Piscine orthoreovirus sequences in escaped farmed Atlantic salmon in Washington and British Columbia

The fish tissue source for all samples is detailed in Table 1.

Washington State samples: Tissue samples were obtained from 17 of the Atlantic salmon caught in the vicinity of the escape by gillnet and purse seine gear in the days following the August 19, 2017, escape (WFC-2011-75 – WFC-2011-110, Table 1). Tissues were also collected from: an Atlantic salmon captured in the eastern Strait of Juan de Fuca near Port Townsend, Washington (WFC-2011-13, Table 1), an angler-caught fish in the nearby Skagit River at river mile 46 (WFC-2011-14, Table 1) and three more fish captured in the Skagit River at river mile 46 in late January, 2018 (WFC-2011-111 – WFC-2011-114, Table 1). In addition, whole egg skeins were collected from two of the Atlantic salmon caught in late January, 2018 (WFC-2011-111, WFC-2011-112). One sample of farmed Atlantic salmon labeled “Product of Iceland” was obtained from a retail store in Redmond, Washington State.

All fish were euthanatized by percussion to the head and then placed on ice immediately after capture/purchase and were either frozen or placed on ice and sampled within 24 h. Samples of gill, heart, liver, head kidney, and/or muscle were individually preserved in RNALater™ (Ambion Inc). Prior to obtaining samples of eggs from each egg skein, each skein was individually rinsed in running tap water for 30 s to remove any surface fluids. Samples of eggs were then taken from the interior of each skein and placed in labeled vials containing RNALater™, as for organ tissue samples. The samples were then shipped under a Canadian Food Inspection Agency (CFIA) importation permit to the Atlantic Veterinary College at the University of Prince Edward Island for testing for PRV and other aquatic animal viruses.

BC-Canada samples: Five Atlantic salmon were captured from September 14 – November 1, 2017, in British Columbia between the lower Fraser River and Johnstone Strait (Table 1, Fig. 1). The commercial fishermen who caught them euthanatized the fish by percussion to the head and placed them on ice for 2–3 days. When received, samples of gill, heart, spleen, liver, and head kidney were pooled to loosely fill a 1.5 ml container of RNALater™ per fish, i.e., all tissue from each fish were combined into a single sample which was shipped on ice to the Atlantic Veterinary College at the University of Prince Edward Island.

At the testing lab, each sample was removed from the RNALater™ preservative and was weighed, chopped into small pieces and directly macerated to a 10% suspension (w/v) in L-15 medium (Winset Inc) with 10x antibiotics/antimycotic (Life Technologies Inc), except for egg samples which were directly macerated in Qiazol (QIAGEN), using Green MagnaLyser beads (Roche Diagnostics). The tissue homogenate was immediately used for total RNA extraction.

In the present study, there was no opportunity to collect back samples of Atlantic salmon organs or eggs from the same environment because the presence of escaped farmed Atlantic salmon in Washington State and BC-Canada was a transient event and did not appear in similar fisheries the following year, 2018.

Total RNA was isolated using a modified total RNA extraction protocol with Trizol Reagent (Life Technologies Inc) and RNeasy mini Kit (QIAGEN) as previously described [6]. Total RNA from eggs was extracted using the RNeasy Lipid Tissue mini kit (QIAGEN) which includes Qiazol. RT-qPCR was run on the LightCycler 480 (Roche Applied Science), version 4.0. The crossing point (Cp) or threshold cycle (Ct) was determined by use of the maximum-second-derivative function on the LightCycler software release 1.5.0. The OneStep RT-PCR kit (QIAGEN) was employed for all RT-qPCR reactions according to the manufacturer’s specifications. Sample RNA quality was based on RT-qPCR for elongation factor 1 alpha (ELF-1α) as internal control targeting Atlantic salmon ELF-1α carried out using Roche LightCycler 480 RNA master Hydrolysis Probe kit (Roche Diagnostics). The primers and probes and reaction conditions used were as previously described [37] and were designed to specifically amplify and detect selected ELF-1α coding sequences in Atlantic salmon, Coho salmon and rainbow trout [37]. The ELF-1α assay was performed in 2 wells per sample, and a sample with Ct value of ≤22.0 was considered of acceptable RNA quality [38]. The RT-qPCR assay for PRV used the primer-probe set sequences and reaction conditions as previously described [6] with minor modifications. Total RNA was denatured at 95 °C for 5 min and immediately placed on ice before use in the RT-qPCR. The PRV assay was performed in 1 well per sample, and samples to be considered PRV-positive had Ct values up to 39.9 and with an exponential curve; Ct values between 40 and 45 were considered suspicious, and a sample was negative if there was no Ct value. The analytical sensitivity of the RT-qPCR assay for PRV was previously determined through serial 10-fold dilutions of an in-vitro transcribed RNA (cRNA) of PRV segment L1 cDNA clone L1-C [6] from 10¹¹ to 10⁰ copies/μl, and had a detection limit of 1 copy. The standard curve based on the serial 10-fold dilutions in 4 replicates at each dilution had a slope of − 3.437, and the Efficiency was 1.954. The cut-off was adjusted to the minimum copy number detected by the assay at Ct value of 40.0 [39]. The positive control for the PRV RT-qPCR assay was the cRNA used at 10^− 4 dilution, which gives a low Ct value (~ Ct of 20). The negative control (no-template-control) was water. Ct values of individual tissue samples from the farmed Atlantic salmon, as tested, were visualized through a box plot showing the median and the quartiles of the different tissue samples. One Way Analysis of variance (ANOVA) [40] was used to analyze differences in Ct values among the different tissue samples.

Selected samples with positive Ct values in the PRV RT-qPCR assay were tested in conventional RT-PCR targeting the full-length genome segment S1 with the PCR primer pair and reaction conditions as previously described [6]. The PCR products were cloned into the pCRII vector using a TOPO TA cloning kit (Invitrogen) in preparation for nucleotide sequencing. Plasmid DNA for sequencing was prepared as described before [41]. DNA sequencing was performed as previously described [42] by ACGT Corporation (Toronto, Ontario, Canada). DNA sequencing was done on plasmid DNA containing the cloned RT-PCR products obtained from reactions using total RNA from tissue samples. Similarity analysis of the obtained DNA sequences was performed using BLAST programs available via the National Center for Biotechnology Information [43]. Analysis to identify putative ORFs and their predicted amino acid sequences was conducted using the Sequence Manipulation suite, version 2 [44].

All PRV S1 segment sequences used in the phylogenetic analyses are described in an additional file in more detail (see Additional file 1: Table S1). The PRV sequences were obtained from selected samples from the escaped farmed Atlantic salmon from the Deepwater Bay, Cypress Island Facility in Puget Sound, Washington State (13 Washington State samples and 2 BC-Canada samples), and from the sample of market-bought fish labeled “Product of Iceland” obtained from a retail store in Redmond, Washington State. A PRV S1 segment sequence WFRC Case-18_01_Sample 7 obtained from separate samples by Western Fisheries Research Centre (WFRC) was included in the analysis. All other PRV sequences were obtained from the GenBank Database [45]. Sequences were processed using ClustalX 2.1 [46]. The multiple sequence alignment was manually verified and adjusted to reach high quality alignment. The phylogenetic trees were generated when positions with gaps were excluded and corrections for multiple substitutions were used. Bootstrapping was performed for 1000 times. In all cases, only the bootstrapping supports higher than 70% were noted. To verify the evolution direction, outgroup sequences were used to determine the root of the phylogenetic trees.

Clark et al. [28] estimated that of the 305,000 farmed salmon in the Deepwater Bay net pen at the time of the catastrophic failure a median number of 253,000 escaped to the marine waters of Bellingham Bay, nearby rivers, and beyond. To estimate the probability that the total number of escaped fish that were infected with PRV was K, given that a total of 253,000 fish escaped, we used our 27 random samples tested for PRV and the 4 fish sampled by Washington Department of Fish and Wildlife as reported in Clark et al. ([28], pp.98), all of which tested positive for PRV (total = 31) in a Bayesian analysis using a beta-binomial likelihood (equation [1]).

(1) P(K|N, S, R) ~ BBN(N, S, R)

where K = the total number infected with PRV, N, the total number escaped (from which samples were obtained (253,000), S = the total number of (random) samples tested for PRV (31), and R = the total number of samples (S) that tested positive for PRV (31), where BBN is the beta-binomial likelihood, N, S, and R are known and K is the unknown to be estimated. The purpose of this analysis was to help estimate the threat that this large escape may pose to wild Pacific salmon and steelhead.

The weight and fork length of the Atlantic salmon caught by fishermen in Washington State and BC-Canada ranged from 1.8–5.4 kg and 62–77 cm, respectively. This is the approximate size range of the Atlantic salmon reported in the Deepwater Bay net pen facility prior to the escape ([28], pp.86–7).

Twenty-eight escaped farmed Atlantic salmon, 23 from Washington State and 5 from British Columbia, and 1 market-bought fish labeled “Product of Iceland” were tested for PRV RNA. For the Washington State fish, tissue samples including gill, heart, liver, spleen or head kidney and eggs (from two fish) were preserved separately and tested individually, whereas samples of gill, heart, spleen, liver, and head kidney were pooled for each of the BC-Canada fish (Table 1). In total, 74 samples were tested for PRV using real-time RT-PCR with TaqMan probe for PRV segment L1 [6] and confirmatory sequencing of PRV segment S1 was done on a subset from these RT-qPCR positive samples. The Ct values for the ELF-1α gene for all samples ranged from 16.91 ± 0.12 to 23.86 ± 0.24, with all samples except ten (from 9 fish) having a Ct value of ≤22.0, which indicated acceptable sample RNA quality [38]. In our experience, the RT-qPCR assay for PRV works well even for samples with ELF-1α Ct values as high as 26.0.

Table 1 summarizes the Ct values of the PRV segment L1 as an indication of PRV load for all samples. All the farmed Atlantic salmon analyzed from Washington State and BC-Canada were positive for PRV (i.e., contained PRV sequences of genome segment L1), as did the purchased sample marketed as “Product of Iceland”. Only one sample of gill tissue from an otherwise PRV-positive fish (WFC-2011-13) was PRV-negative. All 16 samples tested by conventional RT-PCR targeting full-length PRV segment S1 were also positive and generated high quality DNA sequence, confirming PRV infection of the fish that escaped from the Deepwater Bay, Cypress Island facility.

The distribution of Ct values among the different tissues tested in the PRV RT-qPCR assay is shown as a box plot (Fig. 2). Outliers were only evident in the gill samples of WFC-2011-14 (35.60 Ct) and WFC-2011-13 (0 Ct). Both fish had generally high Ct values in corresponding heart and head kidney/liver samples (Table 1). One Way Analysis of variance (ANOVA) [40] showed that differences in Ct values among the different tissue samples were not significant (p = 0.1121), which ruled out cross-contamination among samples. The Ct values for the egg samples, WFC-2011-111 and WFC-2011-112, were 31.10 and 34.78, which was higher than for the corresponding heart samples (Table 1). Egg sample WFC-2011-111 (31.10 Ct) contained enough PRV genetic material to yield full-length S1 sequence, suggesting that the egg was infected with PRV and not simply contaminated with trace amounts of virus during handling. PRV-positive WFC-2011-112-egg sample (34.78 Ct), did not yield a product in conventional RT-PCR for S1. The Ct values of positive control of n = 5 (20.42 ± 0.49) and negative control of n = 5 (0.0) for the PRV RT-qPCR assay were not included in the boxplot of Fig. 2. The Ct value of positive control was significantly different from the fish tissues (p = 0.0026).

While eggs from PRV-positive broodfish may test PRV-negative under commercial hatching conditions, 25% of the hatchlings may test positive at very low levels close to the limit of detection and then 100% negative by first feeding [26]. Five out of fourteen hatcheries tested in Norway had PRV-positive pre-smolts, some with Ct values < 25 [3]. Our data confirms previous reports that it is possible to detect PRV RNA in Atlantic salmon eggs. We provide evidence that PRV concentration in eggs may be lower than in other tissues, and that PRV may be vertically transmitted. Since RT-PCR can give false positive results, a second method is normally used to confirm the results. In case of PRV, only RT-PCR methods are available and used for diagnostic testing. In the present study, both RT-qPCR and conventional RT-PCR were used; RT-qPCR targeted L1 segment whereas conventional RT-PCR targeted S1 segment. The positive RT-qPCR results on selected samples were then confirmed by conventional RT-PCR and sequencing of the PCR product, which also allowed for genotyping of the PRV present in samples. The fact that the PRV detected was related to Icelandic PRV shows that the results obtained by our RT-PCR tests are correct.

The 19 Atlantic salmon caught in Washington State that were sampled in between August 27 and December 31 (Table 1) were also tested for infectious haematopoietic necrosis virus (IHNV), viral haemorrhagic septicaemia virus (VHSV), and infectious salmon anemia virus (ISAV), and were RT-PCR negative for these viruses (data not shown). VHSV and ISAV-HPR0 have been reported in Iceland [47, 48].

The Atlantic salmon sampled in this study and the four sampled by WDFW reported in Clark et al. [28] (total = 31) can reasonably be considered a random sample from the total estimated number of escapees (253,000) due to the range of fishing gear types applied and non-targeted selection of specimens from the fishermen’s catches. Sampling was opportunistic and did not select for size or condition. These fish can therefore be considered exchangeable with the remaining number of the 253,000 escaped fish that by chance were not sampled. We used these numbers in the beta-binomial likelihood to provide a conservative estimate of the prevalence of PRV in the Deepwater Bay farm equation [2]: (2) K ~ BBN (253,000, 31, 31).

The central 95% of the posterior probability distribution of K spanned the interval [230,000, 252,500] (Fig. 3). The mode of distribution is 252,000; the mean is 245,000. Expressing these numbers in terms of the probability (P) that a randomly sampled individual is infected with PRV, the posterior distribution of P spanned the interval [0.91, 0.997], with mean = 0.97 and mode = 0.997. For all practical purposes, all of the fish that escaped were infected with PRV.

Prior to this study, there was only one Icelandic PRV S1 sequence in the GenBank, Accession # KT456505, PRV isolate VT01212014–09 from market-bought fish in BC-Canada labeled “Product of Iceland” [25]. A total of 15 full-length sequences of PRV S1 segment were obtained from the samples of farmed Atlantic salmon tested in this study. Full-length sequence was also obtained from the sample of market-bought fish labeled “Product of Iceland” obtained from a retail store in Redmond, Washington State. We used both Maximum Likelihood and Neighbor Joining methods in phylogenetic analysis and found that they both produced similar trees. The basic sub-genotypes Ia, Ib, and genotype II [5] are all conserved in both trees. It should be noted that the two groups of sub-genotype Ia in the Maximum Likelihood trees are very close and should be considered as one group although visually they appear apart from each other. Also, the bootstrapping values in the Maximum Likelihood trees are very similar to those in the Neighbor Joining trees. Here we show only the trees generated using Maximum Likelihood method. Sequence analyses showed the 16 PRV segment S1 sequences belong to PRV sub-genotype Ia. A phylogenetic tree of these sequences including PRV S1 segment sequence WFRC Case-18_01_Sample 7 obtained from separate samples by WFRC, and selected PRV S1 segment sequences available in the GenBank database (as indicated in Additional file 1: Table S1) is shown in Fig. 4. Thus 107 PRV S1 segment sequences out of the 175 sequences analyzed in this study are included in this phylogenetic tree. All the 175 PRV S1 segment sequences are shown in the phylogenetic tree in an additional file in more detail (see Additional file 2: Figure S1). Although the root of each tree was determined using an outgroup sequence, we did not include the outgroup itself in Fig. 4 and Figure S1, in order to focus on the portion that matters to this analysis. All sequences from the escaped farmed Atlantic salmon fit well inside PRV sub-genotype Ia, and are very similar to GenBank Accession numbers KT456505 (PRV isolate VT01212014–09 from market-bought fish in BC-Canada labeled “Product of Iceland” [25] and HG329893 (Norwegian PRV isolate “195 Aaroy 2007 Wild” [4]). Thus, the PRV strains present in the escaped farmed Atlantic salmon were very similar to PRV in farmed Atlantic salmon in Iceland, the reported country of origin of the Atlantic salmon eggs used in commercial aquaculture in Washington State. This finding of Icelandic PRV in egg samples taken from the escapees in Washington State completes the infection cycle of PRV in farmed Atlantic salmon. It is also the first direct evidence of international movement of PRV via Atlantic salmon eggs.

The sub-genotype grouping in Additional file 2: Figure S1, which includes all PRV isolates, is contrary to the observations by DiCicco et al. [13] that there are some sequences that lie between sub-strains (e.g., JN991012.1, JN991007.1) and therefore the distinction into PRV-1a and 1b is not helpful. It is clearly evident in Additional file 2: Figure S1 and Fig. 4 that Norwegian PRV isolates 1921-S3 (GenBank Accession number JN991007) and 3817-S3 (GenBank Accession number JN991012) of Løvoll et al. [3] belong to sub-genotype Ia, which is distinct from the Norwegian PRV isolate Salmo/GP-2010/NOR (GenBank Accession number GU994022) [1] utilized by DiCicco et al. [13] as the reference sequence PRV-1a but belongs to sub-genotype Ib. At the time of writing this paper only two of these PRV-1a sequences, isolates B5690 and B7274 (GenBank Accession numbers KX851970 and KX851971, respectively) [10] were available in the GenBank Database, and both belong to sub-genotype Ia (Fig. 4). The Norwegian PRV isolate NOR2012-V3621 (GenBank Accession number KY429949) used to reproduce HSMI in an Atlantic salmon challenge study [49] belongs to sub-genotype Ib (Fig. 4). Additional detailed data are in Additional file 1: Table S1 and Additional file 2: Figure S1. Thus, this phylogenetic analysis shows that the second genotype, genotype II, identified by Godoy et al. [7] among PRV isolates from Norway and Chile also includes PRV-2 found in Japan [8] and several PRV isolates (designated PRV-3) found in rainbow trout in Norway and several European countries and in brown trout in Italy [16] and France (Bigarre 2016, unpublished observations) and in rainbow trout [17] and Coho salmon in Chile [18]. We therefore further provide strong support for classification of PRV into two genotypes (I and II) and four sub-genotypes (Ia, Ib, IIa, and IIb).