Citrus flavonoids, β-Glucan and organic acid feed additives decrease relative risk during Yersinia ruckeri O1 biotype 2 infection of rainbow trout (Oncorhynchus mykiss)

Ethics statement

The Danish Animal Experiments Inspectorate, under license no. 2015-15-0201-00645 approved the protocols regarding experimental animals described for this study. The study is thus approved under the Danish law regarding experimental animals.

Rainbow trout

Rainbow trout eggs were acquired from AquaSearch Ova (AquaSearch FRESH, Billund, Denmark, 100% females). Following disinfection with Desamar K30 (Foodtech AG, Uster, Schweiz) according to the manufacturer’s instructions, the eggs were hatched and reared under pathogen-free conditions at the Bornholm Salmon Hatchery (Nexø, Denmark). During the rearing period, the fish were fed commercial pelleted feed.

Experimental feeding

Prior to experimental feeding, the fish were transported to the BioMar Research facility (Hirtshals, Denmark), divided into a total of 30 tanks with a volume of 150 l, each holding 29–32 individuals and allowed to acclimatize for 14 days. Once acclimated, each tank was assigned to one of five experimental feed groups resulting in 6 tanks/dietary group. Prior to the initiation of the experimental work, each feed group was anonymized, and the experiment was performed as a single-blinded setup. This facility has previously been described in detail (Ohtani et al., 2019b). In brief, all tanks were supplied through a closed recirculating water system. The water temperature was 14−15 °C, and filtering was performed through physical filtering, as well as a biofilter. The water was oxygenated and passed through an UV-filter (595 µJ/cm²). At the start of the experimental feeding period, the fish had an average weight of 24.4 g (bulk weights divided by number of individuals for all tanks).

Each experimental feed type was based on the same proprietary base feed (included as control), and supplemented as described in Table 1 to produce group specific two mm pellets. As all supplemented feeds are proprietary, only general descriptions are given. All pellets were produced by BioMar A/S, and the feeding period was 37 days (1.5% bodyweight/day).

As a technical error affected waterflow and oxygen supply to one tank in the control feed group, this tank was eliminated from the experiment and downstream analysis.

Experimental feed performance

The performance of each feed group was evaluated based on weight gain (WG), feed conversion ratio (FCR), specific growth rate (SGR), lipid efficiency ratio (LER and protein efficiency ratio (PER). The equations used are given below:

1) ${\displaystyle WG=\frac{\left(Finalbiomass\left(g\right)-Initialbiomass\left(g\right)\right)}{Initialbiomass\left(g\right)}\times 100\text{\%}}$

2) ${\displaystyle FCR=\frac{Totalconsumedfeed\left(g\right)}{Finalbiomass\left(g\right)-Initialbiomass\left(g\right)+Deadbiomass\left(g\right)}}$

3) ${\displaystyle SGR=\frac{\left(Ln\left(Finalbiomass\left(g\right)\right)\right)-Ln\left(Initialbiomass\left(g\right)\right)}{Growthperiod\left(days\right)}\times 100\text{\%}}$

4) ${\displaystyle LER=\frac{Gainedbiomass\left(g\right)}{Ingestedfat\left(g\right)}}$

5) ${\displaystyle PER=\frac{Gainedbiomass\left(g\right)}{Ingestedprotein\left(g\right)}}$

All calculations are made based on tank-wise bulk weighing of total biomass prior to and following the experimental feeding period. Administered feed was recorded for each tank throughout the experimental period and ingested fat/protein is calculated based on total ingested feed and the content of each component in each feed type.

Experimental infection

Following the initial feeding period, all individuals from four of the six tanks in each experimental feed group were subjected to waterborne experimental bath infection with Y. ruckeri. The individuals from the remaining two tanks were subjected to mock-infection. Experimental feeding continued throughout the infection period.

The bath infection was performed as previously described (Ohtani et al., 2019b). Briefly, cryopreserved Y. ruckeri serotype O1 biotype 2 (strain 07111224) was cultured on blood agar plates for 48 h. Single colonies were then used to inoculate Luria-Bertani broth, which was subsequently incubated at room temperature for 36 h. Prior to infection, the bacteria were harvested by centrifugation and resuspended in clean tank water. Finally, fish from a given tank destined for infection were transferred to a designated infection tank with clean tank water. The infection was then started by the addition of bacterial suspension to a final concentration of 7.5 ×10⁸ CFU/ml. The fish were kept in their respective infection tanks for 3 h, after which they were returned to their holding tanks and monitored closely for 22 days. Mock-infections were performed by transferring fish to separate mock-infection tanks holding clean tank water for 3 h, before returning them to their respective holding tanks, thus mimicking the handling of the infected fish. Experimental feeding continued throughout the infection period, however reduced to 1.1% biomass/day. During the experimental infection period, all fish that met specified humane endpoints (distinct signs of established disease such as loss of equilibrium, protrusion of either or both eyes, haemorrhages along fin bases) were considered moribund, netted and euthanized with an overdose of benzocaine in accordance with the experimental animal license. The terms “survival”, and consequently “mortality”, in the following text reflect the binary nature of survival analyses and should therefore be considered technical, rather than purely descriptive terms. Following euthanasia, a swab was made from the anterior kidney onto 5% blood agar plates to re-isolate the bacterial pathogen in order to satisfy Koch’s postulates.

Statistical analysis

Throughout the analyses in the present study, a 95% confidence level was applied using a threshold α-level of 0.05 for rejection of the null-hypothesis that there is no difference between the groups in question. All statistical analyses were performed using GraphPad Prism^® for Mac (GraphPad, San Diego, USA).

Performance data were tested for underlying Gaussian distribution using the Kolmogorov–Smirnov test with Dallal-Wilkinson-Lilliefor correction. Having confirmed this, the means of each experimental group were compared to that of the control feed group using a one-way ANOVA followed by Dunnett’s post hoc test multiple comparisons test.

Data from the experimental infection were analysed using the Kaplan–Meier survival analysis tool. Moribund fish from which the bacterial pathogen could be re-isolated were computed as mortalities. In case the pathogen could not be re-isolated, the individual was computed as a censored individual. Based on the course of the survival curves, the log-rank (Mantel-Cox) method was chosen for the analysis. Hazard ratios (log-rank method) were calculated pairwise between all experimental groups, as well. Relative percent survival (RPS) was calculated as follows:

6) ${\displaystyle RPS=\left(1-\frac{Mortality\left(supplementedgroup\right)}{Mortality\left(controlgroup\right)}\right)\times 100\text{\%}}$

The Bonferroni correction for multiple comparisons was used when comparing individual survival curves, adjusting the α-level by dividing it with the number of comparisons made within the framework of the dataset.

Experimental feed performance

Feed performance was determined following the experimental feeding period, and the results are shown in Table 2.

No differences were observed between the control group and any of the experimental groups regarding WG, SGR or PER. The group receiving the β-glucan + OA supplement was found to have a statistically significantly lower FCR, as well as a higher LER relative to the control group (P < 0.05).

Experimental infection

Mortalities were observed in all infected tanks between 5- and 21-days post infection (Fig. 1). Log-rank survival curve analysis did not identify any statistically significant differences between replicate tanks in any of the groups (P > 0.05, Fig. S1). Consequently, data from replicate tanks were pooled for each group for all downstream analyses. The results from these analyses are summarized in Fig. 1 and Tables 2 and 3. Group-specific survival curves with 95% confidence intervals are shown in Fig. S2.

A single mortality occurred from one of the mock-infected tanks during the course of the experimental infection. However, given that only one occurred in a total of six tanks, this is considered negligible.

An initial, overall Log-rank survival curve analysis comparing all five survival curves at once did not identify any statistically significant differences between feed groups (P = 0.13). Pairwise comparisons between individual survival curves reached the same overall result, although offering a more nuanced view, as shown in Table 3. While the β-glucan, β-glucan + OA and CF supplemented groups all post P-values <0.05 when compared to the control feed group, correction for multiple comparisons reduces the α-level to 0.005. Their respective survival curves are therefore not significantly different from the control feed group. Comparisons among the supplemented groups resulted in high P-values (>0.5) indicating similar survival curves in these groups.

When subsequently comparing the relative risk of mortality following infection using the hazard ratios comparing the control group and each of the experimental feed groups in turn, all but the yeast cell wall supplemented feed group displayed hazard ratios with 95% confidence intervals <1 (see Fig. 2 and Table 4) These indicate statistically significant reductions in risk following infection relative to the control group (Motulsky, 2014).