of animal positive-strand RNA viruses (55) and has been shown to inhibit double-strand RNA-dependent phosphorylation of the interferon regulatory factor 3 (IRF-3), a key transcription factor for interferon induction (56). The macrodomain of SGPV, to our knowledge, is the first domain of this family to be discovered in a DNA virus, and it might play a similar role as an inhibitor of the interferon pathway.

Conserved poxvirus genes that are missing in SGPV: distinct pathways of membrane biogenesis?

As pointed out above, SGPV lacks numerous genes that are (nearly) fully conserved among the tetrapod-infecting chordopoxviruses, with the implication being that they are lost in SGPV (Table 5). Only three ancestral poxvirus genes appear to have been lost in SGPV; one of these, the gene for the protein phosphatase H1, was, apparently, also independently lost in some entomopoxviruses. The absence in SGPV of the A11, L2, A14, and A17 genes highlights a central functional theme that extends into the longer list of conserved chordopoxvirus genes that are missing in SGPV, namely, membrane biogenesis (57) (Table 5). At least half of the missing genes (14 of 28) encode proteins implicated in this process. Of the seven subunits of a distinct protein complex involved in the association of the viroplasm with membranes, which is required for immature virion formation (58), only one, the protein kinase F10, an ancestral NCLDV protein that is likely to perform multiple functions (59), is represented by an ortholog in SGPV (Table 5; two complex subunits, D2 and A15, are not listed because they appear to have been lost in some other chordopoxviruses as well). Taken together, these findings imply that SGPV employs a pathway of membrane biosynthesis that is distinct from that of other chordopoxviruses. Several uncharacterized SGPV proteins contain predicted transmembrane segments (Table 4) but show no detectable sequence similarity to the sequences of proteins of other poxviruses shown to participate in membrane biogenesis; it appears likely that at least some of these SGPV proteins belong to the putative alternative pathway.

TABLE 5.

Conserved chordopoxvirus genes missing in SGPV

Conserved genea	VACV gene	Known or predicted function	Essentialb	Comment
Genes conserved in chordopoxviruses and entomopoviruses
1178	L5R	Membrane protein, fusion-entry complex subunit	Yes
1181	A11R	Membrane-associated protein implicated in endoplasmic reticulum recruitment for virion morphogenesis	Yes
0040	H1L	Dual-specificity (Ser/Thr and Tyr) protein phosphatase	Yes	Conserved in only two entomopoxviruses
Genes conserved only in chordopoxviruses
1185	A20R	DNA polymerase processivity factor	Yes
1385	I3L	Single-stranded DNA-binding protein essential for replication	Yes
1184	G2R	Late transcription elongation factor	Yes
1172	A12L	Virion core protein	Yes
1177	F17R	DNA-binding virion core protein	Yes
1043	G7L	Virion core protein required for immature virion formation	Yes
1398	A19L	Virion core protein	Yes
0060	G4L	Glutaredoxin involved in the pathway for cytoplasmic disulfide bond formation	Yes
1396	A2.5L	Thioredoxin-like protein involved in the pathway for cytoplasmic disulfide bond formation	Yes
0012	A33R	C-type lectin involved in extracellular virion morphogenesis	No
0268	A25/A26L	A-type inclusion body-like protein	No
0255	O1L	Poorly characterized protein, activator of the extracellular signal-regulated kinase pathway	No
1167	F12L	Protein involved in intracellular enveloped virion maturation and cytoskeleton-dependent virion export	No	Inactivated derivative of DNA polymerase, possibly of bacteriophage origin
1367	G3L	Fusion-entry complex subunit	Yes
1376	H7R	Protein involved in MVc membrane biogenesis	Yes
1380	A14L	Protein involved in MV membrane biogenesis	Yes
1411	A17L	Protein involved in MV membrane biogenesis	Yes
1395	L2R	Protein involved in MV membrane biogenesis	Yes
1366	I5L	MV membrane protein	No
1383	I2L	Membrane protein essential for virus entry	Yes
1391	J1R	Protein involved in MV formation, assembly complex subunit	Yes
1416	D3R	Protein involved in MV formation, assembly complex subunit	Yes
1412	A30L	Protein involved in MV formation, assembly complex subunit	Yes
1392	A9L	Protein involved in MV morphogenesis	Yes
0256	H3L	MV membrane protein involved in cell attachment	No
1415	A14.5L	MV membrane protein that enhances virulence	No

^aNCVOG number.

^bEssentiality was determined for vaccinia virus.

^cMV, mature virion.

Among the other conspicuous gaps in the gene repertoire of SGPV are the single-stranded DNA-binding protein I3 (60, 61) and the DNA polymerase processivity factor A20 (62), two proteins that are essential for VACV DNA replication. Among the predicted SGPV gene products, there are no obvious candidates that could replace these proteins, so the involvement of functionally analogous host proteins seems to be a distinct possibility.

Also missing in SGPV are two components of the thiol-disulfide oxidoreductase pathway, which is essential for the formation of the disulfide bonds in the subunits of the VACV fusion-entry complex, as well as envelope proteins L1 and F9 (63–65). Orthologs of the L1 protein along with the upstream component of the thiol-disulfide oxidoreductase pathway, the E10 protein, are conserved in nearly all NCLDVs (24), with the implication being that the pathway as such is essential. The two missing subunits, the glutaredoxin G4 (66) and the thioredoxin-like protein A2.5 (67), have no orthologs in other NCLDV families either, indicating that the complete oxidoreductase pathway characterized in VACV evolved only after the divergence of SGPV and the rest of the chordopoxviruses. The predicted SGPV thioredoxin (SGPV147) might be responsible, at least in part, for the missing portion of the pathway.

DISCUSSION

Poxvirus infection in salmon was suspected in the 1990s, as TEM showed apoptotic gill epithelial cells with poxvirus-like particles in samples submitted to the Norwegian Veterinary Institute from acute, high-mortality events in freshwater farms with juvenile fish (Dale and Kvellestad, unpublished). Typical poxvirus structures were further characterized in a TEM study of gill disease in Atlantic salmon (4), but no taxonomic assignment was possible in the absence of sequence data. In this study, we confirmed the presence of poxvirus particles and determined the sequence and phylogeny of salmon gill poxvirus, developed qPCR and IHC methods, and analyzed the disease from current as well as archival samples, including the outbreak previously studied by TEM (4).

As there was no experimental model for SGPV disease, we obtained samples from fish with spontaneous cases of the suspect apoptotic gill disease in two hatcheries without other significant disease problems. As controls we included samples from fish involved in several gill disease outbreaks without apoptosis of gill epithelium as well as healthy fish. We found the SGPV infection by qPCR only in the disease cases and could link the SGPV infection in situ to the apoptotic respiratory epithelium by IHC. We found that the infection was widespread in the gills at least 3 days before the onset of severe clinical disease. Mortality coincided with blocking of the respiratory gill surfaces by two different mechanisms. The SGPV infection seems to induce massive apoptosis and detachment of the epithelium, resulting in the acute adherence of the thin gill lamellae. In other fish, an excessive proliferation of the epithelium blocked the respiratory surfaces. These findings indicate that viral replication precedes the gill pathologies that can be likened to atelectasis and solidification of the lungs, respectively. Hypoxia and osmoregulatory disturbances are the expected pathophysiological consequences from such lesions in fish. This is in keeping with the clinical experience that stopping feeding, raising oxygen levels, and avoiding all stress minimize mortality, which otherwise may approach 100% within hours in a tank of fish. Although the classical Koch's postulates remain to be fulfilled, our findings indicate that SGPV causes a distinct disease primarily affecting the gills in salmon. In fish, localized gill infection seems to be the rule for the suspected poxviruses (2–4, 9).

With regard to systemic pathology and infection, we found that hemophagocytosis was associated with severe disease. In infectious salmon anemia, a severe orthomyxoviral disease of salmon, hemophagocytosis is due to virus attachment to the erythrocyte surfaces (68). However, only high or no C_T values were found in organs of the poxvirus-infected salmon with hemophagocytosis, while low C_T values were obtained from the gills in these fish. Hemophagocytosis is possibly a sign of circulatory disturbances aggravating the SGPV disease. It is noteworthy that lethargy, gill pathology, and hemophagocytosis are also reported in koi sleepy disease, associated with a pox virus of gills and seemingly not internal organs (9). Poxviruses are generally epitheliotropic, and skin tissue samples were positive by PCR both in this study and in the study of koi carp (9). However, we found no skin lesions in the salmon, and at present, we cannot exclude the possibility that skin tissue samples carry just virus shed from the gills. IHC seemed to indicate a very narrow cell tropism, since the simple, squamous lamellar epithelium of the gill was infected, while the adjacent stratified epithelium did not show signs of infection. The high mortality due to respiratory SGPV infection appears to be different from that in poxvirus infections in air-breathing vertebrates, where lung pathology is usually seen as part of a generalized infection. This could be related to the fundamental anatomical differences in the respiratory systems in the two vertebrate groups, where fish have their respiratory surface much more exposed to the exterior.

All salmon sampled a week after mortality had subsided in a tank were still infected, as shown by PCR. However, virus levels were generally lower and hemophagocytosis was much less prominent. These findings suggest that although recurrent, acute outbreaks in a tank are not reported by the farmers, SGPV infection may persist. For how long we do not know, and as the reservoir of infection is also unknown, we cannot rule out the possibility of reinfections. However, our archival samples do show that the infection is found not only in freshwater hatcheries but also in the seawater farms that receive the salmon for additional growth. In the cases of combined amoebic gill disease and SGPV disease that we confirmed here, 82% of the fish died (11). This is in agreement with the findings of other studies demonstrating that most gill diseases in the seawater rearing phase are complex, with multiple agents being present (69). Investigation of the role of SGPV in mixed infections will be an important future task, as the gill problems caused by these mixed infections cause considerable losses. To this end, we now have two new diagnostic methods, qPCR and IHC, for the detection of the SGPV in fish tissues, and these are useful for screening and resolving the complex pathology, respectively.

Judging from the archival samples, the SGPV disease emerged in the mid-1990s in Norwegian salmon farms over a wide geographical range. However, SGPV is distinct from other chordopoxviruses that have been analyzed, and its reservoir is unknown. High mortality, like that caused by SGPV in salmon, can be a sign of a new host-agent system with low compatibility; on the other hand, intensive farming may have changed an old host-agent balance. Further studies are needed to clarify the reservoir and host range of SGPV and other aquatic poxviruses beyond salmon farming. The level of production of farmed food fish grew from 13 million to 66 million tons during the period between 1990 and 2012 (70). Epidemics of orthomyxoviral disease virtually stopped salmon production in Chile (71) and demonstrate the importance of disease control in fish farming. Also, for feral fish populations, introduction of new viruses may have serious consequences, as shown by the mass mortalities associated with rhabdoviral disease in the Great Lakes (72). We urgently need more knowledge of fish poxviruses, as the global trade and movement of aquaculture animals are growing. Research on the poxviruses of fishes may also bring cures for the aquaculture industry in the form of vaccines and the development of vectors.

Genome analysis of SGPV established the position of this fish-infecting virus at the base of the chordopoxvirus tree, as could be expected under the assumption of virus-host coevolution. However, the differences between the gene complements of SGPV and those of the rest of the chordopoxviruses are extensive, with SGPV lacking 38 genes otherwise conserved in chordopoxviruses. This difference in gene content is spread unevenly across the functional classes of viral genes. Most of the genes involved in genome replication and expression, as well as core and capsid structure and morphogenesis, are shared by SGPV and other chordopoxviruses. In sharp contrast, the majority of the chordopoxviruses genes implicated in viral membrane biogenesis are missing in SGPV. Chordopoxviruses employ a unique pathway of viral membrane derivation from the membranes of the endoplasmic reticulum of the infected cells that requires the participation of multiple viral proteins (57). The most parsimonious explanation for the absence of most of these proteins in SGPV is that this pathway, at least in its complete form, evolved in viruses infecting tetrapods; however, the alternative scenario, in which the pathway evolved in the ancestral chordopoxvirus but was subsequently lost in SGPV, cannot be ruled out. In addition, the multiprotein complex that in vaccinia virus is involved in viroplasm association with membranes, that is essential for virion maturation, and that so far appears to be conserved in all chordopoxviruses (58) is missing in SGPV. Nevertheless, recognizable poxvirus crescent membranes, immature virions, and mature virions are formed. Thus, SGPV appears to employ a pathway of viral membrane biogenesis similar to that of other chordopoxviruses, despite the absence of key conserved proteins; multiple predicted membrane proteins of SGPV without homologs in other viruses could contribute to this alternative version of membrane biogenesis.

SGPV also lacks the proteins that are involved in the interaction of other chordopoxviruses with host defense systems, such as multiple paralogous genes that encode proteins containing kelch and ankyrin repeats as well as proteins involved in the suppression of host immune mechanisms. The conspicuous exception is the conserved B22R-like giant membrane proteins. However, the SGPV genome encodes numerous uncharacterized genes, many of which encode predicted membrane and secreted proteins as well as predicted nonglobular proteins with low-complexity sequences and simple repeats. Many, if not most, of the protein products of these genes are likely involved in currently unknown interactions with the immunity systems of the fish host. Experimental study of these uncharacterized proteins of SGPV could help the study of the pathogenesis of the gill disease in salmon and, more generally, the mechanisms of the poxvirus-host interaction.