Antioxidant nutrition in Atlantic salmon (Salmo salar) parr and post-smolt, fed diets with high inclusion of plant ingredients and graded levels of micronutrients and selected amino acids

Introduction

The shift from marine to plant-based ingredients in fish feeds leads to changes in dietary concentrations and bioavailability of several nutrients. So far, research on plant-based diets for Atlantic salmon has mainly focused on balancing the amino acid profile and reducing the amounts of antinutrients (Espe et al., 2006; Espe et al., 2007; Krogdahl, Hemre & Mommsen, 2005), while micronutrients have received far less attention. In the present study, two feeding experiments were conducted on salmon, one with parr in freshwater and one with post-smolt in seawater, using a basic diet containing 10% fishmeal and 3.5% fish oil of the total recipe, in addition to plant-based protein and lipid sources. A nutrient premix (NP), containing potentially critical micronutrients, amino acids, taurine and cholesterol, was added at graded levels to this diet and health indicators and biomarkers were used to estimate optimal supplementation of the different nutrients. Hemre et al. (2016) presents data on general fish performance, retention of macronutrients, B-vitamins and amino acid metabolism. Data on selected minerals and vitamin A, D and K are presented by E.-J. Lock, 2016, unpublished data. The present publication is related to antioxidant nutrients and redox regulation.

Vitamin E is the general term for a group of lipid soluble antioxidants, tocopherols and tocotrienols, where α-tocopherol (α-TOH) has the highest biological activity (United States Pharmacopeia 1993). α-TOH is selectively retained in the body of mammals (Kayden & Traber, 1993; Rigotti, 2007) and Atlantic salmon (Hamre, 2011; Hamre, Berge & Lie, 1998), possibly due to a liver TOH binding protein, with high affinity for this isomer (Kayden & Traber, 1993; Yoshida et al., 1992). Vitamin E breaks the chain of lipid peroxidation by reducing fatty acid peroxide radicals, preventing oxidation of new fatty acids (Buettner, 1993). It takes approximately three months for Atlantic salmon parr to adjust their body concentration of α-TOH to the dietary concentration and there is a linear relation between dietary and whole body concentrations when dietary α-TOH ranges between 0 and 300 mg kg⁻¹. The slope is quite reproducible between experiments, at approximately 0.25 between fish concentration on wet weight and feed concentration on dry weight (Hamre et al., 1997; Hamre & Lie, 1995). Vitamin E is added as α-tocopheryl acetate (α-TOAc) in fish feed, thereby protected against oxidation. The requirement is highly variable, in rainbow trout (Oncorhynchus mykiss) dietary requirements between 5 and 100 mg kg⁻¹ have been reported, dependent on dietary composition and experimental conditions. Factors that affect the vitamin E requirement are dietary polyunsaturated fatty acid (PUFA), vitamin C and Se concentrations and oxidized feed (Hamre, 2011). Furthermore, the estimated requirement varies with the response variable used. Hemoglobin (Hb) concentration is a sensitive indicator of overall health and growth performance while indicators of immune function respond at higher dietary levels of vitamin E and will often give very high requirement assessments (Hamre, 2011). In the present study, γ-TOH is used as a representative of the non-α-TOHs, since it is often present in plant oils at high concentrations.

In classical nutrition, vitamin C (ascorbic acid (Asc)) has a well defined role as a cofactor for the enzymes catalyzing the hydroxylation of proline and lysine, necessary for formation of collagen and bone matrix (Barnes & Kodicek, 1972; Gould et al., 1960; Terova et al., 1998). The function of Asc is to keep the iron present at the enzymes’ active site in the reduced state (Meister, 1994). More generally, Asc is a water-soluble antioxidant, it scavenges free radicals and probably participates in recycling oxidized vitamin E (Hamre et al., 1997; Tappel, 1962). The resulting Asc radical or dehydroascorbic acid can in turn be reduced by glutathione (GSH) and ultimately by NADPH produced in energy metabolism (Mãrtensson & Meister, 1991). It was not until stable and fully bioavailable forms of vitamin C were developed that meaningful requirement studies in fish could be designed (Woodward, 1994). Sandnes, Torrissen & Waagbø (1992) performed one of the first of such studies in Atlantic salmon using Ca ascorbate-monophosphate (AP). They found a minimum dietary requirement, based on growth, mortality and skin and backbone hydroxyproline concentration, of 10–20 mg kg⁻¹. Severe deficiency symptoms appeared first after 18 weeks of feeding the non-supplemented diet. Concentration of Asc in the liver was linearly related to dietary AP up to 160 mg kg⁻¹ Asc equivalents. The appearance of vitamin C deficiency symptoms depends on the dietary concentration of vitamin E and is probably affected by other variations in the experimental conditions as well (Dabrowski et al., 2004; Gabaudan & Verlhac, 2001; Sandnes, Torrissen & Waagbø, 1992). It is common for several fish species that dietary Asc required for maximum body or tissue storage surpasses the requirement levels for growth, survival and hydroxyproline concentrations (NRC, 2011). Furthermore, immune response indicators are stimulated by Asc levels far above conventional requirements, which also protect against stress (Trichet, 2010; Waagbø, 1994; Waagbø, 2006).

Selenium (Se) is an essential trace element inserted in selenocysteine (Sec), situated at the active site of Se dependent proteins, termed selenoproteins (Brigelius-Flohé, 1999). The selenoproteins can be grouped in stress-related and housekeeping proteins, the first group responds readily to dietary Se and includes GSH peroxidase (GPX) 1 and 3 (Penglase et al., 2014). In mammals, the dietary Se level at which GPX1 activity and gene expression level off, is used as an indicator of the Se requirement (Sunde et al., 2009; Weiss et al., 1996; Weiss et al., 1997). GPX1 activity in rainbow trout plasma followed a similar trajectory to that in mammalian tissues and leveled off at dietary Se levels (given as Na₂SeO₃) between 0.15 and 0.38 mg kg⁻¹ (Hilton, Hodson & Slinger, 1980). However, in zebrafish, maximum growth correlated with minimum whole body GPX1 activity and mRNA expression at 0.3 mg kg⁻¹ dietary Se. Dietary Se above this level reduced growth and increased GPX1 activity and expression (Penglase et al., 2014). This may indicate that the relation between GPX1 and Se requirement is different in fish and mammals. There was a linear relationship between dietary Se in the range 0.1–1.0 mg kg⁻¹ and whole body Se concentrations in the zebrafish study (Penglase et al., 2014).

The cellular redox environment affects the cell’s fate because gene expression, protein function and molecular pathways are often redox sensitive (Go & Jones, 2013; Jones & Sies, 2015; Kirlin et al., 1999). The key mechanism is that protein cysteine residues switch between oxidized and reduced states corresponding to active and inactive states of the involved proteins (redox switches). An example is that a more oxidized cellular environment induces a redox switch that releases the nuclear factor-erythroid 2-related factor 2 (NFE2L2) transcription factor from a complex with another protein, Kelch Like ECH Associated Protein 1 (KEAP1). Once released, NFE2L2 induces the transcription of at least 50 mammalian genes; many of which code for antioxidants, thiol oxidoreductases and GSH synthesis/recycling genes; that are involved in maintaining the redox balance and/or are involved in redox signaling (reviewed by Ma, 2013). Cellular redox homeostasis is thought to be maintained by redox couples that act as electron buffers due to their ability to readily cycle between oxidized and reduced forms (Jones & Sies, 2015). The major redox couples are reduced/oxidized glutathione (2GSH/GSSG), cysteine (2Cys/CySS) and thioredoxin (Trx(SH)₂/TrxSS) (Huseby, Sundkvist & Svineng, 2009) and the related redox potential (E) is proportional to the ratio between the (reduced)² and oxidized forms of the redox couples, according to the Nernst Equation. GSH/GSSG is present in cells in high concentrations. It may therefore be the most important cellular redox couple and is used as an indicator of tissue redox state in the present study. The GSH based redox potential seems to be strictly regulated in Atlantic salmon (Hamre et al., 2010). A hypothesis of the present study is therefore that GSH/GSSG concentrations and E_GSH are stable in healthy fish and can be used as indicators of fish welfare during the growth phase in salmon. On the other hand, the 2GSH/GSSG redox couple and many genes coding for proteins that maintain the cellular redox system are dynamically regulated during fish embryonic and larval development (Hamre et al., 2014; Penglase et al., 2015; Skjærven et al., 2013; Timme-Laragy et al., 2013).

Here we present data on antioxidant nutrients and redox regulation of two feeding trials (freshwater and seawater) with graded NP levels. Tissue concentrations and retention of vitamin C, E and Se, traditionally viewed as the main antioxidant nutrients, as well as effects of a graded NP on tissue GSH/GSSG concentrations and the related redox potentials, were analyzed. We have also measured the expression of some central genes for regulation of redox homeostasis in the liver; the transcription factor nfe2l2, cuznsod, mnsod, cat, gpx1 and gpx3 which metabolize superoxide anions and H₂O₂, gclc which translates into the rate limiting protein in GSH synthesis, and g6pd and gr involved in keeping GSH in the reduced state.

Materials and Methods

Results

Fish performance is reported in detail by Hemre et al. (2016). Briefly, fish in Trial 1 grew from an initial weight of 18.3 g (± 2.2) to a range of 78.6 g (± 1.9) to 87.3 g (± 4.5). Both fish growth and protein retention increased with increasing dietary NP, while lipid retention decreased, together with liver index and viscera-somatic index. In Trial 2, initial fish size was 228 g (± 4.2) and average final weight was 482 g (± 17). There was no effect of the NP on growth or protein and lipid retention in this trial. Survival was high in both trials, close to 100%, and with no difference between diet groups.

The dietary concentrations of α-TOH, Asc and Se (Fig. S1) were well fitted to first order polynomials in Trial 1 (R² > 0.98). In Trial 2, the diet designated 400% was an outlier, while the other diets showed a good fit to a linear equation (R² > 0.97). The diet planned to contain 400% NP in Trial 2 was recalculated to contain 239% NP, based on the nutrient analyses of the diets. γ-TOH level was similar in all diets, as it was only derived from the feed ingredients (Fig. S1). The dietary concentrations at 100% NP were 118 and 141 mg kg⁻¹ for α-TOH, 67 and 63 mg kg⁻¹ for Asc and 0.62 and 0.79 mg kg⁻¹ for Se, in Trials 1 and 2, respectively. At 0% NP the dietary concentrations were 48 and 76 mg kg⁻¹ for α-TOH, 4.7 and < 5.5 mg kg⁻¹ for Asc and 0.42 and 0.47 mg kg⁻¹ for Se (Table 2).

There was a linear relationship between dietary and whole body concentrations of α-TOH in Trial 1 (R² = 0.94; p (slope = 0) < 10⁻⁴; Fig. 1). In Trial 2, a second order polynomial fitted the data better than a first order polynomial (p = 0.03). The range of whole body concentrations was between 20 and 60 mg kg⁻¹ in Trial 1 and between 30 and 50 mg kg⁻¹ in Trial 2. Whole body γ-TOH concentration varied between 5 and 10 mg kg⁻¹, with a significant positive slope in Trial 1 (p = 0.02) and no effect of diet in Trial 2 (Fig. 1). Whole body Asc concentration followed a second order polynomial in both trials (Trial 1, R² = 0.92, p < 0.0001; Trial 2, R² = 0.88, p = 0.016). Whole body concentration plateaued at a dietary concentration of Asc of 190 mg kg⁻¹ DM and a whole body concentration of 39.5 mg kg⁻¹ in Trial 1. In Trial 2, the plateau was reached at dietary and whole body concentrations of 100 and 14 mg kg⁻¹, respectively. Whole body Se showed a linear relationship with the dietary concentration in both trials (R² > 0.93, p (slope = 0) < 10⁻⁴). The diet dependent whole body concentration was significantly higher in Trial 1 than in Trial 2 (p < 10⁻⁴).

The retention of α-TOH was not affected by diet in Trial 1 and followed a second order polynomial in Trial 2 (p = 0.02) with the highest retentions at intermediate dietary α-TOH concentrations (Fig. 2). Diet did not affect the retention of γ-TOH in Trial 1, while increasing the NP in Trial 2 had a negative effect on γ-TOH retention (p < 0.004). The range of retention of both α- and γ-TOH was 20–30%. In Trial 1, retention of Asc ranged from 11–42% and was negatively linearly related to dietary Asc (R² = 0.34, p (slope = 0) = 0.02). In Trial 2, vitamin C retention was negative at dietary levels below 63 mg kg⁻¹ and max retention was 16% at 89 mg kg⁻¹. A logarithmic function was found to have the best fit to the data (R² = 0.92). Diet did not affect retention of Se, which ranged from 20 to 30%. The retention of Se was higher in Trial 2 than in Trial 1 (p < 10⁻⁴).

GSH and GSSG concentrations were measured in muscle (Fig. 3) and liver (Fig. 4). In muscle in Trial 1, GSH concentrations ranged from 100 to 200 μmol kg⁻¹, GSSG ranged from 0 to 0.4 μmol kg⁻¹ with no effect of diet. There was a tendency that the resulting GSH based redox potential followed a second order polynomial at dietary NP below 150%, so that the tissue became transiently more reduced at supplementation of 25–100% NP than at 150–400% NP (p = 0.058). The redox potential of muscle in Trial 1 ranged from −240 to −190 mV. In Trial 2, muscle GSH and GSSG concentrations ranged from 43 to 80 and 0 to 0.4 μmol kg⁻¹, respectively, and the redox potential from −170 to −220 mV, with no effect of diet. The muscle GSH/GSSG concentrations in these fish were close to quantification limits and some of the data points were removed as outliers. Muscle GSH concentration was higher (p < 0.0001), the redox potential lower (p = 0.002) and GSSG concentration similar in Trial 1 compared to Trial 2. In the liver, GSH, GSSG and the GSH based redox potential were not affected by the diet, however there was a tendency towards a significant second order fit to the data on liver redox potential in Trial 1 (p = 0.066), implicating a higher redox potential in fish fed diets with intermediate NP additions. Liver GSH concentration was higher (p < 0.0001), while GSSG concentration (p = 0.01) and the redox potential (p < 0.0001) were lower in Trial 1 than in Trial 2.

Muscle TBARS (Fig. 5) was negatively correlated to NP addition in both trials (p = 0.004 in Trial 1, p = 0.03 in Trial 2) and the absolute level was similar in the two trials (p = 0.09).

Expression of redox dependent genes was measured in the liver of individual salmon in both trials. PCA plots showed that the dietary groups were not well separated according to expression of the redox related genes (Fig. 6). However, in Trial 1 there was a tendency to grouping of samples related to dietary supplementation of NP, with NP supplementation correlating negatively to g6pd, gclc, gpx1 and gr (−0.38 < R² < −0.43) and positively to gpx3 (R² = 0.39). Gr showed the most covariation with other genes and correlated to g6pd, gpx1 and gclc at 0.79 > R² > 0.49 and to mnsod and cuznsod at R² = 0.39 and 0.34, respectively. Gr was also correlated to nfe2l2 (R² = 0.52), cat was not correlated to the other genes while gpx3 was negatively correlated to the other genes, significant for g6pd (R² = −0.33). In Trial 2, diet was correlated to cuznsod, mnsod and cat (0.39 < R² < 0.24). The expression of gr, g6pd, gpx1 and gclc comprised one distinct group and mnsod, cat and cuznsod another. Gr correlated with g6pd, gpx1 and gclc at 0.67 > R² > 0.37, while gr correlation to mnsod, cat and cuznsod was 0.32 > R² > 0.25. Gpx3 and nfe2l2 were not significantly correlated to the other genes.

Relations between diet and gene expression are given in Fig. S1. In Trial 1, increasing supplementation of NP gave a reduction in gclc, g6pd, gpx1, gr and mnsod expression, while nfe2l2, cat and cuznsod expression were unaffected by diet. In Trial 2, data on nfe2l2 and gclc were fitted to second order polynomials and had higher expression at intermediate NP supplementation. Gpx1 showed the best fit when using a third order polynomial with peak expression at 50 and 100% NP inclusion. Gpx3 had higher expression at low NP inclusion, while mnsod expression increased with increasing NP. Cuznsod expression was slightly lower at intermediate NP inclusion than at high and low inclusion, while G6pd and gr expression were unaffected by the diet. Average expression of gpx1, gpx3 and mnsod was higher in Trial 2 than in Trial 1 (p < 0.0002). Increasing NP inclusion had opposite effects on mnsod expression in the two trials, which decreased in Trial 1 (p = 0.04) and increased in Trial 2 (p = 0.002).

Discussion

The present study showed differences in redox status between parr and post-smolts and effects of supplementation of the NP on redox regulation. However, there were no clear deficiencies of the antioxidant nutrients, vitamin C, vitamin E and Se, even in fish fed the unsupplemented diet. The reason may be that the plant-based feed ingredients contained relatively high levels of vitamin E and Se and that the experimental periods were too short for fish of these body sizes to develop vitamin C deficiency symptoms. The basal feed contained less than 5.5 mg kg⁻¹ Asc, which is below the minimum requirement in Atlantic salmon of 10–20 mg kg⁻¹; however, it took 18 weeks to deplete first-feeding Atlantic salmon for Asc (Sandnes, Torrissen & Waagbø, 1992) and a longer period of time may be needed to deplete larger fish sizes as the post-smolts. The final concentrations of α-TOH and Se in the basal feed were at or above minimum requirements at approximately 50 and 0.3 mg kg⁻¹ diet, respectively (NRC, 2011). Plant oils often have high concentration of tocopherols, however tocopherols are unstable during feed processing and storage and concentrations could vary considerably in the finished feeds (Hamre, Kolås & Sandnes, 2010; Olsvik et al., 2011). Se concentrations in plant ingredients vary according to Se concentration in soil (Alfthan et al., 2015). Therefore, even if α-TOH and Se may be sufficient in some plant ingredients, they should be added to fish feeds to ensure safe supplementation. Commercial fish feeds are commonly supplemented with higher levels of antioxidant nutrients than the minimum requirements given by NRC (2011), which will protect the fish in periods with oxidative stress, such as high water temperatures (Martínez-Álvarez, Morales & Sanz, 2005) elevated water oxygen levels (Lygren, Hamre & Waagbø, 2000), vaccination (Lygren, Hjeltnes & Waagbø, 2001) and disease (Trichet, 2010; Waagbø, 2006). Accordingly, the diets with 100% NP used in this study, contained 118–141 mg kg⁻¹ α-TOH, 63–67 mg kg⁻¹ Asc and 0.62–0.79 mg kg⁻¹ Se, respectively.

The most sensitive biomarker of vitamin E deficiency in salmonids is lowered hemoglobin (Hb) concentration (Hamre, 2011). There was a slight positive correlation between NP addition and Hb concentration in fish from Trial 1 (p (slope = 0) = 0.03) (Hemre et al., 2016), however, the slope was not significantly different from 0 after omitting the 400% NP group (p = 0.44), so fish fed the diets with low NP supplementation were clearly not vitamin E deficient. In Trial 2, there was no effect of diet on Hb (Hemre et al., 2016). Hamre & Lie (1995) found that mortality due to vitamin E deficiency commenced at whole body levels of α-TOH of 5 mg kg⁻¹ wet weight, while the whole body concentration in fish fed the basal diet in this study were 18 and 32 mg kg⁻¹ in Trials 1 and 2, respectively. Accordingly, none of the fish groups in this study experienced vitamin E deficiency, even though the low level of vitamin C in the diet without NP would make the fish more susceptible to low dietary vitamin E (Hamre et al., 1997).

The diets were fed for 12 and 22 weeks in Trials 1 and 2, respectively. The twelve weeks feeding period in Trial 1 should just be enough for the fish to adjust to the α-TOH concentrations of the diet according to previous reports (Hamre, 2011; Hamre & Lie, 1995). However, the slope of whole body concentration to dietary concentration was lower (0.13 vs 0.25) and the intercept with the y-axis was higher (15.6 vs 0) than previously reported, indicating that the steady state whole body α-TOH concentration had not yet been obtained (Hamre, 1995). A relation of body to dietary concentrations of α-TOH has until now only been measured in very young fish and these relationships were always linear (Hamre et al., 1997; Hamre & Lie, 1995). In post-smolts in the present study, the relationship followed a second order polynomial. It is possible that uptake of α-TOH is changing over development, moving from a strict linear relationship just after first-feeding to a saturable mechanism at later stages. If this is the case, the diet 100% NP with 141 mg kg⁻¹ α-TOH would be sufficient to saturate the system, while supplementation above this gave a decreased retention. In summary, supplementing commercial diets with above 150 mg kg⁻¹ α-TOH is recommended to promote fish health and welfare. A fish feed surveillance program run by NIFES in 2015 reported 185–320 mg kg⁻¹ α-TOH in 18 commercial salmon feeds, while two feeds contained as high as 570 mg kg⁻¹ (M. Sanden, 2016, unpublished data) A variable part of this α-TOH would have been derived from the feed ingredients. One have to take into account that α-TOH in the feed ingredients is in the free form, which is easily broken down in salmon intestine and therefore has quite low retention (Meland, 1995). It is therefore a risk involved in including ingredient contribution when deciding dietary α-TOH supplementation.

The non-α-TOHs cannot substitute for α-TOH, because of their lower biological activity (Hamre, 2011; Kayden & Traber, 1993), however they may have specific biological functions in cell signaling (Golli & Azzi, 2010; Wallert et al., 2014). In both trials of this study, the retentions of α- and γ-TOH were similar at 20–30%. Due to lower biological activity, retention of γ-TOH should be lower than α-TOH, but the result can be explained by similar retention of α- and γ-TOH in muscle and adipose tissue, which are the main constituents of the whole body sample. Retention of γ-TOH in other vital organs of salmon, such as gonads, intestine, and gills, is less than 30% compared to α-TOH (Hamre & Lie, 1997).

The whole body concentration of Asc followed a second order equation with respect to Asc supplementation in both trials. It is well known that tissue levels of Asc level off at dietary supplementation that is much higher than the levels where traditional deficiency symptoms are displayed (Gabaudan & Verlhac, 2001; Waagbø & Sandnes, 1996). Since Asc above this minimum requirement has positive effects on immune function, stress resistance and reproduction (Gabaudan & Verlhac, 2001; Waagbø, 1994; Waagbø, 2006), it may be fruitful to supplement more of the vitamin. The whole body concentration levelled off at a dietary Asc concentration of 190 mg kg⁻¹ in Trial 1 and at 63–89 mg kg⁻¹ in Trial 2. The maximum body concentration of Asc in fish from Trial 2 was approximately 30% of that in fish from Trial 1. While the data from Trial 1 follow the expected trajectory based on the literature, the fish in Trial 2 seems to have had an extraordinary high consumption of Asc, with negative retention for fish supplemented below 63 mg kg⁻¹ and maximum retention of 16% at higher supplementation, compared to max 42% retention in Trial 1.

There is probably an optimal ratio between supplementation of α-TOH and Asc, based on the hypothesis that vitamin E is recycled by vitamin C. A buildup of tocopheroxyl radicals would occur if too little vitamin C is present to recycle it, which would have a pro-oxidant effect on the tissues (Hamre et al., 1997). An indication of this principle is the higher mortality due to vitamin C deficiency in fish fed 300 compared to 150 mg kg⁻¹ vitamin E (Hamre et al., 1997). Another indication is that vitamin E and C in tissues from copepods and wild Ballan wrasse are 100 and 500 mg kg⁻¹ dry matter and 9 and 18 mg kg⁻¹ wet weight, respectively (Hamre et al., 2013; Hamre et al., 2008). Theoretically, one should supplement more Asc than vitamin E, although the exact optimal ratio has not yet been identified. Seventeen commercial salmon feeds analyzed in a fish feed surveillance program performed at NIFES in 2015, contained between 118 and 418 mg kg⁻¹ Asc, one feed contained 800 and two feeds contained more than 1,300 mg kg⁻¹ Asc. The concentration of Asc in the feeds was frequently below the concentration of α-TOH (M. Sanden, 2016, unpublished data).

Whole body Se concentration followed a first order relationship with the dietary concentration both in parr and post-smolt, and the retention was not affected by NP supplementation. A similar relationship was found in zebrafish, but this species had higher whole body Se concentration at similar supplementation levels. The difference may be species dependent or perhaps due to the use of Se enriched yeast containing mainly selenomethionine as the Se source in the zebrafish (Penglase et al., 2014) and selenite (Na₂SeO₃) here (Lorentzen, Maage & Julshamn, 1994). The basal diet contained 0.4–0.5 mg kg⁻¹ Se, while the requirement in rainbow trout is estimated at 0.15–0.38 mg kg⁻¹ (Hilton, Hodson & Slinger, 1980). It is therefore unlikely that fish fed the diets used in this study would become Se deficient. Judged by retention efficiency, the Se present in the feed ingredients had similar bioavailability to the added selenite. If gpx1 expression had been affected by low Se in this study, increased NP would probably have increased this response (Hilton, Hodson & Slinger, 1980; Penglase et al., 2014). However, gpx1 was not expressed according to this hypothesis in either trial. Retention and body concentrations of Se were higher in seawater than in fresh water, possibly caused by differences in waterborne Se concentrations. Se toxicity was probably not encountered in the present study, since the diet with the highest NP inclusion contained 1.1 and 1.4 mg kg⁻¹ Se, which is below levels reported to be toxic in fish (Penglase et al., 2014).

To examine whether the graded dietary NP affected redox regulation in the salmon, we measured GSH/GSSG concentrations in liver and muscle and calculated the resulting E_GSH. In Trial 1, there was a tendency (p = 0.06–0.07) that NP supplementation below 150% gave more reduced fish, resulting from higher concentrations of GSH. Considering the above discussion, it is unlikely that this effect would have been caused by low levels of vitamin C, E or Se. The diets also had graded levels of taurine and methionine (Hemre et al., 2016) and at low levels of these nutrients, GSH synthesis or recycling and a resulting lowered E_GSH could have been stimulated to protect against oxidation induced by low taurine (Jong, Azuma & Schaffer, 2012; Penglase et al., 2015), consistent with the relatively higher liver gclc, g6pd and gr expressions at low NP inclusion. In Trial 2, there were no effects of diet on GSH/GSSG concentrations and no correlation between diet and expression of the genes coding for GSH metabolizing enzymes. The higher muscle TBARS at low NP inclusion in both trials confirm that fish fed low levels of micronutrients and selected amino acids may have become oxidized. Overall, the results confirm that the tissue GSH/GSSG concentrations are relatively stable within experiments with salmon, even with large variations in dietary composition, as found by Hamre et al. (2010).

However, the GSH/GSSG concentrations and E_GSH were different in Trials 1 and 2, the fish in Trial 2 being more oxidized, with less GSH both in liver and muscle and more GSSG in the liver, than the fish in Trial 1. Expression of liver gpx1 (cytosol and mitochondria) and gpx3 (extracellular) was also higher in Trial 2, indicating excess activity in removal of H₂O₂ at the expense of GSH. This corresponds with an increased consumption of Asc in Trial 2. These results indicate that the fish in Trial 2 experienced some sort of oxidative stress. The fish did not seem to be diseased, since the growth and feed intake were good and mortalities were negligible (Hemre et al., 2016). The feed was not analyzed for oxidation, however, oxidized feed most often leads to reduced vitamin E retention (Baker & Davies, 1997; Hung, Cho & Slinger, 1981), which was not encountered in the present study. The samples in Trial 2 were taken in June, with rapidly increasing water temperatures and longer day light, resulting in strong growth stimulation in salmon. In cod larvae fed copepods, a similar growth stimulation correlated with more oxidized fish having a higher whole body E_GSH (Karlsen et al., 2015; Penglase et al., 2015). Based on these observations, we hypothesize that redox regulation is involved in surges in growth.

Conclusions

This study gives no indications that diets rich in plant ingredients increase the vitamin C, E or Se requirements in Atlantic salmon. Feeds produced using the current feed ingredients, seem to have sufficient amounts of α-TOH and Se from the feed ingredients alone. However, due to possible variations in ingredient quality, feed processing and farming conditions, feeds should be supplemented with above 150 mg kg⁻¹ α-TOH equivalents. The Se in plant-based feed ingredients may vary and Se supplementation may be warranted in some cases. However, the legal upper limit of Se in fish feeds in the European Union is 0.5 mg kg⁻¹ (Regulation (EC) No 1831/2003) and the law comes into force if the feeds are supplemented. Fish fed the highest concentration of Se (1.4 mg kg⁻¹) did not show any signs of toxicity. Based on body concentrations, Asc supplementation should be above 190 mg kg⁻¹, where the whole body Asc concentration leveled off in Trial 1. At this supplementation level, Asc deficiency can be avoided in periods with oxidative stress and optimal immune function and stress resistance can be obtained. Redox regulation, which includes differential consumption of antioxidant nutrients, seems to change during the production cycle of Atlantic salmon.

Supplemental Information