First description of clinical presentation of piscine orthoreovirus (PRV) infections in salmonid aquaculture in Chile and identification of a second genotype (Genotype II) of PRV

In the Atlantic salmon the affected fish had reduced feed intake, were lethargic, and some swum close to the net pen and in the direction of the water current. An additional movie file shows this in more detail (see Additional file 1). Some of the moribund fish lost equilibrium and either recovered the equilibrium or died. In the coho salmon the affected fish had reduced feed intake, which was followed by appearance of dead fish floating in the surface water mainly in the corner of the net pen.

63 Atlantic salmon were analyzed from two farm sites, which were selected from 17 Atlantic salmon farm sites studied, and 85 coho salmon were analyzed from two farm sites, which were selected from nine coho salmon farm sites studied. The farm sites were chosen in order to increase the probability of detecting HSMI disease in Atlantic salmon or HSMI-like disease in coho salmon with the goal to establish a causal relationship between PRV and HSMI. Thus targeted sampling was carried out by an experienced Veterinarian at peak mortality during a suspected outbreak of HSMI or HSMI-like clinical disease in the absence of any pathogen other than PRV, and the samples for histopathological examination were taken from selected fish showing HSMI or HSMI-like gross pathology. All the samples were collected from 2012 to 2015. Table 1 summarizes the frequency of the significant gross pathology findings noted from Atlantic salmon affected with HSMI and coho salmon affected with HSMI-like disease. In Atlantic salmon the most common lesions related to HSMI were pale and yellow liver in 60.3 %, clotted blood in cardiac cavity (haemopericardium) in 28.6 %, and pale gill and pale heart, each, in 19 % of the fish necropsied (Table 1). The representative gross pathology findings in Atlantic salmon affected with HSMI are shown in Fig. 1a and b. In coho salmon the most common lesions related to HSMI-like disease were yellow liver and haemopericardium in 55.3 %, pale heart in 43.5 %, biliary cholestasis in 31.8 %, ascites in 20.0 % and clotted blood in the abdominal cavity in 17.6 % of the fish necropsied. Unlike the Atlantic salmon, some of the coho salmon also had spinal fracture (22.4 %) and kidney rupture (11.8 %) (Table 1). The representative gross pathology findings in coho salmon affected with HSMI-like disease are shown in Fig. 1c and d.

The clinical presentations described in Atlantic salmon and coho salmon are consistent with a circulatory disturbance similar to that described by Kongtorp et al. [15, 17] and Ferguson et al. [16] for Atlantic salmon affected with HSMI, and by Olsen et al. [23] for rainbow trout affected with HSMI-like disease. In the case of coho salmon, we additionally observed spontaneous spinal fracture and kidney rupture. These lesions were observed in coho salmon that had fast growth, with good body condition and abundant visceral fat, and they have been linked to a non-infectious condition of metabolic type. A gross lesion that was more common in coho salmon was jaundice; this had been previously related to Jaundice coho salmon syndrome, a condition previously associated with infectious salmon anaemia virus [25], however all samples in the present study were negative in the RT-qPCR assay for ISAV. Another disease in Chilean coho salmon related with jaundice is infectious haemolytic anaemia [26]. The main histopathological findings of this disease are haemorrhages in all organs and severe haemosiderosis accompanied by erythrophagocytosis in the kidney and spleen. However, none of the coho salmon cases analyzed in this study had haemosiderosis although erythrophagocytosis was seen in the spleen (see Table 2 below). Jaundice was recently described by Garver et al. [11] in Chinook salmon positive for PRV, but the condition could not be reproduced experimentally although it was possible to transmit the virus to the experimental fish.

Table 2 summarizes the nature and frequency of the significant histopathology lesions in tissues of Atlantic salmon affected with HSMI and coho salmon affected with HSMI-like disease. The microscopic lesions in Atlantic salmon are shown in Fig. 2, and those in coho salmon are shown in Fig. 3. In Atlantic salmon the affected fish presented with epicarditis and myocarditis characterized by infiltration of mononuclear cells, and degeneration and necrosis in muscle fibers in myocardium (Fig. 2a). The coho salmon had similar lesions in the epicardium but unlike in Atlantic salmon, myocarditis was present generally only in the spongy layer (Fig. 3a). In Atlantic salmon the red muscle had multifocal to diffuse infiltration of mononuclear cells and in some cases with degeneration and necrosis (Fig. 2b). The inflammatory response in red muscle in coho salmon was similar but mild to moderate in severity (Fig. 3b). In Atlantic salmon and coho salmon the most frequent lesion in the liver was blood vessels surrounded by a mononuclear inflammatory cell infiltrate (perivasculitis) and necrosis (Figs. 2c and 3c). The spleen and kidney in both species had mononuclear infiltrations; however nephritis was more frequently seen in coho salmon. The posterior gut in both species had mononuclear cell infiltration in the lamina propria and sub mucosa. The histopathological changes in the gill were more evident in coho salmon, and presented as multifocal to diffuse lamellar fusion with interlamellar vesicle formation associated to Paramoeba perurans and epitheliocystis-like epithelial inclusions.

Overall the gross and histopathological lesions in farmed Atlantic salmon in Chile are identical to HSMI lesions described in marine-farmed Atlantic salmon in Norway and Scotland [15, 16], which is the classical presentation of HSMI in farmed Atlantic salmon. The histopathological lesions we describe in coho salmon in the present study are different from those reported in Atlantic salmon [15, 16], and in rainbow trout with the new HSMI-like disease in Norway [27]. In the Chilean coho salmon cases, the heart inflammation is diffuse and present generally in the spongy layer; the red muscle inflammation is mild and not always present (Table 2). Cardiomyopathy (CMS) and Pancreas disease (PD) should be considered in the differential diagnosis. CMS is a viral disease of farmed Atlantic salmon caused by piscine myocarditis virus (PMCV) [21, 28] characterized by epicarditis, massive cell infiltration, degeneration and necrosis of the spongy myocardium and inflammatory infiltration of epicardium [29, 30], but is neither associated with liver necrosis nor myositis like the coho salmon affected by HSMI-like disease in the present study. Moreover, the fish were negative in the RT-qPCR assay for PMCV. The second differential diagnosis, PD, in farmed Atlantic salmon and rainbow trout is caused by salmonid alphavirus (SAV) [27, 31]. The characteristic histopathological findings of PD include cardiomyopathy and skeletal myopathy [32], however, the pancreatic lesions are an important morphological change [27]. All samples analyzed in the present study were negative in the RT-qPCR assay for SAV. Thus, the present study reports coho salmon cases with gross pathology lesions not described previously and a marked inflammation of the spongious layer in the heart and inflammation of the red muscle in the affected fish; both findings not previously described. In order to better describe the disease observed in coho salmon, and further link it to PRV, one would have to reproduce the disease experimentally. However, this is out of the scope of this paper, but would be important for future PRV studies.

Although PRV has been reported in Norway, Ireland, Chile, Canada, and USA, to date HSMI lesions with presence of PRV have only been described in marine-farmed Atlantic salmon in Norway. This is the first report of HSMI lesions with presence of PRV in farmed Atlantic salmon outside of Europe, and the first detection of HSMI-like disease with presence of PRV in coho salmon in Chile. The clinical presentation in coho salmon included signs associated with Jaundice coho salmon syndrome. A similar jaundice syndrome, commonly referred to as yellow fish, which occurs in British Columbia-Canada, was also associated with PRV in Chinook salmon, but the condition could not be reproduced using kidney and liver tissues from affected fish and injecting them into naïve Chinook salmon, sockeye salmon O. nerka, Atlantic salmon, and Pacific herring Clupea pallasii, and only minimal viral replication could be demonstrated by RT-qPCR [11]. Most recently in Norway, HSMI-like disease was described in rainbow trout [23]. It is probable that PRV is also infecting rainbow trout in Chile and more work would be needed to establish if the virus is associated with any form of clinical disease. It will therefore be important to include in the surveillance the rainbow trout tissues, and to use the molecular tools used by Olsen et al. [23] for detecting PRV-related virus, because the virus in rainbow trout in Chile may be presenting differently from that in Norway.

All clinical samples in this study taken for histopathological analysis were also tested for PRV by RT-qPCR assay [33]. The 57 pooled tissue samples from Atlantic salmon from two farm sites and 80 pooled tissue samples from coho salmon from two farm sites were positive for PRV. 36 of the 57 Atlantic salmon tested for PRV and 18 of the 80 coho salmon tested for PRV were subjected to histopathology (Table 2). The distribution of Ct values for Atlantic salmon and coho salmon farms is shown in Fig. 4. The distribution of Ct values in both fish species show several samples with low Ct values (<25) that are indicative of high viral loads, which is consistent with the gross pathology and histopathology findings described above in both cases. The Ct values ranged from 16 to 37 for coho salmon and 18 to 30 for Atlantic salmon. The range was narrower in Atlantic salmon (with a median Ct value of 22.7) compared to coho salmon (median Ct value of 25.75). The data for Atlantic salmon are consistent with the generalized additive logistic regression plot that was described by Løvoll et al. [34] showing the relationship between the probability of samples originating from an HSMI outbreak and PRV RT-qPCR Ct values [35]. It shows that Ct values <25 have a high probability of a sample representing an HSMI outbreak.

All the samples tested yielded the expected PCR product of about 1081 bp. All the PCR products were sequenced and the sequences obtained have been deposited in GenBank [24]; their GenBank Accession numbers are shown in the table in an additional file (see Additional file 2). For each sequence, the sample was sequenced two times at CIBA, once by Macrogen Inc, and once in Canada, and in all four cases of a sample the same sequence was obtained. The table in Additional file 2 shows all the PRV segment S1 nucleotide sequences analysed in this study. Among the Norwegian 180 PRV sequences recently reported by Garseth et al. [5], 60 of them are segment S1 sequences. In the phylogenetic tree prepared by Garseth et al. [5], PRV isolates 131 and 1062 do not belong to any group, indicating that these groups might need to be revised soon. These 60 sequences were merged with the PRV segment S1 sequences from Kibenge et al. [8], some other PRV segment S1 sequences deposited in GenBank [24] by other researchers, and the new Chilean PRV segment S1 sequences obtained in this study (Additional file 2), and a phylogenetic tree was constructed using Maximum Likelihood method, with genome segment S3 of avian reovirus strain 176 (GenBank Accession number AF059720) as the out group sequence [8]. The phylogenetic tree obtained is presented in Fig. 5. The same tree was generated using Neighbor-Joining method (data not shown). Although an outgroup was used to determine the root of the tree, the outgroup itself was not shown in the tree so that we have a detailed graphical display of the tree. In this figure bootstrapping values of more than 60 % are shown. It can be seen that the new Norwegian 60 sequences completely support the sub-genotypes Ia and Ib classification of Kibenge et al. [8]; Garseth et al. [5] Groups II, III, and IV belong to sub-genotype Ia (Norway-Canada subgroup) together with three new Chile PRV cases from coho salmon (GenBank Accession numbers KU131591, KUI31592, KU131593). Moreover, these new Chile PRV cases from coho salmon are most similar to Garseth et al. [5] Group II and the Canadian PRV isolates. Interestingly, a PRV sequence obtained from a market-purchased fish sold in Vancouver, BC, as product from Iceland (GenBank Accession no. KT456505) was distinctly different from Canadian PRV isolates and clustered with Norwegian PRV isolates of sub-genotype Ia (Fig. 5). Garseth et al. [5] Group I (which includes the PRV from sea trout) belongs to sub-genotype Ib (Norway-Chile subgroup) together with the new Chile PRV cases from Atlantic salmon (GenBank Accession numbers KU131597, KUI31599, KU131600, KU131601, KU131603, KU131605) and two new Chile PRV cases from coho salmon (GenBank Accession numbers KU131598, KU131604).

Figure 5 shows that all the Chilean PRV strains from Atlantic salmon grouped as sub-genotype Ib, the Chilean PRV strains from coho salmon were more diversified, grouping in both sub-genotypes Ia and Ib and others forming a distinct new phylogenetic cluster as a separate genotype, Genotype II. In contrast, all the Canadian PRV strains reported to date, regardless of fish species source (i.e., Atlantic salmon, coho salmon and Chinook salmon), group as sub-genotype Ia. The Norwegian PRV-related virus (GenBank Accession no. LN680851) is placed together with the Chilean PRV strains from coho salmon (GenBank Accession numbers KU131595, KU131596) making up the Genotype II cluster. Because the LN680851 sequence is only 561 bp, it distorts the accuracy of the phylogenetic tree (data not shown).

It is interesting to note that the full segment S1 sequence (1081 nucleotides) of PRV Genotype II (GenBank Accession numbers KU131595, KU131596) was amplified using the primers and conditions described by Kibenge et al. [8]. The PRV S1 genome segment is bicistronic, encoding the 330-aa Outer clamp (σ3) protein and a 124-aa non-structural protein designated p13 that induces cytotoxicity [36] and may be relevant for virulence of PRV. Therefore the S1 encoded proteins of PRV sub-genotypes Ia and Ib, and Genotype II were compared in detail. Construction of pairwise sequence identity profiles has been used to define genotypes inside species of viruses [28, 37, 38]. Tables 3 and 4 show the percent sequence identities on segment S1 between 23 selected PRV isolates from sub-genotypes Ia and Ib and the three PRV isolates in Genotype II. Both nucleotide and amino acid identity cut-off values of 97.1 % completely correlated with sub-genotypes Ia and Ib and Genotype II in the σ3 gene. In addition, PRV sub-genotype Ib had less genetic variation (nucleotide identity of 99.6–100 %) than sub-genotype Ia (nucleotide identity of 98.0–99.9 %). The two Chilean PRV S1 sequences of Genotype II had 99.5 % nucleotide identity. The nucleotide and amino acid sequence identities between sub-genotypes Ia and Ib strains were both ≤ 96.4 %, whereas between sub-genotype Ib and Genotype II, the nucleotide sequence identity was ≤ 79.2 % and the amino acid sequence identity was ≤ 78.3 %. PRV sub-genotype Ia and Genotype II had nucleotide sequence identity of ≤ 80.9 % and amino acid sequence identity of ≤ 79.0 % (Table 3). Thus Genotype II was more similar to sub-genotype Ia than to Ib in the σ3 gene. In Table 3, sequence homology of Genotype II sequence KU131596 with the most distant sub-genotype Ib PRV sequence (KU131598) at the nucleotide level is 78.7 %, and the amino acid sequence homology of the σ3 protein is 77.2 %. These comparisons did not include the LN680851 sequence, of which only 561 nt or 187 aa of the σ3 gene are available. In a 494 nt overlap and 164 aa overlap, sub-genotype 1a sequence KU160513 with the lowest identities with Genotype II, had 84.6 and 79.9 % nucleotide and amino acid sequence identities, respectively, with LN680851; sub-genotype 1b sequence KU131598 also with the lowest identities with Genotype II, had 83.4 and 79.3 % nucleotide and amino acid sequence identities, respectively, with LN680851; Genotype II sequence KU131596 with the lowest identities with either sub-genotype 1a or Ib, had 97.8 and 98.2 % nucleotide and amino acid sequence identities, respectively, with LN680851. This analysis confirms the phylogenetic analysis (Fig. 5) showing that all the Chilean PRV strains from Atlantic salmon grouped as sub-genotype Ib whereas the Chilean PRV strains from coho salmon grouped in both sub-genotypes Ia and Ib and Genotype II.

The pairwise amino acid sequence analysis of p13 protein is shown in Table 4. The sequence identity within sub-genotype Ia strains was ≤ 96.0 %; within Ib and Genotype II strains were ≤ 99.2 and ≤ 97.6 %, respectively. The sequence identity between sub-genotypes Ia and Ib strains was ≤ 93.5 %, whereas between sub-genotypes Ia and Genotype II it was ≤ 80.6 %, and between sub-genotypes Ib and Genotype II it was ≤ 78.2 %, Interestingly, some sub-genotypes Ia strains also had 78.2 % amino acid identity with Genotype II strains suggesting that there is no amino acid identity cut-off value between sub-genotypes Ia and Ib and Genotype II for the p13 protein.

Figure 5 suggests that PRV sub-genotype Ib was the original virus that gave rise to sub-genotype Ia and Genotype II later. Although the geographical areas the PRV sub-genotypes Ia and Ib covered in Norway and Chile are overlapping (even in the same fish farms), Ib and Ia appear to spread independently as if they were two different viruses. It is interesting to note that to date we and others [11, 12, 14] have not found Genotype II sequence in any fish species in Canada but this sequence (KU131596 with the lowest identities with either sub-genotype 1a or Ib) had 97.8 and 98.2 % nucleotide and amino acid sequence identities, respectively, with the Norwegian PRV-related virus (GenBank Accession no. LN680851) (Tables 3 and 4). The absence of Genotype II sequence in Canada is not related to differences in RT-qPCR methods used to screen for PRV (for example Haugland et al. [33] versus Løvoll et al. [21]) since we were able to detect it in Chilean farmed coho salmon and to amplify the full segment S1 sequence using the same primers and conditions as we use in Canada [8].