The Atlantic salmon (Stofnfiskur Optimal strain) used in the study were reared from hatching at the VESO Aqualab’s hatchery (Fosslandsosen, Norway). Before the onset of the study, blood plasma was collected from 36 representative fish averaging ~20 grams, all of which tested negative for the presence of circulating antibodies using standard enzyme-linked immunosorbent assay (ELISA). The ELISA assays were based on methods described earlier for Aeromonas salmonicida subsp. salmonicida (28, 29) with exchange of the antigen coat on the ELISA plates to include Vibrio salmonicida, Vibrio anguillarum serotype O1 and 02, V. ordalii, M. viscosa and Yersinia ruckeri. Of the 36 fish analyzed by ELISA, 10 were further analyzed for viral RNA using validated and accredited (Norwegian Standard NS-EN ISO/IEC, 17025) RT-qPCR methodologies (PatoGen AS, Ålesund, Norway). Based on heart samples (see details in section 2.7), the fish were found negative for infectious salmon anaemia virus (ISAV), piscine myocarditis virus (PMCV), piscine reovirus (PRV), SAV and, based on kidney samples, these fish were also found negative for infectious pancreatic necrosis virus (IPNV). Healthy parr without visual deformities were size-graded and anaesthetized using tricaine mesylate at 200 mg/L for 40 seconds (Finquel vet., ScanVacc). The fish in tank A ( Figure 1 ) were inserted intraperitoneally (i.p.) with a passive integrated transponder (PIT) tag and registered in VESO Aqualab’s database. Two weeks later, the fish were anaesthetized again and injected with vaccine combinations or sterile physiological saline (Saline) as listed in Table 1 . The licensed vaccines used were administered in accordance with their respective summary of product characteristics (SPC). The weights of the PIT-tagged fish in tank A were registered and scanned during the vaccination process to register individual and treatment group identity. The fish in tank B were marked on a treatment group level only; as naïve fish by adipose fin-clipping (AFC), or by removal of the right- (RM) or left maxillary (LM) for their immunized counterparts as detailed in Table 1 . The fish in the two tanks were transferred separately 13 weeks later to VESO-Aqualab’s experimental test facility (Namsos, Norway).

The fish were reared at 12 ± 1°C in the hatchery in two 1.5-meter diameter tanks. At the experimental facility, the fish were maintained at 13 ± 1°C in either a 2-meter (tank A) or a 1.5-meter (tank B) tank with tube overflow system ( Figure 1 ). Flow rates were adjusted to ensure that oxygen saturation levels near the outlet remained ≥70%. Removing dead fish and cleaning the tanks was done daily. The fish were starved for minimum 24 hours prior to handling or sampling. All fish were kept sedated to minimize stress during sampling using AQUI-S VET (isoeugenol, MSD Animal Health) according to the product’s label specifications. Moribund and sampled fish removed from the tanks were euthanized using an overdose of benzocaine chloride. By use of automated feeders, the fish were fed a standard commercial extruded pellets without any immunostimulant supplements (Skretting) throughout the study. Feeding rates were 2% body weight per day prior to challenge in Tank A and for the fish in the 10 tanks used in the transmission study. The fish in Tank A were fed ad libitum during the 84-day challenge period. Standard photomanipulation was applied using 12-hours light and 12-hours darkness (12:12) exposure for 6 weeks followed by continuous 24-hour light exposure (24:0) for another 6 weeks before seawater exposure (salinity maintained at 32 ± 3‰). The photomanipulation process was timed so that the fish in tank A were ready for exposure to seawater 1-2 weeks earlier than their counterparts in tank B. All fish were handled in accordance with Norwegian “Regulation on Animal Experimentation”, and the study protocol was approved before start-up by the Norwegian Animal Research Authority (FOTS ID, 27332).

Fish in Tank A were sampled 22 days before the challenge, still in freshwater, 1030 degree-days (dd) post vaccination ( Table 2 ). The challenge was carried out, 1476 dd post vaccination, 9 days after seawater exposure. Briefly, a total of 200 naïve AFC smolts residing in freshwater in tank B were anaesthetized and injected i.p. with 0.1 ml of SAV2 inoculum derived from a PD outbreak in Romsdal, 2011 (Isolate 1; Taksdal et al. (7), GenBank ref. HE863664, MZ395641) containing 3.52 × 10⁴ TCID₅₀/ml, and dropped into tank A containing seawater. These fish (shedders) represented ~19% of the total number of fish in Tank A at the onset of the challenge period (0 days post challenge (dpc)). Dead and moribund fish were removed daily, and their PIT-tag identity scanned and registered in the database throughout the study. Heart samples were collected from all dead PIT-tagged fish between 27 and 53 dpc, and analyzed for relative quantities of SAV RNA by RT-qPCR. Head-kidney smears from the dead PIT-tagged fish were cultured on blood agar with 2% NaCl and incubated at 22°C between 48 and 96 hours. Culture growth was evaluated to detect possible bacterial causes of mortality. Samples were collected at 19, 55 and 84 dpc as shown in Figure 1 ; Table 2 . The cumulative percent mortality during the challenge period was calculated with the denominator adjusted based on the number of fish removed due to samplings at 19 and 55 dpc ( Table 2 ) and by transfer to the 10 tanks used for the transmission studies at 19 dpc ( Figure 1 ).

The experimental design and sampling schedule employed in the transmission studies are illustrated in Figure 1 . At 19 dpc, fish from each treatment group, hereinafter referred to as the ‘pre-challenged fish’, were transferred from tank A into 10 tanks of 1-meter diameter with equal flow rates. At 21 dpc, non-infected fish from tank B, either as naïve (TS1) or vaccinated fish (TS2), were transferred to six of the shedder tanks as shown in Figure 1 . These fish are hereinafter referred to as ‘naïve’ or the ‘vaccinated cohabitants’. With 40 fish in six shedder tanks (1-6) and 20 fish in the remaining four tanks (7-10) at 21 dpc, flow rates were biomass adjusted in the 10 tanks to 1.56 liters/kg fish/minute. As this flow rate resulted in oxygen saturation levels near the tanks outlets of 62%, the flow rate was increased at 23 dpc to 1.86 liters/kg fish/minute in the 10 tanks and until the end of the transmission studies at 47 dpc. The sampling regime of the transmission studies is detailed in Table 3 .

For the neutralization test, blood was collected from the caudal vein of euthanized fish using heparin-coated vacutainers and placed into crushed ice immediately thereafter ( Table 2 ). After centrifugation at, 1000 x g for 10 minutes, plasma samples were retrieved and stored at -80°C until use. The neutralization test was performed as previously described (30) with some modifications. CHSE-214 cells, derived from Chinook salmon (Oncorhynchus tshawytscha) embryo (ECACC CB_91041114), were grown as recommended in section 4.3 of the WOAH manual (https://www.woah.org/fileadmin/Home/eng/Health_standards/aahm/current/2.3.08_SAV.pdf). Fully confluent cells were trypsinized, seeded in a 96-well cell plate, and grown until 80% confluency. Starting with 1:20 dilution, two-fold dilution series of plasma were incubated with SAV2 (same isolate as used in the challenge – see section 2.3) for two hours and then seeded in two parallel wells in the prepared CHSE-214 96 well plate. After 3-4 days of incubation at 15°C, the cell layer was fixed with 80% acetone. SAV-infected cells were visualized using an indirect immunofluorescence test according to the procedure described by Falk et al. (31), but with the use of monoclonal antibody 17H23 directed against the E2 glycoprotein of SAV (32) as the primary antibody and with biotin labelled goat anti-mouse Ig (RayBiotech) and FITC-labelled streptavidin (Invitrogen) as the secondary amplification step. The number of positive cells were counted using a fluorescence microscope. Neutralizing activity was defined as present when more than a 50% reduction in infected cells relative to control wells was observed, as previously described (33). Neutralizing activity in plasma diluted ≥1:20 was recorded as a positive result.

Intraperitoneal side-effects were evaluated in a blinded manner and the degree of visceral adhesions scored on a progressive scale from 0 (no adhesions) to 6 as previously described (34). The number of fish per group used for the side effect evaluations is shown in Table 2 .

In the transmission studies, water samples were collected at 3 time points ( Table 3 ) 5 minutes after the tank’s inlet flow had been turned off. The water was collected using, 1000 ml polyethylene terephthalate bottles (VWR International, LLC), and shipped refrigerated overnight for next-day processing. The relative amount of SAV2 RNA in water samples was quantified as previously described using an RT-ddPCR (RT-droplet digital PCR) method (26). In short, 1000 ml of water from each tank and timepoint was filtrated through a negatively charged nitrocellulose MF hydrophilic membrane filter (MF-Millipore^®) using a peristaltic pump at a flow rate of 200 ml/min. The filter was placed in 2,4 ml lysis buffer in a 60 mm petri-dish and put on a shaker for 30 min at 600 rpm. The membrane filter was removed, and the buffer-sample frozen at -80°C until further analysis. To purify nucleic acids, SAV RNA was extracted using a fully automated MagNA Pure 96 Instrument. The extracted RNA was analyzed by RT-ddPCR to quantify the amount of SAV RNA present.

The RT-qPCR (PCR) analysis measured the relative amounts of SAV RNA in the hearts of fish ( Tables 2 , 3 ). In short, a small tangential portion of the heart (~2 × 2 mm) was cut along the sagittal plane and placed into a tube containing RNAlater (Thermo Fischer Scientific, Waltham, MA, USA). Samples in RNAlater were stored overnight at 4°C and then frozen at -80°C until use. The PCR analysis was carried out using a validated and ISO17025-accredited method (Patogen AS, Ålesund, Norway) previously described (35). The cut-off Cq-value was set to 37.

The TCID₅₀ of SAV in plasma samples was determined as previously described (36) with some modifications. In brief, ten-fold dilution series of plasma were seeded on CHSE-214 cells in four parallel wells (96 well plates). The cell layer was fixed with 80% acetone after 7 days of incubation at 15°C. SAV-infected cells were visualized using an indirect immunofluorescence assay as described above. The TCID₅₀ titres were calculated as described by Kärber (37).

Formalin-fixed samples of heart, pancreas, red and white skeletal muscle were processed for paraffin embedding. For each fish, a single sagittal section was obtained through the heart, which included the ventricle, atrium and bulbus arteriosus. To evaluate the pancreas, a single transverse section was acquired through the pyloric ceca; each of these sections invariably contained multiple islands of exocrine and endocrine pancreatic tissue. Skeletal muscle samples from the lateral line region were microtomed to provide one transverse section and two longitudinal sections per fish, each containing red and white skeletal muscle. Histologic sections (4-6 µm thick) were mounted on glass slides and stained with hematoxylin and eosin using standard methodology. All slides were examined in a blinded manner using brightfield microscopy at various magnifications (20x – 400x) by an experienced anatomic pathologist, certified by the American College of Veterinary Pathologists. Histopathological changes associated with SAV infection were recorded as previously reported (11) and detailed in Table 4 . Each characteristic was scored for severity using a 0-3 scale: 0 = not remarkable, 1 = mild changes, 2 = moderate changes, and 3 = severe changes. Representative pictures and descriptions of the histopathological changes (Grades) for heart, pancreas and muscle used in this study are included as Supplementary Figures S1 – S3 .

Initial analysis was carried out using Pivot tabulations and graphs in Excel^®. The data were transferred and analyzed using Stata/MP 17 for Windows (StataCorp, College Station, TX). The weight gain prior to the challenge was analyzed using ANOVA and a linear regression platform with the treatment group and start weight as explanatory variables. Weight gain post challenge was analyzed using linear regression. Start length was also included as a predictor but dropped as no influence was found. The intraperitoneal side effects levels of the immunized were analyzed using Fisher’s exact test. Only the DNA-PD group produced neutralization titers against SAV2 prior to the challenge while all the groups had high and similar titers at the end of the challenge period (84 dpc). Statistical analysis of this data was therefore deemed unnecessary.

Median regression analysis was used to compare SAV viremia levels and RT-qPCR Cq-values of the hearts. The concentrations of SAV RNA in water between the groups at the first timepoint (21 dpc) were also compared using median regression analysis. The cohabitant fish thereafter added to tanks 1-6 rendered the subsequent SAV RNA in water data limited to descriptive interpretation.

The histopathology data were first examined using tabular and graphical techniques followed by ordinal logistic regression analysis. Two analyses were undertaken by first splitting data across cohabitant (naïve and vaccinated) and pre-challenged fish estimating only vaccine effects. As the different cohabitant combinations influence the infectious dynamics, a second analysis was carried out splitting the data across cohabitant patterns (naïve + pre-challenged, vaccinated + pre-challenged and pre-challenged only) and estimating the effects of vaccine and cohabitants. These analyses are complementary. Linear and median regression models are presented using coefficients with 95% Confidence intervals (95% CI) and corresponding p-values. Ordinal regression models are presented with Odds ratios with 95% CI and corresponding p-values. The term significant in the text refers to a p-value <0.05.

The mean weight of all the fish in tank A (n=1073) at vaccination was 36.4g ±3.6g (min/max 28.0/46.4). When weighed 22 days before the challenge (after, 1030 dd), the Saline group had gained more weight than the immunized groups ( Figure 2 ), as expected when oil-adjuvanted vaccines (OAVs) are used. No significant difference in weight gain (n=75) or visceral adhesions scores (n=50) was found at this time between the immunized groups with mean adhesions scores of 1.56, 1.60 and 1.44 for the DNA-PD, Oil-PD and Control groups, respectively. These side effect levels are within the range reported in the SPC of the OAVs used. Before the challenge, plasma neutralization titers of SAV2 were only observed for the DNA-PD group, (n=30) with 83% prevalence and titers ranging from 1:20 to 1:640 ( Supplementary Figure S4 ). A total of 17 fish (1.6%) died during the immunization period until the challenge, but there were no indications of infectious disease.

No significant differences in weight were found between the treatment groups at either 19 or 55 dpc, while at 84 dpc, the DNA-PD group had significantly outgrown the other groups. With no differences in weight or i.p. side effect scores registered between the immunized groups before the challenge, the significantly higher weight of the DNA-PD group at 84 dpc was likely caused by the difference in protection against PD-associated growth loss as found in similar experiments for SAV3 (11, 12). This finding aligns with the qPCR results of the hearts at 84 dpc with the DNA-PD group harboring significantly less SAV RNA than the other groups, while high and similar levels were measured in all the groups at 55 dpc. Lower cardiac histopathology scores in the DNA-PD group at 84 dpc indicate less impact on heart health in this group. The pathologies of the pancreas at this timepoint were, in contrast, very similar between the groups and markedly milder than found in similar SAV3 studies. Based on the above, the effect against PD-associated growth loss and clearance of SAV2 RNA was significantly higher in the PD-DNA group than the other groups.

Cumulative mortality was low and similar across treatment groups (9.1 to 12.6%) although time to death was notably delayed in the DNA-PD group compared to the other treatment groups. The Cq-values in the hearts of the fish that died between 27 and 53 dpc showed markedly less SAV RNA in the DNA-PD group compared to the other groups. While the underlying mechanisms have not been studied, it can be speculated that these findings result from a higher vaccine efficacy in the DNA-PD group, as supported by the comparative growth and histopathology results. Why the end-point mortality in the Saline group was lower than the Oil-PD and the Control groups is puzzling. The cumulative mortality in the Saline, Oil-PD and Control groups was similar until about 40 dpc, after which the mortality curve for the Saline group flattened while the mortality of the Oil-PD and Control groups was still risimg. There were high SAV2 neutralization titers in all groups at 84 dpc, and one possibility is that the immunity induced by the challenge virus occurred earlier in the naïve Saline group than in the Oil-PD and Control groups. Still, additional time-course studies would be required to understand this.

The SAV2 challenge isolate employed in this study was previously used in a similar challenge model shown to cause markedly less mortality compared to SAV3 (7). Therefore, the mortality levels found in the Efficacy study were higher than expected, and similar to those found in previous vaccine efficacy studies using a SAV3 isolate and the same cohabitation challenge model (11, 12) as employed in this study. In contrast to these studies including a SAV3 challenge, in which 96% (134/140) of the fish were viremic at 19 dpc, only 16% (13/80) were viremic at 19 dpc in the present study. Only 3 hearts of the 13 viremia-positive fish were PCR-positive. This finding agrees with results from an earlier SAV3 infection study demonstrating viremia appearing for a short period before any viral RNA can be detected in the hearts (41). The delay in the onset of the SAV2 infection compared to the aforementioned SAV3 studies is likely due to lower virulence of the SAV2 as previously reported for this genotype (7), as also seen during field outbreaks in Norway (42). Further, the challenge dose each of the shedder fish received in this study was considerably lower (3.52 x 10³ TCID₅₀) than that used in the SAV3 studies (1.26 x 10⁴ TCID₅₀). This may also have contributed to the delayed onset of the SAV2 infection compared to SAV3 challenges (11, 12).

With 94% (75/80) of the fish hearts positive for SAV by PCR at 55 dpc, and after that reductions in the SAV RNA in all the groups at 84 dpc, suggests that the peak in the viremic phase occurred between 19 and 55 dpc. This assumption is supported by the absence of heart and pancreas necrosis and pancreas tissue loss at 19 dpc. Additional evidence supporting this assumption includes the degree of the heart regeneration peaking at 55 dpc and the 85% (67/79) prevalence of PCR-positive heart samples collected in the transmission studies at 47 dpc from all the fish groups (Tanks 7-10) without any cohabitant fish added.

The histopathological changes of the heart and pancreas induced by SAV2 at 55 and 84 dpc were much milder than those registered in the previous SAV3 studies (11, 12). At the same time, the lesion scores in red and white muscles were similar to what was observed in this study. Interestingly, the loss of pancreas tissue at 55 dpc was mostly Grade 1 levels and present in ≤40% of the samples from all groups. At 84 dpc, the pancreas tissue loss scores were reduced by more than half compared to 55 dpc. This is in stark contrast to the markedly more severe and progressive development of pancreas tissue loss in the earlier SAV3 studies. Loss of pancreatic tissue has been suggested to be an important factor causing the growth loss associated with PD (11, 12). This may explain why the growth difference between the DNA-PD and the other groups at 84 dpc, albeit significant, is not as large as in similar treatment groups in earlier SAV3 efficacy studies. This further supports previous studies reporting SAV2 as less virulent than SAV3 (7, 42).

Despite a low prevalence of detectable SAV2 infection in the challenge Tank A at 19 dpc in 3 of the 4 treatment groups, all the pre-challenged fish groups residing in the 10 tanks were infected and developed varying levels of changes in internal organs typical of PD during the 26-day transmission studies. At 21 dpc, water in 7 of the 10 tanks contained SAV RNA ranging from 2.23_Log10 to 5.8_Log10 copies/L. At this timepoint, the levels of SAV RNA in the water containing the DNA-PD group were below the detection limit (0) in two of the tanks (1 and 5) and very low (170 copies/L) in tank 7. These levels were significantly lower than measured in the tanks holding the Control and Saline groups while not significant (p=0.08) compared to the Oil-PD group.

At 34 dpc (=13 dpe), SAV RNA was detected in the water of all 10 tanks, with similar levels (4.08_Log10 to 5.28_Log10 copies/L) within and between the replicate tanks containing the Oil-PD, Control and Saline groups. The corresponding levels for the DNA-PD group were, in contrast, highly variable ranging from 6.03_Log10 in Tank 1 (including naïve cohabitants, TS1) to 2.59_Log10 and 3.05_Log10 SAV RNA copies/L in Tanks 5 (with immunized cohabitants, TS2) and 7 (without any cohabitants), respectively. Of the 20 naïve cohabitant fish sampled from Tanks 1-4 (5 fish/tank, TS1) at 13 dpe, 18 were viremic and 17 with PCR-positive hearts. The results of the heart and pancreas pathologies confirm shedding and effective transmission of the virus to naïve fish in Tanks 1-4 irrespective of the vaccination status of the pre-challenged fish groups.

Of the 5 DNA-PD vaccinated cohabitant fish in Tank 5 (TS2) sampled at 13 dpe, none were viremic, their hearts negative by PCR and key histopathology indicators (pancreas necrosis, tissue loss and heart necrosis) largely absent. Of the 5 Oil-PD cohabitant fish in Tank 6 sampled at 13 dpe (TS2), most were both viremic (3/5), with PCR-positive hearts (4/5) and key histopathological indicator showing early onset of PD. These findings corroborate with lower virus levels in water in tanks containing the DNA-PD group (2.59_Log10 and 3.05_Log10 RNA copies/L), indicative of less viral shedding, concomitantly with PD-DNA vaccinated cohabitant fish being better protected against viral challenge, or a combination of these factors.

At 47 dpc (=26 dpe), SAV RNA was again detected in the water of all 10 tanks, but at much lower levels than the earlier timepoint, suggesting that the viremic phases had largely culminated in all fish groups. This finding is supported by the extensive reduction in prevalence of viremia in the naïve cohabitants in Tanks 1-4 (TS1) from 95% (19/20) at 13 dpe to 10% (6/60) at 26 dpe. In contrast, all the hearts of the 60 naïve fish sampled at 26 dpe were PCR positive with Cq-levels ≤28. Of these, the SAV RNA levels in the Control group were significantly lower than those measured in the DNA-PD and the Oil-PD groups, but not the Saline group. The reason for this finding remains unknown but may be due to varying kinetics of this parameter between groups at given timepoints. However, the prevalences of heart necrosis, pancreas necrosis and loss of pancreas tissue at 26 dpe in the naïve cohabitant cohorts of >90%, ≥40% and >30%, respectively, confirms the successful transmission of PD to these cohorts irrespective of the immune status of the ‘donor group’.

Of the immunized cohabitant fish sampled 26 dpe (47 dpc), the virus was cultured from 7% (1/15) of the fish in the Oil-PD group, while no viremic fish were found from the DNA-PD group. The PCR results from the hearts of these two groups revealed a highly significant difference. The fish in the DNA-PD group were all PCR-negative while 93% (14/15) of the hearts of the Oil-PD group were PCR-positive with Cq-values ranging from 30 to 18 with a median of 23. When coupled with the heart necrosis, pancreas necrosis and loss of pancreas tissue largely absent or low in the DNA-PD immunized cohabitants at 26 dpe ( Figures 9 , 10 ), the results demonstrate that these cohorts, in contrast to their Oil-PD counterparts, effectively resisted against the SAV2 challenge pressure endured in Tank 5. The discrepancy between virus culture-positive and PCR-positive fish in the Oil-PD group might align with an understanding that SAV forms defective viral genomes during replication in vivo (43, 44).

When compared to the PD levels that developed in the naïve cohabitants residing with the pre-challenged DNA-PD (Tank 1), the results signify the value of ‘herd immunity’ in the aquatic environment when employing efficacious vaccines which can inhibit disease transmission and thus, curb the spread of an epizootic.

Water was sampled at three timepoints to quantify SAV RNA shed by the different treatment groups. As such, these measurements offer only snapshot overviews, limiting insights into the dynamic process of viremia that may vary with time based on the relative protection level provided by the vaccine. Future efforts should consider more frequent analysis of SAV RNA in water to better understand the viremia phase associated with PD.

The results of the Efficacy and transmission studies are complementary, demonstrating that the DNA vaccine employed significantly reduces the impact of waterborne SAV2 infection in the presence of susceptible cohorts and further inhibits the spreading of the virus when the neighboring cohorts have been immunized with the same vaccine. Conversely, the fish in the Oil-PD group neither demonstrated significant protection against PD, nor the capacity to curb further transmission of SAV2 to similarly vaccinated cohabitant fish.