Melanized focal changes in skeletal muscle in farmed Atlantic salmon after natural infection with Piscine orthoreovirus (PRV)

Abstract Melanized focal changes in skeletal muscle of farmed Atlantic salmon (Salmo salar) are a major quality problem. The aetiology is unknown, but infection with Piscine orthoreovirus (PRV) has been associated with the condition. Here, we addressed the pathogenesis of red and melanized focal changes and their association with PRV. First, a population of farmed fish (PRV‐negative prior to sea transfer) was sequentially investigated throughout the seawater period. The fish were autopsied and tested for PRV infection. Muscular changes were described by macroscopy and histology, and a classification system was established. Second, in an experimental infection trial, PRV was injected intramuscularly to induce changes. The farmed fish was gradually infected with PRV. Red focal changes occurred throughout the observation period with a low prevalence regardless of PRV status. Melanized changes were highly diverse and their prevalence increased during the trial. Changes of low macroscopic grade and histological category were more prevalent in PRV‐negative fish. Diffuse granulomatous melanized changes only occurred after PRV infection. No muscular changes were observed in the experimentally challenged fish. Our studies do not indicate that PRV infection causes red focal changes, but seems important in the development of granulomatous melanized changes.

The industry has suggested a correlation between red (also known as "bleedings") and melanized focal changes. Compared with the latter, red changes seem rare, but their prevalence is unknown. Presumably, the red changes are located at the same site as the melanized focal changes, which are most frequent in the cranio-ventral part of the abdominal musculature (Mørkøre et al., 2015). Bjørgen et al. (2015) investigated both red and melanized focal changes at slaughter for presence of PRV using both immunohistochemistry (IHC) and RT-qPCR. All samples proved PRVpositive by PCR; however, the virus is ubiquitous in farmed salmon some months after their sea transfer, and as it replicates in erythrocytes (Finstad et al., 2014), the virus can be found in all organs.
Immunohistochemistry demonstrated presence of PRV in macrophage-like cells and in erythrocytes in red focal changes, and in inflamed melanized tissue including even the centre of melanized granulomas. Importantly, transient forms between red and melanized focal changes were also characterized, confirming a link between the two manifestations. Based on their results, Bjørgen et al. (2015) concluded that PRV was a premise for the development of melanized focal changes. Subsequently, these results were challenged by Krasnov, Moghadam, Larsson, Afanasyev, and Mørkøre (2016), suggesting a bacterial component to the pathogenesis, this being based on the detection of transcripts of various prokaryotic rRNA in melanized changes. None of these studies concluded on an initial cause of the condition. The cause of intramuscular bleedings and focal melanization thus remains unknown.
Here, we investigate the sequential development of red and melanized focal muscle changes in a commercial fish population over a period of 15 months. The aim was to reveal the prevalence, severity and anatomical location of red and melanized changes throughout a production period in a seawater farm. Additionally, the melanized changes were systematically classified based on histological differences. We investigated the focal changes for different pathogens with emphasis on PRV, using RT-qPCR, IHC and in situ hybridization (ISH). In addition, we conducted an in vivo experiment trying to induce red and melanized focal changes by injecting PRV in Atlantic salmon. Our studies provide novel insight into the development and nature of a serious quality problem in Atlantic salmon production.  (Table 1).

| Fish and sampling information
At least 600 individuals were autopsied at each time point. In order to thoroughly follow the PRV status of the fish, six minor samplings were conducted in addition to the seven main samplings (E1-E6).
In all of the main samplings, the collected samples included peripheral blood (in heparin) and gill, spleen and muscle samples (on RNAlater) from the first 60 individuals. These samples were investigated for PRV-1 by RT-qPCR. In this work, PRV equals PRV-1. a closer investigation of macroscopic changes and for histological analyses. Representative numbers of the different macroscopic manifestations present at each sampling were selected for in-depth qualitative analysis, including between 5 and 8 control fish (fish with no macroscopic abnormalities). From these selected fish, affected muscle and corresponding unaffected muscle were collected on both formalin and RNAlater. Liver, spleen, head kidney, heart and the sideline containing skin and red and white muscle were collected on formalin. In the additional minor samplings, blood from at least 60 randomly selected fish was collected. All RT-qPCR analyses were performed by PatoGen Analyse.

| Macroscopic registrations
After filleting, the changes were registered according to anatomical location (Figure 1), type and grade of change ( Figure 2). The grade of both red and melanized changes was assessed according to the scoring system used by Mowi, ranging from grade 1 to grade 3, where 1 was very faint discoloration, 2 was a distinct but not severe discoloration and 3 was a prominent and severe discoloration. Grade 1 would probably not have been registered as a quality abnormality at slaughter by some producers, whereas grades 2 and 3 would have implied a quality reduction of the fillet. Representative examples are given in Figure 2. Statistical calculations on the macroscopic registrations were performed with GraphPad Prism © , using Fisher's exact test to calculate the association between sampling groups.

| Histology-standard and special stains
The formalin-fixed samples were dehydrated through graded ethanol baths, cleared in xylene and embedded in paraffin. Sections were made from all collected samples (all organs) and were subsequent to rehydration and stained with haematoxylin and eosin (HE staining). Three red and melanized changes from each sampling were subjected to Gram, Giemsa and periodic acid-Schiff (PAS) staining. Additionally, five active granulomas (grade 9) were stained for mycobacteria (Ziehl-Neelsen's method). All procedures were performed in accordance with Bancroft and Gamble (2008).

| IHC and ISH
Representative changes from each histological category were investigated with IHC for PRV as described previously (Bjørgen et al., 2015), using an antibody targeting the PRV σ1 protein. Briefly, sections were rehydrated and autoclaved. Inhibition was done with phenylhydrazine and blocking with goat normal serum diluted in 5% BSA/TBS (bovine serum albumin/Tris-buffered saline). The primary antibody was diluted in 1% BSA/TBS (dilution 1:700) and incubated for 30 min at room temperature. The sections were further incubated with an anti-mouse secondary antibody (Dako EnVision kit) and developed with AEC to evoke colour (red). PBS was used to wash between each step.
To detect viral RNA in the melanized changes, the slides underwent hybridization with RNAscope probe against a portion of  , 2018). A heart section from PRV-negative fish from a previously published challenge study was also included as a negative control (Finstad, 2014). From the same trial, a heart section from a PRV-positive fish with severe epicarditis was used as a positive control. RNAscope probe against the bacterial gene dapB (#701021) was used as a negative control on duplicates of the same sections to confirm absence of background and/or non-specific cross-reactivity of the assay. In Tank 1, Group 1 was injected intraperitoneally (i.p.) with pelleted blood cells containing high loads of PRV (Lund et al., 2016),

| Experimental challenge study
Group 2 was injected intramuscularly (i.m.) with purified PRV (Lund, 2016), and Groups 3 and 4 were controls given PBS and no treatment, respectively. In Tank 2, Groups 5-7 were given i.p. PRV-infected blood cells, purified PRV and Renibacterium salmoninarum, respectively, that all had been heat-inactivated at 85°C for 25 min.
In Tank 2, Groups 8 and 9 were controls given PBS and no treatment, respectively. The fish were observed for 18 weeks post-challenge (wpc), and six fish of each group were sampled at 3-week intervals. At each sampling, heparinized blood, spleen, kidney, heart and skeletal muscle tissues were collected in RNAlater and in formalin.
The fish were filleted and white muscle examined visually for red or melanized changes. RNA was isolated and RT-qPCR for PRV was performed as described previously (Wessel et al., 2017).

| PRV detection by RT-qPCR
PRV was detected neither in head kidney from 53 fish prior to sea transfer nor in blood from fish from the first three main samplings (I-III) after transfer, that is, during a period of 19 weeks in sea water.

| Macroscopic registrations
Both red and melanized focal changes occurred in all samplings (I-VII). Ninety-two per cent of these changes were in the cranio-ventral and mid-ventral regions of the fillet. Red focal changes macroscopically graded 1 to 3 occurred in all samplings. Their prevalence and severity remained stable throughout the observation period with a prevalence of about 4% for each sampling ( Figure 4). The melanized focal changes first appeared with low frequency and were macroscopically graded as limited severity (grades 1-2). At later samplings, melanized focal changes became more frequent and their macroscopic severity increased ( Figure 4). From Sampling V onwards, melanized focal changes graded 3 were observed ( Figure 4 and Table 2).
A statistically significant increase in the prevalence of melanized focal changes was observed from Sampling I to Sampling II (from 6% to 12%). The prevalence was lower in Sampling III (11%) (not statistically significant, p > 0.05).  The numbers show which categories and which macroscopic grades were present at each of the main samplings (I-VII).

| Red focal changes
Red focal changes from all samplings and with macroscopic grades 1, 2 and 3 displayed tissue changes with some variation, which seemed unrelated to the macroscopic appearance, as all variations described here were in all macroscopic grades. The histological findings varied from severe muscle necrosis with sparse haemorrhage to severe bleedings and severe necrosis. Haemorrhage was both endoand perimysial in all changes. Some changes had scattered intact myocytes among macrophage-like cells and adipocytes (Figure 5a).
Other changes were totally devoid of normal tissue architecture occurring with endomysial haemorrhage, leucocyte infiltrates and activated fibroblasts (Figure 5b). Control muscle from the same anatomical location was devoid of any visible pathological changes ( Figure 5c).

| Melanized focal changes
Melanized focal changes were examined and classified into categories from 1 to 9 indicating the possible sequence at which they develop

| Macroscopic and histological results combined
When combining the macroscopic manifestations with the histological classification, diffuse granulomatous changes were only encountered in macroscopic manifestation grades 2 and 3 (Table 2). Further, in macroscopic manifestations graded 1, individuals with no observed histological changes were present (Table 2). However, grade 1 macroscopic changes displayed big variation in histological appearance, where all categories of histological changes were detected except diffuse granulomatous inflammation (Table 3). In macroscopic changes graded 2, diffuse granulomatous inflammation (category 9) could be detected; however, most of the category 9 changes were classified as grade 3 changes (Table 3).

| Other organs
No pathological changes occurred in the liver, spleen or head kidney at any of the samplings. Sections with pancreatic tissue, peritoneum and pyloric caeca revealed a moderate-to-severe peritonitis in the first two samplings (I and II) with a decreasing severity in the following samplings and a barely detectable level of severity in the last two samplings. This is in accordance with reactions towards intraabdominal vaccination (Mutoloki, 2004). The changes could not be associated with the presence of red and melanized changes. Both affected and apparently unaffected hearts occurred among fish with grade 3 changes.

| Special stains
Gram, Giemsa, PAS and Ziehl-Neelsen staining did not reveal any microorganisms in red or melanized changes.

| IHC and ISH
All granulomatous muscle changes were immunopositive for PRV, in concordance with earlier findings (Bjørgen, 2015), using an antibody targeting PRV protein ( Figure 7A). Other, non-granulomatous, changes could appear both with and without PRV detection. ISH showed positively stained cells in the same locations and a similar distribution pattern as for IHC; however, detection of PRV σ1 proteins was generally more abundant than the detection of PRV RNA ( Figure 7B).

| Experimental challenge study
The Ct values of PRV RNA in blood cells in the fish injected with PRV, that is, Group 1 injected with PRV-infected pelleted blood cells and Group 2 injected with purified PRV, were 22-24 at 3 wpc and approximately 30 at all later samplings. A few fish, mainly in Group 2 (injected with purified PRV), became PRV-negative at 12 wpc. In the cohabitant groups, that is, Groups 3 and 4, there was no detectable virus at 3 wpc, but from 6 wpc and thereafter, all fish in both groups had a Ct of approximately 20 in all samples, with a slight decreasing trend with time. All fish were positive in these groups after 3 wpc.
The fish in the cohabitant groups, that is, those infected by natural route, had higher virus loads than the groups where the virus had been injected. Furthermore, there were no indications of clearing of the virus infection during the experiment for the cohabitant groups.
In the groups of Tank 2, that is, Groups 5-9, where no infective material was administered, there was no detectable virus throughout the experiment.
No melanized focal changes occurred in any of the groups at any sampling after filleting; that is, melanized changes were not induced by experimental infection of PRV either in purified form or as in erythrocytes or as heat-inactivated PRV or heat-inactivated bacteria given i.m. Our results strengthen the hypothesis for a transition from acute red to chronic melanized changes and further support the involvement of PRV in advanced melanized focal changes. In situ hybridization studies on PRV showed a great correlation with previous results (Bjørgen et al., 2015) and argue that PRV both persists and replicates in melanized focal changes, possibly explaining their chronic nature.

| D ISCUSS I ON
Importantly, we also show that red focal changes and less severe melanized changes may occur without PRV presence.
RT-qPCR of blood samples became positive for PRV (2/61) in the fourth minor sampling (E4, 23 weeks post-sea transfer). Thereafter, the prevalence of PRV-positive blood samples increased steadily, and at the end of the observation period, all tested fish were PRVpositive. This indicated that PRV was exposed to the fish population after sea transfer and that it took at least 37 weeks from the initial PRV detection in gills (Sampling II) to widespread distribution and viraemia in the majority of the fish (Sampling VI). PRV belongs to the reoviruses; reo is an acronym for respiratory, enteric orphan, reflecting that these viruses were originally detected in the respiratory and enteric organs and that eventual causation of specific diseases was not evident. Enteric uptake of PRV has been indicated earlier (Hauge et al., 2016); however, these authors found no indication of oral entry as the fish were infected only through the intestine after anal intubation. Our finding that PRV was detected in gills before other sites is consistent with the gill as a port of entry for the virus. However, the Ct values from gills were high at the first positive samplings and one should be cautious to determine a port of entry based on that alone. As blood samples were negative in the Previous studies have reported that unvaccinated farmed fish develop melanized focal changes with the same prevalence at slaughter as vaccinated fish (Berg et al., 2012;Bjørgen et al.,2015;Larsen et al., 2014). However, in our current study, we observed the prevalence at different time points. In Sampling II, the overall prevalence of melanized focal changes more than doubled from Sampling I. The reason behind part of the increase might be attributed to vaccination and peritonitis (Mutoloki et al., 2004). Vaccine-induced side effects (moderate peritonitis) were detected in all fish from Samplings I and II and showed a decreasing severity in the following samplings.
Peritoneal melanization is common in vaccine-induced peritonitis (Poppe & Breck 1997), and one could speculate that this temporary melanization of the peritoneal wall could be registered as a lowgrade melanized change. This fits well with the fact that no severe melanized focal changes (grade 3) were observed at these samplings and that some muscle samples graded 1 showed no histological changes, perhaps indicating only melanization in the peritoneum. We thus believe that vaccination may account for the increased prevalence in Sampling II and that these changes may be of a temporary character without affecting the final prevalence at slaughter.
In some cases, we only encountered melano-macrophages that dispersed among seemingly unaffected myocytes. Their presence most likely explains the observed macroscopic discoloration. Other inflammatory changes were not observed. One may think that such manifestations may indicate the last phases of haematoma clearance.
PRV was not detected in all red focal changes and, therefore, a plausible hypothesis is that in the absence of PRV, the healing process In the current study, 92% of all melanized changes were located in the cranio-ventral and mid-ventral parts of the fillet. It has to be speculated why the majority of changes occur at this site and whether the initial cause might originate in certain anatomical differences throughout the musculature. Interestingly, the percentage of fat content is considerably higher in these locations (47%) as compared to the more dorsal parts of the fillet (9%-18%) (Einen & Skrede, 1998). We did detect a negative imprint of extracellular lipid-like accumulations in both red and melanized focal changes and even in the lumen of blood vessels ( Figure 7). The content of fat in the diet fed to farmed salmon has increased dramatically over the years, from about 15% in the 1980s to 35% today (Tacon, 2005). Concurrently, the amount of marine ingredients has dropped from around 90% in 1990 to 13% in 2013 (Ytrestøyl et al., 2015). The industry reports that the frequency of melanized changes has increased, partly documented in national surveys (Mørkøre et al., 2015). Another important factor reported to have increased over the same period is the occurrence of pancreas disease (PD) which is known to cause changes in white muscle (Mørkøre et al., 2011); however, melanized changes are present with approximately the same prevalence in the northern part of the country where PD has never been diagnosed. The increase in n-6 fatty acids from vegetable oils has caused concerns, and it is well known from humans that a high proportion of n-6 to n-3 fat in the diet shifts the physiological state in the tissues towards the pathogenesis of many diseases: prothrombotic, proinflammatory and proconstrictive (Simopoulos, 2008). In our study, the frequency of inflammatory changes in red and white muscle and myocardium increased with time. This tendency can possibly be associated with the ongoing PRV infection, but also with the growth of the fish and increased content of fat. An important observation from our previous investigation was that a PRV-positive population of fish farmed in in-house tanks was virtually free of both red and melanized changes (Groups F and G in Bjørgen et al., 2015). These fish were considerably smaller than the farmed fish in the other groups as they had a slower growth rate and this may have influenced the storage and distribution of fat in the fillet.
Based on these observations, we speculate that the role of both fat in the diet and its distribution and content in fish should be given attention in future studies on melanized muscle changes.
We tried to establish an experimental in-house model for the induction of melanized focal changes by infecting fish with PRV.
Artificial haemorrhages combined with PRV were induced by intramuscular injection of blood immediately after it was drawn from the same individual. The injection was in the dorsal muscles and not the cranio-ventral muscles and due to practical reasons; the duration of the experiment was 18 weeks, which is much shorter than the production period in sea water. A model for induction of melanized changes would be beneficial for many reasons, but the anticipated multifactorial nature of the development of the melanized changes, including anatomical placement, PRV infection, haemorrhages, myocyte necrosis, duration, feeding procedures and ingredients, should all be considered included (Lund et al., 2018).
The present study has shown that the occurrence of red focal changes in the seawater period seems stable and may turn into melanized focal changes, which accumulate over time. Red and melanized focal changes may be present without PRV being detected, but severe melanized forms were only found in association with PRV infection. The results support that the melanized focal changes in the abdominal region predominantly develop as a consequence of proceeding red focal changes. As these seemingly may occur without PRV, future research should reveal the cause for these changes, and subsequently prevent such changes from developing into chronicity. The prevention of PRV presence and replication seems to be crucial in the latter respect.

ACK N OWLED G EM ENTS
The authors wish to express gratitude to the Global R&D and Technical Department in Mowi ASA for funding and performing the field trial.
The Lura for enthusiastic participation in all samplings. VMD Erik Slagstad is thanked for participation in some of the samplings.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

AUTH O R S' CO NTR I B UTI O N
HB planned the study, sampled material, carried out histological investigations (IHC and ISH) and wrote the manuscript; RH and AK sampled field material and commented on the manuscript; ØO planned the study and commented on the manuscript; DK carried out histological investigations (ISH) and commented on the manuscript; ER planned the experimental study, sampled experimental material and commented on the manuscript; and EOK supervised the study, sampled field material, analysed the results and edited the manuscript.